3DNOW optimization fix
PATCH 1/1 3DNOW optimization fix
As noted by Andreas Franck, the assembler instruction for write prefetch is mistakenly a read prefetch. Below is for linux-2.4.8-ac11 that corrects the error.
Regards
Anders Fugmann
--
include/asm-i386/processor.h Sun Aug 26 01:22:36 2001 renamed from include/asm-i386/processor.h~ Sun Aug 26 01:18:12 2001 +1 -1
@@ -501,7 +502,7 @@ extern inline void prefetchw(const void *x) { - __asm__ __volatile__ ("prefetch (%0)" : : "r"(x)); + __asm__ __volatile__ ("prefetchw (%0)" : : "r"(x)); } #define spin_lock_prefetch(x) prefetchw(x)