signals: do_group_exit(): use signal_group_exit() more consistently
authorOleg Nesterov <oleg@tv-sign.ru>
Wed, 30 Apr 2008 07:52:36 +0000 (00:52 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 30 Apr 2008 15:29:33 +0000 (08:29 -0700)
do_group_exit() checks SIGNAL_GROUP_EXIT to avoid taking sighand->siglock.
Since ed5d2cac114202fe2978a9cbcab8f5032796d538 exec() doesn't set this
flag, we should use signal_group_exit().

This is not needed for correctness, but can speedup the multithreaded exec
and makes the code more consistent.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/exit.c

index ae0f2c4..6d019aa 100644 (file)
@@ -1115,12 +1115,13 @@ asmlinkage long sys_exit(int error_code)
 NORET_TYPE void
 do_group_exit(int exit_code)
 {
+       struct signal_struct *sig = current->signal;
+
        BUG_ON(exit_code & 0x80); /* core dumps don't get here */
 
-       if (current->signal->flags & SIGNAL_GROUP_EXIT)
-               exit_code = current->signal->group_exit_code;
+       if (signal_group_exit(sig))
+               exit_code = sig->group_exit_code;
        else if (!thread_group_empty(current)) {
-               struct signal_struct *const sig = current->signal;
                struct sighand_struct *const sighand = current->sighand;
                spin_lock_irq(&sighand->siglock);
                if (signal_group_exit(sig))