From b10f997bb0f4e5b34d447f498fb85834a40d3acb Mon Sep 17 00:00:00 2001
From: "David S. Miller" <davem@davemloft.net>
Date: Tue, 17 Aug 2010 21:44:13 -0700
Subject: sparc64: Really fix atomic64_t interface types.

Linus noticed that some of the interface arguments
didn't get "int" --> "long" conversion, as needed.

Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/include/asm/atomic_64.h | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/sparc/include/asm/atomic_64.h b/arch/sparc/include/asm/atomic_64.h
index f0c74227c737..bdb2ff880bdd 100644
--- a/arch/sparc/include/asm/atomic_64.h
+++ b/arch/sparc/include/asm/atomic_64.h
@@ -20,14 +20,14 @@
 #define atomic64_set(v, i)	(((v)->counter) = i)
 
 extern void atomic_add(int, atomic_t *);
-extern void atomic64_add(int, atomic64_t *);
+extern void atomic64_add(long, atomic64_t *);
 extern void atomic_sub(int, atomic_t *);
-extern void atomic64_sub(int, atomic64_t *);
+extern void atomic64_sub(long, atomic64_t *);
 
 extern int atomic_add_ret(int, atomic_t *);
-extern long atomic64_add_ret(int, atomic64_t *);
+extern long atomic64_add_ret(long, atomic64_t *);
 extern int atomic_sub_ret(int, atomic_t *);
-extern long atomic64_sub_ret(int, atomic64_t *);
+extern long atomic64_sub_ret(long, atomic64_t *);
 
 #define atomic_dec_return(v) atomic_sub_ret(1, v)
 #define atomic64_dec_return(v) atomic64_sub_ret(1, v)
-- 
cgit v1.2.3-59-g8ed1b


From 9b3bb86acabe0c05923cea1ed3b0bee9439fef4b Mon Sep 17 00:00:00 2001
From: "David S. Miller" <davem@davemloft.net>
Date: Tue, 17 Aug 2010 22:49:26 -0700
Subject: sparc64: Make rwsems 64-bit.

Basically tip-off the powerpc code, use a 64-bit type and atomic64_t
interfaces for the implementation.

This gets us off of the by-hand asm code I wrote, which frankly I
think probably ruins I-cache hit rates.

The idea was the keep the call chains less deep, but anything taking
the rw-semaphores probably is also calling other stuff and therefore
already has allocated a stack-frame.  So no real stack frame savings
ever.

Ben H. has posted patches to make powerpc use 64-bit too and with some
abstractions we can probably use a shared header file somewhere.

With suggestions from Sam Ravnborg.

Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/include/asm/rwsem-const.h |  12 ---
 arch/sparc/include/asm/rwsem.h       | 120 ++++++++++++++++++++++----
 arch/sparc/lib/Makefile              |   2 +-
 arch/sparc/lib/rwsem_64.S            | 163 -----------------------------------
 4 files changed, 104 insertions(+), 193 deletions(-)
 delete mode 100644 arch/sparc/include/asm/rwsem-const.h
 delete mode 100644 arch/sparc/lib/rwsem_64.S

diff --git a/arch/sparc/include/asm/rwsem-const.h b/arch/sparc/include/asm/rwsem-const.h
deleted file mode 100644
index e4c61a18bb28..000000000000
--- a/arch/sparc/include/asm/rwsem-const.h
+++ /dev/null
@@ -1,12 +0,0 @@
-/* rwsem-const.h: RW semaphore counter constants.  */
-#ifndef _SPARC64_RWSEM_CONST_H
-#define _SPARC64_RWSEM_CONST_H
-
-#define RWSEM_UNLOCKED_VALUE		0x00000000
-#define RWSEM_ACTIVE_BIAS		0x00000001
-#define RWSEM_ACTIVE_MASK		0x0000ffff
-#define RWSEM_WAITING_BIAS		(-0x00010000)
-#define RWSEM_ACTIVE_READ_BIAS		RWSEM_ACTIVE_BIAS
-#define RWSEM_ACTIVE_WRITE_BIAS		(RWSEM_WAITING_BIAS + RWSEM_ACTIVE_BIAS)
-
-#endif /* _SPARC64_RWSEM_CONST_H */
diff --git a/arch/sparc/include/asm/rwsem.h b/arch/sparc/include/asm/rwsem.h
index 6e5621006f85..a2b4302869bc 100644
--- a/arch/sparc/include/asm/rwsem.h
+++ b/arch/sparc/include/asm/rwsem.h
@@ -15,16 +15,21 @@
 
 #include <linux/list.h>
 #include <linux/spinlock.h>
-#include <asm/rwsem-const.h>
 
 struct rwsem_waiter;
 
 struct rw_semaphore {
-	signed int count;
-	spinlock_t		wait_lock;
-	struct list_head	wait_list;
+	signed long			count;
+#define RWSEM_UNLOCKED_VALUE		0x00000000L
+#define RWSEM_ACTIVE_BIAS		0x00000001L
+#define RWSEM_ACTIVE_MASK		0xffffffffL
+#define RWSEM_WAITING_BIAS		(-RWSEM_ACTIVE_MASK-1)
+#define RWSEM_ACTIVE_READ_BIAS		RWSEM_ACTIVE_BIAS
+#define RWSEM_ACTIVE_WRITE_BIAS		(RWSEM_WAITING_BIAS + RWSEM_ACTIVE_BIAS)
+	spinlock_t			wait_lock;
+	struct list_head		wait_list;
 #ifdef CONFIG_DEBUG_LOCK_ALLOC
-	struct lockdep_map	dep_map;
+	struct lockdep_map		dep_map;
 #endif
 };
 
@@ -41,6 +46,11 @@ struct rw_semaphore {
 #define DECLARE_RWSEM(name) \
 	struct rw_semaphore name = __RWSEM_INITIALIZER(name)
 
+extern struct rw_semaphore *rwsem_down_read_failed(struct rw_semaphore *sem);
+extern struct rw_semaphore *rwsem_down_write_failed(struct rw_semaphore *sem);
+extern struct rw_semaphore *rwsem_wake(struct rw_semaphore *sem);
+extern struct rw_semaphore *rwsem_downgrade_wake(struct rw_semaphore *sem);
+
 extern void __init_rwsem(struct rw_semaphore *sem, const char *name,
 			 struct lock_class_key *key);
 
@@ -51,27 +61,103 @@ do {								\
 	__init_rwsem((sem), #sem, &__key);			\
 } while (0)
 
-extern void __down_read(struct rw_semaphore *sem);
-extern int __down_read_trylock(struct rw_semaphore *sem);
-extern void __down_write(struct rw_semaphore *sem);
-extern int __down_write_trylock(struct rw_semaphore *sem);
-extern void __up_read(struct rw_semaphore *sem);
-extern void __up_write(struct rw_semaphore *sem);
-extern void __downgrade_write(struct rw_semaphore *sem);
+/*
+ * lock for reading
+ */
+static inline void __down_read(struct rw_semaphore *sem)
+{
+	if (unlikely(atomic64_inc_return((atomic64_t *)(&sem->count)) <= 0L))
+		rwsem_down_read_failed(sem);
+}
+
+static inline int __down_read_trylock(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	while ((tmp = sem->count) >= 0L) {
+		if (tmp == cmpxchg(&sem->count, tmp,
+				   tmp + RWSEM_ACTIVE_READ_BIAS)) {
+			return 1;
+		}
+	}
+	return 0;
+}
 
+/*
+ * lock for writing
+ */
 static inline void __down_write_nested(struct rw_semaphore *sem, int subclass)
 {
-	__down_write(sem);
+	long tmp;
+
+	tmp = atomic64_add_return(RWSEM_ACTIVE_WRITE_BIAS,
+				  (atomic64_t *)(&sem->count));
+	if (unlikely(tmp != RWSEM_ACTIVE_WRITE_BIAS))
+		rwsem_down_write_failed(sem);
 }
 
-static inline int rwsem_atomic_update(int delta, struct rw_semaphore *sem)
+static inline void __down_write(struct rw_semaphore *sem)
 {
-	return atomic_add_return(delta, (atomic_t *)(&sem->count));
+	__down_write_nested(sem, 0);
+}
+
+static inline int __down_write_trylock(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	tmp = cmpxchg(&sem->count, RWSEM_UNLOCKED_VALUE,
+		      RWSEM_ACTIVE_WRITE_BIAS);
+	return tmp == RWSEM_UNLOCKED_VALUE;
 }
 
