diff options
Diffstat (limited to 'drivers/infiniband/ulp')
37 files changed, 2006 insertions, 1770 deletions
diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index 44d8d151ff90..35e9c8a330e2 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -411,6 +411,7 @@ struct ipoib_dev_priv { struct dentry *path_dentry; #endif u64 hca_caps; + u64 kernel_caps; struct ipoib_ethtool_st ethtool; unsigned int max_send_sge; const struct net_device_ops *rn_ops; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c b/drivers/infiniband/ulp/ipoib/ipoib_cm.c index fd9d7f2c4d64..b610d36295bb 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c @@ -465,7 +465,7 @@ static int ipoib_cm_req_handler(struct ib_cm_id *cm_id, goto err_qp; } - psn = prandom_u32() & 0xffffff; + psn = get_random_u32() & 0xffffff; ret = ipoib_cm_modify_rx_qp(dev, cm_id, p->qp, psn); if (ret) goto err_modify; @@ -884,8 +884,8 @@ int ipoib_cm_dev_open(struct net_device *dev) goto err_cm; } - ret = ib_cm_listen(priv->cm.id, cpu_to_be64(IPOIB_CM_IETF_ID | priv->qp->qp_num), - 0); + ret = ib_cm_listen(priv->cm.id, + cpu_to_be64(IPOIB_CM_IETF_ID | priv->qp->qp_num)); if (ret) { pr_warn("%s: failed to listen on ID 0x%llx\n", priv->ca->name, IPOIB_CM_IETF_ID | priv->qp->qp_num); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c b/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c index a09ca21f7dff..8af99b18d361 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c @@ -65,10 +65,10 @@ static void ipoib_get_drvinfo(struct net_device *netdev, ib_get_device_fw_str(priv->ca, drvinfo->fw_version); - strlcpy(drvinfo->bus_info, dev_name(priv->ca->dev.parent), + strscpy(drvinfo->bus_info, dev_name(priv->ca->dev.parent), sizeof(drvinfo->bus_info)); - strlcpy(drvinfo->driver, "ib_ipoib", sizeof(drvinfo->driver)); + strscpy(drvinfo->driver, "ib_ipoib", sizeof(drvinfo->driver)); } static int ipoib_get_coalesce(struct net_device *dev, diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c index 2c3dca41d3bd..ed25061fac62 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c @@ -573,7 +573,7 @@ int ipoib_send(struct net_device *dev, struct sk_buff *skb, unsigned int usable_sge = priv->max_send_sge - !!skb_headlen(skb); if (skb_is_gso(skb)) { - hlen = skb_transport_offset(skb) + tcp_hdrlen(skb); + hlen = skb_tcp_all_headers(skb); phead = skb->data; if (unlikely(!skb_pull(skb, hlen))) { ipoib_warn(priv, "linear data too small\n"); @@ -1109,7 +1109,7 @@ static bool ipoib_dev_addr_changed_valid(struct ipoib_dev_priv *priv) * if he sets the device address back to be based on GID index 0, * he no longer wishs to control it. * - * If the user doesn't control the the device address, + * If the user doesn't control the device address, * IPOIB_FLAG_DEV_ADDR_SET is set and ib_find_gid failed it means * the port GUID has changed and GID at index 0 has changed * so we need to change priv->local_gid and priv->dev->dev_addr diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c index 9934b8bd7f56..ac25fc80fb33 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c @@ -742,7 +742,7 @@ void ipoib_flush_paths(struct net_device *dev) static void path_rec_completion(int status, struct sa_path_rec *pathrec, - void *path_ptr) + int num_prs, void *path_ptr) { struct ipoib_path *path = path_ptr; struct net_device *dev = path->dev; @@ -1664,8 +1664,10 @@ static void ipoib_napi_add(struct net_device *dev) { struct ipoib_dev_priv *priv = ipoib_priv(dev); - netif_napi_add(dev, &priv->recv_napi, ipoib_rx_poll, IPOIB_NUM_WC); - netif_napi_add(dev, &priv->send_napi, ipoib_tx_poll, MAX_SEND_CQE); + netif_napi_add_weight(dev, &priv->recv_napi, ipoib_rx_poll, + IPOIB_NUM_WC); + netif_napi_add_weight(dev, &priv->send_napi, ipoib_tx_poll, + MAX_SEND_CQE); } static void ipoib_napi_del(struct net_device *dev) @@ -1850,11 +1852,12 @@ static void ipoib_parent_unregister_pre(struct net_device *ndev) static void ipoib_set_dev_features(struct ipoib_dev_priv *priv) { priv->hca_caps = priv->ca->attrs.device_cap_flags; + priv->kernel_caps = priv->ca->attrs.kernel_cap_flags; if (priv->hca_caps & IB_DEVICE_UD_IP_CSUM) { priv->dev->hw_features |= NETIF_F_IP_CSUM | NETIF_F_RXCSUM; - if (priv->hca_caps & IB_DEVICE_UD_TSO) + if (priv->kernel_caps & IBK_UD_TSO) priv->dev->hw_features |= NETIF_F_TSO; priv->dev->features |= priv->dev->hw_features; @@ -2201,7 +2204,7 @@ int ipoib_intf_init(struct ib_device *hca, u32 port, const char *name, priv->rn_ops = dev->netdev_ops; - if (hca->attrs.device_cap_flags & IB_DEVICE_VIRTUAL_FUNCTION) + if (hca->attrs.kernel_cap_flags & IBK_VIRTUAL_FUNCTION) dev->netdev_ops = &ipoib_netdev_ops_vf; else dev->netdev_ops = &ipoib_netdev_ops_pf; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_netlink.c b/drivers/infiniband/ulp/ipoib/ipoib_netlink.c index 5b05cf3837da..ea16ba5d8da6 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_netlink.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_netlink.c @@ -32,7 +32,6 @@ #include <linux/netdevice.h> #include <linux/if_arp.h> /* For ARPHRD_xxx */ -#include <linux/module.h> #include <net/rtnetlink.h> #include "ipoib.h" diff --git a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c index 5a150a080ac2..368e5d77416d 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c @@ -197,16 +197,16 @@ int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca) init_attr.send_cq = priv->send_cq; init_attr.recv_cq = priv->recv_cq; - if (priv->hca_caps & IB_DEVICE_UD_TSO) + if (priv->kernel_caps & IBK_UD_TSO) init_attr.create_flags |= IB_QP_CREATE_IPOIB_UD_LSO; - if (priv->hca_caps & IB_DEVICE_BLOCK_MULTICAST_LOOPBACK) + if (priv->kernel_caps & IBK_BLOCK_MULTICAST_LOOPBACK) init_attr.create_flags |= IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK; if (priv->hca_caps & IB_DEVICE_MANAGED_FLOW_STEERING) init_attr.create_flags |= IB_QP_CREATE_NETIF_QP; - if (priv->hca_caps & IB_DEVICE_RDMA_NETDEV_OPA) + if (priv->kernel_caps & IBK_RDMA_NETDEV_OPA) init_attr.create_flags |= IB_QP_CREATE_NETDEV_USE; priv->qp = ib_create_qp(priv->pd, &init_attr); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_vlan.c b/drivers/infiniband/ulp/ipoib/ipoib_vlan.c index 0322dc75396f..4bd161e86f8d 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_vlan.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_vlan.c @@ -30,7 +30,6 @@ * SOFTWARE. */ -#include <linux/module.h> #include <linux/sched/signal.h> #include <linux/init.h> diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c index 776e46ee95da..620ae5b2d80d 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.c +++ b/drivers/infiniband/ulp/iser/iscsi_iser.c @@ -113,10 +113,6 @@ bool iser_pi_enable = false; module_param_named(pi_enable, iser_pi_enable, bool, S_IRUGO); MODULE_PARM_DESC(pi_enable, "Enable T10-PI offload support (default:disabled)"); -int iser_pi_guard; -module_param_named(pi_guard, iser_pi_guard, int, S_IRUGO); -MODULE_PARM_DESC(pi_guard, "T10-PI guard_type [deprecated]"); - static int iscsi_iser_set(const char *val, const struct kernel_param *kp) { int ret; @@ -139,9 +135,8 @@ static int iscsi_iser_set(const char *val, const struct kernel_param *kp) * Notes: In case of data length errors or iscsi PDU completion failures * this routine will signal iscsi layer of connection failure. */ -void -iscsi_iser_recv(struct iscsi_conn *conn, struct iscsi_hdr *hdr, - char *rx_data, int rx_data_len) +void iscsi_iser_recv(struct iscsi_conn *conn, struct iscsi_hdr *hdr, + char *rx_data, int rx_data_len) { int rc = 0; int datalen; @@ -176,8 +171,7 @@ error: * Netes: This routine can't fail, just assign iscsi task * hdr and max hdr size. */ -static int -iscsi_iser_pdu_alloc(struct iscsi_task *task, uint8_t opcode) +static int iscsi_iser_pdu_alloc(struct iscsi_task *task, uint8_t opcode) { struct iscsi_iser_task *iser_task = task->dd_data; @@ -198,9 +192,8 @@ iscsi_iser_pdu_alloc(struct iscsi_task *task, uint8_t opcode) * state mutex to avoid dereferencing the IB device which * may have already been terminated. */ -int -iser_initialize_task_headers(struct iscsi_task *task, - struct iser_tx_desc *tx_desc) +int iser_initialize_task_headers(struct iscsi_task *task, + struct iser_tx_desc *tx_desc) { struct iser_conn *iser_conn = task->conn->dd_data; struct iser_device *device = iser_conn->ib_conn.device; @@ -237,8 +230,7 @@ iser_initialize_task_headers(struct iscsi_task *task, * Return: Returns zero on success or -ENOMEM when failing * to init task headers (dma mapping error). */ -static int -iscsi_iser_task_init(struct iscsi_task *task) +static int iscsi_iser_task_init(struct iscsi_task *task) { struct iscsi_iser_task *iser_task = task->dd_data; int ret; @@ -272,8 +264,8 @@ iscsi_iser_task_init(struct iscsi_task *task) * xmit. * **/ -static int -iscsi_iser_mtask_xmit(struct iscsi_conn *conn, struct iscsi_task *task) +static int iscsi_iser_mtask_xmit(struct iscsi_conn *conn, + struct iscsi_task *task) { int error = 0; @@ -290,9 +282,8 @@ iscsi_iser_mtask_xmit(struct iscsi_conn *conn, struct iscsi_task *task) return error; } -static int -iscsi_iser_task_xmit_unsol_data(struct iscsi_conn *conn, - struct iscsi_task *task) +static int iscsi_iser_task_xmit_unsol_data(struct iscsi_conn *conn, + struct iscsi_task *task) { struct iscsi_r2t_info *r2t = &task->unsol_r2t; struct iscsi_data hdr; @@ -326,8 +317,7 @@ iscsi_iser_task_xmit_unsol_data_exit: * * Return: zero on success or escalates $error on failure. */ -static int -iscsi_iser_task_xmit(struct iscsi_task *task) +static int iscsi_iser_task_xmit(struct iscsi_task *task) { struct iscsi_conn *conn = task->conn; struct iscsi_iser_task *iser_task = task->dd_data; @@ -410,8 +400,7 @@ static void iscsi_iser_cleanup_task(struct iscsi_task *task) * * In addition the error sector is marked. */ -static u8 -iscsi_iser_check_protection(struct iscsi_task *task, sector_t *sector) +static u8 iscsi_iser_check_protection(struct iscsi_task *task, sector_t *sector) { struct iscsi_iser_task *iser_task = task->dd_data; enum iser_data_dir dir = iser_task->dir[ISER_DIR_IN] ? @@ -460,11 +449,9 @@ iscsi_iser_conn_create(struct iscsi_cls_session *cls_session, * -EINVAL in case end-point doesn't exsits anymore or iser connection * state is not UP (teardown already started). */ -static int -iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session, - struct iscsi_cls_conn *cls_conn, - uint64_t transport_eph, - int is_leading) +static int iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session, + struct iscsi_cls_conn *cls_conn, + uint64_t transport_eph, int is_leading) { struct iscsi_conn *conn = cls_conn->dd_data; struct iser_conn *iser_conn; @@ -519,8 +506,7 @@ out: * from this point iscsi must call conn_stop in session/connection * teardown so iser transport must wait for it. */ -static int -iscsi_iser_conn_start(struct iscsi_cls_conn *cls_conn) +static int iscsi_iser_conn_start(struct iscsi_cls_conn *cls_conn) { struct iscsi_conn *iscsi_conn; struct iser_conn *iser_conn; @@ -542,8 +528,7 @@ iscsi_iser_conn_start(struct iscsi_cls_conn *cls_conn) * handle, so we call it under iser the state lock to protect against * this kind of race. */ -static void -iscsi_iser_conn_stop(struct iscsi_cls_conn *cls_conn, int flag) +static void iscsi_iser_conn_stop(struct iscsi_cls_conn *cls_conn, int flag) { struct iscsi_conn *conn = cls_conn->dd_data; struct iser_conn *iser_conn = conn->dd_data; @@ -578,18 +563,16 @@ iscsi_iser_conn_stop(struct iscsi_cls_conn *cls_conn, int flag) * * Removes and free iscsi host. */ -static void -iscsi_iser_session_destroy(struct iscsi_cls_session *cls_session) +static void iscsi_iser_session_destroy(struct iscsi_cls_session *cls_session) { struct Scsi_Host *shost = iscsi_session_to_shost(cls_session); iscsi_session_teardown(cls_session); - iscsi_host_remove(shost); + iscsi_host_remove(shost, false); iscsi_host_free(shost); } -static inline unsigned int -iser_dif_prot_caps(int prot_caps) +static inline unsigned int iser_dif_prot_caps(int prot_caps) { int ret = 0; @@ -667,7 +650,7 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep, SHOST_DIX_GUARD_CRC); } - if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_SG_GAPS_REG)) + if (!(ib_dev->attrs.kernel_cap_flags & IBK_SG_GAPS_REG)) shost->virt_boundary_mask = SZ_4K - 1; if (iscsi_host_add(shost, ib_dev->dev.parent)) { @@ -702,15 +685,14 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep, return cls_session; remove_host: - iscsi_host_remove(shost); + iscsi_host_remove(shost, false); free_host: iscsi_host_free(shost); return NULL; } -static int -iscsi_iser_set_param(struct iscsi_cls_conn *cls_conn, - enum iscsi_param param, char *buf, int buflen) +static int iscsi_iser_set_param(struct iscsi_cls_conn *cls_conn, + enum iscsi_param param, char *buf, int buflen) { int value; @@ -760,8 +742,8 @@ iscsi_iser_set_param(struct iscsi_cls_conn *cls_conn, * * Output connection statistics. */ -static void -iscsi_iser_conn_get_stats(struct iscsi_cls_conn *cls_conn, struct iscsi_stats *stats) +static void iscsi_iser_conn_get_stats(struct iscsi_cls_conn *cls_conn, + struct iscsi_stats *stats) { struct iscsi_conn *conn = cls_conn->dd_data; @@ -812,9 +794,9 @@ static int iscsi_iser_get_ep_param(struct iscsi_endpoint *ep, * Return: iscsi_endpoint created by iscsi layer or ERR_PTR(error) * if fails. */ -static struct iscsi_endpoint * -iscsi_iser_ep_connect(struct Scsi_Host *shost, struct sockaddr *dst_addr, - int non_blocking) +static struct iscsi_endpoint *iscsi_iser_ep_connect(struct Scsi_Host *shost, + struct sockaddr *dst_addr, + int non_blocking) { int err; struct iser_conn *iser_conn; @@ -857,8 +839,7 @@ failure: * or more likely iser connection state transitioned to TEMINATING or * DOWN during the wait period. */ -static int -iscsi_iser_ep_poll(struct iscsi_endpoint *ep, int timeout_ms) +static int iscsi_iser_ep_poll(struct iscsi_endpoint *ep, int timeout_ms) { struct iser_conn *iser_conn = ep->dd_data; int rc; @@ -893,8 +874,7 @@ iscsi_iser_ep_poll(struct iscsi_endpoint *ep, int timeout_ms) * and cleanup or actually call it immediately in case we didn't pass * iscsi conn bind/start stage, thus it is safe. */ -static void -iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep) +static void iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep) { struct iser_conn *iser_conn = ep->dd_data; @@ -991,6 +971,7 @@ static struct scsi_host_template iscsi_iser_sht = { .proc_name = "iscsi_iser", .this_id = -1, .track_queue_depth = 1, + .cmd_size = sizeof(struct iscsi_cmd), }; static struct iscsi_transport iscsi_iser_transport = { diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h index 9f6ac0a09a78..dee8c97ff056 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.h +++ b/drivers/infiniband/ulp/iser/iscsi_iser.h @@ -119,8 +119,6 @@ #define ISER_QP_MAX_RECV_DTOS (ISER_DEF_XMIT_CMDS_MAX) -#define ISER_MIN_POSTED_RX (ISER_DEF_XMIT_CMDS_MAX >> 2) - /* the max TX (send) WR supported by the iSER QP is defined by * * max_send_wr = T * (1 + D) + C ; D is how many inflight dataouts we expect * * to have at max for SCSI command. The tx posting & completion handling code * @@ -148,8 +146,6 @@ - ISER_MAX_RX_MISC_PDUS) / \ (1 + ISER_INFLIGHT_DATAOUTS)) -#define ISER_SIGNAL_CMD_COUNT 32 - /* Constant PDU lengths calculations */ #define ISER_HEADERS_LEN (sizeof(struct iser_ctrl) + sizeof(struct iscsi_hdr)) @@ -207,12 +203,12 @@ struct iser_reg_resources; * * @sge: memory region sg element * @rkey: memory region remote key - * @mem_h: pointer to registration context (FMR/Fastreg) + * @desc: pointer to fast registration context */ struct iser_mem_reg { - struct ib_sge sge; - u32 rkey; - void *mem_h; + struct ib_sge sge; + u32 rkey; + struct iser_fr_desc *desc; }; enum iser_desc_type { @@ -366,11 +362,8 @@ struct iser_fr_pool { * @qp: Connection Queue-pair * @cq: Connection completion queue * @cq_size: The number of max outstanding completions - * @post_recv_buf_count: post receive counter - * @sig_count: send work request signal count - * @rx_wr: receive work request for batch posts * @device: reference to iser device - * @fr_pool: connection fast registration poool + * @fr_pool: connection fast registration pool * @pi_support: Indicate device T10-PI support * @reg_cqe: completion handler */ @@ -379,9 +372,6 @@ struct ib_conn { struct ib_qp *qp; struct ib_cq *cq; u32 cq_size; - int post_recv_buf_count; - u8 sig_count; - struct ib_recv_wr rx_wr[ISER_MIN_POSTED_RX]; struct iser_device *device; struct iser_fr_pool fr_pool; bool pi_support; @@ -397,8 +387,6 @@ struct ib_conn { * @state: connection logical state * @qp_max_recv_dtos: maximum number of data outs, corresponds * to max number of post recvs - * @qp_max_recv_dtos_mask: (qp_max_recv_dtos - 1) - * @min_posted_rx: (qp_max_recv_dtos >> 2) * @max_cmds: maximum cmds allowed for this connection * @name: connection peer portal * @release_work: deffered work for release job @@ -409,7 +397,6 @@ struct ib_conn { * (state is ISER_CONN_UP) * @conn_list: entry in ig conn list * @login_desc: login descriptor - * @rx_desc_head: head of rx_descs cyclic buffer * @rx_descs: rx buffers array (cyclic buffer) * @num_rx_descs: number of rx descriptors * @scsi_sg_tablesize: scsi host sg_tablesize @@ -422,8 +409,6 @@ struct iser_conn { struct iscsi_endpoint *ep; enum iser_conn_state state; unsigned qp_max_recv_dtos; - unsigned qp_max_recv_dtos_mask; - unsigned min_posted_rx; u16 max_cmds; char name[ISER_OBJECT_NAME_SIZE]; struct work_struct release_work; @@ -433,7 +418,6 @@ struct iser_conn { struct completion up_completion; struct list_head conn_list; struct iser_login_desc login_desc; - unsigned int rx_desc_head; struct iser_rx_desc *rx_descs; u32 num_rx_descs; unsigned short scsi_sg_tablesize; @@ -486,7 +470,6 @@ struct iser_global { extern struct iser_global ig; extern int iser_debug_level; extern bool iser_pi_enable; -extern int iser_pi_guard; extern unsigned int iser_max_sectors; extern bool iser_always_reg; @@ -543,18 +526,17 @@ int iser_connect(struct iser_conn *iser_conn, int non_blocking); int iser_post_recvl(struct iser_conn *iser_conn); -int iser_post_recvm(struct iser_conn *iser_conn, int count); -int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc, - bool signal); +int iser_post_recvm(struct iser_conn *iser_conn, + struct iser_rx_desc *rx_desc); +int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc); int iser_dma_map_task_data(struct iscsi_iser_task *iser_task, - struct iser_data_buf *data, enum iser_data_dir iser_dir, enum dma_data_direction dma_dir); void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task, - struct iser_data_buf *data, - enum dma_data_direction dir); + enum iser_data_dir iser_dir, + enum dma_data_direction dma_dir); int iser_initialize_task_headers(struct iscsi_task *task, struct iser_tx_desc *tx_desc); diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c index 27a6f75a9912..7b83f48f60c5 100644 --- a/drivers/infiniband/ulp/iser/iser_initiator.c +++ b/drivers/infiniband/ulp/iser/iser_initiator.c @@ -52,30 +52,17 @@ static int iser_prepare_read_cmd(struct iscsi_task *task) struct iser_mem_reg *mem_reg; int err; struct iser_ctrl *hdr = &iser_task->desc.iser_header; - struct iser_data_buf *buf_in = &iser_task->data[ISER_DIR_IN]; err = iser_dma_map_task_data(iser_task, - buf_in, ISER_DIR_IN, DMA_FROM_DEVICE); if (err) return err; - if (scsi_prot_sg_count(iser_task->sc)) { - struct iser_data_buf *pbuf_in = &iser_task->prot[ISER_DIR_IN]; - - err = iser_dma_map_task_data(iser_task, - pbuf_in, - ISER_DIR_IN, - DMA_FROM_DEVICE); - if (err) - return err; - } - err = iser_reg_mem_fastreg(iser_task, ISER_DIR_IN, false); if (err) { iser_err("Failed to set up Data-IN RDMA\n"); - return err; + goto out_err; } mem_reg = &iser_task->rdma_reg[ISER_DIR_IN]; @@ -88,6 +75,10 @@ static int iser_prepare_read_cmd(struct iscsi_task *task) (unsigned long long)mem_reg->sge.addr); return 0; + +out_err: + iser_dma_unmap_task_data(iser_task, ISER_DIR_IN, DMA_FROM_DEVICE); + return err; } /* Register user buffer memory and initialize passive rdma @@ -95,11 +86,8 @@ static int iser_prepare_read_cmd(struct iscsi_task *task) * task->data[ISER_DIR_OUT].data_len, Protection size * is stored at task->prot[ISER_DIR_OUT].data_len */ -static int -iser_prepare_write_cmd(struct iscsi_task *task, - unsigned int imm_sz, - unsigned int unsol_sz, - unsigned int edtl) +static int iser_prepare_write_cmd(struct iscsi_task *task, unsigned int imm_sz, + unsigned int unsol_sz, unsigned int edtl) { struct iscsi_iser_task *iser_task = task->dd_data; struct iser_mem_reg *mem_reg; @@ -109,28 +97,16 @@ iser_prepare_write_cmd(struct iscsi_task *task, struct ib_sge *tx_dsg = &iser_task->desc.tx_sg[1]; err = iser_dma_map_task_data(iser_task, - buf_out, ISER_DIR_OUT, DMA_TO_DEVICE); if (err) return err; - if (scsi_prot_sg_count(iser_task->sc)) { - struct iser_data_buf *pbuf_out = &iser_task->prot[ISER_DIR_OUT]; - - err = iser_dma_map_task_data(iser_task, - pbuf_out, - ISER_DIR_OUT, - DMA_TO_DEVICE); - if (err) - return err; - } - err = iser_reg_mem_fastreg(iser_task, ISER_DIR_OUT, buf_out->data_len == imm_sz); - if (err != 0) { + if (err) { iser_err("Failed to register write cmd RDMA mem\n"); - return err; + goto out_err; } mem_reg = &iser_task->rdma_reg[ISER_DIR_OUT]; @@ -157,11 +133,15 @@ iser_prepare_write_cmd(struct iscsi_task *task, } return 0; + +out_err: + iser_dma_unmap_task_data(iser_task, ISER_DIR_OUT, DMA_TO_DEVICE); + return err; } /* creates a new tx descriptor and adds header regd buffer */ -static void iser_create_send_desc(struct iser_conn *iser_conn, - struct iser_tx_desc *tx_desc) +static void iser_create_send_desc(struct iser_conn *iser_conn, + struct iser_tx_desc *tx_desc) { struct iser_device *device = iser_conn->ib_conn.device; @@ -247,8 +227,6 @@ int iser_alloc_rx_descriptors(struct iser_conn *iser_conn, struct iser_device *device = ib_conn->device; iser_conn->qp_max_recv_dtos = session->cmds_max; - iser_conn->qp_max_recv_dtos_mask = session->cmds_max - 1; /* cmds_max is 2^N */ - iser_conn->min_posted_rx = iser_conn->qp_max_recv_dtos >> 2; if (iser_alloc_fastreg_pool(ib_conn, session->scsi_cmds_max, iser_conn->pages_per_mr)) @@ -280,7 +258,6 @@ int iser_alloc_rx_descriptors(struct iser_conn *iser_conn, rx_sg->lkey = device->pd->local_dma_lkey; } - iser_conn->rx_desc_head = 0; return 0; rx_desc_dma_map_failed: @@ -322,37 +299,35 @@ void iser_free_rx_descriptors(struct iser_conn *iser_conn) static int iser_post_rx_bufs(struct iscsi_conn *conn, struct iscsi_hdr *req) { struct iser_conn *iser_conn = conn->dd_data; - struct ib_conn *ib_conn = &iser_conn->ib_conn; struct iscsi_session *session = conn->session; + int err = 0; + int i; iser_dbg("req op %x flags %x\n", req->opcode, req->flags); /* check if this is the last login - going to full feature phase */ if ((req->flags & ISCSI_FULL_FEATURE_PHASE) != ISCSI_FULL_FEATURE_PHASE) - return 0; - - /* - * Check that there is one posted recv buffer - * (for the last login response). - */ - WARN_ON(ib_conn->post_recv_buf_count != 1); + goto out; if (session->discovery_sess) { iser_info("Discovery session, re-using login RX buffer\n"); - return 0; - } else - iser_info("Normal session, posting batch of RX %d buffers\n", - iser_conn->min_posted_rx); - - /* Initial post receive buffers */ - if (iser_post_recvm(iser_conn, iser_conn->min_posted_rx)) - return -ENOMEM; + goto out; + } - return 0; -} + iser_info("Normal session, posting batch of RX %d buffers\n", + iser_conn->qp_max_recv_dtos - 1); -static inline bool iser_signal_comp(u8 sig_count) -{ - return ((sig_count % ISER_SIGNAL_CMD_COUNT) == 0); + /* + * Initial post receive buffers. + * There is one already posted recv buffer (for the last login + * response). Therefore, the first recv buffer is skipped here. + */ + for (i = 1; i < iser_conn->qp_max_recv_dtos; i++) { + err = iser_post_recvm(iser_conn, &iser_conn->rx_descs[i]); + if (err) + goto out; + } +out: + return err; } /** @@ -360,8 +335,7 @@ static inline bool iser_signal_comp(u8 sig_count) * @conn: link to matching iscsi connection * @task: SCSI command task */ -int iser_send_command(struct iscsi_conn *conn, - struct iscsi_task *task) +int iser_send_command(struct iscsi_conn *conn, struct iscsi_task *task) { struct iser_conn *iser_conn = conn->dd_data; struct iscsi_iser_task *iser_task = task->dd_data; @@ -371,7 +345,6 @@ int iser_send_command(struct iscsi_conn *conn, struct iscsi_scsi_req *hdr = (struct iscsi_scsi_req *)task->hdr; struct scsi_cmnd *sc = task->sc; struct iser_tx_desc *tx_desc = &iser_task->desc; - u8 sig_count = ++iser_conn->ib_conn.sig_count; edtl = ntohl(hdr->data_length); @@ -418,8 +391,7 @@ int iser_send_command(struct iscsi_conn *conn, iser_task->status = ISER_TASK_STATUS_STARTED; - err = iser_post_send(&iser_conn->ib_conn, tx_desc, - iser_signal_comp(sig_count)); + err = iser_post_send(&iser_conn->ib_conn, tx_desc); if (!err) return 0; @@ -434,8 +406,7 @@ send_command_error: * @task: SCSI command task * @hdr: pointer to the LLD's iSCSI message header */ -int iser_send_data_out(struct iscsi_conn *conn, - struct iscsi_task *task, +int iser_send_data_out(struct iscsi_conn *conn, struct iscsi_task *task, struct iscsi_data *hdr) { struct iser_conn *iser_conn = conn->dd_data; @@ -487,7 +458,7 @@ int iser_send_data_out(struct iscsi_conn *conn, itt, buf_offset, data_seg_len); - err = iser_post_send(&iser_conn->ib_conn, tx_desc, true); + err = iser_post_send(&iser_conn->ib_conn, tx_desc); if (!err) return 0; @@ -497,8 +468,7 @@ send_data_out_error: return err; } -int iser_send_control(struct iscsi_conn *conn, - struct iscsi_task *task) +int iser_send_control(struct iscsi_conn *conn, struct iscsi_task *task) { struct iser_conn *iser_conn = conn->dd_data; struct iscsi_iser_task *iser_task = task->dd_data; @@ -550,7 +520,7 @@ int iser_send_control(struct iscsi_conn *conn, goto send_control_error; } - err = iser_post_send(&iser_conn->ib_conn, mdesc, true); + err = iser_post_send(&iser_conn->ib_conn, mdesc); if (!err) return 0; @@ -567,6 +537,7 @@ void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc) struct iscsi_hdr *hdr; char *data; int length; + bool full_feature_phase; if (unlikely(wc->status != IB_WC_SUCCESS)) { iser_err_comp(wc, "login_rsp"); @@ -580,6 +551,9 @@ void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc) hdr = desc->rsp + sizeof(struct iser_ctrl); data = desc->rsp + ISER_HEADERS_LEN; length = wc->byte_len - ISER_HEADERS_LEN; + full_feature_phase = ((hdr->flags & ISCSI_FULL_FEATURE_PHASE) == + ISCSI_FULL_FEATURE_PHASE) && + (hdr->flags & ISCSI_FLAG_CMD_FINAL); iser_dbg("op 0x%x itt 0x%x dlen %d\n", hdr->opcode, hdr->itt, length); @@ -590,11 +564,15 @@ void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc) desc->rsp_dma, ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE); - ib_conn->post_recv_buf_count--; + if (!full_feature_phase || + iser_conn->iscsi_conn->session->discovery_sess) + return; + + /* Post the first RX buffer that is skipped in iser_post_rx_bufs() */ + iser_post_recvm(iser_conn, iser_conn->rx_descs); } -static inline int -iser_inv_desc(struct iser_fr_desc *desc, u32 rkey) +static inline int iser_inv_desc(struct iser_fr_desc *desc, u32 rkey) { if (unlikely((!desc->sig_protected && rkey != desc->rsc.mr->rkey) || (desc->sig_protected && rkey != desc->rsc.sig_mr->rkey))) { @@ -607,10 +585,8 @@ iser_inv_desc(struct iser_fr_desc *desc, u32 rkey) return 0; } -static int -iser_check_remote_inv(struct iser_conn *iser_conn, - struct ib_wc *wc, - struct iscsi_hdr *hdr) +static int iser_check_remote_inv(struct iser_conn *iser_conn, struct ib_wc *wc, + struct iscsi_hdr *hdr) { if (wc->wc_flags & IB_WC_WITH_INVALIDATE) { struct iscsi_task *task; @@ -631,13 +607,13 @@ iser_check_remote_inv(struct iser_conn *iser_conn, struct iser_fr_desc *desc; if (iser_task->dir[ISER_DIR_IN]) { - desc = iser_task->rdma_reg[ISER_DIR_IN].mem_h; + desc = iser_task->rdma_reg[ISER_DIR_IN].desc; if (unlikely(iser_inv_desc(desc, rkey))) return -EINVAL; } if (iser_task->dir[ISER_DIR_OUT]) { - desc = iser_task->rdma_reg[ISER_DIR_OUT].mem_h; + desc = iser_task->rdma_reg[ISER_DIR_OUT].desc; if (unlikely(iser_inv_desc(desc, rkey))) return -EINVAL; } @@ -657,8 +633,7 @@ void iser_task_rsp(struct ib_cq *cq, struct ib_wc *wc) struct iser_conn *iser_conn = to_iser_conn(ib_conn); struct iser_rx_desc *desc = iser_rx(wc->wr_cqe); struct iscsi_hdr *hdr; - int length; - int outstanding, count, err; + int length, err; if (unlikely(wc->status != IB_WC_SUCCESS)) { iser_err_comp(wc, "task_rsp"); @@ -687,20 +662,9 @@ void iser_task_rsp(struct ib_cq *cq, struct ib_wc *wc) desc->dma_addr, ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); - /* decrementing conn->post_recv_buf_count only --after-- freeing the * - * task eliminates the need to worry on tasks which are completed in * - * parallel to the execution of iser_conn_term. So the code that waits * - * for the posted rx bufs refcount to become zero handles everything */ - ib_conn->post_recv_buf_count--; - - outstanding = ib_conn->post_recv_buf_count; - if (outstanding + iser_conn->min_posted_rx <= iser_conn->qp_max_recv_dtos) { - count = min(iser_conn->qp_max_recv_dtos - outstanding, - iser_conn->min_posted_rx); - err = iser_post_recvm(iser_conn, count); - if (err) - iser_err("posting %d rx bufs err %d\n", count, err); - } + err = iser_post_recvm(iser_conn, desc); + if (err) + iser_err("posting rx buffer err %d\n", err); } void iser_cmd_comp(struct ib_cq *cq, struct ib_wc *wc) @@ -764,27 +728,16 @@ void iser_task_rdma_init(struct iscsi_iser_task *iser_task) void iser_task_rdma_finalize(struct iscsi_iser_task *iser_task) { - int prot_count = scsi_prot_sg_count(iser_task->sc); if (iser_task->dir[ISER_DIR_IN]) { iser_unreg_mem_fastreg(iser_task, ISER_DIR_IN); - iser_dma_unmap_task_data(iser_task, - &iser_task->data[ISER_DIR_IN], + iser_dma_unmap_task_data(iser_task, ISER_DIR_IN, DMA_FROM_DEVICE); - if (prot_count) - iser_dma_unmap_task_data(iser_task, - &iser_task->prot[ISER_DIR_IN], - DMA_FROM_DEVICE); } if (iser_task->dir[ISER_DIR_OUT]) { iser_unreg_mem_fastreg(iser_task, ISER_DIR_OUT); - iser_dma_unmap_task_data(iser_task, - &iser_task->data[ISER_DIR_OUT], + iser_dma_unmap_task_data(iser_task, ISER_DIR_OUT, DMA_TO_DEVICE); - if (prot_count) - iser_dma_unmap_task_data(iser_task, - &iser_task->prot[ISER_DIR_OUT], - DMA_TO_DEVICE); } } diff --git a/drivers/infiniband/ulp/iser/iser_memory.c b/drivers/infiniband/ulp/iser/iser_memory.c index 9776b755d848..29ae2c6a250a 100644 --- a/drivers/infiniband/ulp/iser/iser_memory.c +++ b/drivers/infiniband/ulp/iser/iser_memory.c @@ -30,7 +30,6 @@ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include <linux/module.h> #include <linux/kernel.h> #include <linux/slab.h> #include <linux/mm.h> @@ -44,8 +43,7 @@ void iser_reg_comp(struct ib_cq *cq, struct ib_wc *wc) iser_err_comp(wc, "memreg"); } -static struct iser_fr_desc * -iser_reg_desc_get_fr(struct ib_conn *ib_conn) +static struct iser_fr_desc *iser_reg_desc_get_fr(struct ib_conn *ib_conn) { struct iser_fr_pool *fr_pool = &ib_conn->fr_pool; struct iser_fr_desc *desc; @@ -60,9 +58,8 @@ iser_reg_desc_get_fr(struct ib_conn *ib_conn) return desc; } -static void -iser_reg_desc_put_fr(struct ib_conn *ib_conn, - struct iser_fr_desc *desc) +static void iser_reg_desc_put_fr(struct ib_conn *ib_conn, + struct iser_fr_desc *desc) { struct iser_fr_pool *fr_pool = &ib_conn->fr_pool; unsigned long flags; @@ -73,10 +70,10 @@ iser_reg_desc_put_fr(struct ib_conn *ib_conn, } int iser_dma_map_task_data(struct iscsi_iser_task *iser_task, - struct iser_data_buf *data, - enum iser_data_dir iser_dir, - enum dma_data_direction dma_dir) + enum iser_data_dir iser_dir, + enum dma_data_direction dma_dir) { + struct iser_data_buf *data = &iser_task->data[iser_dir]; struct ib_device *dev; iser_task->dir[iser_dir] = 1; @@ -87,22 +84,44 @@ int iser_dma_map_task_data(struct iscsi_iser_task *iser_task, iser_err("dma_map_sg failed!!!\n"); return -EINVAL; } + + if (scsi_prot_sg_count(iser_task->sc)) { + struct iser_data_buf *pdata = &iser_task->prot[iser_dir]; + + pdata->dma_nents = ib_dma_map_sg(dev, pdata->sg, pdata->size, dma_dir); + if (unlikely(pdata->dma_nents == 0)) { + iser_err("protection dma_map_sg failed!!!