diff mbox series

[1/1] libgcrypt: disable assembly for i386

Message ID 20230702144829.3356214-1-bernd@kuhls.net
State Superseded
Headers show
Series [1/1] libgcrypt: disable assembly for i386 | expand

Commit Message

Bernd Kuhls July 2, 2023, 2:48 p.m. UTC
Fixes:
http://autobuild.buildroot.net/results/3df/3df96fe36802d8c6ac5487bf3d7d7b6c97591d5e/

Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
---
 package/libgcrypt/libgcrypt.mk | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

Yann E. MORIN July 3, 2023, 8:03 p.m. UTC | #1
Bernd, All,

On 2023-07-02 16:48 +0200, Bernd Kuhls spake thusly:
> Fixes:
> http://autobuild.buildroot.net/results/3df/3df96fe36802d8c6ac5487bf3d7d7b6c97591d5e/
> 
> Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
> ---
>  package/libgcrypt/libgcrypt.mk | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/package/libgcrypt/libgcrypt.mk b/package/libgcrypt/libgcrypt.mk
> index 4fffe5360e..bd7b1aa091 100644
> --- a/package/libgcrypt/libgcrypt.mk
> +++ b/package/libgcrypt/libgcrypt.mk
> @@ -21,8 +21,8 @@ LIBGCRYPT_CONF_OPTS = \
>  	$(if $(BR2_OPTIMIZE_0),--disable-ppc-crypto-support,) \
>  	--with-gpg-error-prefix=$(STAGING_DIR)/usr
>  
> -# Libgcrypt doesn't support assembly for coldfire
> -ifeq ($(BR2_m68k_cf),y)
> +# Libgcrypt doesn't support assembly for coldfire & i386

This is relatively weird, especially, since they do have upstream commit
that try to make it work on "old CPUs":

    a5d126c61cc0  configure: fix avx512 check for i386
    f01d4b7a601f  mpi/longlong.h: i386: use tzcnt instruction for trailing zeros

So maybe this is a bug that should be reported and fixed upstream?

Regards,
Yann E. MORIN.

> +ifeq ($(BR2_i386)$(BR2_m68k_cf),y)
>  LIBGCRYPT_CONF_OPTS += --disable-asm
>  endif
>  
> -- 
> 2.39.2
> 
> _______________________________________________
> buildroot mailing list
> buildroot@buildroot.org
> https://lists.buildroot.org/mailman/listinfo/buildroot
Yann E. MORIN July 3, 2023, 8:06 p.m. UTC | #2
Bernd, All,

On 2023-07-03 22:03 +0200, Yann E. MORIN spake thusly:
> On 2023-07-02 16:48 +0200, Bernd Kuhls spake thusly:
> > Fixes:
> > http://autobuild.buildroot.net/results/3df/3df96fe36802d8c6ac5487bf3d7d7b6c97591d5e/
> > 
> > Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
> > ---
> >  package/libgcrypt/libgcrypt.mk | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/package/libgcrypt/libgcrypt.mk b/package/libgcrypt/libgcrypt.mk
> > index 4fffe5360e..bd7b1aa091 100644
> > --- a/package/libgcrypt/libgcrypt.mk
> > +++ b/package/libgcrypt/libgcrypt.mk
> > @@ -21,8 +21,8 @@ LIBGCRYPT_CONF_OPTS = \
> >  	$(if $(BR2_OPTIMIZE_0),--disable-ppc-crypto-support,) \
> >  	--with-gpg-error-prefix=$(STAGING_DIR)/usr
> >  
> > -# Libgcrypt doesn't support assembly for coldfire
> > -ifeq ($(BR2_m68k_cf),y)
> > +# Libgcrypt doesn't support assembly for coldfire & i386
> 
> This is relatively weird, especially, since they do have upstream commit
> that try to make it work on "old CPUs":
> 
>     a5d126c61cc0  configure: fix avx512 check for i386
>     f01d4b7a601f  mpi/longlong.h: i386: use tzcnt instruction for trailing zeros
> 
> So maybe this is a bug that should be reported and fixed upstream?

Anyway: applied to master, thanks.

Regards,
Yann E. MORIN.
Peter Korsgaard July 17, 2023, 11:50 a.m. UTC | #3
>>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes:

 > Bernd, All,
 > On 2023-07-03 22:03 +0200, Yann E. MORIN spake thusly:
 >> On 2023-07-02 16:48 +0200, Bernd Kuhls spake thusly:
 >> > Fixes:
 >> > http://autobuild.buildroot.net/results/3df/3df96fe36802d8c6ac5487bf3d7d7b6c97591d5e/
 >> > 
 >> > Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
 >> > ---
 >> >  package/libgcrypt/libgcrypt.mk | 4 ++--
 >> >  1 file changed, 2 insertions(+), 2 deletions(-)
 >> > 
 >> > diff --git a/package/libgcrypt/libgcrypt.mk b/package/libgcrypt/libgcrypt.mk
 >> > index 4fffe5360e..bd7b1aa091 100644
 >> > --- a/package/libgcrypt/libgcrypt.mk
 >> > +++ b/package/libgcrypt/libgcrypt.mk
 >> > @@ -21,8 +21,8 @@ LIBGCRYPT_CONF_OPTS = \
 >> >  	$(if $(BR2_OPTIMIZE_0),--disable-ppc-crypto-support,) \
 >> >  	--with-gpg-error-prefix=$(STAGING_DIR)/usr
 >> >  
 >> > -# Libgcrypt doesn't support assembly for coldfire
 >> > -ifeq ($(BR2_m68k_cf),y)
 >> > +# Libgcrypt doesn't support assembly for coldfire & i386
 >> 
 >> This is relatively weird, especially, since they do have upstream commit
 >> that try to make it work on "old CPUs":
 >> 
 >> a5d126c61cc0  configure: fix avx512 check for i386
 >> f01d4b7a601f  mpi/longlong.h: i386: use tzcnt instruction for trailing zeros
 >> 
 >> So maybe this is a bug that should be reported and fixed upstream?

 > Anyway: applied to master, thanks.

Committed to 2023.02.x and 2023.05.x, thanks.
diff mbox series

Patch

diff --git a/package/libgcrypt/libgcrypt.mk b/package/libgcrypt/libgcrypt.mk
index 4fffe5360e..bd7b1aa091 100644
--- a/package/libgcrypt/libgcrypt.mk
+++ b/package/libgcrypt/libgcrypt.mk
@@ -21,8 +21,8 @@  LIBGCRYPT_CONF_OPTS = \
 	$(if $(BR2_OPTIMIZE_0),--disable-ppc-crypto-support,) \
 	--with-gpg-error-prefix=$(STAGING_DIR)/usr
 
-# Libgcrypt doesn't support assembly for coldfire
-ifeq ($(BR2_m68k_cf),y)
+# Libgcrypt doesn't support assembly for coldfire & i386
+ifeq ($(BR2_i386)$(BR2_m68k_cf),y)
 LIBGCRYPT_CONF_OPTS += --disable-asm
 endif