From 93a3d05f9d68015f425c8f019c3ba1f489a0c0cd Mon Sep 17 00:00:00 2001
From: Michal Kalderon <michal.kalderon@marvell.com>
Date: Thu, 21 Nov 2019 13:29:57 +0200
Subject: RDMA/qedr: Add kernel capability flags for dpm enabled mode

HW/FW support two types of latency enhancement features.  Until now
user-space implemented only edpm (enhanced dpm).  We add kernel capability
flags to differentiate between current FW in kernel that supports both
ldpm and edpm.  Since edpm is not yet supported for iWARP we add different
flags for iWARP + RoCE.  We also fix bad practice of defining sizes in
rdma-core and pass initialization to kernel, for forward compatibility.

The capability flags are added for backward-forward compatibility between
kernel and rdma-core for qedr.

Before this change there was a field called dpm_enabled which could hold
either 0 or 1 value, this indicated whether RoCE edpm was enabled or
not. We modified this field to be dpm_flags, and bit 1 still holds the
same meaning of RoCE edpm being enabled or not.

Link: https://lore.kernel.org/r/20191121112957.25162-1-michal.kalderon@marvell.com
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/qedr/verbs.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index 4cd292966aa9..115f187f0c86 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -312,7 +312,18 @@ int qedr_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata)
 	}
 	ctx->db_mmap_entry = &entry->rdma_entry;
 
-	uresp.dpm_enabled = dev->user_dpm_enabled;
+	if (!dev->user_dpm_enabled)
+		uresp.dpm_flags = 0;
+	else if (rdma_protocol_iwarp(&dev->ibdev, 1))
+		uresp.dpm_flags = QEDR_DPM_TYPE_IWARP_LEGACY;
+	else
+		uresp.dpm_flags = QEDR_DPM_TYPE_ROCE_ENHANCED |
+				  QEDR_DPM_TYPE_ROCE_LEGACY;
+
+	uresp.dpm_flags |= QEDR_DPM_SIZES_SET;
+	uresp.ldpm_limit_size = QEDR_LDPM_MAX_SIZE;
+	uresp.edpm_trans_size = QEDR_EDPM_TRANS_SIZE;
+
 	uresp.wids_enabled = 1;
 	uresp.wid_count = oparams.wid_count;
 	uresp.db_pa = rdma_user_mmap_get_offset(ctx->db_mmap_entry);
-- 
cgit v1.2.3-59-g8ed1b


From 53bb802315b32bfb2a5a464dfc1f24b5610cad6d Mon Sep 17 00:00:00 2001
From: Selvin Xavier <selvin.xavier@broadcom.com>
Date: Mon, 25 Nov 2019 00:39:34 -0800
Subject: RDMA/bnxt_re: Report more number of completion vectors

Report the the data path MSIx vectors allocated by driver as number of
completion vectors. One interrupt vector is used for Control path. So
reporting one less than the total number of MSIx vectors allocated by the
driver.

Link: https://lore.kernel.org/r/1574671174-5064-7-git-send-email-selvin.xavier@broadcom.com
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/bnxt_re/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index e7e8a0f49464..793c97251588 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -677,7 +677,7 @@ static int bnxt_re_register_ib(struct bnxt_re_dev *rdev)
 
 	bnxt_qplib_get_guid(rdev->netdev->dev_addr, (u8 *)&ibdev->node_guid);
 
-	ibdev->num_comp_vectors	= 1;
+	ibdev->num_comp_vectors	= rdev->num_msix - 1;
 	ibdev->dev.parent = &rdev->en_dev->pdev->dev;
 	ibdev->local_dma_lkey = BNXT_QPLIB_RSVD_LKEY;
 
-- 
cgit v1.2.3-59-g8ed1b


From 4768820243d71d49f1044b3f911ac3d52bdb79af Mon Sep 17 00:00:00 2001
From: Yixian Liu <liuyixian@huawei.com>
Date: Tue, 10 Dec 2019 20:45:02 +0800
Subject: RDMA/hns: Simplify the calculation and usage of wqe idx for post
 verbs

Currently, the wqe idx is calculated repeatly everywhere it is used.  This
patch defines wqe_idx and calculated it only once, then just use it as
needed.

Fixes: 2d40788825ac ("RDMA/hns: Add support for processing send wr and receive wr")
Link: https://lore.kernel.org/r/1575981902-5274-1-git-send-email-liweihang@hisilicon.com
Signed-off-by: Yixian Liu <liuyixian@huawei.com>
Signed-off-by: Weihang Li <liweihang@hisilicon.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_device.h |  3 +-
 drivers/infiniband/hw/hns/hns_roce_hw_v1.c  | 37 +++++++++++--------------
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  | 43 ++++++++++++-----------------
 3 files changed, 35 insertions(+), 48 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 5617434cbfb4..416341ada172 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -423,7 +423,7 @@ struct hns_roce_mr_table {
 struct hns_roce_wq {
 	u64		*wrid;     /* Work request ID */
 	spinlock_t	lock;
-	int		wqe_cnt;  /* WQE num */
+	u32		wqe_cnt;  /* WQE num */
 	int		max_gs;
 	int		offset;
 	int		wqe_shift;	/* WQE size */
@@ -647,7 +647,6 @@ struct hns_roce_qp {
 	u8			sdb_en;
 	u32			doorbell_qpn;
 	u32			sq_signal_bits;
-	u32			sq_next_wqe;
 	struct hns_roce_wq	sq;
 
 	struct ib_umem		*umem;
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
index 2a2b2112f886..a31a21433f65 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
@@ -74,8 +74,8 @@ static int hns_roce_v1_post_send(struct ib_qp *ibqp,
 	unsigned long flags = 0;
 	void *wqe = NULL;
 	__le32 doorbell[2];
+	u32 wqe_idx = 0;
 	int nreq = 0;
-	u32 ind = 0;
 	int ret = 0;
 	u8 *smac;
 	int loopback;
@@ -88,7 +88,7 @@ static int hns_roce_v1_post_send(struct ib_qp *ibqp,
 	}
 
 	spin_lock_irqsave(&qp->sq.lock, flags);
-	ind = qp->sq_next_wqe;
+
 	for (nreq = 0; wr; ++nreq, wr = wr->next) {
 		if (hns_roce_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq)) {
 			ret = -ENOMEM;
@@ -96,6 +96,8 @@ static int hns_roce_v1_post_send(struct ib_qp *ibqp,
 			goto out;
 		}
 
+		wqe_idx = (qp->sq.head + nreq) & (qp->sq.wqe_cnt - 1);
+
 		if (unlikely(wr->num_sge > qp->sq.max_gs)) {
 			dev_err(dev, "num_sge=%d > qp->sq.max_gs=%d\n",
 				wr->num_sge, qp->sq.max_gs);
@@ -104,9 +106,8 @@ static int hns_roce_v1_post_send(struct ib_qp *ibqp,
 			goto out;
 		}
 
-		wqe = get_send_wqe(qp, ind & (qp->sq.wqe_cnt - 1));
-		qp->sq.wrid[(qp->sq.head + nreq) & (qp->sq.wqe_cnt - 1)] =
-								      wr->wr_id;
+		wqe = get_send_wqe(qp, wqe_idx);
+		qp->sq.wrid[wqe_idx] = wr->wr_id;
 
 		/* Corresponding to the RC and RD type wqe process separately */
 		if (ibqp->qp_type == IB_QPT_GSI) {
@@ -210,7 +211,6 @@ static int hns_roce_v1_post_send(struct ib_qp *ibqp,
 				       cpu_to_le32((wr->sg_list[1].addr) >> 32);
 			ud_sq_wqe->l_key1 =
 				       cpu_to_le32(wr->sg_list[1].lkey);
-			ind++;
 		} else if (ibqp->qp_type == IB_QPT_RC) {
 			u32 tmp_len = 0;
 
@@ -308,7 +308,6 @@ static int hns_roce_v1_post_send(struct ib_qp *ibqp,
 				ctrl->flag |= cpu_to_le32(wr->num_sge <<
 					      HNS_ROCE_WQE_SGE_NUM_BIT);
 			}
-			ind++;
 		}
 	}
 
@@ -336,7 +335,6 @@ out:
 		doorbell[1] = sq_db.u32_8;
 
 		hns_roce_write64_k(doorbell, qp->sq.db_reg_l);
-		qp->sq_next_wqe = ind;
 	}
 
 	spin_unlock_irqrestore(&qp->sq.lock, flags);
@@ -348,12 +346,6 @@ static int hns_roce_v1_post_recv(struct ib_qp *ibqp,
 				 const struct ib_recv_wr *wr,
 				 const struct ib_recv_wr **bad_wr)
 {
-	int ret = 0;
-	int nreq = 0;
-	int ind = 0;
-	int i = 0;
-	u32 reg_val;
-	unsigned long flags = 0;
 	struct hns_roce_rq_wqe_ctrl *ctrl = NULL;
 	struct hns_roce_wqe_data_seg *scat = NULL;
 	struct hns_roce_qp *hr_qp = to_hr_qp(ibqp);
@@ -361,9 +353,14 @@ static int hns_roce_v1_post_recv(struct ib_qp *ibqp,
 	struct device *dev = &hr_dev->pdev->dev;
 	struct hns_roce_rq_db rq_db;
 	__le32 doorbell[2] = {0};
+	unsigned long flags = 0;
+	unsigned int wqe_idx;
+	int ret = 0;
+	int nreq = 0;
+	int i = 0;
+	u32 reg_val;
 
 	spin_lock_irqsave(&hr_qp->rq.lock, flags);
-	ind = hr_qp->rq.head & (hr_qp->rq.wqe_cnt - 1);
 
 	for (nreq = 0; wr; ++nreq, wr = wr->next) {
 		if (hns_roce_wq_overflow(&hr_qp->rq, nreq,
@@ -373,6 +370,8 @@ static int hns_roce_v1_post_recv(struct ib_qp *ibqp,
 			goto out;
 		}
 
+		wqe_idx = (hr_qp->rq.head + nreq) & (hr_qp->rq.wqe_cnt - 1);
+
 		if (unlikely(wr->num_sge > hr_qp->rq.max_gs)) {
 			dev_err(dev, "rq:num_sge=%d > qp->sq.max_gs=%d\n",
 				wr->num_sge, hr_qp->rq.max_gs);
@@ -381,7 +380,7 @@ static int hns_roce_v1_post_recv(struct ib_qp *ibqp,
 			goto out;
 		}
 
-		ctrl = get_recv_wqe(hr_qp, ind);
+		ctrl = get_recv_wqe(hr_qp, wqe_idx);
 
 		roce_set_field(ctrl->rwqe_byte_12,
 			       RQ_WQE_CTRL_RWQE_BYTE_12_RWQE_SGE_NUM_M,
@@ -393,9 +392,7 @@ static int hns_roce_v1_post_recv(struct ib_qp *ibqp,
 		for (i = 0; i < wr->num_sge; i++)
 			set_data_seg(scat + i, wr->sg_list + i);
 
-		hr_qp->rq.wrid[ind] = wr->wr_id;
-
-		ind = (ind + 1) & (hr_qp->rq.wqe_cnt - 1);
+		hr_qp->rq.wrid[wqe_idx] = wr->wr_id;
 	}
 
 out:
@@ -2701,7 +2698,6 @@ static int hns_roce_v1_m_sqp(struct ib_qp *ibqp, const struct ib_qp_attr *attr,
 		hr_qp->rq.tail = 0;
 		hr_qp->sq.head = 0;
 		hr_qp->sq.tail = 0;
-		hr_qp->sq_next_wqe = 0;
 	}
 
 	kfree(context);
@@ -3315,7 +3311,6 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr,
 		hr_qp->rq.tail = 0;
 		hr_qp->sq.head = 0;
 		hr_qp->sq.tail = 0;
-		hr_qp->sq_next_wqe = 0;
 	}
 out:
 	kfree(context);
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index cb8071a3e0d5..45ff44dfabc3 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -239,10 +239,10 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 	struct device *dev = hr_dev->dev;
 	struct hns_roce_v2_db sq_db;
 	struct ib_qp_attr attr;
-	unsigned int sge_ind;
 	unsigned int owner_bit;
+	unsigned int sge_idx;
+	unsigned int wqe_idx;
 	unsigned long flags;
-	unsigned int ind;
 	void *wqe = NULL;
 	bool loopback;
 	int attr_mask;
@@ -269,8 +269,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 	}
 
 	spin_lock_irqsave(&qp->sq.lock, flags);
-	ind = qp->sq_next_wqe;
-	sge_ind = qp->next_sge;
+	sge_idx = qp->next_sge;
 
 	for (nreq = 0; wr; ++nreq, wr = wr->next) {
 		if (hns_roce_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq)) {
@@ -279,6 +278,8 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 			goto out;
 		}
 
+		wqe_idx = (qp->sq.head + nreq) & (qp->sq.wqe_cnt - 1);
+
 		if (unlikely(wr->num_sge > qp->sq.max_gs)) {
 			dev_err(dev, "num_sge=%d > qp->sq.max_gs=%d\n",
 				wr->num_sge, qp->sq.max_gs);
@@ -287,10 +288,8 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 			goto out;
 		}
 
-		wqe = get_send_wqe(qp, ind & (qp->sq.wqe_cnt - 1));
-		qp->sq.wrid[(qp->sq.head + nreq) & (qp->sq.wqe_cnt - 1)] =
-								      wr->wr_id;
-
+		wqe = get_send_wqe(qp, wqe_idx);
+		qp->sq.wrid[wqe_idx] = wr->wr_id;
 		owner_bit =
 		       ~(((qp->sq.head + nreq) >> ilog2(qp->sq.wqe_cnt)) & 0x1);
 		tmp_len = 0;
@@ -373,7 +372,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 			roce_set_field(ud_sq_wqe->byte_20,
 				     V2_UD_SEND_WQE_BYTE_20_MSG_START_SGE_IDX_M,
 				     V2_UD_SEND_WQE_BYTE_20_MSG_START_SGE_IDX_S,
-				     sge_ind & (qp->sge.sge_cnt - 1));
+				     sge_idx & (qp->sge.sge_cnt - 1));
 
 			roce_set_field(ud_sq_wqe->byte_24,
 				       V2_UD_SEND_WQE_BYTE_24_UDPSPN_M,
@@ -423,8 +422,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 			memcpy(&ud_sq_wqe->dgid[0], &ah->av.dgid[0],
 			       GID_LEN_V2);
 
-			set_extend_sge(qp, wr, &sge_ind);
-			ind++;
+			set_extend_sge(qp, wr, &sge_idx);
 		} else if (ibqp->qp_type == IB_QPT_RC) {
 			rc_sq_wqe = wqe;
 			memset(rc_sq_wqe, 0, sizeof(*rc_sq_wqe));
@@ -553,12 +551,10 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 					       wr->num_sge);
 			} else if (wr->opcode != IB_WR_REG_MR) {
 				ret = set_rwqe_data_seg(ibqp, wr, rc_sq_wqe,
-							wqe, &sge_ind, bad_wr);
+							wqe, &sge_idx, bad_wr);
 				if (ret)
 					goto out;
 			}
-
-			ind++;
 		} else {
 			dev_err(dev, "Illegal qp_type(0x%x)\n", ibqp->qp_type);
 			spin_unlock_irqrestore(&qp->sq.lock, flags);
@@ -588,8 +584,7 @@ out:
 
 		hns_roce_write64(hr_dev, (__le32 *)&sq_db, qp->sq.db_reg_l);
 
-		qp->sq_next_wqe = ind;
-		qp->next_sge = sge_ind;
+		qp->next_sge = sge_idx;
 
 		if (qp->state == IB_QPS_ERR) {
 			attr_mask = IB_QP_STATE;
@@ -623,13 +618,12 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
 	unsigned long flags;
 	void *wqe = NULL;
 	int attr_mask;
+	u32 wqe_idx;
 	int ret = 0;
 	int nreq;
-	int ind;
 	int i;
 
 	spin_lock_irqsave(&hr_qp->rq.lock, flags);
-	ind = hr_qp->rq.head & (hr_qp->rq.wqe_cnt - 1);
 
 	if (hr_qp->state == IB_QPS_RESET) {
 		spin_unlock_irqrestore(&hr_qp->rq.lock, flags);
@@ -645,6 +639,8 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
 			goto out;
 		}
 
+		wqe_idx = (hr_qp->rq.head + nreq) & (hr_qp->rq.wqe_cnt - 1);
+
 		if (unlikely(wr->num_sge > hr_qp->rq.max_gs)) {
 			dev_err(dev, "rq:num_sge=%d > qp->sq.max_gs=%d\n",
 				wr->num_sge, hr_qp->rq.max_gs);
@@ -653,7 +649,7 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
 			goto out;
 		}
 
-		wqe = get_recv_wqe(hr_qp, ind);
+		wqe = get_recv_wqe(hr_qp, wqe_idx);
 		dseg = (struct hns_roce_v2_wqe_data_seg *)wqe;
 		for (i = 0; i < wr->num_sge; i++) {
 			if (!wr->sg_list[i].length)
@@ -669,8 +665,8 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
 
 		/* rq support inline data */
 		if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) {
-			sge_list = hr_qp->rq_inl_buf.wqe_list[ind].sg_list;
-			hr_qp->rq_inl_buf.wqe_list[ind].sge_cnt =
+			sge_list = hr_qp->rq_inl_buf.wqe_list[wqe_idx].sg_list;
+			hr_qp->rq_inl_buf.wqe_list[wqe_idx].sge_cnt =
 							       (u32)wr->num_sge;
 			for (i = 0; i < wr->num_sge; i++) {
 				sge_list[i].addr =
@@ -679,9 +675,7 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
 			}
 		}
 
-		hr_qp->rq.wrid[ind] = wr->wr_id;
-
-		ind = (ind + 1) & (hr_qp->rq.wqe_cnt - 1);
+		hr_qp->rq.wrid[wqe_idx] = wr->wr_id;
 	}
 
 out:
@@ -4464,7 +4458,6 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp,
 		hr_qp->rq.tail = 0;
 		hr_qp->sq.head = 0;
 		hr_qp->sq.tail = 0;
-		hr_qp->sq_next_wqe = 0;
 		hr_qp->next_sge = 0;
 		if (hr_qp->rq.wqe_cnt)
 			*hr_qp->rdb.db_record = 0;
-- 
cgit v1.2.3-59-g8ed1b


From 58fb0b5625ac976d907427ab639cdbf1aaefdac5 Mon Sep 17 00:00:00 2001
From: Bernard Metzler <bmt@zurich.ibm.com>
Date: Tue, 10 Dec 2019 17:17:29 +0100
Subject: RDMA/siw: Simplify QP representation

Change siw_qp to contain ib_qp. Use rdma_is_kernel_res() on contained
ib_qp to distinguish kernel level from user level applications
resources. Apply same mechanism for kernel/user level application
detection to completion queues.

Link: https://lore.kernel.org/r/20191210161729.31598-1-bmt@zurich.ibm.com
Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/sw/siw/siw.h       | 26 ++++-----------
 drivers/infiniband/sw/siw/siw_cq.c    |  2 +-
 drivers/infiniband/sw/siw/siw_main.c  |  2 +-
 drivers/infiniband/sw/siw/siw_qp.c    | 13 +++-----
 drivers/infiniband/sw/siw/siw_qp_rx.c |  6 ++--
 drivers/infiniband/sw/siw/siw_qp_tx.c |  2 +-
 drivers/infiniband/sw/siw/siw_verbs.c | 61 ++++++++++++++---------------------
 7 files changed, 42 insertions(+), 70 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/siw/siw.h b/drivers/infiniband/sw/siw/siw.h
index b939f489cd46..af5e9f8c0fcd 100644
--- a/drivers/infiniband/sw/siw/siw.h
+++ b/drivers/infiniband/sw/siw/siw.h
@@ -7,6 +7,7 @@
 #define _SIW_H
 
 #include <rdma/ib_verbs.h>
+#include <rdma/restrack.h>
 #include <linux/socket.h>
 #include <linux/skbuff.h>
 #include <crypto/hash.h>
@@ -209,7 +210,6 @@ struct siw_cq {
 	u32 cq_put;
 	u32 cq_get;
 	u32 num_cqe;
-	bool kernel_verbs;
 	struct rdma_user_mmap_entry *cq_entry; /* mmap info for CQE array */
 	u32 id; /* For debugging only */
 };
@@ -254,8 +254,8 @@ struct siw_srq {
 	u32 rq_get;
 	u32 num_rqe; /* max # of wqe's allowed */
 	struct rdma_user_mmap_entry *srq_entry; /* mmap info for SRQ array */
-	char armed; /* inform user if limit hit */
-	char kernel_verbs; /* '1' if kernel client */
+	bool armed:1; /* inform user if limit hit */
+	bool is_kernel_res:1; /* true if kernel client */
 };
 
 struct siw_qp_attrs {
@@ -418,13 +418,11 @@ struct siw_iwarp_tx {
 };
 
 struct siw_qp {
+	struct ib_qp base_qp;
 	struct siw_device *sdev;
-	struct ib_qp *ib_qp;
 	struct kref ref;
-	u32 qp_num;
 	struct list_head devq;
 	int tx_cpu;
-	bool kernel_verbs;
 	struct siw_qp_attrs attrs;
 
 	struct siw_cep *cep;
@@ -472,11 +470,6 @@ struct siw_qp {
 	struct rcu_head rcu;
 };
 
-struct siw_base_qp {
-	struct ib_qp base_qp;
-	struct siw_qp *qp;
-};
-
 /* helper macros */
 #define rx_qp(rx) container_of(rx, struct siw_qp, rx_stream)
 #define tx_qp(tx) container_of(tx, struct siw_qp, tx_ctx)
@@ -572,14 +565,9 @@ static inline struct siw_ucontext *to_siw_ctx(struct ib_ucontext *base_ctx)
 	return container_of(base_ctx, struct siw_ucontext, base_ucontext);
 }
 
-static inline struct siw_base_qp *to_siw_base_qp(struct ib_qp *base_qp)
-{
-	return container_of(base_qp, struct siw_base_qp, base_qp);
-}
-
 static inline struct siw_qp *to_siw_qp(struct ib_qp *base_qp)
 {
-	return to_siw_base_qp(base_qp)->qp;
+	return container_of(base_qp, struct siw_qp, base_qp);
 }
 
 static inline struct siw_cq *to_siw_cq(struct ib_cq *base_cq)
@@ -624,7 +612,7 @@ static inline struct siw_qp *siw_qp_id2obj(struct siw_device *sdev, int id)
 
 static inline u32 qp_id(struct siw_qp *qp)
 {
-	return qp->qp_num;
+	return qp->base_qp.qp_num;
 }
 
 static inline void siw_qp_get(struct siw_qp *qp)
@@ -735,7 +723,7 @@ static inline void siw_crc_skb(struct siw_rx_stream *srx, unsigned int len)
 		  "MEM[0x%08x] %s: " fmt, mem->stag, __func__, ##__VA_ARGS__)
 
 #define siw_dbg_cep(cep, fmt, ...)                                             \
-	ibdev_dbg(&cep->sdev->base_dev, "CEP[0x%pK] %s: " fmt,                  \
+	ibdev_dbg(&cep->sdev->base_dev, "CEP[0x%pK] %s: " fmt,                 \
 		  cep, __func__, ##__VA_ARGS__)
 
 void siw_cq_flush(struct siw_cq *cq);
diff --git a/drivers/infiniband/sw/siw/siw_cq.c b/drivers/infiniband/sw/siw/siw_cq.c
index d8db3bee9da7..d68e37859e73 100644
--- a/drivers/infiniband/sw/siw/siw_cq.c
+++ b/drivers/infiniband/sw/siw/siw_cq.c
@@ -65,7 +65,7 @@ int siw_reap_cqe(struct siw_cq *cq, struct ib_wc *wc)
 		 * reaped here, which do not hold a QP reference
 		 * and do not qualify for memory extension verbs.
 		 */
-		if (likely(cq->kernel_verbs)) {
+		if (likely(rdma_is_kernel_res(&cq->base_cq.res))) {
 			if (cqe->flags & SIW_WQE_REM_INVAL) {
 				wc->ex.invalidate_rkey = cqe->inval_stag;
 				wc->wc_flags = IB_WC_WITH_INVALIDATE;
diff --git a/drivers/infiniband/sw/siw/siw_main.c b/drivers/infiniband/sw/siw/siw_main.c
index c147f0613d95..96ed349c0939 100644
--- a/drivers/infiniband/sw/siw/siw_main.c
+++ b/drivers/infiniband/sw/siw/siw_main.c
@@ -244,7 +244,7 @@ static struct ib_qp *siw_get_base_qp(struct ib_device *base_dev, int id)
 		 * siw_qp_id2obj() increments object reference count
 		 */
 		siw_qp_put(qp);
-		return qp->ib_qp;
+		return &qp->base_qp;
 	}
 	return NULL;
 }
diff --git a/drivers/infiniband/sw/siw/siw_qp.c b/drivers/infiniband/sw/siw/siw_qp.c
index b4317480cee7..875d36d4b1c6 100644
--- a/drivers/infiniband/sw/siw/siw_qp.c
+++ b/drivers/infiniband/sw/siw/siw_qp.c
@@ -1070,8 +1070,8 @@ int siw_sqe_complete(struct siw_qp *qp, struct siw_sqe *sqe, u32 bytes,
 			cqe->imm_data = 0;
 			cqe->bytes = bytes;
 
-			if (cq->kernel_verbs)
-				cqe->base_qp = qp->ib_qp;
+			if (rdma_is_kernel_res(&cq->base_cq.res))
+				cqe->base_qp = &qp->base_qp;
 			else
 				cqe->qp_id = qp_id(qp);
 
@@ -1128,8 +1128,8 @@ int siw_rqe_complete(struct siw_qp *qp, struct siw_rqe *rqe, u32 bytes,
 			cqe->imm_data = 0;
 			cqe->bytes = bytes;
 
-			if (cq->kernel_verbs) {
-				cqe->base_qp = qp->ib_qp;
+			if (rdma_is_kernel_res(&cq->base_cq.res)) {
+				cqe->base_qp = &qp->base_qp;
 				if (inval_stag) {
 					cqe_flags |= SIW_WQE_REM_INVAL;
 					cqe->inval_stag = inval_stag;
@@ -1297,13 +1297,12 @@ void siw_rq_flush(struct siw_qp *qp)
 
 int siw_qp_add(struct siw_device *sdev, struct siw_qp *qp)
 {
-	int rv = xa_alloc(&sdev->qp_xa, &qp->ib_qp->qp_num, qp, xa_limit_32b,
+	int rv = xa_alloc(&sdev->qp_xa, &qp->base_qp.qp_num, qp, xa_limit_32b,
 			  GFP_KERNEL);
 
 	if (!rv) {
 		kref_init(&qp->ref);
 		qp->sdev = sdev;
-		qp->qp_num = qp->ib_qp->qp_num;
 		siw_dbg_qp(qp, "new QP\n");
 	}
 	return rv;
@@ -1312,7 +1311,6 @@ int siw_qp_add(struct siw_device *sdev, struct siw_qp *qp)
 void siw_free_qp(struct kref *ref)
 {
 	struct siw_qp *found, *qp = container_of(ref, struct siw_qp, ref);
-	struct siw_base_qp *siw_base_qp = to_siw_base_qp(qp->ib_qp);
 	struct siw_device *sdev = qp->sdev;
 	unsigned long flags;
 
@@ -1335,5 +1333,4 @@ void siw_free_qp(struct kref *ref)
 	atomic_dec(&sdev->num_qp);
 	siw_dbg_qp(qp, "free QP\n");
 	kfree_rcu(qp, rcu);
-	kfree(siw_base_qp);
 }
diff --git a/drivers/infiniband/sw/siw/siw_qp_rx.c b/drivers/infiniband/sw/siw/siw_qp_rx.c
index c0a887240325..9ccce2909ac4 100644
--- a/drivers/infiniband/sw/siw/siw_qp_rx.c
+++ b/drivers/infiniband/sw/siw/siw_qp_rx.c
@@ -68,7 +68,7 @@ static int siw_rx_umem(struct siw_rx_stream *srx, struct siw_umem *umem,
 			return -EFAULT;
 		}
 		if (srx->mpa_crc_hd) {
-			if (rx_qp(srx)->kernel_verbs) {
+			if (rdma_is_kernel_res(&rx_qp(srx)->base_qp.res)) {
 				crypto_shash_update(srx->mpa_crc_hd,
 					(u8 *)(dest + pg_off), bytes);
 				kunmap_atomic(dest);
@@ -388,7 +388,7 @@ static struct siw_wqe *siw_rqe_get(struct siw_qp *qp)
 				struct siw_rqe *rqe2 = &srq->recvq[off];
 
 				if (!(rqe2->flags & SIW_WQE_VALID)) {
-					srq->armed = 0;
+					srq->armed = false;
 					srq_event = true;
 				}
 			}
@@ -1264,7 +1264,7 @@ static int siw_rdmap_complete(struct siw_qp *qp, int error)
 
 			if (wc_status == SIW_WC_SUCCESS)
 				wc_status = SIW_WC_GENERAL_ERR;
-		} else if (qp->kernel_verbs &&
+		} else if (rdma_is_kernel_res(&qp->base_qp.res) &&
 			   rx_type(wqe) == SIW_OP_READ_LOCAL_INV) {
 			/*
 			 * Handle any STag invalidation request
diff --git a/drivers/infiniband/sw/siw/siw_qp_tx.c b/drivers/infiniband/sw/siw/siw_qp_tx.c
index 5d97bba0ce6d..ae92c8080967 100644
--- a/drivers/infiniband/sw/siw/siw_qp_tx.c
+++ b/drivers/infiniband/sw/siw/siw_qp_tx.c
@@ -817,7 +817,7 @@ static int siw_qp_sq_proc_tx(struct siw_qp *qp, struct siw_wqe *wqe)
 			}
 		} else {
 			wqe->bytes = wqe->sqe.sge[0].length;
-			if (!qp->kernel_verbs) {
+			if (!rdma_is_kernel_res(&qp->base_qp.res)) {
 				if (wqe->bytes > SIW_MAX_INLINE) {
 					rv = -EINVAL;
 					goto tx_error;
diff --git a/drivers/infiniband/sw/siw/siw_verbs.c b/drivers/infiniband/sw/siw/siw_verbs.c
index 5fd6d6499b3d..07e30138aaa1 100644
--- a/drivers/infiniband/sw/siw/siw_verbs.c
+++ b/drivers/infiniband/sw/siw/siw_verbs.c
@@ -303,7 +303,6 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd,
 			    struct ib_udata *udata)
 {
 	struct siw_qp *qp = NULL;
-	struct siw_base_qp *siw_base_qp = NULL;
 	struct ib_device *base_dev = pd->device;
 	struct siw_device *sdev = to_siw_dev(base_dev);
 	struct siw_ucontext *uctx =
@@ -357,26 +356,16 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd,
 		rv = -EINVAL;
 		goto err_out;
 	}
-	siw_base_qp = kzalloc(sizeof(*siw_base_qp), GFP_KERNEL);
-	if (!siw_base_qp) {
-		rv = -ENOMEM;
-		goto err_out;
-	}
 	qp = kzalloc(sizeof(*qp), GFP_KERNEL);
 	if (!qp) {
 		rv = -ENOMEM;
 		goto err_out;
 	}
-	siw_base_qp->qp = qp;
-	qp->ib_qp = &siw_base_qp->base_qp;
-
 	init_rwsem(&qp->state_lock);
 	spin_lock_init(&qp->sq_lock);
 	spin_lock_init(&qp->rq_lock);
 	spin_lock_init(&qp->orq_lock);
 
-	qp->kernel_verbs = !udata;
-
 	rv = siw_qp_add(sdev, qp);
 	if (rv)
 		goto err_out;
@@ -389,10 +378,10 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd,
 	num_sqe = roundup_pow_of_two(attrs->cap.max_send_wr);
 	num_rqe = roundup_pow_of_two(attrs->cap.max_recv_wr);
 
-	if (qp->kernel_verbs)
-		qp->sendq = vzalloc(num_sqe * sizeof(struct siw_sqe));
-	else
+	if (udata)
 		qp->sendq = vmalloc_user(num_sqe * sizeof(struct siw_sqe));
+	else
+		qp->sendq = vzalloc(num_sqe * sizeof(struct siw_sqe));
 
 	if (qp->sendq == NULL) {
 		siw_dbg(base_dev, "SQ size %d alloc failed\n", num_sqe);
@@ -419,13 +408,14 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd,
 		 */
 		qp->srq = to_siw_srq(attrs->srq);
 		qp->attrs.rq_size = 0;
-		siw_dbg(base_dev, "QP [%u]: SRQ attached\n", qp->qp_num);
+		siw_dbg(base_dev, "QP [%u]: SRQ attached\n",
+			qp->base_qp.qp_num);
 	} else if (num_rqe) {
-		if (qp->kernel_verbs)
-			qp->recvq = vzalloc(num_rqe * sizeof(struct siw_rqe));
-		else
+		if (udata)
 			qp->recvq =
 				vmalloc_user(num_rqe * sizeof(struct siw_rqe));
+		else
+			qp->recvq = vzalloc(num_rqe * sizeof(struct siw_rqe));
 
 		if (qp->recvq == NULL) {
 			siw_dbg(base_dev, "RQ size %d alloc failed\n", num_rqe);
@@ -492,13 +482,11 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd,
 	list_add_tail(&qp->devq, &sdev->qp_list);
 	spin_unlock_irqrestore(&sdev->lock, flags);
 
-	return qp->ib_qp;
+	return &qp->base_qp;
 
 err_out_xa:
 	xa_erase(&sdev->qp_xa, qp_id(qp));
 err_out:
-	kfree(siw_base_qp);
-
 	if (qp) {
 		if (uctx) {
 			rdma_user_mmap_entry_remove(qp->sq_entry);
@@ -742,7 +730,7 @@ int siw_post_send(struct ib_qp *base_qp, const struct ib_send_wr *wr,
 	unsigned long flags;
 	int rv = 0;
 
-	if (wr && !qp->kernel_verbs) {
+	if (wr && !rdma_is_kernel_res(&qp->base_qp.res)) {
 		siw_dbg_qp(qp, "wr must be empty for user mapped sq\n");
 		*bad_wr = wr;
 		return -EINVAL;
@@ -939,7 +927,7 @@ int siw_post_send(struct ib_qp *base_qp, const struct ib_send_wr *wr,
 	if (rv <= 0)
 		goto skip_direct_sending;
 
-	if (qp->kernel_verbs) {
+	if (rdma_is_kernel_res(&qp->base_qp.res)) {
 		rv = siw_sq_start(qp);
 	} else {
 		qp->tx_ctx.in_syscall = 1;
@@ -984,8 +972,8 @@ int siw_post_receive(struct ib_qp *base_qp, const struct ib_recv_wr *wr,
 		*bad_wr = wr;
 		return -EOPNOTSUPP; /* what else from errno.h? */
 	}
-	if (!qp->kernel_verbs) {
-		siw_dbg_qp(qp, "no kernel post_recv for user mapped sq\n");
+	if (!rdma_is_kernel_res(&qp->base_qp.res)) {
+		siw_dbg_qp(qp, "no kernel post_recv for user mapped rq\n");
 		*bad_wr = wr;
 		return -EINVAL;
 	}
@@ -1127,14 +1115,13 @@ int siw_create_cq(struct ib_cq *base_cq, const struct ib_cq_init_attr *attr,
 	cq->base_cq.cqe = size;
 	cq->num_cqe = size;
 
-	if (!udata) {
-		cq->kernel_verbs = 1;
-		cq->queue = vzalloc(size * sizeof(struct siw_cqe) +
-				    sizeof(struct siw_cq_ctrl));
-	} else {
+	if (udata)
 		cq->queue = vmalloc_user(size * sizeof(struct siw_cqe) +
 					 sizeof(struct siw_cq_ctrl));
-	}
+	else
+		cq->queue = vzalloc(size * sizeof(struct siw_cqe) +
+				    sizeof(struct siw_cq_ctrl));
+
 	if (cq->queue == NULL) {
 		rv = -ENOMEM;
 		goto err_out;
@@ -1589,9 +1576,9 @@ int siw_create_srq(struct ib_srq *base_srq,
 	srq->num_rqe = roundup_pow_of_two(attrs->max_wr);
 	srq->limit = attrs->srq_limit;
 	if (srq->limit)
-		srq->armed = 1;
+		srq->armed = true;
 
-	srq->kernel_verbs = !udata;
+	srq->is_kernel_res = !udata;
 
 	if (udata)
 		srq->recvq =
@@ -1671,9 +1658,9 @@ int siw_modify_srq(struct ib_srq *base_srq, struct ib_srq_attr *attrs,
 				rv = -EINVAL;
 				goto out;
 			}
-			srq->armed = 1;
+			srq->armed = true;
 		} else {
-			srq->armed = 0;
+			srq->armed = false;
 		}
 		srq->limit = attrs->srq_limit;
 	}
@@ -1745,7 +1732,7 @@ int siw_post_srq_recv(struct ib_srq *base_srq, const struct ib_recv_wr *wr,
 	unsigned long flags;
 	int rv = 0;
 
-	if (unlikely(!srq->kernel_verbs)) {
+	if (unlikely(!srq->is_kernel_res)) {
 		siw_dbg_pd(base_srq->pd,
 			   "[SRQ]: no kernel post_recv for mapped srq\n");
 		rv = -EINVAL;
@@ -1797,7 +1784,7 @@ out:
 void siw_qp_event(struct siw_qp *qp, enum ib_event_type etype)
 {
 	struct ib_event event;
-	struct ib_qp *base_qp = qp->ib_qp;
+	struct ib_qp *base_qp = &qp->base_qp;
 
 	/*
 	 * Do not report asynchronous errors on QP which gets
-- 
cgit v1.2.3-59-g8ed1b


From b5671afe5e39ed71e94eae788bacdcceec69db09 Mon Sep 17 00:00:00 2001
From: Prabhath Sajeepa <psajeepa@purestorage.com>
Date: Thu, 12 Dec 2019 17:11:29 -0700
Subject: IB/mlx5: Fix outstanding_pi index for GSI qps

Commit b0ffeb537f3a ("IB/mlx5: Fix iteration overrun in GSI qps") changed
the way outstanding WRs are tracked for the GSI QP. But the fix did not
cover the case when a call to ib_post_send() fails and updates index to
track outstanding.

Since the prior commmit outstanding_pi should not be bounded otherwise the
loop generate_completions() will fail.

Fixes: b0ffeb537f3a ("IB/mlx5: Fix iteration overrun in GSI qps")
Link: https://lore.kernel.org/r/1576195889-23527-1-git-send-email-psajeepa@purestorage.com
Signed-off-by: Prabhath Sajeepa <psajeepa@purestorage.com>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/gsi.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/gsi.c b/drivers/infiniband/hw/mlx5/gsi.c
index ac4d8d1b9a07..1ae6fd95acaa 100644
--- a/drivers/infiniband/hw/mlx5/gsi.c
+++ b/drivers/infiniband/hw/mlx5/gsi.c
@@ -507,8 +507,7 @@ int mlx5_ib_gsi_post_send(struct ib_qp *qp, const struct ib_send_wr *wr,
 		ret = ib_post_send(tx_qp, &cur_wr.wr, bad_wr);
 		if (ret) {
 			/* Undo the effect of adding the outstanding wr */
-			gsi->outstanding_pi = (gsi->outstanding_pi - 1) %
-					      gsi->cap.max_send_wr;
+			gsi->outstanding_pi--;
 			goto err;
 		}
 		spin_unlock_irqrestore(&gsi->lock, flags);
-- 
cgit v1.2.3-59-g8ed1b


From a242c36951ecd24bc16086940dbe6b522205c461 Mon Sep 17 00:00:00 2001
From: Håkon Bugge <haakon.bugge@oracle.com>
Date: Mon, 16 Dec 2019 13:04:36 +0100
Subject: RDMA/netlink: Do not always generate an ACK for some netlink
 operations
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

In rdma_nl_rcv_skb(), the local variable err is assigned the return value
of the supplied callback function, which could be one of
ib_nl_handle_resolve_resp(), ib_nl_handle_set_timeout(), or
ib_nl_handle_ip_res_resp(). These three functions all return skb->len on
success.

rdma_nl_rcv_skb() is merely a copy of netlink_rcv_skb(). The callback
functions used by the latter have the convention: "Returns 0 on success or
a negative error code".

In particular, the statement (equal for both functions):

   if (nlh->nlmsg_flags & NLM_F_ACK || err)

implies that rdma_nl_rcv_skb() always will ack a message, independent of
the NLM_F_ACK being set in nlmsg_flags or not.

The fix could be to change the above statement, but it is better to keep
the two *_rcv_skb() functions equal in this respect and instead change the
three callback functions in the rdma subsystem to the correct convention.

Fixes: 2ca546b92a02 ("IB/sa: Route SA pathrecord query through netlink")
Fixes: ae43f8286730 ("IB/core: Add IP to GID netlink offload")
Link: https://lore.kernel.org/r/20191216120436.3204814-1-haakon.bugge@oracle.com
Suggested-by: Mark Haywood <mark.haywood@oracle.com>
Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Tested-by: Mark Haywood <mark.haywood@oracle.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/addr.c     | 2 +-
 drivers/infiniband/core/sa_query.c | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c
index 606fa6d86685..1753a9801b70 100644
--- a/drivers/infiniband/core/addr.c
+++ b/drivers/infiniband/core/addr.c
@@ -139,7 +139,7 @@ int ib_nl_handle_ip_res_resp(struct sk_buff *skb,
 	if (ib_nl_is_good_ip_resp(nlh))
 		ib_nl_process_good_ip_rsep(nlh);
 
-	return skb->len;
+	return 0;
 }
 
 static int ib_nl_ip_send_msg(struct rdma_dev_addr *dev_addr,
diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c
index 8917125ea16d..30d4c126a2db 100644
--- a/drivers/infiniband/core/sa_query.c
+++ b/drivers/infiniband/core/sa_query.c
@@ -1068,7 +1068,7 @@ int ib_nl_handle_set_timeout(struct sk_buff *skb,
 	}
 
 settimeout_out:
-	return skb->len;
+	return 0;
 }
 
 static inline int ib_nl_is_good_resolve_resp(const struct nlmsghdr *nlh)
@@ -1139,7 +1139,7 @@ int ib_nl_handle_resolve_resp(struct sk_buff *skb,
 	}
 
 resp_out:
-	return skb->len;
+	return 0;
 }
 
 static void free_sm_ah(struct kref *kref)
-- 
cgit v1.2.3-59-g8ed1b


From 9f48db0d4a08624bb9ba847ea40c8abad753b396 Mon Sep 17 00:00:00 2001
From: Aditya Pakki <pakki001@umn.edu>
Date: Tue, 17 Dec 2019 13:44:37 -0600
Subject: RDMA/srpt: Remove unnecessary assertion in srpt_queue_response

Since ch has already been de-referenced by the time we get to the BUG_ON,
it is useless. The back trace alone is enough to tell what is going on,
delete the redundant BUG_ON.

Link: https://lore.kernel.org/r/20191217194437.25568-1-pakki001@umn.edu
Signed-off-by: Aditya Pakki <pakki001@umn.edu>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/ulp/srpt/ib_srpt.c | 2 --
 1 file changed, 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
index 23c782e3d49a..98552749d71c 100644
--- a/drivers/infiniband/ulp/srpt/ib_srpt.c
+++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
@@ -2810,8 +2810,6 @@ static void srpt_queue_response(struct se_cmd *cmd)
 	int resp_len, ret, i;
 	u8 srp_tm_status;
 
-	BUG_ON(!ch);
-
 	state = ioctx->state;
 	switch (state) {
 	case SRPT_STATE_NEW:
-- 
cgit v1.2.3-59-g8ed1b


From 4d6e8a033f7c6665c7ca5c5cdcf66019baa4b7a2 Mon Sep 17 00:00:00 2001
From: Danit Goldberg <danitg@mellanox.com>
Date: Thu, 19 Dec 2019 15:47:50 +0200
Subject: RDMA/cm: Use RCU synchronization mechanism to protect cm_id_private
 xa_load()

The RCU mechanism is optimized for read-mostly scenarios and therefore
more suitable to protect the cm_id_private to decrease "cm.lock"
congestion.

This patch replaces the existing spinlock locking mechanism and kfree with
RCU mechanism in places where spinlock(cm.lock) protected xa_load
returning the cm_id_priv

In addition, delete the cm_get_id() function as there is no longer a
distinction if the caller already holds the cm_lock.

Remove an open coded version of cm_get_id().

Link: https://lore.kernel.org/r/20191219134750.413429-1-leon@kernel.org
Signed-off-by: Danit Goldberg <danitg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c | 43 +++++++++++++------------------------------
 1 file changed, 13 insertions(+), 30 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 455b3659d84b..2945031dd8a7 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -241,6 +241,7 @@ struct cm_id_private {
 	/* Number of clients sharing this ib_cm_id. Only valid for listeners.
 	 * Protected by the cm.lock spinlock. */
 	int listen_sharecount;
+	struct rcu_head rcu;
 
 	struct ib_mad_send_buf *msg;
 	struct cm_timewait_info *timewait_info;
@@ -593,28 +594,16 @@ static void cm_free_id(__be32 local_id)
 	xa_erase_irq(&cm.local_id_table, cm_local_id(local_id));
 }
 
-static struct cm_id_private * cm_get_id(__be32 local_id, __be32 remote_id)
+static struct cm_id_private *cm_acquire_id(__be32 local_id, __be32 remote_id)
 {
 	struct cm_id_private *cm_id_priv;
 
+	rcu_read_lock();
 	cm_id_priv = xa_load(&cm.local_id_table, cm_local_id(local_id));
-	if (cm_id_priv) {
-		if (cm_id_priv->id.remote_id == remote_id)
-			refcount_inc(&cm_id_priv->refcount);
-		else
-			cm_id_priv = NULL;
-	}
-
-	return cm_id_priv;
-}
-
-static struct cm_id_private * cm_acquire_id(__be32 local_id, __be32 remote_id)
-{
-	struct cm_id_private *cm_id_priv;
-
-	spin_lock_irq(&cm.lock);
-	cm_id_priv = cm_get_id(local_id, remote_id);
-	spin_unlock_irq(&cm.lock);
+	if (!cm_id_priv || cm_id_priv->id.remote_id != remote_id ||
+	    !refcount_inc_not_zero(&cm_id_priv->refcount))
+		cm_id_priv = NULL;
+	rcu_read_unlock();
 
 	return cm_id_priv;
 }
@@ -1089,7 +1078,7 @@ retest:
 	rdma_destroy_ah_attr(&cm_id_priv->av.ah_attr);
 	rdma_destroy_ah_attr(&cm_id_priv->alt_av.ah_attr);
 	kfree(cm_id_priv->private_data);
-	kfree(cm_id_priv);
+	kfree_rcu(cm_id_priv, rcu);
 }
 
 void ib_destroy_cm_id(struct ib_cm_id *cm_id)
@@ -1821,7 +1810,7 @@ static struct cm_id_private * cm_match_req(struct cm_work *work,
 	spin_lock_irq(&cm.lock);
 	timewait_info = cm_insert_remote_id(cm_id_priv->timewait_info);
 	if (timewait_info) {
-		cur_cm_id_priv = cm_get_id(timewait_info->work.local_id,
+		cur_cm_id_priv = cm_acquire_id(timewait_info->work.local_id,
 					   timewait_info->work.remote_id);
 		spin_unlock_irq(&cm.lock);
 		if (cur_cm_id_priv) {
@@ -1835,7 +1824,7 @@ static struct cm_id_private * cm_match_req(struct cm_work *work,
 	timewait_info = cm_insert_remote_qpn(cm_id_priv->timewait_info);
 	if (timewait_info) {
 		cm_cleanup_timewait(cm_id_priv->timewait_info);
-		cur_cm_id_priv = cm_get_id(timewait_info->work.local_id,
+		cur_cm_id_priv = cm_acquire_id(timewait_info->work.local_id,
 					   timewait_info->work.remote_id);
 
 		spin_unlock_irq(&cm.lock);
@@ -2293,7 +2282,7 @@ static int cm_rep_handler(struct cm_work *work)
 		rb_erase(&cm_id_priv->timewait_info->remote_id_node,
 			 &cm.remote_id_table);
 		cm_id_priv->timewait_info->inserted_remote_id = 0;
-		cur_cm_id_priv = cm_get_id(timewait_info->work.local_id,
+		cur_cm_id_priv = cm_acquire_id(timewait_info->work.local_id,
 					   timewait_info->work.remote_id);
 
 		spin_unlock(&cm.lock);
@@ -2788,14 +2777,8 @@ static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
 			spin_unlock_irq(&cm.lock);
 			return NULL;
 		}
-		cm_id_priv = xa_load(&cm.local_id_table,
-				cm_local_id(timewait_info->work.local_id));
-		if (cm_id_priv) {
-			if (cm_id_priv->id.remote_id == remote_id)
-				refcount_inc(&cm_id_priv->refcount);
-			else
-				cm_id_priv = NULL;
-		}
+		cm_id_priv =
+			cm_acquire_id(timewait_info->work.local_id, remote_id);
 		spin_unlock_irq(&cm.lock);
 	} else if (cm_rej_get_msg_rejected(rej_msg) == CM_MSG_RESPONSE_REQ)
 		cm_id_priv = cm_acquire_id(rej_msg->remote_comm_id, 0);
-- 
cgit v1.2.3-59-g8ed1b


From 3593f69c55c53de6e2de8799e4892b8c795f3713 Mon Sep 17 00:00:00 2001
From: Eugene Crosser <evgenii.cherkashin@profitbricks.com>
Date: Thu, 19 Dec 2019 15:48:47 +0200
Subject: RDMA/mlx4: Redo TX checksum offload in line with docs

Ingress checksum offload was not working for IPv6 frames because the
conditional expression that checks validation status passed from the
hardware was not matching the algorithm described in the documentation.

This patch defines L4_CSUM flag (which falls inside the badfcs_enc field
in the existing definition of the CQE layout) and replaces the conditional
expression with the one defined in the "ConnectX(r) Family Programmer's
Manual" document.

Link: https://lore.kernel.org/r/20191219134847.413582-1-leon@kernel.org
Signed-off-by: Eugene Crosser <evgenii.cherkashin@profitbricks.com>
Reviewed-by: Jack Wang <jinpu.wang@profitbricks.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx4/cq.c | 18 +++++++-----------
 include/linux/mlx4/cq.h         |  5 +++++
 2 files changed, 12 insertions(+), 11 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index 306b21281fa2..72eeb9a85bc5 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -568,18 +568,13 @@ static void mlx4_ib_handle_error_cqe(struct mlx4_err_cqe *cqe,
 	wc->vendor_err = cqe->vendor_err_syndrome;
 }
 
-static int mlx4_ib_ipoib_csum_ok(__be16 status, __be16 checksum)
+static int mlx4_ib_ipoib_csum_ok(__be16 status, u8 badfcs_enc, __be16 checksum)
 {
-	return ((status & cpu_to_be16(MLX4_CQE_STATUS_IPV4      |
-				      MLX4_CQE_STATUS_IPV4F     |
-				      MLX4_CQE_STATUS_IPV4OPT   |
-				      MLX4_CQE_STATUS_IPV6      |
-				      MLX4_CQE_STATUS_IPOK)) ==
-		cpu_to_be16(MLX4_CQE_STATUS_IPV4        |
-			    MLX4_CQE_STATUS_IPOK))              &&
-		(status & cpu_to_be16(MLX4_CQE_STATUS_UDP       |
-				      MLX4_CQE_STATUS_TCP))     &&
-		checksum == cpu_to_be16(0xffff);
+	return ((badfcs_enc & MLX4_CQE_STATUS_L4_CSUM) ||
+		((status & cpu_to_be16(MLX4_CQE_STATUS_IPOK)) &&
+		 (status & cpu_to_be16(MLX4_CQE_STATUS_TCP |
+				       MLX4_CQE_STATUS_UDP)) &&
+		 (checksum == cpu_to_be16(0xffff))));
 }
 
 static void use_tunnel_data(struct mlx4_ib_qp *qp, struct mlx4_ib_cq *cq, struct ib_wc *wc,
@@ -855,6 +850,7 @@ repoll:
 		wc->wc_flags	  |= g_mlpath_rqpn & 0x80000000 ? IB_WC_GRH : 0;
 		wc->pkey_index     = be32_to_cpu(cqe->immed_rss_invalid) & 0x7f;
 		wc->wc_flags	  |= mlx4_ib_ipoib_csum_ok(cqe->status,
+					cqe->badfcs_enc,
 					cqe->checksum) ? IB_WC_IP_CSUM_OK : 0;
 		if (is_eth) {
 			wc->slid = 0;
diff --git a/include/linux/mlx4/cq.h b/include/linux/mlx4/cq.h
index 508e8cc5ee86..653d2a0aa44c 100644
--- a/include/linux/mlx4/cq.h
+++ b/include/linux/mlx4/cq.h
@@ -130,6 +130,11 @@ enum {
 	MLX4_CQE_STATUS_IPOK		= 1 << 12,
 };
 
+/* L4_CSUM is logically part of status, but has to checked against badfcs_enc */
+enum {
+	MLX4_CQE_STATUS_L4_CSUM		= 1 << 2,
+};
+
 enum {
 	MLX4_CQE_LLC                     = 1,
 	MLX4_CQE_SNAP                    = 1 << 1,
-- 
cgit v1.2.3-59-g8ed1b


From 2fb3b5ae1ca77109209bd886094224faa37a43e8 Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Thu, 19 Dec 2019 16:19:22 -0500
Subject: IB/hfi1: Add accessor API routines to access context members

This patch adds a set of accessor routines to access context members.

Link: https://lore.kernel.org/r/20191219211922.58387.26548.stgit@awfm-01.aw.intel.com
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/chip.c        |  27 +++----
 drivers/infiniband/hw/hfi1/common.h      |   3 +
 drivers/infiniband/hw/hfi1/driver.c      |  74 +++++++-----------
 drivers/infiniband/hw/hfi1/file_ops.c    |  12 +--
 drivers/infiniband/hw/hfi1/hfi.h         | 129 ++++++++++++++++++++++++++++++-
 drivers/infiniband/hw/hfi1/init.c        |   6 +-
 drivers/infiniband/hw/hfi1/trace_ctxts.h |   2 +-
 drivers/infiniband/hw/hfi1/trace_rx.h    |  13 +---
 drivers/infiniband/hw/hfi1/vnic_main.c   |   2 +-
 9 files changed, 183 insertions(+), 85 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index 9b1fb84a3d45..b4fe37a4db01 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -6862,7 +6862,7 @@ static void rxe_kernel_unfreeze(struct hfi1_devdata *dd)
 		}
 		rcvmask = HFI1_RCVCTRL_CTXT_ENB;
 		/* HFI1_RCVCTRL_TAILUPD_[ENB|DIS] needs to be set explicitly */
-		rcvmask |= rcd->rcvhdrtail_kvaddr ?
+		rcvmask |= hfi1_rcvhdrtail_kvaddr(rcd) ?
 			HFI1_RCVCTRL_TAILUPD_ENB : HFI1_RCVCTRL_TAILUPD_DIS;
 		hfi1_rcvctrl(dd, rcvmask, rcd);
 		hfi1_rcd_put(rcd);
@@ -8394,20 +8394,13 @@ void force_recv_intr(struct hfi1_ctxtdata *rcd)
 static inline int check_packet_present(struct hfi1_ctxtdata *rcd)
 {
 	u32 tail;
-	int present;
-
-	if (!rcd->rcvhdrtail_kvaddr)
-		present = (rcd->seq_cnt ==
-				rhf_rcv_seq(rhf_to_cpu(get_rhf_addr(rcd))));
-	else /* is RDMA rtail */
-		present = (rcd->head != get_rcvhdrtail(rcd));
 
-	if (present)
+	if (hfi1_packet_present(rcd))
 		return 1;
 
 	/* fall back to a CSR read, correct indpendent of DMA_RTAIL */
 	tail = (u32)read_uctxt_csr(rcd->dd, rcd->ctxt, RCV_HDR_TAIL);
-	return rcd->head != tail;
+	return hfi1_rcd_head(rcd) != tail;
 }
 
 /*
@@ -10049,7 +10042,7 @@ u32 lrh_max_header_bytes(struct hfi1_devdata *dd)
 	 * the first kernel context would have been allocated by now so
 	 * we are guaranteed a valid value.
 	 */
-	return (dd->rcd[0]->rcvhdrqentsize - 2/*PBC/RHF*/ + 1/*ICRC*/) << 2;
+	return (get_hdrqentsize(dd->rcd[0]) - 2/*PBC/RHF*/ + 1/*ICRC*/) << 2;
 }
 
 /*
@@ -10094,7 +10087,7 @@ static void set_send_length(struct hfi1_pportdata *ppd)
 		thres = min(sc_percent_to_threshold(dd->vld[i].sc, 50),
 			    sc_mtu_to_threshold(dd->vld[i].sc,
 						dd->vld[i].mtu,
-						dd->rcd[0]->rcvhdrqentsize));
+						get_hdrqentsize(dd->rcd[0])));
 		for (j = 0; j < INIT_SC_PER_VL; j++)
 			sc_set_cr_threshold(
 					pio_select_send_context_vl(dd, j, i),
@@ -11821,7 +11814,7 @@ u32 hdrqempty(struct hfi1_ctxtdata *rcd)
 	head = (read_uctxt_csr(rcd->dd, rcd->ctxt, RCV_HDR_HEAD)
 		& RCV_HDR_HEAD_HEAD_SMASK) >> RCV_HDR_HEAD_HEAD_SHIFT;
 
-	if (rcd->rcvhdrtail_kvaddr)
+	if (hfi1_rcvhdrtail_kvaddr(rcd))
 		tail = get_rcvhdrtail(rcd);
 	else
 		tail = read_uctxt_csr(rcd->dd, rcd->ctxt, RCV_HDR_TAIL);
@@ -11886,13 +11879,13 @@ void hfi1_rcvctrl(struct hfi1_devdata *dd, unsigned int op,
 		/* reset the tail and hdr addresses, and sequence count */
 		write_kctxt_csr(dd, ctxt, RCV_HDR_ADDR,
 				rcd->rcvhdrq_dma);
-		if (rcd->rcvhdrtail_kvaddr)
+		if (hfi1_rcvhdrtail_kvaddr(rcd))
 			write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR,
 					rcd->rcvhdrqtailaddr_dma);
-		rcd->seq_cnt = 1;
+		hfi1_set_seq_cnt(rcd, 1);
 
 		/* reset the cached receive header queue head value */
-		rcd->head = 0;
+		hfi1_set_rcd_head(rcd, 0);
 
 		/*
 		 * Zero the receive header queue so we don't get false
@@ -11972,7 +11965,7 @@ void hfi1_rcvctrl(struct hfi1_devdata *dd, unsigned int op,
 			      IS_RCVAVAIL_START + rcd->ctxt, false);
 		rcvctrl &= ~RCV_CTXT_CTRL_INTR_AVAIL_SMASK;
 	}
-	if ((op & HFI1_RCVCTRL_TAILUPD_ENB) && rcd->rcvhdrtail_kvaddr)
+	if ((op & HFI1_RCVCTRL_TAILUPD_ENB) && hfi1_rcvhdrtail_kvaddr(rcd))
 		rcvctrl |= RCV_CTXT_CTRL_TAIL_UPD_SMASK;
 	if (op & HFI1_RCVCTRL_TAILUPD_DIS) {
 		/* See comment on RcvCtxtCtrl.TailUpd above */
diff --git a/drivers/infiniband/hw/hfi1/common.h b/drivers/infiniband/hw/hfi1/common.h
index d47da7b0438f..40a1ff0c8a8e 100644
--- a/drivers/infiniband/hw/hfi1/common.h
+++ b/drivers/infiniband/hw/hfi1/common.h
@@ -323,6 +323,9 @@ struct diag_pkt {
 /* RHF receive type error - bypass packet errors */
 #define RHF_RTE_BYPASS_NO_ERR		0x0
 
+/* MAX RcvSEQ */
+#define RHF_MAX_SEQ 13
+
 /* IB - LRH header constants */
 #define HFI1_LRH_GRH 0x0003      /* 1. word of IB LRH - next header: GRH */
 #define HFI1_LRH_BTH 0x0002      /* 1. word of IB LRH - next header: BTH */
diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index 01aa1f132f55..cbc521930900 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -411,14 +411,14 @@ drop:
 static inline void init_packet(struct hfi1_ctxtdata *rcd,
 			       struct hfi1_packet *packet)
 {
-	packet->rsize = rcd->rcvhdrqentsize; /* words */
-	packet->maxcnt = rcd->rcvhdrq_cnt * packet->rsize; /* words */
+	packet->rsize = get_hdrqentsize(rcd); /* words */
+	packet->maxcnt = get_hdrq_cnt(rcd) * packet->rsize; /* words */
 	packet->rcd = rcd;
 	packet->updegr = 0;
 	packet->etail = -1;
 	packet->rhf_addr = get_rhf_addr(rcd);
 	packet->rhf = rhf_to_cpu(packet->rhf_addr);
-	packet->rhqoff = rcd->head;
+	packet->rhqoff = hfi1_rcd_head(rcd);
 	packet->numpkt = 0;
 }
 
@@ -551,22 +551,22 @@ static inline void init_ps_mdata(struct ps_mdata *mdata,
 	mdata->maxcnt = packet->maxcnt;
 	mdata->ps_head = packet->rhqoff;
 
-	if (HFI1_CAP_KGET_MASK(rcd->flags, DMA_RTAIL)) {
+	if (get_dma_rtail_setting(rcd)) {
 		mdata->ps_tail = get_rcvhdrtail(rcd);
 		if (rcd->ctxt == HFI1_CTRL_CTXT)
-			mdata->ps_seq = rcd->seq_cnt;
+			mdata->ps_seq = hfi1_seq_cnt(rcd);
 		else
 			mdata->ps_seq = 0; /* not used with DMA_RTAIL */
 	} else {
 		mdata->ps_tail = 0; /* used only with DMA_RTAIL*/
-		mdata->ps_seq = rcd->seq_cnt;
+		mdata->ps_seq = hfi1_seq_cnt(rcd);
 	}
 }
 
 static inline int ps_done(struct ps_mdata *mdata, u64 rhf,
 			  struct hfi1_ctxtdata *rcd)
 {
-	if (HFI1_CAP_KGET_MASK(rcd->flags, DMA_RTAIL))
+	if (get_dma_rtail_setting(rcd))
 		return mdata->ps_head == mdata->ps_tail;
 	return mdata->ps_seq != rhf_rcv_seq(rhf);
 }
@@ -592,11 +592,9 @@ static inline void update_ps_mdata(struct ps_mdata *mdata,
 		mdata->ps_head = 0;
 
 	/* Control context must do seq counting */
-	if (!HFI1_CAP_KGET_MASK(rcd->flags, DMA_RTAIL) ||
-	    (rcd->ctxt == HFI1_CTRL_CTXT)) {
-		if (++mdata->ps_seq > 13)
-			mdata->ps_seq = 1;
-	}
+	if (!get_dma_rtail_setting(rcd) ||
+	    rcd->ctxt == HFI1_CTRL_CTXT)
+		mdata->ps_seq = hfi1_seq_incr_wrap(mdata->ps_seq);
 }
 
 /*
@@ -769,7 +767,7 @@ static inline int process_rcv_packet(struct hfi1_packet *packet, int thread)
 		 * The +2 is the size of the RHF.
 		 */
 		prefetch_range(packet->ebuf,
-			       packet->tlen - ((packet->rcd->rcvhdrqentsize -
+			       packet->tlen - ((get_hdrqentsize(packet->rcd) -
 					       (rhf_hdrq_offset(packet->rhf)
 						+ 2)) * 4));
 	}
@@ -823,7 +821,7 @@ static inline void finish_packet(struct hfi1_packet *packet)
 	 * The only thing we need to do is a final update and call for an
 	 * interrupt
 	 */
-	update_usrhead(packet->rcd, packet->rcd->head, packet->updegr,
+	update_usrhead(packet->rcd, hfi1_rcd_head(packet->rcd), packet->updegr,
 		       packet->etail, rcv_intr_dynamic, packet->numpkt);
 }
 
@@ -832,13 +830,11 @@ static inline void finish_packet(struct hfi1_packet *packet)
  */
 int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *rcd, int thread)
 {
-	u32 seq;
 	int last = RCV_PKT_OK;
 	struct hfi1_packet packet;
 
 	init_packet(rcd, &packet);
-	seq = rhf_rcv_seq(packet.rhf);
-	if (seq != rcd->seq_cnt) {
+	if (last_rcv_seq(rcd, rhf_rcv_seq(packet.rhf))) {
 		last = RCV_PKT_DONE;
 		goto bail;
 	}
@@ -847,15 +843,12 @@ int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *rcd, int thread)
 
 	while (last == RCV_PKT_OK) {
 		last = process_rcv_packet(&packet, thread);
-		seq = rhf_rcv_seq(packet.rhf);
-		if (++rcd->seq_cnt > 13)
-			rcd->seq_cnt = 1;
-		if (seq != rcd->seq_cnt)
+		if (hfi1_seq_incr(rcd, rhf_rcv_seq(packet.rhf)))
 			last = RCV_PKT_DONE;
 		process_rcv_update(last, &packet);
 	}
 	process_rcv_qp_work(&packet);
-	rcd->head = packet.rhqoff;
+	hfi1_set_rcd_head(rcd, packet.rhqoff);
 bail:
 	finish_packet(&packet);
 	return last;
@@ -884,7 +877,7 @@ int handle_receive_interrupt_dma_rtail(struct hfi1_ctxtdata *rcd, int thread)
 		process_rcv_update(last, &packet);
 	}
 	process_rcv_qp_work(&packet);
-	rcd->head = packet.rhqoff;
+	hfi1_set_rcd_head(rcd, packet.rhqoff);
 bail:
 	finish_packet(&packet);
 	return last;
@@ -1019,10 +1012,8 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 
 	init_packet(rcd, &packet);
 
-	if (!HFI1_CAP_KGET_MASK(rcd->flags, DMA_RTAIL)) {
-		u32 seq = rhf_rcv_seq(packet.rhf);
-
-		if (seq != rcd->seq_cnt) {
+	if (!get_dma_rtail_setting(rcd)) {
+		if (last_rcv_seq(rcd, rhf_rcv_seq(packet.rhf))) {
 			last = RCV_PKT_DONE;
 			goto bail;
 		}
@@ -1039,12 +1030,9 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 		 * Control context can potentially receive an invalid
 		 * rhf. Drop such packets.
 		 */
-		if (rcd->ctxt == HFI1_CTRL_CTXT) {
-			u32 seq = rhf_rcv_seq(packet.rhf);
-
-			if (seq != rcd->seq_cnt)
+		if (rcd->ctxt == HFI1_CTRL_CTXT)
+			if (last_rcv_seq(rcd, rhf_rcv_seq(packet.rhf)))
 				skip_pkt = 1;
-		}
 	}
 
 	prescan_rxq(rcd, &packet);
@@ -1074,12 +1062,8 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 			last = process_rcv_packet(&packet, thread);
 		}
 
-		if (!HFI1_CAP_KGET_MASK(rcd->flags, DMA_RTAIL)) {
-			u32 seq = rhf_rcv_seq(packet.rhf);
-
-			if (++rcd->seq_cnt > 13)
-				rcd->seq_cnt = 1;
-			if (seq != rcd->seq_cnt)
+		if (!get_dma_rtail_setting(rcd)) {
+			if (hfi1_seq_incr(rcd, rhf_rcv_seq(packet.rhf)))
 				last = RCV_PKT_DONE;
 			if (needset) {
 				dd_dev_info(dd, "Switching to NO_DMA_RTAIL\n");
@@ -1094,11 +1078,11 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 			 * rhf. Drop such packets.
 			 */
 			if (rcd->ctxt == HFI1_CTRL_CTXT) {
-				u32 seq = rhf_rcv_seq(packet.rhf);
+				bool lseq;
 
-				if (++rcd->seq_cnt > 13)
-					rcd->seq_cnt = 1;
-				if (!last && (seq != rcd->seq_cnt))
+				lseq = hfi1_seq_incr(rcd,
+						     rhf_rcv_seq(packet.rhf));
+				if (!last && lseq)
 					skip_pkt = 1;
 			}
 
@@ -1114,7 +1098,7 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 	}
 
 	process_rcv_qp_work(&packet);
-	rcd->head = packet.rhqoff;
+	hfi1_set_rcd_head(rcd, packet.rhqoff);
 
 bail:
 	/*
@@ -1748,8 +1732,8 @@ void seqfile_dump_rcd(struct seq_file *s, struct hfi1_ctxtdata *rcd)
 	struct ps_mdata mdata;
 
 	seq_printf(s, "Rcd %u: RcvHdr cnt %u entsize %u %s head %llu tail %llu\n",
-		   rcd->ctxt, rcd->rcvhdrq_cnt, rcd->rcvhdrqentsize,
-		   HFI1_CAP_KGET_MASK(rcd->flags, DMA_RTAIL) ?
+		   rcd->ctxt, get_hdrq_cnt(rcd), get_hdrqentsize(rcd),
+		   get_dma_rtail_setting(rcd) ?
 		   "dma_rtail" : "nodma_rtail",
 		   read_uctxt_csr(rcd->dd, rcd->ctxt, RCV_HDR_HEAD) &
 		   RCV_HDR_HEAD_HEAD_MASK,
diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c
index 7c5e3fb22413..bef6946861b2 100644
--- a/drivers/infiniband/hw/hfi1/file_ops.c
+++ b/drivers/infiniband/hw/hfi1/file_ops.c
@@ -505,12 +505,12 @@ static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma)
 			ret = -EINVAL;
 			goto done;
 		}
-		if ((flags & VM_WRITE) || !uctxt->rcvhdrtail_kvaddr) {
+		if ((flags & VM_WRITE) || !hfi1_rcvhdrtail_kvaddr(uctxt)) {
 			ret = -EPERM;
 			goto done;
 		}
 		memlen = PAGE_SIZE;
-		memvirt = (void *)uctxt->rcvhdrtail_kvaddr;
+		memvirt = (void *)hfi1_rcvhdrtail_kvaddr(uctxt);
 		flags &= ~VM_MAYWRITE;
 		break;
 	case SUBCTXT_UREGS:
@@ -1090,7 +1090,7 @@ static void user_init(struct hfi1_ctxtdata *uctxt)
 	 * don't have to wait to be sure the DMA update has happened
 	 * (chip resets head/tail to 0 on transition to enable).
 	 */
-	if (uctxt->rcvhdrtail_kvaddr)
+	if (hfi1_rcvhdrtail_kvaddr(uctxt))
 		clear_rcvhdrtail(uctxt);
 
 	/* Setup J_KEY before enabling the context */
@@ -1154,8 +1154,8 @@ static int get_ctxt_info(struct hfi1_filedata *fd, unsigned long arg, u32 len)
 	cinfo.send_ctxt = uctxt->sc->hw_context;
 
 	cinfo.egrtids = uctxt->egrbufs.alloced;
-	cinfo.rcvhdrq_cnt = uctxt->rcvhdrq_cnt;
-	cinfo.rcvhdrq_entsize = uctxt->rcvhdrqentsize << 2;
+	cinfo.rcvhdrq_cnt = get_hdrq_cnt(uctxt);
+	cinfo.rcvhdrq_entsize = get_hdrqentsize(uctxt) << 2;
 	cinfo.sdma_ring_size = fd->cq->nentries;
 	cinfo.rcvegr_size = uctxt->egrbufs.rcvtid_size;
 
@@ -1543,7 +1543,7 @@ static int manage_rcvq(struct hfi1_ctxtdata *uctxt, u16 subctxt,
 		 * always resets it's tail register back to 0 on a
 		 * transition from disabled to enabled.
 		 */
-		if (uctxt->rcvhdrtail_kvaddr)
+		if (hfi1_rcvhdrtail_kvaddr(uctxt))
 			clear_rcvhdrtail(uctxt);
 		rcvctrl_op = HFI1_RCVCTRL_CTXT_ENB;
 	} else {
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index fc10d65fc3e1..d662fcc9a1b4 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -1507,12 +1507,115 @@ void hfi1_make_ud_req_16B(struct rvt_qp *qp,
 #define RCV_PKT_LIMIT   0x1 /* stop, hit limit, start thread */
 #define RCV_PKT_DONE    0x2 /* stop, no more packets detected */
 
+/**
+ * hfi1_rcd_head - add accessor for rcd head
+ * @rcd: the context
+ */
+static inline u32 hfi1_rcd_head(struct hfi1_ctxtdata *rcd)
+{
+	return rcd->head;
+}
+
+/**
+ * hfi1_set_rcd_head - add accessor for rcd head
+ * @rcd: the context
+ * @head: the new head
+ */
+static inline void hfi1_set_rcd_head(struct hfi1_ctxtdata *rcd, u32 head)
+{
+	rcd->head = head;
+}
+
 /* calculate the current RHF address */
 static inline __le32 *get_rhf_addr(struct hfi1_ctxtdata *rcd)
 {
 	return (__le32 *)rcd->rcvhdrq + rcd->head + rcd->rhf_offset;
 }
 
+/* return DMA_RTAIL configuration */
+static inline bool get_dma_rtail_setting(struct hfi1_ctxtdata *rcd)
+{
+	return !!HFI1_CAP_KGET_MASK(rcd->flags, DMA_RTAIL);
+}
+
+/**
+ * hfi1_seq_incr_wrap - wrapping increment for sequence
+ * @seq: the current sequence number
+ *
+ * Returns: the incremented seq
+ */
+static inline u8 hfi1_seq_incr_wrap(u8 seq)
+{
+	if (++seq > RHF_MAX_SEQ)
+		seq = 1;
+	return seq;
+}
+
+/**
+ * hfi1_seq_cnt - return seq_cnt member
+ * @rcd: the receive context
+ *
+ * Return seq_cnt member
+ */
+static inline u8 hfi1_seq_cnt(struct hfi1_ctxtdata *rcd)
+{
+	return rcd->seq_cnt;
+}
+
+/**
+ * hfi1_set_seq_cnt - return seq_cnt member
+ * @rcd: the receive context
+ *
+ * Return seq_cnt member
+ */
+static inline void hfi1_set_seq_cnt(struct hfi1_ctxtdata *rcd, u8 cnt)
+{
+	rcd->seq_cnt = cnt;
+}
+
+/**
+ * last_rcv_seq - is last
+ * @rcd: the receive context
+ * @seq: sequence
+ *
+ * return true if last packet
+ */
+static inline bool last_rcv_seq(struct hfi1_ctxtdata *rcd, u32 seq)
+{
+	return seq != rcd->seq_cnt;
+}
+
+/**
+ * rcd_seq_incr - increment context sequence number
+ * @rcd: the receive context
+ * @seq: the current sequence number
+ *
+ * Returns: true if the this was the last packet
+ */
+static inline bool hfi1_seq_incr(struct hfi1_ctxtdata *rcd, u32 seq)
+{
+	rcd->seq_cnt = hfi1_seq_incr_wrap(rcd->seq_cnt);
+	return last_rcv_seq(rcd, seq);
+}
+
+/**
+ * get_hdrqentsize - return hdrq entry size
+ * @rcd: the receive context
+ */
+static inline u8 get_hdrqentsize(struct hfi1_ctxtdata *rcd)
+{
+	return rcd->rcvhdrqentsize;
+}
+
+/**
+ * get_hdrq_cnt - return hdrq count
+ * @rcd: the receive context
+ */
+static inline u16 get_hdrq_cnt(struct hfi1_ctxtdata *rcd)
+{
+	return rcd->rcvhdrq_cnt;
+}
+
 int hfi1_reset_device(int);
 
 void receive_interrupt_work(struct work_struct *work);
@@ -2015,9 +2118,21 @@ int hfi1_acquire_user_pages(struct mm_struct *mm, unsigned long vaddr,
 void hfi1_release_user_pages(struct mm_struct *mm, struct page **p,
 			     size_t npages, bool dirty);
 
+/**
+ * hfi1_rcvhdrtail_kvaddr - return tail kvaddr
+ * @rcd - the receive context
+ */
+static inline __le64 *hfi1_rcvhdrtail_kvaddr(const struct hfi1_ctxtdata *rcd)
+{
+	return (__le64 *)rcd->rcvhdrtail_kvaddr;
+}
+
 static inline void clear_rcvhdrtail(const struct hfi1_ctxtdata *rcd)
 {
-	*((u64 *)rcd->rcvhdrtail_kvaddr) = 0ULL;
+	u64 *kv = (u64 *)hfi1_rcvhdrtail_kvaddr(rcd);
+
+	if (kv)
+		*kv = 0ULL;
 }
 
 static inline u32 get_rcvhdrtail(const struct hfi1_ctxtdata *rcd)
@@ -2026,7 +2141,17 @@ static inline u32 get_rcvhdrtail(const struct hfi1_ctxtdata *rcd)
 	 * volatile because it's a DMA target from the chip, routine is
 	 * inlined, and don't want register caching or reordering.
 	 */
-	return (u32)le64_to_cpu(*rcd->rcvhdrtail_kvaddr);
+	return (u32)le64_to_cpu(*hfi1_rcvhdrtail_kvaddr(rcd));
+}
+
+static inline bool hfi1_packet_present(struct hfi1_ctxtdata *rcd)
+{
+	if (likely(!rcd->rcvhdrtail_kvaddr)) {
+		u32 seq = rhf_rcv_seq(rhf_to_cpu(get_rhf_addr(rcd)));
+
+		return !last_rcv_seq(rcd, seq);
+	}
+	return hfi1_rcd_head(rcd) != get_rcvhdrtail(rcd);
 }
 
 /*
diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c
index 26b792bb1027..3e56ee37861d 100644
--- a/drivers/infiniband/hw/hfi1/init.c
+++ b/drivers/infiniband/hw/hfi1/init.c
@@ -154,7 +154,7 @@ static int hfi1_create_kctxt(struct hfi1_devdata *dd,
 	/* Control context must use DMA_RTAIL */
 	if (rcd->ctxt == HFI1_CTRL_CTXT)
 		rcd->flags |= HFI1_CAP_DMA_RTAIL;
-	rcd->seq_cnt = 1;
+	hfi1_set_seq_cnt(rcd, 1);
 
 	rcd->sc = sc_alloc(dd, SC_ACK, rcd->rcvhdrqentsize, dd->node);
 	if (!rcd->sc) {
@@ -1149,9 +1149,9 @@ void hfi1_free_ctxtdata(struct hfi1_devdata *dd, struct hfi1_ctxtdata *rcd)
 		dma_free_coherent(&dd->pcidev->dev, rcvhdrq_size(rcd),
 				  rcd->rcvhdrq, rcd->rcvhdrq_dma);
 		rcd->rcvhdrq = NULL;
-		if (rcd->rcvhdrtail_kvaddr) {
+		if (hfi1_rcvhdrtail_kvaddr(rcd)) {
 			dma_free_coherent(&dd->pcidev->dev, PAGE_SIZE,
-					  (void *)rcd->rcvhdrtail_kvaddr,
+					  (void *)hfi1_rcvhdrtail_kvaddr(rcd),
 					  rcd->rcvhdrqtailaddr_dma);
 			rcd->rcvhdrtail_kvaddr = NULL;
 		}
diff --git a/drivers/infiniband/hw/hfi1/trace_ctxts.h b/drivers/infiniband/hw/hfi1/trace_ctxts.h
index e00c8a7d559c..b5fc5c6cd52f 100644
--- a/drivers/infiniband/hw/hfi1/trace_ctxts.h
+++ b/drivers/infiniband/hw/hfi1/trace_ctxts.h
@@ -80,7 +80,7 @@ TRACE_EVENT(hfi1_uctxtdata,
 			   __entry->credits = uctxt->sc->credits;
 			   __entry->hw_free = le64_to_cpu(*uctxt->sc->hw_free);
 			   __entry->piobase = uctxt->sc->base_addr;
-			   __entry->rcvhdrq_cnt = uctxt->rcvhdrq_cnt;
+			   __entry->rcvhdrq_cnt = get_hdrq_cnt(uctxt);
 			   __entry->rcvhdrq_dma = uctxt->rcvhdrq_dma;
 			   __entry->eager_cnt = uctxt->egrbufs.alloced;
 			   __entry->rcvegr_dma = uctxt->egrbufs.rcvtids[0].dma;
diff --git a/drivers/infiniband/hw/hfi1/trace_rx.h b/drivers/infiniband/hw/hfi1/trace_rx.h
index 3cec960e9674..ebce81c7ac41 100644
--- a/drivers/infiniband/hw/hfi1/trace_rx.h
+++ b/drivers/infiniband/hw/hfi1/trace_rx.h
@@ -107,18 +107,11 @@ TRACE_EVENT(hfi1_receive_interrupt,
 	    TP_fast_assign(DD_DEV_ASSIGN(dd);
 			__entry->ctxt = rcd->ctxt;
 			if (rcd->do_interrupt ==
-			    &handle_receive_interrupt) {
+			    &handle_receive_interrupt)
 				__entry->slow_path = 1;
-				__entry->dma_rtail = 0xFF;
-			} else if (rcd->do_interrupt ==
-					&handle_receive_interrupt_dma_rtail){
-				__entry->dma_rtail = 1;
+			else
 				__entry->slow_path = 0;
-			} else if (rcd->do_interrupt ==
-					&handle_receive_interrupt_nodma_rtail) {
-				__entry->dma_rtail = 0;
-				__entry->slow_path = 0;
-			}
+			__entry->dma_rtail = get_dma_rtail_setting(rcd);
 			),
 	    TP_printk("[%s] ctxt %d SlowPath: %d DmaRtail: %d",
 		      __get_str(dev),
diff --git a/drivers/infiniband/hw/hfi1/vnic_main.c b/drivers/infiniband/hw/hfi1/vnic_main.c
index b49e60e8397d..6b14581b9965 100644
--- a/drivers/infiniband/hw/hfi1/vnic_main.c
+++ b/drivers/infiniband/hw/hfi1/vnic_main.c
@@ -78,7 +78,7 @@ static int setup_vnic_ctxt(struct hfi1_devdata *dd, struct hfi1_ctxtdata *uctxt)
 	if (ret)
 		goto done;
 
-	if (uctxt->rcvhdrtail_kvaddr)
+	if (hfi1_rcvhdrtail_kvaddr(uctxt))
 		clear_rcvhdrtail(uctxt);
 
 	rcvctrl_ops = HFI1_RCVCTRL_CTXT_ENB;
-- 
cgit v1.2.3-59-g8ed1b


From 44ec5aa3c6158122412b3018fdea1474edf0898e Mon Sep 17 00:00:00 2001
From: "Michael J. Ruhl" <michael.j.ruhl@intel.com>
Date: Thu, 19 Dec 2019 16:19:28 -0500
Subject: IB/hfi1: List all receive contexts from debugfs

The current debugfs output for receive contexts (rcds), stops after the
kernel receive contexts have been displayed.  This is not enough
information to fully diagnose packet drops.

Display all of the receive contexts.

Augment the output with some more context information.

Limit the ring buffer header output to 5 entries to avoid overextending
the sequential file output.

Fixes: bf808b5039c ("IB/hfi1: Add kernel receive context info to debugfs")
Link: https://lore.kernel.org/r/20191219211928.58387.20737.stgit@awfm-01.aw.intel.com
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/debugfs.c |  2 +-
 drivers/infiniband/hw/hfi1/driver.c  | 12 +++++++++---
 2 files changed, 10 insertions(+), 4 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/debugfs.c b/drivers/infiniband/hw/hfi1/debugfs.c
index d268bf9c42ee..4633a0ce1a8c 100644
--- a/drivers/infiniband/hw/hfi1/debugfs.c
+++ b/drivers/infiniband/hw/hfi1/debugfs.c
@@ -379,7 +379,7 @@ static void *_rcds_seq_next(struct seq_file *s, void *v, loff_t *pos)
 	struct hfi1_devdata *dd = dd_from_dev(ibd);
 
 	++*pos;
-	if (!dd->rcd || *pos >= dd->n_krcv_queues)
+	if (!dd->rcd || *pos >= dd->num_rcv_contexts)
 		return NULL;
 	return pos;
 }
diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index cbc521930900..8374922ea6e5 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -1726,23 +1726,29 @@ static int process_receive_invalid(struct hfi1_packet *packet)
 	return RHF_RCV_CONTINUE;
 }
 
+#define HFI1_RCVHDR_DUMP_MAX	5
+
 void seqfile_dump_rcd(struct seq_file *s, struct hfi1_ctxtdata *rcd)
 {
 	struct hfi1_packet packet;
 	struct ps_mdata mdata;
+	int i;
 
-	seq_printf(s, "Rcd %u: RcvHdr cnt %u entsize %u %s head %llu tail %llu\n",
+	seq_printf(s, "Rcd %u: RcvHdr cnt %u entsize %u %s ctrl 0x%08llx status 0x%08llx, head %llu tail %llu  sw head %u\n",
 		   rcd->ctxt, get_hdrq_cnt(rcd), get_hdrqentsize(rcd),
 		   get_dma_rtail_setting(rcd) ?
 		   "dma_rtail" : "nodma_rtail",
+		   read_kctxt_csr(rcd->dd, rcd->ctxt, RCV_CTXT_CTRL),
+		   read_kctxt_csr(rcd->dd, rcd->ctxt, RCV_CTXT_STATUS),
 		   read_uctxt_csr(rcd->dd, rcd->ctxt, RCV_HDR_HEAD) &
 		   RCV_HDR_HEAD_HEAD_MASK,
-		   read_uctxt_csr(rcd->dd, rcd->ctxt, RCV_HDR_TAIL));
+		   read_uctxt_csr(rcd->dd, rcd->ctxt, RCV_HDR_TAIL),
+		   rcd->head);
 
 	init_packet(rcd, &packet);
 	init_ps_mdata(&mdata, &packet);
 
-	while (1) {
+	for (i = 0; i < HFI1_RCVHDR_DUMP_MAX; i++) {
 		__le32 *rhf_addr = (__le32 *)rcd->rcvhdrq + mdata.ps_head +
 					 rcd->rhf_offset;
 		struct ib_header *hdr;
-- 
cgit v1.2.3-59-g8ed1b


From 4ad6429d2713924afe10cd960535600b241560a3 Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Thu, 19 Dec 2019 16:19:34 -0500
Subject: IB/rdmavt: Correct comments in rdmavt_qp.h header

Comments need to be with the definition of rvt_restart_sge().

Other comments were duplicated in sw/rdmavt/rc.c and were removed.

Fixes: 385156c5f2a6 ("IB/hfi: Move RC functions into a header file")
Link: https://lore.kernel.org/r/20191219211934.58387.88014.stgit@awfm-01.aw.intel.com
Reviewed-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/sw/rdmavt/rc.c |  9 ++++++++-
 include/rdma/rdmavt_qp.h          | 22 +---------------------
 2 files changed, 9 insertions(+), 22 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/rdmavt/rc.c b/drivers/infiniband/sw/rdmavt/rc.c
index 890d7b760d2e..977906cc0d11 100644
--- a/drivers/infiniband/sw/rdmavt/rc.c
+++ b/drivers/infiniband/sw/rdmavt/rc.c
@@ -195,7 +195,14 @@ void rvt_get_credit(struct rvt_qp *qp, u32 aeth)
 }
 EXPORT_SYMBOL(rvt_get_credit);
 
-/* rvt_restart_sge - rewind the sge state for a wqe */
+/**
+ * rvt_restart_sge - rewind the sge state for a wqe
+ * @ss: the sge state pointer
+ * @wqe: the wqe to rewind
+ * @len: the data length from the start of the wqe in bytes
+ *
+ * Returns the remaining data length.
+ */
 u32 rvt_restart_sge(struct rvt_sge_state *ss, struct rvt_swqe *wqe, u32 len)
 {
 	ss->sge = wqe->sg_list[0];
diff --git a/include/rdma/rdmavt_qp.h b/include/rdma/rdmavt_qp.h
index b550ae89bf85..0d5c70e2d8ab 100644
--- a/include/rdma/rdmavt_qp.h
+++ b/include/rdma/rdmavt_qp.h
@@ -640,34 +640,14 @@ static inline int rvt_cmp_msn(u32 a, u32 b)
 	return (((int)a) - ((int)b)) << 8;
 }
 
-/**
- * rvt_compute_aeth - compute the AETH (syndrome + MSN)
- * @qp: the queue pair to compute the AETH for
- *
- * Returns the AETH.
- */
 __be32 rvt_compute_aeth(struct rvt_qp *qp);
 
-/**
- * rvt_get_credit - flush the send work queue of a QP
- * @qp: the qp who's send work queue to flush
- * @aeth: the Acknowledge Extended Transport Header
- *
- * The QP s_lock should be held.
- */
 void rvt_get_credit(struct rvt_qp *qp, u32 aeth);
 
-/**
- * rvt_restart_sge - rewind the sge state for a wqe
- * @ss: the sge state pointer
- * @wqe: the wqe to rewind
- * @len: the data length from the start of the wqe in bytes
- *
- * Returns the remaining data length.
- */
 u32 rvt_restart_sge(struct rvt_sge_state *ss, struct rvt_swqe *wqe, u32 len);
 
 /**
+ * rvt_div_round_up_mtu - round up divide
  * @qp - the qp pair
  * @len - the length
  *
-- 
cgit v1.2.3-59-g8ed1b


From cbe4b8f0a5766a40563876932cba6c9bf28eb98a Mon Sep 17 00:00:00 2001
From: Artemy Kovalyov <artemyko@mellanox.com>
Date: Sun, 22 Dec 2019 14:46:47 +0200
Subject: IB/mlx5: Unify ODP MR code paths to allow extra flexibility

Building MR translation table in the ODP case requires additional
flexibility, namely random access to DMA addresses. Make both direct and
indirect ODP MR use same code path, separated from the non-ODP MR code
path.

With the restructuring the correct page_shift is now used around
__mlx5_ib_populate_pas().

Fixes: d2183c6f1958 ("RDMA/umem: Move page_shift from ib_umem to ib_odp_umem")
Link: https://lore.kernel.org/r/20191222124649.52300-2-leon@kernel.org
Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/mem.c     | 25 ----------------
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  9 +++---
 drivers/infiniband/hw/mlx5/mr.c      | 58 ++++++++++++++----------------------
 drivers/infiniband/hw/mlx5/odp.c     | 42 ++++++++++++++++++++++++--
 4 files changed, 67 insertions(+), 67 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/mem.c b/drivers/infiniband/hw/mlx5/mem.c
index 048f4e974a61..b90a3649e7d1 100644
--- a/drivers/infiniband/hw/mlx5/mem.c
+++ b/drivers/infiniband/hw/mlx5/mem.c
@@ -101,18 +101,6 @@ void mlx5_ib_cont_pages(struct ib_umem *umem, u64 addr,
 	*count = i;
 }
 
-static u64 umem_dma_to_mtt(dma_addr_t umem_dma)
-{
-	u64 mtt_entry = umem_dma & ODP_DMA_ADDR_MASK;
-
-	if (umem_dma & ODP_READ_ALLOWED_BIT)
-		mtt_entry |= MLX5_IB_MTT_READ;
-	if (umem_dma & ODP_WRITE_ALLOWED_BIT)
-		mtt_entry |= MLX5_IB_MTT_WRITE;
-
-	return mtt_entry;
-}
-
 /*
  * Populate the given array with bus addresses from the umem.
  *
@@ -139,19 +127,6 @@ void __mlx5_ib_populate_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem,
 	struct scatterlist *sg;
 	int entry;
 
-	if (umem->is_odp) {
-		WARN_ON(shift != 0);
-		WARN_ON(access_flags != (MLX5_IB_MTT_READ | MLX5_IB_MTT_WRITE));
-
-		for (i = 0; i < num_pages; ++i) {
-			dma_addr_t pa =
-				to_ib_umem_odp(umem)->dma_list[offset + i];
-
-			pas[i] = cpu_to_be64(umem_dma_to_mtt(pa));
-		}
-		return;
-	}
-
 	i = 0;
 	for_each_sg(umem->sg_head.sgl, sg, umem->nmap, entry) {
 		len = sg_dma_len(sg) >> PAGE_SHIFT;
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index b06f32ff5748..927948328184 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1276,8 +1276,8 @@ void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *ibdev);
 int __init mlx5_ib_odp_init(void);
 void mlx5_ib_odp_cleanup(void);
 void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent);
-void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset,
-			   size_t nentries, struct mlx5_ib_mr *mr, int flags);
+void mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
+			   struct mlx5_ib_mr *mr, int flags);
 
 int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd,
 			       enum ib_uverbs_advise_mr_advice advice,
@@ -1293,9 +1293,8 @@ static inline void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *ibdev) {}
 static inline int mlx5_ib_odp_init(void) { return 0; }
 static inline void mlx5_ib_odp_cleanup(void)				    {}
 static inline void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent) {}
-static inline void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset,
-					 size_t nentries, struct mlx5_ib_mr *mr,
-					 int flags) {}
+static inline void mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
+					 struct mlx5_ib_mr *mr, int flags) {}
 
 static inline int
 mlx5_ib_advise_mr_prefetch(struct ib_pd *pd,
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index ea8bfc3e2d8d..1e38ba0da16a 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -868,36 +868,6 @@ static struct mlx5_ib_mr *alloc_mr_from_cache(
 	return mr;
 }
 
-static inline int populate_xlt(struct mlx5_ib_mr *mr, int idx, int npages,
-			       void *xlt, int page_shift, size_t size,
-			       int flags)
-{
-	struct mlx5_ib_dev *dev = mr->dev;
-	struct ib_umem *umem = mr->umem;
-
-	if (flags & MLX5_IB_UPD_XLT_INDIRECT) {
-		if (!umr_can_use_indirect_mkey(dev))
-			return -EPERM;
-		mlx5_odp_populate_klm(xlt, idx, npages, mr, flags);
-		return npages;
-	}
-
-	npages = min_t(size_t, npages, ib_umem_num_pages(umem) - idx);
-
-	if (!(flags & MLX5_IB_UPD_XLT_ZAP)) {
-		__mlx5_ib_populate_pas(dev, umem, page_shift,
-				       idx, npages, xlt,
-				       MLX5_IB_MTT_PRESENT);
-		/* Clear padding after the pages
-		 * brought from the umem.
-		 */
-		memset(xlt + (npages * sizeof(struct mlx5_mtt)), 0,
-		       size - npages * sizeof(struct mlx5_mtt));
-	}
-
-	return npages;
-}
-
 #define MLX5_MAX_UMR_CHUNK ((1 << (MLX5_MAX_UMR_SHIFT + 4)) - \
 			    MLX5_UMR_MTT_ALIGNMENT)
 #define MLX5_SPARE_UMR_CHUNK 0x10000
@@ -921,6 +891,7 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
 	size_t pages_mapped = 0;
 	size_t pages_to_map = 0;
 	size_t pages_iter = 0;
+	size_t size_to_map = 0;
 	gfp_t gfp;
 	bool use_emergency_page = false;
 
@@ -967,6 +938,15 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
 		goto free_xlt;
 	}
 
+	if (mr->umem->is_odp) {
+		if (!(flags & MLX5_IB_UPD_XLT_INDIRECT)) {
+			struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem);
+			size_t max_pages = ib_umem_odp_num_pages(odp) - idx;
+
+			pages_to_map = min_t(size_t, pages_to_map, max_pages);
+		}
+	}
+
 	sg.addr = dma;
 	sg.lkey = dev->umrc.pd->local_dma_lkey;
 
@@ -989,14 +969,22 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
 	     pages_mapped < pages_to_map && !err;
 	     pages_mapped += pages_iter, idx += pages_iter) {
 		npages = min_t(int, pages_iter, pages_to_map - pages_mapped);
+		size_to_map = npages * desc_size;
 		dma_sync_single_for_cpu(ddev, dma, size, DMA_TO_DEVICE);
-		npages = populate_xlt(mr, idx, npages, xlt,
-				      page_shift, size, flags);
-
+		if (mr->umem->is_odp) {
+			mlx5_odp_populate_xlt(xlt, idx, npages, mr, flags);
+		} else {
+			__mlx5_ib_populate_pas(dev, mr->umem, page_shift, idx,
+					       npages, xlt,
+					       MLX5_IB_MTT_PRESENT);
+			/* Clear padding after the pages
+			 * brought from the umem.
+			 */
+			memset(xlt + size_to_map, 0, size - size_to_map);
+		}
 		dma_sync_single_for_device(ddev, dma, size, DMA_TO_DEVICE);
 
-		sg.length = ALIGN(npages * desc_size,
-				  MLX5_UMR_MTT_ALIGNMENT);
+		sg.length = ALIGN(size_to_map, MLX5_UMR_MTT_ALIGNMENT);
 
 		if (pages_mapped + pages_iter >= pages_to_map) {
 			if (flags & MLX5_IB_UPD_XLT_ENABLE)
diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index f924250f80c2..92da6c4f7ddd 100644
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -93,8 +93,8 @@ struct mlx5_pagefault {
 
 static u64 mlx5_imr_ksm_entries;
 
-void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t idx, size_t nentries,
-			   struct mlx5_ib_mr *imr, int flags)
+static void populate_klm(struct mlx5_klm *pklm, size_t idx, size_t nentries,
+			struct mlx5_ib_mr *imr, int flags)
 {
 	struct mlx5_klm *end = pklm + nentries;
 
@@ -144,6 +144,44 @@ void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t idx, size_t nentries,
 	}
 }
 
+static u64 umem_dma_to_mtt(dma_addr_t umem_dma)
+{
+	u64 mtt_entry = umem_dma & ODP_DMA_ADDR_MASK;
+
+	if (umem_dma & ODP_READ_ALLOWED_BIT)
+		mtt_entry |= MLX5_IB_MTT_READ;
+	if (umem_dma & ODP_WRITE_ALLOWED_BIT)
+		mtt_entry |= MLX5_IB_MTT_WRITE;
+
+	return mtt_entry;
+}
+
+static void populate_mtt(__be64 *pas, size_t idx, size_t nentries,
+			 struct mlx5_ib_mr *mr, int flags)
+{
+	struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem);
+	dma_addr_t pa;
+	size_t i;
+
+	if (flags & MLX5_IB_UPD_XLT_ZAP)
+		return;
+
+	for (i = 0; i < nentries; i++) {
+		pa = odp->dma_list[idx + i];
+		pas[i] = cpu_to_be64(umem_dma_to_mtt(pa));
+	}
+}
+
+void mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
+			   struct mlx5_ib_mr *mr, int flags)
+{
+	if (flags & MLX5_IB_UPD_XLT_INDIRECT) {
+		populate_klm(xlt, idx, nentries, mr, flags);
+	} else {
+		populate_mtt(xlt, idx, nentries, mr, flags);
+	}
+}
+
 static void dma_fence_odp_mr(struct mlx5_ib_mr *mr)
 {
 	struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem);
-- 
cgit v1.2.3-59-g8ed1b


From d07de8bd1709a80a282963ad7b2535148678a9e4 Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@mellanox.com>
Date: Sun, 22 Dec 2019 14:46:48 +0200
Subject: IB/core: Fix ODP get user pages flow

The nr_pages argument of get_user_pages_remote() should always be in terms
of the system page size, not the MR page size. Use PAGE_SIZE instead of
umem_odp->page_shift.

Fixes: 403cd12e2cf7 ("IB/umem: Add contiguous ODP support")
Link: https://lore.kernel.org/r/20191222124649.52300-3-leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/umem_odp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index e42d44e501fd..2e9ee7adab13 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -440,7 +440,7 @@ int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 user_virt,
 
 	while (bcnt > 0) {
 		const size_t gup_num_pages = min_t(size_t,
-				(bcnt + BIT(page_shift) - 1) >> page_shift,
+				ALIGN(bcnt, PAGE_SIZE) / PAGE_SIZE,
 				PAGE_SIZE / sizeof(struct page *));
 
 		down_read(&owning_mm->mmap_sem);
-- 
cgit v1.2.3-59-g8ed1b


From 9ff1b6466a291a33389c4a9c7f3f9b64d62df40a Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@mellanox.com>
Date: Sun, 22 Dec 2019 14:46:49 +0200
Subject: IB/core: Fix ODP with IB_ACCESS_HUGETLB handling

As VMAs for a given range might not be available as part of the
registration phase in ODP.

ib_init_umem_odp() considered the expected page shift value that was
previously set and initializes its internals accordingly.

If memory isn't backed by physical contiguous pages aligned to a hugepage
boundary an error will be set as part of the page fault flow and come back
to the user as some failed RDMA operation.

Fixes: 0008b84ea9af ("IB/umem: Add support to huge ODP")
Link: https://lore.kernel.org/r/20191222124649.52300-4-leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/umem_odp.c | 21 ++++-----------------
 1 file changed, 4 insertions(+), 17 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index 2e9ee7adab13..f42fa31c24a2 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -241,22 +241,10 @@ struct ib_umem_odp *ib_umem_odp_get(struct ib_udata *udata, unsigned long addr,
 	umem_odp->umem.owning_mm = mm = current->mm;
 	umem_odp->notifier.ops = ops;
 
-	umem_odp->page_shift = PAGE_SHIFT;
-	if (access & IB_ACCESS_HUGETLB) {
-		struct vm_area_struct *vma;
-		struct hstate *h;
-
-		down_read(&mm->mmap_sem);
-		vma = find_vma(mm, ib_umem_start(umem_odp));
-		if (!vma || !is_vm_hugetlb_page(vma)) {
-			up_read(&mm->mmap_sem);
-			ret = -EINVAL;
-			goto err_free;
-		}
-		h = hstate_vma(vma);
-		umem_odp->page_shift = huge_page_shift(h);
-		up_read(&mm->mmap_sem);
-	}
+	if (access & IB_ACCESS_HUGETLB)
+		umem_odp->page_shift = HPAGE_SHIFT;
+	else
+		umem_odp->page_shift = PAGE_SHIFT;
 
 	umem_odp->tgid = get_task_pid(current->group_leader, PIDTYPE_PID);
 	ret = ib_init_umem_odp(umem_odp, ops);
@@ -266,7 +254,6 @@ struct ib_umem_odp *ib_umem_odp_get(struct ib_udata *udata, unsigned long addr,
 
 err_put_pid:
 	put_pid(umem_odp->tgid);
-err_free:
 	kfree(umem_odp);
 	return ERR_PTR(ret);
 }
-- 
cgit v1.2.3-59-g8ed1b


From 5369b4828975fcd2d499791b3d25a4e7c0b96baa Mon Sep 17 00:00:00 2001
From: zhengbin <zhengbin13@huawei.com>
Date: Tue, 24 Dec 2019 16:40:08 +0800
Subject: RDMA/siw: use true,false for bool variable

Fixes coccicheck warning:

drivers/infiniband/sw/siw/siw_cm.c:32:18-41: WARNING: Assignment of 0/1 to bool variable

Link: https://lore.kernel.org/r/1577176812-2238-2-git-send-email-zhengbin13@huawei.com
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/sw/siw/siw_cm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/siw/siw_cm.c b/drivers/infiniband/sw/siw/siw_cm.c
index 3bccfef40e7e..0c3f0588346e 100644
--- a/drivers/infiniband/sw/siw/siw_cm.c
+++ b/drivers/infiniband/sw/siw/siw_cm.c
@@ -29,7 +29,7 @@
  * MPA_V2_RDMA_NO_RTR, MPA_V2_RDMA_READ_RTR, MPA_V2_RDMA_WRITE_RTR
  */
 static __be16 rtr_type = MPA_V2_RDMA_READ_RTR | MPA_V2_RDMA_WRITE_RTR;
-static const bool relaxed_ird_negotiation = 1;
+static const bool relaxed_ird_negotiation = true;
 
 static void siw_cm_llp_state_change(struct sock *s);
 static void siw_cm_llp_data_ready(struct sock *s);
-- 
cgit v1.2.3-59-g8ed1b


From d09dbe74e963ffca7bfb0a8c0f98494a864654b6 Mon Sep 17 00:00:00 2001
From: zhengbin <zhengbin13@huawei.com>
Date: Tue, 24 Dec 2019 16:40:09 +0800
Subject: IB/hfi1: use true,false for bool variable

Fixes coccicheck warning:

drivers/infiniband/hw/hfi1/rc.c:2602:1-8: WARNING: Assignment of 0/1 to bool variable

Link: https://lore.kernel.org/r/1577176812-2238-3-git-send-email-zhengbin13@huawei.com
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/rc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c
index 1a3c647675a7..f1734e5e9ac4 100644
--- a/drivers/infiniband/hw/hfi1/rc.c
+++ b/drivers/infiniband/hw/hfi1/rc.c
@@ -2599,7 +2599,7 @@ static noinline int rc_rcv_error(struct ib_other_headers *ohdr, void *data,
 	 * to be sent before sending this one.
 	 */
 	e = NULL;
-	old_req = 1;
+	old_req = true;
 	ibp->rvp.n_rc_dupreq++;
 
 	spin_lock_irqsave(&qp->s_lock, flags);
-- 
cgit v1.2.3-59-g8ed1b


From c934833e772396e20e40715e01144a92faf11735 Mon Sep 17 00:00:00 2001
From: zhengbin <zhengbin13@huawei.com>
Date: Tue, 24 Dec 2019 16:40:10 +0800
Subject: IB/iser: use true,false for bool variable

Fixes coccicheck warning:

drivers/infiniband/ulp/iser/iser_memory.c:530:2-21: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/ulp/iser/iser_verbs.c:1096:2-21: WARNING: Assignment of 0/1 to bool variable

Link: https://lore.kernel.org/r/1577176812-2238-4-git-send-email-zhengbin13@huawei.com
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/ulp/iser/iser_memory.c | 2 +-
 drivers/infiniband/ulp/iser/iser_verbs.c  | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/ulp/iser/iser_memory.c b/drivers/infiniband/ulp/iser/iser_memory.c
index 0f74dc6d12fa..7a8f24de3631 100644
--- a/drivers/infiniband/ulp/iser/iser_memory.c
+++ b/drivers/infiniband/ulp/iser/iser_memory.c
@@ -527,7 +527,7 @@ int iser_reg_rdma_mem(struct iscsi_iser_task *task,
 		if (unlikely(err))
 			goto err_reg;
 
-		desc->sig_protected = 1;
+		desc->sig_protected = true;
 	}
 
 	return 0;
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c
index 1f4a37a3c2b3..127887c6c03f 100644
--- a/drivers/infiniband/ulp/iser/iser_verbs.c
+++ b/drivers/infiniband/ulp/iser/iser_verbs.c
@@ -1093,7 +1093,7 @@ u8 iser_check_task_pi_status(struct iscsi_iser_task *iser_task,
 	int ret;
 
 	if (desc && desc->sig_protected) {
-		desc->sig_protected = 0;
+		desc->sig_protected = false;
 		ret = ib_check_mr_status(desc->rsc.sig_mr,
 					 IB_MR_CHECK_SIG_STATUS, &mr_status);
 		if (ret) {
-- 
cgit v1.2.3-59-g8ed1b


From cf368beb8117e9d818f7b444943637fdc18d79f1 Mon Sep 17 00:00:00 2001
From: zhengbin <zhengbin13@huawei.com>
Date: Tue, 24 Dec 2019 16:40:11 +0800
Subject: RDMA/mlx4: use true,false for bool variable

Fixes coccicheck warning:

drivers/infiniband/hw/mlx4/qp.c:852:2-14: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/hw/mlx4/qp.c:3087:3-10: WARNING: Assignment of 0/1 to bool variable

Link: https://lore.kernel.org/r/1577176812-2238-5-git-send-email-zhengbin13@huawei.com
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx4/qp.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index 85f57b76e446..8d240bc92b6b 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -849,7 +849,7 @@ static void mlx4_ib_release_wqn(struct mlx4_ib_ucontext *context,
 	 * reused for further WQN allocations.
 	 * The next created WQ will allocate a new range.
 	 */
-		range->dirty = 1;
+		range->dirty = true;
 	}
 
 	mutex_unlock(&context->wqn_ranges_mutex);
@@ -3084,7 +3084,7 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, const struct ib_ud_wr *wr,
 		}
 		if (ah->av.eth.vlan != cpu_to_be16(0xffff)) {
 			vlan = be16_to_cpu(ah->av.eth.vlan) & 0x0fff;
-			is_vlan = 1;
+			is_vlan = true;
 		}
 	}
 	err = ib_ud_header_init(send_size, !is_eth, is_eth, is_vlan, is_grh,
-- 
cgit v1.2.3-59-g8ed1b


From 2ab367a70a077818905c92c4cfc7c2bd6bc76a14 Mon Sep 17 00:00:00 2001
From: zhengbin <zhengbin13@huawei.com>
Date: Tue, 24 Dec 2019 16:40:12 +0800
Subject: RDMA/mlx5: use true,false for bool variable

Fixes coccicheck warning:

drivers/infiniband/hw/mlx5/mr.c:150:2-26: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/hw/mlx5/mr.c:1455:2-26: WARNING: Assignment of 0/1 to bool variable
drivers/infiniband/hw/mlx5/qp.c:1874:6-20: WARNING: Assignment of 0/1 to bool variable

Link: https://lore.kernel.org/r/1577176812-2238-6-git-send-email-zhengbin13@huawei.com
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/mr.c | 4 ++--
 drivers/infiniband/hw/mlx5/qp.c | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 1e38ba0da16a..a52784304593 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -147,7 +147,7 @@ static int add_keys(struct mlx5_ib_dev *dev, int c, int num)
 			break;
 		}
 		mr->order = ent->order;
-		mr->allocated_from_cache = 1;
+		mr->allocated_from_cache = true;
 		mr->dev = dev;
 
 		MLX5_SET(mkc, mkc, free, 1);
@@ -1440,7 +1440,7 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 			goto err;
 		}
 
-		mr->allocated_from_cache = 0;
+		mr->allocated_from_cache = false;
 	} else {
 		/*
 		 * Send a UMR WQE
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 7e51870e9e01..04126ef74412 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -1871,7 +1871,7 @@ static void configure_requester_scat_cqe(struct mlx5_ib_dev *dev,
 {
 	enum ib_qp_type qpt = init_attr->qp_type;
 	int scqe_sz;
-	bool allow_scat_cqe = 0;
+	bool allow_scat_cqe = false;
 
 	if (qpt == IB_QPT_UC || qpt == IB_QPT_UD)
 		return;
-- 
cgit v1.2.3-59-g8ed1b


From 6ca18d8927d468c763571f78c9a7387a69ffa020 Mon Sep 17 00:00:00 2001
From: Jiewei Ke <kejiewei.cn@gmail.com>
Date: Fri, 27 Dec 2019 19:36:13 +0800
Subject: RDMA/rxe: Fix error type of mmap_offset

The type of mmap_offset should be u64 instead of int to match the type of
mminfo.offset. If otherwise, after we create several thousands of CQs, it
will run into overflow issues.

