bug#27489: glibc fails to build on i686

[Mark H Weaver]

Mark H Weaver <address@hidden> writes:

> Ricardo Wurmus <address@hidden> writes:
>
>> Since commit 665d6a59161769e10b52ffcbcd5cd2db22f32681 I have to build
>> glibc from source.  Unfortunately, this fails with the following error:
>>
>> …
>> i686-guix-linux-gnu-gcc ../sysdeps/i386/i686/multiarch/strcspn-c.c -c 
>> -std=gnu11 -fgnu89-inline  -O2 -Wall -Werror -Wundef - 
>> In file included from ../sysdeps/x86_64/multiarch/strcspn-c.c:22:0,
>>                  from ../sysdeps/i386/i686/multiarch/strcspn-c.c:2:
>> ../sysdeps/x86_64/multiarch/varshift.h: In function '__m128i_shift_right':
>> ../sysdeps/x86_64/multiarch/varshift.h:26:1: error: SSE vector return 
>> without SSE enabled changes the ABI [-Werror=psabi]
>>  {
>>  ^
>> In file included from 
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.,
>>                  from 
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.,
>>                  from ../sysdeps/x86_64/multiarch/strcspn-c.c:20,
>>                  from ../sysdeps/i386/i686/multiarch/strcspn-c.c:2:
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.0/include/tmmintrin.h:h
>>  _mm_shuffle_epi8 (__m128i __X, __m128i __Y)
>>  ^
>> In file included from ../sysdeps/x86_64/multiarch/strcspn-c.c:22:0,
>>                  from ../sysdeps/i386/i686/multiarch/strcspn-c.c:2:
>> ../sysdeps/x86_64/multiarch/varshift.h:27:10: error: called from here
>>    return _mm_shuffle_epi8 (value,
>>           ^
>> In file included from 
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.,
>>                  from 
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.,
>>                  from 
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.,
>>                  from 
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.,
>>                  from ../sysdeps/x86_64/multiarch/strcspn-c.c:20,
>>                  from ../sysdeps/i386/i686/multiarch/strcspn-c.c:2:
>> /gnu/store/85qsxn71dn6944df5kcvkxg0nm3xdg6z-gcc-cross-boot0-5.4.0-lib/lib/gcc/i686-guix-linux-gnu/5.4.0/include/emmintrin.h:h
>>  _mm_loadu_si128 (__m128i const *__P)
>>  ^
>> In file included from ../sysdeps/x86_64/multiarch/strcspn-c.c:22:0,
>>                  from ../sysdeps/i386/i686/multiarch/strcspn-c.c:2:
>> ../sysdeps/x86_64/multiarch/varshift.h:27:10: error: called from here
>>    return _mm_shuffle_epi8 (value,
>>           ^
>> cc1: all warnings being treated as errors
>> make[4]: *** [/tmp/guix-build-glibc-2.25.drv-0/build/sysd-rules:561: 
>> /tmp/guix-build-glibc-2.25.drv-0/build/string/rtld-strc1
>> make[4]: Leaving directory 
>> '/tmp/guix-build-glibc-2.25.drv-0/glibc-2.25/string'
>> make[3]: *** [../o-iterator.mk:9: 
>> /tmp/guix-build-glibc-2.25.drv-0/build/string/rtld-strchr.os] Error 2
>> make[3]: Leaving directory '/tmp/guix-build-glibc-2.25.drv-0/glibc-2.25/elf'
>> make[2]: *** [Makefile:443: 
>> /tmp/guix-build-glibc-2.25.drv-0/build/elf/rtld-libc.a] Error 2
>> make[2]: Leaving directory '/tmp/guix-build-glibc-2.25.drv-0/glibc-2.25/elf'
>> make[1]: *** [Makefile:215: elf/subdir_lib] Error 2
>> make[1]: Leaving directory '/tmp/guix-build-glibc-2.25.drv-0/glibc-2.25'
>> make: *** [Makefile:9: all] Error 2
>> phase `build' failed after 1749.9 seconds
>> builder for `/gnu/store/ivvdx2m0p6gnmcxmz355z106ffqg9p25-glibc-2.25.drv' 
>> failed with exit code 1
>> guix system: error: build failed: build of 
>> `/gnu/store/ivvdx2m0p6gnmcxmz355z106ffqg9p25-glibc-2.25.drv' failed
>>
>> This means I cannot upgrade my i686 machines.
>
> Hmm.  The patches added by commit 665d6a5916 change only one file:
> rtld.c, so I don't see how that could affect the building of
> strcspn-c.c.
>
> For now, I will try to build this derivation on Hydra, so that it will
> be available as a substitute.

The same failure occurred when Hydra tried to build this.  I'm not sure
what's going on here.  To be continued...

      Mark