package/alsa-lib: force -O2 on i386 with gcc >= 15.x
PATCH 1/1 package/alsa-lib: force -O2 on i386 with gcc >= 15.x
Fixes build error on i386:pcm_dmix_i386.h: In function 'mix_areas_16':
pcm_dmix_i386.h:52:9: error: 'asm' operand has impossible constraints
or there are not enough registers
52 | __asm__ __volatile__ (
Fixes:
https://autobuild.buildroot.net/results/fd9/fd9e6a6f9e2d37df3cfc1ed36707aef9a338f24d/
The oldest build error to be found dates back to Nov 2025:
https://autobuild.buildroot.net/results/3b6/3b606e0c6160563e410535d38a0e7bb09041f223/
so a backport to LTS branches should be considered.
The build error does not occur with gcc 14.4.0 and was reproduced with gcc 16.1.0.
---
Gitlab pipelines passed:
https://gitlab.com/bkuhls/buildroot/-/commits/d022d480a4556f21d86172a593a92288018ed116
package/alsa-lib/alsa-lib.mk | 4 ++++ 1 file changed, 4 insertions(+)
package/alsa-lib/alsa-lib.mk +4 -0
@@ -18,6 +18,10 @@ ALSA_LIB_CONF_OPTS = \ --with-pcm-plugins="$(call qstrip,$(BR2_PACKAGE_ALSA_LIB_PCM_PLUGINS))" \ --with-ctl-plugins="$(call qstrip,$(BR2_PACKAGE_ALSA_LIB_CTL_PLUGINS))" +ifeq ($(BR2_i386)$(BR2_TOOLCHAIN_GCC_AT_LEAST_15),yy) +ALSA_LIB_CFLAGS += -O2 +endif + ifeq ($(BR2_TOOLCHAIN_USES_GLIBC),) ALSA_LIB_CONF_OPTS += --without-versioned endif
Signature · 6 lines
2.47.3 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
Quoted 9 lines · Bernd Kuhls
Fixes build error on i386:
pcm_dmix_i386.h: In function 'mix_areas_16':
pcm_dmix_i386.h:52:9: error: 'asm' operand has impossible constraints
or there are not enough registers
52 |__asm____volatile__(Fixes:
https://autobuild.buildroot.net/results/fd9/fd9e6a6f9e2d37df3cfc1ed36707aef9a338f24d/
Quoted 6 lines
The oldest build error to be found dates back to Nov 2025:
https://autobuild.buildroot.net/results/3b6/3b606e0c6160563e410535d38a0e7bb09041f223/
so a backport to LTS branches should be considered.The build error does not occur with gcc 14.4.0 and was reproduced with gcc 16.1.0.
I have no doubt that passing -O2 works around the issue, but is this the right fix? Especially since it didn't happen with gcc 14.x, it does seem like a regression, no?
Could you make a bug report to GCC? Note that they want the preprocessed source code to be able to easily reproduce.
Thanks a lot!
Thomas
Signature · 7 lines
Thomas Petazzoni, co-owner and CEO, Bootlin Embedded Linux and Kernel engineering and training https://bootlin.com _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot