From 64f2dde3f743c8a1ad8c0a1aa74166c1034afd92 Mon Sep 17 00:00:00 2001
From: "David S. Miller" <davem@davemloft.net>
Date: Wed, 29 Oct 2008 21:25:00 -0700
Subject: sparc64: Run the kernel always in the TSO memory model.

The fact of the matter is, all UltraSPARC-III and later chips only
implement TSO.  They don't implement PSO and RMO memory models at all.

Only the Ultra-I and Ultra-II family chips implement RMO and they are
only helped marginally by using this setting when executing kernel
code.

The big plus to doing this is that we can eliminate all of the non-Sync
memory barriers in the kernel except for the ones used in the optimized
memcpy/memset code (these use block load and store operations which
have their own memory ordering rules).

Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/include/asm/spinlock_64.h | 17 ++++++-----------
 arch/sparc64/kernel/etrap.S          |  6 +++---
 arch/sparc64/kernel/rtrap.S          |  6 +++---
 3 files changed, 12 insertions(+), 17 deletions(-)

diff --git a/arch/sparc/include/asm/spinlock_64.h b/arch/sparc/include/asm/spinlock_64.h
index 120cfe4577c..fbac9d00744 100644
--- a/arch/sparc/include/asm/spinlock_64.h
+++ b/arch/sparc/include/asm/spinlock_64.h
@@ -13,17 +13,12 @@
  * and rebuild your kernel.
  */
 
-/* All of these locking primitives are expected to work properly
- * even in an RMO memory model, which currently is what the kernel
- * runs in.
- *
- * There is another issue.  Because we play games to save cycles
- * in the non-contention case, we need to be extra careful about
- * branch targets into the "spinning" code.  They live in their
- * own section, but the newer V9 branches have a shorter range
- * than the traditional 32-bit sparc branch variants.  The rule
- * is that the branches that go into and out of the spinner sections
- * must be pre-V9 branches.
+/* Because we play games to save cycles in the non-contention case, we
+ * need to be extra careful about branch targets into the "spinning"
+ * code.  They live in their own section, but the newer V9 branches
+ * have a shorter range than the traditional 32-bit sparc branch
+ * variants.  The rule is that the branches that go into and out of
+ * the spinner sections must be pre-V9 branches.
  */
 
 #define __raw_spin_is_locked(lp)	((lp)->lock != 0)
diff --git a/arch/sparc64/kernel/etrap.S b/arch/sparc64/kernel/etrap.S
index 29ce489bc18..786b185e6e3 100644
--- a/arch/sparc64/kernel/etrap.S
+++ b/arch/sparc64/kernel/etrap.S
@@ -16,9 +16,9 @@
 #include <asm/mmu.h>
 
 #define		TASK_REGOFF		(THREAD_SIZE-TRACEREG_SZ-STACKFRAME_SZ)
-#define		ETRAP_PSTATE1		(PSTATE_RMO | PSTATE_PRIV)
+#define		ETRAP_PSTATE1		(PSTATE_TSO | PSTATE_PRIV)
 #define		ETRAP_PSTATE2		\
-		(PSTATE_RMO | PSTATE_PEF | PSTATE_PRIV | PSTATE_IE)
+		(PSTATE_TSO | PSTATE_PEF | PSTATE_PRIV | PSTATE_IE)
 
 /*
  * On entry, %g7 is return address - 0x4.
@@ -130,7 +130,7 @@ etrap_save:	save	%g2, -STACK_BIAS, %sp
 		stx	%g6, [%sp + PTREGS_OFF + PT_V9_G6]
 		stx	%g7, [%sp + PTREGS_OFF + PT_V9_G7]
 		or	%l7, %l0, %l7
-		sethi	%hi(TSTATE_RMO | TSTATE_PEF), %l0
+		sethi	%hi(TSTATE_TSO | TSTATE_PEF), %l0
 		or	%l7, %l0, %l7
 		wrpr	%l2, %tnpc
 		wrpr	%l7, (TSTATE_PRIV | TSTATE_IE), %tstate
diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S
index 97a993c1f7f..6892cf556f4 100644
--- a/arch/sparc64/kernel/rtrap.S
+++ b/arch/sparc64/kernel/rtrap.S
@@ -14,9 +14,9 @@
 #include <asm/visasm.h>
 #include <asm/processor.h>
 
-#define		RTRAP_PSTATE		(PSTATE_RMO|PSTATE_PEF|PSTATE_PRIV|PSTATE_IE)
-#define		RTRAP_PSTATE_IRQOFF	(PSTATE_RMO|PSTATE_PEF|PSTATE_PRIV)
-#define		RTRAP_PSTATE_AG_IRQOFF	(PSTATE_RMO|PSTATE_PEF|PSTATE_PRIV|PSTATE_AG)
+#define		RTRAP_PSTATE		(PSTATE_TSO|PSTATE_PEF|PSTATE_PRIV|PSTATE_IE)
+#define		RTRAP_PSTATE_IRQOFF	(PSTATE_TSO|PSTATE_PEF|PSTATE_PRIV)
+#define		RTRAP_PSTATE_AG_IRQOFF	(PSTATE_TSO|PSTATE_PEF|PSTATE_PRIV|PSTATE_AG)
 
 		.text
 		.align			32
-- 
cgit v1.2.3