netdev - bpf_prog_free_deferred bug. WAS: Re: WARNING in mark_chain

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date:   Wed, 10 Jul 2019 16:29:29 -0700
From:   Andrii Nakryiko <andrii.nakryiko@...il.com>
To:     Hillf Danton <hdanton@...a.com>
Cc:     syzbot <syzbot+f21251a7468cd46efc60@...kaller.appspotmail.com>,
        aaron.f.brown@...el.com, Alexei Starovoitov <ast@...nel.org>,
        bpf <bpf@...r.kernel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        "David S. Miller" <davem@...emloft.net>, hawk@...nel.org,
        intel-wired-lan@...ts.osuosl.org,
        Jakub Kicinski <jakub.kicinski@...ronome.com>,
        jeffrey.t.kirsher@...el.com,
        john fastabend <john.fastabend@...il.com>,
        Martin Lau <kafai@...com>,
        open list <linux-kernel@...r.kernel.org>,
        Networking <netdev@...r.kernel.org>, sasha.neftin@...el.com,
        Song Liu <songliubraving@...com>,
        syzkaller-bugs@...glegroups.com, xdp-newbies@...r.kernel.org,
        Yonghong Song <yhs@...com>
Subject: bpf_prog_free_deferred bug. WAS: Re: WARNING in mark_chain_precision

On Tue, Jul 9, 2019 at 9:08 PM Hillf Danton <hdanton@...a.com> wrote:
>
>
> Mon, 08 Jul 2019 21:25:00 -0700 (PDT)
> > Hello,
> >
> > syzbot has tested the proposed patch but the reproducer still triggered
> > crash:
> > WARNING in bpf_jit_free
> >
> > WARNING: CPU: 0 PID: 9077 at kernel/bpf/core.c:851 bpf_jit_free+0x157/0x1b0
> > Kernel panic - not syncing: panic_on_warn set ...
> > CPU: 0 PID: 9077 Comm: kworker/0:3 Not tainted 5.2.0-rc6+ #1
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
> > Google 01/01/2011
> > Workqueue: events bpf_prog_free_deferred
> > Call Trace:
> >   __dump_stack lib/dump_stack.c:77 [inline]
> >   dump_stack+0x172/0x1f0 lib/dump_stack.c:113
> >   panic+0x2cb/0x744 kernel/panic.c:219
> >   __warn.cold+0x20/0x4d kernel/panic.c:576
> >   report_bug+0x263/0x2b0 lib/bug.c:186
> >   fixup_bug arch/x86/kernel/traps.c:179 [inline]
> >   fixup_bug arch/x86/kernel/traps.c:174 [inline]
> >   do_error_trap+0x11b/0x200 arch/x86/kernel/traps.c:272
> >   do_invalid_op+0x37/0x50 arch/x86/kernel/traps.c:291
> >   invalid_op+0x14/0x20 arch/x86/entry/entry_64.S:986
> > RIP: 0010:bpf_jit_free+0x157/0x1b0
> > Code: 00 fc ff df 48 89 fa 48 c1 ea 03 80 3c 02 00 75 5d 48 b8 00 02 00 00
> > 00 00 ad de 48 39 43 70 0f 84 05 ff ff ff e8 09 7f f4 ff <0f> 0b e9 f9 fe
> > ff ff e8 2d 02 2e 00 e9 d9 fe ff ff 48 89 7d e0 e8
> > RSP: 0018:ffff888084affcb0 EFLAGS: 00010293
> > RAX: ffff88808a622100 RBX: ffff88809639d580 RCX: ffffffff817b0b0d
> > RDX: 0000000000000000 RSI: ffffffff817c4557 RDI: ffff88809639d5f0
> > RBP: ffff888084affcd0 R08: 1ffffffff150daa8 R09: fffffbfff150daa9
> > R10: fffffbfff150daa8 R11: ffffffff8a86d547 R12: ffffc90001921000
> > R13: ffff88809639d5e8 R14: ffff8880a0589800 R15: ffff8880ae834d40
> >   bpf_prog_free_deferred+0x27a/0x350 kernel/bpf/core.c:1982
> >   process_one_work+0x989/0x1790 kernel/workqueue.c:2269
> >   worker_thread+0x98/0xe40 kernel/workqueue.c:2415
> >   kthread+0x354/0x420 kernel/kthread.c:255
> >   ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352
> > Kernel Offset: disabled
> > Rebooting in 86400 seconds..
> >
> >
> > Tested on:
> >
> > commit:         b9321614 bpf: fix precision bit propagation for BPF_ST ins..
> > git tree:       https://github.com/anakryiko/linux bpf-fix-precise-bpf_st
> > console output: https://syzkaller.appspot.com/x/log.txt?x=112f0dfda00000
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=6bb3e6e7997c14f9
> > compiler:       gcc (GCC) 9.0.0 20181231 (experimental)
> >
>
> 1, currently, bpf_prog_put(), the put helper, deletes all kallsyms before
> invoking the free helper, bpf_prog_free(); the latter complains if kallsyms
> are detected not all off.
>
> 2, before commit 538950a1b752 ("soreuseport: setsockopt
> SO_ATTACH_REUSEPORT_[CE]BPF"), __bpf_prog_release() already called the put
> helper or the free one for a given prog depending on its type: put for
> BPF_PROG_TYPE_SOCKET_FILTER.
>
> In the commit we can see bpf_prog_destroy(), __bpf_prog_free(),
> bpf_prog_put(), here and then; Note in __get_bpf() the put for
> !BPF_PROG_TYPE_SOCKET_FILTER.
>
> 3, in commit 113214be7f6c ("bpf: refactor bpf_prog_get and type check into
> helper") bpf_prog_get_type() was added and put in __get_bpf().
>
> In the commit we can see other types like BPF_PROG_TYPE_SCHED_ACT and
> BPF_PROG_TYPE_SCHED_CLS.
>
> 4, in commit 8217ca653ec6 ("bpf: Enable BPF_PROG_TYPE_SK_REUSEPORT bpf prog
> in reuseport selection") sk_reuseport_prog_free() was added without a word
> in the log for it.
>
> 5, enrich prog type in __bpf_prog_release().

Thanks for investigation!

I'm curious what makes BPF_PROG_TYPE_SOCKET_FILTER (and
BPF_PROG_TYPE_SK_REUSEPORT) special, compared to all the other types
of BPF programs. If it's something to do with sockets specifically,
there are a bunch of other programs that deal with sockets, so should
they be handled specially as well (e.g., BPF_PROG_TYPE_CGROUP_SOCK)?

Daniel, do you have any insight here?

>
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -1142,11 +1142,15 @@ static void bpf_release_orig_filter(stru
>
>  static void __bpf_prog_release(struct bpf_prog *prog)
>  {
> -       if (prog->type == BPF_PROG_TYPE_SOCKET_FILTER) {
> +       switch (prog->type) {
> +       case BPF_PROG_TYPE_SOCKET_FILTER:
> +       case BPF_PROG_TYPE_SK_REUSEPORT:
>                 bpf_prog_put(prog);
> -       } else {
> +               break;
> +       default:
>                 bpf_release_orig_filter(prog);
>                 bpf_prog_free(prog);
> +               break;
>         }
>  }
>
> --
>