[PATCH 07/29] x86, tsx: Don't abort immediately in __read/write_lock_failed

[Andi Kleen]

From: Andi Kleen <ak@xxxxxxxxxxxxxxx>

__read/write_lock_failed did execute a PAUSE first thing before
checking the lock. This aborts transactions. Check the lock
state again before executing the pause. This avoids a small
number of extra aborts, and is slightly cheaper too.

Signed-off-by: Andi Kleen <ak@xxxxxxxxxxxxxxx>
---
 arch/x86/lib/rwlock.S |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/arch/x86/lib/rwlock.S b/arch/x86/lib/rwlock.S
index 1cad221..9a6bc12 100644
--- a/arch/x86/lib/rwlock.S
+++ b/arch/x86/lib/rwlock.S
@@ -16,10 +16,12 @@ ENTRY(__write_lock_failed)
 	FRAME
 0:	LOCK_PREFIX
 	WRITE_LOCK_ADD($RW_LOCK_BIAS) (%__lock_ptr)
+	cmpl    $WRITE_LOCK_CMP, (%__lock_ptr)
+	je      2f
 1:	rep; nop
 	cmpl	$WRITE_LOCK_CMP, (%__lock_ptr)
 	jne	1b
-	LOCK_PREFIX
+2:	LOCK_PREFIX
 	WRITE_LOCK_SUB($RW_LOCK_BIAS) (%__lock_ptr)
 	jnz	0b
 	ENDFRAME
@@ -32,10 +34,12 @@ ENTRY(__read_lock_failed)
 	FRAME
 0:	LOCK_PREFIX
 	READ_LOCK_SIZE(inc) (%__lock_ptr)
+	READ_LOCK_SIZE(cmp) $1, (%__lock_ptr)
+	jns  2f
 1:	rep; nop
 	READ_LOCK_SIZE(cmp) $1, (%__lock_ptr)
 	js	1b
-	LOCK_PREFIX
+2:	LOCK_PREFIX
 	READ_LOCK_SIZE(dec) (%__lock_ptr)
 	js	0b
 	ENDFRAME
-- 
1.7.7.6

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/