[PATCH 2/3] clear signal->tty when the last thread exits
Oleg Nesterov
cleared and points to nowhere.
This is OK. Nobody should use signal->tty lockless, and it is no longer
possible to take ->siglock. However this looks wrong even if correct, and
the nice OOPS is better than subtle and hard to find bugs.
Change __exit_signal() to clear signal->tty under ->siglock.
Note: __exit_signal() needs more cleanups. It should not check "sig != NULL"
to detect the all-dead case and we have the same issues with signal->stats.
Signed-off-by: Oleg Nesterov <ol...@redhat.com>
---
kernel/exit.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
--- 34-rc1/kernel/exit.c~9_CLEAR_SIGNAL_TTY 2010-03-19 17:25:36.000000000 +0100
+++ 34-rc1/kernel/exit.c 2010-03-19 18:55:02.000000000 +0100
@@ -81,6 +81,7 @@ static void __exit_signal(struct task_st
{
struct signal_struct *sig = tsk->signal;
struct sighand_struct *sighand;
+ struct tty_struct *tty;
BUG_ON(!sig);
BUG_ON(!atomic_read(&sig->count));
@@ -94,6 +95,8 @@ static void __exit_signal(struct task_st
posix_cpu_timers_exit(tsk);
if (thread_group_leader(tsk)) {
posix_cpu_timers_exit_group(tsk);
+ tty = sig->tty;
+ sig->tty = NULL;
} else {
/*
* If there is any task waiting for the group exit
@@ -148,7 +151,7 @@ static void __exit_signal(struct task_st
* see account_group_exec_runtime().
*/
task_rq_unlock_wait(tsk);
- tty_kref_put(sig->tty);
+ tty_kref_put(tty);
}
}
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majo...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Oleg Nesterov
I didn't get this warning, but the old gcc complains
kernel/exit.c: In function 'release_task':
kernel/exit.c:85: warning: 'tty' may be used uninitialized in this function
This clearly wrong, to the point it blames release_task() instead of
__exit_signal(). But let's make compiler happy anyway, hopefully this
is what it wants.
Signed-off-by: Oleg Nesterov <ol...@redhat.com>
---
kernel/exit.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- 34-rc1/kernel/exit.c~FIX_EXIT_SIGNAL_TTY_WARNING 2010-03-21 18:36:44.000000000 +0100
+++ 34-rc1/kernel/exit.c 2010-03-24 14:59:55.000000000 +0100
@@ -82,7 +82,7 @@ static void __exit_signal(struct task_st
struct signal_struct *sig = tsk->signal;
bool group_dead = thread_group_leader(tsk);
struct sighand_struct *sighand;
- struct tty_struct *tty;
+ struct tty_struct *tty = NULL; /* supress gcc warning */
BUG_ON(!sig);
BUG_ON(!atomic_read(&sig->count));
Andrew Morton
> (fixup for signals-clear-signal-tty-when-the-last-thread-exits.patch)
>
> I didn't get this warning, but the old gcc complains
>
> kernel/exit.c: In function 'release_task':
> kernel/exit.c:85: warning: 'tty' may be used uninitialized in this function
>
> This clearly wrong, to the point it blames release_task() instead of
> __exit_signal(). But let's make compiler happy anyway, hopefully this
> is what it wants.
>
> Signed-off-by: Oleg Nesterov <ol...@redhat.com>
> ---
>
> kernel/exit.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> --- 34-rc1/kernel/exit.c~FIX_EXIT_SIGNAL_TTY_WARNING 2010-03-21 18:36:44.000000000 +0100
> +++ 34-rc1/kernel/exit.c 2010-03-24 14:59:55.000000000 +0100
> @@ -82,7 +82,7 @@ static void __exit_signal(struct task_st
> struct signal_struct *sig = tsk->signal;
> bool group_dead = thread_group_leader(tsk);
> struct sighand_struct *sighand;
> - struct tty_struct *tty;
> + struct tty_struct *tty = NULL; /* supress gcc warning */
uninitialized_var() is a neater way.
(uninitialized_var() will save a teeny bit of .text on old gcc. One
suspects that a newer gcc which is capable of working out that this
variable _isn't_ uninitialized would also be capable of eliding the `= 0').
Oleg Nesterov
>
> > - struct tty_struct *tty;
> > + struct tty_struct *tty = NULL; /* supress gcc warning */
>
> uninitialized_var() is a neater way.
Aha, indeed.
Will resend soon...
Oleg.
Oleg Nesterov
kernel/exit.c: In function 'release_task':
kernel/exit.c:85: warning: 'tty' may be used uninitialized in this function
This clearly wrong, to the point it blames release_task() instead of
__exit_signal(). But let's make compiler happy anyway.
Thanks Andrew, now I know we have the handy uninitialized_var() helper ;)
Signed-off-by: Oleg Nesterov <ol...@redhat.com>
---
kernel/exit.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- 34-rc1/kernel/exit.c~FIX_EXIT_SIGNAL_TTY_WARNING 2010-03-21 18:36:44.000000000 +0100
+++ 34-rc1/kernel/exit.c 2010-03-24 17:36:32.000000000 +0100
@@ -82,7 +82,7 @@ static void __exit_signal(struct task_st
struct signal_struct *sig = tsk->signal;
bool group_dead = thread_group_leader(tsk);
struct sighand_struct *sighand;
- struct tty_struct *tty;
+ struct tty_struct *uninitialized_var(tty);
BUG_ON(!sig);
BUG_ON(!atomic_read(&sig->count));