www-client/firefox-91.4.0-r1 fails to build on x86 with the following error: ERROR: www-client/firefox-91.4.0-r1::gentoo failed (compile phase): (no error message) Attached is output of `emerge --info '=www-client/firefox-91.4.0-r1::gentoo'` output of `emerge -pqv '=www-client/firefox-91.4.0-r1::gentoo'` and the complete build log. Reproducible: Always Steps to Reproduce: 1. emerge -v www-client/firefox-91.4.0-r1 Actual Results: Compilation fails Expected Results: Compilation finishes without error
Created attachment 759376 [details] emerge --info '=www-client/firefox-91.4.0-r1::gentoo'
Created attachment 759377 [details] emerge -pqv '=www-client/firefox-91.4.0-r1::gentoo'
Please attach the full build.log.
Created attachment 759378 [details] Build log
Looks like bug https://bugs.gentoo.org/816975 to me. There's a patch provided to help you, but the main reason seems to be either emulated or cross-compile environment? I'd suggest setting matching CFLAGS and CXXFLAGS (they seem to differ for you), use setarch i686, and/or use -m32 in C(XX)FLAGS. *** This bug has been marked as a duplicate of bug 816975 ***
It's neither emulated nor cross-compile environment. But thanks for the CFLAGS/CXXFLAGS tip. I'll try your suggestion and report back ...
Setting matching CFLAGS and CXXFLAGS does NOT solve the issue. As I've also written before it's neither emulated nor cross-compile environment. New logs are attached. Any other suggestions ?
Created attachment 759530 [details] emerge --info '=www-client/firefox-91.4.0-r1::gentoo'
Created attachment 759531 [details] emerge -pqv '=www-client/firefox-91.4.0-r1::gentoo'
Created attachment 759532 [details] Build log
(In reply to berni.lakeroe from comment #7) > > Any other suggestions ? I gave you plenty in #c5 including a path to a patch if everything else fails. Your CPU is 64-bit so it's not a "real" 32-bit env, also I can't reproduce so it's something in your system. Please check the linked bug for discussion. You may find more success by setting -march=native too. *** This bug has been marked as a duplicate of bug 816975 ***
(In reply to Joonas Niilola from comment #11) > (In reply to berni.lakeroe from comment #7) > > > > Any other suggestions ? > > I gave you plenty in #c5 including a path to a patch if everything else > fails. > > Your CPU is 64-bit so it's not a "real" 32-bit env, also I can't reproduce > so it's something in your system. Please check the linked bug for > discussion. If they're booted into an x86 kernel, I don't think this is really true. For years we had lots of people running x86 on technically-capable-of-64-bit CPUs. If booted in, it should make no difference. > > You may find more success by setting -march=native too. > > *** This bug has been marked as a duplicate of bug 816975 ***
(In reply to Joonas Niilola from comment #11) > (In reply to berni.lakeroe from comment #7) > > > > Any other suggestions ? > > I gave you plenty in #c5 including a path to a patch if everything else > fails. > > Your CPU is 64-bit so it's not a "real" 32-bit env, also I can't reproduce > so it's something in your system. Please check the linked bug for > discussion. > > You may find more success by setting -march=native too. > > *** This bug has been marked as a duplicate of bug 816975 *** Why do you think this is a duplicate of bug 816975 ? I get a completely different error message ...
(In reply to berni.lakeroe from comment #13) > > Why do you think this is a duplicate of bug 816975 ? > I get a completely different error message ... I see this, even in your latest build.log: 78[1G[K[34m168:37.61[m [01m[K/var/tmp/portage/www-client/firefox-91.4.0-r1/work/firefox-91.4.0/modules/fdlibm/src/math_private.h:34:21:[m[K [01;31m[Kerror: [m[Kconflicting declaration '[01m[Ktypedef __double_t double_t[m[K'[m 78[1G[K[34m168:37.61[m 34 | typedef __double_t [01;31m[Kdouble_t[m[K;[m 78[1G[K[34m168:37.61[m | [01;31m[K^~~~~~~~[m[K[m 78[1G[K[34m168:37.61[m In file included from [01m[K/usr/lib/gcc/i686-pc-linux-gnu/11.2.0/include/g++-v11/cmath:45[m[K,[m 78[1G[K[34m168:37.61[m from [01m[K/var/tmp/portage/www-client/firefox-91.4.0-r1/work/firefox_build/dist/system_wrappers/cmath:3[m[K,[m 78[1G[K[34m168:37.61[m from [01m[K/var/tmp/portage/www-client/firefox-91.4.0-r1/work/firefox_build/dist/stl_wrappers/cmath:60[m[K,[m 78[1G[K[34m168:37.61[m from [01m[K/var/tmp/portage/www-client/firefox-91.4.0-r1/work/firefox-91.4.0/modules/fdlibm/src/e_acos.cpp:41[m[K:[m 78[1G[K[34m168:37.62[m [01m[K/usr/include/math.h:156:21:[m[K [01;36m[Knote: [m[Kprevious declaration as '[01m[Ktypedef long double double_t[m[K'[m 78[1G[K[34m168:37.62[m 156 | typedef long double [01;36m[Kdouble_t[m[K;[m 78[1G[K[34m168:37.62[m | [01;36m[K^~~~~~~~[m[K[m 78[1G[K[34m168:37.86[m gmake[4]: *** [/var/tmp/portage/www-client/firefox-91.4.0-r1/work/firefox-91.4.0/config/rules.mk:675: e_acos.o] Error 1[m (don't mind the stupid unicode mess) That was also the exact error in the other bug. Also the provided patch should work for this. Anyway, I offered to carry this patch into our firefox release but haven't heard back from our x86 team to test it doesn't break on *real* x86 hw. I guess I should ping them again.
Ah I see, I simply overlooked this error and now I know what to do. Thanks for clarifying things !
I can confirm now this is a duplicate of bug 816975. After changing typedef double __double_t; to #if defined __FLT_EVAL_METHOD__ && (__FLT_EVAL_METHOD__ == 2) typedef long double __double_t; #else typedef double __double_t; #endif in /var/tmp/portage/www-client/firefox-91.4.0-r1/work/firefox-91.4.0/modules/fdlibm/src/math_private.h it compiles without issues. Setting matching CFLAGS and CXXFLAGS, -march=native or -march=i686 and -m32 did not work for me.