Re: [PATCH] ressources/coreboot/fam15h*: fix building crossgcc 8.3.0 (fixes #64870)

[Adrien 'neox' Bourmault]

Hi and thanks for your answer.

Le dimanche 08 septembre 2024 à 23:22 +0200, Denis 'GNUtoo' Carikli a
écrit :
> - Where does the patch comes from exactly? Can we have a link to it?
> I
>   guess that given that there is a To: they probably comes from the
> GCC
>   mailing list, but I also found it in GCC's git repository (commit
>   331763de7d4850702a0f67298f36017c73cdb103 ("Fix build breakage with
>   latest glibc release")). Also something is strange with that
> commit:
>   $ git describe 331763de7d4850702a0f67298f36017c73cdb103
>   misc/first-auto-changelog-6901-g331763de7d4

This patch has indeed been found in the GCC mailing list.
Here is the mailing list link:
https://gcc.gnu.org/pipermail/gcc-bugs/2021-March/731278.html

Here is the actual commit in the GCC repository:
https://gcc.gnu.org/git/gitweb.cgi?p=gcc.git;h=a5a7cdcaa0c29ee547c41d24f495e9694a6fe7f1

> - What's the status of upstreaming this patch? If I've a given GCC
>   version, can I know if it needs this patch? Or where to check for
> the
>   patch status if possible?

This patch was an answer to a bug report from the GCC development
tracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99264

The changes of the GLIBC that removed the MINSKTSZ constant was
introduced only for systems using the Linux kernel, and while the
changelog is recommanding using sysconf to get the value of
`_SC_MINSTKSZ`. The problem is that it does not allow to get the value
in the preprocessor context.

It has been accepted by GCC and the bug report has been set as RESOLVED
FIXED, meaning it solved the bug.

This patch is needed for all GCC versions from 8 to 9, since this
commit solved the bug for 9, 10 and later versions. It has been adopted
by OpenSUSE for its GCC 8 package:
https://build.opensuse.org/projects/devel:gcc/packages/gcc8/files/gcc8-ada-MINSTKSZ.patch

> 
> The information I'm looking for are typically found in things like
> the
> standard patch header from Debian[1], an example can be found in one
> of
> the GNU Boot patches[2] where the context wasn't enough to have all
> the
> information needed on the patch. 
> 
> This standard header was also reused by another project
> (Yocto/Openembedded) so it's broader than just Debian, but I'm unsure
> how many non-debian distros uses it. 
> 
> So we could also consider requiring this header (in certain
> conditions
> or all the time) for new patches that we bring in GNU Boot later on.

Why not, I'm not against that since it helps clarity and helps to track
things historically.
> 
> Obviously we'd need to modify the documentation/website to make it
> official if we choose to go this way.

Yes, indeed.

Happy hacking!

-- 
Adrien Bourmault
Maintainer, GNU Boot project
Associate member, Free Software Foundation
GPG : E23C26A5DEEEC5FA9CDDD57A57BC26A3687116F6

signature.asc
Description: This is a digitally signed message part