[lfs-support] Glibc and -march=i486 option

Bruce Dubbs bruce.dubbs at gmail.com
Mon Apr 9 14:33:54 PDT 2012


ante wrote:
> On Mon, Apr 9, 2012 at 10:09 PM, Bruce Dubbs <bruce.dubbs at gmail.com> wrote:
>> ante wrote:
>>> Well, I try to ask it once more. Can anybody explain (or give me a link to
>>> relevant discussion) why glibc should be compiled with -march=i486 option?
>>> I remember that in LFS 5.0 it was recommended to clear CFLAGS and CXXFLAGS
>>> during compilation; but why it is recommended -march=i486 option now?
>> To make it consistent for 32-bit intel platforms. There are very few
>> differences in the instruction set between i486 and later. Building for
>> an i686 would probably break if run on an i486 and there would be no
>> significant performance improvements.
> 
> Yes, I understand this. But the system won't run on i486 anyway since
> glibc is the only package built with -march=i486 option. So I suppose
> there must be yet another reason.

It was added 09/15/07.  See

http://wiki.linuxfromscratch.org/lfs/ticket/2018

   -- Bruce




More information about the lfs-support mailing list