systimer01.patch available for review.

Matthew Dillon dillon at apollo.backplane.com
Tue Jan 27 22:33:57 PST 2004


:Shouldn't we either decide on:
:
:- Yes we support i386
:- No we do no support i386
:
:Otherwise you get this dangling piece of code which can bitrot away.
:That's exactly what FreeBSD has been suffering a lot of as well.
:
:If yes, we concentrate on keeping it working, if not we actively work
:towards removing support for it and see if we can now optimise by
:default for i486 and higher.
:
:-- 
:Jeroen Ruigrok van der Werven <asmodai(at)wxs.nl> / asmodai / kita no mono

    Well, what 386 support exists in the source base does not really
    interfere with development so there is no pressing need to rip it
    all out yet.

    But insofar as code generation goes, -mcpu=pentiumpro is now being 
    set for GCC3 (ala Joerg, thank you very much!), and I think that's a
    good idea.   While we could set -march=i486 there isn't much of a point
    doing so because it doesn't really change code generation all that
    much, and we cannot set -march=pentiumpro because it will generate
    CMOVE instructions which are often missing on intel-compatible cpu's
    like the VIA C3.

    We could make a concerted effort to fix the bad inline assembly
    we have in the kernel to make -O3 and friends work properly under gcc3.

					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>





More information about the Kernel mailing list