[PATCH for-next 0/3] RDMA/bnxt_re: Async events update

[PATCH for-next 0/3] RDMA/bnxt_re: Async events update

[Selvin Xavier]

[-- Attachment #1: Type: text/plain, Size: 473 bytes --]

Reports async error events received from the HW.
Please review and apply

Regards,
Selvin

Chandramohan Akula (3):
  RDMA/bnxt_re: Update HW interface headers
  RDMA/bnxt_re: Report async events and errors
  RDMA/bnxt_re: Do not report SRQ error in srq  notification

 drivers/infiniband/hw/bnxt_re/main.c     | 172 ++++++++++++++++++++++++++++---
 drivers/infiniband/hw/bnxt_re/roce_hsi.h |  62 +++++++++++
 2 files changed, 220 insertions(+), 14 deletions(-)

-- 
2.5.5


[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4224 bytes --]

[PATCH for-next 1/3] RDMA/bnxt_re: Update HW interface headers

[Selvin Xavier]

[-- Attachment #1: Type: text/plain, Size: 6787 bytes --]

From: Chandramohan Akula <chandramohan.akula@broadcom.com>

Updating the HW structures for the affiliated event and error
reporting. Newly added interface structures will be used in the
followup patch.

Signed-off-by: Chandramohan Akula <chandramohan.akula@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
---
 drivers/infiniband/hw/bnxt_re/roce_hsi.h | 62 ++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)

diff --git a/drivers/infiniband/hw/bnxt_re/roce_hsi.h b/drivers/infiniband/hw/bnxt_re/roce_hsi.h
index 4a10303..4196d2b 100644
--- a/drivers/infiniband/hw/bnxt_re/roce_hsi.h
+++ b/drivers/infiniband/hw/bnxt_re/roce_hsi.h
@@ -2919,6 +2919,37 @@ struct creq_qp_error_notification {
 	u8	status;
 	u8	req_slow_path_state;
 	u8	req_err_state_reason;
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_NO_ERROR                    0X0UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_OPCODE_ERROR            0X1UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_TIMEOUT_RETRY_LIMIT     0X2UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RNR_TIMEOUT_RETRY_LIMIT 0X3UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_1           0X4UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_2           0X5UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_3           0X6UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_4           0X7UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RX_MEMORY_ERROR         0X8UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_TX_MEMORY_ERROR         0X9UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_READ_RESP_LENGTH        0XAUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_INVALID_READ_RESP       0XBUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ILLEGAL_BIND            0XCUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ILLEGAL_FAST_REG        0XDUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ILLEGAL_INVALIDATE      0XEUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_CMP_ERROR               0XFUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RETRAN_LOCAL_ERROR      0X10UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_WQE_FORMAT_ERROR        0X11UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ORRQ_FORMAT_ERROR       0X12UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_INVALID_AVID_ERROR      0X13UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_AV_DOMAIN_ERROR         0X14UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_CQ_LOAD_ERROR           0X15UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_SERV_TYPE_ERROR         0X16UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_INVALID_OP_ERROR        0X17UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_TX_PCI_ERROR            0X18UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RX_PCI_ERROR            0X19UL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_PROD_WQE_MSMTCH_ERROR   0X1AUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_PSN_RANGE_CHECK_ERROR   0X1BUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RETX_SETUP_ERROR        0X1CUL
+	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_LAST \
+			CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RETX_SETUP_ERROR
 	__le32	xid;
 	u8	v;
 	#define CREQ_QP_ERROR_NOTIFICATION_V     0x1UL
@@ -2928,6 +2959,37 @@ struct creq_qp_error_notification {
 		CREQ_QP_ERROR_NOTIFICATION_EVENT_QP_ERROR_NOTIFICATION
 	u8	res_slow_path_state;
 	u8	res_err_state_reason;
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_NO_ERROR                      0x0UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_EXCEED_MAX                0x1UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PAYLOAD_LENGTH_MISMATCH   0x2UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_EXCEEDS_WQE               0x3UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_OPCODE_ERROR              0x4UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PSN_SEQ_ERROR_RETRY_LIMIT 0x5UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_INVALID_R_KEY          0x6UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_DOMAIN_ERROR           0x7UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_NO_PERMISSION          0x8UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_RANGE_ERROR            0x9UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_INVALID_R_KEY          0xaUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_DOMAIN_ERROR           0xbUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_NO_PERMISSION          0xcUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_RANGE_ERROR            0xdUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_IRRQ_OFLOW                0xeUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_UNSUPPORTED_OPCODE        0xfUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_UNALIGN_ATOMIC            0x10UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_REM_INVALIDATE            0x11UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_MEMORY_ERROR              0x12UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_SRQ_ERROR                 0x13UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_CMP_ERROR                 0x14UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_INVALID_DUP_RKEY          0x15UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_WQE_FORMAT_ERROR          0x16UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_IRRQ_FORMAT_ERROR         0x17UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_CQ_LOAD_ERROR             0x18UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_SRQ_LOAD_ERROR            0x19UL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_PCI_ERROR              0x1bUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_PCI_ERROR              0x1cUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PSN_NOT_FOUND             0x1dUL
+	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_LAST \
+			CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PSN_NOT_FOUND
 	__le16	sq_cons_idx;
 	__le16	rq_cons_idx;
 };
-- 
2.5.5


