# Decent examples: #CFLAGS="-mcpu=athlon-xp -O3 -pipe" #CFLAGS="-march=pentium3 -O3 -pipe" This doesn't match neither the compilation optimization guide [1], nor the devmanual documentation [2], nor ALLOWED_FLAGS in flag-o-matic.eclass. In fact, lots of ebuild replace -O3, plus it actually makes things mostly slower (both compile-time wise and at runtime w/ gcc-4.x). [1] http://www.gentoo.org/doc/en/gcc-optimization.xml#doc_chap2 [2]http://devmanual.gentoo.org/ebuild-writing/functions/src_compile/build-environment/index.html Replace this with -O2, please.
Sending cnf/make.conf.alpha.diff Sending cnf/make.conf.arm.diff Sending cnf/make.conf.ppc.diff Sending cnf/make.conf.ppc64.diff Sending cnf/make.conf.sh.diff Sending cnf/make.conf.sparc.diff Sending cnf/make.conf.x86-fbsd.diff Sending cnf/make.conf.x86.diff Transmitting file data ........ Committed revision 7596.
(In reply to comment #1) > Sending cnf/make.conf.arm.diff > Sending cnf/make.conf.ppc.diff > Sending cnf/make.conf.ppc64.diff > Sending cnf/make.conf.sh.diff Firstly, texts in these files are now at least strange. E. g.:"Don't use -O2. Even -O2 may be risky in some cases." "That option is implied by -O2 -Os and -O2." "Long term testing has shown that -O2 opts can be unreliable on G4's but work on G3 series processors or earlier." "-O2 for the most part seems ok but should be used with caution as for instance app-editors/vim has problems if it is used. -O2 is a good selection." Secondly, if these changes will be backported to branches/2.1.2, then this bug should stay open and should block Bug 181949.
Thanks for r7597.
This bug should block Bug 181949.
This has been released in 2.1.3.6.
