Re: [syzbot] [io-uring?] BUG: unable to handle kernel NULL pointer dereference in io_buffer_select

[Pavel Begunkov <asml.silence@gmail.com>]

On 5/12/25 15:19, Jens Axboe wrote:
> On 5/12/25 8:19 AM, Pavel Begunkov wrote:
>> On 5/12/25 14:56, Jens Axboe wrote:
>> ...>> this line:
>>>>
>>>> tail = smp_load_acquire(&br->tail);
>>>>
>>>> The offset of the tail field is 0xe so bl->buf_ring should be 0. That's
>>>> while it has IOBL_BUF_RING flag set. Same goes for the other report. Also,
>>>> since it's off io_buffer_select(), which looks up the list every time we
>>>> can exclude the req having a dangling pointer.
>>>
>>> It's funky for sure, the other one is regular classic provided buffers.
>>> Interestingly, both reports are for arm32...
>>
>> The other is ring pbuf as well
> 
> True yes, both are pbuf. I can't hit any of this on arm64 or x86-64, fwiw.
> Which is why I thought the arm32 connection might be interesting. Not that
> the arch should matter at all here, but...

Yep, there is a suspicion there might be something on the
mm / allocation side, need to find a good way to narrow it down.
At least we can make syz test patches for us.

-- 
Pavel Begunkov