[MIPS] Fix fpu_save_double on 64-bit.

> Without this fix, _save_fp() in 64-bit kernel is seriously broken.
>
> ffffffff8010bec0 <_save_fp>:
> ffffffff8010bec0:       400d6000        mfc0    t1,c0_status
> ffffffff8010bec4:       000c7140        sll     t2,t0,0x5
> ffffffff8010bec8:       05c10011        bgez    t2,ffffffff8010bf10 <_save_fp+0x50>
> ffffffff8010becc:       00000000        nop
> ffffffff8010bed0:       f4810328        sdc1    $f1,808(a0)
> ...

Fix register usage in fpu_save_double() and make fpu_restore_double()
more symmetric with fpu_save_double().

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

arch/mips/kernel/r4k_switch.S

@@ -75,8 +75,8 @@
 	and	t0, t0, t1
 	LONG_S	t0, ST_OFF(t3)
 
-	fpu_save_double a0 t1 t0 t2		# c0_status passed in t1
-						# clobbers t0 and t2
+	fpu_save_double a0 t0 t1		# c0_status passed in t0
+						# clobbers t1
 1:
 
 	/*
@@ -129,9 +129,9 @@
  */
 LEAF(_save_fp)
 #ifdef CONFIG_64BIT
-	mfc0	t1, CP0_STATUS
+	mfc0	t0, CP0_STATUS
 #endif
-	fpu_save_double a0 t1 t0 t2		# clobbers t1
+	fpu_save_double a0 t0 t1		# clobbers t1
 	jr	ra
 	END(_save_fp)
 
@@ -139,7 +139,10 @@ LEAF(_save_fp)
  * Restore a thread's fp context.
  */
 LEAF(_restore_fp)
-	fpu_restore_double a0, t1		# clobbers t1
+#ifdef CONFIG_64BIT
+	mfc0	t0, CP0_STATUS
+#endif
+	fpu_restore_double a0 t0 t1		# clobbers t1
 	jr	ra
 	END(_restore_fp)
 

include/asm-mips/asmmacro-32.h

@@ -12,7 +12,7 @@
 #include <asm/fpregdef.h>
 #include <asm/mipsregs.h>
 
-	.macro	fpu_save_double thread status tmp1=t0 tmp2
+	.macro	fpu_save_double thread status tmp1=t0
 	cfc1	\tmp1,  fcr31
 	sdc1	$f0,  THREAD_FPR0(\thread)
 	sdc1	$f2,  THREAD_FPR2(\thread)
@@ -70,7 +70,7 @@
 	sw	\tmp, THREAD_FCR31(\thread)
 	.endm
 
-	.macro	fpu_restore_double thread tmp=t0
+	.macro	fpu_restore_double thread status tmp=t0
 	lw	\tmp, THREAD_FCR31(\thread)
 	ldc1	$f0,  THREAD_FPR0(\thread)
 	ldc1	$f2,  THREAD_FPR2(\thread)

include/asm-mips/asmmacro-64.h

@@ -53,12 +53,12 @@
 	sdc1	$f31, THREAD_FPR31(\thread)
 	.endm
 
-	.macro	fpu_save_double thread status tmp1 tmp2
-	sll	\tmp2, \tmp1, 5
-	bgez	\tmp2, 2f
+	.macro	fpu_save_double thread status tmp
+	sll	\tmp, \status, 5
+	bgez	\tmp, 2f
 	fpu_save_16odd \thread
 2:
-	fpu_save_16even \thread \tmp1			# clobbers t1
+	fpu_save_16even \thread \tmp
 	.endm
 
 	.macro	fpu_restore_16even thread tmp=t0
@@ -101,13 +101,12 @@
 	ldc1	$f31, THREAD_FPR31(\thread)
 	.endm
 
-	.macro	fpu_restore_double thread tmp
-	mfc0	t0, CP0_STATUS
-	sll	t1, t0, 5
-	bgez	t1, 1f				# 16 register mode?
+	.macro	fpu_restore_double thread status tmp
+	sll	\tmp, \status, 5
+	bgez	\tmp, 1f				# 16 register mode?
 
-	fpu_restore_16odd a0
-1:	fpu_restore_16even a0, t0		# clobbers t0
+	fpu_restore_16odd \thread
+1:	fpu_restore_16even \thread \tmp
 	.endm
 
 	.macro	cpu_save_nonscratch thread