Re: git: 5e0a4859f28a - main - iwlwifi: Don't compile for gcc before 14

On Mon, 15 Sep 2025, Warner Losh wrote:

> On Mon, Sep 15, 2025, 1:26 PM Bjoern A. Zeeb <bz@freebsd.org> wrote:
>
>> On Mon, 15 Sep 2025, Warner Losh wrote:
>>
>>> The branch main has been updated by imp:
>>>
>>> URL:
>> https://cgit.FreeBSD.org/src/commit/?id=5e0a4859f28ad4869f7a73faf42debc355a370bf
>>>
>>> commit 5e0a4859f28ad4869f7a73faf42debc355a370bf
>>> Author:     Warner Losh <imp@FreeBSD.org>
>>> AuthorDate: 2025-09-14 18:03:16 +0000
>>> Commit:     Warner Losh <imp@FreeBSD.org>
>>> CommitDate: 2025-09-15 03:03:45 +0000
>>>
>>>    iwlwifi: Don't compile for gcc before 14
>>>
>>>    gcc 13 and earlier don't have __builtin_bitcountg. The linux wifi kpi
>>>    uses this unconditionally. While in this one use, it might not be
>>>    needed, I opted to not compile iwlwifi when building gcc12 or 13
>> rather
>>>    than risk breaking it for everbody else.
>>>
>>>    With this change gcc12 builds the kernel. Maybe this will stop jenkins
>>>    email for every commit I make.
>>>
>>>    Sponsored by:           Netflix
>>> ---
>>> sys/modules/Makefile | 3 +++
>>> 1 file changed, 3 insertions(+)
>>>
>>> diff --git a/sys/modules/Makefile b/sys/modules/Makefile
>>> index 5315d518afd8..f9fdbca78869 100644
>>> --- a/sys/modules/Makefile
>>> +++ b/sys/modules/Makefile
>>> @@ -576,7 +576,10 @@ _mlx5ib= mlx5ib
>>>     ${MACHINE_CPUARCH} == "i386"
>>> _ena=         ena
>>> _gve=         gve
>>> +# gcc13 and earlier lack __builtin_bitcountg used by linux emulation
>>
>> (a) I beleive there is no __builtin_bitcountg but you mean
>> __builtin_popcountg
>> both here and in the commit message.
>>
>
> Yes.
>
> (b) That was introduced in 7cbc4d875971860d941cc15d7f42e6cfeffbfe66 for DRM
>>
>> (c) There is no direct use in any LinuxKPI based wireless driver:
>> % grep -r __builtin_popcountg sys/contrib/dev
>> %
>>
>> (4) iwlwifi only uses HWEIGHT32, which was changed by the aforementioned
>> commit
>>      in LinuxKPI.
>>
>
> Yes. I noticed all that. Didn't see a trivial way to fix it right.
>
> (5) Please do it right and in the place where it is actually defined to be
>> used,
>>      in LinuxKPI, and not here as it can be easily fixed there with an
>> #ifdef or
>>      otherwise as there were alternatives on the review if I remember
>> correctly.
>>
>
> Knock yourselves out. I was tired of the CI jobs whining and this was the
> easiest way to make that stop. I've got too many things on my plate to
> refine this more. If you want to support older gcc for this driver, that's
> up to you. The ci jobs have been failing for weeks if not longer

Okay, will do.

But gcc12/13 had been whining for months on a lot of things constantly.
I really don't know how a day to do it right would have made a difference.

-- 
Bjoern A. Zeeb                                                     r15:7