The compiler noticed this: the original code passed the address of a pointer on the stack rather than of a structure to gettimeofday(). It's not clear whether this actually was causing problems. However, the code didn't actually use that info anyway, so this patch simply stops requesting it entirely.

Patch also fixes a couple of warnings about disallowed interspersing of code and variable declarations.

Dan, is this still relevant?

I don't know. If the code is still there and in this state, there's may be a super tiny code-size win here.

Comment on attachment 694984 [details] [diff] [review] fix android IPC apparent stack corruption Review of attachment 694984 [details] [diff] [review]: ----------------------------------------------------------------- rerouting to byron

Comment on attachment 694984 [details] [diff] [review] fix android IPC apparent stack corruption Review of attachment 694984 [details] [diff] [review]: ----------------------------------------------------------------- ::: media/webrtc/signaling/src/sipcc/cpr/android/cpr_android_ipc.c @@ +180,5 @@ > static const char fname[] = "cprCreateMessageQueue"; > cpr_msg_queue_t *msgq; > static int key_id = 100; /* arbitrary starting number */ > + pthread_cond_t _cond = PTHREAD_COND_INITIALIZER; > + pthread_mutex_t _lock = PTHREAD_MUTEX_INITIALIZER; We will leak these if cpr_calloc fails below. Why don't we just do something like msgq->cond = PTHREAD_COND_INITIALIZER? What are the temp variables for anyway?

sdparta removed this function