lto adds 100k more weight with the libgcc_s.so.1 library. The speed is not always what one wants Reproducible: Always
Created attachment 394290 [details, diff] mksh-50d-lto-optional.patch
Recent GCC versions are buggy, you’ll get the shell miscompiled using LTO; for example, the arith-ternary-prec-1 test will fail (bonsaikitten had that the other day for mksh R56c with LTO). At the current point in time, I can only advice *against* using LTO, until compiler maintainers actually support and fix it. (We’ve been having this problem several times, even with -fwhole-program --combine, and GCC is a repeat offender.) Current Hat: mksh upstream
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=6d2258d85eac24a106a262a6d12cb02b434d2f8c commit 6d2258d85eac24a106a262a6d12cb02b434d2f8c Author: Haelwenn (lanodan) Monnier <contact@hacktivis.me> AuthorDate: 2019-03-17 10:23:33 +0000 Commit: Michał Górny <mgorny@gentoo.org> CommitDate: 2019-04-29 13:41:06 +0000 app-shells/mksh: Fix the live ebuild - mksh changed to tarballs so cpio and unpacker is now removed. - Update keywords from release ebuild - Rebuild.sh is useless for us - Bump to EAPI 7 - Remove shell updater as mksh is in baselayout - Remove unstable -c lto - exeinto,doexe → into,dobin - Add test suite dependencies Closes: https://bugs.gentoo.org/show_bug.cgi?id=536992 Closes: https://bugs.gentoo.org/show_bug.cgi?id=631948 Signed-off-by: Haelwenn (lanodan) Monnier <contact@hacktivis.me> Signed-off-by: Michał Górny <mgorny@gentoo.org> app-shells/mksh/mksh-9999.ebuild | 42 ++++++++++++++++++---------------------- 1 file changed, 19 insertions(+), 23 deletions(-)
