session: avoid vpp deadlock due to app crash

In high traffic scenarios, if app crashed or hang on somewhere, app_mq
will quickly accumulate to full, after which vpp worker will try 100
times before giving up allocating slot for every msg. This will cause
vpp main thread barrier sync to fail.

Type: fix

Signed-off-by: liuyacan <liuyacan@corp.netease.com>
Change-Id: I2b2bf2b272c5b3ca7e4a56af179af12bbcde149d

1 files changed, 6 insertions, 0 deletions

diff --git a/src/vnet/session/session_api.c b/src/vnet/session/session_api.c
index 7e7cffbbdd4..00e67dcd2d0 100644
--- a/src/vnet/session/session_api.c
+++ b/src/vnet/session/session_api.c
@@ -93,6 +93,12 @@ mq_try_lock_and_alloc_msg (svm_msg_q_t * app_mq, svm_msg_q_msg_t * msg)
 						SVM_Q_NOWAIT, msg);
       if (!rv)
 	return 0;
+      /*
+       * Break the loop if mq is full, usually this is because the
+       * app has crashed or is hanging on somewhere.
+       */
+      if (rv != -1)
+	break;
       try++;
       usleep (1);
     }