This fixes a futex key reference count bug in futex_lock_pi(),
where a key's reference count is incremented twice but decremented
only once, causing the backing object to not be released.
If the futex is created in a temporary file in an ext3 file system,
this bug causes the file's inode to become an "undead" orphan,
which causes an oops from a BUG_ON() in ext3_put_super() when the
file system is unmounted. glibc's test suite is known to trigger this,
see <http://bugzilla.kernel.org/show_bug.cgi?id=14256>.
The bug is a regression from 2.6.28-git3, namely Peter Zijlstra's
38d47c1b7075bd7ec3881141bb3629da58f88dab "[PATCH] futex: rely on
get_user_pages() for shared futexes". That commit made get_futex_key()
also increment the reference count of the futex key, and updated its
callers to decrement the key's reference count before returning.
Unfortunately the normal exit path in futex_lock_pi() wasn't corrected:
the reference count is incremented by get_futex_key() and queue_lock(),
but the normal exit path only decrements once, via unqueue_me_pi().
The fix is to put_futex_key() after unqueue_me_pi(), since 2.6.31
this is easily done by 'goto out_put_key' rather than 'goto out'.
Signed-off-by: Mikael Pettersson <[email protected]>
Cc: <[email protected]>
---
This patch also works for backports to 2.6.32 and 2.6.31,
but 2.6.30 and 2.6.29 need a "put_futex_key(fshared, &q.key);"
after "unqueue_me_pi(&q);" instead. I can supply patches
for these older kernels if necessary.
kernel/futex.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff -rupN linux-2.6.33-rc5/kernel/futex.c linux-2.6.33-rc5.futex_lock_pi-key-refcnt-fix/kernel/futex.c
--- linux-2.6.33-rc5/kernel/futex.c 2010-01-22 01:21:57.000000000 +0100
+++ linux-2.6.33-rc5.futex_lock_pi-key-refcnt-fix/kernel/futex.c 2010-01-23 21:00:14.000000000 +0100
@@ -1971,7 +1971,7 @@ retry_private:
/* Unqueue and drop the lock */
unqueue_me_pi(&q);
- goto out;
+ goto out_put_key;
out_unlock_put_key:
queue_unlock(&q, hb);
Mikael Pettersson wrote:
> This fixes a futex key reference count bug in futex_lock_pi(),
> where a key's reference count is incremented twice but decremented
> only once, causing the backing object to not be released.
>
> If the futex is created in a temporary file in an ext3 file system,
> this bug causes the file's inode to become an "undead" orphan,
> which causes an oops from a BUG_ON() in ext3_put_super() when the
> file system is unmounted. glibc's test suite is known to trigger this,
> see <http://bugzilla.kernel.org/show_bug.cgi?id=14256>.
>
> The bug is a regression from 2.6.28-git3, namely Peter Zijlstra's
> 38d47c1b7075bd7ec3881141bb3629da58f88dab "[PATCH] futex: rely on
> get_user_pages() for shared futexes". That commit made get_futex_key()
> also increment the reference count of the futex key, and updated its
> callers to decrement the key's reference count before returning.
> Unfortunately the normal exit path in futex_lock_pi() wasn't corrected:
> the reference count is incremented by get_futex_key() and queue_lock(),
> but the normal exit path only decrements once, via unqueue_me_pi().
> The fix is to put_futex_key() after unqueue_me_pi(), since 2.6.31
> this is easily done by 'goto out_put_key' rather than 'goto out'.
>
> Signed-off-by: Mikael Pettersson <[email protected]>
Nice catch. I think we need to do something to make the reference
counting more explicit in futex.c. For now, this is the right fix.
Acked-by: Darren Hart <[email protected]>
> Cc: <[email protected]>
> ---
> This patch also works for backports to 2.6.32 and 2.6.31,
> but 2.6.30 and 2.6.29 need a "put_futex_key(fshared, &q.key);"
> after "unqueue_me_pi(&q);" instead. I can supply patches
> for these older kernels if necessary.
>
> kernel/futex.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff -rupN linux-2.6.33-rc5/kernel/futex.c linux-2.6.33-rc5.futex_lock_pi-key-refcnt-fix/kernel/futex.c
> --- linux-2.6.33-rc5/kernel/futex.c 2010-01-22 01:21:57.000000000 +0100
> +++ linux-2.6.33-rc5.futex_lock_pi-key-refcnt-fix/kernel/futex.c 2010-01-23 21:00:14.000000000 +0100
> @@ -1971,7 +1971,7 @@ retry_private:
> /* Unqueue and drop the lock */
> unqueue_me_pi(&q);
>
> - goto out;
> + goto out_put_key;
>
> out_unlock_put_key:
> queue_unlock(&q, hb);
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
--
Darren Hart
IBM Linux Technology Center
Real-Time Linux Team