Link: https://lore.kernel.org/r/20191227113613.5020-1-kejiewei.cn@gmail.com
Signed-off-by: Jiewei Ke <kejiewei.cn@gmail.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/sw/rxe/rxe_verbs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.h b/drivers/infiniband/sw/rxe/rxe_verbs.h
index 95834206c80c..92de39c4a7c1 100644
--- a/drivers/infiniband/sw/rxe/rxe_verbs.h
+++ b/drivers/infiniband/sw/rxe/rxe_verbs.h
@@ -408,7 +408,7 @@ struct rxe_dev {
 	struct list_head	pending_mmaps;
 
 	spinlock_t		mmap_offset_lock; /* guard mmap_offset */
-	int			mmap_offset;
+	u64			mmap_offset;
 
 	atomic64_t		stats_counters[RXE_NUM_OF_COUNTERS];
 
-- 
cgit v1.2.3-59-g8ed1b


From 04db1580b5e48a79e24aa51ecae0cd4b2296ec23 Mon Sep 17 00:00:00 2001
From: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Date: Mon, 30 Dec 2019 10:24:28 +0800
Subject: RDMA/i40iw: fix a potential NULL pointer dereference

A NULL pointer can be returned by in_dev_get(). Thus add a corresponding
check so that a NULL pointer dereference will be avoided at this place.

Fixes: 8e06af711bf2 ("i40iw: add main, hdr, status")
Link: https://lore.kernel.org/r/1577672668-46499-1-git-send-email-xiyuyang19@fudan.edu.cn
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/i40iw/i40iw_main.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c
index d44cf33df81a..238614370927 100644
--- a/drivers/infiniband/hw/i40iw/i40iw_main.c
+++ b/drivers/infiniband/hw/i40iw/i40iw_main.c
@@ -1225,6 +1225,8 @@ static void i40iw_add_ipv4_addr(struct i40iw_device *iwdev)
 			const struct in_ifaddr *ifa;
 
 			idev = in_dev_get(dev);
+			if (!idev)
+				continue;
 			in_dev_for_each_ifa_rtnl(ifa, idev) {
 				i40iw_debug(&iwdev->sc_dev, I40IW_DEBUG_CM,
 					    "IP=%pI4, vlan_id=%d, MAC=%pM\n", &ifa->ifa_address,
-- 
cgit v1.2.3-59-g8ed1b


From 017d8eada807c6b80905445c0defa5d3b1d108bf Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Thu, 12 Dec 2019 11:38:24 +0200
Subject: RDMA/cm: Delete unused CM LAP functions

Clean the code by deleting LAP functions, which are not called anyway.

Link: https://lore.kernel.org/r/20191212093830.316934-43-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c      | 98 ---------------------------------------
 drivers/infiniband/core/cm_msgs.h | 70 ----------------------------
 2 files changed, 168 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 2945031dd8a7..fae788e38c87 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -3029,104 +3029,6 @@ out:
 	return -EINVAL;
 }
 
-static void cm_format_lap(struct cm_lap_msg *lap_msg,
-			  struct cm_id_private *cm_id_priv,
-			  struct sa_path_rec *alternate_path,
-			  const void *private_data,
-			  u8 private_data_len)
-{
-	bool alt_ext = false;
-
-	if (alternate_path->rec_type == SA_PATH_REC_TYPE_OPA)
-		alt_ext = opa_is_extended_lid(alternate_path->opa.dlid,
-					      alternate_path->opa.slid);
-	cm_format_mad_hdr(&lap_msg->hdr, CM_LAP_ATTR_ID,
-			  cm_form_tid(cm_id_priv));
-	lap_msg->local_comm_id = cm_id_priv->id.local_id;
-	lap_msg->remote_comm_id = cm_id_priv->id.remote_id;
-	cm_lap_set_remote_qpn(lap_msg, cm_id_priv->remote_qpn);
-	/* todo: need remote CM response timeout */
-	cm_lap_set_remote_resp_timeout(lap_msg, 0x1F);
-	lap_msg->alt_local_lid =
-		htons(ntohl(sa_path_get_slid(alternate_path)));
-	lap_msg->alt_remote_lid =
-		htons(ntohl(sa_path_get_dlid(alternate_path)));
-	lap_msg->alt_local_gid = alternate_path->sgid;
-	lap_msg->alt_remote_gid = alternate_path->dgid;
-	if (alt_ext) {
-		lap_msg->alt_local_gid.global.interface_id
-			= OPA_MAKE_ID(be32_to_cpu(alternate_path->opa.slid));
-		lap_msg->alt_remote_gid.global.interface_id
-			= OPA_MAKE_ID(be32_to_cpu(alternate_path->opa.dlid));
-	}
-	cm_lap_set_flow_label(lap_msg, alternate_path->flow_label);
-	cm_lap_set_traffic_class(lap_msg, alternate_path->traffic_class);
-	lap_msg->alt_hop_limit = alternate_path->hop_limit;
-	cm_lap_set_packet_rate(lap_msg, alternate_path->rate);
-	cm_lap_set_sl(lap_msg, alternate_path->sl);
-	cm_lap_set_subnet_local(lap_msg, 1); /* local only... */
-	cm_lap_set_local_ack_timeout(lap_msg,
-		cm_ack_timeout(cm_id_priv->av.port->cm_dev->ack_delay,
-			       alternate_path->packet_life_time));
-
-	if (private_data && private_data_len)
-		memcpy(lap_msg->private_data, private_data, private_data_len);
-}
-
-int ib_send_cm_lap(struct ib_cm_id *cm_id,
-		   struct sa_path_rec *alternate_path,
-		   const void *private_data,
-		   u8 private_data_len)
-{
-	struct cm_id_private *cm_id_priv;
-	struct ib_mad_send_buf *msg;
-	unsigned long flags;
-	int ret;
-
-	if (private_data && private_data_len > IB_CM_LAP_PRIVATE_DATA_SIZE)
-		return -EINVAL;
-
-	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
-	spin_lock_irqsave(&cm_id_priv->lock, flags);
-	if (cm_id->state != IB_CM_ESTABLISHED ||
-	    (cm_id->lap_state != IB_CM_LAP_UNINIT &&
-	     cm_id->lap_state != IB_CM_LAP_IDLE)) {
-		ret = -EINVAL;
-		goto out;
-	}
-
-	ret = cm_init_av_by_path(alternate_path, NULL, &cm_id_priv->alt_av,
-				 cm_id_priv);
-	if (ret)
-		goto out;
-	cm_id_priv->alt_av.timeout =
-			cm_ack_timeout(cm_id_priv->target_ack_delay,
-				       cm_id_priv->alt_av.timeout - 1);
-
-	ret = cm_alloc_msg(cm_id_priv, &msg);
-	if (ret)
-		goto out;
-
-	cm_format_lap((struct cm_lap_msg *) msg->mad, cm_id_priv,
-		      alternate_path, private_data, private_data_len);
-	msg->timeout_ms = cm_id_priv->timeout_ms;
-	msg->context[1] = (void *) (unsigned long) IB_CM_ESTABLISHED;
-
-	ret = ib_post_send_mad(msg, NULL);
-	if (ret) {
-		spin_unlock_irqrestore(&cm_id_priv->lock, flags);
-		cm_free_msg(msg);
-		return ret;
-	}
-
-	cm_id->lap_state = IB_CM_LAP_SENT;
-	cm_id_priv->msg = msg;
-
-out:	spin_unlock_irqrestore(&cm_id_priv->lock, flags);
-	return ret;
-}
-EXPORT_SYMBOL(ib_send_cm_lap);
-
 static void cm_format_path_lid_from_lap(struct cm_lap_msg *lap_msg,
 					struct sa_path_rec *path)
 {
diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h
index 92d7260ac913..23e3d0c6a67c 100644
--- a/drivers/infiniband/core/cm_msgs.h
+++ b/drivers/infiniband/core/cm_msgs.h
@@ -651,101 +651,31 @@ struct cm_lap_msg {
 	u8 private_data[IB_CM_LAP_PRIVATE_DATA_SIZE];
 } __packed;
 
-static inline __be32 cm_lap_get_remote_qpn(struct cm_lap_msg *lap_msg)
-{
-	return cpu_to_be32(be32_to_cpu(lap_msg->offset12) >> 8);
-}
-
-static inline void cm_lap_set_remote_qpn(struct cm_lap_msg *lap_msg, __be32 qpn)
-{
-	lap_msg->offset12 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
-					 (be32_to_cpu(lap_msg->offset12) &
-					  0x000000FF));
-}
-
-static inline u8 cm_lap_get_remote_resp_timeout(struct cm_lap_msg *lap_msg)
-{
-	return (u8) ((be32_to_cpu(lap_msg->offset12) & 0xF8) >> 3);
-}
-
-static inline void cm_lap_set_remote_resp_timeout(struct cm_lap_msg *lap_msg,
-						  u8 resp_timeout)
-{
-	lap_msg->offset12 = cpu_to_be32((resp_timeout << 3) |
-					 (be32_to_cpu(lap_msg->offset12) &
-					  0xFFFFFF07));
-}
-
 static inline __be32 cm_lap_get_flow_label(struct cm_lap_msg *lap_msg)
 {
 	return cpu_to_be32(be32_to_cpu(lap_msg->offset56) >> 12);
 }
 
-static inline void cm_lap_set_flow_label(struct cm_lap_msg *lap_msg,
-					 __be32 flow_label)
-{
-	lap_msg->offset56 = cpu_to_be32(
-				 (be32_to_cpu(lap_msg->offset56) & 0x00000FFF) |
-				 (be32_to_cpu(flow_label) << 12));
-}
-
 static inline u8 cm_lap_get_traffic_class(struct cm_lap_msg *lap_msg)
 {
 	return (u8) be32_to_cpu(lap_msg->offset56);
 }
 
-static inline void cm_lap_set_traffic_class(struct cm_lap_msg *lap_msg,
-					    u8 traffic_class)
-{
-	lap_msg->offset56 = cpu_to_be32(traffic_class |
-					 (be32_to_cpu(lap_msg->offset56) &
-					  0xFFFFFF00));
-}
-
 static inline u8 cm_lap_get_packet_rate(struct cm_lap_msg *lap_msg)
 {
 	return lap_msg->offset61 & 0x3F;
 }
 
-static inline void cm_lap_set_packet_rate(struct cm_lap_msg *lap_msg,
-					  u8 packet_rate)
-{
-	lap_msg->offset61 = (packet_rate & 0x3F) | (lap_msg->offset61 & 0xC0);
-}
-
 static inline u8 cm_lap_get_sl(struct cm_lap_msg *lap_msg)
 {
 	return lap_msg->offset62 >> 4;
 }
 
-static inline void cm_lap_set_sl(struct cm_lap_msg *lap_msg, u8 sl)
-{
-	lap_msg->offset62 = (sl << 4) | (lap_msg->offset62 & 0x0F);
-}
-
-static inline u8 cm_lap_get_subnet_local(struct cm_lap_msg *lap_msg)
-{
-	return (lap_msg->offset62 >> 3) & 0x1;
-}
-
-static inline void cm_lap_set_subnet_local(struct cm_lap_msg *lap_msg,
-					   u8 subnet_local)
-{
-	lap_msg->offset62 = ((subnet_local & 0x1) << 3) |
-			     (lap_msg->offset61 & 0xF7);
-}
 static inline u8 cm_lap_get_local_ack_timeout(struct cm_lap_msg *lap_msg)
 {
 	return lap_msg->offset63 >> 3;
 }
 
-static inline void cm_lap_set_local_ack_timeout(struct cm_lap_msg *lap_msg,
-						u8 local_ack_timeout)
-{
-	lap_msg->offset63 = (local_ack_timeout << 3) |
-			    (lap_msg->offset63 & 0x07);
-}
-
 struct cm_apr_msg {
 	struct ib_mad_hdr hdr;
 
-- 
cgit v1.2.3-59-g8ed1b


From ad9efa05a0ea2fd9960d4e163fad56279cde45b2 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Thu, 12 Dec 2019 11:38:27 +0200
Subject: RDMA/cm: Delete unused CM ARP functions

Clean the code by deleting ARP functions, which are not called anyway.

Link: https://lore.kernel.org/r/20191212093830.316934-46-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c | 66 --------------------------------------------
 include/rdma/ib_cm.h         | 34 -----------------------
 2 files changed, 100 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index fae788e38c87..d396b987b0f2 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -3163,72 +3163,6 @@ deref:	cm_deref_id(cm_id_priv);
 	return -EINVAL;
 }
 
-static void cm_format_apr(struct cm_apr_msg *apr_msg,
-			  struct cm_id_private *cm_id_priv,
-			  enum ib_cm_apr_status status,
-			  void *info,
-			  u8 info_length,
-			  const void *private_data,
-			  u8 private_data_len)
-{
-	cm_format_mad_hdr(&apr_msg->hdr, CM_APR_ATTR_ID, cm_id_priv->tid);
-	apr_msg->local_comm_id = cm_id_priv->id.local_id;
-	apr_msg->remote_comm_id = cm_id_priv->id.remote_id;
-	apr_msg->ap_status = (u8) status;
-
-	if (info && info_length) {
-		apr_msg->info_length = info_length;
-		memcpy(apr_msg->info, info, info_length);
-	}
-
-	if (private_data && private_data_len)
-		memcpy(apr_msg->private_data, private_data, private_data_len);
-}
-
-int ib_send_cm_apr(struct ib_cm_id *cm_id,
-		   enum ib_cm_apr_status status,
-		   void *info,
-		   u8 info_length,
-		   const void *private_data,
-		   u8 private_data_len)
-{
-	struct cm_id_private *cm_id_priv;
-	struct ib_mad_send_buf *msg;
-	unsigned long flags;
-	int ret;
-
-	if ((private_data && private_data_len > IB_CM_APR_PRIVATE_DATA_SIZE) ||
-	    (info && info_length > IB_CM_APR_INFO_LENGTH))
-		return -EINVAL;
-
-	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
-	spin_lock_irqsave(&cm_id_priv->lock, flags);
-	if (cm_id->state != IB_CM_ESTABLISHED ||
-	    (cm_id->lap_state != IB_CM_LAP_RCVD &&
-	     cm_id->lap_state != IB_CM_MRA_LAP_SENT)) {
-		ret = -EINVAL;
-		goto out;
-	}
-
-	ret = cm_alloc_msg(cm_id_priv, &msg);
-	if (ret)
-		goto out;
-
-	cm_format_apr((struct cm_apr_msg *) msg->mad, cm_id_priv, status,
-		      info, info_length, private_data, private_data_len);
-	ret = ib_post_send_mad(msg, NULL);
-	if (ret) {
-		spin_unlock_irqrestore(&cm_id_priv->lock, flags);
-		cm_free_msg(msg);
-		return ret;
-	}
-
-	cm_id->lap_state = IB_CM_LAP_IDLE;
-out:	spin_unlock_irqrestore(&cm_id_priv->lock, flags);
-	return ret;
-}
-EXPORT_SYMBOL(ib_send_cm_apr);
-
 static int cm_apr_handler(struct cm_work *work)
 {
 	struct cm_id_private *cm_id_priv;
diff --git a/include/rdma/ib_cm.h b/include/rdma/ib_cm.h
index b01a8a8d4de9..8ec482e391aa 100644
--- a/include/rdma/ib_cm.h
+++ b/include/rdma/ib_cm.h
@@ -499,21 +499,6 @@ int ib_send_cm_mra(struct ib_cm_id *cm_id,
 		   const void *private_data,
 		   u8 private_data_len);
 
-/**
- * ib_send_cm_lap - Sends a load alternate path request.
- * @cm_id: Connection identifier associated with the load alternate path
- *   message.
- * @alternate_path: A path record that identifies the alternate path to
- *   load.
- * @private_data: Optional user-defined private data sent with the
- *   load alternate path message.
- * @private_data_len: Size of the private data buffer, in bytes.
- */
-int ib_send_cm_lap(struct ib_cm_id *cm_id,
-		   struct sa_path_rec *alternate_path,
-		   const void *private_data,
-		   u8 private_data_len);
-
 /**
  * ib_cm_init_qp_attr - Initializes the QP attributes for use in transitioning
  *   to a specified QP state.
@@ -534,25 +519,6 @@ int ib_cm_init_qp_attr(struct ib_cm_id *cm_id,
 		       struct ib_qp_attr *qp_attr,
 		       int *qp_attr_mask);
 
-/**
- * ib_send_cm_apr - Sends an alternate path response message in response to
- *   a load alternate path request.
- * @cm_id: Connection identifier associated with the alternate path response.
- * @status: Reply status sent with the alternate path response.
- * @info: Optional additional information sent with the alternate path
- *   response.
- * @info_length: Size of the additional information, in bytes.
- * @private_data: Optional user-defined private data sent with the
- *   alternate path response message.
- * @private_data_len: Size of the private data buffer, in bytes.
- */
-int ib_send_cm_apr(struct ib_cm_id *cm_id,
-		   enum ib_cm_apr_status status,
-		   void *info,
-		   u8 info_length,
-		   const void *private_data,
-		   u8 private_data_len);
-
 struct ib_cm_sidr_req_param {
 	struct sa_path_rec	*path;
 	const struct ib_gid_attr *sgid_attr;
-- 
cgit v1.2.3-59-g8ed1b


From ed999f820a6c579298b53270656a0a26c8a6cb87 Mon Sep 17 00:00:00 2001
From: Chuck Lever <chuck.lever@oracle.com>
Date: Wed, 18 Dec 2019 15:18:10 -0500
Subject: RDMA/cma: Add trace points in RDMA Connection Manager

Record state transitions as each connection is established. The IP address
of both peers and the Type of Service is reported. These trace points are
not in performance hot paths.

Also, record each cm_event_handler call to ULPs. This eliminates the need
for each ULP to add its own similar trace point in its CM event handler
function.

These new trace points appear in a new trace subsystem called "rdma_cma".

Sample events:

           <...>-220   [004]   121.430733: cm_id_create:         cm.id=0
           <...>-472   [003]   121.430991: cm_event_handler:     cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 ADDR_RESOLVED (0/0)
           <...>-472   [003]   121.430995: cm_event_done:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 result=0
           <...>-472   [003]   121.431172: cm_event_handler:     cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 ROUTE_RESOLVED (2/0)
           <...>-472   [003]   121.431174: cm_event_done:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 result=0
           <...>-220   [004]   121.433480: cm_qp_create:         cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 pd.id=2 qp_type=RC send_wr=4091 recv_wr=256 qp_num=521 rc=0
           <...>-220   [004]   121.433577: cm_send_req:          cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 qp_num=521
     kworker/1:2-973   [001]   121.436190: cm_send_mra:          cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0
     kworker/1:2-973   [001]   121.436340: cm_send_rtu:          cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0
     kworker/1:2-973   [001]   121.436359: cm_event_handler:     cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 ESTABLISHED (9/0)
     kworker/1:2-973   [001]   121.436365: cm_event_done:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 result=0
           <...>-1975  [005]   123.161954: cm_disconnect:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0
           <...>-1975  [005]   123.161974: cm_sent_dreq:         cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0
           <...>-220   [004]   123.162102: cm_disconnect:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0
     kworker/0:1-13    [000]   123.162391: cm_event_handler:     cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 DISCONNECTED (10/0)
     kworker/0:1-13    [000]   123.162393: cm_event_done:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 result=0
           <...>-220   [004]   123.164456: cm_qp_destroy:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0 qp_num=521
           <...>-220   [004]   123.165290: cm_id_destroy:        cm.id=0 src=192.168.2.51:35090 dst=192.168.2.55:20049 tos=0

Some features to note:
- restracker ID of the rdma_cm_id is tagged on each trace event
- The source and destination IP addresses and TOS are reported
- CM event upcalls are shown with decoded event and status
- CM state transitions are reported
- rdma_cm_id lifetime events are captured
- The latency of ULP CM event handlers is reported
- Lifetime events of associated QPs are reported
- Device removal and insertion is reported

This patch is based on previous work by:

Saeed Mahameed <saeedm@mellanox.com>
Mukesh Kacker <mukesh.kacker@oracle.com>
Ajaykumar Hotchandani <ajaykumar.hotchandani@oracle.com>
Aron Silverton <aron.silverton@oracle.com>
Avinash Repaka <avinash.repaka@oracle.com>
Somasundaram Krishnasamy <somasundaram.krishnasamy@oracle.com>

Link: https://lore.kernel.org/r/20191218201810.30584.3052.stgit@manet.1015granger.net
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/Makefile    |   3 +-
 drivers/infiniband/core/cma.c       |  88 ++++++--
 drivers/infiniband/core/cma_trace.c |  16 ++
 drivers/infiniband/core/cma_trace.h | 391 ++++++++++++++++++++++++++++++++++++
 4 files changed, 475 insertions(+), 23 deletions(-)
 create mode 100644 drivers/infiniband/core/cma_trace.c
 create mode 100644 drivers/infiniband/core/cma_trace.h

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile
index 9a8871e21545..f22555f982e2 100644
--- a/drivers/infiniband/core/Makefile
+++ b/drivers/infiniband/core/Makefile
@@ -20,7 +20,8 @@ ib_cm-y :=			cm.o
 
 iw_cm-y :=			iwcm.o iwpm_util.o iwpm_msg.o
 
-rdma_cm-y :=			cma.o
+CFLAGS_cma_trace.o +=		-I$(src)
+rdma_cm-y :=			cma.o cma_trace.o
 
 rdma_cm-$(CONFIG_INFINIBAND_ADDR_TRANS_CONFIGFS) += cma_configfs.o
 
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 43a6f07e0afe..55a9afacfedd 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -36,6 +36,7 @@
 
 #include "core_priv.h"
 #include "cma_priv.h"
+#include "cma_trace.h"
 
 MODULE_AUTHOR("Sean Hefty");
 MODULE_DESCRIPTION("Generic RDMA CM Agent");
@@ -877,6 +878,7 @@ struct rdma_cm_id *__rdma_create_id(struct net *net,
 	id_priv->id.route.addr.dev_addr.net = get_net(net);
 	id_priv->seq_num &= 0x00ffffff;
 
+	trace_cm_id_create(id_priv);
 	return &id_priv->id;
 }
 EXPORT_SYMBOL(__rdma_create_id);
@@ -928,27 +930,34 @@ int rdma_create_qp(struct rdma_cm_id *id, struct ib_pd *pd,
 	int ret;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
-	if (id->device != pd->device)
-		return -EINVAL;
+	if (id->device != pd->device) {
+		ret = -EINVAL;
+		goto out_err;
+	}
 
 	qp_init_attr->port_num = id->port_num;
 	qp = ib_create_qp(pd, qp_init_attr);
-	if (IS_ERR(qp))
-		return PTR_ERR(qp);
+	if (IS_ERR(qp)) {
+		ret = PTR_ERR(qp);
+		goto out_err;
+	}
 
 	if (id->qp_type == IB_QPT_UD)
 		ret = cma_init_ud_qp(id_priv, qp);
 	else
 		ret = cma_init_conn_qp(id_priv, qp);
 	if (ret)
-		goto err;
+		goto out_destroy;
 
 	id->qp = qp;
 	id_priv->qp_num = qp->qp_num;
 	id_priv->srq = (qp->srq != NULL);
+	trace_cm_qp_create(id_priv, pd, qp_init_attr, 0);
 	return 0;
-err:
+out_destroy:
 	ib_destroy_qp(qp);
+out_err:
+	trace_cm_qp_create(id_priv, pd, qp_init_attr, ret);
 	return ret;
 }
 EXPORT_SYMBOL(rdma_create_qp);
@@ -958,6 +967,7 @@ void rdma_destroy_qp(struct rdma_cm_id *id)
 	struct rdma_id_private *id_priv;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
+	trace_cm_qp_destroy(id_priv);
 	mutex_lock(&id_priv->qp_mutex);
 	ib_destroy_qp(id_priv->id.qp);
 	id_priv->id.qp = NULL;
@@ -1811,6 +1821,7 @@ void rdma_destroy_id(struct rdma_cm_id *id)
 	enum rdma_cm_state state;
 
 	id_priv = container_of(id, struct rdma_id_private, id);
+	trace_cm_id_destroy(id_priv);
 	state = cma_exch(id_priv, RDMA_CM_DESTROYING);
 	cma_cancel_operation(id_priv, state);
 
@@ -1863,6 +1874,7 @@ static int cma_rep_recv(struct rdma_id_private *id_priv)
 	if (ret)
 		goto reject;
 
+	trace_cm_send_rtu(id_priv);
 	ret = ib_send_cm_rtu(id_priv->cm_id.ib, NULL, 0);
 	if (ret)
 		goto reject;
@@ -1871,6 +1883,7 @@ static int cma_rep_recv(struct rdma_id_private *id_priv)
 reject:
 	pr_debug_ratelimited("RDMA CM: CONNECT_ERROR: failed to handle reply. status %d\n", ret);
 	cma_modify_qp_err(id_priv);
+	trace_cm_send_rej(id_priv);
 	ib_send_cm_rej(id_priv->cm_id.ib, IB_CM_REJ_CONSUMER_DEFINED,
 		       NULL, 0, NULL, 0);
 	return ret;
@@ -1890,6 +1903,17 @@ static void cma_set_rep_event_data(struct rdma_cm_event *event,
 	event->param.conn.qp_num = rep_data->remote_qpn;
 }
 
+static int cma_cm_event_handler(struct rdma_id_private *id_priv,
+				struct rdma_cm_event *event)
+{
+	int ret;
+
+	trace_cm_event_handler(id_priv, event);
+	ret = id_priv->id.event_handler(&id_priv->id, event);
+	trace_cm_event_done(id_priv, event, ret);
+	return ret;
+}
+
 static int cma_ib_handler(struct ib_cm_id *cm_id,
 			  const struct ib_cm_event *ib_event)
 {
@@ -1912,8 +1936,10 @@ static int cma_ib_handler(struct ib_cm_id *cm_id,
 		break;
 	case IB_CM_REP_RECEIVED:
 		if (cma_comp(id_priv, RDMA_CM_CONNECT) &&
-		    (id_priv->id.qp_type != IB_QPT_UD))
+		    (id_priv->id.qp_type != IB_QPT_UD)) {
+			trace_cm_send_mra(id_priv);
 			ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0);
+		}
 		if (id_priv->id.qp) {
 			event.status = cma_rep_recv(id_priv);
 			event.event = event.status ? RDMA_CM_EVENT_CONNECT_ERROR :
@@ -1958,7 +1984,7 @@ static int cma_ib_handler(struct ib_cm_id *cm_id,
 		goto out;
 	}
 
-	ret = id_priv->id.event_handler(&id_priv->id, &event);
+	ret = cma_cm_event_handler(id_priv, &event);
 	if (ret) {
 		/* Destroy the CM ID by returning a non-zero value. */
 		id_priv->cm_id.ib = NULL;
@@ -2119,6 +2145,7 @@ static int cma_ib_req_handler(struct ib_cm_id *cm_id,
 	if (IS_ERR(listen_id))
 		return PTR_ERR(listen_id);
 
+	trace_cm_req_handler(listen_id, ib_event->event);
 	if (!cma_ib_check_req_qp_type(&listen_id->id, ib_event)) {
 		ret = -EINVAL;
 		goto net_dev_put;
@@ -2161,7 +2188,7 @@ static int cma_ib_req_handler(struct ib_cm_id *cm_id,
 	 * until we're done accessing it.
 	 */
 	atomic_inc(&conn_id->refcount);
-	ret = conn_id->id.event_handler(&conn_id->id, &event);
+	ret = cma_cm_event_handler(conn_id, &event);
 	if (ret)
 		goto err3;
 	/*
@@ -2170,8 +2197,10 @@ static int cma_ib_req_handler(struct ib_cm_id *cm_id,
 	 */
 	mutex_lock(&lock);
 	if (cma_comp(conn_id, RDMA_CM_CONNECT) &&
-	    (conn_id->id.qp_type != IB_QPT_UD))
+	    (conn_id->id.qp_type != IB_QPT_UD)) {
+		trace_cm_send_mra(cm_id->context);
 		ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0);
+	}
 	mutex_unlock(&lock);
 	mutex_unlock(&conn_id->handler_mutex);
 	mutex_unlock(&listen_id->handler_mutex);
@@ -2286,7 +2315,7 @@ static int cma_iw_handler(struct iw_cm_id *iw_id, struct iw_cm_event *iw_event)
 	event.status = iw_event->status;
 	event.param.conn.private_data = iw_event->private_data;
 	event.param.conn.private_data_len = iw_event->private_data_len;
-	ret = id_priv->id.event_handler(&id_priv->id, &event);
+	ret = cma_cm_event_handler(id_priv, &event);
 	if (ret) {
 		/* Destroy the CM ID by returning a non-zero value. */
 		id_priv->cm_id.iw = NULL;
@@ -2363,7 +2392,7 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id,
 	 * until we're done accessing it.
 	 */
 	atomic_inc(&conn_id->refcount);
-	ret = conn_id->id.event_handler(&conn_id->id, &event);
+	ret = cma_cm_event_handler(conn_id, &event);
 	if (ret) {
 		/* User wants to destroy the CM ID */
 		conn_id->cm_id.iw = NULL;
@@ -2435,6 +2464,7 @@ static int cma_listen_handler(struct rdma_cm_id *id,
 
 	id->context = id_priv->id.context;
 	id->event_handler = id_priv->id.event_handler;
+	trace_cm_event_handler(id_priv, event);
 	return id_priv->id.event_handler(id, event);
 }
 
@@ -2611,7 +2641,7 @@ static void cma_work_handler(struct work_struct *_work)
 	if (!cma_comp_exch(id_priv, work->old_state, work->new_state))
 		goto out;
 
-	if (id_priv->id.event_handler(&id_priv->id, &work->event)) {
+	if (cma_cm_event_handler(id_priv, &work->event)) {
 		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		destroy = 1;
 	}
@@ -2634,7 +2664,7 @@ static void cma_ndev_work_handler(struct work_struct *_work)
 	    id_priv->state == RDMA_CM_DEVICE_REMOVAL)
 		goto out;
 
-	if (id_priv->id.event_handler(&id_priv->id, &work->event)) {
+	if (cma_cm_event_handler(id_priv, &work->event)) {
 		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		destroy = 1;
 	}
@@ -3089,7 +3119,7 @@ static void addr_handler(int status, struct sockaddr *src_addr,
 	} else
 		event.event = RDMA_CM_EVENT_ADDR_RESOLVED;
 
-	if (id_priv->id.event_handler(&id_priv->id, &event)) {
+	if (cma_cm_event_handler(id_priv, &event)) {
 		cma_exch(id_priv, RDMA_CM_DESTROYING);
 		mutex_unlock(&id_priv->handler_mutex);
 		rdma_destroy_id(&id_priv->id);
@@ -3736,7 +3766,7 @@ static int cma_sidr_rep_handler(struct ib_cm_id *cm_id,
 		goto out;
 	}
 
-	ret = id_priv->id.event_handler(&id_priv->id, &event);
+	ret = cma_cm_event_handler(id_priv, &event);
 
 	rdma_destroy_ah_attr(&event.param.ud.ah_attr);
 	if (ret) {
@@ -3800,6 +3830,7 @@ static int cma_resolve_ib_udp(struct rdma_id_private *id_priv,
 	req.timeout_ms = 1 << (CMA_CM_RESPONSE_TIMEOUT - 8);
 	req.max_cm_retries = CMA_MAX_CM_RETRIES;
 
+	trace_cm_send_sidr_req(id_priv);
 	ret = ib_send_cm_sidr_req(id_priv->cm_id.ib, &req);
 	if (ret) {
 		ib_destroy_cm_id(id_priv->cm_id.ib);
@@ -3873,6 +3904,7 @@ static int cma_connect_ib(struct rdma_id_private *id_priv,
 	req.max_cm_retries = CMA_MAX_CM_RETRIES;
 	req.srq = id_priv->srq ? 1 : 0;
 
+	trace_cm_send_req(id_priv);
 	ret = ib_send_cm_req(id_priv->cm_id.ib, &req);
 out:
 	if (ret && !IS_ERR(id)) {
@@ -3986,6 +4018,7 @@ static int cma_accept_ib(struct rdma_id_private *id_priv,
 	rep.rnr_retry_count = min_t(u8, 7, conn_param->rnr_retry_count);
 	rep.srq = id_priv->srq ? 1 : 0;
 
+	trace_cm_send_rep(id_priv);
 	ret = ib_send_cm_rep(id_priv->cm_id.ib, &rep);
 out:
 	return ret;
@@ -4035,6 +4068,7 @@ static int cma_send_sidr_rep(struct rdma_id_private *id_priv,
 	rep.private_data = private_data;
 	rep.private_data_len = private_data_len;
 
+	trace_cm_send_sidr_rep(id_priv);
 	return ib_send_cm_sidr_rep(id_priv->cm_id.ib, &rep);
 }
 
@@ -4120,13 +4154,15 @@ int rdma_reject(struct rdma_cm_id *id, const void *private_data,
 		return -EINVAL;
 
 	if (rdma_cap_ib_cm(id->device, id->port_num)) {
-		if (id->qp_type == IB_QPT_UD)
+		if (id->qp_type == IB_QPT_UD) {
 			ret = cma_send_sidr_rep(id_priv, IB_SIDR_REJECT, 0,
 						private_data, private_data_len);
-		else
+		} else {
+			trace_cm_send_rej(id_priv);
 			ret = ib_send_cm_rej(id_priv->cm_id.ib,
 					     IB_CM_REJ_CONSUMER_DEFINED, NULL,
 					     0, private_data, private_data_len);
+		}
 	} else if (rdma_cap_iw_cm(id->device, id->port_num)) {
 		ret = iw_cm_reject(id_priv->cm_id.iw,
 				   private_data, private_data_len);
@@ -4151,8 +4187,13 @@ int rdma_disconnect(struct rdma_cm_id *id)
 		if (ret)
 			goto out;
 		/* Initiate or respond to a disconnect. */
-		if (ib_send_cm_dreq(id_priv->cm_id.ib, NULL, 0))
-			ib_send_cm_drep(id_priv->cm_id.ib, NULL, 0);
+		trace_cm_disconnect(id_priv);
+		if (ib_send_cm_dreq(id_priv->cm_id.ib, NULL, 0)) {
+			if (!ib_send_cm_drep(id_priv->cm_id.ib, NULL, 0))
+				trace_cm_sent_drep(id_priv);
+		} else {
+			trace_cm_sent_dreq(id_priv);
+		}
 	} else if (rdma_cap_iw_cm(id->device, id->port_num)) {
 		ret = iw_cm_disconnect(id_priv->cm_id.iw, 0);
 	} else
@@ -4218,7 +4259,7 @@ static int cma_ib_mc_handler(int status, struct ib_sa_multicast *multicast)
 	} else
 		event.event = RDMA_CM_EVENT_MULTICAST_ERROR;
 
-	ret = id_priv->id.event_handler(&id_priv->id, &event);
+	ret = cma_cm_event_handler(id_priv, &event);
 
 	rdma_destroy_ah_attr(&event.param.ud.ah_attr);
 	if (ret) {
@@ -4623,6 +4664,7 @@ static void cma_add_one(struct ib_device *device)
 		cma_listen_on_dev(id_priv, cma_dev);
 	mutex_unlock(&lock);
 
+	trace_cm_add_one(device);
 	return;
 
 free_gid_type:
@@ -4653,7 +4695,7 @@ static int cma_remove_id_dev(struct rdma_id_private *id_priv)
 		goto out;
 
 	event.event = RDMA_CM_EVENT_DEVICE_REMOVAL;
-	ret = id_priv->id.event_handler(&id_priv->id, &event);
+	ret = cma_cm_event_handler(id_priv, &event);
 out:
 	mutex_unlock(&id_priv->handler_mutex);
 	return ret;
@@ -4691,6 +4733,8 @@ static void cma_remove_one(struct ib_device *device, void *client_data)
 {
 	struct cma_device *cma_dev = client_data;
 
+	trace_cm_remove_one(device);
+
 	if (!cma_dev)
 		return;
 
diff --git a/drivers/infiniband/core/cma_trace.c b/drivers/infiniband/core/cma_trace.c
new file mode 100644
index 000000000000..b314a281e10e
--- /dev/null
+++ b/drivers/infiniband/core/cma_trace.c
@@ -0,0 +1,16 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Trace points for the RDMA Connection Manager.
+ *
+ * Author: Chuck Lever <chuck.lever@oracle.com>
+ *
+ * Copyright (c) 2019, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#define CREATE_TRACE_POINTS
+
+#include <rdma/rdma_cm.h>
+#include <rdma/ib_cm.h>
+#include "cma_priv.h"
+
+#include "cma_trace.h"
diff --git a/drivers/infiniband/core/cma_trace.h b/drivers/infiniband/core/cma_trace.h
new file mode 100644
index 000000000000..81e36bf13159
--- /dev/null
+++ b/drivers/infiniband/core/cma_trace.h
@@ -0,0 +1,391 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Trace point definitions for the RDMA Connect Manager.
+ *
+ * Author: Chuck Lever <chuck.lever@oracle.com>
+ *
+ * Copyright (c) 2019, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM rdma_cma
+
+#if !defined(_TRACE_RDMA_CMA_H) || defined(TRACE_HEADER_MULTI_READ)
+
+#define _TRACE_RDMA_CMA_H
+
+#include <linux/tracepoint.h>
+#include <trace/events/rdma.h>
+
+/*
+ * enum ib_cm_event_type, from include/rdma/ib_cm.h
+ */
+#define IB_CM_EVENT_LIST			\
+	ib_cm_event(REQ_ERROR)			\
+	ib_cm_event(REQ_RECEIVED)		\
+	ib_cm_event(REP_ERROR)			\
+	ib_cm_event(REP_RECEIVED)		\
+	ib_cm_event(RTU_RECEIVED)		\
+	ib_cm_event(USER_ESTABLISHED)		\
+	ib_cm_event(DREQ_ERROR)			\
+	ib_cm_event(DREQ_RECEIVED)		\
+	ib_cm_event(DREP_RECEIVED)		\
+	ib_cm_event(TIMEWAIT_EXIT)		\
+	ib_cm_event(MRA_RECEIVED)		\
+	ib_cm_event(REJ_RECEIVED)		\
+	ib_cm_event(LAP_ERROR)			\
+	ib_cm_event(LAP_RECEIVED)		\
+	ib_cm_event(APR_RECEIVED)		\
+	ib_cm_event(SIDR_REQ_ERROR)		\
+	ib_cm_event(SIDR_REQ_RECEIVED)		\
+	ib_cm_event_end(SIDR_REP_RECEIVED)
+
+#undef ib_cm_event
+#undef ib_cm_event_end
+
+#define ib_cm_event(x)		TRACE_DEFINE_ENUM(IB_CM_##x);
+#define ib_cm_event_end(x)	TRACE_DEFINE_ENUM(IB_CM_##x);
+
+IB_CM_EVENT_LIST
+
+#undef ib_cm_event
+#undef ib_cm_event_end
+
+#define ib_cm_event(x)		{ IB_CM_##x, #x },
+#define ib_cm_event_end(x)	{ IB_CM_##x, #x }
+
+#define rdma_show_ib_cm_event(x) \
+		__print_symbolic(x, IB_CM_EVENT_LIST)
+
+
+DECLARE_EVENT_CLASS(cma_fsm_class,
+	TP_PROTO(
+		const struct rdma_id_private *id_priv
+	),
+
+	TP_ARGS(id_priv),
+
+	TP_STRUCT__entry(
+		__field(u32, cm_id)
+		__field(u32, tos)
+		__array(unsigned char, srcaddr, sizeof(struct sockaddr_in6))
+		__array(unsigned char, dstaddr, sizeof(struct sockaddr_in6))
+	),
+
+	TP_fast_assign(
+		__entry->cm_id = id_priv->res.id;
+		__entry->tos = id_priv->tos;
+		memcpy(__entry->srcaddr, &id_priv->id.route.addr.src_addr,
+		       sizeof(struct sockaddr_in6));
+		memcpy(__entry->dstaddr, &id_priv->id.route.addr.dst_addr,
+		       sizeof(struct sockaddr_in6));
+	),
+
+	TP_printk("cm.id=%u src=%pISpc dst=%pISpc tos=%u",
+		__entry->cm_id, __entry->srcaddr, __entry->dstaddr, __entry->tos
+	)
+);
+
+#define DEFINE_CMA_FSM_EVENT(name)						\
+		DEFINE_EVENT(cma_fsm_class, cm_##name,				\
+				TP_PROTO(					\
+					const struct rdma_id_private *id_priv	\
+				),						\
+				TP_ARGS(id_priv))
+
+DEFINE_CMA_FSM_EVENT(send_rtu);
+DEFINE_CMA_FSM_EVENT(send_rej);
+DEFINE_CMA_FSM_EVENT(send_mra);
+DEFINE_CMA_FSM_EVENT(send_sidr_req);
+DEFINE_CMA_FSM_EVENT(send_sidr_rep);
+DEFINE_CMA_FSM_EVENT(disconnect);
+DEFINE_CMA_FSM_EVENT(sent_drep);
+DEFINE_CMA_FSM_EVENT(sent_dreq);
+DEFINE_CMA_FSM_EVENT(id_destroy);
+
+TRACE_EVENT(cm_id_create,
+	TP_PROTO(
+		const struct rdma_id_private *id_priv
+	),
+
+	TP_ARGS(id_priv),
+
+	TP_STRUCT__entry(
+		__field(u32, cm_id)
+	),
+
+	TP_fast_assign(
+		__entry->cm_id = id_priv->res.id;
+	),
+
+	TP_printk("cm.id=%u",
+		__entry->cm_id
+	)
+);
+
+DECLARE_EVENT_CLASS(cma_qp_class,
+	TP_PROTO(
+		const struct rdma_id_private *id_priv
+	),
+
+	TP_ARGS(id_priv),
+
+	TP_STRUCT__entry(
+		__field(u32, cm_id)
+		__field(u32, tos)
+		__field(u32, qp_num)
+		__array(unsigned char, srcaddr, sizeof(struct sockaddr_in6))
+		__array(unsigned char, dstaddr, sizeof(struct sockaddr_in6))
+	),
+
+	TP_fast_assign(
+		__entry->cm_id = id_priv->res.id;
+		__entry->tos = id_priv->tos;
+		__entry->qp_num = id_priv->qp_num;
+		memcpy(__entry->srcaddr, &id_priv->id.route.addr.src_addr,
+		       sizeof(struct sockaddr_in6));
+		memcpy(__entry->dstaddr, &id_priv->id.route.addr.dst_addr,
+		       sizeof(struct sockaddr_in6));
+	),
+
+	TP_printk("cm.id=%u src=%pISpc dst=%pISpc tos=%u qp_num=%u",
+		__entry->cm_id, __entry->srcaddr, __entry->dstaddr, __entry->tos,
+		__entry->qp_num
+	)
+);
+
+#define DEFINE_CMA_QP_EVENT(name)						\
+		DEFINE_EVENT(cma_qp_class, cm_##name,				\
+				TP_PROTO(					\
+					const struct rdma_id_private *id_priv	\
+				),						\
+				TP_ARGS(id_priv))
+
+DEFINE_CMA_QP_EVENT(send_req);
+DEFINE_CMA_QP_EVENT(send_rep);
+DEFINE_CMA_QP_EVENT(qp_destroy);
+
+/*
+ * enum ib_wp_type, from include/rdma/ib_verbs.h
+ */
+#define IB_QP_TYPE_LIST				\
+	ib_qp_type(SMI)				\
+	ib_qp_type(GSI)				\
+	ib_qp_type(RC)				\
+	ib_qp_type(UC)				\
+	ib_qp_type(UD)				\
+	ib_qp_type(RAW_IPV6)			\
+	ib_qp_type(RAW_ETHERTYPE)		\
+	ib_qp_type(RAW_PACKET)			\
+	ib_qp_type(XRC_INI)			\
+	ib_qp_type_end(XRC_TGT)
+
+#undef ib_qp_type
+#undef ib_qp_type_end
+
+#define ib_qp_type(x)		TRACE_DEFINE_ENUM(IB_QPT_##x);
+#define ib_qp_type_end(x)	TRACE_DEFINE_ENUM(IB_QPT_##x);
+
+IB_QP_TYPE_LIST
+
+#undef ib_qp_type
+#undef ib_qp_type_end
+
+#define ib_qp_type(x)		{ IB_QPT_##x, #x },
+#define ib_qp_type_end(x)	{ IB_QPT_##x, #x }
+
+#define rdma_show_qp_type(x) \
+		__print_symbolic(x, IB_QP_TYPE_LIST)
+
+
+TRACE_EVENT(cm_qp_create,
+	TP_PROTO(
+		const struct rdma_id_private *id_priv,
+		const struct ib_pd *pd,
+		const struct ib_qp_init_attr *qp_init_attr,
+		int rc
+	),
+
+	TP_ARGS(id_priv, pd, qp_init_attr, rc),
+
+	TP_STRUCT__entry(
+		__field(u32, cm_id)
+		__field(u32, pd_id)
+		__field(u32, tos)
+		__field(u32, qp_num)
+		__field(u32, send_wr)
+		__field(u32, recv_wr)
+		__field(int, rc)
+		__field(unsigned long, qp_type)
+		__array(unsigned char, srcaddr, sizeof(struct sockaddr_in6))
+		__array(unsigned char, dstaddr, sizeof(struct sockaddr_in6))
+	),
+
+	TP_fast_assign(
+		__entry->cm_id = id_priv->res.id;
+		__entry->pd_id = pd->res.id;
+		__entry->tos = id_priv->tos;
+		__entry->send_wr = qp_init_attr->cap.max_send_wr;
+		__entry->recv_wr = qp_init_attr->cap.max_recv_wr;
+		__entry->rc = rc;
+		if (!rc) {
+			__entry->qp_num = id_priv->qp_num;
+			__entry->qp_type = id_priv->id.qp_type;
+		} else {
+			__entry->qp_num = 0;
+			__entry->qp_type = 0;
+		}
+		memcpy(__entry->srcaddr, &id_priv->id.route.addr.src_addr,
+		       sizeof(struct sockaddr_in6));
+		memcpy(__entry->dstaddr, &id_priv->id.route.addr.dst_addr,
+		       sizeof(struct sockaddr_in6));
+	),
+
+	TP_printk("cm.id=%u src=%pISpc dst=%pISpc tos=%u pd.id=%u qp_type=%s"
+		" send_wr=%u recv_wr=%u qp_num=%u rc=%d",
+		__entry->cm_id, __entry->srcaddr, __entry->dstaddr,
+		__entry->tos, __entry->pd_id,
+		rdma_show_qp_type(__entry->qp_type), __entry->send_wr,
+		__entry->recv_wr, __entry->qp_num, __entry->rc
+	)
+);
+
+TRACE_EVENT(cm_req_handler,
+	TP_PROTO(
+		const struct rdma_id_private *id_priv,
+		int event
+	),
+
+	TP_ARGS(id_priv, event),
+
+	TP_STRUCT__entry(
+		__field(u32, cm_id)
+		__field(u32, tos)
+		__field(unsigned long, event)
+		__array(unsigned char, srcaddr, sizeof(struct sockaddr_in6))
+		__array(unsigned char, dstaddr, sizeof(struct sockaddr_in6))
+	),
+
+	TP_fast_assign(
+		__entry->cm_id = id_priv->res.id;
+		__entry->tos = id_priv->tos;
+		__entry->event = event;
+		memcpy(__entry->srcaddr, &id_priv->id.route.addr.src_addr,
+		       sizeof(struct sockaddr_in6));
+		memcpy(__entry->dstaddr, &id_priv->id.route.addr.dst_addr,
+		       sizeof(struct sockaddr_in6));
+	),
+
+	TP_printk("cm.id=%u src=%pISpc dst=%pISpc tos=%u %s (%lu)",
+		__entry->cm_id, __entry->srcaddr, __entry->dstaddr, __entry->tos,
+		rdma_show_ib_cm_event(__entry->event), __entry->event
+	)
+);
+
+TRACE_EVENT(cm_event_handler,
+	TP_PROTO(
+		const struct rdma_id_private *id_priv,
+		const struct rdma_cm_event *event
+	),
+
+	TP_ARGS(id_priv, event),
+
+	TP_STRUCT__entry(
+		__field(u32, cm_id)
+		__field(u32, tos)
+		__field(unsigned long, event)
+		__field(int, status)
+		__array(unsigned char, srcaddr, sizeof(struct sockaddr_in6))
+		__array(unsigned char, dstaddr, sizeof(struct sockaddr_in6))
+	),
+
+	TP_fast_assign(
+		__entry->cm_id = id_priv->res.id;
+		__entry->tos = id_priv->tos;
+		__entry->event = event->event;
+		__entry->status = event->status;
+		memcpy(__entry->srcaddr, &id_priv->id.route.addr.src_addr,
+		       sizeof(struct sockaddr_in6));
+		memcpy(__entry->dstaddr, &id_priv->id.route.addr.dst_addr,
+		       sizeof(struct sockaddr_in6));
+	),
+
+	TP_printk("cm.id=%u src=%pISpc dst=%pISpc tos=%u %s (%lu/%d)",
+		__entry->cm_id, __entry->srcaddr, __entry->dstaddr, __entry->tos,
+		rdma_show_cm_event(__entry->event), __entry->event,
+		__entry->status
+	)
+);
+
+TRACE_EVENT(cm_event_done,
+	TP_PROTO(
+		const struct rdma_id_private *id_priv,
+		const struct rdma_cm_event *event,
+		int result
+	),
+
+	TP_ARGS(id_priv, event, result),
+
+	TP_STRUCT__entry(
+		__field(u32, cm_id)
+		__field(u32, tos)
+		__field(unsigned long, event)
+		__field(int, result)
+		__array(unsigned char, srcaddr, sizeof(struct sockaddr_in6))
+		__array(unsigned char, dstaddr, sizeof(struct sockaddr_in6))
+	),
+
+	TP_fast_assign(
+		__entry->cm_id = id_priv->res.id;
+		__entry->tos = id_priv->tos;
+		__entry->event = event->event;
+		__entry->result = result;
+		memcpy(__entry->srcaddr, &id_priv->id.route.addr.src_addr,
+		       sizeof(struct sockaddr_in6));
+		memcpy(__entry->dstaddr, &id_priv->id.route.addr.dst_addr,
+		       sizeof(struct sockaddr_in6));
+	),
+
+	TP_printk("cm.id=%u src=%pISpc dst=%pISpc tos=%u %s consumer returns %d",
+		__entry->cm_id, __entry->srcaddr, __entry->dstaddr, __entry->tos,
+		rdma_show_cm_event(__entry->event), __entry->result
+	)
+);
+
+DECLARE_EVENT_CLASS(cma_client_class,
+	TP_PROTO(
+		const struct ib_device *device
+	),
+
+	TP_ARGS(device),
+
+	TP_STRUCT__entry(
+		__string(name, device->name)
+	),
+
+	TP_fast_assign(
+		__assign_str(name, device->name);
+	),
+
+	TP_printk("device name=%s",
+		__get_str(name)
+	)
+);
+
+#define DEFINE_CMA_CLIENT_EVENT(name)						\
+		DEFINE_EVENT(cma_client_class, cm_##name,			\
+				TP_PROTO(					\
+					const struct ib_device *device		\
+				),						\
+				TP_ARGS(device))
+
+DEFINE_CMA_CLIENT_EVENT(add_one);
+DEFINE_CMA_CLIENT_EVENT(remove_one);
+
+#endif /* _TRACE_RDMA_CMA_H */
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+#define TRACE_INCLUDE_FILE cma_trace
+
+#include <trace/define_trace.h>
-- 
cgit v1.2.3-59-g8ed1b


From 3e5901cbfcc15da54f6ad148add754e7a2b2a558 Mon Sep 17 00:00:00 2001
From: Chuck Lever <chuck.lever@oracle.com>
Date: Wed, 18 Dec 2019 15:18:15 -0500
Subject: RDMA/core: Trace points for diagnosing completion queue issues

Sample trace events:

   kworker/u29:0-300   [007]   120.042217: cq_alloc:             cq.id=4 nr_cqe=161 comp_vector=2 poll_ctx=WORKQUEUE
          <idle>-0     [002]   120.056292: cq_schedule:          cq.id=4
    kworker/2:1H-482   [002]   120.056402: cq_process:           cq.id=4 wake-up took 109 [us] from interrupt
    kworker/2:1H-482   [002]   120.056407: cq_poll:              cq.id=4 requested 16, returned 1
          <idle>-0     [002]   120.067503: cq_schedule:          cq.id=4
    kworker/2:1H-482   [002]   120.067537: cq_process:           cq.id=4 wake-up took 34 [us] from interrupt
    kworker/2:1H-482   [002]   120.067541: cq_poll:              cq.id=4 requested 16, returned 1
          <idle>-0     [002]   120.067657: cq_schedule:          cq.id=4
    kworker/2:1H-482   [002]   120.067672: cq_process:           cq.id=4 wake-up took 15 [us] from interrupt
    kworker/2:1H-482   [002]   120.067674: cq_poll:              cq.id=4 requested 16, returned 1

 ...

         systemd-1     [002]   122.392653: cq_schedule:          cq.id=4
    kworker/2:1H-482   [002]   122.392688: cq_process:           cq.id=4 wake-up took 35 [us] from interrupt
    kworker/2:1H-482   [002]   122.392693: cq_poll:              cq.id=4 requested 16, returned 16
    kworker/2:1H-482   [002]   122.392836: cq_poll:              cq.id=4 requested 16, returned 16
    kworker/2:1H-482   [002]   122.392970: cq_poll:              cq.id=4 requested 16, returned 16
    kworker/2:1H-482   [002]   122.393083: cq_poll:              cq.id=4 requested 16, returned 16
    kworker/2:1H-482   [002]   122.393195: cq_poll:              cq.id=4 requested 16, returned 3

Several features to note in this output:
 - The WCE count and context type are reported at allocation time
 - The CPU and kworker for each CQ is evident
 - The CQ's restracker ID is tagged on each trace event
 - CQ poll scheduling latency is measured
 - Details about how often single completions occur versus multiple
   completions are evident
 - The cost of the ULP's completion handler is recorded

Link: https://lore.kernel.org/r/20191218201815.30584.3481.stgit@manet.1015granger.net
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/Makefile |   3 +-
 drivers/infiniband/core/cq.c     |  27 +++-
 drivers/infiniband/core/trace.c  |  14 ++
 drivers/infiniband/core/verbs.c  |   4 +
 include/rdma/ib_verbs.h          |   5 +
 include/trace/events/rdma_core.h | 271 +++++++++++++++++++++++++++++++++++++++
 6 files changed, 320 insertions(+), 4 deletions(-)
 create mode 100644 drivers/infiniband/core/trace.c
 create mode 100644 include/trace/events/rdma_core.h

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile
index f22555f982e2..2b86a514c9bf 100644
--- a/drivers/infiniband/core/Makefile
+++ b/drivers/infiniband/core/Makefile
@@ -11,7 +11,8 @@ ib_core-y :=			packer.o ud_header.o verbs.o cq.o rw.o sysfs.o \
 				device.o fmr_pool.o cache.o netlink.o \
 				roce_gid_mgmt.o mr_pool.o addr.o sa_query.o \
 				multicast.o mad.o smi.o agent.o mad_rmpp.o \
-				nldev.o restrack.o counters.o ib_core_uverbs.o
+				nldev.o restrack.o counters.o ib_core_uverbs.o \
+				trace.o
 
 ib_core-$(CONFIG_SECURITY_INFINIBAND) += security.o
 ib_core-$(CONFIG_CGROUP_RDMA) += cgroup.o
diff --git a/drivers/infiniband/core/cq.c b/drivers/infiniband/core/cq.c
index bbfded6d5d3d..4f25b2400694 100644
--- a/drivers/infiniband/core/cq.c
+++ b/drivers/infiniband/core/cq.c
@@ -7,6 +7,8 @@
 #include <linux/slab.h>
 #include <rdma/ib_verbs.h>
 
+#include <trace/events/rdma_core.h>
+
 /* # of WCs to poll for with a single call to ib_poll_cq */
 #define IB_POLL_BATCH			16
 #define IB_POLL_BATCH_DIRECT		8
@@ -41,6 +43,7 @@ static void ib_cq_rdma_dim_work(struct work_struct *w)
 
 	dim->state = DIM_START_MEASURE;
 
+	trace_cq_modify(cq, comps, usec);
 	cq->device->ops.modify_cq(cq, comps, usec);
 }
 
@@ -65,18 +68,29 @@ static void rdma_dim_init(struct ib_cq *cq)
 	INIT_WORK(&dim->work, ib_cq_rdma_dim_work);
 }
 
+static int __poll_cq(struct ib_cq *cq, int num_entries, struct ib_wc *wc)
+{
+	int rc;
+
+	rc = ib_poll_cq(cq, num_entries, wc);
+	trace_cq_poll(cq, num_entries, rc);
+	return rc;
+}
+
 static int __ib_process_cq(struct ib_cq *cq, int budget, struct ib_wc *wcs,
 			   int batch)
 {
 	int i, n, completed = 0;
 
+	trace_cq_process(cq);
+
 	/*
 	 * budget might be (-1) if the caller does not
 	 * want to bound this call, thus we need unsigned
 	 * minimum here.
 	 */
-	while ((n = ib_poll_cq(cq, min_t(u32, batch,
-					 budget - completed), wcs)) > 0) {
+	while ((n = __poll_cq(cq, min_t(u32, batch,
+					budget - completed), wcs)) > 0) {
 		for (i = 0; i < n; i++) {
 			struct ib_wc *wc = &wcs[i];
 
@@ -131,8 +145,10 @@ static int ib_poll_handler(struct irq_poll *iop, int budget)
 	completed = __ib_process_cq(cq, budget, cq->wc, IB_POLL_BATCH);
 	if (completed < budget) {
 		irq_poll_complete(&cq->iop);
-		if (ib_req_notify_cq(cq, IB_POLL_FLAGS) > 0)
+		if (ib_req_notify_cq(cq, IB_POLL_FLAGS) > 0) {
+			trace_cq_reschedule(cq);
 			irq_poll_sched(&cq->iop);
+		}
 	}
 
 	if (dim)
@@ -143,6 +159,7 @@ static int ib_poll_handler(struct irq_poll *iop, int budget)
 
 static void ib_cq_completion_softirq(struct ib_cq *cq, void *private)
 {
+	trace_cq_schedule(cq);
 	irq_poll_sched(&cq->iop);
 }
 
@@ -162,6 +179,7 @@ static void ib_cq_poll_work(struct work_struct *work)
 
 static void ib_cq_completion_workqueue(struct ib_cq *cq, void *private)
 {
+	trace_cq_schedule(cq);
 	queue_work(cq->comp_wq, &cq->work);
 }
 
@@ -239,6 +257,7 @@ struct ib_cq *__ib_alloc_cq_user(struct ib_device *dev, void *private,
 		goto out_destroy_cq;
 	}
 
+	trace_cq_alloc(cq, nr_cqe, comp_vector, poll_ctx);
 	return cq;
 
 out_destroy_cq:
@@ -248,6 +267,7 @@ out_free_wc:
 	kfree(cq->wc);
 out_free_cq:
 	kfree(cq);
+	trace_cq_alloc_error(nr_cqe, comp_vector, poll_ctx, ret);
 	return ERR_PTR(ret);
 }
 EXPORT_SYMBOL(__ib_alloc_cq_user);
@@ -304,6 +324,7 @@ void ib_free_cq_user(struct ib_cq *cq, struct ib_udata *udata)
 		WARN_ON_ONCE(1);
 	}
 
+	trace_cq_free(cq);
 	rdma_restrack_del(&cq->res);
 	cq->device->ops.destroy_cq(cq, udata);
 	if (cq->dim)
diff --git a/drivers/infiniband/core/trace.c b/drivers/infiniband/core/trace.c
new file mode 100644
index 000000000000..6c3514beac4d
--- /dev/null
+++ b/drivers/infiniband/core/trace.c
@@ -0,0 +1,14 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Trace points for core RDMA functions.
+ *
+ * Author: Chuck Lever <chuck.lever@oracle.com>
+ *
+ * Copyright (c) 2019, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#define CREATE_TRACE_POINTS
+
+#include <rdma/ib_verbs.h>
+
+#include <trace/events/rdma_core.h>
diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index dd765e176cdd..289b2f7a9d5e 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -53,6 +53,8 @@
 
 #include "core_priv.h"
 
+#include <trace/events/rdma_core.h>
+
 static int ib_resolve_eth_dmac(struct ib_device *device,
 			       struct rdma_ah_attr *ah_attr);
 
@@ -2744,6 +2746,7 @@ void ib_drain_sq(struct ib_qp *qp)
 		qp->device->ops.drain_sq(qp);
 	else
 		__ib_drain_sq(qp);
+	trace_cq_drain_complete(qp->send_cq);
 }
 EXPORT_SYMBOL(ib_drain_sq);
 
@@ -2772,6 +2775,7 @@ void ib_drain_rq(struct ib_qp *qp)
 		qp->device->ops.drain_rq(qp);
 	else
 		__ib_drain_rq(qp);
+	trace_cq_drain_complete(qp->recv_cq);
 }
 EXPORT_SYMBOL(ib_drain_rq);
 
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 5608e14e3aad..42f28d39f28c 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -1558,6 +1558,11 @@ struct ib_cq {
 	};
 	struct workqueue_struct *comp_wq;
 	struct dim *dim;
+
+	/* updated only by trace points */
+	ktime_t timestamp;
+	bool interrupt;
+
 	/*
 	 * Implementation details of the RDMA core, don't use in drivers:
 	 */
diff --git a/include/trace/events/rdma_core.h b/include/trace/events/rdma_core.h
new file mode 100644
index 000000000000..08f481554e7f
--- /dev/null
+++ b/include/trace/events/rdma_core.h
@@ -0,0 +1,271 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Trace point definitions for core RDMA functions.
+ *
+ * Author: Chuck Lever <chuck.lever@oracle.com>
+ *
+ * Copyright (c) 2019, Oracle and/or its affiliates. All rights reserved.
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM rdma_core
+
+#if !defined(_TRACE_RDMA_CORE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_RDMA_CORE_H
+
+#include <linux/tracepoint.h>
+#include <rdma/ib_verbs.h>
+
+/*
+ * enum ib_poll_context, from include/rdma/ib_verbs.h
+ */
+#define IB_POLL_CTX_LIST			\
+	ib_poll_ctx(DIRECT)			\
+	ib_poll_ctx(SOFTIRQ)			\
+	ib_poll_ctx(WORKQUEUE)			\
+	ib_poll_ctx_end(UNBOUND_WORKQUEUE)
+
+#undef ib_poll_ctx
+#undef ib_poll_ctx_end
+
+#define ib_poll_ctx(x)		TRACE_DEFINE_ENUM(IB_POLL_##x);
+#define ib_poll_ctx_end(x)	TRACE_DEFINE_ENUM(IB_POLL_##x);
+
+IB_POLL_CTX_LIST
+
+#undef ib_poll_ctx
+#undef ib_poll_ctx_end
+
+#define ib_poll_ctx(x)		{ IB_POLL_##x, #x },
+#define ib_poll_ctx_end(x)	{ IB_POLL_##x, #x }
+
+#define rdma_show_ib_poll_ctx(x) \
+		__print_symbolic(x, IB_POLL_CTX_LIST)
+
+/**
+ ** Completion Queue events
+ **/
+
+TRACE_EVENT(cq_schedule,
+	TP_PROTO(
+		struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		cq->timestamp = ktime_get();
+		cq->interrupt = true;
+
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u", __entry->cq_id)
+);
+
+TRACE_EVENT(cq_reschedule,
+	TP_PROTO(
+		struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		cq->timestamp = ktime_get();
+		cq->interrupt = false;
+
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u", __entry->cq_id)
+);
+
+TRACE_EVENT(cq_process,
+	TP_PROTO(
+		const struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(bool, interrupt)
+		__field(s64, latency)
+	),
+
+	TP_fast_assign(
+		ktime_t latency = ktime_sub(ktime_get(), cq->timestamp);
+
+		__entry->cq_id = cq->res.id;
+		__entry->latency = ktime_to_us(latency);
+		__entry->interrupt = cq->interrupt;
+	),
+
+	TP_printk("cq.id=%u wake-up took %lld [us] from %s",
+		__entry->cq_id, __entry->latency,
+		__entry->interrupt ? "interrupt" : "reschedule"
+	)
+);
+
+TRACE_EVENT(cq_poll,
+	TP_PROTO(
+		const struct ib_cq *cq,
+		int requested,
+		int rc
+	),
+
+	TP_ARGS(cq, requested, rc),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, requested)
+		__field(int, rc)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+		__entry->requested = requested;
+		__entry->rc = rc;
+	),
+
+	TP_printk("cq.id=%u requested %d, returned %d",
+		__entry->cq_id, __entry->requested, __entry->rc
+	)
+);
+
+TRACE_EVENT(cq_drain_complete,
+	TP_PROTO(
+		const struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u",
+		__entry->cq_id
+	)
+);
+
+
+TRACE_EVENT(cq_modify,
+	TP_PROTO(
+		const struct ib_cq *cq,
+		u16 comps,
+		u16 usec
+	),
+
+	TP_ARGS(cq, comps, usec),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(unsigned int, comps)
+		__field(unsigned int, usec)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+		__entry->comps = comps;
+		__entry->usec = usec;
+	),
+
+	TP_printk("cq.id=%u comps=%u usec=%u",
+		__entry->cq_id, __entry->comps, __entry->usec
+	)
+);
+
+TRACE_EVENT(cq_alloc,
+	TP_PROTO(
+		const struct ib_cq *cq,
+		int nr_cqe,
+		int comp_vector,
+		enum ib_poll_context poll_ctx
+	),
+
+	TP_ARGS(cq, nr_cqe, comp_vector, poll_ctx),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+		__field(int, nr_cqe)
+		__field(int, comp_vector)
+		__field(unsigned long, poll_ctx)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+		__entry->nr_cqe = nr_cqe;
+		__entry->comp_vector = comp_vector;
+		__entry->poll_ctx = poll_ctx;
+	),
+
+	TP_printk("cq.id=%u nr_cqe=%d comp_vector=%d poll_ctx=%s",
+		__entry->cq_id, __entry->nr_cqe, __entry->comp_vector,
+		rdma_show_ib_poll_ctx(__entry->poll_ctx)
+	)
+);
+
+TRACE_EVENT(cq_alloc_error,
+	TP_PROTO(
+		int nr_cqe,
+		int comp_vector,
+		enum ib_poll_context poll_ctx,
+		int rc
+	),
+
+	TP_ARGS(nr_cqe, comp_vector, poll_ctx, rc),
+
+	TP_STRUCT__entry(
+		__field(int, rc)
+		__field(int, nr_cqe)
+		__field(int, comp_vector)
+		__field(unsigned long, poll_ctx)
+	),
+
+	TP_fast_assign(
+		__entry->rc = rc;
+		__entry->nr_cqe = nr_cqe;
+		__entry->comp_vector = comp_vector;
+		__entry->poll_ctx = poll_ctx;
+	),
+
+	TP_printk("nr_cqe=%d comp_vector=%d poll_ctx=%s rc=%d",
+		__entry->nr_cqe, __entry->comp_vector,
+		rdma_show_ib_poll_ctx(__entry->poll_ctx), __entry->rc
+	)
+);
+
+TRACE_EVENT(cq_free,
+	TP_PROTO(
+		const struct ib_cq *cq
+	),
+
+	TP_ARGS(cq),
+
+	TP_STRUCT__entry(
+		__field(u32, cq_id)
+	),
+
+	TP_fast_assign(
+		__entry->cq_id = cq->res.id;
+	),
+
+	TP_printk("cq.id=%u", __entry->cq_id)
+);
+
+#endif /* _TRACE_RDMA_CORE_H */
+
+#include <trace/define_trace.h>
-- 
cgit v1.2.3-59-g8ed1b


From 622db5b6439ab4d28a44d2d49d58050e59cac781 Mon Sep 17 00:00:00 2001
From: Chuck Lever <chuck.lever@oracle.com>
Date: Wed, 18 Dec 2019 15:18:20 -0500
Subject: RDMA/core: Add trace points to follow MR allocation

Track the lifetime of ib_mr objects. Here's sample output from a test run
with NFS/RDMA:

           <...>-361   [009] 79238.772782: mr_alloc:             pd.id=3 mr.id=11 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.772812: mr_alloc:             pd.id=3 mr.id=12 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.772839: mr_alloc:             pd.id=3 mr.id=13 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.772866: mr_alloc:             pd.id=3 mr.id=14 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.772893: mr_alloc:             pd.id=3 mr.id=15 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.772921: mr_alloc:             pd.id=3 mr.id=16 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.772947: mr_alloc:             pd.id=3 mr.id=17 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.772974: mr_alloc:             pd.id=3 mr.id=18 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.773001: mr_alloc:             pd.id=3 mr.id=19 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.773028: mr_alloc:             pd.id=3 mr.id=20 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79238.773055: mr_alloc:             pd.id=3 mr.id=21 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.270942: mr_alloc:             pd.id=3 mr.id=22 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.270975: mr_alloc:             pd.id=3 mr.id=23 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271007: mr_alloc:             pd.id=3 mr.id=24 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271036: mr_alloc:             pd.id=3 mr.id=25 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271067: mr_alloc:             pd.id=3 mr.id=26 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271095: mr_alloc:             pd.id=3 mr.id=27 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271121: mr_alloc:             pd.id=3 mr.id=28 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271153: mr_alloc:             pd.id=3 mr.id=29 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271181: mr_alloc:             pd.id=3 mr.id=30 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271208: mr_alloc:             pd.id=3 mr.id=31 type=MEM_REG max_num_sg=30 rc=0
           <...>-361   [009] 79240.271236: mr_alloc:             pd.id=3 mr.id=32 type=MEM_REG max_num_sg=30 rc=0
           <...>-4351  [001] 79242.299400: mr_dereg:             mr.id=32
           <...>-4351  [001] 79242.299467: mr_dereg:             mr.id=31
           <...>-4351  [001] 79242.299554: mr_dereg:             mr.id=30
           <...>-4351  [001] 79242.299615: mr_dereg:             mr.id=29
           <...>-4351  [001] 79242.299684: mr_dereg:             mr.id=28
           <...>-4351  [001] 79242.299748: mr_dereg:             mr.id=27
           <...>-4351  [001] 79242.299812: mr_dereg:             mr.id=26
           <...>-4351  [001] 79242.299874: mr_dereg:             mr.id=25
           <...>-4351  [001] 79242.299944: mr_dereg:             mr.id=24
           <...>-4351  [001] 79242.300009: mr_dereg:             mr.id=23
           <...>-4351  [001] 79242.300190: mr_dereg:             mr.id=22
           <...>-4351  [001] 79242.300263: mr_dereg:             mr.id=21
           <...>-4351  [001] 79242.300326: mr_dereg:             mr.id=20
           <...>-4351  [001] 79242.300388: mr_dereg:             mr.id=19
           <...>-4351  [001] 79242.300450: mr_dereg:             mr.id=18
           <...>-4351  [001] 79242.300516: mr_dereg:             mr.id=17
           <...>-4351  [001] 79242.300629: mr_dereg:             mr.id=16
           <...>-4351  [001] 79242.300718: mr_dereg:             mr.id=15
           <...>-4351  [001] 79242.300784: mr_dereg:             mr.id=14
           <...>-4351  [001] 79242.300879: mr_dereg:             mr.id=13
           <...>-4351  [001] 79242.300945: mr_dereg:             mr.id=12
           <...>-4351  [001] 79242.301012: mr_dereg:             mr.id=11

Some features of the output:
- The lifetime and owner PD of each MR is clearly visible.
- The type of MR is captured, as is the SGE array size.
- Failing MR allocation can be recorded.

Link: https://lore.kernel.org/r/20191218201820.30584.34636.stgit@manet.1015granger.net
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/verbs.c  |  39 +++++++++----
 include/trace/events/rdma_core.h | 123 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 151 insertions(+), 11 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index 289b2f7a9d5e..47d54c31eb2a 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -52,6 +52,7 @@
 #include <rdma/rw.h>
 
 #include "core_priv.h"
+#include <trace/events/rdma_core.h>
 
 #include <trace/events/rdma_core.h>
 
@@ -1999,6 +2000,7 @@ int ib_dereg_mr_user(struct ib_mr *mr, struct ib_udata *udata)
 	struct ib_sig_attrs *sig_attrs = mr->sig_attrs;
 	int ret;
 
+	trace_mr_dereg(mr);
 	rdma_restrack_del(&mr->res);
 	ret = mr->device->ops.dereg_mr(mr, udata);
 	if (!ret) {
@@ -2030,11 +2032,16 @@ struct ib_mr *ib_alloc_mr_user(struct ib_pd *pd, enum ib_mr_type mr_type,
 {
 	struct ib_mr *mr;
 
-	if (!pd->device->ops.alloc_mr)
-		return ERR_PTR(-EOPNOTSUPP);
+	if (!pd->device->ops.alloc_mr) {
+		mr = ERR_PTR(-EOPNOTSUPP);
+		goto out;
+	}
 
-	if (WARN_ON_ONCE(mr_type == IB_MR_TYPE_INTEGRITY))
-		return ERR_PTR(-EINVAL);
+	if (mr_type == IB_MR_TYPE_INTEGRITY) {
+		WARN_ON_ONCE(1);
+		mr = ERR_PTR(-EINVAL);
+		goto out;
+	}
 
 	mr = pd->device->ops.alloc_mr(pd, mr_type, max_num_sg, udata);
 	if (!IS_ERR(mr)) {
@@ -2050,6 +2057,8 @@ struct ib_mr *ib_alloc_mr_user(struct ib_pd *pd, enum ib_mr_type mr_type,
 		mr->sig_attrs = NULL;
 	}
 
+out:
+	trace_mr_alloc(pd, mr_type, max_num_sg, mr);
 	return mr;
 }
 EXPORT_SYMBOL(ib_alloc_mr_user);
@@ -2074,21 +2083,27 @@ struct ib_mr *ib_alloc_mr_integrity(struct ib_pd *pd,
 	struct ib_sig_attrs *sig_attrs;
 
 	if (!pd->device->ops.alloc_mr_integrity ||
-	    !pd->device->ops.map_mr_sg_pi)
-		return ERR_PTR(-EOPNOTSUPP);
+	    !pd->device->ops.map_mr_sg_pi) {
+		mr = ERR_PTR(-EOPNOTSUPP);
+		goto out;
+	}
 
-	if (!max_num_meta_sg)
-		return ERR_PTR(-EINVAL);
+	if (!max_num_meta_sg) {
+		mr = ERR_PTR(-EINVAL);
+		goto out;
+	}
 
 	sig_attrs = kzalloc(sizeof(struct ib_sig_attrs), GFP_KERNEL);
-	if (!sig_attrs)
-		return ERR_PTR(-ENOMEM);
+	if (!sig_attrs) {
+		mr = ERR_PTR(-ENOMEM);
+		goto out;
+	}
 
 	mr = pd->device->ops.alloc_mr_integrity(pd, max_num_data_sg,
 						max_num_meta_sg);
 	if (IS_ERR(mr)) {
 		kfree(sig_attrs);
-		return mr;
+		goto out;
 	}
 
 	mr->device = pd->device;
@@ -2102,6 +2117,8 @@ struct ib_mr *ib_alloc_mr_integrity(struct ib_pd *pd,
 	mr->type = IB_MR_TYPE_INTEGRITY;
 	mr->sig_attrs = sig_attrs;
 
+out:
+	trace_mr_integ_alloc(pd, max_num_data_sg, max_num_meta_sg, mr);
 	return mr;
 }
 EXPORT_SYMBOL(ib_alloc_mr_integrity);
diff --git a/include/trace/events/rdma_core.h b/include/trace/events/rdma_core.h
index 08f481554e7f..17642aa54437 100644
--- a/include/trace/events/rdma_core.h
+++ b/include/trace/events/rdma_core.h
@@ -266,6 +266,129 @@ TRACE_EVENT(cq_free,
 	TP_printk("cq.id=%u", __entry->cq_id)
 );
 
+/**
+ ** Memory Region events
+ **/
+
+/*
+ * enum ib_mr_type, from include/rdma/ib_verbs.h
+ */
+#define IB_MR_TYPE_LIST				\
+	ib_mr_type_item(MEM_REG)		\
+	ib_mr_type_item(SG_GAPS)		\
+	ib_mr_type_item(DM)			\
+	ib_mr_type_item(USER)			\
+	ib_mr_type_item(DMA)			\
+	ib_mr_type_end(INTEGRITY)
+
+#undef ib_mr_type_item
+#undef ib_mr_type_end
+
+#define ib_mr_type_item(x)	TRACE_DEFINE_ENUM(IB_MR_TYPE_##x);
+#define ib_mr_type_end(x)	TRACE_DEFINE_ENUM(IB_MR_TYPE_##x);
+
+IB_MR_TYPE_LIST
+
+#undef ib_mr_type_item
+#undef ib_mr_type_end
+
+#define ib_mr_type_item(x)	{ IB_MR_TYPE_##x, #x },
+#define ib_mr_type_end(x)	{ IB_MR_TYPE_##x, #x }
+
+#define rdma_show_ib_mr_type(x) \
+		__print_symbolic(x, IB_MR_TYPE_LIST)
+
+TRACE_EVENT(mr_alloc,
+	TP_PROTO(
+		const struct ib_pd *pd,
+		enum ib_mr_type mr_type,
+		u32 max_num_sg,
+		const struct ib_mr *mr
+	),
+
+	TP_ARGS(pd, mr_type, max_num_sg, mr),
+
+	TP_STRUCT__entry(
+		__field(u32, pd_id)
+		__field(u32, mr_id)
+		__field(u32, max_num_sg)
+		__field(int, rc)
+		__field(unsigned long, mr_type)
+	),
+
+	TP_fast_assign(
+		__entry->pd_id = pd->res.id;
+		if (IS_ERR(mr)) {
+			__entry->mr_id = 0;
+			__entry->rc = PTR_ERR(mr);
+		} else {
+			__entry->mr_id = mr->res.id;
+			__entry->rc = 0;
+		}
+		__entry->max_num_sg = max_num_sg;
+		__entry->mr_type = mr_type;
+	),
+
+	TP_printk("pd.id=%u mr.id=%u type=%s max_num_sg=%u rc=%d",
+		__entry->pd_id, __entry->mr_id,
+		rdma_show_ib_mr_type(__entry->mr_type),
+		__entry->max_num_sg, __entry->rc)
+);
+
+TRACE_EVENT(mr_integ_alloc,
+	TP_PROTO(
+		const struct ib_pd *pd,
+		u32 max_num_data_sg,
+		u32 max_num_meta_sg,
+		const struct ib_mr *mr
+	),
+
+	TP_ARGS(pd, max_num_data_sg, max_num_meta_sg, mr),
+
+	TP_STRUCT__entry(
+		__field(u32, pd_id)
+		__field(u32, mr_id)
+		__field(u32, max_num_data_sg)
+		__field(u32, max_num_meta_sg)
+		__field(int, rc)
+	),
+
+	TP_fast_assign(
+		__entry->pd_id = pd->res.id;
+		if (IS_ERR(mr)) {
+			__entry->mr_id = 0;
+			__entry->rc = PTR_ERR(mr);
+		} else {
+			__entry->mr_id = mr->res.id;
+			__entry->rc = 0;
+		}
+		__entry->max_num_data_sg = max_num_data_sg;
+		__entry->max_num_meta_sg = max_num_meta_sg;
+	),
+
+	TP_printk("pd.id=%u mr.id=%u max_num_data_sg=%u max_num_meta_sg=%u rc=%d",
+		__entry->pd_id, __entry->mr_id, __entry->max_num_data_sg,
+		__entry->max_num_meta_sg, __entry->rc)
+);
+
+TRACE_EVENT(mr_dereg,
+	TP_PROTO(
+		const struct ib_mr *mr
+	),
+
+	TP_ARGS(mr),
+
+	TP_STRUCT__entry(
+		__field(u32, id)
+	),
+
+	TP_fast_assign(
+		__entry->id = mr->res.id;
+	),
+
+	TP_printk("mr.id=%u", __entry->id)
+);
+
 #endif /* _TRACE_RDMA_CORE_H */
 
 #include <trace/define_trace.h>
-- 
cgit v1.2.3-59-g8ed1b


From eca44507c3e908b7362696a4d6a11d90371334c6 Mon Sep 17 00:00:00 2001
From: Wenpeng Liang <liangwenpeng@huawei.com>
Date: Mon, 6 Jan 2020 20:21:10 +0800
Subject: RDMA/hns: Avoid printing address of mtt page

Address of a page shouldn't be printed in case of security issues.

Link: https://lore.kernel.org/r/1578313276-29080-2-git-send-email-liweihang@huawei.com
Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_mr.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c
index 9ad19170c3f9..95765560c1cf 100644
--- a/drivers/infiniband/hw/hns/hns_roce_mr.c
+++ b/drivers/infiniband/hw/hns/hns_roce_mr.c
@@ -1064,8 +1064,8 @@ int hns_roce_ib_umem_write_mtt(struct hns_roce_dev *hr_dev,
 		if (!(npage % (1 << (mtt->page_shift - PAGE_SHIFT)))) {
 			if (page_addr & ((1 << mtt->page_shift) - 1)) {
 				dev_err(dev,
-					"page_addr 0x%llx is not page_shift %d alignment!\n",
-					page_addr, mtt->page_shift);
+					"page_addr is not page_shift %d alignment!\n",
+					mtt->page_shift);
 				ret = -EINVAL;
 				goto out;
 			}
-- 
cgit v1.2.3-59-g8ed1b


From 58e4fc11c1a56756a3fb45a10d5cb09bb68b3030 Mon Sep 17 00:00:00 2001
From: Lijun Ou <oulijun@huawei.com>
Date: Mon, 6 Jan 2020 20:21:11 +0800
Subject: RDMA/hns: Remove unused function hns_roce_init_eq_table()

hns_roce_init_eq_table() is an unused function that only retains its
declaration in driver.

Link: https://lore.kernel.org/r/1578313276-29080-3-git-send-email-liweihang@huawei.com
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_device.h | 1 -
 1 file changed, 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 416341ada172..e447c40fee96 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -1132,7 +1132,6 @@ int hns_roce_mtr_find(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr,
 
 int hns_roce_init_pd_table(struct hns_roce_dev *hr_dev);
 int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev);
-int hns_roce_init_eq_table(struct hns_roce_dev *hr_dev);
 int hns_roce_init_cq_table(struct hns_roce_dev *hr_dev);
 int hns_roce_init_qp_table(struct hns_roce_dev *hr_dev);
 int hns_roce_init_srq_table(struct hns_roce_dev *hr_dev);
-- 
cgit v1.2.3-59-g8ed1b


From 5e049a5d6c83b8a4aea70f02d4fd81f592a2cb49 Mon Sep 17 00:00:00 2001
From: Lijun Ou <oulijun@huawei.com>
Date: Mon, 6 Jan 2020 20:21:12 +0800
Subject: RDMA/hns: Update the value of qp type

The values used to represent service type of RC and UD should be
interchanged according to design of hardware. And it's better to define
these types in enumeration than macros.

Link: https://lore.kernel.org/r/1578313276-29080-4-git-send-email-liweihang@huawei.com
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_device.h | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index e447c40fee96..21751e43aab3 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -107,11 +107,6 @@
 #define NODE_DESC_SIZE				64
 #define DB_REG_OFFSET				0x1000
 
-#define SERV_TYPE_RC				0
-#define SERV_TYPE_RD				1
-#define SERV_TYPE_UC				2
-#define SERV_TYPE_UD				3
-
 /* Configure to HW for PAGE_SIZE larger than 4KB */
 #define PG_SHIFT_OFFSET				(PAGE_SHIFT - 12)
 
@@ -130,6 +125,13 @@
  */
 #define EQ_DEPTH_COEFF				2
 
+enum {
+	SERV_TYPE_RC,
+	SERV_TYPE_UC,
+	SERV_TYPE_RD,
+	SERV_TYPE_UD,
+};
+
 enum {
 	HNS_ROCE_SUPPORT_RQ_RECORD_DB = 1 << 0,
 	HNS_ROCE_SUPPORT_SQ_RECORD_DB = 1 << 1,
-- 
cgit v1.2.3-59-g8ed1b


From 032b0574161fdd381220de7cfd26ad3cb284cbc0 Mon Sep 17 00:00:00 2001
From: Lijun Ou <oulijun@huawei.com>
Date: Mon, 6 Jan 2020 20:21:13 +0800
Subject: RDMA/hns: Delete unnessary parameters in hns_roce_v2_qp_modify()

Current state and new state of qp won't be configured when modifying qp,
so these two redundant parameters should be removed.

Link: https://lore.kernel.org/r/1578313276-29080-5-git-send-email-liweihang@huawei.com
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 45ff44dfabc3..9b9ee22b606b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -3153,8 +3153,6 @@ static int hns_roce_v2_clear_hem(struct hns_roce_dev *hr_dev,
 }
 
 static int hns_roce_v2_qp_modify(struct hns_roce_dev *hr_dev,
-				 enum ib_qp_state cur_state,
-				 enum ib_qp_state new_state,
 				 struct hns_roce_v2_qp_context *context,
 				 struct hns_roce_qp *hr_qp)
 {
@@ -4437,7 +4435,7 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp,
 		       V2_QPC_BYTE_60_QP_ST_S, 0);
 
 	/* SW pass context to HW */
-	ret = hns_roce_v2_qp_modify(hr_dev, cur_state, new_state, ctx, hr_qp);
+	ret = hns_roce_v2_qp_modify(hr_dev, ctx, hr_qp);
 	if (ret) {
 		dev_err(dev, "hns_roce_qp_modify failed(%d)\n", ret);
 		goto out;
-- 
cgit v1.2.3-59-g8ed1b


From 0c53426c7c1a4ad1b2fc647a7085b5d3e7b49025 Mon Sep 17 00:00:00 2001
From: Yixing Liu <liuyixing1@huawei.com>
Date: Mon, 6 Jan 2020 20:21:14 +0800
Subject: RDMA/hns: Remove redundant print information

There are already necessary prints in outer function, prints in
hns_roce_function_clear() may confuse users. So these prints is removed.

Link: https://lore.kernel.org/r/1578313276-29080-6-git-send-email-liweihang@huawei.com
Signed-off-by: Yixing Liu <liuyixing1@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 9b9ee22b606b..d53025b3ae97 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -1248,7 +1248,6 @@ static void hns_roce_function_clear(struct hns_roce_dev *hr_dev)
 	}
 
 out:
-	dev_err(hr_dev->dev, "Func clear fail.\n");
 	hns_roce_func_clr_rst_prc(hr_dev, ret, fclr_write_fail_flag);
 }
 
-- 
cgit v1.2.3-59-g8ed1b


From d800c93bacc7a9c381b5fa9c333fb821ba9df9df Mon Sep 17 00:00:00 2001
From: Wenpeng Liang <liangwenpeng@huawei.com>
Date: Mon, 6 Jan 2020 20:21:15 +0800
Subject: RDMA/hns: Replace custom macros HNS_ROCE_ALIGN_UP

HNS_ROCE_ALIGN_UP can be replaced by round_up() which is defined in
kernel.h.

Link: https://lore.kernel.org/r/1578313276-29080-7-git-send-email-liweihang@huawei.com
Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_device.h |  2 --
 drivers/infiniband/hw/hns/hns_roce_qp.c     | 44 +++++++++++++----------------
 2 files changed, 20 insertions(+), 26 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 21751e43aab3..93c2210fa60e 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -45,8 +45,6 @@
 
 #define HNS_ROCE_MAX_MSG_LEN			0x80000000
 
-#define HNS_ROCE_ALIGN_UP(a, b) ((((a) + (b) - 1) / (b)) * (b))
-
 #define HNS_ROCE_IB_MIN_SQ_STRIDE		6
 
 #define HNS_ROCE_BA_SIZE			(32 * 4096)
diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c
index a6565b674801..c5b01ec8ca6a 100644
--- a/drivers/infiniband/hw/hns/hns_roce_qp.c
+++ b/drivers/infiniband/hw/hns/hns_roce_qp.c
@@ -393,40 +393,38 @@ static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev,
 
 	/* Get buf size, SQ and RQ  are aligned to page_szie */
 	if (hr_dev->caps.max_sq_sg <= 2) {
-		hr_qp->buff_size = HNS_ROCE_ALIGN_UP((hr_qp->rq.wqe_cnt <<
+		hr_qp->buff_size = round_up((hr_qp->rq.wqe_cnt <<
 					     hr_qp->rq.wqe_shift), PAGE_SIZE) +
-				   HNS_ROCE_ALIGN_UP((hr_qp->sq.wqe_cnt <<
+				   round_up((hr_qp->sq.wqe_cnt <<
 					     hr_qp->sq.wqe_shift), PAGE_SIZE);
 
 		hr_qp->sq.offset = 0;
-		hr_qp->rq.offset = HNS_ROCE_ALIGN_UP((hr_qp->sq.wqe_cnt <<
+		hr_qp->rq.offset = round_up((hr_qp->sq.wqe_cnt <<
 					     hr_qp->sq.wqe_shift), PAGE_SIZE);
 	} else {
 		page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT);
 		hr_qp->sge.sge_cnt = ex_sge_num ?
 		   max(page_size / (1 << hr_qp->sge.sge_shift), ex_sge_num) : 0;
-		hr_qp->buff_size = HNS_ROCE_ALIGN_UP((hr_qp->rq.wqe_cnt <<
+		hr_qp->buff_size = round_up((hr_qp->rq.wqe_cnt <<
 					     hr_qp->rq.wqe_shift), page_size) +
-				   HNS_ROCE_ALIGN_UP((hr_qp->sge.sge_cnt <<
+				   round_up((hr_qp->sge.sge_cnt <<
 					     hr_qp->sge.sge_shift), page_size) +
-				   HNS_ROCE_ALIGN_UP((hr_qp->sq.wqe_cnt <<
+				   round_up((hr_qp->sq.wqe_cnt <<
 					     hr_qp->sq.wqe_shift), page_size);
 
 		hr_qp->sq.offset = 0;
 		if (ex_sge_num) {
-			hr_qp->sge.offset = HNS_ROCE_ALIGN_UP(
-							(hr_qp->sq.wqe_cnt <<
-							hr_qp->sq.wqe_shift),
-							page_size);
+			hr_qp->sge.offset = round_up((hr_qp->sq.wqe_cnt <<
+						      hr_qp->sq.wqe_shift),
+						     page_size);
 			hr_qp->rq.offset = hr_qp->sge.offset +
-					HNS_ROCE_ALIGN_UP((hr_qp->sge.sge_cnt <<
-						hr_qp->sge.sge_shift),
-						page_size);
+					   round_up((hr_qp->sge.sge_cnt <<
+						     hr_qp->sge.sge_shift),
+						    page_size);
 		} else {
-			hr_qp->rq.offset = HNS_ROCE_ALIGN_UP(
-							(hr_qp->sq.wqe_cnt <<
-							hr_qp->sq.wqe_shift),
-							page_size);
+			hr_qp->rq.offset = round_up((hr_qp->sq.wqe_cnt <<
+						     hr_qp->sq.wqe_shift),
+						    page_size);
 		}
 	}
 
@@ -593,20 +591,18 @@ static int hns_roce_set_kernel_sq_size(struct hns_roce_dev *hr_dev,
 	/* Get buf size, SQ and RQ are aligned to PAGE_SIZE */
 	page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT);
 	hr_qp->sq.offset = 0;
-	size = HNS_ROCE_ALIGN_UP(hr_qp->sq.wqe_cnt << hr_qp->sq.wqe_shift,
-				 page_size);
+	size = round_up(hr_qp->sq.wqe_cnt << hr_qp->sq.wqe_shift, page_size);
 
 	if (hr_dev->caps.max_sq_sg > 2 && hr_qp->sge.sge_cnt) {
 		hr_qp->sge.sge_cnt = max(page_size/(1 << hr_qp->sge.sge_shift),
-					(u32)hr_qp->sge.sge_cnt);
+					 (u32)hr_qp->sge.sge_cnt);
 		hr_qp->sge.offset = size;
-		size += HNS_ROCE_ALIGN_UP(hr_qp->sge.sge_cnt <<
-					  hr_qp->sge.sge_shift, page_size);
+		size += round_up(hr_qp->sge.sge_cnt << hr_qp->sge.sge_shift,
+				 page_size);
 	}
 
 	hr_qp->rq.offset = size;
-	size += HNS_ROCE_ALIGN_UP((hr_qp->rq.wqe_cnt << hr_qp->rq.wqe_shift),
-				  page_size);
+	size += round_up((hr_qp->rq.wqe_cnt << hr_qp->rq.wqe_shift), page_size);
 	hr_qp->buff_size = size;
 
 	/* Get wr and sge number which send */
-- 
cgit v1.2.3-59-g8ed1b


From 60262b10a96bb186049f1eca4849ccbb2e81f253 Mon Sep 17 00:00:00 2001
From: Lijun Ou <oulijun@huawei.com>
Date: Mon, 6 Jan 2020 20:21:16 +0800
Subject: RDMA/hns: Fix coding style issues

Fix some coding style issuses without changing logic of codes, most of the
modification is unreasonable line breaks and alignments.

