[PATCH RESEND -tip 1/2] locking/atomic/x86: Introduce read64_nonatomic macro to x86_32 with cx8

[Uros Bizjak]

As described in:

  commit e73c4e34a0e9 ("locking/atomic/x86: Introduce arch_atomic64_read_nonatomic() to x86_32")

the value preload before the cmpxchg loop does not need to be atomic.
Introduce read64_nonatomic macro to load the value from atomic_t
location in a faster non-atomic way and use it in atomic64_cx8_32.S.

Signed-off-by: Uros Bizjak <ubizjak@xxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Borislav Petkov <bp@xxxxxxxxx>
Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
Cc: "H. Peter Anvin" <hpa@xxxxxxxxx>
---
 arch/x86/lib/atomic64_cx8_32.S | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/arch/x86/lib/atomic64_cx8_32.S b/arch/x86/lib/atomic64_cx8_32.S
index 90afb488b396..b2eff07d65e4 100644
--- a/arch/x86/lib/atomic64_cx8_32.S
+++ b/arch/x86/lib/atomic64_cx8_32.S
@@ -16,6 +16,11 @@
 	cmpxchg8b (\reg)
 .endm
 
+.macro read64_nonatomic reg
+	movl (\reg), %eax
+	movl 4(\reg), %edx
+.endm
+
 SYM_FUNC_START(atomic64_read_cx8)
 	read64 %ecx
 	RET
@@ -51,7 +56,7 @@ SYM_FUNC_START(atomic64_\func\()_return_cx8)
 	movl %edx, %edi
 	movl %ecx, %ebp
 
-	read64 %ecx
+	read64_nonatomic %ecx
 1:
 	movl %eax, %ebx
 	movl %edx, %ecx
@@ -79,7 +84,7 @@ addsub_return sub sub sbb
 SYM_FUNC_START(atomic64_\func\()_return_cx8)
 	pushl %ebx
 
-	read64 %esi
+	read64_nonatomic %esi
 1:
 	movl %eax, %ebx
 	movl %edx, %ecx
-- 
2.42.0