Fix intermittent mongoose webserver crash on Windows. Occassionally, a new worker thread will be started, but will have not have a connection to handle because another worker thread grabs it first. If no further connections arrive, the new worker thread will proceed to exit. This will crash when accessing the worker's connection's context, which has never been set.

The fix is to use the context that was given to the worker initially in this case.
BUG=none
TEST=none

Review URL: http://codereview.chromium.org/6244002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@73034 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 6 insertions, 6 deletions

diff --git a/third_party/mongoose/mongoose.c b/third_party/mongoose/mongoose.c
index 7e2416f..035532a 100644
--- a/third_party/mongoose/mongoose.c
+++ b/third_party/mongoose/mongoose.c
@@ -4543,12 +4543,12 @@ worker_thread(struct mg_context *ctx)
   }
 
   /* Signal master that we're done with connection and exiting */
-  pthread_mutex_lock(&conn.ctx->thr_mutex);
-  conn.ctx->num_threads--;
-  conn.ctx->num_idle--;
-  pthread_cond_signal(&conn.ctx->thr_cond);
-  assert(conn.ctx->num_threads >= 0);
-  pthread_mutex_unlock(&conn.ctx->thr_mutex);
+  pthread_mutex_lock(&ctx->thr_mutex);
+  ctx->num_threads--;
+  ctx->num_idle--;
+  pthread_cond_signal(&ctx->thr_cond);
+  assert(ctx->num_threads >= 0);
+  pthread_mutex_unlock(&ctx->thr_mutex);
 
   DEBUG_TRACE((DEBUG_MGS_PREFIX "%s: thread %p exiting",
       __func__, (void *) pthread_self()));