-static inline void rwsem_atomic_add(int delta, struct rw_semaphore *sem)
+/*
+ * unlock after reading
+ */
+static inline void __up_read(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	tmp = atomic64_dec_return((atomic64_t *)(&sem->count));
+	if (unlikely(tmp < -1L && (tmp & RWSEM_ACTIVE_MASK) == 0L))
+		rwsem_wake(sem);
+}
+
+/*
+ * unlock after writing
+ */
+static inline void __up_write(struct rw_semaphore *sem)
+{
+	if (unlikely(atomic64_sub_return(RWSEM_ACTIVE_WRITE_BIAS,
+					 (atomic64_t *)(&sem->count)) < 0L))
+		rwsem_wake(sem);
+}
+
+/*
+ * implement atomic add functionality
+ */
+static inline void rwsem_atomic_add(long delta, struct rw_semaphore *sem)
+{
+	atomic64_add(delta, (atomic64_t *)(&sem->count));
+}
+
+/*
+ * downgrade write lock to read lock
+ */
+static inline void __downgrade_write(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	tmp = atomic64_add_return(-RWSEM_WAITING_BIAS, (atomic64_t *)(&sem->count));
+	if (tmp < 0L)
+		rwsem_downgrade_wake(sem);
+}
+
+/*
+ * implement exchange and add functionality
+ */
+static inline long rwsem_atomic_update(long delta, struct rw_semaphore *sem)
 {
-	atomic_add(delta, (atomic_t *)(&sem->count));
+	return atomic64_add_return(delta, (atomic64_t *)(&sem->count));
 }
 
 static inline int rwsem_is_locked(struct rw_semaphore *sem)
diff --git a/arch/sparc/lib/Makefile b/arch/sparc/lib/Makefile
index c4b5e03af115..846d1c4374ea 100644
--- a/arch/sparc/lib/Makefile
+++ b/arch/sparc/lib/Makefile
@@ -15,7 +15,7 @@ lib-$(CONFIG_SPARC32) += divdi3.o udivdi3.o
 lib-$(CONFIG_SPARC32) += copy_user.o locks.o
 lib-y                 += atomic_$(BITS).o
 lib-$(CONFIG_SPARC32) += lshrdi3.o ashldi3.o
-lib-y                 += rwsem_$(BITS).o
+lib-$(CONFIG_SPARC32) += rwsem_32.o
 lib-$(CONFIG_SPARC32) += muldi3.o bitext.o cmpdi2.o
 
 lib-$(CONFIG_SPARC64) += copy_page.o clear_page.o bzero.o
diff --git a/arch/sparc/lib/rwsem_64.S b/arch/sparc/lib/rwsem_64.S
deleted file mode 100644
index 91a7d29a79d5..000000000000
--- a/arch/sparc/lib/rwsem_64.S
+++ /dev/null
@@ -1,163 +0,0 @@
-/* rwsem.S: RW semaphore assembler.
- *
- * Written by David S. Miller (davem@redhat.com), 2001.
- * Derived from asm-i386/rwsem.h
- */
-
-#include <asm/rwsem-const.h>
-
-	.section	.sched.text, "ax"
-
-	.globl		__down_read
-__down_read:
-1:	lduw		[%o0], %g1
-	add		%g1, 1, %g7
-	cas		[%o0], %g1, %g7
-	cmp		%g1, %g7
-	bne,pn		%icc, 1b
-	 add		%g7, 1, %g7
-	cmp		%g7, 0
-	bl,pn		%icc, 3f
-	 nop
-2:
-	retl
-	 nop
-3:
-	save		%sp, -192, %sp
-	call		rwsem_down_read_failed
-	 mov		%i0, %o0
-	ret
-	 restore
-	.size		__down_read, .-__down_read
-
-	.globl		__down_read_trylock
-__down_read_trylock:
-1:	lduw		[%o0], %g1
-	add		%g1, 1, %g7
-	cmp		%g7, 0
-	bl,pn		%icc, 2f
-	 mov		0, %o1
-	cas		[%o0], %g1, %g7
-	cmp		%g1, %g7
-	bne,pn		%icc, 1b
-	 mov		1, %o1
-2:	retl
-	 mov		%o1, %o0
-	.size		__down_read_trylock, .-__down_read_trylock
-
-	.globl		__down_write
-__down_write:
-	sethi		%hi(RWSEM_ACTIVE_WRITE_BIAS), %g1
-	or		%g1, %lo(RWSEM_ACTIVE_WRITE_BIAS), %g1
-1:
-	lduw		[%o0], %g3
-	add		%g3, %g1, %g7
-	cas		[%o0], %g3, %g7
-	cmp		%g3, %g7
-	bne,pn		%icc, 1b
-	 cmp		%g7, 0
-	bne,pn		%icc, 3f
-	 nop
-2:	retl
-	 nop
-3:
-	save		%sp, -192, %sp
-	call		rwsem_down_write_failed
-	 mov		%i0, %o0
-	ret
-	 restore
-	.size		__down_write, .-__down_write
-
-	.globl		__down_write_trylock
-__down_write_trylock:
-	sethi		%hi(RWSEM_ACTIVE_WRITE_BIAS), %g1
-	or		%g1, %lo(RWSEM_ACTIVE_WRITE_BIAS), %g1
-1:
-	lduw		[%o0], %g3
-	cmp		%g3, 0
-	bne,pn		%icc, 2f
-	 mov		0, %o1
-	add		%g3, %g1, %g7
-	cas		[%o0], %g3, %g7
-	cmp		%g3, %g7
-	bne,pn		%icc, 1b
-	 mov		1, %o1
-2:	retl
-	 mov		%o1, %o0
-	.size		__down_write_trylock, .-__down_write_trylock
-
-	.globl		__up_read
-__up_read:
-1:
-	lduw		[%o0], %g1
-	sub		%g1, 1, %g7
-	cas		[%o0], %g1, %g7
-	cmp		%g1, %g7
-	bne,pn		%icc, 1b
-	 cmp		%g7, 0
-	bl,pn		%icc, 3f
-	 nop
-2:	retl
-	 nop
-3:	sethi		%hi(RWSEM_ACTIVE_MASK), %g1
-	sub		%g7, 1, %g7
-	or		%g1, %lo(RWSEM_ACTIVE_MASK), %g1
-	andcc		%g7, %g1, %g0
-	bne,pn		%icc, 2b
-	 nop
-	save		%sp, -192, %sp
-	call		rwsem_wake
-	 mov		%i0, %o0
-	ret
-	 restore
-	.size		__up_read, .-__up_read
-
-	.globl		__up_write
-__up_write:
-	sethi		%hi(RWSEM_ACTIVE_WRITE_BIAS), %g1
-	or		%g1, %lo(RWSEM_ACTIVE_WRITE_BIAS), %g1
-1:
-	lduw		[%o0], %g3
-	sub		%g3, %g1, %g7
-	cas		[%o0], %g3, %g7
-	cmp		%g3, %g7
-	bne,pn		%icc, 1b
-	 sub		%g7, %g1, %g7
-	cmp		%g7, 0
-	bl,pn		%icc, 3f
-	 nop
-2:
-	retl
-	 nop
-3:
-	save		%sp, -192, %sp
-	call		rwsem_wake
-	 mov		%i0, %o0
-	ret
-	 restore
-	.size		__up_write, .-__up_write
-
-	.globl		__downgrade_write
-__downgrade_write:
-	sethi		%hi(RWSEM_WAITING_BIAS), %g1
-	or		%g1, %lo(RWSEM_WAITING_BIAS), %g1
-1:
-	lduw		[%o0], %g3
-	sub		%g3, %g1, %g7
-	cas		[%o0], %g3, %g7
-	cmp		%g3, %g7
-	bne,pn		%icc, 1b
-	 sub		%g7, %g1, %g7
-	cmp		%g7, 0
-	bl,pn		%icc, 3f
-	 nop
-2:
-	retl
-	 nop
-3:
-	save		%sp, -192, %sp
-	call		rwsem_downgrade_wake
-	 mov		%i0, %o0
-	ret
-	 restore
-	.size		__downgrade_write, .-__downgrade_write
-- 
cgit v1.2.3-59-g8ed1b


From 6ec274750c99448c3412bbc10c97ce0c993f8a4e Mon Sep 17 00:00:00 2001
From: Mikulas Patocka <mpatocka@redhat.com>
Date: Wed, 18 Aug 2010 08:03:37 +0000
Subject: sparc64: simple microoptimizations for atomic functions

Simple microoptimizations for sparc64 atomic functions:
Save one instruction by using a delay slot.
Use %g1 instead of %g7, because %g1 is written earlier.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/lib/atomic_64.S | 20 ++++++++------------
 1 file changed, 8 insertions(+), 12 deletions(-)

diff --git a/arch/sparc/lib/atomic_64.S b/arch/sparc/lib/atomic_64.S
index 0268210ca168..703c9c3935b7 100644
--- a/arch/sparc/lib/atomic_64.S
+++ b/arch/sparc/lib/atomic_64.S
@@ -52,10 +52,9 @@ atomic_add_ret: /* %o0 = increment, %o1 = atomic_ptr */
 	cas	[%o1], %g1, %g7
 	cmp	%g1, %g7
 	bne,pn	%icc, 2f
-	 add	%g7, %o0, %g7
-	sra	%g7, 0, %o0
+	 add	%g1, %o0, %g1
 	retl
-	 nop
+	 sra	%g1, 0, %o0
 2:	BACKOFF_SPIN(%o2, %o3, 1b)
 	.size	atomic_add_ret, .-atomic_add_ret
 
