app-antivirus/clamav-0.100.2 build with forced -O0. This is becouse configure --disable-gcc-vcheck always append -O0 (as "unknown broken compiler"). This is degrade perfomance near twice to my current testing flags. I suggest --enable-gcc-vcheck or something like (my current workaround): sed -i -e 's:CFLAGS -O0:CFLAGS:g' "${S}"/configure* "${S}"/m4/reorganization/compiler_checks.m4
Not fixed in 0.102.0. This causes a tremendous performance hit on both startup and during file checks.
Jeroen, it looks like maybe you added that --disable-gcc-vcheck line in commit 48726a08716? It looks to me like that check can only have observable effects with gcc-4.1.0 and gcc-4.4.0, neither of which we have to worry about. So, I think it can be removed. If you added it, and if you don't remember why, then let's just kill it.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=5114e1838d8531ebce400e668efec476b1c0815c commit 5114e1838d8531ebce400e668efec476b1c0815c Author: Michael Orlitzky <mjo@gentoo.org> AuthorDate: 2019-10-16 11:24:26 +0000 Commit: Michael Orlitzky <mjo@gentoo.org> CommitDate: 2019-10-16 11:31:31 +0000 app-antivirus/clamav: remove --disable-gcc-vcheck from configure flags. Our ClamAV builds have passed the --disable-gcc-vcheck flag to its configure script for a while. This is intended to skip a (pointless, on Gentoo) test for old broken versions of GCC, but accidentally leads to "-O0" being inserted into the CFLAGS. That in turn causes noticeable performance degradation. Since the GCC version check is harmless aside from some wasted time, it seems the simplest solution here is to drop "--disable-gcc-vcheck" from the build. This commit adds a new revision of the latest unstable version that does just that. Closes: https://bugs.gentoo.org/670828 Package-Manager: Portage-2.3.76, Repoman-2.3.16 Signed-off-by: Michael Orlitzky <mjo@gentoo.org> .../{clamav-0.102.0.ebuild => clamav-0.102.0-r1.ebuild} | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-)
(In reply to Michael Orlitzky from comment #2) I didn't previously see this comment. > Jeroen, it looks like maybe you added that --disable-gcc-vcheck line in > commit 48726a08716? Looks like it. Sadly I didn't record why I did that and I didn't observe that CFLAGS injection side effect at the time. > It looks to me like that check can only have observable effects with > gcc-4.1.0 and gcc-4.4.0, neither of which we have to worry about. So, I > think it can be removed. > > If you added it, and if you don't remember why, then let's just kill it. Yes, that seems to have worked.