From 4a97d47ef7946cf31b76945c3199b0b5cad6a8ed Mon Sep 17 00:00:00 2001 From: Steve Wise Date: Thu, 26 Apr 2007 15:21:02 -0500 Subject: RDMA/cxgb3: Fix TERM codes Fix TERMINATE layer, type, and ecode values based on conformance testing. Signed-off-by: Steve Wise Signed-off-by: Roland Dreier --- drivers/infiniband/hw/cxgb3/iwch_qp.c | 69 +++++++++++++++++++---------------- 1 file changed, 38 insertions(+), 31 deletions(-) (limited to 'drivers/infiniband/hw/cxgb3') diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c index 0a472c9b44d..714dddbc9a9 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_qp.c +++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c @@ -471,43 +471,62 @@ int iwch_bind_mw(struct ib_qp *qp, return err; } -static void build_term_codes(int t3err, u8 *layer_type, u8 *ecode, int tagged) +static inline void build_term_codes(struct respQ_msg_t *rsp_msg, + u8 *layer_type, u8 *ecode) { - switch (t3err) { + int status = TPT_ERR_INTERNAL_ERR; + int tagged = 0; + int opcode = -1; + int rqtype = 0; + int send_inv = 0; + + if (rsp_msg) { + status = CQE_STATUS(rsp_msg->cqe); + opcode = CQE_OPCODE(rsp_msg->cqe); + rqtype = RQ_TYPE(rsp_msg->cqe); + send_inv = (opcode == T3_SEND_WITH_INV) || + (opcode == T3_SEND_WITH_SE_INV); + tagged = (opcode == T3_RDMA_WRITE) || + (rqtype && (opcode == T3_READ_RESP)); + } + + switch (status) { case TPT_ERR_STAG: - if (tagged == 1) { - *layer_type = LAYER_DDP|DDP_TAGGED_ERR; - *ecode = DDPT_INV_STAG; - } else if (tagged == 2) { + if (send_inv) { + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_OP; + *ecode = RDMAP_CANT_INV_STAG; + } else { *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; *ecode = RDMAP_INV_STAG; } break; case TPT_ERR_PDID: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + if ((opcode == T3_SEND_WITH_INV) || + (opcode == T3_SEND_WITH_SE_INV)) + *ecode = RDMAP_CANT_INV_STAG; + else + *ecode = RDMAP_STAG_NOT_ASSOC; + break; case TPT_ERR_QPID: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + *ecode = RDMAP_STAG_NOT_ASSOC; + break; case TPT_ERR_ACCESS: - if (tagged == 1) { - *layer_type = LAYER_DDP|DDP_TAGGED_ERR; - *ecode = DDPT_STAG_NOT_ASSOC; - } else if (tagged == 2) { - *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; - *ecode = RDMAP_STAG_NOT_ASSOC; - } + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + *ecode = RDMAP_ACC_VIOL; break; case TPT_ERR_WRAP: *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; *ecode = RDMAP_TO_WRAP; break; case TPT_ERR_BOUND: - if (tagged == 1) { + if (tagged) { *layer_type = LAYER_DDP|DDP_TAGGED_ERR; *ecode = DDPT_BASE_BOUNDS; - } else if (tagged == 2) { + } else { *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; *ecode = RDMAP_BASE_BOUNDS; - } else { - *layer_type = LAYER_DDP|DDP_UNTAGGED_ERR; - *ecode = DDPU_MSG_TOOBIG; } break; case TPT_ERR_INVALIDATE_SHARED_MR: @@ -591,8 +610,6 @@ int iwch_post_terminate(struct iwch_qp *qhp, struct respQ_msg_t *rsp_msg) { union t3_wr *wqe; struct terminate_message *term; - int status; - int tagged = 0; struct sk_buff *skb; PDBG("%s %d\n", __FUNCTION__, __LINE__); @@ -610,17 +627,7 @@ int iwch_post_terminate(struct iwch_qp *qhp, struct respQ_msg_t *rsp_msg) /* immediate data starts here. */ term = (struct terminate_message *)wqe->send.sgl; - if (rsp_msg) { - status = CQE_STATUS(rsp_msg->cqe); - if (CQE_OPCODE(rsp_msg->cqe) == T3_RDMA_WRITE) - tagged = 1; - if ((CQE_OPCODE(rsp_msg->cqe) == T3_READ_REQ) || - (CQE_OPCODE(rsp_msg->cqe) == T3_READ_RESP)) - tagged = 2; - } else { - status = TPT_ERR_INTERNAL_ERR; - } - build_term_codes(status, &term->layer_etype, &term->ecode, tagged); + build_term_codes(rsp_msg, &term->layer_etype, &term->ecode); build_fw_riwrh((void *)wqe, T3_WR_SEND, T3_COMPLETION_FLAG | T3_NOTIFY_FLAG, 1, qhp->ep->hwtid, 5); -- cgit v1.2.3 From 1860cdf802310e4a988e0b8fca41cc97da36f779 Mon Sep 17 00:00:00 2001 From: Steve Wise Date: Thu, 26 Apr 2007 15:21:09 -0500 Subject: RDMA/cxgb3: Fail qp creation if the requested max_inline is too large Signed-off-by: Steve Wise Signed-off-by: Roland Dreier --- drivers/infiniband/hw/cxgb3/cxio_wr.h | 1 + drivers/infiniband/hw/cxgb3/iwch_provider.c | 3 +++ 2 files changed, 4 insertions(+) (limited to 'drivers/infiniband/hw/cxgb3') diff --git a/drivers/infiniband/hw/cxgb3/cxio_wr.h b/drivers/infiniband/hw/cxgb3/cxio_wr.h index 90d7b8972cb..ff7290eacef 100644 --- a/drivers/infiniband/hw/cxgb3/cxio_wr.h +++ b/drivers/infiniband/hw/cxgb3/cxio_wr.h @@ -38,6 +38,7 @@ #include "firmware_exports.h" #define T3_MAX_SGE 4 +#define T3_MAX_INLINE 64 #define Q_EMPTY(rptr,wptr) ((rptr)==(wptr)) #define Q_FULL(rptr,wptr,size_log2) ( (((wptr)-(rptr))>>(size_log2)) && \ diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index af28a317016..93038c00713 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -780,6 +780,9 @@ static struct ib_qp *iwch_create_qp(struct ib_pd *pd, if (rqsize > T3_MAX_RQ_SIZE) return ERR_PTR(-EINVAL); + if (attrs->cap.max_inline_data > T3_MAX_INLINE) + return ERR_PTR(-EINVAL); + /* * NOTE: The SQ and total WQ sizes don't need to be * a power of two. However, all the code assumes -- cgit v1.2.3 From 60be4b5966e22040f97db9dada72841bf90479d1 Mon Sep 17 00:00:00 2001 From: Steve Wise Date: Thu, 26 Apr 2007 15:21:15 -0500 Subject: RDMA/cxgb3: Initialize cpu_idx field in cpl_close_listserv_req message Signed-off-by: Steve Wise Signed-off-by: Roland Dreier --- drivers/infiniband/hw/cxgb3/iwch_cm.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/infiniband/hw/cxgb3') diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c index 3b4b0acd707..a02be40ee29 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_cm.c +++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c @@ -1189,6 +1189,7 @@ static int listen_stop(struct iwch_listen_ep *ep) } req = (struct cpl_close_listserv_req *) skb_put(skb, sizeof(*req)); req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + req->cpu_idx = 0; OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_CLOSE_LISTSRV_REQ, ep->stid)); skb->priority = 1; ep->com.tdev->send(ep->com.tdev, skb); -- cgit v1.2.3 From aff9e39d97585486764572ab2f3bf5dfce18c660 Mon Sep 17 00:00:00 2001 From: Steve Wise Date: Thu, 26 Apr 2007 15:21:20 -0500 Subject: RDMA/cxgb3: Support for new abort logic The HW now posts 2 ABORT_RPL and/or PEER_ABORT_REQ messages. We need to handle them by silenty dropping the 1st but mark that we're ready for the final message. This plugs some close races between the uP and HW. Also update the minimum required firmware version. Signed-off-by: Steve Wise Signed-off-by: Roland Dreier --- drivers/infiniband/hw/cxgb3/iwch_cm.c | 18 ++++++++++++++++++ drivers/infiniband/hw/cxgb3/iwch_cm.h | 6 ++++++ 2 files changed, 24 insertions(+) (limited to 'drivers/infiniband/hw/cxgb3') diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c index a02be40ee29..b2faff5abce 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_cm.c +++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c @@ -1109,6 +1109,15 @@ static int abort_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) PDBG("%s ep %p\n", __FUNCTION__, ep); + /* + * We get 2 abort replies from the HW. The first one must + * be ignored except for scribbling that we need one more. + */ + if (!(ep->flags & ABORT_REQ_IN_PROGRESS)) { + ep->flags |= ABORT_REQ_IN_PROGRESS; + return CPL_RET_BUF_DONE; + } + close_complete_upcall(ep); state_set(&ep->com, DEAD); release_ep_resources(ep); @@ -1476,6 +1485,15 @@ static int peer_abort(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) int ret; int state; + /* + * We get 2 peer aborts from the HW. The first one must + * be ignored except for scribbling that we need one more. + */ + if (!(ep->flags & PEER_ABORT_IN_PROGRESS)) { + ep->flags |= PEER_ABORT_IN_PROGRESS; + return CPL_RET_BUF_DONE; + } + if (is_neg_adv_abort(req->status)) { PDBG("%s neg_adv_abort ep %p tid %d\n", __FUNCTION__, ep, ep->hwtid); diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.h b/drivers/infiniband/hw/cxgb3/iwch_cm.h index 0c6f281bd4a..21a388c313c 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_cm.h +++ b/drivers/infiniband/hw/cxgb3/iwch_cm.h @@ -143,6 +143,11 @@ enum iwch_ep_state { DEAD, }; +enum iwch_ep_flags { + PEER_ABORT_IN_PROGRESS = (1 << 0), + ABORT_REQ_IN_PROGRESS = (1 << 1), +}; + struct iwch_ep_common { struct iw_cm_id *cm_id; struct iwch_qp *qp; @@ -181,6 +186,7 @@ struct iwch_ep { u16 plen; u32 ird; u32 ord; + u32 flags; }; static inline struct iwch_ep *to_ep(struct iw_cm_id *cm_id) -- cgit v1.2.3 From f4fd0b224d60044d2da5ca02f8f2b5150c1d8731 Mon Sep 17 00:00:00 2001 From: "Michael S. Tsirkin" Date: Thu, 3 May 2007 13:48:47 +0300 Subject: IB: Add CQ comp_vector support Add a num_comp_vectors member to struct ib_device and extend ib_create_cq() to pass in a comp_vector parameter -- this parallels the userspace libibverbs API. Update all hardware drivers to set num_comp_vectors to 1 and have all ULPs pass 0 for the comp_vector value. Pass the value of num_comp_vectors to userspace rather than hard-coding a value of 1. We want multiple CQ event vector support (via MSI-X or similar for adapters that can generate multiple interrupts), but it's not clear how many vectors we want, or how we want to deal with policy issues such as how to decide which vector to use or how to set up interrupt affinity. This patch is useful for experimenting, since no core changes will be necessary when updating a driver to support multiple vectors, and we know that we want to make at least these changes anyway. Signed-off-by: Michael S. Tsirkin Signed-off-by: Roland Dreier --- drivers/infiniband/hw/cxgb3/iwch_provider.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/infiniband/hw/cxgb3') diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index 93038c00713..78a495f5332 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -139,7 +139,7 @@ static int iwch_destroy_cq(struct ib_cq *ib_cq) return 0; } -static struct ib_cq *iwch_create_cq(struct ib_device *ibdev, int entries, +static struct ib_cq *iwch_create_cq(struct ib_device *ibdev, int entries, int vector, struct ib_ucontext *ib_context, struct ib_udata *udata) { @@ -1110,6 +1110,7 @@ int iwch_register_device(struct iwch_dev *dev) dev->ibdev.node_type = RDMA_NODE_RNIC; memcpy(dev->ibdev.node_desc, IWCH_NODE_DESC, sizeof(IWCH_NODE_DESC)); dev->ibdev.phys_port_cnt = dev->rdev.port_info.nports; + dev->ibdev.num_comp_vectors = 1; dev->ibdev.dma_device = &(dev->rdev.rnic_info.pdev->dev); dev->ibdev.query_device = iwch_query_device; dev->ibdev.query_port = iwch_query_port; -- cgit v1.2.3 From ed23a72778f3dbd465e55b06fe31629e7e1dd2f3 Mon Sep 17 00:00:00 2001 From: Roland Dreier Date: Sun, 6 May 2007 21:02:48 -0700 Subject: IB: Return "maybe missed event" hint from ib_req_notify_cq() The semantics defined by the InfiniBand specification say