@@ -68,10 +67,9 @@ atomic_sub_ret: /* %o0 = decrement, %o1 = atomic_ptr */
 	cas	[%o1], %g1, %g7
 	cmp	%g1, %g7
 	bne,pn	%icc, 2f
-	 sub	%g7, %o0, %g7
-	sra	%g7, 0, %o0
+	 sub	%g1, %o0, %g1
 	retl
-	 nop
+	 sra	%g1, 0, %o0
 2:	BACKOFF_SPIN(%o2, %o3, 1b)
 	.size	atomic_sub_ret, .-atomic_sub_ret
 
@@ -114,10 +112,9 @@ atomic64_add_ret: /* %o0 = increment, %o1 = atomic_ptr */
 	casx	[%o1], %g1, %g7
 	cmp	%g1, %g7
 	bne,pn	%xcc, 2f
-	 add	%g7, %o0, %g7
-	mov	%g7, %o0
-	retl
 	 nop
+	retl
+	 add	%g1, %o0, %o0
 2:	BACKOFF_SPIN(%o2, %o3, 1b)
 	.size	atomic64_add_ret, .-atomic64_add_ret
 
@@ -130,9 +127,8 @@ atomic64_sub_ret: /* %o0 = decrement, %o1 = atomic_ptr */
 	casx	[%o1], %g1, %g7
 	cmp	%g1, %g7
 	bne,pn	%xcc, 2f
-	 sub	%g7, %o0, %g7
-	mov	%g7, %o0
-	retl
 	 nop
+	retl
+	 sub	%g1, %o0, %o0
 2:	BACKOFF_SPIN(%o2, %o3, 1b)
 	.size	atomic64_sub_ret, .-atomic64_sub_ret
-- 
cgit v1.2.3-59-g8ed1b


From 0f58189d4a3ca96d7959501ecb203177efdbc5bd Mon Sep 17 00:00:00 2001
From: "David S. Miller" <davem@davemloft.net>
Date: Wed, 18 Aug 2010 22:53:26 -0700
Subject: sparc64: Make lock backoff really a NOP on UP builds.

As noticed by Mikulas Patocka, the backoff macros don't
completely nop out for UP builds, we still get a
branch always and a delay slot nop.

Fix this by making the branch to the backoff spin loop
selective, then we can nop out the spin loop completely.

Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/include/asm/backoff.h | 11 ++++++++---
 arch/sparc/lib/atomic_64.S       | 16 ++++++++--------
 arch/sparc/lib/bitops.S          | 12 ++++++------
 3 files changed, 22 insertions(+), 17 deletions(-)

diff --git a/arch/sparc/include/asm/backoff.h b/arch/sparc/include/asm/backoff.h
index fa1fdf67e350..db3af0d30fb1 100644
--- a/arch/sparc/include/asm/backoff.h
+++ b/arch/sparc/include/asm/backoff.h
@@ -8,6 +8,9 @@
 #define BACKOFF_SETUP(reg)	\
 	mov	1, reg
 
+#define BACKOFF_LABEL(spin_label, continue_label) \
+	spin_label
+
 #define BACKOFF_SPIN(reg, tmp, label)	\
 	mov	reg, tmp; \
 88:	brnz,pt	tmp, 88b; \
@@ -22,9 +25,11 @@
 #else
 
 #define BACKOFF_SETUP(reg)
-#define BACKOFF_SPIN(reg, tmp, label) \
-	ba,pt	%xcc, label; \
-	 nop;
+
+#define BACKOFF_LABEL(spin_label, continue_label) \
+	continue_label
+
+#define BACKOFF_SPIN(reg, tmp, label)
 
 #endif
 
diff --git a/arch/sparc/lib/atomic_64.S b/arch/sparc/lib/atomic_64.S
index 703c9c3935b7..59186e0fcf39 100644
--- a/arch/sparc/lib/atomic_64.S
+++ b/arch/sparc/lib/atomic_64.S
@@ -21,7 +21,7 @@ atomic_add: /* %o0 = increment, %o1 = atomic_ptr */
 	add	%g1, %o0, %g7
 	cas	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%icc, 2f
+	bne,pn	%icc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 nop
@@ -36,7 +36,7 @@ atomic_sub: /* %o0 = decrement, %o1 = atomic_ptr */
 	sub	%g1, %o0, %g7
 	cas	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%icc, 2f
+	bne,pn	%icc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 nop
@@ -51,7 +51,7 @@ atomic_add_ret: /* %o0 = increment, %o1 = atomic_ptr */
 	add	%g1, %o0, %g7
 	cas	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%icc, 2f
+	bne,pn	%icc, BACKOFF_LABEL(2f, 1b)
 	 add	%g1, %o0, %g1
 	retl
 	 sra	%g1, 0, %o0
@@ -66,7 +66,7 @@ atomic_sub_ret: /* %o0 = decrement, %o1 = atomic_ptr */
 	sub	%g1, %o0, %g7
 	cas	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%icc, 2f
+	bne,pn	%icc, BACKOFF_LABEL(2f, 1b)
 	 sub	%g1, %o0, %g1
 	retl
 	 sra	%g1, 0, %o0
@@ -81,7 +81,7 @@ atomic64_add: /* %o0 = increment, %o1 = atomic_ptr */
 	add	%g1, %o0, %g7
 	casx	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 nop
@@ -96,7 +96,7 @@ atomic64_sub: /* %o0 = decrement, %o1 = atomic_ptr */
 	sub	%g1, %o0, %g7
 	casx	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 nop
@@ -111,7 +111,7 @@ atomic64_add_ret: /* %o0 = increment, %o1 = atomic_ptr */
 	add	%g1, %o0, %g7
 	casx	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 add	%g1, %o0, %o0
@@ -126,7 +126,7 @@ atomic64_sub_ret: /* %o0 = decrement, %o1 = atomic_ptr */
 	sub	%g1, %o0, %g7
 	casx	[%o1], %g1, %g7
 	cmp	%g1, %g7
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 sub	%g1, %o0, %o0
diff --git a/arch/sparc/lib/bitops.S b/arch/sparc/lib/bitops.S
index 2b7228cb8c22..3dc61d5537c0 100644
--- a/arch/sparc/lib/bitops.S
+++ b/arch/sparc/lib/bitops.S
@@ -22,7 +22,7 @@ test_and_set_bit:	/* %o0=nr, %o1=addr */
 	or	%g7, %o2, %g1
 	casx	[%o1], %g7, %g1
 	cmp	%g7, %g1
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 and	%g7, %o2, %g2
 	clr	%o0
 	movrne	%g2, 1, %o0
@@ -45,7 +45,7 @@ test_and_clear_bit:	/* %o0=nr, %o1=addr */
 	andn	%g7, %o2, %g1
 	casx	[%o1], %g7, %g1
 	cmp	%g7, %g1
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 and	%g7, %o2, %g2
 	clr	%o0
 	movrne	%g2, 1, %o0
@@ -68,7 +68,7 @@ test_and_change_bit:	/* %o0=nr, %o1=addr */
 	xor	%g7, %o2, %g1
 	casx	[%o1], %g7, %g1
 	cmp	%g7, %g1
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 and	%g7, %o2, %g2
 	clr	%o0
 	movrne	%g2, 1, %o0
@@ -91,7 +91,7 @@ set_bit:		/* %o0=nr, %o1=addr */
 	or	%g7, %o2, %g1
 	casx	[%o1], %g7, %g1
 	cmp	%g7, %g1
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 nop
@@ -112,7 +112,7 @@ clear_bit:		/* %o0=nr, %o1=addr */
 	andn	%g7, %o2, %g1
 	casx	[%o1], %g7, %g1
 	cmp	%g7, %g1
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 nop
@@ -133,7 +133,7 @@ change_bit:		/* %o0=nr, %o1=addr */
 	xor	%g7, %o2, %g1
 	casx	[%o1], %g7, %g1
 	cmp	%g7, %g1
-	bne,pn	%xcc, 2f
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
 	 nop
 	retl
 	 nop
-- 
cgit v1.2.3-59-g8ed1b


From 019408f9b89c68cd7b8ddb904960dc17ccf7e531 Mon Sep 17 00:00:00 2001
From: Mikulas Patocka <mpatocka@redhat.com>
Date: Thu, 19 Aug 2010 14:15:32 -0700
Subject: sparc64: Fill a missing delay slot.