[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4224 bytes --]

[PATCH for-next 2/3] RDMA/bnxt_re: Report async events and errors

[Selvin Xavier]

[-- Attachment #1: Type: text/plain, Size: 9287 bytes --]

From: Chandramohan Akula <chandramohan.akula@broadcom.com>

Report QP, SRQ and CQ async events and errors.

Signed-off-by: Chandramohan Akula <chandramohan.akula@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
---
 drivers/infiniband/hw/bnxt_re/main.c | 165 +++++++++++++++++++++++++++++++++--
 1 file changed, 156 insertions(+), 9 deletions(-)

diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index c9066aa..7bd18ec 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -970,6 +970,9 @@ static int bnxt_re_handle_unaffi_async_event(struct creq_func_event
 static int bnxt_re_handle_qp_async_event(struct creq_qp_event *qp_event,
 					 struct bnxt_re_qp *qp)
 {
+	struct bnxt_re_srq *srq = container_of(qp->qplib_qp.srq, struct bnxt_re_srq,
+					       qplib_srq);
+	struct creq_qp_error_notification *err_event;
 	struct ib_event event = {};
 	unsigned int flags;
 
@@ -980,14 +983,146 @@ static int bnxt_re_handle_qp_async_event(struct creq_qp_event *qp_event,
 		bnxt_re_unlock_cqs(qp, flags);
 	}
 
-	if (qp->qplib_qp.srq) {
-		event.device = &qp->rdev->ibdev;
-		event.element.qp = &qp->ib_qp;
-		event.event = IB_EVENT_QP_LAST_WQE_REACHED;
+	event.device = &qp->rdev->ibdev;
+	event.element.qp = &qp->ib_qp;
+	event.event = IB_EVENT_QP_FATAL;
+
+	err_event = (struct creq_qp_error_notification *)qp_event;
+
+	switch (err_event->req_err_state_reason) {
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_OPCODE_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_TIMEOUT_RETRY_LIMIT:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RNR_TIMEOUT_RETRY_LIMIT:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_2:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_3:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_INVALID_READ_RESP:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ILLEGAL_BIND:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ILLEGAL_FAST_REG:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ILLEGAL_INVALIDATE:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RETRAN_LOCAL_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_AV_DOMAIN_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_PROD_WQE_MSMTCH_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_PSN_RANGE_CHECK_ERROR:
+		event.event = IB_EVENT_QP_ACCESS_ERR;
+		break;
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_1:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_NAK_ARRIVAL_4:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_READ_RESP_LENGTH:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_WQE_FORMAT_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_ORRQ_FORMAT_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_INVALID_AVID_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_SERV_TYPE_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_INVALID_OP_ERROR:
+		event.event = IB_EVENT_QP_REQ_ERR;
+		break;
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RX_MEMORY_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_TX_MEMORY_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_CMP_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_CQ_LOAD_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_TX_PCI_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RX_PCI_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RETX_SETUP_ERROR:
+		event.event = IB_EVENT_QP_FATAL;
+		break;
+
+	default:
+		break;
 	}
 
