The ebuild has: ``` # Allow elfhack to work in combination with unstripped binaries # when they would normally be larger than 2GiB. append-ldflags "-Wl,--compress-debug-sections=zlib" ``` This interacts kind of poorly with FEATURES=compressdebug as debugedit can't handle it (https://sourceware.org/bugzilla/show_bug.cgi?id=27636). It leads to some noise when merging FF. I wonder if this issue still exists w/ elfhack? (Any references we can find?) If it's still an issue, it's not a big deal, but I thought I'd ask.
Yeah, does look like elfhack is legacy now: # With profile 23.0 elf-hack=legacy is broken with gcc. # With Firefox-115esr elf-hack=relr isn't available (only in rapid). # Solution: Disable build system's elf-hack completely, and add "-z,pack-relative-relocs" # manually with gcc. # # elf-hack configure option isn't available on ppc64/riscv, #916259, #929244, #930046. This is in 115esr's ebuild. I'd like to remove this on next version bump to go through all tests naturally. I'll include testing in x86.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=cef95379ab8792e466d965c27d26a4f095a8feda commit cef95379ab8792e466d965c27d26a4f095a8feda Author: Joonas Niilola <juippis@gentoo.org> AuthorDate: 2024-10-29 10:02:24 +0000 Commit: Joonas Niilola <juippis@gentoo.org> CommitDate: 2024-10-29 11:27:37 +0000 www-client/firefox: add 132.0 - drop 'append-ldflags "-Wl,--compress-debug-sections=zlib"' - drop apulse-related messages - apulse[sdk] should be a 1:1 replacement for pulseaudio, - drop bunch of upstreamed patches, - include a patch in an attempt to enhance results when using gcc to build, - toggle the default 'clang' use flag off. Closes: https://bugs.gentoo.org/939445 Closes: https://bugs.gentoo.org/941878 Signed-off-by: Joonas Niilola <juippis@gentoo.org> www-client/firefox/Manifest | 102 +++ www-client/firefox/firefox-132.0.ebuild | 1336 +++++++++++++++++++++++++++++++ 2 files changed, 1438 insertions(+)
Thank you! BTW, if we need to revert this change, the upcoming debugedit release (>5.0) should solve the issue I was originally hitting, I think.
