Re: [PATCH] xfs: take hm->lock in xfs_ioc_health_monitor() before insert

[Carlos Maiolino <cem@kernel.org>]

On Thu, Sep 10, 2026 at 10:42:53PM +0900, Shigeru Yoshida wrote:
> Carlos Maiolino <cem@kernel.org> writes:
> 
> > On Wed, Sep 09, 2026 at 10:07:56PM -0700, Christoph Hellwig wrote:
> >> On Sun, Sep 06, 2026 at 09:51:50PM -0700, Darrick J. Wong wrote:
> >> > On Sun, Sep 06, 2026 at 08:16:29PM +0530, Deepanshu Kartikey wrote:
> >> > > __xfs_healthmon_insert() asserts that hm->lock is held
> >> > > (lockdep_assert_held), but xfs_ioc_health_monitor() called it right
> >> > > after allocating hm, before ever taking the lock, triggering a
> >> > > lockdep warning.
> >> > > 
> >> > > Take hm->lock around the call.
> >> > > 
> >> > > Fixes: b3a289a2a9397 ("xfs: create event queuing, formatting, and discovery infrastructure")
> >> > > Reported-by: syzbot+ccdf3469f5f653bff7ac@syzkaller.appspotmail.com
> >> > > Closes: https://syzkaller.appspot.com/bug?extid=ccdf3469f5f653bff7ac
> >> > > Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
> >> > 
> >> > Heh, I added a patch just like this one to my own tree.
> >> > Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
> >> 
> >> Can we get this queued up ASAP?  It's also causing a lot of failures in
> >> our zoned XFS CI.
> >> 
> >> 
> >
> > Done. It's in for-next already. Not sure if I'll manage to push the
> > attrs fsblock fix yet this -rc, but this one will go to Linus on
> > Saturday
> 
> I noticed that the second mutex operation in this commit was
> accidentally left as mutex_lock() instead of mutex_unlock(). I've
> sent a fix as a separate patch ("xfs: fix double mutex_lock() call
> in xfs_ioc_health_monitor()"). Feel free to use it if it helps.

Thanks for the info, something went pretty wrong with b4. I didn't touch
the patch manually and somehow it got twisted when I pulled it.
I'll figure it out what went wrong but first let me fix this...

This shouldn't be in the tree at all, so I'll remove the patch for now
as this has not been pulled into the linux-next.


> 
> Thanks,
> Shigeru
> 
>