Re: KASAN: slab-out-of-bounds Read in vhci_hub_control

From: Shuah Khan
Date: Tue Oct 02 2018 - 19:21:23 EST


On 10/02/2018 10:42 AM, Dmitry Vyukov wrote:
> On Tue, Oct 2, 2018 at 6:04 PM, Shuah Khan <shuah@xxxxxxxxxx> wrote:
>> On 09/04/2018 12:52 PM, syzbot wrote:
>>> Hello,
>>>
>>> syzbot found the following crash on:
>>>
>>> HEAD commit: 420f51f4ab6b Merge tag 'arm64-fixes' of git://git.kernel.o..
>>> git tree: upstream
>>> console output: https://syzkaller.appspot.com/x/log.txt?x=126a6f0e400000
>>> kernel config: https://syzkaller.appspot.com/x/.config?x=531a917630d2a492
>>> dashboard link: https://syzkaller.appspot.com/bug?extid=bccc1fe10b70fadc78d0
>>> compiler: gcc (GCC) 8.0.1 20180413 (experimental)
>>> syz repro: https://syzkaller.appspot.com/x/repro.syz?x=121caa46400000
>>> C reproducer: https://syzkaller.appspot.com/x/repro.c?x=14ed8ab6400000
>>
>> C producer doesn't reproduce the problem on 4.19-rc5. Does this C producer
>> depend on state of the machine? i.e what is the status of vhci_hcd - are
>> there any devices attached?
>
> Hi Shuah,
>
> syzbot always runs tests reproducers on a clean machine. There is some
> state are running a Debian wheezy init, but no test/fuzz/stress
> workload is run before the reproducer.
> syzbot also uses VMs, so there are no real devices attached. And it's
> GCE VMs (not qemu), and I think GCE does not even emulate any USB
> devices.
>
> An obvious thing to try would be to use the exact commit and config
> syzbot gave (rather than 4.19-rc5).
> You can also take the image syzbot uses here:
> https://github.com/google/syzkaller/blob/master/docs/syzbot.md#crash-does-not-reproduce
>
>
>> I can see the problem looking at the code and fix is easy. However, I would
>> like be able to reproduce it and verify the fix works. Also this would be a
>> good regression for the driver I could consider adding to selftests.
>
> syzbot can test fixes for bugs with reproducers:
> https://github.com/google/syzkaller/blob/master/docs/syzbot.md#testing-patches
> So it can test your fix. But this obviously won't help with a test.
>

Tried the same config and no luck. Any chance you have the complete dmesg?

thanks,
-- Shuah