Re: [OE-core] [PATCH] site/x86_64-linux: add ac_cv_func_fnmatch_works

2024-06-02 Thread Dmitry Baryshkov
On Thu, 30 May 2024 at 20:41, Khem Raj  wrote:
>
> On Thu, May 30, 2024 at 9:44 AM Dmitry Baryshkov via
> lists.openembedded.org 
> wrote:
> >
> > Other arch-specific caches have ac_cv_func_fnmatch_works. Add it to
> > x86_64 cache too, otherwise cabextract fails to build ([1])
> >
> > [1] 
> > https://autobuilder.yoctoproject.org/typhoon/#/builders/88/builds/3893/steps/14/logs/stdio
> >
> > Signed-off-by: Dmitry Baryshkov 
> > ---
> >  meta/site/x86_64-linux | 3 +++
> >  1 file changed, 3 insertions(+)
> >
> > diff --git a/meta/site/x86_64-linux b/meta/site/x86_64-linux
> > index b9ee141fd5c5..ed9c88fa20f7 100644
> > --- a/meta/site/x86_64-linux
> > +++ b/meta/site/x86_64-linux
> > @@ -83,3 +83,6 @@ ac_cv_func__restgpr_14_x=no
> >
> >  # cvs
> >  cvs_cv_func_printf_ptr=${cvs_cv_func_printf_ptr=yes}
> > +
> > +# cabextract
> > +ac_cv_func_fnmatch_works=${ac_cv_func_fnmatch_works=yes}
>
> I think it will be good to document the underlying reason for this test to
> fail during the autoconf run. It seems to trying fnmatch replacement,
> but then fails to do
> it properly.

Well, there are two issues there:
- testing whether the function is correct isn't going to work for
cross-compilation (including same-arch cross-compile). For this reason
we have ac_cv_func_fnmatch_works in several other site caches.

- cabextract's rpl_fnmatch being broken on glibc-based systems (the
file gets enabled, but then if GLIBC is detected the replacement
function is skipped).

While it might be worth patching the cabextract's fnmatch.c (I'll look
into it), a proper fix is to provide a cached value for the test,
which this patch does.

-- 
With best wishes
Dmitry

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#200201): 
https://lists.openembedded.org/g/openembedded-core/message/200201
Mute This Topic: https://lists.openembedded.org/mt/106392140/21656
Group Owner: openembedded-core+ow...@lists.openembedded.org
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-



Re: [OE-core] [PATCH] site/x86_64-linux: add ac_cv_func_fnmatch_works

2024-05-30 Thread Khem Raj
On Thu, May 30, 2024 at 9:44 AM Dmitry Baryshkov via
lists.openembedded.org 
wrote:
>
> Other arch-specific caches have ac_cv_func_fnmatch_works. Add it to
> x86_64 cache too, otherwise cabextract fails to build ([1])
>
> [1] 
> https://autobuilder.yoctoproject.org/typhoon/#/builders/88/builds/3893/steps/14/logs/stdio
>
> Signed-off-by: Dmitry Baryshkov 
> ---
>  meta/site/x86_64-linux | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/meta/site/x86_64-linux b/meta/site/x86_64-linux
> index b9ee141fd5c5..ed9c88fa20f7 100644
> --- a/meta/site/x86_64-linux
> +++ b/meta/site/x86_64-linux
> @@ -83,3 +83,6 @@ ac_cv_func__restgpr_14_x=no
>
>  # cvs
>  cvs_cv_func_printf_ptr=${cvs_cv_func_printf_ptr=yes}
> +
> +# cabextract
> +ac_cv_func_fnmatch_works=${ac_cv_func_fnmatch_works=yes}

I think it will be good to document the underlying reason for this test to
fail during the autoconf run. It seems to trying fnmatch replacement,
but then fails to do
it properly.

> --
> 2.39.2
>
>
> 
>

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#200037): 
https://lists.openembedded.org/g/openembedded-core/message/200037
Mute This Topic: https://lists.openembedded.org/mt/106392140/21656
Group Owner: openembedded-core+ow...@lists.openembedded.org
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-