that completion events are only generated when a completions is added to a completion queue (CQ) after completion notification is requested. In other words, this means that the following race is possible: while (CQ is not empty) ib_poll_cq(CQ); // new completion is added after while loop is exited ib_req_notify_cq(CQ); // no event is generated for the existing completion To close this race, the IB spec recommends doing another poll of the CQ after requesting notification. However, it is not always possible to arrange code this way (for example, we have found that NAPI for IPoIB cannot poll after requesting notification). Also, some hardware (eg Mellanox HCAs) actually will generate an event for completions added before the call to ib_req_notify_cq() -- which is allowed by the spec, since there's no way for any upper-layer consumer to know exactly when a completion was really added -- so the extra poll of the CQ is just a waste. Motivated by this, we add a new flag "IB_CQ_REPORT_MISSED_EVENTS" for ib_req_notify_cq() so that it can return a hint about whether the a completion may have been added before the request for notification. The return value of ib_req_notify_cq() is extended so: < 0 means an error occurred while requesting notification == 0 means notification was requested successfully, and if IB_CQ_REPORT_MISSED_EVENTS was passed in, then no events were missed and it is safe to wait for another event. > 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was passed in. It means that the consumer must poll the CQ again to make sure it is empty to avoid the race described above. We add a flag to enable this behavior rather than turning it on unconditionally, because checking for missed events may incur significant overhead for some low-level drivers, and consumers that don't care about the results of this test shouldn't be forced to pay for the test. Signed-off-by: Roland Dreier --- drivers/infiniband/hw/cxgb3/cxio_hal.c | 3 +++ drivers/infiniband/hw/cxgb3/iwch_provider.c | 8 +++++--- 2 files changed, 8 insertions(+), 3 deletions(-) (limited to 'drivers/infiniband/hw/cxgb3') diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c index f5e9aeec6f6..76049afc765 100644 --- a/drivers/infiniband/hw/cxgb3/cxio_hal.c +++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c @@ -114,7 +114,10 @@ int cxio_hal_cq_op(struct cxio_rdev *rdev_p, struct t3_cq *cq, return -EIO; } } + + return 1; } + return 0; } diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index 78a495f5332..a891493fd34 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -292,7 +292,7 @@ static int iwch_resize_cq(struct ib_cq *cq, int cqe, struct ib_udata *udata) #endif } -static int iwch_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify notify) +static int iwch_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags) { struct iwch_dev *rhp; struct iwch_cq *chp; @@ -303,7 +303,7 @@ static int iwch_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify notify) chp = to_iwch_cq(ibcq); rhp = chp->rhp; - if (notify == IB_CQ_SOLICITED) + if ((flags & IB_CQ_SOLICITED_MASK) == IB_CQ_SOLICITED) cq_op = CQ_ARM_SE; else cq_op = CQ_ARM_AN; @@ -317,9 +317,11 @@ static int iwch_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify notify) PDBG("%s rptr 0x%x\n", __FUNCTION__, chp->cq.rptr); err = cxio_hal_cq_op(&rhp->rdev, &chp->cq, cq_op, 0); spin_unlock_irqrestore(&chp->lock, flag); - if (err) + if (err < 0) printk(KERN_ERR MOD "Error %d rearming CQID 0x%x\n", err, chp->cq.cqid); + if (err > 0 && !(flags & IB_CQ_REPORT_MISSED_EVENTS)) + err = 0; return err; } -- cgit v1.2.3