If the code were already aligned to 64 bytes, wr instruction would be executed
twice --- once in delay slot and once in the jump target.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/include/asm/system_64.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/sparc/include/asm/system_64.h b/arch/sparc/include/asm/system_64.h
index d24cfe16afc1..e3b65d8cf41b 100644
--- a/arch/sparc/include/asm/system_64.h
+++ b/arch/sparc/include/asm/system_64.h
@@ -106,6 +106,7 @@ do {	__asm__ __volatile__("ba,pt	%%xcc, 1f\n\t" \
  */
 #define write_pic(__p)  					\
 	__asm__ __volatile__("ba,pt	%%xcc, 99f\n\t"		\
+			     " nop\n\t"				\
 			     ".align	64\n"			\
 			  "99:wr	%0, 0x0, %%pic\n\t"	\
 			     "rd	%%pic, %%g0" : : "r" (__p))
-- 
cgit v1.2.3-59-g8ed1b


From 25edd6946a1d74e5e77813c2324a0908c68bcf9e Mon Sep 17 00:00:00 2001
From: "David S. Miller" <davem@davemloft.net>
Date: Mon, 23 Aug 2010 23:10:57 -0700
Subject: sparc64: Get rid of indirect p1275 PROM call buffer.

This is based upon a report by Meelis Roos showing that it's possible
that we'll try to fetch a property that is 32K in size with some
devices.  With the current fixed 3K buffer we use for moving data in
and out of the firmware during PROM calls, that simply won't work.

In fact, it will scramble random kernel data during bootup.

The reasoning behind the temporary buffer is entirely historical.  It
used to be the case that we had problems referencing dynamic kernel
memory (including the stack) early in the boot process before we
explicitly told the firwmare to switch us over to the kernel trap
table.

So what we did was always give the firmware buffers that were locked
into the main kernel image.

But we no longer have problems like that, so get rid of all of this
indirect bounce buffering.

Besides fixing Meelis's bug, this also makes the kernel data about 3K
smaller.

It was also discovered during these conversions that the
implementation of prom_retain() was completely wrong, so that was
fixed here as well.  Currently that interface is not in use.

Reported-by: Meelis Roos <mroos@linux.ee>
Tested-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 arch/sparc/include/asm/oplib_64.h |  27 +---
 arch/sparc/prom/cif.S             |  16 +-
 arch/sparc/prom/console_64.c      |  48 ++++--
 arch/sparc/prom/devops_64.c       |  36 ++++-
 arch/sparc/prom/misc_64.c         | 314 ++++++++++++++++++++++++++------------
 arch/sparc/prom/p1275.c           | 102 +------------
 arch/sparc/prom/tree_64.c         | 210 ++++++++++++++++++-------
 7 files changed, 456 insertions(+), 297 deletions(-)

diff --git a/arch/sparc/include/asm/oplib_64.h b/arch/sparc/include/asm/oplib_64.h
index a5db0317b5fb..3e0b2d62303d 100644
--- a/arch/sparc/include/asm/oplib_64.h
+++ b/arch/sparc/include/asm/oplib_64.h
@@ -185,9 +185,8 @@ extern int prom_getunumber(int syndrome_code,
 			   char *buf, int buflen);
 
 /* Retain physical memory to the caller across soft resets. */
-extern unsigned long prom_retain(const char *name,
-				 unsigned long pa_low, unsigned long pa_high,
-				 long size, long align);
+extern int prom_retain(const char *name, unsigned long size,
+		       unsigned long align, unsigned long *paddr);
 
 /* Load explicit I/D TLB entries into the calling processor. */
 extern long prom_itlb_load(unsigned long index,
@@ -287,26 +286,6 @@ extern void prom_sun4v_guest_soft_state(void);
 extern int prom_ihandle2path(int handle, char *buffer, int bufsize);
 
 /* Client interface level routines. */
-extern long p1275_cmd(const char *, long, ...);
-
-#if 0
-#define P1275_SIZE(x) ((((long)((x) / 32)) << 32) | (x))
-#else
-#define P1275_SIZE(x) x
-#endif
-
-/* We support at most 16 input and 1 output argument */
-#define P1275_ARG_NUMBER		0
-#define P1275_ARG_IN_STRING		1
-#define P1275_ARG_OUT_BUF		2
-#define P1275_ARG_OUT_32B		3
-#define P1275_ARG_IN_FUNCTION		4
-#define P1275_ARG_IN_BUF		5
-#define P1275_ARG_IN_64B		6
-
-#define P1275_IN(x) ((x) & 0xf)
-#define P1275_OUT(x) (((x) << 4) & 0xf0)
-#define P1275_INOUT(i,o) (P1275_IN(i)|P1275_OUT(o))
-#define P1275_ARG(n,x) ((x) << ((n)*3 + 8))
+extern void p1275_cmd_direct(unsigned long *);
 
 #endif /* !(__SPARC64_OPLIB_H) */
diff --git a/arch/sparc/prom/cif.S b/arch/sparc/prom/cif.S
index 5f27ad779c0c..9c86b4b7d429 100644
--- a/arch/sparc/prom/cif.S
+++ b/arch/sparc/prom/cif.S
@@ -9,18 +9,18 @@
 #include <asm/thread_info.h>
 
 	.text
-	.globl	prom_cif_interface
-prom_cif_interface:
-	sethi	%hi(p1275buf), %o0
-	or	%o0, %lo(p1275buf), %o0
-	ldx	[%o0 + 0x010], %o1	! prom_cif_stack
-	save	%o1, -192, %sp
-	ldx	[%i0 + 0x008], %l2	! prom_cif_handler
+	.globl	prom_cif_direct
+prom_cif_direct:
+	sethi	%hi(p1275buf), %o1
+	or	%o1, %lo(p1275buf), %o1
+	ldx	[%o1 + 0x0010], %o2	! prom_cif_stack
+	save	%o2, -192, %sp
+	ldx	[%i1 + 0x0008], %l2	! prom_cif_handler
 	mov	%g4, %l0
 	mov	%g5, %l1
 	mov	%g6, %l3
 	call	%l2
-	 add	%i0, 0x018, %o0		! prom_args
+	 mov	%i0, %o0		! prom_args
 	mov	%l0, %g4
 	mov	%l1, %g5
 	mov	%l3, %g6
diff --git a/arch/sparc/prom/console_64.c b/arch/sparc/prom/console_64.c
index f55d58a8a156..10322dc2f557 100644
--- a/arch/sparc/prom/console_64.c
+++ b/arch/sparc/prom/console_64.c
@@ -21,14 +21,22 @@ extern int prom_stdin, prom_stdout;
 inline int
 prom_nbgetchar(void)
 {
+	unsigned long args[7];
 	char inc;
 
-	if (p1275_cmd("read", P1275_ARG(1,P1275_ARG_OUT_BUF)|
-			      P1275_INOUT(3,1),
-			      prom_stdin, &inc, P1275_SIZE(1)) == 1)
+	args[0] = (unsigned long) "read";
+	args[1] = 3;
+	args[2] = 1;
+	args[3] = (unsigned int) prom_stdin;
+	args[4] = (unsigned long) &inc;
+	args[5] = 1;
+	args[6] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	if (args[6] == 1)
 		return inc;
-	else
-		return -1;
+	return -1;
 }
 
 /* Non blocking put character to console device, returns -1 if
@@ -37,12 +45,22 @@ prom_nbgetchar(void)
 inline int
 prom_nbputchar(char c)
 {
+	unsigned long args[7];
 	char outc;
 	
 	outc = c;
-	if (p1275_cmd("write", P1275_ARG(1,P1275_ARG_IN_BUF)|
-			       P1275_INOUT(3,1),
-			       prom_stdout, &outc, P1275_SIZE(1)) == 1)
+
+	args[0] = (unsigned long) "write";
+	args[1] = 3;
+	args[2] = 1;
+	args[3] = (unsigned int) prom_stdout;
+	args[4] = (unsigned long) &outc;
+	args[5] = 1;
+	args[6] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	if (args[6] == 1)
 		return 0;
 	else
 		return -1;
@@ -67,7 +85,15 @@ prom_putchar(char c)
 void
 prom_puts(const char *s, int len)
 {
-	p1275_cmd("write", P1275_ARG(1,P1275_ARG_IN_BUF)|
-			   P1275_INOUT(3,1),
-			   prom_stdout, s, P1275_SIZE(len));
+	unsigned long args[7];
+
+	args[0] = (unsigned long) "write";
+	args[1] = 3;
+	args[2] = 1;
+	args[3] = (unsigned int) prom_stdout;
+	args[4] = (unsigned long) s;
+	args[5] = len;
+	args[6] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
 }
diff --git a/arch/sparc/prom/devops_64.c b/arch/sparc/prom/devops_64.c
index 9dbd803e46e1..a017119e7ef1 100644
--- a/arch/sparc/prom/devops_64.c
+++ b/arch/sparc/prom/devops_64.c
@@ -18,16 +18,32 @@
 int
 prom_devopen(const char *dstr)
 {
-	return p1275_cmd ("open", P1275_ARG(0,P1275_ARG_IN_STRING)|
-				  P1275_INOUT(1,1),
-				  dstr);
+	unsigned long args[5];
+
+	args[0] = (unsigned long) "open";
+	args[1] = 1;
+	args[2] = 1;
+	args[3] = (unsigned long) dstr;
+	args[4] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[4];
 }
 