\n"); + goto out_unmap; + } + } + return 0; + +out_unmap: + ib_dma_unmap_sg(dev, data->sg, data->size, dma_dir); + return -EINVAL; } + void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task, - struct iser_data_buf *data, - enum dma_data_direction dir) + enum iser_data_dir iser_dir, + enum dma_data_direction dma_dir) { + struct iser_data_buf *data = &iser_task->data[iser_dir]; struct ib_device *dev; dev = iser_task->iser_conn->ib_conn.device->ib_device; - ib_dma_unmap_sg(dev, data->sg, data->size, dir); + ib_dma_unmap_sg(dev, data->sg, data->size, dma_dir); + + if (scsi_prot_sg_count(iser_task->sc)) { + struct iser_data_buf *pdata = &iser_task->prot[iser_dir]; + + ib_dma_unmap_sg(dev, pdata->sg, pdata->size, dma_dir); + } } -static int -iser_reg_dma(struct iser_device *device, struct iser_data_buf *mem, - struct iser_mem_reg *reg) +static int iser_reg_dma(struct iser_device *device, struct iser_data_buf *mem, + struct iser_mem_reg *reg) { struct scatterlist *sg = mem->sg; @@ -133,7 +152,7 @@ void iser_unreg_mem_fastreg(struct iscsi_iser_task *iser_task, struct iser_fr_desc *desc; struct ib_mr_status mr_status; - desc = reg->mem_h; + desc = reg->desc; if (!desc) return; @@ -150,12 +169,12 @@ void iser_unreg_mem_fastreg(struct iscsi_iser_task *iser_task, ib_check_mr_status(desc->rsc.sig_mr, IB_MR_CHECK_SIG_STATUS, &mr_status); } - iser_reg_desc_put_fr(&iser_task->iser_conn->ib_conn, reg->mem_h); - reg->mem_h = NULL; + iser_reg_desc_put_fr(&iser_task->iser_conn->ib_conn, reg->desc); + reg->desc = NULL; } -static void -iser_set_dif_domain(struct scsi_cmnd *sc, struct ib_sig_domain *domain) +static void iser_set_dif_domain(struct scsi_cmnd *sc, + struct ib_sig_domain *domain) { domain->sig_type = IB_SIG_TYPE_T10_DIF; domain->sig.dif.pi_interval = scsi_prot_interval(sc); @@ -171,8 +190,8 @@ iser_set_dif_domain(struct scsi_cmnd *sc, struct ib_sig_domain *domain) domain->sig.dif.ref_remap = true; } -static int -iser_set_sig_attrs(struct scsi_cmnd *sc, struct ib_sig_attrs *sig_attrs) +static int iser_set_sig_attrs(struct scsi_cmnd *sc, + struct ib_sig_attrs *sig_attrs) { switch (scsi_get_prot_op(sc)) { case SCSI_PROT_WRITE_INSERT: @@ -205,8 +224,7 @@ iser_set_sig_attrs(struct scsi_cmnd *sc, struct ib_sig_attrs *sig_attrs) return 0; } -static inline void -iser_set_prot_checks(struct scsi_cmnd *sc, u8 *mask) +static inline void iser_set_prot_checks(struct scsi_cmnd *sc, u8 *mask) { *mask = 0; if (sc->prot_flags & SCSI_PROT_REF_CHECK) @@ -215,11 +233,8 @@ iser_set_prot_checks(struct scsi_cmnd *sc, u8 *mask) *mask |= IB_SIG_CHECK_GUARD; } -static inline void -iser_inv_rkey(struct ib_send_wr *inv_wr, - struct ib_mr *mr, - struct ib_cqe *cqe, - struct ib_send_wr *next_wr) +static inline void iser_inv_rkey(struct ib_send_wr *inv_wr, struct ib_mr *mr, + struct ib_cqe *cqe, struct ib_send_wr *next_wr) { inv_wr->opcode = IB_WR_LOCAL_INV; inv_wr->wr_cqe = cqe; @@ -229,12 +244,11 @@ iser_inv_rkey(struct ib_send_wr *inv_wr, inv_wr->next = next_wr; } -static int -iser_reg_sig_mr(struct iscsi_iser_task *iser_task, - struct iser_data_buf *mem, - struct iser_data_buf *sig_mem, - struct iser_reg_resources *rsc, - struct iser_mem_reg *sig_reg) +static int iser_reg_sig_mr(struct iscsi_iser_task *iser_task, + struct iser_data_buf *mem, + struct iser_data_buf *sig_mem, + struct iser_reg_resources *rsc, + struct iser_mem_reg *sig_reg) { struct iser_tx_desc *tx_desc = &iser_task->desc; struct ib_cqe *cqe = &iser_task->iser_conn->ib_conn.reg_cqe; @@ -335,42 +349,26 @@ static int iser_fast_reg_mr(struct iscsi_iser_task *iser_task, return 0; } -static int -iser_reg_data_sg(struct iscsi_iser_task *task, - struct iser_data_buf *mem, - struct iser_fr_desc *desc, - bool use_dma_key, - struct iser_mem_reg *reg) -{ - struct iser_device *device = task->iser_conn->ib_conn.device; - - if (use_dma_key) - return iser_reg_dma(device, mem, reg); - - return iser_fast_reg_mr(task, mem, &desc->rsc, reg); -} - int iser_reg_mem_fastreg(struct iscsi_iser_task *task, enum iser_data_dir dir, bool all_imm) { struct ib_conn *ib_conn = &task->iser_conn->ib_conn; + struct iser_device *device = ib_conn->device; struct iser_data_buf *mem = &task->data[dir]; struct iser_mem_reg *reg = &task->rdma_reg[dir]; - struct iser_fr_desc *desc = NULL; + struct iser_fr_desc *desc; bool use_dma_key; int err; use_dma_key = mem->dma_nents == 1 && (all_imm || !iser_always_reg) && scsi_get_prot_op(task->sc) == SCSI_PROT_NORMAL; + if (use_dma_key) + return iser_reg_dma(device, mem, reg); - if (!use_dma_key) { - desc = iser_reg_desc_get_fr(ib_conn); - reg->mem_h = desc; - } - + desc = iser_reg_desc_get_fr(ib_conn); if (scsi_get_prot_op(task->sc) == SCSI_PROT_NORMAL) { - err = iser_reg_data_sg(task, mem, desc, use_dma_key, reg); + err = iser_fast_reg_mr(task, mem, &desc->rsc, reg); if (unlikely(err)) goto err_reg; } else { @@ -382,11 +380,12 @@ int iser_reg_mem_fastreg(struct iscsi_iser_task *task, desc->sig_protected = true; } + reg->desc = desc; + return 0; err_reg: - if (desc) - iser_reg_desc_put_fr(ib_conn, desc); + iser_reg_desc_put_fr(ib_conn, desc); return err; } diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index b566f7cb7797..a00ca117303a 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -32,7 +32,6 @@ * SOFTWARE. */ #include <linux/kernel.h> -#include <linux/module.h> #include <linux/slab.h> #include <linux/delay.h> @@ -116,7 +115,7 @@ iser_create_fastreg_desc(struct iser_device *device, if (!desc) return ERR_PTR(-ENOMEM); - if (ib_dev->attrs.device_cap_flags & IB_DEVICE_SG_GAPS_REG) + if (ib_dev->attrs.kernel_cap_flags & IBK_SG_GAPS_REG) mr_type = IB_MR_TYPE_SG_GAPS; else mr_type = IB_MR_TYPE_MEM_REG; @@ -247,6 +246,7 @@ static int iser_create_ib_conn_res(struct ib_conn *ib_conn) device = ib_conn->device; ib_dev = device->ib_device; + /* +1 for drain */ if (ib_conn->pi_support) max_send_wr = ISER_QP_SIG_MAX_REQ_DTOS + 1; else @@ -265,14 +265,15 @@ static int iser_create_ib_conn_res(struct ib_conn *ib_conn) memset(&init_attr, 0, sizeof(init_attr)); init_attr.event_handler = iser_qp_event_callback; - init_attr.qp_context = (void *)ib_conn; - init_attr.send_cq = ib_conn->cq; - init_attr.recv_cq = ib_conn->cq; - init_attr.cap.max_recv_wr = ISER_QP_MAX_RECV_DTOS; + init_attr.qp_context = (void *)ib_conn; + init_attr.send_cq = ib_conn->cq; + init_attr.recv_cq = ib_conn->cq; + /* +1 for drain */ + init_attr.cap.max_recv_wr = ISER_QP_MAX_RECV_DTOS + 1; init_attr.cap.max_send_sge = 2; init_attr.cap.max_recv_sge = 1; - init_attr.sq_sig_type = IB_SIGNAL_REQ_WR; - init_attr.qp_type = IB_QPT_RC; + init_attr.sq_sig_type = IB_SIGNAL_REQ_WR; + init_attr.qp_type = IB_QPT_RC; init_attr.cap.max_send_wr = max_send_wr; if (ib_conn->pi_support) init_attr.create_flags |= IB_QP_CREATE_INTEGRITY_EN; @@ -283,9 +284,8 @@ static int iser_create_ib_conn_res(struct ib_conn *ib_conn) goto out_err; ib_conn->qp = ib_conn->cma_id->qp; - iser_info("setting conn %p cma_id %p qp %p max_send_wr %d\n", - ib_conn, ib_conn->cma_id, - ib_conn->cma_id->qp, max_send_wr); + iser_info("setting conn %p cma_id %p qp %p max_send_wr %d\n", ib_conn, + ib_conn->cma_id, ib_conn->cma_id->qp, max_send_wr); return ret; out_err: @@ -313,7 +313,7 @@ struct iser_device *iser_device_find_by_ib_device(struct rdma_cm_id *cma_id) goto inc_refcnt; device = kzalloc(sizeof *device, GFP_KERNEL); - if (device == NULL) + if (!device) goto out; /* assign this device to the device */ @@ -392,8 +392,7 @@ void iser_release_work(struct work_struct *work) * so the cm_id removal is out of here. It is Safe to * be invoked multiple times. */ -static void iser_free_ib_conn_res(struct iser_conn *iser_conn, - bool destroy) +static void iser_free_ib_conn_res(struct iser_conn *iser_conn, bool destroy) { struct ib_conn *ib_conn = &iser_conn->ib_conn; struct iser_device *device = ib_conn->device; @@ -401,7 +400,7 @@ static void iser_free_ib_conn_res(struct iser_conn *iser_conn, iser_info("freeing conn %p cma_id %p qp %p\n", iser_conn, ib_conn->cma_id, ib_conn->qp); - if (ib_conn->qp != NULL) { + if (ib_conn->qp) { rdma_destroy_qp(ib_conn->cma_id); ib_cq_pool_put(ib_conn->cq, ib_conn->cq_size); ib_conn->qp = NULL; @@ -411,7 +410,7 @@ static void iser_free_ib_conn_res(struct iser_conn *iser_conn, if (iser_conn->rx_descs) iser_free_rx_descriptors(iser_conn); - if (device != NULL) { + if (device) { iser_device_try_release(device); ib_conn->device = NULL; } @@ -445,7 +444,7 @@ void iser_conn_release(struct iser_conn *iser_conn) iser_free_ib_conn_res(iser_conn, true); mutex_unlock(&iser_conn->state_mutex); - if (ib_conn->cma_id != NULL) { + if (ib_conn->cma_id) { rdma_destroy_id(ib_conn->cma_id); ib_conn->cma_id = NULL; } @@ -488,7 +487,7 @@ int iser_conn_terminate(struct iser_conn *iser_conn) iser_conn, err); /* block until all flush errors are consumed */ - ib_drain_sq(ib_conn->qp); + ib_drain_qp(ib_conn->qp); } return 1; @@ -501,13 +500,12 @@ static void iser_connect_error(struct rdma_cm_id *cma_id) { struct iser_conn *iser_conn; - iser_conn = (struct iser_conn *)cma_id->context; + iser_conn = cma_id->context; iser_conn->state = ISER_CONN_TERMINATING; } -static void -iser_calc_scsi_params(struct iser_conn *iser_conn, - unsigned int max_sectors) +static void iser_calc_scsi_params(struct iser_conn *iser_conn, + unsigned int max_sectors) { struct iser_device *device = iser_conn->ib_conn.device; struct ib_device_attr *attr = &device->ib_device->attrs; @@ -521,7 +519,7 @@ iser_calc_scsi_params(struct iser_conn *iser_conn, * (head and tail) for a single page worth data, so one additional * entry is required. */ - if (attr->device_cap_flags & IB_DEVICE_SG_GAPS_REG) + if (attr->kernel_cap_flags & IBK_SG_GAPS_REG) reserved_mr_pages = 0; else reserved_mr_pages = 1; @@ -545,11 +543,11 @@ iser_calc_scsi_params(struct iser_conn *iser_conn, static void iser_addr_handler(struct rdma_cm_id *cma_id) { struct iser_device *device; - struct iser_conn *iser_conn; - struct ib_conn *ib_conn; + struct iser_conn *iser_conn; + struct ib_conn *ib_conn; int ret; - iser_conn = (struct iser_conn *)cma_id->context; + iser_conn = cma_id->context; if (iser_conn->state != ISER_CONN_PENDING) /* bailout */ return; @@ -566,8 +564,8 @@ static void iser_addr_handler(struct rdma_cm_id *cma_id) /* connection T10-PI support */ if (iser_pi_enable) { - if (!(device->ib_device->attrs.device_cap_flags & - IB_DEVICE_INTEGRITY_HANDOVER)) { + if (!(device->ib_device->attrs.kernel_cap_flags & + IBK_INTEGRITY_HANDOVER)) { iser_warn("T10-PI requested but not supported on %s, " "continue without T10-PI\n", dev_name(&ib_conn->device->ib_device->dev)); @@ -593,9 +591,9 @@ static void iser_addr_handler(struct rdma_cm_id *cma_id) static void iser_route_handler(struct rdma_cm_id *cma_id) { struct rdma_conn_param conn_param; - int ret; + int ret; struct iser_cm_hdr req_hdr; - struct iser_conn *iser_conn = (struct iser_conn *)cma_id->context; + struct iser_conn *iser_conn = cma_id->context; struct ib_conn *ib_conn = &iser_conn->ib_conn; struct ib_device *ib_dev = ib_conn->device->ib_device; @@ -609,9 +607,9 @@ static void iser_route_handler(struct rdma_cm_id *cma_id) memset(&conn_param, 0, sizeof conn_param); conn_param.responder_resources = ib_dev->attrs.max_qp_rd_atom; - conn_param.initiator_depth = 1; - conn_param.retry_count = 7; - conn_param.rnr_retry_count = 6; + conn_param.initiator_depth = 1; + conn_param.retry_count = 7; + conn_param.rnr_retry_count = 6; memset(&req_hdr, 0, sizeof(req_hdr)); req_hdr.flags = ISER_ZBVA_NOT_SUP; @@ -638,7 +636,7 @@ static void iser_connected_handler(struct rdma_cm_id *cma_id, struct ib_qp_attr attr; struct ib_qp_init_attr init_attr; - iser_conn = (struct iser_conn *)cma_id->context; + iser_conn = cma_id->context; if (iser_conn->state != ISER_CONN_PENDING) /* bailout */ return; @@ -661,7 +659,7 @@ static void iser_connected_handler(struct rdma_cm_id *cma_id, static void iser_disconnected_handler(struct rdma_cm_id *cma_id) { - struct iser_conn *iser_conn = (struct iser_conn *)cma_id->context; + struct iser_conn *iser_conn = cma_id->context; if (iser_conn_terminate(iser_conn)) { if (iser_conn->iscsi_conn) @@ -675,7 +673,7 @@ static void iser_disconnected_handler(struct rdma_cm_id *cma_id) static void iser_cleanup_handler(struct rdma_cm_id *cma_id, bool destroy) { - struct iser_conn *iser_conn = (struct iser_conn *)cma_id->context; + struct iser_conn *iser_conn = cma_id->context; /* * We are not guaranteed that we visited disconnected_handler @@ -687,12 +685,13 @@ static void iser_cleanup_handler(struct rdma_cm_id *cma_id, complete(&iser_conn->ib_completion); } -static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) +static int iser_cma_handler(struct rdma_cm_id *cma_id, + struct rdma_cm_event *event) { struct iser_conn *iser_conn; int ret = 0; - iser_conn = (struct iser_conn *)cma_id->context; + iser_conn = cma_id->context; iser_info("%s (%d): status %d conn %p id %p\n", rdma_event_msg(event->event), event->event, event->status, cma_id->context, cma_id); @@ -757,7 +756,6 @@ void iser_conn_init(struct iser_conn *iser_conn) INIT_LIST_HEAD(&iser_conn->conn_list); mutex_init(&iser_conn->state_mutex); - ib_conn->post_recv_buf_count = 0; ib_conn->reg_cqe.done = iser_reg_comp; } @@ -765,10 +763,8 @@ void iser_conn_init(struct iser_conn *iser_conn) * starts the process of connecting to the target * sleeps until the connection is established or rejected */ -int iser_connect(struct iser_conn *iser_conn, - struct sockaddr *src_addr, - struct sockaddr *dst_addr, - int non_blocking) +int iser_connect(struct iser_conn *iser_conn, struct sockaddr *src_addr, + struct sockaddr *dst_addr, int non_blocking) { struct ib_conn *ib_conn = &iser_conn->ib_conn; int err = 0; @@ -785,8 +781,7 @@ int iser_connect(struct iser_conn *iser_conn, iser_conn->state = ISER_CONN_PENDING; ib_conn->cma_id = rdma_create_id(&init_net, iser_cma_handler, - (void *)iser_conn, - RDMA_PS_TCP, IB_QPT_RC); + iser_conn, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(ib_conn->cma_id)) { err = PTR_ERR(ib_conn->cma_id); iser_err("rdma_create_id failed: %d\n", err); @@ -829,7 +824,7 @@ int iser_post_recvl(struct iser_conn *iser_conn) struct ib_conn *ib_conn = &iser_conn->ib_conn; struct iser_login_desc *desc = &iser_conn->login_desc; struct ib_recv_wr wr; - int ib_ret; + int ret; desc->sge.addr = desc->rsp_dma; desc->sge.length = ISER_RX_LOGIN_SIZE; @@ -841,46 +836,30 @@ int iser_post_recvl(struct iser_conn *iser_conn) wr.num_sge = 1; wr.next = NULL; - ib_conn->post_recv_buf_count++; - ib_ret = ib_post_recv(ib_conn->qp, &wr, NULL); - if (ib_ret) { - iser_err("ib_post_recv failed ret=%d\n", ib_ret); - ib_conn->post_recv_buf_count--; - } + ret = ib_post_recv(ib_conn->qp, &wr, NULL); + if (unlikely(ret)) + iser_err("ib_post_recv login failed ret=%d\n", ret); - return ib_ret; + return ret; } -int iser_post_recvm(struct iser_conn *iser_conn, int count) +int iser_post_recvm(struct iser_conn *iser_conn, struct iser_rx_desc *rx_desc) { struct ib_conn *ib_conn = &iser_conn->ib_conn; - unsigned int my_rx_head = iser_conn->rx_desc_head; - struct iser_rx_desc *rx_desc; - struct ib_recv_wr *wr; - int i, ib_ret; - - for (wr = ib_conn->rx_wr, i = 0; i < count; i++, wr++) { - rx_desc = &iser_conn->rx_descs[my_rx_head]; - rx_desc->cqe.done = iser_task_rsp; - wr->wr_cqe = &rx_desc->cqe; - wr->sg_list = &rx_desc->rx_sg; - wr->num_sge = 1; - wr->next = wr + 1; - my_rx_head = (my_rx_head + 1) & iser_conn->qp_max_recv_dtos_mask; - } + struct ib_recv_wr wr; + int ret; - wr--; - wr->next = NULL; /* mark end of work requests list */ + rx_desc->cqe.done = iser_task_rsp; + wr.wr_cqe = &rx_desc->cqe; + wr.sg_list = &rx_desc->rx_sg; + wr.num_sge = 1; + wr.next = NULL; - ib_conn->post_recv_buf_count += count; - ib_ret = ib_post_recv(ib_conn->qp, ib_conn->rx_wr, NULL); - if (unlikely(ib_ret)) { - iser_err("ib_post_recv failed ret=%d\n", ib_ret); - ib_conn->post_recv_buf_count -= count; - } else - iser_conn->rx_desc_head = my_rx_head; + ret = ib_post_recv(ib_conn->qp, &wr, NULL); + if (unlikely(ret)) + iser_err("ib_post_recv failed ret=%d\n", ret); - return ib_ret; + return ret; } @@ -888,16 +867,14 @@ int iser_post_recvm(struct iser_conn *iser_conn, int count) * iser_post_send - Initiate a Send DTO operation * @ib_conn: connection RDMA resources * @tx_desc: iSER TX descriptor - * @signal: true to send work request as SIGNALED * * Return: 0 on success, -1 on failure */ -int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc, - bool signal) +int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc) { struct ib_send_wr *wr = &tx_desc->send_wr; struct ib_send_wr *first_wr; - int ib_ret; + int ret; ib_dma_sync_single_for_device(ib_conn->device->ib_device, tx_desc->dma_addr, ISER_HEADERS_LEN, @@ -908,7 +885,7 @@ int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc, wr->sg_list = tx_desc->tx_sg; wr->num_sge = tx_desc->num_sge; wr->opcode = IB_WR_SEND; - wr->send_flags = signal ? IB_SEND_SIGNALED : 0; + wr->send_flags = IB_SEND_SIGNALED; if (tx_desc->inv_wr.next) first_wr = &tx_desc->inv_wr; @@ -917,19 +894,19 @@ int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc, else first_wr = wr; - ib_ret = ib_post_send(ib_conn->qp, first_wr, NULL); - if (unlikely(ib_ret)) + ret = ib_post_send(ib_conn->qp, first_wr, NULL); + if (unlikely(ret)) iser_err("ib_post_send failed, ret:%d opcode:%d\n", - ib_ret, wr->opcode); + ret, wr->opcode); - return ib_ret; + return ret; } u8 iser_check_task_pi_status(struct iscsi_iser_task *iser_task, enum iser_data_dir cmd_dir, sector_t *sector) { struct iser_mem_reg *reg = &iser_task->rdma_reg[cmd_dir]; - struct iser_fr_desc *desc = reg->mem_h; + struct iser_fr_desc *desc = reg->desc; unsigned long sector_size = iser_task->sc->device->sector_size; struct ib_mr_status mr_status; int ret; diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c index 636d590765f9..b360a1527cd1 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.c +++ b/drivers/infiniband/ulp/isert/ib_isert.c @@ -42,11 +42,12 @@ MODULE_PARM_DESC(sg_tablesize, static DEFINE_MUTEX(device_list_mutex); static LIST_HEAD(device_list); +static struct workqueue_struct *isert_login_wq; static struct workqueue_struct *isert_comp_wq; static struct workqueue_struct *isert_release_wq; static int -isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd); +isert_put_response(struct iscsit_conn *conn, struct iscsit_cmd *cmd); static int isert_login_post_recv(struct isert_conn *isert_conn); static int @@ -230,7 +231,7 @@ isert_create_device_ib_res(struct isert_device *device) } /* Check signature cap */ - if (ib_dev->attrs.device_cap_flags & IB_DEVICE_INTEGRITY_HANDOVER) + if (ib_dev->attrs.kernel_cap_flags & IBK_INTEGRITY_HANDOVER) device->pi_capable = true; else device->pi_capable = false; @@ -909,7 +910,7 @@ isert_login_post_recv(struct isert_conn *isert_conn) } static int -isert_put_login_tx(struct iscsi_conn *conn, struct iscsi_login *login, +isert_put_login_tx(struct iscsit_conn *conn, struct iscsi_login *login, u32 length) { struct isert_conn *isert_conn = conn->context; @@ -976,7 +977,7 @@ isert_rx_login_req(struct isert_conn *isert_conn) { struct iser_rx_desc *rx_desc = isert_conn->login_desc; int rx_buflen = isert_conn->login_req_len; - struct iscsi_conn *conn = isert_conn->conn; + struct iscsit_conn *conn = isert_conn->conn; struct iscsi_login *login = conn->conn_login; int size; @@ -1017,24 +1018,24 @@ isert_rx_login_req(struct isert_conn *isert_conn) complete(&isert_conn->login_comp); return; } - schedule_delayed_work(&conn->login_work, 0); + queue_delayed_work(isert_login_wq, &conn->login_work, 0); } -static struct iscsi_cmd -*isert_allocate_cmd(struct iscsi_conn *conn, struct iser_rx_desc *rx_desc) +static struct iscsit_cmd +*isert_allocate_cmd(struct iscsit_conn *conn, struct iser_rx_desc *rx_desc) { struct isert_conn *isert_conn = conn->context; struct isert_cmd *isert_cmd; - struct iscsi_cmd *cmd; + struct iscsit_cmd *cmd; cmd = iscsit_allocate_cmd(conn, TASK_INTERRUPTIBLE); if (!cmd) { - isert_err("Unable to allocate iscsi_cmd + isert_cmd\n"); + isert_err("Unable to allocate iscsit_cmd + isert_cmd\n"); return NULL; } isert_cmd = iscsit_priv_cmd(cmd); isert_cmd->conn = isert_conn; - isert_cmd->iscsi_cmd = cmd; + isert_cmd->iscsit_cmd = cmd; isert_cmd->rx_desc = rx_desc; return cmd; @@ -1042,10 +1043,10 @@ static struct iscsi_cmd static int isert_handle_scsi_cmd(struct isert_conn *isert_conn, - struct isert_cmd *isert_cmd, struct iscsi_cmd *cmd, + struct isert_cmd *isert_cmd, struct iscsit_cmd *cmd, struct iser_rx_desc *rx_desc, unsigned char *buf) { - struct iscsi_conn *conn = isert_conn->conn; + struct iscsit_conn *conn = isert_conn->conn; struct iscsi_scsi_req *hdr = (struct iscsi_scsi_req *)buf; int imm_data, imm_data_len, unsol_data, sg_nents, rc; bool dump_payload = false; @@ -1114,8 +1115,8 @@ isert_handle_iscsi_dataout(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc, unsigned char *buf) { struct scatterlist *sg_start; - struct iscsi_conn *conn = isert_conn->conn; - struct iscsi_cmd *cmd = NULL; + struct iscsit_conn *conn = isert_conn->conn; + struct iscsit_cmd *cmd = NULL; struct iscsi_data *hdr = (struct iscsi_data *)buf; u32 unsol_data_len = ntoh24(hdr->dlength); int rc, sg_nents, sg_off, page_off; @@ -1171,10 +1172,10 @@ isert_handle_iscsi_dataout(struct isert_conn *isert_conn, static int isert_handle_nop_out(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd, - struct iscsi_cmd *cmd, struct iser_rx_desc *rx_desc, + struct iscsit_cmd *cmd, struct iser_rx_desc *rx_desc, unsigned char *buf) { - struct iscsi_conn *conn = isert_conn->conn; + struct iscsit_conn *conn = isert_conn->conn; struct iscsi_nopout *hdr = (struct iscsi_nopout *)buf; int rc; @@ -1190,10 +1191,10 @@ isert_handle_nop_out(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd, static int isert_handle_text_cmd(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd, - struct iscsi_cmd *cmd, struct iser_rx_desc *rx_desc, + struct iscsit_cmd *cmd, struct iser_rx_desc *rx_desc, struct iscsi_text *hdr) { - struct iscsi_conn *conn = isert_conn->conn; + struct iscsit_conn *conn = isert_conn->conn; u32 payload_length = ntoh24(hdr->dlength); int rc; unsigned char *text_in = NULL; @@ -1220,8 +1221,8 @@ isert_rx_opcode(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc, uint32_t write_stag, uint64_t write_va) { struct iscsi_hdr *hdr = isert_get_iscsi_hdr(rx_desc); - struct iscsi_conn *conn = isert_conn->conn; - struct iscsi_cmd *cmd; + struct iscsit_conn *conn = isert_conn->conn; + struct iscsit_cmd *cmd; struct isert_cmd *isert_cmd; int ret = -EINVAL; u8 opcode = (hdr->opcode & ISCSI_OPCODE_MASK); @@ -1404,7 +1405,7 @@ isert_login_recv_done(struct ib_cq *cq, struct ib_wc *wc) static void isert_rdma_rw_ctx_destroy(struct isert_cmd *cmd, struct isert_conn *conn) { - struct se_cmd *se_cmd = &cmd->iscsi_cmd->se_cmd; + struct se_cmd *se_cmd = &cmd->iscsit_cmd->se_cmd; enum dma_data_direction dir = target_reverse_dma_direction(se_cmd); if (!cmd->rw.nr_ops) @@ -1426,9 +1427,9 @@ isert_rdma_rw_ctx_destroy(struct isert_cmd *cmd, struct isert_conn *conn) static void isert_put_cmd(struct isert_cmd *isert_cmd, bool comp_err) { - struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd; + struct iscsit_cmd *cmd = isert_cmd->iscsit_cmd; struct isert_conn *isert_conn = isert_cmd->conn; - struct iscsi_conn *conn = isert_conn->conn; + struct iscsit_conn *conn = isert_conn->conn; struct iscsi_text_rsp *hdr; isert_dbg("Cmd %p\n", isert_cmd); @@ -1575,7 +1576,7 @@ isert_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc) struct isert_device *device = isert_conn->device; struct iser_tx_desc *desc = cqe_to_tx_desc(wc->wr_cqe); struct isert_cmd *isert_cmd = tx_desc_to_cmd(desc); - struct se_cmd *cmd = &isert_cmd->iscsi_cmd->se_cmd; + struct se_cmd *cmd = &isert_cmd->iscsit_cmd->se_cmd; int ret = 0; if (unlikely(wc->status != IB_WC_SUCCESS)) { @@ -1604,7 +1605,7 @@ isert_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc) /* * XXX: isert_put_response() failure is not retried. */ - ret = isert_put_response(isert_conn->conn, isert_cmd->iscsi_cmd); + ret = isert_put_response(isert_conn->conn, isert_cmd->iscsit_cmd); if (ret) pr_warn_ratelimited("isert_put_response() ret: %d\n", ret); } @@ -1617,7 +1618,7 @@ isert_rdma_read_done(struct ib_cq *cq, struct ib_wc *wc) struct isert_device *device = isert_conn->device; struct iser_tx_desc *desc = cqe_to_tx_desc(wc->wr_cqe); struct isert_cmd *isert_cmd = tx_desc_to_cmd(desc); - struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd; + struct iscsit_cmd *cmd = isert_cmd->iscsit_cmd; struct se_cmd *se_cmd = &cmd->se_cmd; int ret = 0; @@ -1662,7 +1663,7 @@ isert_do_control_comp(struct work_struct *work) struct isert_cmd, comp_work); struct isert_conn *isert_conn = isert_cmd->conn; struct ib_device *ib_dev = isert_conn->cm_id->device; - struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd; + struct iscsit_cmd *cmd = isert_cmd->iscsit_cmd; isert_dbg("Cmd %p i_state %d\n", isert_cmd, cmd->i_state); @@ -1720,7 +1721,7 @@ isert_send_done(struct ib_cq *cq, struct ib_wc *wc) isert_dbg("Cmd %p\n", isert_cmd); - switch (isert_cmd->iscsi_cmd->i_state) { + switch (isert_cmd->iscsit_cmd->i_state) { case ISTATE_SEND_TASKMGTRSP: case ISTATE_SEND_LOGOUTRSP: case ISTATE_SEND_REJECT: @@ -1731,7 +1732,7 @@ isert_send_done(struct ib_cq *cq, struct ib_wc *wc) queue_work(isert_comp_wq, &isert_cmd->comp_work); return; default: - isert_cmd->iscsi_cmd->i_state = ISTATE_SENT_STATUS; + isert_cmd->iscsit_cmd->i_state = ISTATE_SENT_STATUS; isert_completion_put(tx_desc, isert_cmd, ib_dev, false); break; } @@ -1755,7 +1756,7 @@ isert_post_response(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd) } static int -isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd) +isert_put_response(struct iscsit_conn *conn, struct iscsit_cmd *cmd) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); struct isert_conn *isert_conn = conn->context; @@ -1806,7 +1807,7 @@ isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd) } static void -isert_aborted_task(struct iscsi_conn *conn, struct iscsi_cmd *cmd) +isert_aborted_task(struct iscsit_conn *conn, struct iscsit_cmd *cmd) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); struct isert_conn *isert_conn = conn->context; @@ -1822,7 +1823,7 @@ isert_aborted_task(struct iscsi_conn *conn, struct iscsi_cmd *cmd) } static enum target_prot_op -isert_get_sup_prot_ops(struct iscsi_conn *conn) +isert_get_sup_prot_ops(struct iscsit_conn *conn) { struct isert_conn *isert_conn = conn->context; struct isert_device *device = isert_conn->device; @@ -1842,7 +1843,7 @@ isert_get_sup_prot_ops(struct iscsi_conn *conn) } static int -isert_put_nopin(struct iscsi_cmd *cmd, struct iscsi_conn *conn, +isert_put_nopin(struct iscsit_cmd *cmd, struct iscsit_conn *conn, bool nopout_response) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); @@ -1862,7 +1863,7 @@ isert_put_nopin(struct iscsi_cmd *cmd, struct iscsi_conn *conn, } static int -isert_put_logout_rsp(struct iscsi_cmd *cmd, struct iscsi_conn *conn) +isert_put_logout_rsp(struct iscsit_cmd *cmd, struct iscsit_conn *conn) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); struct isert_conn *isert_conn = conn->context; @@ -1880,7 +1881,7 @@ isert_put_logout_rsp(struct iscsi_cmd *cmd, struct iscsi_conn *conn) } static int -isert_put_tm_rsp(struct iscsi_cmd *cmd, struct iscsi_conn *conn) +isert_put_tm_rsp(struct iscsit_cmd *cmd, struct iscsit_conn *conn) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); struct isert_conn *isert_conn = conn->context; @@ -1898,7 +1899,7 @@ isert_put_tm_rsp(struct iscsi_cmd *cmd, struct iscsi_conn *conn) } static int -isert_put_reject(struct iscsi_cmd *cmd, struct iscsi_conn *conn) +isert_put_reject(struct iscsit_cmd *cmd, struct iscsit_conn *conn) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); struct isert_conn *isert_conn = conn->context; @@ -1933,7 +1934,7 @@ isert_put_reject(struct iscsi_cmd *cmd, struct iscsi_conn *conn) } static int -isert_put_text_rsp(struct iscsi_cmd *cmd, struct iscsi_conn *conn) +isert_put_text_rsp(struct iscsit_cmd *cmd, struct iscsit_conn *conn) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); struct isert_conn *isert_conn = conn->context; @@ -2035,7 +2036,7 @@ static int isert_rdma_rw_ctx_post(struct isert_cmd *cmd, struct isert_conn *conn, struct ib_cqe *cqe, struct ib_send_wr *chain_wr) { - struct se_cmd *se_cmd = &cmd->iscsi_cmd->se_cmd; + struct se_cmd *se_cmd = &cmd->iscsit_cmd->se_cmd; enum dma_data_direction dir = target_reverse_dma_direction(se_cmd); u8 port_num = conn->cm_id->port_num; u64 addr; @@ -2048,7 +2049,7 @@ isert_rdma_rw_ctx_post(struct isert_cmd *cmd, struct isert_conn *conn, if (dir == DMA_FROM_DEVICE) { addr = cmd->write_va; rkey = cmd->write_stag; - offset = cmd->iscsi_cmd->write_data_done; + offset = cmd->iscsit_cmd->write_data_done; } else { addr = cmd->read_va; rkey = cmd->read_stag; @@ -2088,7 +2089,7 @@ rdma_ctx_post: } static int -isert_put_datain(struct iscsi_conn *conn, struct iscsi_cmd *cmd) +isert_put_datain(struct iscsit_conn *conn, struct iscsit_cmd *cmd) { struct se_cmd *se_cmd = &cmd->se_cmd; struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); @@ -2129,7 +2130,7 @@ isert_put_datain(struct iscsi_conn *conn, struct iscsi_cmd *cmd) } static int -isert_get_dataout(struct iscsi_conn *conn, struct iscsi_cmd *cmd, bool recovery) +isert_get_dataout(struct iscsit_conn *conn, struct iscsit_cmd *cmd, bool recovery) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); int ret; @@ -2147,7 +2148,7 @@ isert_get_dataout(struct iscsi_conn *conn, struct iscsi_cmd *cmd, bool recovery) } static int -isert_immediate_queue(struct iscsi_conn *conn, struct iscsi_cmd *cmd, int state) +isert_immediate_queue(struct iscsit_conn *conn, struct iscsit_cmd *cmd, int state) { struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd); int ret = 0; @@ -2172,7 +2173,7 @@ isert_immediate_queue(struct iscsi_conn *conn, struct iscsi_cmd *cmd, int state) } static int -isert_response_queue(struct iscsi_conn *conn, struct iscsi_cmd *cmd, int state) +isert_response_queue(struct iscsit_conn *conn, struct iscsit_cmd *cmd, int state) { struct isert_conn *isert_conn = conn->context; int ret; @@ -2332,7 +2333,7 @@ isert_rdma_accept(struct isert_conn *isert_conn) } static int -isert_get_login_rx(struct iscsi_conn *conn, struct iscsi_login *login) +isert_get_login_rx(struct iscsit_conn *conn, struct iscsi_login *login) { struct isert_conn *isert_conn = conn->context; int ret; @@ -2348,9 +2349,9 @@ isert_get_login_rx(struct iscsi_conn *conn, struct iscsi_login *login) /* * For login requests after the first PDU, isert_rx_login_req() will - * kick schedule_delayed_work(&conn->login_work) as the packet is - * received, which turns this callback from iscsi_target_do_login_rx() - * into a NOP. + * kick queue_delayed_work(isert_login_wq, &conn->login_work) as + * the packet is received, which turns this callback from + * iscsi_target_do_login_rx() into a NOP. */ if (!login->first_request) return 0; @@ -2368,7 +2369,7 @@ isert_get_login_rx(struct iscsi_conn *conn, struct iscsi_login *login) } static void -isert_set_conn_info(struct iscsi_np *np, struct iscsi_conn *conn, +isert_set_conn_info(struct iscsi_np *np, struct iscsit_conn *conn, struct isert_conn *isert_conn) { struct rdma_cm_id *cm_id = isert_conn->cm_id; @@ -2381,7 +2382,7 @@ isert_set_conn_info(struct iscsi_np *np, struct iscsi_conn *conn, } static int -isert_accept_np(struct iscsi_np *np, struct iscsi_conn *conn) +isert_accept_np(struct iscsi_np *np, struct iscsit_conn *conn) { struct isert_np *isert_np = np->np_context; struct isert_conn *isert_conn; @@ -2489,7 +2490,7 @@ static void isert_release_work(struct work_struct *work) static void isert_wait4logout(struct isert_conn *isert_conn) { - struct iscsi_conn *conn = isert_conn->conn; + struct iscsit_conn *conn = isert_conn->conn; isert_info("conn %p\n", isert_conn); @@ -2501,9 +2502,9 @@ isert_wait4logout(struct isert_conn *isert_conn) } static void -isert_wait4cmds(struct iscsi_conn *conn) +isert_wait4cmds(struct iscsit_conn *conn) { - isert_info("iscsi_conn %p\n", conn); + isert_info("iscsit_conn %p\n", conn); if (conn->sess) { target_stop_session(conn->sess->se_sess); @@ -2521,9 +2522,9 @@ isert_wait4cmds(struct iscsi_conn *conn) * before blocking on the target_wait_for_session_cmds */ static void -isert_put_unsol_pending_cmds(struct iscsi_conn *conn) +isert_put_unsol_pending_cmds(struct iscsit_conn *conn) { - struct iscsi_cmd *cmd, *tmp; + struct iscsit_cmd *cmd, *tmp; static LIST_HEAD(drop_cmd_list); spin_lock_bh(&conn->cmd_lock); @@ -2546,7 +2547,7 @@ isert_put_unsol_pending_cmds(struct iscsi_conn *conn) } } -static void isert_wait_conn(struct iscsi_conn *conn) +static void isert_wait_conn(struct iscsit_conn *conn) { struct isert_conn *isert_conn = conn->context; @@ -2564,7 +2565,7 @@ static void isert_wait_conn(struct iscsi_conn *conn) queue_work(isert_release_wq, &isert_conn->release_work); } -static void isert_free_conn(struct iscsi_conn *conn) +static void isert_free_conn(struct iscsit_conn *conn) { struct isert_conn *isert_conn = conn->context; @@ -2572,7 +2573,7 @@ static void isert_free_conn(struct iscsi_conn *conn) isert_put_conn(isert_conn); } -static void isert_get_rx_pdu(struct iscsi_conn *conn) +static void isert_get_rx_pdu(struct iscsit_conn *conn) { struct completion comp; @@ -2606,20 +2607,23 @@ static struct iscsit_transport iser_target_transport = { static int __init isert_init(void) { - int ret; + isert_login_wq = alloc_workqueue("isert_login_wq", 0, 0); + if (!isert_login_wq) { + isert_err("Unable to allocate isert_login_wq\n"); + return -ENOMEM; + } isert_comp_wq = alloc_workqueue("isert_comp_wq", WQ_UNBOUND | WQ_HIGHPRI, 0); if (!isert_comp_wq) { isert_err("Unable to allocate isert_comp_wq\n"); - return -ENOMEM; + goto destroy_login_wq; } isert_release_wq = alloc_workqueue("isert_release_wq", WQ_UNBOUND, WQ_UNBOUND_MAX_ACTIVE); if (!isert_release_wq) { isert_err("Unable to allocate isert_release_wq\n"); - ret = -ENOMEM; goto destroy_comp_wq; } @@ -2630,17 +2634,20 @@ static int __init isert_init(void) destroy_comp_wq: destroy_workqueue(isert_comp_wq); +destroy_login_wq: + destroy_workqueue(isert_login_wq); - return ret; + return -ENOMEM; } static void __exit isert_exit(void) { - flush_scheduled_work(); + flush_workqueue(isert_login_wq); destroy_workqueue(isert_release_wq); destroy_workqueue(isert_comp_wq); iscsit_unregister_transport(&iser_target_transport); isert_info("iSER_TARGET[0] - Released iser_target_transport\n"); + destroy_workqueue(isert_login_wq); } MODULE_DESCRIPTION("iSER-Target for mainline target infrastructure"); diff --git a/drivers/infiniband/ulp/isert/ib_isert.h b/drivers/infiniband/ulp/isert/ib_isert.h index ca8cfebe26ca..0b2dfd6e7e27 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.h +++ b/drivers/infiniband/ulp/isert/ib_isert.h @@ -146,7 +146,7 @@ struct isert_cmd { u64 pdu_buf_dma; u32 pdu_buf_len; struct isert_conn *conn; - struct iscsi_cmd *iscsi_cmd; + struct iscsit_cmd *iscsit_cmd; struct iser_tx_desc tx_desc; struct iser_rx_desc *rx_desc; struct rdma_rw_ctx rw; @@ -173,7 +173,7 @@ struct isert_conn { u64 login_rsp_dma; struct iser_rx_desc *rx_descs; struct ib_recv_wr rx_wr[ISERT_QP_MAX_RECV_DTOS]; - struct iscsi_conn *conn; + struct iscsit_conn *conn; struct list_head node; struct completion login_comp; struct completion login_req_comp; diff --git a/drivers/infiniband/ulp/opa_vnic/opa_vnic_ethtool.c b/drivers/infiniband/ulp/opa_vnic/opa_vnic_ethtool.c index 42d557dff19d..29b3d8fce3f5 100644 --- a/drivers/infiniband/ulp/opa_vnic/opa_vnic_ethtool.c +++ b/drivers/infiniband/ulp/opa_vnic/opa_vnic_ethtool.c @@ -124,8 +124,8 @@ static struct vnic_stats vnic_gstrings_stats[] = { static void vnic_get_drvinfo(struct net_device *netdev, struct ethtool_drvinfo *drvinfo) { - strlcpy(drvinfo->driver, opa_vnic_driver_name, sizeof(drvinfo->driver)); - strlcpy(drvinfo->bus_info, dev_name(netdev->dev.parent), + strscpy(drvinfo->driver, opa_vnic_driver_name, sizeof(drvinfo->driver)); + strscpy(drvinfo->bus_info, dev_name(netdev->dev.parent), sizeof(drvinfo->bus_info)); } diff --git a/drivers/infiniband/ulp/opa_vnic/opa_vnic_netdev.c b/drivers/infiniband/ulp/opa_vnic/opa_vnic_netdev.c index aeff68f582d3..071f35711468 100644 --- a/drivers/infiniband/ulp/opa_vnic/opa_vnic_netdev.c +++ b/drivers/infiniband/ulp/opa_vnic/opa_vnic_netdev.c @@ -50,7 +50,6 @@ * netdev functionality. */ -#include <linux/module.h> #include <linux/if_vlan.h> #include <linux/crc32.h> diff --git a/drivers/infiniband/ulp/rtrs/Makefile b/drivers/infiniband/ulp/rtrs/Makefile index 3898509be270..5227e7788e1f 100644 --- a/drivers/infiniband/ulp/rtrs/Makefile +++ b/drivers/infiniband/ulp/rtrs/Makefile @@ -1,12 +1,18 @@ # SPDX-License-Identifier: GPL-2.0-or-later +CFLAGS_rtrs-clt-trace.o = -I$(src) + rtrs-client-y := rtrs-clt.o \ rtrs-clt-stats.o \ - rtrs-clt-sysfs.o + rtrs-clt-sysfs.o \ + rtrs-clt-trace.o + +CFLAGS_rtrs-srv-trace.o = -I$(src) rtrs-server-y := rtrs-srv.o \ rtrs-srv-stats.o \ - rtrs-srv-sysfs.o + rtrs-srv-sysfs.o \ + rtrs-srv-trace.o rtrs-core-y := rtrs.o diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-stats.c b/drivers/infiniband/ulp/rtrs/rtrs-clt-stats.c index 76e4352fe3f6..1e6ffafa2db3 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-clt-stats.