[nginx] Core: fixed a race resulting in extra sem_post()'s.

[Roman Arutyunyan]

details:   http://hg.nginx.org/nginx/rev/74edc0ccf27a
branches:  
changeset: 5970:74edc0ccf27a
user:      Roman Arutyunyan <arut at nginx.com>
date:      Wed Feb 04 16:22:43 2015 +0300
description:
Core: fixed a race resulting in extra sem_post()'s.

The mtx->wait counter was not decremented if we were able to obtain the lock
right after incrementing it.  This resulted in unneeded sem_post() calls,
eventually leading to EOVERFLOW errors being logged, "sem_post() failed
while wake shmtx (75: Value too large for defined data type)".

To close the race, mtx->wait is now decremented if we obtain the lock right
after incrementing it in ngx_shmtx_lock().  The result can become -1 if a
concurrent ngx_shmtx_unlock() decrements mtx->wait before the added code does.
However, that only leads to one extra iteration in the next call of
ngx_shmtx_lock().

diffstat:

 src/core/ngx_shmtx.c |  3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diffs (20 lines):

diff -r 3281de8142f5 -r 74edc0ccf27a src/core/ngx_shmtx.c
--- a/src/core/ngx_shmtx.c	Mon Feb 02 21:28:09 2015 +0300
+++ b/src/core/ngx_shmtx.c	Wed Feb 04 16:22:43 2015 +0300
@@ -101,6 +101,7 @@ ngx_shmtx_lock(ngx_shmtx_t *mtx)
             (void) ngx_atomic_fetch_add(mtx->wait, 1);
 
             if (*mtx->lock == 0 && ngx_atomic_cmp_set(mtx->lock, 0, ngx_pid)) {
+                (void) ngx_atomic_fetch_add(mtx->wait, -1);
                 return;
             }
 
@@ -174,7 +175,7 @@ ngx_shmtx_wakeup(ngx_shmtx_t *mtx)
 
         wait = *mtx->wait;
 
-        if (wait == 0) {
+        if ((ngx_atomic_int_t) wait <= 0) {
             return;
         }