On Sat 09-06-12 00:45:35, Wu Fengguang wrote: > On Fri, Jun 08, 2012 at 05:07:36PM +0200, Jan Kara wrote: > > On Fri 08-06-12 10:36:13, Ted Tso wrote: > > > > > > I can reproduce this fairly easily by using ext4 w/o a journal, running > > > under KVM with 1024megs memory, with fsstress (xfstests #13): > > Good catch, thanks! > > > Argh, I wonder how come I didn't hit this. Does attached patch fix the > > problem? > > > diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c > > index 8d2fb8c..41a3ccf 100644 > > --- a/fs/fs-writeback.c > > +++ b/fs/fs-writeback.c > > @@ -664,6 +664,7 @@ static long writeback_sb_inodes(struct super_block *sb, > > /* Wait for I_SYNC. This function drops i_lock... */ > > inode_sleep_on_writeback(inode); > > /* Inode may be gone, start again */ > > + spin_lock(&wb->list_lock); > > continue; > > } > > That looks like the fix. So I pushed it to writeback-for-next. > Thanks for the quick fixing! Attached is a version with proper sign-off and changelog. Just in case you didn't create one yourself already. Honza -- Jan Kara <jack@xxxxxxx> SUSE Labs, CR
>From 98fecbcb9edf39b4ab6aab8a85342f77665cac4d Mon Sep 17 00:00:00 2001 From: Jan Kara <jack@xxxxxxx> Date: Fri, 8 Jun 2012 17:08:03 +0200 Subject: [PATCH] writeback: Fix lock imbalance in writeback_sb_inodes() Fix bug introduced by 169ebd90. We have to have wb_list_lock locked when restarting writeback loop after having waited for inode writeback. Signed-off-by: Jan Kara <jack@xxxxxxx> --- fs/fs-writeback.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c index 8d2fb8c..41a3ccf 100644 --- a/fs/fs-writeback.c +++ b/fs/fs-writeback.c @@ -664,6 +664,7 @@ static long writeback_sb_inodes(struct super_block *sb, /* Wait for I_SYNC. This function drops i_lock... */ inode_sleep_on_writeback(inode); /* Inode may be gone, start again */ + spin_lock(&wb->list_lock); continue; } inode->i_state |= I_SYNC; -- 1.7.1
