From 80d5e8a235624cd3c0e24be7c070fd6f445e590d Mon Sep 17 00:00:00 2001 From: Bart Van Assche Date: Wed, 10 Jul 2013 17:36:35 +0200 Subject: IB/srp: Let srp_abort() return FAST_IO_FAIL if TL offline If the transport layer is offline it is more appropriate to let srp_abort() return FAST_IO_FAIL instead of SUCCESS. Reported-by: Sebastian Riemer Acked-by: David Dillow Signed-off-by: Bart Van Assche Signed-off-by: Roland Dreier --- drivers/infiniband/ulp/srp/ib_srp.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c index 9d8b46eafdb2..f93baf8254c4 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.c +++ b/drivers/infiniband/ulp/srp/ib_srp.c @@ -1753,8 +1753,7 @@ static int srp_abort(struct scsi_cmnd *scmnd) if (!req || !srp_claim_req(target, req, scmnd)) return FAILED; if (srp_send_tsk_mgmt(target, req->index, scmnd->device->lun, - SRP_TSK_ABORT_TASK) == 0 || - target->transport_offline) + SRP_TSK_ABORT_TASK) == 0) ret = SUCCESS; else if (target->transport_offline) ret = FAST_IO_FAIL; -- cgit v1.2.3-59-g8ed1b From 288dde9f23b6726c1e8147bf635721372bf77b16 Mon Sep 17 00:00:00 2001 From: Moshe Lazer Date: Wed, 10 Jul 2013 14:31:03 +0300 Subject: mlx5_core: Adjust hca_cap.uar_page_sz to conform to Connect-IB spec Sparse reported an endianness bug in the assignment to hca_cap.uar_page_sz. Fix the declaration of this field to be __be16 (which is what is in the firmware spec), renaming the field to log_uar_pg_size to conform to the spec, which fixes the endianness bug reported by sparse. Reported-by: Fengguang Wu Signed-off-by: Moshe Lazer Signed-off-by: Or Gerlitz Signed-off-by: Roland Dreier --- drivers/net/ethernet/mellanox/mlx5/core/main.c | 2 +- include/linux/mlx5/device.h | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'drivers') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c index f21cc397d1bc..12242de2b0e3 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c @@ -212,7 +212,7 @@ static int handle_hca_cap(struct mlx5_core_dev *dev) set_ctx->hca_cap.log_max_qp = dev->profile->log_max_qp; memset(&set_out, 0, sizeof(set_out)); - set_ctx->hca_cap.uar_page_sz = cpu_to_be16(PAGE_SHIFT - 12); + set_ctx->hca_cap.log_uar_page_sz = cpu_to_be16(PAGE_SHIFT - 12); set_ctx->hdr.opcode = cpu_to_be16(MLX5_CMD_OP_SET_HCA_CAP); err = mlx5_cmd_exec(dev, set_ctx, sizeof(*set_ctx), &set_out, sizeof(set_out)); diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h index 51390915e538..8de8d8f22384 100644 --- a/include/linux/mlx5/device.h +++ b/include/linux/mlx5/device.h @@ -317,8 +317,8 @@ struct mlx5_hca_cap { u8 log_max_pd; u8 rsvd25; u8 log_max_xrcd; - u8 rsvd26[40]; - __be32 uar_page_sz; + u8 rsvd26[42]; + __be16 log_uar_page_sz; u8 rsvd27[28]; u8 log_msx_atomic_size_qp; u8 rsvd28[2]; -- cgit v1.2.3-59-g8ed1b From 308c813b19cb676df7e5e70b5f014fa56e918677 Mon Sep 17 00:00:00 2001 From: Mike Marciniszyn Date: Wed, 3 Jul 2013 13:50:28 -0400 Subject: IB/qib: Fix module-level leak The vzalloc()'ed field physshadow is leaked on module unload. This patch adds vfree after the sibling page shadow is freed. Reported-by: Dean Luick Reviewed-by: Dean Luick Signed-off-by: Mike Marciniszyn Signed-off-by: Roland Dreier --- drivers/infiniband/hw/qib/qib_init.