 /* Close the device described by device handle 'dhandle'. */
 int
 prom_devclose(int dhandle)
 {
-	p1275_cmd ("close", P1275_INOUT(1,0), dhandle);
+	unsigned long args[4];
+
+	args[0] = (unsigned long) "close";
+	args[1] = 1;
+	args[2] = 0;
+	args[3] = (unsigned int) dhandle;
+
+	p1275_cmd_direct(args);
+
 	return 0;
 }
 
@@ -37,5 +53,15 @@ prom_devclose(int dhandle)
 void
 prom_seek(int dhandle, unsigned int seekhi, unsigned int seeklo)
 {
-	p1275_cmd ("seek", P1275_INOUT(3,1), dhandle, seekhi, seeklo);
+	unsigned long args[7];
+
+	args[0] = (unsigned long) "seek";
+	args[1] = 3;
+	args[2] = 1;
+	args[3] = (unsigned int) dhandle;
+	args[4] = seekhi;
+	args[5] = seeklo;
+	args[6] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
 }
diff --git a/arch/sparc/prom/misc_64.c b/arch/sparc/prom/misc_64.c
index 39fc6af21b7c..6cb1581d6aef 100644
--- a/arch/sparc/prom/misc_64.c
+++ b/arch/sparc/prom/misc_64.c
@@ -20,10 +20,17 @@
 
 int prom_service_exists(const char *service_name)
 {
-	int err = p1275_cmd("test", P1275_ARG(0, P1275_ARG_IN_STRING) |
-			    P1275_INOUT(1, 1), service_name);
+	unsigned long args[5];
 
-	if (err)
+	args[0] = (unsigned long) "test";
+	args[1] = 1;
+	args[2] = 1;
+	args[3] = (unsigned long) service_name;
+	args[4] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	if (args[4])
 		return 0;
 	return 1;
 }
@@ -31,30 +38,47 @@ int prom_service_exists(const char *service_name)
 void prom_sun4v_guest_soft_state(void)
 {
 	const char *svc = "SUNW,soft-state-supported";
+	unsigned long args[3];
 
 	if (!prom_service_exists(svc))
 		return;
-	p1275_cmd(svc, P1275_INOUT(0, 0));
+	args[0] = (unsigned long) svc;
+	args[1] = 0;
+	args[2] = 0;
+	p1275_cmd_direct(args);
 }
 
 /* Reset and reboot the machine with the command 'bcommand'. */
 void prom_reboot(const char *bcommand)
 {
+	unsigned long args[4];
+
 #ifdef CONFIG_SUN_LDOMS
 	if (ldom_domaining_enabled)
 		ldom_reboot(bcommand);
 #endif
-	p1275_cmd("boot", P1275_ARG(0, P1275_ARG_IN_STRING) |
-		  P1275_INOUT(1, 0), bcommand);
+	args[0] = (unsigned long) "boot";
+	args[1] = 1;
+	args[2] = 0;
+	args[3] = (unsigned long) bcommand;
+
+	p1275_cmd_direct(args);
 }
 
 /* Forth evaluate the expression contained in 'fstring'. */
 void prom_feval(const char *fstring)
 {
+	unsigned long args[5];
+
 	if (!fstring || fstring[0] == 0)
 		return;
-	p1275_cmd("interpret", P1275_ARG(0, P1275_ARG_IN_STRING) |
-		  P1275_INOUT(1, 1), fstring);
+	args[0] = (unsigned long) "interpret";
+	args[1] = 1;
+	args[2] = 1;
+	args[3] = (unsigned long) fstring;
+	args[4] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
 }
 EXPORT_SYMBOL(prom_feval);
 
@@ -68,6 +92,7 @@ extern void smp_release(void);
  */
 void prom_cmdline(void)
 {
+	unsigned long args[3];
 	unsigned long flags;
 
 	local_irq_save(flags);
@@ -76,7 +101,11 @@ void prom_cmdline(void)
 	smp_capture();
 #endif
 
-	p1275_cmd("enter", P1275_INOUT(0, 0));
+	args[0] = (unsigned long) "enter";
+	args[1] = 0;
+	args[2] = 0;
+
+	p1275_cmd_direct(args);
 
 #ifdef CONFIG_SMP
 	smp_release();
@@ -90,22 +119,32 @@ void prom_cmdline(void)
  */
 void notrace prom_halt(void)
 {
+	unsigned long args[3];
+
 #ifdef CONFIG_SUN_LDOMS
 	if (ldom_domaining_enabled)
 		ldom_power_off();
 #endif
 again:
-	p1275_cmd("exit", P1275_INOUT(0, 0));
+	args[0] = (unsigned long) "exit";
+	args[1] = 0;
+	args[2] = 0;
+	p1275_cmd_direct(args);
 	goto again; /* PROM is out to get me -DaveM */
 }
 
 void prom_halt_power_off(void)
 {
+	unsigned long args[3];
+
 #ifdef CONFIG_SUN_LDOMS
 	if (ldom_domaining_enabled)
 		ldom_power_off();
 #endif
-	p1275_cmd("SUNW,power-off", P1275_INOUT(0, 0));
+	args[0] = (unsigned long) "SUNW,power-off";
+	args[1] = 0;
+	args[2] = 0;
+	p1275_cmd_direct(args);
 
 	/* if nothing else helps, we just halt */
 	prom_halt();
@@ -114,10 +153,15 @@ void prom_halt_power_off(void)
 /* Set prom sync handler to call function 'funcp'. */
 void prom_setcallback(callback_func_t funcp)
 {
+	unsigned long args[5];
 	if (!funcp)
 		return;
-	p1275_cmd("set-callback", P1275_ARG(0, P1275_ARG_IN_FUNCTION) |
-		  P1275_INOUT(1, 1), funcp);
+	args[0] = (unsigned long) "set-callback";
+	args[1] = 1;
+	args[2] = 1;
+	args[3] = (unsigned long) funcp;
+	args[4] = (unsigned long) -1;
+	p1275_cmd_direct(args);
 }
 
 /* Get the idprom and stuff it into buffer 'idbuf'.  Returns the
@@ -173,57 +217,61 @@ static int prom_get_memory_ihandle(void)
 }
 
 /* Load explicit I/D TLB entries. */
+static long tlb_load(const char *type, unsigned long index,
+		     unsigned long tte_data, unsigned long vaddr)
+{
+	unsigned long args[9];
+
+	args[0] = (unsigned long) prom_callmethod_name;
+	args[1] = 5;
+	args[2] = 1;
+	args[3] = (unsigned long) type;
+	args[4] = (unsigned int) prom_get_mmu_ihandle();
+	args[5] = vaddr;
+	args[6] = tte_data;
+	args[7] = index;
+	args[8] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (long) args[8];
+}
+
 long prom_itlb_load(unsigned long index,
 		    unsigned long tte_data,
 		    unsigned long vaddr)
 {
-	return p1275_cmd(prom_callmethod_name,
-			 (P1275_ARG(0, P1275_ARG_IN_STRING) |
-			  P1275_ARG(2, P1275_ARG_IN_64B) |
-			  P1275_ARG(3, P1275_ARG_IN_64B) |
-			  P1275_INOUT(5, 1)),
-			 "SUNW,itlb-load",
-			 prom_get_mmu_ihandle(),
-			 /* And then our actual args are pushed backwards. */
-			 vaddr,
-			 tte_data,
-			 index);
+	return tlb_load("SUNW,itlb-load", index, tte_data, vaddr);
 }
 
 long prom_dtlb_load(unsigned long index,
 		    unsigned long tte_data,
 		    unsigned long vaddr)
 {
-	return p1275_cmd(prom_callmethod_name,
-			 (P1275_ARG(0, P1275_ARG_IN_STRING) |
-			  P1275_ARG(2, P1275_ARG_IN_64B) |
-			  P1275_ARG(3, P1275_ARG_IN_64B) |
-			  P1275_INOUT(5, 1)),
-			 "SUNW,dtlb-load",
-			 prom_get_mmu_ihandle(),
-			 /* And then our actual args are pushed backwards. */
-			 vaddr,
-			 tte_data,
-			 index);
+	return tlb_load("SUNW,dtlb-load", index, tte_data, vaddr);
 }
 
