Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 22 Aug 2022 15:52:32 +0100
From:   Matthew Wilcox <willy@infradead.org>
To:     Siddh Raman Pant <code@siddh.me>
Cc:     david <david@fromorbit.com>, djwong <djwong@kernel.org>,
        fgheet255t <fgheet255t@gmail.com>, hch <hch@infradead.org>,
        linux-ext4 <linux-ext4@vger.kernel.org>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        linux-xfs <linux-xfs@vger.kernel.org>,
        riteshh <riteshh@linux.ibm.com>,
        syzbot+a8e049cd3abd342936b6 
        <syzbot+a8e049cd3abd342936b6@syzkaller.appspotmail.com>,
        syzkaller-bugs <syzkaller-bugs@googlegroups.com>
Subject: Re: [syzbot] WARNING in iomap_iter
Message-ID: <YwOYMJrvBuoVye7R@casper.infradead.org>
References: <182c028abf0.2dc6f7c973088.2963173753499991828@siddh.me>
 <20220821114816.24193-1-code@siddh.me>
 <YwOWiDKhVxm7m0fa@casper.infradead.org>
 <182c607e79a.820e4a7012709.6365464609772129416@siddh.me>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <182c607e79a.820e4a7012709.6365464609772129416@siddh.me>
Precedence: bulk

On Mon, Aug 22, 2022 at 08:19:43PM +0530, Siddh Raman Pant wrote:
> On Mon, 22 Aug 2022 20:15:28 +0530  Matthew Wilcox  wrote:
> > On Sun, Aug 21, 2022 at 05:18:16PM +0530, Siddh Raman Pant wrote:
> > > @@ -979,9 +979,15 @@ loop_set_status_from_info(struct loop_device *lo,
> > >  
> > >       lo->lo_offset = info->lo_offset;
> > >       lo->lo_sizelimit = info->lo_sizelimit;
> > > +     lo->lo_flags = info->lo_flags;
> > > +
> > > +     /* loff_t/int vars are assigned __u64/__u32 vars (respectively) */
> > > +     if (lo->lo_offset < 0 || lo->lo_sizelimit < 0 || lo->lo_flags < 0)
> > > +             return -EOVERFLOW;
> > 
> > Why would you check lo_flags?  That really, really should be an unsigned
> > type.
> 
> I agree, but the loop_device struct has (see line 54 of loop.c):
>         int             lo_flags;
> 
> Thus, I checked for it, as we are not changing any types.

But it's not an integer.  It's a bitfield.  Nobody checks lo_flags for
"is it less than zero".  That makes it very different from lo_offset.