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'drivers') diff --git a/drivers/infiniband/hw/qib/qib_init.c b/drivers/infiniband/hw/qib/qib_init.c index fdae42973056..36e048e0e1d9 100644 --- a/drivers/infiniband/hw/qib/qib_init.c +++ b/drivers/infiniband/hw/qib/qib_init.c @@ -1350,7 +1350,7 @@ static void cleanup_device_data(struct qib_devdata *dd) if (dd->pageshadow) { struct page **tmpp = dd->pageshadow; dma_addr_t *tmpd = dd->physshadow; - int i, cnt = 0; + int i; for (ctxt = 0; ctxt < dd->cfgctxts; ctxt++) { int ctxt_tidbase = ctxt * dd->rcvtidcnt; @@ -1363,13 +1363,13 @@ static void cleanup_device_data(struct qib_devdata *dd) PAGE_SIZE, PCI_DMA_FROMDEVICE); qib_release_user_pages(&tmpp[i], 1); tmpp[i] = NULL; - cnt++; } } - tmpp = dd->pageshadow; dd->pageshadow = NULL; vfree(tmpp); + dd->physshadow = NULL; + vfree(tmpd); } /* -- cgit v1.2.3-59-g8ed1b From 0b3ddf380ca7aa6a009cc3e1944933fff8113b6a Mon Sep 17 00:00:00 2001 From: Dean Luick Date: Thu, 11 Jul 2013 15:32:14 -0400 Subject: IB/qib: Log all SDMA errors unconditionally This patch adds code to log SDMA errors for supportability purposes. Signed-off-by: Dean Luick Signed-off-by: Mike Marciniszyn Signed-off-by: Roland Dreier --- drivers/infiniband/hw/qib/qib.h | 2 +- drivers/infiniband/hw/qib/qib_iba7322.c | 114 ++++++++++++++++++++++++++++++++ drivers/infiniband/hw/qib/qib_sdma.c | 56 ++++++++++++++++ 3 files changed, 171 insertions(+), 1 deletion(-) (limited to 'drivers') diff --git a/drivers/infiniband/hw/qib/qib.h b/drivers/infiniband/hw/qib/qib.h index 5453e2b36567..4a9af795b88f 100644 --- a/drivers/infiniband/hw/qib/qib.h +++ b/drivers/infiniband/hw/qib/qib.h @@ -1348,7 +1348,7 @@ static inline int __qib_sdma_running(struct qib_pportdata *ppd) return ppd->sdma_state.current_state == qib_sdma_state_s99_running; } int qib_sdma_running(struct qib_pportdata *); - +void dump_sdma_state(struct qib_pportdata *ppd); void __qib_sdma_process_event(struct qib_pportdata *, enum qib_sdma_events); void qib_sdma_process_event(struct qib_pportdata *, enum qib_sdma_events); diff --git a/drivers/infiniband/hw/qib/qib_iba7322.c b/drivers/infiniband/hw/qib/qib_iba7322.c index f7c4b44b1f93..21e8b09d4bf8 100644 --- a/drivers/infiniband/hw/qib/qib_iba7322.c +++ b/drivers/infiniband/hw/qib/qib_iba7322.c @@ -83,6 +83,7 @@ static void ibsd_wr_allchans(struct qib_pportdata *, int, unsigned, unsigned); static void serdes_7322_los_enable(struct qib_pportdata *, int); static int serdes_7322_init_old(struct qib_pportdata *); static int serdes_7322_init_new(struct qib_pportdata *); +static void dump_sdma_7322_state(struct qib_pportdata *); #define BMASK(msb, lsb) (((1 << ((msb) + 1 - (lsb))) - 1) << (lsb)) @@ -652,6 +653,7 @@ struct qib_chippport_specific { u8 ibmalfusesnap; struct qib_qsfp_data qsfp_data; char epmsgbuf[192]; /* for port error interrupt msg buffer */ + char sdmamsgbuf[192]; /* for per-port sdma error messages */ }; static struct { @@ -1601,6 +1603,15 @@ static void sdma_7322_p_errors(struct qib_pportdata *ppd, u64 errs) spin_lock_irqsave(&ppd->sdma_lock, flags); + if (errs != QIB_E_P_SDMAHALT) { + /* SDMA errors have QIB_E_P_SDMAHALT and another bit set */ + qib_dev_porterr(dd, ppd->port, + "SDMA %s 0x%016llx %s\n", + qib_sdma_state_names[ppd->sdma_state.current_state], + errs, ppd->cpspec->sdmamsgbuf); + dump_sdma_7322_state(ppd); + } + switch (ppd->sdma_state.current_state) { case qib_sdma_state_s00_hw_down: break; @@ -2156,6 +2167,29 @@ static void