 int prom_map(int mode, unsigned long size,
 	     unsigned long vaddr, unsigned long paddr)
 {
-	int ret = p1275_cmd(prom_callmethod_name,
-			    (P1275_ARG(0, P1275_ARG_IN_STRING) |
-			     P1275_ARG(3, P1275_ARG_IN_64B) |
-			     P1275_ARG(4, P1275_ARG_IN_64B) |
-			     P1275_ARG(6, P1275_ARG_IN_64B) |
-			     P1275_INOUT(7, 1)),
-			    prom_map_name,
-			    prom_get_mmu_ihandle(),
-			    mode,
-			    size,
-			    vaddr,
-			    0,
-			    paddr);
-
+	unsigned long args[11];
+	int ret;
+
+	args[0] = (unsigned long) prom_callmethod_name;
+	args[1] = 7;
+	args[2] = 1;
+	args[3] = (unsigned long) prom_map_name;
+	args[4] = (unsigned int) prom_get_mmu_ihandle();
+	args[5] = (unsigned int) mode;
+	args[6] = size;
+	args[7] = vaddr;
+	args[8] = 0;
+	args[9] = paddr;
+	args[10] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	ret = (int) args[10];
 	if (ret == 0)
 		ret = -1;
 	return ret;
@@ -231,40 +279,51 @@ int prom_map(int mode, unsigned long size,
 
 void prom_unmap(unsigned long size, unsigned long vaddr)
 {
-	p1275_cmd(prom_callmethod_name,
-		  (P1275_ARG(0, P1275_ARG_IN_STRING) |
-		   P1275_ARG(2, P1275_ARG_IN_64B) |
-		   P1275_ARG(3, P1275_ARG_IN_64B) |
-		   P1275_INOUT(4, 0)),
-		  prom_unmap_name,
-		  prom_get_mmu_ihandle(),
-		  size,
-		  vaddr);
+	unsigned long args[7];
+
+	args[0] = (unsigned long) prom_callmethod_name;
+	args[1] = 4;
+	args[2] = 0;
+	args[3] = (unsigned long) prom_unmap_name;
+	args[4] = (unsigned int) prom_get_mmu_ihandle();
+	args[5] = size;
+	args[6] = vaddr;
+
+	p1275_cmd_direct(args);
 }
 
 /* Set aside physical memory which is not touched or modified
  * across soft resets.
  */
-unsigned long prom_retain(const char *name,
-			  unsigned long pa_low, unsigned long pa_high,
-			  long size, long align)
+int prom_retain(const char *name, unsigned long size,
+		unsigned long align, unsigned long *paddr)
 {
-	/* XXX I don't think we return multiple values correctly.
-	 * XXX OBP supposedly returns pa_low/pa_high here, how does
-	 * XXX it work?
+	unsigned long args[11];
+
+	args[0] = (unsigned long) prom_callmethod_name;
+	args[1] = 5;
+	args[2] = 3;
+	args[3] = (unsigned long) "SUNW,retain";
+	args[4] = (unsigned int) prom_get_memory_ihandle();
+	args[5] = align;
+	args[6] = size;
+	args[7] = (unsigned long) name;
+	args[8] = (unsigned long) -1;
+	args[9] = (unsigned long) -1;
+	args[10] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	if (args[8])
+		return (int) args[8];
+
+	/* Next we get "phys_high" then "phys_low".  On 64-bit
+	 * the phys_high cell is don't care since the phys_low
+	 * cell has the full value.
 	 */
+	*paddr = args[10];
 
-	/* If align is zero, the pa_low/pa_high args are passed,
-	 * else they are not.
-	 */
-	if (align == 0)
-		return p1275_cmd("SUNW,retain",
-				 (P1275_ARG(0, P1275_ARG_IN_BUF) | P1275_INOUT(5, 2)),
-				 name, pa_low, pa_high, size, align);
-	else
-		return p1275_cmd("SUNW,retain",
-				 (P1275_ARG(0, P1275_ARG_IN_BUF) | P1275_INOUT(3, 2)),
-				 name, size, align);
+	return 0;
 }
 
 /* Get "Unumber" string for the SIMM at the given
@@ -277,62 +336,129 @@ int prom_getunumber(int syndrome_code,
 		    unsigned long phys_addr,
 		    char *buf, int buflen)
 {
-	return p1275_cmd(prom_callmethod_name,
-			 (P1275_ARG(0, P1275_ARG_IN_STRING)	|
-			  P1275_ARG(3, P1275_ARG_OUT_BUF)	|
-			  P1275_ARG(6, P1275_ARG_IN_64B)	|
-			  P1275_INOUT(8, 2)),
-			 "SUNW,get-unumber", prom_get_memory_ihandle(),
-			 buflen, buf, P1275_SIZE(buflen),
-			 0, phys_addr, syndrome_code);
+	unsigned long args[12];
+
+	args[0] = (unsigned long) prom_callmethod_name;
+	args[1] = 7;
+	args[2] = 2;
+	args[3] = (unsigned long) "SUNW,get-unumber";
+	args[4] = (unsigned int) prom_get_memory_ihandle();
+	args[5] = buflen;
+	args[6] = (unsigned long) buf;
+	args[7] = 0;
+	args[8] = phys_addr;
+	args[9] = (unsigned int) syndrome_code;
+	args[10] = (unsigned long) -1;
+	args[11] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[10];
 }
 
 /* Power management extensions. */
 void prom_sleepself(void)
 {
-	p1275_cmd("SUNW,sleep-self", P1275_INOUT(0, 0));
+	unsigned long args[3];
+
+	args[0] = (unsigned long) "SUNW,sleep-self";
+	args[1] = 0;
+	args[2] = 0;
+	p1275_cmd_direct(args);
 }
 
 int prom_sleepsystem(void)
 {
-	return p1275_cmd("SUNW,sleep-system", P1275_INOUT(0, 1));
+	unsigned long args[4];
+
+	args[0] = (unsigned long) "SUNW,sleep-system";
+	args[1] = 0;
+	args[2] = 1;
+	args[3] = (unsigned long) -1;
+	p1275_cmd_direct(args);
+
+	return (int) args[3];
 }
 
 int prom_wakeupsystem(void)
 {
-	return p1275_cmd("SUNW,wakeup-system", P1275_INOUT(0, 1));
+	unsigned long args[4];
+
+	args[0] = (unsigned long) "SUNW,wakeup-system";
+	args[1] = 0;
+	args[2] = 1;
+	args[3] = (unsigned long) -1;
+	p1275_cmd_direct(args);
+
+	return (int) args[3];
 }
 
 #ifdef CONFIG_SMP
 void prom_startcpu(int cpunode, unsigned long pc, unsigned long arg)
 {
-	p1275_cmd("SUNW,start-cpu", P1275_INOUT(3, 0), cpunode, pc, arg);
+	unsigned long args[6];
+
+	args[0] = (unsigned long) "SUNW,start-cpu";
+	args[1] = 3;
+	args[2] = 0;
+	args[3] = (unsigned int) cpunode;
+	args[4] = pc;
+	args[5] = arg;
+	p1275_cmd_direct(args);
 }
 
 void prom_startcpu_cpuid(int cpuid, unsigned long pc, unsigned long arg)
 {
-	p1275_cmd("SUNW,start-cpu-by-cpuid", P1275_INOUT(3, 0),
-		  cpuid, pc, arg);
+	unsigned long args[6];
+
+	args[0] = (unsigned long) "SUNW,start-cpu-by-cpuid";
+	args[1] = 3;
+	args[2] = 0;
+	args[3] = (unsigned int) cpuid;
+	args[4] = pc;
+	args[5] = arg;
+	p1275_cmd_direct(args);
 }
 
 void prom_stopcpu_cpuid(int cpuid)
 {
-	p1275_cmd("SUNW,stop-cpu-by-cpuid", P1275_INOUT(1, 0),
-		  cpuid);
+	unsigned long args[4];
+
+	args[0] = (unsigned long) "SUNW,stop-cpu-by-cpuid";
+	args[1] = 1;
+	args[2] = 0;
+	args[3] = (unsigned int) cpuid;
+	p1275_cmd_direct(args);
 }
 
 void prom_stopself(void)
 {
-	p1275_cmd("SUNW,stop-self", P1275_INOUT(0, 0));
+	unsigned long args[3];
+
+	args[0] = (unsigned long) "SUNW,stop-self";
+	args[1] = 0;
+	args[2] = 0;
+	p1275_cmd_direct(args);
 }
 
 void prom_idleself(void)
 {
-	p1275_cmd("SUNW,idle-self", P1275_INOUT(0, 0));
+	unsigned long args[3];
+
+	args[0] = (unsigned long) "SUNW,idle-self";
+	args[1] = 0;
+	args[2] = 0;
+	p1275_cmd_direct(args);
 }
 
 void prom_resumecpu(int cpunode)
 {
-	p1275_cmd("SUNW,resume-cpu", P1275_INOUT(1, 0), cpunode);
+	unsigned long args[4];
+
+	args[0] = (unsigned long) "SUNW,resume-cpu";
+	args[1] = 1;
+	args[2] = 0;
+	args[3] = (unsigned int) cpunode;
+	p1275_cmd_direct(args);
 }
 #endif
diff --git a/arch/sparc/prom/p1275.c b/arch/sparc/prom/p1275.c
index 2d8b70d397f1..fa6e4e219b9c 100644
--- a/arch/sparc/prom/p1275.c
+++ b/arch/sparc/prom/p1275.c
@@ -22,13 +22,11 @@ struct {
 	long prom_callback;			/* 0x00 */
 	void (*prom_cif_handler)(long *);	/* 0x08 */
 	unsigned long prom_cif_stack;		/* 0x10 */
-	unsigned long prom_args [23];		/* 0x18 */
-	char prom_buffer [3000];
 } p1275buf;
 
 extern void prom_world(int);
 
