On Wed, Jul 31, 2024 at 06:46:39PM +0200, Christophe JAILLET wrote:
> Le 31/07/2024 à 00:23, Kent Overstreet a écrit :
> > On Tue, Jul 30, 2024 at 02:45:17PM GMT, Dan Carpenter wrote:
> > > Hello Kent Overstreet,
> > >
> > > Commit ffcbec607613 ("bcachefs: Kill opts.buckets_nouse") from Apr 6,
> > > 2024 (linux-next), leads to the following Smatch static checker
> > > warning:
> > >
> > > fs/bcachefs/super.c:1202 bch2_dev_free()
> > > warn: 'ca->buckets_nouse' double freed
> >
> > yup, that's a bug
> >
> > -- >8 --
> > Subject: [PATCH] bcachefs: Fix double free of ca->buckets_nouse
> >
> > Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> > Fixes: ffcbec6076 ("bcachefs: Kill opts.buckets_nouse")
> > Signed-off-by: Kent Overstreet <kent.overstreet@xxxxxxxxx>
> >
> > diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c
> > index 0542953a7a..6d0ce3d734 100644
> > --- a/fs/bcachefs/super.c
> > +++ b/fs/bcachefs/super.c
> > @@ -1195,7 +1195,6 @@ static void bch2_dev_free(struct bch_dev *ca)
> > if (ca->kobj.state_in_sysfs)
> > kobject_del(&ca->kobj);
> > - kfree(ca->buckets_nouse);
>
> Hi,
>
> unrelated, but apparently this should also have been kvfree().
>
> Dan, do you already have a check for such cases where kvfree() or kfree()
> are called when the other one should be used instead?
>
No. I don't. I've been meaning to write one...
KTODO: write a smatch check for calling kfree instead of kvfree()
I'll probably do it in the next week or two. The min/max changes are
higher priority right now. :P #painful
regards,
dan carpenter
