Re: [syzbot] [bpf?] BUG: unable to handle kernel paging request in bpf_prog_ADDR (2)

[Stanislav Fomichev]

On Mon, Apr 22, 2024 at 3:37 AM Björn Töpel <bjorn@kernelorg> wrote:
>
> On Sat, 20 Apr 2024 at 00:42, Stanislav Fomichev <sdf@xxxxxxxxxx> wrote:
> >
> > On 04/19, Alexei Starovoitov wrote:
> > > On Mon, Apr 8, 2024 at 8:53 PM syzbot
> > > <syzbot+838346b979830606c854@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> > > >
> > > > Hello,
> > > >
> > > > syzbot found the following issue on:
> > > >
> > > > HEAD commit:    fe46a7dd189e Merge tag 'sound-6.9-rc1' of git://gitkernel..
> > > > git tree:       upstream
> > > > console+strace: https://syzkaller.appspot.com/x/log.txt?x=12596223180000
> > > > kernel config:  https://syzkaller.appspot.com/x/.config?x=4d90a36f0cab495a
> > > > dashboard link: https://syzkaller.appspot.com/bug?extid=838346b979830606c854
> > > > compiler:       Debian clang version 15.0.6, GNU ld (GNU Binutils for Debian) 2.40
> > > > syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=134ecbb5180000
> > > > C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=141a8b3d180000
> > > >
> > > > Downloadable assets:
> > > > disk image: https://storage.googleapis.com/syzbot-assets/f6c04726a2ae/disk-fe46a7dd.raw.xz
> > > > vmlinux: https://storage.googleapis.com/syzbot-assets/09c26ce901ea/vmlinux-fe46a7dd.xz
> > > > kernel image: https://storage.googleapis.com/syzbot-assets/134acf7f5322/bzImage-fe46a7dd.xz
> > > >
> > > > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > > > Reported-by: syzbot+838346b979830606c854@xxxxxxxxxxxxxxxxxxxxxxxxx
> > > >
> > > > BUG: unable to handle page fault for address: 0000001000000112
> > > > #PF: supervisor read access in kernel mode
> > > > #PF: error_code(0x0000) - not-present page
> > > > PGD 800000002e7b1067 P4D 800000002e7b1067 PUD 0
> > > > Oops: 0000 [#1] PREEMPT SMP KASAN PTI
> > > > CPU: 0 PID: 5060 Comm: syz-executor351 Not tainted 6.8.0-syzkaller-08951-gfe46a7dd189e #0
> > > > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/27/2024
> > > > RIP: 0010:bpf_prog_a8e24a805b35c61b+0x19/0x1e
> > > > Code: cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc cc f3 0f 1e fa 0f 1f 44 00 00 66 90 55 48 89 e5 f3 0f 1e fa 31 c0 48 8b 7f 18 <8b> 7f 00 c9 c3 cc cc cc cc cc cc 40 03 00 00 cc cc cc cc cc cc cc
> > > > RSP: 0018:ffffc90003b07b30 EFLAGS: 00010246
> > > > RAX: 0000000000000000 RBX: ffffc90000ace048 RCX: ffff88802aa89e00
> > > > RDX: 0000000000000000 RSI: ffffc90000ace048 RDI: 0000001000000112
> > > > RBP: ffffc90003b07b30 R08: ffffffff81bf633c R09: 1ffffffff2595ca0
> > > > R10: dffffc0000000000 R11: ffffffffa000095c R12: ffffc90000ace030
> > > > R13: ffff88802ac3ae28 R14: dffffc0000000000 R15: ffff88802ac3ae28
> > > > FS:  000055558f759380(0000) GS:ffff8880b9400000(0000) knlGS:0000000000000000
> > > > CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > > > CR2: 0000001000000112 CR3: 0000000077cfa000 CR4: 00000000003506f0
> > > > DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> > > > DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> > > > Call Trace:
> > > >  <TASK>
> > > >  bpf_dispatcher_nop_func include/linux/bpf.h:1234 [inline]
> > > >  __bpf_prog_run include/linux/filter.h:657 [inline]
> > > >  bpf_prog_run include/linux/filter.h:664 [inline]
> > > >  bpf_prog_run_array_cg kernel/bpf/cgroup.c:51 [inline]
> > > >  __cgroup_bpf_run_filter_setsockopt+0x6fa/0x1040 kernel/bpf/cgroup.c:1830
> > > >  do_sock_setsockopt+0x6b4/0x720 net/socket.c:2293
> > > >  __sys_setsockopt+0x1ae/0x250 net/socket.c:2334
> > > >  __do_sys_setsockopt net/socket.c:2343 [inline]
> > > >  __se_sys_setsockopt net/socket.c:2340 [inline]
> > > >  __x64_sys_setsockopt+0xb5/0xd0 net/socket.c:2340
> > > >  do_syscall_64+0xfb/0x240
> > > >  entry_SYSCALL_64_after_hwframe+0x6d/0x75
> > >
> > > This one looks interesting.
> > > But I cannot reproduce it.
> > >
> > > Bjorn or Stan,
> > >
> > > Could you take a look?
> > >
> > > Probably a race in xdp dispatcher setup or the way cgroup-lsm
> > > logic is doing it.
> >
> > Managed to repro it by hacking the C reproducer to attach bpf prog
> > to /sys/fs/cgroup instead of syzkallers custom path. Will try to
> > poke it a bit more..
>
> Stan, did you get anywhere? Please share your hack, where you manage
> to reproduce the issue.

Yes, I think the problem is more naive. The syszbot reproducer manages to attach
cgroup_skb program to a cgroup_sockopt hook :-/. I'll try to send a
patch this week
to fix it (need to write a proper selftest as well).

> Cheers,
> Björn