Re: [PATCH] io_uring: add missing IORING_MAP_OFF_ZCRX_REGION in io_uring_mmap
From: Jens Axboe
Date: Fri Feb 21 2025 - 12:29:16 EST
On 2/21/25 10:06 AM, Pavel Begunkov wrote:
> On 2/21/25 16:39, Jens Axboe wrote:
>> On 2/21/25 5:20 AM, Pavel Begunkov wrote:
>>> On 2/21/25 09:10, lizetao wrote:
>>>> Hi,
>>>>
>>>>> -----Original Message-----
>>>>> From: Bui Quang Minh <minhquangbui99@xxxxxxxxx>
>>>>> Sent: Friday, February 21, 2025 5:00 PM
>>>>> To: io-uring@xxxxxxxxxxxxxxx
>>>>> Cc: Jens Axboe <axboe@xxxxxxxxx>; Pavel Begunkov
>>>>> <asml.silence@xxxxxxxxx>; David Wei <dw@xxxxxxxxxxx>; linux-
>>>>> kernel@xxxxxxxxxxxxxxx; Bui Quang Minh <minhquangbui99@xxxxxxxxx>
>>>>> Subject: [PATCH] io_uring: add missing IORING_MAP_OFF_ZCRX_REGION in
>>>>> io_uring_mmap
>>>>>
>>>>> Allow user to mmap the kernel allocated zerocopy-rx refill queue.
>>>>>
>>>>
>>>> Maybe fixed-tag should be added here.
>>>
>>> No need, it's not strictly a fix, and whlist it's not yet sent to
>>> linus, the tags only cause confusion when hashes change, e.g. on rebase.
>>
>> I do like using fixes still, if only to provide a link to the original
>> commit without needing to dig for it. And yeah there's the occasional
>> rebase where I forget to update the sha, but those get discovered pretty
>> quick.
>
> Maybe a "Link" tag would be better or some more inconsequential
> "Refers-to", but otherwise you can call it a feature and avoid the
> hassle of fixing it up, and people getting spammed by tooling,
> and Stephen having to write about broken hashes.
Having the sha is nice though so people can just look it up. But yeah
the tag doesn't really exist, and I think we have way too many tags
already, which is why I just use Fixes for stuff like that too. Link
is good for email discussion, though 99% of the time it ends up just
being a patch posting and not really interesting...
--
Jens Axboe