James Phillips wrote:
I laughed at your question because I remember reading somewhere that using aggressive optimization options is a good way to find compiler bugs. I think that extends of optimizations for "new" CPU architectures as well.
I also heard kernel code avoids MMX instructions for some reason: it may have to do with interrupt handling (fewer registers=faster?). x86 (and AMD64) processors are backwards compatible, so you don't strictly need the latest instructions.
Ah , i've just read it may not be safe to use MMX and SSE instructions
in kernel code.
So my CFLAGS is much too agressive , i'll notice this.
Thank you all.
Aaron Lewis - PGP: 0xA476D2E9
irc: A4r0n on freenode
firstname.lastname@example.org mailing list
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"