Date:   Sun, 20 Nov 2022 08:04:13 -0800
From:   Eric Dumazet <edumazet@...gle.com>
To:     Hillf Danton <hdanton@...a.com>
Cc:     Shigeru Yoshida <syoshida@...hat.com>, kuba@...nel.org,
        pabeni@...hat.com, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org, syzkaller-bugs@...glegroups.com,
        syzbot+106f9b687cd64ee70cd1@...kaller.appspotmail.com
Subject: Re: [PATCH v2] net: tun: Fix use-after-free in tun_detach()

On Sun, Nov 20, 2022 at 2:49 AM Hillf Danton <hdanton@...a.com> wrote:
>
> On 20 Nov 2022 18:02:13 +0900 Shigeru Yoshida <syoshida@...hat.com>
> > syzbot reported use-after-free in tun_detach() [1].  This causes call
> > trace like below:
> >
> > ==================================================================
> > BUG: KASAN: use-after-free in notifier_call_chain+0x1ee/0x200 kernel/notifier.c:75
> > Read of size 8 at addr ffff88807324e2a8 by task syz-executor.0/3673
> >
> > CPU: 0 PID: 3673 Comm: syz-executor.0 Not tainted 6.1.0-rc5-syzkaller-00044-gcc675d22e422 #0
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022
> > Call Trace:
> >  <TASK>
> >  __dump_stack lib/dump_stack.c:88 [inline]
> >  dump_stack_lvl+0xd1/0x138 lib/dump_stack.c:106
> >  print_address_description mm/kasan/report.c:284 [inline]
> >  print_report+0x15e/0x461 mm/kasan/report.c:395
> >  kasan_report+0xbf/0x1f0 mm/kasan/report.c:495
> >  notifier_call_chain+0x1ee/0x200 kernel/notifier.c:75
> >  call_netdevice_notifiers_info+0x86/0x130 net/core/dev.c:1942
> >  call_netdevice_notifiers_extack net/core/dev.c:1983 [inline]
> >  call_netdevice_notifiers net/core/dev.c:1997 [inline]
> >  netdev_wait_allrefs_any net/core/dev.c:10237 [inline]
> >  netdev_run_todo+0xbc6/0x1100 net/core/dev.c:10351
> >  tun_detach drivers/net/tun.c:704 [inline]
> >  tun_chr_close+0xe4/0x190 drivers/net/tun.c:3467
> >  __fput+0x27c/0xa90 fs/file_table.c:320
> >  task_work_run+0x16f/0x270 kernel/task_work.c:179
> >  exit_task_work include/linux/task_work.h:38 [inline]
> >  do_exit+0xb3d/0x2a30 kernel/exit.c:820
> >  do_group_exit+0xd4/0x2a0 kernel/exit.c:950
> >  get_signal+0x21b1/0x2440 kernel/signal.c:2858
> >  arch_do_signal_or_restart+0x86/0x2300 arch/x86/kernel/signal.c:869
> >  exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
> >  exit_to_user_mode_prepare+0x15f/0x250 kernel/entry/common.c:203
> >  __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
> >  syscall_exit_to_user_mode+0x1d/0x50 kernel/entry/common.c:296
> >  do_syscall_64+0x46/0xb0 arch/x86/entry/common.c:86
> >  entry_SYSCALL_64_after_hwframe+0x63/0xcd
> >
> > The cause of the issue is that sock_put() from __tun_detach() drops
> > last reference count for struct net, and then notifier_call_chain()
> > from netdev_state_change() accesses that struct net.
>
> Correct. IOW the race between netdev_run_todo() and cleanup_net() is behind
> the uaf report from syzbot.
>
> >
> > This patch fixes the issue by calling sock_put() from tun_detach()
> > after all necessary accesses for the struct net has done.
>
> Thanks for your fix.
>
> But tun is not special wrt netdev_run_todo() and call_netdevice_notifiers(),
> so the correct fix should be making netdev grab another hold on net and
> invoking put_net() in the path of netdev_run_todo().

Well, this is not going to work. Unless I am missing something.

1) A netns is destroyed when its refcount reaches zero.

   if (refcount_dec_and_test(&net->ns.count))
        __put_net(net);

    This transition is final.

    (It is illegal to attempt a refcount_inc() from this state)

2) All its netdevices are unregistered.

   Trying to get a reference on netns in netdevice dismantle path
would immediately trigger a refcount_t warning.

Powered by blists - more mailing lists