-	if (event.device && qp->ib_qp.event_handler)
+	switch (err_event->res_err_state_reason) {
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_EXCEED_MAX:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PAYLOAD_LENGTH_MISMATCH:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PSN_SEQ_ERROR_RETRY_LIMIT:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_INVALID_R_KEY:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_DOMAIN_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_NO_PERMISSION:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_RANGE_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_INVALID_R_KEY:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_DOMAIN_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_NO_PERMISSION:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_RANGE_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_UNALIGN_ATOMIC:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PSN_NOT_FOUND:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_INVALID_DUP_RKEY:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_IRRQ_FORMAT_ERROR:
+		event.event = IB_EVENT_QP_ACCESS_ERR;
+		break;
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_EXCEEDS_WQE:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_WQE_FORMAT_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_UNSUPPORTED_OPCODE:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_REM_INVALIDATE:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_OPCODE_ERROR:
+		event.event = IB_EVENT_QP_REQ_ERR;
+		break;
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_IRRQ_OFLOW:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_CMP_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_CQ_LOAD_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_TX_PCI_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_RX_PCI_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_MEMORY_ERROR:
+		event.event = IB_EVENT_QP_FATAL;
+		break;
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_SRQ_LOAD_ERROR:
+	case CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_SRQ_ERROR:
+		if (srq)
+			event.event = IB_EVENT_SRQ_ERR;
+		break;
+	default:
+		break;
+	}
+
+	if (err_event->res_err_state_reason || err_event->req_err_state_reason) {
+		dev_err_once(rdev_to_dev(qp->rdev),
+			     "%s %s qp_id: %d cons (%d %d) req (%d %d) res (%d %d)\n",
+			     __func__, rdma_is_kernel_res(&qp->ib_qp.res) ? "kernel" : "user",
+			     qp->qplib_qp.id,
+			     err_event->sq_cons_idx,
+			     err_event->rq_cons_idx,
+			     err_event->req_slow_path_state,
+			     err_event->req_err_state_reason,
+			     err_event->res_slow_path_state,
+			     err_event->res_err_state_reason);
+	} else {
+		if (srq)
+			event.event = IB_EVENT_QP_LAST_WQE_REACHED;
+	}
+
+	if (event.event == IB_EVENT_SRQ_ERR && srq->ib_srq.event_handler)  {
+		(*srq->ib_srq.event_handler)(&event,
+				srq->ib_srq.srq_context);
+	} else if (event.device && qp->ib_qp.event_handler) {
 		qp->ib_qp.event_handler(&event, qp->ib_qp.qp_context);
+	}
+
+	return 0;
+}
+
+static int bnxt_re_handle_cq_async_error(void *event, struct bnxt_re_cq *cq)
+{
+	struct creq_cq_error_notification *cqerr;
+	struct ib_event ibevent = {};
+
+	cqerr = event;
+	switch (cqerr->cq_err_reason) {
+	case CREQ_CQ_ERROR_NOTIFICATION_CQ_ERR_REASON_REQ_CQ_INVALID_ERROR:
+	case CREQ_CQ_ERROR_NOTIFICATION_CQ_ERR_REASON_REQ_CQ_OVERFLOW_ERROR:
+	case CREQ_CQ_ERROR_NOTIFICATION_CQ_ERR_REASON_REQ_CQ_LOAD_ERROR:
+	case CREQ_CQ_ERROR_NOTIFICATION_CQ_ERR_REASON_RES_CQ_INVALID_ERROR:
+	case CREQ_CQ_ERROR_NOTIFICATION_CQ_ERR_REASON_RES_CQ_OVERFLOW_ERROR:
+	case CREQ_CQ_ERROR_NOTIFICATION_CQ_ERR_REASON_RES_CQ_LOAD_ERROR:
+		ibevent.event = IB_EVENT_CQ_ERR;
+	default:
+		break;
+	}
+
+	if (ibevent.event == IB_EVENT_CQ_ERR && cq->ib_cq.event_handler) {
+		ibevent.element.cq = &cq->ib_cq;
+		ibevent.device = &cq->rdev->ibdev;
+
+		dev_err_once(rdev_to_dev(cq->rdev),
+			     "%s err reason %d\n", __func__, cqerr->cq_err_reason);
+		cq->ib_cq.event_handler(&ibevent, cq->ib_cq.cq_context);
+	}
 
 	return 0;
 }
@@ -995,6 +1130,10 @@ static int bnxt_re_handle_qp_async_event(struct creq_qp_event *qp_event,
 static int bnxt_re_handle_affi_async_event(struct creq_qp_event *affi_async,
 					   void *obj)
 {
+	struct bnxt_qplib_qp *lib_qp;
+	struct bnxt_qplib_cq *lib_cq;
+	struct bnxt_re_qp *qp;
+	struct bnxt_re_cq *cq;
 	int rc = 0;
 	u8 event;
 