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-stats.c @@ -13,8 +13,8 @@ void rtrs_clt_update_wc_stats(struct rtrs_clt_con *con) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); - struct rtrs_clt_stats *stats = sess->stats; + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); + struct rtrs_clt_stats *stats = clt_path->stats; struct rtrs_clt_stats_pcpu *s; int cpu; @@ -32,11 +32,7 @@ void rtrs_clt_update_wc_stats(struct rtrs_clt_con *con) void rtrs_clt_inc_failover_cnt(struct rtrs_clt_stats *stats) { - struct rtrs_clt_stats_pcpu *s; - - s = get_cpu_ptr(stats->pcpu_stats); - s->rdma.failover_cnt++; - put_cpu_ptr(stats->pcpu_stats); + this_cpu_inc(stats->pcpu_stats->rdma.failover_cnt); } int rtrs_clt_stats_migration_from_cnt_to_str(struct rtrs_clt_stats *stats, char *buf) @@ -169,19 +165,15 @@ int rtrs_clt_reset_all_stats(struct rtrs_clt_stats *s, bool enable) static inline void rtrs_clt_update_rdma_stats(struct rtrs_clt_stats *stats, size_t size, int d) { - struct rtrs_clt_stats_pcpu *s; - - s = get_cpu_ptr(stats->pcpu_stats); - s->rdma.dir[d].cnt++; - s->rdma.dir[d].size_total += size; - put_cpu_ptr(stats->pcpu_stats); + this_cpu_inc(stats->pcpu_stats->rdma.dir[d].cnt); + this_cpu_add(stats->pcpu_stats->rdma.dir[d].size_total, size); } void rtrs_clt_update_all_stats(struct rtrs_clt_io_req *req, int dir) { struct rtrs_clt_con *con = req->con; - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); - struct rtrs_clt_stats *stats = sess->stats; + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); + struct rtrs_clt_stats *stats = clt_path->stats; unsigned int len; len = req->usr_len + req->data_len; diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c b/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c index 0e69180c3771..d3c436ead694 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c @@ -16,21 +16,21 @@ #define MIN_MAX_RECONN_ATT -1 #define MAX_MAX_RECONN_ATT 9999 -static void rtrs_clt_sess_release(struct kobject *kobj) +static void rtrs_clt_path_release(struct kobject *kobj) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); - free_sess(sess); + free_path(clt_path); } static struct kobj_type ktype_sess = { .sysfs_ops = &kobj_sysfs_ops, - .release = rtrs_clt_sess_release + .release = rtrs_clt_path_release }; -static void rtrs_clt_sess_stats_release(struct kobject *kobj) +static void rtrs_clt_path_stats_release(struct kobject *kobj) { struct rtrs_clt_stats *stats; @@ -43,14 +43,15 @@ static void rtrs_clt_sess_stats_release(struct kobject *kobj) static struct kobj_type ktype_stats = { .sysfs_ops = &kobj_sysfs_ops, - .release = rtrs_clt_sess_stats_release, + .release = rtrs_clt_path_stats_release, }; static ssize_t max_reconnect_attempts_show(struct device *dev, struct device_attribute *attr, char *page) { - struct rtrs_clt *clt = container_of(dev, struct rtrs_clt, dev); + struct rtrs_clt_sess *clt = container_of(dev, struct rtrs_clt_sess, + dev); return sysfs_emit(page, "%d\n", rtrs_clt_get_max_reconnect_attempts(clt)); @@ -63,7 +64,8 @@ static ssize_t max_reconnect_attempts_store(struct device *dev, { int value; int ret; - struct rtrs_clt *clt = container_of(dev, struct rtrs_clt, dev); + struct rtrs_clt_sess *clt = container_of(dev, struct rtrs_clt_sess, + dev); ret = kstrtoint(buf, 10, &value); if (ret) { @@ -90,9 +92,9 @@ static ssize_t mpath_policy_show(struct device *dev, struct device_attribute *attr, char *page) { - struct rtrs_clt *clt; + struct rtrs_clt_sess *clt; - clt = container_of(dev, struct rtrs_clt, dev); + clt = container_of(dev, struct rtrs_clt_sess, dev); switch (clt->mp_policy) { case MP_POLICY_RR: @@ -114,12 +116,12 @@ static ssize_t mpath_policy_store(struct device *dev, const char *buf, size_t count) { - struct rtrs_clt *clt; + struct rtrs_clt_sess *clt; int value; int ret; size_t len = 0; - clt = container_of(dev, struct rtrs_clt, dev); + clt = container_of(dev, struct rtrs_clt_sess, dev); ret = kstrtoint(buf, 10, &value); if (!ret && (value == MP_POLICY_RR || @@ -154,8 +156,7 @@ static DEVICE_ATTR_RW(mpath_policy); static ssize_t add_path_show(struct device *dev, struct device_attribute *attr, char *page) { - return sysfs_emit( - page, + return sysfs_emit(page, "Usage: echo [<source addr>@]<destination addr> > %s\n\n*addr ::= [ ip:<ipv4|ipv6> | gid:<gid> ]\n", attr->attr.name); } @@ -169,12 +170,12 @@ static ssize_t add_path_store(struct device *dev, .src = &srcaddr, .dst = &dstaddr }; - struct rtrs_clt *clt; + struct rtrs_clt_sess *clt; const char *nl; size_t len; int err; - clt = container_of(dev, struct rtrs_clt, dev); + clt = container_of(dev, struct rtrs_clt_sess, dev); nl = strchr(buf, '\n'); if (nl) @@ -197,10 +198,10 @@ static DEVICE_ATTR_RW(add_path); static ssize_t rtrs_clt_state_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); - if (sess->state == RTRS_CLT_CONNECTED) + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); + if (clt_path->state == RTRS_CLT_CONNECTED) return sysfs_emit(page, "connected\n"); return sysfs_emit(page, "disconnected\n"); @@ -219,16 +220,16 @@ static ssize_t rtrs_clt_reconnect_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; int ret; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); if (!sysfs_streq(buf, "1")) { - rtrs_err(sess->clt, "%s: unknown value: '%s'\n", + rtrs_err(clt_path->clt, "%s: unknown value: '%s'\n", attr->attr.name, buf); return -EINVAL; } - ret = rtrs_clt_reconnect_from_sysfs(sess); + ret = rtrs_clt_reconnect_from_sysfs(clt_path); if (ret) return ret; @@ -249,15 +250,15 @@ static ssize_t rtrs_clt_disconnect_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); if (!sysfs_streq(buf, "1")) { - rtrs_err(sess->clt, "%s: unknown value: '%s'\n", + rtrs_err(clt_path->clt, "%s: unknown value: '%s'\n", attr->attr.name, buf); return -EINVAL; } - rtrs_clt_close_conns(sess, true); + rtrs_clt_close_conns(clt_path, true); return count; } @@ -276,16 +277,16 @@ static ssize_t rtrs_clt_remove_path_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; int ret; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); if (!sysfs_streq(buf, "1")) { - rtrs_err(sess->clt, "%s: unknown value: '%s'\n", + rtrs_err(clt_path->clt, "%s: unknown value: '%s'\n", attr->attr.name, buf); return -EINVAL; } - ret = rtrs_clt_remove_path_from_sysfs(sess, &attr->attr); + ret = rtrs_clt_remove_path_from_sysfs(clt_path, &attr->attr); if (ret) return ret; @@ -333,11 +334,11 @@ static ssize_t rtrs_clt_hca_port_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = container_of(kobj, typeof(*sess), kobj); + clt_path = container_of(kobj, typeof(*clt_path), kobj); - return sysfs_emit(page, "%u\n", sess->hca_port); + return sysfs_emit(page, "%u\n", clt_path->hca_port); } static struct kobj_attribute rtrs_clt_hca_port_attr = @@ -347,11 +348,11 @@ static ssize_t rtrs_clt_hca_name_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); - return sysfs_emit(page, "%s\n", sess->hca_name); + return sysfs_emit(page, "%s\n", clt_path->hca_name); } static struct kobj_attribute rtrs_clt_hca_name_attr = @@ -361,12 +362,12 @@ static ssize_t rtrs_clt_cur_latency_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); return sysfs_emit(page, "%lld ns\n", - ktime_to_ns(sess->s.hb_cur_latency)); + ktime_to_ns(clt_path->s.hb_cur_latency)); } static struct kobj_attribute rtrs_clt_cur_latency_attr = @@ -376,11 +377,11 @@ static ssize_t rtrs_clt_src_addr_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; int len; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); - len = sockaddr_to_str((struct sockaddr *)&sess->s.src_addr, page, + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); + len = sockaddr_to_str((struct sockaddr *)&clt_path->s.src_addr, page, PAGE_SIZE); len += sysfs_emit_at(page, len, "\n"); return len; @@ -393,11 +394,11 @@ static ssize_t rtrs_clt_dst_addr_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; int len; - sess = container_of(kobj, struct rtrs_clt_sess, kobj); - len = sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, page, + clt_path = container_of(kobj, struct rtrs_clt_path, kobj); + len = sockaddr_to_str((struct sockaddr *)&clt_path->s.dst_addr, page, PAGE_SIZE); len += sysfs_emit_at(page, len, "\n"); return len; @@ -406,7 +407,7 @@ static ssize_t rtrs_clt_dst_addr_show(struct kobject *kobj, static struct kobj_attribute rtrs_clt_dst_addr_attr = __ATTR(dst_addr, 0444, rtrs_clt_dst_addr_show, NULL); -static struct attribute *rtrs_clt_sess_attrs[] = { +static struct attribute *rtrs_clt_path_attrs[] = { &rtrs_clt_hca_name_attr.attr, &rtrs_clt_hca_port_attr.attr, &rtrs_clt_src_addr_attr.attr, @@ -419,42 +420,43 @@ static struct attribute *rtrs_clt_sess_attrs[] = { NULL, }; -static const struct attribute_group rtrs_clt_sess_attr_group = { - .attrs = rtrs_clt_sess_attrs, +static const struct attribute_group rtrs_clt_path_attr_group = { + .attrs = rtrs_clt_path_attrs, }; -int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess) +int rtrs_clt_create_path_files(struct rtrs_clt_path *clt_path) { - struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_sess *clt = clt_path->clt; char str[NAME_MAX]; int err; struct rtrs_addr path = { - .src = &sess->s.src_addr, - .dst = &sess->s.dst_addr, + .src = &clt_path->s.src_addr, + .dst = &clt_path->s.dst_addr, }; rtrs_addr_to_str(&path, str, sizeof(str)); - err = kobject_init_and_add(&sess->kobj, &ktype_sess, clt->kobj_paths, + err = kobject_init_and_add(&clt_path->kobj, &ktype_sess, + clt->kobj_paths, "%s", str); if (err) { pr_err("kobject_init_and_add: %d\n", err); - kobject_put(&sess->kobj); + kobject_put(&clt_path->kobj); return err; } - err = sysfs_create_group(&sess->kobj, &rtrs_clt_sess_attr_group); + err = sysfs_create_group(&clt_path->kobj, &rtrs_clt_path_attr_group); if (err) { pr_err("sysfs_create_group(): %d\n", err); goto put_kobj; } - err = kobject_init_and_add(&sess->stats->kobj_stats, &ktype_stats, - &sess->kobj, "stats"); + err = kobject_init_and_add(&clt_path->stats->kobj_stats, &ktype_stats, + &clt_path->kobj, "stats"); if (err) { pr_err("kobject_init_and_add: %d\n", err); - kobject_put(&sess->stats->kobj_stats); + kobject_put(&clt_path->stats->kobj_stats); goto remove_group; } - err = sysfs_create_group(&sess->stats->kobj_stats, + err = sysfs_create_group(&clt_path->stats->kobj_stats, &rtrs_clt_stats_attr_group); if (err) { pr_err("failed to create stats sysfs group, err: %d\n", err); @@ -464,25 +466,25 @@ int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess) return 0; put_kobj_stats: - kobject_del(&sess->stats->kobj_stats); - kobject_put(&sess->stats->kobj_stats); + kobject_del(&clt_path->stats->kobj_stats); + kobject_put(&clt_path->stats->kobj_stats); remove_group: - sysfs_remove_group(&sess->kobj, &rtrs_clt_sess_attr_group); + sysfs_remove_group(&clt_path->kobj, &rtrs_clt_path_attr_group); put_kobj: - kobject_del(&sess->kobj); - kobject_put(&sess->kobj); + kobject_del(&clt_path->kobj); + kobject_put(&clt_path->kobj); return err; } -void rtrs_clt_destroy_sess_files(struct rtrs_clt_sess *sess, +void rtrs_clt_destroy_path_files(struct rtrs_clt_path *clt_path, const struct attribute *sysfs_self) { - kobject_del(&sess->stats->kobj_stats); - kobject_put(&sess->stats->kobj_stats); + kobject_del(&clt_path->stats->kobj_stats); + kobject_put(&clt_path->stats->kobj_stats); if (sysfs_self) - sysfs_remove_file_self(&sess->kobj, sysfs_self); - kobject_del(&sess->kobj); + sysfs_remove_file_self(&clt_path->kobj, sysfs_self); + kobject_del(&clt_path->kobj); } static struct attribute *rtrs_clt_attrs[] = { @@ -496,12 +498,12 @@ static const struct attribute_group rtrs_clt_attr_group = { .attrs = rtrs_clt_attrs, }; -int rtrs_clt_create_sysfs_root_files(struct rtrs_clt *clt) +int rtrs_clt_create_sysfs_root_files(struct rtrs_clt_sess *clt) { return sysfs_create_group(&clt->dev.kobj, &rtrs_clt_attr_group); } -void rtrs_clt_destroy_sysfs_root(struct rtrs_clt *clt) +void rtrs_clt_destroy_sysfs_root(struct rtrs_clt_sess *clt) { sysfs_remove_group(&clt->dev.kobj, &rtrs_clt_attr_group); diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.c b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.c new file mode 100644 index 000000000000..f14fa1f36ce8 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.c @@ -0,0 +1,15 @@ +// SPDX-License-Identifier: GPL-2.0+ +/* + * RDMA Network Block Driver + * + * Copyright (c) 2022 1&1 IONOS SE. All rights reserved. + */ +#include "rtrs.h" +#include "rtrs-clt.h" + +/* + * We include this last to have the helpers above available for the trace + * event implementations. + */ +#define CREATE_TRACE_POINTS +#include "rtrs-clt-trace.h" diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h new file mode 100644 index 000000000000..7738e2676855 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-trace.h @@ -0,0 +1,86 @@ +/* SPDX-License-Identifier: GPL-2.0+ */ +/* + * RDMA Network Block Driver + * + * Copyright (c) 2022 1&1 IONOS SE. All rights reserved. + */ +#undef TRACE_SYSTEM +#define TRACE_SYSTEM rtrs_clt + +#if !defined(_TRACE_RTRS_CLT_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_RTRS_CLT_H + +#include <linux/tracepoint.h> + +struct rtrs_clt_path; +struct rtrs_clt_sess; + +TRACE_DEFINE_ENUM(RTRS_CLT_CONNECTING); +TRACE_DEFINE_ENUM(RTRS_CLT_CONNECTING_ERR); +TRACE_DEFINE_ENUM(RTRS_CLT_RECONNECTING); +TRACE_DEFINE_ENUM(RTRS_CLT_CONNECTED); +TRACE_DEFINE_ENUM(RTRS_CLT_CLOSING); +TRACE_DEFINE_ENUM(RTRS_CLT_CLOSED); +TRACE_DEFINE_ENUM(RTRS_CLT_DEAD); + +#define show_rtrs_clt_state(x) \ + __print_symbolic(x, \ + { RTRS_CLT_CONNECTING, "CONNECTING" }, \ + { RTRS_CLT_CONNECTING_ERR, "CONNECTING_ERR" }, \ + { RTRS_CLT_RECONNECTING, "RECONNECTING" }, \ + { RTRS_CLT_CONNECTED, "CONNECTED" }, \ + { RTRS_CLT_CLOSING, "CLOSING" }, \ + { RTRS_CLT_CLOSED, "CLOSED" }, \ + { RTRS_CLT_DEAD, "DEAD" }) + +DECLARE_EVENT_CLASS(rtrs_clt_conn_class, + TP_PROTO(struct rtrs_clt_path *clt_path), + + TP_ARGS(clt_path), + + TP_STRUCT__entry( + __field(int, state) + __field(int, reconnect_attempts) + __field(int, max_reconnect_attempts) + __field(int, fail_cnt) + __field(int, success_cnt) + __array(char, sessname, NAME_MAX) + ), + + TP_fast_assign( + struct rtrs_clt_sess *clt = clt_path->clt; + + __entry->state = clt_path->state; + __entry->reconnect_attempts = clt_path->reconnect_attempts; + __entry->max_reconnect_attempts = clt->max_reconnect_attempts; + __entry->fail_cnt = clt_path->stats->reconnects.fail_cnt; + __entry->success_cnt = clt_path->stats->reconnects.successful_cnt; + memcpy(__entry->sessname, kobject_name(&clt_path->kobj), NAME_MAX); + ), + + TP_printk("RTRS-CLT: sess='%s' state=%s attempts='%d' max-attempts='%d' fail='%d' success='%d'", + __entry->sessname, + show_rtrs_clt_state(__entry->state), + __entry->reconnect_attempts, + __entry->max_reconnect_attempts, + __entry->fail_cnt, + __entry->success_cnt + ) +); + +#define DEFINE_CLT_CONN_EVENT(name) \ +DEFINE_EVENT(rtrs_clt_conn_class, rtrs_##name, \ + TP_PROTO(struct rtrs_clt_path *clt_path), \ + TP_ARGS(clt_path)) + +DEFINE_CLT_CONN_EVENT(clt_reconnect_work); +DEFINE_CLT_CONN_EVENT(clt_close_conns); +DEFINE_CLT_CONN_EVENT(rdma_error_recovery); + +#endif /* _TRACE_RTRS_CLT_H */ + +#undef TRACE_INCLUDE_PATH +#define TRACE_INCLUDE_PATH . +#define TRACE_INCLUDE_FILE rtrs-clt-trace +#include <trace/define_trace.h> + diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt.c b/drivers/infiniband/ulp/rtrs/rtrs-clt.c index 15c0077dd27e..8546b8816524 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-clt.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt.c @@ -16,6 +16,7 @@ #include "rtrs-clt.h" #include "rtrs-log.h" +#include "rtrs-clt-trace.h" #define RTRS_CONNECT_TIMEOUT_MS 30000 /* @@ -46,21 +47,24 @@ static struct rtrs_rdma_dev_pd dev_pd = { static struct workqueue_struct *rtrs_wq; static struct class *rtrs_clt_dev_class; -static inline bool rtrs_clt_is_connected(const struct rtrs_clt *clt) +static inline bool rtrs_clt_is_connected(const struct rtrs_clt_sess *clt) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; bool connected = false; rcu_read_lock(); - list_for_each_entry_rcu(sess, &clt->paths_list, s.entry) - connected |= READ_ONCE(sess->state) == RTRS_CLT_CONNECTED; + list_for_each_entry_rcu(clt_path, &clt->paths_list, s.entry) + if (READ_ONCE(clt_path->state) == RTRS_CLT_CONNECTED) { + connected = true; + break; + } rcu_read_unlock(); return connected; } static struct rtrs_permit * -__rtrs_get_permit(struct rtrs_clt *clt, enum rtrs_clt_con_type con_type) +__rtrs_get_permit(struct rtrs_clt_sess *clt, enum rtrs_clt_con_type con_type) { size_t max_depth = clt->queue_depth; struct rtrs_permit *permit; @@ -87,7 +91,7 @@ __rtrs_get_permit(struct rtrs_clt *clt, enum rtrs_clt_con_type con_type) return permit; } -static inline void __rtrs_put_permit(struct rtrs_clt *clt, +static inline void __rtrs_put_permit(struct rtrs_clt_sess *clt, struct rtrs_permit *permit) { clear_bit_unlock(permit->mem_id, clt->permits_map); @@ -107,7 +111,7 @@ static inline void __rtrs_put_permit(struct rtrs_clt *clt, * Context: * Can sleep if @wait == RTRS_PERMIT_WAIT */ -struct rtrs_permit *rtrs_clt_get_permit(struct rtrs_clt *clt, +struct rtrs_permit *rtrs_clt_get_permit(struct rtrs_clt_sess *clt, enum rtrs_clt_con_type con_type, enum wait_type can_wait) { @@ -142,7 +146,8 @@ EXPORT_SYMBOL(rtrs_clt_get_permit); * Context: * Does not matter */ -void rtrs_clt_put_permit(struct rtrs_clt *clt, struct rtrs_permit *permit) +void rtrs_clt_put_permit(struct rtrs_clt_sess *clt, + struct rtrs_permit *permit) { if (WARN_ON(!test_bit(permit->mem_id, clt->permits_map))) return; @@ -163,29 +168,29 @@ EXPORT_SYMBOL(rtrs_clt_put_permit); /** * rtrs_permit_to_clt_con() - returns RDMA connection pointer by the permit - * @sess: client session pointer + * @clt_path: client path pointer * @permit: permit for the allocation of the RDMA buffer * Note: * IO connection starts from 1. * 0 connection is for user messages. */ static -struct rtrs_clt_con *rtrs_permit_to_clt_con(struct rtrs_clt_sess *sess, +struct rtrs_clt_con *rtrs_permit_to_clt_con(struct rtrs_clt_path *clt_path, struct rtrs_permit *permit) { int id = 0; if (permit->con_type == RTRS_IO_CON) - id = (permit->cpu_id % (sess->s.irq_con_num - 1)) + 1; + id = (permit->cpu_id % (clt_path->s.irq_con_num - 1)) + 1; - return to_clt_con(sess->s.con[id]); + return to_clt_con(clt_path->s.con[id]); } /** * rtrs_clt_change_state() - change the session state through session state * machine. * - * @sess: client session to change the state of. + * @clt_path: client path to change the state of. * @new_state: state to change to. * * returns true if sess's state is changed to new state, otherwise return false. @@ -193,15 +198,15 @@ struct rtrs_clt_con *rtrs_permit_to_clt_con(struct rtrs_clt_sess *sess, * Locks: * state_wq lock must be hold. */ -static bool rtrs_clt_change_state(struct rtrs_clt_sess *sess, +static bool rtrs_clt_change_state(struct rtrs_clt_path *clt_path, enum rtrs_clt_state new_state) { enum rtrs_clt_state old_state; bool changed = false; - lockdep_assert_held(&sess->state_wq.lock); + lockdep_assert_held(&clt_path->state_wq.lock); - old_state = sess->state; + old_state = clt_path->state; switch (new_state) { case RTRS_CLT_CONNECTING: switch (old_state) { @@ -275,51 +280,45 @@ static bool rtrs_clt_change_state(struct rtrs_clt_sess *sess, break; } if (changed) { - sess->state = new_state; - wake_up_locked(&sess->state_wq); + clt_path->state = new_state; + wake_up_locked(&clt_path->state_wq); } return changed; } -static bool rtrs_clt_change_state_from_to(struct rtrs_clt_sess *sess, +static bool rtrs_clt_change_state_from_to(struct rtrs_clt_path *clt_path, enum rtrs_clt_state old_state, enum rtrs_clt_state new_state) { bool changed = false; - spin_lock_irq(&sess->state_wq.lock); - if (sess->state == old_state) - changed = rtrs_clt_change_state(sess, new_state); - spin_unlock_irq(&sess->state_wq.lock); + spin_lock_irq(&clt_path->state_wq.lock); + if (clt_path->state == old_state) + changed = rtrs_clt_change_state(clt_path, new_state); + spin_unlock_irq(&clt_path->state_wq.lock); return changed; } +static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_path *clt_path); static void rtrs_rdma_error_recovery(struct rtrs_clt_con *con) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); + + trace_rtrs_rdma_error_recovery(clt_path); - if (rtrs_clt_change_state_from_to(sess, + if (rtrs_clt_change_state_from_to(clt_path, RTRS_CLT_CONNECTED, RTRS_CLT_RECONNECTING)) { - struct rtrs_clt *clt = sess->clt; - unsigned int delay_ms; - - /* - * Normal scenario, reconnect if we were successfully connected - */ - delay_ms = clt->reconnect_delay_sec * 1000; - queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, - msecs_to_jiffies(delay_ms + - prandom_u32() % RTRS_RECONNECT_SEED)); + queue_work(rtrs_wq, &clt_path->err_recovery_work); } else { /* * Error can happen just on establishing new connection, * so notify waiter with error state, waiter is responsible * for cleaning the rest and reconnect if needed. */ - rtrs_clt_change_state_from_to(sess, + rtrs_clt_change_state_from_to(clt_path, RTRS_CLT_CONNECTING, RTRS_CLT_CONNECTING_ERR); } @@ -330,7 +329,7 @@ static void rtrs_clt_fast_reg_done(struct ib_cq *cq, struct ib_wc *wc) struct rtrs_clt_con *con = to_clt_con(wc->qp->qp_context); if (wc->status != IB_WC_SUCCESS) { - rtrs_err(con->c.sess, "Failed IB_WR_REG_MR: %s\n", + rtrs_err(con->c.path, "Failed IB_WR_REG_MR: %s\n", ib_wc_status_msg(wc->status)); rtrs_rdma_error_recovery(con); } @@ -350,7 +349,7 @@ static void rtrs_clt_inv_rkey_done(struct ib_cq *cq, struct ib_wc *wc) struct rtrs_clt_con *con = to_clt_con(wc->qp->qp_context); if (wc->status != IB_WC_SUCCESS) { - rtrs_err(con->c.sess, "Failed IB_WR_LOCAL_INV: %s\n", + rtrs_err(con->c.path, "Failed IB_WR_LOCAL_INV: %s\n", ib_wc_status_msg(wc->status)); rtrs_rdma_error_recovery(con); } @@ -380,14 +379,14 @@ static void complete_rdma_req(struct rtrs_clt_io_req *req, int errno, bool notify, bool can_wait) { struct rtrs_clt_con *con = req->con; - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; int err; if (WARN_ON(!req->in_use)) return; if (WARN_ON(!req->con)) return; - sess = to_clt_sess(con->c.sess); + clt_path = to_clt_path(con->c.path); if (req->sg_cnt) { if (req->dir == DMA_FROM_DEVICE && req->need_inv) { @@ -417,7 +416,7 @@ static void complete_rdma_req(struct rtrs_clt_io_req *req, int errno, refcount_inc(&req->ref); err = rtrs_inv_rkey(req); if (err) { - rtrs_err(con->c.sess, "Send INV WR key=%#x: %d\n", + rtrs_err(con->c.path, "Send INV WR key=%#x: %d\n", req->mr->rkey, err); } else if (can_wait) { wait_for_completion(&req->inv_comp); @@ -433,21 +432,21 @@ static void complete_rdma_req(struct rtrs_clt_io_req *req, int errno, if (!refcount_dec_and_test(&req->ref)) return; } - ib_dma_unmap_sg(sess->s.dev->ib_dev, req->sglist, + ib_dma_unmap_sg(clt_path->s.dev->ib_dev, req->sglist, req->sg_cnt, req->dir); } if (!refcount_dec_and_test(&req->ref)) return; if (req->mp_policy == MP_POLICY_MIN_INFLIGHT) - atomic_dec(&sess->stats->inflight); + atomic_dec(&clt_path->stats->inflight); req->in_use = false; req->con = NULL; if (errno) { - rtrs_err_rl(con->c.sess, "IO request failed: error=%d path=%s [%s:%u] notify=%d\n", - errno, kobject_name(&sess->kobj), sess->hca_name, - sess->hca_port, notify); + rtrs_err_rl(con->c.path, "IO request failed: error=%d path=%s [%s:%u] notify=%d\n", + errno, kobject_name(&clt_path->kobj), clt_path->hca_name, + clt_path->hca_port, notify); } if (notify) @@ -459,12 +458,12 @@ static int rtrs_post_send_rdma(struct rtrs_clt_con *con, struct rtrs_rbuf *rbuf, u32 off, u32 imm, struct ib_send_wr *wr) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); enum ib_send_flags flags; struct ib_sge sge; if (!req->sg_size) { - rtrs_wrn(con->c.sess, + rtrs_wrn(con->c.path, "Doing RDMA Write failed, no data supplied\n"); return -EINVAL; } @@ -472,16 +471,17 @@ static int rtrs_post_send_rdma(struct rtrs_clt_con *con, /* user data and user message in the first list element */ sge.addr = req->iu->dma_addr; sge.length = req->sg_size; - sge.lkey = sess->s.dev->ib_pd->local_dma_lkey; + sge.lkey = clt_path->s.dev->ib_pd->local_dma_lkey; /* * From time to time we have to post signalled sends, * or send queue will fill up and only QP reset can help. */ - flags = atomic_inc_return(&con->c.wr_cnt) % sess->s.signal_interval ? + flags = atomic_inc_return(&con->c.wr_cnt) % clt_path->s.signal_interval ? 0 : IB_SEND_SIGNALED; - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, req->iu->dma_addr, + ib_dma_sync_single_for_device(clt_path->s.dev->ib_dev, + req->iu->dma_addr, req->sg_size, DMA_TO_DEVICE); return rtrs_iu_post_rdma_write_imm(&con->c, req->iu, &sge, 1, @@ -489,15 +489,15 @@ static int rtrs_post_send_rdma(struct rtrs_clt_con *con, imm, flags, wr, NULL); } -static void process_io_rsp(struct rtrs_clt_sess *sess, u32 msg_id, +static void process_io_rsp(struct rtrs_clt_path *clt_path, u32 msg_id, s16 errno, bool w_inval) { struct rtrs_clt_io_req *req; - if (WARN_ON(msg_id >= sess->queue_depth)) + if (WARN_ON(msg_id >= clt_path->queue_depth)) return; - req = &sess->reqs[msg_id]; + req = &clt_path->reqs[msg_id]; /* Drop need_inv if server responded with send with invalidation */ req->need_inv &= !w_inval; complete_rdma_req(req, errno, true, false); @@ -507,21 +507,21 @@ static void rtrs_clt_recv_done(struct rtrs_clt_con *con, struct ib_wc *wc) { struct rtrs_iu *iu; int err; - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); - WARN_ON((sess->flags & RTRS_MSG_NEW_RKEY_F) == 0); + WARN_ON((clt_path->flags & RTRS_MSG_NEW_RKEY_F) == 0); iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); err = rtrs_iu_post_recv(&con->c, iu); if (err) { - rtrs_err(con->c.sess, "post iu failed %d\n", err); + rtrs_err(con->c.path, "post iu failed %d\n", err); rtrs_rdma_error_recovery(con); } } static void rtrs_clt_rkey_rsp_done(struct rtrs_clt_con *con, struct ib_wc *wc) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); struct rtrs_msg_rkey_rsp *msg; u32 imm_type, imm_payload; bool w_inval = false; @@ -529,25 +529,26 @@ static void rtrs_clt_rkey_rsp_done(struct rtrs_clt_con *con, struct ib_wc *wc) u32 buf_id; int err; - WARN_ON((sess->flags & RTRS_MSG_NEW_RKEY_F) == 0); + WARN_ON((clt_path->flags & RTRS_MSG_NEW_RKEY_F) == 0); iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); if (wc->byte_len < sizeof(*msg)) { - rtrs_err(con->c.sess, "rkey response is malformed: size %d\n", + rtrs_err(con->c.path, "rkey response is malformed: size %d\n", wc->byte_len); goto out; } - ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr, + ib_dma_sync_single_for_cpu(clt_path->s.dev->ib_dev, iu->dma_addr, iu->size, DMA_FROM_DEVICE); msg = iu->buf; if (le16_to_cpu(msg->type) != RTRS_MSG_RKEY_RSP) { - rtrs_err(sess->clt, "rkey response is malformed: type %d\n", + rtrs_err(clt_path->clt, + "rkey response is malformed: type %d\n", le16_to_cpu(msg->type)); goto out; } buf_id = le16_to_cpu(msg->buf_id); - if (WARN_ON(buf_id >= sess->queue_depth)) + if (WARN_ON(buf_id >= clt_path->queue_depth)) goto out; rtrs_from_imm(be32_to_cpu(wc->ex.imm_data), &imm_type, &imm_payload); @@ -560,10 +561,10 @@ static void rtrs_clt_rkey_rsp_done(struct rtrs_clt_con *con, struct ib_wc *wc) if (WARN_ON(buf_id != msg_id)) goto out; - sess->rbufs[buf_id].rkey = le32_to_cpu(msg->rkey); - process_io_rsp(sess, msg_id, err, w_inval); + clt_path->rbufs[buf_id].rkey = le32_to_cpu(msg->rkey); + process_io_rsp(clt_path, msg_id, err, w_inval); } - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, iu->dma_addr, + ib_dma_sync_single_for_device(clt_path->s.dev->ib_dev, iu->dma_addr, iu->size, DMA_FROM_DEVICE); return rtrs_clt_recv_done(con, wc); out: @@ -600,14 +601,14 @@ static int rtrs_post_recv_empty_x2(struct rtrs_con *con, struct ib_cqe *cqe) static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) { struct rtrs_clt_con *con = to_clt_con(wc->qp->qp_context); - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); u32 imm_type, imm_payload; bool w_inval = false; int err; if (wc->status != IB_WC_SUCCESS) { if (wc->status != IB_WC_WR_FLUSH_ERR) { - rtrs_err(sess->clt, "RDMA failed: %s\n", + rtrs_err(clt_path->clt, "RDMA failed: %s\n", ib_wc_status_msg(wc->status)); rtrs_rdma_error_recovery(con); } @@ -632,21 +633,21 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) w_inval = (imm_type == RTRS_IO_RSP_W_INV_IMM); rtrs_from_io_rsp_imm(imm_payload, &msg_id, &err); - process_io_rsp(sess, msg_id, err, w_inval); + process_io_rsp(clt_path, msg_id, err, w_inval); } else if (imm_type == RTRS_HB_MSG_IMM) { WARN_ON(con->c.cid); - rtrs_send_hb_ack(&sess->s); - if (sess->flags & RTRS_MSG_NEW_RKEY_F) + rtrs_send_hb_ack(&clt_path->s); + if (clt_path->flags & RTRS_MSG_NEW_RKEY_F) return rtrs_clt_recv_done(con, wc); } else if (imm_type == RTRS_HB_ACK_IMM) { WARN_ON(con->c.cid); - sess->s.hb_missed_cnt = 0; - sess->s.hb_cur_latency = - ktime_sub(ktime_get(), sess->s.hb_last_sent); - if (sess->flags & RTRS_MSG_NEW_RKEY_F) + clt_path->s.hb_missed_cnt = 0; + clt_path->s.hb_cur_latency = + ktime_sub(ktime_get(), clt_path->s.hb_last_sent); + if (clt_path->flags & RTRS_MSG_NEW_RKEY_F) return rtrs_clt_recv_done(con, wc); } else { - rtrs_wrn(con->c.sess, "Unknown IMM type %u\n", + rtrs_wrn(con->c.path, "Unknown IMM type %u\n", imm_type); } if (w_inval) @@ -658,7 +659,7 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) else err = rtrs_post_recv_empty(&con->c, &io_comp_cqe); if (err) { - rtrs_err(con->c.sess, "rtrs_post_recv_empty(): %d\n", + rtrs_err(con->c.path, "rtrs_post_recv_empty(): %d\n", err); rtrs_rdma_error_recovery(con); } @@ -670,7 +671,7 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) WARN_ON(!