Re: BUG: unable to handle kernel NULL pointer dereference in sidtab_search_core

[Paul Moore]

On Fri, Dec 22, 2017 at 11:59 AM, Dmitry Vyukov <dvyukov@xxxxxxxxxx> wrote:
> On Fri, Dec 22, 2017 at 5:03 PM, Paul Moore <paul@xxxxxxxxxxxxxx> wrote:
>>>>>> wrote:
>>>>>>> Hello,
>>>>>>>
>>>>>>> syzkaller hit the following crash on
>>>>>>> 6084b576dca2e898f5c101baef151f7bfdbb606d
>>>>>>> git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/master
>>>>>>> compiler: gcc (GCC) 7.1.1 20170620
>>>>>>> .config is attached
>>>>>>> Raw console output is attached.
>>>>>>> C reproducer is attached
>>>>>>> syzkaller reproducer is attached. See https://goo.gl/kgGztJ
>>>>>>> for information about syzkaller reproducers
>>>>>>>
>>>>>>>
>>>>>>> SELinux: security_compute_sid:  unrecognized SID 1
>>>>>>> SELinux: security_compute_sid:  unrecognized SID 1
>>>>>>> SELinux: security_compute_sid:  unrecognized SID 1
>>>>>>> SELinux: security_compute_sid:  unrecognized SID 1
>>>>>>> SELinux: security_compute_sid:  unrecognized SID 1
>>>>>>> BUG: unable to handle kernel NULL pointer dereference at 0000000000000001
>>>>>>> IP: sidtab_search_core+0x88/0x110 security/selinux/ss/sidtab.c:100
>>>>>>> PGD 0 P4D 0
>>>>>>> Oops: 0000 [#1] SMP
>>>>>>> Dumping ftrace buffer:
>>>>>>>    (ftrace buffer empty)
>>>>>>> Modules linked in:
>>>>>>> CPU: 1 PID: 4252 Comm: kworker/u4:1 Not tainted 4.15.0-rc3-next-20171214+
>>>>>>> #67
>>>>>>> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
>>>>>>> Google 01/01/2011
>>>>>>> RIP: 0010:sidtab_search_core+0x88/0x110 security/selinux/ss/sidtab.c:100
>>>>>>> RSP: 0018:ffffc900028abc18 EFLAGS: 00010293
>>>>>>> RAX: ffff8802131a87c0 RBX: 0000000000000001 RCX: ffffffff8165d978
>>>>>>> RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffffffff83fd17a0
>>>>>>> RBP: ffffc900028abc40 R08: 0000000000000001 R09: 0000000000000001
>>>>>>> R10: ffffc900028abbe0 R11: 0000000000000000 R12: 0000000000000001
>>>>>>> R13: 0000000000000001 R14: 0000000000000000 R15: ffff880214d93800
>>>>>>> FS:  0000000000000000(0000) GS:ffff88021fd00000(0000) knlGS:0000000000000000
>>>>>>> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>>>>>>> CR2: 0000000000000001 CR3: 0000000214e31000 CR4: 00000000001406e0
>>>>>>> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>>>>>>> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>>>>>>> Call Trace:
>>>>>>>  sidtab_search+0x1f/0x30 security/selinux/ss/sidtab.c:111
>>>>>>>  security_compute_sid.part.11+0xe2/0x710 security/selinux/ss/services.c:1618
>>>>>>>  security_compute_sid+0x92/0xa0 security/selinux/ss/services.c:1598
>>>>>>>  security_transition_sid+0x57/0x70 security/selinux/ss/services.c:1764
>>>>>>>  selinux_bprm_set_creds+0x215/0x2f0 security/selinux/hooks.c:2423
>>>>>>>  security_bprm_set_creds+0x41/0x60 security/security.c:332
>>>>>>>  prepare_binprm+0xae/0x1f0 fs/exec.c:1561
>>>>>>>  do_execveat_common.isra.30+0x6f7/0xb90 fs/exec.c:1784
>>>>>>>  do_execve+0x31/0x40 fs/exec.c:1848
>>>>>>>  call_usermodehelper_exec_async+0x104/0x190 kernel/umh.c:100
>>>>>>>  ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:524
>>>>>>> Code: 8b 5b 50 48 85 db 75 e5 e8 e6 c9 c5 ff 49 8b 5f 18 48 85 db 75 10 eb
>>>>>>> 43 e8 d6 c9 c5 ff 48 8b 5b 50 48 85 db 74 35 e8 c8 c9 c5 ff <44> 8b 23 41 83
>>>>>>> fc 02 76 e4 e8 ba c9 c5 ff 41 83 fc 03 75 1c 48
>>>>>>> RIP: sidtab_search_core+0x88/0x110 security/selinux/ss/sidtab.c:100 RSP:
>>>>>>> ffffc900028abc18
>>>>>>> CR2: 0000000000000001
>>>>>>> ---[ end trace 571c0ea6c6959387 ]---
>>>>>>> Kernel panic - not syncing: Fatal exception
>>>>>>> Dumping ftrace buffer:
>>>>>>>    (ftrace buffer empty)
>>>>>>> Kernel Offset: disabled
>>>>>>> Rebooting in 86400 seconds..
>>>>>>
>>>>>> Based on the reproducer and the stack trace, I'm guessing the system
>>>>>> is attempting to load a kernel module for a a defined, but unloaded,
>>>>>> protocol.  Looking quickly at the SELinux bprm and sidtab code,
>>>>>> nothing obvious is jumping out at me.  Considering the number of false
>>>>>> positives I've been seeing from syzbot lately, I'm assuming this is
>>>>>> more of the same.
>>>>>
>>>>> Hi Paul,
>>>>>
>>>>> What are these false positives? Please elaborate.
>>>>> There is no single false positive that I am aware of. All the ones
>>>>> that were debugged are real kernel bugs.
>>>>
>>>> I've replied to several of the syzbot automated reports with the
>>>> "invalid" response.  I haven't been keeping track, but it seems like
>>>> approximately 50% of the SELinux related reports don't make sense upon
>>>> inspection.
>>>
>>> Can you please point me to some of these bugs? I don't see anything
>>> like this in my inbox, in google group nor in database.
>>
>> Not easily, no.  I don't keep track of these reports once I've
>> responded to the syzbot mail.
>
> There must be traces of this in database and on mailing lists (even if
> you drop syzkaller-bugs@ syzbot will re-add it). So far I did not find
> any traces...

<sigh>

Okay, so I dug through my mail and I found at least two instances (see
below).  There may be more, but I'm getting tired of dealing with
syzbot.  I'm a big fan of increased testing, especially automated
testing, but right now syzbot seems to be costing me a
disproportionate amount of time.

* I sent an invalid reply to
bot+c06fb26dfcd8220259619bf322a79eb0887147c8 on November 3rd, I don't
see an acknowledgement from the robot but I may have deleted it.

* I sent an invalid reply to
bot+fcb30800d359ef1113a6edda069a77c2eebb9bf1 on December 6th, it was
acknowledged by the robot.

-- 
paul moore
www.paul-moore.com