Re: [PATCH liburing] test/io_uring_passthrough: add test for vectored fixed-buffers

[Jens Axboe <axboe@kernel.dk>]

On 5/21/25 8:46 AM, Pavel Begunkov wrote:
> On 5/21/25 15:14, Jens Axboe wrote:
>> On 5/21/25 8:05 AM, Pavel Begunkov wrote:
>>>> About the use of io_uring_prep_read/write*() helpers ? you're right,
>>>> they don?t really add much here since the passthrough command handles
>>>> the fields directly. I?ll work on a cleanup patch to remove those and
>>>> simplify the submission code.
>>>
>>> I don't care about the test itself much, but it means there
>>> are lots of unused fields for the nvme commands that are not
>>> checked by the kernel and hence can't be reused in the future.
>>> That's not great
>>
>> It's still a pretty recent addition, no reason we can't add the checks
>> and get them back to stable as well. At least that would open the door
>> for more easy future expansion.
> 
> nvme passthrough? It has been around for a while
> 
> commit 456cba386e94f22fa1b1426303fdcac9e66b1417
> Author: Kanchan Joshi <joshi.k@samsung.com>
> Date:   Wed May 11 11:17:48 2022 +0530
> 
>     nvme: wire-up uring-cmd support for io-passthru on char-device.
> 
> 
> and if people followed the test for initialising sqes, it'll start
> failing for them.

Huh yes, that's older than I thought, we'd need to got back to 6.1
stable for that. Which isn't a huge deal, but also seems like the
risk would be too high at that point. Unfortunately lots of folks use
odd ball distro kernels that don't diligently pull stable fixes.

-- 
Jens Axboe