-extern void prom_cif_interface(void);
+extern void prom_cif_direct(unsigned long *args);
 extern void prom_cif_callback(void);
 
 /*
@@ -36,114 +34,20 @@ extern void prom_cif_callback(void);
  */
 DEFINE_RAW_SPINLOCK(prom_entry_lock);
 
-long p1275_cmd(const char *service, long fmt, ...)
+void p1275_cmd_direct(unsigned long *args)
 {
-	char *p, *q;
 	unsigned long flags;
-	int nargs, nrets, i;
-	va_list list;
-	long attrs, x;
-	
-	p = p1275buf.prom_buffer;
 
 	raw_local_save_flags(flags);
 	raw_local_irq_restore(PIL_NMI);
 	raw_spin_lock(&prom_entry_lock);
 
-	p1275buf.prom_args[0] = (unsigned long)p;		/* service */
-	strcpy (p, service);
-	p = (char *)(((long)(strchr (p, 0) + 8)) & ~7);
-	p1275buf.prom_args[1] = nargs = (fmt & 0x0f);		/* nargs */
-	p1275buf.prom_args[2] = nrets = ((fmt & 0xf0) >> 4); 	/* nrets */
-	attrs = fmt >> 8;
-	va_start(list, fmt);
-	for (i = 0; i < nargs; i++, attrs >>= 3) {
-		switch (attrs & 0x7) {
-		case P1275_ARG_NUMBER:
-			p1275buf.prom_args[i + 3] =
-						(unsigned)va_arg(list, long);
-			break;
-		case P1275_ARG_IN_64B:
-			p1275buf.prom_args[i + 3] =
-				va_arg(list, unsigned long);
-			break;
-		case P1275_ARG_IN_STRING:
-			strcpy (p, va_arg(list, char *));
-			p1275buf.prom_args[i + 3] = (unsigned long)p;
-			p = (char *)(((long)(strchr (p, 0) + 8)) & ~7);
-			break;
-		case P1275_ARG_OUT_BUF:
-			(void) va_arg(list, char *);
-			p1275buf.prom_args[i + 3] = (unsigned long)p;
-			x = va_arg(list, long);
-			i++; attrs >>= 3;
-			p = (char *)(((long)(p + (int)x + 7)) & ~7);
-			p1275buf.prom_args[i + 3] = x;
-			break;
-		case P1275_ARG_IN_BUF:
-			q = va_arg(list, char *);
-			p1275buf.prom_args[i + 3] = (unsigned long)p;
-			x = va_arg(list, long);
-			i++; attrs >>= 3;
-			memcpy (p, q, (int)x);
-			p = (char *)(((long)(p + (int)x + 7)) & ~7);
-			p1275buf.prom_args[i + 3] = x;
-			break;
-		case P1275_ARG_OUT_32B:
-			(void) va_arg(list, char *);
-			p1275buf.prom_args[i + 3] = (unsigned long)p;
-			p += 32;
-			break;
-		case P1275_ARG_IN_FUNCTION:
-			p1275buf.prom_args[i + 3] =
-					(unsigned long)prom_cif_callback;
-			p1275buf.prom_callback = va_arg(list, long);
-			break;
-		}
-	}
-	va_end(list);
-
 	prom_world(1);
-	prom_cif_interface();
+	prom_cif_direct(args);
 	prom_world(0);
 
-	attrs = fmt >> 8;
-	va_start(list, fmt);
-	for (i = 0; i < nargs; i++, attrs >>= 3) {
-		switch (attrs & 0x7) {
-		case P1275_ARG_NUMBER:
-			(void) va_arg(list, long);
-			break;
-		case P1275_ARG_IN_STRING:
-			(void) va_arg(list, char *);
-			break;
-		case P1275_ARG_IN_FUNCTION:
-			(void) va_arg(list, long);
-			break;
-		case P1275_ARG_IN_BUF:
-			(void) va_arg(list, char *);
-			(void) va_arg(list, long);
-			i++; attrs >>= 3;
-			break;
-		case P1275_ARG_OUT_BUF:
-			p = va_arg(list, char *);
-			x = va_arg(list, long);
-			memcpy (p, (char *)(p1275buf.prom_args[i + 3]), (int)x);
-			i++; attrs >>= 3;
-			break;
-		case P1275_ARG_OUT_32B:
-			p = va_arg(list, char *);
-			memcpy (p, (char *)(p1275buf.prom_args[i + 3]), 32);
-			break;
-		}
-	}
-	va_end(list);
-	x = p1275buf.prom_args [nargs + 3];
-
 	raw_spin_unlock(&prom_entry_lock);
 	raw_local_irq_restore(flags);
-
-	return x;
 }
 
 void prom_cif_init(void *cif_handler, void *cif_stack)
diff --git a/arch/sparc/prom/tree_64.c b/arch/sparc/prom/tree_64.c
index 3c0d2dd9f693..9d3f9137a43a 100644
--- a/arch/sparc/prom/tree_64.c
+++ b/arch/sparc/prom/tree_64.c
@@ -16,22 +16,39 @@
 #include <asm/oplib.h>
 #include <asm/ldc.h>
 
+static int prom_node_to_node(const char *type, int node)
+{
+	unsigned long args[5];
+
+	args[0] = (unsigned long) type;
+	args[1] = 1;
+	args[2] = 1;
+	args[3] = (unsigned int) node;
+	args[4] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[4];
+}
+
 /* Return the child of node 'node' or zero if no this node has no
  * direct descendent.
  */
 inline int __prom_getchild(int node)
 {
-	return p1275_cmd ("child", P1275_INOUT(1, 1), node);
+	return prom_node_to_node("child", node);
 }
 
 inline int prom_getchild(int node)
 {
 	int cnode;
 
-	if(node == -1) return 0;
+	if (node == -1)
+		return 0;
 	cnode = __prom_getchild(node);
-	if(cnode == -1) return 0;
-	return (int)cnode;
+	if (cnode == -1)
+		return 0;
+	return cnode;
 }
 EXPORT_SYMBOL(prom_getchild);
 
@@ -39,10 +56,12 @@ inline int prom_getparent(int node)
 {
 	int cnode;
 
-	if(node == -1) return 0;
-	cnode = p1275_cmd ("parent", P1275_INOUT(1, 1), node);
-	if(cnode == -1) return 0;
-	return (int)cnode;
+	if (node == -1)
+		return 0;
+	cnode = prom_node_to_node("parent", node);
+	if (cnode == -1)
+		return 0;
+	return cnode;
 }
 
 /* Return the next sibling of node 'node' or zero if no more siblings
@@ -50,7 +69,7 @@ inline int prom_getparent(int node)
  */
 inline int __prom_getsibling(int node)
 {
-	return p1275_cmd(prom_peer_name, P1275_INOUT(1, 1), node);
+	return prom_node_to_node(prom_peer_name, node);
 }
 
 inline int prom_getsibling(int node)
@@ -72,11 +91,21 @@ EXPORT_SYMBOL(prom_getsibling);
  */
 inline int prom_getproplen(int node, const char *prop)
 {
-	if((!node) || (!prop)) return -1;
-	return p1275_cmd ("getproplen", 
-			  P1275_ARG(1,P1275_ARG_IN_STRING)|
-			  P1275_INOUT(2, 1), 
-			  node, prop);
+	unsigned long args[6];
+
+	if (!node || !prop)
+		return -1;
+
+	args[0] = (unsigned long) "getproplen";
+	args[1] = 2;
+	args[2] = 1;
+	args[3] = (unsigned int) node;
+	args[4] = (unsigned long) prop;
+	args[5] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[5];
 }
 EXPORT_SYMBOL(prom_getproplen);
 
