Summary: | www-client/firefox-49.0_beta8 fails to build on musl-amd64 with 'pthread_setname_np was not declared in this scope' | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | tt_1 <herrtimson> |
Component: | Overlays | Assignee: | Gentoo musl team <musl> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | blueness, felix.janda, gentoo, mozilla, tsmksubc |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 430702 | ||
Attachments: |
compressed build.log
emerge.info patch for sys-libs/musl |
Created attachment 445132 [details]
emerge.info
musl doesn't have pthread_setname_np. "np" stands for non-portable, and indeed for different systems the number and type of arguments differs. To make firefox happy, musl would need to implement the same variant as glibc or Mac OSX. In http://openwall.com/lists/musl/2016/04/27/2 the musl author says that there are no major objections to adding it. Created attachment 445634 [details, diff]
patch for sys-libs/musl
The patch for musl solves the compile error of firefox. Is it a copy from glibc? No, the function is written from scratch using the man page. An improved version of this patch is currently being reviewed on the musl mailing list. I was also able to build and run www-client/firefox-49.0_beta10 against musl-1.1.15 with the improved version of Felix's patch from http://www.openwall.com/lists/musl/2016/09/17/1 . So if musl is what needs the patch, is there anything I should do before I commit firefox-49.0 to the tree? The patch has been commited upstream at http://git.musl-libc.org/cgit/musl/commit/?id=8fb28b0b3e7a5e958fb844722a4b2ef9bc244af1 this has been fixed in musl-1.1.16 |
Created attachment 445130 [details] compressed build.log pthread_setname_np and musl don't seem to like each other?