Just learned the hard way that at least for non-volatile asm()s gcc
indeed does what the documentation says: It may move it across jumps
(i.e. ahead of the cpu_has() check). While the documentation claims
that this can also happen for volatile asm()s, if that was the case
we'd have many more problems in our code (and e,g, Linux would too).
Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
--- a/xen/include/asm-x86/random.h
+++ b/xen/include/asm-x86/random.h
@@ -8,7 +8,7 @@ static inline unsigned int arch_get_rand
unsigned int val = 0;
if ( cpu_has(¤t_cpu_data, X86_FEATURE_RDRAND) )
- asm ( ".byte 0x0f,0xc7,0xf0" : "+a" (val) );
+ __asm__ __volatile__ ( ".byte 0x0f,0xc7,0xf0" : "+a" (val) );