kevin/linux
1
Fork 0
Code

locking/atomic/x86: Introduce the read64_nonatomic macro to x86_32 with cx8

Browse Source 
As described in commit:

  e73c4e34a0 ("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 the read64_nonatomic assembly macro to load the value from a
atomic_t location in a faster non-atomic way and use it in
atomic64_cx8_32.S.

Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lore.kernel.org/r/20240605181424.3228-1-ubizjak@gmail.com
This commit is contained in:
Uros Bizjak 2024-06-05 20:13:15 +02:00 committed by Ingo Molnar
parent 51835949dd
commit 6e30a7c98a
1 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
arch/x86/lib/atomic64_cx8_32.S
View File

@ -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