| Summary: | >=www-client/firefox-72.0: building with USE=pgo and USE=-clang fails | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | darkstalker <slayerbeast> |
| Component: | Current packages | Assignee: | Mozilla Gentoo Team <mozilla> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | daggs, dechcaudron, geggen54, gentoo, jstein, mail, nanikata15, zohran.londais |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | https://bugzilla.mozilla.org/show_bug.cgi?id=1601903 | ||
| See Also: | https://bugs.gentoo.org/show_bug.cgi?id=703256 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
build log
emerge --info emerge --info emerge --info build.log Fix broken patchset, readd missing llvm/lto patches |
||
Created attachment 613856 [details]
emerge --info
On my system Firefox 73 failed to build with +lto so I removed that and it built. So I don't think it's just "pgo lto". It's +lto alone although I know OP said it builds with lto alone. Created attachment 613902 [details]
emerge --info
I have the same problem.
In order to compile use ld.gold adding -Wl,-fuse-ld=gold to LDFLAGS Ciao Luigi Can confirm the problem. Adding -Wl,-fuse-ld=gold to LDFLAGS did not help - build aborted with same error. Disabling pgo and lto did make it build. (In reply to Luigi 'Comio' Mantellini from comment #4) > In order to compile use ld.gold adding -Wl,-fuse-ld=gold to LDFLAGS > Ciao > > Luigi Didn't help neither. Created attachment 614088 [details]
emerge --info
Can confirm same behavior. Attaching 'emerge --info'.
Created attachment 614090 [details]
build.log
the profile server is not generating any data, this is what the failure is trying to indicate. I will look at the issue, but as I have stated multiple times I do not use pgo so someone will have to test any possible fix I might find. (In reply to Jory A. Pratt from comment #9) > the profile server is not generating any data, this is what the failure is > trying to indicate. I will look at the issue, but as I have stated multiple > times I do not use pgo so someone will have to test any possible fix I might > find. I can also confirm the issue and I can also check any possible fix. The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=3755724da101dd492d6a2a61083945ad92933666 commit 3755724da101dd492d6a2a61083945ad92933666 Author: Thomas Deutschmann <whissi@gentoo.org> AuthorDate: 2020-02-24 16:38:40 +0000 Commit: Thomas Deutschmann <whissi@gentoo.org> CommitDate: 2020-02-24 16:52:46 +0000 profiles: p.use.mask >=www-client/firefox-72[pgo] USE=pgo is broken since firefox-72. Bug: https://bugs.gentoo.org/709594 Bug: https://bugs.gentoo.org/709654 Signed-off-by: Thomas Deutschmann <whissi@gentoo.org> profiles/arch/amd64/package.use.mask | 4 ++-- profiles/arch/powerpc/ppc64/64le/package.use.mask | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) Wait, pgo is working perfectly fine for me, on both 72 and 73, but I'm using clang (which firefox tend to test a lot more given they use it for official binaries, I didn't try 73.0.1 yet though) Is it really necessary to mask it for both or do I have to unmask it manually now? Sorry, I didn't test with USE=clang. I'll check what we can do here without forcing user interaction for users without USE=clang. Yeah I understand. Given can't do that kind of conditional mask I imagine would need ebuild changes but then users would need to remove the pgo flag instead unless let it stay "enabled" and simply not get used (but that's misleading, so I don't know). On another note, I can report that 73.0.1 built fine with clang+lto+pgo(unmasked) That wasn't very nice at all. builds and runs just fine with