Re: git: a5a3daf88af5 - main - fixup! libmlx5: Extend mlx5dv_get_qp() to return UAR mapping info

On 9/16/26 23:07, Konstantin Belousov wrote:
> On Wed, Sep 16, 2026 at 02:12:39PM +0000, Konstantin Belousov wrote:
>> The branch main has been updated by kib:
>>
>> URL: https://cgit.FreeBSD.org/src/commit/?id=a5a3daf88af5e4d30952a7fc8b4d27d83458f484
>>
>> commit a5a3daf88af5e4d30952a7fc8b4d27d83458f484
>> Author:     Ariel Ehrenberg <aehrenberg@nvidia.com>
>> AuthorDate: 2026-07-28 12:02:36 +0000
>> Commit:     Konstantin Belousov <kib@FreeBSD.org>
>> CommitDate: 2026-09-16 14:12:06 +0000
>>
>>      fixup! libmlx5: Extend mlx5dv_get_qp() to return UAR mapping info
> To answer some questions I saw on private comm channel:
> 
> The form of commits was intended, I initially have a plan to squash fixes
> into the corresponding commits.  But after some thought, I decided to
> keep the original commits intact, and push fixes as separate changes.
> This preserves the authorship and does not attribute changes not made
> by the initial author, potentially putting him off the hook.
> 
> Might be I should have edited the 'fixup!' rebase markers.  Note that
> using the 'Fixes:' tags is impossible because I cannot know the final
> hashes in  advance.
> 
> This patch series as a whole was validated by Nvidia verification for
> ConnectX RDMA.

FWIW, when I recently pushed a set of changes to iw_cxgbe I used
Co-authored-by: tags to acknowledge changes made by other authors
while keeping the Author field as the "main" author, including when
cherry-picking changes from the Linux driver rather than using
separate fixup commits.  I think the specific phrase fixup! is also
best if reserved for commits you intend to squash (e.g those created
with git commit --fixup for use with a later git rebase --autosquash).
If you want to keep them as separate commits it would be best to
reword the title to avoid confusion.

-- 
John Baldwin