| View previous topic :: View next topic |
| Author |
Message |
lars_the_bear
Tux's lil' helper
Joined: 05 Jun 2024
Posts: 115
|
 Posted: Thu Jun 06, 2024 10:20 am Post subject: [solved] Firefox installation failed :/ |
 |
|
Hi folks
Well, this is a stinker :/
After nearly four hours of compiling, the firefox build failed, with this compiler error (and many others much like it):
| Code: |
...
88:32.28 In file included from /var/tmp/portage/www-client/firefox-115.11.0/work/firefox-115.11.0/gfx/skia/skia/src/core/SkOpts.cpp:46:
88:32.28 /var/tmp/portage/www-client/firefox-115.11.0/work/firefox-115.11.0/gfx/skia/skia/src/opts/SkRasterPipeline_opts.h:406:40: error: always_inline function '_mm256_fmadd_ps' requires target feature 'fma', but would be inlined into function 'mad' that is compiled without support for 'fma'
88:32.28 406 | SI F mad(F f, F m, F a) { return _mm256_fmadd_ps(f, m, a); }
|
I'm not certain, but I think this is a bug/limitation in GCC, triggered by using '--march=native'. It's trying to use macros associated with the AVX instruction set, but support for them has not been built into the various GCC libraries. Still, at compile time, the compiler is trying to use them. The CPU does support AVX (at least, /proc/cpuinfo says it does), but GCC has done something freaky.
I think, from earlier experience, that this problem might be solved by removing '--march=native'; but the use of this flag is recommended in the Gentoo set-up instructions. It's not been a problem until now.
I'm currently trying to rebuild like this, without --march=native :
# CFLAGS=-O2 CXXFLAGS=-O2 emerge firefox
I guess I'll know in another four hours or so whether it's successful.
What I'm getting around to is: is the use of `--march=native` generally such a good idea? Am I likely to run into other problems using this? How much difference is it likely to make, if I don't?
BR, Lars
Last edited by lars_the_bear on Thu Jun 06, 2024 9:00 pm; edited 1 time in total |
|
| Back to top |
|
 |
lars_the_bear
Tux's lil' helper
Joined: 05 Jun 2024
Posts: 115
|
| Posted: Thu Jun 06, 2024 1:49 pm Post subject: |
|
|
For the record,
# CFLAGS=-O2 CXXFLAGS=-O2 emerge firefox
worked, and firefox is now running. So I guess '--march=native' was the culprit.
Weird. But I've seen this kind of thing before, so I wasn't entirely surprised.
BR, Lars. |
|
| Back to top |
|
|
pietinger
Moderator
Joined: 17 Oct 2006
Posts: 4495
Location: Bavaria
|
| Posted: Thu Jun 06, 2024 2:01 pm Post subject: |
|
|
It would be interesting to see what happens if you use the CPU architecture instead of “native”. Example from my box (intel 13th gen):
| Code: | COMMON_FLAGS="-march=raptorlake -O2 -pipe"
CFLAGS="${COMMON_FLAGS}"
CXXFLAGS="${COMMON_FLAGS}" |
_________________
https://wiki.gentoo.org/wiki/User:Pietinger |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 21964
|
|
| Back to top |
|
|
lars_the_bear
Tux's lil' helper
Joined: 05 Jun 2024
Posts: 115
|
| Posted: Thu Jun 06, 2024 2:57 pm Post subject: |
|
|
| pietinger wrote: | | It would be interesting to see what happens if you use the CPU architecture instead of “native”. |
Indeed. But, since it takes two hours to get to the failure point, I don't think I'm going to have the patience to try 
BR, Lars. |
|
| Back to top |
|
|
pjp
Administrator
Joined: 16 Apr 2002
Posts: 20140
|
| Posted: Thu Jun 06, 2024 4:10 pm Post subject: |
|
|
| lars_the_bear wrote: | Indeed. But, since it takes two hours to get to the failure point, I don't think I'm going to have the patience to try ;)
BR, Lars. |
You may be somewhat surprised how much more frequently firefox seems to need updates when compiling it :) If it works, the next one will be along soon enough.
_________________
Quis separabit? Quo animo? |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Installing Gentoo
