[MIPS] Update/fix futex assembly

    
 o Implement futex_atomic_op_inuser() operation
 o Don't use the R10000-ll/sc bug workaround version for every processor.
   branch likely is deprecated and some historic ll/sc processors don't
   implement it.  In any case it's slow.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

include/asm-mips/futex.h

@@ -7,6 +7,7 @@
 #include <linux/futex.h>
 #include <asm/errno.h>
 #include <asm/uaccess.h>
+#include <asm/war.h>
 
 #ifdef CONFIG_SMP
 #define __FUTEX_SMP_SYNC "	sync					\n"
@@ -16,11 +17,12 @@
 
 #define __futex_atomic_op(insn, ret, oldval, uaddr, oparg)		\
 {									\
+	if (cpu_has_llsc && R10000_LLSC_WAR) {				\
 		__asm__ __volatile__(					\
 		"	.set	push				\n"	\
 		"	.set	noat				\n"	\
 		"	.set	mips3				\n"	\
-	"1:	ll	%1, (%3)	# __futex_atomic_op1	\n"	\
+		"1:	ll	%1, (%3)	# __futex_atomic_op	\n" \
 		"	.set	mips0				\n"	\
 		"	" insn	"				\n"	\
 		"	.set	mips3				\n"	\
@@ -40,6 +42,33 @@
 		"	.previous				\n"	\
 		: "=r" (ret), "=r" (oldval)				\
 		: "0" (0), "r" (uaddr), "Jr" (oparg), "i" (-EFAULT));	\
+	} else if (cpu_has_llsc) {					\
+		__asm__ __volatile__(					\
+		"	.set	push				\n"	\
+		"	.set	noat				\n"	\
+		"	.set	mips3				\n"	\
+		"1:	ll	%1, (%3)	# __futex_atomic_op	\n" \
+		"	.set	mips0				\n"	\
+		"	" insn	"				\n"	\
+		"	.set	mips3				\n"	\
+		"2:	sc	$1, (%3)			\n"	\
+		"	beqz	$1, 1b				\n"	\
+		__FUTEX_SMP_SYNC					\
+		"3:						\n"	\
+		"	.set	pop				\n"	\
+		"	.set	mips0				\n"	\
+		"	.section .fixup,\"ax\"			\n"	\
+		"4:	li	%0, %5				\n"	\
+		"	j	2b				\n"	\
+		"	.previous				\n"	\
+		"	.section __ex_table,\"a\"		\n"	\
+		"	"__UA_ADDR "\t1b, 4b			\n"	\
+		"	"__UA_ADDR "\t2b, 4b			\n"	\
+		"	.previous				\n"	\
+		: "=r" (ret), "=r" (oldval)				\
+		: "0" (0), "r" (uaddr), "Jr" (oparg), "i" (-EFAULT));	\
+	} else								\
+		ret = -ENOSYS;						\
 }
 
 static inline int
@@ -102,7 +131,69 @@ futex_atomic_op_inuser (int encoded_op, int __user *uaddr)
 static inline int
 futex_atomic_cmpxchg_inatomic(int __user *uaddr, int oldval, int newval)
 {
+	int retval;
+
+	if (!access_ok(VERIFY_WRITE, uaddr, sizeof(int)))
+		return -EFAULT;
+
+	if (cpu_has_llsc && R10000_LLSC_WAR) {
+		__asm__ __volatile__(
+		"# futex_atomic_cmpxchg_inatomic			\n"
+		"	.set	push					\n"
+		"	.set	noat					\n"
+		"	.set	mips3					\n"
+		"1:	ll	%0, %2					\n"
+		"	bne	%0, %z3, 3f				\n"
+		"	.set	mips0					\n"
+		"	move	$1, %z4					\n"
+		"	.set	mips3					\n"
+		"2:	sc	$1, %1					\n"
+		"	beqzl	$1, 1b					\n"
+		__FUTEX_SMP_SYNC
+		"3:							\n"
+		"	.set	pop					\n"
+		"	.section .fixup,\"ax\"				\n"
+		"4:	li	%0, %5					\n"
+		"	j	3b					\n"
+		"	.previous					\n"
+		"	.section __ex_table,\"a\"			\n"
+		"	"__UA_ADDR "\t1b, 4b				\n"
+		"	"__UA_ADDR "\t2b, 4b				\n"
+		"	.previous					\n"
+		: "=&r" (retval), "=R" (*uaddr)
+		: "R" (*uaddr), "Jr" (oldval), "Jr" (newval), "i" (-EFAULT)
+		: "memory");
+	} else if (cpu_has_llsc) {
+		__asm__ __volatile__(
+		"# futex_atomic_cmpxchg_inatomic			\n"
+		"	.set	push					\n"
+		"	.set	noat					\n"
+		"	.set	mips3					\n"
+		"1:	ll	%0, %2					\n"
+		"	bne	%0, %z3, 3f				\n"
+		"	.set	mips0					\n"
+		"	move	$1, %z4					\n"
+		"	.set	mips3					\n"
+		"2:	sc	$1, %1					\n"
+		"	beqz	$1, 1b					\n"
+		__FUTEX_SMP_SYNC
+		"3:							\n"
+		"	.set	pop					\n"
+		"	.section .fixup,\"ax\"				\n"
+		"4:	li	%0, %5					\n"
+		"	j	3b					\n"
+		"	.previous					\n"
+		"	.section __ex_table,\"a\"			\n"
+		"	"__UA_ADDR "\t1b, 4b				\n"
+		"	"__UA_ADDR "\t2b, 4b				\n"
+		"	.previous					\n"
+		: "=&r" (retval), "=R" (*uaddr)
+		: "R" (*uaddr), "Jr" (oldval), "Jr" (newval), "i" (-EFAULT)
+		: "memory");
+	} else
 		return -ENOSYS;
+
+	return retval;
 }
 
 #endif