(wc->wc_flags & IB_WC_WITH_INVALIDATE || wc->wc_flags & IB_WC_WITH_IMM)); WARN_ON(wc->wr_cqe->done != rtrs_clt_rdma_done); - if (sess->flags & RTRS_MSG_NEW_RKEY_F) { + if (clt_path->flags & RTRS_MSG_NEW_RKEY_F) { if (wc->wc_flags & IB_WC_WITH_INVALIDATE) return rtrs_clt_recv_done(con, wc); @@ -685,7 +686,7 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) break; default: - rtrs_wrn(sess->clt, "Unexpected WC type: %d\n", wc->opcode); + rtrs_wrn(clt_path->clt, "Unexpected WC type: %d\n", wc->opcode); return; } } @@ -693,10 +694,10 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) static int post_recv_io(struct rtrs_clt_con *con, size_t q_size) { int err, i; - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); for (i = 0; i < q_size; i++) { - if (sess->flags & RTRS_MSG_NEW_RKEY_F) { + if (clt_path->flags & RTRS_MSG_NEW_RKEY_F) { struct rtrs_iu *iu = &con->rsp_ius[i]; err = rtrs_iu_post_recv(&con->c, iu); @@ -710,16 +711,16 @@ static int post_recv_io(struct rtrs_clt_con *con, size_t q_size) return 0; } -static int post_recv_sess(struct rtrs_clt_sess *sess) +static int post_recv_path(struct rtrs_clt_path *clt_path) { size_t q_size = 0; int err, cid; - for (cid = 0; cid < sess->s.con_num; cid++) { + for (cid = 0; cid < clt_path->s.con_num; cid++) { if (cid == 0) q_size = SERVICE_CON_QUEUE_DEPTH; else - q_size = sess->queue_depth; + q_size = clt_path->queue_depth; /* * x2 for RDMA read responses + FR key invalidations, @@ -727,9 +728,10 @@ static int post_recv_sess(struct rtrs_clt_sess *sess) */ q_size *= 2; - err = post_recv_io(to_clt_con(sess->s.con[cid]), q_size); + err = post_recv_io(to_clt_con(clt_path->s.con[cid]), q_size); if (err) { - rtrs_err(sess->clt, "post_recv_io(), err: %d\n", err); + rtrs_err(clt_path->clt, "post_recv_io(), err: %d\n", + err); return err; } } @@ -740,29 +742,29 @@ static int post_recv_sess(struct rtrs_clt_sess *sess) struct path_it { int i; struct list_head skip_list; - struct rtrs_clt *clt; - struct rtrs_clt_sess *(*next_path)(struct path_it *it); + struct rtrs_clt_sess *clt; + struct rtrs_clt_path *(*next_path)(struct path_it *it); }; -/** - * list_next_or_null_rr_rcu - get next list element in round-robin fashion. +/* + * rtrs_clt_get_next_path_or_null - get clt path from the list or return NULL * @head: the head for the list. - * @ptr: the list head to take the next element from. - * @type: the type of the struct this is embedded in. - * @memb: the name of the list_head within the struct. + * @clt_path: The element to take the next clt_path from. * - * Next element returned in round-robin fashion, i.e. head will be skipped, + * Next clt path returned in round-robin fashion, i.e. head will be skipped, * but if list is observed as empty, NULL will be returned. * - * This primitive may safely run concurrently with the _rcu list-mutation + * This function may safely run concurrently with the _rcu list-mutation * primitives such as list_add_rcu() as long as it's guarded by rcu_read_lock(). */ -#define list_next_or_null_rr_rcu(head, ptr, type, memb) \ -({ \ - list_next_or_null_rcu(head, ptr, type, memb) ?: \ - list_next_or_null_rcu(head, READ_ONCE((ptr)->next), \ - type, memb); \ -}) +static inline struct rtrs_clt_path * +rtrs_clt_get_next_path_or_null(struct list_head *head, struct rtrs_clt_path *clt_path) +{ + return list_next_or_null_rcu(head, &clt_path->s.entry, typeof(*clt_path), s.entry) ?: + list_next_or_null_rcu(head, + READ_ONCE((&clt_path->s.entry)->next), + typeof(*clt_path), s.entry); +} /** * get_next_path_rr() - Returns path in round-robin fashion. @@ -773,11 +775,11 @@ struct path_it { * Locks: * rcu_read_lock() must be hold. */ -static struct rtrs_clt_sess *get_next_path_rr(struct path_it *it) +static struct rtrs_clt_path *get_next_path_rr(struct path_it *it) { - struct rtrs_clt_sess __rcu **ppcpu_path; - struct rtrs_clt_sess *path; - struct rtrs_clt *clt; + struct rtrs_clt_path __rcu **ppcpu_path; + struct rtrs_clt_path *path; + struct rtrs_clt_sess *clt; clt = it->clt; @@ -793,10 +795,8 @@ static struct rtrs_clt_sess *get_next_path_rr(struct path_it *it) path = list_first_or_null_rcu(&clt->paths_list, typeof(*path), s.entry); else - path = list_next_or_null_rr_rcu(&clt->paths_list, - &path->s.entry, - typeof(*path), - s.entry); + path = rtrs_clt_get_next_path_or_null(&clt->paths_list, path); + rcu_assign_pointer(*ppcpu_path, path); return path; @@ -811,26 +811,26 @@ static struct rtrs_clt_sess *get_next_path_rr(struct path_it *it) * Locks: * rcu_read_lock() must be hold. */ -static struct rtrs_clt_sess *get_next_path_min_inflight(struct path_it *it) +static struct rtrs_clt_path *get_next_path_min_inflight(struct path_it *it) { - struct rtrs_clt_sess *min_path = NULL; - struct rtrs_clt *clt = it->clt; - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *min_path = NULL; + struct rtrs_clt_sess *clt = it->clt; + struct rtrs_clt_path *clt_path; int min_inflight = INT_MAX; int inflight; - list_for_each_entry_rcu(sess, &clt->paths_list, s.entry) { - if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTED) + list_for_each_entry_rcu(clt_path, &clt->paths_list, s.entry) { + if (READ_ONCE(clt_path->state) != RTRS_CLT_CONNECTED) continue; - if (!list_empty(raw_cpu_ptr(sess->mp_skip_entry))) + if (!list_empty(raw_cpu_ptr(clt_path->mp_skip_entry))) continue; - inflight = atomic_read(&sess->stats->inflight); + inflight = atomic_read(&clt_path->stats->inflight); if (inflight < min_inflight) { min_inflight = inflight; - min_path = sess; + min_path = clt_path; } } @@ -862,26 +862,26 @@ static struct rtrs_clt_sess *get_next_path_min_inflight(struct path_it *it) * Therefore the caller MUST check the returned * path is NULL and trigger the IO error. */ -static struct rtrs_clt_sess *get_next_path_min_latency(struct path_it *it) +static struct rtrs_clt_path *get_next_path_min_latency(struct path_it *it) { - struct rtrs_clt_sess *min_path = NULL; - struct rtrs_clt *clt = it->clt; - struct rtrs_clt_sess *sess; - ktime_t min_latency = INT_MAX; + struct rtrs_clt_path *min_path = NULL; + struct rtrs_clt_sess *clt = it->clt; + struct rtrs_clt_path *clt_path; + ktime_t min_latency = KTIME_MAX; ktime_t latency; - list_for_each_entry_rcu(sess, &clt->paths_list, s.entry) { - if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTED) + list_for_each_entry_rcu(clt_path, &clt->paths_list, s.entry) { + if (READ_ONCE(clt_path->state) != RTRS_CLT_CONNECTED) continue; - if (!list_empty(raw_cpu_ptr(sess->mp_skip_entry))) + if (!list_empty(raw_cpu_ptr(clt_path->mp_skip_entry))) continue; - latency = sess->s.hb_cur_latency; + latency = clt_path->s.hb_cur_latency; if (latency < min_latency) { min_latency = latency; - min_path = sess; + min_path = clt_path; } } @@ -895,7 +895,7 @@ static struct rtrs_clt_sess *get_next_path_min_latency(struct path_it *it) return min_path; } -static inline void path_it_init(struct path_it *it, struct rtrs_clt *clt) +static inline void path_it_init(struct path_it *it, struct rtrs_clt_sess *clt) { INIT_LIST_HEAD(&it->skip_list); it->clt = clt; @@ -913,7 +913,7 @@ static inline void path_it_deinit(struct path_it *it) { struct list_head *skip, *tmp; /* - * The skip_list is used only for the MIN_INFLIGHT policy. + * The skip_list is used only for the MIN_INFLIGHT and MIN_LATENCY policies. * We need to remove paths from it, so that next IO can insert * paths (->mp_skip_entry) into a skip_list again. */ @@ -928,7 +928,7 @@ static inline void path_it_deinit(struct path_it *it) * the corresponding buffer of rtrs_iu (req->iu->buf), which later on will * also hold the control message of rtrs. * @req: an io request holding information about IO. - * @sess: client session + * @clt_path: client path * @conf: conformation callback function to notify upper layer. * @permit: permit for allocation of RDMA remote buffer * @priv: private pointer @@ -940,7 +940,7 @@ static inline void path_it_deinit(struct path_it *it) * @dir: direction of the IO. */ static void rtrs_clt_init_req(struct rtrs_clt_io_req *req, - struct rtrs_clt_sess *sess, + struct rtrs_clt_path *clt_path, void (*conf)(void *priv, int errno), struct rtrs_permit *permit, void *priv, const struct kvec *vec, size_t usr_len, @@ -958,13 +958,13 @@ static void rtrs_clt_init_req(struct rtrs_clt_io_req *req, req->sg_cnt = sg_cnt; req->priv = priv; req->dir = dir; - req->con = rtrs_permit_to_clt_con(sess, permit); + req->con = rtrs_permit_to_clt_con(clt_path, permit); req->conf = conf; req->need_inv = false; req->need_inv_comp = false; req->inv_errno = 0; refcount_set(&req->ref, 1); - req->mp_policy = sess->clt->mp_policy; + req->mp_policy = clt_path->clt->mp_policy; iov_iter_kvec(&iter, READ, vec, 1, usr_len); len = _copy_from_iter(req->iu->buf, usr_len, &iter); @@ -974,7 +974,7 @@ static void rtrs_clt_init_req(struct rtrs_clt_io_req *req, } static struct rtrs_clt_io_req * -rtrs_clt_get_req(struct rtrs_clt_sess *sess, +rtrs_clt_get_req(struct rtrs_clt_path *clt_path, void (*conf)(void *priv, int errno), struct rtrs_permit *permit, void *priv, const struct kvec *vec, size_t usr_len, @@ -983,14 +983,14 @@ rtrs_clt_get_req(struct rtrs_clt_sess *sess, { struct rtrs_clt_io_req *req; - req = &sess->reqs[permit->mem_id]; - rtrs_clt_init_req(req, sess, conf, permit, priv, vec, usr_len, + req = &clt_path->reqs[permit->mem_id]; + rtrs_clt_init_req(req, clt_path, conf, permit, priv, vec, usr_len, sg, sg_cnt, data_len, dir); return req; } static struct rtrs_clt_io_req * -rtrs_clt_get_copy_req(struct rtrs_clt_sess *alive_sess, +rtrs_clt_get_copy_req(struct rtrs_clt_path *alive_path, struct rtrs_clt_io_req *fail_req) { struct rtrs_clt_io_req *req; @@ -999,8 +999,8 @@ rtrs_clt_get_copy_req(struct rtrs_clt_sess *alive_sess, .iov_len = fail_req->usr_len }; - req = &alive_sess->reqs[fail_req->permit->mem_id]; - rtrs_clt_init_req(req, alive_sess, fail_req->conf, fail_req->permit, + req = &alive_path->reqs[fail_req->permit->mem_id]; + rtrs_clt_init_req(req, alive_path, fail_req->conf, fail_req->permit, fail_req->priv, &vec, fail_req->usr_len, fail_req->sglist, fail_req->sg_cnt, fail_req->data_len, fail_req->dir); @@ -1010,10 +1010,11 @@ rtrs_clt_get_copy_req(struct rtrs_clt_sess *alive_sess, static int rtrs_post_rdma_write_sg(struct rtrs_clt_con *con, struct rtrs_clt_io_req *req, struct rtrs_rbuf *rbuf, bool fr_en, - u32 size, u32 imm, struct ib_send_wr *wr, + u32 count, u32 size, u32 imm, + struct ib_send_wr *wr, struct ib_send_wr *tail) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); struct ib_sge *sge = req->sge; enum ib_send_flags flags; struct scatterlist *sg; @@ -1030,25 +1031,26 @@ static int rtrs_post_rdma_write_sg(struct rtrs_clt_con *con, num_sge = 2; ptail = tail; } else { - for_each_sg(req->sglist, sg, req->sg_cnt, i) { + for_each_sg(req->sglist, sg, count, i) { sge[i].addr = sg_dma_address(sg); sge[i].length = sg_dma_len(sg); - sge[i].lkey = sess->s.dev->ib_pd->local_dma_lkey; + sge[i].lkey = clt_path->s.dev->ib_pd->local_dma_lkey; } - num_sge = 1 + req->sg_cnt; + num_sge = 1 + count; } sge[i].addr = req->iu->dma_addr; sge[i].length = size; - sge[i].lkey = sess->s.dev->ib_pd->local_dma_lkey; + sge[i].lkey = clt_path->s.dev->ib_pd->local_dma_lkey; /* * From time to time we have to post signalled sends, * or send queue will fill up and only QP reset can help. */ - flags = atomic_inc_return(&con->c.wr_cnt) % sess->s.signal_interval ? + flags = atomic_inc_return(&con->c.wr_cnt) % clt_path->s.signal_interval ? 0 : IB_SEND_SIGNALED; - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, req->iu->dma_addr, + ib_dma_sync_single_for_device(clt_path->s.dev->ib_dev, + req->iu->dma_addr, size, DMA_TO_DEVICE); return rtrs_iu_post_rdma_write_imm(&con->c, req->iu, sge, num_sge, @@ -1074,8 +1076,8 @@ static int rtrs_map_sg_fr(struct rtrs_clt_io_req *req, size_t count) static int rtrs_clt_write_req(struct rtrs_clt_io_req *req) { struct rtrs_clt_con *con = req->con; - struct rtrs_sess *s = con->c.sess; - struct rtrs_clt_sess *sess = to_clt_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_clt_path *clt_path = to_clt_path(s); struct rtrs_msg_rdma_write *msg; struct rtrs_rbuf *rbuf; @@ -1088,13 +1090,13 @@ static int rtrs_clt_write_req(struct rtrs_clt_io_req *req) const size_t tsize = sizeof(*msg) + req->data_len + req->usr_len; - if (tsize > sess->chunk_size) { + if (tsize > clt_path->chunk_size) { rtrs_wrn(s, "Write request failed, size too big %zu > %d\n", - tsize, sess->chunk_size); + tsize, clt_path->chunk_size); return -EMSGSIZE; } if (req->sg_cnt) { - count = ib_dma_map_sg(sess->s.dev->ib_dev, req->sglist, + count = ib_dma_map_sg(clt_path->s.dev->ib_dev, req->sglist, req->sg_cnt, req->dir); if (!count) { rtrs_wrn(s, "Write request failed, map failed\n"); @@ -1111,7 +1113,7 @@ static int rtrs_clt_write_req(struct rtrs_clt_io_req *req) imm = rtrs_to_io_req_imm(imm); buf_id = req->permit->mem_id; req->sg_size = tsize; - rbuf = &sess->rbufs[buf_id]; + rbuf = &clt_path->rbufs[buf_id]; if (count) { ret = rtrs_map_sg_fr(req, count); @@ -1119,7 +1121,7 @@ static int rtrs_clt_write_req(struct rtrs_clt_io_req *req) rtrs_err_rl(s, "Write request failed, failed to map fast reg. data, err: %d\n", ret); - ib_dma_unmap_sg(sess->s.dev->ib_dev, req->sglist, + ib_dma_unmap_sg(clt_path->s.dev->ib_dev, req->sglist, req->sg_cnt, req->dir); return ret; } @@ -1147,18 +1149,18 @@ static int rtrs_clt_write_req(struct rtrs_clt_io_req *req) */ rtrs_clt_update_all_stats(req, WRITE); - ret = rtrs_post_rdma_write_sg(req->con, req, rbuf, fr_en, + ret = rtrs_post_rdma_write_sg(req->con, req, rbuf, fr_en, count, req->usr_len + sizeof(*msg), imm, wr, &inv_wr); if (ret) { rtrs_err_rl(s, "Write request failed: error=%d path=%s [%s:%u]\n", - ret, kobject_name(&sess->kobj), sess->hca_name, - sess->hca_port); + ret, kobject_name(&clt_path->kobj), clt_path->hca_name, + clt_path->hca_port); if (req->mp_policy == MP_POLICY_MIN_INFLIGHT) - atomic_dec(&sess->stats->inflight); + atomic_dec(&clt_path->stats->inflight); if (req->sg_cnt) - ib_dma_unmap_sg(sess->s.dev->ib_dev, req->sglist, + ib_dma_unmap_sg(clt_path->s.dev->ib_dev, req->sglist, req->sg_cnt, req->dir); } @@ -1168,10 +1170,10 @@ static int rtrs_clt_write_req(struct rtrs_clt_io_req *req) static int rtrs_clt_read_req(struct rtrs_clt_io_req *req) { struct rtrs_clt_con *con = req->con; - struct rtrs_sess *s = con->c.sess; - struct rtrs_clt_sess *sess = to_clt_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_clt_path *clt_path = to_clt_path(s); struct rtrs_msg_rdma_read *msg; - struct rtrs_ib_dev *dev = sess->s.dev; + struct rtrs_ib_dev *dev = clt_path->s.dev; struct ib_reg_wr rwr; struct ib_send_wr *wr = NULL; @@ -1181,10 +1183,10 @@ static int rtrs_clt_read_req(struct rtrs_clt_io_req *req) const size_t tsize = sizeof(*msg) + req->data_len + req->usr_len; - if (tsize > sess->chunk_size) { + if (tsize > clt_path->chunk_size) { rtrs_wrn(s, "Read request failed, message size is %zu, bigger than CHUNK_SIZE %d\n", - tsize, sess->chunk_size); + tsize, clt_path->chunk_size); return -EMSGSIZE; } @@ -1254,15 +1256,15 @@ static int rtrs_clt_read_req(struct rtrs_clt_io_req *req) */ rtrs_clt_update_all_stats(req, READ); - ret = rtrs_post_send_rdma(req->con, req, &sess->rbufs[buf_id], + ret = rtrs_post_send_rdma(req->con, req, &clt_path->rbufs[buf_id], req->data_len, imm, wr); if (ret) { rtrs_err_rl(s, "Read request failed: error=%d path=%s [%s:%u]\n", - ret, kobject_name(&sess->kobj), sess->hca_name, - sess->hca_port); + ret, kobject_name(&clt_path->kobj), clt_path->hca_name, + clt_path->hca_port); if (req->mp_policy == MP_POLICY_MIN_INFLIGHT) - atomic_dec(&sess->stats->inflight); + atomic_dec(&clt_path->stats->inflight); req->need_inv = false; if (req->sg_cnt) ib_dma_unmap_sg(dev->ib_dev, req->sglist, @@ -1277,21 +1279,21 @@ static int rtrs_clt_read_req(struct rtrs_clt_io_req *req) * @clt: clt context * @fail_req: a failed io request. */ -static int rtrs_clt_failover_req(struct rtrs_clt *clt, +static int rtrs_clt_failover_req(struct rtrs_clt_sess *clt, struct rtrs_clt_io_req *fail_req) { - struct rtrs_clt_sess *alive_sess; + struct rtrs_clt_path *alive_path; struct rtrs_clt_io_req *req; int err = -ECONNABORTED; struct path_it it; rcu_read_lock(); for (path_it_init(&it, clt); - (alive_sess = it.next_path(&it)) && it.i < it.clt->paths_num; + (alive_path = it.next_path(&it)) && it.i < it.clt->paths_num; it.i++) { - if (READ_ONCE(alive_sess->state) != RTRS_CLT_CONNECTED) + if (READ_ONCE(alive_path->state) != RTRS_CLT_CONNECTED) continue; - req = rtrs_clt_get_copy_req(alive_sess, fail_req); + req = rtrs_clt_get_copy_req(alive_path, fail_req); if (req->dir == DMA_TO_DEVICE) err = rtrs_clt_write_req(req); else @@ -1301,7 +1303,7 @@ static int rtrs_clt_failover_req(struct rtrs_clt *clt, continue; } /* Success path */ - rtrs_clt_inc_failover_cnt(alive_sess->stats); + rtrs_clt_inc_failover_cnt(alive_path->stats); break; } path_it_deinit(&it); @@ -1310,16 +1312,16 @@ static int rtrs_clt_failover_req(struct rtrs_clt *clt, return err; } -static void fail_all_outstanding_reqs(struct rtrs_clt_sess *sess) +static void fail_all_outstanding_reqs(struct rtrs_clt_path *clt_path) { - struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_sess *clt = clt_path->clt; struct rtrs_clt_io_req *req; int i, err; - if (!sess->reqs) + if (!clt_path->reqs) return; - for (i = 0; i < sess->queue_depth; ++i) { - req = &sess->reqs[i]; + for (i = 0; i < clt_path->queue_depth; ++i) { + req = &clt_path->reqs[i]; if (!req->in_use) continue; @@ -1337,38 +1339,39 @@ static void fail_all_outstanding_reqs(struct rtrs_clt_sess *sess) } } -static void free_sess_reqs(struct rtrs_clt_sess *sess) +static void free_path_reqs(struct rtrs_clt_path *clt_path) { struct rtrs_clt_io_req *req; int i; - if (!sess->reqs) + if (!clt_path->reqs) return; - for (i = 0; i < sess->queue_depth; ++i) { - req = &sess->reqs[i]; + for (i = 0; i < clt_path->queue_depth; ++i) { + req = &clt_path->reqs[i]; if (req->mr) ib_dereg_mr(req->mr); kfree(req->sge); - rtrs_iu_free(req->iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(req->iu, clt_path->s.dev->ib_dev, 1); } - kfree(sess->reqs); - sess->reqs = NULL; + kfree(clt_path->reqs); + clt_path->reqs = NULL; } -static int alloc_sess_reqs(struct rtrs_clt_sess *sess) +static int alloc_path_reqs(struct rtrs_clt_path *clt_path) { struct rtrs_clt_io_req *req; int i, err = -ENOMEM; - sess->reqs = kcalloc(sess->queue_depth, sizeof(*sess->reqs), - GFP_KERNEL); - if (!sess->reqs) + clt_path->reqs = kcalloc(clt_path->queue_depth, + sizeof(*clt_path->reqs), + GFP_KERNEL); + if (!clt_path->reqs) return -ENOMEM; - for (i = 0; i < sess->queue_depth; ++i) { - req = &sess->reqs[i]; - req->iu = rtrs_iu_alloc(1, sess->max_hdr_size, GFP_KERNEL, - sess->s.dev->ib_dev, + for (i = 0; i < clt_path->queue_depth; ++i) { + req = &clt_path->reqs[i]; + req->iu = rtrs_iu_alloc(1, clt_path->max_hdr_size, GFP_KERNEL, + clt_path->s.dev->ib_dev, DMA_TO_DEVICE, rtrs_clt_rdma_done); if (!req->iu) @@ -1378,13 +1381,14 @@ static int alloc_sess_reqs(struct rtrs_clt_sess *sess) if (!req->sge) goto out; - req->mr = ib_alloc_mr(sess->s.dev->ib_pd, IB_MR_TYPE_MEM_REG, - sess->max_pages_per_mr); + req->mr = ib_alloc_mr(clt_path->s.dev->ib_pd, + IB_MR_TYPE_MEM_REG, + clt_path->max_pages_per_mr); if (IS_ERR(req->mr)) { err = PTR_ERR(req->mr); req->mr = NULL; - pr_err("Failed to alloc sess->max_pages_per_mr %d\n", - sess->max_pages_per_mr); + pr_err("Failed to alloc clt_path->max_pages_per_mr %d\n", + clt_path->max_pages_per_mr); goto out; } @@ -1394,18 +1398,17 @@ static int alloc_sess_reqs(struct rtrs_clt_sess *sess) return 0; out: - free_sess_reqs(sess); + free_path_reqs(clt_path); return err; } -static int alloc_permits(struct rtrs_clt *clt) +static int alloc_permits(struct rtrs_clt_sess *clt) { unsigned int chunk_bits; int err, i; - clt->permits_map = kcalloc(BITS_TO_LONGS(clt->queue_depth), - sizeof(long), GFP_KERNEL); + clt->permits_map = bitmap_zalloc(clt->queue_depth, GFP_KERNEL); if (!clt->permits_map) { err = -ENOMEM; goto out_err; @@ -1427,33 +1430,31 @@ static int alloc_permits(struct rtrs_clt *clt) return 0; err_map: - kfree(clt->permits_map); + bitmap_free(clt->permits_map); clt->permits_map = NULL; out_err: return err; } -static void free_permits(struct rtrs_clt *clt) +static void free_permits(struct rtrs_clt_sess *clt) { - if (clt->permits_map) { - size_t sz = clt->queue_depth; - + if (clt->permits_map) wait_event(clt->permits_wait, - find_first_bit(clt->permits_map, sz) >= sz); - } - kfree(clt->permits_map); + bitmap_empty(clt->permits_map, clt->queue_depth)); + + bitmap_free(clt->permits_map); clt->permits_map = NULL; kfree(clt->permits); clt->permits = NULL; } -static void query_fast_reg_mode(struct rtrs_clt_sess *sess) +static void query_fast_reg_mode(struct rtrs_clt_path *clt_path) { struct ib_device *ib_dev; u64 max_pages_per_mr; int mr_page_shift; - ib_dev = sess->s.dev->ib_dev; + ib_dev = clt_path->s.dev->ib_dev; /* * Use the smallest page size supported by the HCA, down to a @@ -1463,24 +1464,24 @@ static void query_fast_reg_mode(struct rtrs_clt_sess *sess) mr_page_shift = max(12, ffs(ib_dev->attrs.page_size_cap) - 1); max_pages_per_mr = ib_dev->attrs.max_mr_size; do_div(max_pages_per_mr, (1ull << mr_page_shift)); - sess->max_pages_per_mr = - min3(sess->max_pages_per_mr, (u32)max_pages_per_mr, + clt_path->max_pages_per_mr = + min3(clt_path->max_pages_per_mr, (u32)max_pages_per_mr, ib_dev->attrs.max_fast_reg_page_list_len); - sess->clt->max_segments = - min(sess->max_pages_per_mr, sess->clt->max_segments); + clt_path->clt->max_segments = + min(clt_path->max_pages_per_mr, clt_path->clt->max_segments); } -static bool rtrs_clt_change_state_get_old(struct rtrs_clt_sess *sess, +static bool rtrs_clt_change_state_get_old(struct rtrs_clt_path *clt_path, enum rtrs_clt_state new_state, enum rtrs_clt_state *old_state) { bool changed; - spin_lock_irq(&sess->state_wq.lock); + spin_lock_irq(&clt_path->state_wq.lock); if (old_state) - *old_state = sess->state; - changed = rtrs_clt_change_state(sess, new_state); - spin_unlock_irq(&sess->state_wq.lock); + *old_state = clt_path->state; + changed = rtrs_clt_change_state(clt_path, new_state); + spin_unlock_irq(&clt_path->state_wq.lock); return changed; } @@ -1492,9 +1493,9 @@ static void rtrs_clt_hb_err_handler(struct rtrs_con *c) rtrs_rdma_error_recovery(con); } -static void rtrs_clt_init_hb(struct rtrs_clt_sess *sess) +static void rtrs_clt_init_hb(struct rtrs_clt_path *clt_path) { - rtrs_init_hb(&sess->s, &io_comp_cqe, + rtrs_init_hb(&clt_path->s, &io_comp_cqe, RTRS_HB_INTERVAL_MS, RTRS_HB_MISSED_MAX, rtrs_clt_hb_err_handler, @@ -1504,17 +1505,32 @@ static void rtrs_clt_init_hb(struct rtrs_clt_sess *sess) static void rtrs_clt_reconnect_work(struct work_struct *work); static void rtrs_clt_close_work(struct work_struct *work); -static struct rtrs_clt_sess *alloc_sess(struct rtrs_clt *clt, +static void rtrs_clt_err_recovery_work(struct work_struct *work) +{ + struct rtrs_clt_path *clt_path; + struct rtrs_clt_sess *clt; + int delay_ms; + + clt_path = container_of(work, struct rtrs_clt_path, err_recovery_work); + clt = clt_path->clt; + delay_ms = clt->reconnect_delay_sec * 1000; + rtrs_clt_stop_and_destroy_conns(clt_path); + queue_delayed_work(rtrs_wq, &clt_path->reconnect_dwork, + msecs_to_jiffies(delay_ms + + prandom_u32_max(RTRS_RECONNECT_SEED))); +} + +static struct rtrs_clt_path *alloc_path(struct rtrs_clt_sess *clt, const struct rtrs_addr *path, size_t con_num, u32 nr_poll_queues) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; int err = -ENOMEM; int cpu; size_t total_con; - sess = kzalloc(sizeof(*sess), GFP_KERNEL); - if (!sess) + clt_path = kzalloc(sizeof(*clt_path), GFP_KERNEL); + if (!clt_path) goto err; /* @@ -1522,20 +1538,21 @@ static struct rtrs_clt_sess *alloc_sess(struct rtrs_clt *clt, * +1: Extra connection for user messages */ total_con = con_num + nr_poll_queues + 1; - sess->s.con = kcalloc(total_con, sizeof(*sess->s.con), GFP_KERNEL); - if (!sess->s.con) - goto err_free_sess; + clt_path->s.con = kcalloc(total_con, sizeof(*clt_path->s.con), + GFP_KERNEL); + if (!clt_path->s.con) + goto err_free_path; - sess->s.con_num = total_con; - sess->s.irq_con_num = con_num + 1; + clt_path->s.con_num = total_con; + clt_path->s.irq_con_num = con_num + 1; - sess->stats = kzalloc(sizeof(*sess->stats), GFP_KERNEL); - if (!sess->stats) + clt_path->stats = kzalloc(sizeof(*clt_path->stats), GFP_KERNEL); + if (!clt_path->stats) goto err_free_con; - mutex_init(&sess->init_mutex); - uuid_gen(&sess->s.uuid); - memcpy(&sess->s.dst_addr, path->dst, + mutex_init(&clt_path->init_mutex); + uuid_gen(&clt_path->s.uuid); + memcpy(&clt_path->s.dst_addr, path->dst, rdma_addr_size((struct sockaddr *)path->dst)); /* @@ -1544,53 +1561,55 @@ static struct rtrs_clt_sess *alloc_sess(struct rtrs_clt *clt, * the sess->src_addr will contain only zeros, which is then fine. */ if (path->src) - memcpy(&sess->s.src_addr, path->src, + memcpy(&clt_path->s.src_addr, path->src, rdma_addr_size((struct sockaddr *)path->src)); - strscpy(sess->s.sessname, clt->sessname, sizeof(sess->s.sessname)); - sess->clt = clt; - sess->max_pages_per_mr = RTRS_MAX_SEGMENTS; - init_waitqueue_head(&sess->state_wq); - sess->state = RTRS_CLT_CONNECTING; - atomic_set(&sess->connected_cnt, 0); - INIT_WORK(&sess->close_work, rtrs_clt_close_work); - INIT_DELAYED_WORK(&sess->reconnect_dwork, rtrs_clt_reconnect_work); - rtrs_clt_init_hb(sess); - - sess->mp_skip_entry = alloc_percpu(typeof(*sess->mp_skip_entry)); - if (!sess->mp_skip_entry) + strscpy(clt_path->s.sessname, clt->sessname, + sizeof(clt_path->s.sessname)); + clt_path->clt = clt; + clt_path->max_pages_per_mr = RTRS_MAX_SEGMENTS; + init_waitqueue_head(&clt_path->state_wq); + clt_path->state = RTRS_CLT_CONNECTING; + atomic_set(&clt_path->connected_cnt, 0); + INIT_WORK(&clt_path->close_work, rtrs_clt_close_work); + INIT_WORK(&clt_path->err_recovery_work, rtrs_clt_err_recovery_work); + INIT_DELAYED_WORK(&clt_path->reconnect_dwork, rtrs_clt_reconnect_work); + rtrs_clt_init_hb(clt_path); + + clt_path->mp_skip_entry = alloc_percpu(typeof(*clt_path->mp_skip_entry)); + if (!clt_path->mp_skip_entry) goto err_free_stats; for_each_possible_cpu(cpu) - INIT_LIST_HEAD(per_cpu_ptr(sess->mp_skip_entry, cpu)); + INIT_LIST_HEAD(per_cpu_ptr(clt_path->mp_skip_entry, cpu)); - err = rtrs_clt_init_stats(sess->stats); + err = rtrs_clt_init_stats(clt_path->stats); if (err) goto err_free_percpu; - return sess; + return clt_path; err_free_percpu: - free_percpu(sess->mp_skip_entry); + free_percpu(clt_path->mp_skip_entry); err_free_stats: - kfree(sess->stats); + kfree(clt_path->stats); err_free_con: - kfree(sess->s.con); -err_free_sess: - kfree(sess); + kfree(clt_path->s.con); +err_free_path: + kfree(clt_path); err: return ERR_PTR(err); } -void free_sess(struct rtrs_clt_sess *sess) +void free_path(struct rtrs_clt_path *clt_path) { - free_percpu(sess->mp_skip_entry); - mutex_destroy(&sess->init_mutex); - kfree(sess->s.con); - kfree(sess->rbufs); - kfree(sess); + free_percpu(clt_path->mp_skip_entry); + mutex_destroy(&clt_path->init_mutex); + kfree(clt_path->s.con); + kfree(clt_path->rbufs); + kfree(clt_path); } -static int create_con(struct rtrs_clt_sess *sess, unsigned int cid) +static int create_con(struct rtrs_clt_path *clt_path, unsigned int cid) { struct rtrs_clt_con *con; @@ -1601,28 +1620,28 @@ static int create_con(struct rtrs_clt_sess *sess, unsigned int cid) /* Map first two connections to the first CPU */ con->cpu = (cid ? cid - 1 : 0) % nr_cpu_ids; con->c.cid = cid; - con->c.sess = &sess->s; + con->c.path = &clt_path->s; /* Align with srv, init as 1 */ atomic_set(&con->c.wr_cnt, 1); mutex_init(&con->con_mutex); - sess->s.con[cid] = &con->c; + clt_path->s.con[cid] = &con->c; return 0; } static void destroy_con(struct rtrs_clt_con *con) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); - sess->s.con[con->c.cid] = NULL; + clt_path->s.con[con->c.cid] = NULL; mutex_destroy(&con->con_mutex); kfree(con); } static int create_con_cq_qp(struct rtrs_clt_con *con) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); u32 max_send_wr, max_recv_wr, cq_num, max_send_sge, wr_limit; int err, cq_vector; struct rtrs_msg_rkey_rsp *rsp; @@ -1631,7 +1650,7 @@ static int create_con_cq_qp(struct rtrs_clt_con *con) if (con->c.cid == 0) { max_send_sge = 1; /* We must be the first here */ - if (WARN_ON(sess->s.dev)) + if (WARN_ON(clt_path->s.dev)) return -EINVAL; /* @@ -1639,16 +1658,16 @@ static int create_con_cq_qp(struct rtrs_clt_con *con) * Be careful not to close user connection before ib dev * is gracefully put. */ - sess->s.dev = rtrs_ib_dev_find_or_add(con->c.cm_id->device, + clt_path->s.dev = rtrs_ib_dev_find_or_add(con->c.cm_id->device, &dev_pd); - if (!sess->s.dev) { - rtrs_wrn(sess->clt, + if (!clt_path->s.dev) { + rtrs_wrn(clt_path->clt, "rtrs_ib_dev_find_get_or_add(): no memory\n"); return -ENOMEM; } - sess->s.dev_ref = 1; - query_fast_reg_mode(sess); - wr_limit = sess->s.dev->ib_dev->attrs.max_qp_wr; + clt_path->s.dev_ref = 1; + query_fast_reg_mode(clt_path); + wr_limit = clt_path->s.dev->ib_dev->attrs.max_qp_wr; /* * Two (request + registration) completion for send * Two for recv if always_invalidate is set on server @@ -1665,27 +1684,28 @@ static int create_con_cq_qp(struct rtrs_clt_con *con) * This is always true if user connection (cid == 0) is * established first. */ - if (WARN_ON(!sess->s.dev)) + if (WARN_ON(!clt_path->s.dev)) return -EINVAL; - if (WARN_ON(!sess->queue_depth)) + if (WARN_ON(!clt_path->queue_depth)) return -EINVAL; - wr_limit = sess->s.dev->ib_dev->attrs.max_qp_wr; + wr_limit = clt_path->s.dev->ib_dev->attrs.max_qp_wr; /* Shared between connections */ - sess->s.dev_ref++; + clt_path->s.dev_ref++; max_send_wr = min_t(int, wr_limit, /* QD * (REQ + RSP + FR REGS or INVS) + drain */ - sess->queue_depth * 3 + 1); + clt_path->queue_depth * 3 + 1); max_recv_wr = min_t(int, wr_limit, - sess->queue_depth * 3 + 1); + clt_path->queue_depth * 3 + 1); max_send_sge = 2; } atomic_set(&con->c.sq_wr_avail, max_send_wr); cq_num = max_send_wr + max_recv_wr; /* alloc iu to recv new rkey reply when server reports flags set */ - if (sess->flags & RTRS_MSG_NEW_RKEY_F || con->c.cid == 0) { + if (clt_path->flags & RTRS_MSG_NEW_RKEY_F || con->c.cid == 0) { con->rsp_ius = rtrs_iu_alloc(cq_num, sizeof(*rsp), - GFP_KERNEL, sess->s.dev->ib_dev, + GFP_KERNEL, + clt_path->s.dev->ib_dev, DMA_FROM_DEVICE, rtrs_clt_rdma_done); if (!con->rsp_ius) @@ -1693,13 +1713,13 @@ static int create_con_cq_qp(struct rtrs_clt_con *con) con->queue_num = cq_num; } cq_num = max_send_wr + max_recv_wr; - cq_vector = con->cpu % sess->s.dev->ib_dev->num_comp_vectors; - if (con->c.cid >= sess->s.irq_con_num) - err = rtrs_cq_qp_create(&sess->s, &con->c, max_send_sge, + cq_vector = con->cpu % clt_path->s.dev->ib_dev->num_comp_vectors; + if (con->c.cid >= clt_path->s.irq_con_num) + err = rtrs_cq_qp_create(&clt_path->s, &con->c, max_send_sge, cq_vector, cq_num, max_send_wr, max_recv_wr, IB_POLL_DIRECT); else - err = rtrs_cq_qp_create(&sess->s, &con->c, max_send_sge, + err = rtrs_cq_qp_create(&clt_path->s, &con->c, max_send_sge, cq_vector, cq_num, max_send_wr, max_recv_wr, IB_POLL_SOFTIRQ); /* @@ -1711,7 +1731,7 @@ static int create_con_cq_qp(struct rtrs_clt_con *con) static void destroy_con_cq_qp(struct rtrs_clt_con *con) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); /* * Be careful here: destroy_con_cq_qp() can be called even @@ -1720,13 +1740,14 @@ static void destroy_con_cq_qp(struct rtrs_clt_con *con) lockdep_assert_held(&con->con_mutex); rtrs_cq_qp_destroy(&con->c); if (con->rsp_ius) { - rtrs_iu_free(con->rsp_ius, sess->s.dev->ib_dev, con->queue_num); + rtrs_iu_free(con->rsp_ius, clt_path->s.dev->ib_dev, + con->queue_num); con->rsp_ius = NULL; con->queue_num = 0; } - if (sess->s.dev_ref && !--sess->s.dev_ref) { - rtrs_ib_dev_put(sess->s.dev); - sess->s.dev = NULL; + if (clt_path->s.dev_ref && !--clt_path->s.dev_ref) { + rtrs_ib_dev_put(clt_path->s.dev); + clt_path->s.dev = NULL; } } @@ -1745,7 +1766,7 @@ static void destroy_cm(struct rtrs_clt_con *con) static int rtrs_rdma_addr_resolved(struct rtrs_clt_con *con) { - struct rtrs_sess *s = con->c.sess; + struct rtrs_path *s = con->c.path; int err; mutex_lock(&con->con_mutex); @@ -1764,8 +1785,8 @@ static int rtrs_rdma_addr_resolved(struct rtrs_clt_con *con) static int rtrs_rdma_route_resolved(struct rtrs_clt_con *con) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); - struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); + struct rtrs_clt_sess *clt = clt_path->clt; struct rtrs_msg_conn_req msg; struct rdma_conn_param param; @@ -1782,11 +1803,11 @@ static int rtrs_rdma_route_resolved(struct rtrs_clt_con *con) .magic = cpu_to_le16(RTRS_MAGIC), .version = cpu_to_le16(RTRS_PROTO_VER), .cid = cpu_to_le16(con->c.cid), - .cid_num = cpu_to_le16(sess->s.con_num), - .recon_cnt = cpu_to_le16(sess->s.recon_cnt), + .cid_num = cpu_to_le16(clt_path->s.con_num), + .recon_cnt = cpu_to_le16(clt_path->s.recon_cnt), }; - msg.first_conn = sess->for_new_clt ? FIRST_CONN : 0; - uuid_copy(&msg.sess_uuid, &sess->s.uuid); + msg.first_conn = clt_path->for_new_clt ? FIRST_CONN : 0; + uuid_copy(&msg.sess_uuid, &clt_path->s.uuid); uuid_copy(&msg.paths_uuid, &clt->paths_uuid); err = rdma_connect_locked(con->c.cm_id, ¶m); @@ -1799,8 +1820,8 @@ static int rtrs_rdma_route_resolved(struct rtrs_clt_con *con) static int rtrs_rdma_conn_established(struct rtrs_clt_con *con, struct rdma_cm_event *ev) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); - struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); + struct rtrs_clt_sess *clt = clt_path->clt; const struct rtrs_msg_conn_rsp *msg; u16 version, queue_depth; int errno; @@ -1831,31 +1852,32 @@ static int rtrs_rdma_conn_established(struct rtrs_clt_con *con, if (con->c.cid == 0) { queue_depth = le16_to_cpu(msg->queue_depth); - if (sess->queue_depth > 0 && queue_depth != sess->queue_depth) { + if (clt_path->queue_depth > 0 && queue_depth != clt_path->queue_depth) { rtrs_err(clt, "Error: queue depth changed\n"); /* * Stop any more reconnection attempts */ - sess->reconnect_attempts = -1; + clt_path->reconnect_attempts = -1; rtrs_err(clt, "Disabling auto-reconnect. Trigger a manual reconnect after issue is resolved\n"); return -ECONNRESET; } - if (!sess->rbufs) { - sess->rbufs = kcalloc(queue_depth, sizeof(*sess->rbufs), - GFP_KERNEL); - if (!sess->rbufs) + if (!clt_path->rbufs) { + clt_path->rbufs = kcalloc(queue_depth, + sizeof(*clt_path->rbufs), + GFP_KERNEL); + if (!clt_path->rbufs) return -ENOMEM; } - sess->queue_depth = queue_depth; - sess->s.signal_interval = min_not_zero(queue_depth, + clt_path->queue_depth = queue_depth; + clt_path->s.signal_interval = min_not_zero(queue_depth, (unsigned short) SERVICE_CON_QUEUE_DEPTH); - sess->max_hdr_size = le32_to_cpu(msg->max_hdr_size); - sess->max_io_size = le32_to_cpu(msg->max_io_size); - sess->flags = le32_to_cpu(msg->flags); - sess->chunk_size = sess->max_io_size + sess->max_hdr_size; + clt_path->max_hdr_size = le32_to_cpu(msg->max_hdr_size); + clt_path->max_io_size = le32_to_cpu(msg->max_io_size); + clt_path->flags = le32_to_cpu(msg->flags); + clt_path->chunk_size = clt_path->max_io_size + clt_path->max_hdr_size; /* * Global IO size is always a minimum. @@ -1866,20 +1888,20 @@ static int rtrs_rdma_conn_established(struct rtrs_clt_con *con, * connections in parallel, use lock. */ mutex_lock(&clt->paths_mutex); - clt->queue_depth = sess->queue_depth; - clt->max_io_size = min_not_zero(sess->max_io_size, + clt->queue_depth = clt_path->queue_depth; + clt->max_io_size = min_not_zero(clt_path->max_io_size, clt->max_io_size); mutex_unlock(&clt->paths_mutex); /* * Cache the hca_port and hca_name for sysfs */ - sess->hca_port = con->c.cm_id->port_num; - scnprintf(sess->hca_name, sizeof(sess->hca_name), - sess->s.dev->ib_dev->name); - sess->s.src_addr = con->c.cm_id->route.addr.src_addr; + clt_path->hca_port = con->c.cm_id->port_num; + scnprintf(clt_path->hca_name, sizeof(clt_path->hca_name), + clt_path->s.dev->ib_dev->name); + clt_path->s.src_addr = con->c.cm_id->route.addr.src_addr; /* set for_new_clt, to allow future reconnect on any path */ - sess->for_new_clt = 1; + clt_path->for_new_clt = 1; } return 0; @@ -1887,16 +1909,16 @@ static int rtrs_rdma_conn_established(struct rtrs_clt_con *con, static inline void flag_success_on_conn(struct rtrs_clt_con *con) { - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); - atomic_inc(&sess->connected_cnt); + atomic_inc(&clt_path->connected_cnt); con->cm_err = 1; } static int rtrs_rdma_conn_rejected(struct rtrs_clt_con *con, struct rdma_cm_event *ev) { - struct rtrs_sess *s = con->c.sess; + struct rtrs_path *s = con->c.path; const struct rtrs_msg_conn_rsp *msg; const char *rej_msg; int status, errno; @@ -1924,23 +1946,25 @@ static int rtrs_rdma_conn_rejected(struct rtrs_clt_con *con, return -ECONNRESET; } -void rtrs_clt_close_conns(struct rtrs_clt_sess *sess, bool wait) +void rtrs_clt_close_conns(struct rtrs_clt_path *clt_path, bool wait) { - if (rtrs_clt_change_state_get_old(sess, RTRS_CLT_CLOSING, NULL)) - queue_work(rtrs_wq, &sess->close_work); + trace_rtrs_clt_close_conns(clt_path); + + if (rtrs_clt_change_state_get_old(clt_path, RTRS_CLT_CLOSING, NULL)) + queue_work(rtrs_wq, &clt_path->close_work); if (wait) - flush_work(&sess->close_work); + flush_work(&clt_path->close_work); } static inline void flag_error_on_conn(struct rtrs_clt_con *con, int cm_err) { if (con->cm_err == 1) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = to_clt_sess(con->c.sess); - if (atomic_dec_and_test(&sess->connected_cnt)) + clt_path = to_clt_path(con->c.path); + if (atomic_dec_and_test(&clt_path->connected_cnt)) - wake_up(&sess->state_wq); + wake_up(&clt_path->state_wq); } con->cm_err = cm_err; } @@ -1949,8 +1973,8 @@ static int rtrs_clt_rdma_cm_handler(struct rdma_cm_id *cm_id, struct rdma_cm_event *ev) { struct rtrs_clt_con *con = cm_id->context; - struct rtrs_sess *s = con->c.sess; - struct rtrs_clt_sess *sess = to_clt_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_clt_path *clt_path = to_clt_path(s); int cm_err = 0; switch (ev->event) { @@ -1968,7 +1992,7 @@ static int rtrs_clt_rdma_cm_handler(struct rdma_cm_id *cm_id, * i.e. wake up without state change, but we set cm_err. */ flag_success_on_conn(con); - wake_up(&sess->state_wq); + wake_up(&clt_path->state_wq); return 0; } break; @@ -1997,7 +2021,7 @@ static int rtrs_clt_rdma_cm_handler(struct rdma_cm_id *cm_id, /* * Device removal is a special case. Queue close and return 0. */ - rtrs_clt_close_conns(sess, false); + rtrs_clt_close_conns(clt_path, false); return 0; default: rtrs_err(s, "Unexpected RDMA CM error (CM event: %s, err: %d)\n", @@ -2020,13 +2044,13 @@ static int rtrs_clt_rdma_cm_handler(struct rdma_cm_id *cm_id, static int create_cm(struct rtrs_clt_con *con) { - struct rtrs_sess *s = con->c.sess; - struct rtrs_clt_sess *sess = to_clt_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_clt_path *clt_path = to_clt_path(s); struct rdma_cm_id *cm_id; int err; cm_id = rdma_create_id(&init_net, rtrs_clt_rdma_cm_handler, con, - sess->s.dst_addr.ss_family == AF_IB ? + clt_path->s.dst_addr.ss_family == AF_IB ? RDMA_PS_IB : RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(cm_id)) { err = PTR_ERR(cm_id); @@ -2042,8 +2066,8 @@ static int create_cm(struct rtrs_clt_con *con) rtrs_err(s, "Set address reuse failed, err: %d\n", err); goto destroy_cm; } - err = rdma_resolve_addr(cm_id, (struct sockaddr *)&sess->s.src_addr, - (struct sockaddr *)&sess->s.dst_addr, + err = rdma_resolve_addr(cm_id, (struct sockaddr *)&clt_path->s.src_addr, + (struct sockaddr *)&clt_path->s.dst_addr, RTRS_CONNECT_TIMEOUT_MS); if (err) { rtrs_err(s, "Failed to resolve address, err: %d\n", err); @@ -2055,8 +2079,8 @@ static int create_cm(struct rtrs_clt_con *con) * or session state was really changed to error by device removal. */ err = wait_event_interruptible_timeout( - sess->state_wq, - con->cm_err || sess->state != RTRS_CLT_CONNECTING, + clt_path->state_wq, + con->cm_err || clt_path->state != RTRS_CLT_CONNECTING, msecs_to_jiffies(RTRS_CONNECT_TIMEOUT_MS)); if (err == 0 || err == -ERESTARTSYS) { if (err == 0) @@ -2068,7 +2092,7 @@ static int create_cm(struct rtrs_clt_con *con) err = con->cm_err; goto errr; } - if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTING) { + if (READ_ONCE(clt_path->state) != RTRS_CLT_CONNECTING) { /* Device removal */ err = -ECONNABORTED; goto errr; @@ -2087,9 +2111,9 @@ destroy_cm: return err; } -static void rtrs_clt_sess_up(struct rtrs_clt_sess *sess) +static void rtrs_clt_path_up(struct rtrs_clt_path *clt_path) { - struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_sess *clt = clt_path->clt; int up; /* @@ -2113,19 +2137,19 @@ static void rtrs_clt_sess_up(struct rtrs_clt_sess *sess) mutex_unlock(&clt->paths_ev_mutex); /* Mark session as established */ - sess->established = true; - sess->reconnect_attempts = 0; - sess->stats->reconnects.successful_cnt++; + clt_path->established = true; + clt_path->reconnect_attempts = 0; + clt_path->stats->reconnects.successful_cnt++; } -static void rtrs_clt_sess_down(struct rtrs_clt_sess *sess) +static void rtrs_clt_path_down(struct rtrs_clt_path *clt_path) { - struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_sess *clt = clt_path->clt; - if (!sess->established) + if (!clt_path->established) return; - sess->established = false; + clt_path->established = false; mutex_lock(&clt->paths_ev_mutex); WARN_ON(!clt->paths_up); if (--clt->paths_up == 0) @@ -2133,19 +2157,19 @@ static void rtrs_clt_sess_down(struct rtrs_clt_sess *sess) mutex_unlock(&clt->paths_ev_mutex); } -static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_sess *sess) +static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_path *clt_path) { struct rtrs_clt_con *con; unsigned int cid; - WARN_ON(READ_ONCE(sess->state) == RTRS_CLT_CONNECTED); + WARN_ON(READ_ONCE(clt_path->state) == RTRS_CLT_CONNECTED); /* * Possible race with rtrs_clt_open(), when DEVICE_REMOVAL comes * exactly in between. Start destroying after it finishes. */ - mutex_lock(&sess->init_mutex); - mutex_unlock(&sess->init_mutex); + mutex_lock(&clt_path->init_mutex); + mutex_unlock(&clt_path->init_mutex); /* * All IO paths must observe !CONNECTED state before we @@ -2153,7 +2177,7 @@ static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_sess *sess) */ synchronize_rcu(); - rtrs_stop_hb(&sess->s); + rtrs_stop_hb(&clt_path->s); /* * The order it utterly crucial: firstly disconnect and complete all @@ -2162,15 +2186,15 @@ static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_sess *sess) * eventually notify upper layer about session disconnection. */ - for (cid = 0; cid < sess->s.con_num; cid++) { - if (!sess->s.con[cid]) + for (cid = 0; cid < clt_path->s.con_num; cid++) { + if (!clt_path->s.con[cid]) break; - con = to_clt_con(sess->s.con[cid]); + con = to_clt_con(clt_path->s.con[cid]); stop_cm(con); } - fail_all_outstanding_reqs(sess); - free_sess_reqs(sess); - rtrs_clt_sess_down(sess); + fail_all_outstanding_reqs(clt_path); + free_path_reqs(clt_path); + rtrs_clt_path_down(clt_path); /* * Wait for graceful shutdown, namely when peer side invokes @@ -2180,13 +2204,14 @@ static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_sess *sess) * since CM does not fire anything. That is fine, we are not in * hurry. */ - wait_event_timeout(sess->state_wq, !atomic_read(&sess->connected_cnt), + wait_event_timeout(clt_path->state_wq, + !atomic_read(&clt_path->connected_cnt), msecs_to_jiffies(RTRS_CONNECT_TIMEOUT_MS)); - for (cid = 0; cid < sess->s.con_num; cid++) { - if (!sess->s.con[cid]) + for (cid = 0; cid < clt_path->s.con_num; cid++) { + if (!clt_path->s.con[cid]) break; - con = to_clt_con(sess->s.con[cid]); + con = to_clt_con(clt_path->s.con[cid]); mutex_lock(&con->con_mutex); destroy_con_cq_qp(con); mutex_unlock(&con->con_mutex); @@ -2195,26 +2220,15 @@ static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_sess *sess) } } -static inline bool xchg_sessions(struct rtrs_clt_sess __rcu **rcu_ppcpu_path, - struct rtrs_clt_sess *sess, - struct rtrs_clt_sess *next) -{ - struct rtrs_clt_sess **ppcpu_path; - - /* Call cmpxchg() without sparse warnings */ - ppcpu_path = (typeof(ppcpu_path))rcu_ppcpu_path; - return sess == cmpxchg(ppcpu_path, sess, next); -} - -static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_sess *sess) +static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_path *clt_path) { - struct rtrs_clt *clt = sess->clt; - struct rtrs_clt_sess *next; + struct rtrs_clt_sess *clt = clt_path->clt; + struct rtrs_clt_path *next; bool wait_for_grace = false; int cpu; mutex_lock(&clt->paths_mutex); - list_del_rcu(&sess->s.entry); + list_del_rcu(&clt_path->s.entry); /* Make sure everybody observes path removal. */ synchronize_rcu(); @@ -2255,8 +2269,7 @@ static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_sess *sess) * removed. If @sess is the last element, then @next is NULL. */ rcu_read_lock(); - next = list_next_or_null_rr_rcu(&clt->paths_list, &sess->s.entry, - typeof(*next), s.entry); + next = rtrs_clt_get_next_path_or_null(&clt->paths_list, clt_path); rcu_read_unlock(); /* @@ -2264,11 +2277,11 @@ static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_sess *sess) * removed, so change the pointer manually. */ for_each_possible_cpu(cpu) { - struct rtrs_clt_sess __rcu **ppcpu_path; + struct rtrs_clt_path __rcu **ppcpu_path; ppcpu_path = per_cpu_ptr(clt->pcpu_path, cpu); if (rcu_dereference_protected(*ppcpu_path, - lockdep_is_held(&clt->paths_mutex)) != sess) + lockdep_is_held(&clt->paths_mutex)) != clt_path) /* * synchronize_rcu() was called just after deleting * entry from the list, thus IO code path cannot @@ -2281,7 +2294,8 @@ static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_sess *sess) * We race with IO code path, which also changes pointer, * thus we have to be careful not to overwrite it. */ - if (xchg_sessions(ppcpu_path, sess, next)) + if (try_cmpxchg((struct rtrs_clt_path **)ppcpu_path, &clt_path, + next)) /* * @ppcpu_path was successfully replaced with @next, * that means that someone could also pick up the @@ -2296,29 +2310,30 @@ static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_sess *sess) mutex_unlock(&clt->paths_mutex); } -static void rtrs_clt_add_path_to_arr(struct rtrs_clt_sess *sess) +static void rtrs_clt_add_path_to_arr(struct rtrs_clt_path *clt_path) { - struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_sess *clt = clt_path->clt; mutex_lock(&clt->paths_mutex); clt->paths_num++; - list_add_tail_rcu(&sess->s.entry, &clt->paths_list); + list_add_tail_rcu(&clt_path->s.entry, &clt->paths_list); mutex_unlock(&clt->paths_mutex); } static void rtrs_clt_close_work(struct work_struct *work) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = container_of(work, struct rtrs_clt_sess, close_work); + clt_path = container_of(work, struct rtrs_clt_path, close_work); - cancel_delayed_work_sync(&sess->reconnect_dwork); - rtrs_clt_stop_and_destroy_conns(sess); - rtrs_clt_change_state_get_old(sess, RTRS_CLT_CLOSED, NULL); + cancel_work_sync(&clt_path->err_recovery_work); + cancel_delayed_work_sync(&clt_path->reconnect_dwork); + rtrs_clt_stop_and_destroy_conns(clt_path); + rtrs_clt_change_state_get_old(clt_path, RTRS_CLT_CLOSED, NULL); } -static int init_conns(struct rtrs_clt_sess *sess) +static int init_conns(struct rtrs_clt_path *clt_path) { unsigned int cid; int err; @@ -2328,31 +2343,31 @@ static int init_conns(struct rtrs_clt_sess *sess) * to avoid clashes with previous sessions not yet closed * sessions on a server side. */ - sess->s.recon_cnt++; + clt_path->s.recon_cnt++; /* Establish all RDMA connections */ - for (cid = 0; cid < sess->s.con_num; cid++) { - err = create_con(sess, cid); + for (cid = 0; cid < clt_path->s.con_num; cid++) { + err = create_con(clt_path, cid); if (err) goto destroy; - err = create_cm(to_clt_con(sess->s.con[cid])); + err = create_cm(to_clt_con(clt_path->s.con[cid])); if (err) { - destroy_con(to_clt_con(sess->s.con[cid])); + destroy_con(to_clt_con(clt_path->s.con[cid])); goto destroy; } } - err = alloc_sess_reqs(sess); + err = alloc_path_reqs(clt_path); if (err) goto destroy; - rtrs_start_hb(&sess->s); + rtrs_start_hb(&clt_path->s); return 0; destroy: while (cid--) { - struct rtrs_clt_con *con = to_clt_con(sess->s.con[cid]); + struct rtrs_clt_con *con = to_clt_con(clt_path->s.con[cid]); stop_cm(con); @@ -2367,7 +2382,7 @@ destroy: * doing rdma_resolve_addr(), switch to CONNECTION_ERR state * manually to keep reconnecting. */ - rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING_ERR, NULL); + rtrs_clt_change_state_get_old(clt_path, RTRS_CLT_CONNECTING_ERR, NULL); return err; } @@ -2375,31 +2390,32 @@ destroy: static void rtrs_clt_info_req_done(struct ib_cq *cq, struct ib_wc *wc) { struct rtrs_clt_con *con = to_clt_con(wc->qp->qp_context); - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); struct rtrs_iu *iu; iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); - rtrs_iu_free(iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(iu, clt_path->s.dev->ib_dev, 1); if (wc->status != IB_WC_SUCCESS) { - rtrs_err(sess->clt, "Sess info request send failed: %s\n", + rtrs_err(clt_path->clt, "Path info request send failed: %s\n", ib_wc_status_msg(wc->status)); - rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING_ERR, NULL); + rtrs_clt_change_state_get_old(clt_path, RTRS_CLT_CONNECTING_ERR, NULL); return; } rtrs_clt_update_wc_stats(con); } -static int process_info_rsp(struct rtrs_clt_sess *sess, +static int process_info_rsp(struct rtrs_clt_path *clt_path, const struct rtrs_msg_info_rsp *msg) { unsigned int sg_cnt, total_len; int i, sgi; sg_cnt = le16_to_cpu(msg->sg_cnt); - if (!sg_cnt || (sess->queue_depth % sg_cnt)) { - rtrs_err(sess->clt, "Incorrect sg_cnt %d, is not multiple\n", + if (!sg_cnt || (clt_path->queue_depth % sg_cnt)) { + rtrs_err(clt_path->clt, + "Incorrect sg_cnt %d, is not multiple\n", sg_cnt); return -EINVAL; } @@ -2408,15 +2424,15 @@ static int process_info_rsp(struct rtrs_clt_sess *sess, * Check if IB immediate data size is enough to hold the mem_id and * the offset inside the memory chunk. */ - if ((ilog2(sg_cnt - 1) + 1) + (ilog2(sess->chunk_size - 1) + 1) > + if ((ilog2(sg_cnt - 1) + 1) + (ilog2(clt_path->chunk_size - 1) + 1) > MAX_IMM_PAYL_BITS) { - rtrs_err(sess->clt, + rtrs_err(clt_path->clt, "RDMA immediate size (%db) not enough to encode %d buffers of size %dB\n", - MAX_IMM_PAYL_BITS, sg_cnt, sess->chunk_size); + MAX_IMM_PAYL_BITS, sg_cnt, clt_path->chunk_size); return -EINVAL; } total_len = 0; - for (sgi = 0, i = 0; sgi < sg_cnt && i < sess->queue_depth; sgi++) { + for (sgi = 0, i = 0; sgi < sg_cnt && i < clt_path->queue_depth; sgi++) { const struct rtrs_sg_desc *desc = &msg->desc[sgi]; u32 len, rkey; u64 addr; @@ -2427,26 +2443,28 @@ static int process_info_rsp(struct rtrs_clt_sess *sess, total_len += len; - if (!len || (len % sess->chunk_size)) { - rtrs_err(sess->clt, "Incorrect [%d].len %d\n", sgi, + if (!len || (len % clt_path->chunk_size)) { + rtrs_err(clt_path->clt, "Incorrect [%d].len %d\n", + sgi, len); return -EINVAL; } - for ( ; len && i < sess->queue_depth; i++) { - sess->rbufs[i].addr = addr; - sess->rbufs[i].rkey = rkey; + for ( ; len && i < clt_path->queue_depth; i++) { + clt_path->rbufs[i].addr = addr; + clt_path->rbufs[i].rkey = rkey; - len -= sess->chunk_size; - addr += sess->chunk_size; + len -= clt_path->chunk_size; + addr += clt_path->chunk_size; } } /* Sanity check */ - if (sgi != sg_cnt || i != sess->queue_depth) { - rtrs_err(sess->clt, "Incorrect sg vector, not fully mapped\n"); + if (sgi != sg_cnt || i != clt_path->queue_depth) { + rtrs_err(clt_path->clt, + "Incorrect sg vector, not fully mapped\n"); return -EINVAL; } - if (total_len != sess->chunk_size * sess->queue_depth) { - rtrs_err(sess->clt, "Incorrect total_len %d\n", total_len); + if (total_len != clt_path->chunk_size * clt_path->queue_depth) { + rtrs_err(clt_path->clt, "Incorrect total_len %d\n", total_len); return -EINVAL; } @@ -2456,7 +2474,7 @@ static int process_info_rsp(struct rtrs_clt_sess *sess, static void rtrs_clt_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc) { struct rtrs_clt_con *con = to_clt_con(wc->qp->qp_context); - struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_path *clt_path = to_clt_path(con->c.path); struct rtrs_msg_info_rsp *msg; enum rtrs_clt_state state; struct rtrs_iu *iu; @@ -2468,37 +2486,37 @@ static void rtrs_clt_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc) WARN_ON(con->c.cid); iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); if (wc->status != IB_WC_SUCCESS) { - rtrs_err(sess->clt, "Sess info response recv failed: %s\n", + rtrs_err(clt_path->clt, "Path info response recv failed: %s\n", ib_wc_status_msg(wc->status)); goto out; } WARN_ON(wc->opcode != IB_WC_RECV); if (wc->byte_len < sizeof(*msg)) { - rtrs_err(sess->clt, "Sess info response is malformed: size %d\n", + rtrs_err(clt_path->clt, "Path info response is malformed: size %d\n", wc->byte_len); goto out; } - ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr, + ib_dma_sync_single_for_cpu(clt_path->s.dev->ib_dev, iu->dma_addr, iu->size, DMA_FROM_DEVICE); msg = iu->buf; if (le16_to_cpu(msg->type) != RTRS_MSG_INFO_RSP) { - rtrs_err(sess->clt, "Sess info response is malformed: type %d\n", + rtrs_err(clt_path->clt, "Path info response is malformed: type %d\n", le16_to_cpu(msg->type)); goto out; } rx_sz = sizeof(*msg); rx_sz += sizeof(msg->desc[0]) * le16_to_cpu(msg->sg_cnt); if (wc->byte_len < rx_sz) { - rtrs_err(sess->clt, "Sess info response is malformed: size %d\n", + rtrs_err(clt_path->clt, "Path info response is malformed: size %d\n", wc->byte_len); goto out; } - err = process_info_rsp(sess, msg); + err = process_info_rsp(clt_path, msg); if (err) goto out; - err = post_recv_sess(sess); + err = post_recv_path(clt_path); if (err) goto out; @@ -2506,25 +2524,25 @@ static void rtrs_clt_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc) out: rtrs_clt_update_wc_stats(con); - rtrs_iu_free(iu, sess->s.dev->ib_dev, 1); - rtrs_clt_change_state_get_old(sess, state, NULL); + rtrs_iu_free(iu, clt_path->s.dev->ib_dev, 1); + rtrs_clt_change_state_get_old(clt_path, state, NULL); } -static int rtrs_send_sess_info(struct rtrs_clt_sess *sess) +static int rtrs_send_path_info(struct rtrs_clt_path *clt_path) { - struct rtrs_clt_con *usr_con = to_clt_con(sess->s.con[0]); + struct rtrs_clt_con *usr_con = to_clt_con(clt_path->s.con[0]); struct rtrs_msg_info_req *msg; struct rtrs_iu *tx_iu, *rx_iu; size_t rx_sz; int err; rx_sz = sizeof(struct rtrs_msg_info_rsp); - rx_sz += sizeof(struct rtrs_sg_desc) * sess->queue_depth; + rx_sz += sizeof(struct rtrs_sg_desc) * clt_path->queue_depth; tx_iu = rtrs_iu_alloc(1, sizeof(struct rtrs_msg_info_req), GFP_KERNEL, - sess->s.dev->ib_dev, DMA_TO_DEVICE, + clt_path->s.dev->ib_dev, DMA_TO_DEVICE, rtrs_clt_info_req_done); - rx_iu = rtrs_iu_alloc(1, rx_sz, GFP_KERNEL, sess->s.dev->ib_dev, + rx_iu = rtrs_iu_alloc(1, rx_sz, GFP_KERNEL, clt_path->s.dev->ib_dev, DMA_FROM_DEVICE, rtrs_clt_info_rsp_done); if (!tx_iu || !rx_iu) { err = -ENOMEM; @@ -2533,33 +2551,34 @@ static int rtrs_send_sess_info(struct rtrs_clt_sess *sess) /* Prepare for getting info response */ err = rtrs_iu_post_recv(&usr_con->c, rx_iu); if (err) { - rtrs_err(sess->clt, "rtrs_iu_post_recv(), err: %d\n", err); + rtrs_err(clt_path->clt, "rtrs_iu_post_recv(), err: %d\n", err); goto out; } rx_iu = NULL; msg = tx_iu->buf; msg->type = cpu_to_le16(RTRS_MSG_INFO_REQ); - memcpy(msg->sessname, sess->s.sessname, sizeof(msg->sessname)); + memcpy(msg->pathname, clt_path->s.sessname, sizeof(msg->pathname)); - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, tx_iu->dma_addr, + ib_dma_sync_single_for_device(clt_path->s.dev->ib_dev, + tx_iu->dma_addr, tx_iu->size, DMA_TO_DEVICE); /* Send info request */ err = rtrs_iu_post_send(&usr_con->c, tx_iu, sizeof(*msg), NULL); if (err) { - rtrs_err(sess->clt, "rtrs_iu_post_send(), err: %d\n", err); + rtrs_err(clt_path->clt, "rtrs_iu_post_send(), err: %d\n", err); goto out; } tx_iu = NULL; /* Wait for state change */ - wait_event_interruptible_timeout(sess->state_wq, - sess->state != RTRS_CLT_CONNECTING, + wait_event_interruptible_timeout(clt_path->state_wq, + clt_path->state != RTRS_CLT_CONNECTING, msecs_to_jiffies( RTRS_CONNECT_TIMEOUT_MS)); - if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTED) { - if (READ_ONCE(sess->state) == RTRS_CLT_CONNECTING_ERR) + if (READ_ONCE(clt_path->state) != RTRS_CLT_CONNECTED) { + if (READ_ONCE(clt_path->state) == RTRS_CLT_CONNECTING_ERR) err = -ECONNRESET; else err = -ETIMEDOUT; @@ -2567,82 +2586,81 @@ static int rtrs_send_sess_info(struct rtrs_clt_sess *sess) out: if (tx_iu) - rtrs_iu_free(tx_iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(tx_iu, clt_path->s.dev->ib_dev, 1); if (rx_iu) - rtrs_iu_free(rx_iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(rx_iu, clt_path->s.dev->ib_dev, 1); if (err) /* If we've never taken async path because of malloc problems */ - rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING_ERR, NULL); + rtrs_clt_change_state_get_old(clt_path, + RTRS_CLT_CONNECTING_ERR, NULL); return err; } /** - * init_sess() - establishes all session connections and does handshake - * @sess: client session. + * init_path() - establishes all path connections and does handshake + * @clt_path: client path. * In case of error full close or reconnect procedure should be taken, * because reconnect or close async works can be started. */ -static int init_sess(struct rtrs_clt_sess *sess) +static int init_path(struct rtrs_clt_path *clt_path) { int err; char str[NAME_MAX]; struct rtrs_addr path = { - .src = &sess->s.src_addr, - .dst = &sess->s.dst_addr, + .src = &clt_path->s.src_addr, + .dst = &clt_path->s.dst_addr, }; rtrs_addr_to_str(&path, str, sizeof(str)); - mutex_lock(&sess->init_mutex); - err = init_conns(sess); + mutex_lock(&clt_path->init_mutex); + err = init_conns(clt_path); if (err) { - rtrs_err(sess->clt, + rtrs_err(clt_path->clt, "init_conns() failed: err=%d path=%s [%s:%u]\n", err, - str, sess->hca_name, sess->hca_port); + str, clt_path->hca_name, clt_path->hca_port); goto out; } - err = rtrs_send_sess_info(sess); + err = rtrs_send_path_info(clt_path); if (err) { - rtrs_err( - sess->clt, - "rtrs_send_sess_info() failed: err=%d path=%s [%s:%u]\n", - err, str, sess->hca_name, sess->hca_port); + rtrs_err(clt_path->clt, + "rtrs_send_path_info() failed: err=%d path=%s [%s:%u]\n", + err, str, clt_path->hca_name, clt_path->hca_port); goto out; } - rtrs_clt_sess_up(sess); + rtrs_clt_path_up(clt_path); out: - mutex_unlock(&sess->init_mutex); + mutex_unlock(&clt_path->init_mutex); return err; } static void rtrs_clt_reconnect_work(struct work_struct *work) { - struct rtrs_clt_sess *sess; - struct rtrs_clt *clt; - unsigned int delay_ms; + struct rtrs_clt_path *clt_path; + struct rtrs_clt_sess *clt; int err; - sess = container_of(to_delayed_work(work), struct rtrs_clt_sess, - reconnect_dwork); - clt = sess->clt; + clt_path = container_of(to_delayed_work(work), struct rtrs_clt_path, + reconnect_dwork); + clt = clt_path->clt; - if (READ_ONCE(sess->state) != RTRS_CLT_RECONNECTING) + trace_rtrs_clt_reconnect_work(clt_path); + + if (READ_ONCE(clt_path->state) != RTRS_CLT_RECONNECTING) return; - if (sess->reconnect_attempts >= clt->max_reconnect_attempts) { - /* Close a session completely if max attempts is reached */ - rtrs_clt_close_conns(sess, false); + if (clt_path->reconnect_attempts >= clt->max_reconnect_attempts) { + /* Close a path completely if max attempts is reached */ + rtrs_clt_close_conns(clt_path, false); return; } - sess->reconnect_attempts++; + clt_path->reconnect_attempts++; - /* Stop everything */ - rtrs_clt_stop_and_destroy_conns(sess); msleep(RTRS_RECONNECT_BACKOFF); - if (rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING, NULL)) { - err = init_sess(sess); + if (rtrs_clt_change_state_get_old(clt_path, RTRS_CLT_CONNECTING, NULL)) { + err = init_path(clt_path); if (err) goto reconnect_again; } @@ -2650,31 +2668,30 @@ static void rtrs_clt_reconnect_work(struct work_struct *work) return; reconnect_again: - if (rtrs_clt_change_state_get_old(sess, RTRS_CLT_RECONNECTING, NULL)) { - sess->stats->reconnects.fail_cnt++; - delay_ms = clt->reconnect_delay_sec * 1000; - queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, - msecs_to_jiffies(delay_ms + - prandom_u32() % - RTRS_RECONNECT_SEED)); + if (rtrs_clt_change_state_get_old(clt_path, RTRS_CLT_RECONNECTING, NULL)) { + clt_path->stats->reconnects.fail_cnt++; + queue_work(rtrs_wq, &clt_path->err_recovery_work); } } static void rtrs_clt_dev_release(struct device *dev) { - struct rtrs_clt *clt = container_of(dev, struct rtrs_clt, dev); + struct rtrs_clt_sess *clt = container_of(dev, struct rtrs_clt_sess, + dev); + mutex_destroy(&clt->paths_ev_mutex); + mutex_destroy(&clt->paths_mutex); kfree(clt); } -static struct rtrs_clt *alloc_clt(const char *sessname, size_t paths_num, +static struct rtrs_clt_sess *alloc_clt(const char *sessname, size_t paths_num, u16 port, size_t pdu_sz, void *priv, void (*link_ev)(void *priv, enum rtrs_clt_link_ev ev), unsigned int reconnect_delay_sec, unsigned int max_reconnect_attempts) { - struct rtrs_clt *clt; + struct rtrs_clt_sess *clt; int err; if (!paths_num || paths_num > MAX_PATHS_NUM) @@ -2693,6 +2710,8 @@ static struct rtrs_clt *alloc_clt(const char *sessname, size_t paths_num, return ERR_PTR(-ENOMEM); } + clt->dev.class = rtrs_clt_dev_class; + clt->dev.release = rtrs_clt_dev_release; uuid_gen(&clt->paths_uuid); INIT_LIST_HEAD_RCU(&clt->paths_list); clt->paths_num = paths_num; @@ -2709,60 +2728,58 @@ static struct rtrs_clt *alloc_clt(const char *sessname, size_t paths_num, init_waitqueue_head(&clt->permits_wait); mutex_init(&clt->paths_ev_mutex); mutex_init(&clt->paths_mutex); + device_initialize(&clt->dev); - clt->dev.class = rtrs_clt_dev_class; - clt->dev.release = rtrs_clt_dev_release; err = dev_set_name(&clt->dev, "%s", sessname); if (err) - goto err; + goto err_put; + /* * Suppress user space notification until * sysfs files are created */ dev_set_uevent_suppress(&clt->dev, true); - err = device_register(&clt->dev); - if (err) { - put_device(&clt->dev); - goto err; - } + err = device_add(&clt->dev); + if (err) + goto err_put; clt->kobj_paths = kobject_create_and_add("paths", &clt->dev.kobj); if (!clt->kobj_paths) { err = -ENOMEM; - goto err_dev; + goto err_del; } err = rtrs_clt_create_sysfs_root_files(clt); if (err) { kobject_del(clt->kobj_paths); kobject_put(clt->kobj_paths); - goto err_dev; + goto err_del; } dev_set_uevent_suppress(&clt->dev, false); kobject_uevent(&clt->dev.kobj, KOBJ_ADD); return clt; -err_dev: - device_unregister(&clt->dev); -err: +err_del: + device_del(&clt->dev); +err_put: free_percpu(clt->pcpu_path); - kfree(clt); + put_device(&clt->dev); return ERR_PTR(err); } -static void free_clt(struct rtrs_clt *clt) +static void free_clt(struct rtrs_clt_sess *clt) { - free_permits(clt); free_percpu(clt->pcpu_path); - mutex_destroy(&clt->paths_ev_mutex); - mutex_destroy(&clt->paths_mutex); - /* release callback will free clt in last put */ + + /* + * release callback will free clt and destroy mutexes in last put + */ device_unregister(&clt->dev); } /** - * rtrs_clt_open() - Open a session to an RTRS server + * rtrs_clt_open() - Open a path to an RTRS server * @ops: holds the link event callback and the private pointer. - * @sessname: name of the session + * @pathname: name of the path to an RTRS server * @paths: Paths to be established defined by their src and dst addresses * @paths_num: Number of elements in the @paths array * @port: port to be used by the RTRS session @@ -2777,24 +2794,24 @@ static void free_clt(struct rtrs_clt *clt) * * Return a valid pointer on success otherwise PTR_ERR. */ -struct rtrs_clt *rtrs_clt_open(struct rtrs_clt_ops *ops, - const char *sessname, +struct rtrs_clt_sess *rtrs_clt_open(struct rtrs_clt_ops *ops, + const char *pathname, const struct rtrs_addr *paths, size_t paths_num, u16 port, size_t pdu_sz, u8 reconnect_delay_sec, s16 max_reconnect_attempts, u32 nr_poll_queues) { - struct rtrs_clt_sess *sess, *tmp; - struct rtrs_clt *clt; + struct rtrs_clt_path *clt_path, *tmp; + struct rtrs_clt_sess *clt; int err, i; - if (strchr(sessname, '/') || strchr(sessname, '.')) { - pr_err("sessname cannot contain / and .\n"); + if (strchr(pathname, '/') || strchr(pathname, '.')) { + pr_err("pathname cannot contain / and .\n"); err = -EINVAL; goto out; } - clt = alloc_clt(sessname, paths_num, port, pdu_sz, ops->priv, + clt = alloc_clt(pathname, paths_num, port, pdu_sz, ops->priv, ops->link_ev, reconnect_delay_sec, max_reconnect_attempts); @@ -2803,49 +2820,49 @@ struct rtrs_clt *rtrs_clt_open(struct rtrs_clt_ops *ops, goto out; } for (i = 0; i < paths_num; i++) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; - sess = alloc_sess(clt, &paths[i], nr_cpu_ids, + clt_path = alloc_path(clt, &paths[i], nr_cpu_ids, nr_poll_queues); - if (IS_ERR(sess)) { - err = PTR_ERR(sess); - goto close_all_sess; + if (IS_ERR(clt_path)) { + err = PTR_ERR(clt_path); + goto close_all_path; } if (!i) - sess->for_new_clt = 1; - list_add_tail_rcu(&sess->s.entry, &clt->paths_list); + clt_path->for_new_clt = 1; + list_add_tail_rcu(&clt_path->s.entry, &clt->paths_list); - err = init_sess(sess); + err = init_path(clt_path); if (err) { - list_del_rcu(&sess->s.entry); - rtrs_clt_close_conns(sess, true); - free_percpu(sess->stats->pcpu_stats); - kfree(sess->stats); - free_sess(sess); - goto close_all_sess; + list_del_rcu(&clt_path->s.entry); + rtrs_clt_close_conns(clt_path, true); + free_percpu(clt_path->stats->pcpu_stats); + kfree(clt_path->stats); + free_path(clt_path); + goto close_all_path; } - err = rtrs_clt_create_sess_files(sess); + err = rtrs_clt_create_path_files(clt_path); if (err) { - list_del_rcu(&sess->s.entry); - rtrs_clt_close_conns(sess, true); - free_percpu(sess->stats->pcpu_stats); - kfree(sess->stats); - free_sess(sess); - goto close_all_sess; + list_del_rcu(&clt_path->s.entry); + rtrs_clt_close_conns(clt_path, true); + free_percpu(clt_path->stats->pcpu_stats); + kfree(clt_path->stats); + free_path(clt_path); + goto close_all_path; } } err = alloc_permits(clt); if (err) - goto close_all_sess; + goto close_all_path; return clt; -close_all_sess: - list_for_each_entry_safe(sess, tmp, &clt->paths_list, s.entry) { - rtrs_clt_destroy_sess_files(sess, NULL); - rtrs_clt_close_conns(sess, true); - kobject_put(&sess->kobj); +close_all_path: + list_for_each_entry_safe(clt_path, tmp, &clt->paths_list, s.entry) { + rtrs_clt_destroy_path_files(clt_path, NULL); + rtrs_clt_close_conns(clt_path, true); + kobject_put(&clt_path->kobj); } rtrs_clt_destroy_sysfs_root(clt); free_clt(clt); @@ -2856,37 +2873,40 @@ out: EXPORT_SYMBOL(rtrs_clt_open); /** - * rtrs_clt_close() - Close a session + * rtrs_clt_close() - Close a path * @clt: Session handle. Session is freed upon return. */ -void rtrs_clt_close(struct rtrs_clt *clt) +void rtrs_clt_close(struct rtrs_clt_sess *clt) { - struct rtrs_clt_sess *sess, *tmp; + struct rtrs_clt_path *clt_path, *tmp; /* Firstly forbid sysfs access */ rtrs_clt_destroy_sysfs_root(clt); /* Now it is safe to iterate over all paths without locks */ - list_for_each_entry_safe(sess, tmp, &clt->paths_list, s.entry) { - rtrs_clt_close_conns(sess, true); - rtrs_clt_destroy_sess_files(sess, NULL); - kobject_put(&sess->kobj); + list_for_each_entry_safe(clt_path, tmp, &clt->paths_list, s.entry) { + rtrs_clt_close_conns(clt_path, true); + rtrs_clt_destroy_path_files(clt_path, NULL); + kobject_put(&clt_path->kobj); } + free_permits(clt); free_clt(clt); } EXPORT_SYMBOL(rtrs_clt_close); -int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_sess *sess) +int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_path *clt_path) { enum rtrs_clt_state old_state; int err = -EBUSY; bool changed; - changed = rtrs_clt_change_state_get_old(sess, RTRS_CLT_RECONNECTING, + changed = rtrs_clt_change_state_get_old(clt_path, + RTRS_CLT_RECONNECTING, &old_state); if (changed) { - sess->reconnect_attempts = 0; - queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, 0); + clt_path->reconnect_attempts = 0; + rtrs_clt_stop_and_destroy_conns(clt_path); + queue_delayed_work(rtrs_wq, &clt_path->reconnect_dwork, 0); } if (changed || old_state == RTRS_CLT_RECONNECTING) { /* @@ -2894,15 +2914,15 @@ int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_sess *sess) * execution, so do the flush if we have queued something * right now or work is pending. */ - flush_delayed_work(&sess->reconnect_dwork); - err = (READ_ONCE(sess->state) == + flush_delayed_work(&clt_path->reconnect_dwork); + err = (READ_ONCE(clt_path->state) == RTRS_CLT_CONNECTED ? 0 : -ENOTCONN); } return err; } -int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_sess *sess, +int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_path *clt_path, const struct attribute *sysfs_self) { enum rtrs_clt_state old_state; @@ -2918,27 +2938,27 @@ int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_sess *sess, * removing the path. */ do { - rtrs_clt_close_conns(sess, true); - changed = rtrs_clt_change_state_get_old(sess, + rtrs_clt_close_conns(clt_path, true); + changed = rtrs_clt_change_state_get_old(clt_path, RTRS_CLT_DEAD, &old_state); } while (!changed && old_state != RTRS_CLT_DEAD); if (changed) { - rtrs_clt_remove_path_from_arr(sess); - rtrs_clt_destroy_sess_files(sess, sysfs_self); - kobject_put(&sess->kobj); + rtrs_clt_remove_path_from_arr(clt_path); + rtrs_clt_destroy_path_files(clt_path, sysfs_self); + kobject_put(&clt_path->kobj); } return 0; } -void rtrs_clt_set_max_reconnect_attempts(struct rtrs_clt *clt, int value) +void rtrs_clt_set_max_reconnect_attempts(struct rtrs_clt_sess *clt, int value) { clt->max_reconnect_attempts = (unsigned int)value; } -int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt *clt) +int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt_sess *clt) { return (int)clt->max_reconnect_attempts; } @@ -2968,12 +2988,12 @@ int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt *clt) * On dir=WRITE rtrs client will rdma write data in sg to server side. */ int rtrs_clt_request(int dir, struct rtrs_clt_req_ops *ops, - struct rtrs_clt *clt, struct rtrs_permit *permit, - const struct kvec *vec, size_t nr, size_t data_len, - struct scatterlist *sg, unsigned int sg_cnt) + struct rtrs_clt_sess *clt, struct rtrs_permit *permit, + const struct kvec *vec, size_t nr, size_t data_len, + struct scatterlist *sg, unsigned int sg_cnt) { struct rtrs_clt_io_req *req; - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; enum dma_data_direction dma_dir; int err = -ECONNABORTED, i; @@ -2995,19 +3015,19 @@ int rtrs_clt_request(int dir, struct rtrs_clt_req_ops *ops, rcu_read_lock(); for (path_it_init(&it, clt); - (sess = it.next_path(&it)) && it.i < it.clt->paths_num; it.i++) { - if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTED) + (clt_path = it.next_path(&it)) && it.i < it.clt->paths_num; it.i++) { + if (READ_ONCE(clt_path->state) != RTRS_CLT_CONNECTED) continue; - if (usr_len + hdr_len > sess->max_hdr_size) { - rtrs_wrn_rl(sess->clt, + if (usr_len + hdr_len > clt_path->max_hdr_size) { + rtrs_wrn_rl(clt_path->clt, "%s request failed, user message size is %zu and header length %zu, but max size is %u\n", dir == READ ? "Read" : "Write", - usr_len, hdr_len, sess->max_hdr_size); + usr_len, hdr_len, clt_path->max_hdr_size); err = -EMSGSIZE; break; } - req = rtrs_clt_get_req(sess, ops->conf_fn, permit, ops->priv, + req = rtrs_clt_get_req(clt_path, ops->conf_fn, permit, ops->priv, vec, usr_len, sg, sg_cnt, data_len, dma_dir); if (dir == READ) @@ -3028,21 +3048,21 @@ int rtrs_clt_request(int dir, struct rtrs_clt_req_ops *ops, } EXPORT_SYMBOL(rtrs_clt_request); -int rtrs_clt_rdma_cq_direct(struct rtrs_clt *clt, unsigned int index) +int rtrs_clt_rdma_cq_direct(struct rtrs_clt_sess *clt, unsigned int index) { /* If no path, return -1 for block layer not to try again */ int cnt = -1; struct rtrs_con *con; - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; struct path_it it; rcu_read_lock(); for (path_it_init(&it, clt); - (sess = it.next_path(&it)) && it.i < it.clt->paths_num; it.i++) { - if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTED) + (clt_path = it.next_path(&it)) && it.i < it.clt->paths_num; it.i++) { + if (READ_ONCE(clt_path->state) != RTRS_CLT_CONNECTED) continue; - con = sess->s.con[index + 1]; + con = clt_path->s.con[index + 1]; cnt = ib_process_cq_direct(con->cq, -1); if (cnt) break; @@ -3062,7 +3082,7 @@ EXPORT_SYMBOL(rtrs_clt_rdma_cq_direct); * 0 on success * -ECOMM no connection to the server */ -int rtrs_clt_query(struct rtrs_clt *clt, struct rtrs_attrs *attr) +int rtrs_clt_query(struct rtrs_clt_sess *clt, struct rtrs_attrs *attr) { if (!rtrs_clt_is_connected(clt)) return -ECOMM; @@ -3077,15 +3097,15 @@ int rtrs_clt_query(struct rtrs_clt *clt, struct rtrs_attrs *attr) } EXPORT_SYMBOL(rtrs_clt_query); -int rtrs_clt_create_path_from_sysfs(struct rtrs_clt *clt, +int rtrs_clt_create_path_from_sysfs(struct rtrs_clt_sess *clt, struct rtrs_addr *addr) { - struct rtrs_clt_sess *sess; + struct rtrs_clt_path *clt_path; int err; - sess = alloc_sess(clt, addr, nr_cpu_ids, 0); - if (IS_ERR(sess)) - return PTR_ERR(sess); + clt_path = alloc_path(clt, addr, nr_cpu_ids, 0); + if (IS_ERR(clt_path)) + return PTR_ERR(clt_path); mutex_lock(&clt->paths_mutex); if (clt->paths_num == 0) { @@ -3094,7 +3114,7 @@ int rtrs_clt_create_path_from_sysfs(struct rtrs_clt *clt, * the addition of the first path is like a new session for * the storage server */ - sess->for_new_clt = 1; + clt_path->for_new_clt = 1; } mutex_unlock(&clt->paths_mutex); @@ -3104,24 +3124,24 @@ int rtrs_clt_create_path_from_sysfs(struct rtrs_clt *clt, * IO will never grab it. Also it is very important to add * path before init, since init fires LINK_CONNECTED event. */ - rtrs_clt_add_path_to_arr(sess); + rtrs_clt_add_path_to_arr(clt_path); - err = init_sess(sess); + err = init_path(clt_path); if (err) - goto close_sess; + goto close_path; - err = rtrs_clt_create_sess_files(sess); + err = rtrs_clt_create_path_files(clt_path); if (err) - goto close_sess; + goto close_path; return 0; -close_sess: - rtrs_clt_remove_path_from_arr(sess); - rtrs_clt_close_conns(sess, true); - free_percpu(sess->stats->pcpu_stats); - kfree(sess->stats); - free_sess(sess); +close_path: + rtrs_clt_remove_path_from_arr(clt_path); + rtrs_clt_close_conns(clt_path, true); + free_percpu(clt_path->stats->pcpu_stats); + kfree(clt_path->stats); + free_path(clt_path); return err; } diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt.h b/drivers/infiniband/ulp/rtrs/rtrs-clt.h index 9afffccff973..f848c0392d98 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-clt.h +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt.h @@ -124,9 +124,9 @@ struct rtrs_rbuf { u32 rkey; }; -struct rtrs_clt_sess { - struct rtrs_sess s; - struct rtrs_clt *clt; +struct rtrs_clt_path { + struct rtrs_path s; + struct rtrs_clt_sess *clt; wait_queue_head_t state_wq; enum rtrs_clt_state state; atomic_t connected_cnt; @@ -134,6 +134,7 @@ struct rtrs_clt_sess { struct rtrs_clt_io_req *reqs; struct delayed_work reconnect_dwork; struct work_struct close_work; + struct work_struct err_recovery_work; unsigned int reconnect_attempts; bool established; struct rtrs_rbuf *rbufs; @@ -153,10 +154,10 @@ struct rtrs_clt_sess { *mp_skip_entry; }; -struct rtrs_clt { +struct rtrs_clt_sess { struct list_head paths_list; /* rcu protected list */ size_t paths_num; - struct rtrs_clt_sess + struct rtrs_clt_path __rcu * __percpu *pcpu_path; uuid_t paths_uuid; int paths_up; @@ -186,31 +187,32 @@ static inline struct rtrs_clt_con *to_clt_con(struct rtrs_con *c) return container_of(c, struct rtrs_clt_con, c); } -static inline struct rtrs_clt_sess *to_clt_sess(struct rtrs_sess *s) +static inline struct rtrs_clt_path *to_clt_path(struct rtrs_path *s) { - return container_of(s, struct rtrs_clt_sess, s); + return container_of(s, struct rtrs_clt_path, s); } -static inline int permit_size(struct rtrs_clt *clt) +static inline int permit_size(struct rtrs_clt_sess *clt) { return sizeof(struct rtrs_permit) + clt->pdu_sz; } -static inline struct rtrs_permit *get_permit(struct rtrs_clt *clt, int idx) +static inline struct rtrs_permit *get_permit(struct rtrs_clt_sess *clt, + int idx) { return (struct rtrs_permit *)(clt->permits + permit_size(clt) * idx); } -int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_sess *sess); -void rtrs_clt_close_conns(struct rtrs_clt_sess *sess, bool wait); -int rtrs_clt_create_path_from_sysfs(struct rtrs_clt *clt, +int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_path *path); +void rtrs_clt_close_conns(struct rtrs_clt_path *clt_path, bool wait); +int rtrs_clt_create_path_from_sysfs(struct rtrs_clt_sess *clt, struct rtrs_addr *addr); -int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_sess *sess, +int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_path *path, const struct attribute *sysfs_self); -void rtrs_clt_set_max_reconnect_attempts(struct rtrs_clt *clt, int value); -int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt *clt); -void free_sess(struct rtrs_clt_sess *sess); +void rtrs_clt_set_max_reconnect_attempts(struct rtrs_clt_sess *clt, int value); +int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt_sess *clt); +void free_path(struct rtrs_clt_path *clt_path); /* rtrs-clt-stats.c */ @@ -239,11 +241,11 @@ ssize_t rtrs_clt_reset_all_help(struct rtrs_clt_stats *stats, /* rtrs-clt-sysfs.c */ -int rtrs_clt_create_sysfs_root_files(struct rtrs_clt *clt); -void rtrs_clt_destroy_sysfs_root(struct rtrs_clt *clt); +int rtrs_clt_create_sysfs_root_files(struct rtrs_clt_sess *clt); +void rtrs_clt_destroy_sysfs_root(struct rtrs_clt_sess *clt); -int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess); -void rtrs_clt_destroy_sess_files(struct rtrs_clt_sess *sess, +int rtrs_clt_create_path_files(struct rtrs_clt_path *clt_path); +void rtrs_clt_destroy_path_files(struct rtrs_clt_path *clt_path, const struct attribute *sysfs_self); #endif /* RTRS_CLT_H */ diff --git a/drivers/infiniband/ulp/rtrs/rtrs-pri.h b/drivers/infiniband/ulp/rtrs/rtrs-pri.h index 78eac9a4f703..a2420eecaf5a 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-pri.h +++ b/drivers/infiniband/ulp/rtrs/rtrs-pri.h @@ -23,6 +23,16 @@ #define RTRS_PROTO_VER_STRING __stringify(RTRS_PROTO_VER_MAJOR) "." \ __stringify(RTRS_PROTO_VER_MINOR) +/* + * Max IB immediate data size is 2^28 (MAX_IMM_PAYL_BITS) + * and the minimum chunk size is 4096 (2^12). + * So the maximum sess_queue_depth is 65535 (2^16 - 1) in theory + * since queue_depth in rtrs_msg_conn_rsp is defined as le16. + * Therefore the pratical max value of sess_queue_depth is + * somewhere between 1 and 65535 and it depends on the system. + */ +#define MAX_SESS_QUEUE_DEPTH 65535 + enum rtrs_imm_const { MAX_IMM_TYPE_BITS = 4, MAX_IMM_TYPE_MASK = ((1 << MAX_IMM_TYPE_BITS) - 1), @@ -46,16 +56,6 @@ enum { MAX_PATHS_NUM = 128, - /* - * Max IB immediate data size is 2^28 (MAX_IMM_PAYL_BITS) - * and the minimum chunk size is 4096 (2^12). - * So the maximum sess_queue_depth is 65536 (2^16) in theory. - * But mempool_create, create_qp and ib_post_send fail with - * "cannot allocate memory" error if sess_queue_depth is too big. - * Therefore the pratical max value of sess_queue_depth is - * somewhere between 1 and 65534 and it depends on the system. - */ - MAX_SESS_QUEUE_DEPTH = 65535, MIN_CHUNK_SIZE = 8192, RTRS_HB_INTERVAL_MS = 5000, @@ -90,7 +90,7 @@ struct rtrs_ib_dev { }; struct rtrs_con { - struct rtrs_sess *sess; + struct rtrs_path *path; struct ib_qp *qp; struct ib_cq *cq; struct rdma_cm_id *cm_id; @@ -100,7 +100,7 @@ struct rtrs_con { atomic_t sq_wr_avail; }; -struct rtrs_sess { +struct rtrs_path { struct list_head entry; struct sockaddr_storage dst_addr; struct sockaddr_storage src_addr; @@ -229,11 +229,11 @@ struct rtrs_msg_conn_rsp { /** * struct rtrs_msg_info_req * @type: @RTRS_MSG_INFO_REQ - * @sessname: Session name chosen by client + * @pathname: Path name chosen by client */ struct rtrs_msg_info_req { __le16 type; - u8 sessname[NAME_MAX]; + u8 pathname[NAME_MAX]; u8 reserved[15]; }; @@ -313,19 +313,19 @@ int rtrs_iu_post_rdma_write_imm(struct rtrs_con *con, struct rtrs_iu *iu, int rtrs_post_recv_empty(struct rtrs_con *con, struct ib_cqe *cqe); -int rtrs_cq_qp_create(struct rtrs_sess *sess, struct rtrs_con *con, +int rtrs_cq_qp_create(struct rtrs_path *path, struct rtrs_con *con, u32 max_send_sge, int cq_vector, int nr_cqe, u32 max_send_wr, u32 max_recv_wr, enum ib_poll_context poll_ctx); void rtrs_cq_qp_destroy(struct rtrs_con *con); -void rtrs_init_hb(struct rtrs_sess *sess, struct ib_cqe *cqe, +void rtrs_init_hb(struct rtrs_path *path, struct ib_cqe *cqe, unsigned int interval_ms, unsigned int missed_max, void (*err_handler)(struct rtrs_con *con), struct workqueue_struct *wq); -void rtrs_start_hb(struct rtrs_sess *sess); -void rtrs_stop_hb(struct rtrs_sess *sess); -void rtrs_send_hb_ack(struct rtrs_sess *sess); +void rtrs_start_hb(struct rtrs_path *path); +void rtrs_stop_hb(struct rtrs_path *path); +void rtrs_send_hb_ack(struct rtrs_path *path); void rtrs_rdma_dev_pd_init(enum ib_pd_flags pd_flags, struct rtrs_rdma_dev_pd *pool); diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-stats.c b/drivers/infiniband/ulp/rtrs/rtrs-srv-stats.c index 44b1c1652131..2aff1213a19d 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-srv-stats.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-stats.c @@ -14,9 +14,14 @@ int rtrs_srv_reset_rdma_stats(struct rtrs_srv_stats *stats, bool enable) { if (enable) { - struct rtrs_srv_stats_rdma_stats *r = &stats->rdma_stats; + int cpu; + struct rtrs_srv_stats_rdma_stats *r; + + for_each_possible_cpu(cpu) { + r = per_cpu_ptr(stats->rdma_stats, cpu); + memset(r, 0, sizeof(*r)); + } - memset(r, 0, sizeof(*r)); return 0; } @@ -25,11 +30,22 @@ int rtrs_srv_reset_rdma_stats(struct rtrs_srv_stats *stats, bool enable) ssize_t rtrs_srv_stats_rdma_to_str(struct rtrs_srv_stats *stats, char *page) { - struct rtrs_srv_stats_rdma_stats *r = &stats->rdma_stats; + int cpu; + struct rtrs_srv_stats_rdma_stats sum; + struct rtrs_srv_stats_rdma_stats *r; + + memset(&sum, 0, sizeof(sum)); + + for_each_possible_cpu(cpu) { + r = per_cpu_ptr(stats->rdma_stats, cpu); + + sum.dir[READ].cnt += r->dir[READ].cnt; + sum.dir[READ].size_total += r->dir[READ].size_total; + sum.dir[WRITE].cnt += r->dir[WRITE].cnt; + sum.dir[WRITE].size_total += r->dir[WRITE].size_total; + } - return sysfs_emit(page, "%lld %lld %lld %lldn %u\n", - (s64)atomic64_read(&r->dir[READ].cnt), - (s64)atomic64_read(&r->dir[READ].size_total), - (s64)atomic64_read(&r->dir[WRITE].cnt), - (s64)atomic64_read(&r->dir[WRITE].size_total), 0); + return sysfs_emit(page, "%llu %llu %llu %llu\n", + sum.dir[READ].cnt, sum.dir[READ].size_total, + sum.dir[WRITE].cnt, sum.dir[WRITE].size_total); } diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c b/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c index 9c43ce5ba1c1..2a3c9ac64a42 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c @@ -15,10 +15,10 @@ static void rtrs_srv_release(struct kobject *kobj) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; - sess = container_of(kobj, struct rtrs_srv_sess, kobj); - kfree(sess); + srv_path = container_of(kobj, struct rtrs_srv_path, kobj); + kfree(srv_path); } static struct kobj_type ktype = { @@ -36,24 +36,25 @@ static ssize_t rtrs_srv_disconnect_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { - struct rtrs_srv_sess *sess; - struct rtrs_sess *s; + struct rtrs_srv_path *srv_path; + struct rtrs_path *s; char str[MAXHOSTNAMELEN]; - sess = container_of(kobj, struct rtrs_srv_sess, kobj); - s = &sess->s; + srv_path = container_of(kobj, struct rtrs_srv_path, kobj); + s = &srv_path->s; if (!sysfs_streq(buf, "1")) { rtrs_err(s, "%s: invalid value: '%s'\n", attr->attr.name, buf); return -EINVAL; } - sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, str, sizeof(str)); + sockaddr_to_str((struct sockaddr *)&srv_path->s.dst_addr, str, + sizeof(str)); rtrs_info(s, "disconnect for path %s requested\n", str); /* first remove sysfs itself to avoid deadlock */ - sysfs_remove_file_self(&sess->kobj, &attr->attr); - close_sess(sess); + sysfs_remove_file_self(&srv_path->kobj, &attr->attr); + close_path(srv_path); return count; } @@ -66,11 +67,11 @@ static ssize_t rtrs_srv_hca_port_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; struct rtrs_con *usr_con; - sess = container_of(kobj, typeof(*sess), kobj); - usr_con = sess->s.con[0]; + srv_path = container_of(kobj, typeof(*srv_path), kobj); + usr_con = srv_path->s.con[0]; return sysfs_emit(page, "%u\n", usr_con->cm_id->port_num); } @@ -82,11 +83,11 @@ static ssize_t rtrs_srv_hca_name_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; - sess = container_of(kobj, struct rtrs_srv_sess, kobj); + srv_path = container_of(kobj, struct rtrs_srv_path, kobj); - return sysfs_emit(page, "%s\n", sess->s.dev->ib_dev->name); + return sysfs_emit(page, "%s\n", srv_path->s.dev->ib_dev->name); } static struct kobj_attribute rtrs_srv_hca_name_attr = @@ -96,11 +97,11 @@ static ssize_t rtrs_srv_src_addr_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; int cnt; - sess = container_of(kobj, struct rtrs_srv_sess, kobj); - cnt = sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, + srv_path = container_of(kobj, struct rtrs_srv_path, kobj); + cnt = sockaddr_to_str((struct sockaddr *)&srv_path->s.dst_addr, page, PAGE_SIZE); return cnt + sysfs_emit_at(page, cnt, "\n"); } @@ -112,11 +113,11 @@ static ssize_t rtrs_srv_dst_addr_show(struct kobject *kobj, struct kobj_attribute *attr, char *page) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; int len; - sess = container_of(kobj, struct rtrs_srv_sess, kobj); - len = sockaddr_to_str((struct sockaddr *)&sess->s.src_addr, page, + srv_path = container_of(kobj, struct rtrs_srv_path, kobj); + len = sockaddr_to_str((struct sockaddr *)&srv_path->s.src_addr, page, PAGE_SIZE); len += sysfs_emit_at(page, len, "\n"); return len; @@ -125,7 +126,7 @@ static ssize_t rtrs_srv_dst_addr_show(struct kobject *kobj, static struct kobj_attribute rtrs_srv_dst_addr_attr = __ATTR(dst_addr, 0444, rtrs_srv_dst_addr_show, NULL); -static struct attribute *rtrs_srv_sess_attrs[] = { +static struct attribute *rtrs_srv_path_attrs[] = { &rtrs_srv_hca_name_attr.attr, &rtrs_srv_hca_port_attr.attr, &rtrs_srv_src_addr_attr.attr, @@ -134,8 +135,8 @@ static struct attribute *rtrs_srv_sess_attrs[] = { NULL, }; -static const struct attribute_group rtrs_srv_sess_attr_group = { - .attrs = rtrs_srv_sess_attrs, +static const struct attribute_group rtrs_srv_path_attr_group = { + .attrs = rtrs_srv_path_attrs, }; STAT_ATTR(struct rtrs_srv_stats, rdma, @@ -151,9 +152,9 @@ static const struct attribute_group rtrs_srv_stats_attr_group = { .attrs = rtrs_srv_stats_attrs, }; -static int rtrs_srv_create_once_sysfs_root_folders(struct rtrs_srv_sess *sess) +static int rtrs_srv_create_once_sysfs_root_folders(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; int err = 0; mutex_lock(&srv->paths_mutex); @@ -164,7 +165,7 @@ static int rtrs_srv_create_once_sysfs_root_folders(struct rtrs_srv_sess *sess) goto unlock; } srv->dev.class = rtrs_dev_class; - err = dev_set_name(&srv->dev, "%s", sess->s.sessname); + err = dev_set_name(&srv->dev, "%s", srv_path->s.sessname); if (err) goto unlock; @@ -196,9 +197,9 @@ unlock: } static void -rtrs_srv_destroy_once_sysfs_root_folders(struct rtrs_srv_sess *sess) +rtrs_srv_destroy_once_sysfs_root_folders(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; mutex_lock(&srv->paths_mutex); if (!--srv->dev_ref) { @@ -213,33 +214,35 @@ rtrs_srv_destroy_once_sysfs_root_folders(struct rtrs_srv_sess *sess) } } -static void rtrs_srv_sess_stats_release(struct kobject *kobj) +static void rtrs_srv_path_stats_release(struct kobject *kobj) { struct rtrs_srv_stats *stats; stats = container_of(kobj, struct rtrs_srv_stats, kobj_stats); + free_percpu(stats->rdma_stats); + kfree(stats); } static struct kobj_type ktype_stats = { .sysfs_ops = &kobj_sysfs_ops, - .release = rtrs_srv_sess_stats_release, + .release = rtrs_srv_path_stats_release, }; -static int rtrs_srv_create_stats_files(struct rtrs_srv_sess *sess) +static int rtrs_srv_create_stats_files(struct rtrs_srv_path *srv_path) { int err; - struct rtrs_sess *s = &sess->s; + struct rtrs_path *s = &srv_path->s; - err = kobject_init_and_add(&sess->stats->kobj_stats, &ktype_stats, - &sess->kobj, "stats"); + err = kobject_init_and_add(&srv_path->stats->kobj_stats, &ktype_stats, + &srv_path->kobj, "stats"); if (err) { rtrs_err(s, "kobject_init_and_add(): %d\n", err); - kobject_put(&sess->stats->kobj_stats); + kobject_put(&srv_path->stats->kobj_stats); return err; } - err = sysfs_create_group(&sess->stats->kobj_stats, + err = sysfs_create_group(&srv_path->stats->kobj_stats, &rtrs_srv_stats_attr_group); if (err) { rtrs_err(s, "sysfs_create_group(): %d\n", err); @@ -249,64 +252,64 @@ static int rtrs_srv_create_stats_files(struct rtrs_srv_sess *sess) return 0; err: - kobject_del(&sess->stats->kobj_stats); - kobject_put(&sess->stats->kobj_stats); + kobject_del(&srv_path->stats->kobj_stats); + kobject_put(&srv_path->stats->kobj_stats); return err; } -int rtrs_srv_create_sess_files(struct rtrs_srv_sess *sess) +int rtrs_srv_create_path_files(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; - struct rtrs_sess *s = &sess->s; + struct rtrs_srv_sess *srv = srv_path->srv; + struct rtrs_path *s = &srv_path->s; char str[NAME_MAX]; int err; struct rtrs_addr path = { - .src = &sess->s.dst_addr, - .dst = &sess->s.src_addr, + .src = &srv_path->s.dst_addr, + .dst = &srv_path->s.src_addr, }; rtrs_addr_to_str(&path, str, sizeof(str)); - err = rtrs_srv_create_once_sysfs_root_folders(sess); + err = rtrs_srv_create_once_sysfs_root_folders(srv_path); if (err) return err; - err = kobject_init_and_add(&sess->kobj, &ktype, srv->kobj_paths, + err = kobject_init_and_add(&srv_path->kobj, &ktype, srv->kobj_paths, "%s", str); if (err) { rtrs_err(s, "kobject_init_and_add(): %d\n", err); goto destroy_root; } - err = sysfs_create_group(&sess->kobj, &rtrs_srv_sess_attr_group); + err = sysfs_create_group(&srv_path->kobj, &rtrs_srv_path_attr_group); if (err) { rtrs_err(s, "sysfs_create_group(): %d\n", err); goto put_kobj; } - err = rtrs_srv_create_stats_files(sess); + err = rtrs_srv_create_stats_files(srv_path); if (err) goto remove_group; return 0; remove_group: - sysfs_remove_group(&sess->kobj, &rtrs_srv_sess_attr_group); + sysfs_remove_group(&srv_path->kobj, &rtrs_srv_path_attr_group); put_kobj: - kobject_del(&sess->kobj); + kobject_del(&srv_path->kobj); destroy_root: - kobject_put(&sess->kobj); - rtrs_srv_destroy_once_sysfs_root_folders(sess); + kobject_put(&srv_path->kobj); + rtrs_srv_destroy_once_sysfs_root_folders(srv_path); return err; } -void rtrs_srv_destroy_sess_files(struct rtrs_srv_sess *sess) +void rtrs_srv_destroy_path_files(struct rtrs_srv_path *srv_path) { - if (sess->kobj.state_in_sysfs) { - kobject_del(&sess->stats->kobj_stats); - kobject_put(&sess->stats->kobj_stats); - sysfs_remove_group(&sess->kobj, &rtrs_srv_sess_attr_group); - kobject_put(&sess->kobj); + if (srv_path->kobj.state_in_sysfs) { + kobject_del(&srv_path->stats->kobj_stats); + kobject_put(&srv_path->stats->kobj_stats); + sysfs_remove_group(&srv_path->kobj, &rtrs_srv_path_attr_group); + kobject_put(&srv_path->kobj); - rtrs_srv_destroy_once_sysfs_root_folders(sess); + rtrs_srv_destroy_once_sysfs_root_folders(srv_path); } } diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.c b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.c new file mode 100644 index 000000000000..29ca59ceb0dd --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.c @@ -0,0 +1,16 @@ +// SPDX-License-Identifier: GPL-2.0+ +/* + * RDMA Network Block Driver + * + * Copyright (c) 2022 1&1 IONOS SE. All rights reserved. + */ +#include "rtrs.h" +#include "rtrs-pri.h" +#include "rtrs-srv.h" + +/* + * We include this last to have the helpers above available for the trace + * event implementations. + */ +#define CREATE_TRACE_POINTS +#include "rtrs-srv-trace.h" diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h new file mode 100644 index 000000000000..587d3e033081 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-trace.h @@ -0,0 +1,88 @@ +/* SPDX-License-Identifier: GPL-2.0+ */ +/* + * RDMA Network Block Driver + * + * Copyright (c) 2022 1&1 IONOS SE. All rights reserved. + */ +#undef TRACE_SYSTEM +#define TRACE_SYSTEM rtrs_srv + +#if !defined(_TRACE_RTRS_SRV_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_RTRS_SRV_H + +#include <linux/tracepoint.h> + +struct rtrs_srv_op; +struct rtrs_srv_con; +struct rtrs_srv_path; + +TRACE_DEFINE_ENUM(RTRS_SRV_CONNECTING); +TRACE_DEFINE_ENUM(RTRS_SRV_CONNECTED); +TRACE_DEFINE_ENUM(RTRS_SRV_CLOSING); +TRACE_DEFINE_ENUM(RTRS_SRV_CLOSED); + +#define show_rtrs_srv_state(x) \ + __print_symbolic(x, \ + { RTRS_SRV_CONNECTING, "CONNECTING" }, \ + { RTRS_SRV_CONNECTED, "CONNECTED" }, \ + { RTRS_SRV_CLOSING, "CLOSING" }, \ + { RTRS_SRV_CLOSED, "CLOSED" }) + +TRACE_EVENT(send_io_resp_imm, + TP_PROTO(struct rtrs_srv_op *id, + bool need_inval, + bool always_invalidate, + int errno), + + TP_ARGS(id, need_inval, always_invalidate, errno), + + TP_STRUCT__entry( + __field(u8, dir) + __field(bool, need_inval) + __field(bool, always_invalidate) + __field(u32, msg_id) + __field(int, wr_cnt) + __field(u32, signal_interval) + __field(int, state) + __field(int, errno) + __array(char, sessname, NAME_MAX) + ), + + TP_fast_assign( + struct rtrs_srv_con *con = id->con; + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); + + __entry->dir = id->dir; + __entry->state = srv_path->state; + __entry->errno = errno; + __entry->need_inval = need_inval; + __entry->always_invalidate = always_invalidate; + __entry->msg_id = id->msg_id; + __entry->wr_cnt = atomic_read(&con->c.wr_cnt); + __entry->signal_interval = s->signal_interval; + memcpy(__entry->sessname, kobject_name(&srv_path->kobj), NAME_MAX); + ), + + TP_printk("sess='%s' state='%s' dir=%s err='%d' inval='%d' glob-inval='%d' msgid='%u' wrcnt='%d' sig-interval='%u'", + __entry->sessname, + show_rtrs_srv_state(__entry->state), + __print_symbolic(__entry->dir, + { READ, "READ" }, + { WRITE, "WRITE" }), + __entry->errno, + __entry->need_inval, + __entry->always_invalidate, + __entry->msg_id, + __entry->wr_cnt, + __entry->signal_interval + ) +); + +#endif /* _TRACE_RTRS_SRV_H */ + +#undef TRACE_INCLUDE_PATH +#define TRACE_INCLUDE_PATH . +#define TRACE_INCLUDE_FILE rtrs-srv-trace +#include <trace/define_trace.h> + diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv.c b/drivers/infiniband/ulp/rtrs/rtrs-srv.c index 7df71f8cf149..22d7ba05e9fe 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-srv.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv.c @@ -11,12 +11,12 @@ #define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt #include <linux/module.h> -#include <linux/mempool.h> #include "rtrs-srv.h" #include "rtrs-log.h" #include <rdma/ib_cm.h> #include <rdma/ib_verbs.h> +#include "rtrs-srv-trace.h" MODULE_DESCRIPTION("RDMA Transport Server"); MODULE_LICENSE("GPL"); @@ -26,11 +26,7 @@ MODULE_LICENSE("GPL"); #define DEFAULT_SESS_QUEUE_DEPTH 512 #define MAX_HDR_SIZE PAGE_SIZE -/* We guarantee to serve 10 paths at least */ -#define CHUNK_POOL_SZ 10 - static struct rtrs_rdma_dev_pd dev_pd; -static mempool_t *chunk_pool; struct class *rtrs_dev_class; static struct rtrs_srv_ib_ctx ib_ctx; @@ -62,19 +58,14 @@ static inline struct rtrs_srv_con *to_srv_con(struct rtrs_con *c) return container_of(c, struct rtrs_srv_con, c); } -static inline struct rtrs_srv_sess *to_srv_sess(struct rtrs_sess *s) -{ - return container_of(s, struct rtrs_srv_sess, s); -} - -static bool rtrs_srv_change_state(struct rtrs_srv_sess *sess, +static bool rtrs_srv_change_state(struct rtrs_srv_path *srv_path, enum rtrs_srv_state new_state) { enum rtrs_srv_state old_state; bool changed = false; - spin_lock_irq(&sess->state_lock); - old_state = sess->state; + spin_lock_irq(&srv_path->state_lock); + old_state = srv_path->state; switch (new_state) { case RTRS_SRV_CONNECTED: if (old_state == RTRS_SRV_CONNECTING) @@ -93,8 +84,8 @@ static bool rtrs_srv_change_state(struct rtrs_srv_sess *sess, break; } if (changed) - sess->state = new_state; - spin_unlock_irq(&sess->state_lock); + srv_path->state = new_state; + spin_unlock_irq(&srv_path->state_lock); return changed; } @@ -106,16 +97,16 @@ static void free_id(struct rtrs_srv_op *id) kfree(id); } -static void rtrs_srv_free_ops_ids(struct rtrs_srv_sess *sess) +static void rtrs_srv_free_ops_ids(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; int i; - if (sess->ops_ids) { + if (srv_path->ops_ids) { for (i = 0; i < srv->queue_depth; i++) - free_id(sess->ops_ids[i]); - kfree(sess->ops_ids); - sess->ops_ids = NULL; + free_id(srv_path->ops_ids[i]); + kfree(srv_path->ops_ids); + srv_path->ops_ids = NULL; } } @@ -127,21 +118,24 @@ static struct ib_cqe io_comp_cqe = { static inline void rtrs_srv_inflight_ref_release(struct percpu_ref *ref) { - struct rtrs_srv_sess *sess = container_of(ref, struct rtrs_srv_sess, ids_inflight_ref); + struct rtrs_srv_path *srv_path = container_of(ref, + struct rtrs_srv_path, + ids_inflight_ref); - percpu_ref_exit(&sess->ids_inflight_ref); - complete(&sess->complete_done); + percpu_ref_exit(&srv_path->ids_inflight_ref); + complete(&srv_path->complete_done); } -static int rtrs_srv_alloc_ops_ids(struct rtrs_srv_sess *sess) +static int rtrs_srv_alloc_ops_ids(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; struct rtrs_srv_op *id; int i, ret; - sess->ops_ids = kcalloc(srv->queue_depth, sizeof(*sess->ops_ids), - GFP_KERNEL); - if (!sess->ops_ids) + srv_path->ops_ids = kcalloc(srv->queue_depth, + sizeof(*srv_path->ops_ids), + GFP_KERNEL); + if (!srv_path->ops_ids) goto err; for (i = 0; i < srv->queue_depth; ++i) { @@ -149,44 +143,44 @@ static int rtrs_srv_alloc_ops_ids(struct rtrs_srv_sess *sess) if (!id) goto err; - sess->ops_ids[i] = id; + srv_path->ops_ids[i] = id; } - ret = percpu_ref_init(&sess->ids_inflight_ref, + ret = percpu_ref_init(&srv_path->ids_inflight_ref, rtrs_srv_inflight_ref_release, 0, GFP_KERNEL); if (ret) { pr_err("Percpu reference init failed\n"); goto err; } - init_completion(&sess->complete_done); + init_completion(&srv_path->complete_done); return 0; err: - rtrs_srv_free_ops_ids(sess); + rtrs_srv_free_ops_ids(srv_path); return -ENOMEM; } -static inline void rtrs_srv_get_ops_ids(struct rtrs_srv_sess *sess) +static inline void rtrs_srv_get_ops_ids(struct rtrs_srv_path *srv_path) { - percpu_ref_get(&sess->ids_inflight_ref); + percpu_ref_get(&srv_path->ids_inflight_ref); } -static inline void rtrs_srv_put_ops_ids(struct rtrs_srv_sess *sess) +static inline void rtrs_srv_put_ops_ids(struct rtrs_srv_path *srv_path) { - percpu_ref_put(&sess->ids_inflight_ref); + percpu_ref_put(&srv_path->ids_inflight_ref); } static void rtrs_srv_reg_mr_done(struct ib_cq *cq, struct ib_wc *wc) { struct rtrs_srv_con *con = to_srv_con(wc->qp->qp_context); - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); if (wc->status != IB_WC_SUCCESS) { rtrs_err(s, "REG MR failed: %s\n", ib_wc_status_msg(wc->status)); - close_sess(sess); + close_path(srv_path); return; } } @@ -197,9 +191,9 @@ static struct ib_cqe local_reg_cqe = { static int rdma_write_sg(struct rtrs_srv_op *id) { - struct rtrs_sess *s = id->con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); - dma_addr_t dma_addr = sess->dma_addr[id->msg_id]; + struct rtrs_path *s = id->con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); + dma_addr_t dma_addr = srv_path->dma_addr[id->msg_id]; struct rtrs_srv_mr *srv_mr; struct ib_send_wr inv_wr; struct ib_rdma_wr imm_wr; @@ -233,7 +227,7 @@ static int rdma_write_sg(struct rtrs_srv_op *id) return -EINVAL; } - plist->lkey = sess->s.dev->ib_pd->local_dma_lkey; + plist->lkey = srv_path->s.dev->ib_pd->local_dma_lkey; offset += plist->length; wr->wr.sg_list = plist; @@ -284,7 +278,7 @@ static int rdma_write_sg(struct rtrs_srv_op *id) if (always_invalidate) { struct rtrs_msg_rkey_rsp *msg; - srv_mr = &sess->mrs[id->msg_id]; + srv_mr = &srv_path->mrs[id->msg_id]; rwr.wr.opcode = IB_WR_REG_MR; rwr.wr.wr_cqe = &local_reg_cqe; rwr.wr.num_sge = 0; @@ -300,11 +294,11 @@ static int rdma_write_sg(struct rtrs_srv_op *id) list.addr = srv_mr->iu->dma_addr; list.length = sizeof(*msg); - list.lkey = sess->s.dev->ib_pd->local_dma_lkey; + list.lkey = srv_path->s.dev->ib_pd->local_dma_lkey; imm_wr.wr.sg_list = &list; imm_wr.wr.num_sge = 1; imm_wr.wr.opcode = IB_WR_SEND_WITH_IMM; - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, + ib_dma_sync_single_for_device(srv_path->s.dev->ib_dev, srv_mr->iu->dma_addr, srv_mr->iu->size, DMA_TO_DEVICE); } else { @@ -317,7 +311,7 @@ static int rdma_write_sg(struct rtrs_srv_op *id) 0, need_inval)); imm_wr.wr.wr_cqe = &io_comp_cqe; - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, dma_addr, + ib_dma_sync_single_for_device(srv_path->s.dev->ib_dev, dma_addr, offset, DMA_BIDIRECTIONAL); err = ib_post_send(id->con->c.qp, &id->tx_wr.wr, NULL); @@ -341,8 +335,8 @@ static int rdma_write_sg(struct rtrs_srv_op *id) static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, int errno) { - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); struct ib_send_wr inv_wr, *wr = NULL; struct ib_rdma_wr imm_wr; struct ib_reg_wr rwr; @@ -377,6 +371,8 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, } } + trace_send_io_resp_imm(id, need_inval, always_invalidate, errno); + if (need_inval && always_invalidate) { wr = &inv_wr; inv_wr.next = &rwr.wr; @@ -402,7 +398,7 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, struct ib_sge list; struct rtrs_msg_rkey_rsp *msg; - srv_mr = &sess->mrs[id->msg_id]; + srv_mr = &srv_path->mrs[id->msg_id]; rwr.wr.next = &imm_wr.wr; rwr.wr.opcode = IB_WR_REG_MR; rwr.wr.wr_cqe = &local_reg_cqe; @@ -419,11 +415,11 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, list.addr = srv_mr->iu->dma_addr; list.length = sizeof(*msg); - list.lkey = sess->s.dev->ib_pd->local_dma_lkey; + list.lkey = srv_path->s.dev->ib_pd->local_dma_lkey; imm_wr.wr.sg_list = &list; imm_wr.wr.num_sge = 1; imm_wr.wr.opcode = IB_WR_SEND_WITH_IMM; - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, + ib_dma_sync_single_for_device(srv_path->s.dev->ib_dev, srv_mr->iu->dma_addr, srv_mr->iu->size, DMA_TO_DEVICE); } else { @@ -444,11 +440,11 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, return err; } -void close_sess(struct rtrs_srv_sess *sess) +void close_path(struct rtrs_srv_path *srv_path) { - if (rtrs_srv_change_state(sess, RTRS_SRV_CLOSING)) - queue_work(rtrs_wq, &sess->close_work); - WARN_ON(sess->state != RTRS_SRV_CLOSING); + if (rtrs_srv_change_state(srv_path, RTRS_SRV_CLOSING)) + queue_work(rtrs_wq, &srv_path->close_work); + WARN_ON(srv_path->state != RTRS_SRV_CLOSING); } static inline const char *rtrs_srv_state_str(enum rtrs_srv_state state) @@ -480,35 +476,35 @@ static inline const char *rtrs_srv_state_str(enum rtrs_srv_state state) */ bool rtrs_srv_resp_rdma(struct rtrs_srv_op *id, int status) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; struct rtrs_srv_con *con; - struct rtrs_sess *s; + struct rtrs_path *s; int err; if (WARN_ON(!id)) return true; con = id->con; - s = con->c.sess; - sess = to_srv_sess(s); + s = con->c.path; + srv_path = to_srv_path(s); id->status = status; - if (sess->state != RTRS_SRV_CONNECTED) { + if (srv_path->state != RTRS_SRV_CONNECTED) { rtrs_err_rl(s, - "Sending I/O response failed, session %s is disconnected, sess state %s\n", - kobject_name(&sess->kobj), - rtrs_srv_state_str(sess->state)); + "Sending I/O response failed, server path %s is disconnected, path state %s\n", + kobject_name(&srv_path->kobj), + rtrs_srv_state_str(srv_path->state)); goto out; } if (always_invalidate) { - struct rtrs_srv_mr *mr = &sess->mrs[id->msg_id]; + struct rtrs_srv_mr *mr = &srv_path->mrs[id->msg_id]; ib_update_fast_reg_key(mr->mr, ib_inc_rkey(mr->mr->rkey)); } if (atomic_sub_return(1, &con->c.sq_wr_avail) < 0) { - rtrs_err(s, "IB send queue full: sess=%s cid=%d\n", - kobject_name(&sess->kobj), + rtrs_err(s, "IB send queue full: srv_path=%s cid=%d\n", + kobject_name(&srv_path->kobj), con->c.cid); atomic_add(1, &con->c.sq_wr_avail); spin_lock(&con->rsp_wr_wait_lock); @@ -523,12 +519,12 @@ bool rtrs_srv_resp_rdma(struct rtrs_srv_op *id, int status) err = rdma_write_sg(id); if (err) { - rtrs_err_rl(s, "IO response failed: %d: sess=%s\n", err, - kobject_name(&sess->kobj)); - close_sess(sess); + rtrs_err_rl(s, "IO response failed: %d: srv_path=%s\n", err, + kobject_name(&srv_path->kobj)); + close_path(srv_path); } out: - rtrs_srv_put_ops_ids(sess); + rtrs_srv_put_ops_ids(srv_path); return true; } EXPORT_SYMBOL(rtrs_srv_resp_rdma); @@ -538,33 +534,33 @@ EXPORT_SYMBOL(rtrs_srv_resp_rdma); * @srv: Session pointer * @priv: The private pointer that is associated with the session. */ -void rtrs_srv_set_sess_priv(struct rtrs_srv *srv, void *priv) +void rtrs_srv_set_sess_priv(struct rtrs_srv_sess *srv, void *priv) { srv->priv = priv; } EXPORT_SYMBOL(rtrs_srv_set_sess_priv); -static void unmap_cont_bufs(struct rtrs_srv_sess *sess) +static void unmap_cont_bufs(struct rtrs_srv_path *srv_path) { int i; - for (i = 0; i < sess->mrs_num; i++) { + for (i = 0; i < srv_path->mrs_num; i++) { struct rtrs_srv_mr *srv_mr; - srv_mr = &sess->mrs[i]; - rtrs_iu_free(srv_mr->iu, sess->s.dev->ib_dev, 1); + srv_mr = &srv_path->mrs[i]; + rtrs_iu_free(srv_mr->iu, srv_path->s.dev->ib_dev, 1); ib_dereg_mr(srv_mr->mr); - ib_dma_unmap_sg(sess->s.dev->ib_dev, srv_mr->sgt.sgl, + ib_dma_unmap_sg(srv_path->s.dev->ib_dev, srv_mr->sgt.sgl, srv_mr->sgt.nents, DMA_BIDIRECTIONAL); sg_free_table(&srv_mr->sgt); } - kfree(sess->mrs); + kfree(srv_path->mrs); } -static int map_cont_bufs(struct rtrs_srv_sess *sess) +static int map_cont_bufs(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; - struct rtrs_sess *ss = &sess->s; + struct rtrs_srv_sess *srv = srv_path->srv; + struct rtrs_path *ss = &srv_path->s; int i, mri, err, mrs_num; unsigned int chunk_bits; int chunks_per_mr = 1; @@ -581,23 +577,23 @@ static int map_cont_bufs(struct rtrs_srv_sess *sess) mrs_num = srv->queue_depth; } else { chunks_per_mr = - sess->s.dev->ib_dev->attrs.max_fast_reg_page_list_len; + srv_path->s.dev->ib_dev->attrs.max_fast_reg_page_list_len; mrs_num = DIV_ROUND_UP(srv->queue_depth, chunks_per_mr); chunks_per_mr = DIV_ROUND_UP(srv->queue_depth, mrs_num); } - sess->mrs = kcalloc(mrs_num, sizeof(*sess->mrs), GFP_KERNEL); - if (!sess->mrs) + srv_path->mrs = kcalloc(mrs_num, sizeof(*srv_path->mrs), GFP_KERNEL); + if (!srv_path->mrs) return -ENOMEM; - sess->mrs_num = mrs_num; + srv_path->mrs_num = mrs_num; for (mri = 0; mri < mrs_num; mri++) { - struct rtrs_srv_mr *srv_mr = &sess->mrs[mri]; + struct rtrs_srv_mr *srv_mr = &srv_path->mrs[mri]; struct sg_table *sgt = &srv_mr->sgt; struct scatterlist *s; struct ib_mr *mr; - int nr, chunks; + int nr, nr_sgt, chunks; chunks = chunks_per_mr * mri; if (!always_invalidate) @@ -612,19 +608,19 @@ static int map_cont_bufs(struct rtrs_srv_sess *sess) sg_set_page(s, srv->chunks[chunks + i], max_chunk_size, 0); - nr = ib_dma_map_sg(sess->s.dev->ib_dev, sgt->sgl, + nr_sgt = ib_dma_map_sg(srv_path->s.dev->ib_dev, sgt->sgl, sgt->nents, DMA_BIDIRECTIONAL); - if (nr < sgt->nents) { - err = nr < 0 ? nr : -EINVAL; + if (!nr_sgt) { + err = -EINVAL; goto free_sg; } - mr = ib_alloc_mr(sess->s.dev->ib_pd, IB_MR_TYPE_MEM_REG, - sgt->nents); + mr = ib_alloc_mr(srv_path->s.dev->ib_pd, IB_MR_TYPE_MEM_REG, + nr_sgt); if (IS_ERR(mr)) { err = PTR_ERR(mr); goto unmap_sg; } - nr = ib_map_mr_sg(mr, sgt->sgl, sgt->nents, + nr = ib_map_mr_sg(mr, sgt->sgl, nr_sgt, NULL, max_chunk_size); if (nr < 0 || nr < sgt->nents) { err = nr < 0 ? nr : -EINVAL; @@ -634,7 +630,7 @@ static int map_cont_bufs(struct rtrs_srv_sess *sess) if (always_invalidate) { srv_mr->iu = rtrs_iu_alloc(1, sizeof(struct rtrs_msg_rkey_rsp), - GFP_KERNEL, sess->s.dev->ib_dev, + GFP_KERNEL, srv_path->s.dev->ib_dev, DMA_TO_DEVICE, rtrs_srv_rdma_done); if (!srv_mr->iu) { err = -ENOMEM; @@ -643,8 +639,8 @@ static int map_cont_bufs(struct rtrs_srv_sess *sess) } } /* Eventually dma addr for each chunk can be cached */ - for_each_sg(sgt->sgl, s, sgt->orig_nents, i) - sess->dma_addr[chunks + i] = sg_dma_address(s); + for_each_sg(sgt->sgl, s, nr_sgt, i) + srv_path->dma_addr[chunks + i] = sg_dma_address(s); ib_update_fast_reg_key(mr, ib_inc_rkey(mr->rkey)); srv_mr->mr = mr; @@ -652,75 +648,75 @@ static int map_cont_bufs(struct rtrs_srv_sess *sess) continue; err: while (mri--) { - srv_mr = &sess->mrs[mri]; + srv_mr = &srv_path->mrs[mri]; sgt = &srv_mr->sgt; mr = srv_mr->mr; - rtrs_iu_free(srv_mr->iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(srv_mr->iu, srv_path->s.dev->ib_dev, 1); dereg_mr: ib_dereg_mr(mr); unmap_sg: - ib_dma_unmap_sg(sess->s.dev->ib_dev, sgt->sgl, + ib_dma_unmap_sg(srv_path->s.dev->ib_dev, sgt->sgl, sgt->nents, DMA_BIDIRECTIONAL); free_sg: sg_free_table(sgt); } - kfree(sess->mrs); + kfree(srv_path->mrs); return err; } chunk_bits = ilog2(srv->queue_depth - 1) + 1; - sess->mem_bits = (MAX_IMM_PAYL_BITS - chunk_bits); + srv_path->mem_bits = (MAX_IMM_PAYL_BITS - chunk_bits); return 0; } static void rtrs_srv_hb_err_handler(struct rtrs_con *c) { - close_sess(to_srv_sess(c->sess)); + close_path(to_srv_path(c->path)); } -static void rtrs_srv_init_hb(struct rtrs_srv_sess *sess) +static void rtrs_srv_init_hb(struct rtrs_srv_path *srv_path) { - rtrs_init_hb(&sess->s, &io_comp_cqe, + rtrs_init_hb(&srv_path->s, &io_comp_cqe, RTRS_HB_INTERVAL_MS, RTRS_HB_MISSED_MAX, rtrs_srv_hb_err_handler, rtrs_wq); } -static void rtrs_srv_start_hb(struct rtrs_srv_sess *sess) +static void rtrs_srv_start_hb(struct rtrs_srv_path *srv_path) { - rtrs_start_hb(&sess->s); + rtrs_start_hb(&srv_path->s); } -static void rtrs_srv_stop_hb(struct rtrs_srv_sess *sess) +static void rtrs_srv_stop_hb(struct rtrs_srv_path *srv_path) { - rtrs_stop_hb(&sess->s); + rtrs_stop_hb(&srv_path->s); } static void rtrs_srv_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc) { struct rtrs_srv_con *con = to_srv_con(wc->qp->qp_context); - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); struct rtrs_iu *iu; iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); - rtrs_iu_free(iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(iu, srv_path->s.dev->ib_dev, 1); if (wc->status != IB_WC_SUCCESS) { rtrs_err(s, "Sess info response send failed: %s\n", ib_wc_status_msg(wc->status)); - close_sess(sess); + close_path(srv_path); return; } WARN_ON(wc->opcode != IB_WC_SEND); } -static void rtrs_srv_sess_up(struct rtrs_srv_sess *sess) +static void rtrs_srv_path_up(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; struct rtrs_srv_ctx *ctx = srv->ctx; int up; @@ -731,18 +727,18 @@ static void rtrs_srv_sess_up(struct rtrs_srv_sess *sess) mutex_unlock(&srv->paths_ev_mutex); /* Mark session as established */ - sess->established = true; + srv_path->established = true; } -static void rtrs_srv_sess_down(struct rtrs_srv_sess *sess) +static void rtrs_srv_path_down(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; struct rtrs_srv_ctx *ctx = srv->ctx; - if (!sess->established) + if (!srv_path->established) return; - sess->established = false; + srv_path->established = false; mutex_lock(&srv->paths_ev_mutex); WARN_ON(!srv->paths_up); if (--srv->paths_up == 0) @@ -750,11 +746,11 @@ static void rtrs_srv_sess_down(struct rtrs_srv_sess *sess) mutex_unlock(&srv->paths_ev_mutex); } -static bool exist_sessname(struct rtrs_srv_ctx *ctx, - const char *sessname, const uuid_t *path_uuid) +static bool exist_pathname(struct rtrs_srv_ctx *ctx, + const char *pathname, const uuid_t *path_uuid) { - struct rtrs_srv *srv; - struct rtrs_srv_sess *sess; + struct rtrs_srv_sess *srv; + struct rtrs_srv_path *srv_path; bool found = false; mutex_lock(&ctx->srv_mutex); @@ -767,9 +763,9 @@ static bool exist_sessname(struct rtrs_srv_ctx *ctx, continue; } - list_for_each_entry(sess, &srv->paths_list, s.entry) { - if (strlen(sess->s.sessname) == strlen(sessname) && - !strcmp(sess->s.sessname, sessname)) { + list_for_each_entry(srv_path, &srv->paths_list, s.entry) { + if (strlen(srv_path->s.sessname) == strlen(pathname) && + !strcmp(srv_path->s.sessname, pathname)) { found = true; break; } @@ -782,14 +778,14 @@ static bool exist_sessname(struct rtrs_srv_ctx *ctx, return found; } -static int post_recv_sess(struct rtrs_srv_sess *sess); +static int post_recv_path(struct rtrs_srv_path *srv_path); static int rtrs_rdma_do_reject(struct rdma_cm_id *cm_id, int errno); static int process_info_req(struct rtrs_srv_con *con, struct rtrs_msg_info_req *msg) { - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); struct ib_send_wr *reg_wr = NULL; struct rtrs_msg_info_rsp *rsp; struct rtrs_iu *tx_iu; @@ -797,31 +793,32 @@ static int process_info_req(struct rtrs_srv_con *con, int mri, err; size_t tx_sz; - err = post_recv_sess(sess); + err = post_recv_path(srv_path); if (err) { - rtrs_err(s, "post_recv_sess(), err: %d\n", err); + rtrs_err(s, "post_recv_path(), err: %d\n", err); return err; } - if (strchr(msg->sessname, '/') || strchr(msg->sessname, '.')) { - rtrs_err(s, "sessname cannot contain / and .