@@ -87,19 +116,25 @@ EXPORT_SYMBOL(prom_getproplen);
 inline int prom_getproperty(int node, const char *prop,
 			    char *buffer, int bufsize)
 {
+	unsigned long args[8];
 	int plen;
 
 	plen = prom_getproplen(node, prop);
-	if ((plen > bufsize) || (plen == 0) || (plen == -1)) {
+	if ((plen > bufsize) || (plen == 0) || (plen == -1))
 		return -1;
-	} else {
-		/* Ok, things seem all right. */
-		return p1275_cmd(prom_getprop_name, 
-				 P1275_ARG(1,P1275_ARG_IN_STRING)|
-				 P1275_ARG(2,P1275_ARG_OUT_BUF)|
-				 P1275_INOUT(4, 1), 
-				 node, prop, buffer, P1275_SIZE(plen));
-	}
+
+	args[0] = (unsigned long) prom_getprop_name;
+	args[1] = 4;
+	args[2] = 1;
+	args[3] = (unsigned int) node;
+	args[4] = (unsigned long) prop;
+	args[5] = (unsigned long) buffer;
+	args[6] = bufsize;
+	args[7] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[7];
 }
 EXPORT_SYMBOL(prom_getproperty);
 
@@ -110,7 +145,7 @@ inline int prom_getint(int node, const char *prop)
 {
 	int intprop;
 
-	if(prom_getproperty(node, prop, (char *) &intprop, sizeof(int)) != -1)
+	if (prom_getproperty(node, prop, (char *) &intprop, sizeof(int)) != -1)
 		return intprop;
 
 	return -1;
@@ -126,7 +161,8 @@ int prom_getintdefault(int node, const char *property, int deflt)
 	int retval;
 
 	retval = prom_getint(node, property);
-	if(retval == -1) return deflt;
+	if (retval == -1)
+		return deflt;
 
 	return retval;
 }
@@ -138,7 +174,8 @@ int prom_getbool(int node, const char *prop)
 	int retval;
 
 	retval = prom_getproplen(node, prop);
-	if(retval == -1) return 0;
+	if (retval == -1)
+		return 0;
 	return 1;
 }
 EXPORT_SYMBOL(prom_getbool);
@@ -152,7 +189,8 @@ void prom_getstring(int node, const char *prop, char *user_buf, int ubuf_size)
 	int len;
 
 	len = prom_getproperty(node, prop, user_buf, ubuf_size);
-	if(len != -1) return;
+	if (len != -1)
+		return;
 	user_buf[0] = 0;
 }
 EXPORT_SYMBOL(prom_getstring);
@@ -164,7 +202,8 @@ int prom_nodematch(int node, const char *name)
 {
 	char namebuf[128];
 	prom_getproperty(node, "name", namebuf, sizeof(namebuf));
-	if(strcmp(namebuf, name) == 0) return 1;
+	if (strcmp(namebuf, name) == 0)
+		return 1;
 	return 0;
 }
 
@@ -190,16 +229,29 @@ int prom_searchsiblings(int node_start, const char *nodename)
 }
 EXPORT_SYMBOL(prom_searchsiblings);
 
+static const char *prom_nextprop_name = "nextprop";
+
 /* Return the first property type for node 'node'.
  * buffer should be at least 32B in length
  */
 inline char *prom_firstprop(int node, char *buffer)
 {
+	unsigned long args[7];
+
 	*buffer = 0;
-	if(node == -1) return buffer;
-	p1275_cmd ("nextprop", P1275_ARG(2,P1275_ARG_OUT_32B)|
-			       P1275_INOUT(3, 0), 
-			       node, (char *) 0x0, buffer);
+	if (node == -1)
+		return buffer;
+
+	args[0] = (unsigned long) prom_nextprop_name;
+	args[1] = 3;
+	args[2] = 1;
+	args[3] = (unsigned int) node;
+	args[4] = 0;
+	args[5] = (unsigned long) buffer;
+	args[6] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
 	return buffer;
 }
 EXPORT_SYMBOL(prom_firstprop);
@@ -210,9 +262,10 @@ EXPORT_SYMBOL(prom_firstprop);
  */
 inline char *prom_nextprop(int node, const char *oprop, char *buffer)
 {
+	unsigned long args[7];
 	char buf[32];
 
-	if(node == -1) {
+	if (node == -1) {
 		*buffer = 0;
 		return buffer;
 	}
@@ -220,10 +273,17 @@ inline char *prom_nextprop(int node, const char *oprop, char *buffer)
 		strcpy (buf, oprop);
 		oprop = buf;
 	}
-	p1275_cmd ("nextprop", P1275_ARG(1,P1275_ARG_IN_STRING)|
-				    P1275_ARG(2,P1275_ARG_OUT_32B)|
-				    P1275_INOUT(3, 0), 
-				    node, oprop, buffer); 
+
+	args[0] = (unsigned long) prom_nextprop_name;
+	args[1] = 3;
+	args[2] = 1;
+	args[3] = (unsigned int) node;
+	args[4] = (unsigned long) oprop;
+	args[5] = (unsigned long) buffer;
+	args[6] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
 	return buffer;
 }
 EXPORT_SYMBOL(prom_nextprop);
@@ -231,12 +291,19 @@ EXPORT_SYMBOL(prom_nextprop);
 int
 prom_finddevice(const char *name)
 {
+	unsigned long args[5];
+
 	if (!name)
 		return 0;
-	return p1275_cmd(prom_finddev_name,
-			 P1275_ARG(0,P1275_ARG_IN_STRING)|
-			 P1275_INOUT(1, 1), 
-			 name);
+	args[0] = (unsigned long) "finddevice";
+	args[1] = 1;
+	args[2] = 1;
+	args[3] = (unsigned long) name;
+	args[4] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[4];
 }
 EXPORT_SYMBOL(prom_finddevice);
 
@@ -247,7 +314,7 @@ int prom_node_has_property(int node, const char *prop)
 	*buf = 0;
 	do {
 		prom_nextprop(node, buf, buf);
-		if(!strcmp(buf, prop))
+		if (!strcmp(buf, prop))
 			return 1;
 	} while (*buf);
 	return 0;
@@ -260,6 +327,8 @@ EXPORT_SYMBOL(prom_node_has_property);
 int
 prom_setprop(int node, const char *pname, char *value, int size)
 {
+	unsigned long args[8];
+
 	if (size == 0)
 		return 0;
 	if ((pname == 0) || (value == 0))
@@ -271,19 +340,37 @@ prom_setprop(int node, const char *pname, char *value, int size)
 		return 0;
 	}
 #endif
-	return p1275_cmd ("setprop", P1275_ARG(1,P1275_ARG_IN_STRING)|
-					  P1275_ARG(2,P1275_ARG_IN_BUF)|
-					  P1275_INOUT(4, 1), 
-					  node, pname, value, P1275_SIZE(size));
+	args[0] = (unsigned long) "setprop";
+	args[1] = 4;
+	args[2] = 1;
+	args[3] = (unsigned int) node;
+	args[4] = (unsigned long) pname;
+	args[5] = (unsigned long) value;
+	args[6] = size;
+	args[7] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[7];
 }
 EXPORT_SYMBOL(prom_setprop);
 
 inline int prom_inst2pkg(int inst)
 {
+	unsigned long args[5];
 	int node;
 	
-	node = p1275_cmd ("instance-to-package", P1275_INOUT(1, 1), inst);
-	if (node == -1) return 0;
+	args[0] = (unsigned long) "instance-to-package";
+	args[1] = 1;
+	args[2] = 1;
+	args[3] = (unsigned int) inst;
+	args[4] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	node = (int) args[4];
+	if (node == -1)
+		return 0;
 	return node;
 }
 
@@ -296,17 +383,28 @@ prom_pathtoinode(const char *path)
 	int node, inst;
 
 	inst = prom_devopen (path);
-	if (inst == 0) return 0;
-	node = prom_inst2pkg (inst);
-	prom_devclose (inst);
-	if (node == -1) return 0;
+	if (inst == 0)
+		return 0;
+	node = prom_inst2pkg(inst);
+	prom_devclose(inst);
+	if (node == -1)
+		return 0;
 	return node;
 }
 
 int prom_ihandle2path(int handle, char *buffer, int bufsize)
 {
-	return p1275_cmd("instance-to-path",
-			 P1275_ARG(1,P1275_ARG_OUT_BUF)|
-			 P1275_INOUT(3, 1),
-			 handle, buffer, P1275_SIZE(bufsize));
+	unsigned long args[7];
+
+	args[0] = (unsigned long) "instance-to-path";
+	args[1] = 3;
+	args[2] = 1;
+	args[3] = (unsigned int) handle;
+	args[4] = (unsigned long) buffer;
+	args[5] = bufsize;
+	args[6] = (unsigned long) -1;
+
+	p1275_cmd_direct(args);
+
+	return (int) args[6];
 }
-- 
cgit v1.2.3-59-g8ed1b