Received-SPF: pass (google.com: domain of linux-kernel+bounces-198510-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Date: Mon, 3 Jun 2024 10:51:51 +1000
From: Dave Chinner <david@fromorbit.com>
To: syzbot <syzbot+981b8efffb3d71c46bef@syzkaller.appspotmail.com>
Cc: akpm@linux-foundation.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, muchun.song@linux.dev, roman.gushchin@linux.dev,
	syzkaller-bugs@googlegroups.com, zhengqi.arch@bytedance.com
Subject: Re: [syzbot] [mm?] UBSAN: shift-out-of-bounds in do_shrink_slab
Message-ID: <Zl0Tp/1PYaIAgJJz@dread.disaster.area>
References: <0000000000003051770619cec4f8@google.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <0000000000003051770619cec4f8@google.com>

On Sat, Jun 01, 2024 at 12:08:25AM -0700, syzbot wrote:
> Hello,
> 
> syzbot found the following issue on:
> 
> HEAD commit:    6dc544b66971 Add linux-next specific files for 20240528
> git tree:       linux-next
> console output: https://syzkaller.appspot.com/x/log.txt?x=14c7f806980000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=6a363b35598e573d
> dashboard link: https://syzkaller.appspot.com/bug?extid=981b8efffb3d71c46bef
> compiler:       Debian clang version 15.0.6, GNU ld (GNU Binutils for Debian) 2.40
> 
> Unfortunately, I don't have any reproducer for this issue yet.
> 
> Downloadable assets:
> disk image: https://storage.googleapis.com/syzbot-assets/334699ab67f8/disk-6dc544b6.raw.xz
> vmlinux: https://storage.googleapis.com/syzbot-assets/4ca32b2218ce/vmlinux-6dc544b6.xz
> kernel image: https://storage.googleapis.com/syzbot-assets/400bc5f019b3/bzImage-6dc544b6.xz
> 
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+981b8efffb3d71c46bef@syzkaller.appspotmail.com
> 
> ------------[ cut here ]------------
> UBSAN: shift-out-of-bounds in mm/shrinker.c:406:18
> shift exponent -1 is negative
> CPU: 0 PID: 5278 Comm: syz-executor.1 Not tainted 6.10.0-rc1-next-20240528-syzkaller #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 04/02/2024
> Call Trace:
>  <TASK>
>  __dump_stack lib/dump_stack.c:88 [inline]
>  dump_stack_lvl+0x241/0x360 lib/dump_stack.c:114
>  ubsan_epilogue lib/ubsan.c:231 [inline]
>  __ubsan_handle_shift_out_of_bounds+0x3c8/0x420 lib/ubsan.c:468
>  do_shrink_slab+0xe26/0x1160 mm/shrinker.c:406

	total_scan = nr >> priority;

Ok, that means the shrinker has been passed a priority of -1 from
the core memory reclaim code. That means it is more likely that
something has gone wrong with the higher level struct scan_control
sc->priority handling, not something in teh shrinker code itself.

>  shrink_slab_memcg mm/shrinker.c:548 [inline]
>  shrink_slab+0x87c/0x14d0 mm/shrinker.c:626
>  shrink_node_memcgs mm/vmscan.c:5923 [inline]
>  shrink_node+0xb82/0x4150 mm/vmscan.c:5961
>  shrink_zones mm/vmscan.c:6205 [inline]
>  do_try_to_free_pages+0x789/0x1cb0 mm/vmscan.c:6267

This has a loop that does:

	do {
	.....
		shrink_zones(zonelist, sc);
	.....
	} while (--sc->priority >= 0);

and all the callers initialise sc->priority to DEF_PRIORITY.  Hence
I can't see how we get shrink_zones() gets called with sc->priority
== -1 from here or anywhere else that decrements sc->priority. This
needs someone with more core mm reclaim expertise than I have to
triage this further.

-Dave.
-- 
Dave Chinner
david@fromorbit.com