qib_7322_handle_hwerrors(struct qib_devdata *dd, char *msg, qib_dev_err(dd, "%s hardware error\n", msg); + if (hwerrs & + (SYM_MASK(HwErrMask, SDmaMemReadErrMask_0) | + SYM_MASK(HwErrMask, SDmaMemReadErrMask_1))) { + int pidx = 0; + int err; + unsigned long flags; + struct qib_pportdata *ppd = dd->pport; + for (; pidx < dd->num_pports; ++pidx, ppd++) { + err = 0; + if (pidx == 0 && (hwerrs & + SYM_MASK(HwErrMask, SDmaMemReadErrMask_0))) + err++; + if (pidx == 1 && (hwerrs & + SYM_MASK(HwErrMask, SDmaMemReadErrMask_1))) + err++; + if (err) { + spin_lock_irqsave(&ppd->sdma_lock, flags); + dump_sdma_7322_state(ppd); + spin_unlock_irqrestore(&ppd->sdma_lock, flags); + } + } + } + if (isfatal && !dd->diag_client) { qib_dev_err(dd, "Fatal Hardware Error, no longer usable, SN %.16s\n", @@ -6753,6 +6787,86 @@ static void qib_sdma_set_7322_desc_cnt(struct qib_pportdata *ppd, unsigned cnt) qib_write_kreg_port(ppd, krp_senddmadesccnt, cnt); } +/* + * sdma_lock should be acquired before calling this routine + */ +static void dump_sdma_7322_state(struct qib_pportdata *ppd) +{ + u64 reg, reg1, reg2; + + reg = qib_read_kreg_port(ppd, krp_senddmastatus); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmastatus: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_sendctrl); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA sendctrl: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmabase); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmabase: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmabufmask0); + reg1 = qib_read_kreg_port(ppd, krp_senddmabufmask1); + reg2 = qib_read_kreg_port(ppd, krp_senddmabufmask2); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmabufmask 0:%llx 1:%llx 2:%llx\n", + reg, reg1, reg2); + + /* get bufuse bits, clear them, and print them again if non-zero */ + reg = qib_read_kreg_port(ppd, krp_senddmabuf_use0); + qib_write_kreg_port(ppd, krp_senddmabuf_use0, reg); + reg1 = qib_read_kreg_port(ppd, krp_senddmabuf_use1); + qib_write_kreg_port(ppd, krp_senddmabuf_use0, reg1); + reg2 = qib_read_kreg_port(ppd, krp_senddmabuf_use2); + qib_write_kreg_port(ppd, krp_senddmabuf_use0, reg2); + /* 0 and 1 should always be zero, so print as short form */ + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA current senddmabuf_use 0:%llx 1:%llx 2:%llx\n", + reg, reg1, reg2); + reg = qib_read_kreg_port(ppd, krp_senddmabuf_use0); + reg1 = qib_read_kreg_port(ppd, krp_senddmabuf_use1); + reg2 = qib_read_kreg_port(ppd, krp_senddmabuf_use2); + /* 0 and 1 should always be zero, so print as short form */ + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA cleared senddmabuf_use 0:%llx 1:%llx 2:%llx\n", + reg, reg1, reg2); + + reg = qib_read_kreg_port(ppd, krp_senddmatail); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmatail: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmahead); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmahead: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmaheadaddr); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmaheadaddr: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmalengen); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmalengen: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmadesccnt); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmadesccnt: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmaidlecnt); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmaidlecnt: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmaprioritythld); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmapriorityhld: 