Link: https://lore.kernel.org/r/1578313276-29080-8-git-send-email-liweihang@huawei.com
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Lang Cheng <chenglang@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 178 +++++++++++------------------
 drivers/infiniband/hw/hns/hns_roce_main.c  |  57 +++++----
 2 files changed, 92 insertions(+), 143 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index d53025b3ae97..00fd40fd8380 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -63,20 +63,15 @@ static void set_frmr_seg(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe,
 	struct hns_roce_mr *mr = to_hr_mr(wr->mr);
 
 	/* use ib_access_flags */
-	roce_set_bit(rc_sq_wqe->byte_4,
-		     V2_RC_FRMR_WQE_BYTE_4_BIND_EN_S,
+	roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_BIND_EN_S,
 		     wr->access & IB_ACCESS_MW_BIND ? 1 : 0);
-	roce_set_bit(rc_sq_wqe->byte_4,
-		     V2_RC_FRMR_WQE_BYTE_4_ATOMIC_S,
+	roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_ATOMIC_S,
 		     wr->access & IB_ACCESS_REMOTE_ATOMIC ? 1 : 0);
-	roce_set_bit(rc_sq_wqe->byte_4,
-		     V2_RC_FRMR_WQE_BYTE_4_RR_S,
+	roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_RR_S,
 		     wr->access & IB_ACCESS_REMOTE_READ ? 1 : 0);
-	roce_set_bit(rc_sq_wqe->byte_4,
-		     V2_RC_FRMR_WQE_BYTE_4_RW_S,
+	roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_RW_S,
 		     wr->access & IB_ACCESS_REMOTE_WRITE ? 1 : 0);
-	roce_set_bit(rc_sq_wqe->byte_4,
-		     V2_RC_FRMR_WQE_BYTE_4_LW_S,
+	roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_LW_S,
 		     wr->access & IB_ACCESS_LOCAL_WRITE ? 1 : 0);
 
 	/* Data structure reuse may lead to confusion */
@@ -1371,8 +1366,7 @@ static int hns_roce_query_pf_timer_resource(struct hns_roce_dev *hr_dev)
 	return 0;
 }
 
-static int hns_roce_set_vf_switch_param(struct hns_roce_dev *hr_dev,
-						  int vf_id)
+static int hns_roce_set_vf_switch_param(struct hns_roce_dev *hr_dev, int vf_id)
 {
 	struct hns_roce_cmq_desc desc;
 	struct hns_roce_vf_switch *swt;
@@ -1381,13 +1375,12 @@ static int hns_roce_set_vf_switch_param(struct hns_roce_dev *hr_dev,
 	swt = (struct hns_roce_vf_switch *)desc.data;
 	hns_roce_cmq_setup_basic_desc(&desc, HNS_SWITCH_PARAMETER_CFG, true);
 	swt->rocee_sel |= cpu_to_le32(HNS_ICL_SWITCH_CMD_ROCEE_SEL);
-	roce_set_field(swt->fun_id,
-			VF_SWITCH_DATA_FUN_ID_VF_ID_M,
-			VF_SWITCH_DATA_FUN_ID_VF_ID_S,
-			vf_id);
+	roce_set_field(swt->fun_id, VF_SWITCH_DATA_FUN_ID_VF_ID_M,
+		       VF_SWITCH_DATA_FUN_ID_VF_ID_S, vf_id);
 	ret = hns_roce_cmq_send(hr_dev, &desc, 1);
 	if (ret)
 		return ret;
+
 	desc.flag =
 		cpu_to_le16(HNS_ROCE_CMD_FLAG_NO_INTR | HNS_ROCE_CMD_FLAG_IN);
 	desc.flag &= cpu_to_le16(~HNS_ROCE_CMD_FLAG_WR);
@@ -1811,37 +1804,32 @@ static int hns_roce_config_link_table(struct hns_roce_dev *hr_dev,
 			req_a->base_addr_h =
 				cpu_to_le32(link_tbl->table.map >> 32);
 			roce_set_field(req_a->depth_pgsz_init_en,
-				       CFG_LLM_QUE_DEPTH_M,
-				       CFG_LLM_QUE_DEPTH_S,
+				       CFG_LLM_QUE_DEPTH_M, CFG_LLM_QUE_DEPTH_S,
 				       link_tbl->npages);
 			roce_set_field(req_a->depth_pgsz_init_en,
-				       CFG_LLM_QUE_PGSZ_M,
-				       CFG_LLM_QUE_PGSZ_S,
+				       CFG_LLM_QUE_PGSZ_M, CFG_LLM_QUE_PGSZ_S,
 				       link_tbl->pg_sz);
 			req_a->head_ba_l = cpu_to_le32(entry[0].blk_ba0);
 			req_a->head_ba_h_nxtptr =
 				cpu_to_le32(entry[0].blk_ba1_nxt_ptr);
-			roce_set_field(req_a->head_ptr,
-				       CFG_LLM_HEAD_PTR_M,
+			roce_set_field(req_a->head_ptr, CFG_LLM_HEAD_PTR_M,
 				       CFG_LLM_HEAD_PTR_S, 0);
 		} else {
 			req_b->tail_ba_l =
 				cpu_to_le32(entry[page_num - 1].blk_ba0);
-			roce_set_field(req_b->tail_ba_h,
-				       CFG_LLM_TAIL_BA_H_M,
+			roce_set_field(req_b->tail_ba_h, CFG_LLM_TAIL_BA_H_M,
 				       CFG_LLM_TAIL_BA_H_S,
 				       entry[page_num - 1].blk_ba1_nxt_ptr &
-				       HNS_ROCE_LINK_TABLE_BA1_M);
-			roce_set_field(req_b->tail_ptr,
-				       CFG_LLM_TAIL_PTR_M,
+					       HNS_ROCE_LINK_TABLE_BA1_M);
+			roce_set_field(req_b->tail_ptr, CFG_LLM_TAIL_PTR_M,
 				       CFG_LLM_TAIL_PTR_S,
 				       (entry[page_num - 2].blk_ba1_nxt_ptr &
-				       HNS_ROCE_LINK_TABLE_NXT_PTR_M) >>
-				       HNS_ROCE_LINK_TABLE_NXT_PTR_S);
+					HNS_ROCE_LINK_TABLE_NXT_PTR_M) >>
+					       HNS_ROCE_LINK_TABLE_NXT_PTR_S);
 		}
 	}
-	roce_set_field(req_a->depth_pgsz_init_en,
-		       CFG_LLM_INIT_EN_M, CFG_LLM_INIT_EN_S, 1);
+	roce_set_field(req_a->depth_pgsz_init_en, CFG_LLM_INIT_EN_M,
+		       CFG_LLM_INIT_EN_S, 1);
 
 	return hns_roce_cmq_send(hr_dev, desc, 2);
 }
@@ -2134,11 +2122,9 @@ static int hns_roce_config_sgid_table(struct hns_roce_dev *hr_dev,
 
 	hns_roce_cmq_setup_basic_desc(&desc, HNS_ROCE_OPC_CFG_SGID_TB, false);
 
-	roce_set_field(sgid_tb->table_idx_rsv,
-		       CFG_SGID_TB_TABLE_IDX_M,
+	roce_set_field(sgid_tb->table_idx_rsv, CFG_SGID_TB_TABLE_IDX_M,
 		       CFG_SGID_TB_TABLE_IDX_S, gid_index);
-	roce_set_field(sgid_tb->vf_sgid_type_rsv,
-		       CFG_SGID_TB_VF_SGID_TYPE_M,
+	roce_set_field(sgid_tb->vf_sgid_type_rsv, CFG_SGID_TB_VF_SGID_TYPE_M,
 		       CFG_SGID_TB_VF_SGID_TYPE_S, sgid_type);
 
 	p = (u32 *)&gid->raw[0];
@@ -2409,11 +2395,10 @@ static int hns_roce_v2_mw_write_mtpt(void *mb_buf, struct hns_roce_mw *mw)
 		       V2_MPT_BYTE_4_MPT_ST_S, V2_MPT_ST_FREE);
 	roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PD_M,
 		       V2_MPT_BYTE_4_PD_S, mw->pdn);
-	roce_set_field(mpt_entry->byte_4_pd_hop_st,
-		       V2_MPT_BYTE_4_PBL_HOP_NUM_M,
+	roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PBL_HOP_NUM_M,
 		       V2_MPT_BYTE_4_PBL_HOP_NUM_S,
-		       mw->pbl_hop_num == HNS_ROCE_HOP_NUM_0 ?
-		       0 : mw->pbl_hop_num);
+		       mw->pbl_hop_num == HNS_ROCE_HOP_NUM_0 ? 0 :
+							       mw->pbl_hop_num);
 	roce_set_field(mpt_entry->byte_4_pd_hop_st,
 		       V2_MPT_BYTE_4_PBL_BA_PG_SZ_M,
 		       V2_MPT_BYTE_4_PBL_BA_PG_SZ_S,
@@ -2554,8 +2539,7 @@ static void hns_roce_v2_write_cqc(struct hns_roce_dev *hr_dev,
 	roce_set_field(cq_context->byte_4_pg_ceqn, V2_CQC_BYTE_4_ARM_ST_M,
 		       V2_CQC_BYTE_4_ARM_ST_S, REG_NXT_CEQE);
 	roce_set_field(cq_context->byte_4_pg_ceqn, V2_CQC_BYTE_4_SHIFT_M,
-		       V2_CQC_BYTE_4_SHIFT_S,
-		       ilog2(hr_cq->cq_depth));
+		       V2_CQC_BYTE_4_SHIFT_S, ilog2(hr_cq->cq_depth));
 	roce_set_field(cq_context->byte_4_pg_ceqn, V2_CQC_BYTE_4_CEQN_M,
 		       V2_CQC_BYTE_4_CEQN_S, hr_cq->vector);
 
@@ -3829,13 +3813,11 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 	/* Configure GID index */
 	port_num = rdma_ah_get_port_num(&attr->ah_attr);
 	roce_set_field(context->byte_20_smac_sgid_idx,
-		       V2_QPC_BYTE_20_SGID_IDX_M,
-		       V2_QPC_BYTE_20_SGID_IDX_S,
+		       V2_QPC_BYTE_20_SGID_IDX_M, V2_QPC_BYTE_20_SGID_IDX_S,
 		       hns_get_gid_index(hr_dev, port_num - 1,
 					 grh->sgid_index));
 	roce_set_field(qpc_mask->byte_20_smac_sgid_idx,
-		       V2_QPC_BYTE_20_SGID_IDX_M,
-		       V2_QPC_BYTE_20_SGID_IDX_S, 0);
+		       V2_QPC_BYTE_20_SGID_IDX_M, V2_QPC_BYTE_20_SGID_IDX_S, 0);
 	memcpy(&(context->dmac), dmac, sizeof(u32));
 	roce_set_field(context->byte_52_udpspn_dmac, V2_QPC_BYTE_52_DMAC_M,
 		       V2_QPC_BYTE_52_DMAC_S, *((u16 *)(&dmac[4])));
@@ -4225,8 +4207,7 @@ static int hns_roce_v2_set_opt_fields(struct ib_qp *ibqp,
 
 		roce_set_field(context->byte_212_lsn,
 			       V2_QPC_BYTE_212_RETRY_CNT_M,
-			       V2_QPC_BYTE_212_RETRY_CNT_S,
-			       attr->retry_cnt);
+			       V2_QPC_BYTE_212_RETRY_CNT_S, attr->retry_cnt);
 		roce_set_field(qpc_mask->byte_212_lsn,
 			       V2_QPC_BYTE_212_RETRY_CNT_M,
 			       V2_QPC_BYTE_212_RETRY_CNT_S, 0);
@@ -5145,8 +5126,7 @@ static int hns_roce_v2_ceq_int(struct hns_roce_dev *hr_dev,
 		 */
 		dma_rmb();
 
-		cqn = roce_get_field(ceqe->comp,
-				     HNS_ROCE_V2_CEQE_COMP_CQN_M,
+		cqn = roce_get_field(ceqe->comp, HNS_ROCE_V2_CEQE_COMP_CQN_M,
 				     HNS_ROCE_V2_CEQE_COMP_CQN_S);
 
 		hns_roce_cq_completion(hr_dev, cqn);
@@ -5399,126 +5379,98 @@ static void hns_roce_config_eqc(struct hns_roce_dev *hr_dev,
 		eq->eqe_ba = eq->l0_dma;
 
 	/* set eqc state */
-	roce_set_field(eqc->byte_4,
-		       HNS_ROCE_EQC_EQ_ST_M,
-		       HNS_ROCE_EQC_EQ_ST_S,
+	roce_set_field(eqc->byte_4, HNS_ROCE_EQC_EQ_ST_M, HNS_ROCE_EQC_EQ_ST_S,
 		       HNS_ROCE_V2_EQ_STATE_VALID);
 
 	/* set eqe hop num */
-	roce_set_field(eqc->byte_4,
-		       HNS_ROCE_EQC_HOP_NUM_M,
+	roce_set_field(eqc->byte_4, HNS_ROCE_EQC_HOP_NUM_M,
 		       HNS_ROCE_EQC_HOP_NUM_S, eq->hop_num);
 
 	/* set eqc over_ignore */
-	roce_set_field(eqc->byte_4,
-		       HNS_ROCE_EQC_OVER_IGNORE_M,
+	roce_set_field(eqc->byte_4, HNS_ROCE_EQC_OVER_IGNORE_M,
 		       HNS_ROCE_EQC_OVER_IGNORE_S, eq->over_ignore);
 
 	/* set eqc coalesce */
-	roce_set_field(eqc->byte_4,
-		       HNS_ROCE_EQC_COALESCE_M,
+	roce_set_field(eqc->byte_4, HNS_ROCE_EQC_COALESCE_M,
 		       HNS_ROCE_EQC_COALESCE_S, eq->coalesce);
 
 	/* set eqc arm_state */
-	roce_set_field(eqc->byte_4,
-		       HNS_ROCE_EQC_ARM_ST_M,
+	roce_set_field(eqc->byte_4, HNS_ROCE_EQC_ARM_ST_M,
 		       HNS_ROCE_EQC_ARM_ST_S, eq->arm_st);
 
 	/* set eqn */
-	roce_set_field(eqc->byte_4,
-		       HNS_ROCE_EQC_EQN_M,
-		       HNS_ROCE_EQC_EQN_S, eq->eqn);
+	roce_set_field(eqc->byte_4, HNS_ROCE_EQC_EQN_M, HNS_ROCE_EQC_EQN_S,
+		       eq->eqn);
 
 	/* set eqe_cnt */
-	roce_set_field(eqc->byte_4,
-		       HNS_ROCE_EQC_EQE_CNT_M,
-		       HNS_ROCE_EQC_EQE_CNT_S,
-		       HNS_ROCE_EQ_INIT_EQE_CNT);
+	roce_set_field(eqc->byte_4, HNS_ROCE_EQC_EQE_CNT_M,
+		       HNS_ROCE_EQC_EQE_CNT_S, HNS_ROCE_EQ_INIT_EQE_CNT);
 
 	/* set eqe_ba_pg_sz */
-	roce_set_field(eqc->byte_8,
-		       HNS_ROCE_EQC_BA_PG_SZ_M,
+	roce_set_field(eqc->byte_8, HNS_ROCE_EQC_BA_PG_SZ_M,
 		       HNS_ROCE_EQC_BA_PG_SZ_S,
 		       eq->eqe_ba_pg_sz + PG_SHIFT_OFFSET);
 
 	/* set eqe_buf_pg_sz */
-	roce_set_field(eqc->byte_8,
-		       HNS_ROCE_EQC_BUF_PG_SZ_M,
+	roce_set_field(eqc->byte_8, HNS_ROCE_EQC_BUF_PG_SZ_M,
 		       HNS_ROCE_EQC_BUF_PG_SZ_S,
 		       eq->eqe_buf_pg_sz + PG_SHIFT_OFFSET);
 
 	/* set eq_producer_idx */
-	roce_set_field(eqc->byte_8,
-		       HNS_ROCE_EQC_PROD_INDX_M,
-		       HNS_ROCE_EQC_PROD_INDX_S,
-		       HNS_ROCE_EQ_INIT_PROD_IDX);
+	roce_set_field(eqc->byte_8, HNS_ROCE_EQC_PROD_INDX_M,
+		       HNS_ROCE_EQC_PROD_INDX_S, HNS_ROCE_EQ_INIT_PROD_IDX);
 
 	/* set eq_max_cnt */
-	roce_set_field(eqc->byte_12,
-		       HNS_ROCE_EQC_MAX_CNT_M,
+	roce_set_field(eqc->byte_12, HNS_ROCE_EQC_MAX_CNT_M,
 		       HNS_ROCE_EQC_MAX_CNT_S, eq->eq_max_cnt);
 
 	/* set eq_period */
-	roce_set_field(eqc->byte_12,
-		       HNS_ROCE_EQC_PERIOD_M,
+	roce_set_field(eqc->byte_12, HNS_ROCE_EQC_PERIOD_M,
 		       HNS_ROCE_EQC_PERIOD_S, eq->eq_period);
 
 	/* set eqe_report_timer */
-	roce_set_field(eqc->eqe_report_timer,
-		       HNS_ROCE_EQC_REPORT_TIMER_M,
+	roce_set_field(eqc->eqe_report_timer, HNS_ROCE_EQC_REPORT_TIMER_M,
 		       HNS_ROCE_EQC_REPORT_TIMER_S,
 		       HNS_ROCE_EQ_INIT_REPORT_TIMER);
 
 	/* set eqe_ba [34:3] */
-	roce_set_field(eqc->eqe_ba0,
-		       HNS_ROCE_EQC_EQE_BA_L_M,
+	roce_set_field(eqc->eqe_ba0, HNS_ROCE_EQC_EQE_BA_L_M,
 		       HNS_ROCE_EQC_EQE_BA_L_S, eq->eqe_ba >> 3);
 
 	/* set eqe_ba [64:35] */
-	roce_set_field(eqc->eqe_ba1,
-		       HNS_ROCE_EQC_EQE_BA_H_M,
+	roce_set_field(eqc->eqe_ba1, HNS_ROCE_EQC_EQE_BA_H_M,
 		       HNS_ROCE_EQC_EQE_BA_H_S, eq->eqe_ba >> 35);
 
 	/* set eq shift */
-	roce_set_field(eqc->byte_28,
-		       HNS_ROCE_EQC_SHIFT_M,
-		       HNS_ROCE_EQC_SHIFT_S, eq->shift);
+	roce_set_field(eqc->byte_28, HNS_ROCE_EQC_SHIFT_M, HNS_ROCE_EQC_SHIFT_S,
+		       eq->shift);
 
 	/* set eq MSI_IDX */
-	roce_set_field(eqc->byte_28,
-		       HNS_ROCE_EQC_MSI_INDX_M,
-		       HNS_ROCE_EQC_MSI_INDX_S,
-		       HNS_ROCE_EQ_INIT_MSI_IDX);
+	roce_set_field(eqc->byte_28, HNS_ROCE_EQC_MSI_INDX_M,
+		       HNS_ROCE_EQC_MSI_INDX_S, HNS_ROCE_EQ_INIT_MSI_IDX);
 
 	/* set cur_eqe_ba [27:12] */
-	roce_set_field(eqc->byte_28,
-		       HNS_ROCE_EQC_CUR_EQE_BA_L_M,
+	roce_set_field(eqc->byte_28, HNS_ROCE_EQC_CUR_EQE_BA_L_M,
 		       HNS_ROCE_EQC_CUR_EQE_BA_L_S, eq->cur_eqe_ba >> 12);
 
 	/* set cur_eqe_ba [59:28] */
-	roce_set_field(eqc->byte_32,
-		       HNS_ROCE_EQC_CUR_EQE_BA_M_M,
+	roce_set_field(eqc->byte_32, HNS_ROCE_EQC_CUR_EQE_BA_M_M,
 		       HNS_ROCE_EQC_CUR_EQE_BA_M_S, eq->cur_eqe_ba >> 28);
 
 	/* set cur_eqe_ba [63:60] */
-	roce_set_field(eqc->byte_36,
-		       HNS_ROCE_EQC_CUR_EQE_BA_H_M,
+	roce_set_field(eqc->byte_36, HNS_ROCE_EQC_CUR_EQE_BA_H_M,
 		       HNS_ROCE_EQC_CUR_EQE_BA_H_S, eq->cur_eqe_ba >> 60);
 
 	/* set eq consumer idx */
-	roce_set_field(eqc->byte_36,
-		       HNS_ROCE_EQC_CONS_INDX_M,
-		       HNS_ROCE_EQC_CONS_INDX_S,
-		       HNS_ROCE_EQ_INIT_CONS_IDX);
+	roce_set_field(eqc->byte_36, HNS_ROCE_EQC_CONS_INDX_M,
+		       HNS_ROCE_EQC_CONS_INDX_S, HNS_ROCE_EQ_INIT_CONS_IDX);
 
 	/* set nex_eqe_ba[43:12] */
-	roce_set_field(eqc->nxt_eqe_ba0,
-		       HNS_ROCE_EQC_NXT_EQE_BA_L_M,
+	roce_set_field(eqc->nxt_eqe_ba0, HNS_ROCE_EQC_NXT_EQE_BA_L_M,
 		       HNS_ROCE_EQC_NXT_EQE_BA_L_S, eq->nxt_eqe_ba >> 12);
 
 	/* set nex_eqe_ba[63:44] */
-	roce_set_field(eqc->nxt_eqe_ba1,
-		       HNS_ROCE_EQC_NXT_EQE_BA_H_M,
+	roce_set_field(eqc->nxt_eqe_ba1, HNS_ROCE_EQC_NXT_EQE_BA_H_M,
 		       HNS_ROCE_EQC_NXT_EQE_BA_H_S, eq->nxt_eqe_ba >> 44);
 }
 
@@ -5818,18 +5770,16 @@ static int __hns_roce_request_irq(struct hns_roce_dev *hr_dev, int irq_num,
 
 	/* irq contains: abnormal + AEQ + CEQ */
 	for (j = 0; j < other_num; j++)
-		snprintf((char *)hr_dev->irq_names[j],
-			 HNS_ROCE_INT_NAME_LEN, "hns-abn-%d", j);
+		snprintf((char *)hr_dev->irq_names[j], HNS_ROCE_INT_NAME_LEN,
+			 "hns-abn-%d", j);
 
 	for (j = other_num; j < (other_num + aeq_num); j++)
-		snprintf((char *)hr_dev->irq_names[j],
-			 HNS_ROCE_INT_NAME_LEN, "hns-aeq-%d",
-			 j - other_num);
+		snprintf((char *)hr_dev->irq_names[j], HNS_ROCE_INT_NAME_LEN,
+			 "hns-aeq-%d", j - other_num);
 
 	for (j = (other_num + aeq_num); j < irq_num; j++)
-		snprintf((char *)hr_dev->irq_names[j],
-			 HNS_ROCE_INT_NAME_LEN, "hns-ceq-%d",
-			 j - other_num - aeq_num);
+		snprintf((char *)hr_dev->irq_names[j], HNS_ROCE_INT_NAME_LEN,
+			 "hns-ceq-%d", j - other_num - aeq_num);
 
 	for (j = 0; j < irq_num; j++) {
 		if (j < other_num)
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 854ef6e74788..84e4707337a9 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -90,7 +90,7 @@ static int hns_roce_add_gid(const struct ib_gid_attr *attr, void **context)
 static int hns_roce_del_gid(const struct ib_gid_attr *attr, void **context)
 {
 	struct hns_roce_dev *hr_dev = to_hr_dev(attr->device);
-	struct ib_gid_attr zattr = { };
+	struct ib_gid_attr zattr = {};
 	u8 port = attr->port_num - 1;
 	int ret;
 
@@ -259,11 +259,12 @@ static int hns_roce_query_port(struct ib_device *ib_dev, u8 port_num,
 
 	mtu = iboe_get_mtu(net_dev->mtu);
 	props->active_mtu = mtu ? min(props->max_mtu, mtu) : IB_MTU_256;
-	props->state = (netif_running(net_dev) && netif_carrier_ok(net_dev)) ?
-			IB_PORT_ACTIVE : IB_PORT_DOWN;
-	props->phys_state = (props->state == IB_PORT_ACTIVE) ?
-			     IB_PORT_PHYS_STATE_LINK_UP :
-			     IB_PORT_PHYS_STATE_DISABLED;
+	props->state = netif_running(net_dev) && netif_carrier_ok(net_dev) ?
+			       IB_PORT_ACTIVE :
+			       IB_PORT_DOWN;
+	props->phys_state = props->state == IB_PORT_ACTIVE ?
+				    IB_PORT_PHYS_STATE_LINK_UP :
+				    IB_PORT_PHYS_STATE_DISABLED;
 
 	spin_unlock_irqrestore(&hr_dev->iboe.lock, flags);
 
@@ -481,13 +482,13 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev)
 
 	ib_dev = &hr_dev->ib_dev;
 
-	ib_dev->node_type		= RDMA_NODE_IB_CA;
-	ib_dev->dev.parent		= dev;
+	ib_dev->node_type = RDMA_NODE_IB_CA;
+	ib_dev->dev.parent = dev;
 
-	ib_dev->phys_port_cnt		= hr_dev->caps.num_ports;
-	ib_dev->local_dma_lkey		= hr_dev->caps.reserved_lkey;
-	ib_dev->num_comp_vectors	= hr_dev->caps.num_comp_vectors;
-	ib_dev->uverbs_cmd_mask		=
+	ib_dev->phys_port_cnt = hr_dev->caps.num_ports;
+	ib_dev->local_dma_lkey = hr_dev->caps.reserved_lkey;
+	ib_dev->num_comp_vectors = hr_dev->caps.num_comp_vectors;
+	ib_dev->uverbs_cmd_mask =
 		(1ULL << IB_USER_VERBS_CMD_GET_CONTEXT) |
 		(1ULL << IB_USER_VERBS_CMD_QUERY_DEVICE) |
 		(1ULL << IB_USER_VERBS_CMD_QUERY_PORT) |
@@ -503,8 +504,7 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev)
 		(1ULL << IB_USER_VERBS_CMD_QUERY_QP) |
 		(1ULL << IB_USER_VERBS_CMD_DESTROY_QP);
 
-	ib_dev->uverbs_ex_cmd_mask |=
-		(1ULL << IB_USER_VERBS_EX_CMD_MODIFY_CQ);
+	ib_dev->uverbs_ex_cmd_mask |= (1ULL << IB_USER_VERBS_EX_CMD_MODIFY_CQ);
 
 	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_REREG_MR) {
 		ib_dev->uverbs_cmd_mask |= (1ULL << IB_USER_VERBS_CMD_REREG_MR);
@@ -589,11 +589,13 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev)
 
 	if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) {
 		ret = hns_roce_init_hem_table(hr_dev,
-				      &hr_dev->mr_table.mtt_cqe_table,
-				      HEM_TYPE_CQE, hr_dev->caps.mtt_entry_sz,
-				      hr_dev->caps.num_cqe_segs, 1);
+					      &hr_dev->mr_table.mtt_cqe_table,
+					      HEM_TYPE_CQE,
+					      hr_dev->caps.mtt_entry_sz,
+					      hr_dev->caps.num_cqe_segs, 1);
 		if (ret) {
-			dev_err(dev, "Failed to init MTT CQE context memory, aborting.\n");
+			dev_err(dev,
+				"Failed to init CQE context memory, aborting.\n");
 			goto err_unmap_cqe;
 		}
 	}
@@ -633,7 +635,7 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev)
 					      hr_dev->caps.num_qps, 1);
 		if (ret) {
 			dev_err(dev,
-			       "Failed to init trrl_table memory, aborting.\n");
+				"Failed to init trrl_table memory, aborting.\n");
 			goto err_unmap_irrl;
 		}
 	}
@@ -653,7 +655,7 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev)
 					      hr_dev->caps.num_srqs, 1);
 		if (ret) {
 			dev_err(dev,
-			      "Failed to init SRQ context memory, aborting.\n");
+				"Failed to init SRQ context memory, aborting.\n");
 			goto err_unmap_cq;
 		}
 	}
@@ -692,33 +694,31 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev)
 					      hr_dev->caps.num_qps, 1);
 		if (ret) {
 			dev_err(dev,
-			      "Failed to init SCC context memory, aborting.\n");
+				"Failed to init SCC context memory, aborting.\n");
 			goto err_unmap_idx;
 		}
 	}
 
 	if (hr_dev->caps.qpc_timer_entry_sz) {
-		ret = hns_roce_init_hem_table(hr_dev,
-					      &hr_dev->qpc_timer_table,
+		ret = hns_roce_init_hem_table(hr_dev, &hr_dev->qpc_timer_table,
 					      HEM_TYPE_QPC_TIMER,
 					      hr_dev->caps.qpc_timer_entry_sz,
 					      hr_dev->caps.num_qpc_timer, 1);
 		if (ret) {
 			dev_err(dev,
-			      "Failed to init QPC timer memory, aborting.\n");
+				"Failed to init QPC timer memory, aborting.\n");
 			goto err_unmap_ctx;
 		}
 	}
 
 	if (hr_dev->caps.cqc_timer_entry_sz) {
-		ret = hns_roce_init_hem_table(hr_dev,
-					      &hr_dev->cqc_timer_table,
+		ret = hns_roce_init_hem_table(hr_dev, &hr_dev->cqc_timer_table,
 					      HEM_TYPE_CQC_TIMER,
 					      hr_dev->caps.cqc_timer_entry_sz,
 					      hr_dev->caps.num_cqc_timer, 1);
 		if (ret) {
 			dev_err(dev,
-			      "Failed to init CQC timer memory, aborting.\n");
+				"Failed to init CQC timer memory, aborting.\n");
 			goto err_unmap_qpc_timer;
 		}
 	}
@@ -727,8 +727,7 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev)
 
 err_unmap_qpc_timer:
 	if (hr_dev->caps.qpc_timer_entry_sz)
-		hns_roce_cleanup_hem_table(hr_dev,
-					   &hr_dev->qpc_timer_table);
+		hns_roce_cleanup_hem_table(hr_dev, &hr_dev->qpc_timer_table);
 
 err_unmap_ctx:
 	if (hr_dev->caps.sccc_entry_sz)
-- 
cgit v1.2.3-59-g8ed1b


From 4cca96a8d9da0ed8217cfdf2aec0c3c8b88e8911 Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Thu, 12 Dec 2019 13:30:21 +0200
Subject: IB/mlx5: Do reverse sequence during device removal

When IB device profile initialization completes, device is marked as
active.

However, IB device is not marked inactive, during device removal flow. It
should be the mirror of the add flow.

Hence, mark it inactive during remove sequence.

Link: https://lore.kernel.org/r/20191212113024.336702-2-leon@kernel.org
Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 997cbfe4b90c..05b557ddd0d8 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6725,6 +6725,8 @@ void __mlx5_ib_remove(struct mlx5_ib_dev *dev,
 		      const struct mlx5_ib_profile *profile,
 		      int stage)
 {
+	dev->ib_active = false;
+
 	/* Number of stages to cleanup */
 	while (stage) {
 		stage--;
-- 
cgit v1.2.3-59-g8ed1b


From 6b57cea9221b0247ad5111b348522625e489a8e4 Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Thu, 12 Dec 2019 13:30:22 +0200
Subject: IB/core: Let IB core distribute cache update events

Currently when the low level driver notifies Pkey, GID, and port change
events they are notified to the registered handlers in the order they are
registered.

IB core and other ULPs such as IPoIB are interested in GID, LID, Pkey
change events.

Since all GID queries done by ULPs are serviced by IB core, and the IB
core deferes cache updates to a work queue, it is possible for other
clients to see stale cache data when they handle their own events.

For example, the below call tree shows how ipoib will call
rdma_query_gid() concurrently with the update to the cache sitting in the
WQ.

mlx5_ib_handle_event()
  ib_dispatch_event()
    ib_cache_event()
       queue_work() -> slow cache update

    [..]
    ipoib_event()
     queue_work()
       [..]
       work handler
         ipoib_ib_dev_flush_light()
           __ipoib_ib_dev_flush()
              ipoib_dev_addr_changed_valid()
                rdma_query_gid() <- Returns old GID, cache not updated.

Move all the event dispatch to a work queue so that the cache update is
always done before any clients are notified.

Fixes: f35faa4ba956 ("IB/core: Simplify ib_query_gid to always refer to cache")
Link: https://lore.kernel.org/r/20191212113024.336702-3-leon@kernel.org
Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cache.c     | 121 ++++++++++++++++++++++--------------
 drivers/infiniband/core/core_priv.h |   1 +
 drivers/infiniband/core/device.c    |  33 +++-------
 include/rdma/ib_verbs.h             |   9 ++-
 4 files changed, 92 insertions(+), 72 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c
index d535995711c3..e55f345799e4 100644
--- a/drivers/infiniband/core/cache.c
+++ b/drivers/infiniband/core/cache.c
@@ -51,9 +51,8 @@ struct ib_pkey_cache {
 
 struct ib_update_work {
 	struct work_struct work;
-	struct ib_device  *device;
-	u8                 port_num;
-	bool		   enforce_security;
+	struct ib_event event;
+	bool enforce_security;
 };
 
 union ib_gid zgid;
@@ -130,7 +129,7 @@ static void dispatch_gid_change_event(struct ib_device *ib_dev, u8 port)
 	event.element.port_num	= port;
 	event.event		= IB_EVENT_GID_CHANGE;
 
-	ib_dispatch_event(&event);
+	ib_dispatch_event_clients(&event);
 }
 
 static const char * const gid_type_str[] = {
@@ -1381,9 +1380,8 @@ err:
 	return ret;
 }
 
-static void ib_cache_update(struct ib_device *device,
-			    u8                port,
-			    bool	      enforce_security)
+static int
+ib_cache_update(struct ib_device *device, u8 port, bool enforce_security)
 {
 	struct ib_port_attr       *tprops = NULL;
 	struct ib_pkey_cache      *pkey_cache = NULL, *old_pkey_cache;
@@ -1391,11 +1389,11 @@ static void ib_cache_update(struct ib_device *device,
 	int                        ret;
 
 	if (!rdma_is_port_valid(device, port))
-		return;
+		return -EINVAL;
 
 	tprops = kmalloc(sizeof *tprops, GFP_KERNEL);
 	if (!tprops)
-		return;
+		return -ENOMEM;
 
 	ret = ib_query_port(device, port, tprops);
 	if (ret) {
@@ -1413,8 +1411,10 @@ static void ib_cache_update(struct ib_device *device,
 	pkey_cache = kmalloc(struct_size(pkey_cache, table,
 					 tprops->pkey_tbl_len),
 			     GFP_KERNEL);
-	if (!pkey_cache)
+	if (!pkey_cache) {
+		ret = -ENOMEM;
 		goto err;
+	}
 
 	pkey_cache->table_len = tprops->pkey_tbl_len;
 
@@ -1446,50 +1446,84 @@ static void ib_cache_update(struct ib_device *device,
 
 	kfree(old_pkey_cache);
 	kfree(tprops);
-	return;
+	return 0;
 
 err:
 	kfree(pkey_cache);
 	kfree(tprops);
+	return ret;
+}
+
+static void ib_cache_event_task(struct work_struct *_work)
+{
+	struct ib_update_work *work =
+		container_of(_work, struct ib_update_work, work);
+	int ret;
+
+	/* Before distributing the cache update event, first sync
+	 * the cache.
+	 */
+	ret = ib_cache_update(work->event.device, work->event.element.port_num,
+			      work->enforce_security);
+
+	/* GID event is notified already for individual GID entries by
+	 * dispatch_gid_change_event(). Hence, notifiy for rest of the
+	 * events.
+	 */
+	if (!ret && work->event.event != IB_EVENT_GID_CHANGE)
+		ib_dispatch_event_clients(&work->event);
+
+	kfree(work);
 }
 
-static void ib_cache_task(struct work_struct *_work)
+static void ib_generic_event_task(struct work_struct *_work)
 {
 	struct ib_update_work *work =
 		container_of(_work, struct ib_update_work, work);
 
-	ib_cache_update(work->device,
-			work->port_num,
-			work->enforce_security);
+	ib_dispatch_event_clients(&work->event);
 	kfree(work);
 }
 
-static void ib_cache_event(struct ib_event_handler *handler,
-			   struct ib_event *event)
+static bool is_cache_update_event(const struct ib_event *event)
+{
+	return (event->event == IB_EVENT_PORT_ERR    ||
+		event->event == IB_EVENT_PORT_ACTIVE ||
+		event->event == IB_EVENT_LID_CHANGE  ||
+		event->event == IB_EVENT_PKEY_CHANGE ||
+		event->event == IB_EVENT_CLIENT_REREGISTER ||
+		event->event == IB_EVENT_GID_CHANGE);
+}
+
+/**
+ * ib_dispatch_event - Dispatch an asynchronous event
+ * @event:Event to dispatch
+ *
+ * Low-level drivers must call ib_dispatch_event() to dispatch the
+ * event to all registered event handlers when an asynchronous event
+ * occurs.
+ */
+void ib_dispatch_event(const struct ib_event *event)
 {
 	struct ib_update_work *work;
 
-	if (event->event == IB_EVENT_PORT_ERR    ||
-	    event->event == IB_EVENT_PORT_ACTIVE ||
-	    event->event == IB_EVENT_LID_CHANGE  ||
-	    event->event == IB_EVENT_PKEY_CHANGE ||
-	    event->event == IB_EVENT_CLIENT_REREGISTER ||
-	    event->event == IB_EVENT_GID_CHANGE) {
-		work = kmalloc(sizeof *work, GFP_ATOMIC);
-		if (work) {
-			INIT_WORK(&work->work, ib_cache_task);
-			work->device   = event->device;
-			work->port_num = event->element.port_num;
-			if (event->event == IB_EVENT_PKEY_CHANGE ||
-			    event->event == IB_EVENT_GID_CHANGE)
-				work->enforce_security = true;
-			else
-				work->enforce_security = false;
-
-			queue_work(ib_wq, &work->work);
-		}
-	}
+	work = kzalloc(sizeof(*work), GFP_ATOMIC);
+	if (!work)
+		return;
+
+	if (is_cache_update_event(event))
+		INIT_WORK(&work->work, ib_cache_event_task);
+	else
+		INIT_WORK(&work->work, ib_generic_event_task);
+
+	work->event = *event;
+	if (event->event == IB_EVENT_PKEY_CHANGE ||
+	    event->event == IB_EVENT_GID_CHANGE)
+		work->enforce_security = true;
+
+	queue_work(ib_wq, &work->work);
 }
+EXPORT_SYMBOL(ib_dispatch_event);
 
 int ib_cache_setup_one(struct ib_device *device)
 {
@@ -1505,9 +1539,6 @@ int ib_cache_setup_one(struct ib_device *device)
 	rdma_for_each_port (device, p)
 		ib_cache_update(device, p, true);
 
-	INIT_IB_EVENT_HANDLER(&device->cache.event_handler,
-			      device, ib_cache_event);
-	ib_register_event_handler(&device->cache.event_handler);
 	return 0;
 }
 
@@ -1529,14 +1560,12 @@ void ib_cache_release_one(struct ib_device *device)
 
 void ib_cache_cleanup_one(struct ib_device *device)
 {
-	/* The cleanup function unregisters the event handler,
-	 * waits for all in-progress workqueue elements and cleans
-	 * up the GID cache. This function should be called after
-	 * the device was removed from the devices list and all
-	 * clients were removed, so the cache exists but is
+	/* The cleanup function waits for all in-progress workqueue
+	 * elements and cleans up the GID cache. This function should be
+	 * called after the device was removed from the devices list and
+	 * all clients were removed, so the cache exists but is
 	 * non-functional and shouldn't be updated anymore.
 	 */
-	ib_unregister_event_handler(&device->cache.event_handler);
 	flush_workqueue(ib_wq);
 	gid_table_cleanup_one(device);
 
diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h
index 3645e092e1c7..d657d90e618b 100644
--- a/drivers/infiniband/core/core_priv.h
+++ b/drivers/infiniband/core/core_priv.h
@@ -149,6 +149,7 @@ unsigned long roce_gid_type_mask_support(struct ib_device *ib_dev, u8 port);
 int ib_cache_setup_one(struct ib_device *device);
 void ib_cache_cleanup_one(struct ib_device *device);
 void ib_cache_release_one(struct ib_device *device);
+void ib_dispatch_event_clients(struct ib_event *event);
 
 #ifdef CONFIG_CGROUP_RDMA
 void ib_device_register_rdmacg(struct ib_device *device);
diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 84dd74fe13b8..c38b2b0b078a 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -588,6 +588,7 @@ struct ib_device *_ib_alloc_device(size_t size)
 
 	INIT_LIST_HEAD(&device->event_handler_list);
 	spin_lock_init(&device->event_handler_lock);
+	init_rwsem(&device->event_handler_rwsem);
 	mutex_init(&device->unregistration_lock);
 	/*
 	 * client_data needs to be alloc because we don't want our mark to be
@@ -1931,17 +1932,15 @@ EXPORT_SYMBOL(ib_set_client_data);
  *
  * ib_register_event_handler() registers an event handler that will be
  * called back when asynchronous IB events occur (as defined in
- * chapter 11 of the InfiniBand Architecture Specification).  This
- * callback may occur in interrupt context.
+ * chapter 11 of the InfiniBand Architecture Specification). This
+ * callback occurs in workqueue context.
  */
 void ib_register_event_handler(struct ib_event_handler *event_handler)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
+	down_write(&event_handler->device->event_handler_rwsem);
 	list_add_tail(&event_handler->list,
 		      &event_handler->device->event_handler_list);
-	spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
+	up_write(&event_handler->device->event_handler_rwsem);
 }
 EXPORT_SYMBOL(ib_register_event_handler);
 
@@ -1954,35 +1953,23 @@ EXPORT_SYMBOL(ib_register_event_handler);
  */
 void ib_unregister_event_handler(struct ib_event_handler *event_handler)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(&event_handler->device->event_handler_lock, flags);
+	down_write(&event_handler->device->event_handler_rwsem);
 	list_del(&event_handler->list);
-	spin_unlock_irqrestore(&event_handler->device->event_handler_lock, flags);
+	up_write(&event_handler->device->event_handler_rwsem);
 }
 EXPORT_SYMBOL(ib_unregister_event_handler);
 
-/**
- * ib_dispatch_event - Dispatch an asynchronous event
- * @event:Event to dispatch
- *
- * Low-level drivers must call ib_dispatch_event() to dispatch the
- * event to all registered event handlers when an asynchronous event
- * occurs.
- */
-void ib_dispatch_event(struct ib_event *event)
+void ib_dispatch_event_clients(struct ib_event *event)
 {
-	unsigned long flags;
 	struct ib_event_handler *handler;
 
-	spin_lock_irqsave(&event->device->event_handler_lock, flags);
+	down_read(&event->device->event_handler_rwsem);
 
 	list_for_each_entry(handler, &event->device->event_handler_list, list)
 		handler->handler(handler, event);
 
-	spin_unlock_irqrestore(&event->device->event_handler_lock, flags);
+	up_read(&event->device->event_handler_rwsem);
 }
-EXPORT_SYMBOL(ib_dispatch_event);
 
 static int iw_query_port(struct ib_device *device,
 			   u8 port_num,
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 42f28d39f28c..f36fb657518f 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2154,7 +2154,6 @@ struct ib_port_cache {
 
 struct ib_cache {
 	rwlock_t                lock;
-	struct ib_event_handler event_handler;
 };
 
 struct ib_port_immutable {
@@ -2632,7 +2631,11 @@ struct ib_device {
 	struct rcu_head rcu_head;
 
 	struct list_head              event_handler_list;
-	spinlock_t                    event_handler_lock;
+	/* Protects event_handler_list */
+	struct rw_semaphore event_handler_rwsem;
+
+	/* Protects QP's event_handler calls and open_qp list */
+	spinlock_t event_handler_lock;
 
 	struct rw_semaphore	      client_data_rwsem;
 	struct xarray                 client_data;
@@ -2947,7 +2950,7 @@ bool ib_modify_qp_is_ok(enum ib_qp_state cur_state, enum ib_qp_state next_state,
 
 void ib_register_event_handler(struct ib_event_handler *event_handler);
 void ib_unregister_event_handler(struct ib_event_handler *event_handler);
-void ib_dispatch_event(struct ib_event *event);
+void ib_dispatch_event(const struct ib_event *event);
 
 int ib_query_port(struct ib_device *device,
 		  u8 port_num, struct ib_port_attr *port_attr);
-- 
cgit v1.2.3-59-g8ed1b


From 17e1064632512db419cb9bb4555aec1763969b7d Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Thu, 12 Dec 2019 13:30:23 +0200
Subject: IB/core: Cut down single member ib_cache structure

Given that ib_cache structure has only single member now, merge the cache
lock directly in the ib_device.

Link: https://lore.kernel.org/r/20191212113024.336702-4-leon@kernel.org
Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cache.c | 30 +++++++++++++++---------------
 include/rdma/ib_verbs.h         |  7 ++-----
 2 files changed, 17 insertions(+), 20 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c
index e55f345799e4..17bfedd24cc3 100644
--- a/drivers/infiniband/core/cache.c
+++ b/drivers/infiniband/core/cache.c
@@ -1033,7 +1033,7 @@ int ib_get_cached_pkey(struct ib_device *device,
 	if (!rdma_is_port_valid(device, port_num))
 		return -EINVAL;
 
-	read_lock_irqsave(&device->cache.lock, flags);
+	read_lock_irqsave(&device->cache_lock, flags);
 
 	cache = device->port_data[port_num].cache.pkey;
 
@@ -1042,7 +1042,7 @@ int ib_get_cached_pkey(struct ib_device *device,
 	else
 		*pkey = cache->table[index];
 
-	read_unlock_irqrestore(&device->cache.lock, flags);
+	read_unlock_irqrestore(&device->cache_lock, flags);
 
 	return ret;
 }
@@ -1057,9 +1057,9 @@ int ib_get_cached_subnet_prefix(struct ib_device *device,
 	if (!rdma_is_port_valid(device, port_num))
 		return -EINVAL;
 
-	read_lock_irqsave(&device->cache.lock, flags);
+	read_lock_irqsave(&device->cache_lock, flags);
 	*sn_pfx = device->port_data[port_num].cache.subnet_prefix;
-	read_unlock_irqrestore(&device->cache.lock, flags);
+	read_unlock_irqrestore(&device->cache_lock, flags);
 
 	return 0;
 }
@@ -1079,7 +1079,7 @@ int ib_find_cached_pkey(struct ib_device *device,
 	if (!rdma_is_port_valid(device, port_num))
 		return -EINVAL;
 
-	read_lock_irqsave(&device->cache.lock, flags);
+	read_lock_irqsave(&device->cache_lock, flags);
 
 	cache = device->port_data[port_num].cache.pkey;
 
@@ -1100,7 +1100,7 @@ int ib_find_cached_pkey(struct ib_device *device,
 		ret = 0;
 	}
 
-	read_unlock_irqrestore(&device->cache.lock, flags);
+	read_unlock_irqrestore(&device->cache_lock, flags);
 
 	return ret;
 }
@@ -1119,7 +1119,7 @@ int ib_find_exact_cached_pkey(struct ib_device *device,
 	if (!rdma_is_port_valid(device, port_num))
 		return -EINVAL;
 
-	read_lock_irqsave(&device->cache.lock, flags);
+	read_lock_irqsave(&device->cache_lock, flags);
 
 	cache = device->port_data[port_num].cache.pkey;
 
@@ -1132,7 +1132,7 @@ int ib_find_exact_cached_pkey(struct ib_device *device,
 			break;
 		}
 
-	read_unlock_irqrestore(&device->cache.lock, flags);
+	read_unlock_irqrestore(&device->cache_lock, flags);
 
 	return ret;
 }
@@ -1148,9 +1148,9 @@ int ib_get_cached_lmc(struct ib_device *device,
 	if (!rdma_is_port_valid(device, port_num))
 		return -EINVAL;
 
-	read_lock_irqsave(&device->cache.lock, flags);
+	read_lock_irqsave(&device->cache_lock, flags);
 	*lmc = device->port_data[port_num].cache.lmc;
-	read_unlock_irqrestore(&device->cache.lock, flags);
+	read_unlock_irqrestore(&device->cache_lock, flags);
 
 	return ret;
 }
@@ -1166,9 +1166,9 @@ int ib_get_cached_port_state(struct ib_device   *device,
 	if (!rdma_is_port_valid(device, port_num))
 		return -EINVAL;
 
-	read_lock_irqsave(&device->cache.lock, flags);
+	read_lock_irqsave(&device->cache_lock, flags);
 	*port_state = device->port_data[port_num].cache.port_state;
-	read_unlock_irqrestore(&device->cache.lock, flags);
+	read_unlock_irqrestore(&device->cache_lock, flags);
 
 	return ret;
 }
@@ -1428,7 +1428,7 @@ ib_cache_update(struct ib_device *device, u8 port, bool enforce_security)
 		}
 	}
 
-	write_lock_irq(&device->cache.lock);
+	write_lock_irq(&device->cache_lock);
 
 	old_pkey_cache = device->port_data[port].cache.pkey;
 
@@ -1437,7 +1437,7 @@ ib_cache_update(struct ib_device *device, u8 port, bool enforce_security)
 	device->port_data[port].cache.port_state = tprops->state;
 
 	device->port_data[port].cache.subnet_prefix = tprops->subnet_prefix;
-	write_unlock_irq(&device->cache.lock);
+	write_unlock_irq(&device->cache_lock);
 
 	if (enforce_security)
 		ib_security_cache_change(device,
@@ -1530,7 +1530,7 @@ int ib_cache_setup_one(struct ib_device *device)
 	unsigned int p;
 	int err;
 
-	rwlock_init(&device->cache.lock);
+	rwlock_init(&device->cache_lock);
 
 	err = gid_table_setup_one(device);
 	if (err)
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index f36fb657518f..37dac147a946 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2152,10 +2152,6 @@ struct ib_port_cache {
 	enum ib_port_state     port_state;
 };
 
-struct ib_cache {
-	rwlock_t                lock;
-};
-
 struct ib_port_immutable {
 	int                           pkey_tbl_len;
 	int                           gid_tbl_len;
@@ -2641,7 +2637,8 @@ struct ib_device {
 	struct xarray                 client_data;
 	struct mutex                  unregistration_lock;
 
-	struct ib_cache               cache;
+	/* Synchronize GID, Pkey cache entries, subnet prefix, LMC */
+	rwlock_t cache_lock;
 	/**
 	 * port_data is indexed by port number
 	 */
-- 
cgit v1.2.3-59-g8ed1b


From 40adf686128856c4add948fb002d43e2c507d1aa Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Thu, 12 Dec 2019 13:30:24 +0200
Subject: IB/core: Rename event_handler_lock to qp_open_list_lock

This lock is used to protect the qp->open_list linked list. As a side
effect it seems to also globally serialize the qp event_handler, but it
isn't clear if that is a deliberate design.

Link: https://lore.kernel.org/r/20191212113024.336702-5-leon@kernel.org
Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/device.c |  2 +-
 drivers/infiniband/core/verbs.c  | 12 ++++++------
 include/rdma/ib_verbs.h          |  2 +-
 3 files changed, 8 insertions(+), 8 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index c38b2b0b078a..e4c3ea727eac 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -587,7 +587,7 @@ struct ib_device *_ib_alloc_device(size_t size)
 	rdma_init_coredev(&device->coredev, device, &init_net);
 
 	INIT_LIST_HEAD(&device->event_handler_list);
-	spin_lock_init(&device->event_handler_lock);
+	spin_lock_init(&device->qp_open_list_lock);
 	init_rwsem(&device->event_handler_rwsem);
 	mutex_init(&device->unregistration_lock);
 	/*
diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index 47d54c31eb2a..78b27aff2846 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -1056,11 +1056,11 @@ static void __ib_shared_qp_event_handler(struct ib_event *event, void *context)
 	struct ib_qp *qp = context;
 	unsigned long flags;
 
-	spin_lock_irqsave(&qp->device->event_handler_lock, flags);
+	spin_lock_irqsave(&qp->device->qp_open_list_lock, flags);
 	list_for_each_entry(event->element.qp, &qp->open_list, open_list)
 		if (event->element.qp->event_handler)
 			event->element.qp->event_handler(event, event->element.qp->qp_context);
-	spin_unlock_irqrestore(&qp->device->event_handler_lock, flags);
+	spin_unlock_irqrestore(&qp->device->qp_open_list_lock, flags);
 }
 
 static void __ib_insert_xrcd_qp(struct ib_xrcd *xrcd, struct ib_qp *qp)
@@ -1097,9 +1097,9 @@ static struct ib_qp *__ib_open_qp(struct ib_qp *real_qp,
 	qp->qp_num = real_qp->qp_num;
 	qp->qp_type = real_qp->qp_type;
 
-	spin_lock_irqsave(&real_qp->device->event_handler_lock, flags);
+	spin_lock_irqsave(&real_qp->device->qp_open_list_lock, flags);
 	list_add(&qp->open_list, &real_qp->open_list);
-	spin_unlock_irqrestore(&real_qp->device->event_handler_lock, flags);
+	spin_unlock_irqrestore(&real_qp->device->qp_open_list_lock, flags);
 
 	return qp;
 }
@@ -1827,9 +1827,9 @@ int ib_close_qp(struct ib_qp *qp)
 	if (real_qp == qp)
 		return -EINVAL;
 
-	spin_lock_irqsave(&real_qp->device->event_handler_lock, flags);
+	spin_lock_irqsave(&real_qp->device->qp_open_list_lock, flags);
 	list_del(&qp->open_list);
-	spin_unlock_irqrestore(&real_qp->device->event_handler_lock, flags);
+	spin_unlock_irqrestore(&real_qp->device->qp_open_list_lock, flags);
 
 	atomic_dec(&real_qp->usecnt);
 	if (qp->qp_sec)
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 37dac147a946..cea4e198701e 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -2631,7 +2631,7 @@ struct ib_device {
 	struct rw_semaphore event_handler_rwsem;
 
 	/* Protects QP's event_handler calls and open_qp list */
-	spinlock_t event_handler_lock;
+	spinlock_t qp_open_list_lock;
 
 	struct rw_semaphore	      client_data_rwsem;
 	struct xarray                 client_data;
-- 
cgit v1.2.3-59-g8ed1b


From 74f75cda754eb69a77f910ceb5bc85f8e9ba56a5 Mon Sep 17 00:00:00 2001
From: Arnd Bergmann <arnd@arndb.de>
Date: Thu, 9 Jan 2020 09:47:30 +0100
Subject: IB/core: Fix build failure without hugepages

HPAGE_SHIFT is only defined on architectures that support hugepages:

drivers/infiniband/core/umem_odp.c: In function 'ib_umem_odp_get':
drivers/infiniband/core/umem_odp.c:245:26: error: 'HPAGE_SHIFT' undeclared (first use in this function); did you mean 'PAGE_SHIFT'?

Enclose this in an #ifdef.

Fixes: 9ff1b6466a29 ("IB/core: Fix ODP with IB_ACCESS_HUGETLB handling")
Link: https://lore.kernel.org/r/20200109084740.2872079-1-arnd@arndb.de
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/umem_odp.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index f42fa31c24a2..b9baf7d0a5cb 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -241,10 +241,11 @@ struct ib_umem_odp *ib_umem_odp_get(struct ib_udata *udata, unsigned long addr,
 	umem_odp->umem.owning_mm = mm = current->mm;
 	umem_odp->notifier.ops = ops;
 
+	umem_odp->page_shift = PAGE_SHIFT;
+#ifdef CONFIG_HUGETLB_PAGE
 	if (access & IB_ACCESS_HUGETLB)
 		umem_odp->page_shift = HPAGE_SHIFT;
-	else
-		umem_odp->page_shift = PAGE_SHIFT;
+#endif
 
 	umem_odp->tgid = get_task_pid(current->group_leader, PIDTYPE_PID);
 	ret = ib_init_umem_odp(umem_odp, ops);
-- 
cgit v1.2.3-59-g8ed1b


From 14e23bd6d22123f6f3b2747701fa6cd4c6d05873 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:22:03 +0200
Subject: RDMA/core: Fix locking in ib_uverbs_event_read
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This should not be using ib_dev to test for disassociation, during
disassociation is_closed is set under lock and the waitq is triggered.

Instead check is_closed and be sure to re-obtain the lock to test the
value after the wait_event returns.

Fixes: 036b10635739 ("IB/uverbs: Enable device removal when there are active user space applications")
Link: https://lore.kernel.org/r/1578504126-9400-12-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Håkon Bugge <haakon.bugge@oracle.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs_main.c | 32 ++++++++++++++------------------
 1 file changed, 14 insertions(+), 18 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 970d8e31dd65..8f5de4dcad97 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -220,7 +220,6 @@ void ib_uverbs_release_file(struct kref *ref)
 }
 
 static ssize_t ib_uverbs_event_read(struct ib_uverbs_event_queue *ev_queue,
-				    struct ib_uverbs_file *uverbs_file,
 				    struct file *filp, char __user *buf,
 				    size_t count, loff_t *pos,
 				    size_t eventsz)
@@ -238,19 +237,16 @@ static ssize_t ib_uverbs_event_read(struct ib_uverbs_event_queue *ev_queue,
 
 		if (wait_event_interruptible(ev_queue->poll_wait,
 					     (!list_empty(&ev_queue->event_list) ||
-			/* The barriers built into wait_event_interruptible()
-			 * and wake_up() guarentee this will see the null set
-			 * without using RCU
-			 */
-					     !uverbs_file->device->ib_dev)))
+					      ev_queue->is_closed)))
 			return -ERESTARTSYS;
 
+		spin_lock_irq(&ev_queue->lock);
+
 		/* If device was disassociated and no event exists set an error */
-		if (list_empty(&ev_queue->event_list) &&
-		    !uverbs_file->device->ib_dev)
+		if (list_empty(&ev_queue->event_list) && ev_queue->is_closed) {
+			spin_unlock_irq(&ev_queue->lock);
 			return -EIO;
-
-		spin_lock_irq(&ev_queue->lock);
+		}
 	}
 
 	event = list_entry(ev_queue->event_list.next, struct ib_uverbs_event, list);
@@ -285,8 +281,7 @@ static ssize_t ib_uverbs_async_event_read(struct file *filp, char __user *buf,
 {
 	struct ib_uverbs_async_event_file *file = filp->private_data;
 
-	return ib_uverbs_event_read(&file->ev_queue, file->uverbs_file, filp,
-				    buf, count, pos,
+	return ib_uverbs_event_read(&file->ev_queue, filp, buf, count, pos,
 				    sizeof(struct ib_uverbs_async_event_desc));
 }
 
@@ -296,9 +291,8 @@ static ssize_t ib_uverbs_comp_event_read(struct file *filp, char __user *buf,
 	struct ib_uverbs_completion_event_file *comp_ev_file =
 		filp->private_data;
 
-	return ib_uverbs_event_read(&comp_ev_file->ev_queue,
-				    comp_ev_file->uobj.ufile, filp,
-				    buf, count, pos,
+	return ib_uverbs_event_read(&comp_ev_file->ev_queue, filp, buf, count,
+				    pos,
 				    sizeof(struct ib_uverbs_comp_event_desc));
 }
 
@@ -321,7 +315,9 @@ static __poll_t ib_uverbs_event_poll(struct ib_uverbs_event_queue *ev_queue,
 static __poll_t ib_uverbs_async_event_poll(struct file *filp,
 					       struct poll_table_struct *wait)
 {
-	return ib_uverbs_event_poll(filp->private_data, filp, wait);
+	struct ib_uverbs_async_event_file *file = filp->private_data;
+
+	return ib_uverbs_event_poll(&file->ev_queue, filp, wait);
 }
 
 static __poll_t ib_uverbs_comp_event_poll(struct file *filp,
@@ -335,9 +331,9 @@ static __poll_t ib_uverbs_comp_event_poll(struct file *filp,
 
 static int ib_uverbs_async_event_fasync(int fd, struct file *filp, int on)
 {
-	struct ib_uverbs_event_queue *ev_queue = filp->private_data;
+	struct ib_uverbs_async_event_file *file = filp->private_data;
 
-	return fasync_helper(fd, filp, on, &ev_queue->async_queue);
+	return fasync_helper(fd, filp, on, &file->ev_queue.async_queue);
 }
 
 static int ib_uverbs_comp_event_fasync(int fd, struct file *filp, int on)
-- 
cgit v1.2.3-59-g8ed1b


From de730f71919b1ade900a6c8a91a9fe7011b1b53b Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Mon, 6 Jan 2020 08:41:44 -0500
Subject: IB/hfi1: Move chip specific functions to chip.c

Move routines and defines associated with hdrq size validation to a chip
specific routine since the limits are specific to the device.

Fix incorrect value for min size 2 -> 32

CSR writes should also be in chip.c.

Create a chip routine to write the hdrq specific CSRs and call as
appropriate.

Link: https://lore.kernel.org/r/20200106134144.119356.74312.stgit@awfm-01.aw.intel.com
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/chip.c | 78 +++++++++++++++++++++++++++++++++++++++
 drivers/infiniband/hw/hfi1/chip.h |  6 +++
 drivers/infiniband/hw/hfi1/init.c | 72 ++----------------------------------
 3 files changed, 87 insertions(+), 69 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index b4fe37a4db01..860615d3685c 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -11858,6 +11858,84 @@ static u32 encoded_size(u32 size)
 	return 0x1;	/* if invalid, go with the minimum size */
 }
 
+/**
+ * encode_rcv_header_entry_size - return chip specific encoding for size
+ * @size: size in dwords
+ *
+ * Convert a receive header entry size that to the encoding used in the CSR.
+ *
+ * Return a zero if the given size is invalid, otherwise the encoding.
+ */
+u8 encode_rcv_header_entry_size(u8 size)
+{
+	/* there are only 3 valid receive header entry sizes */
+	if (size == 2)
+		return 1;
+	if (size == 16)
+		return 2;
+	if (size == 32)
+		return 4;
+	return 0; /* invalid */
+}
+
+/**
+ * hfi1_validate_rcvhdrcnt - validate hdrcnt
+ * @dd: the device data
+ * @thecnt: the header count
+ */
+int hfi1_validate_rcvhdrcnt(struct hfi1_devdata *dd, uint thecnt)
+{
+	if (thecnt <= HFI1_MIN_HDRQ_EGRBUF_CNT) {
+		dd_dev_err(dd, "Receive header queue count too small\n");
+		return -EINVAL;
+	}
+
+	if (thecnt > HFI1_MAX_HDRQ_EGRBUF_CNT) {
+		dd_dev_err(dd,
+			   "Receive header queue count cannot be greater than %u\n",
+			   HFI1_MAX_HDRQ_EGRBUF_CNT);
+		return -EINVAL;
+	}
+
+	if (thecnt % HDRQ_INCREMENT) {
+		dd_dev_err(dd, "Receive header queue count %d must be divisible by %lu\n",
+			   thecnt, HDRQ_INCREMENT);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/**
+ * set_hdrq_regs - set header queue registers for context
+ * @dd: the device data
+ * @ctxt: the context
+ * @entsize: the dword entry size
+ * @hdrcnt: the number of header entries
+ */
+void set_hdrq_regs(struct hfi1_devdata *dd, u8 ctxt, u8 entsize, u16 hdrcnt)
+{
+	u64 reg;
+
+	reg = (((u64)hdrcnt >> HDRQ_SIZE_SHIFT) & RCV_HDR_CNT_CNT_MASK) <<
+	      RCV_HDR_CNT_CNT_SHIFT;
+	write_kctxt_csr(dd, ctxt, RCV_HDR_CNT, reg);
+	reg = ((u64)encode_rcv_header_entry_size(entsize) &
+	       RCV_HDR_ENT_SIZE_ENT_SIZE_MASK) <<
+	      RCV_HDR_ENT_SIZE_ENT_SIZE_SHIFT;
+	write_kctxt_csr(dd, ctxt, RCV_HDR_ENT_SIZE, reg);
+	reg = ((u64)DEFAULT_RCVHDRSIZE & RCV_HDR_SIZE_HDR_SIZE_MASK) <<
+	      RCV_HDR_SIZE_HDR_SIZE_SHIFT;
+	write_kctxt_csr(dd, ctxt, RCV_HDR_SIZE, reg);
+
+	/*
+	 * Program dummy tail address for every receive context
+	 * before enabling any receive context
+	 */
+	write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR,
+			dd->rcvhdrtail_dummy_dma);
+}
+
 void hfi1_rcvctrl(struct hfi1_devdata *dd, unsigned int op,
 		  struct hfi1_ctxtdata *rcd)
 {
diff --git a/drivers/infiniband/hw/hfi1/chip.h b/drivers/infiniband/hw/hfi1/chip.h
index 4ca5ac8d7e9e..05b713098567 100644
--- a/drivers/infiniband/hw/hfi1/chip.h
+++ b/drivers/infiniband/hw/hfi1/chip.h
@@ -358,6 +358,8 @@
 #define MAX_EAGER_BUFFER       (256 * 1024)
 #define MAX_EAGER_BUFFER_TOTAL (64 * (1 << 20)) /* max per ctxt 64MB */
 #define MAX_EXPECTED_BUFFER    (2048 * 1024)
+#define HFI1_MIN_HDRQ_EGRBUF_CNT 32
+#define HFI1_MAX_HDRQ_EGRBUF_CNT 16352
 
 /*
  * Receive expected base and count and eager base and count increment -
@@ -699,6 +701,10 @@ static inline u32 chip_rcv_array_count(struct hfi1_devdata *dd)
 	return read_csr(dd, RCV_ARRAY_CNT);
 }
 
+u8 encode_rcv_header_entry_size(u8 size);
+int hfi1_validate_rcvhdrcnt(struct hfi1_devdata *dd, uint thecnt);
+void set_hdrq_regs(struct hfi1_devdata *dd, u8 ctxt, u8 entsize, u16 hdrcnt);
+
 u64 create_pbc(struct hfi1_pportdata *ppd, u64 flags, int srate_mbs, u32 vl,
 	       u32 dw_len);
 
diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c
index 3e56ee37861d..81c35516bf40 100644
--- a/drivers/infiniband/hw/hfi1/init.c
+++ b/drivers/infiniband/hw/hfi1/init.c
@@ -78,8 +78,6 @@
  */
 #define HFI1_MIN_USER_CTXT_BUFCNT 7
 
-#define HFI1_MIN_HDRQ_EGRBUF_CNT 2
-#define HFI1_MAX_HDRQ_EGRBUF_CNT 16352
 #define HFI1_MIN_EAGER_BUFFER_SIZE (4 * 1024) /* 4KB */
 #define HFI1_MAX_EAGER_BUFFER_SIZE (256 * 1024) /* 256KB */
 
@@ -122,8 +120,6 @@ unsigned int user_credit_return_threshold = 33;	/* default is 33% */
 module_param(user_credit_return_threshold, uint, S_IRUGO);
 MODULE_PARM_DESC(user_credit_return_threshold, "Credit return threshold for user send contexts, return when unreturned credits passes this many blocks (in percent of allocated blocks, 0 is off)");
 
-static inline u64 encode_rcv_header_entry_size(u16 size);
-
 DEFINE_XARRAY_FLAGS(hfi1_dev_table, XA_FLAGS_ALLOC | XA_FLAGS_LOCK_IRQ);
 
 static int hfi1_create_kctxt(struct hfi1_devdata *dd,
@@ -510,23 +506,6 @@ void hfi1_free_ctxt(struct hfi1_ctxtdata *rcd)
 	hfi1_rcd_put(rcd);
 }
 
-/*
- * Convert a receive header entry size that to the encoding used in the CSR.
- *
- * Return a zero if the given size is invalid.
- */
-static inline u64 encode_rcv_header_entry_size(u16 size)
-{
-	/* there are only 3 valid receive header entry sizes */
-	if (size == 2)
-		return 1;
-	if (size == 16)
-		return 2;
-	else if (size == 32)
-		return 4;
-	return 0; /* invalid */
-}
-
 /*
  * Select the largest ccti value over all SLs to determine the intra-
  * packet gap for the link.
@@ -1611,29 +1590,6 @@ static void postinit_cleanup(struct hfi1_devdata *dd)
 	hfi1_free_devdata(dd);
 }
 
-static int init_validate_rcvhdrcnt(struct hfi1_devdata *dd, uint thecnt)
-{
-	if (thecnt <= HFI1_MIN_HDRQ_EGRBUF_CNT) {
-		dd_dev_err(dd, "Receive header queue count too small\n");
-		return -EINVAL;
-	}
-
-	if (thecnt > HFI1_MAX_HDRQ_EGRBUF_CNT) {
-		dd_dev_err(dd,
-			   "Receive header queue count cannot be greater than %u\n",
-			   HFI1_MAX_HDRQ_EGRBUF_CNT);
-		return -EINVAL;
-	}
-
-	if (thecnt % HDRQ_INCREMENT) {
-		dd_dev_err(dd, "Receive header queue count %d must be divisible by %lu\n",
-			   thecnt, HDRQ_INCREMENT);
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
 static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 {
 	int ret = 0, j, pidx, initfail;
@@ -1661,7 +1617,7 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	}
 
 	/* Validate some global module parameters */
-	ret = init_validate_rcvhdrcnt(dd, rcvhdrcnt);
+	ret = hfi1_validate_rcvhdrcnt(dd, rcvhdrcnt);
 	if (ret)
 		goto bail;
 
@@ -1842,7 +1798,6 @@ static void shutdown_one(struct pci_dev *pdev)
 int hfi1_create_rcvhdrq(struct hfi1_devdata *dd, struct hfi1_ctxtdata *rcd)
 {
 	unsigned amt;
-	u64 reg;
 
 	if (!rcd->rcvhdrq) {
 		gfp_t gfp_flags;
@@ -1874,30 +1829,9 @@ int hfi1_create_rcvhdrq(struct hfi1_devdata *dd, struct hfi1_ctxtdata *rcd)
 				goto bail_free;
 		}
 	}
-	/*
-	 * These values are per-context:
-	 *	RcvHdrCnt
-	 *	RcvHdrEntSize
-	 *	RcvHdrSize
-	 */
-	reg = ((u64)(rcd->rcvhdrq_cnt >> HDRQ_SIZE_SHIFT)
-			& RCV_HDR_CNT_CNT_MASK)
-		<< RCV_HDR_CNT_CNT_SHIFT;
-	write_kctxt_csr(dd, rcd->ctxt, RCV_HDR_CNT, reg);
-	reg = (encode_rcv_header_entry_size(rcd->rcvhdrqentsize)
-			& RCV_HDR_ENT_SIZE_ENT_SIZE_MASK)
-		<< RCV_HDR_ENT_SIZE_ENT_SIZE_SHIFT;
-	write_kctxt_csr(dd, rcd->ctxt, RCV_HDR_ENT_SIZE, reg);
-	reg = ((u64)DEFAULT_RCVHDRSIZE & RCV_HDR_SIZE_HDR_SIZE_MASK)
-		<< RCV_HDR_SIZE_HDR_SIZE_SHIFT;
-	write_kctxt_csr(dd, rcd->ctxt, RCV_HDR_SIZE, reg);
 
-	/*
-	 * Program dummy tail address for every receive context
-	 * before enabling any receive context
-	 */
-	write_kctxt_csr(dd, rcd->ctxt, RCV_HDR_TAIL_ADDR,
-			dd->rcvhdrtail_dummy_dma);
+	set_hdrq_regs(rcd->dd, rcd->ctxt, rcd->rcvhdrqentsize,
+		      rcd->rcvhdrq_cnt);
 
 	return 0;
 
-- 
cgit v1.2.3-59-g8ed1b


From 01c7fc501b8f13a1c5d126d6015ef0908527516e Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Mon, 6 Jan 2020 08:41:51 -0500
Subject: IB/hfi1: Add fast and slow handlers for receive context

This patch eliminate special cases by adding a fast_handler member to the
receive context and changes to the fast handler as specified in the new
variable. Initialize the variable as soon as the setting for dma tail is
known when the context is created.

Setting fast path is called every time when any context has entered slow
path. Add function to check if contexts is using fast path and do not set
fast path when it is already done to improve RCD fastpath setting.

Link: https://lore.kernel.org/r/20200106134150.119356.87558.stgit@awfm-01.aw.intel.com
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/driver.c   | 66 +++++++----------------------------
 drivers/infiniband/hw/hfi1/hfi.h      | 41 +++++++++++++++++++++-
 drivers/infiniband/hw/hfi1/init.c     |  5 +++
 drivers/infiniband/hw/hfi1/trace_rx.h |  6 +---
 4 files changed, 58 insertions(+), 60 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index 8374922ea6e5..3671191b5aed 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -883,9 +883,8 @@ bail:
 	return last;
 }
 
-static inline void set_nodma_rtail(struct hfi1_devdata *dd, u16 ctxt)
+static void set_all_fastpath(struct hfi1_devdata *dd, struct hfi1_ctxtdata *rcd)
 {
-	struct hfi1_ctxtdata *rcd;
 	u16 i;
 
 	/*
@@ -893,50 +892,17 @@ static inline void set_nodma_rtail(struct hfi1_devdata *dd, u16 ctxt)
 	 * interrupt handler only for that context. Otherwise, switch
 	 * interrupt handler for all statically allocated kernel contexts.
 	 */
-	if (ctxt >= dd->first_dyn_alloc_ctxt) {
-		rcd = hfi1_rcd_get_by_index_safe(dd, ctxt);
-		if (rcd) {
-			rcd->do_interrupt =
-				&handle_receive_interrupt_nodma_rtail;
-			hfi1_rcd_put(rcd);
-		}
-		return;
-	}
-
-	for (i = HFI1_CTRL_CTXT + 1; i < dd->first_dyn_alloc_ctxt; i++) {
-		rcd = hfi1_rcd_get_by_index(dd, i);
-		if (rcd)
-			rcd->do_interrupt =
-				&handle_receive_interrupt_nodma_rtail;
+	if (rcd->ctxt >= dd->first_dyn_alloc_ctxt && !rcd->is_vnic) {
+		hfi1_rcd_get(rcd);
+		hfi1_set_fast(rcd);
 		hfi1_rcd_put(rcd);
-	}
-}
-
-static inline void set_dma_rtail(struct hfi1_devdata *dd, u16 ctxt)
-{
-	struct hfi1_ctxtdata *rcd;
-	u16 i;
-
-	/*
-	 * For dynamically allocated kernel contexts (like vnic) switch
-	 * interrupt handler only for that context. Otherwise, switch
-	 * interrupt handler for all statically allocated kernel contexts.
-	 */
-	if (ctxt >= dd->first_dyn_alloc_ctxt) {
-		rcd = hfi1_rcd_get_by_index_safe(dd, ctxt);
-		if (rcd) {
-			rcd->do_interrupt =
-				&handle_receive_interrupt_dma_rtail;
-			hfi1_rcd_put(rcd);
-		}
 		return;
 	}
 
-	for (i = HFI1_CTRL_CTXT + 1; i < dd->first_dyn_alloc_ctxt; i++) {
+	for (i = HFI1_CTRL_CTXT + 1; i < dd->num_rcv_contexts; i++) {
 		rcd = hfi1_rcd_get_by_index(dd, i);
-		if (rcd)
-			rcd->do_interrupt =
-				&handle_receive_interrupt_dma_rtail;
+		if (rcd && (i < dd->first_dyn_alloc_ctxt || rcd->is_vnic))
+			hfi1_set_fast(rcd);
 		hfi1_rcd_put(rcd);
 	}
 }
@@ -952,7 +918,7 @@ void set_all_slowpath(struct hfi1_devdata *dd)
 		if (!rcd)
 			continue;
 		if (i < dd->first_dyn_alloc_ctxt || rcd->is_vnic)
-			rcd->do_interrupt = &handle_receive_interrupt;
+			rcd->do_interrupt = rcd->slow_handler;
 
 		hfi1_rcd_put(rcd);
 	}
@@ -1065,11 +1031,6 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 		if (!get_dma_rtail_setting(rcd)) {
 			if (hfi1_seq_incr(rcd, rhf_rcv_seq(packet.rhf)))
 				last = RCV_PKT_DONE;
-			if (needset) {
-				dd_dev_info(dd, "Switching to NO_DMA_RTAIL\n");
-				set_nodma_rtail(dd, rcd->ctxt);
-				needset = 0;
-			}
 		} else {
 			if (packet.rhqoff == hdrqtail)
 				last = RCV_PKT_DONE;
@@ -1085,15 +1046,12 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 				if (!last && lseq)
 					skip_pkt = 1;
 			}
-
-			if (needset) {
-				dd_dev_info(dd,
-					    "Switching to DMA_RTAIL\n");
-				set_dma_rtail(dd, rcd->ctxt);
-				needset = 0;
-			}
 		}
 
+		if (needset) {
+			needset = false;
+			set_all_fastpath(dd, rcd);
+		}
 		process_rcv_update(last, &packet);
 	}
 
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index d662fcc9a1b4..0c643bcd9b17 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -197,6 +197,8 @@ struct exp_tid_set {
 	u32 count;
 };
 
+struct hfi1_ctxtdata;
+typedef int (*intr_handler)(struct hfi1_ctxtdata *rcd, int data);
 typedef int (*rhf_rcv_function_ptr)(struct hfi1_packet *packet);
 
 struct tid_queue {
@@ -226,7 +228,11 @@ struct hfi1_ctxtdata {
 	 * be valid. Worst case is we process an extra interrupt and up to 64
 	 * packets with the wrong interrupt handler.
 	 */
-	int (*do_interrupt)(struct hfi1_ctxtdata *rcd, int threaded);
+	intr_handler do_interrupt;
+	/** fast handler after autoactive */
+	intr_handler fast_handler;
+	/** slow handler */
+	intr_handler slow_handler;
 	/* verbs rx_stats per rcd */
 	struct hfi1_opcode_stats_perctx *opstats;
 	/* clear interrupt mask */
@@ -1616,6 +1622,39 @@ static inline u16 get_hdrq_cnt(struct hfi1_ctxtdata *rcd)
 	return rcd->rcvhdrq_cnt;
 }
 
+/**
+ * hfi1_is_slowpath - check if this context is slow path
+ * @rcd: the receive context
+ */
+static inline bool hfi1_is_slowpath(struct hfi1_ctxtdata *rcd)
+{
+	return rcd->do_interrupt == rcd->slow_handler;
+}
+
+/**
+ * hfi1_is_fastpath - check if this context is fast path
+ * @rcd: the receive context
+ */
+static inline bool hfi1_is_fastpath(struct hfi1_ctxtdata *rcd)
+{
+	if (rcd->ctxt == HFI1_CTRL_CTXT)
+		return false;
+
+	return rcd->do_interrupt == rcd->fast_handler;
+}
+
+/**
+ * hfi1_set_fast - change to the fast handler
+ * @rcd: the receive context
+ */
+static inline void hfi1_set_fast(struct hfi1_ctxtdata *rcd)
+{
+	if (unlikely(!rcd))
+		return;
+	if (unlikely(!hfi1_is_fastpath(rcd)))
+		rcd->do_interrupt = rcd->fast_handler;
+}
+
 int hfi1_reset_device(int);
 
 void receive_interrupt_work(struct work_struct *work);
diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c
index 81c35516bf40..925f75f94487 100644
--- a/drivers/infiniband/hw/hfi1/init.c
+++ b/drivers/infiniband/hw/hfi1/init.c
@@ -150,6 +150,11 @@ static int hfi1_create_kctxt(struct hfi1_devdata *dd,
 	/* Control context must use DMA_RTAIL */
 	if (rcd->ctxt == HFI1_CTRL_CTXT)
 		rcd->flags |= HFI1_CAP_DMA_RTAIL;
+	rcd->fast_handler = get_dma_rtail_setting(rcd) ?
+				handle_receive_interrupt_dma_rtail :
+				handle_receive_interrupt_nodma_rtail;
+	rcd->slow_handler = handle_receive_interrupt;
+
 	hfi1_set_seq_cnt(rcd, 1);
 
 	rcd->sc = sc_alloc(dd, SC_ACK, rcd->rcvhdrqentsize, dd->node);
diff --git a/drivers/infiniband/hw/hfi1/trace_rx.h b/drivers/infiniband/hw/hfi1/trace_rx.h
index ebce81c7ac41..168079ed122c 100644
--- a/drivers/infiniband/hw/hfi1/trace_rx.h
+++ b/drivers/infiniband/hw/hfi1/trace_rx.h
@@ -106,11 +106,7 @@ TRACE_EVENT(hfi1_receive_interrupt,
 			     ),
 	    TP_fast_assign(DD_DEV_ASSIGN(dd);
 			__entry->ctxt = rcd->ctxt;
-			if (rcd->do_interrupt ==
-			    &handle_receive_interrupt)
-				__entry->slow_path = 1;
-			else
-				__entry->slow_path = 0;
+			__entry->slow_path = hfi1_is_slowpath(rcd);
 			__entry->dma_rtail = get_dma_rtail_setting(rcd);
 			),
 	    TP_printk("[%s] ctxt %d SlowPath: %d DmaRtail: %d",
-- 
cgit v1.2.3-59-g8ed1b


From 7b8a8b72c9eaa631d2065ad50f15eb448b43c352 Mon Sep 17 00:00:00 2001
From: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Date: Mon, 6 Jan 2020 08:41:57 -0500
Subject: IB/hfi1: Move common receive IRQ code to function

Tracing interrupts, incrementing interrupt counter and ASPM are part that
will be reused by HFI1 receive IRQ handlers.

Create common function to have shared code in one place.

Link: https://lore.kernel.org/r/20200106134157.119356.32656.stgit@awfm-01.aw.intel.com
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/chip.c | 82 +++++++++++++++++++++++++--------------
 1 file changed, 52 insertions(+), 30 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index 860615d3685c..9e2bf9959c96 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -8403,6 +8403,55 @@ static inline int check_packet_present(struct hfi1_ctxtdata *rcd)
 	return hfi1_rcd_head(rcd) != tail;
 }
 
+/**
+ * Common code for receive contexts interrupt handlers.
+ * Update traces, increment kernel IRQ counter and
+ * setup ASPM when needed.
+ */
+static void receive_interrupt_common(struct hfi1_ctxtdata *rcd)
+{
+	struct hfi1_devdata *dd = rcd->dd;
+
+	trace_hfi1_receive_interrupt(dd, rcd);
+	this_cpu_inc(*dd->int_counter);
+	aspm_ctx_disable(rcd);
+}
+
+/**
+ * __hfi1_rcd_eoi_intr() - Make HW issue receive interrupt
+ * when there are packets present in the queue. When calling
+ * with interrupts enabled please use hfi1_rcd_eoi_intr.
+ *
+ * @rcd: valid receive context
+ */
+static void __hfi1_rcd_eoi_intr(struct hfi1_ctxtdata *rcd)
+{
+	clear_recv_intr(rcd);
+	if (check_packet_present(rcd))
+		force_recv_intr(rcd);
+}
+
+/**
+ * hfi1_rcd_eoi_intr() - End of Interrupt processing action
+ *
+ * @rcd: Ptr to hfi1_ctxtdata of receive context
+ *
+ *  Hold IRQs so we can safely clear the interrupt and
+ *  recheck for a packet that may have arrived after the previous
+ *  check and the interrupt clear.  If a packet arrived, force another
+ *  interrupt. This routine can be called at the end of receive packet
+ *  processing in interrupt service routines, interrupt service thread
+ *  and softirqs
+ */
+static void hfi1_rcd_eoi_intr(struct hfi1_ctxtdata *rcd)
+{
+	unsigned long flags;
+
+	local_irq_save(flags);
+	__hfi1_rcd_eoi_intr(rcd);
+	local_irq_restore(flags);
+}
+
 /*
  * Receive packet IRQ handler.  This routine expects to be on its own IRQ.
  * This routine will try to handle packets immediately (latency), but if
@@ -8414,13 +8463,9 @@ static inline int check_packet_present(struct hfi1_ctxtdata *rcd)
 irqreturn_t receive_context_interrupt(int irq, void *data)
 {
 	struct hfi1_ctxtdata *rcd = data;
-	struct hfi1_devdata *dd = rcd->dd;
 	int disposition;
-	int present;
 
-	trace_hfi1_receive_interrupt(dd, rcd);
-	this_cpu_inc(*dd->int_counter);
-	aspm_ctx_disable(rcd);
+	receive_interrupt_common(rcd);
 
 	/* receive interrupt remains blocked while processing packets */
 	disposition = rcd->do_interrupt(rcd, 0);
@@ -8433,17 +8478,7 @@ irqreturn_t receive_context_interrupt(int irq, void *data)
 	if (disposition == RCV_PKT_LIMIT)
 		return IRQ_WAKE_THREAD;
 
-	/*
-	 * The packet processor detected no more packets.  Clear the receive
-	 * interrupt and recheck for a packet packet that may have arrived
-	 * after the previous check and interrupt clear.  If a packet arrived,
-	 * force another interrupt.
-	 */
-	clear_recv_intr(rcd);
-	present = check_packet_present(rcd);
-	if (present)
-		force_recv_intr(rcd);
-
+	__hfi1_rcd_eoi_intr(rcd);
 	return IRQ_HANDLED;
 }
 
@@ -8454,24 +8489,11 @@ irqreturn_t receive_context_interrupt(int irq, void *data)
 irqreturn_t receive_context_thread(int irq, void *data)
 {
 	struct hfi1_ctxtdata *rcd = data;
-	int present;
 
 	/* receive interrupt is still blocked from the IRQ handler */
 	(void)rcd->do_interrupt(rcd, 1);
 
-	/*
-	 * The packet processor will only return if it detected no more
-	 * packets.  Hold IRQs here so we can safely clear the interrupt and
-	 * recheck for a packet that may have arrived after the previous
-	 * check and the interrupt clear.  If a packet arrived, force another
-	 * interrupt.
-	 */
-	local_irq_disable();
-	clear_recv_intr(rcd);
-	present = check_packet_present(rcd);
-	if (present)
-		force_recv_intr(rcd);
-	local_irq_enable();
+	hfi1_rcd_eoi_intr(rcd);
 
 	return IRQ_HANDLED;
 }
-- 
cgit v1.2.3-59-g8ed1b


From cd47b594dbef7c573c0790026a0ea48273ed2b14 Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Mon, 6 Jan 2020 08:42:04 -0500
Subject: IB/hfi1: IB/hfi1: Add an API to handle special case drop

This patch pushes special case drop logic into an API to be shared by all
interrupt handlers.

Additionally, convert do_drop to a bool.

Link: https://lore.kernel.org/r/20200106134203.119356.36962.stgit@awfm-01.aw.intel.com
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/driver.c |  6 +-----
 drivers/infiniband/hw/hfi1/hfi.h    | 21 ++++++++++++++++++++-
 drivers/infiniband/hw/hfi1/init.c   |  4 ++--
 3 files changed, 23 insertions(+), 8 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index 3671191b5aed..bbc74583628c 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -1004,11 +1004,7 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 	prescan_rxq(rcd, &packet);
 
 	while (last == RCV_PKT_OK) {
-		if (unlikely(dd->do_drop &&
-			     atomic_xchg(&dd->drop_packet, DROP_PACKET_OFF) ==
-			     DROP_PACKET_ON)) {
-			dd->do_drop = 0;
-
+		if (hfi1_need_drop(dd)) {
 			/* On to the next packet */
 			packet.rhqoff += packet.rsize;
 			packet.rhf_addr = (__le32 *)rcd->rcvhdrq +
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index 0c643bcd9b17..92125434f820 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -1316,7 +1316,7 @@ struct hfi1_devdata {
 	struct err_info_constraint err_info_xmit_constraint;
 
 	atomic_t drop_packet;
-	u8 do_drop;
+	bool do_drop;
 	u8 err_info_uncorrectable;
 	u8 err_info_fmconfig;
 
@@ -2462,6 +2462,25 @@ static inline bool is_integrated(struct hfi1_devdata *dd)
 	return dd->pcidev->device == PCI_DEVICE_ID_INTEL1;
 }
 
+/**
+ * hfi1_need_drop - detect need for drop
+ * @dd: - the device
+ *
+ * In some cases, the first packet needs to be dropped.
+ *
+ * Return true is the current packet needs to be dropped and false otherwise.
+ */
+static inline bool hfi1_need_drop(struct hfi1_devdata *dd)
+{
+	if (unlikely(dd->do_drop &&
+		     atomic_xchg(&dd->drop_packet, DROP_PACKET_OFF) ==
+		     DROP_PACKET_ON)) {
+		dd->do_drop = false;
+		return true;
+	}
+	return false;
+}
+
 int hfi1_tempsense_rd(struct hfi1_devdata *dd, struct hfi1_temp *temp);
 
 #define DD_DEV_ENTRY(dd)       __string(dev, dev_name(&(dd)->pcidev->dev))
diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c
index 925f75f94487..e3acda7a0800 100644
--- a/drivers/infiniband/hw/hfi1/init.c
+++ b/drivers/infiniband/hw/hfi1/init.c
@@ -876,10 +876,10 @@ int hfi1_init(struct hfi1_devdata *dd, int reinit)
 
 	if (is_ax(dd)) {
 		atomic_set(&dd->drop_packet, DROP_PACKET_ON);
-		dd->do_drop = 1;
+		dd->do_drop = true;
 	} else {
 		atomic_set(&dd->drop_packet, DROP_PACKET_OFF);
-		dd->do_drop = 0;
+		dd->do_drop = false;
 	}
 
 	/* make sure the link is not "up" */
-- 
cgit v1.2.3-59-g8ed1b


From 62661038c33b96ed949e751780a69f4d3608bad0 Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Mon, 6 Jan 2020 08:42:10 -0500
Subject: IB/hfi1: Create API for auto activate

Add an auto activate routine for use by the interrupt handler.

Link: https://lore.kernel.org/r/20200106134210.119356.43079.stgit@awfm-01.aw.intel.com
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/driver.c | 37 +++++++++++++++++++++++--------------
 1 file changed, 23 insertions(+), 14 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index bbc74583628c..46c1be0c4b80 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -924,11 +924,8 @@ void set_all_slowpath(struct hfi1_devdata *dd)
 	}
 }
 
-static inline int set_armed_to_active(struct hfi1_ctxtdata *rcd,
-				      struct hfi1_packet *packet,
-				      struct hfi1_devdata *dd)
+static bool __set_armed_to_active(struct hfi1_packet *packet)
 {
-	struct work_struct *lsaw = &rcd->ppd->linkstate_active_work;
 	u8 etype = rhf_rcv_type(packet->rhf);
 	u8 sc = SC15_PACKET;
 
@@ -943,19 +940,34 @@ static inline int set_armed_to_active(struct hfi1_ctxtdata *rcd,
 		sc = hfi1_16B_get_sc(hdr);
 	}
 	if (sc != SC15_PACKET) {
-		int hwstate = driver_lstate(rcd->ppd);
+		int hwstate = driver_lstate(packet->rcd->ppd);
+		struct work_struct *lsaw =
+				&packet->rcd->ppd->linkstate_active_work;
 
 		if (hwstate != IB_PORT_ACTIVE) {
-			dd_dev_info(dd,
+			dd_dev_info(packet->rcd->dd,
 				    "Unexpected link state %s\n",
 				    opa_lstate_name(hwstate));
-			return 0;
+			return false;
 		}
 
-		queue_work(rcd->ppd->link_wq, lsaw);
-		return 1;
+		queue_work(packet->rcd->ppd->link_wq, lsaw);
+		return true;
 	}
-	return 0;
+	return false;
+}
+
+/**
+ * armed to active - the fast path for armed to active
+ * @packet: the packet structure
+ *
+ * Return true if packet processing needs to bail.
+ */
+static bool set_armed_to_active(struct hfi1_packet *packet)
+{
+	if (likely(packet->rcd->ppd->host_link_state != HLS_UP_ARMED))
+		return false;
+	return __set_armed_to_active(packet);
 }
 
 /*
@@ -1016,10 +1028,7 @@ int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread)
 			last = skip_rcv_packet(&packet, thread);
 			skip_pkt = 0;
 		} else {
-			/* Auto activate link on non-SC15 packet receive */
-			if (unlikely(rcd->ppd->host_link_state ==
-				     HLS_UP_ARMED) &&
-			    set_armed_to_active(rcd, &packet, dd))
+			if (set_armed_to_active(&packet))
 				goto bail;
 			last = process_rcv_packet(&packet, thread);
 		}
-- 
cgit v1.2.3-59-g8ed1b


From 13d2a8384bd9b2dc3f737b19e3d74c845590e6eb Mon Sep 17 00:00:00 2001
From: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Date: Mon, 6 Jan 2020 08:42:16 -0500
Subject: IB/hfi1: Decouple IRQ name from type

IRQ name was connected to IRQ type, this is not sufficient and it would be
better to use name as argument to msix_request_irq instead of assigning it
to variables when function is called.

Index argument was required to generate name and now it can be removed.

To generate name correctly helpers function were added and updated.

Link: https://lore.kernel.org/r/20200106134216.119356.44478.stgit@awfm-01.aw.intel.com
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/msix.c | 106 +++++++++++++++++++++-----------------
 drivers/infiniband/hw/hfi1/msix.h |   1 +
 2 files changed, 59 insertions(+), 48 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/msix.c b/drivers/infiniband/hw/hfi1/msix.c
index d920b165d696..4a620cf80588 100644
--- a/drivers/infiniband/hw/hfi1/msix.c
+++ b/drivers/infiniband/hw/hfi1/msix.c
@@ -115,13 +115,11 @@ int msix_initialize(struct hfi1_devdata *dd)
  */
 static int msix_request_irq(struct hfi1_devdata *dd, void *arg,
 			    irq_handler_t handler, irq_handler_t thread,
-			    u32 idx, enum irq_type type)
+			    enum irq_type type, const char *name)
 {
 	unsigned long nr;
 	int irq;
 	int ret;
-	const char *err_info;
-	char name[MAX_NAME_SIZE];
 	struct hfi1_msix_entry *me;
 
 	/* Allocate an MSIx vector */
@@ -135,43 +133,15 @@ static int msix_request_irq(struct hfi1_devdata *dd, void *arg,
 	if (nr == dd->msix_info.max_requested)
 		return -ENOSPC;
 
-	/* Specific verification and determine the name */
-	switch (type) {
-	case IRQ_GENERAL:
-		/* general interrupt must be MSIx vector 0 */
-		if (nr) {
-			spin_lock(&dd->msix_info.msix_lock);
-			__clear_bit(nr, dd->msix_info.in_use_msix);
-			spin_unlock(&dd->msix_info.msix_lock);
-			dd_dev_err(dd, "Invalid index %lu for GENERAL IRQ\n",
-				   nr);
-			return -EINVAL;
-		}
-		snprintf(name, sizeof(name), DRIVER_NAME "_%d", dd->unit);
-		err_info = "general";
-		break;
-	case IRQ_SDMA:
-		snprintf(name, sizeof(name), DRIVER_NAME "_%d sdma%d",
-			 dd->unit, idx);
-		err_info = "sdma";
-		break;
-	case IRQ_RCVCTXT:
-		snprintf(name, sizeof(name), DRIVER_NAME "_%d kctxt%d",
-			 dd->unit, idx);
-		err_info = "receive context";
-		break;
-	case IRQ_OTHER:
-	default:
+	if (type < IRQ_SDMA && type >= IRQ_OTHER)
 		return -EINVAL;
-	}
-	name[sizeof(name) - 1] = 0;
 
 	irq = pci_irq_vector(dd->pcidev, nr);
 	ret = pci_request_irq(dd->pcidev, nr, handler, thread, arg, name);
 	if (ret) {
 		dd_dev_err(dd,
-			   "%s: request for IRQ %d failed, MSIx %d, err %d\n",
-			   err_info, irq, idx, ret);
+			   "%s: request for IRQ %d failed, MSIx %lu, err %d\n",
+			   name, irq, nr, ret);
 		spin_lock(&dd->msix_info.msix_lock);
 		__clear_bit(nr, dd->msix_info.in_use_msix);
 		spin_unlock(&dd->msix_info.msix_lock);
@@ -195,17 +165,13 @@ static int msix_request_irq(struct hfi1_devdata *dd, void *arg,
 	return nr;
 }
 
-/**
- * msix_request_rcd_irq() - Helper function for RCVAVAIL IRQs
- * @rcd: valid rcd context
- *
- */
-int msix_request_rcd_irq(struct hfi1_ctxtdata *rcd)
+static int msix_request_rcd_irq_common(struct hfi1_ctxtdata *rcd,
+				       irq_handler_t handler,
+				       irq_handler_t thread,
+				       const char *name)
 {
-	int nr;
-
-	nr = msix_request_irq(rcd->dd, rcd, receive_context_interrupt,
-			      receive_context_thread, rcd->ctxt, IRQ_RCVCTXT);
+	int nr = msix_request_irq(rcd->dd, rcd, handler, thread,
+				  IRQ_RCVCTXT, name);
 	if (nr < 0)
 		return nr;
 
@@ -221,6 +187,22 @@ int msix_request_rcd_irq(struct hfi1_ctxtdata *rcd)
 	return 0;
 }
 
+/**
+ * msix_request_rcd_irq() - Helper function for RCVAVAIL IRQs
+ * @rcd: valid rcd context
+ *
+ */
+int msix_request_rcd_irq(struct hfi1_ctxtdata *rcd)
+{
+	char name[MAX_NAME_SIZE];
+
+	snprintf(name, sizeof(name), DRIVER_NAME "_%d kctxt%d",
+		 rcd->dd->unit, rcd->ctxt);
+
+	return msix_request_rcd_irq_common(rcd, receive_context_interrupt,
+					   receive_context_thread, name);
+}
+
 /**
  * msix_request_smda_ira() - Helper for getting SDMA IRQ resources
  * @sde: valid sdma engine
@@ -229,9 +211,12 @@ int msix_request_rcd_irq(struct hfi1_ctxtdata *rcd)
 int msix_request_sdma_irq(struct sdma_engine *sde)
 {
 	int nr;
+	char name[MAX_NAME_SIZE];
 
+	snprintf(name, sizeof(name), DRIVER_NAME "_%d sdma%d",
+		 sde->dd->unit, sde->this_idx);
 	nr = msix_request_irq(sde->dd, sde, sdma_interrupt, NULL,
-			      sde->this_idx, IRQ_SDMA);
+			      IRQ_SDMA, name);
 	if (nr < 0)
 		return nr;
 	sde->msix_intr = nr;
@@ -240,6 +225,32 @@ int msix_request_sdma_irq(struct sdma_engine *sde)
 	return 0;
 }
 
+/**
+ * msix_request_general_irq(void) - Helper for getting general IRQ
+ * resources
+ * @dd: valid device data
+ */
+int msix_request_general_irq(struct hfi1_devdata *dd)
+{
+	int nr;
+	char name[MAX_NAME_SIZE];
+
+	snprintf(name, sizeof(name), DRIVER_NAME "_%d", dd->unit);
+	nr = msix_request_irq(dd, dd, general_interrupt, NULL, IRQ_GENERAL,
+			      name);
+	if (nr < 0)
+		return nr;
+
+	/* general interrupt must be MSIx vector 0 */
+	if (nr) {
+		msix_free_irq(dd, (u8)nr);
+		dd_dev_err(dd, "Invalid index %d for GENERAL IRQ\n", nr);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 /**
  * enable_sdma_src() - Helper to enable SDMA IRQ srcs
  * @dd: valid devdata structure
@@ -265,10 +276,9 @@ static void enable_sdma_srcs(struct hfi1_devdata *dd, int i)
 int msix_request_irqs(struct hfi1_devdata *dd)
 {
 	int i;
-	int ret;
+	int ret = msix_request_general_irq(dd);
 
-	ret = msix_request_irq(dd, dd, general_interrupt, NULL, 0, IRQ_GENERAL);
-	if (ret < 0)
+	if (ret)
 		return ret;
 
 	for (i = 0; i < dd->num_sdma; i++) {
diff --git a/drivers/infiniband/hw/hfi1/msix.h b/drivers/infiniband/hw/hfi1/msix.h
index a514881632a4..1a02ab7971c8 100644
--- a/drivers/infiniband/hw/hfi1/msix.h
+++ b/drivers/infiniband/hw/hfi1/msix.h
@@ -54,6 +54,7 @@
 int msix_initialize(struct hfi1_devdata *dd);
 int msix_request_irqs(struct hfi1_devdata *dd);
 void msix_clean_up_interrupts(struct hfi1_devdata *dd);
+int msix_request_general_irq(struct hfi1_devdata *dd);
 int msix_request_rcd_irq(struct hfi1_ctxtdata *rcd);
 int msix_request_sdma_irq(struct sdma_engine *sde);
 void msix_free_irq(struct hfi1_devdata *dd, u8 msix_intr);
-- 
cgit v1.2.3-59-g8ed1b


From d791d294edd2f96e5200882f789e8cb4ed431991 Mon Sep 17 00:00:00 2001
From: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Date: Mon, 6 Jan 2020 08:42:22 -0500
Subject: IB/hfi1: Return void in packet receiving functions

Packet receiving functions returns int value, and yet the return values
are not used at all.

This patch converts the functions to return void.

Link: https://lore.kernel.org/r/20200106134222.119356.84098.stgit@awfm-01.aw.intel.com
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/driver.c | 41 +++++++++++++++----------------------
 drivers/infiniband/hw/hfi1/hfi.h    |  2 +-
 2 files changed, 18 insertions(+), 25 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index 46c1be0c4b80..0b9fbfa8b7fb 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -1553,23 +1553,22 @@ void handle_eflags(struct hfi1_packet *packet)
  * The following functions are called by the interrupt handler. They are type
  * specific handlers for each packet type.
  */
-static int process_receive_ib(struct hfi1_packet *packet)
+static void process_receive_ib(struct hfi1_packet *packet)
 {
 	if (hfi1_setup_9B_packet(packet))
-		return RHF_RCV_CONTINUE;
+		return;
 
 	if (unlikely(hfi1_dbg_should_fault_rx(packet)))
-		return RHF_RCV_CONTINUE;
+		return;
 
 	trace_hfi1_rcvhdr(packet);
 
 	if (unlikely(rhf_err_flags(packet->rhf))) {
 		handle_eflags(packet);
-		return RHF_RCV_CONTINUE;
+		return;
 	}
 
 	hfi1_ib_rcv(packet);
-	return RHF_RCV_CONTINUE;
 }
 
 static inline bool hfi1_is_vnic_packet(struct hfi1_packet *packet)
@@ -1585,23 +1584,23 @@ static inline bool hfi1_is_vnic_packet(struct hfi1_packet *packet)
 	return false;
 }
 
-static int process_receive_bypass(struct hfi1_packet *packet)
+static void process_receive_bypass(struct hfi1_packet *packet)
 {
 	struct hfi1_devdata *dd = packet->rcd->dd;
 
 	if (hfi1_is_vnic_packet(packet)) {
 		hfi1_vnic_bypass_rcv(packet);
-		return RHF_RCV_CONTINUE;
+		return;
 	}
 
 	if (hfi1_setup_bypass_packet(packet))
-		return RHF_RCV_CONTINUE;
+		return;
 
 	trace_hfi1_rcvhdr(packet);
 
 	if (unlikely(rhf_err_flags(packet->rhf))) {
 		handle_eflags(packet);
-		return RHF_RCV_CONTINUE;
+		return;
 	}
 
 	if (hfi1_16B_get_l2(packet->hdr) == 0x2) {
@@ -1624,17 +1623,16 @@ static int process_receive_bypass(struct hfi1_packet *packet)
 				(OPA_EI_STATUS_SMASK | BAD_L2_ERR);
 		}
 	}
-	return RHF_RCV_CONTINUE;
 }
 
-static int process_receive_error(struct hfi1_packet *packet)
+static void process_receive_error(struct hfi1_packet *packet)
 {
 	/* KHdrHCRCErr -- KDETH packet with a bad HCRC */
 	if (unlikely(
 		 hfi1_dbg_fault_suppress_err(&packet->rcd->dd->verbs_dev) &&
 		 (rhf_rcv_type_err(packet->rhf) == RHF_RCV_TYPE_ERROR ||
 		  packet->rhf & RHF_DC_ERR)))
-		return RHF_RCV_CONTINUE;
+		return;
 
 	hfi1_setup_ib_header(packet);
 	handle_eflags(packet);
@@ -1642,32 +1640,29 @@ static int process_receive_error(struct hfi1_packet *packet)
 	if (unlikely(rhf_err_flags(packet->rhf)))
 		dd_dev_err(packet->rcd->dd,
 			   "Unhandled error packet received. Dropping.\n");
-
-	return RHF_RCV_CONTINUE;
 }
 
-static int kdeth_process_expected(struct hfi1_packet *packet)
+static void kdeth_process_expected(struct hfi1_packet *packet)
 {
 	hfi1_setup_9B_packet(packet);
 	if (unlikely(hfi1_dbg_should_fault_rx(packet)))
-		return RHF_RCV_CONTINUE;
+		return;
 
 	if (unlikely(rhf_err_flags(packet->rhf))) {
 		struct hfi1_ctxtdata *rcd = packet->rcd;
 
 		if (hfi1_handle_kdeth_eflags(rcd, rcd->ppd, packet))
-			return RHF_RCV_CONTINUE;
+			return;
 	}
 
 	hfi1_kdeth_expected_rcv(packet);
-	return RHF_RCV_CONTINUE;
 }
 
-static int kdeth_process_eager(struct hfi1_packet *packet)
+static void kdeth_process_eager(struct hfi1_packet *packet)
 {
 	hfi1_setup_9B_packet(packet);
 	if (unlikely(hfi1_dbg_should_fault_rx(packet)))
-		return RHF_RCV_CONTINUE;
+		return;
 
 	trace_hfi1_rcvhdr(packet);
 	if (unlikely(rhf_err_flags(packet->rhf))) {
@@ -1675,18 +1670,16 @@ static int kdeth_process_eager(struct hfi1_packet *packet)
 
 		show_eflags_errs(packet);
 		if (hfi1_handle_kdeth_eflags(rcd, rcd->ppd, packet))
-			return RHF_RCV_CONTINUE;
+			return;
 	}
 
 	hfi1_kdeth_eager_rcv(packet);
-	return RHF_RCV_CONTINUE;
 }
 
-static int process_receive_invalid(struct hfi1_packet *packet)
+static void process_receive_invalid(struct hfi1_packet *packet)
 {
 	dd_dev_err(packet->rcd->dd, "Invalid packet type %d. Dropping\n",
 		   rhf_rcv_type(packet->rhf));
-	return RHF_RCV_CONTINUE;
 }
 
 #define HFI1_RCVHDR_DUMP_MAX	5
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index 92125434f820..752dc087cc42 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -199,7 +199,7 @@ struct exp_tid_set {
 
 struct hfi1_ctxtdata;
 typedef int (*intr_handler)(struct hfi1_ctxtdata *rcd, int data);
-typedef int (*rhf_rcv_function_ptr)(struct hfi1_packet *packet);
+typedef void (*rhf_rcv_function_ptr)(struct hfi1_packet *packet);
 
 struct tid_queue {
 	struct list_head queue_head;
-- 
cgit v1.2.3-59-g8ed1b


From 5ffd048698ea5139743acd45e8ab388a683642b8 Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Mon, 6 Jan 2020 08:42:28 -0500
Subject: IB/hfi1: Add software counter for ctxt0 seq drop

All other code paths increment some form of drop counter.

This was missed in the original implementation.

Fixes: 82c2611daaf0 ("staging/rdma/hfi1: Handle packets with invalid RHF on context 0")
Link: https://lore.kernel.org/r/20200106134228.119356.96828.stgit@awfm-01.aw.intel.com
Reviewed-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/chip.c   | 10 ++++++++++
 drivers/infiniband/hw/hfi1/chip.h   |  1 +
 drivers/infiniband/hw/hfi1/driver.c |  1 +
 drivers/infiniband/hw/hfi1/hfi.h    |  2 ++
 4 files changed, 14 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index 9e2bf9959c96..012f3331b8b4 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -1685,6 +1685,14 @@ static u64 access_sw_pio_drain(const struct cntr_entry *entry,
 	return dd->verbs_dev.n_piodrain;
 }
 
+static u64 access_sw_ctx0_seq_drop(const struct cntr_entry *entry,
+				   void *context, int vl, int mode, u64 data)
+{
+	struct hfi1_devdata *dd = context;
+
+	return dd->ctx0_seq_drop;
+}
+
 static u64 access_sw_vtx_wait(const struct cntr_entry *entry,
 			      void *context, int vl, int mode, u64 data)
 {
@@ -4249,6 +4257,8 @@ static struct cntr_entry dev_cntrs[DEV_CNTR_LAST] = {
 			    access_sw_cpu_intr),
 [C_SW_CPU_RCV_LIM] = CNTR_ELEM("RcvLimit", 0, 0, CNTR_NORMAL,
 			    access_sw_cpu_rcv_limit),
+[C_SW_CTX0_SEQ_DROP] = CNTR_ELEM("SeqDrop0", 0, 0, CNTR_NORMAL,
+			    access_sw_ctx0_seq_drop),
 [C_SW_VTX_WAIT] = CNTR_ELEM("vTxWait", 0, 0, CNTR_NORMAL,
 			    access_sw_vtx_wait),
 [C_SW_PIO_WAIT] = CNTR_ELEM("PioWait", 0, 0, CNTR_NORMAL,
diff --git a/drivers/infiniband/hw/hfi1/chip.h b/drivers/infiniband/hw/hfi1/chip.h
index 05b713098567..93c9c8c0ac30 100644
--- a/drivers/infiniband/hw/hfi1/chip.h
+++ b/drivers/infiniband/hw/hfi1/chip.h
@@ -932,6 +932,7 @@ enum {
 	C_DC_PG_STS_TX_MBE_CNT,
 	C_SW_CPU_INTR,
 	C_SW_CPU_RCV_LIM,
+	C_SW_CTX0_SEQ_DROP,
 	C_SW_VTX_WAIT,
 	C_SW_PIO_WAIT,
 	C_SW_PIO_DRAIN,
diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index 0b9fbfa8b7fb..049d15befe58 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -732,6 +732,7 @@ static noinline int skip_rcv_packet(struct hfi1_packet *packet, int thread)
 {
 	int ret;
 
+	packet->rcd->dd->ctx0_seq_drop++;
 	/* Set up for the next packet */
 	packet->rhqoff += packet->rsize;
 	if (packet->rhqoff >= packet->maxcnt)
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index 752dc087cc42..6365e8ffed9d 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -1159,6 +1159,8 @@ struct hfi1_devdata {
 
 	char *boardname; /* human readable board info */
 
+	u64 ctx0_seq_drop;
+
 	/* reset value */
 	u64 z_int_counter;
 	u64 z_rcv_limit;
-- 
cgit v1.2.3-59-g8ed1b


From 2c9d4e26d1ab27ceae2ded2ffe930f8e5f5b2a89 Mon Sep 17 00:00:00 2001
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
Date: Mon, 6 Jan 2020 08:42:35 -0500
Subject: IB/hfi1: Add RcvShortLengthErrCnt to hfi1stats

This counter, RxShrErr, is required for error analysis and debug.

Fixes: 7724105686e7 ("IB/hfi1: add driver files")
Link: https://lore.kernel.org/r/20200106134235.119356.29123.stgit@awfm-01.aw.intel.com
Reviewed-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/chip.c           | 1 +
 drivers/infiniband/hw/hfi1/chip.h           | 1 +
 drivers/infiniband/hw/hfi1/chip_registers.h | 1 +
 3 files changed, 3 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index 012f3331b8b4..e0b1238d31df 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -4114,6 +4114,7 @@ def_access_ibp_counter(rc_crwaits);
 static struct cntr_entry dev_cntrs[DEV_CNTR_LAST] = {
 [C_RCV_OVF] = RXE32_DEV_CNTR_ELEM(RcvOverflow, RCV_BUF_OVFL_CNT, CNTR_SYNTH),
 [C_RX_LEN_ERR] = RXE32_DEV_CNTR_ELEM(RxLenErr, RCV_LENGTH_ERR_CNT, CNTR_SYNTH),
+[C_RX_SHORT_ERR] = RXE32_DEV_CNTR_ELEM(RxShrErr, RCV_SHORT_ERR_CNT, CNTR_SYNTH),
 [C_RX_ICRC_ERR] = RXE32_DEV_CNTR_ELEM(RxICrcErr, RCV_ICRC_ERR_CNT, CNTR_SYNTH),
 [C_RX_EBP] = RXE32_DEV_CNTR_ELEM(RxEbpCnt, RCV_EBP_CNT, CNTR_SYNTH),
 [C_RX_TID_FULL] = RXE32_DEV_CNTR_ELEM(RxTIDFullEr, RCV_TID_FULL_ERR_CNT,
diff --git a/drivers/infiniband/hw/hfi1/chip.h b/drivers/infiniband/hw/hfi1/chip.h
index 93c9c8c0ac30..725509261016 100644
--- a/drivers/infiniband/hw/hfi1/chip.h
+++ b/drivers/infiniband/hw/hfi1/chip.h
@@ -865,6 +865,7 @@ static inline int idx_from_vl(int vl)
 enum {
 	C_RCV_OVF = 0,
 	C_RX_LEN_ERR,
+	C_RX_SHORT_ERR,
 	C_RX_ICRC_ERR,
 	C_RX_EBP,
 	C_RX_TID_FULL,
diff --git a/drivers/infiniband/hw/hfi1/chip_registers.h b/drivers/infiniband/hw/hfi1/chip_registers.h
index ab3589d17aee..fb3ec9bff7a2 100644
--- a/drivers/infiniband/hw/hfi1/chip_registers.h
+++ b/drivers/infiniband/hw/hfi1/chip_registers.h
@@ -381,6 +381,7 @@
 #define DC_LCB_STS_LINK_TRANSFER_ACTIVE (DC_LCB_CSRS + 0x000000000468)
 #define DC_LCB_STS_ROUND_TRIP_LTP_CNT (DC_LCB_CSRS + 0x0000000004B0)
 #define RCV_LENGTH_ERR_CNT 0
+#define RCV_SHORT_ERR_CNT 2
 #define RCV_ICRC_ERR_CNT 6
 #define RCV_EBP_CNT 9
 #define RCV_BUF_OVFL_CNT 10
-- 
cgit v1.2.3-59-g8ed1b


From 468d020e2f02867b8ec561461a1689cd4365e493 Mon Sep 17 00:00:00 2001
From: Lijun Ou <oulijun@huawei.com>
Date: Thu, 9 Jan 2020 20:10:52 +0800
Subject: RDMA/hns: Bugfix for posting a wqe with sge

Driver should first check whether the sge is valid, then fill the valid
sge and the caculated total into hardware, otherwise invalid sges will
cause an error.

Fixes: 52e3b42a2f58 ("RDMA/hns: Filter for zero length of sge in hip08 kernel mode")
Fixes: 7bdee4158b37 ("RDMA/hns: Fill sq wqe context of ud type in hip08")
Link: https://lore.kernel.org/r/1578571852-13704-1-git-send-email-liweihang@huawei.com
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 41 ++++++++++++++++++------------
 1 file changed, 25 insertions(+), 16 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 00fd40fd8380..fb16908652e7 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -105,7 +105,7 @@ static void set_atomic_seg(struct hns_roce_wqe_atomic_seg *aseg,
 }
 
 static void set_extend_sge(struct hns_roce_qp *qp, const struct ib_send_wr *wr,
-			   unsigned int *sge_ind)
+			   unsigned int *sge_ind, int valid_num_sge)
 {
 	struct hns_roce_v2_wqe_data_seg *dseg;
 	struct ib_sge *sg;
@@ -118,7 +118,7 @@ static void set_extend_sge(struct hns_roce_qp *qp, const struct ib_send_wr *wr,
 
 	if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC)
 		num_in_wqe = HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE;
-	extend_sge_num = wr->num_sge - num_in_wqe;
+	extend_sge_num = valid_num_sge - num_in_wqe;
 	sg = wr->sg_list + num_in_wqe;
 	shift = qp->hr_buf.page_shift;
 
@@ -154,14 +154,16 @@ static void set_extend_sge(struct hns_roce_qp *qp, const struct ib_send_wr *wr,
 static int set_rwqe_data_seg(struct ib_qp *ibqp, const struct ib_send_wr *wr,
 			     struct hns_roce_v2_rc_send_wqe *rc_sq_wqe,
 			     void *wqe, unsigned int *sge_ind,
+			     int valid_num_sge,
 			     const struct ib_send_wr **bad_wr)
 {
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device);
 	struct hns_roce_v2_wqe_data_seg *dseg = wqe;
 	struct hns_roce_qp *qp = to_hr_qp(ibqp);
+	int j = 0;
 	int i;
 
-	if (wr->send_flags & IB_SEND_INLINE && wr->num_sge) {
+	if (wr->send_flags & IB_SEND_INLINE && valid_num_sge) {
 		if (le32_to_cpu(rc_sq_wqe->msg_len) >
 		    hr_dev->caps.max_sq_inline) {
 			*bad_wr = wr;
@@ -185,7 +187,7 @@ static int set_rwqe_data_seg(struct ib_qp *ibqp, const struct ib_send_wr *wr,
 		roce_set_bit(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_INLINE_S,
 			     1);
 	} else {
-		if (wr->num_sge <= HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) {
+		if (valid_num_sge <= HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) {
 			for (i = 0; i < wr->num_sge; i++) {
 				if (likely(wr->sg_list[i].length)) {
 					set_data_seg_v2(dseg, wr->sg_list + i);
@@ -198,19 +200,21 @@ static int set_rwqe_data_seg(struct ib_qp *ibqp, const struct ib_send_wr *wr,
 				     V2_RC_SEND_WQE_BYTE_20_MSG_START_SGE_IDX_S,
 				     (*sge_ind) & (qp->sge.sge_cnt - 1));
 
-			for (i = 0; i < HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE; i++) {
+			for (i = 0; i < wr->num_sge &&
+			     j < HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE; i++) {
 				if (likely(wr->sg_list[i].length)) {
 					set_data_seg_v2(dseg, wr->sg_list + i);
 					dseg++;
+					j++;
 				}
 			}
 
-			set_extend_sge(qp, wr, sge_ind);
+			set_extend_sge(qp, wr, sge_ind, valid_num_sge);
 		}
 
 		roce_set_field(rc_sq_wqe->byte_16,
 			       V2_RC_SEND_WQE_BYTE_16_SGE_NUM_M,
-			       V2_RC_SEND_WQE_BYTE_16_SGE_NUM_S, wr->num_sge);
+			       V2_RC_SEND_WQE_BYTE_16_SGE_NUM_S, valid_num_sge);
 	}
 
 	return 0;
@@ -238,6 +242,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 	unsigned int sge_idx;
 	unsigned int wqe_idx;
 	unsigned long flags;
+	int valid_num_sge;
 	void *wqe = NULL;
 	bool loopback;
 	int attr_mask;
@@ -287,8 +292,16 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 		qp->sq.wrid[wqe_idx] = wr->wr_id;
 		owner_bit =
 		       ~(((qp->sq.head + nreq) >> ilog2(qp->sq.wqe_cnt)) & 0x1);
+		valid_num_sge = 0;
 		tmp_len = 0;
 
+		for (i = 0; i < wr->num_sge; i++) {
+			if (likely(wr->sg_list[i].length)) {
+				tmp_len += wr->sg_list[i].length;
+				valid_num_sge++;
+			}
+		}
+
 		/* Corresponding to the QP type, wqe process separately */
 		if (ibqp->qp_type == IB_QPT_GSI) {
 			ud_sq_wqe = wqe;
@@ -324,9 +337,6 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 				       V2_UD_SEND_WQE_BYTE_4_OPCODE_S,
 				       HNS_ROCE_V2_WQE_OP_SEND);
 
-			for (i = 0; i < wr->num_sge; i++)
-				tmp_len += wr->sg_list[i].length;
-
 			ud_sq_wqe->msg_len =
 			 cpu_to_le32(le32_to_cpu(ud_sq_wqe->msg_len) + tmp_len);
 
@@ -362,7 +372,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 			roce_set_field(ud_sq_wqe->byte_16,
 				       V2_UD_SEND_WQE_BYTE_16_SGE_NUM_M,
 				       V2_UD_SEND_WQE_BYTE_16_SGE_NUM_S,
-				       wr->num_sge);
+				       valid_num_sge);
 
 			roce_set_field(ud_sq_wqe->byte_20,
 				     V2_UD_SEND_WQE_BYTE_20_MSG_START_SGE_IDX_M,
@@ -417,12 +427,10 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 			memcpy(&ud_sq_wqe->dgid[0], &ah->av.dgid[0],
 			       GID_LEN_V2);
 
-			set_extend_sge(qp, wr, &sge_idx);
+			set_extend_sge(qp, wr, &sge_idx, valid_num_sge);
 		} else if (ibqp->qp_type == IB_QPT_RC) {
 			rc_sq_wqe = wqe;
 			memset(rc_sq_wqe, 0, sizeof(*rc_sq_wqe));
-			for (i = 0; i < wr->num_sge; i++)
-				tmp_len += wr->sg_list[i].length;
 
 			rc_sq_wqe->msg_len =
 			 cpu_to_le32(le32_to_cpu(rc_sq_wqe->msg_len) + tmp_len);
@@ -543,10 +551,11 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 				roce_set_field(rc_sq_wqe->byte_16,
 					       V2_RC_SEND_WQE_BYTE_16_SGE_NUM_M,
 					       V2_RC_SEND_WQE_BYTE_16_SGE_NUM_S,
-					       wr->num_sge);
+					       valid_num_sge);
 			} else if (wr->opcode != IB_WR_REG_MR) {
 				ret = set_rwqe_data_seg(ibqp, wr, rc_sq_wqe,
-							wqe, &sge_idx, bad_wr);
+							wqe, &sge_idx,
+							valid_num_sge, bad_wr);
 				if (ret)
 					goto out;
 			}
-- 
cgit v1.2.3-59-g8ed1b


From 626903e9355bdf8d401fc0ac7e7407862c642710 Mon Sep 17 00:00:00 2001
From: Xi Wang <wangxi11@huawei.com>
Date: Thu, 9 Jan 2020 20:20:12 +0800
Subject: RDMA/hns: Add support for reporting wc as software mode

When hardware is in resetting stage, we may can't poll back all the
expected work completions as the hardware won't generate cqe anymore.

This patch allows the driver to compose the expected wc instead of the
hardware during resetting stage. Once the hardware finished resetting, we
can poll cq from hardware again.

