Re: kernel BUG at fs/inode.c:LINE!
From: Dmitry Vyukov
Date: Fri Apr 12 2019 - 07:04:42 EST
On Thu, Apr 11, 2019 at 4:23 AM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> On Thu, Apr 11, 2019 at 08:50:17AM +0800, Ian Kent wrote:
> > On Wed, 2019-04-10 at 14:41 +0200, Dmitry Vyukov wrote:
> > > On Wed, Apr 10, 2019 at 2:12 PM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> > > >
> > > > On Wed, Apr 10, 2019 at 08:07:15PM +0800, Ian Kent wrote:
> > > >
> > > > > > I'm unable to find a branch matching the line numbers.
> > > > > >
> > > > > > Given that, on the face of it, the scenario is impossible I'm
> > > > > > seeking clarification on what linux-next to look at for the
> > > > > > sake of accuracy.
> > > > > >
> > > > > > So I'm wondering if this testing done using the master branch
> > > > > > or one of the daily branches one would use to check for conflicts
> > > > > > before posting?
> > > > >
> > > > > Sorry those are tags not branches.
> > > >
> > > > FWIW, that's next-20181214; it is what master had been in mid-December
> > > > and master is rebased every day. Can it be reproduced with the current
> > > > tree?
> > >
> > > From the info on the dashboard we know that it happened only once on
> > > d14b746c (the second one is result of reproducing the first one). So
> > > it was either fixed or just hard to trigger.
> > Looking at the source of tag next-20181214 in linux-next-history I see
> > this is mistake I made due to incorrect error handling which I fixed
> > soon after (there was in fact a double iput()).
> Right - "autofs: fix possible inode leak in autofs_fill_super()" had been
> broken (and completely pointless), leading to double iput() in that failure
> case. And yes, double iput() can trigger that BUG_ON(), and with non-zero
> odds do so with that stack trace.
> As far as I'm concerned, case closed - bug had been in a misguided "fix"
> for inexistent leak (coming from misreading the calling conventions for
> d_make_root()), introduced in -next at next-20181130 and kicked out of
> there in next-20181219. Dropped by Ian's request in
> Message-ID: <66d497c00cffb3e4109ca0d5287c8277954d7132.camel@xxxxxxxxxx>
> which has fixed that crap. Moreover, that posting had been in reply to
> that very syzcaller report, AFAICS.
> I don't know how to tell the bot to STFU and close the report in this
> situation; up to you, folks.
Please see the following for this:
> syzbot will keep track of this bug report. See:
> https://goo.gl/tpsmEJ#bug-status-tracking for how to communicate with syzbot.
There are just 3 operations: mark as fixed by a commit, mark as
invalid, mark as duplicate.
I won't be always around. Tracking statuses of bug reports is in the
interests of kernel quality.
> As an aside, the cause of that bug is that d_make_root() calling conventions
> are insufficiently documented. All we have is
> || d_alloc_root() is gone, along with a lot of bugs caused by code
> ||misusing it. Replacement: d_make_root(inode). The difference is,
> ||d_make_root() drops the reference to inode if dentry allocation fails.
> in Documentation/filesystems/porting, and that's not good enough. Anyone
> willing to take a shot at that? FWIW, the calling conventions are:
> d_make_root(inode) normally allocates and returns a new dentry.
> On failure NULL is returned. A reference to inode is consumed in all
> cases (on success it is transferred to new dentry, on failure it is
> dropped), so failure handling does not need anything done to inode.
> d_make_root(NULL) quietly returns NULL, which further simplifies the
> error handling in typical caller. Usually it's something like
> inode = foofs_new_inode(....);
> s->s_root = d_make_inode(inode);
> if (!s->s_root)
> bugger off, no need to undo inode allocation
> We do not need to check if foofs_new_inode() has returned NULL and we
> do not need any special cleanups in case of failure - not for the
> undoing the inode allocation.
> If anyone cares to convert that into coherent (and printable) documentation,
> patches are welcome...