0x%016llx\n", reg); + + reg = qib_read_kreg_port(ppd, krp_senddmareloadcnt); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA senddmareloadcnt: 0x%016llx\n", reg); + + dump_sdma_state(ppd); +} + static struct sdma_set_state_action sdma_7322_action_table[] = { [qib_sdma_state_s00_hw_down] = { .go_s99_running_tofalse = 1, diff --git a/drivers/infiniband/hw/qib/qib_sdma.c b/drivers/infiniband/hw/qib/qib_sdma.c index 3fc514431212..32162d355370 100644 --- a/drivers/infiniband/hw/qib/qib_sdma.c +++ b/drivers/infiniband/hw/qib/qib_sdma.c @@ -708,6 +708,62 @@ unlock: return ret; } +/* + * sdma_lock should be acquired before calling this routine + */ +void dump_sdma_state(struct qib_pportdata *ppd) +{ + struct qib_sdma_desc *descq; + struct qib_sdma_txreq *txp, *txpnext; + __le64 *descqp; + u64 desc[2]; + dma_addr_t addr; + u16 gen, dwlen, dwoffset; + u16 head, tail, cnt; + + head = ppd->sdma_descq_head; + tail = ppd->sdma_descq_tail; + cnt = qib_sdma_descq_freecnt(ppd); + descq = ppd->sdma_descq; + + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA ppd->sdma_descq_head: %u\n", head); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA ppd->sdma_descq_tail: %u\n", tail); + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA sdma_descq_freecnt: %u\n", cnt); + + /* print info for each entry in the descriptor queue */ + while (head != tail) { + char flags[6] = { 'x', 'x', 'x', 'x', 'x', 0 }; + + descqp = &descq[head].qw[0]; + desc[0] = le64_to_cpu(descqp[0]); + desc[1] = le64_to_cpu(descqp[1]); + flags[0] = (desc[0] & 1<<15) ? 'I' : '-'; + flags[1] = (desc[0] & 1<<14) ? 'L' : 'S'; + flags[2] = (desc[0] & 1<<13) ? 'H' : '-'; + flags[3] = (desc[0] & 1<<12) ? 'F' : '-'; + flags[4] = (desc[0] & 1<<11) ? 'L' : '-'; + addr = (desc[1] << 32) | ((desc[0] >> 32) & 0xfffffffcULL); + gen = (desc[0] >> 30) & 3ULL; + dwlen = (desc[0] >> 14) & (0x7ffULL << 2); + dwoffset = (desc[0] & 0x7ffULL) << 2; + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA sdmadesc[%u]: flags:%s addr:0x%016llx gen:%u len:%u bytes offset:%u bytes\n", + head, flags, addr, gen, dwlen, dwoffset); + if (++head == ppd->sdma_descq_cnt) + head = 0; + } + + /* print dma descriptor indices from the TX requests */ + list_for_each_entry_safe(txp, txpnext, &ppd->sdma_activelist, + list) + qib_dev_porterr(ppd->dd, ppd->port, + "SDMA txp->start_idx: %u txp->next_descq_idx: %u\n", + txp->start_idx, txp->next_descq_idx); +} + void qib_sdma_process_event(struct qib_pportdata *ppd, enum qib_sdma_events event) { -- cgit v1.2.3-59-g8ed1b From 5e631a03af7eaa55b9ef7fa7611144c2c698c6c6 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Wed, 10 Jul 2013 13:58:59 +0300 Subject: mlx5: Return -EFAULT instead of -EPERM For copy_to/from_user() failure, the correct error code is -EFAULT not -EPERM. Signed-off-by: Dan Carpenter Acked-by: Or Gerlitz Signed-off-by: Roland Dreier --- drivers/infiniband/hw/mlx5/mr.c | 8 ++++---- drivers/net/ethernet/mellanox/mlx5/core/cmd.c | 10 +++++----- drivers/net/ethernet/mellanox/mlx5/core/debugfs.c | 12 ++++-------- 3 files changed, 13 insertions(+), 17 deletions(-) (limited