Link: https://lore.kernel.org/r/1578572412-25756-1-git-send-email-liweihang@huawei.com
Signed-off-by: Xi Wang <wangxi11@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_cq.c     |   2 +
 drivers/infiniband/hw/hns/hns_roce_device.h |  17 +++
 drivers/infiniband/hw/hns/hns_roce_hw_v1.c  |  14 ++-
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  | 158 +++++++++++++++++++++++-----
 drivers/infiniband/hw/hns/hns_roce_main.c   |  47 +++++++++
 drivers/infiniband/hw/hns/hns_roce_qp.c     |  48 ++++++++-
 6 files changed, 252 insertions(+), 34 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_cq.c b/drivers/infiniband/hw/hns/hns_roce_cq.c
index af1d8823b3f0..61f53a85767b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_cq.c
+++ b/drivers/infiniband/hw/hns/hns_roce_cq.c
@@ -370,6 +370,8 @@ int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr,
 	hr_cq->buf.size = hr_cq->cq_depth * hr_dev->caps.cq_entry_sz;
 	hr_cq->buf.page_shift = PAGE_SHIFT + hr_dev->caps.cqe_buf_pg_sz;
 	spin_lock_init(&hr_cq->lock);
+	INIT_LIST_HEAD(&hr_cq->sq_list);
+	INIT_LIST_HEAD(&hr_cq->rq_list);
 
 	if (udata) {
 		ret = create_user_cq(hr_dev, hr_cq, udata, &resp);
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 93c2210fa60e..8d2bcd883d00 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -498,6 +498,10 @@ struct hns_roce_cq {
 	u32				vector;
 	atomic_t			refcount;
 	struct completion		free;
+	struct list_head		sq_list; /* all qps on this send cq */
+	struct list_head		rq_list; /* all qps on this recv cq */
+	int				is_armed; /* cq is armed */
+	struct list_head		node; /* all armed cqs are on a list */
 };
 
 struct hns_roce_idx_que {
@@ -681,6 +685,9 @@ struct hns_roce_qp {
 	u32			next_sge;
 
 	struct hns_roce_rinl_buf rq_inl_buf;
+	struct list_head	node;		/* all qps are on a list */
+	struct list_head	rq_node;	/* all recv qps are on a list */
+	struct list_head	sq_node;	/* all send qps are on a list */
 };
 
 struct hns_roce_ib_iboe {
@@ -910,6 +917,12 @@ struct hns_roce_dfx_hw {
 			      int *buffer);
 };
 
+enum hns_roce_device_state {
+	HNS_ROCE_DEVICE_STATE_INITED,
+	HNS_ROCE_DEVICE_STATE_RST_DOWN,
+	HNS_ROCE_DEVICE_STATE_UNINIT,
+};
+
 struct hns_roce_hw {
 	int (*reset)(struct hns_roce_dev *hr_dev, bool enable);
 	int (*cmq_init)(struct hns_roce_dev *hr_dev);
@@ -992,6 +1005,9 @@ struct hns_roce_dev {
 	bool			dis_db;
 	unsigned long		reset_cnt;
 	struct hns_roce_ib_iboe iboe;
+	enum hns_roce_device_state state;
+	struct list_head	qp_list; /* list of all qps on this dev */
+	spinlock_t		qp_list_lock; /* protect qp_list */
 
 	struct list_head        pgdir_list;
 	struct mutex            pgdir_mutex;
@@ -1255,6 +1271,7 @@ void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type);
 void hns_roce_qp_event(struct hns_roce_dev *hr_dev, u32 qpn, int event_type);
 void hns_roce_srq_event(struct hns_roce_dev *hr_dev, u32 srqn, int event_type);
 int hns_get_gid_index(struct hns_roce_dev *hr_dev, u8 port, int gid_index);
+void hns_roce_handle_device_err(struct hns_roce_dev *hr_dev);
 int hns_roce_init(struct hns_roce_dev *hr_dev);
 void hns_roce_exit(struct hns_roce_dev *hr_dev);
 
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
index a31a21433f65..c6e66586e533 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
@@ -3609,14 +3609,18 @@ int hns_roce_v1_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
 	if (ret)
 		return ret;
 
-	send_cq = to_hr_cq(hr_qp->ibqp.send_cq);
-	recv_cq = to_hr_cq(hr_qp->ibqp.recv_cq);
+	send_cq = hr_qp->ibqp.send_cq ? to_hr_cq(hr_qp->ibqp.send_cq) : NULL;
+	recv_cq = hr_qp->ibqp.recv_cq ? to_hr_cq(hr_qp->ibqp.recv_cq) : NULL;
 
 	hns_roce_lock_cqs(send_cq, recv_cq);
 	if (!udata) {
-		__hns_roce_v1_cq_clean(recv_cq, hr_qp->qpn, hr_qp->ibqp.srq ?
-				       to_hr_srq(hr_qp->ibqp.srq) : NULL);
-		if (send_cq != recv_cq)
+		if (recv_cq)
+			__hns_roce_v1_cq_clean(recv_cq, hr_qp->qpn,
+					       (hr_qp->ibqp.srq ?
+						to_hr_srq(hr_qp->ibqp.srq) :
+						NULL));
+
+		if (send_cq && send_cq != recv_cq)
 			__hns_roce_v1_cq_clean(send_cq, hr_qp->qpn, NULL);
 	}
 	hns_roce_unlock_cqs(send_cq, recv_cq);
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index fb16908652e7..f1e0ba6a11d5 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -225,6 +225,30 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp,
 				 int attr_mask, enum ib_qp_state cur_state,
 				 enum ib_qp_state new_state);
 
+static int check_send_valid(struct hns_roce_dev *hr_dev,
+			    struct hns_roce_qp *hr_qp)
+{
+	struct ib_qp *ibqp = &hr_qp->ibqp;
+	struct device *dev = hr_dev->dev;
+
+	if (unlikely(ibqp->qp_type != IB_QPT_RC &&
+		     ibqp->qp_type != IB_QPT_GSI &&
+		     ibqp->qp_type != IB_QPT_UD)) {
+		dev_err(dev, "Not supported QP(0x%x)type!\n", ibqp->qp_type);
+		return -EOPNOTSUPP;
+	} else if (unlikely(hr_qp->state == IB_QPS_RESET ||
+		   hr_qp->state == IB_QPS_INIT ||
+		   hr_qp->state == IB_QPS_RTR)) {
+		dev_err(dev, "Post WQE fail, QP state %d!\n", hr_qp->state);
+		return -EINVAL;
+	} else if (unlikely(hr_dev->state >= HNS_ROCE_DEVICE_STATE_RST_DOWN)) {
+		dev_err(dev, "Post WQE fail, dev state %d!\n", hr_dev->state);
+		return -EIO;
+	}
+
+	return 0;
+}
+
 static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 				 const struct ib_send_wr *wr,
 				 const struct ib_send_wr **bad_wr)
@@ -247,28 +271,21 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 	bool loopback;
 	int attr_mask;
 	u32 tmp_len;
-	int ret = 0;
 	u32 hr_op;
 	u8 *smac;
 	int nreq;
+	int ret;
 	int i;
 
-	if (unlikely(ibqp->qp_type != IB_QPT_RC &&
-		     ibqp->qp_type != IB_QPT_GSI &&
-		     ibqp->qp_type != IB_QPT_UD)) {
-		dev_err(dev, "Not supported QP(0x%x)type!\n", ibqp->qp_type);
-		*bad_wr = wr;
-		return -EOPNOTSUPP;
-	}
+	spin_lock_irqsave(&qp->sq.lock, flags);
 
-	if (unlikely(qp->state == IB_QPS_RESET || qp->state == IB_QPS_INIT ||
-		     qp->state == IB_QPS_RTR)) {
-		dev_err(dev, "Post WQE fail, QP state %d err!\n", qp->state);
+	ret = check_send_valid(hr_dev, qp);
+	if (ret) {
 		*bad_wr = wr;
-		return -EINVAL;
+		nreq = 0;
+		goto out;
 	}
 
-	spin_lock_irqsave(&qp->sq.lock, flags);
 	sge_idx = qp->next_sge;
 
 	for (nreq = 0; wr; ++nreq, wr = wr->next) {
@@ -609,6 +626,17 @@ out:
 	return ret;
 }
 
+static int check_recv_valid(struct hns_roce_dev *hr_dev,
+			    struct hns_roce_qp *hr_qp)
+{
+	if (unlikely(hr_dev->state >= HNS_ROCE_DEVICE_STATE_RST_DOWN))
+		return -EIO;
+	else if (hr_qp->state == IB_QPS_RESET)
+		return -EINVAL;
+
+	return 0;
+}
+
 static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
 				 const struct ib_recv_wr *wr,
 				 const struct ib_recv_wr **bad_wr)
@@ -623,16 +651,17 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
 	void *wqe = NULL;
 	int attr_mask;
 	u32 wqe_idx;
-	int ret = 0;
 	int nreq;
+	int ret;
 	int i;
 
 	spin_lock_irqsave(&hr_qp->rq.lock, flags);
 
-	if (hr_qp->state == IB_QPS_RESET) {
-		spin_unlock_irqrestore(&hr_qp->rq.lock, flags);
+	ret = check_recv_valid(hr_dev, hr_qp);
+	if (ret) {
 		*bad_wr = wr;
-		return -EINVAL;
+		nreq = 0;
+		goto out;
 	}
 
 	for (nreq = 0; wr; ++nreq, wr = wr->next) {
@@ -2673,6 +2702,55 @@ static int hns_roce_handle_recv_inl_wqe(struct hns_roce_v2_cqe *cqe,
 	return 0;
 }
 
+static int sw_comp(struct hns_roce_qp *hr_qp, struct hns_roce_wq *wq,
+		   int num_entries, struct ib_wc *wc)
+{
+	unsigned int left;
+	int npolled = 0;
+
+	left = wq->head - wq->tail;
+	if (left == 0)
+		return 0;
+
+	left = min_t(unsigned int, (unsigned int)num_entries, left);
+	while (npolled < left) {
+		wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)];
+		wc->status = IB_WC_WR_FLUSH_ERR;
+		wc->vendor_err = 0;
+		wc->qp = &hr_qp->ibqp;
+
+		wq->tail++;
+		wc++;
+		npolled++;
+	}
+
+	return npolled;
+}
+
+static int hns_roce_v2_sw_poll_cq(struct hns_roce_cq *hr_cq, int num_entries,
+				  struct ib_wc *wc)
+{
+	struct hns_roce_qp *hr_qp;
+	int npolled = 0;
+
+	list_for_each_entry(hr_qp, &hr_cq->sq_list, sq_node) {
+		npolled += sw_comp(hr_qp, &hr_qp->sq,
+				   num_entries - npolled, wc + npolled);
+		if (npolled >= num_entries)
+			goto out;
+	}
+
+	list_for_each_entry(hr_qp, &hr_cq->rq_list, rq_node) {
+		npolled += sw_comp(hr_qp, &hr_qp->rq,
+				   num_entries - npolled, wc + npolled);
+		if (npolled >= num_entries)
+			goto out;
+	}
+
+out:
+	return npolled;
+}
+
 static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq,
 				struct hns_roce_qp **cur_qp, struct ib_wc *wc)
 {
@@ -2953,6 +3031,7 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq,
 static int hns_roce_v2_poll_cq(struct ib_cq *ibcq, int num_entries,
 			       struct ib_wc *wc)
 {
+	struct hns_roce_dev *hr_dev = to_hr_dev(ibcq->device);
 	struct hns_roce_cq *hr_cq = to_hr_cq(ibcq);
 	struct hns_roce_qp *cur_qp = NULL;
 	unsigned long flags;
@@ -2960,6 +3039,18 @@ static int hns_roce_v2_poll_cq(struct ib_cq *ibcq, int num_entries,
 
 	spin_lock_irqsave(&hr_cq->lock, flags);
 
+	/*
+	 * When the device starts to reset, the state is RST_DOWN. At this time,
+	 * there may still be some valid CQEs in the hardware that are not
+	 * polled. Therefore, it is not allowed to switch to the software mode
+	 * immediately. When the state changes to UNINIT, CQE no longer exists
+	 * in the hardware, and then switch to software mode.
+	 */
+	if (hr_dev->state == HNS_ROCE_DEVICE_STATE_UNINIT) {
+		npolled = hns_roce_v2_sw_poll_cq(hr_cq, num_entries, wc);
+		goto out;
+	}
+
 	for (npolled = 0; npolled < num_entries; ++npolled) {
 		if (hns_roce_v2_poll_one(hr_cq, &cur_qp, wc + npolled))
 			break;
@@ -2971,6 +3062,7 @@ static int hns_roce_v2_poll_cq(struct ib_cq *ibcq, int num_entries,
 		hns_roce_v2_cq_set_ci(hr_cq, hr_cq->cons_index);
 	}
 
+out:
 	spin_unlock_irqrestore(&hr_cq->lock, flags);
 
 	return npolled;
@@ -4629,6 +4721,7 @@ static int hns_roce_v2_destroy_qp_common(struct hns_roce_dev *hr_dev,
 {
 	struct hns_roce_cq *send_cq, *recv_cq;
 	struct ib_device *ibdev = &hr_dev->ib_dev;
+	unsigned long flags;
 	int ret = 0;
 
 	if (hr_qp->ibqp.qp_type == IB_QPT_RC && hr_qp->state != IB_QPS_RESET) {
@@ -4639,21 +4732,32 @@ static int hns_roce_v2_destroy_qp_common(struct hns_roce_dev *hr_dev,
 			ibdev_err(ibdev, "modify QP to Reset failed.\n");
 	}
 
-	send_cq = to_hr_cq(hr_qp->ibqp.send_cq);
-	recv_cq = to_hr_cq(hr_qp->ibqp.recv_cq);
+	send_cq = hr_qp->ibqp.send_cq ? to_hr_cq(hr_qp->ibqp.send_cq) : NULL;
+	recv_cq = hr_qp->ibqp.recv_cq ? to_hr_cq(hr_qp->ibqp.recv_cq) : NULL;
 
+	spin_lock_irqsave(&hr_dev->qp_list_lock, flags);
 	hns_roce_lock_cqs(send_cq, recv_cq);
 
+	list_del(&hr_qp->node);
+	list_del(&hr_qp->sq_node);
+	list_del(&hr_qp->rq_node);
+
 	if (!udata) {
-		__hns_roce_v2_cq_clean(recv_cq, hr_qp->qpn, hr_qp->ibqp.srq ?
-				       to_hr_srq(hr_qp->ibqp.srq) : NULL);
-		if (send_cq != recv_cq)
+		if (recv_cq)
+			__hns_roce_v2_cq_clean(recv_cq, hr_qp->qpn,
+					       (hr_qp->ibqp.srq ?
+						to_hr_srq(hr_qp->ibqp.srq) :
+						NULL));
+
+		if (send_cq && send_cq != recv_cq)
 			__hns_roce_v2_cq_clean(send_cq, hr_qp->qpn, NULL);
+
 	}
 
 	hns_roce_qp_remove(hr_dev, hr_qp);
 
 	hns_roce_unlock_cqs(send_cq, recv_cq);
+	spin_unlock_irqrestore(&hr_dev->qp_list_lock, flags);
 
 	hns_roce_qp_free(hr_dev, hr_qp);
 
@@ -6397,6 +6501,10 @@ static void __hns_roce_hw_v2_uninit_instance(struct hnae3_handle *handle,
 		return;
 
 	handle->priv = NULL;
+
+	hr_dev->state = HNS_ROCE_DEVICE_STATE_UNINIT;
+	hns_roce_handle_device_err(hr_dev);
+
 	hns_roce_exit(hr_dev);
 	kfree(hr_dev->priv);
 	ib_dealloc_device(&hr_dev->ib_dev);
@@ -6458,7 +6566,6 @@ static void hns_roce_hw_v2_uninit_instance(struct hnae3_handle *handle,
 static int hns_roce_hw_v2_reset_notify_down(struct hnae3_handle *handle)
 {
 	struct hns_roce_dev *hr_dev;
-	struct ib_event event;
 
 	if (handle->rinfo.instance_state != HNS_ROCE_STATE_INITED) {
 		set_bit(HNS_ROCE_RST_DIRECT_RETURN, &handle->rinfo.state);
@@ -6476,10 +6583,7 @@ static int hns_roce_hw_v2_reset_notify_down(struct hnae3_handle *handle)
 	hr_dev->active = false;
 	hr_dev->dis_db = true;
 
-	event.event = IB_EVENT_DEVICE_FATAL;
-	event.device = &hr_dev->ib_dev;
-	event.element.port_num = 1;
-	ib_dispatch_event(&event);
+	hr_dev->state = HNS_ROCE_DEVICE_STATE_RST_DOWN;
 
 	return 0;
 }
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 84e4707337a9..6e589f2ca35e 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -862,6 +862,50 @@ err_uar_table_free:
 	return ret;
 }
 
+static void check_and_get_armed_cq(struct list_head *cq_list, struct ib_cq *cq)
+{
+	struct hns_roce_cq *hr_cq = to_hr_cq(cq);
+	unsigned long flags;
+
+	spin_lock_irqsave(&hr_cq->lock, flags);
+	if (cq->comp_handler) {
+		if (!hr_cq->is_armed) {
+			hr_cq->is_armed = 1;
+			list_add_tail(&hr_cq->node, cq_list);
+		}
+	}
+	spin_unlock_irqrestore(&hr_cq->lock, flags);
+}
+
+void hns_roce_handle_device_err(struct hns_roce_dev *hr_dev)
+{
+	struct hns_roce_qp *hr_qp;
+	struct hns_roce_cq *hr_cq;
+	struct list_head cq_list;
+	unsigned long flags_qp;
+	unsigned long flags;
+
+	INIT_LIST_HEAD(&cq_list);
+
+	spin_lock_irqsave(&hr_dev->qp_list_lock, flags);
+	list_for_each_entry(hr_qp, &hr_dev->qp_list, node) {
+		spin_lock_irqsave(&hr_qp->sq.lock, flags_qp);
+		if (hr_qp->sq.tail != hr_qp->sq.head)
+			check_and_get_armed_cq(&cq_list, hr_qp->ibqp.send_cq);
+		spin_unlock_irqrestore(&hr_qp->sq.lock, flags_qp);
+
+		spin_lock_irqsave(&hr_qp->rq.lock, flags_qp);
+		if ((!hr_qp->ibqp.srq) && (hr_qp->rq.tail != hr_qp->rq.head))
+			check_and_get_armed_cq(&cq_list, hr_qp->ibqp.recv_cq);
+		spin_unlock_irqrestore(&hr_qp->rq.lock, flags_qp);
+	}
+
+	list_for_each_entry(hr_cq, &cq_list, node)
+		hns_roce_cq_completion(hr_dev, hr_cq->cqn);
+
+	spin_unlock_irqrestore(&hr_dev->qp_list_lock, flags);
+}
+
 int hns_roce_init(struct hns_roce_dev *hr_dev)
 {
 	int ret;
@@ -932,6 +976,9 @@ int hns_roce_init(struct hns_roce_dev *hr_dev)
 		}
 	}
 
+	INIT_LIST_HEAD(&hr_dev->qp_list);
+	spin_lock_init(&hr_dev->qp_list_lock);
+
 	ret = hns_roce_register_device(hr_dev);
 	if (ret)
 		goto error_failed_register_device;
diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c
index c5b01ec8ca6a..7c8de1e0d48a 100644
--- a/drivers/infiniband/hw/hns/hns_roce_qp.c
+++ b/drivers/infiniband/hw/hns/hns_roce_qp.c
@@ -677,6 +677,29 @@ static void free_rq_inline_buf(struct hns_roce_qp *hr_qp)
 	kfree(hr_qp->rq_inl_buf.wqe_list);
 }
 
+static void add_qp_to_list(struct hns_roce_dev *hr_dev,
+			   struct hns_roce_qp *hr_qp,
+			   struct ib_cq *send_cq, struct ib_cq *recv_cq)
+{
+	struct hns_roce_cq *hr_send_cq, *hr_recv_cq;
+	unsigned long flags;
+
+	hr_send_cq = send_cq ? to_hr_cq(send_cq) : NULL;
+	hr_recv_cq = recv_cq ? to_hr_cq(recv_cq) : NULL;
+
+	spin_lock_irqsave(&hr_dev->qp_list_lock, flags);
+	hns_roce_lock_cqs(hr_send_cq, hr_recv_cq);
+
+	list_add_tail(&hr_qp->node, &hr_dev->qp_list);
+	if (hr_send_cq)
+		list_add_tail(&hr_qp->sq_node, &hr_send_cq->sq_list);
+	if (hr_recv_cq)
+		list_add_tail(&hr_qp->rq_node, &hr_recv_cq->rq_list);
+
+	hns_roce_unlock_cqs(hr_send_cq, hr_recv_cq);
+	spin_unlock_irqrestore(&hr_dev->qp_list_lock, flags);
+}
+
 static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
 				     struct ib_pd *ib_pd,
 				     struct ib_qp_init_attr *init_attr,
@@ -946,6 +969,9 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
 	}
 
 	hr_qp->event = hns_roce_ib_qp_event;
+
+	add_qp_to_list(hr_dev, hr_qp, init_attr->send_cq, init_attr->recv_cq);
+
 	hns_roce_free_buf_list(buf_list, hr_qp->region_cnt);
 
 	return 0;
@@ -1228,7 +1254,16 @@ out:
 void hns_roce_lock_cqs(struct hns_roce_cq *send_cq, struct hns_roce_cq *recv_cq)
 		       __acquires(&send_cq->lock) __acquires(&recv_cq->lock)
 {
-	if (send_cq == recv_cq) {
+	if (unlikely(send_cq == NULL && recv_cq == NULL)) {
+		__acquire(&send_cq->lock);
+		__acquire(&recv_cq->lock);
+	} else if (unlikely(send_cq != NULL && recv_cq == NULL)) {
+		spin_lock_irq(&send_cq->lock);
+		__acquire(&recv_cq->lock);
+	} else if (unlikely(send_cq == NULL && recv_cq != NULL)) {
+		spin_lock_irq(&recv_cq->lock);
+		__acquire(&send_cq->lock);
+	} else if (send_cq == recv_cq) {
 		spin_lock_irq(&send_cq->lock);
 		__acquire(&recv_cq->lock);
 	} else if (send_cq->cqn < recv_cq->cqn) {
@@ -1244,7 +1279,16 @@ void hns_roce_unlock_cqs(struct hns_roce_cq *send_cq,
 			 struct hns_roce_cq *recv_cq) __releases(&send_cq->lock)
 			 __releases(&recv_cq->lock)
 {
-	if (send_cq == recv_cq) {
+	if (unlikely(send_cq == NULL && recv_cq == NULL)) {
+		__release(&recv_cq->lock);
+		__release(&send_cq->lock);
+	} else if (unlikely(send_cq != NULL && recv_cq == NULL)) {
+		__release(&recv_cq->lock);
+		spin_unlock(&send_cq->lock);
+	} else if (unlikely(send_cq == NULL && recv_cq != NULL)) {
+		__release(&send_cq->lock);
+		spin_unlock(&recv_cq->lock);
+	} else if (send_cq == recv_cq) {
 		__release(&recv_cq->lock);
 		spin_unlock_irq(&send_cq->lock);
 	} else if (send_cq->cqn < recv_cq->cqn) {
-- 
cgit v1.2.3-59-g8ed1b


From 1e123d96b85fd41278efc994f8096e1df86fd427 Mon Sep 17 00:00:00 2001
From: Guoqing Jiang <guoqing.jiang@cloud.ionos.com>
Date: Thu, 9 Jan 2020 14:40:43 +0100
Subject: RDMA/core: Remove err in iw_query_port

Since we can return device->ops.query_port directly, so no need to keep
those lines.

Link: https://lore.kernel.org/r/20200109134043.15568-1-guoqing.jiang@cloud.ionos.com
Signed-off-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/device.c | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index e4c3ea727eac..f6c255202d7f 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -1977,7 +1977,6 @@ static int iw_query_port(struct ib_device *device,
 {
 	struct in_device *inetdev;
 	struct net_device *netdev;
-	int err;
 
 	memset(port_attr, 0, sizeof(*port_attr));
 
@@ -2008,11 +2007,7 @@ static int iw_query_port(struct ib_device *device,
 	}
 
 	dev_put(netdev);
-	err = device->ops.query_port(device, port_num, port_attr);
-	if (err)
-		return err;
-
-	return 0;
+	return device->ops.query_port(device, port_num, port_attr);
 }
 
 static int __ib_query_port(struct ib_device *device,
-- 
cgit v1.2.3-59-g8ed1b


From f164be8c03663034416d019c355fbbd2dbd189d7 Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@mellanox.com>
Date: Thu, 12 Dec 2019 13:09:26 +0200
Subject: IB/mlx5: Extend caps stage to handle VAR capabilities

Extend caps stage to handle VAR capabilities.

Link: https://lore.kernel.org/r/20191212110928.334995-4-leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c    | 40 ++++++++++++++++++++++++++++++++++--
 drivers/infiniband/hw/mlx5/mlx5_ib.h | 10 +++++++++
 2 files changed, 48 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 05b557ddd0d8..e6aa836f1b55 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6335,6 +6335,35 @@ static const struct ib_device_ops mlx5_ib_dev_dm_ops = {
 	.reg_dm_mr = mlx5_ib_reg_dm_mr,
 };
 
+static int mlx5_ib_init_var_table(struct mlx5_ib_dev *dev)
+{
+	struct mlx5_core_dev *mdev = dev->mdev;
+	struct mlx5_var_table *var_table = &dev->var_table;
+	u8 log_doorbell_bar_size;
+	u8 log_doorbell_stride;
+	u64 bar_size;
+
+	log_doorbell_bar_size = MLX5_CAP_DEV_VDPA_EMULATION(mdev,
+					log_doorbell_bar_size);
+	log_doorbell_stride = MLX5_CAP_DEV_VDPA_EMULATION(mdev,
+					log_doorbell_stride);
+	var_table->hw_start_addr = dev->mdev->bar_addr +
+				MLX5_CAP64_DEV_VDPA_EMULATION(mdev,
+					doorbell_bar_offset);
+	bar_size = (1ULL << log_doorbell_bar_size) * 4096;
+	var_table->stride_size = 1ULL << log_doorbell_stride;
+	var_table->num_var_hw_entries = bar_size / var_table->stride_size;
+	mutex_init(&var_table->bitmap_lock);
+	var_table->bitmap = bitmap_zalloc(var_table->num_var_hw_entries,
+					  GFP_KERNEL);
+	return (var_table->bitmap) ? 0 : -ENOMEM;
+}
+
+static void mlx5_ib_stage_caps_cleanup(struct mlx5_ib_dev *dev)
+{
+	bitmap_free(dev->var_table.bitmap);
+}
+
 static int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev)
 {
 	struct mlx5_core_dev *mdev = dev->mdev;
@@ -6422,6 +6451,13 @@ static int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev)
 	     MLX5_CAP_GEN(dev->mdev, disable_local_lb_mc)))
 		mutex_init(&dev->lb.mutex);
 
+	if (MLX5_CAP_GEN_64(dev->mdev, general_obj_types) &
+			MLX5_GENERAL_OBJ_TYPES_CAP_VIRTIO_NET_Q) {
+		err = mlx5_ib_init_var_table(dev);
+		if (err)
+			return err;
+	}
+
 	dev->ib_dev.use_cq_dim = true;
 
 	return 0;
@@ -6772,7 +6808,7 @@ static const struct mlx5_ib_profile pf_profile = {
 		     mlx5_ib_stage_flow_db_cleanup),
 	STAGE_CREATE(MLX5_IB_STAGE_CAPS,
 		     mlx5_ib_stage_caps_init,
-		     NULL),
+		     mlx5_ib_stage_caps_cleanup),
 	STAGE_CREATE(MLX5_IB_STAGE_NON_DEFAULT_CB,
 		     mlx5_ib_stage_non_default_cb,
 		     NULL),
@@ -6829,7 +6865,7 @@ const struct mlx5_ib_profile raw_eth_profile = {
 		     mlx5_ib_stage_flow_db_cleanup),
 	STAGE_CREATE(MLX5_IB_STAGE_CAPS,
 		     mlx5_ib_stage_caps_init,
-		     NULL),
+		     mlx5_ib_stage_caps_cleanup),
 	STAGE_CREATE(MLX5_IB_STAGE_NON_DEFAULT_CB,
 		     mlx5_ib_stage_raw_eth_non_default_cb,
 		     NULL),
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 927948328184..d3ec11fb59a6 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -959,6 +959,15 @@ struct mlx5_devx_event_table {
 	struct xarray event_xa;
 };
 
+struct mlx5_var_table {
+	/* serialize updating the bitmap */
+	struct mutex bitmap_lock;
+	unsigned long *bitmap;
+	u64 hw_start_addr;
+	u32 stride_size;
+	u64 num_var_hw_entries;
+};
+
 struct mlx5_ib_dev {
 	struct ib_device		ib_dev;
 	struct mlx5_core_dev		*mdev;
@@ -1013,6 +1022,7 @@ struct mlx5_ib_dev {
 	struct mlx5_srq_table   srq_table;
 	struct mlx5_async_ctx   async_ctx;
 	struct mlx5_devx_event_table devx_event_table;
+	struct mlx5_var_table var_table;
 
 	struct xarray sig_mrs;
 };
-- 
cgit v1.2.3-59-g8ed1b


From 7be76bef320b1f1d1b8dc87d3d5a03f3a2421a43 Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@mellanox.com>
Date: Thu, 12 Dec 2019 13:09:27 +0200
Subject: IB/mlx5: Introduce VAR object and its alloc/destroy methods

Introduce VAR object and its alloc/destroy KABI methods. The internal
implementation uses the IB core API to manage mmap/munamp calls.

Link: https://lore.kernel.org/r/20191212110928.334995-5-leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c        | 157 +++++++++++++++++++++++++++++++
 drivers/infiniband/hw/mlx5/mlx5_ib.h     |   7 ++
 include/uapi/rdma/mlx5_user_ioctl_cmds.h |  17 ++++
 3 files changed, 181 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index e6aa836f1b55..433ed48425ad 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2078,6 +2078,7 @@ static void mlx5_ib_mmap_free(struct rdma_user_mmap_entry *entry)
 {
 	struct mlx5_user_mmap_entry *mentry = to_mmmap(entry);
 	struct mlx5_ib_dev *dev = to_mdev(entry->ucontext->device);
+	struct mlx5_var_table *var_table = &dev->var_table;
 	struct mlx5_ib_dm *mdm;
 
 	switch (mentry->mmap_flag) {
@@ -2087,6 +2088,12 @@ static void mlx5_ib_mmap_free(struct rdma_user_mmap_entry *entry)
 				       mdm->size);
 		kfree(mdm);
 		break;
+	case MLX5_IB_MMAP_TYPE_VAR:
+		mutex_lock(&var_table->bitmap_lock);
+		clear_bit(mentry->page_idx, var_table->bitmap);
+		mutex_unlock(&var_table->bitmap_lock);
+		kfree(mentry);
+		break;
 	default:
 		WARN_ON(true);
 	}
@@ -2255,6 +2262,15 @@ static int mlx5_ib_mmap_offset(struct mlx5_ib_dev *dev,
 	return ret;
 }
 
+static u64 mlx5_entry_to_mmap_offset(struct mlx5_user_mmap_entry *entry)
+{
+	u16 cmd = entry->rdma_entry.start_pgoff >> 16;
+	u16 index = entry->rdma_entry.start_pgoff & 0xFFFF;
+
+	return (((index >> 8) << 16) | (cmd << MLX5_IB_MMAP_CMD_SHIFT) |
+		(index & 0xFF)) << PAGE_SHIFT;
+}
+
 static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma)
 {
 	struct mlx5_ib_ucontext *context = to_mucontext(ibcontext);
@@ -6034,6 +6050,145 @@ static void mlx5_ib_cleanup_multiport_master(struct mlx5_ib_dev *dev)
 	mlx5_nic_vport_disable_roce(dev->mdev);
 }
 
+static int var_obj_cleanup(struct ib_uobject *uobject,
+			   enum rdma_remove_reason why,
+			   struct uverbs_attr_bundle *attrs)
+{
+	struct mlx5_user_mmap_entry *obj = uobject->object;
+
+	rdma_user_mmap_entry_remove(&obj->rdma_entry);
+	return 0;
+}
+
+static struct mlx5_user_mmap_entry *
+alloc_var_entry(struct mlx5_ib_ucontext *c)
+{
+	struct mlx5_user_mmap_entry *entry;
+	struct mlx5_var_table *var_table;
+	u32 page_idx;
+	int err;
+
+	var_table = &to_mdev(c->ibucontext.device)->var_table;
+	entry = kzalloc(sizeof(*entry), GFP_KERNEL);
+	if (!entry)
+		return ERR_PTR(-ENOMEM);
+
+	mutex_lock(&var_table->bitmap_lock);
+	page_idx = find_first_zero_bit(var_table->bitmap,
+				       var_table->num_var_hw_entries);
+	if (page_idx >= var_table->num_var_hw_entries) {
+		err = -ENOSPC;
+		mutex_unlock(&var_table->bitmap_lock);
+		goto end;
+	}
+
+	set_bit(page_idx, var_table->bitmap);
+	mutex_unlock(&var_table->bitmap_lock);
+
+	entry->address = var_table->hw_start_addr +
+				(page_idx * var_table->stride_size);
+	entry->page_idx = page_idx;
+	entry->mmap_flag = MLX5_IB_MMAP_TYPE_VAR;
+
+	err = rdma_user_mmap_entry_insert_range(
+		&c->ibucontext, &entry->rdma_entry, var_table->stride_size,
+		MLX5_IB_MMAP_OFFSET_START << 16,
+		(MLX5_IB_MMAP_OFFSET_END << 16) + (1UL << 16) - 1);
+	if (err)
+		goto err_insert;
+
+	return entry;
+
+err_insert:
+	mutex_lock(&var_table->bitmap_lock);
+	clear_bit(page_idx, var_table->bitmap);
+	mutex_unlock(&var_table->bitmap_lock);
+end:
+	kfree(entry);
+	return ERR_PTR(err);
+}
+
+static int UVERBS_HANDLER(MLX5_IB_METHOD_VAR_OBJ_ALLOC)(
+	struct uverbs_attr_bundle *attrs)
+{
+	struct ib_uobject *uobj = uverbs_attr_get_uobject(
+		attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_HANDLE);
+	struct mlx5_ib_ucontext *c;
+	struct mlx5_user_mmap_entry *entry;
+	u64 mmap_offset;
+	u32 length;
+	int err;
+
+	c = to_mucontext(ib_uverbs_get_ucontext(attrs));
+	if (IS_ERR(c))
+		return PTR_ERR(c);
+
+	entry = alloc_var_entry(c);
+	if (IS_ERR(entry))
+		return PTR_ERR(entry);
+
+	mmap_offset = mlx5_entry_to_mmap_offset(entry);
+	length = entry->rdma_entry.npages * PAGE_SIZE;
+	uobj->object = entry;
+
+	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_OFFSET,
+			     &mmap_offset, sizeof(mmap_offset));
+	if (err)
+		goto err;
+
+	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_PAGE_ID,
+			     &entry->page_idx, sizeof(entry->page_idx));
+	if (err)
+		goto err;
+
+	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_LENGTH,
+			     &length, sizeof(length));
+	if (err)
+		goto err;
+
+	return 0;
+
+err:
+	rdma_user_mmap_entry_remove(&entry->rdma_entry);
+	return err;
+}
+
+DECLARE_UVERBS_NAMED_METHOD(
+	MLX5_IB_METHOD_VAR_OBJ_ALLOC,
+	UVERBS_ATTR_IDR(MLX5_IB_ATTR_VAR_OBJ_ALLOC_HANDLE,
+			MLX5_IB_OBJECT_VAR,
+			UVERBS_ACCESS_NEW,
+			UA_MANDATORY),
+	UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_VAR_OBJ_ALLOC_PAGE_ID,
+			   UVERBS_ATTR_TYPE(u32),
+			   UA_MANDATORY),
+	UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_LENGTH,
+			   UVERBS_ATTR_TYPE(u32),
+			   UA_MANDATORY),
+	UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_OFFSET,
+			    UVERBS_ATTR_TYPE(u64),
+			    UA_MANDATORY));
+
+DECLARE_UVERBS_NAMED_METHOD_DESTROY(
+	MLX5_IB_METHOD_VAR_OBJ_DESTROY,
+	UVERBS_ATTR_IDR(MLX5_IB_ATTR_VAR_OBJ_DESTROY_HANDLE,
+			MLX5_IB_OBJECT_VAR,
+			UVERBS_ACCESS_DESTROY,
+			UA_MANDATORY));
+
+DECLARE_UVERBS_NAMED_OBJECT(MLX5_IB_OBJECT_VAR,
+			    UVERBS_TYPE_ALLOC_IDR(var_obj_cleanup),
+			    &UVERBS_METHOD(MLX5_IB_METHOD_VAR_OBJ_ALLOC),
+			    &UVERBS_METHOD(MLX5_IB_METHOD_VAR_OBJ_DESTROY));
+
+static bool var_is_supported(struct ib_device *device)
+{
+	struct mlx5_ib_dev *dev = to_mdev(device);
+
+	return (MLX5_CAP_GEN_64(dev->mdev, general_obj_types) &
+			MLX5_GENERAL_OBJ_TYPES_CAP_VIRTIO_NET_Q);
+}
+
 ADD_UVERBS_ATTRIBUTES_SIMPLE(
 	mlx5_ib_dm,
 	UVERBS_OBJECT_DM,
@@ -6064,6 +6219,8 @@ static const struct uapi_definition mlx5_ib_defs[] = {
 	UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION,
 				&mlx5_ib_flow_action),
 	UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_DM, &mlx5_ib_dm),
+	UAPI_DEF_CHAIN_OBJ_TREE_NAMED(MLX5_IB_OBJECT_VAR,
+				UAPI_DEF_IS_OBJ_SUPPORTED(var_is_supported)),
 	{}
 };
 
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index d3ec11fb59a6..676462c8909f 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -71,6 +71,11 @@
 
 #define MLX5_MKEY_PAGE_SHIFT_MASK __mlx5_mask(mkc, log_page_size)
 
+enum {
+	MLX5_IB_MMAP_OFFSET_START = 9,
+	MLX5_IB_MMAP_OFFSET_END = 255,
+};
+
 enum {
 	MLX5_IB_MMAP_CMD_SHIFT	= 8,
 	MLX5_IB_MMAP_CMD_MASK	= 0xff,
@@ -120,6 +125,7 @@ enum {
 
 enum mlx5_ib_mmap_type {
 	MLX5_IB_MMAP_TYPE_MEMIC = 1,
+	MLX5_IB_MMAP_TYPE_VAR = 2,
 };
 
 #define MLX5_LOG_SW_ICM_BLOCK_SIZE(dev)                                        \
@@ -563,6 +569,7 @@ struct mlx5_user_mmap_entry {
 	struct rdma_user_mmap_entry rdma_entry;
 	u8 mmap_flag;
 	u64 address;
+	u32 page_idx;
 };
 
 struct mlx5_ib_dm {
diff --git a/include/uapi/rdma/mlx5_user_ioctl_cmds.h b/include/uapi/rdma/mlx5_user_ioctl_cmds.h
index 20d88307f75f..afe7da6f2b8e 100644
--- a/include/uapi/rdma/mlx5_user_ioctl_cmds.h
+++ b/include/uapi/rdma/mlx5_user_ioctl_cmds.h
@@ -115,6 +115,22 @@ enum mlx5_ib_devx_obj_methods {
 	MLX5_IB_METHOD_DEVX_OBJ_ASYNC_QUERY,
 };
 
+enum mlx5_ib_var_alloc_attrs {
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_HANDLE = (1U << UVERBS_ID_NS_SHIFT),
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_OFFSET,
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_LENGTH,
+	MLX5_IB_ATTR_VAR_OBJ_ALLOC_PAGE_ID,
+};
+
+enum mlx5_ib_var_obj_destroy_attrs {
+	MLX5_IB_ATTR_VAR_OBJ_DESTROY_HANDLE = (1U << UVERBS_ID_NS_SHIFT),
+};
+
+enum mlx5_ib_var_obj_methods {
+	MLX5_IB_METHOD_VAR_OBJ_ALLOC = (1U << UVERBS_ID_NS_SHIFT),
+	MLX5_IB_METHOD_VAR_OBJ_DESTROY,
+};
+
 enum mlx5_ib_devx_umem_reg_attrs {
 	MLX5_IB_ATTR_DEVX_UMEM_REG_HANDLE = (1U << UVERBS_ID_NS_SHIFT),
 	MLX5_IB_ATTR_DEVX_UMEM_REG_ADDR,
@@ -156,6 +172,7 @@ enum mlx5_ib_objects {
 	MLX5_IB_OBJECT_FLOW_MATCHER,
 	MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD,
 	MLX5_IB_OBJECT_DEVX_ASYNC_EVENT_FD,
+	MLX5_IB_OBJECT_VAR,
 };
 
 enum mlx5_ib_flow_matcher_create_attrs {
-- 
cgit v1.2.3-59-g8ed1b


From 3f59b6c3e600f9665dcf5b8e566cd7b778f03045 Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@mellanox.com>
Date: Thu, 12 Dec 2019 13:09:28 +0200
Subject: IB/mlx5: Add mmap support for VAR

Add mmap support for VAR, it uses the 'offset' command mode with
involvement of IB core APIs to find the previously allocated mmap entry.

Link: https://lore.kernel.org/r/20191212110928.334995-6-leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 433ed48425ad..97bcf01960ae 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2253,7 +2253,10 @@ static int mlx5_ib_mmap_offset(struct mlx5_ib_dev *dev,
 
 	mentry = to_mmmap(entry);
 	pfn = (mentry->address >> PAGE_SHIFT);
-	prot = pgprot_writecombine(vma->vm_page_prot);
+	if (mentry->mmap_flag == MLX5_IB_MMAP_TYPE_VAR)
+		prot = pgprot_noncached(vma->vm_page_prot);
+	else
+		prot = pgprot_writecombine(vma->vm_page_prot);
 	ret = rdma_user_mmap_io(ucontext, vma, pfn,
 				entry->npages * PAGE_SIZE,
 				prot,
-- 
cgit v1.2.3-59-g8ed1b


From 8bdf9dd984c18375d1090ddeb1792511f619c5c1 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Mon, 13 Jan 2020 14:33:10 +0000
Subject: RDMA/uverbs: Remove needs_kfree_rcu from uverbs_obj_type_class

After device disassociation the uapi_objects are destroyed and freed,
however it is still possible that core code can be holding a kref on the
uobject. When it finally goes to uverbs_uobject_free() via the kref_put()
it can trigger a use-after-free on the uapi_object.

Since needs_kfree_rcu is a micro optimization that only benefits file
uobjects, just get rid of it. There is no harm in using kfree_rcu even if
it isn't required, and the number of involved objects is small.

Link: https://lore.kernel.org/r/20200113143306.GA28717@ziepe.ca
Signed-off-by: Michael Guralnik <michaelgur@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/rdma_core.c | 23 +----------------------
 include/rdma/uverbs_types.h         |  1 -
 2 files changed, 1 insertion(+), 23 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index 6c72773faf29..17bdbe38fdfa 100644
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -49,13 +49,7 @@ void uverbs_uobject_get(struct ib_uobject *uobject)
 
 static void uverbs_uobject_free(struct kref *ref)
 {
-	struct ib_uobject *uobj =
-		container_of(ref, struct ib_uobject, ref);
-
-	if (uobj->uapi_object->type_class->needs_kfree_rcu)
-		kfree_rcu(uobj, rcu);
-	else
-		kfree(uobj);
+	kfree_rcu(container_of(ref, struct ib_uobject, ref), rcu);
 }
 
 void uverbs_uobject_put(struct ib_uobject *uobject)
@@ -744,20 +738,6 @@ const struct uverbs_obj_type_class uverbs_idr_class = {
 	.lookup_put = lookup_put_idr_uobject,
 	.destroy_hw = destroy_hw_idr_uobject,
 	.remove_handle = remove_handle_idr_uobject,
-	/*
-	 * When we destroy an object, we first just lock it for WRITE and
-	 * actually DESTROY it in the finalize stage. So, the problematic
-	 * scenario is when we just started the finalize stage of the
-	 * destruction (nothing was executed yet). Now, the other thread
-	 * fetched the object for READ access, but it didn't lock it yet.
-	 * The DESTROY thread continues and starts destroying the object.
-	 * When the other thread continue - without the RCU, it would
-	 * access freed memory. However, the rcu_read_lock delays the free
-	 * until the rcu_read_lock of the READ operation quits. Since the
-	 * exclusive lock of the object is still taken by the DESTROY flow, the
-	 * READ operation will get -EBUSY and it'll just bail out.
-	 */
-	.needs_kfree_rcu = true,
 };
 EXPORT_SYMBOL(uverbs_idr_class);
 
@@ -920,7 +900,6 @@ const struct uverbs_obj_type_class uverbs_fd_class = {
 	.lookup_put = lookup_put_fd_uobject,
 	.destroy_hw = destroy_hw_fd_uobject,
 	.remove_handle = remove_handle_fd_uobject,
-	.needs_kfree_rcu = false,
 };
 EXPORT_SYMBOL(uverbs_fd_class);
 
diff --git a/include/rdma/uverbs_types.h b/include/rdma/uverbs_types.h
index d57a5ba00c74..0b0f5a5f392d 100644
--- a/include/rdma/uverbs_types.h
+++ b/include/rdma/uverbs_types.h
@@ -98,7 +98,6 @@ struct uverbs_obj_type_class {
 				       enum rdma_remove_reason why,
 				       struct uverbs_attr_bundle *attrs);
 	void (*remove_handle)(struct ib_uobject *uobj);
-	u8    needs_kfree_rcu;
 };
 
 struct uverbs_obj_type {
-- 
cgit v1.2.3-59-g8ed1b


From 6898d1c661d79f4707d8ba82991b2195822780ca Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:21:53 +0200
Subject: RDMA/mlx5: Use RCU and direct refcounts to keep memory alive

dispatch_event_fd() runs from a notifier with minimal locking, and relies
on RCU and a file refcount to keep the uobject and eventfd alive.

As the next patch wants to remove the file_operations release function
from the drivers, re-organize things so that the devx_event_notifier()
path uses the existing RCU to manage the lifetime of the uobject and
eventfd.

Move the refcount puts to a call_rcu so that the objects are guaranteed to
exist and remove the indirect file refcount.

Link: https://lore.kernel.org/r/1578504126-9400-2-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/rdma_core.c | 11 ++++++-----
 drivers/infiniband/core/rdma_core.h | 15 ---------------
 drivers/infiniband/hw/mlx5/devx.c   | 34 +++++++++++++++++-----------------
 include/rdma/uverbs_types.h         | 12 ++++++++++++
 4 files changed, 35 insertions(+), 37 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index 17bdbe38fdfa..aef6fb823206 100644
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -42,20 +42,21 @@
 #include "core_priv.h"
 #include "rdma_core.h"
 
-void uverbs_uobject_get(struct ib_uobject *uobject)
-{
-	kref_get(&uobject->ref);
-}
-
 static void uverbs_uobject_free(struct kref *ref)
 {
 	kfree_rcu(container_of(ref, struct ib_uobject, ref), rcu);
 }
 
+/*
+ * In order to indicate we no longer needs this uobject, uverbs_uobject_put
+ * is called. When the reference count is decreased, the uobject is freed.
+ * For example, this is used when attaching a completion channel to a CQ.
+ */
 void uverbs_uobject_put(struct ib_uobject *uobject)
 {
 	kref_put(&uobject->ref, uverbs_uobject_free);
 }
+EXPORT_SYMBOL(uverbs_uobject_put);
 
 static int uverbs_try_lock_object(struct ib_uobject *uobj,
 				  enum rdma_lookup_mode mode)
diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h
index e63fbda25e1d..d5d58a10bb28 100644
--- a/drivers/infiniband/core/rdma_core.h
+++ b/drivers/infiniband/core/rdma_core.h
@@ -50,21 +50,6 @@ void uverbs_destroy_ufile_hw(struct ib_uverbs_file *ufile,
 
 int uobj_destroy(struct ib_uobject *uobj, struct uverbs_attr_bundle *attrs);
 
-/*
- * uverbs_uobject_get is called in order to increase the reference count on
- * an uobject. This is useful when a handler wants to keep the uobject's memory
- * alive, regardless if this uobject is still alive in the context's objects
- * repository. Objects are put via uverbs_uobject_put.
- */
-void uverbs_uobject_get(struct ib_uobject *uobject);
-
-/*
- * In order to indicate we no longer needs this uobject, uverbs_uobject_put
- * is called. When the reference count is decreased, the uobject is freed.
- * For example, this is used when attaching a completion channel to a CQ.
- */
-void uverbs_uobject_put(struct ib_uobject *uobject);
-
 /* Indicate this fd is no longer used by this consumer, but its memory isn't
  * necessarily released yet. When the last reference is put, we release the
  * memory. After this call is executed, calling uverbs_uobject_get isn't
diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c
index 9d0a18cf9e5e..968fff018fc0 100644
--- a/drivers/infiniband/hw/mlx5/devx.c
+++ b/drivers/infiniband/hw/mlx5/devx.c
@@ -72,7 +72,6 @@ struct devx_event_subscription {
 	struct rcu_head	rcu;
 	u64 cookie;
 	struct devx_async_event_file *ev_file;
-	struct file *filp; /* Upon hot unplug we need a direct access to */
 	struct eventfd_ctx *eventfd;
 };
 
@@ -2032,6 +2031,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_SUBSCRIBE_EVENT)(
 			goto err;
 
 		list_add_tail(&event_sub->event_list, &sub_list);
+		uverbs_uobject_get(&ev_file->uobj);
 		if (use_eventfd) {
 			event_sub->eventfd =
 				eventfd_ctx_fdget(redirect_fd);
@@ -2045,7 +2045,6 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_SUBSCRIBE_EVENT)(
 
 		event_sub->cookie = cookie;
 		event_sub->ev_file = ev_file;
-		event_sub->filp = fd_uobj->object;
 		/* May be needed upon cleanup the devx object/subscription */
 		event_sub->xa_key_level1 = key_level1;
 		event_sub->xa_key_level2 = obj_id;
@@ -2099,7 +2098,7 @@ err:
 
 		if (event_sub->eventfd)
 			eventfd_ctx_put(event_sub->eventfd);
-
+		uverbs_uobject_put(&event_sub->ev_file->uobj);
 		kfree(event_sub);
 	}
 
@@ -2361,17 +2360,10 @@ static void dispatch_event_fd(struct list_head *fd_list,
 	struct devx_event_subscription *item;
 
 	list_for_each_entry_rcu(item, fd_list, xa_list) {
-		if (!get_file_rcu(item->filp))
-			continue;
-
-		if (item->eventfd) {
+		if (item->eventfd)
 			eventfd_signal(item->eventfd, 1);
-			fput(item->filp);
-			continue;
-		}
-
-		deliver_event(item, data);
-		fput(item->filp);
+		else
+			deliver_event(item, data);
 	}
 }
 
@@ -2653,6 +2645,17 @@ static __poll_t devx_async_event_poll(struct file *filp,
 	return pollflags;
 }
 
+static void devx_free_subscription(struct rcu_head *rcu)
+{
+	struct devx_event_subscription *event_sub =
+		container_of(rcu, struct devx_event_subscription, rcu);
+
+	if (event_sub->eventfd)
+		eventfd_ctx_put(event_sub->eventfd);
+	uverbs_uobject_put(&event_sub->ev_file->uobj);
+	kfree(event_sub);
+}
+
 static int devx_async_event_close(struct inode *inode, struct file *filp)
 {
 	struct devx_async_event_file *ev_file = filp->private_data;
@@ -2665,12 +2668,9 @@ static int devx_async_event_close(struct inode *inode, struct file *filp)
 	list_for_each_entry_safe(event_sub, event_sub_tmp,
 				 &ev_file->subscribed_events_list, file_list) {
 		devx_cleanup_subscription(dev, event_sub);
-		if (event_sub->eventfd)
-			eventfd_ctx_put(event_sub->eventfd);
-
 		list_del_rcu(&event_sub->file_list);
 		/* subscription may not be used by the read API any more */
-		kfree_rcu(event_sub, rcu);
+		call_rcu(&event_sub->rcu, devx_free_subscription);
 	}
 
 	mutex_unlock(&dev->devx_event_table.event_xa_lock);
diff --git a/include/rdma/uverbs_types.h b/include/rdma/uverbs_types.h
index 0b0f5a5f392d..ca65d4704b4b 100644
--- a/include/rdma/uverbs_types.h
+++ b/include/rdma/uverbs_types.h
@@ -144,6 +144,18 @@ void rdma_alloc_abort_uobject(struct ib_uobject *uobj,
 int __must_check rdma_alloc_commit_uobject(struct ib_uobject *uobj,
 					   struct uverbs_attr_bundle *attrs);
 
+/*
+ * uverbs_uobject_get is called in order to increase the reference count on
+ * an uobject. This is useful when a handler wants to keep the uobject's memory
+ * alive, regardless if this uobject is still alive in the context's objects
+ * repository. Objects are put via uverbs_uobject_put.
+ */
+static inline void uverbs_uobject_get(struct ib_uobject *uobject)
+{
+	kref_get(&uobject->ref);
+}
+void uverbs_uobject_put(struct ib_uobject *uobject);
+
 struct uverbs_obj_fd_type {
 	/*
 	 * In fd based objects, uverbs_obj_type_ops points to generic
-- 
cgit v1.2.3-59-g8ed1b


From f7c8416ccea52b41e29227b3a5066540f51ee471 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:21:54 +0200
Subject: RDMA/core: Simplify destruction of FD uobjects

FD uobjects have a weird split between the struct file and uobject
world. Simplify this to make them pure uobjects and use a generic release
method for all struct file operations.

This fixes the control flow so that mlx5_cmd_cleanup_async_ctx() is always
called before erasing the linked list contents to make the concurrancy
simpler to understand.

For this to work the uobject destruction must fence anything that it is
cleaning up - the design must not rely on struct file lifetime.

Only deliver_event() relies on the struct file to when adding new events
to the queue, add a is_destroyed check under lock to block it.

Link: https://lore.kernel.org/r/1578504126-9400-3-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/rdma_core.c        |  34 +++++----
 drivers/infiniband/core/rdma_core.h        |   8 ---
 drivers/infiniband/core/uverbs_main.c      |  23 +-----
 drivers/infiniband/core/uverbs_std_types.c |  23 +++---
 drivers/infiniband/core/uverbs_uapi.c      |   6 +-
 drivers/infiniband/hw/mlx5/devx.c          | 111 +++++++++++++----------------
 include/rdma/uverbs_types.h                |  12 ++--
 7 files changed, 94 insertions(+), 123 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index aef6fb823206..0ed0341b8e30 100644
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -353,9 +353,9 @@ lookup_get_fd_uobject(const struct uverbs_api_object *obj,
 
 	uobject = f->private_data;
 	/*
-	 * fget(id) ensures we are not currently running uverbs_close_fd,
-	 * and the caller is expected to ensure that uverbs_close_fd is never
-	 * done while a call top lookup is possible.
+	 * fget(id) ensures we are not currently running
+	 * uverbs_uobject_fd_release(), and the caller is expected to ensure
+	 * that release is never done while a call to lookup is possible.
 	 */
 	if (f->f_op != fd_type->fops) {
 		fput(f);
@@ -548,7 +548,7 @@ static int __must_check destroy_hw_fd_uobject(struct ib_uobject *uobj,
 {
 	const struct uverbs_obj_fd_type *fd_type = container_of(
 		uobj->uapi_object->type_attrs, struct uverbs_obj_fd_type, type);
-	int ret = fd_type->context_closed(uobj, why);
+	int ret = fd_type->destroy_object(uobj, why);
 
 	if (ib_is_destroy_retryable(ret, why, uobj))
 		return ret;
@@ -587,9 +587,9 @@ static int alloc_commit_fd_uobject(struct ib_uobject *uobj)
 
 	/*
 	 * The kref for uobj is moved into filp->private data and put in
-	 * uverbs_close_fd(). Once alloc_commit() succeeds uverbs_close_fd()
-	 * must be guaranteed to be called from the provided fops release
-	 * callback.
+	 * uverbs_close_fd(). Once alloc_commit() succeeds
+	 * uverbs_uobject_fd_release() must be guaranteed to be called from
+	 * the provided fops release callback.
 	 */
 	filp = anon_inode_getfile(fd_type->name,
 				  fd_type->fops,
@@ -600,7 +600,7 @@ static int alloc_commit_fd_uobject(struct ib_uobject *uobj)
 
 	uobj->object = filp;
 
-	/* Matching put will be done in uverbs_close_fd() */
+	/* Matching put will be done in uverbs_uobject_fd_release() */
 	kref_get(&uobj->ufile->ref);
 
 	/* This shouldn't be used anymore. Use the file object instead */
@@ -608,7 +608,7 @@ static int alloc_commit_fd_uobject(struct ib_uobject *uobj)
 
 	/*
 	 * NOTE: Once we install the file we loose ownership of our kref on
-	 * uobj. It will be put by uverbs_close_fd()
+	 * uobj. It will be put by uverbs_uobject_fd_release()
 	 */
 	fd_install(fd, filp);
 
@@ -676,7 +676,10 @@ static void lookup_put_fd_uobject(struct ib_uobject *uobj,
 	struct file *filp = uobj->object;
 
 	WARN_ON(mode != UVERBS_LOOKUP_READ);
-	/* This indirectly calls uverbs_close_fd and free the object */
+	/*
+	 * This indirectly calls uverbs_uobject_fd_release() and free the
+	 * object
+	 */
 	fput(filp);
 }
 
@@ -742,9 +745,13 @@ const struct uverbs_obj_type_class uverbs_idr_class = {
 };
 EXPORT_SYMBOL(uverbs_idr_class);
 
-void uverbs_close_fd(struct file *f)
+/*
+ * Users of UVERBS_TYPE_ALLOC_FD should set this function as the struct
+ * file_operations release method.
+ */
+int uverbs_uobject_fd_release(struct inode *inode, struct file *filp)
 {
-	struct ib_uobject *uobj = f->private_data;
+	struct ib_uobject *uobj = filp->private_data;
 	struct ib_uverbs_file *ufile = uobj->ufile;
 	struct uverbs_attr_bundle attrs = {
 		.context = uobj->context,
@@ -768,8 +775,9 @@ void uverbs_close_fd(struct file *f)
 
 	/* Pairs with filp->private_data in alloc_begin_fd_uobject */
 	uverbs_uobject_put(uobj);
+	return 0;
 }
-EXPORT_SYMBOL(uverbs_close_fd);
+EXPORT_SYMBOL(uverbs_uobject_fd_release);
 
 /*
  * Drop the ucontext off the ufile and completely disconnect it from the
diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h
index d5d58a10bb28..92694253e776 100644
--- a/drivers/infiniband/core/rdma_core.h
+++ b/drivers/infiniband/core/rdma_core.h
@@ -50,14 +50,6 @@ void uverbs_destroy_ufile_hw(struct ib_uverbs_file *ufile,
 
 int uobj_destroy(struct ib_uobject *uobj, struct uverbs_attr_bundle *attrs);
 
-/* Indicate this fd is no longer used by this consumer, but its memory isn't
- * necessarily released yet. When the last reference is put, we release the
- * memory. After this call is executed, calling uverbs_uobject_get isn't
- * allowed.
- * This must be called from the release file_operations of the file!
- */
-void uverbs_close_fd(struct file *f);
-
 /*
  * Get an ib_uobject that corresponds to the given id from ufile, assuming
  * the object is from the given type. Lock it to the required access when
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 8f5de4dcad97..da56fda259fd 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -373,32 +373,11 @@ static int ib_uverbs_async_event_close(struct inode *inode, struct file *filp)
 	return 0;
 }
 
-static int ib_uverbs_comp_event_close(struct inode *inode, struct file *filp)
-{
-	struct ib_uobject *uobj = filp->private_data;
-	struct ib_uverbs_completion_event_file *file = container_of(
-		uobj, struct ib_uverbs_completion_event_file, uobj);
-	struct ib_uverbs_event *entry, *tmp;
-
-	spin_lock_irq(&file->ev_queue.lock);
-	list_for_each_entry_safe(entry, tmp, &file->ev_queue.event_list, list) {
-		if (entry->counter)
-			list_del(&entry->obj_list);
-		kfree(entry);
-	}
-	file->ev_queue.is_closed = 1;
-	spin_unlock_irq(&file->ev_queue.lock);
-
-	uverbs_close_fd(filp);
-
-	return 0;
-}
-
 const struct file_operations uverbs_event_fops = {
 	.owner	 = THIS_MODULE,
 	.read	 = ib_uverbs_comp_event_read,
 	.poll    = ib_uverbs_comp_event_poll,
-	.release = ib_uverbs_comp_event_close,
+	.release = uverbs_uobject_fd_release,
 	.fasync  = ib_uverbs_comp_event_fasync,
 	.llseek	 = no_llseek,
 };
diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c
index 35b2e2c640cc..def038a0fe77 100644
--- a/drivers/infiniband/core/uverbs_std_types.c
+++ b/drivers/infiniband/core/uverbs_std_types.c
@@ -202,22 +202,29 @@ static int uverbs_free_pd(struct ib_uobject *uobject,
 	return 0;
 }
 
-static int uverbs_hot_unplug_completion_event_file(struct ib_uobject *uobj,
-						   enum rdma_remove_reason why)
+static int
+uverbs_completion_event_file_destroy_uobj(struct ib_uobject *uobj,
+					  enum rdma_remove_reason why)
 {
-	struct ib_uverbs_completion_event_file *comp_event_file =
+	struct ib_uverbs_completion_event_file *file =
 		container_of(uobj, struct ib_uverbs_completion_event_file,
 			     uobj);
-	struct ib_uverbs_event_queue *event_queue = &comp_event_file->ev_queue;
+	struct ib_uverbs_event_queue *event_queue = &file->ev_queue;
+	struct ib_uverbs_event *entry, *tmp;
 
 	spin_lock_irq(&event_queue->lock);
 	event_queue->is_closed = 1;
 	spin_unlock_irq(&event_queue->lock);
+	wake_up_interruptible(&event_queue->poll_wait);
+	kill_fasync(&event_queue->async_queue, SIGIO, POLL_IN);
 
-	if (why == RDMA_REMOVE_DRIVER_REMOVE) {
-		wake_up_interruptible(&event_queue->poll_wait);
-		kill_fasync(&event_queue->async_queue, SIGIO, POLL_IN);
+	spin_lock_irq(&event_queue->lock);
+	list_for_each_entry_safe(entry, tmp, &event_queue->event_list, list) {
+		if (entry->counter)
+			list_del(&entry->obj_list);
+		kfree(entry);
 	}
+	spin_unlock_irq(&event_queue->lock);
 	return 0;
 };
 
@@ -230,7 +237,7 @@ EXPORT_SYMBOL(uverbs_destroy_def_handler);
 DECLARE_UVERBS_NAMED_OBJECT(
 	UVERBS_OBJECT_COMP_CHANNEL,
 	UVERBS_TYPE_ALLOC_FD(sizeof(struct ib_uverbs_completion_event_file),
-			     uverbs_hot_unplug_completion_event_file,
+			     uverbs_completion_event_file_destroy_uobj,
 			     &uverbs_event_fops,
 			     "[infinibandevent]",
 			     O_RDONLY));
diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c
index 00c547887132..9b84a126187a 100644
--- a/drivers/infiniband/core/uverbs_uapi.c
+++ b/drivers/infiniband/core/uverbs_uapi.c
@@ -195,9 +195,9 @@ static int uapi_merge_obj_tree(struct uverbs_api *uapi,
 		 * disassociation, and the FD types require the driver to use
 		 * struct file_operations.owner to prevent the driver module
 		 * code from unloading while the file is open. This provides
-		 * enough safety that uverbs_close_fd() will continue to work.
-		 * Drivers using FD are responsible to handle disassociation of
-		 * the device on their own.
+		 * enough safety that uverbs_uobject_fd_release() will
+		 * continue to work.  Drivers using FD are responsible to
+		 * handle disassociation of the device on their own.
 		 */
 		if (WARN_ON(is_driver &&
 			    obj->type_attrs->type_class != &uverbs_idr_class &&
diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c
index 968fff018fc0..02125d82b96d 100644
--- a/drivers/infiniband/hw/mlx5/devx.c
+++ b/drivers/infiniband/hw/mlx5/devx.c
@@ -2328,6 +2328,9 @@ static int deliver_event(struct devx_event_subscription *event_sub,
 			return 0;
 		}
 
+		/* is_destroyed is ignored here because we don't have any memory
+		 * allocation to clean up for the omit_data case
+		 */
 		list_add_tail(&event_sub->event_list, &ev_file->event_list);
 		spin_unlock_irqrestore(&ev_file->lock, flags);
 		wake_up_interruptible(&ev_file->poll_wait);
@@ -2347,7 +2350,10 @@ static int deliver_event(struct devx_event_subscription *event_sub,
 	memcpy(event_data->hdr.out_data, data, sizeof(struct mlx5_eqe));
 
 	spin_lock_irqsave(&ev_file->lock, flags);
-	list_add_tail(&event_data->list, &ev_file->event_list);
+	if (!ev_file->is_destroyed)
+		list_add_tail(&event_data->list, &ev_file->event_list);
+	else
+		kfree(event_data);
 	spin_unlock_irqrestore(&ev_file->lock, flags);
 	wake_up_interruptible(&ev_file->poll_wait);
 
@@ -2501,23 +2507,6 @@ static ssize_t devx_async_cmd_event_read(struct file *filp, char __user *buf,
 	return ret;
 }
 
-static int devx_async_cmd_event_close(struct inode *inode, struct file *filp)
-{
-	struct ib_uobject *uobj = filp->private_data;
-	struct devx_async_cmd_event_file *comp_ev_file = container_of(
-		uobj, struct devx_async_cmd_event_file, uobj);
-	struct devx_async_data *entry, *tmp;
-
-	spin_lock_irq(&comp_ev_file->ev_queue.lock);
-	list_for_each_entry_safe(entry, tmp,
-				 &comp_ev_file->ev_queue.event_list, list)
-		kvfree(entry);
-	spin_unlock_irq(&comp_ev_file->ev_queue.lock);
-
-	uverbs_close_fd(filp);
-	return 0;
-}
-
 static __poll_t devx_async_cmd_event_poll(struct file *filp,
 					      struct poll_table_struct *wait)
 {
@@ -2541,7 +2530,7 @@ static const struct file_operations devx_async_cmd_event_fops = {
 	.owner	 = THIS_MODULE,
 	.read	 = devx_async_cmd_event_read,
 	.poll    = devx_async_cmd_event_poll,
-	.release = devx_async_cmd_event_close,
+	.release = uverbs_uobject_fd_release,
 	.llseek	 = no_llseek,
 };
 
@@ -2656,78 +2645,74 @@ static void devx_free_subscription(struct rcu_head *rcu)
 	kfree(event_sub);
 }
 
-static int devx_async_event_close(struct inode *inode, struct file *filp)
-{
-	struct devx_async_event_file *ev_file = filp->private_data;
-	struct devx_event_subscription *event_sub, *event_sub_tmp;
-	struct devx_async_event_data *entry, *tmp;
-	struct mlx5_ib_dev *dev = ev_file->dev;
-
-	mutex_lock(&dev->devx_event_table.event_xa_lock);
-	/* delete the subscriptions which are related to this FD */
-	list_for_each_entry_safe(event_sub, event_sub_tmp,
-				 &ev_file->subscribed_events_list, file_list) {
-		devx_cleanup_subscription(dev, event_sub);
-		list_del_rcu(&event_sub->file_list);
-		/* subscription may not be used by the read API any more */
-		call_rcu(&event_sub->rcu, devx_free_subscription);
-	}
-
-	mutex_unlock(&dev->devx_event_table.event_xa_lock);
-
-	/* free the pending events allocation */
-	if (!ev_file->omit_data) {
-		spin_lock_irq(&ev_file->lock);
-		list_for_each_entry_safe(entry, tmp,
-					 &ev_file->event_list, list)
-			kfree(entry); /* read can't come any more */
-		spin_unlock_irq(&ev_file->lock);
-	}
-
-	uverbs_close_fd(filp);
-	put_device(&dev->ib_dev.dev);
-	return 0;
-}
-
 static const struct file_operations devx_async_event_fops = {
 	.owner	 = THIS_MODULE,
 	.read	 = devx_async_event_read,
 	.poll    = devx_async_event_poll,
-	.release = devx_async_event_close,
+	.release = uverbs_uobject_fd_release,
 	.llseek	 = no_llseek,
 };
 
-static int devx_hot_unplug_async_cmd_event_file(struct ib_uobject *uobj,
-						   enum rdma_remove_reason why)
+static int devx_async_cmd_event_destroy_uobj(struct ib_uobject *uobj,
+					     enum rdma_remove_reason why)
 {
 	struct devx_async_cmd_event_file *comp_ev_file =
 		container_of(uobj, struct devx_async_cmd_event_file,
 			     uobj);
 	struct devx_async_event_queue *ev_queue = &comp_ev_file->ev_queue;
+	struct devx_async_data *entry, *tmp;
 
 	spin_lock_irq(&ev_queue->lock);
 	ev_queue->is_destroyed = 1;
 	spin_unlock_irq(&ev_queue->lock);
-
-	if (why == RDMA_REMOVE_DRIVER_REMOVE)
-		wake_up_interruptible(&ev_queue->poll_wait);
+	wake_up_interruptible(&ev_queue->poll_wait);
 
 	mlx5_cmd_cleanup_async_ctx(&comp_ev_file->async_ctx);
+
+	spin_lock_irq(&comp_ev_file->ev_queue.lock);
+	list_for_each_entry_safe(entry, tmp,
+				 &comp_ev_file->ev_queue.event_list, list)
+		kvfree(entry);
+	spin_unlock_irq(&comp_ev_file->ev_queue.lock);
 	return 0;
 };
 
-static int devx_hot_unplug_async_event_file(struct ib_uobject *uobj,
-					    enum rdma_remove_reason why)
+static int devx_async_event_destroy_uobj(struct ib_uobject *uobj,
+					 enum rdma_remove_reason why)
 {
 	struct devx_async_event_file *ev_file =
 		container_of(uobj, struct devx_async_event_file,
 			     uobj);
+	struct devx_event_subscription *event_sub, *event_sub_tmp;
+	struct devx_async_event_data *entry, *tmp;
+	struct mlx5_ib_dev *dev = ev_file->dev;
 
 	spin_lock_irq(&ev_file->lock);
 	ev_file->is_destroyed = 1;
 	spin_unlock_irq(&ev_file->lock);
-
 	wake_up_interruptible(&ev_file->poll_wait);
+
+	mutex_lock(&dev->devx_event_table.event_xa_lock);
+	/* delete the subscriptions which are related to this FD */
+	list_for_each_entry_safe(event_sub, event_sub_tmp,
+				 &ev_file->subscribed_events_list, file_list) {
+		devx_cleanup_subscription(dev, event_sub);
+		list_del_rcu(&event_sub->file_list);
+		/* subscription may not be used by the read API any more */
+		call_rcu(&event_sub->rcu, devx_free_subscription);
+	}
+	mutex_unlock(&dev->devx_event_table.event_xa_lock);
+
+	/* free the pending events allocation */
+	if (!ev_file->omit_data) {
+		spin_lock_irq(&ev_file->lock);
+		list_for_each_entry_safe(entry, tmp,
+					 &ev_file->event_list, list)
+			kfree(entry); /* read can't come any more */
+		spin_unlock_irq(&ev_file->lock);
+	}
+
+	put_device(&dev->ib_dev.dev);
 	return 0;
 };
 
@@ -2913,7 +2898,7 @@ DECLARE_UVERBS_NAMED_METHOD(
 DECLARE_UVERBS_NAMED_OBJECT(
 	MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD,
 	UVERBS_TYPE_ALLOC_FD(sizeof(struct devx_async_cmd_event_file),
-			     devx_hot_unplug_async_cmd_event_file,
+			     devx_async_cmd_event_destroy_uobj,
 			     &devx_async_cmd_event_fops, "[devx_async_cmd]",
 			     O_RDONLY),
 	&UVERBS_METHOD(MLX5_IB_METHOD_DEVX_ASYNC_CMD_FD_ALLOC));
@@ -2931,7 +2916,7 @@ DECLARE_UVERBS_NAMED_METHOD(
 DECLARE_UVERBS_NAMED_OBJECT(
 	MLX5_IB_OBJECT_DEVX_ASYNC_EVENT_FD,
 	UVERBS_TYPE_ALLOC_FD(sizeof(struct devx_async_event_file),
-			     devx_hot_unplug_async_event_file,
+			     devx_async_event_destroy_uobj,
 			     &devx_async_event_fops, "[devx_async_event]",
 			     O_RDONLY),
 	&UVERBS_METHOD(MLX5_IB_METHOD_DEVX_ASYNC_EVENT_FD_ALLOC));
diff --git a/include/rdma/uverbs_types.h b/include/rdma/uverbs_types.h
index ca65d4704b4b..b633fa2543e3 100644
--- a/include/rdma/uverbs_types.h
+++ b/include/rdma/uverbs_types.h
@@ -161,11 +161,11 @@ struct uverbs_obj_fd_type {
 	 * In fd based objects, uverbs_obj_type_ops points to generic
 	 * fd operations. In order to specialize the underlying types (e.g.
 	 * completion_channel), we use fops, name and flags for fd creation.
-	 * context_closed is called when the context is closed either when
-	 * the driver is removed or the process terminated.
+	 * destroy_object is called when the uobject is to be destroyed,
+	 * because the driver is removed or the FD is closed.
 	 */
 	struct uverbs_obj_type  type;
-	int (*context_closed)(struct ib_uobject *uobj,
+	int (*destroy_object)(struct ib_uobject *uobj,
 			      enum rdma_remove_reason why);
 	const struct file_operations	*fops;
 	const char			*name;
@@ -174,11 +174,11 @@ struct uverbs_obj_fd_type {
 
 extern const struct uverbs_obj_type_class uverbs_idr_class;
 extern const struct uverbs_obj_type_class uverbs_fd_class;
-void uverbs_close_fd(struct file *f);
+int uverbs_uobject_fd_release(struct inode *inode, struct file *filp);
 
 #define UVERBS_BUILD_BUG_ON(cond) (sizeof(char[1 - 2 * !!(cond)]) -	\
 				   sizeof(char))
-#define UVERBS_TYPE_ALLOC_FD(_obj_size, _context_closed, _fops, _name, _flags)\
+#define UVERBS_TYPE_ALLOC_FD(_obj_size, _destroy_object, _fops, _name, _flags) \
 	((&((const struct uverbs_obj_fd_type)				\
 	 {.type = {							\
 		.type_class = &uverbs_fd_class,				\
@@ -186,7 +186,7 @@ void uverbs_close_fd(struct file *f);
 			UVERBS_BUILD_BUG_ON((_obj_size) <               \
 					    sizeof(struct ib_uobject)), \
 	 },								\
-	 .context_closed = _context_closed,				\
+	 .destroy_object = _destroy_object,				\
 	 .fops = _fops,							\
 	 .name = _name,							\
 	 .flags = _flags}))->type)
-- 
cgit v1.2.3-59-g8ed1b


From 93887e66ff3c1a62c36a7f0eca9445063399b39d Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:21:55 +0200
Subject: RDMA/mlx5: Simplify devx async commands

With the new FD structure the async commands do not need to hold any
references while running. The existing mlx5_cmd_exec_cb() and
mlx5_cmd_cleanup_async_ctx() provide enough synchronization to ensure
that all outstanding commands are completed before the uobject can be
destructed.

Remove the now confusing get_file() and the type erasure of the
devx_async_cmd_event_file.

Link: https://lore.kernel.org/r/1578504126-9400-4-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/devx.c | 24 ++++++++++--------------
 1 file changed, 10 insertions(+), 14 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c
index 02125d82b96d..7bb91d24d394 100644
--- a/drivers/infiniband/hw/mlx5/devx.c
+++ b/drivers/infiniband/hw/mlx5/devx.c
@@ -30,7 +30,7 @@ enum devx_obj_flags {
 struct devx_async_data {
 	struct mlx5_ib_dev *mdev;
 	struct list_head list;
-	struct ib_uobject *fd_uobj;
+	struct devx_async_cmd_event_file *ev_file;
 	struct mlx5_async_work cb_work;
 	u16 cmd_out_len;
 	/* must be last field in this structure */
@@ -1673,21 +1673,20 @@ static void devx_query_callback(int status, struct mlx5_async_work *context)
 {
 	struct devx_async_data *async_data =
 		container_of(context, struct devx_async_data, cb_work);
-	struct ib_uobject *fd_uobj = async_data->fd_uobj;
-	struct devx_async_cmd_event_file *ev_file;
-	struct devx_async_event_queue *ev_queue;
+	struct devx_async_cmd_event_file *ev_file = async_data->ev_file;
+	struct devx_async_event_queue *ev_queue = &ev_file->ev_queue;
 	unsigned long flags;
 
-	ev_file = container_of(fd_uobj, struct devx_async_cmd_event_file,
-			       uobj);
-	ev_queue = &ev_file->ev_queue;
-
+	/*
+	 * Note that if the struct devx_async_cmd_event_file uobj begins to be
+	 * destroyed it will block at mlx5_cmd_cleanup_async_ctx() until this
+	 * routine returns, ensuring that it always remains valid here.
+	 */
 	spin_lock_irqsave(&ev_queue->lock, flags);
 	list_add_tail(&async_data->list, &ev_queue->event_list);
 	spin_unlock_irqrestore(&ev_queue->lock, flags);
 
 	wake_up_interruptible(&ev_queue->poll_wait);
-	fput(fd_uobj->object);
 }
 
 #define MAX_ASYNC_BYTES_IN_USE (1024 * 1024) /* 1MB */
@@ -1756,9 +1755,8 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_ASYNC_QUERY)(
 
 	async_data->cmd_out_len = cmd_out_len;
 	async_data->mdev = mdev;
-	async_data->fd_uobj = fd_uobj;
+	async_data->ev_file = ev_file;
 
-	get_file(fd_uobj->object);
 	MLX5_SET(general_obj_in_cmd_hdr, cmd_in, uid, uid);
 	err = mlx5_cmd_exec_cb(&ev_file->async_ctx, cmd_in,
 		    uverbs_attr_get_len(attrs,
@@ -1768,12 +1766,10 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_ASYNC_QUERY)(
 		    devx_query_callback, &async_data->cb_work);
 
 	if (err)
-		goto cb_err;
+		goto free_async;
 
 	return 0;
 
-cb_err:
-	fput(fd_uobj->object);
 free_async:
 	kvfree(async_data);
 sub_bytes:
-- 
cgit v1.2.3-59-g8ed1b


From 849e149063bd10eb6211c14617491a0bc9516c2f Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:21:56 +0200
Subject: RDMA/core: Do not allow alloc_commit to fail

This is a left over from an earlier version that creates a lot of
complexity for error unwind, particularly for FD uobjects.

The only reason this was done is so that anon_inode_get_file() could be
called with the final fops and a fully setup uobject. Both need to be
setup since unwinding anon_inode_get_file() via fput will call the
driver's release().

Now that the driver does not provide release, we no longer need to worry
about this complicated sequence, simply create the struct file at the
start and allow the core code's release function to deal with the abort
case.

This allows all the confusing error paths around commit to be removed.

Link: https://lore.kernel.org/r/1578504126-9400-5-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/rdma_core.c    | 118 ++++++++++++++++-----------------
 drivers/infiniband/core/rdma_core.h    |  21 +-----
 drivers/infiniband/core/uverbs_cmd.c   |  40 ++++++-----
 drivers/infiniband/core/uverbs_ioctl.c |  45 ++++---------
 include/rdma/uverbs_std_types.h        |  10 ---
 include/rdma/uverbs_types.h            |   6 +-
 6 files changed, 99 insertions(+), 141 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index 0ed0341b8e30..a9f5263c9559 100644
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -130,7 +130,11 @@ static int uverbs_destroy_uobject(struct ib_uobject *uobj,
 	lockdep_assert_held(&ufile->hw_destroy_rwsem);
 	assert_uverbs_usecnt(uobj, UVERBS_LOOKUP_WRITE);
 
-	if (uobj->object) {
+	if (reason == RDMA_REMOVE_ABORT) {
+		WARN_ON(!list_empty(&uobj->list));
+		WARN_ON(!uobj->context);
+		uobj->uapi_object->type_class->alloc_abort(uobj);
+	} else if (uobj->object) {
 		ret = uobj->uapi_object->type_class->destroy_hw(uobj, reason,
 								attrs);
 		if (ret) {
@@ -146,12 +150,6 @@ static int uverbs_destroy_uobject(struct ib_uobject *uobj,
 		uobj->object = NULL;
 	}
 
-	if (reason == RDMA_REMOVE_ABORT) {
-		WARN_ON(!list_empty(&uobj->list));
-		WARN_ON(!uobj->context);
-		uobj->uapi_object->type_class->alloc_abort(uobj);
-	}
-
 	uobj->context = NULL;
 
 	/*
@@ -450,22 +448,40 @@ static struct ib_uobject *
 alloc_begin_fd_uobject(const struct uverbs_api_object *obj,
 		       struct ib_uverbs_file *ufile)
 {
+	const struct uverbs_obj_fd_type *fd_type =
+		container_of(obj->type_attrs, struct uverbs_obj_fd_type, type);
 	int new_fd;
 	struct ib_uobject *uobj;
+	struct file *filp;
+
+	if (WARN_ON(fd_type->fops->release != &uverbs_uobject_fd_release))
+		return ERR_PTR(-EINVAL);
 
 	new_fd = get_unused_fd_flags(O_CLOEXEC);
 	if (new_fd < 0)
 		return ERR_PTR(new_fd);
 
 	uobj = alloc_uobj(ufile, obj);
-	if (IS_ERR(uobj)) {
-		put_unused_fd(new_fd);
-		return uobj;
+	if (IS_ERR(uobj))
+		goto err_fd;
+
+	/* Note that uverbs_uobject_fd_release() is called during abort */
+	filp = anon_inode_getfile(fd_type->name, fd_type->fops, NULL,
+				  fd_type->flags);
+	if (IS_ERR(filp)) {
+		uobj = ERR_CAST(filp);
+		goto err_uobj;
 	}
+	uobj->object = filp;
 
 	uobj->id = new_fd;
 	uobj->ufile = ufile;
+	return uobj;
 
+err_uobj:
+	uverbs_uobject_put(uobj);
+err_fd:
+	put_unused_fd(new_fd);
 	return uobj;
 }
 
@@ -539,6 +555,9 @@ static void remove_handle_idr_uobject(struct ib_uobject *uobj)
 
 static void alloc_abort_fd_uobject(struct ib_uobject *uobj)
 {
+	struct file *filp = uobj->object;
+
+	fput(filp);
 	put_unused_fd(uobj->id);
 }
 
@@ -560,7 +579,7 @@ static void remove_handle_fd_uobject(struct ib_uobject *uobj)
 {
 }
 
-static int alloc_commit_idr_uobject(struct ib_uobject *uobj)
+static void alloc_commit_idr_uobject(struct ib_uobject *uobj)
 {
 	struct ib_uverbs_file *ufile = uobj->ufile;
 	void *old;
@@ -574,31 +593,12 @@ static int alloc_commit_idr_uobject(struct ib_uobject *uobj)
 	 */
 	old = xa_store(&ufile->idr, uobj->id, uobj, GFP_KERNEL);
 	WARN_ON(old != NULL);
-
-	return 0;
 }
 
-static int alloc_commit_fd_uobject(struct ib_uobject *uobj)
+static void alloc_commit_fd_uobject(struct ib_uobject *uobj)
 {
-	const struct uverbs_obj_fd_type *fd_type = container_of(
-		uobj->uapi_object->type_attrs, struct uverbs_obj_fd_type, type);
 	int fd = uobj->id;
-	struct file *filp;
-
-	/*
-	 * The kref for uobj is moved into filp->private data and put in
-	 * uverbs_close_fd(). Once alloc_commit() succeeds
-	 * uverbs_uobject_fd_release() must be guaranteed to be called from
-	 * the provided fops release callback.
-	 */
-	filp = anon_inode_getfile(fd_type->name,
-				  fd_type->fops,
-				  uobj,
-				  fd_type->flags);
-	if (IS_ERR(filp))
-		return PTR_ERR(filp);
-
-	uobj->object = filp;
+	struct file *filp = uobj->object;
 
 	/* Matching put will be done in uverbs_uobject_fd_release() */
 	kref_get(&uobj->ufile->ref);
@@ -610,9 +610,8 @@ static int alloc_commit_fd_uobject(struct ib_uobject *uobj)
 	 * NOTE: Once we install the file we loose ownership of our kref on
 	 * uobj. It will be put by uverbs_uobject_fd_release()
 	 */
+	filp->private_data = uobj;
 	fd_install(fd, filp);
-
-	return 0;
 }
 
 /*
@@ -620,19 +619,13 @@ static int alloc_commit_fd_uobject(struct ib_uobject *uobj)
  * caller can no longer assume uobj is valid. If this function fails it
  * destroys the uboject, including the attached HW object.
  */
-int __must_check rdma_alloc_commit_uobject(struct ib_uobject *uobj,
-					   struct uverbs_attr_bundle *attrs)
+void rdma_alloc_commit_uobject(struct ib_uobject *uobj,
+			       struct uverbs_attr_bundle *attrs)
 {
 	struct ib_uverbs_file *ufile = attrs->ufile;
-	int ret;
 
 	/* alloc_commit consumes the uobj kref */
-	ret = uobj->uapi_object->type_class->alloc_commit(uobj);
-	if (ret) {
-		uverbs_destroy_uobject(uobj, RDMA_REMOVE_ABORT, attrs);
-		up_read(&ufile->hw_destroy_rwsem);
-		return ret;
-	}
+	uobj->uapi_object->type_class->alloc_commit(uobj);
 
 	/* kref is held so long as the uobj is on the uobj list. */
 	uverbs_uobject_get(uobj);
@@ -645,8 +638,6 @@ int __must_check rdma_alloc_commit_uobject(struct ib_uobject *uobj,
 
 	/* Matches the down_read in rdma_alloc_begin_uobject */
 	up_read(&ufile->hw_destroy_rwsem);
-
-	return 0;
 }
 
 /*
@@ -658,7 +649,6 @@ void rdma_alloc_abort_uobject(struct ib_uobject *uobj,
 {
 	struct ib_uverbs_file *ufile = uobj->ufile;
 
-	uobj->object = NULL;
 	uverbs_destroy_uobject(uobj, RDMA_REMOVE_ABORT, attrs);
 
 	/* Matches the down_read in rdma_alloc_begin_uobject */
@@ -751,14 +741,23 @@ EXPORT_SYMBOL(uverbs_idr_class);
  */
 int uverbs_uobject_fd_release(struct inode *inode, struct file *filp)
 {
-	struct ib_uobject *uobj = filp->private_data;
-	struct ib_uverbs_file *ufile = uobj->ufile;
-	struct uverbs_attr_bundle attrs = {
-		.context = uobj->context,
-		.ufile = ufile,
-	};
+	struct ib_uverbs_file *ufile;
+	struct ib_uobject *uobj;
+
+	/*
+	 * This can only happen if the fput came from alloc_abort_fd_uobject()
+	 */
+	if (!filp->private_data)
+		return 0;
+	uobj = filp->private_data;
+	ufile = uobj->ufile;
 
 	if (down_read_trylock(&ufile->hw_destroy_rwsem)) {
+		struct uverbs_attr_bundle attrs = {
+			.context = uobj->context,
+			.ufile = ufile,
+		};
+
 		/*
 		 * lookup_get_fd_uobject holds the kref on the struct file any
 		 * time a FD uobj is locked, which prevents this release
@@ -770,7 +769,7 @@ int uverbs_uobject_fd_release(struct inode *inode, struct file *filp)
 		up_read(&ufile->hw_destroy_rwsem);
 	}
 
-	/* Matches the get in alloc_begin_fd_uobject */
+	/* Matches the get in alloc_commit_fd_uobject() */
 	kref_put(&ufile->ref, ib_uverbs_release_file);
 
 	/* Pairs with filp->private_data in alloc_begin_fd_uobject */
@@ -938,12 +937,10 @@ uverbs_get_uobject_from_file(u16 object_id, enum uverbs_obj_access access,
 	}
 }
 
-int uverbs_finalize_object(struct ib_uobject *uobj,
-			   enum uverbs_obj_access access, bool commit,
-			   struct uverbs_attr_bundle *attrs)
+void uverbs_finalize_object(struct ib_uobject *uobj,
+			    enum uverbs_obj_access access, bool commit,
+			    struct uverbs_attr_bundle *attrs)
 {
-	int ret = 0;
-
 	/*
 	 * refcounts should be handled at the object level and not at the
 	 * uobject level. Refcounts of the objects themselves are done in
@@ -963,14 +960,11 @@ int uverbs_finalize_object(struct ib_uobject *uobj,
 		break;
 	case UVERBS_ACCESS_NEW:
 		if (commit)
-			ret = rdma_alloc_commit_uobject(uobj, attrs);
+			rdma_alloc_commit_uobject(uobj, attrs);
 		else
 			rdma_alloc_abort_uobject(uobj, attrs);
 		break;
 	default:
 		WARN_ON(true);
-		ret = -EOPNOTSUPP;
 	}
-
-	return ret;
 }
diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h
index 92694253e776..29f905e8c2a8 100644
--- a/drivers/infiniband/core/rdma_core.h
+++ b/drivers/infiniband/core/rdma_core.h
@@ -63,24 +63,9 @@ struct ib_uobject *
 uverbs_get_uobject_from_file(u16 object_id, enum uverbs_obj_access access,
 			     s64 id, struct uverbs_attr_bundle *attrs);
 
-/*
- * Note that certain finalize stages could return a status:
- *   (a) alloc_commit could return a failure if the object is committed at the
- *       same time when the context is destroyed.
- *   (b) remove_commit could fail if the object wasn't destroyed successfully.
- * Since multiple objects could be finalized in one transaction, it is very NOT
- * recommended to have several finalize actions which have side effects.
- * For example, it's NOT recommended to have a certain action which has both
- * a commit action and a destroy action or two destroy objects in the same
- * action. The rule of thumb is to have one destroy or commit action with
- * multiple lookups.
- * The first non zero return value of finalize_object is returned from this
- * function. For example, this could happen when we couldn't destroy an
- * object.
- */
-int uverbs_finalize_object(struct ib_uobject *uobj,
-			   enum uverbs_obj_access access, bool commit,
-			   struct uverbs_attr_bundle *attrs);
+void uverbs_finalize_object(struct ib_uobject *uobj,
+			    enum uverbs_obj_access access, bool commit,
+			    struct uverbs_attr_bundle *attrs);
 
 int uverbs_output_written(const struct uverbs_attr_bundle *bundle, size_t idx);
 
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 06ed32c8662f..74f6ae475ca0 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -446,7 +446,8 @@ static int ib_uverbs_alloc_pd(struct uverbs_attr_bundle *attrs)
 	if (ret)
 		goto err_copy;
 
-	return uobj_alloc_commit(uobj, attrs);
+	rdma_alloc_commit_uobject(uobj, attrs);
+	return 0;
 
 err_copy:
 	ib_dealloc_pd_user(pd, uverbs_get_cleared_udata(attrs));
@@ -642,7 +643,8 @@ static int ib_uverbs_open_xrcd(struct uverbs_attr_bundle *attrs)
 
 	mutex_unlock(&ibudev->xrcd_tree_mutex);
 
-	return uobj_alloc_commit(&obj->uobject, attrs);
+	rdma_alloc_commit_uobject(&obj->uobject, attrs);
+	return 0;
 
 err_copy:
 	if (inode) {
@@ -774,7 +776,8 @@ static int ib_uverbs_reg_mr(struct uverbs_attr_bundle *attrs)
 
 	uobj_put_obj_read(pd);
 
-	return uobj_alloc_commit(uobj, attrs);
+	rdma_alloc_commit_uobject(uobj, attrs);
+	return 0;
 
 err_copy:
 	ib_dereg_mr_user(mr, uverbs_get_cleared_udata(attrs));
@@ -928,7 +931,8 @@ static int ib_uverbs_alloc_mw(struct uverbs_attr_bundle *attrs)
 		goto err_copy;
 
 	uobj_put_obj_read(pd);
-	return uobj_alloc_commit(uobj, attrs);
+	rdma_alloc_commit_uobject(uobj, attrs);
+	return 0;
 
 err_copy:
 	uverbs_dealloc_mw(mw);
@@ -980,7 +984,8 @@ static int ib_uverbs_create_comp_channel(struct uverbs_attr_bundle *attrs)
 		return ret;
 	}
 
-	return uobj_alloc_commit(uobj, attrs);
+	rdma_alloc_commit_uobject(uobj, attrs);
+	return 0;
 }
 
 static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs,
@@ -1049,9 +1054,7 @@ static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs,
 	if (ret)
 		goto err_cb;
 
-	ret = uobj_alloc_commit(&obj->uobject, attrs);
-	if (ret)
-		return ERR_PTR(ret);
+	rdma_alloc_commit_uobject(&obj->uobject, attrs);
 	return obj;
 
 err_cb:
@@ -1491,7 +1494,8 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 	if (ind_tbl)
 		uobj_put_obj_read(ind_tbl);
 
-	return uobj_alloc_commit(&obj->uevent.uobject, attrs);
+	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
+	return 0;
 err_cb:
 	ib_destroy_qp_user(qp, uverbs_get_cleared_udata(attrs));
 
@@ -1623,7 +1627,8 @@ static int ib_uverbs_open_qp(struct uverbs_attr_bundle *attrs)
 	qp->uobject = &obj->uevent.uobject;
 	uobj_put_read(xrcd_uobj);
 
-	return uobj_alloc_commit(&obj->uevent.uobject, attrs);
+	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
+	return 0;
 
 err_destroy:
 	ib_destroy_qp_user(qp, uverbs_get_cleared_udata(attrs));
@@ -2465,7 +2470,8 @@ static int ib_uverbs_create_ah(struct uverbs_attr_bundle *attrs)
 		goto err_copy;
 
 	uobj_put_obj_read(pd);
-	return uobj_alloc_commit(uobj, attrs);
+	rdma_alloc_commit_uobject(uobj, attrs);
+	return 0;
 
 err_copy:
 	rdma_destroy_ah_user(ah, RDMA_DESTROY_AH_SLEEPABLE,
@@ -2977,7 +2983,8 @@ static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs)
 
 	uobj_put_obj_read(pd);
 	uobj_put_obj_read(cq);
-	return uobj_alloc_commit(&obj->uevent.uobject, attrs);
+	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
+	return 0;
 
 err_copy:
 	ib_destroy_wq(wq, uverbs_get_cleared_udata(attrs));
@@ -3151,7 +3158,8 @@ static int ib_uverbs_ex_create_rwq_ind_table(struct uverbs_attr_bundle *attrs)
 	for (j = 0; j < num_read_wqs; j++)
 		uobj_put_obj_read(wqs[j]);
 
-	return uobj_alloc_commit(uobj, attrs);
+	rdma_alloc_commit_uobject(uobj, attrs);
+	return 0;
 
 err_copy:
 	ib_destroy_rwq_ind_table(rwq_ind_tbl);
@@ -3329,7 +3337,8 @@ static int ib_uverbs_ex_create_flow(struct uverbs_attr_bundle *attrs)
 	kfree(flow_attr);
 	if (cmd.flow_attr.num_of_specs)
 		kfree(kern_flow_attr);
-	return uobj_alloc_commit(uobj, attrs);
+	rdma_alloc_commit_uobject(uobj, attrs);
+	return 0;
 err_copy:
 	if (!qp->device->ops.destroy_flow(flow_id))
 		atomic_dec(&qp->usecnt);
@@ -3477,7 +3486,8 @@ static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs,
 		uobj_put_obj_read(attr.ext.cq);
 
 	uobj_put_obj_read(pd);
-	return uobj_alloc_commit(&obj->uevent.uobject, attrs);
+	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
+	return 0;
 
 err_copy:
 	ib_destroy_srq_user(srq, uverbs_get_cleared_udata(attrs));
diff --git a/drivers/infiniband/core/uverbs_ioctl.c b/drivers/infiniband/core/uverbs_ioctl.c
index 269938f59d3f..538affbc517e 100644
--- a/drivers/infiniband/core/uverbs_ioctl.c
+++ b/drivers/infiniband/core/uverbs_ioctl.c
@@ -220,24 +220,17 @@ static int uverbs_process_idrs_array(struct bundle_priv *pbundle,
 	return ret;
 }
 
-static int uverbs_free_idrs_array(const struct uverbs_api_attr *attr_uapi,
-				  struct uverbs_objs_arr_attr *attr,
-				  bool commit, struct uverbs_attr_bundle *attrs)
+static void uverbs_free_idrs_array(const struct uverbs_api_attr *attr_uapi,
+				   struct uverbs_objs_arr_attr *attr,
+				   bool commit,
+				   struct uverbs_attr_bundle *attrs)
 {
 	const struct uverbs_attr_spec *spec = &attr_uapi->spec;
-	int current_ret;
-	int ret = 0;
 	size_t i;
 
-	for (i = 0; i != attr->len; i++) {
-		current_ret = uverbs_finalize_object(attr->uobjects[i],
-						     spec->u2.objs_arr.access,
-						     commit, attrs);
-		if (!ret)
-			ret = current_ret;
-	}
-
-	return ret;
+	for (i = 0; i != attr->len; i++)
+		uverbs_finalize_object(attr->uobjects[i],
+				       spec->u2.objs_arr.access, commit, attrs);
 }
 
 static int uverbs_process_attr(struct bundle_priv *pbundle,
@@ -495,26 +488,22 @@ static int ib_uverbs_run_method(struct bundle_priv *pbundle,
 	return ret;
 }
 
-static int bundle_destroy(struct bundle_priv *pbundle, bool commit)
+static void bundle_destroy(struct bundle_priv *pbundle, bool commit)
 {
 	unsigned int key_bitmap_len = pbundle->method_elm->key_bitmap_len;
 	struct bundle_alloc_head *memblock;
 	unsigned int i;
-	int ret = 0;
 
 	/* fast path for simple uobjects */
 	i = -1;
 	while ((i = find_next_bit(pbundle->uobj_finalize, key_bitmap_len,
 				  i + 1)) < key_bitmap_len) {
 		struct uverbs_attr *attr = &pbundle->bundle.attrs[i];
-		int current_ret;
 
-		current_ret = uverbs_finalize_object(
+		uverbs_finalize_object(
 			attr->obj_attr.uobject,
 			attr->obj_attr.attr_elm->spec.u.obj.access, commit,
 			&pbundle->bundle);
-		if (!ret)
-			ret = current_ret;
 	}
 
 	i = -1;
@@ -523,7 +512,6 @@ static int bundle_destroy(struct bundle_priv *pbundle, bool commit)
 		struct uverbs_attr *attr = &pbundle->bundle.attrs[i];
 		const struct uverbs_api_attr *attr_uapi;
 		void __rcu **slot;
-		int current_ret;
 
 		slot = uapi_get_attr_for_method(
 			pbundle,
@@ -534,11 +522,8 @@ static int bundle_destroy(struct bundle_priv *pbundle, bool commit)
 		attr_uapi = rcu_dereference_protected(*slot, true);
 
 		if (attr_uapi->spec.type == UVERBS_ATTR_TYPE_IDRS_ARRAY) {
-			current_ret = uverbs_free_idrs_array(
-				attr_uapi, &attr->objs_arr_attr, commit,
-				&pbundle->bundle);
-			if (!ret)
-				ret = current_ret;
+			uverbs_free_idrs_array(attr_uapi, &attr->objs_arr_attr,
+					       commit, &pbundle->bundle);
 		}
 	}
 
@@ -548,8 +533,6 @@ static int bundle_destroy(struct bundle_priv *pbundle, bool commit)
 		memblock = memblock->next;
 		kvfree(tmp);
 	}
-
-	return ret;
 }
 
 static int ib_uverbs_cmd_verbs(struct ib_uverbs_file *ufile,
@@ -562,7 +545,6 @@ static int ib_uverbs_cmd_verbs(struct ib_uverbs_file *ufile,
 	struct bundle_priv *pbundle;
 	struct bundle_priv onstack;
 	void __rcu **slot;
-	int destroy_ret;
 	int ret;
 
 	if (unlikely(hdr->driver_id != uapi->driver_id))
@@ -610,10 +592,7 @@ static int ib_uverbs_cmd_verbs(struct ib_uverbs_file *ufile,
 	memset(pbundle->spec_finalize, 0, sizeof(pbundle->spec_finalize));
 
 	ret = ib_uverbs_run_method(pbundle, hdr->num_attrs);
-	destroy_ret = bundle_destroy(pbundle, ret == 0);
-	if (unlikely(destroy_ret && !ret))
-		return destroy_ret;
-
+	bundle_destroy(pbundle, ret == 0);
 	return ret;
 }
 
diff --git a/include/rdma/uverbs_std_types.h b/include/rdma/uverbs_std_types.h
index 05eabfd5d0d3..c6bcaad4a8e0 100644
--- a/include/rdma/uverbs_std_types.h
+++ b/include/rdma/uverbs_std_types.h
@@ -104,16 +104,6 @@ static inline void uobj_put_write(struct ib_uobject *uobj)
 	rdma_lookup_put_uobject(uobj, UVERBS_LOOKUP_WRITE);
 }
 
-static inline int __must_check
-uobj_alloc_commit(struct ib_uobject *uobj, struct uverbs_attr_bundle *attrs)
-{
-	int ret = rdma_alloc_commit_uobject(uobj, attrs);
-
-	if (ret)
-		return ret;
-	return 0;
-}
-
 static inline void uobj_alloc_abort(struct ib_uobject *uobj,
 				    struct uverbs_attr_bundle *attrs)
 {
diff --git a/include/rdma/uverbs_types.h b/include/rdma/uverbs_types.h
index b633fa2543e3..01d59df4e9e7 100644
--- a/include/rdma/uverbs_types.h
+++ b/include/rdma/uverbs_types.h
@@ -85,7 +85,7 @@ struct uverbs_obj_type_class {
 	struct ib_uobject *(*alloc_begin)(const struct uverbs_api_object *obj,
 					  struct ib_uverbs_file *ufile);
 	/* This consumes the kref on uobj */
-	int (*alloc_commit)(struct ib_uobject *uobj);
+	void (*alloc_commit)(struct ib_uobject *uobj);
 	/* This does not consume the kref on uobj */
 	void (*alloc_abort)(struct ib_uobject *uobj);
 
@@ -141,8 +141,8 @@ struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_api_object *obj,
 					    struct uverbs_attr_bundle *attrs);
 void rdma_alloc_abort_uobject(struct ib_uobject *uobj,
 			      struct uverbs_attr_bundle *attrs);
-int __must_check rdma_alloc_commit_uobject(struct ib_uobject *uobj,
-					   struct uverbs_attr_bundle *attrs);
+void rdma_alloc_commit_uobject(struct ib_uobject *uobj,
+			       struct uverbs_attr_bundle *attrs);
 
 /*
  * uverbs_uobject_get is called in order to increase the reference count on
-- 
cgit v1.2.3-59-g8ed1b


From 4ec1dcfcdf3756555adb5122c2443ea27741296a Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:21:57 +0200
Subject: RDMA/core: Make ib_ucq_object use ib_uevent_object

Any uobject that sends events into the async_event_file should be using
ib_uevent_object so it can use the standard uevent based helper
functions. CQ pushes events into both the async_event and the comp_channel
in an open coded way. Move the async events related stuff to
ib_uevent_object.

Link: https://lore.kernel.org/r/1578504126-9400-6-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs.h              |  5 ++---
 drivers/infiniband/core/uverbs_cmd.c          | 20 +++++++++-----------
 drivers/infiniband/core/uverbs_main.c         | 17 ++++++-----------
 drivers/infiniband/core/uverbs_std_types_cq.c | 18 ++++++++----------
 4 files changed, 25 insertions(+), 35 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index 63f7f7db5902..9fa044699cf0 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -183,6 +183,7 @@ struct ib_uverbs_mcast_entry {
 
 struct ib_uevent_object {
 	struct ib_uobject	uobject;
+	/* List member for ib_uverbs_async_event_file list */
 	struct list_head	event_list;
 	u32			events_reported;
 };
@@ -210,11 +211,9 @@ struct ib_uwq_object {
 };
 
 struct ib_ucq_object {
-	struct ib_uobject	uobject;
+	struct ib_uevent_object uevent;
 	struct list_head	comp_list;
-	struct list_head	async_list;
 	u32			comp_events_reported;
-	u32			async_events_reported;
 };
 
 extern const struct file_operations uverbs_event_fops;
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 74f6ae475ca0..3a2a27830ae2 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1015,11 +1015,9 @@ static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs,
 		}
 	}
 
-	obj->uobject.user_handle = cmd->user_handle;
-	obj->comp_events_reported  = 0;
-	obj->async_events_reported = 0;
+	obj->uevent.uobject.user_handle = cmd->user_handle;
 	INIT_LIST_HEAD(&obj->comp_list);
-	INIT_LIST_HEAD(&obj->async_list);
+	INIT_LIST_HEAD(&obj->uevent.event_list);
 
 	attr.cqe = cmd->cqe;
 	attr.comp_vector = cmd->comp_vector;
@@ -1031,7 +1029,7 @@ static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs,
 		goto err_file;
 	}
 	cq->device        = ib_dev;
-	cq->uobject       = &obj->uobject;
+	cq->uobject       = &obj->uevent.uobject;
 	cq->comp_handler  = ib_uverbs_comp_handler;
 	cq->event_handler = ib_uverbs_cq_event_handler;
 	cq->cq_context    = ev_file ? &ev_file->ev_queue : NULL;
@@ -1041,9 +1039,9 @@ static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs,
 	if (ret)
 		goto err_free;
 
-	obj->uobject.object = cq;
+	obj->uevent.uobject.object = cq;
 	memset(&resp, 0, sizeof resp);
-	resp.base.cq_handle = obj->uobject.id;
+	resp.base.cq_handle = obj->uevent.uobject.id;
 	resp.base.cqe       = cq->cqe;
 	resp.response_length = uverbs_response_length(attrs, sizeof(resp));
 
@@ -1054,7 +1052,7 @@ static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs,
 	if (ret)
 		goto err_cb;
 
-	rdma_alloc_commit_uobject(&obj->uobject, attrs);
+	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
 	return obj;
 
 err_cb:
@@ -1067,7 +1065,7 @@ err_file:
 		ib_uverbs_release_ucq(attrs->ufile, ev_file, obj);
 
 err:
-	uobj_alloc_abort(&obj->uobject, attrs);
+	uobj_alloc_abort(&obj->uevent.uobject, attrs);
 
 	return ERR_PTR(ret);
 }
@@ -1261,10 +1259,10 @@ static int ib_uverbs_destroy_cq(struct uverbs_attr_bundle *attrs)
 	if (IS_ERR(uobj))
 		return PTR_ERR(uobj);
 
-	obj = container_of(uobj, struct ib_ucq_object, uobject);
+	obj = container_of(uobj, struct ib_ucq_object, uevent.uobject);
 	memset(&resp, 0, sizeof(resp));
 	resp.comp_events_reported  = obj->comp_events_reported;
-	resp.async_events_reported = obj->async_events_reported;
+	resp.async_events_reported = obj->uevent.events_reported;
 
 	uobj_put_destroy(uobj);
 
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index da56fda259fd..940bb291a57d 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -150,12 +150,7 @@ void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
 		uverbs_uobject_put(&ev_file->uobj);
 	}
 
-	spin_lock_irq(&file->async_file->ev_queue.lock);
-	list_for_each_entry_safe(evt, tmp, &uobj->async_list, obj_list) {
-		list_del(&evt->list);
-		kfree(evt);
-	}
-	spin_unlock_irq(&file->async_file->ev_queue.lock);
+	ib_uverbs_release_uevent(file, &uobj->uevent);
 }
 
 void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
@@ -413,7 +408,7 @@ void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context)
 		return;
 	}
 
-	uobj = container_of(cq->uobject, struct ib_ucq_object, uobject);
+	uobj = container_of(cq->uobject, struct ib_ucq_object, uevent.uobject);
 
 	entry->desc.comp.cq_handle = cq->uobject->user_handle;
 	entry->counter		   = &uobj->comp_events_reported;
@@ -462,12 +457,12 @@ static void ib_uverbs_async_handler(struct ib_uverbs_file *file,
 
 void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr)
 {
-	struct ib_ucq_object *uobj = container_of(event->element.cq->uobject,
-						  struct ib_ucq_object, uobject);
+	struct ib_uevent_object *uobj = container_of(
+		event->element.cq->uobject, struct ib_uevent_object, uobject);
 
 	ib_uverbs_async_handler(uobj->uobject.ufile, uobj->uobject.user_handle,
-				event->event, &uobj->async_list,
-				&uobj->async_events_reported);
+				event->event, &uobj->event_list,
+				&uobj->events_reported);
 }
 
 void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr)
diff --git a/drivers/infiniband/core/uverbs_std_types_cq.c b/drivers/infiniband/core/uverbs_std_types_cq.c
index e39fe6a8aac4..fbc605a95e51 100644
--- a/drivers/infiniband/core/uverbs_std_types_cq.c
+++ b/drivers/infiniband/core/uverbs_std_types_cq.c
@@ -41,7 +41,7 @@ static int uverbs_free_cq(struct ib_uobject *uobject,
 	struct ib_cq *cq = uobject->object;
 	struct ib_uverbs_event_queue *ev_queue = cq->cq_context;
 	struct ib_ucq_object *ucq =
-		container_of(uobject, struct ib_ucq_object, uobject);
+		container_of(uobject, struct ib_ucq_object, uevent.uobject);
 	int ret;
 
 	ret = ib_destroy_cq_user(cq, &attrs->driver_udata);
@@ -63,7 +63,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)(
 {
 	struct ib_ucq_object *obj = container_of(
 		uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_CQ_HANDLE),
-		typeof(*obj), uobject);
+		typeof(*obj), uevent.uobject);
 	struct ib_device *ib_dev = attrs->context->device;
 	int ret;
 	u64 user_handle;
@@ -106,10 +106,8 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)(
 		goto err_event_file;
 	}
 
-	obj->comp_events_reported  = 0;
-	obj->async_events_reported = 0;
 	INIT_LIST_HEAD(&obj->comp_list);
-	INIT_LIST_HEAD(&obj->async_list);
+	INIT_LIST_HEAD(&obj->uevent.event_list);
 
 	cq = rdma_zalloc_drv_obj(ib_dev, ib_cq);
 	if (!cq) {
@@ -118,7 +116,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)(
 	}
 
 	cq->device        = ib_dev;
-	cq->uobject       = &obj->uobject;
+	cq->uobject       = &obj->uevent.uobject;
 	cq->comp_handler  = ib_uverbs_comp_handler;
 	cq->event_handler = ib_uverbs_cq_event_handler;
 	cq->cq_context    = ev_file ? &ev_file->ev_queue : NULL;
@@ -129,8 +127,8 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)(
 	if (ret)
 		goto err_free;
 
-	obj->uobject.object = cq;
-	obj->uobject.user_handle = user_handle;
+	obj->uevent.uobject.object = cq;
+	obj->uevent.uobject.user_handle = user_handle;
 	rdma_restrack_uadd(&cq->res);
 
 	ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_CQ_RESP_CQE, &cq->cqe,
@@ -182,10 +180,10 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_DESTROY)(
 	struct ib_uobject *uobj =
 		uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_CQ_HANDLE);
 	struct ib_ucq_object *obj =
-		container_of(uobj, struct ib_ucq_object, uobject);
+		container_of(uobj, struct ib_ucq_object, uevent.uobject);
 	struct ib_uverbs_destroy_cq_resp resp = {
 		.comp_events_reported = obj->comp_events_reported,
-		.async_events_reported = obj->async_events_reported
+		.async_events_reported = obj->uevent.events_reported
 	};
 
 	return uverbs_copy_to(attrs, UVERBS_ATTR_DESTROY_CQ_RESP, &resp,
-- 
cgit v1.2.3-59-g8ed1b


From 5bd48c18c8cea0154800b40ec75201fa71684312 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:21:58 +0200
Subject: RDMA/core: Do not erase the type of ib_cq.uobject

This is a struct ib_ucq_object pointer, instead of using container_of()
all over the place just store it with its actual type.