lto/pgo/clang. The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=0658199715254b723aefcdea8641ccde7d316a36 commit 0658199715254b723aefcdea8641ccde7d316a36 Author: Thomas Deutschmann <whissi@gentoo.org> AuthorDate: 2020-02-24 22:33:45 +0000 Commit: Thomas Deutschmann <whissi@gentoo.org> CommitDate: 2020-02-24 22:35:19 +0000 profiles: Revert "profiles: p.use.mask >=www-client/firefox-72[pgo]" This reverts commit 3755724da101dd492d6a2a61083945ad92933666. p.use.mask was too strict because USE=pgo works when using USE=clang. Bug: https://bugs.gentoo.org/709594 Bug: https://bugs.gentoo.org/709654 Signed-off-by: Thomas Deutschmann <whissi@gentoo.org> profiles/arch/amd64/package.use.mask | 4 ++-- profiles/arch/powerpc/ppc64/64le/package.use.mask | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=5059bb531cc73e428042a83fe07191b2d98d40c4 commit 5059bb531cc73e428042a83fe07191b2d98d40c4 Author: Thomas Deutschmann <whissi@gentoo.org> AuthorDate: 2020-02-24 22:31:24 +0000 Commit: Thomas Deutschmann <whissi@gentoo.org> CommitDate: 2020-02-24 22:35:18 +0000 www-client/firefox: block USE=pgo when USE=-clang USE=pgo is currently known to be broken when using GCC. Instead of silently ignoring USE=pgo when using GCC, we are forcing a user decision so user has the chance to enable USE=clang in case PGO is important for user without wasting 2+ hours commpiling firefox just to find out that PGO which was requested was ignored. Bug: https://bugs.gentoo.org/709594 Bug: https://bugs.gentoo.org/709654 Package-Manager: Portage-2.3.89, Repoman-2.3.20 Signed-off-by: Thomas Deutschmann <whissi@gentoo.org> www-client/firefox/firefox-73.0.1.ebuild | 8 ++++++++ 1 file changed, 8 insertions(+) *** Bug 709594 has been marked as a duplicate of this bug. *** I don't understand this bug report as I have a firefox-72.0 binary package, working fine too, iirc.
eix firefox -I
[I] www-client/firefox
Installed versions: 73.0.1^td(21:04:52 27/02/2020)(clang custom-cflags jack lto pgo startup-notification system-av1 system-harfbuzz system-icu system-jpeg system-libevent system-libvpx system-sqlite system-webp wayland -bindist -custom-optimization -debug -eme-free -geckodriver -gmp-autoupdate -hardened -hwaccel -pulseaudio -screenshot -selinux -test -wifi CPU_FLAGS_ARM="-neon" CPU_FLAGS_X86="avx2" L10N="fr -ach -af -an -ar -ast -az -be -bg -bn -br -bs -ca -cak -cs -cy -da -de -dsb -el -en-CA -en-GB -eo -es-AR -es-CL -es-ES -es-MX -et -eu -fa -ff -fi -fy -ga -gd -gl -gn -gu -he -hi -hr -hsb -hu -hy -ia -id -is -it -ja -ka -kab -kk -km -kn -ko -lij -lt -lv -mk -mr -ms -my -nb -nl -nn -oc -pa -pl -pt-BR -pt-PT -rm -ro -ru -si -sk -sl -son -sq -sr -sv -ta -te -th -tr -uk -ur -uz -vi -xh -zh-CN -zh-TW")
emerge -p -K =firefox-72.0.2
These are the packages that would be merged, in reverse order:
Calculating dependencies ...... done!
[binary UD~] www-client/firefox-72.0.2 [73.0.1] USE="-clang*"
*** Bug 716576 has been marked as a duplicate of this bug. *** Created attachment 631224 [details, diff]
Fix broken patchset, readd missing llvm/lto patches
(In reply to Jory A. Pratt from comment #20) > Created attachment 631224 [details, diff] [details, diff] > Fix broken patchset, readd missing llvm/lto patches This seems to do the trick for making USE=lto,-pgo,-clang firefox build successfully, as long as one restores the two patches that Thomas has removed as part of an old version security cleanup 2 days ago. *** Bug 717432 has been marked as a duplicate of this bug. *** syncned repo to 777a84c4c5cc4240b97ec11ce431ddf704f6c534, 75.0 builds well. |
Created attachment 613854 [details] build log Compiling firefox 73 with gcc and USE="lto pgo" fails. Using only lto works fine.