to 'drivers') diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index e2daa8f02476..bd41df95b6f0 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -171,7 +171,7 @@ static ssize_t size_write(struct file *filp, const char __user *buf, int c; if (copy_from_user(lbuf, buf, sizeof(lbuf))) - return -EPERM; + return -EFAULT; c = order2idx(dev, ent->order); lbuf[sizeof(lbuf) - 1] = 0; @@ -208,7 +208,7 @@ static ssize_t size_read(struct file *filp, char __user *buf, size_t count, return err; if (copy_to_user(buf, lbuf, err)) - return -EPERM; + return -EFAULT; *pos += err; @@ -233,7 +233,7 @@ static ssize_t limit_write(struct file *filp, const char __user *buf, int c; if (copy_from_user(lbuf, buf, sizeof(lbuf))) - return -EPERM; + return -EFAULT; c = order2idx(dev, ent->order); lbuf[sizeof(lbuf) - 1] = 0; @@ -270,7 +270,7 @@ static ssize_t limit_read(struct file *filp, char __user *buf, size_t count, return err; if (copy_to_user(buf, lbuf, err)) - return -EPERM; + return -EFAULT; *pos += err; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c index c1c0eef89694..205753a04cfc 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c @@ -693,7 +693,7 @@ static ssize_t dbg_write(struct file *filp, const char __user *buf, return -ENOMEM; if (copy_from_user(lbuf, buf, sizeof(lbuf))) - return -EPERM; + return -EFAULT; lbuf[sizeof(lbuf) - 1] = 0; @@ -889,7 +889,7 @@ static ssize_t data_write(struct file *filp, const char __user *buf, return -ENOMEM; if (copy_from_user(ptr, buf, count)) { - err = -EPERM; + err = -EFAULT; goto out; } dbg->in_msg = ptr; @@ -919,7 +919,7 @@ static ssize_t data_read(struct file *filp, char __user *buf, size_t count, copy = min_t(int, count, dbg->outlen); if (copy_to_user(buf, dbg->out_msg, copy)) - return -EPERM; + return -EFAULT; *pos += copy; @@ -949,7 +949,7 @@ static ssize_t outlen_read(struct file *filp, char __user *buf, size_t count, return err; if (copy_to_user(buf, &outlen, err)) - return -EPERM; + return -EFAULT; *pos += err; @@ -974,7 +974,7 @@ static ssize_t outlen_write(struct file *filp, const char __user *buf, dbg->outlen = 0; if (copy_from_user(outlen_str, buf, count)) - return -EPERM; + return -EFAULT; outlen_str[7] = 0; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c index a550a8ef94a1..4273c06e2e96 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c @@ -148,7 +148,6 @@ static ssize_t average_read(struct file *filp, char __user *buf, size_t count, struct mlx5_cmd_stats *stats; u64 field = 0; int ret; - int err; char tbuf[22]; if (*pos) @@ -161,9 +160,8 @@ static ssize_t average_read(struct file *filp, char __user *buf, size_t count, spin_unlock(&stats->lock); ret = snprintf(tbuf, sizeof(tbuf), "%llu\n", field); if (ret > 0) { - err = copy_to_user(buf, tbuf, ret); - if (err) - return err; + if (copy_to_user(buf, tbuf, ret)) + return -EFAULT; } *pos += ret; @@ -418,7 +416,6 @@ static ssize_t dbg_read(struct file *filp, char __user *buf, size_t count, char tbuf[18]; u64 field; int ret; - int err; if (*pos) return 0; @@ -445,9 +442,8 @@ static ssize_t dbg_read(struct file *filp, char __user *buf, size_t count, ret = snprintf(tbuf, sizeof(tbuf), "0x%llx\n", field); if (ret > 0) { - err = copy_to_user(buf, tbuf, ret); - if (err) - return err; + if (copy_to_user(buf, tbuf, ret)) + return -EFAULT; } *pos += ret; -- cgit v1.2.3-59-g8ed1b