Link: https://lore.kernel.org/r/1578504126-9400-7-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/nldev.c               |  3 +-
 drivers/infiniband/core/uverbs_cmd.c          | 40 +++++++++++++++++----------
 drivers/infiniband/core/uverbs_main.c         |  7 ++---
 drivers/infiniband/core/uverbs_std_types_cq.c |  2 +-
 include/rdma/ib_verbs.h                       |  4 ++-
 5 files changed, 34 insertions(+), 22 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
index cbf6041a5d4a..37b433aa7306 100644
--- a/drivers/infiniband/core/nldev.c
+++ b/drivers/infiniband/core/nldev.c
@@ -41,6 +41,7 @@
 #include "core_priv.h"
 #include "cma_priv.h"
 #include "restrack.h"
+#include "uverbs.h"
 
 typedef int (*res_fill_func_t)(struct sk_buff*, bool,
 			       struct rdma_restrack_entry*, uint32_t);
@@ -599,7 +600,7 @@ static int fill_res_cq_entry(struct sk_buff *msg, bool has_cap_net_admin,
 		goto err;
 	if (!rdma_is_kernel_res(res) &&
 	    nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_CTXN,
-			cq->uobject->context->res.id))
+			cq->uobject->uevent.uobject.context->res.id))
 		goto err;
 
 	if (fill_res_name_pid(msg, res))
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 3a2a27830ae2..b08679a5da25 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1029,7 +1029,7 @@ static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs,
 		goto err_file;
 	}
 	cq->device        = ib_dev;
-	cq->uobject       = &obj->uevent.uobject;
+	cq->uobject       = obj;
 	cq->comp_handler  = ib_uverbs_comp_handler;
 	cq->event_handler = ib_uverbs_cq_event_handler;
 	cq->cq_context    = ev_file ? &ev_file->ev_queue : NULL;
@@ -1134,7 +1134,8 @@ static int ib_uverbs_resize_cq(struct uverbs_attr_bundle *attrs)
 
 	ret = uverbs_response(attrs, &resp, sizeof(resp));
 out:
-	uobj_put_obj_read(cq);
+	rdma_lookup_put_uobject(&cq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 
 	return ret;
 }
@@ -1217,7 +1218,8 @@ static int ib_uverbs_poll_cq(struct uverbs_attr_bundle *attrs)
 		ret = uverbs_output_written(attrs, UVERBS_ATTR_CORE_OUT);
 
 out_put:
-	uobj_put_obj_read(cq);
+	rdma_lookup_put_uobject(&cq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	return ret;
 }
 
@@ -1238,8 +1240,8 @@ static int ib_uverbs_req_notify_cq(struct uverbs_attr_bundle *attrs)
 	ib_req_notify_cq(cq, cmd.solicited_only ?
 			 IB_CQ_SOLICITED : IB_CQ_NEXT_COMP);
 
-	uobj_put_obj_read(cq);
-
+	rdma_lookup_put_uobject(&cq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	return 0;
 }
 
@@ -1484,9 +1486,11 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 	if (pd)
 		uobj_put_obj_read(pd);
 	if (scq)
-		uobj_put_obj_read(scq);
+		rdma_lookup_put_uobject(&scq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 	if (rcq && rcq != scq)
-		uobj_put_obj_read(rcq);
+		rdma_lookup_put_uobject(&rcq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 	if (srq)
 		uobj_put_obj_read(srq);
 	if (ind_tbl)
@@ -1503,9 +1507,11 @@ err_put:
 	if (pd)
 		uobj_put_obj_read(pd);
 	if (scq)
-		uobj_put_obj_read(scq);
+		rdma_lookup_put_uobject(&scq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 	if (rcq && rcq != scq)
-		uobj_put_obj_read(rcq);
+		rdma_lookup_put_uobject(&rcq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 	if (srq)
 		uobj_put_obj_read(srq);
 	if (ind_tbl)
@@ -2980,14 +2986,16 @@ static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs)
 		goto err_copy;
 
 	uobj_put_obj_read(pd);
-	uobj_put_obj_read(cq);
+	rdma_lookup_put_uobject(&cq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
 	return 0;
 
 err_copy:
 	ib_destroy_wq(wq, uverbs_get_cleared_udata(attrs));
 err_put_cq:
-	uobj_put_obj_read(cq);
+	rdma_lookup_put_uobject(&cq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 err_put_pd:
 	uobj_put_obj_read(pd);
 err_uobj:
@@ -3481,7 +3489,8 @@ static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs,
 		uobj_put_read(xrcd_uobj);
 
 	if (ib_srq_has_cq(cmd->srq_type))
-		uobj_put_obj_read(attr.ext.cq);
+		rdma_lookup_put_uobject(&attr.ext.cq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 
 	uobj_put_obj_read(pd);
 	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
@@ -3498,7 +3507,8 @@ err_put:
 
 err_put_cq:
 	if (ib_srq_has_cq(cmd->srq_type))
-		uobj_put_obj_read(attr.ext.cq);
+		rdma_lookup_put_uobject(&attr.ext.cq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 
 err_put_xrcd:
 	if (cmd->srq_type == IB_SRQT_XRC) {
@@ -3714,8 +3724,8 @@ static int ib_uverbs_ex_modify_cq(struct uverbs_attr_bundle *attrs)
 
 	ret = rdma_set_cq_moderation(cq, cmd.attr.cq_count, cmd.attr.cq_period);
 
-	uobj_put_obj_read(cq);
-
+	rdma_lookup_put_uobject(&cq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	return ret;
 }
 
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 940bb291a57d..2849bf40bc97 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -408,9 +408,9 @@ void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context)
 		return;
 	}
 
-	uobj = container_of(cq->uobject, struct ib_ucq_object, uevent.uobject);
+	uobj = cq->uobject;
 
-	entry->desc.comp.cq_handle = cq->uobject->user_handle;
+	entry->desc.comp.cq_handle = cq->uobject->uevent.uobject.user_handle;
 	entry->counter		   = &uobj->comp_events_reported;
 
 	list_add_tail(&entry->list, &ev_queue->event_list);
@@ -457,8 +457,7 @@ static void ib_uverbs_async_handler(struct ib_uverbs_file *file,
 
 void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr)
 {
-	struct ib_uevent_object *uobj = container_of(
-		event->element.cq->uobject, struct ib_uevent_object, uobject);
+	struct ib_uevent_object *uobj = &event->element.cq->uobject->uevent;
 
 	ib_uverbs_async_handler(uobj->uobject.ufile, uobj->uobject.user_handle,
 				event->event, &uobj->event_list,
diff --git a/drivers/infiniband/core/uverbs_std_types_cq.c b/drivers/infiniband/core/uverbs_std_types_cq.c
index fbc605a95e51..a41c758042cc 100644
--- a/drivers/infiniband/core/uverbs_std_types_cq.c
+++ b/drivers/infiniband/core/uverbs_std_types_cq.c
@@ -116,7 +116,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)(
 	}
 
 	cq->device        = ib_dev;
-	cq->uobject       = &obj->uevent.uobject;
+	cq->uobject       = obj;
 	cq->comp_handler  = ib_uverbs_comp_handler;
 	cq->event_handler = ib_uverbs_cq_event_handler;
 	cq->cq_context    = ev_file ? &ev_file->ev_queue : NULL;
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index cea4e198701e..e9ab986ab323 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -77,6 +77,8 @@ extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
 extern struct workqueue_struct *ib_comp_unbound_wq;
 
+struct ib_ucq_object;
+
 __printf(3, 4) __cold
 void ibdev_printk(const char *level, const struct ib_device *ibdev,
 		  const char *format, ...);
@@ -1544,7 +1546,7 @@ enum ib_poll_context {
 
 struct ib_cq {
 	struct ib_device       *device;
-	struct ib_uobject      *uobject;
+	struct ib_ucq_object   *uobject;
 	ib_comp_handler   	comp_handler;
 	void                  (*event_handler)(struct ib_event *, void *);
 	void                   *cq_context;
-- 
cgit v1.2.3-59-g8ed1b


From 620d3f8176cbb3a9c0c7962a05fb15310a9998d4 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:21:59 +0200
Subject: RDMA/core: Do not erase the type of ib_qp.uobject

This is a struct ib_uqp_object pointer, instead of using container_of()
all over the place just store it with its actual type.

Link: https://lore.kernel.org/r/1578504126-9400-8-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/core_priv.h   |  2 +-
 drivers/infiniband/core/uverbs_cmd.c  | 34 +++++++++++++++++++++-------------
 drivers/infiniband/core/uverbs_main.c |  3 +--
 include/rdma/ib_verbs.h               |  3 ++-
 4 files changed, 25 insertions(+), 17 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h
index d657d90e618b..b1457b3464d3 100644
--- a/drivers/infiniband/core/core_priv.h
+++ b/drivers/infiniband/core/core_priv.h
@@ -321,7 +321,7 @@ static inline struct ib_qp *_ib_create_qp(struct ib_device *dev,
 					  struct ib_pd *pd,
 					  struct ib_qp_init_attr *attr,
 					  struct ib_udata *udata,
-					  struct ib_uobject *uobj)
+					  struct ib_uqp_object *uobj)
 {
 	enum ib_qp_type qp_type = attr->qp_type;
 	struct ib_qp *qp;
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index b08679a5da25..4d84d083847c 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1424,7 +1424,7 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 		qp = ib_create_qp(pd, &attr);
 	else
 		qp = _ib_create_qp(device, pd, &attr, &attrs->driver_udata,
-				   &obj->uevent.uobject);
+				   obj);
 
 	if (IS_ERR(qp)) {
 		ret = PTR_ERR(qp);
@@ -1457,7 +1457,7 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 			atomic_inc(&ind_tbl->usecnt);
 	} else {
 		/* It is done in _ib_create_qp for other QP types */
-		qp->uobject = &obj->uevent.uobject;
+		qp->uobject = obj;
 	}
 
 	obj->uevent.uobject.object = qp;
@@ -1628,7 +1628,7 @@ static int ib_uverbs_open_qp(struct uverbs_attr_bundle *attrs)
 
 	obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, uobject);
 	atomic_inc(&obj->uxrcd->refcnt);
-	qp->uobject = &obj->uevent.uobject;
+	qp->uobject = obj;
 	uobj_put_read(xrcd_uobj);
 
 	rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs);
@@ -1693,7 +1693,8 @@ static int ib_uverbs_query_qp(struct uverbs_attr_bundle *attrs)
 
 	ret = ib_query_qp(qp, attr, cmd.attr_mask, init_attr);
 
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 
 	if (ret)
 		goto out;
@@ -1930,7 +1931,8 @@ static int modify_qp(struct uverbs_attr_bundle *attrs,
 				      &attrs->driver_udata);
 
 release_qp:
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 out:
 	kfree(attr);
 
@@ -2194,7 +2196,8 @@ static int ib_uverbs_post_send(struct uverbs_attr_bundle *attrs)
 		ret = ret2;
 
 out_put:
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 
 	while (wr) {
 		if (is_ud && ud_wr(wr)->ah)
@@ -2336,7 +2339,8 @@ static int ib_uverbs_post_recv(struct uverbs_attr_bundle *attrs)
 	resp.bad_wr = 0;
 	ret = qp->device->ops.post_recv(qp->real_qp, wr, &bad_wr);
 
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	if (ret) {
 		for (next = wr; next; next = next->next) {
 			++resp.bad_wr;
@@ -2517,7 +2521,7 @@ static int ib_uverbs_attach_mcast(struct uverbs_attr_bundle *attrs)
 	if (!qp)
 		return -EINVAL;
 
-	obj = container_of(qp->uobject, struct ib_uqp_object, uevent.uobject);
+	obj = qp->uobject;
 
 	mutex_lock(&obj->mcast_lock);
 	list_for_each_entry(mcast, &obj->mcast_list, list)
@@ -2544,7 +2548,8 @@ static int ib_uverbs_attach_mcast(struct uverbs_attr_bundle *attrs)
 
 out_put:
 	mutex_unlock(&obj->mcast_lock);
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 
 	return ret;
 }
@@ -2566,7 +2571,7 @@ static int ib_uverbs_detach_mcast(struct uverbs_attr_bundle *attrs)
 	if (!qp)
 		return -EINVAL;
 
-	obj = container_of(qp->uobject, struct ib_uqp_object, uevent.uobject);
+	obj = qp->uobject;
 	mutex_lock(&obj->mcast_lock);
 
 	list_for_each_entry(mcast, &obj->mcast_list, list)
@@ -2587,7 +2592,8 @@ static int ib_uverbs_detach_mcast(struct uverbs_attr_bundle *attrs)
 
 out_put:
 	mutex_unlock(&obj->mcast_lock);
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	return ret;
 }
 
@@ -3339,7 +3345,8 @@ static int ib_uverbs_ex_create_flow(struct uverbs_attr_bundle *attrs)
 	if (err)
 		goto err_copy;
 
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	kfree(flow_attr);
 	if (cmd.flow_attr.num_of_specs)
 		kfree(kern_flow_attr);
@@ -3353,7 +3360,8 @@ err_free:
 err_free_flow_attr:
 	kfree(flow_attr);
 err_put:
-	uobj_put_obj_read(qp);
+	rdma_lookup_put_uobject(&qp->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 err_uobj:
 	uobj_alloc_abort(uobj, attrs);
 err_free_attr:
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 2849bf40bc97..12da68fca1c1 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -472,8 +472,7 @@ void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr)
 	if (!event->element.qp->uobject)
 		return;
 
-	uobj = container_of(event->element.qp->uobject,
-			    struct ib_uevent_object, uobject);
+	uobj = &event->element.qp->uobject->uevent;
 
 	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
 				event->event, &uobj->event_list,
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index e9ab986ab323..b4dbc5f9636a 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -72,6 +72,7 @@
 #define IB_FW_VERSION_NAME_MAX	ETHTOOL_FWVERS_LEN
 
 struct ib_umem_odp;
+struct ib_uqp_object;
 
 extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
@@ -1735,7 +1736,7 @@ struct ib_qp {
 	atomic_t		usecnt;
 	struct list_head	open_list;
 	struct ib_qp           *real_qp;
-	struct ib_uobject      *uobject;
+	struct ib_uqp_object   *uobject;
 	void                  (*event_handler)(struct ib_event *, void *);
 	void		       *qp_context;
 	/* sgid_attrs associated with the AV's */
-- 
cgit v1.2.3-59-g8ed1b


From 9fbe334c6a67c3c09f187e4b9b0e6eaf0ad31429 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:22:00 +0200
Subject: RDMA/core: Do not erase the type of ib_srq.uobject

This is a struct ib_usrq_object pointer, instead of using container_of()
all over the place just store it with its actual type.

Link: https://lore.kernel.org/r/1578504126-9400-9-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs_cmd.c  | 17 +++++++++++------
 drivers/infiniband/core/uverbs_main.c |  3 +--
 include/rdma/ib_verbs.h               |  3 ++-
 3 files changed, 14 insertions(+), 9 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 4d84d083847c..8350e023d3b5 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1492,7 +1492,8 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 		rdma_lookup_put_uobject(&rcq->uobject->uevent.uobject,
 					UVERBS_LOOKUP_READ);
 	if (srq)
-		uobj_put_obj_read(srq);
+		rdma_lookup_put_uobject(&srq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 	if (ind_tbl)
 		uobj_put_obj_read(ind_tbl);
 
@@ -1513,7 +1514,8 @@ err_put:
 		rdma_lookup_put_uobject(&rcq->uobject->uevent.uobject,
 					UVERBS_LOOKUP_READ);
 	if (srq)
-		uobj_put_obj_read(srq);
+		rdma_lookup_put_uobject(&srq->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 	if (ind_tbl)
 		uobj_put_obj_read(ind_tbl);
 
@@ -2390,7 +2392,8 @@ static int ib_uverbs_post_srq_recv(struct uverbs_attr_bundle *attrs)
 	resp.bad_wr = 0;
 	ret = srq->device->ops.post_srq_recv(srq, wr, &bad_wr);
 
-	uobj_put_obj_read(srq);
+	rdma_lookup_put_uobject(&srq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 
 	if (ret)
 		for (next = wr; next; next = next->next) {
@@ -3458,7 +3461,7 @@ static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs,
 	srq->device        = pd->device;
 	srq->pd            = pd;
 	srq->srq_type	   = cmd->srq_type;
-	srq->uobject       = &obj->uevent.uobject;
+	srq->uobject       = obj;
 	srq->event_handler = attr.event_handler;
 	srq->srq_context   = attr.srq_context;
 
@@ -3584,7 +3587,8 @@ static int ib_uverbs_modify_srq(struct uverbs_attr_bundle *attrs)
 	ret = srq->device->ops.modify_srq(srq, &attr, cmd.attr_mask,
 					  &attrs->driver_udata);
 
-	uobj_put_obj_read(srq);
+	rdma_lookup_put_uobject(&srq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 
 	return ret;
 }
@@ -3607,7 +3611,8 @@ static int ib_uverbs_query_srq(struct uverbs_attr_bundle *attrs)
 
 	ret = ib_query_srq(srq, &attr);
 
-	uobj_put_obj_read(srq);
+	rdma_lookup_put_uobject(&srq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 
 	if (ret)
 		return ret;
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 12da68fca1c1..a501f76bc2f9 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -493,8 +493,7 @@ void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr)
 {
 	struct ib_uevent_object *uobj;
 
-	uobj = container_of(event->element.srq->uobject,
-			    struct ib_uevent_object, uobject);
+	uobj = &event->element.srq->uobject->uevent;
 
 	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
 				event->event, &uobj->event_list,
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index b4dbc5f9636a..7990b55b1b40 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -73,6 +73,7 @@
 
 struct ib_umem_odp;
 struct ib_uqp_object;
+struct ib_usrq_object;
 
 extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
@@ -1575,7 +1576,7 @@ struct ib_cq {
 struct ib_srq {
 	struct ib_device       *device;
 	struct ib_pd	       *pd;
-	struct ib_uobject      *uobject;
+	struct ib_usrq_object  *uobject;
 	void		      (*event_handler)(struct ib_event *, void *);
 	void		       *srq_context;
 	enum ib_srq_type	srq_type;
-- 
cgit v1.2.3-59-g8ed1b


From e04dd13159b0ddc0ff7f5e110bf99af3c65fabd3 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:22:01 +0200
Subject: RDMA/core: Do not erase the type of ib_wq.uobject

This is a struct ib_uwq_object pointer, instead of using container_of()
all over the place just store it with its actual type.

Link: https://lore.kernel.org/r/1578504126-9400-10-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs_cmd.c  | 13 ++++++++-----
 drivers/infiniband/core/uverbs_main.c |  3 +--
 include/rdma/ib_verbs.h               |  3 ++-
 3 files changed, 11 insertions(+), 8 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 8350e023d3b5..66f86b4e5e81 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -2971,7 +2971,7 @@ static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs)
 		goto err_put_cq;
 	}
 
-	wq->uobject = &obj->uevent.uobject;
+	wq->uobject = obj;
 	obj->uevent.uobject.object = wq;
 	wq->wq_type = wq_init_attr.wq_type;
 	wq->cq = cq;
@@ -2981,7 +2981,7 @@ static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs)
 	atomic_set(&wq->usecnt, 0);
 	atomic_inc(&pd->usecnt);
 	atomic_inc(&cq->usecnt);
-	wq->uobject = &obj->uevent.uobject;
+	wq->uobject = obj;
 	obj->uevent.uobject.object = wq;
 
 	memset(&resp, 0, sizeof(resp));
@@ -3070,7 +3070,8 @@ static int ib_uverbs_ex_modify_wq(struct uverbs_attr_bundle *attrs)
 	}
 	ret = wq->device->ops.modify_wq(wq, &wq_attr, cmd.attr_mask,
 					&attrs->driver_udata);
-	uobj_put_obj_read(wq);
+	rdma_lookup_put_uobject(&wq->uobject->uevent.uobject,
+				UVERBS_LOOKUP_READ);
 	return ret;
 }
 
@@ -3171,7 +3172,8 @@ static int ib_uverbs_ex_create_rwq_ind_table(struct uverbs_attr_bundle *attrs)
 	kfree(wqs_handles);
 
 	for (j = 0; j < num_read_wqs; j++)
-		uobj_put_obj_read(wqs[j]);
+		rdma_lookup_put_uobject(&wqs[j]->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 
 	rdma_alloc_commit_uobject(uobj, attrs);
 	return 0;
@@ -3182,7 +3184,8 @@ err_uobj:
 	uobj_alloc_abort(uobj, attrs);
 put_wqs:
 	for (j = 0; j < num_read_wqs; j++)
-		uobj_put_obj_read(wqs[j]);
+		rdma_lookup_put_uobject(&wqs[j]->uobject->uevent.uobject,
+					UVERBS_LOOKUP_READ);
 err_free:
 	kfree(wqs_handles);
 	kfree(wqs);
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index a501f76bc2f9..87f7a94adf57 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -481,8 +481,7 @@ void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr)
 
 void ib_uverbs_wq_event_handler(struct ib_event *event, void *context_ptr)
 {
-	struct ib_uevent_object *uobj = container_of(event->element.wq->uobject,
-						  struct ib_uevent_object, uobject);
+	struct ib_uevent_object *uobj = &event->element.wq->uobject->uevent;
 
 	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
 				event->event, &uobj->event_list,
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 7990b55b1b40..d8031f6f327e 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -74,6 +74,7 @@
 struct ib_umem_odp;
 struct ib_uqp_object;
 struct ib_usrq_object;
+struct ib_uwq_object;
 
 extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
@@ -1621,7 +1622,7 @@ enum ib_wq_state {
 
 struct ib_wq {
 	struct ib_device       *device;
-	struct ib_uobject      *uobject;
+	struct ib_uwq_object   *uobject;
 	void		    *wq_context;
 	void		    (*event_handler)(struct ib_event *, void *);
 	struct ib_pd	       *pd;
-- 
cgit v1.2.3-59-g8ed1b


From 817d65765069d1e7dee97473a6465947db1bea80 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:22:02 +0200
Subject: RDMA/core: Simplify type usage for ib_uverbs_async_handler()

This function works on an ib_uverbs_async_file. Accept that as a parameter
instead of the struct ib_uverbs_file.

Consoldiate all the callers working from an ib_uevent_object to a single
function and locate the async_file directly from the struct ib_uobject
instead of using context_ptr.

Link: https://lore.kernel.org/r/1578504126-9400-11-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs.h      |  2 -
 drivers/infiniband/core/uverbs_cmd.c  |  9 +---
 drivers/infiniband/core/uverbs_main.c | 79 +++++++++++++++--------------------
 3 files changed, 34 insertions(+), 56 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index 9fa044699cf0..8384b66b661c 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -235,8 +235,6 @@ void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr);
 void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr);
 void ib_uverbs_wq_event_handler(struct ib_event *event, void *context_ptr);
 void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr);
-void ib_uverbs_event_handler(struct ib_event_handler *handler,
-			     struct ib_event *event);
 int ib_uverbs_dealloc_xrcd(struct ib_uobject *uobject, struct ib_xrcd *xrcd,
 			   enum rdma_remove_reason why,
 			   struct uverbs_attr_bundle *attrs);
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 66f86b4e5e81..51117e784d40 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1378,7 +1378,6 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 	}
 
 	attr.event_handler = ib_uverbs_qp_event_handler;
-	attr.qp_context    = attrs->ufile;
 	attr.send_cq       = scq;
 	attr.recv_cq       = rcq;
 	attr.srq           = srq;
@@ -1394,7 +1393,6 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 	attr.cap.max_recv_sge    = cmd->max_recv_sge;
 	attr.cap.max_inline_data = cmd->max_inline_data;
 
-	obj->uevent.events_reported     = 0;
 	INIT_LIST_HEAD(&obj->uevent.event_list);
 	INIT_LIST_HEAD(&obj->mcast_list);
 
@@ -1442,7 +1440,6 @@ static int create_qp(struct uverbs_attr_bundle *attrs,
 		qp->srq		  = attr.srq;
 		qp->rwq_ind_tbl	  = ind_tbl;
 		qp->event_handler = attr.event_handler;
-		qp->qp_context	  = attr.qp_context;
 		qp->qp_type	  = attr.qp_type;
 		atomic_set(&qp->usecnt, 0);
 		atomic_inc(&pd->usecnt);
@@ -1577,7 +1574,7 @@ static int ib_uverbs_open_qp(struct uverbs_attr_bundle *attrs)
 	struct ib_xrcd		       *xrcd;
 	struct ib_uobject	       *uninitialized_var(xrcd_uobj);
 	struct ib_qp                   *qp;
-	struct ib_qp_open_attr          attr;
+	struct ib_qp_open_attr          attr = {};
 	int ret;
 	struct ib_device *ib_dev;
 
@@ -1603,11 +1600,9 @@ static int ib_uverbs_open_qp(struct uverbs_attr_bundle *attrs)
 	}
 
 	attr.event_handler = ib_uverbs_qp_event_handler;
-	attr.qp_context    = attrs->ufile;
 	attr.qp_num        = cmd.qpn;
 	attr.qp_type       = cmd.qp_type;
 
-	obj->uevent.events_reported = 0;
 	INIT_LIST_HEAD(&obj->uevent.event_list);
 	INIT_LIST_HEAD(&obj->mcast_list);
 
@@ -2962,7 +2957,6 @@ static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs)
 	wq_init_attr.wq_type = cmd.wq_type;
 	wq_init_attr.event_handler = ib_uverbs_wq_event_handler;
 	wq_init_attr.create_flags = cmd.create_flags;
-	obj->uevent.events_reported = 0;
 	INIT_LIST_HEAD(&obj->uevent.event_list);
 
 	wq = pd->device->ops.create_wq(pd, &wq_init_attr, &attrs->driver_udata);
@@ -3452,7 +3446,6 @@ static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs,
 	attr.attr.max_sge   = cmd->max_sge;
 	attr.attr.srq_limit = cmd->srq_limit;
 
-	obj->uevent.events_reported = 0;
 	INIT_LIST_HEAD(&obj->uevent.event_list);
 
 	srq = rdma_zalloc_drv_obj(ib_dev, ib_srq);
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 87f7a94adf57..97770e72392f 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -421,92 +421,79 @@ void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context)
 	kill_fasync(&ev_queue->async_queue, SIGIO, POLL_IN);
 }
 
-static void ib_uverbs_async_handler(struct ib_uverbs_file *file,
-				    __u64 element, __u64 event,
-				    struct list_head *obj_list,
-				    u32 *counter)
+static void
+ib_uverbs_async_handler(struct ib_uverbs_async_event_file *async_file,
+			__u64 element, __u64 event, struct list_head *obj_list,
+			u32 *counter)
 {
 	struct ib_uverbs_event *entry;
 	unsigned long flags;
 
-	spin_lock_irqsave(&file->async_file->ev_queue.lock, flags);
-	if (file->async_file->ev_queue.is_closed) {
-		spin_unlock_irqrestore(&file->async_file->ev_queue.lock, flags);
+	spin_lock_irqsave(&async_file->ev_queue.lock, flags);
+	if (async_file->ev_queue.is_closed) {
+		spin_unlock_irqrestore(&async_file->ev_queue.lock, flags);
 		return;
 	}
 
 	entry = kmalloc(sizeof(*entry), GFP_ATOMIC);
 	if (!entry) {
-		spin_unlock_irqrestore(&file->async_file->ev_queue.lock, flags);
+		spin_unlock_irqrestore(&async_file->ev_queue.lock, flags);
 		return;
 	}
 
-	entry->desc.async.element    = element;
+	entry->desc.async.element = element;
 	entry->desc.async.event_type = event;
-	entry->desc.async.reserved   = 0;
-	entry->counter               = counter;
+	entry->desc.async.reserved = 0;
+	entry->counter = counter;
 
-	list_add_tail(&entry->list, &file->async_file->ev_queue.event_list);
+	list_add_tail(&entry->list, &async_file->ev_queue.event_list);
 	if (obj_list)
 		list_add_tail(&entry->obj_list, obj_list);
-	spin_unlock_irqrestore(&file->async_file->ev_queue.lock, flags);
+	spin_unlock_irqrestore(&async_file->ev_queue.lock, flags);
 
-	wake_up_interruptible(&file->async_file->ev_queue.poll_wait);
-	kill_fasync(&file->async_file->ev_queue.async_queue, SIGIO, POLL_IN);
+	wake_up_interruptible(&async_file->ev_queue.poll_wait);
+	kill_fasync(&async_file->ev_queue.async_queue, SIGIO, POLL_IN);
 }
 
-void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr)
+static void uverbs_uobj_event(struct ib_uevent_object *eobj,
+			      struct ib_event *event)
 {
-	struct ib_uevent_object *uobj = &event->element.cq->uobject->uevent;
+	ib_uverbs_async_handler(eobj->uobject.ufile->async_file,
+				eobj->uobject.user_handle, event->event,
+				&eobj->event_list, &eobj->events_reported);
+}
 
-	ib_uverbs_async_handler(uobj->uobject.ufile, uobj->uobject.user_handle,
-				event->event, &uobj->event_list,
-				&uobj->events_reported);
+void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr)
+{
+	uverbs_uobj_event(&event->element.cq->uobject->uevent, event);
 }
 
 void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr)
 {
-	struct ib_uevent_object *uobj;
-
 	/* for XRC target qp's, check that qp is live */
 	if (!event->element.qp->uobject)
 		return;
 
-	uobj = &event->element.qp->uobject->uevent;
-
-	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
-				event->event, &uobj->event_list,
-				&uobj->events_reported);
+	uverbs_uobj_event(&event->element.qp->uobject->uevent, event);
 }
 
 void ib_uverbs_wq_event_handler(struct ib_event *event, void *context_ptr)
 {
-	struct ib_uevent_object *uobj = &event->element.wq->uobject->uevent;
-
-	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
-				event->event, &uobj->event_list,
-				&uobj->events_reported);
+	uverbs_uobj_event(&event->element.wq->uobject->uevent, event);
 }
 
 void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr)
 {
-	struct ib_uevent_object *uobj;
-
-	uobj = &event->element.srq->uobject->uevent;
-
-	ib_uverbs_async_handler(context_ptr, uobj->uobject.user_handle,
-				event->event, &uobj->event_list,
-				&uobj->events_reported);
+	uverbs_uobj_event(&event->element.srq->uobject->uevent, event);
 }
 
-void ib_uverbs_event_handler(struct ib_event_handler *handler,
-			     struct ib_event *event)
+static void ib_uverbs_event_handler(struct ib_event_handler *handler,
+				    struct ib_event *event)
 {
-	struct ib_uverbs_file *file =
-		container_of(handler, struct ib_uverbs_file, event_handler);
-
-	ib_uverbs_async_handler(file, event->element.port_num, event->event,
-				NULL, NULL);
+	ib_uverbs_async_handler(
+		container_of(handler, struct ib_uverbs_file, event_handler)
+			->async_file,
+		event->element.port_num, event->event, NULL, NULL);
 }
 
 void ib_uverbs_free_async_event_file(struct ib_uverbs_file *file)
-- 
cgit v1.2.3-59-g8ed1b


From 39e83af817610e80a9582df94d44173bcb6f85e4 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:22:04 +0200
Subject: RDMA/core: Remove the ufile arg from rdma_alloc_begin_uobject

Now that all callers provide a non-NULL attrs the ufile is redundant.
Adjust things so that the context handling is done inside alloc_uobj,
and the ib_uverbs_get_ucontext_file() is avoided if we already have the
context.

Link: https://lore.kernel.org/r/1578504126-9400-13-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/rdma_core.c | 36 +++++++++++++++++++-----------------
 include/rdma/uverbs_std_types.h     |  3 +--
 include/rdma/uverbs_types.h         |  3 +--
 3 files changed, 21 insertions(+), 21 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index a9f5263c9559..50251da123f7 100644
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -256,15 +256,20 @@ int __uobj_perform_destroy(const struct uverbs_api_object *obj, u32 id,
 }
 
 /* alloc_uobj must be undone by uverbs_destroy_uobject() */
-static struct ib_uobject *alloc_uobj(struct ib_uverbs_file *ufile,
+static struct ib_uobject *alloc_uobj(struct uverbs_attr_bundle *attrs,
 				     const struct uverbs_api_object *obj)
 {
+	struct ib_uverbs_file *ufile = attrs->ufile;
 	struct ib_uobject *uobj;
-	struct ib_ucontext *ucontext;
 
-	ucontext = ib_uverbs_get_ucontext_file(ufile);
-	if (IS_ERR(ucontext))
-		return ERR_CAST(ucontext);
+	if (!attrs->context) {
+		struct ib_ucontext *ucontext =
+			ib_uverbs_get_ucontext_file(ufile);
+
+		if (IS_ERR(ucontext))
+			return ERR_CAST(ucontext);
+		attrs->context = ucontext;
+	}
 
 	uobj = kzalloc(obj->type_attrs->obj_size, GFP_KERNEL);
 	if (!uobj)
@@ -274,7 +279,7 @@ static struct ib_uobject *alloc_uobj(struct ib_uverbs_file *ufile,
 	 * The object is added to the list in the commit stage.
 	 */
 	uobj->ufile = ufile;
-	uobj->context = ucontext;
+	uobj->context = attrs->context;
 	INIT_LIST_HEAD(&uobj->list);
 	uobj->uapi_object = obj;
 	/*
@@ -417,12 +422,12 @@ free:
 
 static struct ib_uobject *
 alloc_begin_idr_uobject(const struct uverbs_api_object *obj,
-			struct ib_uverbs_file *ufile)
+			struct uverbs_attr_bundle *attrs)
 {
 	int ret;
 	struct ib_uobject *uobj;
 
-	uobj = alloc_uobj(ufile, obj);
+	uobj = alloc_uobj(attrs, obj);
 	if (IS_ERR(uobj))
 		return uobj;
 
@@ -438,7 +443,7 @@ alloc_begin_idr_uobject(const struct uverbs_api_object *obj,
 	return uobj;
 
 remove:
-	xa_erase(&ufile->idr, uobj->id);
+	xa_erase(&attrs->ufile->idr, uobj->id);
 uobj_put:
 	uverbs_uobject_put(uobj);
 	return ERR_PTR(ret);
@@ -446,7 +451,7 @@ uobj_put:
 
 static struct ib_uobject *
 alloc_begin_fd_uobject(const struct uverbs_api_object *obj,
-		       struct ib_uverbs_file *ufile)
+		       struct uverbs_attr_bundle *attrs)
 {
 	const struct uverbs_obj_fd_type *fd_type =
 		container_of(obj->type_attrs, struct uverbs_obj_fd_type, type);
@@ -461,7 +466,7 @@ alloc_begin_fd_uobject(const struct uverbs_api_object *obj,
 	if (new_fd < 0)
 		return ERR_PTR(new_fd);
 
-	uobj = alloc_uobj(ufile, obj);
+	uobj = alloc_uobj(attrs, obj);
 	if (IS_ERR(uobj))
 		goto err_fd;
 
@@ -475,7 +480,6 @@ alloc_begin_fd_uobject(const struct uverbs_api_object *obj,
 	uobj->object = filp;
 
 	uobj->id = new_fd;
-	uobj->ufile = ufile;
 	return uobj;
 
 err_uobj:
@@ -486,9 +490,9 @@ err_fd:
 }
 
 struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_api_object *obj,
-					    struct ib_uverbs_file *ufile,
 					    struct uverbs_attr_bundle *attrs)
 {
+	struct ib_uverbs_file *ufile = attrs->ufile;
 	struct ib_uobject *ret;
 
 	if (IS_ERR(obj))
@@ -502,13 +506,11 @@ struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_api_object *obj,
 	if (!down_read_trylock(&ufile->hw_destroy_rwsem))
 		return ERR_PTR(-EIO);
 
-	ret = obj->type_class->alloc_begin(obj, ufile);
+	ret = obj->type_class->alloc_begin(obj, attrs);
 	if (IS_ERR(ret)) {
 		up_read(&ufile->hw_destroy_rwsem);
 		return ret;
 	}
-	if (attrs)
-		attrs->context = ret->context;
 	return ret;
 }
 
@@ -930,7 +932,7 @@ uverbs_get_uobject_from_file(u16 object_id, enum uverbs_obj_access access,
 		return rdma_lookup_get_uobject(obj, attrs->ufile, id,
 					       UVERBS_LOOKUP_WRITE, attrs);
 	case UVERBS_ACCESS_NEW:
-		return rdma_alloc_begin_uobject(obj, attrs->ufile, attrs);
+		return rdma_alloc_begin_uobject(obj, attrs);
 	default:
 		WARN_ON(true);
 		return ERR_PTR(-EOPNOTSUPP);
diff --git a/include/rdma/uverbs_std_types.h b/include/rdma/uverbs_std_types.h
index c6bcaad4a8e0..1b28ce1aba07 100644
--- a/include/rdma/uverbs_std_types.h
+++ b/include/rdma/uverbs_std_types.h
@@ -114,8 +114,7 @@ static inline struct ib_uobject *
 __uobj_alloc(const struct uverbs_api_object *obj,
 	     struct uverbs_attr_bundle *attrs, struct ib_device **ib_dev)
 {
-	struct ib_uobject *uobj =
-		rdma_alloc_begin_uobject(obj, attrs->ufile, attrs);
+	struct ib_uobject *uobj = rdma_alloc_begin_uobject(obj, attrs);
 
 	if (!IS_ERR(uobj))
 		*ib_dev = attrs->context->device;
diff --git a/include/rdma/uverbs_types.h b/include/rdma/uverbs_types.h
index 01d59df4e9e7..f1cbdae67250 100644
--- a/include/rdma/uverbs_types.h
+++ b/include/rdma/uverbs_types.h
@@ -83,7 +83,7 @@ enum rdma_lookup_mode {
  */
 struct uverbs_obj_type_class {
 	struct ib_uobject *(*alloc_begin)(const struct uverbs_api_object *obj,
-					  struct ib_uverbs_file *ufile);
+					  struct uverbs_attr_bundle *attrs);
 	/* This consumes the kref on uobj */
 	void (*alloc_commit)(struct ib_uobject *uobj);
 	/* This does not consume the kref on uobj */
@@ -137,7 +137,6 @@ struct ib_uobject *rdma_lookup_get_uobject(const struct uverbs_api_object *obj,
 void rdma_lookup_put_uobject(struct ib_uobject *uobj,
 			     enum rdma_lookup_mode mode);
 struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_api_object *obj,
-					    struct ib_uverbs_file *ufile,
 					    struct uverbs_attr_bundle *attrs);
 void rdma_alloc_abort_uobject(struct ib_uobject *uobj,
 			      struct uverbs_attr_bundle *attrs);
-- 
cgit v1.2.3-59-g8ed1b


From 3e032c0e92aa0c4c0b46c5e2d6d41706c8fce488 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:22:05 +0200
Subject: RDMA/core: Make ib_uverbs_async_event_file into a uobject

This makes async events aligned with completion events as both are full
uobjects of FD type and use the same uobject lifecycle.

A bunch of duplicate code is consolidated and the general flow between the
two FDs is now very similar.

Link: https://lore.kernel.org/r/1578504126-9400-14-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/Makefile                   |   3 +-
 drivers/infiniband/core/rdma_core.h                |   1 +
 drivers/infiniband/core/uverbs.h                   |  15 +--
 drivers/infiniband/core/uverbs_cmd.c               |  30 ++---
 drivers/infiniband/core/uverbs_main.c              | 130 ++++-----------------
 drivers/infiniband/core/uverbs_std_types.c         |  25 ++--
 .../infiniband/core/uverbs_std_types_async_fd.c    |  33 ++++++
 drivers/infiniband/core/uverbs_uapi.c              |   1 +
 include/uapi/rdma/ib_user_ioctl_cmds.h             |   1 +
 9 files changed, 96 insertions(+), 143 deletions(-)
 create mode 100644 drivers/infiniband/core/uverbs_std_types_async_fd.c

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile
index 2b86a514c9bf..d1b14887960e 100644
--- a/drivers/infiniband/core/Makefile
+++ b/drivers/infiniband/core/Makefile
@@ -35,6 +35,7 @@ ib_uverbs-y :=			uverbs_main.o uverbs_cmd.o uverbs_marshall.o \
 				uverbs_std_types_cq.o \
 				uverbs_std_types_flow_action.o uverbs_std_types_dm.o \
 				uverbs_std_types_mr.o uverbs_std_types_counters.o \
-				uverbs_uapi.o uverbs_std_types_device.o
+				uverbs_uapi.o uverbs_std_types_device.o \
+				uverbs_std_types_async_fd.o
 ib_uverbs-$(CONFIG_INFINIBAND_USER_MEM) += umem.o
 ib_uverbs-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += umem_odp.o
diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h
index 29f905e8c2a8..33978e0f1262 100644
--- a/drivers/infiniband/core/rdma_core.h
+++ b/drivers/infiniband/core/rdma_core.h
@@ -151,6 +151,7 @@ void uapi_compute_bundle_size(struct uverbs_api_ioctl_method *method_elm,
 			      unsigned int num_attrs);
 void uverbs_user_mmap_disassociate(struct ib_uverbs_file *ufile);
 
+extern const struct uapi_definition uverbs_def_obj_async_fd[];
 extern const struct uapi_definition uverbs_def_obj_counters[];
 extern const struct uapi_definition uverbs_def_obj_cq[];
 extern const struct uapi_definition uverbs_def_obj_device[];
diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index 8384b66b661c..ccde5d20a6cf 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -111,7 +111,6 @@ struct ib_uverbs_device {
 	struct srcu_struct			disassociate_srcu;
 	struct mutex				lists_mutex; /* protect lists */
 	struct list_head			uverbs_file_list;
-	struct list_head			uverbs_events_file_list;
 	struct uverbs_api			*uapi;
 };
 
@@ -124,10 +123,9 @@ struct ib_uverbs_event_queue {
 };
 
 struct ib_uverbs_async_event_file {
+	struct ib_uobject			uobj;
 	struct ib_uverbs_event_queue		ev_queue;
-	struct ib_uverbs_file		       *uverbs_file;
-	struct kref				ref;
-	struct list_head			list;
+	struct ib_event_handler			event_handler;
 };
 
 struct ib_uverbs_completion_event_file {
@@ -144,8 +142,7 @@ struct ib_uverbs_file {
 	 * ucontext_lock held
 	 */
 	struct ib_ucontext		       *ucontext;
-	struct ib_event_handler			event_handler;
-	struct ib_uverbs_async_event_file       *async_file;
+	struct ib_uverbs_async_event_file      *async_file;
 	struct list_head			list;
 
 	/*
@@ -217,10 +214,10 @@ struct ib_ucq_object {
 };
 
 extern const struct file_operations uverbs_event_fops;
+extern const struct file_operations uverbs_async_event_fops;
 void ib_uverbs_init_event_queue(struct ib_uverbs_event_queue *ev_queue);
-struct file *ib_uverbs_alloc_async_event_file(struct ib_uverbs_file *uverbs_file,
-					      struct ib_device *ib_dev);
-void ib_uverbs_free_async_event_file(struct ib_uverbs_file *uverbs_file);
+void ib_uverbs_init_async_event_file(struct ib_uverbs_async_event_file *ev_file);
+void ib_uverbs_free_event_queue(struct ib_uverbs_event_queue *event_queue);
 void ib_uverbs_flow_resources_free(struct ib_uflow_resources *uflow_res);
 
 void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 51117e784d40..ced1384d316b 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -209,9 +209,9 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 	struct ib_uverbs_get_context      cmd;
 	struct ib_uverbs_get_context_resp resp;
 	struct ib_ucontext		 *ucontext;
-	struct file			 *filp;
 	struct ib_rdmacg_object		 cg_obj;
 	struct ib_device *ib_dev;
+	struct ib_uobject *uobj;
 	int ret;
 
 	ret = uverbs_request(attrs, &cmd, sizeof(cmd));
@@ -254,30 +254,28 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 
 	xa_init_flags(&ucontext->mmap_xa, XA_FLAGS_ALLOC);
 
-	ret = get_unused_fd_flags(O_CLOEXEC);
-	if (ret < 0)
+	uobj = uobj_alloc(UVERBS_OBJECT_ASYNC_EVENT, attrs, &ib_dev);
+	if (IS_ERR(uobj)) {
+		ret = PTR_ERR(uobj);
 		goto err_free;
-	resp.async_fd = ret;
-
-	filp = ib_uverbs_alloc_async_event_file(file, ib_dev);
-	if (IS_ERR(filp)) {
-		ret = PTR_ERR(filp);
-		goto err_fd;
 	}
 
+	resp.async_fd = uobj->id;
 	resp.num_comp_vectors = file->device->num_comp_vectors;
 
 	ret = uverbs_response(attrs, &resp, sizeof(resp));
 	if (ret)
-		goto err_file;
+		goto err_uobj;
 
 	ret = ib_dev->ops.alloc_ucontext(ucontext, &attrs->driver_udata);
 	if (ret)
-		goto err_file;
+		goto err_uobj;
 
 	rdma_restrack_uadd(&ucontext->res);
 
-	fd_install(resp.async_fd, filp);
+	ib_uverbs_init_async_event_file(
+		container_of(uobj, struct ib_uverbs_async_event_file, uobj));
+	rdma_alloc_commit_uobject(uobj, attrs);
 
 	/*
 	 * Make sure that ib_uverbs_get_ucontext() sees the pointer update
@@ -289,12 +287,8 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 
 	return 0;
 
-err_file:
-	ib_uverbs_free_async_event_file(file);
-	fput(filp);
-
-err_fd:
-	put_unused_fd(resp.async_fd);
+err_uobj:
+	rdma_alloc_abort_uobject(uobj, attrs);
 
 err_free:
 	kfree(ucontext);
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 97770e72392f..121e65f69c0b 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -125,14 +125,6 @@ static void ib_uverbs_release_dev(struct device *device)
 	kfree(dev);
 }
 
-static void ib_uverbs_release_async_event_file(struct kref *ref)
-{
-	struct ib_uverbs_async_event_file *file =
-		container_of(ref, struct ib_uverbs_async_event_file, ref);
-
-	kfree(file);
-}
-
 void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
 			  struct ib_uverbs_completion_event_file *ev_file,
 			  struct ib_ucq_object *uobj)
@@ -203,8 +195,7 @@ void ib_uverbs_release_file(struct kref *ref)
 		ib_uverbs_comp_dev(file->device);
 
 	if (file->async_file)
-		kref_put(&file->async_file->ref,
-			 ib_uverbs_release_async_event_file);
+		uverbs_uobject_put(&file->async_file->uobj);
 	put_device(&file->device->dev);
 
 	if (file->disassociate_page)
@@ -339,35 +330,6 @@ static int ib_uverbs_comp_event_fasync(int fd, struct file *filp, int on)
 	return fasync_helper(fd, filp, on, &comp_ev_file->ev_queue.async_queue);
 }
 
-static int ib_uverbs_async_event_close(struct inode *inode, struct file *filp)
-{
-	struct ib_uverbs_async_event_file *file = filp->private_data;
-	struct ib_uverbs_file *uverbs_file = file->uverbs_file;
-	struct ib_uverbs_event *entry, *tmp;
-	int closed_already = 0;
-
-	mutex_lock(&uverbs_file->device->lists_mutex);
-	spin_lock_irq(&file->ev_queue.lock);
-	closed_already = file->ev_queue.is_closed;
-	file->ev_queue.is_closed = 1;
-	list_for_each_entry_safe(entry, tmp, &file->ev_queue.event_list, list) {
-		if (entry->counter)
-			list_del(&entry->obj_list);
-		kfree(entry);
-	}
-	spin_unlock_irq(&file->ev_queue.lock);
-	if (!closed_already) {
-		list_del(&file->list);
-		ib_unregister_event_handler(&uverbs_file->event_handler);
-	}
-	mutex_unlock(&uverbs_file->device->lists_mutex);
-
-	kref_put(&uverbs_file->ref, ib_uverbs_release_file);
-	kref_put(&file->ref, ib_uverbs_release_async_event_file);
-
-	return 0;
-}
-
 const struct file_operations uverbs_event_fops = {
 	.owner	 = THIS_MODULE,
 	.read	 = ib_uverbs_comp_event_read,
@@ -377,11 +339,11 @@ const struct file_operations uverbs_event_fops = {
 	.llseek	 = no_llseek,
 };
 
-static const struct file_operations uverbs_async_event_fops = {
+const struct file_operations uverbs_async_event_fops = {
 	.owner	 = THIS_MODULE,
 	.read	 = ib_uverbs_async_event_read,
 	.poll    = ib_uverbs_async_event_poll,
-	.release = ib_uverbs_async_event_close,
+	.release = uverbs_uobject_fd_release,
 	.fasync  = ib_uverbs_async_event_fasync,
 	.llseek	 = no_llseek,
 };
@@ -491,17 +453,11 @@ static void ib_uverbs_event_handler(struct ib_event_handler *handler,
 				    struct ib_event *event)
 {
 	ib_uverbs_async_handler(
-		container_of(handler, struct ib_uverbs_file, event_handler)
-			->async_file,
+		container_of(handler, struct ib_uverbs_async_event_file,
+			     event_handler),
 		event->element.port_num, event->event, NULL, NULL);
 }
 
-void ib_uverbs_free_async_event_file(struct ib_uverbs_file *file)
-{
-	kref_put(&file->async_file->ref, ib_uverbs_release_async_event_file);
-	file->async_file = NULL;
-}
-
 void ib_uverbs_init_event_queue(struct ib_uverbs_event_queue *ev_queue)
 {
 	spin_lock_init(&ev_queue->lock);
@@ -511,45 +467,23 @@ void ib_uverbs_init_event_queue(struct ib_uverbs_event_queue *ev_queue)
 	ev_queue->async_queue = NULL;
 }
 
-struct file *ib_uverbs_alloc_async_event_file(struct ib_uverbs_file *uverbs_file,
-					      struct ib_device	*ib_dev)
+void ib_uverbs_init_async_event_file(
+	struct ib_uverbs_async_event_file *async_file)
 {
-	struct ib_uverbs_async_event_file *ev_file;
-	struct file *filp;
-
-	ev_file = kzalloc(sizeof(*ev_file), GFP_KERNEL);
-	if (!ev_file)
-		return ERR_PTR(-ENOMEM);
-
-	ib_uverbs_init_event_queue(&ev_file->ev_queue);
-	ev_file->uverbs_file = uverbs_file;
-	kref_get(&ev_file->uverbs_file->ref);
-	kref_init(&ev_file->ref);
-	filp = anon_inode_getfile("[infinibandevent]", &uverbs_async_event_fops,
-				  ev_file, O_RDONLY);
-	if (IS_ERR(filp))
-		goto err_put_refs;
-
-	mutex_lock(&uverbs_file->device->lists_mutex);
-	list_add_tail(&ev_file->list,
-		      &uverbs_file->device->uverbs_events_file_list);
-	mutex_unlock(&uverbs_file->device->lists_mutex);
-
-	WARN_ON(uverbs_file->async_file);
-	uverbs_file->async_file = ev_file;
-	kref_get(&uverbs_file->async_file->ref);
-	INIT_IB_EVENT_HANDLER(&uverbs_file->event_handler,
-			      ib_dev,
-			      ib_uverbs_event_handler);
-	ib_register_event_handler(&uverbs_file->event_handler);
-	/* At that point async file stuff was fully set */
+	struct ib_uverbs_file *uverbs_file = async_file->uobj.ufile;
+	struct ib_device *ib_dev = async_file->uobj.context->device;
 
-	return filp;
+	ib_uverbs_init_event_queue(&async_file->ev_queue);
 
-err_put_refs:
-	kref_put(&ev_file->uverbs_file->ref, ib_uverbs_release_file);
-	kref_put(&ev_file->ref, ib_uverbs_release_async_event_file);
-	return filp;
+	if (!WARN_ON(uverbs_file->async_file)) {
+		uverbs_file->async_file = async_file;
+		/* Pairs with the put in ib_uverbs_release_file */
+		uverbs_uobject_get(&async_file->uobj);
+	}
+
+	INIT_IB_EVENT_HANDLER(&async_file->event_handler, ib_dev,
+			      ib_uverbs_event_handler);
+	ib_register_event_handler(&async_file->event_handler);
 }
 
 static ssize_t verify_hdr(struct ib_uverbs_cmd_hdr *hdr,
@@ -1178,7 +1112,6 @@ static void ib_uverbs_add_one(struct ib_device *device)
 	mutex_init(&uverbs_dev->xrcd_tree_mutex);
 	mutex_init(&uverbs_dev->lists_mutex);
 	INIT_LIST_HEAD(&uverbs_dev->uverbs_file_list);
-	INIT_LIST_HEAD(&uverbs_dev->uverbs_events_file_list);
 	rcu_assign_pointer(uverbs_dev->ib_dev, device);
 	uverbs_dev->num_comp_vectors = device->num_comp_vectors;
 
@@ -1223,7 +1156,6 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 					struct ib_device *ib_dev)
 {
 	struct ib_uverbs_file *file;
-	struct ib_uverbs_async_event_file *event_file;
 	struct ib_event event;
 
 	/* Pending running commands to terminate */
@@ -1246,31 +1178,15 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 		 */
 		mutex_unlock(&uverbs_dev->lists_mutex);
 
-		ib_uverbs_event_handler(&file->event_handler, &event);
+		if (file->async_file)
+			ib_uverbs_event_handler(
+				&file->async_file->event_handler, &event);
+
 		uverbs_destroy_ufile_hw(file, RDMA_REMOVE_DRIVER_REMOVE);
 		kref_put(&file->ref, ib_uverbs_release_file);
 
 		mutex_lock(&uverbs_dev->lists_mutex);
 	}
-
-	while (!list_empty(&uverbs_dev->uverbs_events_file_list)) {
-		event_file = list_first_entry(&uverbs_dev->
-					      uverbs_events_file_list,
-					      struct ib_uverbs_async_event_file,
-					      list);
-		spin_lock_irq(&event_file->ev_queue.lock);
-		event_file->ev_queue.is_closed = 1;
-		spin_unlock_irq(&event_file->ev_queue.lock);
-
-		list_del(&event_file->list);
-		ib_unregister_event_handler(
-			&event_file->uverbs_file->event_handler);
-		event_file->uverbs_file->event_handler.device =
-			NULL;
-
-		wake_up_interruptible(&event_file->ev_queue.poll_wait);
-		kill_fasync(&event_file->ev_queue.async_queue, SIGIO, POLL_IN);
-	}
 	mutex_unlock(&uverbs_dev->lists_mutex);
 
 	uverbs_disassociate_api(uverbs_dev->uapi);
diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c
index def038a0fe77..efe70bcf79b1 100644
--- a/drivers/infiniband/core/uverbs_std_types.c
+++ b/drivers/infiniband/core/uverbs_std_types.c
@@ -202,17 +202,15 @@ static int uverbs_free_pd(struct ib_uobject *uobject,
 	return 0;
 }
 
-static int
-uverbs_completion_event_file_destroy_uobj(struct ib_uobject *uobj,
-					  enum rdma_remove_reason why)
+void ib_uverbs_free_event_queue(struct ib_uverbs_event_queue *event_queue)
 {
-	struct ib_uverbs_completion_event_file *file =
-		container_of(uobj, struct ib_uverbs_completion_event_file,
-			     uobj);
-	struct ib_uverbs_event_queue *event_queue = &file->ev_queue;
 	struct ib_uverbs_event *entry, *tmp;
 
 	spin_lock_irq(&event_queue->lock);
+	/*
+	 * The user must ensure that no new items are added to the event_list
+	 * once is_closed is set.
+	 */
 	event_queue->is_closed = 1;
 	spin_unlock_irq(&event_queue->lock);
 	wake_up_interruptible(&event_queue->poll_wait);
@@ -225,8 +223,19 @@ uverbs_completion_event_file_destroy_uobj(struct ib_uobject *uobj,
 		kfree(entry);
 	}
 	spin_unlock_irq(&event_queue->lock);
+}
+
+static int
+uverbs_completion_event_file_destroy_uobj(struct ib_uobject *uobj,
+					  enum rdma_remove_reason why)
+{
+	struct ib_uverbs_completion_event_file *file =
+		container_of(uobj, struct ib_uverbs_completion_event_file,
+			     uobj);
+
+	ib_uverbs_free_event_queue(&file->ev_queue);
 	return 0;
-};
+}
 
 int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs)
 {
diff --git a/drivers/infiniband/core/uverbs_std_types_async_fd.c b/drivers/infiniband/core/uverbs_std_types_async_fd.c
new file mode 100644
index 000000000000..31ff96898b06
--- /dev/null
+++ b/drivers/infiniband/core/uverbs_std_types_async_fd.c
@@ -0,0 +1,33 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/*
+ * Copyright (c) 2019, Mellanox Technologies inc.  All rights reserved.
+ */
+
+#include <rdma/uverbs_std_types.h>
+#include <rdma/uverbs_ioctl.h>
+#include "rdma_core.h"
+#include "uverbs.h"
+
+static int uverbs_async_event_destroy_uobj(struct ib_uobject *uobj,
+					   enum rdma_remove_reason why)
+{
+	struct ib_uverbs_async_event_file *event_file =
+		container_of(uobj, struct ib_uverbs_async_event_file, uobj);
+
+	ib_unregister_event_handler(&event_file->event_handler);
+	ib_uverbs_free_event_queue(&event_file->ev_queue);
+	return 0;
+}
+
+DECLARE_UVERBS_NAMED_OBJECT(
+	UVERBS_OBJECT_ASYNC_EVENT,
+	UVERBS_TYPE_ALLOC_FD(sizeof(struct ib_uverbs_async_event_file),
+			     uverbs_async_event_destroy_uobj,
+			     &uverbs_async_event_fops,
+			     "[infinibandevent]",
+			     O_RDONLY));
+
+const struct uapi_definition uverbs_def_obj_async_fd[] = {
+	UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_ASYNC_EVENT),
+	{}
+};
diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c
index 9b84a126187a..3f121ac31e0a 100644
--- a/drivers/infiniband/core/uverbs_uapi.c
+++ b/drivers/infiniband/core/uverbs_uapi.c
@@ -626,6 +626,7 @@ void uverbs_destroy_api(struct uverbs_api *uapi)
 }
 
 static const struct uapi_definition uverbs_core_api[] = {
+	UAPI_DEF_CHAIN(uverbs_def_obj_async_fd),
 	UAPI_DEF_CHAIN(uverbs_def_obj_counters),
 	UAPI_DEF_CHAIN(uverbs_def_obj_cq),
 	UAPI_DEF_CHAIN(uverbs_def_obj_device),
diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h
index 64f0e3aacd3f..9cfadb5120d9 100644
--- a/include/uapi/rdma/ib_user_ioctl_cmds.h
+++ b/include/uapi/rdma/ib_user_ioctl_cmds.h
@@ -56,6 +56,7 @@ enum uverbs_default_objects {
 	UVERBS_OBJECT_FLOW_ACTION,
 	UVERBS_OBJECT_DM,
 	UVERBS_OBJECT_COUNTERS,
+	UVERBS_OBJECT_ASYNC_EVENT,
 };
 
 enum {
-- 
cgit v1.2.3-59-g8ed1b


From 5c55cfd6a553d008fcd54e4a4e3fed1340ee5090 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 19:22:06 +0200
Subject: RDMA/core: Use READ_ONCE for ib_ufile.async_file

The writer for async_file holds the ucontext_lock, while the readers are
left unlocked. Most readers rely on an implicit locking, either by having
a uobject (which cannot be created before a context) or by holding the
ib_ufile kref.

However ib_uverbs_free_hw_resources() has no implicit lock and has a
possible race. Make this all clear and sane by using READ_ONCE
consistently.

Link: https://lore.kernel.org/r/1578504126-9400-15-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs.h              |  6 ++----
 drivers/infiniband/core/uverbs_cmd.c          |  2 +-
 drivers/infiniband/core/uverbs_main.c         | 29 +++++++++++----------------
 drivers/infiniband/core/uverbs_std_types.c    |  6 +++---
 drivers/infiniband/core/uverbs_std_types_cq.c |  1 -
 5 files changed, 18 insertions(+), 26 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index ccde5d20a6cf..aaa5c7550913 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -220,11 +220,9 @@ void ib_uverbs_init_async_event_file(struct ib_uverbs_async_event_file *ev_file)
 void ib_uverbs_free_event_queue(struct ib_uverbs_event_queue *event_queue);
 void ib_uverbs_flow_resources_free(struct ib_uflow_resources *uflow_res);
 
-void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
-			   struct ib_uverbs_completion_event_file *ev_file,
+void ib_uverbs_release_ucq(struct ib_uverbs_completion_event_file *ev_file,
 			   struct ib_ucq_object *uobj);
-void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
-			      struct ib_uevent_object *uobj);
+void ib_uverbs_release_uevent(struct ib_uevent_object *uobj);
 void ib_uverbs_release_file(struct kref *ref);
 
 void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context);
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index ced1384d316b..29b1b5ad8836 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1056,7 +1056,7 @@ err_free:
 	kfree(cq);
 err_file:
 	if (ev_file)
-		ib_uverbs_release_ucq(attrs->ufile, ev_file, obj);
+		ib_uverbs_release_ucq(ev_file, obj);
 
 err:
 	uobj_alloc_abort(&obj->uevent.uobject, attrs);
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 121e65f69c0b..1f279b0a8e49 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -125,9 +125,8 @@ static void ib_uverbs_release_dev(struct device *device)
 	kfree(dev);
 }
 
-void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
-			  struct ib_uverbs_completion_event_file *ev_file,
-			  struct ib_ucq_object *uobj)
+void ib_uverbs_release_ucq(struct ib_uverbs_completion_event_file *ev_file,
+			   struct ib_ucq_object *uobj)
 {
 	struct ib_uverbs_event *evt, *tmp;
 
@@ -142,20 +141,21 @@ void ib_uverbs_release_ucq(struct ib_uverbs_file *file,
 		uverbs_uobject_put(&ev_file->uobj);
 	}
 
-	ib_uverbs_release_uevent(file, &uobj->uevent);
+	ib_uverbs_release_uevent(&uobj->uevent);
 }
 
-void ib_uverbs_release_uevent(struct ib_uverbs_file *file,
-			      struct ib_uevent_object *uobj)
+void ib_uverbs_release_uevent(struct ib_uevent_object *uobj)
 {
+	struct ib_uverbs_async_event_file *async_file =
+		READ_ONCE(uobj->uobject.ufile->async_file);
 	struct ib_uverbs_event *evt, *tmp;
 
-	spin_lock_irq(&file->async_file->ev_queue.lock);
+	spin_lock_irq(&async_file->ev_queue.lock);
 	list_for_each_entry_safe(evt, tmp, &uobj->event_list, obj_list) {
 		list_del(&evt->list);
 		kfree(evt);
 	}
-	spin_unlock_irq(&file->async_file->ev_queue.lock);
+	spin_unlock_irq(&async_file->ev_queue.lock);
 }
 
 void ib_uverbs_detach_umcast(struct ib_qp *qp,
@@ -420,7 +420,7 @@ ib_uverbs_async_handler(struct ib_uverbs_async_event_file *async_file,
 static void uverbs_uobj_event(struct ib_uevent_object *eobj,
 			      struct ib_event *event)
 {
-	ib_uverbs_async_handler(eobj->uobject.ufile->async_file,
+	ib_uverbs_async_handler(READ_ONCE(eobj->uobject.ufile->async_file),
 				eobj->uobject.user_handle, event->event,
 				&eobj->event_list, &eobj->events_reported);
 }
@@ -476,9 +476,9 @@ void ib_uverbs_init_async_event_file(
 	ib_uverbs_init_event_queue(&async_file->ev_queue);
 
 	if (!WARN_ON(uverbs_file->async_file)) {
-		uverbs_file->async_file = async_file;
 		/* Pairs with the put in ib_uverbs_release_file */
 		uverbs_uobject_get(&async_file->uobj);
+		smp_store_release(&uverbs_file->async_file, async_file);
 	}
 
 	INIT_IB_EVENT_HANDLER(&async_file->event_handler, ib_dev,
@@ -1156,13 +1156,9 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 					struct ib_device *ib_dev)
 {
 	struct ib_uverbs_file *file;
-	struct ib_event event;
 
 	/* Pending running commands to terminate */
 	uverbs_disassociate_api_pre(uverbs_dev);
-	event.event = IB_EVENT_DEVICE_FATAL;
-	event.element.port_num = 0;
-	event.device = ib_dev;
 
 	mutex_lock(&uverbs_dev->lists_mutex);
 	while (!list_empty(&uverbs_dev->uverbs_file_list)) {
@@ -1178,9 +1174,8 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 		 */
 		mutex_unlock(&uverbs_dev->lists_mutex);
 
-		if (file->async_file)
-			ib_uverbs_event_handler(
-				&file->async_file->event_handler, &event);
+		ib_uverbs_async_handler(READ_ONCE(file->async_file), 0,
+					IB_EVENT_DEVICE_FATAL, NULL, NULL);
 
 		uverbs_destroy_ufile_hw(file, RDMA_REMOVE_DRIVER_REMOVE);
 		kref_put(&file->ref, ib_uverbs_release_file);
diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c
index efe70bcf79b1..994d8744b246 100644
--- a/drivers/infiniband/core/uverbs_std_types.c
+++ b/drivers/infiniband/core/uverbs_std_types.c
@@ -105,7 +105,7 @@ static int uverbs_free_qp(struct ib_uobject *uobject,
 	if (uqp->uxrcd)
 		atomic_dec(&uqp->uxrcd->refcnt);
 
-	ib_uverbs_release_uevent(attrs->ufile, &uqp->uevent);
+	ib_uverbs_release_uevent(&uqp->uevent);
 	return ret;
 }
 
@@ -138,7 +138,7 @@ static int uverbs_free_wq(struct ib_uobject *uobject,
 	if (ib_is_destroy_retryable(ret, why, uobject))
 		return ret;
 
-	ib_uverbs_release_uevent(attrs->ufile, &uwq->uevent);
+	ib_uverbs_release_uevent(&uwq->uevent);
 	return ret;
 }
 
@@ -163,7 +163,7 @@ static int uverbs_free_srq(struct ib_uobject *uobject,
 		atomic_dec(&us->uxrcd->refcnt);
 	}
 
-	ib_uverbs_release_uevent(attrs->ufile, uevent);
+	ib_uverbs_release_uevent(uevent);
 	return ret;
 }
 
diff --git a/drivers/infiniband/core/uverbs_std_types_cq.c b/drivers/infiniband/core/uverbs_std_types_cq.c
index a41c758042cc..da4110a0eea2 100644
--- a/drivers/infiniband/core/uverbs_std_types_cq.c
+++ b/drivers/infiniband/core/uverbs_std_types_cq.c
@@ -49,7 +49,6 @@ static int uverbs_free_cq(struct ib_uobject *uobject,
 		return ret;
 
 	ib_uverbs_release_ucq(
-		attrs->ufile,
 		ev_queue ? container_of(ev_queue,
 					struct ib_uverbs_completion_event_file,
 					ev_queue) :
-- 
cgit v1.2.3-59-g8ed1b


From a91e093cad37c60dc16ca382ed6f869c4c99fb77 Mon Sep 17 00:00:00 2001
From: Weihang Li <liweihang@huawei.com>
Date: Sat, 11 Jan 2020 18:32:39 +0800
Subject: RDMA/hns: Remove some redundant variables related to capabilities

In struct hns_roce_caps, max_srq_sg and max_srqwqes is unused, and
max_srqs has the same effect with num_srqs. So remove them from this
structrue.

Link: https://lore.kernel.org/r/1578738761-3176-2-git-send-email-liweihang@huawei.com
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_device.h | 3 ---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  | 3 ---
 drivers/infiniband/hw/hns/hns_roce_main.c   | 2 +-
 3 files changed, 1 insertion(+), 7 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 8d2bcd883d00..7459ebb0bc36 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -800,10 +800,8 @@ struct hns_roce_caps {
 	int             reserved_qps;
 	int		num_qpc_timer;
 	int		num_cqc_timer;
-	u32		max_srq_sg;
 	int		num_srqs;
 	u32		max_wqes;
-	u32		max_srqs;
 	u32		max_srq_wrs;
 	u32		max_srq_sges;
 	u32		max_sq_desc_sz;
@@ -817,7 +815,6 @@ struct hns_roce_caps {
 	u32		min_wqes;
 	int		reserved_cqs;
 	int		reserved_srqs;
-	u32		max_srqwqes;
 	int		num_aeq_vectors;
 	int		num_comp_vectors;
 	int		num_other_vectors;
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index f1e0ba6a11d5..3b13290d0946 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -1668,12 +1668,10 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
 	caps->num_srqs		= HNS_ROCE_V2_MAX_SRQ_NUM;
 	caps->min_cqes		= HNS_ROCE_MIN_CQE_NUM;
 	caps->max_cqes		= HNS_ROCE_V2_MAX_CQE_NUM;
-	caps->max_srqwqes	= HNS_ROCE_V2_MAX_SRQWQE_NUM;
 	caps->max_sq_sg		= HNS_ROCE_V2_MAX_SQ_SGE_NUM;
 	caps->max_extend_sg	= HNS_ROCE_V2_MAX_EXTEND_SGE_NUM;
 	caps->max_rq_sg		= HNS_ROCE_V2_MAX_RQ_SGE_NUM;
 	caps->max_sq_inline	= HNS_ROCE_V2_MAX_SQ_INLINE;
-	caps->max_srq_sg	= HNS_ROCE_V2_MAX_SRQ_SGE_NUM;
 	caps->num_uars		= HNS_ROCE_V2_UAR_NUM;
 	caps->phy_num_uars	= HNS_ROCE_V2_PHY_UAR_NUM;
 	caps->num_aeq_vectors	= HNS_ROCE_V2_AEQE_VEC_NUM;
@@ -1761,7 +1759,6 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
 	caps->local_ca_ack_delay = 0;
 	caps->max_mtu = IB_MTU_4096;
 
-	caps->max_srqs		= HNS_ROCE_V2_MAX_SRQ;
 	caps->max_srq_wrs	= HNS_ROCE_V2_MAX_SRQ_WR;
 	caps->max_srq_sges	= HNS_ROCE_V2_MAX_SRQ_SGE;
 
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 6e589f2ca35e..d0031d559213 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -210,7 +210,7 @@ static int hns_roce_query_device(struct ib_device *ib_dev,
 	props->max_pkeys = 1;
 	props->local_ca_ack_delay = hr_dev->caps.local_ca_ack_delay;
 	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) {
-		props->max_srq = hr_dev->caps.max_srqs;
+		props->max_srq = hr_dev->caps.num_srqs;
 		props->max_srq_wr = hr_dev->caps.max_srq_wrs;
 		props->max_srq_sge = hr_dev->caps.max_srq_sges;
 	}
-- 
cgit v1.2.3-59-g8ed1b


From ba6bb7e974212ef9c114ad94708fb88a82e240ea Mon Sep 17 00:00:00 2001
From: Lijun Ou <oulijun@huawei.com>
Date: Sat, 11 Jan 2020 18:32:40 +0800
Subject: RDMA/hns: Add interfaces to get pf capabilities from firmware

pf capabilities are set by default for hip08 previously which should
depends on different types of hardware. So add new interfaces to get them
from firmware.

Link: https://lore.kernel.org/r/1578738761-3176-3-git-send-email-liweihang@huawei.com
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_device.h |   6 +
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  | 367 ++++++++++++++++++++++++++++
 drivers/infiniband/hw/hns/hns_roce_hw_v2.h  | 154 ++++++++++++
 3 files changed, 527 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 7459ebb0bc36..a7c4ff975c28 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -898,6 +898,12 @@ struct hns_roce_caps {
 	u32		tpq_buf_pg_sz;
 	u32		chunk_sz;	/* chunk size in non multihop mode */
 	u64		flags;
+	u16		default_ceq_max_cnt;
+	u16		default_ceq_period;
+	u16		default_aeq_max_cnt;
+	u16		default_aeq_period;
+	u16		default_aeq_arm_st;
+	u16		default_ceq_arm_st;
 };
 
 struct hns_roce_work {
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 3b13290d0946..b0c799d84b36 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -1598,6 +1598,373 @@ static int hns_roce_v2_set_bt(struct hns_roce_dev *hr_dev)
 	return hns_roce_cmq_send(hr_dev, &desc, 1);
 }
 
+static void set_default_caps(struct hns_roce_dev *hr_dev)
+{
+	struct hns_roce_caps *caps = &hr_dev->caps;
+
+	caps->num_qps		= HNS_ROCE_V2_MAX_QP_NUM;
+	caps->max_wqes		= HNS_ROCE_V2_MAX_WQE_NUM;
+	caps->num_cqs		= HNS_ROCE_V2_MAX_CQ_NUM;
+	caps->num_srqs		= HNS_ROCE_V2_MAX_SRQ_NUM;
+	caps->min_cqes		= HNS_ROCE_MIN_CQE_NUM;
+	caps->max_cqes		= HNS_ROCE_V2_MAX_CQE_NUM;
+	caps->max_sq_sg		= HNS_ROCE_V2_MAX_SQ_SGE_NUM;
+	caps->max_extend_sg	= HNS_ROCE_V2_MAX_EXTEND_SGE_NUM;
+	caps->max_rq_sg		= HNS_ROCE_V2_MAX_RQ_SGE_NUM;
+	caps->max_sq_inline	= HNS_ROCE_V2_MAX_SQ_INLINE;
+	caps->num_uars		= HNS_ROCE_V2_UAR_NUM;
+	caps->phy_num_uars	= HNS_ROCE_V2_PHY_UAR_NUM;
+	caps->num_aeq_vectors	= HNS_ROCE_V2_AEQE_VEC_NUM;
+	caps->num_comp_vectors	= HNS_ROCE_V2_COMP_VEC_NUM;
+	caps->num_other_vectors = HNS_ROCE_V2_ABNORMAL_VEC_NUM;
+	caps->num_mtpts		= HNS_ROCE_V2_MAX_MTPT_NUM;
+	caps->num_mtt_segs	= HNS_ROCE_V2_MAX_MTT_SEGS;
+	caps->num_cqe_segs	= HNS_ROCE_V2_MAX_CQE_SEGS;
+	caps->num_srqwqe_segs	= HNS_ROCE_V2_MAX_SRQWQE_SEGS;
+	caps->num_idx_segs	= HNS_ROCE_V2_MAX_IDX_SEGS;
+	caps->num_pds		= HNS_ROCE_V2_MAX_PD_NUM;
+	caps->max_qp_init_rdma	= HNS_ROCE_V2_MAX_QP_INIT_RDMA;
+	caps->max_qp_dest_rdma	= HNS_ROCE_V2_MAX_QP_DEST_RDMA;
+	caps->max_sq_desc_sz	= HNS_ROCE_V2_MAX_SQ_DESC_SZ;
+	caps->max_rq_desc_sz	= HNS_ROCE_V2_MAX_RQ_DESC_SZ;
+	caps->max_srq_desc_sz	= HNS_ROCE_V2_MAX_SRQ_DESC_SZ;
+	caps->qpc_entry_sz	= HNS_ROCE_V2_QPC_ENTRY_SZ;
+	caps->irrl_entry_sz	= HNS_ROCE_V2_IRRL_ENTRY_SZ;
+	caps->trrl_entry_sz	= HNS_ROCE_V2_TRRL_ENTRY_SZ;
+	caps->cqc_entry_sz	= HNS_ROCE_V2_CQC_ENTRY_SZ;
+	caps->srqc_entry_sz	= HNS_ROCE_V2_SRQC_ENTRY_SZ;
+	caps->mtpt_entry_sz	= HNS_ROCE_V2_MTPT_ENTRY_SZ;
+	caps->mtt_entry_sz	= HNS_ROCE_V2_MTT_ENTRY_SZ;
+	caps->idx_entry_sz	= HNS_ROCE_V2_IDX_ENTRY_SZ;
+	caps->cq_entry_sz	= HNS_ROCE_V2_CQE_ENTRY_SIZE;
+	caps->page_size_cap	= HNS_ROCE_V2_PAGE_SIZE_SUPPORTED;
+	caps->reserved_lkey	= 0;
+	caps->reserved_pds	= 0;
+	caps->reserved_mrws	= 1;
+	caps->reserved_uars	= 0;
+	caps->reserved_cqs	= 0;
+	caps->reserved_srqs	= 0;
+	caps->reserved_qps	= HNS_ROCE_V2_RSV_QPS;
+
+	caps->qpc_ba_pg_sz	= 0;
+	caps->qpc_buf_pg_sz	= 0;
+	caps->qpc_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
+	caps->srqc_ba_pg_sz	= 0;
+	caps->srqc_buf_pg_sz	= 0;
+	caps->srqc_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
+	caps->cqc_ba_pg_sz	= 0;
+	caps->cqc_buf_pg_sz	= 0;
+	caps->cqc_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
+	caps->mpt_ba_pg_sz	= 0;
+	caps->mpt_buf_pg_sz	= 0;
+	caps->mpt_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
+	caps->mtt_ba_pg_sz	= 0;
+	caps->mtt_buf_pg_sz	= 0;
+	caps->mtt_hop_num	= HNS_ROCE_MTT_HOP_NUM;
+	caps->wqe_sq_hop_num	= HNS_ROCE_SQWQE_HOP_NUM;
+	caps->wqe_sge_hop_num	= HNS_ROCE_EXT_SGE_HOP_NUM;
+	caps->wqe_rq_hop_num	= HNS_ROCE_RQWQE_HOP_NUM;
+	caps->cqe_ba_pg_sz	= HNS_ROCE_BA_PG_SZ_SUPPORTED_256K;
+	caps->cqe_buf_pg_sz	= 0;
+	caps->cqe_hop_num	= HNS_ROCE_CQE_HOP_NUM;
+	caps->srqwqe_ba_pg_sz	= 0;
+	caps->srqwqe_buf_pg_sz	= 0;
+	caps->srqwqe_hop_num	= HNS_ROCE_SRQWQE_HOP_NUM;
+	caps->idx_ba_pg_sz	= 0;
+	caps->idx_buf_pg_sz	= 0;
+	caps->idx_hop_num	= HNS_ROCE_IDX_HOP_NUM;
+	caps->chunk_sz		= HNS_ROCE_V2_TABLE_CHUNK_SIZE;
+
+	caps->flags		= HNS_ROCE_CAP_FLAG_REREG_MR |
+				  HNS_ROCE_CAP_FLAG_ROCE_V1_V2 |
+				  HNS_ROCE_CAP_FLAG_RQ_INLINE |
+				  HNS_ROCE_CAP_FLAG_RECORD_DB |
+				  HNS_ROCE_CAP_FLAG_SQ_RECORD_DB;
+
+	caps->pkey_table_len[0] = 1;
+	caps->gid_table_len[0]	= HNS_ROCE_V2_GID_INDEX_NUM;
+	caps->ceqe_depth	= HNS_ROCE_V2_COMP_EQE_NUM;
+	caps->aeqe_depth	= HNS_ROCE_V2_ASYNC_EQE_NUM;
+	caps->local_ca_ack_delay = 0;
+	caps->max_mtu = IB_MTU_4096;
+
+	caps->max_srq_wrs	= HNS_ROCE_V2_MAX_SRQ_WR;
+	caps->max_srq_sges	= HNS_ROCE_V2_MAX_SRQ_SGE;
+
+	if (hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP08_B) {
+		caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC | HNS_ROCE_CAP_FLAG_MW |
+			       HNS_ROCE_CAP_FLAG_SRQ | HNS_ROCE_CAP_FLAG_FRMR |
+			       HNS_ROCE_CAP_FLAG_QP_FLOW_CTRL;
+
+		caps->num_qpc_timer	  = HNS_ROCE_V2_MAX_QPC_TIMER_NUM;
+		caps->qpc_timer_entry_sz  = HNS_ROCE_V2_QPC_TIMER_ENTRY_SZ;
+		caps->qpc_timer_ba_pg_sz  = 0;
+		caps->qpc_timer_buf_pg_sz = 0;
+		caps->qpc_timer_hop_num   = HNS_ROCE_HOP_NUM_0;
+		caps->num_cqc_timer	  = HNS_ROCE_V2_MAX_CQC_TIMER_NUM;
+		caps->cqc_timer_entry_sz  = HNS_ROCE_V2_CQC_TIMER_ENTRY_SZ;
+		caps->cqc_timer_ba_pg_sz  = 0;
+		caps->cqc_timer_buf_pg_sz = 0;
+		caps->cqc_timer_hop_num   = HNS_ROCE_HOP_NUM_0;
+
+		caps->sccc_entry_sz	  = HNS_ROCE_V2_SCCC_ENTRY_SZ;
+		caps->sccc_ba_pg_sz	  = 0;
+		caps->sccc_buf_pg_sz	  = 0;
+		caps->sccc_hop_num	  = HNS_ROCE_SCCC_HOP_NUM;
+	}
+}
+
+static void calc_pg_sz(int obj_num, int obj_size, int hop_num, int ctx_bt_num,
+		       int *buf_page_size, int *bt_page_size, u32 hem_type)
+{
+	u64 obj_per_chunk;
+	int bt_chunk_size = 1 << PAGE_SHIFT;
+	int buf_chunk_size = 1 << PAGE_SHIFT;
+	int obj_per_chunk_default = buf_chunk_size / obj_size;
+
+	*buf_page_size = 0;
+	*bt_page_size = 0;
+
+	switch (hop_num) {
+	case 3:
+		obj_per_chunk = ctx_bt_num * (bt_chunk_size / BA_BYTE_LEN) *
+				(bt_chunk_size / BA_BYTE_LEN) *
+				(bt_chunk_size / BA_BYTE_LEN) *
+				 obj_per_chunk_default;
+		break;
+	case 2:
+		obj_per_chunk = ctx_bt_num * (bt_chunk_size / BA_BYTE_LEN) *
+				(bt_chunk_size / BA_BYTE_LEN) *
+				 obj_per_chunk_default;
+		break;
+	case 1:
+		obj_per_chunk = ctx_bt_num * (bt_chunk_size / BA_BYTE_LEN) *
+				obj_per_chunk_default;
+		break;
+	case HNS_ROCE_HOP_NUM_0:
+		obj_per_chunk = ctx_bt_num * obj_per_chunk_default;
+		break;
+	default:
+		pr_err("Table %d not support hop_num = %d!\n", hem_type,
+			hop_num);
+		return;
+	}
+
+	if (hem_type >= HEM_TYPE_MTT)
+		*bt_page_size = ilog2(DIV_ROUND_UP(obj_num, obj_per_chunk));
+	else
+		*buf_page_size = ilog2(DIV_ROUND_UP(obj_num, obj_per_chunk));
+}
+
+static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev)
+{
+	struct hns_roce_cmq_desc desc[HNS_ROCE_QUERY_PF_CAPS_CMD_NUM];
+	struct hns_roce_caps *caps = &hr_dev->caps;
+	struct hns_roce_query_pf_caps_a *resp_a;
+	struct hns_roce_query_pf_caps_b *resp_b;
+	struct hns_roce_query_pf_caps_c *resp_c;
+	struct hns_roce_query_pf_caps_d *resp_d;
+	struct hns_roce_query_pf_caps_e *resp_e;
+	int ctx_hop_num;
+	int pbl_hop_num;
+	int ret;
+	int i;
+
+	for (i = 0; i < HNS_ROCE_QUERY_PF_CAPS_CMD_NUM; i++) {
+		hns_roce_cmq_setup_basic_desc(&desc[i],
+					      HNS_ROCE_OPC_QUERY_PF_CAPS_NUM,
+					      true);
+		if (i < (HNS_ROCE_QUERY_PF_CAPS_CMD_NUM - 1))
+			desc[i].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT);
+		else
+			desc[i].flag &= ~cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT);
+	}
+
+	ret = hns_roce_cmq_send(hr_dev, desc, HNS_ROCE_QUERY_PF_CAPS_CMD_NUM);
+	if (ret)
+		return ret;
+
+	resp_a = (struct hns_roce_query_pf_caps_a *)desc[0].data;
+	resp_b = (struct hns_roce_query_pf_caps_b *)desc[1].data;
+	resp_c = (struct hns_roce_query_pf_caps_c *)desc[2].data;
+	resp_d = (struct hns_roce_query_pf_caps_d *)desc[3].data;
+	resp_e = (struct hns_roce_query_pf_caps_e *)desc[4].data;
+
+	caps->local_ca_ack_delay     = resp_a->local_ca_ack_delay;
+	caps->max_sq_sg		     = le16_to_cpu(resp_a->max_sq_sg);
+	caps->max_sq_inline	     = le16_to_cpu(resp_a->max_sq_inline);
+	caps->max_rq_sg		     = le16_to_cpu(resp_a->max_rq_sg);
+	caps->max_extend_sg	     = le32_to_cpu(resp_a->max_extend_sg);
+	caps->num_qpc_timer	     = le16_to_cpu(resp_a->num_qpc_timer);
+	caps->num_cqc_timer	     = le16_to_cpu(resp_a->num_cqc_timer);
+	caps->max_srq_sges	     = le16_to_cpu(resp_a->max_srq_sges);
+	caps->num_aeq_vectors	     = resp_a->num_aeq_vectors;
+	caps->num_other_vectors	     = resp_a->num_other_vectors;
+	caps->max_sq_desc_sz	     = resp_a->max_sq_desc_sz;
+	caps->max_rq_desc_sz	     = resp_a->max_rq_desc_sz;
+	caps->max_srq_desc_sz	     = resp_a->max_srq_desc_sz;
+	caps->cq_entry_sz	     = resp_a->cq_entry_sz;
+
+	caps->mtpt_entry_sz	     = resp_b->mtpt_entry_sz;
+	caps->irrl_entry_sz	     = resp_b->irrl_entry_sz;
+	caps->trrl_entry_sz	     = resp_b->trrl_entry_sz;
+	caps->cqc_entry_sz	     = resp_b->cqc_entry_sz;
+	caps->srqc_entry_sz	     = resp_b->srqc_entry_sz;
+	caps->idx_entry_sz	     = resp_b->idx_entry_sz;
+	caps->sccc_entry_sz	     = resp_b->scc_ctx_entry_sz;
+	caps->max_mtu		     = resp_b->max_mtu;
+	caps->qpc_entry_sz	     = le16_to_cpu(resp_b->qpc_entry_sz);
+	caps->min_cqes		     = resp_b->min_cqes;
+	caps->min_wqes		     = resp_b->min_wqes;
+	caps->page_size_cap	     = le32_to_cpu(resp_b->page_size_cap);
+	caps->pkey_table_len[0]	     = resp_b->pkey_table_len;
+	caps->phy_num_uars	     = resp_b->phy_num_uars;
+	ctx_hop_num		     = resp_b->ctx_hop_num;
+	pbl_hop_num		     = resp_b->pbl_hop_num;
+
+	caps->num_pds = 1 << roce_get_field(resp_c->cap_flags_num_pds,
+					    V2_QUERY_PF_CAPS_C_NUM_PDS_M,
+					    V2_QUERY_PF_CAPS_C_NUM_PDS_S);
+	caps->flags = roce_get_field(resp_c->cap_flags_num_pds,
+				     V2_QUERY_PF_CAPS_C_CAP_FLAGS_M,
+				     V2_QUERY_PF_CAPS_C_CAP_FLAGS_S);
+	caps->num_cqs = 1 << roce_get_field(resp_c->max_gid_num_cqs,
+					    V2_QUERY_PF_CAPS_C_NUM_CQS_M,
+					    V2_QUERY_PF_CAPS_C_NUM_CQS_S);
+	caps->gid_table_len[0] = roce_get_field(resp_c->max_gid_num_cqs,
+						V2_QUERY_PF_CAPS_C_MAX_GID_M,
+						V2_QUERY_PF_CAPS_C_MAX_GID_S);
+	caps->max_cqes = 1 << roce_get_field(resp_c->cq_depth,
+					     V2_QUERY_PF_CAPS_C_CQ_DEPTH_M,
+					     V2_QUERY_PF_CAPS_C_CQ_DEPTH_S);
+	caps->num_mtpts = 1 << roce_get_field(resp_c->num_mrws,
+					      V2_QUERY_PF_CAPS_C_NUM_MRWS_M,
+					      V2_QUERY_PF_CAPS_C_NUM_MRWS_S);
+	caps->num_qps = 1 << roce_get_field(resp_c->ord_num_qps,
+					    V2_QUERY_PF_CAPS_C_NUM_QPS_M,
+					    V2_QUERY_PF_CAPS_C_NUM_QPS_S);
+	caps->max_qp_init_rdma = roce_get_field(resp_c->ord_num_qps,
+						V2_QUERY_PF_CAPS_C_MAX_ORD_M,
+						V2_QUERY_PF_CAPS_C_MAX_ORD_S);
+	caps->max_qp_dest_rdma = caps->max_qp_init_rdma;
+	caps->max_wqes = 1 << le16_to_cpu(resp_c->sq_depth);
+	caps->num_srqs = 1 << roce_get_field(resp_d->wq_hop_num_max_srqs,
+					     V2_QUERY_PF_CAPS_D_NUM_SRQS_M,
+					     V2_QUERY_PF_CAPS_D_NUM_SRQS_S);
+	caps->max_srq_wrs = 1 << le16_to_cpu(resp_d->srq_depth);
+	caps->ceqe_depth = 1 << roce_get_field(resp_d->num_ceqs_ceq_depth,
+					       V2_QUERY_PF_CAPS_D_CEQ_DEPTH_M,
+					       V2_QUERY_PF_CAPS_D_CEQ_DEPTH_S);
+	caps->num_comp_vectors = roce_get_field(resp_d->num_ceqs_ceq_depth,
+						V2_QUERY_PF_CAPS_D_NUM_CEQS_M,
+						V2_QUERY_PF_CAPS_D_NUM_CEQS_S);
+	caps->aeqe_depth = 1 << roce_get_field(resp_d->arm_st_aeq_depth,
+					       V2_QUERY_PF_CAPS_D_AEQ_DEPTH_M,
+					       V2_QUERY_PF_CAPS_D_AEQ_DEPTH_S);
+	caps->default_aeq_arm_st = roce_get_field(resp_d->arm_st_aeq_depth,
+					    V2_QUERY_PF_CAPS_D_AEQ_ARM_ST_M,
+					    V2_QUERY_PF_CAPS_D_AEQ_ARM_ST_S);
+	caps->default_ceq_arm_st = roce_get_field(resp_d->arm_st_aeq_depth,
+					    V2_QUERY_PF_CAPS_D_CEQ_ARM_ST_M,
+					    V2_QUERY_PF_CAPS_D_CEQ_ARM_ST_S);
+	caps->reserved_pds = roce_get_field(resp_d->num_uars_rsv_pds,
+					    V2_QUERY_PF_CAPS_D_RSV_PDS_M,
+					    V2_QUERY_PF_CAPS_D_RSV_PDS_S);
+	caps->num_uars = 1 << roce_get_field(resp_d->num_uars_rsv_pds,
+					     V2_QUERY_PF_CAPS_D_NUM_UARS_M,
+					     V2_QUERY_PF_CAPS_D_NUM_UARS_S);
+	caps->reserved_qps = roce_get_field(resp_d->rsv_uars_rsv_qps,
+					    V2_QUERY_PF_CAPS_D_RSV_QPS_M,
+					    V2_QUERY_PF_CAPS_D_RSV_QPS_S);
+	caps->reserved_uars = roce_get_field(resp_d->rsv_uars_rsv_qps,
+					     V2_QUERY_PF_CAPS_D_RSV_UARS_M,
+					     V2_QUERY_PF_CAPS_D_RSV_UARS_S);
+	caps->reserved_mrws = roce_get_field(resp_e->chunk_size_shift_rsv_mrws,
+					     V2_QUERY_PF_CAPS_E_RSV_MRWS_M,
+					     V2_QUERY_PF_CAPS_E_RSV_MRWS_S);
+	caps->chunk_sz = 1 << roce_get_field(resp_e->chunk_size_shift_rsv_mrws,
+					 V2_QUERY_PF_CAPS_E_CHUNK_SIZE_SHIFT_M,
+					 V2_QUERY_PF_CAPS_E_CHUNK_SIZE_SHIFT_S);
+	caps->reserved_cqs = roce_get_field(resp_e->rsv_cqs,
+					    V2_QUERY_PF_CAPS_E_RSV_CQS_M,
+					    V2_QUERY_PF_CAPS_E_RSV_CQS_S);
+	caps->reserved_srqs = roce_get_field(resp_e->rsv_srqs,
+					     V2_QUERY_PF_CAPS_E_RSV_SRQS_M,
+					     V2_QUERY_PF_CAPS_E_RSV_SRQS_S);
+	caps->reserved_lkey = roce_get_field(resp_e->rsv_lkey,
+					     V2_QUERY_PF_CAPS_E_RSV_LKEYS_M,
+					     V2_QUERY_PF_CAPS_E_RSV_LKEYS_S);
+	caps->default_ceq_max_cnt = le16_to_cpu(resp_e->ceq_max_cnt);
+	caps->default_ceq_period = le16_to_cpu(resp_e->ceq_period);
+	caps->default_aeq_max_cnt = le16_to_cpu(resp_e->aeq_max_cnt);
+	caps->default_aeq_period = le16_to_cpu(resp_e->aeq_period);
+
+	caps->qpc_timer_entry_sz = HNS_ROCE_V2_QPC_TIMER_ENTRY_SZ;
+	caps->cqc_timer_entry_sz = HNS_ROCE_V2_CQC_TIMER_ENTRY_SZ;
+	caps->mtt_entry_sz = HNS_ROCE_V2_MTT_ENTRY_SZ;
+	caps->num_mtt_segs = HNS_ROCE_V2_MAX_MTT_SEGS;
+	caps->mtt_ba_pg_sz = 0;
+	caps->num_cqe_segs = HNS_ROCE_V2_MAX_CQE_SEGS;
+	caps->num_srqwqe_segs = HNS_ROCE_V2_MAX_SRQWQE_SEGS;
+	caps->num_idx_segs = HNS_ROCE_V2_MAX_IDX_SEGS;
+
+	caps->qpc_hop_num = ctx_hop_num;
+	caps->srqc_hop_num = ctx_hop_num;
+	caps->cqc_hop_num = ctx_hop_num;
+	caps->mpt_hop_num = ctx_hop_num;
+	caps->mtt_hop_num = pbl_hop_num;
+	caps->cqe_hop_num = pbl_hop_num;
+	caps->srqwqe_hop_num = pbl_hop_num;
+	caps->idx_hop_num = pbl_hop_num;
+	caps->wqe_sq_hop_num = roce_get_field(resp_d->wq_hop_num_max_srqs,
+					  V2_QUERY_PF_CAPS_D_SQWQE_HOP_NUM_M,
+					  V2_QUERY_PF_CAPS_D_SQWQE_HOP_NUM_S);
+	caps->wqe_sge_hop_num = roce_get_field(resp_d->wq_hop_num_max_srqs,
+					  V2_QUERY_PF_CAPS_D_EX_SGE_HOP_NUM_M,
+					  V2_QUERY_PF_CAPS_D_EX_SGE_HOP_NUM_S);
+	caps->wqe_rq_hop_num = roce_get_field(resp_d->wq_hop_num_max_srqs,
+					  V2_QUERY_PF_CAPS_D_RQWQE_HOP_NUM_M,
+					  V2_QUERY_PF_CAPS_D_RQWQE_HOP_NUM_S);
+
+	calc_pg_sz(caps->num_qps, caps->qpc_entry_sz, caps->qpc_hop_num,
+		   caps->qpc_bt_num, &caps->qpc_buf_pg_sz, &caps->qpc_ba_pg_sz,
+		   HEM_TYPE_QPC);
+	calc_pg_sz(caps->num_mtpts, caps->mtpt_entry_sz, caps->mpt_hop_num,
+		   caps->mpt_bt_num, &caps->mpt_buf_pg_sz, &caps->mpt_ba_pg_sz,
+		   HEM_TYPE_MTPT);
+	calc_pg_sz(caps->num_cqs, caps->cqc_entry_sz, caps->cqc_hop_num,
+		   caps->cqc_bt_num, &caps->cqc_buf_pg_sz, &caps->cqc_ba_pg_sz,
+		   HEM_TYPE_CQC);
+	calc_pg_sz(caps->num_srqs, caps->srqc_entry_sz, caps->srqc_hop_num,
+		   caps->srqc_bt_num, &caps->srqc_buf_pg_sz,
+		   &caps->srqc_ba_pg_sz, HEM_TYPE_SRQC);
+
+	if (hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP08_B) {
+		caps->sccc_hop_num = ctx_hop_num;
+		caps->qpc_timer_hop_num = HNS_ROCE_HOP_NUM_0;
+		caps->cqc_timer_hop_num = HNS_ROCE_HOP_NUM_0;
+
+		calc_pg_sz(caps->num_qps, caps->sccc_entry_sz,
+			   caps->sccc_hop_num, caps->sccc_bt_num,
+			   &caps->sccc_buf_pg_sz, &caps->sccc_ba_pg_sz,
+			   HEM_TYPE_SCCC);
+		calc_pg_sz(caps->num_cqc_timer, caps->cqc_timer_entry_sz,
+			   caps->cqc_timer_hop_num, caps->cqc_timer_bt_num,
+			   &caps->cqc_timer_buf_pg_sz,
+			   &caps->cqc_timer_ba_pg_sz, HEM_TYPE_CQC_TIMER);
+	}
+
+	calc_pg_sz(caps->num_cqe_segs, caps->mtt_entry_sz, caps->cqe_hop_num,
+		   1, &caps->cqe_buf_pg_sz, &caps->cqe_ba_pg_sz, HEM_TYPE_CQE);
+	calc_pg_sz(caps->num_srqwqe_segs, caps->mtt_entry_sz,
+		   caps->srqwqe_hop_num, 1, &caps->srqwqe_buf_pg_sz,
+		   &caps->srqwqe_ba_pg_sz, HEM_TYPE_SRQWQE);
+	calc_pg_sz(caps->num_idx_segs, caps->idx_entry_sz, caps->idx_hop_num,
+		   1, &caps->idx_buf_pg_sz, &caps->idx_ba_pg_sz, HEM_TYPE_IDX);
+
+	return 0;
+}
+
 static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
 {
 	struct hns_roce_caps *caps = &hr_dev->caps;
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index 76a14db7028d..c56bd3182d86 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -85,6 +85,7 @@
 #define HNS_ROCE_V2_SRQC_ENTRY_SZ		64
 #define HNS_ROCE_V2_MTPT_ENTRY_SZ		64
 #define HNS_ROCE_V2_MTT_ENTRY_SZ		64
+#define HNS_ROCE_V2_IDX_ENTRY_SZ		4
 #define HNS_ROCE_V2_CQE_ENTRY_SIZE		32
 #define HNS_ROCE_V2_SCCC_ENTRY_SZ		32
 #define HNS_ROCE_V2_QPC_TIMER_ENTRY_SZ		PAGE_SIZE
@@ -109,6 +110,9 @@
 #define HNS_ROCE_PBL_HOP_NUM			2
 #define HNS_ROCE_EQE_HOP_NUM			2
 #define HNS_ROCE_IDX_HOP_NUM			1
+#define HNS_ROCE_SQWQE_HOP_NUM			2
+#define HNS_ROCE_EXT_SGE_HOP_NUM		1
+#define HNS_ROCE_RQWQE_HOP_NUM			2
 
 #define HNS_ROCE_V2_GID_INDEX_NUM		256
 
@@ -237,6 +241,7 @@ enum hns_roce_opcode_type {
 	HNS_ROCE_OPC_CFG_EXT_LLM			= 0x8403,
 	HNS_ROCE_OPC_CFG_TMOUT_LLM			= 0x8404,
 	HNS_ROCE_OPC_QUERY_PF_TIMER_RES			= 0x8406,
+	HNS_ROCE_OPC_QUERY_PF_CAPS_NUM                  = 0x8408,
 	HNS_ROCE_OPC_CFG_SGID_TB			= 0x8500,
 	HNS_ROCE_OPC_CFG_SMAC_TB			= 0x8501,
 	HNS_ROCE_OPC_POST_MB				= 0x8504,
@@ -1569,6 +1574,155 @@ struct hns_roce_cfg_smac_tb {
 #define CFG_SMAC_TB_VF_SMAC_H_S 0
 #define CFG_SMAC_TB_VF_SMAC_H_M GENMASK(15, 0)
 
+#define HNS_ROCE_QUERY_PF_CAPS_CMD_NUM 5
+struct hns_roce_query_pf_caps_a {
+	u8 number_ports;
+	u8 local_ca_ack_delay;
+	__le16 max_sq_sg;
+	__le16 max_sq_inline;
+	__le16 max_rq_sg;
+	__le32 max_extend_sg;
+	__le16 num_qpc_timer;
+	__le16 num_cqc_timer;
+	__le16 max_srq_sges;
+	u8 num_aeq_vectors;
+	u8 num_other_vectors;
+	u8 max_sq_desc_sz;
+	u8 max_rq_desc_sz;
+	u8 max_srq_desc_sz;
+	u8 cq_entry_sz;
+};
+
+struct hns_roce_query_pf_caps_b {
+	u8 mtpt_entry_sz;
+	u8 irrl_entry_sz;
+	u8 trrl_entry_sz;
+	u8 cqc_entry_sz;
+	u8 srqc_entry_sz;
+	u8 idx_entry_sz;
+	u8 scc_ctx_entry_sz;
+	u8 max_mtu;
+	__le16 qpc_entry_sz;
+	__le16 qpc_timer_entry_sz;
+	__le16 cqc_timer_entry_sz;
+	u8 min_cqes;
+	u8 min_wqes;
+	__le32 page_size_cap;
+	u8 pkey_table_len;
+	u8 phy_num_uars;
+	u8 ctx_hop_num;
+	u8 pbl_hop_num;
+};
+
+struct hns_roce_query_pf_caps_c {
+	__le32 cap_flags_num_pds;
+	__le32 max_gid_num_cqs;
+	__le32 cq_depth;
+	__le32 num_mrws;
+	__le32 ord_num_qps;
+	__le16 sq_depth;
+	__le16 rq_depth;
+};
+
+#define V2_QUERY_PF_CAPS_C_NUM_PDS_S 0
+#define V2_QUERY_PF_CAPS_C_NUM_PDS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_C_CAP_FLAGS_S 20
+#define V2_QUERY_PF_CAPS_C_CAP_FLAGS_M GENMASK(31, 20)
+
+#define V2_QUERY_PF_CAPS_C_NUM_CQS_S 0
+#define V2_QUERY_PF_CAPS_C_NUM_CQS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_C_MAX_GID_S 20
+#define V2_QUERY_PF_CAPS_C_MAX_GID_M GENMASK(28, 20)
+
+#define V2_QUERY_PF_CAPS_C_CQ_DEPTH_S 0
+#define V2_QUERY_PF_CAPS_C_CQ_DEPTH_M GENMASK(22, 0)
+
+#define V2_QUERY_PF_CAPS_C_NUM_MRWS_S 0
+#define V2_QUERY_PF_CAPS_C_NUM_MRWS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_C_NUM_QPS_S 0
+#define V2_QUERY_PF_CAPS_C_NUM_QPS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_C_MAX_ORD_S 20
+#define V2_QUERY_PF_CAPS_C_MAX_ORD_M GENMASK(27, 20)
+
+struct hns_roce_query_pf_caps_d {
+	__le32 wq_hop_num_max_srqs;
+	__le16 srq_depth;
+	__le16 rsv;
+	__le32 num_ceqs_ceq_depth;
+	__le32 arm_st_aeq_depth;
+	__le32 num_uars_rsv_pds;
+	__le32 rsv_uars_rsv_qps;
+};
+#define V2_QUERY_PF_CAPS_D_NUM_SRQS_S 0
+#define V2_QUERY_PF_CAPS_D_NUM_SRQS_M GENMASK(20, 0)
+
+#define V2_QUERY_PF_CAPS_D_RQWQE_HOP_NUM_S 20
+#define V2_QUERY_PF_CAPS_D_RQWQE_HOP_NUM_M GENMASK(21, 20)
+
+#define V2_QUERY_PF_CAPS_D_EX_SGE_HOP_NUM_S 22
+#define V2_QUERY_PF_CAPS_D_EX_SGE_HOP_NUM_M GENMASK(23, 22)
+
+#define V2_QUERY_PF_CAPS_D_SQWQE_HOP_NUM_S 24
+#define V2_QUERY_PF_CAPS_D_SQWQE_HOP_NUM_M GENMASK(25, 24)
+
+
+#define V2_QUERY_PF_CAPS_D_CEQ_DEPTH_S 0
+#define V2_QUERY_PF_CAPS_D_CEQ_DEPTH_M GENMASK(21, 0)
+
+#define V2_QUERY_PF_CAPS_D_NUM_CEQS_S 22
+#define V2_QUERY_PF_CAPS_D_NUM_CEQS_M GENMASK(31, 22)
+
+#define V2_QUERY_PF_CAPS_D_AEQ_DEPTH_S 0
+#define V2_QUERY_PF_CAPS_D_AEQ_DEPTH_M GENMASK(21, 0)
+
+#define V2_QUERY_PF_CAPS_D_AEQ_ARM_ST_S 22
+#define V2_QUERY_PF_CAPS_D_AEQ_ARM_ST_M GENMASK(23, 22)
+
+#define V2_QUERY_PF_CAPS_D_CEQ_ARM_ST_S 24
+#define V2_QUERY_PF_CAPS_D_CEQ_ARM_ST_M GENMASK(25, 24)
+
+#define V2_QUERY_PF_CAPS_D_RSV_PDS_S 0
+#define V2_QUERY_PF_CAPS_D_RSV_PDS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_D_NUM_UARS_S 20
+#define V2_QUERY_PF_CAPS_D_NUM_UARS_M GENMASK(27, 20)
+
+#define V2_QUERY_PF_CAPS_D_RSV_QPS_S 0
+#define V2_QUERY_PF_CAPS_D_RSV_QPS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_D_RSV_UARS_S 20
+#define V2_QUERY_PF_CAPS_D_RSV_UARS_M GENMASK(27, 20)
+
+struct hns_roce_query_pf_caps_e {
+	__le32 chunk_size_shift_rsv_mrws;
+	__le32 rsv_cqs;
+	__le32 rsv_srqs;
+	__le32 rsv_lkey;
+	__le16 ceq_max_cnt;
+	__le16 ceq_period;
+	__le16 aeq_max_cnt;
+	__le16 aeq_period;
+};
+
+#define V2_QUERY_PF_CAPS_E_RSV_MRWS_S 0
+#define V2_QUERY_PF_CAPS_E_RSV_MRWS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_E_CHUNK_SIZE_SHIFT_S 20
+#define V2_QUERY_PF_CAPS_E_CHUNK_SIZE_SHIFT_M GENMASK(31, 20)
+
+#define V2_QUERY_PF_CAPS_E_RSV_CQS_S 0
+#define V2_QUERY_PF_CAPS_E_RSV_CQS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_E_RSV_SRQS_S 0
+#define V2_QUERY_PF_CAPS_E_RSV_SRQS_M GENMASK(19, 0)
+
+#define V2_QUERY_PF_CAPS_E_RSV_LKEYS_S 0
+#define V2_QUERY_PF_CAPS_E_RSV_LKEYS_M GENMASK(19, 0)
+
 struct hns_roce_cmq_desc {
 	__le16 opcode;
 	__le16 flag;
-- 
cgit v1.2.3-59-g8ed1b


From 80a785701660ba3a36b0f72e1674b82ca7ca3ce0 Mon Sep 17 00:00:00 2001
From: Lijun Ou <oulijun@huawei.com>
Date: Sat, 11 Jan 2020 18:32:41 +0800
Subject: RDMA/hns: Get pf capabilities from firmware

Get pf capabilities from firmware according to different hardwares, if it
fails, all capabilities will be set with a default value.

