[dpdk-dev] [PATCH v2 5/5] eal: fix race condition in IPC requests

[Anatoly Burakov]

Unlocking the action list before sending message and locking it
again aftterwards introduces a window where a response might
arrive before we have a chance to start waiting on a condition,
resulting in timeouts on valid messages.

Fixes: 783b6e54971d ("eal: add synchronous multi-process communication")
Cc: jianfeng.tan at intel.com

Signed-off-by: Anatoly Burakov <anatoly.burakov at intel.com>
---

Notes:
    v2: added this patch

 lib/librte_eal/common/eal_common_proc.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/lib/librte_eal/common/eal_common_proc.c b/lib/librte_eal/common/eal_common_proc.c
index 3a1088e..da7930f 100644
--- a/lib/librte_eal/common/eal_common_proc.c
+++ b/lib/librte_eal/common/eal_common_proc.c
@@ -561,10 +561,10 @@ mp_request_one(const char *dst, struct rte_mp_msg *req,
 	exist = find_sync_request(dst, req->name);
 	if (!exist)
 		TAILQ_INSERT_TAIL(&sync_requests.requests, &sync_req, next);
-	pthread_mutex_unlock(&sync_requests.lock);
 	if (exist) {
 		RTE_LOG(ERR, EAL, "A pending request %s:%s\n", dst, req->name);
 		rte_errno = EEXIST;
+		pthread_mutex_unlock(&sync_requests.lock);
 		return -1;
 	}
 
@@ -578,7 +578,6 @@ mp_request_one(const char *dst, struct rte_mp_msg *req,
 
 	reply->nb_sent++;
 
-	pthread_mutex_lock(&sync_requests.lock);
 	do {
 		pthread_cond_timedwait(&sync_req.cond, &sync_requests.lock, ts);
 		/* Check spurious wakeups */
-- 
2.7.4