On Tue, 2012-04-24 at 23:35 +0100, Ken Moffat wrote: > On Tue, Apr 24, 2012 at 10:42:40PM +0100, Matt Burgess wrote: > > On Tue, 2012-04-24 at 16:18 -0500, Bruce Dubbs wrote: > > > Ken Moffat wrote: > > > > > > > Thanks. So, why do we do it on 32-bit x86 ? My memory says that > > > > Greg proposed it, something along the lines of "it's what upstream > > > > recommend", but I could be mistaken. > > > > > > I can't recall and I don't want to research it right now. > > > > http://wiki.linuxfromscratch.org/lfs/ticket/2299 and the related commit at > > http://wiki.linuxfromscratch.org/lfs/changeset/8885 - in short Fedora > > and DIY did it, so we copied them. > > > > Regards, > > > > Matt. > > > Thanks, Matt - I got as far as r8885 but took a break and didn't > find the ticket. I don't particularly care what DIY do or did, but > if Fedora did it, well, must be the right thing. So, when is LFS > moving to systemd and initrd ? (that will be the time I leave, I > think ;-) > > But, my current failure to build with -O3 notwithstanding, why only > on m486 and not x86_64 ?
No idea, Ken. Looking at Fedora, they've always built x86_64 with O3 as well (as well as many other arches too). Maybe it was simply because, at the time, LFS only targetted/tested x86 and therefore to err on the side of caution, we guarded the O3 flag in case it caused build failures on arches we weren't testing at the time? My personal view on this is that we should just drop that and use whatever upstream consider to be the correct CFLAGS, which from my current logs appears to be O2 on x86_64. Regards, Matt. -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
