1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
|
http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=e3a384578f677c05d812d99c2c92aa13670bd06a
Upstream-Status: Pending
Remove the `q' suffix on x86-64 atomic instructions.
We don't need the `q' suffix on x86_64 atomic instructions for AO_t,
which is defined as "unsigned long". "unsigned long" is 32bit for x32
and 64bit for x86-64. The register operand in x86-64 atomic instructions
is sufficient to properly determine the register size.
Received this patch from H.J. Lu <hjl.tools@gmail.com>
Signed-Off-By: Nitin A Kamble <nitin.a.kamble@intel.com> 2011/12/02
--- a/src/atomic_ops/sysdeps/gcc/x86_64.h
+++ b/src/atomic_ops/sysdeps/gcc/x86_64.h
@@ -56,7 +56,7 @@ AO_fetch_and_add_full (volatile AO_t *p,
{
AO_t result;
- __asm__ __volatile__ ("lock; xaddq %0, %1" :
+ __asm__ __volatile__ ("lock; xadd %0, %1" :
"=r" (result), "=m" (*p) : "0" (incr), "m" (*p)
: "memory");
return result;
@@ -106,7 +106,7 @@ AO_int_fetch_and_add_full (volatile unsi
AO_INLINE void
AO_or_full (volatile AO_t *p, AO_t incr)
{
- __asm__ __volatile__ ("lock; orq %1, %0" :
+ __asm__ __volatile__ ("lock; or %1, %0" :
"=m" (*p) : "r" (incr), "m" (*p) : "memory");
}
@@ -131,7 +131,7 @@ AO_compare_and_swap_full(volatile AO_t *
AO_t old, AO_t new_val)
{
char result;
- __asm__ __volatile__("lock; cmpxchgq %3, %0; setz %1"
+ __asm__ __volatile__("lock; cmpxchg %3, %0; setz %1"
: "=m"(*addr), "=q"(result)
: "m"(*addr), "r" (new_val), "a"(old) : "memory");
return (int) result;
|
