Re: [PATCH] io_uring: Add NULL checks for current->io_uring

[Jens Axboe <axboe@kernel.dk>]

On 1/11/23 7:10 PM, Jia-Ju Bai wrote:
> 
> 
> On 2023/1/11 22:49, Jens Axboe wrote:
>> On 1/11/23 3:19 AM, Jia-Ju Bai wrote:
>>> As described in a previous commit 998b30c3948e, current->io_uring could
>>> be NULL, and thus a NULL check is required for this variable.
>>>
>>> In the same way, other functions that access current->io_uring also
>>> require NULL checks of this variable.
>> This seems odd. Have you actually seen traces of this, or is it just
>> based on "guess it can be NULL sometimes, check it in all spots"?
>>
> 
> Thanks for the reply!
> I checked the previous commit and inferred that there may be some problems.
> I am not quite sure of this, and thus want to listen to your opinions :)

I'd invite you to look over each of them separately, and see if that path
could potentially lead to current->io_uring == NULL and thus being an
issue. I think that'd be a useful exercise, and you never know that you
might find :-)

-- 
Jens Axboe