[PATCH] tgtd: fix memory access overrun on array of pthread_t type data

bs_thread_close function accesses memory outside allocated region
because it reads each array element with an index before confirming
that the index is within the range of the array.

This fixes the issue and gets rid of the following warning detected by
valgrind:

 ==10848== Invalid read of size 8
 ==10848==    at 0x42AF91: bs_thread_close (bs.c:461)
 ==10848==    by 0x42BD94: bs_sheepdog_exit (bs_sheepdog.c:1275)
 ==10848==    by 0x4183B7: tgt_device_destroy (target.c:739)
 ==10848==    by 0x41608D: mtask_received (mgmt.c:251)
 ==10848==    by 0x4164CB: mtask_recv_send_handler (mgmt.c:670)
 ==10848==    by 0x4141F8: event_loop (tgtd.c:428)
 ==10848==    by 0x414899: main (tgtd.c:611)

Signed-off-by: Ryusuke Konishi <[email protected]>
---
 usr/bs.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/usr/bs.c b/usr/bs.c
index b0ee66f..13d3b4e 100644
--- a/usr/bs.c
+++ b/usr/bs.c
@@ -458,7 +458,7 @@ void bs_thread_close(struct bs_thread_info *info)
 	info->stop = 1;
 	pthread_cond_broadcast(&info->pending_cond);
 
-	for (i = 0; info->worker_thread[i] && i < info->nr_worker_threads; i++)
+	for (i = 0; i < info->nr_worker_threads && info->worker_thread[i]; i++)
 		pthread_join(info->worker_thread[i], NULL);
 
 	pthread_cond_destroy(&info->pending_cond);
-- 
1.7.9.3