Re: [PATCH for-next 2/2] nvme: optimise io_uring passthrough completion

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

On 5/17/23 14:53, Christoph Hellwig wrote:
> On Wed, May 17, 2023 at 02:30:47PM +0100, Pavel Begunkov wrote:
>> Aside that you decided to ignore the third point, that's a
>> generic interface, not nvme specific, there are patches for
>> net cmds, someone even tried to use it for drm. How do you
>> think new users are supposed to appear if the only helper
>> doing the job can hang the userspace for their use case?
>> Well, then maybe it'll remain nvme/ublk specific with such
>> an approach.
> 
> New users can add new code when it's actualy needed.  We don't
> bloat the kernel for maybe in the future crap as a policy.

Let me put it for you this way, it's an absolutely horrendous
idea to leave the old innocently looking name, i.e.
io_uring_cmd_complete_in_task(), and add there a bunch of
restrictions no new user would care about, that's called
shooting yourself in the leg.

So, we need to rename the function, which, again, for absolutely
no reason adds dependency on ublk. Why doing that instead of
waiting until ublk is converted? That's a big mystery.

-- 
Pavel Begunkov