[PATCH v2 1/1] x86: In x86-64 barrier_nospec can always be lfence

[David Laight]

When barrier_nospec() was added the defintion was copied from the
one used to synchronise rdtsc.

On very old cpu rdtsc was a synchronising instruction.
When this change X86_FEATURE_LFENCE_RDTSC (and a MFENCE copy) were
(probably) added so lfence/mfence could be added to synchronise rdtsc.
For old cpu (I think the code checks XMM2) no barrier was added.

I'm not sure why that code was used for barrier_nospec().
Or why rdtsc ended up being synchronised by barrier_nospec().
lfence is the right instruction (well as good as you get).

In any case all x86-64 cpu support XMM2 and lfence so there is
to point using alternative().
Separate the 32bit and 64bit definitions but leave the barrier
missing on old 32bit cpu.

Signed-off-by: David Laight <david.laight.linux@xxxxxxxxx>

v2: use a explicit lfence rather than __rmb().
    Update commit message text w.r.t rdtsc.
Signed-off-by: David Laight <david.laight.linux@xxxxxxxxx>
---
 arch/x86/include/asm/barrier.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/barrier.h b/arch/x86/include/asm/barrier.h
index 7b44b3c4cce1..b9af75624cf5 100644
--- a/arch/x86/include/asm/barrier.h
+++ b/arch/x86/include/asm/barrier.h
@@ -45,7 +45,11 @@
 	__mask; })
 
 /* Prevent speculative execution past this barrier. */
-#define barrier_nospec() alternative("", "lfence", X86_FEATURE_LFENCE_RDTSC)
+#ifdef CONFIG_X86_32
+#define barrier_nospec() alternative("", "lfence", X86_FEATURE_XMM2)
+#else
+#define barrier_nospec() asm volatile("lfence":::"memory")
+#endif
 
 #define __dma_rmb()	barrier()
 #define __dma_wmb()	barrier()
-- 
2.39.5