On Sun, May 01, 2011 at 10:35:38AM +0800, Hugh Dickins wrote:
> Yesterday's mmotm hangs at startup, and with lockdep it reports:
> BUG: spinlock recursion on CPU#1, blkid/284 - with bdi_lock_two()
> called from bdev_inode_switch_bdi() in the backtrace. It appears
> that this function is sometimes called with new the same as old.
>
> Signed-off-by: Hugh Dickins <hughd@xxxxxxxxxx>
Thanks!
Reviewed-by: Wu Fengguang <fengguang.wu@xxxxxxxxx>
> Fix to
> writeback-split-inode_wb_list_lock-into-bdi_writebacklist_lock.patch
>
> fs/block_dev.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> --- 2.6.39-rc5-mm1/fs/block_dev.c 2011-04-29 18:20:09.183314733 -0700
> +++ linux/fs/block_dev.c 2011-04-30 17:55:45.718785263 -0700
> @@ -57,6 +57,8 @@ static void bdev_inode_switch_bdi(struct
> {
> struct backing_dev_info *old = inode->i_data.backing_dev_info;
>
> + if (dst == old)
> + return;
nitpick: it could help to add a comment
/* avoid spinlock recursion */
to indicate that's not merely an optional optimization, but indeed
required for correctness.
Thanks,
Fengguang
> bdi_lock_two(&old->wb, &dst->wb);
> spin_lock(&inode->i_lock);
> inode->i_data.backing_dev_info = dst;
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