@@ -1002,11 +1141,19 @@ static int bnxt_re_handle_affi_async_event(struct creq_qp_event *affi_async,
 		return rc; /* QP was already dead, still return success */
 
 	event = affi_async->event;
-	if (event == CREQ_QP_EVENT_EVENT_QP_ERROR_NOTIFICATION) {
-		struct bnxt_qplib_qp *lib_qp = obj;
-		struct bnxt_re_qp *qp = container_of(lib_qp, struct bnxt_re_qp,
-						     qplib_qp);
+	switch (event) {
+	case CREQ_QP_EVENT_EVENT_QP_ERROR_NOTIFICATION:
+		lib_qp = obj;
+		qp = container_of(lib_qp, struct bnxt_re_qp, qplib_qp);
 		rc = bnxt_re_handle_qp_async_event(affi_async, qp);
+		break;
+	case CREQ_QP_EVENT_EVENT_CQ_ERROR_NOTIFICATION:
+		lib_cq = obj;
+		cq = container_of(lib_cq, struct bnxt_re_cq, qplib_cq);
+		rc = bnxt_re_handle_cq_async_error(affi_async, cq);
+		break;
+	default:
+		rc = -EINVAL;
 	}
 	return rc;
 }
-- 
2.5.5


[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4224 bytes --]

[PATCH for-next 3/3] RDMA/bnxt_re: Do not report SRQ error in srq notification

[Selvin Xavier]

[-- Attachment #1: Type: text/plain, Size: 1185 bytes --]

From: Chandramohan Akula <chandramohan.akula@broadcom.com>

In the SRQ notification handler, do not report the SRQ_ERROR
in the default event case, as there was no error.

Signed-off-by: Chandramohan Akula <chandramohan.akula@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
---
 drivers/infiniband/hw/bnxt_re/main.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c
index 7bd18ec..b6a2d83 100644
--- a/drivers/infiniband/hw/bnxt_re/main.c
+++ b/drivers/infiniband/hw/bnxt_re/main.c
@@ -1187,13 +1187,10 @@ static int bnxt_re_srqn_handler(struct bnxt_qplib_nq *nq,
 
 	ib_event.device = &srq->rdev->ibdev;
 	ib_event.element.srq = &srq->ib_srq;
-	if (event == NQ_SRQ_EVENT_EVENT_SRQ_THRESHOLD_EVENT)
-		ib_event.event = IB_EVENT_SRQ_LIMIT_REACHED;
-	else
-		ib_event.event = IB_EVENT_SRQ_ERR;
 
 	if (srq->ib_srq.event_handler) {
-		/* Lock event_handler? */
+		if (event == NQ_SRQ_EVENT_EVENT_SRQ_THRESHOLD_EVENT)
+			ib_event.event = IB_EVENT_SRQ_LIMIT_REACHED;
 		(*srq->ib_srq.event_handler)(&ib_event,
 					     srq->ib_srq.srq_context);
 	}
-- 
2.5.5


[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4224 bytes --]

Re: [PATCH for-next 1/3] RDMA/bnxt_re: Update HW interface headers

[Leon Romanovsky]

On Wed, Oct 04, 2023 at 10:31:27PM -0700, Selvin Xavier wrote:
> From: Chandramohan Akula <chandramohan.akula@broadcom.com>
> 
> Updating the HW structures for the affiliated event and error
> reporting. Newly added interface structures will be used in the
> followup patch.
> 
> Signed-off-by: Chandramohan Akula <chandramohan.akula@broadcom.com>
> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
> ---
>  drivers/infiniband/hw/bnxt_re/roce_hsi.h | 62 ++++++++++++++++++++++++++++++++
>  1 file changed, 62 insertions(+)

<...>

> +	#define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_LAST \
> +			CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RETX_SETUP_ERROR
>  	__le32	xid;

<...>

> +	#define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_LAST \
> +			CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PSN_NOT_FOUND

These defines are not used.

Thanks

>  	__le16	sq_cons_idx;
>  	__le16	rq_cons_idx;
>  };
> -- 
> 2.5.5
> 

Re: [PATCH for-next 2/3] RDMA/bnxt_re: Report async events and errors

[Leon Romanovsky]

On Wed, Oct 04, 2023 at 10:31:28PM -0700, Selvin Xavier wrote:
> From: Chandramohan Akula <chandramohan.akula@broadcom.com>
> 
> Report QP, SRQ and CQ async events and errors.
> 
> Signed-off-by: Chandramohan Akula <chandramohan.akula@broadcom.com>
> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
> ---
>  drivers/infiniband/hw/bnxt_re/main.c | 165 +++++++++++++++++++++++++++++++++--
>  1 file changed, 156 insertions(+), 9 deletions(-)

