x86: atomic64: Improve atomic64_read()

Linus noticed that the 32-bit version of atomic64_read() was
being overly complex with re-reading the value and doing a
retry loop over that.

Instead we can just rely on cmpxchg8b returning either the new
value or returning the current value.

We can use any 'old' value, which will be faster as it can be
loaded via immediates. Using some value that is not equal to
the real value in memory the instruction gets faster.

This also has the advantage that the CPU could avoid dirtying
the cacheline.

Reported-by: Linus Torvalds <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: David Howells <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Arnd Bergmann <[email protected]>
LKML-Reference: <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>

1 files changed, 2 insertions, 6 deletions

diff --git a/arch/x86/lib/atomic64_32.c b/arch/x86/lib/atomic64_32.c
index d21e725d3d84..afa5d444918b 100644
--- a/arch/x86/lib/atomic64_32.c
+++ b/arch/x86/lib/atomic64_32.c
@@ -67,13 +67,9 @@ void atomic64_set(atomic64_t *ptr, u64 new_val)
  */
 u64 atomic64_read(atomic64_t *ptr)
 {
-	u64 curr_val;
+	u64 old = 1LL << 32;
 
-	do {
-		curr_val = __atomic64_read(ptr);
-	} while (atomic64_cmpxchg(ptr, curr_val, curr_val) != curr_val);
-
-	return curr_val;
+	return cmpxchg8b(&ptr->counter, old, old);
 }
 
 /**