I posted here asking for advice wether to list this as a bug: http://forums.gentoo.org/viewtopic-t-433403-highlight-.html Noone replied, so here goes. Wether you have a K6, K6-II, K6-II+ or K6-III processor, trying to optimize the kernel always results in -march=k6. I tried changing Makefile.cpu to compile with -march=k6-2 (on a K6-2) and it worked just fine. Why not allow GCC to -march properly?
We don't write kernels, we distribute them; you need to ask upstream. That said, there's really no reliable way to say if a CPU is K6, K6-2 or K6-III and gcc tends to produce broken code if set incorrectly (see Bug 24379 and others).
Well, I guess I can understand if the mainline kernel guys wouldn't want to make this change if it leads to hundreds of e-mails from people not getting their kernel compiled from ignorance of help text that should be written, accopanying the new K6-options. But specific patches are applied to the sources making them the Gentoo sources. And Gentoo is all about choice, right? The way Gentoo-sources are now we're not being given the CHOICE to -march=k6-2, we have to find the spot in a Makefile and change it to our liking. I can do it but everyone can't. So I protest to this being marked as resolved and hope for a simple patch and helptext warning that what seems to be a K6-2 might not be etc. Meanwhile, I'll post a bug upstream as well and see what they say about it.
(In reply to comment #2) > And Gentoo is all about choice, right? The way Gentoo-sources are now we're not > being given the CHOICE to -march=k6-2, we have to find the spot in a Makefile > and change it to our liking. I can do it but everyone can't. Yeah, better safe than sorry, broken kernel sucks. That's developers choice. > So I protest to this being marked as resolved and hope for a simple patch and > helptext warning that what seems to be a K6-2 might not be etc. Won't happen, this is not a Gentoo issue at all. > Meanwhile, I'll post a bug upstream as well and see what they say about it. Good luck. ;) Closing.