Link: https://lore.kernel.org/r/1578738761-3176-4-git-send-email-liweihang@huawei.com
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 113 +----------------------------
 drivers/infiniband/hw/hns/hns_roce_hw_v2.h |   2 +
 2 files changed, 6 insertions(+), 109 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index b0c799d84b36..a523bfd89646 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -2029,127 +2029,22 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev)
 	hr_dev->vendor_part_id = hr_dev->pci_dev->device;
 	hr_dev->sys_image_guid = be64_to_cpu(hr_dev->ib_dev.node_guid);
 
-	caps->num_qps		= HNS_ROCE_V2_MAX_QP_NUM;
-	caps->max_wqes		= HNS_ROCE_V2_MAX_WQE_NUM;
-	caps->num_cqs		= HNS_ROCE_V2_MAX_CQ_NUM;
-	caps->num_srqs		= HNS_ROCE_V2_MAX_SRQ_NUM;
-	caps->min_cqes		= HNS_ROCE_MIN_CQE_NUM;
-	caps->max_cqes		= HNS_ROCE_V2_MAX_CQE_NUM;
-	caps->max_sq_sg		= HNS_ROCE_V2_MAX_SQ_SGE_NUM;
-	caps->max_extend_sg	= HNS_ROCE_V2_MAX_EXTEND_SGE_NUM;
-	caps->max_rq_sg		= HNS_ROCE_V2_MAX_RQ_SGE_NUM;
-	caps->max_sq_inline	= HNS_ROCE_V2_MAX_SQ_INLINE;
-	caps->num_uars		= HNS_ROCE_V2_UAR_NUM;
-	caps->phy_num_uars	= HNS_ROCE_V2_PHY_UAR_NUM;
-	caps->num_aeq_vectors	= HNS_ROCE_V2_AEQE_VEC_NUM;
-	caps->num_comp_vectors	= HNS_ROCE_V2_COMP_VEC_NUM;
-	caps->num_other_vectors	= HNS_ROCE_V2_ABNORMAL_VEC_NUM;
-	caps->num_mtpts		= HNS_ROCE_V2_MAX_MTPT_NUM;
 	caps->num_mtt_segs	= HNS_ROCE_V2_MAX_MTT_SEGS;
 	caps->num_cqe_segs	= HNS_ROCE_V2_MAX_CQE_SEGS;
 	caps->num_srqwqe_segs	= HNS_ROCE_V2_MAX_SRQWQE_SEGS;
 	caps->num_idx_segs	= HNS_ROCE_V2_MAX_IDX_SEGS;
-	caps->num_pds		= HNS_ROCE_V2_MAX_PD_NUM;
-	caps->max_qp_init_rdma	= HNS_ROCE_V2_MAX_QP_INIT_RDMA;
-	caps->max_qp_dest_rdma	= HNS_ROCE_V2_MAX_QP_DEST_RDMA;
-	caps->max_sq_desc_sz	= HNS_ROCE_V2_MAX_SQ_DESC_SZ;
-	caps->max_rq_desc_sz	= HNS_ROCE_V2_MAX_RQ_DESC_SZ;
-	caps->max_srq_desc_sz	= HNS_ROCE_V2_MAX_SRQ_DESC_SZ;
-	caps->qpc_entry_sz	= HNS_ROCE_V2_QPC_ENTRY_SZ;
-	caps->irrl_entry_sz	= HNS_ROCE_V2_IRRL_ENTRY_SZ;
-	caps->trrl_entry_sz	= HNS_ROCE_V2_TRRL_ENTRY_SZ;
-	caps->cqc_entry_sz	= HNS_ROCE_V2_CQC_ENTRY_SZ;
-	caps->srqc_entry_sz	= HNS_ROCE_V2_SRQC_ENTRY_SZ;
-	caps->mtpt_entry_sz	= HNS_ROCE_V2_MTPT_ENTRY_SZ;
-	caps->mtt_entry_sz	= HNS_ROCE_V2_MTT_ENTRY_SZ;
-	caps->idx_entry_sz	= 4;
-	caps->cq_entry_sz	= HNS_ROCE_V2_CQE_ENTRY_SIZE;
-	caps->page_size_cap	= HNS_ROCE_V2_PAGE_SIZE_SUPPORTED;
-	caps->reserved_lkey	= 0;
-	caps->reserved_pds	= 0;
-	caps->reserved_mrws	= 1;
-	caps->reserved_uars	= 0;
-	caps->reserved_cqs	= 0;
-	caps->reserved_srqs	= 0;
-	caps->reserved_qps	= HNS_ROCE_V2_RSV_QPS;
 
-	caps->qpc_ba_pg_sz	= 0;
-	caps->qpc_buf_pg_sz	= 0;
-	caps->qpc_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
-	caps->srqc_ba_pg_sz	= 0;
-	caps->srqc_buf_pg_sz	= 0;
-	caps->srqc_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
-	caps->cqc_ba_pg_sz	= 0;
-	caps->cqc_buf_pg_sz	= 0;
-	caps->cqc_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
-	caps->mpt_ba_pg_sz	= 0;
-	caps->mpt_buf_pg_sz	= 0;
-	caps->mpt_hop_num	= HNS_ROCE_CONTEXT_HOP_NUM;
-	caps->pbl_ba_pg_sz	= 2;
+	caps->pbl_ba_pg_sz	= HNS_ROCE_BA_PG_SZ_SUPPORTED_16K;
 	caps->pbl_buf_pg_sz	= 0;
 	caps->pbl_hop_num	= HNS_ROCE_PBL_HOP_NUM;
-	caps->mtt_ba_pg_sz	= 0;
-	caps->mtt_buf_pg_sz	= 0;
-	caps->mtt_hop_num	= HNS_ROCE_MTT_HOP_NUM;
-	caps->wqe_sq_hop_num	= 2;
-	caps->wqe_sge_hop_num	= 1;
-	caps->wqe_rq_hop_num	= 2;
-	caps->cqe_ba_pg_sz	= 6;
-	caps->cqe_buf_pg_sz	= 0;
-	caps->cqe_hop_num	= HNS_ROCE_CQE_HOP_NUM;
-	caps->srqwqe_ba_pg_sz	= 0;
-	caps->srqwqe_buf_pg_sz	= 0;
-	caps->srqwqe_hop_num	= HNS_ROCE_SRQWQE_HOP_NUM;
-	caps->idx_ba_pg_sz	= 0;
-	caps->idx_buf_pg_sz	= 0;
-	caps->idx_hop_num	= HNS_ROCE_IDX_HOP_NUM;
 	caps->eqe_ba_pg_sz	= 0;
 	caps->eqe_buf_pg_sz	= 0;
 	caps->eqe_hop_num	= HNS_ROCE_EQE_HOP_NUM;
 	caps->tsq_buf_pg_sz	= 0;
-	caps->chunk_sz		= HNS_ROCE_V2_TABLE_CHUNK_SIZE;
-
-	caps->flags		= HNS_ROCE_CAP_FLAG_REREG_MR |
-				  HNS_ROCE_CAP_FLAG_ROCE_V1_V2 |
-				  HNS_ROCE_CAP_FLAG_RQ_INLINE |
-				  HNS_ROCE_CAP_FLAG_RECORD_DB |
-				  HNS_ROCE_CAP_FLAG_SQ_RECORD_DB;
-
-	if (hr_dev->pci_dev->revision == 0x21)
-		caps->flags |= HNS_ROCE_CAP_FLAG_MW |
-			       HNS_ROCE_CAP_FLAG_FRMR;
 
-	caps->pkey_table_len[0] = 1;
-	caps->gid_table_len[0] = HNS_ROCE_V2_GID_INDEX_NUM;
-	caps->ceqe_depth	= HNS_ROCE_V2_COMP_EQE_NUM;
-	caps->aeqe_depth	= HNS_ROCE_V2_ASYNC_EQE_NUM;
-	caps->local_ca_ack_delay = 0;
-	caps->max_mtu = IB_MTU_4096;
-
-	caps->max_srq_wrs	= HNS_ROCE_V2_MAX_SRQ_WR;
-	caps->max_srq_sges	= HNS_ROCE_V2_MAX_SRQ_SGE;
-
-	if (hr_dev->pci_dev->revision == 0x21) {
-		caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC |
-			       HNS_ROCE_CAP_FLAG_SRQ |
-			       HNS_ROCE_CAP_FLAG_QP_FLOW_CTRL;
-
-		caps->num_qpc_timer	  = HNS_ROCE_V2_MAX_QPC_TIMER_NUM;
-		caps->qpc_timer_entry_sz  = HNS_ROCE_V2_QPC_TIMER_ENTRY_SZ;
-		caps->qpc_timer_ba_pg_sz  = 0;
-		caps->qpc_timer_buf_pg_sz = 0;
-		caps->qpc_timer_hop_num   = HNS_ROCE_HOP_NUM_0;
-		caps->num_cqc_timer	  = HNS_ROCE_V2_MAX_CQC_TIMER_NUM;
-		caps->cqc_timer_entry_sz  = HNS_ROCE_V2_CQC_TIMER_ENTRY_SZ;
-		caps->cqc_timer_ba_pg_sz  = 0;
-		caps->cqc_timer_buf_pg_sz = 0;
-		caps->cqc_timer_hop_num   = HNS_ROCE_HOP_NUM_0;
-
-		caps->sccc_entry_sz	= HNS_ROCE_V2_SCCC_ENTRY_SZ;
-		caps->sccc_ba_pg_sz	= 0;
-		caps->sccc_buf_pg_sz    = 0;
-		caps->sccc_hop_num	= HNS_ROCE_SCCC_HOP_NUM;
-	}
+	ret = hns_roce_query_pf_caps(hr_dev);
+	if (ret)
+		set_default_caps(hr_dev);
 
 	ret = hns_roce_v2_set_bt(hr_dev);
 	if (ret)
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index c56bd3182d86..539cea28bda9 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -114,6 +114,8 @@
 #define HNS_ROCE_EXT_SGE_HOP_NUM		1
 #define HNS_ROCE_RQWQE_HOP_NUM			2
 
+#define HNS_ROCE_BA_PG_SZ_SUPPORTED_256K	6
+#define HNS_ROCE_BA_PG_SZ_SUPPORTED_16K		2
 #define HNS_ROCE_V2_GID_INDEX_NUM		256
 
 #define HNS_ROCE_V2_TABLE_CHUNK_SIZE		(1 << 18)
-- 
cgit v1.2.3-59-g8ed1b


From 7db82697b8bf05ae56d02bf8da998bcd1122531d Mon Sep 17 00:00:00 2001
From: Jiaran Zhang <zhangjiaran@huawei.com>
Date: Wed, 15 Jan 2020 09:42:26 +0800
Subject: RDMA/hns: Add support for extended atomic in userspace

To support extended atomic operations including cmp & swap and fetch & add
of 8 bytes, 16 bytes, 32 bytes, 64 bytes in userspace, some field in qpc
should be configured.

Link: https://lore.kernel.org/r/1579052546-11746-1-git-send-email-liweihang@huawei.com
Signed-off-by: Jiaran Zhang <zhangjiaran@huawei.com>
Signed-off-by: Weihang Li <liweihang@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 16 +++++++++++++++-
 drivers/infiniband/hw/hns/hns_roce_hw_v2.h |  3 ++-
 2 files changed, 17 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index a523bfd89646..12c4cd8e9378 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -1630,7 +1630,7 @@ static void set_default_caps(struct hns_roce_dev *hr_dev)
 	caps->max_srq_desc_sz	= HNS_ROCE_V2_MAX_SRQ_DESC_SZ;
 	caps->qpc_entry_sz	= HNS_ROCE_V2_QPC_ENTRY_SZ;
 	caps->irrl_entry_sz	= HNS_ROCE_V2_IRRL_ENTRY_SZ;
-	caps->trrl_entry_sz	= HNS_ROCE_V2_TRRL_ENTRY_SZ;
+	caps->trrl_entry_sz	= HNS_ROCE_V2_EXT_ATOMIC_TRRL_ENTRY_SZ;
 	caps->cqc_entry_sz	= HNS_ROCE_V2_CQC_ENTRY_SZ;
 	caps->srqc_entry_sz	= HNS_ROCE_V2_SRQC_ENTRY_SZ;
 	caps->mtpt_entry_sz	= HNS_ROCE_V2_MTPT_ENTRY_SZ;
@@ -3545,6 +3545,9 @@ static void set_access_flags(struct hns_roce_qp *hr_qp,
 	roce_set_bit(context->byte_76_srqn_op_en, V2_QPC_BYTE_76_ATE_S,
 		     !!(access_flags & IB_ACCESS_REMOTE_ATOMIC));
 	roce_set_bit(qpc_mask->byte_76_srqn_op_en, V2_QPC_BYTE_76_ATE_S, 0);
+	roce_set_bit(context->byte_76_srqn_op_en, V2_QPC_BYTE_76_EXT_ATE_S,
+		     !!(access_flags & IB_ACCESS_REMOTE_ATOMIC));
+	roce_set_bit(qpc_mask->byte_76_srqn_op_en, V2_QPC_BYTE_76_EXT_ATE_S, 0);
 }
 
 static void set_qpc_wqe_cnt(struct hns_roce_qp *hr_qp,
@@ -3912,6 +3915,12 @@ static void modify_qp_init_to_init(struct ib_qp *ibqp,
 			     IB_ACCESS_REMOTE_ATOMIC));
 		roce_set_bit(qpc_mask->byte_76_srqn_op_en, V2_QPC_BYTE_76_ATE_S,
 			     0);
+		roce_set_bit(context->byte_76_srqn_op_en,
+			     V2_QPC_BYTE_76_EXT_ATE_S,
+			     !!(attr->qp_access_flags &
+				IB_ACCESS_REMOTE_ATOMIC));
+		roce_set_bit(qpc_mask->byte_76_srqn_op_en,
+			     V2_QPC_BYTE_76_EXT_ATE_S, 0);
 	} else {
 		roce_set_bit(context->byte_76_srqn_op_en, V2_QPC_BYTE_76_RRE_S,
 			     !!(hr_qp->access_flags & IB_ACCESS_REMOTE_READ));
@@ -3927,6 +3936,11 @@ static void modify_qp_init_to_init(struct ib_qp *ibqp,
 			     !!(hr_qp->access_flags & IB_ACCESS_REMOTE_ATOMIC));
 		roce_set_bit(qpc_mask->byte_76_srqn_op_en, V2_QPC_BYTE_76_ATE_S,
 			     0);
+		roce_set_bit(context->byte_76_srqn_op_en,
+			     V2_QPC_BYTE_76_EXT_ATE_S,
+			     !!(hr_qp->access_flags & IB_ACCESS_REMOTE_ATOMIC));
+		roce_set_bit(qpc_mask->byte_76_srqn_op_en,
+			     V2_QPC_BYTE_76_EXT_ATE_S, 0);
 	}
 
 	roce_set_field(context->byte_16_buf_ba_pg_sz, V2_QPC_BYTE_16_PD_M,
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index 539cea28bda9..2a117ff6a6be 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -81,6 +81,7 @@
 #define HNS_ROCE_V2_QPC_ENTRY_SZ		256
 #define HNS_ROCE_V2_IRRL_ENTRY_SZ		64
 #define HNS_ROCE_V2_TRRL_ENTRY_SZ		48
+#define HNS_ROCE_V2_EXT_ATOMIC_TRRL_ENTRY_SZ	100
 #define HNS_ROCE_V2_CQC_ENTRY_SZ		64
 #define HNS_ROCE_V2_SRQC_ENTRY_SZ		64
 #define HNS_ROCE_V2_MTPT_ENTRY_SZ		64
@@ -650,7 +651,7 @@ struct hns_roce_v2_qp_context {
 #define	V2_QPC_BYTE_76_ATE_S 27
 
 #define	V2_QPC_BYTE_76_RQIE_S 28
-
+#define	V2_QPC_BYTE_76_EXT_ATE_S 29
 #define	V2_QPC_BYTE_76_RQ_VLAN_EN_S 30
 #define	V2_QPC_BYTE_80_RX_CQN_S 0
 #define V2_QPC_BYTE_80_RX_CQN_M GENMASK(23, 0)
-- 
cgit v1.2.3-59-g8ed1b


From 57f63f371b5e9169c83964db8a9060776ec87a79 Mon Sep 17 00:00:00 2001
From: Gal Pressman <galpress@amazon.com>
Date: Tue, 14 Jan 2020 10:57:03 +0200
Subject: RDMA/efa: Device definitions documentation updates

Various clarifications and updates to the documentation of the device
definitions.
No functional changes in this patch.

Link: https://lore.kernel.org/r/20200114085706.82229-4-galpress@amazon.com
Reviewed-by: Firas JahJah <firasj@amazon.com>
Reviewed-by: Yossi Leybovich <sleybo@amazon.com>
Signed-off-by: Gal Pressman <galpress@amazon.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/efa/efa_admin_cmds_defs.h | 37 ++++++++++++++++---------
 1 file changed, 24 insertions(+), 13 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h b/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h
index e96bcb16bd2b..74b787a90660 100644
--- a/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h
+++ b/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h
@@ -160,10 +160,16 @@ struct efa_admin_create_qp_resp {
 	/* Common Admin Queue completion descriptor */
 	struct efa_admin_acq_common_desc acq_common_desc;
 
-	/* Opaque handle to be used for consequent operations on the QP */
+	/*
+	 * Opaque handle to be used for consequent admin operations on the
+	 * QP
+	 */
 	u32 qp_handle;
 
-	/* QP number in the given EFA virtual device */
+	/*
+	 * QP number in the given EFA virtual device. Least-significant bits
+	 *    (as needed according to max_qp) carry unique QP ID
+	 */
 	u16 qp_num;
 
 	/* MBZ */
@@ -286,6 +292,7 @@ struct efa_admin_create_ah_cmd {
 	/* PD number */
 	u16 pd;
 
+	/* MBZ */
 	u16 reserved;
 };
 
@@ -296,6 +303,7 @@ struct efa_admin_create_ah_resp {
 	/* Target interface address handle (opaque) */
 	u16 ah;
 
+	/* MBZ */
 	u16 reserved;
 };
 
@@ -372,6 +380,7 @@ struct efa_admin_reg_mr_cmd {
 	 */
 	u8 permissions;
 
+	/* MBZ */
 	u16 reserved16_w5;
 
 	/* number of pages in PBL (redundant, could be calculated) */
@@ -419,20 +428,20 @@ struct efa_admin_create_cq_cmd {
 	struct efa_admin_aq_common_desc aq_common_desc;
 
 	/*
-	 * 4:0 : reserved5
+	 * 4:0 : reserved5 - MBZ
 	 * 5 : interrupt_mode_enabled - if set, cq operates
 	 *    in interrupt mode (i.e. CQ events and MSI-X are
 	 *    generated), otherwise - polling
 	 * 6 : virt - If set, ring base address is virtual
 	 *    (IOVA returned by MR registration)
-	 * 7 : reserved6
+	 * 7 : reserved6 - MBZ
 	 */
 	u8 cq_caps_1;
 
 	/*
 	 * 4:0 : cq_entry_size_words - size of CQ entry in
 	 *    32-bit words, valid values: 4, 8.
-	 * 7:5 : reserved7
+	 * 7:5 : reserved7 - MBZ
 	 */
 	u8 cq_caps_2;
 
@@ -478,6 +487,7 @@ struct efa_admin_destroy_cq_cmd {
 
 	u16 cq_idx;
 
+	/* MBZ */
 	u16 reserved1;
 };
 
@@ -530,7 +540,7 @@ struct efa_admin_get_set_feature_common_desc {
 	/*
 	 * 1:0 : select - 0x1 - current value; 0x3 - default
 	 *    value
-	 * 7:3 : reserved3
+	 * 7:3 : reserved3 - MBZ
 	 */
 	u8 flags;
 
@@ -557,10 +567,10 @@ struct efa_admin_feature_device_attr_desc {
 	/* Bar used for SQ and RQ doorbells */
 	u16 db_bar;
 
-	/* Indicates how many bits are used physical address access */
+	/* Indicates how many bits are used on physical address access */
 	u8 phys_addr_width;
 
-	/* Indicates how many bits are used virtual address access */
+	/* Indicates how many bits are used on virtual address access */
 	u8 virt_addr_width;
 
 	/*
@@ -578,27 +588,28 @@ struct efa_admin_feature_queue_attr_desc {
 	/* The maximum number of queue pairs supported */
 	u32 max_qp;
 
+	/* Maximum number of WQEs per Send Queue */
 	u32 max_sq_depth;
 
-	/* max send wr used in inline-buf */
+	/* Maximum size of data that can be sent inline in a Send WQE */
 	u32 inline_buf_size;
 
+	/* Maximum number of buffer descriptors per Recv Queue */
 	u32 max_rq_depth;
 
 	/* The maximum number of completion queues supported per VF */
 	u32 max_cq;
 
+	/* Maximum number of CQEs per Completion Queue */
 	u32 max_cq_depth;
 
 	/* Number of sub-CQs to be created for each CQ */
 	u16 sub_cqs_per_cq;
 
+	/* MBZ */
 	u16 reserved;
 
-	/*
-	 * Maximum number of SGEs (buffs) allowed for a single send work
-	 *    queue element (WQE)
-	 */
+	/* Maximum number of SGEs (buffers) allowed for a single send WQE */
 	u16 max_wr_send_sges;
 
 	/* Maximum number of SGEs allowed for a single recv WQE */
-- 
cgit v1.2.3-59-g8ed1b


From f5f5ddbe73dc4d781d367bc5e2c8e79fc418d233 Mon Sep 17 00:00:00 2001
From: Gal Pressman <galpress@amazon.com>
Date: Tue, 14 Jan 2020 10:57:04 +0200
Subject: RDMA/efa: Remove {} brackets from single statement if

The {} brackets are not needed according to the Linux coding style.

Link: https://lore.kernel.org/r/20200114085706.82229-5-galpress@amazon.com
Reviewed-by: Daniel Kranzdorf <dkkranzd@amazon.com>
Reviewed-by: Firas JahJah <firasj@amazon.com>
Signed-off-by: Gal Pressman <galpress@amazon.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/efa/efa_verbs.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c
index 50c22575aed6..ce89c0b9c315 100644
--- a/drivers/infiniband/hw/efa/efa_verbs.c
+++ b/drivers/infiniband/hw/efa/efa_verbs.c
@@ -1608,13 +1608,12 @@ static int __efa_mmap(struct efa_dev *dev, struct efa_ucontext *ucontext,
 		err = -EINVAL;
 	}
 
-	if (err) {
+	if (err)
 		ibdev_dbg(
 			&dev->ibdev,
 			"Couldn't mmap address[%#llx] length[%#zx] mmap_flag[%d] err[%d]\n",
 			entry->address, rdma_entry->npages * PAGE_SIZE,
 			entry->mmap_flag, err);
-	}
 
 	rdma_user_mmap_entry_put(rdma_entry);
 	return err;
-- 
cgit v1.2.3-59-g8ed1b


From 0428c6ef8a3c167b852804756357425902766120 Mon Sep 17 00:00:00 2001
From: Gal Pressman <galpress@amazon.com>
Date: Tue, 14 Jan 2020 10:57:05 +0200
Subject: RDMA/efa: Remove unused ucontext parameter from
 efa_qp_user_mmap_entries_remove

The ucontext parameter is unused, remove it.

Link: https://lore.kernel.org/r/20200114085706.82229-6-galpress@amazon.com
Reviewed-by: Firas JahJah <firasj@amazon.com>
Reviewed-by: Yossi Leybovich <sleybo@amazon.com>
Signed-off-by: Gal Pressman <galpress@amazon.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/efa/efa_verbs.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c
index ce89c0b9c315..9a6cff718c49 100644
--- a/drivers/infiniband/hw/efa/efa_verbs.c
+++ b/drivers/infiniband/hw/efa/efa_verbs.c
@@ -387,8 +387,7 @@ static int efa_destroy_qp_handle(struct efa_dev *dev, u32 qp_handle)
 	return efa_com_destroy_qp(&dev->edev, &params);
 }
 
-static void efa_qp_user_mmap_entries_remove(struct efa_ucontext *uctx,
-					    struct efa_qp *qp)
+static void efa_qp_user_mmap_entries_remove(struct efa_qp *qp)
 {
 	rdma_user_mmap_entry_remove(qp->rq_mmap_entry);
 	rdma_user_mmap_entry_remove(qp->rq_db_mmap_entry);
@@ -398,8 +397,6 @@ static void efa_qp_user_mmap_entries_remove(struct efa_ucontext *uctx,
 
 int efa_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
 {
-	struct efa_ucontext *ucontext = rdma_udata_to_drv_context(udata,
-		struct efa_ucontext, ibucontext);
 	struct efa_dev *dev = to_edev(ibqp->pd->device);
 	struct efa_qp *qp = to_eqp(ibqp);
 	int err;
@@ -418,7 +415,7 @@ int efa_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
 				 DMA_TO_DEVICE);
 	}
 
-	efa_qp_user_mmap_entries_remove(ucontext, qp);
+	efa_qp_user_mmap_entries_remove(qp);
 	kfree(qp);
 	return 0;
 }
@@ -510,7 +507,7 @@ static int qp_mmap_entries_setup(struct efa_qp *qp,
 	return 0;
 
 err_remove_mmap:
-	efa_qp_user_mmap_entries_remove(ucontext, qp);
+	efa_qp_user_mmap_entries_remove(qp);
 
 	return -ENOMEM;
 }
@@ -719,7 +716,7 @@ struct ib_qp *efa_create_qp(struct ib_pd *ibpd,
 	return &qp->ibqp;
 
 err_remove_mmap_entries:
-	efa_qp_user_mmap_entries_remove(ucontext, qp);
+	efa_qp_user_mmap_entries_remove(qp);
 err_destroy_qp:
 	efa_destroy_qp_handle(dev, create_qp_resp.qp_handle);
 err_free_mapped:
-- 
cgit v1.2.3-59-g8ed1b


From 4e8d683f11767dc6cd8eaac6b599928f343b4cdb Mon Sep 17 00:00:00 2001
From: Rao Shoaib <rao.shoaib@oracle.com>
Date: Mon, 13 Jan 2020 16:41:19 -0800
Subject: Introduce maximum WQE size to check limits

Introduce maximum WQE size to impose limits on max SGE's and inline data

Link: https://lore.kernel.org/r/1578962480-17814-2-git-send-email-rao.shoaib@oracle.com
Signed-off-by: Rao Shoaib <rao.shoaib@oracle.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/sw/rxe/rxe_param.h | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/rxe/rxe_param.h b/drivers/infiniband/sw/rxe/rxe_param.h
index 353c6668249e..f59616b02477 100644
--- a/drivers/infiniband/sw/rxe/rxe_param.h
+++ b/drivers/infiniband/sw/rxe/rxe_param.h
@@ -34,6 +34,8 @@
 #ifndef RXE_PARAM_H
 #define RXE_PARAM_H
 
+#include <uapi/rdma/rdma_user_rxe.h>
+
 static inline enum ib_mtu rxe_mtu_int_to_enum(int mtu)
 {
 	if (mtu < 256)
@@ -64,7 +66,6 @@ enum rxe_device_param {
 	RXE_PAGE_SIZE_CAP		= 0xfffff000,
 	RXE_MAX_QP			= 0x10000,
 	RXE_MAX_QP_WR			= 0x4000,
-	RXE_MAX_INLINE_DATA		= 400,
 	RXE_DEVICE_CAP_FLAGS		= IB_DEVICE_BAD_PKEY_CNTR
 					| IB_DEVICE_BAD_QKEY_CNTR
 					| IB_DEVICE_AUTO_PATH_MIG
@@ -77,6 +78,10 @@ enum rxe_device_param {
 					| IB_DEVICE_MEM_MGT_EXTENSIONS
 					| IB_DEVICE_ALLOW_USER_UNREG,
 	RXE_MAX_SGE			= 32,
+	RXE_MAX_WQE_SIZE		= sizeof(struct rxe_send_wqe) +
+					  sizeof(struct ib_sge) * RXE_MAX_SGE,
+	RXE_MAX_INLINE_DATA		= RXE_MAX_WQE_SIZE -
+					  sizeof(struct rxe_send_wqe),
 	RXE_MAX_SGE_RD			= 32,
 	RXE_MAX_CQ			= 16384,
 	RXE_MAX_LOG_CQE			= 15,
-- 
cgit v1.2.3-59-g8ed1b


From 363824f92a9259b025f05ab038eb47224377f9e9 Mon Sep 17 00:00:00 2001
From: Rao Shoaib <rao.shoaib@oracle.com>
Date: Mon, 13 Jan 2020 16:41:20 -0800
Subject: RDMA/rxe: Compute the maximum sges and inline size based on the WQE
 size

The SGE buffer size and max_inline data should be derived from the size of
the WQE. Each value individually sets the WQE size, so compute the actual
sizes based on the actual WQE size and configure the QP with the maximums.

Also fix the missing return of the actual maximum capability to the caller.

Link: https://lore.kernel.org/r/1578962480-17814-3-git-send-email-rao.shoaib@oracle.com
Signed-off-by: Rao Shoaib <rao.shoaib@oracle.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/sw/rxe/rxe_qp.c | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/sw/rxe/rxe_qp.c b/drivers/infiniband/sw/rxe/rxe_qp.c
index e2c6d1cedf41..ec21f616ac98 100644
--- a/drivers/infiniband/sw/rxe/rxe_qp.c
+++ b/drivers/infiniband/sw/rxe/rxe_qp.c
@@ -237,19 +237,17 @@ static int rxe_qp_init_req(struct rxe_dev *rxe, struct rxe_qp *qp,
 	 */
 	qp->src_port = RXE_ROCE_V2_SPORT +
 		(hash_32_generic(qp_num(qp), 14) & 0x3fff);
-
 	qp->sq.max_wr		= init->cap.max_send_wr;
-	qp->sq.max_sge		= init->cap.max_send_sge;
-	qp->sq.max_inline	= init->cap.max_inline_data;
 
-	wqe_size = max_t(int, sizeof(struct rxe_send_wqe) +
-			 qp->sq.max_sge * sizeof(struct ib_sge),
-			 sizeof(struct rxe_send_wqe) +
-			 qp->sq.max_inline);
+	/* These caps are limited by rxe_qp_chk_cap() done by the caller */
+	wqe_size = max_t(int, init->cap.max_send_sge * sizeof(struct ib_sge),
+			 init->cap.max_inline_data);
+	qp->sq.max_sge = init->cap.max_send_sge =
+		wqe_size / sizeof(struct ib_sge);
+	qp->sq.max_inline = init->cap.max_inline_data = wqe_size;
+	wqe_size += sizeof(struct rxe_send_wqe);
 
-	qp->sq.queue = rxe_queue_init(rxe,
-				      &qp->sq.max_wr,
-				      wqe_size);
+	qp->sq.queue = rxe_queue_init(rxe, &qp->sq.max_wr, wqe_size);
 	if (!qp->sq.queue)
 		return -ENOMEM;
 
-- 
cgit v1.2.3-59-g8ed1b


From 0fbb37dd82998b5c83355997b3bdba2806968ac7 Mon Sep 17 00:00:00 2001
From: Sergey Gorenko <sergeygo@mellanox.com>
Date: Wed, 15 Jan 2020 13:30:55 +0000
Subject: IB/srp: Never use immediate data if it is disabled by a user

Some SRP targets that do not support specification SRP-2, put the garbage
to the reserved bits of the SRP login response.  The problem was not
detected for a long time because the SRP initiator ignored those bits. But
now one of them is used as SRP_LOGIN_RSP_IMMED_SUPP. And it causes a
critical error on the target when the initiator sends immediate data.

The ib_srp module has a use_imm_date parameter to enable or disable
immediate data manually. But it does not help in the above case, because
use_imm_date is ignored at handling the SRP login response. The problem is
definitely caused by a bug on the target side, but the initiator's
behavior also does not look correct.  The initiator should not use
immediate data if use_imm_date is disabled by a user.

This commit adds an additional checking of use_imm_date at the handling of
SRP login response to avoid unexpected use of immediate data.

Fixes: 882981f4a411 ("RDMA/srp: Add support for immediate data")
Link: https://lore.kernel.org/r/20200115133055.30232-1-sergeygo@mellanox.com
Signed-off-by: Sergey Gorenko <sergeygo@mellanox.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/ulp/srp/ib_srp.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c
index b7f7a5f7bd98..cd1181c39ed2 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++ b/drivers/infiniband/ulp/srp/ib_srp.c
@@ -2546,7 +2546,8 @@ static void srp_cm_rep_handler(struct ib_cm_id *cm_id,
 	if (lrsp->opcode == SRP_LOGIN_RSP) {
 		ch->max_ti_iu_len = be32_to_cpu(lrsp->max_ti_iu_len);
 		ch->req_lim       = be32_to_cpu(lrsp->req_lim_delta);
-		ch->use_imm_data  = lrsp->rsp_flags & SRP_LOGIN_RSP_IMMED_SUPP;
+		ch->use_imm_data  = srp_use_imm_data &&
+			(lrsp->rsp_flags & SRP_LOGIN_RSP_IMMED_SUPP);
 		ch->max_it_iu_len = srp_max_it_iu_len(target->cmd_sg_cnt,
 						      ch->use_imm_data,
 						      target->max_it_iu_size);
-- 
cgit v1.2.3-59-g8ed1b


From d680e88e2013186e696665cbf2056fb32b781e41 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 20:05:32 +0200
Subject: RDMA/core: Add UVERBS_METHOD_ASYNC_EVENT_ALLOC

Allow the async FD to be allocated separately from the context.

This is necessary to introduce the ioctl to create a context, as an ioctl
should only ever create a single uobject at a time.

If multiple async FDs are created then the first one is used to deliver
affiliated events from any ib_uevent_object, with all subsequent ones will
receive only unaffiliated events.

Link: https://lore.kernel.org/r/1578506740-22188-3-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs_main.c              |  4 +++-
 .../infiniband/core/uverbs_std_types_async_fd.c    | 23 +++++++++++++++++++++-
 include/uapi/rdma/ib_user_ioctl_cmds.h             |  8 ++++++++
 3 files changed, 33 insertions(+), 2 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index 1f279b0a8e49..fb9e75257607 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -475,7 +475,9 @@ void ib_uverbs_init_async_event_file(
 
 	ib_uverbs_init_event_queue(&async_file->ev_queue);
 
-	if (!WARN_ON(uverbs_file->async_file)) {
+	/* The first async_event_file becomes the default one for the file. */
+	lockdep_assert_held(&uverbs_file->ucontext_lock);
+	if (!uverbs_file->async_file) {
 		/* Pairs with the put in ib_uverbs_release_file */
 		uverbs_uobject_get(&async_file->uobj);
 		smp_store_release(&uverbs_file->async_file, async_file);
diff --git a/drivers/infiniband/core/uverbs_std_types_async_fd.c b/drivers/infiniband/core/uverbs_std_types_async_fd.c
index 31ff96898b06..484dba136950 100644
--- a/drivers/infiniband/core/uverbs_std_types_async_fd.c
+++ b/drivers/infiniband/core/uverbs_std_types_async_fd.c
@@ -8,6 +8,19 @@
 #include "rdma_core.h"
 #include "uverbs.h"
 
+static int UVERBS_HANDLER(UVERBS_METHOD_ASYNC_EVENT_ALLOC)(
+	struct uverbs_attr_bundle *attrs)
+{
+	struct ib_uobject *uobj =
+		uverbs_attr_get_uobject(attrs, UVERBS_METHOD_ASYNC_EVENT_ALLOC);
+
+	mutex_lock(&attrs->ufile->ucontext_lock);
+	ib_uverbs_init_async_event_file(
+		container_of(uobj, struct ib_uverbs_async_event_file, uobj));
+	mutex_unlock(&attrs->ufile->ucontext_lock);
+	return 0;
+}
+
 static int uverbs_async_event_destroy_uobj(struct ib_uobject *uobj,
 					   enum rdma_remove_reason why)
 {
@@ -19,13 +32,21 @@ static int uverbs_async_event_destroy_uobj(struct ib_uobject *uobj,
 	return 0;
 }
 
+DECLARE_UVERBS_NAMED_METHOD(
+	UVERBS_METHOD_ASYNC_EVENT_ALLOC,
+	UVERBS_ATTR_FD(UVERBS_ATTR_ASYNC_EVENT_ALLOC_FD_HANDLE,
+		       UVERBS_OBJECT_ASYNC_EVENT,
+		       UVERBS_ACCESS_NEW,
+		       UA_MANDATORY));
+
 DECLARE_UVERBS_NAMED_OBJECT(
 	UVERBS_OBJECT_ASYNC_EVENT,
 	UVERBS_TYPE_ALLOC_FD(sizeof(struct ib_uverbs_async_event_file),
 			     uverbs_async_event_destroy_uobj,
 			     &uverbs_async_event_fops,
 			     "[infinibandevent]",
-			     O_RDONLY));
+			     O_RDONLY),
+	&UVERBS_METHOD(UVERBS_METHOD_ASYNC_EVENT_ALLOC));
 
 const struct uapi_definition uverbs_def_obj_async_fd[] = {
 	UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_ASYNC_EVENT),
diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h
index 9cfadb5120d9..498955c576f3 100644
--- a/include/uapi/rdma/ib_user_ioctl_cmds.h
+++ b/include/uapi/rdma/ib_user_ioctl_cmds.h
@@ -242,4 +242,12 @@ enum uverbs_attrs_flow_destroy_ids {
 	UVERBS_ATTR_DESTROY_FLOW_HANDLE,
 };
 
+enum uverbs_method_async_event {
+	UVERBS_METHOD_ASYNC_EVENT_ALLOC,
+};
+
+enum uverbs_attrs_async_event_create {
+	UVERBS_ATTR_ASYNC_EVENT_ALLOC_FD_HANDLE,
+};
+
 #endif
-- 
cgit v1.2.3-59-g8ed1b


From da57db25677f8566a354b0d0e4973a0fdcabcf84 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 20:05:33 +0200
Subject: RDMA/core: Remove ucontext_lock from the uverbs_destry_ufile_hw()
 path

This lock only serializes ucontext creation. Instead of checking the
ucontext_lock during destruction hold the existing hw_destroy_rwsem during
creation, which is the standard pattern for object creation.

The simplification of locking is needed for the next patch.

Link: https://lore.kernel.org/r/1578506740-22188-4-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/rdma_core.c  | 21 +--------------------
 drivers/infiniband/core/uverbs_cmd.c |  5 ++++-
 2 files changed, 5 insertions(+), 21 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index 50251da123f7..5128cb16bb48 100644
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -845,9 +845,7 @@ static int __uverbs_cleanup_ufile(struct ib_uverbs_file *ufile,
 }
 
 /*
- * Destroy the uncontext and every uobject associated with it. If called with
- * reason != RDMA_REMOVE_CLOSE this will not return until the destruction has
- * been completed and ufile->ucontext is NULL.
+ * Destroy the uncontext and every uobject associated with it.
  *
  * This is internally locked and can be called in parallel from multiple
  * contexts.
@@ -855,22 +853,6 @@ static int __uverbs_cleanup_ufile(struct ib_uverbs_file *ufile,
 void uverbs_destroy_ufile_hw(struct ib_uverbs_file *ufile,
 			     enum rdma_remove_reason reason)
 {
-	if (reason == RDMA_REMOVE_CLOSE) {
-		/*
-		 * During destruction we might trigger something that
-		 * synchronously calls release on any file descriptor. For
-		 * this reason all paths that come from file_operations
-		 * release must use try_lock. They can progress knowing that
-		 * there is an ongoing uverbs_destroy_ufile_hw that will clean
-		 * up the driver resources.
-		 */
-		if (!mutex_trylock(&ufile->ucontext_lock))
-			return;
-
-	} else {
-		mutex_lock(&ufile->ucontext_lock);
-	}
-
 	down_write(&ufile->hw_destroy_rwsem);
 
 	/*
@@ -899,7 +881,6 @@ void uverbs_destroy_ufile_hw(struct ib_uverbs_file *ufile,
 
 done:
 	up_write(&ufile->hw_destroy_rwsem);
-	mutex_unlock(&ufile->ucontext_lock);
 }
 
 const struct uverbs_obj_type_class uverbs_fd_class = {
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 29b1b5ad8836..d71ffe44b8ae 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -218,6 +218,8 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 	if (ret)
 		return ret;
 
+	if (!down_read_trylock(&file->hw_destroy_rwsem))
+		return -EIO;
 	mutex_lock(&file->ucontext_lock);
 	ib_dev = srcu_dereference(file->device->ib_dev,
 				  &file->device->disassociate_srcu);
@@ -284,7 +286,7 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 	smp_store_release(&file->ucontext, ucontext);
 
 	mutex_unlock(&file->ucontext_lock);
-
+	up_read(&file->hw_destroy_rwsem);
 	return 0;
 
 err_uobj:
@@ -298,6 +300,7 @@ err_alloc:
 
 err:
 	mutex_unlock(&file->ucontext_lock);
+	up_read(&file->hw_destroy_rwsem);
 	return ret;
 }
 
-- 
cgit v1.2.3-59-g8ed1b


From a1123418ba1078037d9fecb72573ff7222dfe201 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 8 Jan 2020 20:05:34 +0200
Subject: RDMA/uverbs: Add ioctl command to get a device context

Allow future extensions of the get context command through the uverbs
ioctl kabi.

Unlike the uverbs version this does not return an async_fd as well, that
has to be done with another command.

Link: https://lore.kernel.org/r/1578506740-22188-5-git-send-email-yishaih@mellanox.com
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs.h                   |   3 +
 drivers/infiniband/core/uverbs_cmd.c               | 132 ++++++++++++---------
 drivers/infiniband/core/uverbs_main.c              |   9 +-
 .../infiniband/core/uverbs_std_types_async_fd.c    |   2 -
 drivers/infiniband/core/uverbs_std_types_device.c  |  30 +++++
 include/uapi/rdma/ib_user_ioctl_cmds.h             |   5 +
 6 files changed, 119 insertions(+), 62 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index aaa5c7550913..4d4cec46d251 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -220,6 +220,9 @@ void ib_uverbs_init_async_event_file(struct ib_uverbs_async_event_file *ev_file)
 void ib_uverbs_free_event_queue(struct ib_uverbs_event_queue *event_queue);
 void ib_uverbs_flow_resources_free(struct ib_uflow_resources *uflow_res);
 
+int ib_alloc_ucontext(struct uverbs_attr_bundle *attrs);
+int ib_init_ucontext(struct uverbs_attr_bundle *attrs);
+
 void ib_uverbs_release_ucq(struct ib_uverbs_completion_event_file *ev_file,
 			   struct ib_ucq_object *uobj);
 void ib_uverbs_release_uevent(struct ib_uevent_object *uobj);
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index d71ffe44b8ae..c8693f5231dd 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -203,104 +203,118 @@ _ib_uverbs_lookup_comp_file(s32 fd, struct uverbs_attr_bundle *attrs)
 #define ib_uverbs_lookup_comp_file(_fd, _ufile)                                \
 	_ib_uverbs_lookup_comp_file((_fd)*typecheck(s32, _fd), _ufile)
 
-static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
+int ib_alloc_ucontext(struct uverbs_attr_bundle *attrs)
 {
-	struct ib_uverbs_file *file = attrs->ufile;
-	struct ib_uverbs_get_context      cmd;
-	struct ib_uverbs_get_context_resp resp;
-	struct ib_ucontext		 *ucontext;
-	struct ib_rdmacg_object		 cg_obj;
+	struct ib_uverbs_file *ufile = attrs->ufile;
+	struct ib_ucontext *ucontext;
 	struct ib_device *ib_dev;
-	struct ib_uobject *uobj;
-	int ret;
 
-	ret = uverbs_request(attrs, &cmd, sizeof(cmd));
-	if (ret)
-		return ret;
+	ib_dev = srcu_dereference(ufile->device->ib_dev,
+				  &ufile->device->disassociate_srcu);
+	if (!ib_dev)
+		return -EIO;
+
+	ucontext = rdma_zalloc_drv_obj(ib_dev, ib_ucontext);
+	if (!ucontext)
+		return -ENOMEM;
+
+	ucontext->res.type = RDMA_RESTRACK_CTX;
+	ucontext->device = ib_dev;
+	ucontext->ufile = ufile;
+	xa_init_flags(&ucontext->mmap_xa, XA_FLAGS_ALLOC);
+	attrs->context = ucontext;
+	return 0;
+}
+
+int ib_init_ucontext(struct uverbs_attr_bundle *attrs)
+{
+	struct ib_ucontext *ucontext = attrs->context;
+	struct ib_uverbs_file *file = attrs->ufile;
+	int ret;
 
 	if (!down_read_trylock(&file->hw_destroy_rwsem))
 		return -EIO;
 	mutex_lock(&file->ucontext_lock);
-	ib_dev = srcu_dereference(file->device->ib_dev,
-				  &file->device->disassociate_srcu);
-	if (!ib_dev) {
-		ret = -EIO;
-		goto err;
-	}
-
 	if (file->ucontext) {
 		ret = -EINVAL;
 		goto err;
 	}
 
-	ret = ib_rdmacg_try_charge(&cg_obj, ib_dev, RDMACG_RESOURCE_HCA_HANDLE);
+	ret = ib_rdmacg_try_charge(&ucontext->cg_obj, ucontext->device,
+				   RDMACG_RESOURCE_HCA_HANDLE);
 	if (ret)
 		goto err;
 
-	ucontext = rdma_zalloc_drv_obj(ib_dev, ib_ucontext);
-	if (!ucontext) {
-		ret = -ENOMEM;
-		goto err_alloc;
-	}
+	ret = ucontext->device->ops.alloc_ucontext(ucontext,
+						   &attrs->driver_udata);
+	if (ret)
+		goto err_uncharge;
 
-	attrs->context = ucontext;
+	rdma_restrack_uadd(&ucontext->res);
 
-	ucontext->res.type = RDMA_RESTRACK_CTX;
-	ucontext->device = ib_dev;
-	ucontext->cg_obj = cg_obj;
-	/* ufile is required when some objects are released */
-	ucontext->ufile = file;
+	/*
+	 * Make sure that ib_uverbs_get_ucontext() sees the pointer update
+	 * only after all writes to setup the ucontext have completed
+	 */
+	smp_store_release(&file->ucontext, ucontext);
+
+	mutex_unlock(&file->ucontext_lock);
+	up_read(&file->hw_destroy_rwsem);
+	return 0;
 
-	ucontext->closing = false;
-	ucontext->cleanup_retryable = false;
+err_uncharge:
+	ib_rdmacg_uncharge(&ucontext->cg_obj, ucontext->device,
+			   RDMACG_RESOURCE_HCA_HANDLE);
+err:
+	mutex_unlock(&file->ucontext_lock);
+	up_read(&file->hw_destroy_rwsem);
+	return ret;
+}
 
-	xa_init_flags(&ucontext->mmap_xa, XA_FLAGS_ALLOC);
+static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
+{
+	struct ib_uverbs_get_context_resp resp;
+	struct ib_uverbs_get_context cmd;
+	struct ib_device *ib_dev;
+	struct ib_uobject *uobj;
+	int ret;
+
+	ret = uverbs_request(attrs, &cmd, sizeof(cmd));
+	if (ret)
+		return ret;
+
+	ret = ib_alloc_ucontext(attrs);
+	if (ret)
+		return ret;
 
 	uobj = uobj_alloc(UVERBS_OBJECT_ASYNC_EVENT, attrs, &ib_dev);
 	if (IS_ERR(uobj)) {
 		ret = PTR_ERR(uobj);
-		goto err_free;
+		goto err_ucontext;
 	}
 
-	resp.async_fd = uobj->id;
-	resp.num_comp_vectors = file->device->num_comp_vectors;
-
+	resp = (struct ib_uverbs_get_context_resp){
+		.num_comp_vectors = attrs->ufile->device->num_comp_vectors,
+		.async_fd = uobj->id,
+	};
 	ret = uverbs_response(attrs, &resp, sizeof(resp));
 	if (ret)
 		goto err_uobj;
 
-	ret = ib_dev->ops.alloc_ucontext(ucontext, &attrs->driver_udata);
+	ret = ib_init_ucontext(attrs);
 	if (ret)
 		goto err_uobj;
 
-	rdma_restrack_uadd(&ucontext->res);
-
 	ib_uverbs_init_async_event_file(
 		container_of(uobj, struct ib_uverbs_async_event_file, uobj));
 	rdma_alloc_commit_uobject(uobj, attrs);
-
-	/*
-	 * Make sure that ib_uverbs_get_ucontext() sees the pointer update
-	 * only after all writes to setup the ucontext have completed
-	 */
-	smp_store_release(&file->ucontext, ucontext);
-
-	mutex_unlock(&file->ucontext_lock);
-	up_read(&file->hw_destroy_rwsem);
 	return 0;
 
 err_uobj:
 	rdma_alloc_abort_uobject(uobj, attrs);
-
-err_free:
-	kfree(ucontext);
-
-err_alloc:
-	ib_rdmacg_uncharge(&cg_obj, ib_dev, RDMACG_RESOURCE_HCA_HANDLE);
-
-err:
-	mutex_unlock(&file->ucontext_lock);
-	up_read(&file->hw_destroy_rwsem);
+err_ucontext:
+	kfree(attrs->context);
+	attrs->context = NULL;
 	return ret;
 }
 
diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index fb9e75257607..2d4083bf4a04 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c
@@ -150,6 +150,9 @@ void ib_uverbs_release_uevent(struct ib_uevent_object *uobj)
 		READ_ONCE(uobj->uobject.ufile->async_file);
 	struct ib_uverbs_event *evt, *tmp;
 
+	if (!async_file)
+		return;
+
 	spin_lock_irq(&async_file->ev_queue.lock);
 	list_for_each_entry_safe(evt, tmp, &uobj->event_list, obj_list) {
 		list_del(&evt->list);
@@ -391,6 +394,9 @@ ib_uverbs_async_handler(struct ib_uverbs_async_event_file *async_file,
 	struct ib_uverbs_event *entry;
 	unsigned long flags;
 
+	if (!async_file)
+		return;
+
 	spin_lock_irqsave(&async_file->ev_queue.lock, flags);
 	if (async_file->ev_queue.is_closed) {
 		spin_unlock_irqrestore(&async_file->ev_queue.lock, flags);
@@ -476,12 +482,13 @@ void ib_uverbs_init_async_event_file(
 	ib_uverbs_init_event_queue(&async_file->ev_queue);
 
 	/* The first async_event_file becomes the default one for the file. */
-	lockdep_assert_held(&uverbs_file->ucontext_lock);
+	mutex_lock(&uverbs_file->ucontext_lock);
 	if (!uverbs_file->async_file) {
 		/* Pairs with the put in ib_uverbs_release_file */
 		uverbs_uobject_get(&async_file->uobj);
 		smp_store_release(&uverbs_file->async_file, async_file);
 	}
+	mutex_unlock(&uverbs_file->ucontext_lock);
 
 	INIT_IB_EVENT_HANDLER(&async_file->event_handler, ib_dev,
 			      ib_uverbs_event_handler);
diff --git a/drivers/infiniband/core/uverbs_std_types_async_fd.c b/drivers/infiniband/core/uverbs_std_types_async_fd.c
index 484dba136950..82ec0806b34b 100644
--- a/drivers/infiniband/core/uverbs_std_types_async_fd.c
+++ b/drivers/infiniband/core/uverbs_std_types_async_fd.c
@@ -14,10 +14,8 @@ static int UVERBS_HANDLER(UVERBS_METHOD_ASYNC_EVENT_ALLOC)(
 	struct ib_uobject *uobj =
 		uverbs_attr_get_uobject(attrs, UVERBS_METHOD_ASYNC_EVENT_ALLOC);
 
-	mutex_lock(&attrs->ufile->ucontext_lock);
 	ib_uverbs_init_async_event_file(
 		container_of(uobj, struct ib_uverbs_async_event_file, uobj));
-	mutex_unlock(&attrs->ufile->ucontext_lock);
 	return 0;
 }
 
diff --git a/drivers/infiniband/core/uverbs_std_types_device.c b/drivers/infiniband/core/uverbs_std_types_device.c
index 2a3f2f01028d..2c5943548760 100644
--- a/drivers/infiniband/core/uverbs_std_types_device.c
+++ b/drivers/infiniband/core/uverbs_std_types_device.c
@@ -200,6 +200,35 @@ static int UVERBS_HANDLER(UVERBS_METHOD_QUERY_PORT)(
 					     &resp, sizeof(resp));
 }
 
+static int UVERBS_HANDLER(UVERBS_METHOD_GET_CONTEXT)(
+	struct uverbs_attr_bundle *attrs)
+{
+	u32 num_comp = attrs->ufile->device->num_comp_vectors;
+	int ret;
+
+	ret = uverbs_copy_to(attrs, UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
+			     &num_comp, sizeof(num_comp));
+	if (IS_UVERBS_COPY_ERR(ret))
+		return ret;
+
+	ret = ib_alloc_ucontext(attrs);
+	if (ret)
+		return ret;
+	ret = ib_init_ucontext(attrs);
+	if (ret) {
+		kfree(attrs->context);
+		attrs->context = NULL;
+		return ret;
+	}
+	return 0;
+}
+
+DECLARE_UVERBS_NAMED_METHOD(
+	UVERBS_METHOD_GET_CONTEXT,
+	UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
+			    UVERBS_ATTR_TYPE(u32), UA_OPTIONAL),
+	UVERBS_ATTR_UHW());
+
 DECLARE_UVERBS_NAMED_METHOD(
 	UVERBS_METHOD_INFO_HANDLES,
 	/* Also includes any device specific object ids */
@@ -220,6 +249,7 @@ DECLARE_UVERBS_NAMED_METHOD(
 		UA_MANDATORY));
 
 DECLARE_UVERBS_GLOBAL_METHODS(UVERBS_OBJECT_DEVICE,
+			      &UVERBS_METHOD(UVERBS_METHOD_GET_CONTEXT),
 			      &UVERBS_METHOD(UVERBS_METHOD_INVOKE_WRITE),
 			      &UVERBS_METHOD(UVERBS_METHOD_INFO_HANDLES),
 			      &UVERBS_METHOD(UVERBS_METHOD_QUERY_PORT));
diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h
index 498955c576f3..da6c63c8cd32 100644
--- a/include/uapi/rdma/ib_user_ioctl_cmds.h
+++ b/include/uapi/rdma/ib_user_ioctl_cmds.h
@@ -68,6 +68,7 @@ enum uverbs_methods_device {
 	UVERBS_METHOD_INVOKE_WRITE,
 	UVERBS_METHOD_INFO_HANDLES,
 	UVERBS_METHOD_QUERY_PORT,
+	UVERBS_METHOD_GET_CONTEXT,
 };
 
 enum uverbs_attrs_invoke_write_cmd_attr_ids {
@@ -81,6 +82,10 @@ enum uverbs_attrs_query_port_cmd_attr_ids {
 	UVERBS_ATTR_QUERY_PORT_RESP,
 };
 
+enum uverbs_attrs_get_context_attr_ids {
+	UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
+};
+
 enum uverbs_attrs_create_cq_cmd_attr_ids {
 	UVERBS_ATTR_CREATE_CQ_HANDLE,
 	UVERBS_ATTR_CREATE_CQ_CQE,
-- 
cgit v1.2.3-59-g8ed1b


From 86dd738cf20cee4ac316274743e5707c4ccb6039 Mon Sep 17 00:00:00 2001
From: Michael Guralnik <michaelgur@mellanox.com>
Date: Wed, 8 Jan 2020 20:05:37 +0200
Subject: RDMA/efa: Allow passing of optional access flags for MR registration

As part of adding a range of optional access flags that drivers need to be
able to accept, mask this range inside efa driver.  This will prevent the
driver from failing when an access flag from that range is passed.

Link: https://lore.kernel.org/r/1578506740-22188-8-git-send-email-yishaih@mellanox.com
Signed-off-by: Michael Guralnik <michaelgur@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/efa/efa_verbs.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c
index 9a6cff718c49..a556572058ff 100644
--- a/drivers/infiniband/hw/efa/efa_verbs.c
+++ b/drivers/infiniband/hw/efa/efa_verbs.c
@@ -1367,6 +1367,7 @@ struct ib_mr *efa_reg_mr(struct ib_pd *ibpd, u64 start, u64 length,
 		IB_ACCESS_LOCAL_WRITE |
 		(is_rdma_read_cap(dev) ? IB_ACCESS_REMOTE_READ : 0);
 
+	access_flags &= ~IB_UVERBS_ACCESS_OPTIONAL_RANGE;
 	if (access_flags & ~supp_access_flags) {
 		ibdev_dbg(&dev->ibdev,
 			  "Unsupported access flags[%#x], supported[%#x]\n",
-- 
cgit v1.2.3-59-g8ed1b


From 811646998e3e64af19524c8e99b790a4043c5f9b Mon Sep 17 00:00:00 2001
From: Michael Guralnik <michaelgur@mellanox.com>
Date: Wed, 8 Jan 2020 20:05:39 +0200
Subject: RDMA/core: Add the core support field to METHOD_GET_CONTEXT

Add the core support field to METHOD_GET_CONTEXT, this field should
represent capabilities that are not device-specific.

Return support for optional access flags for memory regions. User-space
will use this capability to mask the optional access flags for
unsupporting kernels.

Link: https://lore.kernel.org/r/1578506740-22188-10-git-send-email-yishaih@mellanox.com
Signed-off-by: Michael Guralnik <michaelgur@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs_std_types_device.c | 8 ++++++++
 include/uapi/rdma/ib_user_ioctl_cmds.h            | 1 +
 include/uapi/rdma/ib_user_ioctl_verbs.h           | 4 ++++
 3 files changed, 13 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs_std_types_device.c b/drivers/infiniband/core/uverbs_std_types_device.c
index 2c5943548760..ae4a59d6f9b1 100644
--- a/drivers/infiniband/core/uverbs_std_types_device.c
+++ b/drivers/infiniband/core/uverbs_std_types_device.c
@@ -204,6 +204,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_GET_CONTEXT)(
 	struct uverbs_attr_bundle *attrs)
 {
 	u32 num_comp = attrs->ufile->device->num_comp_vectors;
+	u64 core_support = IB_UVERBS_CORE_SUPPORT_OPTIONAL_MR_ACCESS;
 	int ret;
 
 	ret = uverbs_copy_to(attrs, UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
@@ -211,6 +212,11 @@ static int UVERBS_HANDLER(UVERBS_METHOD_GET_CONTEXT)(
 	if (IS_UVERBS_COPY_ERR(ret))
 		return ret;
 
+	ret = uverbs_copy_to(attrs, UVERBS_ATTR_GET_CONTEXT_CORE_SUPPORT,
+			     &core_support, sizeof(core_support));
+	if (IS_UVERBS_COPY_ERR(ret))
+		return ret;
+
 	ret = ib_alloc_ucontext(attrs);
 	if (ret)
 		return ret;
@@ -227,6 +233,8 @@ DECLARE_UVERBS_NAMED_METHOD(
 	UVERBS_METHOD_GET_CONTEXT,
 	UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
 			    UVERBS_ATTR_TYPE(u32), UA_OPTIONAL),
+	UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_GET_CONTEXT_CORE_SUPPORT,
+			    UVERBS_ATTR_TYPE(u64), UA_OPTIONAL),
 	UVERBS_ATTR_UHW());
 
 DECLARE_UVERBS_NAMED_METHOD(
diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h
index da6c63c8cd32..d4ddbe4e696c 100644
--- a/include/uapi/rdma/ib_user_ioctl_cmds.h
+++ b/include/uapi/rdma/ib_user_ioctl_cmds.h
@@ -84,6 +84,7 @@ enum uverbs_attrs_query_port_cmd_attr_ids {
 
 enum uverbs_attrs_get_context_attr_ids {
 	UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS,
+	UVERBS_ATTR_GET_CONTEXT_CORE_SUPPORT,
 };
 
 enum uverbs_attrs_create_cq_cmd_attr_ids {
diff --git a/include/uapi/rdma/ib_user_ioctl_verbs.h b/include/uapi/rdma/ib_user_ioctl_verbs.h
index 2a165f40ee38..a640bb814be0 100644
--- a/include/uapi/rdma/ib_user_ioctl_verbs.h
+++ b/include/uapi/rdma/ib_user_ioctl_verbs.h
@@ -44,6 +44,10 @@
 #define IB_UVERBS_ACCESS_OPTIONAL_FIRST (1 << 20)
 #define IB_UVERBS_ACCESS_OPTIONAL_LAST (1 << 29)
 
+enum ib_uverbs_core_support {
+	IB_UVERBS_CORE_SUPPORT_OPTIONAL_MR_ACCESS = 1 << 0,
+};
+
 enum ib_uverbs_access_flags {
 	IB_UVERBS_ACCESS_LOCAL_WRITE = 1 << 0,
 	IB_UVERBS_ACCESS_REMOTE_WRITE = 1 << 1,
-- 
cgit v1.2.3-59-g8ed1b


From d6de0bb1850f6fcefd4f9fed2de69c0915a2c8a9 Mon Sep 17 00:00:00 2001
From: Michael Guralnik <michaelgur@mellanox.com>
Date: Wed, 8 Jan 2020 20:05:40 +0200
Subject: RDMA/mlx5: Set relaxed ordering when requested

Enable relaxed ordering in the mkey context when requested. As relaxed
ordering is not currently supported in UMR, disable UMR usage for relaxed
ordering MRs.

Link: https://lore.kernel.org/r/1578506740-22188-11-git-send-email-yishaih@mellanox.com
Signed-off-by: Michael Guralnik <michaelgur@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  5 ++++-
 drivers/infiniband/hw/mlx5/mr.c      | 19 +++++++++++++++++--
 drivers/infiniband/hw/mlx5/odp.c     |  2 +-
 drivers/infiniband/hw/mlx5/qp.c      |  2 +-
 4 files changed, 23 insertions(+), 5 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 676462c8909f..aa14d3c8abd9 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1523,7 +1523,7 @@ int mlx5_ib_qp_set_counter(struct ib_qp *qp, struct rdma_counter *counter);
 u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num);
 
 static inline bool mlx5_ib_can_use_umr(struct mlx5_ib_dev *dev,
-				       bool do_modify_atomic)
+				       bool do_modify_atomic, int access_flags)
 {
 	if (MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled))
 		return false;
@@ -1533,6 +1533,9 @@ static inline bool mlx5_ib_can_use_umr(struct mlx5_ib_dev *dev,
 	    MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled))
 		return false;
 
+	if (access_flags & IB_ACCESS_RELAXED_ORDERING)
+		return false;
+
 	return true;
 }
 
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index a52784304593..1fcae0735e01 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -661,12 +661,21 @@ int mlx5_mr_cache_cleanup(struct mlx5_ib_dev *dev)
 static void set_mkc_access_pd_addr_fields(void *mkc, int acc, u64 start_addr,
 					  struct ib_pd *pd)
 {
+	struct mlx5_ib_dev *dev = to_mdev(pd->device);
+
 	MLX5_SET(mkc, mkc, a, !!(acc & IB_ACCESS_REMOTE_ATOMIC));
 	MLX5_SET(mkc, mkc, rw, !!(acc & IB_ACCESS_REMOTE_WRITE));
 	MLX5_SET(mkc, mkc, rr, !!(acc & IB_ACCESS_REMOTE_READ));
 	MLX5_SET(mkc, mkc, lw, !!(acc & IB_ACCESS_LOCAL_WRITE));
 	MLX5_SET(mkc, mkc, lr, 1);
 
+	if (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write))
+		MLX5_SET(mkc, mkc, relaxed_ordering_write,
+			 !!(acc & IB_ACCESS_RELAXED_ORDERING));
+	if (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read))
+		MLX5_SET(mkc, mkc, relaxed_ordering_read,
+			 !!(acc & IB_ACCESS_RELAXED_ORDERING));
+
 	MLX5_SET(mkc, mkc, pd, to_mpd(pd)->pdn);
 	MLX5_SET(mkc, mkc, qpn, 0xffffff);
 	MLX5_SET64(mkc, mkc, start_addr, start_addr);
@@ -1063,6 +1072,12 @@ static struct mlx5_ib_mr *reg_create(struct ib_mr *ibmr, struct ib_pd *pd,
 	mkc = MLX5_ADDR_OF(create_mkey_in, in, memory_key_mkey_entry);
 	MLX5_SET(mkc, mkc, free, !populate);
 	MLX5_SET(mkc, mkc, access_mode_1_0, MLX5_MKC_ACCESS_MODE_MTT);
+	if (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write))
+		MLX5_SET(mkc, mkc, relaxed_ordering_write,
+			 !!(access_flags & IB_ACCESS_RELAXED_ORDERING));
+	if (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read))
+		MLX5_SET(mkc, mkc, relaxed_ordering_read,
+			 !!(access_flags & IB_ACCESS_RELAXED_ORDERING));
 	MLX5_SET(mkc, mkc, a, !!(access_flags & IB_ACCESS_REMOTE_ATOMIC));
 	MLX5_SET(mkc, mkc, rw, !!(access_flags & IB_ACCESS_REMOTE_WRITE));
 	MLX5_SET(mkc, mkc, rr, !!(access_flags & IB_ACCESS_REMOTE_READ));
@@ -1251,7 +1266,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 	if (err < 0)
 		return ERR_PTR(err);
 
-	use_umr = mlx5_ib_can_use_umr(dev, true);
+	use_umr = mlx5_ib_can_use_umr(dev, true, access_flags);
 
 	if (order <= mr_cache_max_order(dev) && use_umr) {
 		mr = alloc_mr_from_cache(pd, umem, virt_addr, length, ncont,
@@ -1419,7 +1434,7 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 			goto err;
 	}
 
-	if (!mlx5_ib_can_use_umr(dev, true) ||
+	if (!mlx5_ib_can_use_umr(dev, true, access_flags) ||
 	    (flags & IB_MR_REREG_TRANS && !use_umr_mtt_update(mr, addr, len))) {
 		/*
 		 * UMR can't be used - MKey needs to be replaced.
diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index 92da6c4f7ddd..409dffb5b0c0 100644
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -380,7 +380,7 @@ void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 	memset(caps, 0, sizeof(*caps));
 
 	if (!MLX5_CAP_GEN(dev->mdev, pg) ||
-	    !mlx5_ib_can_use_umr(dev, true))
+	    !mlx5_ib_can_use_umr(dev, true, 0))
 		return;
 
 	caps->general_caps = IB_ODP_SUPPORT;
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 04126ef74412..425efa1349b0 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -4823,7 +4823,7 @@ static int set_reg_wr(struct mlx5_ib_qp *qp,
 	bool atomic = wr->access & IB_ACCESS_REMOTE_ATOMIC;
 	u8 flags = 0;
 
-	if (!mlx5_ib_can_use_umr(dev, atomic)) {
+	if (!mlx5_ib_can_use_umr(dev, atomic, wr->access)) {
 		mlx5_ib_warn(to_mdev(qp->ibqp.device),
 			     "Fast update of %s for MR is disabled\n",
 			     (MLX5_CAP_GEN(dev->mdev,
-- 
cgit v1.2.3-59-g8ed1b


From d7fab916374c8f5488ca792c57ea8b72e8631213 Mon Sep 17 00:00:00 2001
From: Avihai Horon <avihaih@mellanox.com>
Date: Wed, 15 Jan 2020 16:54:59 +0200
Subject: IB/mlx5: Expose RoCE accelerator counters

Introduce the following RoCE accelerator counters:
* roce_adp_retrans - number of adaptive retransmission for RoCE traffic.
* roce_adp_retrans_to - number of times RoCE traffic reached time out
  due to adaptive retransmission.
* roce_slow_restart - number of times RoCE slow restart was used.
* roce_slow_restart_cnps - number of times RoCE slow restart
  generate CNP packets.
* roce_slow_restart_trans - number of times RoCE slow restart change
  state to slow restart.

Link: https://lore.kernel.org/r/20200115145459.83280-3-leon@kernel.org
Signed-off-by: Avihai Horon <avihaih@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 97bcf01960ae..5d41a2c69400 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -5370,6 +5370,14 @@ static const struct mlx5_ib_counter extended_err_cnts[] = {
 	INIT_Q_COUNTER(req_cqe_flush_error),
 };
 
+static const struct mlx5_ib_counter roce_accl_cnts[] = {
+	INIT_Q_COUNTER(roce_adp_retrans),
+	INIT_Q_COUNTER(roce_adp_retrans_to),
+	INIT_Q_COUNTER(roce_slow_restart),
+	INIT_Q_COUNTER(roce_slow_restart_cnps),
+	INIT_Q_COUNTER(roce_slow_restart_trans),
+};
+
 #define INIT_EXT_PPCNT_COUNTER(_name)		\
 	{ .name = #_name, .offset =	\
 	MLX5_BYTE_OFF(ppcnt_reg, \
@@ -5418,6 +5426,9 @@ static int __mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev,
 	if (MLX5_CAP_GEN(dev->mdev, enhanced_error_q_counters))
 		num_counters += ARRAY_SIZE(extended_err_cnts);
 
+	if (MLX5_CAP_GEN(dev->mdev, roce_accl))
+		num_counters += ARRAY_SIZE(roce_accl_cnts);
+
 	cnts->num_q_counters = num_counters;
 
 	if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) {
@@ -5478,6 +5489,13 @@ static void mlx5_ib_fill_counters(struct mlx5_ib_dev *dev,
 		}
 	}
 
+	if (MLX5_CAP_GEN(dev->mdev, roce_accl)) {
+		for (i = 0; i < ARRAY_SIZE(roce_accl_cnts); i++, j++) {
+			names[j] = roce_accl_cnts[i].name;
+			offsets[j] = roce_accl_cnts[i].offset;
+		}
+	}
+
 	if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) {
 		for (i = 0; i < ARRAY_SIZE(cong_cnts); i++, j++) {
 			names[j] = cong_cnts[i].name;
-- 
cgit v1.2.3-59-g8ed1b


From eaad647e5cc27f7b46a27f3b85b14c4c8a64bffa Mon Sep 17 00:00:00 2001
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
Date: Wed, 15 Jan 2020 10:50:50 +0200
Subject: IB/mlx4: Fix memory leak in add_gid error flow

In procedure mlx4_ib_add_gid(), if the driver is unable to update the FW
gid table, there is a memory leak in the driver's copy of the gid table:
the gid entry's context buffer is not freed.

If such an error occurs, free the entry's context buffer, and mark the
entry as available (by setting its context pointer to NULL).

Fixes: e26be1bfef81 ("IB/mlx4: Implement ib_device callbacks")
Link: https://lore.kernel.org/r/20200115085050.73746-1-leon@kernel.org
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx4/main.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 34055cbab38c..2f5d9b181848 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -246,6 +246,13 @@ static int mlx4_ib_update_gids(struct gid_entry *gids,
 	return mlx4_ib_update_gids_v1(gids, ibdev, port_num);
 }
 
+static void free_gid_entry(struct gid_entry *entry)
+{
+	memset(&entry->gid, 0, sizeof(entry->gid));
+	kfree(entry->ctx);
+	entry->ctx = NULL;
+}
+
 static int mlx4_ib_add_gid(const struct ib_gid_attr *attr, void **context)
 {
 	struct mlx4_ib_dev *ibdev = to_mdev(attr->device);
@@ -313,6 +320,8 @@ static int mlx4_ib_add_gid(const struct ib_gid_attr *attr, void **context)
 				     GFP_ATOMIC);
 		if (!gids) {
 			ret = -ENOMEM;
+			*context = NULL;
+			free_gid_entry(&port_gid_table->gids[free]);
 		} else {
 			for (i = 0; i < MLX4_MAX_PORT_GIDS; i++) {
 				memcpy(&gids[i].gid, &port_gid_table->gids[i].gid, sizeof(union ib_gid));
@@ -324,6 +333,12 @@ static int mlx4_ib_add_gid(const struct ib_gid_attr *attr, void **context)
 
 	if (!ret && hw_update) {
 		ret = mlx4_ib_update_gids(gids, ibdev, attr->port_num);
+		if (ret) {
+			spin_lock_bh(&iboe->lock);
+			*context = NULL;
+			free_gid_entry(&port_gid_table->gids[free]);
+			spin_unlock_bh(&iboe->lock);
+		}
 		kfree(gids);
 	}
 
@@ -353,10 +368,7 @@ static int mlx4_ib_del_gid(const struct ib_gid_attr *attr, void **context)
 		if (!ctx->refcount) {
 			unsigned int real_index = ctx->real_index;
 
-			memset(&port_gid_table->gids[real_index].gid, 0,
-			       sizeof(port_gid_table->gids[real_index].gid));
-			kfree(port_gid_table->gids[real_index].ctx);
-			port_gid_table->gids[real_index].ctx = NULL;
+			free_gid_entry(&port_gid_table->gids[real_index]);
 			hw_update = 1;
 		}
 	}
-- 
cgit v1.2.3-59-g8ed1b


From 6b3712c0246ca7b2b8fa05eab2362cf267410f7e Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Wed, 15 Jan 2020 20:20:44 +0000
Subject: RDMA/core: Ensure that rdma_user_mmap_entry_remove() is a fence

The set of entry->driver_removed is missing locking, protect it with
xa_lock() which is held by the only reader.

Otherwise readers may continue to see driver_removed = false after
rdma_user_mmap_entry_remove() returns and may continue to try and
establish new mmaps.

Fixes: 3411f9f01b76 ("RDMA/core: Create mmap database and cookie helper functions")
Link: https://lore.kernel.org/r/20200115202041.GA17199@ziepe.ca
Reviewed-by: Gal Pressman <galpress@amazon.com>
Acked-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/ib_core_uverbs.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/ib_core_uverbs.c b/drivers/infiniband/core/ib_core_uverbs.c
index b7cb59844ece..b51bd7087a88 100644
--- a/drivers/infiniband/core/ib_core_uverbs.c
+++ b/drivers/infiniband/core/ib_core_uverbs.c
@@ -232,7 +232,9 @@ void rdma_user_mmap_entry_remove(struct rdma_user_mmap_entry *entry)
 	if (!entry)
 		return;
 
+	xa_lock(&entry->ucontext->mmap_xa);
 	entry->driver_removed = true;
+	xa_unlock(&entry->ucontext->mmap_xa);
 	kref_put(&entry->ref, rdma_user_mmap_entry_free);
 }
 EXPORT_SYMBOL(rdma_user_mmap_entry_remove);
-- 
cgit v1.2.3-59-g8ed1b


From 4bbd4923d1f5627b0c47a9d7dfb5cc91224cfe0c Mon Sep 17 00:00:00 2001
From: Danit Goldberg <danitg@mellanox.com>
Date: Thu, 16 Jan 2020 14:00:48 +0200
Subject: IB/mlx5: Return the administrative GUID if exists

A user can change the operational GUID (a.k.a affective GUID) through
link/infiniband. Therefore it is preferred to return the currently set
GUID if it exists instead of the operational.

This way the PF can query which VF GUID will be set in the next bind.  In
order to align with MAC address, zero is returned if administrative GUID
is not set.

For example, before setting administrative GUID:
 $ ip link show
 ib0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 4092 qdisc mq state UP mode DEFAULT group default qlen 256
 link/infiniband 00:00:00:08:fe:80:00:00:00:00:00:00:52:54:00:c0:fe:12:34:55 brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff
 vf 0     link/infiniband 00:00:00:08:fe:80:00:00:00:00:00:00:52:54:00:c0:fe:12:34:55 brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff,
 spoof checking off, NODE_GUID 00:00:00:00:00:00:00:00, PORT_GUID 00:00:00:00:00:00:00:00, link-state auto, trust off, query_rss off

Then:

 $ ip link set ib0 vf 0 node_guid 11:00:af:21:cb:05:11:00
 $ ip link set ib0 vf 0 port_guid 22:11:af:21:cb:05:11:00

After setting administrative GUID:
 $ ip link show
 ib0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 4092 qdisc mq state UP mode DEFAULT group default qlen 256
 link/infiniband 00:00:00:08:fe:80:00:00:00:00:00:00:52:54:00:c0:fe:12:34:55 brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff
 vf 0     link/infiniband 00:00:00:08:fe:80:00:00:00:00:00:00:52:54:00:c0:fe:12:34:55 brd 00:ff:ff:ff:ff:12:40:1b:ff:ff:00:00:00:00:00:00:ff:ff:ff:ff,
 spoof checking off, NODE_GUID 11:00:af:21:cb:05:11:00, PORT_GUID 22:11:af:21:cb:05:11:00, link-state auto, trust off, query_rss off

Fixes: 9c0015ef0928 ("IB/mlx5: Implement callbacks for getting VFs GUID attributes")
Link: https://lore.kernel.org/r/20200116120048.12744-1-leon@kernel.org
Signed-off-by: Danit Goldberg <danitg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/ib_virt.c | 28 ++++++++++++----------------
 include/linux/mlx5/driver.h          |  5 +++++
 2 files changed, 17 insertions(+), 16 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx5/ib_virt.c b/drivers/infiniband/hw/mlx5/ib_virt.c
index 4f0edd4832bd..b61165359954 100644
--- a/drivers/infiniband/hw/mlx5/ib_virt.c
+++ b/drivers/infiniband/hw/mlx5/ib_virt.c
@@ -164,8 +164,10 @@ static int set_vf_node_guid(struct ib_device *device, int vf, u8 port, u64 guid)
 	in->field_select = MLX5_HCA_VPORT_SEL_NODE_GUID;
 	in->node_guid = guid;
 	err = mlx5_core_modify_hca_vport_context(mdev, 1, 1, vf + 1, in);
-	if (!err)
+	if (!err) {
 		vfs_ctx[vf].node_guid = guid;
+		vfs_ctx[vf].node_guid_valid = 1;
+	}
 	kfree(in);
 	return err;
 }
@@ -185,8 +187,10 @@ static int set_vf_port_guid(struct ib_device *device, int vf, u8 port, u64 guid)
 	in->field_select = MLX5_HCA_VPORT_SEL_PORT_GUID;
 	in->port_guid = guid;
 	err = mlx5_core_modify_hca_vport_context(mdev, 1, 1, vf + 1, in);
-	if (!err)
+	if (!err) {
 		vfs_ctx[vf].port_guid = guid;
+		vfs_ctx[vf].port_guid_valid = 1;
+	}
 	kfree(in);
 	return err;
 }
@@ -208,20 +212,12 @@ int mlx5_ib_get_vf_guid(struct ib_device *device, int vf, u8 port,
 {
 	struct mlx5_ib_dev *dev = to_mdev(device);
 	struct mlx5_core_dev *mdev = dev->mdev;
-	struct mlx5_hca_vport_context *rep;
-	int err;
-
-	rep = kzalloc(sizeof(*rep), GFP_KERNEL);
-	if (!rep)
-		return -ENOMEM;
+	struct mlx5_vf_context *vfs_ctx = mdev->priv.sriov.vfs_ctx;
 
-	err = mlx5_query_hca_vport_context(mdev, 1, 1, vf+1, rep);
-	if (err)
-		goto ex;
+	node_guid->guid =
+		vfs_ctx[vf].node_guid_valid ? vfs_ctx[vf].node_guid : 0;
+	port_guid->guid =
+		vfs_ctx[vf].port_guid_valid ? vfs_ctx[vf].port_guid : 0;
 
-	port_guid->guid = rep->port_guid;
-	node_guid->guid = rep->node_guid;
-ex:
-	kfree(rep);
-	return err;
+	return 0;
 }
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 27200dea0297..a24937fc56b9 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -461,6 +461,11 @@ struct mlx5_vf_context {
 	int	enabled;
 	u64	port_guid;
 	u64	node_guid;
+	/* Valid bits are used to validate administrative guid only.
+	 * Enabled after ndo_set_vf_guid
+	 */
+	u8	port_guid_valid:1;
+	u8	node_guid_valid:1;
 	enum port_state_policy	policy;
 };
 
-- 
cgit v1.2.3-59-g8ed1b


From 792a7c1f2e74bba878a1836cfbe9611f98576d3b Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Thu, 16 Jan 2020 13:00:31 -0400
Subject: RDMA/cm: Add accessors for CM_REQ transport_type

Access the two fields through wrappers, like all other fields, to make it
clearer what is happening.

Link: https://lore.kernel.org/r/20200116170037.30109-2-jgg@ziepe.ca
Tested-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm_msgs.h | 41 +++++++++++++++++++++++++++------------
 1 file changed, 29 insertions(+), 12 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h
index 23e3d0c6a67c..9af9a3212074 100644
--- a/drivers/infiniband/core/cm_msgs.h
+++ b/drivers/infiniband/core/cm_msgs.h
@@ -124,14 +124,37 @@ static inline void cm_req_set_remote_resp_timeout(struct cm_req_msg *req_msg,
 					  0xFFFFFF07));
 }
 
+static inline u8 cm_req_get_transport_type(struct cm_req_msg *req_msg)
+{
+	return (u8) ((be32_to_cpu(req_msg->offset40) & 0x06) >> 1);
+}
+
+static inline void cm_req_set_transport_type(struct cm_req_msg *req_msg, u8 val)
+{
+	req_msg->offset40 =
+		cpu_to_be32((be32_to_cpu(req_msg->offset40) & 0xFFFFFFF9) |
+		(val << 1));
+}
+
+static inline u8 cm_req_get_transport_type_ex(struct cm_req_msg *req_msg)
+{
+	return req_msg->offset51 & 0x7;
+}
+
+static inline void cm_req_set_transport_type_ex(struct cm_req_msg *req_msg,
+						u8 val)
+{
+	req_msg->offset51 = (req_msg->offset51 & 0xF8) | val;
+}
+
 static inline enum ib_qp_type cm_req_get_qp_type(struct cm_req_msg *req_msg)
 {
-	u8 transport_type = (u8) (be32_to_cpu(req_msg->offset40) & 0x06) >> 1;
+	u8 transport_type = cm_req_get_transport_type(req_msg);
 	switch(transport_type) {
 	case 0: return IB_QPT_RC;
 	case 1: return IB_QPT_UC;
 	case 3:
-		switch (req_msg->offset51 & 0x7) {
+		switch (cm_req_get_transport_type_ex(req_msg)) {
 		case 1: return IB_QPT_XRC_TGT;
 		default: return 0;
 		}
@@ -144,20 +167,14 @@ static inline void cm_req_set_qp_type(struct cm_req_msg *req_msg,
 {
 	switch(qp_type) {
 	case IB_QPT_UC:
-		req_msg->offset40 = cpu_to_be32((be32_to_cpu(
-						  req_msg->offset40) &
-						   0xFFFFFFF9) | 0x2);
+		cm_req_set_transport_type(req_msg, 1);
 		break;
 	case IB_QPT_XRC_INI:
-		req_msg->offset40 = cpu_to_be32((be32_to_cpu(
-						 req_msg->offset40) &
-						   0xFFFFFFF9) | 0x6);
-		req_msg->offset51 = (req_msg->offset51 & 0xF8) | 1;
+		cm_req_set_transport_type(req_msg, 3);
+		cm_req_set_transport_type_ex(req_msg, 1);
 		break;
 	default:
-		req_msg->offset40 = cpu_to_be32(be32_to_cpu(
-						 req_msg->offset40) &
-						  0xFFFFFFF9);
+		cm_req_set_transport_type(req_msg, 0);
 	}
 }
 
-- 
cgit v1.2.3-59-g8ed1b


From d05d4ac4c9316a2e2ea5be434a7f7a021ec44a3c Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@mellanox.com>
Date: Thu, 16 Jan 2020 13:00:32 -0400
Subject: RDMA/cm: Add SET/GET implementations to hide IBA wire format

There is no separation between RDMA-CM wire format as it is declared in
IBTA and kernel logic which implements needed support. Such situation
causes to many mistakes in conversion between big-endian (wire format)
and CPU format used by kernel. It also mixes RDMA core code with
combination of uXX and beXX variables.

The idea that all accesses to IBA definitions will go through special
GET/SET macros to ensure that no conversion mistakes are made. The
shifting and masking required to read the value is automatically deduced
using the field offset description from the tables in the IBA
specification.

This starts with the CM MADs described in IBTA release 1.3 volume 1.

To confirm that the new macros behave the same as the old accessors a
self-test is included in this patch.

Each macro replacing a straightforward struct field compile-time tests
that the new field has the same offsetof() and width as the old field.

For the fields with accessor functions a runtime test, the 'all ones'
value is placed in a dummy message and read back in several ways to
confirm that both approaches give identical results.

Later patches in this series delete the self test.

This creates a tested table of new field name, old field name(s) and some
meta information like BE coding for the functions which will be used in
the next patches.

Link: https://lore.kernel.org/r/20200116170037.30109-3-jgg@ziepe.ca
Link: https://lore.kernel.org/r/20191212093830.316934-5-leon@kernel.org
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Tested-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c      | 267 ++++++++++++++++++++++++++++++++++++++
 drivers/infiniband/core/cm_msgs.h |   1 +
 include/rdma/iba.h                | 146 +++++++++++++++++++++
 include/rdma/ibta_vol1_c12.h      | 208 +++++++++++++++++++++++++++++
 4 files changed, 622 insertions(+)
 create mode 100644 include/rdma/iba.h
 create mode 100644 include/rdma/ibta_vol1_c12.h

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index d396b987b0f2..7f609979e4de 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -4253,10 +4253,277 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data)
 	kfree(cm_dev);
 }
 
+/*
+ * Check at compile time that the byte offset and length of field old_name in
+ * the struct matches the byte offset and length in the new macro.
+ */
+#define _IBA_CHECK_OFF(old_name, field_struct, field_offset, mask, bits) \
+	static_assert(offsetof(field_struct, old_name) == (field_offset));     \
+	static_assert(bits == sizeof(((field_struct *)0)->old_name) * 8)
+#define IBA_CHECK_OFF(field, old_name) _IBA_CHECK_OFF(old_name, field)
+
+IBA_CHECK_OFF(CM_REQ_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_REQ_SERVICE_ID, service_id);
+IBA_CHECK_OFF(CM_REQ_LOCAL_CA_GUID, local_ca_guid);
+IBA_CHECK_OFF(CM_REQ_LOCAL_Q_KEY, local_qkey);
+IBA_CHECK_OFF(CM_REQ_PARTITION_KEY, pkey);
+IBA_CHECK_OFF(CM_REQ_PRIMARY_LOCAL_PORT_LID, primary_local_lid);
+IBA_CHECK_OFF(CM_REQ_PRIMARY_REMOTE_PORT_LID, primary_remote_lid);
+IBA_CHECK_OFF(CM_REQ_PRIMARY_LOCAL_PORT_GID, primary_local_gid);
+IBA_CHECK_OFF(CM_REQ_PRIMARY_REMOTE_PORT_GID, primary_remote_gid);
+IBA_CHECK_OFF(CM_REQ_PRIMARY_TRAFFIC_CLASS, primary_traffic_class);
+IBA_CHECK_OFF(CM_REQ_PRIMARY_HOP_LIMIT, primary_hop_limit);
+IBA_CHECK_OFF(CM_REQ_ALTERNATE_LOCAL_PORT_LID, alt_local_lid);
+IBA_CHECK_OFF(CM_REQ_ALTERNATE_REMOTE_PORT_LID, alt_remote_lid);
+IBA_CHECK_OFF(CM_REQ_ALTERNATE_LOCAL_PORT_GID, alt_local_gid);
+IBA_CHECK_OFF(CM_REQ_ALTERNATE_REMOTE_PORT_GID, alt_remote_gid);
+IBA_CHECK_OFF(CM_REQ_ALTERNATE_TRAFFIC_CLASS, alt_traffic_class);
+IBA_CHECK_OFF(CM_REQ_ALTERNATE_HOP_LIMIT, alt_hop_limit);
+IBA_CHECK_OFF(CM_REQ_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_MRA_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_MRA_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_MRA_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_REJ_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_REJ_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_REJ_REASON, reason);
+IBA_CHECK_OFF(CM_REJ_ARI, ari);
+IBA_CHECK_OFF(CM_REJ_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_REP_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_REP_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_REP_LOCAL_Q_KEY, local_qkey);
+IBA_CHECK_OFF(CM_REP_RESPONDER_RESOURCES, resp_resources);
+IBA_CHECK_OFF(CM_REP_INITIATOR_DEPTH, initiator_depth);
+IBA_CHECK_OFF(CM_REP_LOCAL_CA_GUID, local_ca_guid);
+IBA_CHECK_OFF(CM_REP_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_RTU_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_RTU_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_RTU_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_DREQ_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_DREQ_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_DREQ_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_DREP_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_DREP_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_DREP_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_LAP_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_LAP_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_LAP_ALTERNATE_LOCAL_PORT_LID, alt_local_lid);
+IBA_CHECK_OFF(CM_LAP_ALTERNATE_REMOTE_PORT_LID, alt_remote_lid);
+IBA_CHECK_OFF(CM_LAP_ALTERNATE_LOCAL_PORT_GID, alt_local_gid);
+IBA_CHECK_OFF(CM_LAP_ALTERNATE_REMOTE_PORT_GID, alt_remote_gid);
+IBA_CHECK_OFF(CM_LAP_ALTERNATE_HOP_LIMIT, alt_hop_limit);
+IBA_CHECK_OFF(CM_LAP_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_APR_LOCAL_COMM_ID, local_comm_id);
+IBA_CHECK_OFF(CM_APR_REMOTE_COMM_ID, remote_comm_id);
+IBA_CHECK_OFF(CM_APR_ADDITIONAL_INFORMATION_LENGTH, info_length);
+IBA_CHECK_OFF(CM_APR_AR_STATUS, ap_status);
+IBA_CHECK_OFF(CM_APR_ADDITIONAL_INFORMATION, info);
+IBA_CHECK_OFF(CM_APR_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_SIDR_REQ_REQUESTID, request_id);
+IBA_CHECK_OFF(CM_SIDR_REQ_PARTITION_KEY, pkey);
+IBA_CHECK_OFF(CM_SIDR_REQ_SERVICEID, service_id);
+IBA_CHECK_OFF(CM_SIDR_REQ_PRIVATE_DATA, private_data);
+IBA_CHECK_OFF(CM_SIDR_REP_REQUESTID, request_id);
+IBA_CHECK_OFF(CM_SIDR_REP_STATUS, status);
+IBA_CHECK_OFF(CM_SIDR_REP_ADDITIONAL_INFORMATION_LENGTH, info_length);
+IBA_CHECK_OFF(CM_SIDR_REP_SERVICEID, service_id);
+IBA_CHECK_OFF(CM_SIDR_REP_Q_KEY, qkey);
+IBA_CHECK_OFF(CM_SIDR_REP_ADDITIONAL_INFORMATION, info);
+IBA_CHECK_OFF(CM_SIDR_REP_PRIVATE_DATA, private_data);
+
+/*
+ * Check that the new macro gets the same bits as the old get function.
+ *  - IBA_SET() IBA_GET and old get_fn all agree on the field width.
+ *    The field width should match what IBA_SET truncates to
+ *  - Reading from an all ones data should not return extra bits
+ *  - Setting '1' should be the same (ie no endian problems)
+ */
+/* defeat builtin_constant checks */
+u64 cm_global_all_ones = 0xffffffffffffffffULL;
+#define _IBA_CHECK_GET(fn, field_struct, field_offset, mask, bits)             \
+	({                                                                     \
+		field_struct *lmsg = (field_struct *)msg;                      \
+		unsigned long long all_ones;                                   \
+		static_assert(sizeof(*lmsg) <= sizeof(msg));                   \
+                                                                               \
+		bitmap_zero(msg, nbits);                                       \
+		_IBA_SET(field_struct, field_offset, mask, bits, lmsg,         \
+			 cm_global_all_ones);                                  \
+		all_ones = (1ULL << bitmap_weight(msg, nbits)) - 1;            \
+		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
+		    all_ones) {                                                \
+			printk("Failed #1 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+		if (fn != all_ones) {                                          \
+			printk("Failed #2 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+                                                                               \
+		bitmap_fill(msg, nbits);                                       \
+		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
+		    all_ones) {                                                \
+			printk("Failed #3 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+		if (fn != all_ones) {                                          \
+			printk("Failed #4 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+                                                                               \
+		_IBA_SET(field_struct, field_offset, mask, bits, lmsg, 0);     \
+		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
+		    0) {                                                       \
+			printk("Failed #5 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+		if (fn != 0) {                                                 \
+			printk("Failed #6 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+		_IBA_SET(field_struct, field_offset, mask, bits, lmsg, 1);     \
+		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
+		    1) {                                                       \
+			printk("Failed #7 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+		if (fn != 1) {                                                 \
+			printk("Failed #8 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+	})
+#define IBA_CHECK_GET(field, fn_name) _IBA_CHECK_GET(fn_name(lmsg), field)
+#define IBA_CHECK_GET_BE(field, fn_name) _IBA_CHECK_GET(be32_to_cpu(fn_name(lmsg)), field)
+
+/*
+ * Write the all ones value using the old setter and check that the new getter
+ * reads it back.
+ */
+#define _IBA_CHECK_SET(fn, field_struct, field_offset, mask, bits)             \
+	({                                                                     \
+		field_struct *lmsg = (field_struct *)msg;                      \
+		unsigned long long all_ones;                                   \
+		static_assert(sizeof(*lmsg) <= sizeof(msg));                   \
+                                                                               \
+		bitmap_zero(msg, nbits);                                       \
+		_IBA_SET(field_struct, field_offset, mask, bits, lmsg,         \
+			 cm_global_all_ones);                                  \
+		all_ones = (1ULL << bitmap_weight(msg, nbits)) - 1;            \
+		bitmap_zero(msg, nbits);                                       \
+		fn;                                                            \
+		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
+		    all_ones) {                                                \
+			printk("Failed #9 line=%u\n", __LINE__);               \
+			return;                                                \
+		}                                                              \
+		all_ones = 1;                                                  \
+		fn;                                                            \
+		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
+		    1) {                                                       \
+			printk("Failed #10 line=%u\n", __LINE__);              \
+			return;                                                \
+		}                                                              \
+	})
+
+#define IBA_CHECK_SET(field, fn_name) _IBA_CHECK_SET(fn_name(lmsg, all_ones), field)
+#define IBA_CHECK_SET_BE(field, fn_name)                                       \
+	_IBA_CHECK_SET(fn_name(lmsg, cpu_to_be32(all_ones)), field)
+
+static void self_test(void)
+{
+	unsigned long msg[256/4];
+	const unsigned int nbits = sizeof(msg) * 8;
+
+	printk("Running CM extractor self test\n");
+	IBA_CHECK_GET_BE(CM_REQ_LOCAL_QPN, cm_req_get_local_qpn);
+	IBA_CHECK_SET_BE(CM_REQ_LOCAL_QPN, cm_req_set_local_qpn);
+	IBA_CHECK_GET(CM_REQ_RESPONDER_RESOURCES, cm_req_get_resp_res);
+	IBA_CHECK_SET(CM_REQ_RESPONDER_RESOURCES, cm_req_set_resp_res);
+	IBA_CHECK_GET(CM_REQ_INITIATOR_DEPTH, cm_req_get_init_depth);
+	IBA_CHECK_SET(CM_REQ_INITIATOR_DEPTH, cm_req_set_init_depth);
+	IBA_CHECK_GET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, cm_req_get_remote_resp_timeout);
+	IBA_CHECK_SET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, cm_req_set_remote_resp_timeout);
+	IBA_CHECK_GET(CM_REQ_TRANSPORT_SERVICE_TYPE, cm_req_get_transport_type);
+	IBA_CHECK_SET(CM_REQ_TRANSPORT_SERVICE_TYPE, cm_req_set_transport_type);
+	IBA_CHECK_GET(CM_REQ_END_TO_END_FLOW_CONTROL, cm_req_get_flow_ctrl);
+	IBA_CHECK_SET(CM_REQ_END_TO_END_FLOW_CONTROL, cm_req_set_flow_ctrl);
+	IBA_CHECK_GET_BE(CM_REQ_STARTING_PSN, cm_req_get_starting_psn);
+	IBA_CHECK_SET_BE(CM_REQ_STARTING_PSN, cm_req_set_starting_psn);
+	IBA_CHECK_GET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, cm_req_get_local_resp_timeout);
+	IBA_CHECK_SET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, cm_req_set_local_resp_timeout);
+	IBA_CHECK_GET(CM_REQ_RETRY_COUNT, cm_req_get_retry_count);
+	IBA_CHECK_SET(CM_REQ_RETRY_COUNT, cm_req_set_retry_count);
+	IBA_CHECK_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, cm_req_get_path_mtu);
+	IBA_CHECK_SET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, cm_req_set_path_mtu);
+	IBA_CHECK_GET(CM_REQ_RNR_RETRY_COUNT, cm_req_get_rnr_retry_count);
+	IBA_CHECK_SET(CM_REQ_RNR_RETRY_COUNT, cm_req_set_rnr_retry_count);
+	IBA_CHECK_GET(CM_REQ_MAX_CM_RETRIES, cm_req_get_max_cm_retries);
+	IBA_CHECK_SET(CM_REQ_MAX_CM_RETRIES, cm_req_set_max_cm_retries);
+	IBA_CHECK_GET(CM_REQ_SRQ, cm_req_get_srq);
+	IBA_CHECK_SET(CM_REQ_SRQ, cm_req_set_srq);
+	IBA_CHECK_GET(CM_REQ_EXTENDED_TRANSPORT_TYPE, cm_req_get_transport_type_ex);
+	IBA_CHECK_SET(CM_REQ_EXTENDED_TRANSPORT_TYPE, cm_req_set_transport_type_ex);
+	IBA_CHECK_GET_BE(CM_REQ_PRIMARY_FLOW_LABEL, cm_req_get_primary_flow_label);
+	IBA_CHECK_SET_BE(CM_REQ_PRIMARY_FLOW_LABEL, cm_req_set_primary_flow_label);
+	IBA_CHECK_GET(CM_REQ_PRIMARY_PACKET_RATE, cm_req_get_primary_packet_rate);
+	IBA_CHECK_SET(CM_REQ_PRIMARY_PACKET_RATE, cm_req_set_primary_packet_rate);
+	IBA_CHECK_GET(CM_REQ_PRIMARY_SL, cm_req_get_primary_sl);
+	IBA_CHECK_SET(CM_REQ_PRIMARY_SL, cm_req_set_primary_sl);
+	IBA_CHECK_GET(CM_REQ_PRIMARY_SUBNET_LOCAL, cm_req_get_primary_subnet_local);
+	IBA_CHECK_SET(CM_REQ_PRIMARY_SUBNET_LOCAL, cm_req_set_primary_subnet_local);
+	IBA_CHECK_GET(CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT, cm_req_get_primary_local_ack_timeout);
+	IBA_CHECK_SET(CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT, cm_req_set_primary_local_ack_timeout);
+	IBA_CHECK_GET_BE(CM_REQ_ALTERNATE_FLOW_LABEL, cm_req_get_alt_flow_label);
+	IBA_CHECK_SET_BE(CM_REQ_ALTERNATE_FLOW_LABEL, cm_req_set_alt_flow_label);
+	IBA_CHECK_GET(CM_REQ_ALTERNATE_PACKET_RATE, cm_req_get_alt_packet_rate);
+	IBA_CHECK_SET(CM_REQ_ALTERNATE_PACKET_RATE, cm_req_set_alt_packet_rate);
+	IBA_CHECK_GET(CM_REQ_ALTERNATE_SL, cm_req_get_alt_sl);
+	IBA_CHECK_SET(CM_REQ_ALTERNATE_SL, cm_req_set_alt_sl);
+	IBA_CHECK_GET(CM_REQ_ALTERNATE_SUBNET_LOCAL, cm_req_get_alt_subnet_local);
+	IBA_CHECK_SET(CM_REQ_ALTERNATE_SUBNET_LOCAL, cm_req_set_alt_subnet_local);
+	IBA_CHECK_GET(CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT, cm_req_get_alt_local_ack_timeout);
+	IBA_CHECK_SET(CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT, cm_req_set_alt_local_ack_timeout);
+	IBA_CHECK_GET(CM_MRA_MESSAGE_MRAED, cm_mra_get_msg_mraed);
+	IBA_CHECK_SET(CM_MRA_MESSAGE_MRAED, cm_mra_set_msg_mraed);
+	IBA_CHECK_GET(CM_MRA_SERVICE_TIMEOUT, cm_mra_get_service_timeout);
+	IBA_CHECK_SET(CM_MRA_SERVICE_TIMEOUT, cm_mra_set_service_timeout);
+	IBA_CHECK_GET(CM_REJ_MESSAGE_REJECTED, cm_rej_get_msg_rejected);
+	IBA_CHECK_SET(CM_REJ_MESSAGE_REJECTED, cm_rej_set_msg_rejected);
+	IBA_CHECK_GET(CM_REJ_REJECTED_INFO_LENGTH, cm_rej_get_reject_info_len);
+	IBA_CHECK_SET(CM_REJ_REJECTED_INFO_LENGTH, cm_rej_set_reject_info_len);
+	IBA_CHECK_GET_BE(CM_REP_LOCAL_QPN, cm_rep_get_local_qpn);
+	IBA_CHECK_SET_BE(CM_REP_LOCAL_QPN, cm_rep_set_local_qpn);
+	IBA_CHECK_GET_BE(CM_REP_LOCAL_EE_CONTEXT_NUMBER, cm_rep_get_local_eecn);
+	IBA_CHECK_SET_BE(CM_REP_LOCAL_EE_CONTEXT_NUMBER, cm_rep_set_local_eecn);
+	IBA_CHECK_GET_BE(CM_REP_STARTING_PSN, cm_rep_get_starting_psn);
+	IBA_CHECK_SET_BE(CM_REP_STARTING_PSN, cm_rep_set_starting_psn);
+	IBA_CHECK_GET(CM_REP_TARGET_ACK_DELAY, cm_rep_get_target_ack_delay);
+	IBA_CHECK_SET(CM_REP_TARGET_ACK_DELAY, cm_rep_set_target_ack_delay);
+	IBA_CHECK_GET(CM_REP_FAILOVER_ACCEPTED, cm_rep_get_failover);
+	IBA_CHECK_SET(CM_REP_FAILOVER_ACCEPTED, cm_rep_set_failover);
+	IBA_CHECK_GET(CM_REP_END_TO_END_FLOW_CONTROL, cm_rep_get_flow_ctrl);
+	IBA_CHECK_SET(CM_REP_END_TO_END_FLOW_CONTROL, cm_rep_set_flow_ctrl);
+	IBA_CHECK_GET(CM_REP_RNR_RETRY_COUNT, cm_rep_get_rnr_retry_count);
+	IBA_CHECK_SET(CM_REP_RNR_RETRY_COUNT, cm_rep_set_rnr_retry_count);
+	IBA_CHECK_GET(CM_REP_SRQ, cm_rep_get_srq);
+	IBA_CHECK_SET(CM_REP_SRQ, cm_rep_set_srq);
+	IBA_CHECK_GET_BE(CM_DREQ_REMOTE_QPN_EECN, cm_dreq_get_remote_qpn);
+	IBA_CHECK_SET_BE(CM_DREQ_REMOTE_QPN_EECN, cm_dreq_set_remote_qpn);
+	IBA_CHECK_GET_BE(CM_LAP_ALTERNATE_FLOW_LABEL, cm_lap_get_flow_label);
+	IBA_CHECK_GET(CM_LAP_ALTERNATE_TRAFFIC_CLASS, cm_lap_get_traffic_class);
+	IBA_CHECK_GET(CM_LAP_ALTERNATE_PACKET_RATE, cm_lap_get_packet_rate);
+	IBA_CHECK_GET(CM_LAP_ALTERNATE_SL, cm_lap_get_sl);
+	IBA_CHECK_GET(CM_LAP_ALTERNATE_LOCAL_ACK_TIMEOUT, cm_lap_get_local_ack_timeout);
+	IBA_CHECK_GET_BE(CM_SIDR_REP_QPN, cm_sidr_rep_get_qpn);
+	IBA_CHECK_SET_BE(CM_SIDR_REP_QPN, cm_sidr_rep_set_qpn);
+	printk("Success!\n");
+}
+
 static int __init ib_cm_init(void)
 {
 	int ret;
 
+	self_test();
+
 	INIT_LIST_HEAD(&cm.device_list);
 	rwlock_init(&cm.device_lock);
 	spin_lock_init(&cm.lock);
diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h
index 9af9a3212074..bf62461d801f 100644
--- a/drivers/infiniband/core/cm_msgs.h
+++ b/drivers/infiniband/core/cm_msgs.h
@@ -8,6 +8,7 @@
 #ifndef CM_MSGS_H
 #define CM_MSGS_H
 