\n"); + if (strchr(msg->pathname, '/') || strchr(msg->pathname, '.')) { + rtrs_err(s, "pathname cannot contain / and .\n"); return -EINVAL; } - if (exist_sessname(sess->srv->ctx, - msg->sessname, &sess->srv->paths_uuid)) { - rtrs_err(s, "sessname is duplicated: %s\n", msg->sessname); + if (exist_pathname(srv_path->srv->ctx, + msg->pathname, &srv_path->srv->paths_uuid)) { + rtrs_err(s, "pathname is duplicated: %s\n", msg->pathname); return -EPERM; } - strscpy(sess->s.sessname, msg->sessname, sizeof(sess->s.sessname)); + strscpy(srv_path->s.sessname, msg->pathname, + sizeof(srv_path->s.sessname)); - rwr = kcalloc(sess->mrs_num, sizeof(*rwr), GFP_KERNEL); + rwr = kcalloc(srv_path->mrs_num, sizeof(*rwr), GFP_KERNEL); if (!rwr) return -ENOMEM; tx_sz = sizeof(*rsp); - tx_sz += sizeof(rsp->desc[0]) * sess->mrs_num; - tx_iu = rtrs_iu_alloc(1, tx_sz, GFP_KERNEL, sess->s.dev->ib_dev, + tx_sz += sizeof(rsp->desc[0]) * srv_path->mrs_num; + tx_iu = rtrs_iu_alloc(1, tx_sz, GFP_KERNEL, srv_path->s.dev->ib_dev, DMA_TO_DEVICE, rtrs_srv_info_rsp_done); if (!tx_iu) { err = -ENOMEM; @@ -830,10 +827,10 @@ static int process_info_req(struct rtrs_srv_con *con, rsp = tx_iu->buf; rsp->type = cpu_to_le16(RTRS_MSG_INFO_RSP); - rsp->sg_cnt = cpu_to_le16(sess->mrs_num); + rsp->sg_cnt = cpu_to_le16(srv_path->mrs_num); - for (mri = 0; mri < sess->mrs_num; mri++) { - struct ib_mr *mr = sess->mrs[mri].mr; + for (mri = 0; mri < srv_path->mrs_num; mri++) { + struct ib_mr *mr = srv_path->mrs[mri].mr; rsp->desc[mri].addr = cpu_to_le64(mr->iova); rsp->desc[mri].key = cpu_to_le32(mr->rkey); @@ -854,13 +851,13 @@ static int process_info_req(struct rtrs_srv_con *con, reg_wr = &rwr[mri].wr; } - err = rtrs_srv_create_sess_files(sess); + err = rtrs_srv_create_path_files(srv_path); if (err) goto iu_free; - kobject_get(&sess->kobj); - get_device(&sess->srv->dev); - rtrs_srv_change_state(sess, RTRS_SRV_CONNECTED); - rtrs_srv_start_hb(sess); + kobject_get(&srv_path->kobj); + get_device(&srv_path->srv->dev); + rtrs_srv_change_state(srv_path, RTRS_SRV_CONNECTED); + rtrs_srv_start_hb(srv_path); /* * We do not account number of established connections at the current @@ -868,9 +865,10 @@ static int process_info_req(struct rtrs_srv_con *con, * all connections are successfully established. Thus, simply notify * listener with a proper event if we are the first path. */ - rtrs_srv_sess_up(sess); + rtrs_srv_path_up(srv_path); - ib_dma_sync_single_for_device(sess->s.dev->ib_dev, tx_iu->dma_addr, + ib_dma_sync_single_for_device(srv_path->s.dev->ib_dev, + tx_iu->dma_addr, tx_iu->size, DMA_TO_DEVICE); /* Send info response */ @@ -878,7 +876,7 @@ static int process_info_req(struct rtrs_srv_con *con, if (err) { rtrs_err(s, "rtrs_iu_post_send(), err: %d\n", err); iu_free: - rtrs_iu_free(tx_iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(tx_iu, srv_path->s.dev->ib_dev, 1); } rwr_free: kfree(rwr); @@ -889,8 +887,8 @@ rwr_free: static void rtrs_srv_info_req_done(struct ib_cq *cq, struct ib_wc *wc) { struct rtrs_srv_con *con = to_srv_con(wc->qp->qp_context); - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); struct rtrs_msg_info_req *msg; struct rtrs_iu *iu; int err; @@ -910,7 +908,7 @@ static void rtrs_srv_info_req_done(struct ib_cq *cq, struct ib_wc *wc) wc->byte_len); goto close; } - ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr, + ib_dma_sync_single_for_cpu(srv_path->s.dev->ib_dev, iu->dma_addr, iu->size, DMA_FROM_DEVICE); msg = iu->buf; if (le16_to_cpu(msg->type) != RTRS_MSG_INFO_REQ) { @@ -923,22 +921,22 @@ static void rtrs_srv_info_req_done(struct ib_cq *cq, struct ib_wc *wc) goto close; out: - rtrs_iu_free(iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(iu, srv_path->s.dev->ib_dev, 1); return; close: - close_sess(sess); + close_path(srv_path); goto out; } static int post_recv_info_req(struct rtrs_srv_con *con) { - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); struct rtrs_iu *rx_iu; int err; rx_iu = rtrs_iu_alloc(1, sizeof(struct rtrs_msg_info_req), - GFP_KERNEL, sess->s.dev->ib_dev, + GFP_KERNEL, srv_path->s.dev->ib_dev, DMA_FROM_DEVICE, rtrs_srv_info_req_done); if (!rx_iu) return -ENOMEM; @@ -946,7 +944,7 @@ static int post_recv_info_req(struct rtrs_srv_con *con) err = rtrs_iu_post_recv(&con->c, rx_iu); if (err) { rtrs_err(s, "rtrs_iu_post_recv(), err: %d\n", err); - rtrs_iu_free(rx_iu, sess->s.dev->ib_dev, 1); + rtrs_iu_free(rx_iu, srv_path->s.dev->ib_dev, 1); return err; } @@ -966,20 +964,20 @@ static int post_recv_io(struct rtrs_srv_con *con, size_t q_size) return 0; } -static int post_recv_sess(struct rtrs_srv_sess *sess) +static int post_recv_path(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; - struct rtrs_sess *s = &sess->s; + struct rtrs_srv_sess *srv = srv_path->srv; + struct rtrs_path *s = &srv_path->s; size_t q_size; int err, cid; - for (cid = 0; cid < sess->s.con_num; cid++) { + for (cid = 0; cid < srv_path->s.con_num; cid++) { if (cid == 0) q_size = SERVICE_CON_QUEUE_DEPTH; else q_size = srv->queue_depth; - err = post_recv_io(to_srv_con(sess->s.con[cid]), q_size); + err = post_recv_io(to_srv_con(srv_path->s.con[cid]), q_size); if (err) { rtrs_err(s, "post_recv_io(), err: %d\n", err); return err; @@ -993,9 +991,9 @@ static void process_read(struct rtrs_srv_con *con, struct rtrs_msg_rdma_read *msg, u32 buf_id, u32 off) { - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); - struct rtrs_srv *srv = sess->srv; + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); + struct rtrs_srv_sess *srv = srv_path->srv; struct rtrs_srv_ctx *ctx = srv->ctx; struct rtrs_srv_op *id; @@ -1003,10 +1001,10 @@ static void process_read(struct rtrs_srv_con *con, void *data; int ret; - if (sess->state != RTRS_SRV_CONNECTED) { + if (srv_path->state != RTRS_SRV_CONNECTED) { rtrs_err_rl(s, "Processing read request failed, session is disconnected, sess state %s\n", - rtrs_srv_state_str(sess->state)); + rtrs_srv_state_str(srv_path->state)); return; } if (msg->sg_cnt != 1 && msg->sg_cnt != 0) { @@ -1014,9 +1012,9 @@ static void process_read(struct rtrs_srv_con *con, "Processing read request failed, invalid message\n"); return; } - rtrs_srv_get_ops_ids(sess); - rtrs_srv_update_rdma_stats(sess->stats, off, READ); - id = sess->ops_ids[buf_id]; + rtrs_srv_get_ops_ids(srv_path); + rtrs_srv_update_rdma_stats(srv_path->stats, off, READ); + id = srv_path->ops_ids[buf_id]; id->con = con; id->dir = READ; id->msg_id = buf_id; @@ -1024,7 +1022,7 @@ static void process_read(struct rtrs_srv_con *con, usr_len = le16_to_cpu(msg->usr_len); data_len = off - usr_len; data = page_address(srv->chunks[buf_id]); - ret = ctx->ops.rdma_ev(srv->priv, id, READ, data, data_len, + ret = ctx->ops.rdma_ev(srv->priv, id, data, data_len, data + data_len, usr_len); if (ret) { @@ -1042,18 +1040,18 @@ send_err_msg: rtrs_err_rl(s, "Sending err msg for failed RDMA-Write-Req failed, msg_id %d, err: %d\n", buf_id, ret); - close_sess(sess); + close_path(srv_path); } - rtrs_srv_put_ops_ids(sess); + rtrs_srv_put_ops_ids(srv_path); } static void process_write(struct rtrs_srv_con *con, struct rtrs_msg_rdma_write *req, u32 buf_id, u32 off) { - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); - struct rtrs_srv *srv = sess->srv; + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); + struct rtrs_srv_sess *srv = srv_path->srv; struct rtrs_srv_ctx *ctx = srv->ctx; struct rtrs_srv_op *id; @@ -1061,15 +1059,15 @@ static void process_write(struct rtrs_srv_con *con, void *data; int ret; - if (sess->state != RTRS_SRV_CONNECTED) { + if (srv_path->state != RTRS_SRV_CONNECTED) { rtrs_err_rl(s, "Processing write request failed, session is disconnected, sess state %s\n", - rtrs_srv_state_str(sess->state)); + rtrs_srv_state_str(srv_path->state)); return; } - rtrs_srv_get_ops_ids(sess); - rtrs_srv_update_rdma_stats(sess->stats, off, WRITE); - id = sess->ops_ids[buf_id]; + rtrs_srv_get_ops_ids(srv_path); + rtrs_srv_update_rdma_stats(srv_path->stats, off, WRITE); + id = srv_path->ops_ids[buf_id]; id->con = con; id->dir = WRITE; id->msg_id = buf_id; @@ -1077,7 +1075,7 @@ static void process_write(struct rtrs_srv_con *con, usr_len = le16_to_cpu(req->usr_len); data_len = off - usr_len; data = page_address(srv->chunks[buf_id]); - ret = ctx->ops.rdma_ev(srv->priv, id, WRITE, data, data_len, + ret = ctx->ops.rdma_ev(srv->priv, id, data, data_len, data + data_len, usr_len); if (ret) { rtrs_err_rl(s, @@ -1094,20 +1092,21 @@ send_err_msg: rtrs_err_rl(s, "Processing write request failed, sending I/O response failed, msg_id %d, err: %d\n", buf_id, ret); - close_sess(sess); + close_path(srv_path); } - rtrs_srv_put_ops_ids(sess); + rtrs_srv_put_ops_ids(srv_path); } static void process_io_req(struct rtrs_srv_con *con, void *msg, u32 id, u32 off) { - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); struct rtrs_msg_rdma_hdr *hdr; unsigned int type; - ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, sess->dma_addr[id], + ib_dma_sync_single_for_cpu(srv_path->s.dev->ib_dev, + srv_path->dma_addr[id], max_chunk_size, DMA_BIDIRECTIONAL); hdr = msg; type = le16_to_cpu(hdr->type); @@ -1129,7 +1128,7 @@ static void process_io_req(struct rtrs_srv_con *con, void *msg, return; err: - close_sess(sess); + close_path(srv_path); } static void rtrs_srv_inv_rkey_done(struct ib_cq *cq, struct ib_wc *wc) @@ -1137,16 +1136,16 @@ static void rtrs_srv_inv_rkey_done(struct ib_cq *cq, struct ib_wc *wc) struct rtrs_srv_mr *mr = container_of(wc->wr_cqe, typeof(*mr), inv_cqe); struct rtrs_srv_con *con = to_srv_con(wc->qp->qp_context); - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); - struct rtrs_srv *srv = sess->srv; + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); + struct rtrs_srv_sess *srv = srv_path->srv; u32 msg_id, off; void *data; if (wc->status != IB_WC_SUCCESS) { rtrs_err(s, "Failed IB_WR_LOCAL_INV: %s\n", ib_wc_status_msg(wc->status)); - close_sess(sess); + close_path(srv_path); } msg_id = mr->msg_id; off = mr->msg_off; @@ -1194,9 +1193,9 @@ static void rtrs_rdma_process_wr_wait_list(struct rtrs_srv_con *con) static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) { struct rtrs_srv_con *con = to_srv_con(wc->qp->qp_context); - struct rtrs_sess *s = con->c.sess; - struct rtrs_srv_sess *sess = to_srv_sess(s); - struct rtrs_srv *srv = sess->srv; + struct rtrs_path *s = con->c.path; + struct rtrs_srv_path *srv_path = to_srv_path(s); + struct rtrs_srv_sess *srv = srv_path->srv; u32 imm_type, imm_payload; int err; @@ -1206,7 +1205,7 @@ static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) "%s (wr_cqe: %p, type: %d, vendor_err: 0x%x, len: %u)\n", ib_wc_status_msg(wc->status), wc->wr_cqe, wc->opcode, wc->vendor_err, wc->byte_len); - close_sess(sess); + close_path(srv_path); } return; } @@ -1222,7 +1221,7 @@ static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) err = rtrs_post_recv_empty(&con->c, &io_comp_cqe); if (err) { rtrs_err(s, "rtrs_post_recv(), err: %d\n", err); - close_sess(sess); + close_path(srv_path); break; } rtrs_from_imm(be32_to_cpu(wc->ex.imm_data), @@ -1231,16 +1230,16 @@ static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) u32 msg_id, off; void *data; - msg_id = imm_payload >> sess->mem_bits; - off = imm_payload & ((1 << sess->mem_bits) - 1); + msg_id = imm_payload >> srv_path->mem_bits; + off = imm_payload & ((1 << srv_path->mem_bits) - 1); if (msg_id >= srv->queue_depth || off >= max_chunk_size) { rtrs_err(s, "Wrong msg_id %u, off %u\n", msg_id, off); - close_sess(sess); + close_path(srv_path); return; } if (always_invalidate) { - struct rtrs_srv_mr *mr = &sess->mrs[msg_id]; + struct rtrs_srv_mr *mr = &srv_path->mrs[msg_id]; mr->msg_off = off; mr->msg_id = msg_id; @@ -1248,7 +1247,7 @@ static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) if (err) { rtrs_err(s, "rtrs_post_recv(), err: %d\n", err); - close_sess(sess); + close_path(srv_path); break; } } else { @@ -1257,10 +1256,10 @@ static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) } } else if (imm_type == RTRS_HB_MSG_IMM) { WARN_ON(con->c.cid); - rtrs_send_hb_ack(&sess->s); + rtrs_send_hb_ack(&srv_path->s); } else if (imm_type == RTRS_HB_ACK_IMM) { WARN_ON(con->c.cid); - sess->s.hb_missed_cnt = 0; + srv_path->s.hb_missed_cnt = 0; } else { rtrs_wrn(s, "Unknown IMM type %u\n", imm_type); } @@ -1284,22 +1283,23 @@ static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) } /** - * rtrs_srv_get_sess_name() - Get rtrs_srv peer hostname. + * rtrs_srv_get_path_name() - Get rtrs_srv peer hostname. * @srv: Session - * @sessname: Sessname buffer + * @pathname: Pathname buffer * @len: Length of sessname buffer */ -int rtrs_srv_get_sess_name(struct rtrs_srv *srv, char *sessname, size_t len) +int rtrs_srv_get_path_name(struct rtrs_srv_sess *srv, char *pathname, + size_t len) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; int err = -ENOTCONN; mutex_lock(&srv->paths_mutex); - list_for_each_entry(sess, &srv->paths_list, s.entry) { - if (sess->state != RTRS_SRV_CONNECTED) + list_for_each_entry(srv_path, &srv->paths_list, s.entry) { + if (srv_path->state != RTRS_SRV_CONNECTED) continue; - strscpy(sessname, sess->s.sessname, - min_t(size_t, sizeof(sess->s.sessname), len)); + strscpy(pathname, srv_path->s.sessname, + min_t(size_t, sizeof(srv_path->s.sessname), len)); err = 0; break; } @@ -1307,50 +1307,51 @@ int rtrs_srv_get_sess_name(struct rtrs_srv *srv, char *sessname, size_t len) return err; } -EXPORT_SYMBOL(rtrs_srv_get_sess_name); +EXPORT_SYMBOL(rtrs_srv_get_path_name); /** * rtrs_srv_get_queue_depth() - Get rtrs_srv qdepth. * @srv: Session */ -int rtrs_srv_get_queue_depth(struct rtrs_srv *srv) +int rtrs_srv_get_queue_depth(struct rtrs_srv_sess *srv) { return srv->queue_depth; } EXPORT_SYMBOL(rtrs_srv_get_queue_depth); -static int find_next_bit_ring(struct rtrs_srv_sess *sess) +static int find_next_bit_ring(struct rtrs_srv_path *srv_path) { - struct ib_device *ib_dev = sess->s.dev->ib_dev; + struct ib_device *ib_dev = srv_path->s.dev->ib_dev; int v; - v = cpumask_next(sess->cur_cq_vector, &cq_affinity_mask); + v = cpumask_next(srv_path->cur_cq_vector, &cq_affinity_mask); if (v >= nr_cpu_ids || v >= ib_dev->num_comp_vectors) v = cpumask_first(&cq_affinity_mask); return v; } -static int rtrs_srv_get_next_cq_vector(struct rtrs_srv_sess *sess) +static int rtrs_srv_get_next_cq_vector(struct rtrs_srv_path *srv_path) { - sess->cur_cq_vector = find_next_bit_ring(sess); + srv_path->cur_cq_vector = find_next_bit_ring(srv_path); - return sess->cur_cq_vector; + return srv_path->cur_cq_vector; } static void rtrs_srv_dev_release(struct device *dev) { - struct rtrs_srv *srv = container_of(dev, struct rtrs_srv, dev); + struct rtrs_srv_sess *srv = container_of(dev, struct rtrs_srv_sess, + dev); kfree(srv); } -static void free_srv(struct rtrs_srv *srv) +static void free_srv(struct rtrs_srv_sess *srv) { int i; WARN_ON(refcount_read(&srv->refcount)); for (i = 0; i < srv->queue_depth; i++) - mempool_free(srv->chunks[i], chunk_pool); + __free_pages(srv->chunks[i], get_order(max_chunk_size)); kfree(srv->chunks); mutex_destroy(&srv->paths_mutex); mutex_destroy(&srv->paths_ev_mutex); @@ -1358,11 +1359,11 @@ static void free_srv(struct rtrs_srv *srv) put_device(&srv->dev); } -static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, +static struct rtrs_srv_sess *get_or_create_srv(struct rtrs_srv_ctx *ctx, const uuid_t *paths_uuid, bool first_conn) { - struct rtrs_srv *srv; + struct rtrs_srv_sess *srv; int i; mutex_lock(&ctx->srv_mutex); @@ -1403,7 +1404,8 @@ static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, goto err_free_srv; for (i = 0; i < srv->queue_depth; i++) { - srv->chunks[i] = mempool_alloc(chunk_pool, GFP_KERNEL); + srv->chunks[i] = alloc_pages(GFP_KERNEL, + get_order(max_chunk_size)); if (!srv->chunks[i]) goto err_free_chunks; } @@ -1416,7 +1418,7 @@ static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, err_free_chunks: while (i--) - mempool_free(srv->chunks[i], chunk_pool); + __free_pages(srv->chunks[i], get_order(max_chunk_size)); kfree(srv->chunks); err_free_srv: @@ -1424,7 +1426,7 @@ err_free_srv: return ERR_PTR(-ENOMEM); } -static void put_srv(struct rtrs_srv *srv) +static void put_srv(struct rtrs_srv_sess *srv) { if (refcount_dec_and_test(&srv->refcount)) { struct rtrs_srv_ctx *ctx = srv->ctx; @@ -1438,23 +1440,23 @@ static void put_srv(struct rtrs_srv *srv) } } -static void __add_path_to_srv(struct rtrs_srv *srv, - struct rtrs_srv_sess *sess) +static void __add_path_to_srv(struct rtrs_srv_sess *srv, + struct rtrs_srv_path *srv_path) { - list_add_tail(&sess->s.entry, &srv->paths_list); + list_add_tail(&srv_path->s.entry, &srv->paths_list); srv->paths_num++; WARN_ON(srv->paths_num >= MAX_PATHS_NUM); } -static void del_path_from_srv(struct rtrs_srv_sess *sess) +static void del_path_from_srv(struct rtrs_srv_path *srv_path) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; if (WARN_ON(!srv)) return; mutex_lock(&srv->paths_mutex); - list_del(&sess->s.entry); + list_del(&srv_path->s.entry); WARN_ON(!srv->paths_num); srv->paths_num--; mutex_unlock(&srv->paths_mutex); @@ -1484,47 +1486,48 @@ static int sockaddr_cmp(const struct sockaddr *a, const struct sockaddr *b) } } -static bool __is_path_w_addr_exists(struct rtrs_srv *srv, +static bool __is_path_w_addr_exists(struct rtrs_srv_sess *srv, struct rdma_addr *addr) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; - list_for_each_entry(sess, &srv->paths_list, s.entry) - if (!sockaddr_cmp((struct sockaddr *)&sess->s.dst_addr, + list_for_each_entry(srv_path, &srv->paths_list, s.entry) + if (!sockaddr_cmp((struct sockaddr *)&srv_path->s.dst_addr, (struct sockaddr *)&addr->dst_addr) && - !sockaddr_cmp((struct sockaddr *)&sess->s.src_addr, + !sockaddr_cmp((struct sockaddr *)&srv_path->s.src_addr, (struct sockaddr *)&addr->src_addr)) return true; return false; } -static void free_sess(struct rtrs_srv_sess *sess) +static void free_path(struct rtrs_srv_path *srv_path) { - if (sess->kobj.state_in_sysfs) { - kobject_del(&sess->kobj); - kobject_put(&sess->kobj); + if (srv_path->kobj.state_in_sysfs) { + kobject_del(&srv_path->kobj); + kobject_put(&srv_path->kobj); } else { - kfree(sess->stats); - kfree(sess); + free_percpu(srv_path->stats->rdma_stats); + kfree(srv_path->stats); + kfree(srv_path); } } static void rtrs_srv_close_work(struct work_struct *work) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; struct rtrs_srv_con *con; int i; - sess = container_of(work, typeof(*sess), close_work); + srv_path = container_of(work, typeof(*srv_path), close_work); - rtrs_srv_destroy_sess_files(sess); - rtrs_srv_stop_hb(sess); + rtrs_srv_destroy_path_files(srv_path); + rtrs_srv_stop_hb(srv_path); - for (i = 0; i < sess->s.con_num; i++) { - if (!sess->s.con[i]) + for (i = 0; i < srv_path->s.con_num; i++) { + if (!srv_path->s.con[i]) continue; - con = to_srv_con(sess->s.con[i]); + con = to_srv_con(srv_path->s.con[i]); rdma_disconnect(con->c.cm_id); ib_drain_qp(con->c.qp); } @@ -1533,41 +1536,41 @@ static void rtrs_srv_close_work(struct work_struct *work) * Degrade ref count to the usual model with a single shared * atomic_t counter */ - percpu_ref_kill(&sess->ids_inflight_ref); + percpu_ref_kill(&srv_path->ids_inflight_ref); /* Wait for all completion */ - wait_for_completion(&sess->complete_done); + wait_for_completion(&srv_path->complete_done); /* Notify upper layer if we are the last path */ - rtrs_srv_sess_down(sess); + rtrs_srv_path_down(srv_path); - unmap_cont_bufs(sess); - rtrs_srv_free_ops_ids(sess); + unmap_cont_bufs(srv_path); + rtrs_srv_free_ops_ids(srv_path); - for (i = 0; i < sess->s.con_num; i++) { - if (!sess->s.con[i]) + for (i = 0; i < srv_path->s.con_num; i++) { + if (!srv_path->s.con[i]) continue; - con = to_srv_con(sess->s.con[i]); + con = to_srv_con(srv_path->s.con[i]); rtrs_cq_qp_destroy(&con->c); rdma_destroy_id(con->c.cm_id); kfree(con); } - rtrs_ib_dev_put(sess->s.dev); + rtrs_ib_dev_put(srv_path->s.dev); - del_path_from_srv(sess); - put_srv(sess->srv); - sess->srv = NULL; - rtrs_srv_change_state(sess, RTRS_SRV_CLOSED); + del_path_from_srv(srv_path); + put_srv(srv_path->srv); + srv_path->srv = NULL; + rtrs_srv_change_state(srv_path, RTRS_SRV_CLOSED); - kfree(sess->dma_addr); - kfree(sess->s.con); - free_sess(sess); + kfree(srv_path->dma_addr); + kfree(srv_path->s.con); + free_path(srv_path); } -static int rtrs_rdma_do_accept(struct rtrs_srv_sess *sess, +static int rtrs_rdma_do_accept(struct rtrs_srv_path *srv_path, struct rdma_cm_id *cm_id) { - struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_sess *srv = srv_path->srv; struct rtrs_msg_conn_rsp msg; struct rdma_conn_param param; int err; @@ -1615,25 +1618,25 @@ static int rtrs_rdma_do_reject(struct rdma_cm_id *cm_id, int errno) return errno; } -static struct rtrs_srv_sess * -__find_sess(struct rtrs_srv *srv, const uuid_t *sess_uuid) +static struct rtrs_srv_path * +__find_path(struct rtrs_srv_sess *srv, const uuid_t *sess_uuid) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; - list_for_each_entry(sess, &srv->paths_list, s.entry) { - if (uuid_equal(&sess->s.uuid, sess_uuid)) - return sess; + list_for_each_entry(srv_path, &srv->paths_list, s.entry) { + if (uuid_equal(&srv_path->s.uuid, sess_uuid)) + return srv_path; } return NULL; } -static int create_con(struct rtrs_srv_sess *sess, +static int create_con(struct rtrs_srv_path *srv_path, struct rdma_cm_id *cm_id, unsigned int cid) { - struct rtrs_srv *srv = sess->srv; - struct rtrs_sess *s = &sess->s; + struct rtrs_srv_sess *srv = srv_path->srv; + struct rtrs_path *s = &srv_path->s; struct rtrs_srv_con *con; u32 cq_num, max_send_wr, max_recv_wr, wr_limit; @@ -1648,10 +1651,10 @@ static int create_con(struct rtrs_srv_sess *sess, spin_lock_init(&con->rsp_wr_wait_lock); INIT_LIST_HEAD(&con->rsp_wr_wait_list); con->c.cm_id = cm_id; - con->c.sess = &sess->s; + con->c.path = &srv_path->s; con->c.cid = cid; atomic_set(&con->c.wr_cnt, 1); - wr_limit = sess->s.dev->ib_dev->attrs.max_qp_wr; + wr_limit = srv_path->s.dev->ib_dev->attrs.max_qp_wr; if (con->c.cid == 0) { /* @@ -1684,10 +1687,10 @@ static int create_con(struct rtrs_srv_sess *sess, } cq_num = max_send_wr + max_recv_wr; atomic_set(&con->c.sq_wr_avail, max_send_wr); - cq_vector = rtrs_srv_get_next_cq_vector(sess); + cq_vector = rtrs_srv_get_next_cq_vector(srv_path); /* TODO: SOFTIRQ can be faster, but be careful with softirq context */ - err = rtrs_cq_qp_create(&sess->s, &con->c, 1, cq_vector, cq_num, + err = rtrs_cq_qp_create(&srv_path->s, &con->c, 1, cq_vector, cq_num, max_send_wr, max_recv_wr, IB_POLL_WORKQUEUE); if (err) { @@ -1699,8 +1702,8 @@ static int create_con(struct rtrs_srv_sess *sess, if (err) goto free_cqqp; } - WARN_ON(sess->s.con[cid]); - sess->s.con[cid] = &con->c; + WARN_ON(srv_path->s.con[cid]); + srv_path->s.con[cid] = &con->c; /* * Change context from server to current connection. The other @@ -1719,13 +1722,13 @@ err: return err; } -static struct rtrs_srv_sess *__alloc_sess(struct rtrs_srv *srv, +static struct rtrs_srv_path *__alloc_path(struct rtrs_srv_sess *srv, struct rdma_cm_id *cm_id, unsigned int con_num, unsigned int recon_cnt, const uuid_t *uuid) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; int err = -ENOMEM; char str[NAME_MAX]; struct rtrs_addr path; @@ -1739,74 +1742,82 @@ static struct rtrs_srv_sess *__alloc_sess(struct rtrs_srv *srv, pr_err("Path with same addr exists\n"); goto err; } - sess = kzalloc(sizeof(*sess), GFP_KERNEL); - if (!sess) + srv_path = kzalloc(sizeof(*srv_path), GFP_KERNEL); + if (!srv_path) goto err; - sess->stats = kzalloc(sizeof(*sess->stats), GFP_KERNEL); - if (!sess->stats) + srv_path->stats = kzalloc(sizeof(*srv_path->stats), GFP_KERNEL); + if (!srv_path->stats) goto err_free_sess; - sess->stats->sess = sess; - - sess->dma_addr = kcalloc(srv->queue_depth, sizeof(*sess->dma_addr), - GFP_KERNEL); - if (!sess->dma_addr) + srv_path->stats->rdma_stats = alloc_percpu(struct rtrs_srv_stats_rdma_stats); + if (!srv_path->stats->rdma_stats) goto err_free_stats; - sess->s.con = kcalloc(con_num, sizeof(*sess->s.con), GFP_KERNEL); - if (!sess->s.con) + srv_path->stats->srv_path = srv_path; + + srv_path->dma_addr = kcalloc(srv->queue_depth, + sizeof(*srv_path->dma_addr), + GFP_KERNEL); + if (!srv_path->dma_addr) + goto err_free_percpu; + + srv_path->s.con = kcalloc(con_num, sizeof(*srv_path->s.con), + GFP_KERNEL); + if (!srv_path->s.con) goto err_free_dma_addr; - sess->state = RTRS_SRV_CONNECTING; - sess->srv = srv; - sess->cur_cq_vector = -1; - sess->s.dst_addr = cm_id->route.addr.dst_addr; - sess->s.src_addr = cm_id->route.addr.src_addr; + srv_path->state = RTRS_SRV_CONNECTING; + srv_path->srv = srv; + srv_path->cur_cq_vector = -1; + srv_path->s.dst_addr = cm_id->route.addr.dst_addr; + srv_path->s.src_addr = cm_id->route.addr.src_addr; /* temporary until receiving session-name from client */ - path.src = &sess->s.src_addr; - path.dst = &sess->s.dst_addr; + path.src = &srv_path->s.src_addr; + path.dst = &srv_path->s.dst_addr; rtrs_addr_to_str(&path, str, sizeof(str)); - strscpy(sess->s.sessname, str, sizeof(sess->s.sessname)); - - sess->s.con_num = con_num; - sess->s.irq_con_num = con_num; - sess->s.recon_cnt = recon_cnt; - uuid_copy(&sess->s.uuid, uuid); - spin_lock_init(&sess->state_lock); - INIT_WORK(&sess->close_work, rtrs_srv_close_work); - rtrs_srv_init_hb(sess); - - sess->s.dev = rtrs_ib_dev_find_or_add(cm_id->device, &dev_pd); - if (!sess->s.dev) { + strscpy(srv_path->s.sessname, str, sizeof(srv_path->s.sessname)); + + srv_path->s.con_num = con_num; + srv_path->s.irq_con_num = con_num; + srv_path->s.recon_cnt = recon_cnt; + uuid_copy(&srv_path->s.uuid, uuid); + spin_lock_init(&srv_path->state_lock); + INIT_WORK(&srv_path->close_work, rtrs_srv_close_work); + rtrs_srv_init_hb(srv_path); + + srv_path->s.dev = rtrs_ib_dev_find_or_add(cm_id->device, &dev_pd); + if (!srv_path->s.dev) { err = -ENOMEM; goto err_free_con; } - err = map_cont_bufs(sess); + err = map_cont_bufs(srv_path); if (err) goto err_put_dev; - err = rtrs_srv_alloc_ops_ids(sess); + err = rtrs_srv_alloc_ops_ids(srv_path); if (err) goto err_unmap_bufs; - __add_path_to_srv(srv, sess); + __add_path_to_srv(srv, srv_path); - return sess; + return srv_path; err_unmap_bufs: - unmap_cont_bufs(sess); + unmap_cont_bufs(srv_path); err_put_dev: - rtrs_ib_dev_put(sess->s.dev); + rtrs_ib_dev_put(srv_path->s.dev); err_free_con: - kfree(sess->s.con); + kfree(srv_path->s.con); err_free_dma_addr: - kfree(sess->dma_addr); + kfree(srv_path->dma_addr); +err_free_percpu: + free_percpu(srv_path->stats->rdma_stats); err_free_stats: - kfree(sess->stats); + kfree(srv_path->stats); err_free_sess: - kfree(sess); + kfree(srv_path); err: return ERR_PTR(err); } @@ -1816,8 +1827,8 @@ static int rtrs_rdma_connect(struct rdma_cm_id *cm_id, size_t len) { struct rtrs_srv_ctx *ctx = cm_id->context; - struct rtrs_srv_sess *sess; - struct rtrs_srv *srv; + struct rtrs_srv_path *srv_path; + struct rtrs_srv_sess *srv; u16 version, con_num, cid; u16 recon_cnt; @@ -1857,16 +1868,16 @@ static int rtrs_rdma_connect(struct rdma_cm_id *cm_id, goto reject_w_err; } mutex_lock(&srv->paths_mutex); - sess = __find_sess(srv, &msg->sess_uuid); - if (sess) { - struct rtrs_sess *s = &sess->s; + srv_path = __find_path(srv, &msg->sess_uuid); + if (srv_path) { + struct rtrs_path *s = &srv_path->s; /* Session already holds a reference */ put_srv(srv); - if (sess->state != RTRS_SRV_CONNECTING) { + if (srv_path->state != RTRS_SRV_CONNECTING) { rtrs_err(s, "Session in wrong state: %s\n", - rtrs_srv_state_str(sess->state)); + rtrs_srv_state_str(srv_path->state)); mutex_unlock(&srv->paths_mutex); goto reject_w_err; } @@ -1886,19 +1897,19 @@ static int rtrs_rdma_connect(struct rdma_cm_id *cm_id, goto reject_w_err; } } else { - sess = __alloc_sess(srv, cm_id, con_num, recon_cnt, + srv_path = __alloc_path(srv, cm_id, con_num, recon_cnt, &msg->sess_uuid); - if (IS_ERR(sess)) { + if (IS_ERR(srv_path)) { mutex_unlock(&srv->paths_mutex); put_srv(srv); - err = PTR_ERR(sess); + err = PTR_ERR(srv_path); pr_err("RTRS server session allocation failed: %d\n", err); goto reject_w_err; } } - err = create_con(sess, cm_id, cid); + err = create_con(srv_path, cm_id, cid); if (err) { - rtrs_err((&sess->s), "create_con(), error %d\n", err); + rtrs_err((&srv_path->s), "create_con(), error %d\n", err); rtrs_rdma_do_reject(cm_id, err); /* * Since session has other connections we follow normal way @@ -1907,9 +1918,9 @@ static int rtrs_rdma_connect(struct rdma_cm_id *cm_id, */ goto close_and_return_err; } - err = rtrs_rdma_do_accept(sess, cm_id); + err = rtrs_rdma_do_accept(srv_path, cm_id); if (err) { - rtrs_err((&sess->s), "rtrs_rdma_do_accept(), error %d\n", err); + rtrs_err((&srv_path->s), "rtrs_rdma_do_accept(), error %d\n", err); rtrs_rdma_do_reject(cm_id, err); /* * Since current connection was successfully added to the @@ -1929,7 +1940,7 @@ reject_w_err: close_and_return_err: mutex_unlock(&srv->paths_mutex); - close_sess(sess); + close_path(srv_path); return err; } @@ -1937,14 +1948,14 @@ close_and_return_err: static int rtrs_srv_rdma_cm_handler(struct rdma_cm_id *cm_id, struct rdma_cm_event *ev) { - struct rtrs_srv_sess *sess = NULL; - struct rtrs_sess *s = NULL; + struct rtrs_srv_path *srv_path = NULL; + struct rtrs_path *s = NULL; if (ev->event != RDMA_CM_EVENT_CONNECT_REQUEST) { struct rtrs_con *c = cm_id->context; - s = c->sess; - sess = to_srv_sess(s); + s = c->path; + srv_path = to_srv_path(s); } switch (ev->event) { @@ -1968,7 +1979,7 @@ static int rtrs_srv_rdma_cm_handler(struct rdma_cm_id *cm_id, case RDMA_CM_EVENT_ADDR_CHANGE: case RDMA_CM_EVENT_TIMEWAIT_EXIT: case RDMA_CM_EVENT_DEVICE_REMOVAL: - close_sess(sess); + close_path(srv_path); break; default: pr_err("Ignoring unexpected CM event %s, err %d\n", @@ -2176,23 +2187,23 @@ struct rtrs_srv_ctx *rtrs_srv_open(struct rtrs_srv_ops *ops, u16 port) } EXPORT_SYMBOL(rtrs_srv_open); -static void close_sessions(struct rtrs_srv *srv) +static void close_paths(struct rtrs_srv_sess *srv) { - struct rtrs_srv_sess *sess; + struct rtrs_srv_path *srv_path; mutex_lock(&srv->paths_mutex); - list_for_each_entry(sess, &srv->paths_list, s.entry) - close_sess(sess); + list_for_each_entry(srv_path, &srv->paths_list, s.entry) + close_path(srv_path); mutex_unlock(&srv->paths_mutex); } static void close_ctx(struct rtrs_srv_ctx *ctx) { - struct rtrs_srv *srv; + struct rtrs_srv_sess *srv; mutex_lock(&ctx->srv_mutex); list_for_each_entry(srv, &ctx->srv_list, ctx_list) - close_sessions(srv); + close_paths(srv); mutex_unlock(&ctx->srv_mutex); flush_workqueue(rtrs_wq); } @@ -2256,14 +2267,10 @@ static int __init rtrs_server_init(void) err); return err; } - chunk_pool = mempool_create_page_pool(sess_queue_depth * CHUNK_POOL_SZ, - get_order(max_chunk_size)); - if (!chunk_pool) - return -ENOMEM; rtrs_dev_class = class_create(THIS_MODULE, "rtrs-server"); if (IS_ERR(rtrs_dev_class)) { err = PTR_ERR(rtrs_dev_class); - goto out_chunk_pool; + goto out_err; } rtrs_wq = alloc_workqueue("rtrs_server_wq", 0, 0); if (!rtrs_wq) { @@ -2275,9 +2282,7 @@ static int __init rtrs_server_init(void) out_dev_class: class_destroy(rtrs_dev_class); -out_chunk_pool: - mempool_destroy(chunk_pool); - +out_err: return err; } @@ -2285,7 +2290,6 @@ static void __exit rtrs_server_exit(void) { destroy_workqueue(rtrs_wq); class_destroy(rtrs_dev_class); - mempool_destroy(chunk_pool); rtrs_rdma_dev_pd_deinit(&dev_pd); } diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv.h b/drivers/infiniband/ulp/rtrs/rtrs-srv.h index 7d403c12faf3..2f8a638e36fa 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-srv.h +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv.h @@ -12,6 +12,7 @@ #include <linux/device.h> #include <linux/refcount.h> +#include <linux/percpu.h> #include "rtrs-pri.h" /* @@ -29,15 +30,15 @@ enum rtrs_srv_state { */ struct rtrs_srv_stats_rdma_stats { struct { - atomic64_t cnt; - atomic64_t size_total; + u64 cnt; + u64 size_total; } dir[2]; }; struct rtrs_srv_stats { - struct kobject kobj_stats; - struct rtrs_srv_stats_rdma_stats rdma_stats; - struct rtrs_srv_sess *sess; + struct kobject kobj_stats; + struct rtrs_srv_stats_rdma_stats __percpu *rdma_stats; + struct rtrs_srv_path *srv_path; }; struct rtrs_srv_con { @@ -71,9 +72,9 @@ struct rtrs_srv_mr { struct rtrs_iu *iu; /* send buffer for new rkey msg */ }; -struct rtrs_srv_sess { - struct rtrs_sess s; - struct rtrs_srv *srv; +struct rtrs_srv_path { + struct rtrs_path s; + struct rtrs_srv_sess *srv; struct work_struct close_work; enum rtrs_srv_state state; spinlock_t state_lock; @@ -90,7 +91,12 @@ struct rtrs_srv_sess { struct rtrs_srv_stats *stats; }; -struct rtrs_srv { +static inline struct rtrs_srv_path *to_srv_path(struct rtrs_path *s) +{ + return container_of(s, struct rtrs_srv_path, s); +} + +struct rtrs_srv_sess { struct list_head paths_list; int paths_up; struct mutex paths_ev_mutex; @@ -125,13 +131,13 @@ struct rtrs_srv_ib_ctx { extern struct class *rtrs_dev_class; -void close_sess(struct rtrs_srv_sess *sess); +void close_path(struct rtrs_srv_path *srv_path); static inline void rtrs_srv_update_rdma_stats(struct rtrs_srv_stats *s, size_t size, int d) { - atomic64_inc(&s->rdma_stats.dir[d].cnt); - atomic64_add(size, &s->rdma_stats.dir[d].size_total); + this_cpu_inc(s->rdma_stats->dir[d].cnt); + this_cpu_add(s->rdma_stats->dir[d].size_total, size); } /* functions which are implemented in rtrs-srv-stats.c */ @@ -142,7 +148,7 @@ ssize_t rtrs_srv_reset_all_help(struct rtrs_srv_stats *stats, char *page, size_t len); /* functions which are implemented in rtrs-srv-sysfs.c */ -int rtrs_srv_create_sess_files(struct rtrs_srv_sess *sess); -void rtrs_srv_destroy_sess_files(struct rtrs_srv_sess *sess); +int rtrs_srv_create_path_files(struct rtrs_srv_path *srv_path); +void rtrs_srv_destroy_path_files(struct rtrs_srv_path *srv_path); #endif /* RTRS_SRV_H */ diff --git a/drivers/infiniband/ulp/rtrs/rtrs.c b/drivers/infiniband/ulp/rtrs/rtrs.c index 37952c8e768c..ed324b47d93a 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs.c +++ b/drivers/infiniband/ulp/rtrs/rtrs.c @@ -69,16 +69,16 @@ EXPORT_SYMBOL_GPL(rtrs_iu_free); int rtrs_iu_post_recv(struct rtrs_con *con, struct rtrs_iu *iu) { - struct rtrs_sess *sess = con->sess; + struct rtrs_path *path = con->path; struct ib_recv_wr wr; struct ib_sge list; list.addr = iu->dma_addr; list.length = iu->size; - list.lkey = sess->dev->ib_pd->local_dma_lkey; + list.lkey = path->dev->ib_pd->local_dma_lkey; if (list.length == 0) { - rtrs_wrn(con->sess, + rtrs_wrn(con->path, "Posting receive work request failed, sg list is empty\n"); return -EINVAL; } @@ -126,7 +126,7 @@ static int rtrs_post_send(struct ib_qp *qp, struct ib_send_wr *head, int rtrs_iu_post_send(struct rtrs_con *con, struct rtrs_iu *iu, size_t size, struct ib_send_wr *head) { - struct rtrs_sess *sess = con->sess; + struct rtrs_path *path = con->path; struct ib_send_wr wr; struct ib_sge list; @@ -135,7 +135,7 @@ int rtrs_iu_post_send(struct rtrs_con *con, struct rtrs_iu *iu, size_t size, list.addr = iu->dma_addr; list.length = size; - list.lkey = sess->dev->ib_pd->local_dma_lkey; + list.lkey = path->dev->ib_pd->local_dma_lkey; wr = (struct ib_send_wr) { .wr_cqe = &iu->cqe, @@ -175,7 +175,7 @@ int rtrs_iu_post_rdma_write_imm(struct rtrs_con *con, struct rtrs_iu *iu, * length error */ for (i = 0; i < num_sge; i++) - if (WARN_ON(sge[i].length == 0)) + if (WARN_ONCE(sge[i].length == 0, "sg %d is zero length\n", i)) return -EINVAL; return rtrs_post_send(con->qp, head, &wr.wr, tail); @@ -188,11 +188,11 @@ static int rtrs_post_rdma_write_imm_empty(struct rtrs_con *con, struct ib_send_wr *head) { struct ib_rdma_wr wr; - struct rtrs_sess *sess = con->sess; + struct rtrs_path *path = con->path; enum ib_send_flags sflags; atomic_dec_if_positive(&con->sq_wr_avail); - sflags = (atomic_inc_return(&con->wr_cnt) % sess->signal_interval) ? + sflags = (atomic_inc_return(&con->wr_cnt) % path->signal_interval) ? 0 : IB_SEND_SIGNALED; wr = (struct ib_rdma_wr) { @@ -211,12 +211,12 @@ static void qp_event_handler(struct ib_event *ev, void *ctx) switch (ev->event) { case IB_EVENT_COMM_EST: - rtrs_info(con->sess, "QP event %s (%d) received\n", + rtrs_info(con->path, "QP event %s (%d) received\n", ib_event_msg(ev->event), ev->event); rdma_notify(con->cm_id, IB_EVENT_COMM_EST); break; default: - rtrs_info(con->sess, "Unhandled QP event %s (%d) received\n", + rtrs_info(con->path, "Unhandled QP event %s (%d) received\n", ib_event_msg(ev->event), ev->event); break; } @@ -224,7 +224,7 @@ static void qp_event_handler(struct ib_event *ev, void *ctx) static bool is_pollqueue(struct rtrs_con *con) { - return con->cid >= con->sess->irq_con_num; + return con->cid >= con->path->irq_con_num; } static int create_cq(struct rtrs_con *con, int cq_vector, int nr_cqe, @@ -240,7 +240,7 @@ static int create_cq(struct rtrs_con *con, int cq_vector, int nr_cqe, cq = ib_cq_pool_get(cm_id->device, nr_cqe, cq_vector, poll_ctx); if (IS_ERR(cq)) { - rtrs_err(con->sess, "Creating completion queue failed, errno: %ld\n", + rtrs_err(con->path, "Creating completion queue failed, errno: %ld\n", PTR_ERR(cq)); return PTR_ERR(cq); } @@ -271,7 +271,7 @@ static int create_qp(struct rtrs_con *con, struct ib_pd *pd, ret = rdma_create_qp(cm_id, pd, &init_attr); if (ret) { - rtrs_err(con->sess, "Creating QP failed, err: %d\n", ret); + rtrs_err(con->path, "Creating QP failed, err: %d\n", ret); return ret; } con->qp = cm_id->qp; @@ -290,7 +290,7 @@ static void destroy_cq(struct rtrs_con *con) con->cq = NULL; } -int rtrs_cq_qp_create(struct rtrs_sess *sess, struct rtrs_con *con, +int rtrs_cq_qp_create(struct rtrs_path *path, struct rtrs_con *con, u32 max_send_sge, int cq_vector, int nr_cqe, u32 max_send_wr, u32 max_recv_wr, enum ib_poll_context poll_ctx) @@ -301,13 +301,13 @@ int rtrs_cq_qp_create(struct rtrs_sess *sess, struct rtrs_con *con, if (err) return err; - err = create_qp(con, sess->dev->ib_pd, max_send_wr, max_recv_wr, + err = create_qp(con, path->dev->ib_pd, max_send_wr, max_recv_wr, max_send_sge); if (err) { destroy_cq(con); return err; } - con->sess = sess; + con->path = path; return 0; } @@ -323,24 +323,24 @@ void rtrs_cq_qp_destroy(struct rtrs_con *con) } EXPORT_SYMBOL_GPL(rtrs_cq_qp_destroy); -static void schedule_hb(struct rtrs_sess *sess) +static void schedule_hb(struct rtrs_path *path) { - queue_delayed_work(sess->hb_wq, &sess->hb_dwork, - msecs_to_jiffies(sess->hb_interval_ms)); + queue_delayed_work(path->hb_wq, &path->hb_dwork, + msecs_to_jiffies(path->hb_interval_ms)); } -void rtrs_send_hb_ack(struct rtrs_sess *sess) +void rtrs_send_hb_ack(struct rtrs_path *path) { - struct rtrs_con *usr_con = sess->con[0]; + struct rtrs_con *usr_con = path->con[0]; u32 imm; int err; imm = rtrs_to_imm(RTRS_HB_ACK_IMM, 0); - err = rtrs_post_rdma_write_imm_empty(usr_con, sess->hb_cqe, imm, + err = rtrs_post_rdma_write_imm_empty(usr_con, path->hb_cqe, imm, NULL); if (err) { - rtrs_err(sess, "send HB ACK failed, errno: %d\n", err); - sess->hb_err_handler(usr_con); + rtrs_err(path, "send HB ACK failed, errno: %d\n", err); + path->hb_err_handler(usr_con); return; } } @@ -349,63 +349,63 @@ EXPORT_SYMBOL_GPL(rtrs_send_hb_ack); static void hb_work(struct work_struct *work) { struct rtrs_con *usr_con; - struct rtrs_sess *sess; + struct rtrs_path *path; u32 imm; int err; - sess = container_of(to_delayed_work(work), typeof(*sess), hb_dwork); - usr_con = sess->con[0]; + path = container_of(to_delayed_work(work), typeof(*path), hb_dwork); + usr_con = path->con[0]; - if (sess->hb_missed_cnt > sess->hb_missed_max) { - rtrs_err(sess, "HB missed max reached.\n"); - sess->hb_err_handler(usr_con); + if (path->hb_missed_cnt > path->hb_missed_max) { + rtrs_err(path, "HB missed max reached.\n"); + path->hb_err_handler(usr_con); return; } - if (sess->hb_missed_cnt++) { + if (path->hb_missed_cnt++) { /* Reschedule work without sending hb */ - schedule_hb(sess); + schedule_hb(path); return; } - sess->hb_last_sent = ktime_get(); + path->hb_last_sent = ktime_get(); imm = rtrs_to_imm(RTRS_HB_MSG_IMM, 0); - err = rtrs_post_rdma_write_imm_empty(usr_con, sess->hb_cqe, imm, + err = rtrs_post_rdma_write_imm_empty(usr_con, path->hb_cqe, imm, NULL); if (err) { - rtrs_err(sess, "HB send failed, errno: %d\n", err); - sess->hb_err_handler(usr_con); + rtrs_err(path, "HB send failed, errno: %d\n", err); + path->hb_err_handler(usr_con); return; } - schedule_hb(sess); + schedule_hb(path); } -void rtrs_init_hb(struct rtrs_sess *sess, struct ib_cqe *cqe, +void rtrs_init_hb(struct rtrs_path *path, struct ib_cqe *cqe, unsigned int interval_ms, unsigned int missed_max, void (*err_handler)(struct rtrs_con *con), struct workqueue_struct *wq) { - sess->hb_cqe = cqe; - sess->hb_interval_ms = interval_ms; - sess->hb_err_handler = err_handler; - sess->hb_wq = wq; - sess->hb_missed_max = missed_max; - sess->hb_missed_cnt = 0; - INIT_DELAYED_WORK(&sess->hb_dwork, hb_work); + path->hb_cqe = cqe; + path->hb_interval_ms = interval_ms; + path->hb_err_handler = err_handler; + path->hb_wq = wq; + path->hb_missed_max = missed_max; + path->hb_missed_cnt = 0; + INIT_DELAYED_WORK(&path->hb_dwork, hb_work); } EXPORT_SYMBOL_GPL(rtrs_init_hb); -void rtrs_start_hb(struct rtrs_sess *sess) +void rtrs_start_hb(struct rtrs_path *path) { - schedule_hb(sess); + schedule_hb(path); } EXPORT_SYMBOL_GPL(rtrs_start_hb); -void rtrs_stop_hb(struct rtrs_sess *sess) +void rtrs_stop_hb(struct rtrs_path *path) { - cancel_delayed_work_sync(&sess->hb_dwork); - sess->hb_missed_cnt = 0; + cancel_delayed_work_sync(&path->hb_dwork); + path->hb_missed_cnt = 0; } EXPORT_SYMBOL_GPL(rtrs_stop_hb); @@ -479,7 +479,6 @@ static int rtrs_str_to_sockaddr(const char *addr, size_t len, */ int sockaddr_to_str(const struct sockaddr *addr, char *buf, size_t len) { - switch (addr->sa_family) { case AF_IB: return scnprintf(buf, len, "gid:%pI6", diff --git a/drivers/infiniband/ulp/rtrs/rtrs.h b/drivers/infiniband/ulp/rtrs/rtrs.h index 859c79685daf..b48b53a7c143 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs.h +++ b/drivers/infiniband/ulp/rtrs/rtrs.h @@ -13,9 +13,9 @@ #include <linux/scatterlist.h> struct rtrs_permit; -struct rtrs_clt; +struct rtrs_clt_sess; struct rtrs_srv_ctx; -struct rtrs_srv; +struct rtrs_srv_sess; struct rtrs_srv_op; /* @@ -52,14 +52,14 @@ struct rtrs_clt_ops { void (*link_ev)(void *priv, enum rtrs_clt_link_ev ev); }; -struct rtrs_clt *rtrs_clt_open(struct rtrs_clt_ops *ops, - const char *sessname, +struct rtrs_clt_sess *rtrs_clt_open(struct rtrs_clt_ops *ops, + const char *pathname, const struct rtrs_addr *paths, size_t path_cnt, u16 port, size_t pdu_sz, u8 reconnect_delay_sec, s16 max_reconnect_attempts, u32 nr_poll_queues); -void rtrs_clt_close(struct rtrs_clt *sess); +void rtrs_clt_close(struct rtrs_clt_sess *clt); enum wait_type { RTRS_PERMIT_NOWAIT = 0, @@ -77,11 +77,12 @@ enum rtrs_clt_con_type { RTRS_IO_CON }; -struct rtrs_permit *rtrs_clt_get_permit(struct rtrs_clt *sess, - enum rtrs_clt_con_type con_type, - enum wait_type wait); +struct rtrs_permit *rtrs_clt_get_permit(struct rtrs_clt_sess *sess, + enum rtrs_clt_con_type con_type, + enum wait_type wait); -void rtrs_clt_put_permit(struct rtrs_clt *sess, struct rtrs_permit *permit); +void rtrs_clt_put_permit(struct rtrs_clt_sess *sess, + struct rtrs_permit *permit); /** * rtrs_clt_req_ops - it holds the request confirmation callback @@ -98,10 +99,10 @@ struct rtrs_clt_req_ops { }; int rtrs_clt_request(int dir, struct rtrs_clt_req_ops *ops, - struct rtrs_clt *sess, struct rtrs_permit *permit, + struct rtrs_clt_sess *sess, struct rtrs_permit *permit, const struct kvec *vec, size_t nr, size_t len, struct scatterlist *sg, unsigned int sg_cnt); -int rtrs_clt_rdma_cq_direct(struct rtrs_clt *clt, unsigned int index); +int rtrs_clt_rdma_cq_direct(struct rtrs_clt_sess *clt, unsigned int index); /** * rtrs_attrs - RTRS session attributes @@ -112,7 +113,7 @@ struct rtrs_attrs { u32 max_segments; }; -int rtrs_clt_query(struct rtrs_clt *sess, struct rtrs_attrs *attr); +int rtrs_clt_query(struct rtrs_clt_sess *sess, struct rtrs_attrs *attr); /* * Here goes RTRS server API @@ -138,7 +139,6 @@ struct rtrs_srv_ops { * @priv: Private data set by rtrs_srv_set_sess_priv() * @id: internal RTRS operation id - * @dir: READ/WRITE * @data: Pointer to (bidirectional) rdma memory area: * - in case of %RTRS_SRV_RDMA_EV_RECV contains * data sent by the client @@ -150,7 +150,7 @@ struct rtrs_srv_ops { * @usrlen: Size of the user message */ int (*rdma_ev)(void *priv, - struct rtrs_srv_op *id, int dir, + struct rtrs_srv_op *id, void *data, size_t datalen, const void *usr, size_t usrlen); /** @@ -163,7 +163,7 @@ struct rtrs_srv_ops { * @priv: Private data from user if previously set with * rtrs_srv_set_sess_priv() */ - int (*link_ev)(struct rtrs_srv *sess, enum rtrs_srv_link_ev ev, + int (*link_ev)(struct rtrs_srv_sess *sess, enum rtrs_srv_link_ev ev, void *priv); }; @@ -173,11 +173,12 @@ void rtrs_srv_close(struct rtrs_srv_ctx *ctx); bool rtrs_srv_resp_rdma(struct rtrs_srv_op *id, int errno); -void rtrs_srv_set_sess_priv(struct rtrs_srv *sess, void *priv); +void rtrs_srv_set_sess_priv(struct rtrs_srv_sess *sess, void *priv); -int rtrs_srv_get_sess_name(struct rtrs_srv *sess, char *sessname, size_t len); +int rtrs_srv_get_path_name(struct rtrs_srv_sess *sess, char *pathname, + size_t len); -int rtrs_srv_get_queue_depth(struct rtrs_srv *sess); +int rtrs_srv_get_queue_depth(struct rtrs_srv_sess *sess); int rtrs_addr_to_sockaddr(const char *str, size_t len, u16 port, struct rtrs_addr *addr); diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c index e174e853f8a4..1075c2ac8fe2 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.c +++ b/drivers/infiniband/ulp/srp/ib_srp.c @@ -430,7 +430,7 @@ static struct srp_fr_pool *srp_create_fr_pool(struct ib_device *device, spin_lock_init(&pool->lock); INIT_LIST_HEAD(&pool->free_list); - if (device->attrs.device_cap_flags & IB_DEVICE_SG_GAPS_REG) + if (device->attrs.kernel_cap_flags & IBK_SG_GAPS_REG) mr_type = IB_MR_TYPE_SG_GAPS; else mr_type = IB_MR_TYPE_MEM_REG; @@ -699,7 +699,7 @@ static void srp_free_ch_ib(struct srp_target_port *target, static void srp_path_rec_completion(int status, struct sa_path_rec *pathrec, - void *ch_ptr) + int num_paths, void *ch_ptr) { struct srp_rdma_ch *ch = ch_ptr; struct srp_target_port *target = ch->target; @@ -1282,8 +1282,7 @@ struct srp_terminate_context { int scsi_result; }; -static bool srp_terminate_cmd(struct scsi_cmnd *scmnd, void *context_ptr, - bool reserved) +static bool srp_terminate_cmd(struct scsi_cmnd *scmnd, void *context_ptr) { struct srp_terminate_context *context = context_ptr; struct srp_target_port *target = context->srp_target; @@ -1962,7 +1961,8 @@ static void srp_process_rsp(struct srp_rdma_ch *ch, struct srp_rsp *rsp) if (scmnd) { req = scsi_cmd_priv(scmnd); scmnd = srp_claim_req(ch, req, NULL, scmnd); - } else { + } + if (!scmnd) { shost_printk(KERN_ERR, target->scsi_host, "Null scmnd for RSP w/tag %#016llx received on ch %td / QP %#x\n", rsp->tag, ch - target->ch, ch->qp->qp_num); @@ -2789,7 +2789,7 @@ static int srp_send_tsk_mgmt(struct srp_rdma_ch *ch, u64 req_tag, u64 lun, static int srp_abort(struct scsi_cmnd *scmnd) { struct srp_target_port *target = host_to_target(scmnd->device->host); - struct srp_request *req = (struct srp_request *) scmnd->host_scribble; + struct srp_request *req = scsi_cmd_priv(scmnd); u32 tag; u16 ch_idx; struct srp_rdma_ch *ch; @@ -2797,8 +2797,6 @@ static int srp_abort(struct scsi_cmnd *scmnd) shost_printk(KERN_ERR, target->scsi_host, "SRP abort called\n"); - if (!req) - return SUCCESS; tag = blk_mq_unique_tag(scsi_cmd_to_rq(scmnd)); ch_idx = blk_mq_unique_tag_to_hwq(tag); if (WARN_ON_ONCE(ch_idx >= target->ch_count)) @@ -2991,7 +2989,7 @@ static ssize_t local_ib_port_show(struct device *dev, { struct srp_target_port *target = host_to_target(class_to_shost(dev)); - return sysfs_emit(buf, "%d\n", target->srp_host->port); + return sysfs_emit(buf, "%u\n", target->srp_host->port); } static DEVICE_ATTR_RO(local_ib_port); @@ -3179,11 +3177,16 @@ static void srp_release_dev(struct device *dev) struct srp_host *host = container_of(dev, struct srp_host, dev); - complete(&host->released); + kfree(host); } +static struct attribute *srp_class_attrs[]; + +ATTRIBUTE_GROUPS(srp_class); + static struct class srp_class = { .name = "infiniband_srp", + .dev_groups = srp_class_groups, .dev_release = srp_release_dev }; @@ -3650,7 +3653,7 @@ static ssize_t add_target_store(struct device *dev, target_host->max_cmd_len = sizeof ((struct srp_cmd *) (void *) 0L)->cdb; target_host->max_segment_size = ib_dma_max_seg_size(ibdev); - if (!(ibdev->attrs.device_cap_flags & IB_DEVICE_SG_GAPS_REG)) + if (!(ibdev->attrs.kernel_cap_flags & IBK_SG_GAPS_REG)) target_host->virt_boundary_mask = ~srp_dev->mr_page_mask; target = host_to_target(target_host); @@ -3706,8 +3709,8 @@ static ssize_t add_target_store(struct device *dev, } if (srp_dev->use_fast_reg) { - bool gaps_reg = (ibdev->attrs.device_cap_flags & - IB_DEVICE_SG_GAPS_REG); + bool gaps_reg = ibdev->attrs.kernel_cap_flags & + IBK_SG_GAPS_REG; max_sectors_per_mr = srp_dev->max_pages_per_mr << (ilog2(srp_dev->mr_page_size) - 9); @@ -3884,12 +3887,19 @@ static ssize_t port_show(struct device *dev, struct device_attribute *attr, { struct srp_host *host = container_of(dev, struct srp_host, dev); - return sysfs_emit(buf, "%d\n", host->port); + return sysfs_emit(buf, "%u\n", host->port); } static DEVICE_ATTR_RO(port); -static struct srp_host *srp_add_port(struct srp_device *device, u8 port) +static struct attribute *srp_class_attrs[] = { + &dev_attr_add_target.attr, + &dev_attr_ibdev.attr, + &dev_attr_port.attr, + NULL +}; + +static struct srp_host *srp_add_port(struct srp_device *device, u32 port) { struct srp_host *host; @@ -3899,33 +3909,24 @@ static struct srp_host *srp_add_port(struct srp_device *device, u8 port) INIT_LIST_HEAD(&host->target_list); spin_lock_init(&host->target_lock); - init_completion(&host->released); mutex_init(&host->add_target_mutex); host->srp_dev = device; host->port = port; + device_initialize(&host->dev); host->dev.class = &srp_class; host->dev.parent = device->dev->dev.parent; - dev_set_name(&host->dev, "srp-%s-%d", dev_name(&device->dev->dev), - port); - - if (device_register(&host->dev)) - goto free_host; - if (device_create_file(&host->dev, &dev_attr_add_target)) - goto err_class; - if (device_create_file(&host->dev, &dev_attr_ibdev)) - goto err_class; - if (device_create_file(&host->dev, &dev_attr_port)) - goto err_class; + if (dev_set_name(&host->dev, "srp-%s-%u", dev_name(&device->dev->dev), + port)) + goto put_host; + if (device_add(&host->dev)) + goto put_host; return host; -err_class: - device_unregister(&host->dev); - -free_host: - kfree(host); - +put_host: + device_del(&host->dev); + put_device(&host->dev); return NULL; } @@ -3937,7 +3938,7 @@ static void srp_rename_dev(struct ib_device *device, void *client_data) list_for_each_entry_safe(host, tmp_host, &srp_dev->dev_list, list) { char name[IB_DEVICE_NAME_MAX + 8]; - snprintf(name, sizeof(name), "srp-%s-%d", + snprintf(name, sizeof(name), "srp-%s-%u", dev_name(&device->dev), host->port); device_rename(&host->dev, name); } @@ -3949,7 +3950,7 @@ static int srp_add_one(struct ib_device *device) struct ib_device_attr *attr = &device->attrs; struct srp_host *host; int mr_page_shift; - unsigned int p; + u32 p; u64 max_pages_per_mr; unsigned int flags = 0; @@ -4031,12 +4032,11 @@ static void srp_remove_one(struct ib_device *device, void *client_data) srp_dev = client_data; list_for_each_entry_safe(host, tmp_host, &srp_dev->dev_list, list) { - device_unregister(&host->dev); /* - * Wait for the sysfs entry to go away, so that no new - * target ports can be created. + * Remove the add_target sysfs entry so that no new target ports + * can be created. */ - wait_for_completion(&host->released); + device_del(&host->dev); /* * Remove all target ports. @@ -4047,12 +4047,14 @@ static void srp_remove_one(struct ib_device *device, void *client_data) spin_unlock(&host->target_lock); /* - * Wait for tl_err and target port removal tasks. + * srp_queue_remove_work() queues a call to + * srp_remove_target(). The latter function cancels + * target->tl_err_work so waiting for the remove works to + * finish is sufficient. */ - flush_workqueue(system_long_wq); flush_workqueue(srp_remove_wq); - kfree(host); + put_device(&host->dev); } ib_dealloc_pd(srp_dev->pd); diff --git a/drivers/infiniband/ulp/srp/ib_srp.h b/drivers/infiniband/ulp/srp/ib_srp.h index abccddeea1e3..00b0068fda20 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.h +++ b/drivers/infiniband/ulp/srp/ib_srp.h @@ -92,6 +92,9 @@ enum srp_iu_type { }; /* + * RDMA adapter in the initiator system. + * + * @dev_list: List of RDMA ports associated with this RDMA adapter (srp_host). * @mr_page_mask: HCA memory registration page mask. * @mr_page_size: HCA memory registration page size. * @mr_max_size: Maximum size in bytes of a single FR registration request. @@ -109,13 +112,18 @@ struct srp_device { bool use_fast_reg; }; +/* + * One port of an RDMA adapter in the initiator system. + * + * @target_list: List of connected target ports (struct srp_target_port). + * @target_lock: Protects @target_list. + */ struct srp_host { struct srp_device *srp_dev; - u8 port; + u32 port; struct device dev; struct list_head target_list; spinlock_t target_lock; - struct completion released; struct list_head list; struct mutex add_target_mutex; }; @@ -183,7 +191,7 @@ struct srp_rdma_ch { }; /** - * struct srp_target_port + * struct srp_target_port - RDMA port in the SRP target system * @comp_vector: Completion vector used by the first RDMA channel created for * this target port. */ diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index f86ee1c4b970..3c3fae738c3e 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -565,12 +565,9 @@ static int srpt_refresh_port(struct srpt_port *sport) if (ret) return ret; - sport->port_guid_id.wwn.priv = sport; - srpt_format_guid(sport->port_guid_id.name, - sizeof(sport->port_guid_id.name), + srpt_format_guid(sport->guid_name, ARRAY_SIZE(sport->guid_name), &sport->gid.global.interface_id); - sport->port_gid_id.wwn.priv = sport; - snprintf(sport->port_gid_id.name, sizeof(sport->port_gid_id.name), + snprintf(sport->gid_name, ARRAY_SIZE(sport->gid_name), "0x%016llx%016llx", be64_to_cpu(sport->gid.global.subnet_prefix), be64_to_cpu(sport->gid.global.interface_id)); @@ -1424,7 +1421,7 @@ static int srpt_build_cmd_rsp(struct srpt_rdma_ch *ch, srp_rsp->flags |= SRP_RSP_FLAG_SNSVALID; srp_rsp->sense_data_len = cpu_to_be32(sense_data_len); - memcpy(srp_rsp + 1, sense_data, sense_data_len); + memcpy(srp_rsp->data, sense_data, sense_data_len); } return sizeof(*srp_rsp) + sense_data_len; @@ -2221,13 +2218,13 @@ static int srpt_cm_req_recv(struct srpt_device *const sdev, ch->zw_cqe.done = srpt_zerolength_write_done; INIT_WORK(&ch->release_work, srpt_release_channel_work); ch->sport = sport; - if (ib_cm_id) { - ch->ib_cm.cm_id = ib_cm_id; - ib_cm_id->context = ch; - } else { + if (rdma_cm_id) { ch->using_rdma_cm = true; ch->rdma_cm.cm_id = rdma_cm_id; rdma_cm_id->context = ch; + } else { + ch->ib_cm.cm_id = ib_cm_id; + ib_cm_id->context = ch; } /* * ch->rq_size should be at least as large as the initiator queue @@ -2303,7 +2300,7 @@ static int srpt_cm_req_recv(struct srpt_device *const sdev, goto free_recv_ring; } - strlcpy(ch->sess_name, src_addr, sizeof(ch->sess_name)); + strscpy(ch->sess_name, src_addr, sizeof(ch->sess_name)); snprintf(i_port_id, sizeof(i_port_id), "0x%016llx%016llx", be64_to_cpu(*(__be64 *)nexus->i_port_id), be64_to_cpu(*(__be64 *)(nexus->i_port_id + 8))); @@ -2314,31 +2311,35 @@ static int srpt_cm_req_recv(struct srpt_device *const sdev, tag_num = ch->rq_size; tag_size = 1; /* ib_srpt does not use se_sess->sess_cmd_map */ - mutex_lock(&sport->port_guid_id.mutex); - list_for_each_entry(stpg, &sport->port_guid_id.tpg_list, entry) { - if (!IS_ERR_OR_NULL(ch->sess)) - break; - ch->sess = target_setup_session(&stpg->tpg, tag_num, + if (sport->guid_id) { + mutex_lock(&sport->guid_id->mutex); + list_for_each_entry(stpg, &sport->guid_id->tpg_list, entry) { + if (!IS_ERR_OR_NULL(ch->sess)) + break; + ch->sess = target_setup_session(&stpg->tpg, tag_num, tag_size, TARGET_PROT_NORMAL, ch->sess_name, ch, NULL); + } + mutex_unlock(&sport->guid_id->mutex); } - mutex_unlock(&sport->port_guid_id.mutex); - mutex_lock(&sport->port_gid_id.mutex); - list_for_each_entry(stpg, &sport->port_gid_id.tpg_list, entry) { - if (!IS_ERR_OR_NULL(ch->sess)) - break; - ch->sess = target_setup_session(&stpg->tpg, tag_num, + if (sport->gid_id) { + mutex_lock(&sport->gid_id->mutex); + list_for_each_entry(stpg, &sport->gid_id->tpg_list, entry) { + if (!IS_ERR_OR_NULL(ch->sess)) + break; + ch->sess = target_setup_session(&stpg->tpg, tag_num, tag_size, TARGET_PROT_NORMAL, i_port_id, ch, NULL); - if (!IS_ERR_OR_NULL(ch->sess)) - break; - /* Retry without leading "0x" */ - ch->sess = target_setup_session(&stpg->tpg, tag_num, + if (!IS_ERR_OR_NULL(ch->sess)) + break; + /* Retry without leading "0x" */ + ch->sess = target_setup_session(&stpg->tpg, tag_num, tag_size, TARGET_PROT_NORMAL, i_port_id + 2, ch, NULL); + } + mutex_unlock(&sport->gid_id->mutex); } - mutex_unlock(&sport->port_gid_id.mutex); if (IS_ERR_OR_NULL(ch->sess)) { WARN_ON_ONCE(ch->sess == NULL); @@ -2983,7 +2984,12 @@ static int srpt_release_sport(struct srpt_port *sport) return 0; } -static struct se_wwn *__srpt_lookup_wwn(const char *name) +struct port_and_port_id { + struct srpt_port *sport; + struct srpt_port_id **port_id; +}; + +static struct port_and_port_id __srpt_lookup_port(const char *name) { struct ib_device *dev; struct srpt_device *sdev; @@ -2998,25 +3004,38 @@ static struct se_wwn *__srpt_lookup_wwn(const char *name) for (i = 0; i < dev->phys_port_cnt; i++) { sport = &sdev->port[i]; - if (strcmp(sport->port_guid_id.name, name) == 0) - return &sport->port_guid_id.wwn; - if (strcmp(sport->port_gid_id.name, name) == 0) - return &sport->port_gid_id.wwn; + if (strcmp(sport->guid_name, name) == 0) { + kref_get(&sdev->refcnt); + return (struct port_and_port_id){ + sport, &sport->guid_id}; + } + if (strcmp(sport->gid_name, name) == 0) { + kref_get(&sdev->refcnt); + return (struct port_and_port_id){ + sport, &sport->gid_id}; + } } } - return NULL; + return (struct port_and_port_id){}; } -static struct se_wwn *srpt_lookup_wwn(const char *name) +/** + * srpt_lookup_port() - Look up an RDMA port by name + * @name: ASCII port name + * + * Increments the RDMA port reference count if an RDMA port pointer is returned. + * The caller must drop that reference count by calling srpt_port_put_ref(). + */ +static struct port_and_port_id srpt_lookup_port(const char *name) { - struct se_wwn *wwn; + struct port_and_port_id papi; spin_lock(&srpt_dev_lock); - wwn = __srpt_lookup_wwn(name); + papi = __srpt_lookup_port(name); spin_unlock(&srpt_dev_lock); - return wwn; + return papi; } static void srpt_free_srq(struct srpt_device *sdev) @@ -3101,6 +3120,18 @@ static int srpt_use_srq(struct srpt_device *sdev, bool use_srq) return ret; } +static void srpt_free_sdev(struct kref *refcnt) +{ + struct srpt_device *sdev = container_of(refcnt, typeof(*sdev), refcnt); + + kfree(sdev); +} + +static void srpt_sdev_put(struct srpt_device *sdev) +{ + kref_put(&sdev->refcnt, srpt_free_sdev); +} + /** * srpt_add_one - InfiniBand device addition callback function * @device: Describes a HCA. @@ -3119,6 +3150,7 @@ static int srpt_add_one(struct ib_device *device) if (!sdev) return -ENOMEM; + kref_init(&sdev->refcnt); sdev->device = device; mutex_init(&sdev->sdev_mutex); @@ -3159,7 +3191,7 @@ static int srpt_add_one(struct ib_device *device) * if this HCA is gone bad and replaced by different HCA */ ret = sdev->cm_id ? - ib_cm_listen(sdev->cm_id, cpu_to_be64(srpt_service_guid), 0) : + ib_cm_listen(sdev->cm_id, cpu_to_be64(srpt_service_guid)) : 0; if (ret < 0) { pr_err("ib_cm_listen() failed: %d (cm_id state = %d)\n", ret, @@ -3182,10 +3214,6 @@ static int srpt_add_one(struct ib_device *device) sport->port_attrib.srp_sq_size = DEF_SRPT_SQ_SIZE; sport->port_attrib.use_srq = false; INIT_WORK(&sport->work, srpt_refresh_port_work); - mutex_init(&sport->port_guid_id.mutex); - INIT_LIST_HEAD(&sport->port_guid_id.tpg_list); - mutex_init(&sport->port_gid_id.mutex); - INIT_LIST_HEAD(&sport->port_gid_id.tpg_list); ret = srpt_refresh_port(sport); if (ret) { @@ -3214,7 +3242,7 @@ err_ring: srpt_free_srq(sdev); ib_dealloc_pd(sdev->pd); free_dev: - kfree(sdev); + srpt_sdev_put(sdev); pr_info("%s(%s) failed.\n", __func__, dev_name(&device->dev)); return ret; } @@ -3258,7 +3286,7 @@ static void srpt_remove_one(struct ib_device *device, void *client_data) ib_dealloc_pd(sdev->pd); - kfree(sdev); + srpt_sdev_put(sdev); } static struct ib_client srpt_client = { @@ -3286,10 +3314,10 @@ static struct srpt_port_id *srpt_wwn_to_sport_id(struct se_wwn *wwn) { struct srpt_port *sport = wwn->priv; - if (wwn == &sport->port_guid_id.wwn) - return &sport->port_guid_id; - if (wwn == &sport->port_gid_id.wwn) - return &sport->port_gid_id; + if (sport->guid_id && &sport->guid_id->wwn == wwn) + return sport->guid_id; + if (sport->gid_id && &sport->gid_id->wwn == wwn) + return sport->gid_id; WARN_ON_ONCE(true); return NULL; } @@ -3774,7 +3802,31 @@ static struct se_wwn *srpt_make_tport(struct target_fabric_configfs *tf, struct config_group *group, const char *name) { - return srpt_lookup_wwn(name) ? : ERR_PTR(-EINVAL); + struct port_and_port_id papi = srpt_lookup_port(name); + struct srpt_port *sport = papi.sport; + struct srpt_port_id *port_id; + + if (!papi.port_id) + return ERR_PTR(-EINVAL); + if (*papi.port_id) { + /* Attempt to create a directory that already exists. */ + WARN_ON_ONCE(true); + return &(*papi.port_id)->wwn; + } + port_id = kzalloc(sizeof(*port_id), GFP_KERNEL); + if (!port_id) { + srpt_sdev_put(sport->sdev); + return ERR_PTR(-ENOMEM); + } + mutex_init(&port_id->mutex); + INIT_LIST_HEAD(&port_id->tpg_list); + port_id->wwn.priv = sport; + memcpy(port_id->name, port_id == sport->guid_id ? sport->guid_name : + sport->gid_name, ARRAY_SIZE(port_id->name)); + + *papi.port_id = port_id; + + return &port_id->wwn; } /** @@ -3783,6 +3835,18 @@ static struct se_wwn *srpt_make_tport(struct target_fabric_configfs *tf, */ static void srpt_drop_tport(struct se_wwn *wwn) { + struct srpt_port_id *port_id = container_of(wwn, typeof(*port_id), wwn); + struct srpt_port *sport = wwn->priv; + + if (sport->guid_id == port_id) + sport->guid_id = NULL; + else if (sport->gid_id == port_id) + sport->gid_id = NULL; + else + WARN_ON_ONCE(true); + + srpt_sdev_put(sport->sdev); + kfree(port_id); } static ssize_t srpt_wwn_version_show(struct config_item *item, char *buf) diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.h b/drivers/infiniband/ulp/srpt/ib_srpt.h index 76e66f630c17..4c46b301eea1 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.h +++ b/drivers/infiniband/ulp/srpt/ib_srpt.h @@ -376,7 +376,7 @@ struct srpt_tpg { }; /** - * struct srpt_port_id - information about an RDMA port name + * struct srpt_port_id - LIO RDMA port information * @mutex: Protects @tpg_list changes. * @tpg_list: TPGs associated with the RDMA port name. * @wwn: WWN associated with the RDMA port name. @@ -393,7 +393,7 @@ struct srpt_port_id { }; /** - * struct srpt_port - information associated by SRPT with a single IB port + * struct srpt_port - SRPT RDMA port information * @sdev: backpointer to the HCA information. * @mad_agent: per-port management datagram processing information. * @enabled: Whether or not this target port is enabled. @@ -402,8 +402,10 @@ struct srpt_port_id { * @lid: cached value of the port's lid. * @gid: cached value of the port's gid. * @work: work structure for refreshing the aforementioned cached values. - * @port_guid_id: target port GUID - * @port_gid_id: target port GID + * @guid_name: port name in GUID format. + * @guid_id: LIO target port information for the port name in GUID format. + * @gid_name: port name in GID format. + * @gid_id: LIO target port information for the port name in GID format. * @port_attrib: Port attributes that can be accessed through configfs. * @refcount: Number of objects associated with this port. * @freed_channels: Completion that will be signaled once @refcount becomes 0. @@ -419,8 +421,10 @@ struct srpt_port { u32 lid; union ib_gid gid; struct work_struct work; - struct srpt_port_id port_guid_id; - struct srpt_port_id port_gid_id; + char guid_name[64]; + struct srpt_port_id *guid_id; + char gid_name[64]; + struct srpt_port_id *gid_id; struct srpt_port_attrib port_attrib; atomic_t refcount; struct completion *freed_channels; @@ -430,6 +434,7 @@ struct srpt_port { /** * struct srpt_device - information associated by SRPT with a single HCA + * @refcnt: Reference count for this device. * @device: Backpointer to the struct ib_device managed by the IB core. * @pd: IB protection domain. * @lkey: L_Key (local key) with write access to all local memory. @@ -445,6 +450,7 @@ struct srpt_port { * @port: Information about the ports owned by this HCA. */ struct srpt_device { + struct kref refcnt; struct ib_device *device; struct ib_pd *pd; u32 lkey; |