Re: 3.0-rcX BUG at fs/btrfs/ioctl.c:432 - bisected

[Sage Weil]

On Fri, 10 Jun 2011, Josef Bacik wrote:
> On 06/10/2011 02:35 PM, Sage Weil wrote:
> > On Fri, 10 Jun 2011, Josef Bacik wrote:
> >> On 06/10/2011 02:14 PM, Sage Weil wrote:
> >>> On Fri, 10 Jun 2011, Sage Weil wrote:
> >>>> On Fri, 10 Jun 2011, Chris Mason wrote:
> >>>>> Excerpts from Jim Schutt's message of 2011-06-10 13:06:22 -0400:
> >>>>>
> >>>>> [ two different btrfs crashes ]
> >>>>>
> >>>>> I think your two crashes in btrfs were from the uninit variables and
> >>>>> those should be fixed in rc2.
> >>>>>
> >>>>>> When I did my bisection, my criteria for success/failure was
> >>>>>> "did mkcephfs succeed?".  When I apply this criteria to a recent
> >>>>>> linus kernel (e.g. 06e86849cf4019), which includes the fix you
> >>>>>> mentioned (aa0467d8d2a00e), I get still a different failure mode,
> >>>>>> which doesn't actually reference btrfs:
> >>>>>>
> >>>>>> [  276.364178] BUG: unable to handle kernel NULL pointer dereference at 000000000000000a
> >>>>>> [  276.365127] IP: [<ffffffffa05434b1>] journal_start+0x3e/0x9c [jbd]
> >>>>>
> >>>>> Looking at the resulting code in the oops, we're here in journal_start:
> >>>>>
> >>>>>         if (handle) {
> >>>>>                 J_ASSERT(handle->h_transaction->t_journal == journal);
> >>>>>
> >>>>> handle comes from current->journal_info, and we're doing a deref on
> >>>>> handle->h_transaction, which is probably 0xa.
> >>>>>
> >>>>> So, we're leaving crud in current->journal_info and ext3 is finding it.
> >>>>>
> >>>>> Perhaps its from ceph starting a transaction but leaving it running?
> >>>>> The bug came with Josef's transaction performance fixes, but it is
> >>>>> probably a mixture of his code with the ioctls ceph is using.
> >>>>
> >>>> Ah, yeah, that's the problem.  We saw a similar problem a while back with 
> >>>> the start/stop transaction ioctls.  In this case, create_snapshot is doing
> >>>>
> >>>> 	trans = btrfs_start_transaction(root->fs_info->extent_root, 5);
> >>>> 	if (IS_ERR(trans)) {
> >>>> 		ret = PTR_ERR(trans);
> >>>> 		goto fail;
> >>>> 	}
> >>>>
> >>>> which sets current->journal_info.  Then
> >>>>
> >>>> 	ret = btrfs_snap_reserve_metadata(trans, pending_snapshot);
> >>>> 	BUG_ON(ret);
> >>>>
> >>>> 	list_add(&pending_snapshot->list,
> >>>> 		 &trans->transaction->pending_snapshots);
> >>>> 	if (async_transid) {
> >>>> 		*async_transid = trans->transid;
> >>>> 		ret = btrfs_commit_transaction_async(trans,
> >>>> 				     root->fs_info->extent_root, 1);
> >>>> 	} else {
> >>>> 		ret = btrfs_commit_transaction(trans,
> >>>> 				       root->fs_info->extent_root);
> >>>> 	}
> >>>>
> >>>> but the async snap creation ioctl takes the async path, which runs 
> >>>> btrfs_commit_transaction in a worker thread.
> >>>>
> >>>> I'm not sure what the right thing to do is here is... can whatever is in 
> >>>> journal_info be attached to trans instead in 
> >>>> btrfs_commit_transaction_async()?
> >>>
> >>> It looks like it's not used for anything in btrfs, actually; it's just set 
> >>> and cleared.  What's the point of that?
> >>>
> >>
> >> It is used now, check the beginning of start_transaction().  Thanks,
> > 
> > Oh I see, okay.
> > 
> > So clearing it in btrfs_commit_transaction_async should be fine then, 
> > right?  When btrfs_commit_transaction runs in the other thread it won't 
> > care that current->journal_info is NULL.
> > 
> 
> Oh yeah your patch is good :),

Okay cool.  Here's the fix with a proper changelog and a little 
use-after-free paranoia.

Thanks!
sage