+#include <rdma/ibta_vol1_c12.h>
 #include <rdma/ib_mad.h>
 #include <rdma/ib_cm.h>
 
diff --git a/include/rdma/iba.h b/include/rdma/iba.h
new file mode 100644
index 000000000000..6a1115b02a0d
--- /dev/null
+++ b/include/rdma/iba.h
@@ -0,0 +1,146 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/*
+ * Copyright (c) 2020, Mellanox Technologies inc.  All rights reserved.
+ */
+#ifndef _IBA_DEFS_H_
+#define _IBA_DEFS_H_
+
+#include <linux/kernel.h>
+#include <linux/bitfield.h>
+#include <asm/unaligned.h>
+
+static inline u32 _iba_get8(const u8 *ptr)
+{
+	return *ptr;
+}
+
+static inline void _iba_set8(u8 *ptr, u32 mask, u32 prep_value)
+{
+	*ptr = (*ptr & ~mask) | prep_value;
+}
+
+static inline u16 _iba_get16(const __be16 *ptr)
+{
+	return be16_to_cpu(*ptr);
+}
+
+static inline void _iba_set16(__be16 *ptr, u16 mask, u16 prep_value)
+{
+	*ptr = cpu_to_be16((be16_to_cpu(*ptr) & ~mask) | prep_value);
+}
+
+static inline u32 _iba_get32(const __be32 *ptr)
+{
+	return be32_to_cpu(*ptr);
+}
+
+static inline void _iba_set32(__be32 *ptr, u32 mask, u32 prep_value)
+{
+	*ptr = cpu_to_be32((be32_to_cpu(*ptr) & ~mask) | prep_value);
+}
+
+static inline u64 _iba_get64(const __be64 *ptr)
+{
+	/*
+	 * The mads are constructed so that 32 bit and smaller are naturally
+	 * aligned, everything larger has a max alignment of 4 bytes.
+	 */
+	return be64_to_cpu(get_unaligned(ptr));
+}
+
+static inline void _iba_set64(__be64 *ptr, u64 mask, u64 prep_value)
+{
+	put_unaligned(cpu_to_be64((_iba_get64(ptr) & ~mask) | prep_value), ptr);
+}
+
+#define _IBA_SET(field_struct, field_offset, field_mask, num_bits, ptr, value) \
+	({                                                                     \
+		field_struct *_ptr = ptr;                                      \
+		_iba_set##num_bits((void *)_ptr + (field_offset), field_mask,  \
+				   FIELD_PREP(field_mask, value));             \
+	})
+#define IBA_SET(field, ptr, value) _IBA_SET(field, ptr, value)
+
+#define _IBA_GET_MEM_PTR(field_struct, field_offset, type, num_bits, ptr)      \
+	({                                                                     \
+		field_struct *_ptr = ptr;                                      \
+		(type *)((void *)_ptr + (field_offset));                       \
+	})
+#define IBA_GET_MEM_PTR(field, ptr) _IBA_GET_MEM_PTR(field, ptr)
+
+/* FIXME: A set should always set the entire field, meaning we should zero the trailing bytes */
+#define _IBA_SET_MEM(field_struct, field_offset, type, num_bits, ptr, in,      \
+		     bytes)                                                    \
+	({                                                                     \
+		const type *_in_ptr = in;                                      \
+		WARN_ON(bytes * 8 > num_bits);                                 \
+		if (in && bytes)                                               \
+			memcpy(_IBA_GET_MEM_PTR(field_struct, field_offset,    \
+						type, num_bits, ptr),          \
+			       _in_ptr, bytes);                                \
+	})
+#define IBA_SET_MEM(field, ptr, in, bytes) _IBA_SET_MEM(field, ptr, in, bytes)
+
+#define _IBA_GET(field_struct, field_offset, field_mask, num_bits, ptr)        \
+	({                                                                     \
+		const field_struct *_ptr = ptr;                                \
+		(u##num_bits) FIELD_GET(                                       \
+			field_mask, _iba_get##num_bits((const void *)_ptr +    \
+						       (field_offset)));       \
+	})
+#define IBA_GET(field, ptr) _IBA_GET(field, ptr)
+
+#define _IBA_GET_MEM(field_struct, field_offset, type, num_bits, ptr, out,     \
+		     bytes)                                                    \
+	({                                                                     \
+		type *_out_ptr = out;                                          \
+		WARN_ON(bytes * 8 > num_bits);                                 \
+		if (out && bytes)                                              \
+			memcpy(_out_ptr,                                       \
+			       _IBA_GET_MEM_PTR(field_struct, field_offset,    \
+						type, num_bits, ptr),          \
+			       bytes);                                         \
+	})
+#define IBA_GET_MEM(field, ptr, out, bytes) _IBA_GET_MEM(field, ptr, out, bytes)
+
+/*
+ * The generated list becomes the parameters to the macros, the order is:
+ *  - struct this applies to
+ *  - starting offset of the max
+ *  - GENMASK or GENMASK_ULL in CPU order
+ *  - The width of data the mask operations should work on, in bits
+ */
+
+/*
+ * Extraction using a tabular description like table 106. bit_offset is from
+ * the Byte[Bit] notation.
+ */
+#define IBA_FIELD_BLOC(field_struct, byte_offset, bit_offset, num_bits)        \
+	field_struct, byte_offset,                                             \
+		GENMASK(7 - (bit_offset), 7 - (bit_offset) - (num_bits - 1)),  \
+		8
+#define IBA_FIELD8_LOC(field_struct, byte_offset, num_bits)                    \
+	IBA_FIELD_BLOC(field_struct, byte_offset, 0, num_bits)
+
+#define IBA_FIELD16_LOC(field_struct, byte_offset, num_bits)                   \
+	field_struct, (byte_offset)&0xFFFE,                                    \
+		GENMASK(15 - (((byte_offset) % 2) * 8),                        \
+			15 - (((byte_offset) % 2) * 8) - (num_bits - 1)),      \
+		16
+
+#define IBA_FIELD32_LOC(field_struct, byte_offset, num_bits)                   \
+	field_struct, (byte_offset)&0xFFFC,                                    \
+		GENMASK(31 - (((byte_offset) % 4) * 8),                        \
+			31 - (((byte_offset) % 4) * 8) - (num_bits - 1)),      \
+		32
+
+#define IBA_FIELD64_LOC(field_struct, byte_offset)                             \
+	field_struct, byte_offset, GENMASK_ULL(63, 0), 64
+/*
+ * In IBTA spec, everything that is more than 64bits is multiple
+ * of bytes without leftover bits.
+ */
+#define IBA_FIELD_MLOC(field_struct, byte_offset, num_bits, type)              \
+	field_struct, byte_offset, type, num_bits
+
+#endif /* _IBA_DEFS_H_ */
diff --git a/include/rdma/ibta_vol1_c12.h b/include/rdma/ibta_vol1_c12.h
new file mode 100644
index 000000000000..916db5c27dc5
--- /dev/null
+++ b/include/rdma/ibta_vol1_c12.h
@@ -0,0 +1,208 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/*
+ * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved.
+ *
+ * This file is IBTA volume 1, chapter 12 declarations:
+ *  CHAPTER 12: COMMUNICATION MANAGEMENT
+ */
+#ifndef _IBTA_VOL1_C12_H_
+#define _IBTA_VOL1_C12_H_
+
+#include <rdma/iba.h>
+
+#define CM_FIELD_BLOC(field_struct, byte_offset, bits_offset, width)           \
+	IBA_FIELD_BLOC(field_struct,                                           \
+		       (byte_offset + sizeof(struct ib_mad_hdr)), bits_offset, \
+		       width)
+#define CM_FIELD8_LOC(field_struct, byte_offset, width)                        \
+	IBA_FIELD8_LOC(field_struct,                                           \
+		       (byte_offset + sizeof(struct ib_mad_hdr)), width)
+#define CM_FIELD16_LOC(field_struct, byte_offset, width)                       \
+	IBA_FIELD16_LOC(field_struct,                                          \
+			(byte_offset + sizeof(struct ib_mad_hdr)), width)
+#define CM_FIELD32_LOC(field_struct, byte_offset, width)                       \
+	IBA_FIELD32_LOC(field_struct,                                          \
+			(byte_offset + sizeof(struct ib_mad_hdr)), width)
+#define CM_FIELD64_LOC(field_struct, byte_offset)                              \
+	IBA_FIELD64_LOC(field_struct, (byte_offset + sizeof(struct ib_mad_hdr)))
+#define CM_FIELD_MLOC(field_struct, byte_offset, width, type)                  \
+	IBA_FIELD_MLOC(field_struct,                                           \
+		       (byte_offset + sizeof(struct ib_mad_hdr)), width, type)
+#define CM_STRUCT(field_struct, total_len)                                     \
+	static_assert((total_len) % 32 == 0);
+
+/* Table 106 REQ Message Contents */
+#define CM_REQ_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_req_msg, 0, 32)
+#define CM_REQ_SERVICE_ID CM_FIELD64_LOC(struct cm_req_msg, 8)
+#define CM_REQ_LOCAL_CA_GUID CM_FIELD64_LOC(struct cm_req_msg, 16)
+#define CM_REQ_LOCAL_Q_KEY CM_FIELD32_LOC(struct cm_req_msg, 28, 32)
+#define CM_REQ_LOCAL_QPN CM_FIELD32_LOC(struct cm_req_msg, 32, 24)
+#define CM_REQ_RESPONDER_RESOURCES CM_FIELD8_LOC(struct cm_req_msg, 35, 8)
+#define CM_REQ_LOCAL_EECN CM_FIELD32_LOC(struct cm_req_msg, 36, 24)
+#define CM_REQ_INITIATOR_DEPTH CM_FIELD8_LOC(struct cm_req_msg, 39, 8)
+#define CM_REQ_REMOTE_EECN CM_FIELD32_LOC(struct cm_req_msg, 40, 24)
+#define CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT                                      \
+	CM_FIELD8_LOC(struct cm_req_msg, 43, 5)
+#define CM_REQ_TRANSPORT_SERVICE_TYPE CM_FIELD_BLOC(struct cm_req_msg, 43, 5, 2)
+#define CM_REQ_END_TO_END_FLOW_CONTROL                                         \
+	CM_FIELD_BLOC(struct cm_req_msg, 43, 7, 1)
+#define CM_REQ_STARTING_PSN CM_FIELD32_LOC(struct cm_req_msg, 44, 24)
+#define CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT CM_FIELD8_LOC(struct cm_req_msg, 47, 5)
+#define CM_REQ_RETRY_COUNT CM_FIELD_BLOC(struct cm_req_msg, 47, 5, 3)
+#define CM_REQ_PARTITION_KEY CM_FIELD16_LOC(struct cm_req_msg, 48, 16)
+#define CM_REQ_PATH_PACKET_PAYLOAD_MTU CM_FIELD8_LOC(struct cm_req_msg, 50, 4)
+#define CM_REQ_RDC_EXISTS CM_FIELD_BLOC(struct cm_req_msg, 50, 4, 1)
+#define CM_REQ_RNR_RETRY_COUNT CM_FIELD_BLOC(struct cm_req_msg, 50, 5, 3)
+#define CM_REQ_MAX_CM_RETRIES CM_FIELD8_LOC(struct cm_req_msg, 51, 4)
+#define CM_REQ_SRQ CM_FIELD_BLOC(struct cm_req_msg, 51, 4, 1)
+#define CM_REQ_EXTENDED_TRANSPORT_TYPE                                         \
+	CM_FIELD_BLOC(struct cm_req_msg, 51, 5, 3)
+#define CM_REQ_PRIMARY_LOCAL_PORT_LID CM_FIELD16_LOC(struct cm_req_msg, 52, 16)
+#define CM_REQ_PRIMARY_REMOTE_PORT_LID CM_FIELD16_LOC(struct cm_req_msg, 54, 16)
+#define CM_REQ_PRIMARY_LOCAL_PORT_GID                                          \
+	CM_FIELD_MLOC(struct cm_req_msg, 56, 128, union ib_gid)
+#define CM_REQ_PRIMARY_REMOTE_PORT_GID                                         \
+	CM_FIELD_MLOC(struct cm_req_msg, 72, 128, union ib_gid)
+#define CM_REQ_PRIMARY_FLOW_LABEL CM_FIELD32_LOC(struct cm_req_msg, 88, 20)
+#define CM_REQ_PRIMARY_PACKET_RATE CM_FIELD_BLOC(struct cm_req_msg, 91, 2, 6)
+#define CM_REQ_PRIMARY_TRAFFIC_CLASS CM_FIELD8_LOC(struct cm_req_msg, 92, 8)
+#define CM_REQ_PRIMARY_HOP_LIMIT CM_FIELD8_LOC(struct cm_req_msg, 93, 8)
+#define CM_REQ_PRIMARY_SL CM_FIELD8_LOC(struct cm_req_msg, 94, 4)
+#define CM_REQ_PRIMARY_SUBNET_LOCAL CM_FIELD_BLOC(struct cm_req_msg, 94, 4, 1)
+#define CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT CM_FIELD8_LOC(struct cm_req_msg, 95, 5)
+#define CM_REQ_ALTERNATE_LOCAL_PORT_LID                                        \
+	CM_FIELD16_LOC(struct cm_req_msg, 96, 16)
+#define CM_REQ_ALTERNATE_REMOTE_PORT_LID                                       \
+	CM_FIELD16_LOC(struct cm_req_msg, 98, 16)
+#define CM_REQ_ALTERNATE_LOCAL_PORT_GID                                        \
+	CM_FIELD_MLOC(struct cm_req_msg, 100, 128, union ib_gid)
+#define CM_REQ_ALTERNATE_REMOTE_PORT_GID                                       \
+	CM_FIELD_MLOC(struct cm_req_msg, 116, 128, union ib_gid)
+#define CM_REQ_ALTERNATE_FLOW_LABEL CM_FIELD32_LOC(struct cm_req_msg, 132, 20)
+#define CM_REQ_ALTERNATE_PACKET_RATE CM_FIELD_BLOC(struct cm_req_msg, 135, 2, 6)
+#define CM_REQ_ALTERNATE_TRAFFIC_CLASS CM_FIELD8_LOC(struct cm_req_msg, 136, 8)
+#define CM_REQ_ALTERNATE_HOP_LIMIT CM_FIELD8_LOC(struct cm_req_msg, 137, 8)
+#define CM_REQ_ALTERNATE_SL CM_FIELD8_LOC(struct cm_req_msg, 138, 4)
+#define CM_REQ_ALTERNATE_SUBNET_LOCAL                                          \
+	CM_FIELD_BLOC(struct cm_req_msg, 138, 4, 1)
+#define CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT                                     \
+	CM_FIELD8_LOC(struct cm_req_msg, 139, 5)
+#define CM_REQ_SAP_SUPPORTED CM_FIELD_BLOC(struct cm_req_msg, 139, 5, 1)
+#define CM_REQ_PRIVATE_DATA CM_FIELD_MLOC(struct cm_req_msg, 140, 736, void)
+CM_STRUCT(struct cm_req_msg, 140 * 8 + 736);
+
+/* Table 107 MRA Message Contents */
+#define CM_MRA_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_mra_msg, 0, 32)
+#define CM_MRA_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_mra_msg, 4, 32)
+#define CM_MRA_MESSAGE_MRAED CM_FIELD8_LOC(struct cm_mra_msg, 8, 2)
+#define CM_MRA_SERVICE_TIMEOUT CM_FIELD8_LOC(struct cm_mra_msg, 9, 5)
+#define CM_MRA_PRIVATE_DATA CM_FIELD_MLOC(struct cm_mra_msg, 10, 1776, void)
+CM_STRUCT(struct cm_mra_msg, 10 * 8 + 1776);
+
+/* Table 108 REJ Message Contents */
+#define CM_REJ_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_rej_msg, 0, 32)
+#define CM_REJ_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_rej_msg, 4, 32)
+#define CM_REJ_MESSAGE_REJECTED CM_FIELD8_LOC(struct cm_rej_msg, 8, 2)
+#define CM_REJ_REJECTED_INFO_LENGTH CM_FIELD8_LOC(struct cm_rej_msg, 9, 7)
+#define CM_REJ_REASON CM_FIELD16_LOC(struct cm_rej_msg, 10, 16)
+#define CM_REJ_ARI CM_FIELD_MLOC(struct cm_rej_msg, 12, 576, void)
+#define CM_REJ_PRIVATE_DATA CM_FIELD_MLOC(struct cm_rej_msg, 84, 1184, void)
+CM_STRUCT(struct cm_rej_msg, 84 * 8 + 1184);
+
+/* Table 110 REP Message Contents */
+#define CM_REP_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_rep_msg, 0, 32)
+#define CM_REP_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_rep_msg, 4, 32)
+#define CM_REP_LOCAL_Q_KEY CM_FIELD32_LOC(struct cm_rep_msg, 8, 32)
+#define CM_REP_LOCAL_QPN CM_FIELD32_LOC(struct cm_rep_msg, 12, 24)
+#define CM_REP_LOCAL_EE_CONTEXT_NUMBER CM_FIELD32_LOC(struct cm_rep_msg, 16, 24)
+#define CM_REP_STARTING_PSN CM_FIELD32_LOC(struct cm_rep_msg, 20, 24)
+#define CM_REP_RESPONDER_RESOURCES CM_FIELD8_LOC(struct cm_rep_msg, 24, 8)
+#define CM_REP_INITIATOR_DEPTH CM_FIELD8_LOC(struct cm_rep_msg, 25, 8)
+#define CM_REP_TARGET_ACK_DELAY CM_FIELD8_LOC(struct cm_rep_msg, 26, 5)
+#define CM_REP_FAILOVER_ACCEPTED CM_FIELD_BLOC(struct cm_rep_msg, 26, 5, 2)
+#define CM_REP_END_TO_END_FLOW_CONTROL                                         \
+	CM_FIELD_BLOC(struct cm_rep_msg, 26, 7, 1)
+#define CM_REP_RNR_RETRY_COUNT CM_FIELD8_LOC(struct cm_rep_msg, 27, 3)
+#define CM_REP_SRQ CM_FIELD_BLOC(struct cm_rep_msg, 27, 3, 1)
+#define CM_REP_LOCAL_CA_GUID CM_FIELD64_LOC(struct cm_rep_msg, 28)
+#define CM_REP_PRIVATE_DATA CM_FIELD_MLOC(struct cm_rep_msg, 36, 1568, void)
+CM_STRUCT(struct cm_rep_msg, 36 * 8 + 1568);
+
+/* Table 111 RTU Message Contents */
+#define CM_RTU_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_rtu_msg, 0, 32)
+#define CM_RTU_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_rtu_msg, 4, 32)
+#define CM_RTU_PRIVATE_DATA CM_FIELD_MLOC(struct cm_rtu_msg, 8, 1792, void)
+CM_STRUCT(struct cm_rtu_msg, 8 * 8 + 1792);
+
+/* Table 112 DREQ Message Contents */
+#define CM_DREQ_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_dreq_msg, 0, 32)
+#define CM_DREQ_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_dreq_msg, 4, 32)
+#define CM_DREQ_REMOTE_QPN_EECN CM_FIELD32_LOC(struct cm_dreq_msg, 8, 24)
+#define CM_DREQ_PRIVATE_DATA CM_FIELD_MLOC(struct cm_dreq_msg, 12, 1760, void)
+CM_STRUCT(struct cm_dreq_msg, 12 * 8 + 1760);
+
+/* Table 113 DREP Message Contents */
+#define CM_DREP_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_drep_msg, 0, 32)
+#define CM_DREP_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_drep_msg, 4, 32)
+#define CM_DREP_PRIVATE_DATA CM_FIELD_MLOC(struct cm_drep_msg, 8, 1792, void)
+CM_STRUCT(struct cm_drep_msg, 8 * 8 + 1792);
+
+/* Table 115 LAP Message Contents */
+#define CM_LAP_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_lap_msg, 0, 32)
+#define CM_LAP_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_lap_msg, 4, 32)
+#define CM_LAP_REMOTE_QPN_EECN CM_FIELD32_LOC(struct cm_lap_msg, 12, 24)
+#define CM_LAP_REMOTE_CM_RESPONSE_TIMEOUT                                      \
+	CM_FIELD8_LOC(struct cm_lap_msg, 15, 5)
+#define CM_LAP_ALTERNATE_LOCAL_PORT_LID                                        \
+	CM_FIELD16_LOC(struct cm_lap_msg, 20, 16)
+#define CM_LAP_ALTERNATE_REMOTE_PORT_LID                                       \
+	CM_FIELD16_LOC(struct cm_lap_msg, 22, 16)
+#define CM_LAP_ALTERNATE_LOCAL_PORT_GID                                        \
+	CM_FIELD_MLOC(struct cm_lap_msg, 24, 128, union ib_gid)
+#define CM_LAP_ALTERNATE_REMOTE_PORT_GID                                       \
+	CM_FIELD_MLOC(struct cm_lap_msg, 40, 128, union ib_gid)
+#define CM_LAP_ALTERNATE_FLOW_LABEL CM_FIELD32_LOC(struct cm_lap_msg, 56, 20)
+#define CM_LAP_ALTERNATE_TRAFFIC_CLASS CM_FIELD8_LOC(struct cm_lap_msg, 59, 8)
+#define CM_LAP_ALTERNATE_HOP_LIMIT CM_FIELD8_LOC(struct cm_lap_msg, 60, 8)
+#define CM_LAP_ALTERNATE_PACKET_RATE CM_FIELD_BLOC(struct cm_lap_msg, 61, 2, 6)
+#define CM_LAP_ALTERNATE_SL CM_FIELD8_LOC(struct cm_lap_msg, 62, 4)
+#define CM_LAP_ALTERNATE_SUBNET_LOCAL CM_FIELD_BLOC(struct cm_lap_msg, 62, 4, 1)
+#define CM_LAP_ALTERNATE_LOCAL_ACK_TIMEOUT                                     \
+	CM_FIELD8_LOC(struct cm_lap_msg, 63, 5)
+#define CM_LAP_PRIVATE_DATA CM_FIELD_MLOC(struct cm_lap_msg, 64, 1344, void)
+CM_STRUCT(struct cm_lap_msg, 64 * 8 + 1344);
+
+/* Table 116 APR Message Contents */
+#define CM_APR_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_apr_msg, 0, 32)
+#define CM_APR_REMOTE_COMM_ID CM_FIELD32_LOC(struct cm_apr_msg, 4, 32)
+#define CM_APR_ADDITIONAL_INFORMATION_LENGTH                                   \
+	CM_FIELD8_LOC(struct cm_apr_msg, 8, 8)
+#define CM_APR_AR_STATUS CM_FIELD8_LOC(struct cm_apr_msg, 9, 8)
+#define CM_APR_ADDITIONAL_INFORMATION                                          \
+	CM_FIELD_MLOC(struct cm_apr_msg, 12, 576, void)
+#define CM_APR_PRIVATE_DATA CM_FIELD_MLOC(struct cm_apr_msg, 84, 1184, void)
+CM_STRUCT(struct cm_apr_msg, 84 * 8 + 1184);
+
+/* Table 119 SIDR_REQ Message Contents */
+#define CM_SIDR_REQ_REQUESTID CM_FIELD32_LOC(struct cm_sidr_req_msg, 0, 32)
+#define CM_SIDR_REQ_PARTITION_KEY CM_FIELD16_LOC(struct cm_sidr_req_msg, 4, 16)
+#define CM_SIDR_REQ_SERVICEID CM_FIELD64_LOC(struct cm_sidr_req_msg, 8)
+#define CM_SIDR_REQ_PRIVATE_DATA                                               \
+	CM_FIELD_MLOC(struct cm_sidr_req_msg, 16, 1728, void)
+CM_STRUCT(struct cm_sidr_req_msg, 16 * 8 + 1728);
+
+/* Table 120 SIDR_REP Message Contents */
+#define CM_SIDR_REP_REQUESTID CM_FIELD32_LOC(struct cm_sidr_rep_msg, 0, 32)
+#define CM_SIDR_REP_STATUS CM_FIELD8_LOC(struct cm_sidr_rep_msg, 4, 8)
+#define CM_SIDR_REP_ADDITIONAL_INFORMATION_LENGTH                              \
+	CM_FIELD8_LOC(struct cm_sidr_rep_msg, 5, 8)
+#define CM_SIDR_REP_QPN CM_FIELD32_LOC(struct cm_sidr_rep_msg, 8, 24)
+#define CM_SIDR_REP_SERVICEID CM_FIELD64_LOC(struct cm_sidr_rep_msg, 12)
+#define CM_SIDR_REP_Q_KEY CM_FIELD32_LOC(struct cm_sidr_rep_msg, 20, 32)
+#define CM_SIDR_REP_ADDITIONAL_INFORMATION                                     \
+	CM_FIELD_MLOC(struct cm_sidr_rep_msg, 24, 576, void)
+#define CM_SIDR_REP_PRIVATE_DATA                                               \
+	CM_FIELD_MLOC(struct cm_sidr_rep_msg, 96, 1088, void)
+CM_STRUCT(struct cm_sidr_rep_msg, 96 * 8 + 1088);
+
+#endif /* _IBTA_VOL1_C12_H_ */
-- 
cgit v1.2.3-59-g8ed1b


From b6bbee6889300372e8d84373de0afbd980155f90 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Thu, 16 Jan 2020 13:00:33 -0400
Subject: RDMA/cm: Use IBA functions for simple get/set acessors

Use a Coccinelle spatch to replace CM helper functions with IBA_GET/SET
versions. Applied with

$ spatch --sp-file edits.sp --in-place drivers/infiniband/core/cm.c

The spatch file was generated using the template pattern:

@@
expression val;
{struct} *msg;
@@
- {old_setter}
+ IBA_SET({new_name}, msg, val)
@@
{struct} *msg;
@@
- {old_getter}
+ IBA_GET({new_name}, msg)

Iterated for every IBA_CHECK_GET()/IBA_CHECK_GET() pairing. Touched up
with clang-format after.

Link: https://lore.kernel.org/r/20200116170037.30109-4-jgg@ziepe.ca
Tested-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c      | 242 ++++++++++---------------
 drivers/infiniband/core/cm_msgs.h | 371 +-------------------------------------
 2 files changed, 104 insertions(+), 509 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 7f609979e4de..a1a968160132 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -1255,23 +1255,26 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 	req_msg->service_id = param->service_id;
 	req_msg->local_ca_guid = cm_id_priv->id.device->node_guid;
 	cm_req_set_local_qpn(req_msg, cpu_to_be32(param->qp_num));
-	cm_req_set_init_depth(req_msg, param->initiator_depth);
-	cm_req_set_remote_resp_timeout(req_msg,
-				       param->remote_cm_response_timeout);
+	IBA_SET(CM_REQ_INITIATOR_DEPTH, req_msg, param->initiator_depth);
+	IBA_SET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, req_msg,
+		param->remote_cm_response_timeout);
 	cm_req_set_qp_type(req_msg, param->qp_type);
-	cm_req_set_flow_ctrl(req_msg, param->flow_control);
+	IBA_SET(CM_REQ_END_TO_END_FLOW_CONTROL, req_msg, param->flow_control);
 	cm_req_set_starting_psn(req_msg, cpu_to_be32(param->starting_psn));
-	cm_req_set_local_resp_timeout(req_msg,
-				      param->local_cm_response_timeout);
+	IBA_SET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, req_msg,
+		param->local_cm_response_timeout);
 	req_msg->pkey = param->primary_path->pkey;
-	cm_req_set_path_mtu(req_msg, param->primary_path->mtu);
-	cm_req_set_max_cm_retries(req_msg, param->max_cm_retries);
+	IBA_SET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg,
+		param->primary_path->mtu);
+	IBA_SET(CM_REQ_MAX_CM_RETRIES, req_msg, param->max_cm_retries);
 
 	if (param->qp_type != IB_QPT_XRC_INI) {
-		cm_req_set_resp_res(req_msg, param->responder_resources);
-		cm_req_set_retry_count(req_msg, param->retry_count);
-		cm_req_set_rnr_retry_count(req_msg, param->rnr_retry_count);
-		cm_req_set_srq(req_msg, param->srq);
+		IBA_SET(CM_REQ_RESPONDER_RESOURCES, req_msg,
+			param->responder_resources);
+		IBA_SET(CM_REQ_RETRY_COUNT, req_msg, param->retry_count);
+		IBA_SET(CM_REQ_RNR_RETRY_COUNT, req_msg,
+			param->rnr_retry_count);
+		IBA_SET(CM_REQ_SRQ, req_msg, param->srq);
 	}
 
 	req_msg->primary_local_gid = pri_path->sgid;
@@ -1293,12 +1296,13 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 		req_msg->primary_remote_lid = IB_LID_PERMISSIVE;
 	}
 	cm_req_set_primary_flow_label(req_msg, pri_path->flow_label);
-	cm_req_set_primary_packet_rate(req_msg, pri_path->rate);
+	IBA_SET(CM_REQ_PRIMARY_PACKET_RATE, req_msg, pri_path->rate);
 	req_msg->primary_traffic_class = pri_path->traffic_class;
 	req_msg->primary_hop_limit = pri_path->hop_limit;
-	cm_req_set_primary_sl(req_msg, pri_path->sl);
-	cm_req_set_primary_subnet_local(req_msg, (pri_path->hop_limit <= 1));
-	cm_req_set_primary_local_ack_timeout(req_msg,
+	IBA_SET(CM_REQ_PRIMARY_SL, req_msg, pri_path->sl);
+	IBA_SET(CM_REQ_PRIMARY_SUBNET_LOCAL, req_msg,
+		(pri_path->hop_limit <= 1));
+	IBA_SET(CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT, req_msg,
 		cm_ack_timeout(cm_id_priv->av.port->cm_dev->ack_delay,
 			       pri_path->packet_life_time));
 
@@ -1328,12 +1332,13 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 		}
 		cm_req_set_alt_flow_label(req_msg,
 					  alt_path->flow_label);
-		cm_req_set_alt_packet_rate(req_msg, alt_path->rate);
+		IBA_SET(CM_REQ_ALTERNATE_PACKET_RATE, req_msg, alt_path->rate);
 		req_msg->alt_traffic_class = alt_path->traffic_class;
 		req_msg->alt_hop_limit = alt_path->hop_limit;
-		cm_req_set_alt_sl(req_msg, alt_path->sl);
-		cm_req_set_alt_subnet_local(req_msg, (alt_path->hop_limit <= 1));
-		cm_req_set_alt_local_ack_timeout(req_msg,
+		IBA_SET(CM_REQ_ALTERNATE_SL, req_msg, alt_path->sl);
+		IBA_SET(CM_REQ_ALTERNATE_SUBNET_LOCAL, req_msg,
+			(alt_path->hop_limit <= 1));
+		IBA_SET(CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT, req_msg,
 			cm_ack_timeout(cm_id_priv->av.port->cm_dev->ack_delay,
 				       alt_path->packet_life_time));
 	}
@@ -1473,11 +1478,11 @@ static int cm_issue_rej(struct cm_port *port,
 	cm_format_mad_hdr(&rej_msg->hdr, CM_REJ_ATTR_ID, rcv_msg->hdr.tid);
 	rej_msg->remote_comm_id = rcv_msg->local_comm_id;
 	rej_msg->local_comm_id = rcv_msg->remote_comm_id;
-	cm_rej_set_msg_rejected(rej_msg, msg_rejected);
+	IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, msg_rejected);
 	rej_msg->reason = cpu_to_be16(reason);
 
 	if (ari && ari_length) {
-		cm_rej_set_reject_info_len(rej_msg, ari_length);
+		IBA_SET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg, ari_length);
 		memcpy(rej_msg->ari, ari, ari_length);
 	}
 
@@ -1548,14 +1553,14 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 	primary_path->traffic_class = req_msg->primary_traffic_class;
 	primary_path->reversible = 1;
 	primary_path->pkey = req_msg->pkey;
-	primary_path->sl = cm_req_get_primary_sl(req_msg);
+	primary_path->sl = IBA_GET(CM_REQ_PRIMARY_SL, req_msg);
 	primary_path->mtu_selector = IB_SA_EQ;
-	primary_path->mtu = cm_req_get_path_mtu(req_msg);
+	primary_path->mtu = IBA_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg);
 	primary_path->rate_selector = IB_SA_EQ;
-	primary_path->rate = cm_req_get_primary_packet_rate(req_msg);
+	primary_path->rate = IBA_GET(CM_REQ_PRIMARY_PACKET_RATE, req_msg);
 	primary_path->packet_life_time_selector = IB_SA_EQ;
 	primary_path->packet_life_time =
-		cm_req_get_primary_local_ack_timeout(req_msg);
+		IBA_GET(CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT, req_msg);
 	primary_path->packet_life_time -= (primary_path->packet_life_time > 0);
 	primary_path->service_id = req_msg->service_id;
 	if (sa_path_is_roce(primary_path))
@@ -1569,14 +1574,15 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 		alt_path->traffic_class = req_msg->alt_traffic_class;
 		alt_path->reversible = 1;
 		alt_path->pkey = req_msg->pkey;
-		alt_path->sl = cm_req_get_alt_sl(req_msg);
+		alt_path->sl = IBA_GET(CM_REQ_ALTERNATE_SL, req_msg);
 		alt_path->mtu_selector = IB_SA_EQ;
-		alt_path->mtu = cm_req_get_path_mtu(req_msg);
+		alt_path->mtu =
+			IBA_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg);
 		alt_path->rate_selector = IB_SA_EQ;
-		alt_path->rate = cm_req_get_alt_packet_rate(req_msg);
+		alt_path->rate = IBA_GET(CM_REQ_ALTERNATE_PACKET_RATE, req_msg);
 		alt_path->packet_life_time_selector = IB_SA_EQ;
 		alt_path->packet_life_time =
-			cm_req_get_alt_local_ack_timeout(req_msg);
+			IBA_GET(CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT, req_msg);
 		alt_path->packet_life_time -= (alt_path->packet_life_time > 0);
 		alt_path->service_id = req_msg->service_id;
 
@@ -1658,16 +1664,16 @@ static void cm_format_req_event(struct cm_work *work,
 	param->remote_qpn = be32_to_cpu(cm_req_get_local_qpn(req_msg));
 	param->qp_type = cm_req_get_qp_type(req_msg);
 	param->starting_psn = be32_to_cpu(cm_req_get_starting_psn(req_msg));
-	param->responder_resources = cm_req_get_init_depth(req_msg);
-	param->initiator_depth = cm_req_get_resp_res(req_msg);
+	param->responder_resources = IBA_GET(CM_REQ_INITIATOR_DEPTH, req_msg);
+	param->initiator_depth = IBA_GET(CM_REQ_RESPONDER_RESOURCES, req_msg);
 	param->local_cm_response_timeout =
-					cm_req_get_remote_resp_timeout(req_msg);
-	param->flow_control = cm_req_get_flow_ctrl(req_msg);
+		IBA_GET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, req_msg);
+	param->flow_control = IBA_GET(CM_REQ_END_TO_END_FLOW_CONTROL, req_msg);
 	param->remote_cm_response_timeout =
-					cm_req_get_local_resp_timeout(req_msg);
-	param->retry_count = cm_req_get_retry_count(req_msg);
-	param->rnr_retry_count = cm_req_get_rnr_retry_count(req_msg);
-	param->srq = cm_req_get_srq(req_msg);
+		IBA_GET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, req_msg);
+	param->retry_count = IBA_GET(CM_REQ_RETRY_COUNT, req_msg);
+	param->rnr_retry_count = IBA_GET(CM_REQ_RNR_RETRY_COUNT, req_msg);
+	param->srq = IBA_GET(CM_REQ_SRQ, req_msg);
 	param->ppath_sgid_attr = cm_id_priv->av.ah_attr.grh.sgid_attr;
 	work->cm_event.private_data = &req_msg->private_data;
 }
@@ -1703,10 +1709,10 @@ static void cm_format_mra(struct cm_mra_msg *mra_msg,
 			  const void *private_data, u8 private_data_len)
 {
 	cm_format_mad_hdr(&mra_msg->hdr, CM_MRA_ATTR_ID, cm_id_priv->tid);
-	cm_mra_set_msg_mraed(mra_msg, msg_mraed);
+	IBA_SET(CM_MRA_MESSAGE_MRAED, mra_msg, msg_mraed);
 	mra_msg->local_comm_id = cm_id_priv->id.local_id;
 	mra_msg->remote_comm_id = cm_id_priv->id.remote_id;
-	cm_mra_set_service_timeout(mra_msg, service_timeout);
+	IBA_SET(CM_MRA_SERVICE_TIMEOUT, mra_msg, service_timeout);
 
 	if (private_data && private_data_len)
 		memcpy(mra_msg->private_data, private_data, private_data_len);
@@ -1726,26 +1732,27 @@ static void cm_format_rej(struct cm_rej_msg *rej_msg,
 	switch(cm_id_priv->id.state) {
 	case IB_CM_REQ_RCVD:
 		rej_msg->local_comm_id = 0;
-		cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_REQ);
+		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, CM_MSG_RESPONSE_REQ);
 		break;
 	case IB_CM_MRA_REQ_SENT:
 		rej_msg->local_comm_id = cm_id_priv->id.local_id;
-		cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_REQ);
+		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, CM_MSG_RESPONSE_REQ);
 		break;
 	case IB_CM_REP_RCVD:
 	case IB_CM_MRA_REP_SENT:
 		rej_msg->local_comm_id = cm_id_priv->id.local_id;
-		cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_REP);
+		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, CM_MSG_RESPONSE_REP);
 		break;
 	default:
 		rej_msg->local_comm_id = cm_id_priv->id.local_id;
-		cm_rej_set_msg_rejected(rej_msg, CM_MSG_RESPONSE_OTHER);
+		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg,
+			CM_MSG_RESPONSE_OTHER);
 		break;
 	}
 
 	rej_msg->reason = cpu_to_be16(reason);
 	if (ari && ari_length) {
-		cm_rej_set_reject_info_len(rej_msg, ari_length);
+		IBA_SET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg, ari_length);
 		memcpy(rej_msg->ari, ari, ari_length);
 	}
 
@@ -1866,20 +1873,20 @@ out:
  */
 static void cm_process_routed_req(struct cm_req_msg *req_msg, struct ib_wc *wc)
 {
-	if (!cm_req_get_primary_subnet_local(req_msg)) {
+	if (!IBA_GET(CM_REQ_PRIMARY_SUBNET_LOCAL, req_msg)) {
 		if (req_msg->primary_local_lid == IB_LID_PERMISSIVE) {
 			req_msg->primary_local_lid = ib_lid_be16(wc->slid);
-			cm_req_set_primary_sl(req_msg, wc->sl);
+			IBA_SET(CM_REQ_PRIMARY_SL, req_msg, wc->sl);
 		}
 
 		if (req_msg->primary_remote_lid == IB_LID_PERMISSIVE)
 			req_msg->primary_remote_lid = cpu_to_be16(wc->dlid_path_bits);
 	}
 
-	if (!cm_req_get_alt_subnet_local(req_msg)) {
+	if (!IBA_GET(CM_REQ_ALTERNATE_SUBNET_LOCAL, req_msg)) {
 		if (req_msg->alt_local_lid == IB_LID_PERMISSIVE) {
 			req_msg->alt_local_lid = ib_lid_be16(wc->slid);
-			cm_req_set_alt_sl(req_msg, wc->sl);
+			IBA_SET(CM_REQ_ALTERNATE_SL, req_msg, wc->sl);
 		}
 
 		if (req_msg->alt_remote_lid == IB_LID_PERMISSIVE)
@@ -1989,16 +1996,18 @@ static int cm_req_handler(struct cm_work *work)
 	}
 	cm_id_priv->tid = req_msg->hdr.tid;
 	cm_id_priv->timeout_ms = cm_convert_to_ms(
-					cm_req_get_local_resp_timeout(req_msg));
-	cm_id_priv->max_cm_retries = cm_req_get_max_cm_retries(req_msg);
+		IBA_GET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, req_msg));
+	cm_id_priv->max_cm_retries = IBA_GET(CM_REQ_MAX_CM_RETRIES, req_msg);
 	cm_id_priv->remote_qpn = cm_req_get_local_qpn(req_msg);
-	cm_id_priv->initiator_depth = cm_req_get_resp_res(req_msg);
-	cm_id_priv->responder_resources = cm_req_get_init_depth(req_msg);
-	cm_id_priv->path_mtu = cm_req_get_path_mtu(req_msg);
+	cm_id_priv->initiator_depth =
+		IBA_GET(CM_REQ_RESPONDER_RESOURCES, req_msg);
+	cm_id_priv->responder_resources =
+		IBA_GET(CM_REQ_INITIATOR_DEPTH, req_msg);
+	cm_id_priv->path_mtu = IBA_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg);
 	cm_id_priv->pkey = req_msg->pkey;
 	cm_id_priv->sq_psn = cm_req_get_starting_psn(req_msg);
-	cm_id_priv->retry_count = cm_req_get_retry_count(req_msg);
-	cm_id_priv->rnr_retry_count = cm_req_get_rnr_retry_count(req_msg);
+	cm_id_priv->retry_count = IBA_GET(CM_REQ_RETRY_COUNT, req_msg);
+	cm_id_priv->rnr_retry_count = IBA_GET(CM_REQ_RNR_RETRY_COUNT, req_msg);
 	cm_id_priv->qp_type = cm_req_get_qp_type(req_msg);
 
 	cm_format_req_event(work, cm_id_priv, &listen_cm_id_priv->id);
@@ -2025,19 +2034,20 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg,
 	rep_msg->remote_comm_id = cm_id_priv->id.remote_id;
 	cm_rep_set_starting_psn(rep_msg, cpu_to_be32(param->starting_psn));
 	rep_msg->resp_resources = param->responder_resources;
-	cm_rep_set_target_ack_delay(rep_msg,
-				    cm_id_priv->av.port->cm_dev->ack_delay);
-	cm_rep_set_failover(rep_msg, param->failover_accepted);
-	cm_rep_set_rnr_retry_count(rep_msg, param->rnr_retry_count);
+	IBA_SET(CM_REP_TARGET_ACK_DELAY, rep_msg,
+		cm_id_priv->av.port->cm_dev->ack_delay);
+	IBA_SET(CM_REP_FAILOVER_ACCEPTED, rep_msg, param->failover_accepted);
+	IBA_SET(CM_REP_RNR_RETRY_COUNT, rep_msg, param->rnr_retry_count);
 	rep_msg->local_ca_guid = cm_id_priv->id.device->node_guid;
 
 	if (cm_id_priv->qp_type != IB_QPT_XRC_TGT) {
 		rep_msg->initiator_depth = param->initiator_depth;
-		cm_rep_set_flow_ctrl(rep_msg, param->flow_control);
-		cm_rep_set_srq(rep_msg, param->srq);
+		IBA_SET(CM_REP_END_TO_END_FLOW_CONTROL, rep_msg,
+			param->flow_control);
+		IBA_SET(CM_REP_SRQ, rep_msg, param->srq);
 		cm_rep_set_local_qpn(rep_msg, cpu_to_be32(param->qp_num));
 	} else {
-		cm_rep_set_srq(rep_msg, 1);
+		IBA_SET(CM_REP_SRQ, rep_msg, 1);
 		cm_rep_set_local_eecn(rep_msg, cpu_to_be32(param->qp_num));
 	}
 
@@ -2176,11 +2186,11 @@ static void cm_format_rep_event(struct cm_work *work, enum ib_qp_type qp_type)
 	param->starting_psn = be32_to_cpu(cm_rep_get_starting_psn(rep_msg));
 	param->responder_resources = rep_msg->initiator_depth;
 	param->initiator_depth = rep_msg->resp_resources;
-	param->target_ack_delay = cm_rep_get_target_ack_delay(rep_msg);
-	param->failover_accepted = cm_rep_get_failover(rep_msg);
-	param->flow_control = cm_rep_get_flow_ctrl(rep_msg);
-	param->rnr_retry_count = cm_rep_get_rnr_retry_count(rep_msg);
-	param->srq = cm_rep_get_srq(rep_msg);
+	param->target_ack_delay = IBA_GET(CM_REP_TARGET_ACK_DELAY, rep_msg);
+	param->failover_accepted = IBA_GET(CM_REP_FAILOVER_ACCEPTED, rep_msg);
+	param->flow_control = IBA_GET(CM_REP_END_TO_END_FLOW_CONTROL, rep_msg);
+	param->rnr_retry_count = IBA_GET(CM_REP_RNR_RETRY_COUNT, rep_msg);
+	param->srq = IBA_GET(CM_REP_SRQ, rep_msg);
 	work->cm_event.private_data = &rep_msg->private_data;
 }
 
@@ -2311,8 +2321,9 @@ static int cm_rep_handler(struct cm_work *work)
 	cm_id_priv->initiator_depth = rep_msg->resp_resources;
 	cm_id_priv->responder_resources = rep_msg->initiator_depth;
 	cm_id_priv->sq_psn = cm_rep_get_starting_psn(rep_msg);
-	cm_id_priv->rnr_retry_count = cm_rep_get_rnr_retry_count(rep_msg);
-	cm_id_priv->target_ack_delay = cm_rep_get_target_ack_delay(rep_msg);
+	cm_id_priv->rnr_retry_count = IBA_GET(CM_REP_RNR_RETRY_COUNT, rep_msg);
+	cm_id_priv->target_ack_delay =
+		IBA_GET(CM_REP_TARGET_ACK_DELAY, rep_msg);
 	cm_id_priv->av.timeout =
 			cm_ack_timeout(cm_id_priv->target_ack_delay,
 				       cm_id_priv->av.timeout - 1);
@@ -2756,7 +2767,7 @@ static void cm_format_rej_event(struct cm_work *work)
 	rej_msg = (struct cm_rej_msg *)work->mad_recv_wc->recv_buf.mad;
 	param = &work->cm_event.param.rej_rcvd;
 	param->ari = rej_msg->ari;
-	param->ari_length = cm_rej_get_reject_info_len(rej_msg);
+	param->ari_length = IBA_GET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg);
 	param->reason = __be16_to_cpu(rej_msg->reason);
 	work->cm_event.private_data = &rej_msg->private_data;
 }
@@ -2780,7 +2791,8 @@ static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
 		cm_id_priv =
 			cm_acquire_id(timewait_info->work.local_id, remote_id);
 		spin_unlock_irq(&cm.lock);
-	} else if (cm_rej_get_msg_rejected(rej_msg) == CM_MSG_RESPONSE_REQ)
+	} else if (IBA_GET(CM_REJ_MESSAGE_REJECTED, rej_msg) ==
+		   CM_MSG_RESPONSE_REQ)
 		cm_id_priv = cm_acquire_id(rej_msg->remote_comm_id, 0);
 	else
 		cm_id_priv = cm_acquire_id(rej_msg->remote_comm_id, remote_id);
@@ -2941,7 +2953,7 @@ EXPORT_SYMBOL(ib_send_cm_mra);
 
 static struct cm_id_private * cm_acquire_mraed_id(struct cm_mra_msg *mra_msg)
 {
-	switch (cm_mra_get_msg_mraed(mra_msg)) {
+	switch (IBA_GET(CM_MRA_MESSAGE_MRAED, mra_msg)) {
 	case CM_MSG_RESPONSE_REQ:
 		return cm_acquire_id(mra_msg->remote_comm_id, 0);
 	case CM_MSG_RESPONSE_REP:
@@ -2966,28 +2978,31 @@ static int cm_mra_handler(struct cm_work *work)
 
 	work->cm_event.private_data = &mra_msg->private_data;
 	work->cm_event.param.mra_rcvd.service_timeout =
-					cm_mra_get_service_timeout(mra_msg);
-	timeout = cm_convert_to_ms(cm_mra_get_service_timeout(mra_msg)) +
+		IBA_GET(CM_MRA_SERVICE_TIMEOUT, mra_msg);
+	timeout = cm_convert_to_ms(IBA_GET(CM_MRA_SERVICE_TIMEOUT, mra_msg)) +
 		  cm_convert_to_ms(cm_id_priv->av.timeout);
 
 	spin_lock_irq(&cm_id_priv->lock);
 	switch (cm_id_priv->id.state) {
 	case IB_CM_REQ_SENT:
-		if (cm_mra_get_msg_mraed(mra_msg) != CM_MSG_RESPONSE_REQ ||
+		if (IBA_GET(CM_MRA_MESSAGE_MRAED, mra_msg) !=
+			    CM_MSG_RESPONSE_REQ ||
 		    ib_modify_mad(cm_id_priv->av.port->mad_agent,
 				  cm_id_priv->msg, timeout))
 			goto out;
 		cm_id_priv->id.state = IB_CM_MRA_REQ_RCVD;
 		break;
 	case IB_CM_REP_SENT:
-		if (cm_mra_get_msg_mraed(mra_msg) != CM_MSG_RESPONSE_REP ||
+		if (IBA_GET(CM_MRA_MESSAGE_MRAED, mra_msg) !=
+			    CM_MSG_RESPONSE_REP ||
 		    ib_modify_mad(cm_id_priv->av.port->mad_agent,
 				  cm_id_priv->msg, timeout))
 			goto out;
 		cm_id_priv->id.state = IB_CM_MRA_REP_RCVD;
 		break;
 	case IB_CM_ESTABLISHED:
-		if (cm_mra_get_msg_mraed(mra_msg) != CM_MSG_RESPONSE_OTHER ||
+		if (IBA_GET(CM_MRA_MESSAGE_MRAED, mra_msg) !=
+			    CM_MSG_RESPONSE_OTHER ||
 		    cm_id_priv->id.lap_state != IB_CM_LAP_SENT ||
 		    ib_modify_mad(cm_id_priv->av.port->mad_agent,
 				  cm_id_priv->msg, timeout)) {
@@ -3054,16 +3069,17 @@ static void cm_format_path_from_lap(struct cm_id_private *cm_id_priv,
 	path->sgid = lap_msg->alt_remote_gid;
 	path->flow_label = cm_lap_get_flow_label(lap_msg);
 	path->hop_limit = lap_msg->alt_hop_limit;
-	path->traffic_class = cm_lap_get_traffic_class(lap_msg);
+	path->traffic_class = IBA_GET(CM_LAP_ALTERNATE_TRAFFIC_CLASS, lap_msg);
 	path->reversible = 1;
 	path->pkey = cm_id_priv->pkey;
-	path->sl = cm_lap_get_sl(lap_msg);
+	path->sl = IBA_GET(CM_LAP_ALTERNATE_SL, lap_msg);
 	path->mtu_selector = IB_SA_EQ;
 	path->mtu = cm_id_priv->path_mtu;
 	path->rate_selector = IB_SA_EQ;
-	path->rate = cm_lap_get_packet_rate(lap_msg);
+	path->rate = IBA_GET(CM_LAP_ALTERNATE_PACKET_RATE, lap_msg);
 	path->packet_life_time_selector = IB_SA_EQ;
-	path->packet_life_time = cm_lap_get_local_ack_timeout(lap_msg);
+	path->packet_life_time =
+		IBA_GET(CM_LAP_ALTERNATE_LOCAL_ACK_TIMEOUT, lap_msg);
 	path->packet_life_time -= (path->packet_life_time > 0);
 	cm_format_path_lid_from_lap(lap_msg, path);
 }
@@ -4436,83 +4452,21 @@ static void self_test(void)
 	printk("Running CM extractor self test\n");
 	IBA_CHECK_GET_BE(CM_REQ_LOCAL_QPN, cm_req_get_local_qpn);
 	IBA_CHECK_SET_BE(CM_REQ_LOCAL_QPN, cm_req_set_local_qpn);
-	IBA_CHECK_GET(CM_REQ_RESPONDER_RESOURCES, cm_req_get_resp_res);
-	IBA_CHECK_SET(CM_REQ_RESPONDER_RESOURCES, cm_req_set_resp_res);
-	IBA_CHECK_GET(CM_REQ_INITIATOR_DEPTH, cm_req_get_init_depth);
-	IBA_CHECK_SET(CM_REQ_INITIATOR_DEPTH, cm_req_set_init_depth);
-	IBA_CHECK_GET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, cm_req_get_remote_resp_timeout);
-	IBA_CHECK_SET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, cm_req_set_remote_resp_timeout);
-	IBA_CHECK_GET(CM_REQ_TRANSPORT_SERVICE_TYPE, cm_req_get_transport_type);
-	IBA_CHECK_SET(CM_REQ_TRANSPORT_SERVICE_TYPE, cm_req_set_transport_type);
-	IBA_CHECK_GET(CM_REQ_END_TO_END_FLOW_CONTROL, cm_req_get_flow_ctrl);
-	IBA_CHECK_SET(CM_REQ_END_TO_END_FLOW_CONTROL, cm_req_set_flow_ctrl);
 	IBA_CHECK_GET_BE(CM_REQ_STARTING_PSN, cm_req_get_starting_psn);
 	IBA_CHECK_SET_BE(CM_REQ_STARTING_PSN, cm_req_set_starting_psn);
-	IBA_CHECK_GET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, cm_req_get_local_resp_timeout);
-	IBA_CHECK_SET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, cm_req_set_local_resp_timeout);
-	IBA_CHECK_GET(CM_REQ_RETRY_COUNT, cm_req_get_retry_count);
-	IBA_CHECK_SET(CM_REQ_RETRY_COUNT, cm_req_set_retry_count);
-	IBA_CHECK_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, cm_req_get_path_mtu);
-	IBA_CHECK_SET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, cm_req_set_path_mtu);
-	IBA_CHECK_GET(CM_REQ_RNR_RETRY_COUNT, cm_req_get_rnr_retry_count);
-	IBA_CHECK_SET(CM_REQ_RNR_RETRY_COUNT, cm_req_set_rnr_retry_count);
-	IBA_CHECK_GET(CM_REQ_MAX_CM_RETRIES, cm_req_get_max_cm_retries);
-	IBA_CHECK_SET(CM_REQ_MAX_CM_RETRIES, cm_req_set_max_cm_retries);
-	IBA_CHECK_GET(CM_REQ_SRQ, cm_req_get_srq);
-	IBA_CHECK_SET(CM_REQ_SRQ, cm_req_set_srq);
-	IBA_CHECK_GET(CM_REQ_EXTENDED_TRANSPORT_TYPE, cm_req_get_transport_type_ex);
-	IBA_CHECK_SET(CM_REQ_EXTENDED_TRANSPORT_TYPE, cm_req_set_transport_type_ex);
 	IBA_CHECK_GET_BE(CM_REQ_PRIMARY_FLOW_LABEL, cm_req_get_primary_flow_label);
 	IBA_CHECK_SET_BE(CM_REQ_PRIMARY_FLOW_LABEL, cm_req_set_primary_flow_label);
-	IBA_CHECK_GET(CM_REQ_PRIMARY_PACKET_RATE, cm_req_get_primary_packet_rate);
-	IBA_CHECK_SET(CM_REQ_PRIMARY_PACKET_RATE, cm_req_set_primary_packet_rate);
-	IBA_CHECK_GET(CM_REQ_PRIMARY_SL, cm_req_get_primary_sl);
-	IBA_CHECK_SET(CM_REQ_PRIMARY_SL, cm_req_set_primary_sl);
-	IBA_CHECK_GET(CM_REQ_PRIMARY_SUBNET_LOCAL, cm_req_get_primary_subnet_local);
-	IBA_CHECK_SET(CM_REQ_PRIMARY_SUBNET_LOCAL, cm_req_set_primary_subnet_local);
-	IBA_CHECK_GET(CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT, cm_req_get_primary_local_ack_timeout);
-	IBA_CHECK_SET(CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT, cm_req_set_primary_local_ack_timeout);
 	IBA_CHECK_GET_BE(CM_REQ_ALTERNATE_FLOW_LABEL, cm_req_get_alt_flow_label);
 	IBA_CHECK_SET_BE(CM_REQ_ALTERNATE_FLOW_LABEL, cm_req_set_alt_flow_label);
-	IBA_CHECK_GET(CM_REQ_ALTERNATE_PACKET_RATE, cm_req_get_alt_packet_rate);
-	IBA_CHECK_SET(CM_REQ_ALTERNATE_PACKET_RATE, cm_req_set_alt_packet_rate);
-	IBA_CHECK_GET(CM_REQ_ALTERNATE_SL, cm_req_get_alt_sl);
-	IBA_CHECK_SET(CM_REQ_ALTERNATE_SL, cm_req_set_alt_sl);
-	IBA_CHECK_GET(CM_REQ_ALTERNATE_SUBNET_LOCAL, cm_req_get_alt_subnet_local);
-	IBA_CHECK_SET(CM_REQ_ALTERNATE_SUBNET_LOCAL, cm_req_set_alt_subnet_local);
-	IBA_CHECK_GET(CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT, cm_req_get_alt_local_ack_timeout);
-	IBA_CHECK_SET(CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT, cm_req_set_alt_local_ack_timeout);
-	IBA_CHECK_GET(CM_MRA_MESSAGE_MRAED, cm_mra_get_msg_mraed);
-	IBA_CHECK_SET(CM_MRA_MESSAGE_MRAED, cm_mra_set_msg_mraed);
-	IBA_CHECK_GET(CM_MRA_SERVICE_TIMEOUT, cm_mra_get_service_timeout);
-	IBA_CHECK_SET(CM_MRA_SERVICE_TIMEOUT, cm_mra_set_service_timeout);
-	IBA_CHECK_GET(CM_REJ_MESSAGE_REJECTED, cm_rej_get_msg_rejected);
-	IBA_CHECK_SET(CM_REJ_MESSAGE_REJECTED, cm_rej_set_msg_rejected);
-	IBA_CHECK_GET(CM_REJ_REJECTED_INFO_LENGTH, cm_rej_get_reject_info_len);
-	IBA_CHECK_SET(CM_REJ_REJECTED_INFO_LENGTH, cm_rej_set_reject_info_len);
 	IBA_CHECK_GET_BE(CM_REP_LOCAL_QPN, cm_rep_get_local_qpn);
 	IBA_CHECK_SET_BE(CM_REP_LOCAL_QPN, cm_rep_set_local_qpn);
 	IBA_CHECK_GET_BE(CM_REP_LOCAL_EE_CONTEXT_NUMBER, cm_rep_get_local_eecn);
 	IBA_CHECK_SET_BE(CM_REP_LOCAL_EE_CONTEXT_NUMBER, cm_rep_set_local_eecn);
 	IBA_CHECK_GET_BE(CM_REP_STARTING_PSN, cm_rep_get_starting_psn);
 	IBA_CHECK_SET_BE(CM_REP_STARTING_PSN, cm_rep_set_starting_psn);
-	IBA_CHECK_GET(CM_REP_TARGET_ACK_DELAY, cm_rep_get_target_ack_delay);
-	IBA_CHECK_SET(CM_REP_TARGET_ACK_DELAY, cm_rep_set_target_ack_delay);
-	IBA_CHECK_GET(CM_REP_FAILOVER_ACCEPTED, cm_rep_get_failover);
-	IBA_CHECK_SET(CM_REP_FAILOVER_ACCEPTED, cm_rep_set_failover);
-	IBA_CHECK_GET(CM_REP_END_TO_END_FLOW_CONTROL, cm_rep_get_flow_ctrl);
-	IBA_CHECK_SET(CM_REP_END_TO_END_FLOW_CONTROL, cm_rep_set_flow_ctrl);
-	IBA_CHECK_GET(CM_REP_RNR_RETRY_COUNT, cm_rep_get_rnr_retry_count);
-	IBA_CHECK_SET(CM_REP_RNR_RETRY_COUNT, cm_rep_set_rnr_retry_count);
-	IBA_CHECK_GET(CM_REP_SRQ, cm_rep_get_srq);
-	IBA_CHECK_SET(CM_REP_SRQ, cm_rep_set_srq);
 	IBA_CHECK_GET_BE(CM_DREQ_REMOTE_QPN_EECN, cm_dreq_get_remote_qpn);
 	IBA_CHECK_SET_BE(CM_DREQ_REMOTE_QPN_EECN, cm_dreq_set_remote_qpn);
 	IBA_CHECK_GET_BE(CM_LAP_ALTERNATE_FLOW_LABEL, cm_lap_get_flow_label);
-	IBA_CHECK_GET(CM_LAP_ALTERNATE_TRAFFIC_CLASS, cm_lap_get_traffic_class);
-	IBA_CHECK_GET(CM_LAP_ALTERNATE_PACKET_RATE, cm_lap_get_packet_rate);
-	IBA_CHECK_GET(CM_LAP_ALTERNATE_SL, cm_lap_get_sl);
-	IBA_CHECK_GET(CM_LAP_ALTERNATE_LOCAL_ACK_TIMEOUT, cm_lap_get_local_ack_timeout);
 	IBA_CHECK_GET_BE(CM_SIDR_REP_QPN, cm_sidr_rep_get_qpn);
 	IBA_CHECK_SET_BE(CM_SIDR_REP_QPN, cm_sidr_rep_set_qpn);
 	printk("Success!\n");
diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h
index bf62461d801f..d30586b1b8a4 100644
--- a/drivers/infiniband/core/cm_msgs.h
+++ b/drivers/infiniband/core/cm_msgs.h
@@ -87,75 +87,14 @@ static inline void cm_req_set_local_qpn(struct cm_req_msg *req_msg, __be32 qpn)
 					  0x000000FF));
 }
 
-static inline u8 cm_req_get_resp_res(struct cm_req_msg *req_msg)
-{
-	return (u8) be32_to_cpu(req_msg->offset32);
-}
-
-static inline void cm_req_set_resp_res(struct cm_req_msg *req_msg, u8 resp_res)
-{
-	req_msg->offset32 = cpu_to_be32(resp_res |
-					(be32_to_cpu(req_msg->offset32) &
-					 0xFFFFFF00));
-}
-
-static inline u8 cm_req_get_init_depth(struct cm_req_msg *req_msg)
-{
-	return (u8) be32_to_cpu(req_msg->offset36);
-}
-
-static inline void cm_req_set_init_depth(struct cm_req_msg *req_msg,
-					 u8 init_depth)
-{
-	req_msg->offset36 = cpu_to_be32(init_depth |
-					(be32_to_cpu(req_msg->offset36) &
-					 0xFFFFFF00));
-}
-
-static inline u8 cm_req_get_remote_resp_timeout(struct cm_req_msg *req_msg)
-{
-	return (u8) ((be32_to_cpu(req_msg->offset40) & 0xF8) >> 3);
-}
-
-static inline void cm_req_set_remote_resp_timeout(struct cm_req_msg *req_msg,
-						  u8 resp_timeout)
-{
-	req_msg->offset40 = cpu_to_be32((resp_timeout << 3) |
-					 (be32_to_cpu(req_msg->offset40) &
-					  0xFFFFFF07));
-}
-
-static inline u8 cm_req_get_transport_type(struct cm_req_msg *req_msg)
-{
-	return (u8) ((be32_to_cpu(req_msg->offset40) & 0x06) >> 1);
-}
-
-static inline void cm_req_set_transport_type(struct cm_req_msg *req_msg, u8 val)
-{
-	req_msg->offset40 =
-		cpu_to_be32((be32_to_cpu(req_msg->offset40) & 0xFFFFFFF9) |
-		(val << 1));
-}
-
-static inline u8 cm_req_get_transport_type_ex(struct cm_req_msg *req_msg)
-{
-	return req_msg->offset51 & 0x7;
-}
-
-static inline void cm_req_set_transport_type_ex(struct cm_req_msg *req_msg,
-						u8 val)
-{
-	req_msg->offset51 = (req_msg->offset51 & 0xF8) | val;
-}
-
 static inline enum ib_qp_type cm_req_get_qp_type(struct cm_req_msg *req_msg)
 {
-	u8 transport_type = cm_req_get_transport_type(req_msg);
+	u8 transport_type = IBA_GET(CM_REQ_TRANSPORT_SERVICE_TYPE, req_msg);
 	switch(transport_type) {
 	case 0: return IB_QPT_RC;
 	case 1: return IB_QPT_UC;
 	case 3:
-		switch (cm_req_get_transport_type_ex(req_msg)) {
+		switch (IBA_GET(CM_REQ_EXTENDED_TRANSPORT_TYPE, req_msg)) {
 		case 1: return IB_QPT_XRC_TGT;
 		default: return 0;
 		}
@@ -168,30 +107,17 @@ static inline void cm_req_set_qp_type(struct cm_req_msg *req_msg,
 {
 	switch(qp_type) {
 	case IB_QPT_UC:
-		cm_req_set_transport_type(req_msg, 1);
+		IBA_SET(CM_REQ_TRANSPORT_SERVICE_TYPE, req_msg, 1);
 		break;
 	case IB_QPT_XRC_INI:
-		cm_req_set_transport_type(req_msg, 3);
-		cm_req_set_transport_type_ex(req_msg, 1);
+		IBA_SET(CM_REQ_TRANSPORT_SERVICE_TYPE, req_msg, 3);
+		IBA_SET(CM_REQ_EXTENDED_TRANSPORT_TYPE, req_msg, 1);
 		break;
 	default:
-		cm_req_set_transport_type(req_msg, 0);
+		IBA_SET(CM_REQ_TRANSPORT_SERVICE_TYPE, req_msg, 0);
 	}
 }
 
-static inline u8 cm_req_get_flow_ctrl(struct cm_req_msg *req_msg)
-{
-	return be32_to_cpu(req_msg->offset40) & 0x1;
-}
-
-static inline void cm_req_set_flow_ctrl(struct cm_req_msg *req_msg,
-					u8 flow_ctrl)
-{
-	req_msg->offset40 = cpu_to_be32((flow_ctrl & 0x1) |
-					 (be32_to_cpu(req_msg->offset40) &
-					  0xFFFFFFFE));
-}
-
 static inline __be32 cm_req_get_starting_psn(struct cm_req_msg *req_msg)
 {
 	return cpu_to_be32(be32_to_cpu(req_msg->offset44) >> 8);
@@ -204,74 +130,6 @@ static inline void cm_req_set_starting_psn(struct cm_req_msg *req_msg,
 			    (be32_to_cpu(req_msg->offset44) & 0x000000FF));
 }
 
-static inline u8 cm_req_get_local_resp_timeout(struct cm_req_msg *req_msg)
-{
-	return (u8) ((be32_to_cpu(req_msg->offset44) & 0xF8) >> 3);
-}
-
-static inline void cm_req_set_local_resp_timeout(struct cm_req_msg *req_msg,
-						 u8 resp_timeout)
-{
-	req_msg->offset44 = cpu_to_be32((resp_timeout << 3) |
-			    (be32_to_cpu(req_msg->offset44) & 0xFFFFFF07));
-}
-
-static inline u8 cm_req_get_retry_count(struct cm_req_msg *req_msg)
-{
-	return (u8) (be32_to_cpu(req_msg->offset44) & 0x7);
-}
-
-static inline void cm_req_set_retry_count(struct cm_req_msg *req_msg,
-					  u8 retry_count)
-{
-	req_msg->offset44 = cpu_to_be32((retry_count & 0x7) |
-			    (be32_to_cpu(req_msg->offset44) & 0xFFFFFFF8));
-}
-
-static inline u8 cm_req_get_path_mtu(struct cm_req_msg *req_msg)
-{
-	return req_msg->offset50 >> 4;
-}
-
-static inline void cm_req_set_path_mtu(struct cm_req_msg *req_msg, u8 path_mtu)
-{
-	req_msg->offset50 = (u8) ((req_msg->offset50 & 0xF) | (path_mtu << 4));
-}
-
-static inline u8 cm_req_get_rnr_retry_count(struct cm_req_msg *req_msg)
-{
-	return req_msg->offset50 & 0x7;
-}
-
-static inline void cm_req_set_rnr_retry_count(struct cm_req_msg *req_msg,
-					      u8 rnr_retry_count)
-{
-	req_msg->offset50 = (u8) ((req_msg->offset50 & 0xF8) |
-				  (rnr_retry_count & 0x7));
-}
-
-static inline u8 cm_req_get_max_cm_retries(struct cm_req_msg *req_msg)
-{
-	return req_msg->offset51 >> 4;
-}
-
-static inline void cm_req_set_max_cm_retries(struct cm_req_msg *req_msg,
-					     u8 retries)
-{
-	req_msg->offset51 = (u8) ((req_msg->offset51 & 0xF) | (retries << 4));
-}
-
-static inline u8 cm_req_get_srq(struct cm_req_msg *req_msg)
-{
-	return (req_msg->offset51 & 0x8) >> 3;
-}
-
-static inline void cm_req_set_srq(struct cm_req_msg *req_msg, u8 srq)
-{
-	req_msg->offset51 = (u8) ((req_msg->offset51 & 0xF7) |
-				  ((srq & 0x1) << 3));
-}
-
 static inline __be32 cm_req_get_primary_flow_label(struct cm_req_msg *req_msg)
 {
 	return cpu_to_be32(be32_to_cpu(req_msg->primary_offset88) >> 12);
@@ -286,54 +144,6 @@ static inline void cm_req_set_primary_flow_label(struct cm_req_msg *req_msg,
 				     (be32_to_cpu(flow_label) << 12));
 }
 
-static inline u8 cm_req_get_primary_packet_rate(struct cm_req_msg *req_msg)
-{
-	return (u8) (be32_to_cpu(req_msg->primary_offset88) & 0x3F);
-}
-
-static inline void cm_req_set_primary_packet_rate(struct cm_req_msg *req_msg,
-						  u8 rate)
-{
-	req_msg->primary_offset88 = cpu_to_be32(
-				    (be32_to_cpu(req_msg->primary_offset88) &
-				     0xFFFFFFC0) | (rate & 0x3F));
-}
-
-static inline u8 cm_req_get_primary_sl(struct cm_req_msg *req_msg)
-{
-	return (u8) (req_msg->primary_offset94 >> 4);
-}
-
-static inline void cm_req_set_primary_sl(struct cm_req_msg *req_msg, u8 sl)
-{
-	req_msg->primary_offset94 = (u8) ((req_msg->primary_offset94 & 0x0F) |
-					  (sl << 4));
-}
-
-static inline u8 cm_req_get_primary_subnet_local(struct cm_req_msg *req_msg)
-{
-	return (u8) ((req_msg->primary_offset94 & 0x08) >> 3);
-}
-
-static inline void cm_req_set_primary_subnet_local(struct cm_req_msg *req_msg,
-						   u8 subnet_local)
-{
-	req_msg->primary_offset94 = (u8) ((req_msg->primary_offset94 & 0xF7) |
-					  ((subnet_local & 0x1) << 3));
-}
-
-static inline u8 cm_req_get_primary_local_ack_timeout(struct cm_req_msg *req_msg)
-{
-	return (u8) (req_msg->primary_offset95 >> 3);
-}
-
-static inline void cm_req_set_primary_local_ack_timeout(struct cm_req_msg *req_msg,
-							u8 local_ack_timeout)
-{
-	req_msg->primary_offset95 = (u8) ((req_msg->primary_offset95 & 0x07) |
-					  (local_ack_timeout << 3));
-}
-
 static inline __be32 cm_req_get_alt_flow_label(struct cm_req_msg *req_msg)
 {
 	return cpu_to_be32(be32_to_cpu(req_msg->alt_offset132) >> 12);
@@ -348,54 +158,6 @@ static inline void cm_req_set_alt_flow_label(struct cm_req_msg *req_msg,
 				  (be32_to_cpu(flow_label) << 12));
 }
 
-static inline u8 cm_req_get_alt_packet_rate(struct cm_req_msg *req_msg)
-{
-	return (u8) (be32_to_cpu(req_msg->alt_offset132) & 0x3F);
-}
-
-static inline void cm_req_set_alt_packet_rate(struct cm_req_msg *req_msg,
-					      u8 rate)
-{
-	req_msg->alt_offset132 = cpu_to_be32(
-				 (be32_to_cpu(req_msg->alt_offset132) &
-				  0xFFFFFFC0) | (rate & 0x3F));
-}
-
-static inline u8 cm_req_get_alt_sl(struct cm_req_msg *req_msg)
-{
-	return (u8) (req_msg->alt_offset138 >> 4);
-}
-
-static inline void cm_req_set_alt_sl(struct cm_req_msg *req_msg, u8 sl)
-{
-	req_msg->alt_offset138 = (u8) ((req_msg->alt_offset138 & 0x0F) |
-				       (sl << 4));
-}
-
-static inline u8 cm_req_get_alt_subnet_local(struct cm_req_msg *req_msg)
-{
-	return (u8) ((req_msg->alt_offset138 & 0x08) >> 3);
-}
-
-static inline void cm_req_set_alt_subnet_local(struct cm_req_msg *req_msg,
-					       u8 subnet_local)
-{
-	req_msg->alt_offset138 = (u8) ((req_msg->alt_offset138 & 0xF7) |
-				       ((subnet_local & 0x1) << 3));
-}
-
-static inline u8 cm_req_get_alt_local_ack_timeout(struct cm_req_msg *req_msg)
-{
-	return (u8) (req_msg->alt_offset139 >> 3);
-}
-
-static inline void cm_req_set_alt_local_ack_timeout(struct cm_req_msg *req_msg,
-						    u8 local_ack_timeout)
-{
-	req_msg->alt_offset139 = (u8) ((req_msg->alt_offset139 & 0x07) |
-				       (local_ack_timeout << 3));
-}
-
 /* Message REJected or MRAed */
 enum cm_msg_response {
 	CM_MSG_RESPONSE_REQ = 0x0,
@@ -417,28 +179,6 @@ enum cm_msg_response {
 
 } __packed;
 
-static inline u8 cm_mra_get_msg_mraed(struct cm_mra_msg *mra_msg)
-{
-	return (u8) (mra_msg->offset8 >> 6);
-}
-
-static inline void cm_mra_set_msg_mraed(struct cm_mra_msg *mra_msg, u8 msg)
-{
-	mra_msg->offset8 = (u8) ((mra_msg->offset8 & 0x3F) | (msg << 6));
-}
-
-static inline u8 cm_mra_get_service_timeout(struct cm_mra_msg *mra_msg)
-{
-	return (u8) (mra_msg->offset9 >> 3);
-}
-
-static inline void cm_mra_set_service_timeout(struct cm_mra_msg *mra_msg,
-					      u8 service_timeout)
-{
-	mra_msg->offset9 = (u8) ((mra_msg->offset9 & 0x07) |
-				 (service_timeout << 3));
-}
-
 struct cm_rej_msg {
 	struct ib_mad_hdr hdr;
 
@@ -455,27 +195,6 @@ struct cm_rej_msg {
 
 } __packed;
 
-static inline u8 cm_rej_get_msg_rejected(struct cm_rej_msg *rej_msg)
-{
-	return (u8) (rej_msg->offset8 >> 6);
-}
-
-static inline void cm_rej_set_msg_rejected(struct cm_rej_msg *rej_msg, u8 msg)
-{
-	rej_msg->offset8 = (u8) ((rej_msg->offset8 & 0x3F) | (msg << 6));
-}
-
-static inline u8 cm_rej_get_reject_info_len(struct cm_rej_msg *rej_msg)
-{
-	return (u8) (rej_msg->offset9 >> 1);
-}
-
-static inline void cm_rej_set_reject_info_len(struct cm_rej_msg *rej_msg,
-					      u8 len)
-{
-	rej_msg->offset9 = (u8) ((rej_msg->offset9 & 0x1) | (len << 1));
-}
-
 struct cm_rep_msg {
 	struct ib_mad_hdr hdr;
 
@@ -540,64 +259,6 @@ static inline void cm_rep_set_starting_psn(struct cm_rep_msg *rep_msg,
 			    (be32_to_cpu(rep_msg->offset20) & 0x000000FF));
 }
 
-static inline u8 cm_rep_get_target_ack_delay(struct cm_rep_msg *rep_msg)
-{
-	return (u8) (rep_msg->offset26 >> 3);
-}
-
-static inline void cm_rep_set_target_ack_delay(struct cm_rep_msg *rep_msg,
-					       u8 target_ack_delay)
-{
-	rep_msg->offset26 = (u8) ((rep_msg->offset26 & 0x07) |
-				  (target_ack_delay << 3));
-}
-
-static inline u8 cm_rep_get_failover(struct cm_rep_msg *rep_msg)
-{
-	return (u8) ((rep_msg->offset26 & 0x06) >> 1);
-}
-
-static inline void cm_rep_set_failover(struct cm_rep_msg *rep_msg, u8 failover)
-{
-	rep_msg->offset26 = (u8) ((rep_msg->offset26 & 0xF9) |
-				  ((failover & 0x3) << 1));
-}
-
-static inline u8 cm_rep_get_flow_ctrl(struct cm_rep_msg *rep_msg)
-{
-	return (u8) (rep_msg->offset26 & 0x01);
-}
-
-static inline void cm_rep_set_flow_ctrl(struct cm_rep_msg *rep_msg,
-					    u8 flow_ctrl)
-{
-	rep_msg->offset26 = (u8) ((rep_msg->offset26 & 0xFE) |
-				  (flow_ctrl & 0x1));
-}
-
-static inline u8 cm_rep_get_rnr_retry_count(struct cm_rep_msg *rep_msg)
-{
-	return (u8) (rep_msg->offset27 >> 5);
-}
-
-static inline void cm_rep_set_rnr_retry_count(struct cm_rep_msg *rep_msg,
-					      u8 rnr_retry_count)
-{
-	rep_msg->offset27 = (u8) ((rep_msg->offset27 & 0x1F) |
-				  (rnr_retry_count << 5));
-}
-
-static inline u8 cm_rep_get_srq(struct cm_rep_msg *rep_msg)
-{
-	return (u8) ((rep_msg->offset27 >> 4) & 0x1);
-}
-
-static inline void cm_rep_set_srq(struct cm_rep_msg *rep_msg, u8 srq)
-{
-	rep_msg->offset27 = (u8) ((rep_msg->offset27 & 0xEF) |
-				  ((srq & 0x1) << 4));
-}
-
 struct cm_rtu_msg {
 	struct ib_mad_hdr hdr;
 
@@ -674,26 +335,6 @@ static inline __be32 cm_lap_get_flow_label(struct cm_lap_msg *lap_msg)
 	return cpu_to_be32(be32_to_cpu(lap_msg->offset56) >> 12);
 }
 
-static inline u8 cm_lap_get_traffic_class(struct cm_lap_msg *lap_msg)
-{
-	return (u8) be32_to_cpu(lap_msg->offset56);
-}
-
-static inline u8 cm_lap_get_packet_rate(struct cm_lap_msg *lap_msg)
-{
-	return lap_msg->offset61 & 0x3F;
-}
-
-static inline u8 cm_lap_get_sl(struct cm_lap_msg *lap_msg)
-{
-	return lap_msg->offset62 >> 4;
-}
-
-static inline u8 cm_lap_get_local_ack_timeout(struct cm_lap_msg *lap_msg)
-{
-	return lap_msg->offset63 >> 3;
-}
-
 struct cm_apr_msg {
 	struct ib_mad_hdr hdr;
 
-- 
cgit v1.2.3-59-g8ed1b


From 01adb7f46fe455975d9aec07966e72958bc20f8e Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Thu, 16 Jan 2020 13:00:34 -0400
Subject: RDMA/cm: Use IBA functions for swapping get/set acessors

Use a Coccinelle spatch script to replace CM helper functions that
return/accept BE values with IBA_GET/SET versions. Applied with

$ spatch --sp-file edits.sp --in-place drivers/infiniband/core/cm.c

The spatch file was generated using the template pattern:

@@
expression val;
{struct} *msg;
@@
- {old_setter}(msg, val)
+ IBA_SET({new_name}, msg, be{bits}_to_cpu(val))
@@
{struct} *msg;
@@
- {old_getter}(msg)
+ cpu_to_be{bits}(IBA_GET({new_name}, msg))

Iterated for every IBA_CHECK_GET_BE()/IBA_CHECK_SET_BE() pairing.

And the below iterated over all byte sizes to remove doubled byte swaps:

@@
expression val;
@@
-be{bits}_to_cpu(cpu_to_be{bits}(val))
+val

(and __be_to_cpu and ntoh varients)

Touched up with clang-format after.

Link: https://lore.kernel.org/r/20200116170037.30109-5-jgg@ziepe.ca
Tested-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c      | 186 +++++++-------------------------------
 drivers/infiniband/core/cm_msgs.h | 119 +-----------------------
 2 files changed, 36 insertions(+), 269 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index a1a968160132..976bb85b6aa6 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -1254,13 +1254,13 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 	req_msg->local_comm_id = cm_id_priv->id.local_id;
 	req_msg->service_id = param->service_id;
 	req_msg->local_ca_guid = cm_id_priv->id.device->node_guid;
-	cm_req_set_local_qpn(req_msg, cpu_to_be32(param->qp_num));
+	IBA_SET(CM_REQ_LOCAL_QPN, req_msg, param->qp_num);
 	IBA_SET(CM_REQ_INITIATOR_DEPTH, req_msg, param->initiator_depth);
 	IBA_SET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, req_msg,
 		param->remote_cm_response_timeout);
 	cm_req_set_qp_type(req_msg, param->qp_type);
 	IBA_SET(CM_REQ_END_TO_END_FLOW_CONTROL, req_msg, param->flow_control);
-	cm_req_set_starting_psn(req_msg, cpu_to_be32(param->starting_psn));
+	IBA_SET(CM_REQ_STARTING_PSN, req_msg, param->starting_psn);
 	IBA_SET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, req_msg,
 		param->local_cm_response_timeout);
 	req_msg->pkey = param->primary_path->pkey;
@@ -1295,7 +1295,8 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 		req_msg->primary_local_lid = IB_LID_PERMISSIVE;
 		req_msg->primary_remote_lid = IB_LID_PERMISSIVE;
 	}
-	cm_req_set_primary_flow_label(req_msg, pri_path->flow_label);
+	IBA_SET(CM_REQ_PRIMARY_FLOW_LABEL, req_msg,
+		be32_to_cpu(pri_path->flow_label));
 	IBA_SET(CM_REQ_PRIMARY_PACKET_RATE, req_msg, pri_path->rate);
 	req_msg->primary_traffic_class = pri_path->traffic_class;
 	req_msg->primary_hop_limit = pri_path->hop_limit;
@@ -1330,8 +1331,8 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 			req_msg->alt_local_lid = IB_LID_PERMISSIVE;
 			req_msg->alt_remote_lid = IB_LID_PERMISSIVE;
 		}
-		cm_req_set_alt_flow_label(req_msg,
-					  alt_path->flow_label);
+		IBA_SET(CM_REQ_ALTERNATE_FLOW_LABEL, req_msg,
+			be32_to_cpu(alt_path->flow_label));
 		IBA_SET(CM_REQ_ALTERNATE_PACKET_RATE, req_msg, alt_path->rate);
 		req_msg->alt_traffic_class = alt_path->traffic_class;
 		req_msg->alt_hop_limit = alt_path->hop_limit;
@@ -1437,8 +1438,8 @@ int ib_send_cm_req(struct ib_cm_id *cm_id,
 	cm_id_priv->msg->timeout_ms = cm_id_priv->timeout_ms;
 	cm_id_priv->msg->context[1] = (void *) (unsigned long) IB_CM_REQ_SENT;
 
-	cm_id_priv->local_qpn = cm_req_get_local_qpn(req_msg);
-	cm_id_priv->rq_psn = cm_req_get_starting_psn(req_msg);
+	cm_id_priv->local_qpn = cpu_to_be32(IBA_GET(CM_REQ_LOCAL_QPN, req_msg));
+	cm_id_priv->rq_psn = cpu_to_be32(IBA_GET(CM_REQ_STARTING_PSN, req_msg));
 
 	spin_lock_irqsave(&cm_id_priv->lock, flags);
 	ret = ib_post_send_mad(cm_id_priv->msg, NULL);
@@ -1548,7 +1549,8 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 {
 	primary_path->dgid = req_msg->primary_local_gid;
 	primary_path->sgid = req_msg->primary_remote_gid;
-	primary_path->flow_label = cm_req_get_primary_flow_label(req_msg);
+	primary_path->flow_label =
+		cpu_to_be32(IBA_GET(CM_REQ_PRIMARY_FLOW_LABEL, req_msg));
 	primary_path->hop_limit = req_msg->primary_hop_limit;
 	primary_path->traffic_class = req_msg->primary_traffic_class;
 	primary_path->reversible = 1;
@@ -1569,7 +1571,8 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 	if (cm_req_has_alt_path(req_msg)) {
 		alt_path->dgid = req_msg->alt_local_gid;
 		alt_path->sgid = req_msg->alt_remote_gid;
-		alt_path->flow_label = cm_req_get_alt_flow_label(req_msg);
+		alt_path->flow_label = cpu_to_be32(
+			IBA_GET(CM_REQ_ALTERNATE_FLOW_LABEL, req_msg));
 		alt_path->hop_limit = req_msg->alt_hop_limit;
 		alt_path->traffic_class = req_msg->alt_traffic_class;
 		alt_path->reversible = 1;
@@ -1661,9 +1664,9 @@ static void cm_format_req_event(struct cm_work *work,
 	}
 	param->remote_ca_guid = req_msg->local_ca_guid;
 	param->remote_qkey = be32_to_cpu(req_msg->local_qkey);
-	param->remote_qpn = be32_to_cpu(cm_req_get_local_qpn(req_msg));
+	param->remote_qpn = IBA_GET(CM_REQ_LOCAL_QPN, req_msg);
 	param->qp_type = cm_req_get_qp_type(req_msg);
-	param->starting_psn = be32_to_cpu(cm_req_get_starting_psn(req_msg));
+	param->starting_psn = IBA_GET(CM_REQ_STARTING_PSN, req_msg);
 	param->responder_resources = IBA_GET(CM_REQ_INITIATOR_DEPTH, req_msg);
 	param->initiator_depth = IBA_GET(CM_REQ_RESPONDER_RESOURCES, req_msg);
 	param->local_cm_response_timeout =
@@ -1924,7 +1927,8 @@ static int cm_req_handler(struct cm_work *work)
 	}
 	cm_id_priv->timewait_info->work.remote_id = req_msg->local_comm_id;
 	cm_id_priv->timewait_info->remote_ca_guid = req_msg->local_ca_guid;
-	cm_id_priv->timewait_info->remote_qpn = cm_req_get_local_qpn(req_msg);
+	cm_id_priv->timewait_info->remote_qpn =
+		cpu_to_be32(IBA_GET(CM_REQ_LOCAL_QPN, req_msg));
 
 	listen_cm_id_priv = cm_match_req(work, cm_id_priv);
 	if (!listen_cm_id_priv) {
@@ -1998,14 +2002,15 @@ static int cm_req_handler(struct cm_work *work)
 	cm_id_priv->timeout_ms = cm_convert_to_ms(
 		IBA_GET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, req_msg));
 	cm_id_priv->max_cm_retries = IBA_GET(CM_REQ_MAX_CM_RETRIES, req_msg);
-	cm_id_priv->remote_qpn = cm_req_get_local_qpn(req_msg);
+	cm_id_priv->remote_qpn =
+		cpu_to_be32(IBA_GET(CM_REQ_LOCAL_QPN, req_msg));
 	cm_id_priv->initiator_depth =
 		IBA_GET(CM_REQ_RESPONDER_RESOURCES, req_msg);
 	cm_id_priv->responder_resources =
 		IBA_GET(CM_REQ_INITIATOR_DEPTH, req_msg);
 	cm_id_priv->path_mtu = IBA_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg);
 	cm_id_priv->pkey = req_msg->pkey;
-	cm_id_priv->sq_psn = cm_req_get_starting_psn(req_msg);
+	cm_id_priv->sq_psn = cpu_to_be32(IBA_GET(CM_REQ_STARTING_PSN, req_msg));
 	cm_id_priv->retry_count = IBA_GET(CM_REQ_RETRY_COUNT, req_msg);
 	cm_id_priv->rnr_retry_count = IBA_GET(CM_REQ_RNR_RETRY_COUNT, req_msg);
 	cm_id_priv->qp_type = cm_req_get_qp_type(req_msg);
@@ -2032,7 +2037,7 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg,
 	cm_format_mad_hdr(&rep_msg->hdr, CM_REP_ATTR_ID, cm_id_priv->tid);
 	rep_msg->local_comm_id = cm_id_priv->id.local_id;
 	rep_msg->remote_comm_id = cm_id_priv->id.remote_id;
-	cm_rep_set_starting_psn(rep_msg, cpu_to_be32(param->starting_psn));
+	IBA_SET(CM_REP_STARTING_PSN, rep_msg, param->starting_psn);
 	rep_msg->resp_resources = param->responder_resources;
 	IBA_SET(CM_REP_TARGET_ACK_DELAY, rep_msg,
 		cm_id_priv->av.port->cm_dev->ack_delay);
@@ -2045,10 +2050,10 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg,
 		IBA_SET(CM_REP_END_TO_END_FLOW_CONTROL, rep_msg,
 			param->flow_control);
 		IBA_SET(CM_REP_SRQ, rep_msg, param->srq);
-		cm_rep_set_local_qpn(rep_msg, cpu_to_be32(param->qp_num));
+		IBA_SET(CM_REP_LOCAL_QPN, rep_msg, param->qp_num);
 	} else {
 		IBA_SET(CM_REP_SRQ, rep_msg, 1);
-		cm_rep_set_local_eecn(rep_msg, cpu_to_be32(param->qp_num));
+		IBA_SET(CM_REP_LOCAL_EE_CONTEXT_NUMBER, rep_msg, param->qp_num);
 	}
 
 	if (param->private_data && param->private_data_len)
@@ -2099,7 +2104,7 @@ int ib_send_cm_rep(struct ib_cm_id *cm_id,
 	cm_id_priv->msg = msg;
 	cm_id_priv->initiator_depth = param->initiator_depth;
 	cm_id_priv->responder_resources = param->responder_resources;
-	cm_id_priv->rq_psn = cm_rep_get_starting_psn(rep_msg);
+	cm_id_priv->rq_psn = cpu_to_be32(IBA_GET(CM_REP_STARTING_PSN, rep_msg));
 	cm_id_priv->local_qpn = cpu_to_be32(param->qp_num & 0xFFFFFF);
 
 out:	spin_unlock_irqrestore(&cm_id_priv->lock, flags);
@@ -2183,7 +2188,7 @@ static void cm_format_rep_event(struct cm_work *work, enum ib_qp_type qp_type)
 	param->remote_ca_guid = rep_msg->local_ca_guid;
 	param->remote_qkey = be32_to_cpu(rep_msg->local_qkey);
 	param->remote_qpn = be32_to_cpu(cm_rep_get_qpn(rep_msg, qp_type));
-	param->starting_psn = be32_to_cpu(cm_rep_get_starting_psn(rep_msg));
+	param->starting_psn = IBA_GET(CM_REP_STARTING_PSN, rep_msg);
 	param->responder_resources = rep_msg->initiator_depth;
 	param->initiator_depth = rep_msg->resp_resources;
 	param->target_ack_delay = IBA_GET(CM_REP_TARGET_ACK_DELAY, rep_msg);
@@ -2320,7 +2325,7 @@ static int cm_rep_handler(struct cm_work *work)
 	cm_id_priv->remote_qpn = cm_rep_get_qpn(rep_msg, cm_id_priv->qp_type);
 	cm_id_priv->initiator_depth = rep_msg->resp_resources;
 	cm_id_priv->responder_resources = rep_msg->initiator_depth;
-	cm_id_priv->sq_psn = cm_rep_get_starting_psn(rep_msg);
+	cm_id_priv->sq_psn = cpu_to_be32(IBA_GET(CM_REP_STARTING_PSN, rep_msg));
 	cm_id_priv->rnr_retry_count = IBA_GET(CM_REP_RNR_RETRY_COUNT, rep_msg);
 	cm_id_priv->target_ack_delay =
 		IBA_GET(CM_REP_TARGET_ACK_DELAY, rep_msg);
@@ -2431,7 +2436,8 @@ static void cm_format_dreq(struct cm_dreq_msg *dreq_msg,
 			  cm_form_tid(cm_id_priv));
 	dreq_msg->local_comm_id = cm_id_priv->id.local_id;
 	dreq_msg->remote_comm_id = cm_id_priv->id.remote_id;
-	cm_dreq_set_remote_qpn(dreq_msg, cm_id_priv->remote_qpn);
+	IBA_SET(CM_DREQ_REMOTE_QPN_EECN, dreq_msg,
+		be32_to_cpu(cm_id_priv->remote_qpn));
 
 	if (private_data && private_data_len)
 		memcpy(dreq_msg->private_data, private_data, private_data_len);
@@ -2599,7 +2605,8 @@ static int cm_dreq_handler(struct cm_work *work)
 	work->cm_event.private_data = &dreq_msg->private_data;
 
 	spin_lock_irq(&cm_id_priv->lock);
-	if (cm_id_priv->local_qpn != cm_dreq_get_remote_qpn(dreq_msg))
+	if (cm_id_priv->local_qpn !=
+	    cpu_to_be32(IBA_GET(CM_DREQ_REMOTE_QPN_EECN, dreq_msg)))
 		goto unlock;
 
 	switch (cm_id_priv->id.state) {
@@ -3067,7 +3074,8 @@ static void cm_format_path_from_lap(struct cm_id_private *cm_id_priv,
 {
 	path->dgid = lap_msg->alt_local_gid;
 	path->sgid = lap_msg->alt_remote_gid;
-	path->flow_label = cm_lap_get_flow_label(lap_msg);
+	path->flow_label =
+		cpu_to_be32(IBA_GET(CM_LAP_ALTERNATE_FLOW_LABEL, lap_msg));
 	path->hop_limit = lap_msg->alt_hop_limit;
 	path->traffic_class = IBA_GET(CM_LAP_ALTERNATE_TRAFFIC_CLASS, lap_msg);
 	path->reversible = 1;
@@ -3423,7 +3431,7 @@ static void cm_format_sidr_rep(struct cm_sidr_rep_msg *sidr_rep_msg,
 			  cm_id_priv->tid);
 	sidr_rep_msg->request_id = cm_id_priv->id.remote_id;
 	sidr_rep_msg->status = param->status;
-	cm_sidr_rep_set_qpn(sidr_rep_msg, cpu_to_be32(param->qp_num));
+	IBA_SET(CM_SIDR_REP_QPN, sidr_rep_msg, param->qp_num);
 	sidr_rep_msg->service_id = cm_id_priv->id.service_id;
 	sidr_rep_msg->qkey = cpu_to_be32(param->qkey);
 
@@ -3494,7 +3502,7 @@ static void cm_format_sidr_rep_event(struct cm_work *work,
 	param = &work->cm_event.param.sidr_rep_rcvd;
 	param->status = sidr_rep_msg->status;
 	param->qkey = be32_to_cpu(sidr_rep_msg->qkey);
-	param->qpn = be32_to_cpu(cm_sidr_rep_get_qpn(sidr_rep_msg));
+	param->qpn = IBA_GET(CM_SIDR_REP_QPN, sidr_rep_msg);
 	param->info = &sidr_rep_msg->info;
 	param->info_len = sidr_rep_msg->info_length;
 	param->sgid_attr = cm_id_priv->av.ah_attr.grh.sgid_attr;
@@ -4346,138 +4354,10 @@ IBA_CHECK_OFF(CM_SIDR_REP_Q_KEY, qkey);
 IBA_CHECK_OFF(CM_SIDR_REP_ADDITIONAL_INFORMATION, info);
 IBA_CHECK_OFF(CM_SIDR_REP_PRIVATE_DATA, private_data);
 
-/*
- * Check that the new macro gets the same bits as the old get function.
- *  - IBA_SET() IBA_GET and old get_fn all agree on the field width.
- *    The field width should match what IBA_SET truncates to
- *  - Reading from an all ones data should not return extra bits
- *  - Setting '1' should be the same (ie no endian problems)
- */
-/* defeat builtin_constant checks */
-u64 cm_global_all_ones = 0xffffffffffffffffULL;
-#define _IBA_CHECK_GET(fn, field_struct, field_offset, mask, bits)             \
-	({                                                                     \
-		field_struct *lmsg = (field_struct *)msg;                      \
-		unsigned long long all_ones;                                   \
-		static_assert(sizeof(*lmsg) <= sizeof(msg));                   \
-                                                                               \
-		bitmap_zero(msg, nbits);                                       \
-		_IBA_SET(field_struct, field_offset, mask, bits, lmsg,         \
-			 cm_global_all_ones);                                  \
-		all_ones = (1ULL << bitmap_weight(msg, nbits)) - 1;            \
-		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
-		    all_ones) {                                                \
-			printk("Failed #1 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-		if (fn != all_ones) {                                          \
-			printk("Failed #2 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-                                                                               \
-		bitmap_fill(msg, nbits);                                       \
-		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
-		    all_ones) {                                                \
-			printk("Failed #3 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-		if (fn != all_ones) {                                          \
-			printk("Failed #4 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-                                                                               \
-		_IBA_SET(field_struct, field_offset, mask, bits, lmsg, 0);     \
-		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
-		    0) {                                                       \
-			printk("Failed #5 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-		if (fn != 0) {                                                 \
-			printk("Failed #6 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-		_IBA_SET(field_struct, field_offset, mask, bits, lmsg, 1);     \
-		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
-		    1) {                                                       \
-			printk("Failed #7 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-		if (fn != 1) {                                                 \
-			printk("Failed #8 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-	})
-#define IBA_CHECK_GET(field, fn_name) _IBA_CHECK_GET(fn_name(lmsg), field)
-#define IBA_CHECK_GET_BE(field, fn_name) _IBA_CHECK_GET(be32_to_cpu(fn_name(lmsg)), field)
-
-/*
- * Write the all ones value using the old setter and check that the new getter
- * reads it back.
- */
-#define _IBA_CHECK_SET(fn, field_struct, field_offset, mask, bits)             \
-	({                                                                     \
-		field_struct *lmsg = (field_struct *)msg;                      \
-		unsigned long long all_ones;                                   \
-		static_assert(sizeof(*lmsg) <= sizeof(msg));                   \
-                                                                               \
-		bitmap_zero(msg, nbits);                                       \
-		_IBA_SET(field_struct, field_offset, mask, bits, lmsg,         \
-			 cm_global_all_ones);                                  \
-		all_ones = (1ULL << bitmap_weight(msg, nbits)) - 1;            \
-		bitmap_zero(msg, nbits);                                       \
-		fn;                                                            \
-		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
-		    all_ones) {                                                \
-			printk("Failed #9 line=%u\n", __LINE__);               \
-			return;                                                \
-		}                                                              \
-		all_ones = 1;                                                  \
-		fn;                                                            \
-		if (_IBA_GET(field_struct, field_offset, mask, bits, lmsg) !=  \
-		    1) {                                                       \
-			printk("Failed #10 line=%u\n", __LINE__);              \
-			return;                                                \
-		}                                                              \
-	})
-
-#define IBA_CHECK_SET(field, fn_name) _IBA_CHECK_SET(fn_name(lmsg, all_ones), field)
-#define IBA_CHECK_SET_BE(field, fn_name)                                       \
-	_IBA_CHECK_SET(fn_name(lmsg, cpu_to_be32(all_ones)), field)
-
-static void self_test(void)
-{
-	unsigned long msg[256/4];
-	const unsigned int nbits = sizeof(msg) * 8;
-
-	printk("Running CM extractor self test\n");
-	IBA_CHECK_GET_BE(CM_REQ_LOCAL_QPN, cm_req_get_local_qpn);
-	IBA_CHECK_SET_BE(CM_REQ_LOCAL_QPN, cm_req_set_local_qpn);
-	IBA_CHECK_GET_BE(CM_REQ_STARTING_PSN, cm_req_get_starting_psn);
-	IBA_CHECK_SET_BE(CM_REQ_STARTING_PSN, cm_req_set_starting_psn);
-	IBA_CHECK_GET_BE(CM_REQ_PRIMARY_FLOW_LABEL, cm_req_get_primary_flow_label);
-	IBA_CHECK_SET_BE(CM_REQ_PRIMARY_FLOW_LABEL, cm_req_set_primary_flow_label);
-	IBA_CHECK_GET_BE(CM_REQ_ALTERNATE_FLOW_LABEL, cm_req_get_alt_flow_label);
-	IBA_CHECK_SET_BE(CM_REQ_ALTERNATE_FLOW_LABEL, cm_req_set_alt_flow_label);
-	IBA_CHECK_GET_BE(CM_REP_LOCAL_QPN, cm_rep_get_local_qpn);
-	IBA_CHECK_SET_BE(CM_REP_LOCAL_QPN, cm_rep_set_local_qpn);
-	IBA_CHECK_GET_BE(CM_REP_LOCAL_EE_CONTEXT_NUMBER, cm_rep_get_local_eecn);
-	IBA_CHECK_SET_BE(CM_REP_LOCAL_EE_CONTEXT_NUMBER, cm_rep_set_local_eecn);
-	IBA_CHECK_GET_BE(CM_REP_STARTING_PSN, cm_rep_get_starting_psn);
-	IBA_CHECK_SET_BE(CM_REP_STARTING_PSN, cm_rep_set_starting_psn);
-	IBA_CHECK_GET_BE(CM_DREQ_REMOTE_QPN_EECN, cm_dreq_get_remote_qpn);
-	IBA_CHECK_SET_BE(CM_DREQ_REMOTE_QPN_EECN, cm_dreq_set_remote_qpn);
-	IBA_CHECK_GET_BE(CM_LAP_ALTERNATE_FLOW_LABEL, cm_lap_get_flow_label);
-	IBA_CHECK_GET_BE(CM_SIDR_REP_QPN, cm_sidr_rep_get_qpn);
-	IBA_CHECK_SET_BE(CM_SIDR_REP_QPN, cm_sidr_rep_set_qpn);
-	printk("Success!\n");
-}
-
 static int __init ib_cm_init(void)
 {
 	int ret;
 
-	self_test();
-
 	INIT_LIST_HEAD(&cm.device_list);
 	rwlock_init(&cm.device_lock);
 	spin_lock_init(&cm.lock);
diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h
index d30586b1b8a4..86ab6952d5d8 100644
--- a/drivers/infiniband/core/cm_msgs.h
+++ b/drivers/infiniband/core/cm_msgs.h
@@ -75,18 +75,6 @@ struct cm_req_msg {
 
 } __packed;
 
-static inline __be32 cm_req_get_local_qpn(struct cm_req_msg *req_msg)
-{
-	return cpu_to_be32(be32_to_cpu(req_msg->offset32) >> 8);
-}
-
-static inline void cm_req_set_local_qpn(struct cm_req_msg *req_msg, __be32 qpn)
-{
-	req_msg->offset32 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
-					 (be32_to_cpu(req_msg->offset32) &
-					  0x000000FF));
-}
-
 static inline enum ib_qp_type cm_req_get_qp_type(struct cm_req_msg *req_msg)
 {
 	u8 transport_type = IBA_GET(CM_REQ_TRANSPORT_SERVICE_TYPE, req_msg);
@@ -118,46 +106,6 @@ static inline void cm_req_set_qp_type(struct cm_req_msg *req_msg,
 	}
 }
 
-static inline __be32 cm_req_get_starting_psn(struct cm_req_msg *req_msg)
-{
-	return cpu_to_be32(be32_to_cpu(req_msg->offset44) >> 8);
-}
-
-static inline void cm_req_set_starting_psn(struct cm_req_msg *req_msg,
-					   __be32 starting_psn)
-{
-	req_msg->offset44 = cpu_to_be32((be32_to_cpu(starting_psn) << 8) |
-			    (be32_to_cpu(req_msg->offset44) & 0x000000FF));
-}
-
-static inline __be32 cm_req_get_primary_flow_label(struct cm_req_msg *req_msg)
-{
-	return cpu_to_be32(be32_to_cpu(req_msg->primary_offset88) >> 12);
-}
-
-static inline void cm_req_set_primary_flow_label(struct cm_req_msg *req_msg,
-						 __be32 flow_label)
-{
-	req_msg->primary_offset88 = cpu_to_be32(
-				    (be32_to_cpu(req_msg->primary_offset88) &
-				     0x00000FFF) |
-				     (be32_to_cpu(flow_label) << 12));
-}
-
-static inline __be32 cm_req_get_alt_flow_label(struct cm_req_msg *req_msg)
-{
-	return cpu_to_be32(be32_to_cpu(req_msg->alt_offset132) >> 12);
-}
-
-static inline void cm_req_set_alt_flow_label(struct cm_req_msg *req_msg,
-					     __be32 flow_label)
-{
-	req_msg->alt_offset132 = cpu_to_be32(
-				 (be32_to_cpu(req_msg->alt_offset132) &
-				  0x00000FFF) |
-				  (be32_to_cpu(flow_label) << 12));
-}
-
 /* Message REJected or MRAed */
 enum cm_msg_response {
 	CM_MSG_RESPONSE_REQ = 0x0,
@@ -219,44 +167,12 @@ struct cm_rep_msg {
 
 } __packed;
 
-static inline __be32 cm_rep_get_local_qpn(struct cm_rep_msg *rep_msg)
-{
-	return cpu_to_be32(be32_to_cpu(rep_msg->offset12) >> 8);
-}
-
-static inline void cm_rep_set_local_qpn(struct cm_rep_msg *rep_msg, __be32 qpn)
-{
-	rep_msg->offset12 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
-			    (be32_to_cpu(rep_msg->offset12) & 0x000000FF));
-}
-
-static inline __be32 cm_rep_get_local_eecn(struct cm_rep_msg *rep_msg)
-{
-	return cpu_to_be32(be32_to_cpu(rep_msg->offset16) >> 8);
-}
-
-static inline void cm_rep_set_local_eecn(struct cm_rep_msg *rep_msg, __be32 eecn)
-{
-	rep_msg->offset16 = cpu_to_be32((be32_to_cpu(eecn) << 8) |
-			    (be32_to_cpu(rep_msg->offset16) & 0x000000FF));
-}
-
 static inline __be32 cm_rep_get_qpn(struct cm_rep_msg *rep_msg, enum ib_qp_type qp_type)
 {
 	return (qp_type == IB_QPT_XRC_INI) ?
-		cm_rep_get_local_eecn(rep_msg) : cm_rep_get_local_qpn(rep_msg);
-}
-
-static inline __be32 cm_rep_get_starting_psn(struct cm_rep_msg *rep_msg)
-{
-	return cpu_to_be32(be32_to_cpu(rep_msg->offset20) >> 8);
-}
-
-static inline void cm_rep_set_starting_psn(struct cm_rep_msg *rep_msg,
-					   __be32 starting_psn)
-{
-	rep_msg->offset20 = cpu_to_be32((be32_to_cpu(starting_psn) << 8) |
-			    (be32_to_cpu(rep_msg->offset20) & 0x000000FF));
+		       cpu_to_be32(IBA_GET(CM_REP_LOCAL_EE_CONTEXT_NUMBER,
+					   rep_msg)) :
+		       cpu_to_be32(IBA_GET(CM_REP_LOCAL_QPN, rep_msg));
 }
 
 struct cm_rtu_msg {
@@ -281,17 +197,6 @@ struct cm_dreq_msg {
 
 } __packed;
 
-static inline __be32 cm_dreq_get_remote_qpn(struct cm_dreq_msg *dreq_msg)
-{
-	return cpu_to_be32(be32_to_cpu(dreq_msg->offset8) >> 8);
-}
-
-static inline void cm_dreq_set_remote_qpn(struct cm_dreq_msg *dreq_msg, __be32 qpn)
-{
-	dreq_msg->offset8 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
-			    (be32_to_cpu(dreq_msg->offset8) & 0x000000FF));
-}
-
 struct cm_drep_msg {
 	struct ib_mad_hdr hdr;
 
@@ -330,11 +235,6 @@ struct cm_lap_msg {
 	u8 private_data[IB_CM_LAP_PRIVATE_DATA_SIZE];
 } __packed;
 
-static inline __be32 cm_lap_get_flow_label(struct cm_lap_msg *lap_msg)
-{
-	return cpu_to_be32(be32_to_cpu(lap_msg->offset56) >> 12);
-}
-
 struct cm_apr_msg {
 	struct ib_mad_hdr hdr;
 
@@ -376,17 +276,4 @@ struct cm_sidr_rep_msg {
 	u8 private_data[IB_CM_SIDR_REP_PRIVATE_DATA_SIZE];
 } __packed;
 
-static inline __be32 cm_sidr_rep_get_qpn(struct cm_sidr_rep_msg *sidr_rep_msg)
-{
-	return cpu_to_be32(be32_to_cpu(sidr_rep_msg->offset8) >> 8);
-}
-
-static inline void cm_sidr_rep_set_qpn(struct cm_sidr_rep_msg *sidr_rep_msg,
-				       __be32 qpn)
-{
-	sidr_rep_msg->offset8 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
-					(be32_to_cpu(sidr_rep_msg->offset8) &
-					 0x000000FF));
-}
-
 #endif /* CM_MSGS_H */
-- 
cgit v1.2.3-59-g8ed1b


From 91b60a7128d96244794beb9b324eb39273872da2 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Thu, 16 Jan 2020 13:00:35 -0400
Subject: RDMA/cm: Use IBA functions for simple structure members

Use a Coccinelle spatch script to replace use of simple CM structure
members with IBA_GET/SET versions. Applied with

$ spatch --sp-file edits.sp --in-place drivers/infiniband/core/cm.c

The spatch file was generated using the template pattern:

@@
expression val;
{struct} *msg;
@@
- msg->{old_name} = val
+ IBA_SET({new_name}, msg, be{bits}_to_cpu(val))
@@
{struct} *msg;
@@
- msg->{old_name}
+ cpu_to_be{bits}(IBA_GET({new_name}, msg))

Iterated for every IBA_CHECK_OFF that isn't a CM_FIELD_MLOC.

And the below iterated over all byte sizes to remove doubled byte swaps:

@@
expression val;
@@
-be{bits}_to_cpu(cpu_to_be{bits}(val))
+val

(and __be_to_cpu and ntoh varients)

Touched up with clang-format after.

