Re: general protection fault in rds_ib_get_mr

[Santosh Shilimkar <santosh.shilimkar@oracle.com>]

On 7/5/2018 12:06 AM, Johannes Thumshirn wrote:
> On Wed, Jul 04, 2018 at 05:32:20PM -0700, santosh.shilimkar@oracle.com wrote:
>> On 7/4/18 2:35 PM, Eric Biggers wrote:
>>> On Sun, May 13, 2018 at 10:46:37PM -0700, santosh.shilimkar@oracle.com wrote:
>>>> On 5/13/18 2:10 PM, Eric Biggers wrote:
>>>>> On Wed, Mar 21, 2018 at 09:00:01AM -0700, syzbot wrote:
>>>>
>>>> [...]
>>>>
>>>>
>>>>> Still reproducible on Linus' tree (commit 66e1c94db3cd4) and linux-next
>>>>> (next-20180511).  Here's a simplified reproducer:
>>>>>
>>>> Thanks for the test case !!
>>>>
>>
>> [...]
>>
>>>
>>> Ping; this RDS bug is still unfixed.  The same reproducer I gave earlier still
>>> works, even when run as a non-root user.  I tested upstream commit
>>> fc36def997cfd6 (v4.18-rc3-113-gfc36def997cfd), and linux-next commit
>>> bce40927669338 (next-20180704).
>>>
>> Yeah I haven't sent a fix for this yet. Downstream we have refactored
>> MR code and I wanted to get that upstream. With that this bug will
>> also get addressed. I was hoping to get that series but since some
>> optimisation is still pending investigations its getting delayed.
>>
>> Will find out if I can fix this panic in meantime with smaller change.
>> Thanks for reminder.
> 
> Oh please sent bug fixes _before_ refactoring in that area. Otherwise
> it's a pain for everybody backporting the fix to a stable kernel.
> 
OK. we will look into it if an interim fix can be made....