Skip to contentSkip to conversation

Opteron bug

DISCUSSION Manfred Spraul 2 participants 2 messages Jun 20, 2004

lore
Share & tools
lore
#0 lore raw
Quoted 3 lines · Andi

The kernel never uses backwards REP prefixes.

Huh? memmove uses backwards rep movsb on x86-64, at least in 2.6.5:

http://lxr.linux.no/source/arch/x86_64/lib/memmove.c?v=2.6.5;a=x86_64#L8

i386 is similar:

http://lxr.linux.no/source/include/asm-i386/string.h?v=2.6.5#L297

But the bug appears to be so rare that it shouldn't matter - lets wait for the microcode update. Btw, is there a runtime microcode updater for Opteron cpus, similar to the Intel microcode driver?

--
Manfred

#1 lore raw
Quoted 6 lines · On Sun, 20 Jun 2004 15:07:59 +0200 Manfred Spraul <[email protected]>…
Quoted 3 lines · Andi

The kernel never uses backwards REP prefixes.

Huh? memmove uses backwards rep movsb on x86-64, at least in 2.6.5:

Hmm, you're right. I forgot that. Ok, I will change it, although it's unlikely to be a real problem.

Quoted 3 lines

But the bug appears to be so rare that it shouldn't matter - lets wait for the microcode update. Btw, is there a runtime microcode updater for Opteron cpus, similar to the Intel microcode driver?

AMD seems to only release microcode updates to BIOS vendors.

-Andi