<...>

> +	if (err_event->res_err_state_reason || err_event->req_err_state_reason) {
> +		dev_err_once(rdev_to_dev(qp->rdev),
> +			     "%s %s qp_id: %d cons (%d %d) req (%d %d) res (%d %d)\n",
> +			     __func__, rdma_is_kernel_res(&qp->ib_qp.res) ? "kernel" : "user",
> +			     qp->qplib_qp.id,
> +			     err_event->sq_cons_idx,
> +			     err_event->rq_cons_idx,
> +			     err_event->req_slow_path_state,
> +			     err_event->req_err_state_reason,
> +			     err_event->res_slow_path_state,
> +			     err_event->res_err_state_reason);

<...>

> +		dev_err_once(rdev_to_dev(cq->rdev),
> +			     "%s err reason %d\n", __func__, cqerr->cq_err_reason);
> +		cq->ib_cq.event_handler(&ibevent, cq->ib_cq.cq_context);

ibdev_*() prints please.

Thanks

Re: [PATCH for-next 1/3] RDMA/bnxt_re: Update HW interface headers

[Selvin Xavier]

[-- Attachment #1: Type: text/plain, Size: 1344 bytes --]

On Mon, Oct 9, 2023 at 11:39 AM Leon Romanovsky <leon@kernel.org> wrote:
>
> On Wed, Oct 04, 2023 at 10:31:27PM -0700, Selvin Xavier wrote:
> > From: Chandramohan Akula <chandramohan.akula@broadcom.com>
> >
> > Updating the HW structures for the affiliated event and error
> > reporting. Newly added interface structures will be used in the
> > followup patch.
> >
> > Signed-off-by: Chandramohan Akula <chandramohan.akula@broadcom.com>
> > Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
> > ---
> >  drivers/infiniband/hw/bnxt_re/roce_hsi.h | 62 ++++++++++++++++++++++++++++++++
> >  1 file changed, 62 insertions(+)
>
> <...>
>
> > +     #define CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_LAST \
> > +                     CREQ_QP_ERROR_NOTIFICATION_REQ_ERR_STATE_REASON_REQ_RETX_SETUP_ERROR
> >       __le32  xid;
>
> <...>
>
> > +     #define CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_LAST \
> > +                     CREQ_QP_ERROR_NOTIFICATION_RES_ERR_STATE_REASON_RES_PSN_NOT_FOUND
>
> These defines are not used.
These defines are from an autogenerated file and it indicates the LAST
define from this set. It won't be used. I will remove it from this
patch
Thanks
>
> Thanks
>
> >       __le16  sq_cons_idx;
> >       __le16  rq_cons_idx;
> >  };
> > --
> > 2.5.5
> >
>
>

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4224 bytes --]

Re: [PATCH for-next 2/3] RDMA/bnxt_re: Report async events and errors

[Selvin Xavier]

[-- Attachment #1: Type: text/plain, Size: 1606 bytes --]

On Mon, Oct 9, 2023 at 11:40 AM Leon Romanovsky <leon@kernel.org> wrote:
>
> On Wed, Oct 04, 2023 at 10:31:28PM -0700, Selvin Xavier wrote:
> > From: Chandramohan Akula <chandramohan.akula@broadcom.com>
> >
> > Report QP, SRQ and CQ async events and errors.
> >
> > Signed-off-by: Chandramohan Akula <chandramohan.akula@broadcom.com>
> > Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
> > ---
> >  drivers/infiniband/hw/bnxt_re/main.c | 165 +++++++++++++++++++++++++++++++++--
> >  1 file changed, 156 insertions(+), 9 deletions(-)
>
> <...>
>
> > +     if (err_event->res_err_state_reason || err_event->req_err_state_reason) {
> > +             dev_err_once(rdev_to_dev(qp->rdev),
> > +                          "%s %s qp_id: %d cons (%d %d) req (%d %d) res (%d %d)\n",
> > +                          __func__, rdma_is_kernel_res(&qp->ib_qp.res) ? "kernel" : "user",
> > +                          qp->qplib_qp.id,
> > +                          err_event->sq_cons_idx,
> > +                          err_event->rq_cons_idx,
> > +                          err_event->req_slow_path_state,
> > +                          err_event->req_err_state_reason,
> > +                          err_event->res_slow_path_state,
> > +                          err_event->res_err_state_reason);
>
> <...>
>
> > +             dev_err_once(rdev_to_dev(cq->rdev),
> > +                          "%s err reason %d\n", __func__, cqerr->cq_err_reason);
> > +             cq->ib_cq.event_handler(&ibevent, cq->ib_cq.cq_context);
>
> ibdev_*() prints please.
will do
>
> Thanks

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4224 bytes --]