DPDK patches and discussions
 help / color / mirror / Atom feed
From: Ajit Khaparde <ajit.khaparde@broadcom.com>
To: dev@dpdk.org
Cc: ferruh.yigit@intel.com,
	Somnath Kotur <somnath.kotur@broadcom.com>,
	stable@dpdk.org, Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Subject: [dpdk-dev] [PATCH 5/7] net/bnxt: fix race between interrupt handler and dev config
Date: Sun, 13 Oct 2019 17:01:18 -0700	[thread overview]
Message-ID: <20191014000120.3020-6-ajit.khaparde@broadcom.com> (raw)
In-Reply-To: <20191014000120.3020-1-ajit.khaparde@broadcom.com>

From: Somnath Kotur <somnath.kotur@broadcom.com>

Fix a possible that while interrupt handler is in the middle of processing,
dev_configure_op() might come in and try to destroy/realloc the default
completion ring on which interrupt completions arrive.
Synchronize the two so they don't race each other

Fixes: 7bc8e9a227cc ("net/bnxt: support async link notification")
Cc: stable@dpdk.org

Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
---
 drivers/net/bnxt/bnxt.h        |  1 +
 drivers/net/bnxt/bnxt_ethdev.c | 19 +++++++++++++++++--
 drivers/net/bnxt/bnxt_irq.c    |  7 ++++++-
 3 files changed, 24 insertions(+), 3 deletions(-)

diff --git a/drivers/net/bnxt/bnxt.h b/drivers/net/bnxt/bnxt.h
index 189f693b06..e5bdf63602 100644
--- a/drivers/net/bnxt/bnxt.h
+++ b/drivers/net/bnxt/bnxt.h
@@ -574,6 +574,7 @@ struct bnxt {
 	void				*hwrm_short_cmd_req_addr;
 	rte_iova_t			hwrm_short_cmd_req_dma_addr;
 	rte_spinlock_t			hwrm_lock;
+	pthread_mutex_t			def_cp_lock;
 	uint16_t			max_req_len;
 	uint16_t			max_resp_len;
 	uint16_t                        hwrm_max_ext_req_len;
diff --git a/drivers/net/bnxt/bnxt_ethdev.c b/drivers/net/bnxt/bnxt_ethdev.c
index b58a875c7d..e7ec99e15d 100644
--- a/drivers/net/bnxt/bnxt_ethdev.c
+++ b/drivers/net/bnxt/bnxt_ethdev.c
@@ -627,6 +627,9 @@ static int bnxt_dev_configure_op(struct rte_eth_dev *eth_dev)
 		 * resource reservation. This will ensure the resource counts
 		 * are calculated correctly.
 		 */
+
+		pthread_mutex_lock(&bp->def_cp_lock);
+
 		if (!BNXT_HAS_NQ(bp) && bp->async_cp_ring) {
 			bnxt_disable_int(bp);
 			bnxt_free_cp_ring(bp, bp->async_cp_ring);
@@ -635,15 +638,20 @@ static int bnxt_dev_configure_op(struct rte_eth_dev *eth_dev)
 		rc = bnxt_hwrm_func_reserve_vf_resc(bp, false);
 		if (rc) {
 			PMD_DRV_LOG(ERR, "HWRM resource alloc fail:%x\n", rc);
+			pthread_mutex_unlock(&bp->def_cp_lock);
 			return -ENOSPC;
 		}
 
 		if (!BNXT_HAS_NQ(bp) && bp->async_cp_ring) {
 			rc = bnxt_alloc_async_cp_ring(bp);
-			if (rc)
+			if (rc) {
+				pthread_mutex_unlock(&bp->def_cp_lock);
 				return rc;
+			}
 			bnxt_enable_int(bp);
 		}
+
+		pthread_mutex_unlock(&bp->def_cp_lock);
 	} else {
 		/* legacy driver needs to get updated values */
 		rc = bnxt_hwrm_func_qcaps(bp);
@@ -4585,8 +4593,14 @@ bnxt_init_locks(struct bnxt *bp)
 	int err;
 
 	err = pthread_mutex_init(&bp->flow_lock, NULL);
-	if (err)
+	if (err) {
 		PMD_DRV_LOG(ERR, "Unable to initialize flow_lock\n");
+		return err;
+	}
+
+	err = pthread_mutex_init(&bp->def_cp_lock, NULL);
+	if (err)
+		PMD_DRV_LOG(ERR, "Unable to initialize def_cp_lock\n");
 	return err;
 }
 
@@ -4735,6 +4749,7 @@ static void
 bnxt_uninit_locks(struct bnxt *bp)
 {
 	pthread_mutex_destroy(&bp->flow_lock);
+	pthread_mutex_destroy(&bp->def_cp_lock);
 }
 
 static int
diff --git a/drivers/net/bnxt/bnxt_irq.c b/drivers/net/bnxt/bnxt_irq.c
index 6c5bcfb623..4feb637e22 100644
--- a/drivers/net/bnxt/bnxt_irq.c
+++ b/drivers/net/bnxt/bnxt_irq.c
@@ -30,9 +30,12 @@ void bnxt_int_handler(void *param)
 		return;
 
 	raw_cons = cpr->cp_raw_cons;
+	pthread_mutex_lock(&bp->def_cp_lock);
 	while (1) {
-		if (!cpr || !cpr->cp_ring_struct || !cpr->cp_db.doorbell)
+		if (!cpr || !cpr->cp_ring_struct || !cpr->cp_db.doorbell) {
+			pthread_mutex_unlock(&bp->def_cp_lock);
 			return;
+		}
 
 		cons = RING_CMP(cpr->cp_ring_struct, raw_cons);
 		cmp = &cpr->cp_desc_ring[cons];
@@ -49,6 +52,8 @@ void bnxt_int_handler(void *param)
 		bnxt_db_nq_arm(cpr);
 	else
 		B_CP_DB_REARM(cpr, cpr->cp_raw_cons);
+
+	pthread_mutex_unlock(&bp->def_cp_lock);
 }
 
 int bnxt_free_int(struct bnxt *bp)
-- 
2.20.1 (Apple Git-117)


  parent reply	other threads:[~2019-10-14  0:02 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-10-14  0:01 [dpdk-dev] [PATCH 0/7] patchset for bnxt Ajit Khaparde
2019-10-14  0:01 ` [dpdk-dev] [PATCH 1/7] net/bnxt: fix dereference before null check Ajit Khaparde
2019-10-14  0:01 ` [dpdk-dev] [PATCH 2/7] net/bnxt: reset Rx allocation state on Rx queue restart Ajit Khaparde
2019-10-14  0:01 ` [dpdk-dev] [PATCH 3/7] net/bnxt: cleanup comments in source Ajit Khaparde
2019-10-14  0:01 ` [dpdk-dev] [PATCH 4/7] net/bnxt: move macro definitions to header file Ajit Khaparde
2019-10-14  0:01 ` Ajit Khaparde [this message]
2019-10-14  0:01 ` [dpdk-dev] [PATCH 6/7] net/bnxt: fix endianness while retrieving MTU from FW Ajit Khaparde
2019-10-14  0:01 ` [dpdk-dev] [PATCH 7/7] net/bnxt: fix error handling Ajit Khaparde
2019-10-14  3:58 ` [dpdk-dev] [PATCH 0/7] patchset for bnxt Ajit Khaparde

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191014000120.3020-6-ajit.khaparde@broadcom.com \
    --to=ajit.khaparde@broadcom.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=kalesh-anakkur.purayil@broadcom.com \
    --cc=somnath.kotur@broadcom.com \
    --cc=stable@dpdk.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).