Link: https://lore.kernel.org/r/20200116170037.30109-6-jgg@ziepe.ca
Tested-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c | 384 +++++++++++++++++++++++++++----------------
 1 file changed, 245 insertions(+), 139 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 976bb85b6aa6..047ee560046e 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -1251,9 +1251,11 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 	cm_format_mad_hdr(&req_msg->hdr, CM_REQ_ATTR_ID,
 			  cm_form_tid(cm_id_priv));
 
-	req_msg->local_comm_id = cm_id_priv->id.local_id;
-	req_msg->service_id = param->service_id;
-	req_msg->local_ca_guid = cm_id_priv->id.device->node_guid;
+	IBA_SET(CM_REQ_LOCAL_COMM_ID, req_msg,
+		be32_to_cpu(cm_id_priv->id.local_id));
+	IBA_SET(CM_REQ_SERVICE_ID, req_msg, be64_to_cpu(param->service_id));
+	IBA_SET(CM_REQ_LOCAL_CA_GUID, req_msg,
+		be64_to_cpu(cm_id_priv->id.device->node_guid));
 	IBA_SET(CM_REQ_LOCAL_QPN, req_msg, param->qp_num);
 	IBA_SET(CM_REQ_INITIATOR_DEPTH, req_msg, param->initiator_depth);
 	IBA_SET(CM_REQ_REMOTE_CM_RESPONSE_TIMEOUT, req_msg,
@@ -1263,7 +1265,8 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 	IBA_SET(CM_REQ_STARTING_PSN, req_msg, param->starting_psn);
 	IBA_SET(CM_REQ_LOCAL_CM_RESPONSE_TIMEOUT, req_msg,
 		param->local_cm_response_timeout);
-	req_msg->pkey = param->primary_path->pkey;
+	IBA_SET(CM_REQ_PARTITION_KEY, req_msg,
+		be16_to_cpu(param->primary_path->pkey));
 	IBA_SET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg,
 		param->primary_path->mtu);
 	IBA_SET(CM_REQ_MAX_CM_RETRIES, req_msg, param->max_cm_retries);
@@ -1286,20 +1289,26 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 			= OPA_MAKE_ID(be32_to_cpu(pri_path->opa.dlid));
 	}
 	if (pri_path->hop_limit <= 1) {
-		req_msg->primary_local_lid = pri_ext ? 0 :
-			htons(ntohl(sa_path_get_slid(pri_path)));
-		req_msg->primary_remote_lid = pri_ext ? 0 :
-			htons(ntohl(sa_path_get_dlid(pri_path)));
+		IBA_SET(CM_REQ_PRIMARY_LOCAL_PORT_LID, req_msg,
+			be16_to_cpu(pri_ext ? 0 :
+					      htons(ntohl(sa_path_get_slid(
+						      pri_path)))));
+		IBA_SET(CM_REQ_PRIMARY_REMOTE_PORT_LID, req_msg,
+			be16_to_cpu(pri_ext ? 0 :
+					      htons(ntohl(sa_path_get_dlid(
+						      pri_path)))));
 	} else {
 		/* Work-around until there's a way to obtain remote LID info */
-		req_msg->primary_local_lid = IB_LID_PERMISSIVE;
-		req_msg->primary_remote_lid = IB_LID_PERMISSIVE;
+		IBA_SET(CM_REQ_PRIMARY_LOCAL_PORT_LID, req_msg,
+			be16_to_cpu(IB_LID_PERMISSIVE));
+		IBA_SET(CM_REQ_PRIMARY_REMOTE_PORT_LID, req_msg,
+			be16_to_cpu(IB_LID_PERMISSIVE));
 	}
 	IBA_SET(CM_REQ_PRIMARY_FLOW_LABEL, req_msg,
 		be32_to_cpu(pri_path->flow_label));
 	IBA_SET(CM_REQ_PRIMARY_PACKET_RATE, req_msg, pri_path->rate);
-	req_msg->primary_traffic_class = pri_path->traffic_class;
-	req_msg->primary_hop_limit = pri_path->hop_limit;
+	IBA_SET(CM_REQ_PRIMARY_TRAFFIC_CLASS, req_msg, pri_path->traffic_class);
+	IBA_SET(CM_REQ_PRIMARY_HOP_LIMIT, req_msg, pri_path->hop_limit);
 	IBA_SET(CM_REQ_PRIMARY_SL, req_msg, pri_path->sl);
 	IBA_SET(CM_REQ_PRIMARY_SUBNET_LOCAL, req_msg,
 		(pri_path->hop_limit <= 1));
@@ -1323,19 +1332,29 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 				= OPA_MAKE_ID(be32_to_cpu(alt_path->opa.dlid));
 		}
 		if (alt_path->hop_limit <= 1) {
-			req_msg->alt_local_lid = alt_ext ? 0 :
-				htons(ntohl(sa_path_get_slid(alt_path)));
-			req_msg->alt_remote_lid = alt_ext ? 0 :
-				htons(ntohl(sa_path_get_dlid(alt_path)));
+			IBA_SET(CM_REQ_ALTERNATE_LOCAL_PORT_LID, req_msg,
+				be16_to_cpu(
+					alt_ext ? 0 :
+						  htons(ntohl(sa_path_get_slid(
+							  alt_path)))));
+			IBA_SET(CM_REQ_ALTERNATE_REMOTE_PORT_LID, req_msg,
+				be16_to_cpu(
+					alt_ext ? 0 :
+						  htons(ntohl(sa_path_get_dlid(
+							  alt_path)))));
 		} else {
-			req_msg->alt_local_lid = IB_LID_PERMISSIVE;
-			req_msg->alt_remote_lid = IB_LID_PERMISSIVE;
+			IBA_SET(CM_REQ_ALTERNATE_LOCAL_PORT_LID, req_msg,
+				be16_to_cpu(IB_LID_PERMISSIVE));
+			IBA_SET(CM_REQ_ALTERNATE_REMOTE_PORT_LID, req_msg,
+				be16_to_cpu(IB_LID_PERMISSIVE));
 		}
 		IBA_SET(CM_REQ_ALTERNATE_FLOW_LABEL, req_msg,
 			be32_to_cpu(alt_path->flow_label));
 		IBA_SET(CM_REQ_ALTERNATE_PACKET_RATE, req_msg, alt_path->rate);
-		req_msg->alt_traffic_class = alt_path->traffic_class;
-		req_msg->alt_hop_limit = alt_path->hop_limit;
+		IBA_SET(CM_REQ_ALTERNATE_TRAFFIC_CLASS, req_msg,
+			alt_path->traffic_class);
+		IBA_SET(CM_REQ_ALTERNATE_HOP_LIMIT, req_msg,
+			alt_path->hop_limit);
 		IBA_SET(CM_REQ_ALTERNATE_SL, req_msg, alt_path->sl);
 		IBA_SET(CM_REQ_ALTERNATE_SUBNET_LOCAL, req_msg,
 			(alt_path->hop_limit <= 1));
@@ -1477,10 +1496,12 @@ static int cm_issue_rej(struct cm_port *port,
 	rej_msg = (struct cm_rej_msg *) msg->mad;
 
 	cm_format_mad_hdr(&rej_msg->hdr, CM_REJ_ATTR_ID, rcv_msg->hdr.tid);
-	rej_msg->remote_comm_id = rcv_msg->local_comm_id;
-	rej_msg->local_comm_id = rcv_msg->remote_comm_id;
+	IBA_SET(CM_REJ_REMOTE_COMM_ID, rej_msg,
+		IBA_GET(CM_REJ_LOCAL_COMM_ID, rcv_msg));
+	IBA_SET(CM_REJ_LOCAL_COMM_ID, rej_msg,
+		IBA_GET(CM_REJ_REMOTE_COMM_ID, rcv_msg));
 	IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, msg_rejected);
-	rej_msg->reason = cpu_to_be16(reason);
+	IBA_SET(CM_REJ_REASON, rej_msg, reason);
 
 	if (ari && ari_length) {
 		IBA_SET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg, ari_length);
@@ -1496,7 +1517,8 @@ static int cm_issue_rej(struct cm_port *port,
 
 static bool cm_req_has_alt_path(struct cm_req_msg *req_msg)
 {
-	return ((req_msg->alt_local_lid) ||
+	return ((cpu_to_be16(
+			IBA_GET(CM_REQ_ALTERNATE_LOCAL_PORT_LID, req_msg))) ||
 		(ib_is_opa_gid(&req_msg->alt_local_gid)));
 }
 
@@ -1517,9 +1539,11 @@ static void cm_format_path_lid_from_req(struct cm_req_msg *req_msg,
 
 	if (primary_path->rec_type != SA_PATH_REC_TYPE_OPA) {
 		sa_path_set_dlid(primary_path,
-				 ntohs(req_msg->primary_local_lid));
+				 IBA_GET(CM_REQ_PRIMARY_LOCAL_PORT_LID,
+					 req_msg));
 		sa_path_set_slid(primary_path,
-				 ntohs(req_msg->primary_remote_lid));
+				 IBA_GET(CM_REQ_PRIMARY_REMOTE_PORT_LID,
+					 req_msg));
 	} else {
 		lid = opa_get_lid_from_gid(&req_msg->primary_local_gid);
 		sa_path_set_dlid(primary_path, lid);
@@ -1532,8 +1556,12 @@ static void cm_format_path_lid_from_req(struct cm_req_msg *req_msg,
 		return;
 
 	if (alt_path->rec_type != SA_PATH_REC_TYPE_OPA) {
-		sa_path_set_dlid(alt_path, ntohs(req_msg->alt_local_lid));
-		sa_path_set_slid(alt_path, ntohs(req_msg->alt_remote_lid));
+		sa_path_set_dlid(alt_path,
+				 IBA_GET(CM_REQ_ALTERNATE_LOCAL_PORT_LID,
+					 req_msg));
+		sa_path_set_slid(alt_path,
+				 IBA_GET(CM_REQ_ALTERNATE_REMOTE_PORT_LID,
+					 req_msg));
 	} else {
 		lid = opa_get_lid_from_gid(&req_msg->alt_local_gid);
 		sa_path_set_dlid(alt_path, lid);
@@ -1551,10 +1579,12 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 	primary_path->sgid = req_msg->primary_remote_gid;
 	primary_path->flow_label =
 		cpu_to_be32(IBA_GET(CM_REQ_PRIMARY_FLOW_LABEL, req_msg));
-	primary_path->hop_limit = req_msg->primary_hop_limit;
-	primary_path->traffic_class = req_msg->primary_traffic_class;
+	primary_path->hop_limit = IBA_GET(CM_REQ_PRIMARY_HOP_LIMIT, req_msg);
+	primary_path->traffic_class =
+		IBA_GET(CM_REQ_PRIMARY_TRAFFIC_CLASS, req_msg);
 	primary_path->reversible = 1;
-	primary_path->pkey = req_msg->pkey;
+	primary_path->pkey =
+		cpu_to_be16(IBA_GET(CM_REQ_PARTITION_KEY, req_msg));
 	primary_path->sl = IBA_GET(CM_REQ_PRIMARY_SL, req_msg);
 	primary_path->mtu_selector = IB_SA_EQ;
 	primary_path->mtu = IBA_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg);
@@ -1564,7 +1594,8 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 	primary_path->packet_life_time =
 		IBA_GET(CM_REQ_PRIMARY_LOCAL_ACK_TIMEOUT, req_msg);
 	primary_path->packet_life_time -= (primary_path->packet_life_time > 0);
-	primary_path->service_id = req_msg->service_id;
+	primary_path->service_id =
+		cpu_to_be64(IBA_GET(CM_REQ_SERVICE_ID, req_msg));
 	if (sa_path_is_roce(primary_path))
 		primary_path->roce.route_resolved = false;
 
@@ -1573,10 +1604,13 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 		alt_path->sgid = req_msg->alt_remote_gid;
 		alt_path->flow_label = cpu_to_be32(
 			IBA_GET(CM_REQ_ALTERNATE_FLOW_LABEL, req_msg));
-		alt_path->hop_limit = req_msg->alt_hop_limit;
-		alt_path->traffic_class = req_msg->alt_traffic_class;
+		alt_path->hop_limit =
+			IBA_GET(CM_REQ_ALTERNATE_HOP_LIMIT, req_msg);
+		alt_path->traffic_class =
+			IBA_GET(CM_REQ_ALTERNATE_TRAFFIC_CLASS, req_msg);
 		alt_path->reversible = 1;
-		alt_path->pkey = req_msg->pkey;
+		alt_path->pkey =
+			cpu_to_be16(IBA_GET(CM_REQ_PARTITION_KEY, req_msg));
 		alt_path->sl = IBA_GET(CM_REQ_ALTERNATE_SL, req_msg);
 		alt_path->mtu_selector = IB_SA_EQ;
 		alt_path->mtu =
@@ -1587,7 +1621,8 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 		alt_path->packet_life_time =
 			IBA_GET(CM_REQ_ALTERNATE_LOCAL_ACK_TIMEOUT, req_msg);
 		alt_path->packet_life_time -= (alt_path->packet_life_time > 0);
-		alt_path->service_id = req_msg->service_id;
+		alt_path->service_id =
+			cpu_to_be64(IBA_GET(CM_REQ_SERVICE_ID, req_msg));
 
 		if (sa_path_is_roce(alt_path))
 			alt_path->roce.route_resolved = false;
@@ -1662,8 +1697,9 @@ static void cm_format_req_event(struct cm_work *work,
 	} else {
 		param->alternate_path = NULL;
 	}
-	param->remote_ca_guid = req_msg->local_ca_guid;
-	param->remote_qkey = be32_to_cpu(req_msg->local_qkey);
+	param->remote_ca_guid =
+		cpu_to_be64(IBA_GET(CM_REQ_LOCAL_CA_GUID, req_msg));
+	param->remote_qkey = IBA_GET(CM_REQ_LOCAL_Q_KEY, req_msg);
 	param->remote_qpn = IBA_GET(CM_REQ_LOCAL_QPN, req_msg);
 	param->qp_type = cm_req_get_qp_type(req_msg);
 	param->starting_psn = IBA_GET(CM_REQ_STARTING_PSN, req_msg);
@@ -1713,8 +1749,10 @@ static void cm_format_mra(struct cm_mra_msg *mra_msg,
 {
 	cm_format_mad_hdr(&mra_msg->hdr, CM_MRA_ATTR_ID, cm_id_priv->tid);
 	IBA_SET(CM_MRA_MESSAGE_MRAED, mra_msg, msg_mraed);
-	mra_msg->local_comm_id = cm_id_priv->id.local_id;
-	mra_msg->remote_comm_id = cm_id_priv->id.remote_id;
+	IBA_SET(CM_MRA_LOCAL_COMM_ID, mra_msg,
+		be32_to_cpu(cm_id_priv->id.local_id));
+	IBA_SET(CM_MRA_REMOTE_COMM_ID, mra_msg,
+		be32_to_cpu(cm_id_priv->id.remote_id));
 	IBA_SET(CM_MRA_SERVICE_TIMEOUT, mra_msg, service_timeout);
 
 	if (private_data && private_data_len)
@@ -1730,30 +1768,34 @@ static void cm_format_rej(struct cm_rej_msg *rej_msg,
 			  u8 private_data_len)
 {
 	cm_format_mad_hdr(&rej_msg->hdr, CM_REJ_ATTR_ID, cm_id_priv->tid);
-	rej_msg->remote_comm_id = cm_id_priv->id.remote_id;
+	IBA_SET(CM_REJ_REMOTE_COMM_ID, rej_msg,
+		be32_to_cpu(cm_id_priv->id.remote_id));
 
 	switch(cm_id_priv->id.state) {
 	case IB_CM_REQ_RCVD:
-		rej_msg->local_comm_id = 0;
+		IBA_SET(CM_REJ_LOCAL_COMM_ID, rej_msg, be32_to_cpu(0));
 		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, CM_MSG_RESPONSE_REQ);
 		break;
 	case IB_CM_MRA_REQ_SENT:
-		rej_msg->local_comm_id = cm_id_priv->id.local_id;
+		IBA_SET(CM_REJ_LOCAL_COMM_ID, rej_msg,
+			be32_to_cpu(cm_id_priv->id.local_id));
 		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, CM_MSG_RESPONSE_REQ);
 		break;
 	case IB_CM_REP_RCVD:
 	case IB_CM_MRA_REP_SENT:
-		rej_msg->local_comm_id = cm_id_priv->id.local_id;
+		IBA_SET(CM_REJ_LOCAL_COMM_ID, rej_msg,
+			be32_to_cpu(cm_id_priv->id.local_id));
 		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg, CM_MSG_RESPONSE_REP);
 		break;
 	default:
-		rej_msg->local_comm_id = cm_id_priv->id.local_id;
+		IBA_SET(CM_REJ_LOCAL_COMM_ID, rej_msg,
+			be32_to_cpu(cm_id_priv->id.local_id));
 		IBA_SET(CM_REJ_MESSAGE_REJECTED, rej_msg,
 			CM_MSG_RESPONSE_OTHER);
 		break;
 	}
 
-	rej_msg->reason = cpu_to_be16(reason);
+	IBA_SET(CM_REJ_REASON, rej_msg, reason);
 	if (ari && ari_length) {
 		IBA_SET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg, ari_length);
 		memcpy(rej_msg->ari, ari, ari_length);
@@ -1850,8 +1892,9 @@ static struct cm_id_private * cm_match_req(struct cm_work *work,
 	}
 
 	/* Find matching listen request. */
-	listen_cm_id_priv = cm_find_listen(cm_id_priv->id.device,
-					   req_msg->service_id);
+	listen_cm_id_priv = cm_find_listen(
+		cm_id_priv->id.device,
+		cpu_to_be64(IBA_GET(CM_REQ_SERVICE_ID, req_msg)));
 	if (!listen_cm_id_priv) {
 		cm_cleanup_timewait(cm_id_priv->timewait_info);
 		spin_unlock_irq(&cm.lock);
@@ -1877,23 +1920,31 @@ out:
 static void cm_process_routed_req(struct cm_req_msg *req_msg, struct ib_wc *wc)
 {
 	if (!IBA_GET(CM_REQ_PRIMARY_SUBNET_LOCAL, req_msg)) {
-		if (req_msg->primary_local_lid == IB_LID_PERMISSIVE) {
-			req_msg->primary_local_lid = ib_lid_be16(wc->slid);
+		if (cpu_to_be16(IBA_GET(CM_REQ_PRIMARY_LOCAL_PORT_LID,
+					req_msg)) == IB_LID_PERMISSIVE) {
+			IBA_SET(CM_REQ_PRIMARY_LOCAL_PORT_LID, req_msg,
+				be16_to_cpu(ib_lid_be16(wc->slid)));
 			IBA_SET(CM_REQ_PRIMARY_SL, req_msg, wc->sl);
 		}
 
-		if (req_msg->primary_remote_lid == IB_LID_PERMISSIVE)
-			req_msg->primary_remote_lid = cpu_to_be16(wc->dlid_path_bits);
+		if (cpu_to_be16(IBA_GET(CM_REQ_PRIMARY_REMOTE_PORT_LID,
+					req_msg)) == IB_LID_PERMISSIVE)
+			IBA_SET(CM_REQ_PRIMARY_REMOTE_PORT_LID, req_msg,
+				wc->dlid_path_bits);
 	}
 
 	if (!IBA_GET(CM_REQ_ALTERNATE_SUBNET_LOCAL, req_msg)) {
-		if (req_msg->alt_local_lid == IB_LID_PERMISSIVE) {
-			req_msg->alt_local_lid = ib_lid_be16(wc->slid);
+		if (cpu_to_be16(IBA_GET(CM_REQ_ALTERNATE_LOCAL_PORT_LID,
+					req_msg)) == IB_LID_PERMISSIVE) {
+			IBA_SET(CM_REQ_ALTERNATE_LOCAL_PORT_LID, req_msg,
+				be16_to_cpu(ib_lid_be16(wc->slid)));
 			IBA_SET(CM_REQ_ALTERNATE_SL, req_msg, wc->sl);
 		}
 
-		if (req_msg->alt_remote_lid == IB_LID_PERMISSIVE)
-			req_msg->alt_remote_lid = cpu_to_be16(wc->dlid_path_bits);
+		if (cpu_to_be16(IBA_GET(CM_REQ_ALTERNATE_REMOTE_PORT_LID,
+					req_msg)) == IB_LID_PERMISSIVE)
+			IBA_SET(CM_REQ_ALTERNATE_REMOTE_PORT_LID, req_msg,
+				wc->dlid_path_bits);
 	}
 }
 
@@ -1913,7 +1964,8 @@ static int cm_req_handler(struct cm_work *work)
 		return PTR_ERR(cm_id);
 
 	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
-	cm_id_priv->id.remote_id = req_msg->local_comm_id;
+	cm_id_priv->id.remote_id =
+		cpu_to_be32(IBA_GET(CM_REQ_LOCAL_COMM_ID, req_msg));
 	ret = cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
 				      work->mad_recv_wc->recv_buf.grh,
 				      &cm_id_priv->av);
@@ -1925,8 +1977,10 @@ static int cm_req_handler(struct cm_work *work)
 		ret = PTR_ERR(cm_id_priv->timewait_info);
 		goto destroy;
 	}
-	cm_id_priv->timewait_info->work.remote_id = req_msg->local_comm_id;
-	cm_id_priv->timewait_info->remote_ca_guid = req_msg->local_ca_guid;
+	cm_id_priv->timewait_info->work.remote_id =
+		cpu_to_be32(IBA_GET(CM_REQ_LOCAL_COMM_ID, req_msg));
+	cm_id_priv->timewait_info->remote_ca_guid =
+		cpu_to_be64(IBA_GET(CM_REQ_LOCAL_CA_GUID, req_msg));
 	cm_id_priv->timewait_info->remote_qpn =
 		cpu_to_be32(IBA_GET(CM_REQ_LOCAL_QPN, req_msg));
 
@@ -1940,7 +1994,8 @@ static int cm_req_handler(struct cm_work *work)
 
 	cm_id_priv->id.cm_handler = listen_cm_id_priv->id.cm_handler;
 	cm_id_priv->id.context = listen_cm_id_priv->id.context;
-	cm_id_priv->id.service_id = req_msg->service_id;
+	cm_id_priv->id.service_id =
+		cpu_to_be64(IBA_GET(CM_REQ_SERVICE_ID, req_msg));
 	cm_id_priv->id.service_mask = ~cpu_to_be64(0);
 
 	cm_process_routed_req(req_msg, work->mad_recv_wc->wc);
@@ -2009,7 +2064,7 @@ static int cm_req_handler(struct cm_work *work)
 	cm_id_priv->responder_resources =
 		IBA_GET(CM_REQ_INITIATOR_DEPTH, req_msg);
 	cm_id_priv->path_mtu = IBA_GET(CM_REQ_PATH_PACKET_PAYLOAD_MTU, req_msg);
-	cm_id_priv->pkey = req_msg->pkey;
+	cm_id_priv->pkey = cpu_to_be16(IBA_GET(CM_REQ_PARTITION_KEY, req_msg));
 	cm_id_priv->sq_psn = cpu_to_be32(IBA_GET(CM_REQ_STARTING_PSN, req_msg));
 	cm_id_priv->retry_count = IBA_GET(CM_REQ_RETRY_COUNT, req_msg);
 	cm_id_priv->rnr_retry_count = IBA_GET(CM_REQ_RNR_RETRY_COUNT, req_msg);
@@ -2035,18 +2090,23 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg,
 			  struct ib_cm_rep_param *param)
 {
 	cm_format_mad_hdr(&rep_msg->hdr, CM_REP_ATTR_ID, cm_id_priv->tid);
-	rep_msg->local_comm_id = cm_id_priv->id.local_id;
-	rep_msg->remote_comm_id = cm_id_priv->id.remote_id;
+	IBA_SET(CM_REP_LOCAL_COMM_ID, rep_msg,
+		be32_to_cpu(cm_id_priv->id.local_id));
+	IBA_SET(CM_REP_REMOTE_COMM_ID, rep_msg,
+		be32_to_cpu(cm_id_priv->id.remote_id));
 	IBA_SET(CM_REP_STARTING_PSN, rep_msg, param->starting_psn);
-	rep_msg->resp_resources = param->responder_resources;
+	IBA_SET(CM_REP_RESPONDER_RESOURCES, rep_msg,
+		param->responder_resources);
 	IBA_SET(CM_REP_TARGET_ACK_DELAY, rep_msg,
 		cm_id_priv->av.port->cm_dev->ack_delay);
 	IBA_SET(CM_REP_FAILOVER_ACCEPTED, rep_msg, param->failover_accepted);
 	IBA_SET(CM_REP_RNR_RETRY_COUNT, rep_msg, param->rnr_retry_count);
-	rep_msg->local_ca_guid = cm_id_priv->id.device->node_guid;
+	IBA_SET(CM_REP_LOCAL_CA_GUID, rep_msg,
+		be64_to_cpu(cm_id_priv->id.device->node_guid));
 
 	if (cm_id_priv->qp_type != IB_QPT_XRC_TGT) {
-		rep_msg->initiator_depth = param->initiator_depth;
+		IBA_SET(CM_REP_INITIATOR_DEPTH, rep_msg,
+			param->initiator_depth);
 		IBA_SET(CM_REP_END_TO_END_FLOW_CONTROL, rep_msg,
 			param->flow_control);
 		IBA_SET(CM_REP_SRQ, rep_msg, param->srq);
@@ -2118,8 +2178,10 @@ static void cm_format_rtu(struct cm_rtu_msg *rtu_msg,
 			  u8 private_data_len)
 {
 	cm_format_mad_hdr(&rtu_msg->hdr, CM_RTU_ATTR_ID, cm_id_priv->tid);
-	rtu_msg->local_comm_id = cm_id_priv->id.local_id;
-	rtu_msg->remote_comm_id = cm_id_priv->id.remote_id;
+	IBA_SET(CM_RTU_LOCAL_COMM_ID, rtu_msg,
+		be32_to_cpu(cm_id_priv->id.local_id));
+	IBA_SET(CM_RTU_REMOTE_COMM_ID, rtu_msg,
+		be32_to_cpu(cm_id_priv->id.remote_id));
 
 	if (private_data && private_data_len)
 		memcpy(rtu_msg->private_data, private_data, private_data_len);
@@ -2185,12 +2247,13 @@ static void cm_format_rep_event(struct cm_work *work, enum ib_qp_type qp_type)
 
 	rep_msg = (struct cm_rep_msg *)work->mad_recv_wc->recv_buf.mad;
 	param = &work->cm_event.param.rep_rcvd;
-	param->remote_ca_guid = rep_msg->local_ca_guid;
-	param->remote_qkey = be32_to_cpu(rep_msg->local_qkey);
+	param->remote_ca_guid =
+		cpu_to_be64(IBA_GET(CM_REP_LOCAL_CA_GUID, rep_msg));
+	param->remote_qkey = IBA_GET(CM_REP_LOCAL_Q_KEY, rep_msg);
 	param->remote_qpn = be32_to_cpu(cm_rep_get_qpn(rep_msg, qp_type));
 	param->starting_psn = IBA_GET(CM_REP_STARTING_PSN, rep_msg);
-	param->responder_resources = rep_msg->initiator_depth;
-	param->initiator_depth = rep_msg->resp_resources;
+	param->responder_resources = IBA_GET(CM_REP_INITIATOR_DEPTH, rep_msg);
+	param->initiator_depth = IBA_GET(CM_REP_RESPONDER_RESOURCES, rep_msg);
 	param->target_ack_delay = IBA_GET(CM_REP_TARGET_ACK_DELAY, rep_msg);
 	param->failover_accepted = IBA_GET(CM_REP_FAILOVER_ACCEPTED, rep_msg);
 	param->flow_control = IBA_GET(CM_REP_END_TO_END_FLOW_CONTROL, rep_msg);
@@ -2207,8 +2270,9 @@ static void cm_dup_rep_handler(struct cm_work *work)
 	int ret;
 
 	rep_msg = (struct cm_rep_msg *) work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(rep_msg->remote_comm_id,
-				   rep_msg->local_comm_id);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_REP_REMOTE_COMM_ID, rep_msg)),
+		cpu_to_be32(IBA_GET(CM_REP_LOCAL_COMM_ID, rep_msg)));
 	if (!cm_id_priv)
 		return;
 
@@ -2252,11 +2316,12 @@ static int cm_rep_handler(struct cm_work *work)
 	struct cm_timewait_info *timewait_info;
 
 	rep_msg = (struct cm_rep_msg *)work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(rep_msg->remote_comm_id, 0);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_REP_REMOTE_COMM_ID, rep_msg)), 0);
 	if (!cm_id_priv) {
 		cm_dup_rep_handler(work);
 		pr_debug("%s: remote_comm_id %d, no cm_id_priv\n", __func__,
-			 be32_to_cpu(rep_msg->remote_comm_id));
+			 IBA_GET(CM_REP_REMOTE_COMM_ID, rep_msg));
 		return -EINVAL;
 	}
 
@@ -2270,15 +2335,18 @@ static int cm_rep_handler(struct cm_work *work)
 	default:
 		spin_unlock_irq(&cm_id_priv->lock);
 		ret = -EINVAL;
-		pr_debug("%s: cm_id_priv->id.state: %d, local_comm_id %d, remote_comm_id %d\n",
-			 __func__, cm_id_priv->id.state,
-			 be32_to_cpu(rep_msg->local_comm_id),
-			 be32_to_cpu(rep_msg->remote_comm_id));
+		pr_debug(
+			"%s: cm_id_priv->id.state: %d, local_comm_id %d, remote_comm_id %d\n",
+			__func__, cm_id_priv->id.state,
+			IBA_GET(CM_REP_LOCAL_COMM_ID, rep_msg),
+			IBA_GET(CM_REP_REMOTE_COMM_ID, rep_msg));
 		goto error;
 	}
 
-	cm_id_priv->timewait_info->work.remote_id = rep_msg->local_comm_id;
-	cm_id_priv->timewait_info->remote_ca_guid = rep_msg->local_ca_guid;
+	cm_id_priv->timewait_info->work.remote_id =
+		cpu_to_be32(IBA_GET(CM_REP_LOCAL_COMM_ID, rep_msg));
+	cm_id_priv->timewait_info->remote_ca_guid =
+		cpu_to_be64(IBA_GET(CM_REP_LOCAL_CA_GUID, rep_msg));
 	cm_id_priv->timewait_info->remote_qpn = cm_rep_get_qpn(rep_msg, cm_id_priv->qp_type);
 
 	spin_lock(&cm.lock);
@@ -2288,7 +2356,7 @@ static int cm_rep_handler(struct cm_work *work)
 		spin_unlock_irq(&cm_id_priv->lock);
 		ret = -EINVAL;
 		pr_debug("%s: Failed to insert remote id %d\n", __func__,
-			 be32_to_cpu(rep_msg->remote_comm_id));
+			 IBA_GET(CM_REP_REMOTE_COMM_ID, rep_msg));
 		goto error;
 	}
 	/* Check for a stale connection. */
@@ -2306,9 +2374,10 @@ static int cm_rep_handler(struct cm_work *work)
 			     IB_CM_REJ_STALE_CONN, CM_MSG_RESPONSE_REP,
 			     NULL, 0);
 		ret = -EINVAL;
-		pr_debug("%s: Stale connection. local_comm_id %d, remote_comm_id %d\n",
-			 __func__, be32_to_cpu(rep_msg->local_comm_id),
-			 be32_to_cpu(rep_msg->remote_comm_id));
+		pr_debug(
+			"%s: Stale connection. local_comm_id %d, remote_comm_id %d\n",
+			__func__, IBA_GET(CM_REP_LOCAL_COMM_ID, rep_msg),
+			IBA_GET(CM_REP_REMOTE_COMM_ID, rep_msg));
 
 		if (cur_cm_id_priv) {
 			cm_id = &cur_cm_id_priv->id;
@@ -2321,10 +2390,13 @@ static int cm_rep_handler(struct cm_work *work)
 	spin_unlock(&cm.lock);
 
 	cm_id_priv->id.state = IB_CM_REP_RCVD;
-	cm_id_priv->id.remote_id = rep_msg->local_comm_id;
+	cm_id_priv->id.remote_id =
+		cpu_to_be32(IBA_GET(CM_REP_LOCAL_COMM_ID, rep_msg));
 	cm_id_priv->remote_qpn = cm_rep_get_qpn(rep_msg, cm_id_priv->qp_type);
-	cm_id_priv->initiator_depth = rep_msg->resp_resources;
-	cm_id_priv->responder_resources = rep_msg->initiator_depth;
+	cm_id_priv->initiator_depth =
+		IBA_GET(CM_REP_RESPONDER_RESOURCES, rep_msg);
+	cm_id_priv->responder_resources =
+		IBA_GET(CM_REP_INITIATOR_DEPTH, rep_msg);
 	cm_id_priv->sq_psn = cpu_to_be32(IBA_GET(CM_REP_STARTING_PSN, rep_msg));
 	cm_id_priv->rnr_retry_count = IBA_GET(CM_REP_RNR_RETRY_COUNT, rep_msg);
 	cm_id_priv->target_ack_delay =
@@ -2394,8 +2466,9 @@ static int cm_rtu_handler(struct cm_work *work)
 	int ret;
 
 	rtu_msg = (struct cm_rtu_msg *)work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(rtu_msg->remote_comm_id,
-				   rtu_msg->local_comm_id);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_RTU_REMOTE_COMM_ID, rtu_msg)),
+		cpu_to_be32(IBA_GET(CM_RTU_LOCAL_COMM_ID, rtu_msg)));
 	if (!cm_id_priv)
 		return -EINVAL;
 
@@ -2434,8 +2507,10 @@ static void cm_format_dreq(struct cm_dreq_msg *dreq_msg,
 {
 	cm_format_mad_hdr(&dreq_msg->hdr, CM_DREQ_ATTR_ID,
 			  cm_form_tid(cm_id_priv));
-	dreq_msg->local_comm_id = cm_id_priv->id.local_id;
-	dreq_msg->remote_comm_id = cm_id_priv->id.remote_id;
+	IBA_SET(CM_DREQ_LOCAL_COMM_ID, dreq_msg,
+		be32_to_cpu(cm_id_priv->id.local_id));
+	IBA_SET(CM_DREQ_REMOTE_COMM_ID, dreq_msg,
+		be32_to_cpu(cm_id_priv->id.remote_id));
 	IBA_SET(CM_DREQ_REMOTE_QPN_EECN, dreq_msg,
 		be32_to_cpu(cm_id_priv->remote_qpn));
 
@@ -2500,8 +2575,10 @@ static void cm_format_drep(struct cm_drep_msg *drep_msg,
 			  u8 private_data_len)
 {
 	cm_format_mad_hdr(&drep_msg->hdr, CM_DREP_ATTR_ID, cm_id_priv->tid);
-	drep_msg->local_comm_id = cm_id_priv->id.local_id;
-	drep_msg->remote_comm_id = cm_id_priv->id.remote_id;
+	IBA_SET(CM_DREP_LOCAL_COMM_ID, drep_msg,
+		be32_to_cpu(cm_id_priv->id.local_id));
+	IBA_SET(CM_DREP_REMOTE_COMM_ID, drep_msg,
+		be32_to_cpu(cm_id_priv->id.remote_id));
 
 	if (private_data && private_data_len)
 		memcpy(drep_msg->private_data, private_data, private_data_len);
@@ -2572,8 +2649,10 @@ static int cm_issue_drep(struct cm_port *port,
 	drep_msg = (struct cm_drep_msg *) msg->mad;
 
 	cm_format_mad_hdr(&drep_msg->hdr, CM_DREP_ATTR_ID, dreq_msg->hdr.tid);
-	drep_msg->remote_comm_id = dreq_msg->local_comm_id;
-	drep_msg->local_comm_id = dreq_msg->remote_comm_id;
+	IBA_SET(CM_DREP_REMOTE_COMM_ID, drep_msg,
+		IBA_GET(CM_DREQ_LOCAL_COMM_ID, dreq_msg));
+	IBA_SET(CM_DREP_LOCAL_COMM_ID, drep_msg,
+		IBA_GET(CM_DREQ_REMOTE_COMM_ID, dreq_msg));
 
 	ret = ib_post_send_mad(msg, NULL);
 	if (ret)
@@ -2590,15 +2669,17 @@ static int cm_dreq_handler(struct cm_work *work)
 	int ret;
 
 	dreq_msg = (struct cm_dreq_msg *)work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(dreq_msg->remote_comm_id,
-				   dreq_msg->local_comm_id);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_DREQ_REMOTE_COMM_ID, dreq_msg)),
+		cpu_to_be32(IBA_GET(CM_DREQ_LOCAL_COMM_ID, dreq_msg)));
 	if (!cm_id_priv) {
 		atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES].
 				counter[CM_DREQ_COUNTER]);
 		cm_issue_drep(work->port, work->mad_recv_wc);
-		pr_debug("%s: no cm_id_priv, local_comm_id %d, remote_comm_id %d\n",
-			 __func__, be32_to_cpu(dreq_msg->local_comm_id),
-			 be32_to_cpu(dreq_msg->remote_comm_id));
+		pr_debug(
+			"%s: no cm_id_priv, local_comm_id %d, remote_comm_id %d\n",
+			__func__, IBA_GET(CM_DREQ_LOCAL_COMM_ID, dreq_msg),
+			IBA_GET(CM_DREQ_REMOTE_COMM_ID, dreq_msg));
 		return -EINVAL;
 	}
 
@@ -2672,8 +2753,9 @@ static int cm_drep_handler(struct cm_work *work)
 	int ret;
 
 	drep_msg = (struct cm_drep_msg *)work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(drep_msg->remote_comm_id,
-				   drep_msg->local_comm_id);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_DREP_REMOTE_COMM_ID, drep_msg)),
+		cpu_to_be32(IBA_GET(CM_DREP_LOCAL_COMM_ID, drep_msg)));
 	if (!cm_id_priv)
 		return -EINVAL;
 
@@ -2775,7 +2857,7 @@ static void cm_format_rej_event(struct cm_work *work)
 	param = &work->cm_event.param.rej_rcvd;
 	param->ari = rej_msg->ari;
 	param->ari_length = IBA_GET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg);
-	param->reason = __be16_to_cpu(rej_msg->reason);
+	param->reason = IBA_GET(CM_REJ_REASON, rej_msg);
 	work->cm_event.private_data = &rej_msg->private_data;
 }
 
@@ -2785,9 +2867,9 @@ static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
 	struct cm_id_private *cm_id_priv;
 	__be32 remote_id;
 
-	remote_id = rej_msg->local_comm_id;
+	remote_id = cpu_to_be32(IBA_GET(CM_REJ_LOCAL_COMM_ID, rej_msg));
 
-	if (__be16_to_cpu(rej_msg->reason) == IB_CM_REJ_TIMEOUT) {
+	if (IBA_GET(CM_REJ_REASON, rej_msg) == IB_CM_REJ_TIMEOUT) {
 		spin_lock_irq(&cm.lock);
 		timewait_info = cm_find_remote_id( *((__be64 *) rej_msg->ari),
 						  remote_id);
@@ -2800,9 +2882,13 @@ static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
 		spin_unlock_irq(&cm.lock);
 	} else if (IBA_GET(CM_REJ_MESSAGE_REJECTED, rej_msg) ==
 		   CM_MSG_RESPONSE_REQ)
-		cm_id_priv = cm_acquire_id(rej_msg->remote_comm_id, 0);
+		cm_id_priv = cm_acquire_id(
+			cpu_to_be32(IBA_GET(CM_REJ_REMOTE_COMM_ID, rej_msg)),
+			0);
 	else
-		cm_id_priv = cm_acquire_id(rej_msg->remote_comm_id, remote_id);
+		cm_id_priv = cm_acquire_id(
+			cpu_to_be32(IBA_GET(CM_REJ_REMOTE_COMM_ID, rej_msg)),
+			remote_id);
 
 	return cm_id_priv;
 }
@@ -2830,7 +2916,7 @@ static int cm_rej_handler(struct cm_work *work)
 		/* fall through */
 	case IB_CM_REQ_RCVD:
 	case IB_CM_MRA_REQ_SENT:
-		if (__be16_to_cpu(rej_msg->reason) == IB_CM_REJ_STALE_CONN)
+		if (IBA_GET(CM_REJ_REASON, rej_msg) == IB_CM_REJ_STALE_CONN)
 			cm_enter_timewait(cm_id_priv);
 		else
 			cm_reset_to_idle(cm_id_priv);
@@ -2962,11 +3048,14 @@ static struct cm_id_private * cm_acquire_mraed_id(struct cm_mra_msg *mra_msg)
 {
 	switch (IBA_GET(CM_MRA_MESSAGE_MRAED, mra_msg)) {
 	case CM_MSG_RESPONSE_REQ:
-		return cm_acquire_id(mra_msg->remote_comm_id, 0);
+		return cm_acquire_id(
+			cpu_to_be32(IBA_GET(CM_MRA_REMOTE_COMM_ID, mra_msg)),
+			0);
 	case CM_MSG_RESPONSE_REP:
 	case CM_MSG_RESPONSE_OTHER:
-		return cm_acquire_id(mra_msg->remote_comm_id,
-				     mra_msg->local_comm_id);
+		return cm_acquire_id(
+			cpu_to_be32(IBA_GET(CM_MRA_REMOTE_COMM_ID, mra_msg)),
+			cpu_to_be32(IBA_GET(CM_MRA_LOCAL_COMM_ID, mra_msg)));
 	default:
 		return NULL;
 	}
@@ -3057,8 +3146,10 @@ static void cm_format_path_lid_from_lap(struct cm_lap_msg *lap_msg,
 	u32 lid;
 
 	if (path->rec_type != SA_PATH_REC_TYPE_OPA) {
-		sa_path_set_dlid(path, ntohs(lap_msg->alt_local_lid));
-		sa_path_set_slid(path, ntohs(lap_msg->alt_remote_lid));
+		sa_path_set_dlid(path, IBA_GET(CM_LAP_ALTERNATE_LOCAL_PORT_LID,
+					       lap_msg));
+		sa_path_set_slid(path, IBA_GET(CM_LAP_ALTERNATE_REMOTE_PORT_LID,
+					       lap_msg));
 	} else {
 		lid = opa_get_lid_from_gid(&lap_msg->alt_local_gid);
 		sa_path_set_dlid(path, lid);
@@ -3076,7 +3167,7 @@ static void cm_format_path_from_lap(struct cm_id_private *cm_id_priv,
 	path->sgid = lap_msg->alt_remote_gid;
 	path->flow_label =
 		cpu_to_be32(IBA_GET(CM_LAP_ALTERNATE_FLOW_LABEL, lap_msg));
-	path->hop_limit = lap_msg->alt_hop_limit;
+	path->hop_limit = IBA_GET(CM_LAP_ALTERNATE_HOP_LIMIT, lap_msg);
 	path->traffic_class = IBA_GET(CM_LAP_ALTERNATE_TRAFFIC_CLASS, lap_msg);
 	path->reversible = 1;
 	path->pkey = cm_id_priv->pkey;
@@ -3109,8 +3200,9 @@ static int cm_lap_handler(struct cm_work *work)
 
 	/* todo: verify LAP request and send reject APR if invalid. */
 	lap_msg = (struct cm_lap_msg *)work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(lap_msg->remote_comm_id,
-				   lap_msg->local_comm_id);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_LAP_REMOTE_COMM_ID, lap_msg)),
+		cpu_to_be32(IBA_GET(CM_LAP_LOCAL_COMM_ID, lap_msg)));
 	if (!cm_id_priv)
 		return -EINVAL;
 
@@ -3201,14 +3293,17 @@ static int cm_apr_handler(struct cm_work *work)
 		return -EINVAL;
 
 	apr_msg = (struct cm_apr_msg *)work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(apr_msg->remote_comm_id,
-				   apr_msg->local_comm_id);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_APR_REMOTE_COMM_ID, apr_msg)),
+		cpu_to_be32(IBA_GET(CM_APR_LOCAL_COMM_ID, apr_msg)));
 	if (!cm_id_priv)
 		return -EINVAL; /* Unmatched reply. */
 
-	work->cm_event.param.apr_rcvd.ap_status = apr_msg->ap_status;
+	work->cm_event.param.apr_rcvd.ap_status =
+		IBA_GET(CM_APR_AR_STATUS, apr_msg);
 	work->cm_event.param.apr_rcvd.apr_info = &apr_msg->info;
-	work->cm_event.param.apr_rcvd.info_len = apr_msg->info_length;
+	work->cm_event.param.apr_rcvd.info_len =
+		IBA_GET(CM_APR_ADDITIONAL_INFORMATION_LENGTH, apr_msg);
 	work->cm_event.private_data = &apr_msg->private_data;
 
 	spin_lock_irq(&cm_id_priv->lock);
@@ -3281,9 +3376,12 @@ static void cm_format_sidr_req(struct cm_sidr_req_msg *sidr_req_msg,
 {
 	cm_format_mad_hdr(&sidr_req_msg->hdr, CM_SIDR_REQ_ATTR_ID,
 			  cm_form_tid(cm_id_priv));
-	sidr_req_msg->request_id = cm_id_priv->id.local_id;
-	sidr_req_msg->pkey = param->path->pkey;
-	sidr_req_msg->service_id = param->service_id;
+	IBA_SET(CM_SIDR_REQ_REQUESTID, sidr_req_msg,
+		be32_to_cpu(cm_id_priv->id.local_id));
+	IBA_SET(CM_SIDR_REQ_PARTITION_KEY, sidr_req_msg,
+		be16_to_cpu(param->path->pkey));
+	IBA_SET(CM_SIDR_REQ_SERVICEID, sidr_req_msg,
+		be64_to_cpu(param->service_id));
 
 	if (param->private_data && param->private_data_len)
 		memcpy(sidr_req_msg->private_data, param->private_data,
@@ -3351,9 +3449,10 @@ static void cm_format_sidr_req_event(struct cm_work *work,
 	sidr_req_msg = (struct cm_sidr_req_msg *)
 				work->mad_recv_wc->recv_buf.mad;
 	param = &work->cm_event.param.sidr_req_rcvd;
-	param->pkey = __be16_to_cpu(sidr_req_msg->pkey);
+	param->pkey = IBA_GET(CM_SIDR_REQ_PARTITION_KEY, sidr_req_msg);
 	param->listen_id = listen_id;
-	param->service_id = sidr_req_msg->service_id;
+	param->service_id =
+		cpu_to_be64(IBA_GET(CM_SIDR_REQ_SERVICEID, sidr_req_msg));
 	param->bth_pkey = cm_get_bth_pkey(work);
 	param->port = work->port->port_num;
 	param->sgid_attr = rx_cm_id->av.ah_attr.grh.sgid_attr;
@@ -3385,7 +3484,8 @@ static int cm_sidr_req_handler(struct cm_work *work)
 	if (ret)
 		goto out;
 
-	cm_id_priv->id.remote_id = sidr_req_msg->request_id;
+	cm_id_priv->id.remote_id =
+		cpu_to_be32(IBA_GET(CM_SIDR_REQ_REQUESTID, sidr_req_msg));
 	cm_id_priv->tid = sidr_req_msg->hdr.tid;
 	atomic_inc(&cm_id_priv->work_count);
 
@@ -3398,8 +3498,9 @@ static int cm_sidr_req_handler(struct cm_work *work)
 		goto out; /* Duplicate message. */
 	}
 	cm_id_priv->id.state = IB_CM_SIDR_REQ_RCVD;
-	cur_cm_id_priv = cm_find_listen(cm_id->device,
-					sidr_req_msg->service_id);
+	cur_cm_id_priv = cm_find_listen(
+		cm_id->device,
+		cpu_to_be64(IBA_GET(CM_SIDR_REQ_SERVICEID, sidr_req_msg)));
 	if (!cur_cm_id_priv) {
 		spin_unlock_irq(&cm.lock);
 		cm_reject_sidr_req(cm_id_priv, IB_SIDR_UNSUPPORTED);
@@ -3411,7 +3512,8 @@ static int cm_sidr_req_handler(struct cm_work *work)
 
 	cm_id_priv->id.cm_handler = cur_cm_id_priv->id.cm_handler;
 	cm_id_priv->id.context = cur_cm_id_priv->id.context;
-	cm_id_priv->id.service_id = sidr_req_msg->service_id;
+	cm_id_priv->id.service_id =
+		cpu_to_be64(IBA_GET(CM_SIDR_REQ_SERVICEID, sidr_req_msg));
 	cm_id_priv->id.service_mask = ~cpu_to_be64(0);
 
 	cm_format_sidr_req_event(work, cm_id_priv, &cur_cm_id_priv->id);
@@ -3429,11 +3531,13 @@ static void cm_format_sidr_rep(struct cm_sidr_rep_msg *sidr_rep_msg,
 {
 	cm_format_mad_hdr(&sidr_rep_msg->hdr, CM_SIDR_REP_ATTR_ID,
 			  cm_id_priv->tid);
-	sidr_rep_msg->request_id = cm_id_priv->id.remote_id;
-	sidr_rep_msg->status = param->status;
+	IBA_SET(CM_SIDR_REP_REQUESTID, sidr_rep_msg,
+		be32_to_cpu(cm_id_priv->id.remote_id));
+	IBA_SET(CM_SIDR_REP_STATUS, sidr_rep_msg, param->status);
 	IBA_SET(CM_SIDR_REP_QPN, sidr_rep_msg, param->qp_num);
-	sidr_rep_msg->service_id = cm_id_priv->id.service_id;
-	sidr_rep_msg->qkey = cpu_to_be32(param->qkey);
+	IBA_SET(CM_SIDR_REP_SERVICEID, sidr_rep_msg,
+		be64_to_cpu(cm_id_priv->id.service_id));
+	IBA_SET(CM_SIDR_REP_Q_KEY, sidr_rep_msg, param->qkey);
 
 	if (param->info && param->info_length)
 		memcpy(sidr_rep_msg->info, param->info, param->info_length);
@@ -3500,11 +3604,12 @@ static void cm_format_sidr_rep_event(struct cm_work *work,
 	sidr_rep_msg = (struct cm_sidr_rep_msg *)
 				work->mad_recv_wc->recv_buf.mad;
 	param = &work->cm_event.param.sidr_rep_rcvd;
-	param->status = sidr_rep_msg->status;
-	param->qkey = be32_to_cpu(sidr_rep_msg->qkey);
+	param->status = IBA_GET(CM_SIDR_REP_STATUS, sidr_rep_msg);
+	param->qkey = IBA_GET(CM_SIDR_REP_Q_KEY, sidr_rep_msg);
 	param->qpn = IBA_GET(CM_SIDR_REP_QPN, sidr_rep_msg);
 	param->info = &sidr_rep_msg->info;
-	param->info_len = sidr_rep_msg->info_length;
+	param->info_len = IBA_GET(CM_SIDR_REP_ADDITIONAL_INFORMATION_LENGTH,
+				  sidr_rep_msg);
 	param->sgid_attr = cm_id_priv->av.ah_attr.grh.sgid_attr;
 	work->cm_event.private_data = &sidr_rep_msg->private_data;
 }
@@ -3516,7 +3621,8 @@ static int cm_sidr_rep_handler(struct cm_work *work)
 
 	sidr_rep_msg = (struct cm_sidr_rep_msg *)
 				work->mad_recv_wc->recv_buf.mad;
-	cm_id_priv = cm_acquire_id(sidr_rep_msg->request_id, 0);
+	cm_id_priv = cm_acquire_id(
+		cpu_to_be32(IBA_GET(CM_SIDR_REP_REQUESTID, sidr_rep_msg)), 0);
 	if (!cm_id_priv)
 		return -EINVAL; /* Unmatched reply. */
 
-- 
cgit v1.2.3-59-g8ed1b


From 4ca662a30a3c9c8a320549f7ef3066921e36f0b4 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Thu, 16 Jan 2020 13:00:36 -0400
Subject: RDMA/cm: Use IBA functions for complex structure members

Use a Coccinelle spatch to replace CM structure members used as
structures, arrays, or pointers with IBA_GET/SET versions. Applied with

$ spatch --sp-file edits.sp --in-place drivers/infiniband/core/cm.c

The spatch file was generated using the template pattern:

@@
expression src;
expression len;
{struct} *msg;
@@
- memcpy(msg->{old_name}, src, len)
+ IBA_SET_MEM({new_name}, msg, src, len)
@@
{struct} *msg;
identifier x;
@@
- msg->{old_name}.x
+ IBA_GET_MEM_PTR({new_name}, msg)->x
@@
{struct} *msg;
@@
- &msg->{old_name}
+ IBA_GET_MEM_PTR({new_name}, msg)

For GIDs:
@@
{struct} *msg;
@@
- msg->{old_name}
+ *IBA_GET_MEM_PTR({new_name}, msg)

For non-GIDs:
@@
{struct} *msg;
@@
- msg->{old_name}
+ IBA_GET_MEM_PTR({new_name}, msg)

Iterated for every remaining IBA_CHECK_OFF()/IBA_CHECK_GET()
pairing. Touched up with clang-format after.

Link: https://lore.kernel.org/r/20200116170037.30109-7-jgg@ziepe.ca
Tested-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c | 171 +++++++++++++++++++++++++++----------------
 1 file changed, 107 insertions(+), 64 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 047ee560046e..5ccd59f1ebb8 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -1280,13 +1280,17 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 		IBA_SET(CM_REQ_SRQ, req_msg, param->srq);
 	}
 
-	req_msg->primary_local_gid = pri_path->sgid;
-	req_msg->primary_remote_gid = pri_path->dgid;
+	*IBA_GET_MEM_PTR(CM_REQ_PRIMARY_LOCAL_PORT_GID, req_msg) =
+		pri_path->sgid;
+	*IBA_GET_MEM_PTR(CM_REQ_PRIMARY_REMOTE_PORT_GID, req_msg) =
+		pri_path->dgid;
 	if (pri_ext) {
-		req_msg->primary_local_gid.global.interface_id
-			= OPA_MAKE_ID(be32_to_cpu(pri_path->opa.slid));
-		req_msg->primary_remote_gid.global.interface_id
-			= OPA_MAKE_ID(be32_to_cpu(pri_path->opa.dlid));
+		IBA_GET_MEM_PTR(CM_REQ_PRIMARY_LOCAL_PORT_GID, req_msg)
+			->global.interface_id =
+			OPA_MAKE_ID(be32_to_cpu(pri_path->opa.slid));
+		IBA_GET_MEM_PTR(CM_REQ_PRIMARY_REMOTE_PORT_GID, req_msg)
+			->global.interface_id =
+			OPA_MAKE_ID(be32_to_cpu(pri_path->opa.dlid));
 	}
 	if (pri_path->hop_limit <= 1) {
 		IBA_SET(CM_REQ_PRIMARY_LOCAL_PORT_LID, req_msg,
@@ -1323,13 +1327,19 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 			alt_ext = opa_is_extended_lid(alt_path->opa.dlid,
 						      alt_path->opa.slid);
 
-		req_msg->alt_local_gid = alt_path->sgid;
-		req_msg->alt_remote_gid = alt_path->dgid;
+		*IBA_GET_MEM_PTR(CM_REQ_ALTERNATE_LOCAL_PORT_GID, req_msg) =
+			alt_path->sgid;
+		*IBA_GET_MEM_PTR(CM_REQ_ALTERNATE_REMOTE_PORT_GID, req_msg) =
+			alt_path->dgid;
 		if (alt_ext) {
-			req_msg->alt_local_gid.global.interface_id
-				= OPA_MAKE_ID(be32_to_cpu(alt_path->opa.slid));
-			req_msg->alt_remote_gid.global.interface_id
-				= OPA_MAKE_ID(be32_to_cpu(alt_path->opa.dlid));
+			IBA_GET_MEM_PTR(CM_REQ_ALTERNATE_LOCAL_PORT_GID,
+					req_msg)
+				->global.interface_id =
+				OPA_MAKE_ID(be32_to_cpu(alt_path->opa.slid));
+			IBA_GET_MEM_PTR(CM_REQ_ALTERNATE_REMOTE_PORT_GID,
+					req_msg)
+				->global.interface_id =
+				OPA_MAKE_ID(be32_to_cpu(alt_path->opa.dlid));
 		}
 		if (alt_path->hop_limit <= 1) {
 			IBA_SET(CM_REQ_ALTERNATE_LOCAL_PORT_LID, req_msg,
@@ -1364,8 +1374,8 @@ static void cm_format_req(struct cm_req_msg *req_msg,
 	}
 
 	if (param->private_data && param->private_data_len)
-		memcpy(req_msg->private_data, param->private_data,
-		       param->private_data_len);
+		IBA_SET_MEM(CM_REQ_PRIVATE_DATA, req_msg, param->private_data,
+			    param->private_data_len);
 }
 
 static int cm_validate_req_param(struct ib_cm_req_param *param)
@@ -1505,7 +1515,7 @@ static int cm_issue_rej(struct cm_port *port,
 
 	if (ari && ari_length) {
 		IBA_SET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg, ari_length);
-		memcpy(rej_msg->ari, ari, ari_length);
+		IBA_SET_MEM(CM_REJ_ARI, rej_msg, ari, ari_length);
 	}
 
 	ret = ib_post_send_mad(msg, NULL);
@@ -1519,7 +1529,8 @@ static bool cm_req_has_alt_path(struct cm_req_msg *req_msg)
 {
 	return ((cpu_to_be16(
 			IBA_GET(CM_REQ_ALTERNATE_LOCAL_PORT_LID, req_msg))) ||
-		(ib_is_opa_gid(&req_msg->alt_local_gid)));
+		(ib_is_opa_gid(IBA_GET_MEM_PTR(CM_REQ_ALTERNATE_LOCAL_PORT_GID,
+					       req_msg))));
 }
 
 static void cm_path_set_rec_type(struct ib_device *ib_device, u8 port_num,
@@ -1545,10 +1556,12 @@ static void cm_format_path_lid_from_req(struct cm_req_msg *req_msg,
 				 IBA_GET(CM_REQ_PRIMARY_REMOTE_PORT_LID,
 					 req_msg));
 	} else {
-		lid = opa_get_lid_from_gid(&req_msg->primary_local_gid);
+		lid = opa_get_lid_from_gid(IBA_GET_MEM_PTR(
+			CM_REQ_PRIMARY_LOCAL_PORT_GID, req_msg));
 		sa_path_set_dlid(primary_path, lid);
 
-		lid = opa_get_lid_from_gid(&req_msg->primary_remote_gid);
+		lid = opa_get_lid_from_gid(IBA_GET_MEM_PTR(
+			CM_REQ_PRIMARY_REMOTE_PORT_GID, req_msg));
 		sa_path_set_slid(primary_path, lid);
 	}
 
@@ -1563,10 +1576,12 @@ static void cm_format_path_lid_from_req(struct cm_req_msg *req_msg,
 				 IBA_GET(CM_REQ_ALTERNATE_REMOTE_PORT_LID,
 					 req_msg));
 	} else {
-		lid = opa_get_lid_from_gid(&req_msg->alt_local_gid);
+		lid = opa_get_lid_from_gid(IBA_GET_MEM_PTR(
+			CM_REQ_ALTERNATE_LOCAL_PORT_GID, req_msg));
 		sa_path_set_dlid(alt_path, lid);
 
-		lid = opa_get_lid_from_gid(&req_msg->alt_remote_gid);
+		lid = opa_get_lid_from_gid(IBA_GET_MEM_PTR(
+			CM_REQ_ALTERNATE_REMOTE_PORT_GID, req_msg));
 		sa_path_set_slid(alt_path, lid);
 	}
 }
@@ -1575,8 +1590,10 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 				     struct sa_path_rec *primary_path,
 				     struct sa_path_rec *alt_path)
 {
-	primary_path->dgid = req_msg->primary_local_gid;
-	primary_path->sgid = req_msg->primary_remote_gid;
+	primary_path->dgid =
+		*IBA_GET_MEM_PTR(CM_REQ_PRIMARY_LOCAL_PORT_GID, req_msg);
+	primary_path->sgid =
+		*IBA_GET_MEM_PTR(CM_REQ_PRIMARY_REMOTE_PORT_GID, req_msg);
 	primary_path->flow_label =
 		cpu_to_be32(IBA_GET(CM_REQ_PRIMARY_FLOW_LABEL, req_msg));
 	primary_path->hop_limit = IBA_GET(CM_REQ_PRIMARY_HOP_LIMIT, req_msg);
@@ -1600,8 +1617,10 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg,
 		primary_path->roce.route_resolved = false;
 
 	if (cm_req_has_alt_path(req_msg)) {
-		alt_path->dgid = req_msg->alt_local_gid;
-		alt_path->sgid = req_msg->alt_remote_gid;
+		alt_path->dgid = *IBA_GET_MEM_PTR(
+			CM_REQ_ALTERNATE_LOCAL_PORT_GID, req_msg);
+		alt_path->sgid = *IBA_GET_MEM_PTR(
+			CM_REQ_ALTERNATE_REMOTE_PORT_GID, req_msg);
 		alt_path->flow_label = cpu_to_be32(
 			IBA_GET(CM_REQ_ALTERNATE_FLOW_LABEL, req_msg));
 		alt_path->hop_limit =
@@ -1714,7 +1733,8 @@ static void cm_format_req_event(struct cm_work *work,
 	param->rnr_retry_count = IBA_GET(CM_REQ_RNR_RETRY_COUNT, req_msg);
 	param->srq = IBA_GET(CM_REQ_SRQ, req_msg);
 	param->ppath_sgid_attr = cm_id_priv->av.ah_attr.grh.sgid_attr;
-	work->cm_event.private_data = &req_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_REQ_PRIVATE_DATA, req_msg);
 }
 
 static void cm_process_work(struct cm_id_private *cm_id_priv,
@@ -1756,7 +1776,8 @@ static void cm_format_mra(struct cm_mra_msg *mra_msg,
 	IBA_SET(CM_MRA_SERVICE_TIMEOUT, mra_msg, service_timeout);
 
 	if (private_data && private_data_len)
-		memcpy(mra_msg->private_data, private_data, private_data_len);
+		IBA_SET_MEM(CM_MRA_PRIVATE_DATA, mra_msg, private_data,
+			    private_data_len);
 }
 
 static void cm_format_rej(struct cm_rej_msg *rej_msg,
@@ -1798,11 +1819,12 @@ static void cm_format_rej(struct cm_rej_msg *rej_msg,
 	IBA_SET(CM_REJ_REASON, rej_msg, reason);
 	if (ari && ari_length) {
 		IBA_SET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg, ari_length);
-		memcpy(rej_msg->ari, ari, ari_length);
+		IBA_SET_MEM(CM_REJ_ARI, rej_msg, ari, ari_length);
 	}
 
 	if (private_data && private_data_len)
-		memcpy(rej_msg->private_data, private_data, private_data_len);
+		IBA_SET_MEM(CM_REJ_PRIVATE_DATA, rej_msg, private_data,
+			    private_data_len);
 }
 
 static void cm_dup_req_handler(struct cm_work *work,
@@ -2012,10 +2034,11 @@ static int cm_req_handler(struct cm_work *work)
 		work->path[0].rec_type =
 			sa_conv_gid_to_pathrec_type(gid_attr->gid_type);
 	} else {
-		cm_path_set_rec_type(work->port->cm_dev->ib_device,
-				     work->port->port_num,
-				     &work->path[0],
-				     &req_msg->primary_local_gid);
+		cm_path_set_rec_type(
+			work->port->cm_dev->ib_device, work->port->port_num,
+			&work->path[0],
+			IBA_GET_MEM_PTR(CM_REQ_PRIMARY_LOCAL_PORT_GID,
+					req_msg));
 	}
 	if (cm_req_has_alt_path(req_msg))
 		work->path[1].rec_type = work->path[0].rec_type;
@@ -2117,8 +2140,8 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg,
 	}
 
 	if (param->private_data && param->private_data_len)
-		memcpy(rep_msg->private_data, param->private_data,
-		       param->private_data_len);
+		IBA_SET_MEM(CM_REP_PRIVATE_DATA, rep_msg, param->private_data,
+			    param->private_data_len);
 }
 
 int ib_send_cm_rep(struct ib_cm_id *cm_id,
@@ -2184,7 +2207,8 @@ static void cm_format_rtu(struct cm_rtu_msg *rtu_msg,
 		be32_to_cpu(cm_id_priv->id.remote_id));
 
 	if (private_data && private_data_len)
-		memcpy(rtu_msg->private_data, private_data, private_data_len);
+		IBA_SET_MEM(CM_RTU_PRIVATE_DATA, rtu_msg, private_data,
+			    private_data_len);
 }
 
 int ib_send_cm_rtu(struct ib_cm_id *cm_id,
@@ -2259,7 +2283,8 @@ static void cm_format_rep_event(struct cm_work *work, enum ib_qp_type qp_type)
 	param->flow_control = IBA_GET(CM_REP_END_TO_END_FLOW_CONTROL, rep_msg);
 	param->rnr_retry_count = IBA_GET(CM_REP_RNR_RETRY_COUNT, rep_msg);
 	param->srq = IBA_GET(CM_REP_SRQ, rep_msg);
-	work->cm_event.private_data = &rep_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_REP_PRIVATE_DATA, rep_msg);
 }
 
 static void cm_dup_rep_handler(struct cm_work *work)
@@ -2472,7 +2497,8 @@ static int cm_rtu_handler(struct cm_work *work)
 	if (!cm_id_priv)
 		return -EINVAL;
 
-	work->cm_event.private_data = &rtu_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_RTU_PRIVATE_DATA, rtu_msg);
 
 	spin_lock_irq(&cm_id_priv->lock);
 	if (cm_id_priv->id.state != IB_CM_REP_SENT &&
@@ -2515,7 +2541,8 @@ static void cm_format_dreq(struct cm_dreq_msg *dreq_msg,
 		be32_to_cpu(cm_id_priv->remote_qpn));
 
 	if (private_data && private_data_len)
-		memcpy(dreq_msg->private_data, private_data, private_data_len);
+		IBA_SET_MEM(CM_DREQ_PRIVATE_DATA, dreq_msg, private_data,
+			    private_data_len);
 }
 
 int ib_send_cm_dreq(struct ib_cm_id *cm_id,
@@ -2581,7 +2608,8 @@ static void cm_format_drep(struct cm_drep_msg *drep_msg,
 		be32_to_cpu(cm_id_priv->id.remote_id));
 
 	if (private_data && private_data_len)
-		memcpy(drep_msg->private_data, private_data, private_data_len);
+		IBA_SET_MEM(CM_DREP_PRIVATE_DATA, drep_msg, private_data,
+			    private_data_len);
 }
 
 int ib_send_cm_drep(struct ib_cm_id *cm_id,
@@ -2683,7 +2711,8 @@ static int cm_dreq_handler(struct cm_work *work)
 		return -EINVAL;
 	}
 
-	work->cm_event.private_data = &dreq_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_DREQ_PRIVATE_DATA, dreq_msg);
 
 	spin_lock_irq(&cm_id_priv->lock);
 	if (cm_id_priv->local_qpn !=
@@ -2759,7 +2788,8 @@ static int cm_drep_handler(struct cm_work *work)
 	if (!cm_id_priv)
 		return -EINVAL;
 
-	work->cm_event.private_data = &drep_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_DREP_PRIVATE_DATA, drep_msg);
 
 	spin_lock_irq(&cm_id_priv->lock);
 	if (cm_id_priv->id.state != IB_CM_DREQ_SENT &&
@@ -2855,10 +2885,11 @@ static void cm_format_rej_event(struct cm_work *work)
 
 	rej_msg = (struct cm_rej_msg *)work->mad_recv_wc->recv_buf.mad;
 	param = &work->cm_event.param.rej_rcvd;
-	param->ari = rej_msg->ari;
+	param->ari = IBA_GET_MEM_PTR(CM_REJ_ARI, rej_msg);
 	param->ari_length = IBA_GET(CM_REJ_REJECTED_INFO_LENGTH, rej_msg);
 	param->reason = IBA_GET(CM_REJ_REASON, rej_msg);
-	work->cm_event.private_data = &rej_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_REJ_PRIVATE_DATA, rej_msg);
 }
 
 static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
@@ -2871,8 +2902,9 @@ static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
 
 	if (IBA_GET(CM_REJ_REASON, rej_msg) == IB_CM_REJ_TIMEOUT) {
 		spin_lock_irq(&cm.lock);
-		timewait_info = cm_find_remote_id( *((__be64 *) rej_msg->ari),
-						  remote_id);
+		timewait_info = cm_find_remote_id(
+			*((__be64 *)IBA_GET_MEM_PTR(CM_REJ_ARI, rej_msg)),
+			remote_id);
 		if (!timewait_info) {
 			spin_unlock_irq(&cm.lock);
 			return NULL;
@@ -3072,7 +3104,8 @@ static int cm_mra_handler(struct cm_work *work)
 	if (!cm_id_priv)
 		return -EINVAL;
 
-	work->cm_event.private_data = &mra_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_MRA_PRIVATE_DATA, mra_msg);
 	work->cm_event.param.mra_rcvd.service_timeout =
 		IBA_GET(CM_MRA_SERVICE_TIMEOUT, mra_msg);
 	timeout = cm_convert_to_ms(IBA_GET(CM_MRA_SERVICE_TIMEOUT, mra_msg)) +
@@ -3151,10 +3184,12 @@ static void cm_format_path_lid_from_lap(struct cm_lap_msg *lap_msg,
 		sa_path_set_slid(path, IBA_GET(CM_LAP_ALTERNATE_REMOTE_PORT_LID,
 					       lap_msg));
 	} else {
-		lid = opa_get_lid_from_gid(&lap_msg->alt_local_gid);
+		lid = opa_get_lid_from_gid(IBA_GET_MEM_PTR(
+			CM_LAP_ALTERNATE_LOCAL_PORT_GID, lap_msg));
 		sa_path_set_dlid(path, lid);
 
-		lid = opa_get_lid_from_gid(&lap_msg->alt_remote_gid);
+		lid = opa_get_lid_from_gid(IBA_GET_MEM_PTR(
+			CM_LAP_ALTERNATE_REMOTE_PORT_GID, lap_msg));
 		sa_path_set_slid(path, lid);
 	}
 }
@@ -3163,8 +3198,9 @@ static void cm_format_path_from_lap(struct cm_id_private *cm_id_priv,
 				    struct sa_path_rec *path,
 				    struct cm_lap_msg *lap_msg)
 {
-	path->dgid = lap_msg->alt_local_gid;
-	path->sgid = lap_msg->alt_remote_gid;
+	path->dgid = *IBA_GET_MEM_PTR(CM_LAP_ALTERNATE_LOCAL_PORT_GID, lap_msg);
+	path->sgid =
+		*IBA_GET_MEM_PTR(CM_LAP_ALTERNATE_REMOTE_PORT_GID, lap_msg);
 	path->flow_label =
 		cpu_to_be32(IBA_GET(CM_LAP_ALTERNATE_FLOW_LABEL, lap_msg));
 	path->hop_limit = IBA_GET(CM_LAP_ALTERNATE_HOP_LIMIT, lap_msg);
@@ -3209,12 +3245,13 @@ static int cm_lap_handler(struct cm_work *work)
 	param = &work->cm_event.param.lap_rcvd;
 	memset(&work->path[0], 0, sizeof(work->path[1]));
 	cm_path_set_rec_type(work->port->cm_dev->ib_device,
-			     work->port->port_num,
-			     &work->path[0],
-			     &lap_msg->alt_local_gid);
+			     work->port->port_num, &work->path[0],
+			     IBA_GET_MEM_PTR(CM_LAP_ALTERNATE_LOCAL_PORT_GID,
+					     lap_msg));
 	param->alternate_path = &work->path[0];
 	cm_format_path_from_lap(cm_id_priv, param->alternate_path, lap_msg);
-	work->cm_event.private_data = &lap_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_LAP_PRIVATE_DATA, lap_msg);
 
 	spin_lock_irq(&cm_id_priv->lock);
 	if (cm_id_priv->id.state != IB_CM_ESTABLISHED)
@@ -3301,10 +3338,12 @@ static int cm_apr_handler(struct cm_work *work)
 
 	work->cm_event.param.apr_rcvd.ap_status =
 		IBA_GET(CM_APR_AR_STATUS, apr_msg);
-	work->cm_event.param.apr_rcvd.apr_info = &apr_msg->info;
+	work->cm_event.param.apr_rcvd.apr_info =
+		IBA_GET_MEM_PTR(CM_APR_ADDITIONAL_INFORMATION, apr_msg);
 	work->cm_event.param.apr_rcvd.info_len =
 		IBA_GET(CM_APR_ADDITIONAL_INFORMATION_LENGTH, apr_msg);
-	work->cm_event.private_data = &apr_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_APR_PRIVATE_DATA, apr_msg);
 
 	spin_lock_irq(&cm_id_priv->lock);
 	if (cm_id_priv->id.state != IB_CM_ESTABLISHED ||
@@ -3384,8 +3423,8 @@ static void cm_format_sidr_req(struct cm_sidr_req_msg *sidr_req_msg,
 		be64_to_cpu(param->service_id));
 
 	if (param->private_data && param->private_data_len)
-		memcpy(sidr_req_msg->private_data, param->private_data,
-		       param->private_data_len);
+		IBA_SET_MEM(CM_SIDR_REQ_PRIVATE_DATA, sidr_req_msg,
+			    param->private_data, param->private_data_len);
 }
 
 int ib_send_cm_sidr_req(struct ib_cm_id *cm_id,
@@ -3456,7 +3495,8 @@ static void cm_format_sidr_req_event(struct cm_work *work,
 	param->bth_pkey = cm_get_bth_pkey(work);
 	param->port = work->port->port_num;
 	param->sgid_attr = rx_cm_id->av.ah_attr.grh.sgid_attr;
-	work->cm_event.private_data = &sidr_req_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_SIDR_REQ_PRIVATE_DATA, sidr_req_msg);
 }
 
 static int cm_sidr_req_handler(struct cm_work *work)
@@ -3540,11 +3580,12 @@ static void cm_format_sidr_rep(struct cm_sidr_rep_msg *sidr_rep_msg,
 	IBA_SET(CM_SIDR_REP_Q_KEY, sidr_rep_msg, param->qkey);
 
 	if (param->info && param->info_length)
-		memcpy(sidr_rep_msg->info, param->info, param->info_length);
+		IBA_SET_MEM(CM_SIDR_REP_ADDITIONAL_INFORMATION, sidr_rep_msg,
+			    param->info, param->info_length);
 
 	if (param->private_data && param->private_data_len)
-		memcpy(sidr_rep_msg->private_data, param->private_data,
-		       param->private_data_len);
+		IBA_SET_MEM(CM_SIDR_REP_PRIVATE_DATA, sidr_rep_msg,
+			    param->private_data, param->private_data_len);
 }
 
 int ib_send_cm_sidr_rep(struct ib_cm_id *cm_id,
@@ -3607,11 +3648,13 @@ static void cm_format_sidr_rep_event(struct cm_work *work,
 	param->status = IBA_GET(CM_SIDR_REP_STATUS, sidr_rep_msg);
 	param->qkey = IBA_GET(CM_SIDR_REP_Q_KEY, sidr_rep_msg);
 	param->qpn = IBA_GET(CM_SIDR_REP_QPN, sidr_rep_msg);
-	param->info = &sidr_rep_msg->info;
+	param->info = IBA_GET_MEM_PTR(CM_SIDR_REP_ADDITIONAL_INFORMATION,
+				      sidr_rep_msg);
 	param->info_len = IBA_GET(CM_SIDR_REP_ADDITIONAL_INFORMATION_LENGTH,
 				  sidr_rep_msg);
 	param->sgid_attr = cm_id_priv->av.ah_attr.grh.sgid_attr;
-	work->cm_event.private_data = &sidr_rep_msg->private_data;
+	work->cm_event.private_data =
+		IBA_GET_MEM_PTR(CM_SIDR_REP_PRIVATE_DATA, sidr_rep_msg);
 }
 
 static int cm_sidr_rep_handler(struct cm_work *work)
-- 
cgit v1.2.3-59-g8ed1b


From 13e0af1801f2e74639b4eadb69ed1fad2cf802f7 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Thu, 16 Jan 2020 13:00:37 -0400
Subject: RDMA/cm: Remove CM message structs

All accesses now use the new IBA acessor scheme, so delete the structs
entirely and generate the structures from the schema file.

Link: https://lore.kernel.org/r/20200116170037.30109-8-jgg@ziepe.ca
Tested-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cm.c      |  77 --------------
 drivers/infiniband/core/cm_msgs.h | 211 --------------------------------------
 include/rdma/ibta_vol1_c12.h      |   7 +-
 3 files changed, 6 insertions(+), 289 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 5ccd59f1ebb8..68cc1b2d6824 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -4426,83 +4426,6 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data)
 	kfree(cm_dev);
 }
 
-/*
- * Check at compile time that the byte offset and length of field old_name in
- * the struct matches the byte offset and length in the new macro.
- */
-#define _IBA_CHECK_OFF(old_name, field_struct, field_offset, mask, bits) \
-	static_assert(offsetof(field_struct, old_name) == (field_offset));     \
-	static_assert(bits == sizeof(((field_struct *)0)->old_name) * 8)
-#define IBA_CHECK_OFF(field, old_name) _IBA_CHECK_OFF(old_name, field)
-
-IBA_CHECK_OFF(CM_REQ_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_REQ_SERVICE_ID, service_id);
-IBA_CHECK_OFF(CM_REQ_LOCAL_CA_GUID, local_ca_guid);
-IBA_CHECK_OFF(CM_REQ_LOCAL_Q_KEY, local_qkey);
-IBA_CHECK_OFF(CM_REQ_PARTITION_KEY, pkey);
-IBA_CHECK_OFF(CM_REQ_PRIMARY_LOCAL_PORT_LID, primary_local_lid);
-IBA_CHECK_OFF(CM_REQ_PRIMARY_REMOTE_PORT_LID, primary_remote_lid);
-IBA_CHECK_OFF(CM_REQ_PRIMARY_LOCAL_PORT_GID, primary_local_gid);
-IBA_CHECK_OFF(CM_REQ_PRIMARY_REMOTE_PORT_GID, primary_remote_gid);
-IBA_CHECK_OFF(CM_REQ_PRIMARY_TRAFFIC_CLASS, primary_traffic_class);
-IBA_CHECK_OFF(CM_REQ_PRIMARY_HOP_LIMIT, primary_hop_limit);
-IBA_CHECK_OFF(CM_REQ_ALTERNATE_LOCAL_PORT_LID, alt_local_lid);
-IBA_CHECK_OFF(CM_REQ_ALTERNATE_REMOTE_PORT_LID, alt_remote_lid);
-IBA_CHECK_OFF(CM_REQ_ALTERNATE_LOCAL_PORT_GID, alt_local_gid);
-IBA_CHECK_OFF(CM_REQ_ALTERNATE_REMOTE_PORT_GID, alt_remote_gid);
-IBA_CHECK_OFF(CM_REQ_ALTERNATE_TRAFFIC_CLASS, alt_traffic_class);
-IBA_CHECK_OFF(CM_REQ_ALTERNATE_HOP_LIMIT, alt_hop_limit);
-IBA_CHECK_OFF(CM_REQ_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_MRA_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_MRA_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_MRA_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_REJ_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_REJ_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_REJ_REASON, reason);
-IBA_CHECK_OFF(CM_REJ_ARI, ari);
-IBA_CHECK_OFF(CM_REJ_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_REP_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_REP_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_REP_LOCAL_Q_KEY, local_qkey);
-IBA_CHECK_OFF(CM_REP_RESPONDER_RESOURCES, resp_resources);
-IBA_CHECK_OFF(CM_REP_INITIATOR_DEPTH, initiator_depth);
-IBA_CHECK_OFF(CM_REP_LOCAL_CA_GUID, local_ca_guid);
-IBA_CHECK_OFF(CM_REP_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_RTU_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_RTU_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_RTU_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_DREQ_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_DREQ_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_DREQ_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_DREP_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_DREP_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_DREP_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_LAP_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_LAP_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_LAP_ALTERNATE_LOCAL_PORT_LID, alt_local_lid);
-IBA_CHECK_OFF(CM_LAP_ALTERNATE_REMOTE_PORT_LID, alt_remote_lid);
-IBA_CHECK_OFF(CM_LAP_ALTERNATE_LOCAL_PORT_GID, alt_local_gid);
-IBA_CHECK_OFF(CM_LAP_ALTERNATE_REMOTE_PORT_GID, alt_remote_gid);
-IBA_CHECK_OFF(CM_LAP_ALTERNATE_HOP_LIMIT, alt_hop_limit);
-IBA_CHECK_OFF(CM_LAP_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_APR_LOCAL_COMM_ID, local_comm_id);
-IBA_CHECK_OFF(CM_APR_REMOTE_COMM_ID, remote_comm_id);
-IBA_CHECK_OFF(CM_APR_ADDITIONAL_INFORMATION_LENGTH, info_length);
-IBA_CHECK_OFF(CM_APR_AR_STATUS, ap_status);
-IBA_CHECK_OFF(CM_APR_ADDITIONAL_INFORMATION, info);
-IBA_CHECK_OFF(CM_APR_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_SIDR_REQ_REQUESTID, request_id);
-IBA_CHECK_OFF(CM_SIDR_REQ_PARTITION_KEY, pkey);
-IBA_CHECK_OFF(CM_SIDR_REQ_SERVICEID, service_id);
-IBA_CHECK_OFF(CM_SIDR_REQ_PRIVATE_DATA, private_data);
-IBA_CHECK_OFF(CM_SIDR_REP_REQUESTID, request_id);
-IBA_CHECK_OFF(CM_SIDR_REP_STATUS, status);
-IBA_CHECK_OFF(CM_SIDR_REP_ADDITIONAL_INFORMATION_LENGTH, info_length);
-IBA_CHECK_OFF(CM_SIDR_REP_SERVICEID, service_id);
-IBA_CHECK_OFF(CM_SIDR_REP_Q_KEY, qkey);
-IBA_CHECK_OFF(CM_SIDR_REP_ADDITIONAL_INFORMATION, info);
-IBA_CHECK_OFF(CM_SIDR_REP_PRIVATE_DATA, private_data);
-
 static int __init ib_cm_init(void)
 {
 	int ret;
diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h
index 86ab6952d5d8..0cc40656b5c5 100644
--- a/drivers/infiniband/core/cm_msgs.h
+++ b/drivers/infiniband/core/cm_msgs.h
@@ -19,62 +19,6 @@
 
 #define IB_CM_CLASS_VERSION	2 /* IB specification 1.2 */
 
-struct cm_req_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 rsvd4;
-	__be64 service_id;
-	__be64 local_ca_guid;
-	__be32 rsvd24;
-	__be32 local_qkey;
-	/* local QPN:24, responder resources:8 */
-	__be32 offset32;
-	/* local EECN:24, initiator depth:8 */
-	__be32 offset36;
-	/*
-	 * remote EECN:24, remote CM response timeout:5,
-	 * transport service type:2, end-to-end flow control:1
-	 */
-	__be32 offset40;
-	/* starting PSN:24, local CM response timeout:5, retry count:3 */
-	__be32 offset44;
-	__be16 pkey;
-	/* path MTU:4, RDC exists:1, RNR retry count:3. */
-	u8 offset50;
-	/* max CM Retries:4, SRQ:1, extended transport type:3 */
-	u8 offset51;
-
-	__be16 primary_local_lid;
-	__be16 primary_remote_lid;
-	union ib_gid primary_local_gid;
-	union ib_gid primary_remote_gid;
-	/* flow label:20, rsvd:6, packet rate:6 */
-	__be32 primary_offset88;
-	u8 primary_traffic_class;
-	u8 primary_hop_limit;
-	/* SL:4, subnet local:1, rsvd:3 */
-	u8 primary_offset94;
-	/* local ACK timeout:5, rsvd:3 */
-	u8 primary_offset95;
-
-	__be16 alt_local_lid;
-	__be16 alt_remote_lid;
-	union ib_gid alt_local_gid;
-	union ib_gid alt_remote_gid;
-	/* flow label:20, rsvd:6, packet rate:6 */
-	__be32 alt_offset132;
-	u8 alt_traffic_class;
-	u8 alt_hop_limit;
-	/* SL:4, subnet local:1, rsvd:3 */
-	u8 alt_offset138;
-	/* local ACK timeout:5, rsvd:3 */
-	u8 alt_offset139;
-
-	u32 private_data[IB_CM_REQ_PRIVATE_DATA_SIZE / sizeof(u32)];
-
-} __packed;
-
 static inline enum ib_qp_type cm_req_get_qp_type(struct cm_req_msg *req_msg)
 {
 	u8 transport_type = IBA_GET(CM_REQ_TRANSPORT_SERVICE_TYPE, req_msg);
@@ -113,60 +57,6 @@ enum cm_msg_response {
 	CM_MSG_RESPONSE_OTHER = 0x2
 };
 
- struct cm_mra_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-	/* message MRAed:2, rsvd:6 */
-	u8 offset8;
-	/* service timeout:5, rsvd:3 */
-	u8 offset9;
-
-	u8 private_data[IB_CM_MRA_PRIVATE_DATA_SIZE];
-
-} __packed;
-
-struct cm_rej_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-	/* message REJected:2, rsvd:6 */
-	u8 offset8;
-	/* reject info length:7, rsvd:1. */
-	u8 offset9;
-	__be16 reason;
-	u8 ari[IB_CM_REJ_ARI_LENGTH];
-
-	u8 private_data[IB_CM_REJ_PRIVATE_DATA_SIZE];
-
-} __packed;
-
-struct cm_rep_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-	__be32 local_qkey;
-	/* local QPN:24, rsvd:8 */
-	__be32 offset12;
-	/* local EECN:24, rsvd:8 */
-	__be32 offset16;
-	/* starting PSN:24 rsvd:8 */
-	__be32 offset20;
-	u8 resp_resources;
-	u8 initiator_depth;
-	/* target ACK delay:5, failover accepted:2, end-to-end flow control:1 */
-	u8 offset26;
-	/* RNR retry count:3, SRQ:1, rsvd:5 */
-	u8 offset27;
-	__be64 local_ca_guid;
-
-	u8 private_data[IB_CM_REP_PRIVATE_DATA_SIZE];
-
-} __packed;
-
 static inline __be32 cm_rep_get_qpn(struct cm_rep_msg *rep_msg, enum ib_qp_type qp_type)
 {
 	return (qp_type == IB_QPT_XRC_INI) ?
@@ -175,105 +65,4 @@ static inline __be32 cm_rep_get_qpn(struct cm_rep_msg *rep_msg, enum ib_qp_type
 		       cpu_to_be32(IBA_GET(CM_REP_LOCAL_QPN, rep_msg));
 }
 
-struct cm_rtu_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-
-	u8 private_data[IB_CM_RTU_PRIVATE_DATA_SIZE];
-
-} __packed;
-
-struct cm_dreq_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-	/* remote QPN/EECN:24, rsvd:8 */
-	__be32 offset8;
-
-	u8 private_data[IB_CM_DREQ_PRIVATE_DATA_SIZE];
-
-} __packed;
-
-struct cm_drep_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-
-	u8 private_data[IB_CM_DREP_PRIVATE_DATA_SIZE];
-
-} __packed;
-
-struct cm_lap_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-
-	__be32 rsvd8;
-	/* remote QPN/EECN:24, remote CM response timeout:5, rsvd:3 */
-	__be32 offset12;
-	__be32 rsvd16;
-
-	__be16 alt_local_lid;
-	__be16 alt_remote_lid;
-	union ib_gid alt_local_gid;
-	union ib_gid alt_remote_gid;
-	/* flow label:20, rsvd:4, traffic class:8 */
-	__be32 offset56;
-	u8 alt_hop_limit;
-	/* rsvd:2, packet rate:6 */
-	u8 offset61;
-	/* SL:4, subnet local:1, rsvd:3 */
-	u8 offset62;
-	/* local ACK timeout:5, rsvd:3 */
-	u8 offset63;
-
-	u8 private_data[IB_CM_LAP_PRIVATE_DATA_SIZE];
-} __packed;
-
-struct cm_apr_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 local_comm_id;
-	__be32 remote_comm_id;
-
-	u8 info_length;
-	u8 ap_status;
-	__be16 rsvd;
-	u8 info[IB_CM_APR_INFO_LENGTH];
-
-	u8 private_data[IB_CM_APR_PRIVATE_DATA_SIZE];
-} __packed;
-
-struct cm_sidr_req_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 request_id;
-	__be16 pkey;
-	__be16 rsvd;
-	__be64 service_id;
-
-	u32 private_data[IB_CM_SIDR_REQ_PRIVATE_DATA_SIZE / sizeof(u32)];
-} __packed;
-
-struct cm_sidr_rep_msg {
-	struct ib_mad_hdr hdr;
-
-	__be32 request_id;
-	u8 status;
-	u8 info_length;
-	__be16 rsvd;
-	/* QPN:24, rsvd:8 */
-	__be32 offset8;
-	__be64 service_id;
-	__be32 qkey;
-	u8 info[IB_CM_SIDR_REP_INFO_LENGTH];
-
-	u8 private_data[IB_CM_SIDR_REP_PRIVATE_DATA_SIZE];
-} __packed;
-
 #endif /* CM_MSGS_H */
diff --git a/include/rdma/ibta_vol1_c12.h b/include/rdma/ibta_vol1_c12.h
index 916db5c27dc5..269904425d3f 100644
--- a/include/rdma/ibta_vol1_c12.h
+++ b/include/rdma/ibta_vol1_c12.h
@@ -29,7 +29,12 @@
 	IBA_FIELD_MLOC(field_struct,                                           \
 		       (byte_offset + sizeof(struct ib_mad_hdr)), width, type)
 #define CM_STRUCT(field_struct, total_len)                                     \
-	static_assert((total_len) % 32 == 0);
+	field_struct                                                           \
+	{                                                                      \
+		struct ib_mad_hdr hdr;                                         \
+		u32 _data[(total_len) / 32 +                                   \
+			  BUILD_BUG_ON_ZERO((total_len) % 32 != 0)];           \
+	}
 
 /* Table 106 REQ Message Contents */
 #define CM_REQ_LOCAL_COMM_ID CM_FIELD32_LOC(struct cm_req_msg, 0, 32)
-- 
cgit v1.2.3-59-g8ed1b


From 79ba4f9310673a8870fda5a062a558923fa65c93 Mon Sep 17 00:00:00 2001
From: Nathan Chancellor <natechancellor@gmail.com>
Date: Thu, 16 Jan 2020 15:26:58 -0700
Subject: IB/hfi1: Fix logical condition in msix_request_irq

Clang warns:

drivers/infiniband/hw/hfi1/msix.c:136:22: warning: overlapping
comparisons always evaluate to false [-Wtautological-overlap-compare]
        if (type < IRQ_SDMA && type >= IRQ_OTHER)
            ~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~
1 warning generated.

It is impossible for something to be less than 0 (IRQ_SDMA) and greater
than or equal to 3 (IRQ_OTHER) at the same time. A logical OR should
have been used to keep the same logic as before.

Link: https://lore.kernel.org/r/20200116222658.5285-1-natechancellor@gmail.com
Link: https://github.com/ClangBuiltLinux/linux/issues/841
Fixes: 13d2a8384bd9 ("IB/hfi1: Decouple IRQ name from type")
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Acked-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/msix.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/hfi1/msix.c b/drivers/infiniband/hw/hfi1/msix.c
index 4a620cf80588..db82db497b2c 100644
--- a/drivers/infiniband/hw/hfi1/msix.c
+++ b/drivers/infiniband/hw/hfi1/msix.c
@@ -133,7 +133,7 @@ static int msix_request_irq(struct hfi1_devdata *dd, void *arg,
 	if (nr == dd->msix_info.max_requested)
 		return -ENOSPC;
 
-	if (type < IRQ_SDMA && type >= IRQ_OTHER)
+	if (type < IRQ_SDMA || type >= IRQ_OTHER)
 		return -EINVAL;
 
 	irq = pci_irq_vector(dd->pcidev, nr);
-- 
cgit v1.2.3-59-g8ed1b


From 7f04c71f1fc2618f2aeb2d41c65e3b01ee20b3d5 Mon Sep 17 00:00:00 2001
From: Dillon Brock <dab9861@gmail.com>
Date: Sat, 18 Jan 2020 11:25:42 -0500
Subject: IB/opa_vnic: Spelling correction of 'erorr' to 'error'

Correcting a minor spelling mistake in the comments.

Link: https://lore.kernel.org/r/20200118162542.15188-1-dab9861@gmail.com
Signed-off-by: Dillon Brock <dab9861@gmail.com>
Acked-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h b/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h
index e4c9bf2ef7e2..4480092c68e0 100644
--- a/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h
+++ b/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h
@@ -358,7 +358,7 @@ struct opa_veswport_summary_counters {
  * @rx_drop_state: received packets in non-forwarding port state
  * @rx_logic: other receive errors
  *
- * All the above are counters of corresponding erorr conditions.
+ * All the above are counters of corresponding error conditions.
  */
 struct opa_veswport_error_counters {
 	__be16  vp_instance;
-- 
cgit v1.2.3-59-g8ed1b


From ea660ad7c1c476fd6e5e3b17780d47159db71dea Mon Sep 17 00:00:00 2001
From: Håkon Bugge <haakon.bugge@oracle.com>
Date: Thu, 23 Jan 2020 16:55:21 +0100
Subject: IB/mlx4: Fix leak in id_map_find_del
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Using CX-3 virtual functions, either from a bare-metal machine or
pass-through from a VM, MAD packets are proxied through the PF driver.

Since the VF drivers have separate name spaces for MAD Transaction Ids
(TIDs), the PF driver has to re-map the TIDs and keep the book keeping in
a cache.

Following the RDMA Connection Manager (CM) protocol, it is clear when an
entry has to evicted from the cache. When a DREP is sent from
mlx4_ib_multiplex_cm_handler(), id_map_find_del() is called. Similar when
a REJ is received by the mlx4_ib_demux_cm_handler(), id_map_find_del() is
called.

This function wipes out the TID in use from the IDR or XArray and removes
the id_map_entry from the table.

In short, it does everything except the topping of the cake, which is to
remove the entry from the list and free it. In other words, for the REJ
case enumerated above, one id_map_entry will be leaked.

For the other case above, a DREQ has been received first. The reception of
the DREQ will trigger queuing of a delayed work to delete the
id_map_entry, for the case where the VM doesn't send back a DREP.

In the normal case, the VM _will_ send back a DREP, and id_map_find_del()
will be called.

But this scenario introduces a secondary leak. First, when the DREQ is
received, a delayed work is queued. The VM will then return a DREP, which
will call id_map_find_del(). As stated above, this will free the TID used
from the XArray or IDR. Now, there is window where that particular TID can
be re-allocated, lets say by an outgoing REQ. This TID will later be wiped
out by the delayed work, when the function id_map_ent_timeout() is
called. But the id_map_entry allocated by the outgoing REQ will not be
de-allocated, and we have a leak.

Both leaks are fixed by removing the id_map_find_del() function and only
using schedule_delayed(). Of course, a check in schedule_delayed() to see
if the work already has been queued, has been added.

Another benefit of always using the delayed version for deleting entries,
is that we do get a TimeWait effect; a TID no longer in use, will occupy
the XArray or IDR for CM_CLEANUP_CACHE_TIMEOUT time, without any ability
of being re-used for that time period.

Fixes: 3cf69cc8dbeb ("IB/mlx4: Add CM paravirtualization")
Link: https://lore.kernel.org/r/20200123155521.1212288-1-haakon.bugge@oracle.com
Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Signed-off-by: Manjunath Patil <manjunath.b.patil@oracle.com>
Reviewed-by: Rama Nichanamatlu <rama.nichanamatlu@oracle.com>
Reviewed-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx4/cm.c | 29 +++--------------------------
 1 file changed, 3 insertions(+), 26 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/mlx4/cm.c b/drivers/infiniband/hw/mlx4/cm.c
index ecd6cadd529a..b591861934b3 100644
--- a/drivers/infiniband/hw/mlx4/cm.c
+++ b/drivers/infiniband/hw/mlx4/cm.c
@@ -186,23 +186,6 @@ out:
 	kfree(ent);
 }
 
-static void id_map_find_del(struct ib_device *ibdev, int pv_cm_id)
-{
-	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
-	struct rb_root *sl_id_map = &sriov->sl_id_map;
-	struct id_map_entry *ent, *found_ent;
-
-	spin_lock(&sriov->id_map_lock);
-	ent = xa_erase(&sriov->pv_id_table, pv_cm_id);
-	if (!ent)
-		goto out;
-	found_ent = id_map_find_by_sl_id(ibdev, ent->slave_id, ent->sl_cm_id);
-	if (found_ent && found_ent == ent)
-		rb_erase(&found_ent->node, sl_id_map);
-out:
-	spin_unlock(&sriov->id_map_lock);
-}
-
 static void sl_id_map_add(struct ib_device *ibdev, struct id_map_entry *new)
 {
 	struct rb_root *sl_id_map = &to_mdev(ibdev)->sriov.sl_id_map;
@@ -294,7 +277,7 @@ static void schedule_delayed(struct ib_device *ibdev, struct id_map_entry *id)
 	spin_lock(&sriov->id_map_lock);
 	spin_lock_irqsave(&sriov->going_down_lock, flags);
 	/*make sure that there is no schedule inside the scheduled work.*/
-	if (!sriov->is_going_down) {
+	if (!sriov->is_going_down && !id->scheduled_delete) {
 		id->scheduled_delete = 1;
 		schedule_delayed_work(&id->timeout, CM_CLEANUP_CACHE_TIMEOUT);
 	}
@@ -341,9 +324,6 @@ cont:
 
 	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
 		schedule_delayed(ibdev, id);
-	else if (mad->mad_hdr.attr_id == CM_DREP_ATTR_ID)
-		id_map_find_del(ibdev, pv_cm_id);
-
 	return 0;
 }
 
@@ -382,12 +362,9 @@ int mlx4_ib_demux_cm_handler(struct ib_device *ibdev, int port, int *slave,
 		*slave = id->slave_id;
 	set_remote_comm_id(mad, id->sl_cm_id);
 
-	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
+	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID ||
+	    mad->mad_hdr.attr_id == CM_REJ_ATTR_ID)
 		schedule_delayed(ibdev, id);
-	else if (mad->mad_hdr.attr_id == CM_REJ_ATTR_ID ||
-			mad->mad_hdr.attr_id == CM_DREP_ATTR_ID) {
-		id_map_find_del(ibdev, (int) pv_cm_id);
-	}
 
 	return 0;
 }
-- 
cgit v1.2.3-59-g8ed1b


From 36798d5ae1af62e830c5e045b2e41ce038690c61 Mon Sep 17 00:00:00 2001
From: Artemy Kovalyov <artemyko@mellanox.com>
Date: Tue, 28 Jan 2020 15:56:12 +0200
Subject: RDMA/umem: Fix ib_umem_find_best_pgsz()

Except for the last entry, the ending iova alignment sets the maximum
possible page size as the low bits of the iova must be zero when starting
the next chunk.

Fixes: 4a35339958f1 ("RDMA/umem: Add API to find best driver supported page size in an MR")
Link: https://lore.kernel.org/r/20200128135612.174820-1-leon@kernel.org
Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Tested-by: Gal Pressman <galpress@amazon.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/umem.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c
index 146f98fbf22b..ac4738d2e0dd 100644
--- a/drivers/infiniband/core/umem.c
+++ b/drivers/infiniband/core/umem.c
@@ -166,10 +166,13 @@ unsigned long ib_umem_find_best_pgsz(struct ib_umem *umem,
 		 * for any address.
 		 */
 		mask |= (sg_dma_address(sg) + pgoff) ^ va;
-		if (i && i != (umem->nmap - 1))
-			/* restrict by length as well for interior SGEs */
-			mask |= sg_dma_len(sg);
 		va += sg_dma_len(sg) - pgoff;
+		/* Except for the last entry, the ending iova alignment sets
+		 * the maximum possible page size as the low bits of the iova
+		 * must be zero when starting the next chunk.
+		 */
+		if (i != (umem->nmap - 1))
+			mask |= va;
 		pgoff = 0;
 	}
 	best_pg_bit = rdma_find_pg_bit(mask, pgsz_bitmap);
-- 
cgit v1.2.3-59-g8ed1b


From b4fb4cc5ba83b20dae13cef116c33648e81d2f44 Mon Sep 17 00:00:00 2001
From: Parav Pandit <parav@mellanox.com>
Date: Sun, 26 Jan 2020 16:26:46 +0200
Subject: RDMA/cma: Fix unbalanced cm_id reference count during address resolve

Below commit missed the AF_IB and loopback code flow in
rdma_resolve_addr().  This leads to an unbalanced cm_id refcount in
cma_work_handler() which puts the refcount which was not incremented prior
to queuing the work.

A call trace is observed with such code flow:

 BUG: unable to handle kernel NULL pointer dereference at (null)
 [<ffffffff96b67e16>] __mutex_lock_slowpath+0x166/0x1d0
 [<ffffffff96b6715f>] mutex_lock+0x1f/0x2f
 [<ffffffffc0beabb5>] cma_work_handler+0x25/0xa0
 [<ffffffff964b9ebf>] process_one_work+0x17f/0x440
 [<ffffffff964baf56>] worker_thread+0x126/0x3c0

Hence, hold the cm_id reference when scheduling the resolve work item.

Fixes: 722c7b2bfead ("RDMA/{cma, core}: Avoid callback on rdma_addr_cancel()")
Link: https://lore.kernel.org/r/20200126142652.104803-2-leon@kernel.org
Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/cma.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 55a9afacfedd..72f032160c4b 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -3148,6 +3148,7 @@ static int cma_resolve_loopback(struct rdma_id_private *id_priv)
 	rdma_addr_get_sgid(&id_priv->id.route.addr.dev_addr, &gid);
 	rdma_addr_set_dgid(&id_priv->id.route.addr.dev_addr, &gid);
 
+	atomic_inc(&id_priv->refcount);
 	cma_init_resolve_addr_work(work, id_priv);
 	queue_work(cma_wq, &work->work);
 	return 0;
@@ -3174,6 +3175,7 @@ static int cma_resolve_ib_addr(struct rdma_id_private *id_priv)
 	rdma_addr_set_dgid(&id_priv->id.route.addr.dev_addr, (union ib_gid *)
 		&(((struct sockaddr_ib *) &id_priv->id.route.addr.dst_addr)->sib_addr));
 
+	atomic_inc(&id_priv->refcount);
 	cma_init_resolve_addr_work(work, id_priv);
 	queue_work(cma_wq, &work->work);
 	return 0;
-- 
cgit v1.2.3-59-g8ed1b


From ba19e16651a5c67fa8b53ce4fb5bfbab8d326b6c Mon Sep 17 00:00:00 2001
From: Gal Pressman <galpress@amazon.com>
Date: Wed, 29 Jan 2020 09:18:03 +0200
Subject: RDMA/efa: Mask access flags with the correct optional range

The uapi value IB_UVERBS_ACCESS_OPTIONAL_RANGE shouldn't be used inside
the driver, use IB_ACCESS_OPTIONAL instead.

Fixes: 86dd738cf20c ("RDMA/efa: Allow passing of optional access flags for MR registration")
Link: https://lore.kernel.org/r/20200129071803.40117-1-galpress@amazon.com
Signed-off-by: Gal Pressman <galpress@amazon.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/efa/efa_verbs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c
index 4c7b7800a861..ec5545870554 100644
--- a/drivers/infiniband/hw/efa/efa_verbs.c
+++ b/drivers/infiniband/hw/efa/efa_verbs.c
@@ -1367,7 +1367,7 @@ struct ib_mr *efa_reg_mr(struct ib_pd *ibpd, u64 start, u64 length,
 		IB_ACCESS_LOCAL_WRITE |
 		(is_rdma_read_cap(dev) ? IB_ACCESS_REMOTE_READ : 0);
 
-	access_flags &= ~IB_UVERBS_ACCESS_OPTIONAL_RANGE;
+	access_flags &= ~IB_ACCESS_OPTIONAL;
 	if (access_flags & ~supp_access_flags) {
 		ibdev_dbg(&dev->ibdev,
 			  "Unsupported access flags[%#x], supported[%#x]\n",
-- 
cgit v1.2.3-59-g8ed1b


From 8889f6fa35884d09f24734e10fea0c9ddcbc6429 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@mellanox.com>
Date: Thu, 30 Jan 2020 11:21:21 -0400
Subject: RDMA/core: Make the entire API tree static

Compilation of mlx5 driver without CONFIG_INFINIBAND_USER_ACCESS generates
the following error.

on x86_64:

 ld: drivers/infiniband/hw/mlx5/main.o: in function `mlx5_ib_handler_MLX5_IB_METHOD_VAR_OBJ_ALLOC':
 main.c:(.text+0x186d): undefined reference to `ib_uverbs_get_ucontext_file'
 ld: drivers/infiniband/hw/mlx5/main.o:(.rodata+0x2480): undefined reference to `uverbs_idr_class'
 ld: drivers/infiniband/hw/mlx5/main.o:(.rodata+0x24d8): undefined reference to `uverbs_destroy_def_handler'

This is happening because some parts of the UAPI description are not
static. This is a hold over from earlier code that relied on struct
pointers to refer to object types, now object types are referenced by
number. Remove the unused globals and add statics to the remaining UAPI
description elements.

Remove the redundent #ifdefs around mlx5_ib_*defs and obsolete
mlx5_ib_get_devx_tree().

The compiler now trims alot more unused code, including the above
problematic definitions when !CONFIG_INFINIBAND_USER_ACCESS.

Fixes: 7be76bef320b ("IB/mlx5: Introduce VAR object and its alloc/destroy methods")
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/core/uverbs.h     | 17 -----------------
 drivers/infiniband/hw/mlx5/main.c    |  2 --
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  7 +++----
 include/rdma/uverbs_named_ioctl.h    |  6 +++---
 4 files changed, 6 insertions(+), 26 deletions(-)

(limited to 'drivers/infiniband')

diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h
index 4d4cec46d251..7df71983212d 100644
--- a/drivers/infiniband/core/uverbs.h
+++ b/drivers/infiniband/core/uverbs.h
@@ -271,23 +271,6 @@ int ib_uverbs_kern_spec_to_ib_spec_filter(enum ib_flow_spec_type type,
 					  size_t kern_filter_sz,
 					  union ib_flow_spec *ib_spec);
 
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_DEVICE);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_PD);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_MR);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_COMP_CHANNEL);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_CQ);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_QP);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_AH);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_MW);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_SRQ);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_FLOW);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_WQ);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_RWQ_IND_TBL);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_XRCD);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_FLOW_ACTION);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_DM);
-extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_COUNTERS);
-
 /*
  * ib_uverbs_query_port_resp.port_cap_flags started out as just a copy of the
  * PortInfo CapabilityMask, but was extended with unique bits.
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 01fc09f3ddd3..0ca958143280 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -6247,10 +6247,8 @@ ADD_UVERBS_ATTRIBUTES_SIMPLE(
 			     enum mlx5_ib_uapi_flow_action_flags));
 
 static const struct uapi_definition mlx5_ib_defs[] = {
-#if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)
 	UAPI_DEF_CHAIN(mlx5_ib_devx_defs),
 	UAPI_DEF_CHAIN(mlx5_ib_flow_defs),
-#endif
 
 	UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION,
 				&mlx5_ib_flow_action),
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 7b019bd4de4b..d9bffcc93587 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1379,14 +1379,14 @@ int mlx5_ib_fill_res_entry(struct sk_buff *msg,
 int mlx5_ib_fill_stat_entry(struct sk_buff *msg,
 			    struct rdma_restrack_entry *res);
 
+extern const struct uapi_definition mlx5_ib_devx_defs[];
+extern const struct uapi_definition mlx5_ib_flow_defs[];
+
 #if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)
 int mlx5_ib_devx_create(struct mlx5_ib_dev *dev, bool is_user);
 void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid);
 void mlx5_ib_devx_init_event_table(struct mlx5_ib_dev *dev);
 void mlx5_ib_devx_cleanup_event_table(struct mlx5_ib_dev *dev);
-const struct uverbs_object_tree_def *mlx5_ib_get_devx_tree(void);
-extern const struct uapi_definition mlx5_ib_devx_defs[];
-extern const struct uapi_definition mlx5_ib_flow_defs[];
 struct mlx5_ib_flow_handler *mlx5_ib_raw_fs_rule_add(
 	struct mlx5_ib_dev *dev, struct mlx5_ib_flow_matcher *fs_matcher,
 	struct mlx5_flow_context *flow_context,
@@ -1394,7 +1394,6 @@ struct mlx5_ib_flow_handler *mlx5_ib_raw_fs_rule_add(
 	void *cmd_in, int inlen, int dest_id, int dest_type);
 bool mlx5_ib_devx_is_flow_dest(void *obj, int *dest_id, int *dest_type);
 bool mlx5_ib_devx_is_flow_counter(void *obj, u32 offset, u32 *counter_id);
-int mlx5_ib_get_flow_trees(const struct uverbs_object_tree_def **root);
 void mlx5_ib_destroy_flow_action_raw(struct mlx5_ib_flow_action *maction);
 #else
 static inline int
diff --git a/include/rdma/uverbs_named_ioctl.h b/include/rdma/uverbs_named_ioctl.h
index 3447bfe356d6..6ae6cf8e4c2e 100644
--- a/include/rdma/uverbs_named_ioctl.h
+++ b/include/rdma/uverbs_named_ioctl.h
@@ -76,7 +76,7 @@
 #define DECLARE_UVERBS_NAMED_OBJECT(_object_id, _type_attrs, ...)              \
 	static const struct uverbs_method_def *const UVERBS_OBJECT_METHODS(    \
 		_object_id)[] = { __VA_ARGS__ };                               \
-	const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {           \
+	static const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {    \
 		.id = _object_id,                                              \
 		.type_attrs = &_type_attrs,                                    \
 		.num_methods = ARRAY_SIZE(UVERBS_OBJECT_METHODS(_object_id)),  \
@@ -88,10 +88,10 @@
  * identify all uapi methods with a (object,method) tuple. However, they have
  * no type pointer.
  */
-#define DECLARE_UVERBS_GLOBAL_METHODS(_object_id, ...)	\
+#define DECLARE_UVERBS_GLOBAL_METHODS(_object_id, ...)                         \
 	static const struct uverbs_method_def *const UVERBS_OBJECT_METHODS(    \
 		_object_id)[] = { __VA_ARGS__ };                               \
-	const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {           \
+	static const struct uverbs_object_def UVERBS_OBJECT(_object_id) = {    \
 		.id = _object_id,                                              \
 		.num_methods = ARRAY_SIZE(UVERBS_OBJECT_METHODS(_object_id)),  \
 		.methods = &UVERBS_OBJECT_METHODS(_object_id)                  \
-- 
cgit v1.2.3-59-g8ed1b