powerpc: Define 32/64 bit asm macros and use them in fpu.S

These macros help in writing assembly code that works for both ppc32
and ppc64.  With this we now have a common fpu.S.  This takes out
load_up_fpu from head_64.S.

Signed-off-by: Paul Mackerras <paulus@samba.org>

arch/powerpc/kernel/fpu.S

@@ -27,13 +27,9 @@
  * Load up this task's FP registers from its thread_struct,
  * enable the FPU for the current task and return to the task.
  */
-	.globl	load_up_fpu
+_GLOBAL(load_up_fpu)
-load_up_fpu:
 	mfmsr	r5
 	ori	r5,r5,MSR_FP
-#ifdef CONFIG_PPC64BRIDGE
-	clrldi	r5,r5,1			/* turn off 64-bit mode */
-#endif /* CONFIG_PPC64BRIDGE */
 	SYNC
 	MTMSRD(r5)			/* enable use of fpu now */
 	isync
@@ -43,67 +39,57 @@ load_up_fpu:
  * to another.  Instead we call giveup_fpu in switch_to.
  */
 #ifndef CONFIG_SMP
-	tophys(r6,0)			/* get __pa constant */
+	LOADBASE(r3, last_task_used_math)
-	addis	r3,r6,last_task_used_math@ha
+	tophys(r3,r3)
-	lwz	r4,last_task_used_math@l(r3)
+	LDL	r4,OFF(last_task_used_math)(r3)
-	cmpwi	0,r4,0
+	CMPI	0,r4,0
 	beq	1f
-	add	r4,r4,r6
+	tophys(r4,r4)
 	addi	r4,r4,THREAD		/* want last_task_used_math->thread */
 	SAVE_32FPRS(0, r4)
 	mffs	fr0
 	stfd	fr0,THREAD_FPSCR-4(r4)
-	lwz	r5,PT_REGS(r4)
+	LDL	r5,PT_REGS(r4)
-	add	r5,r5,r6
+	tophys(r5,r5)
-	lwz	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
+	LDL	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
 	li	r10,MSR_FP|MSR_FE0|MSR_FE1
 	andc	r4,r4,r10		/* disable FP for previous task */
-	stw	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
+	STL	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
 1:
 #endif /* CONFIG_SMP */
 	/* enable use of FP after return */
+#ifdef CONFIG_PPC32
 	mfspr	r5,SPRN_SPRG3		/* current task's THREAD (phys) */
 	lwz	r4,THREAD_FPEXC_MODE(r5)
 	ori	r9,r9,MSR_FP		/* enable FP for current */
 	or	r9,r9,r4
+#else
+	ld	r4,PACACURRENT(r13)
+	addi	r5,r4,THREAD		/* Get THREAD */
+	ld	r4,THREAD_FPEXC_MODE(r5)
+	ori	r12,r12,MSR_FP
+	or	r12,r12,r4
+	std	r12,_MSR(r1)
+#endif
 	lfd	fr0,THREAD_FPSCR-4(r5)
 	mtfsf	0xff,fr0
 	REST_32FPRS(0, r5)
 #ifndef CONFIG_SMP
 	subi	r4,r5,THREAD
-	sub	r4,r4,r6
+	tovirt(r4,r4)
-	stw	r4,last_task_used_math@l(r3)
+	STL	r4,OFF(last_task_used_math)(r3)
 #endif /* CONFIG_SMP */
 	/* restore registers and return */
 	/* we haven't used ctr or xer or lr */
 	b	fast_exception_return
 
-/*
- * FP unavailable trap from kernel - print a message, but let
- * the task use FP in the kernel until it returns to user mode.
- */
- 	.globl	KernelFP
-KernelFP:
-	lwz	r3,_MSR(r1)
-	ori	r3,r3,MSR_FP
-	stw	r3,_MSR(r1)		/* enable use of FP after return */
-	lis	r3,86f@h
-	ori	r3,r3,86f@l
-	mr	r4,r2			/* current */
-	lwz	r5,_NIP(r1)
-	bl	printk
-	b	ret_from_except
-86:	.string	"floating point used in kernel (task=%p, pc=%x)\n"
-	.align	4,0
-
 /*
  * giveup_fpu(tsk)
  * Disable FP for the task given as the argument,
  * and save the floating-point registers in its thread_struct.
  * Enables the FPU for use in the kernel on return.
  */
-	.globl	giveup_fpu
+_GLOBAL(giveup_fpu)
-giveup_fpu:
 	mfmsr	r5
 	ori	r5,r5,MSR_FP
 	SYNC_601
@@ -111,23 +97,23 @@ giveup_fpu:
 	MTMSRD(r5)			/* enable use of fpu now */
 	SYNC_601
 	isync
-	cmpwi	0,r3,0
+	CMPI	0,r3,0
 	beqlr-				/* if no previous owner, done */
 	addi	r3,r3,THREAD	        /* want THREAD of task */
-	lwz	r5,PT_REGS(r3)
+	LDL	r5,PT_REGS(r3)
-	cmpwi	0,r5,0
+	CMPI	0,r5,0
 	SAVE_32FPRS(0, r3)
 	mffs	fr0
 	stfd	fr0,THREAD_FPSCR-4(r3)
 	beq	1f
-	lwz	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
+	LDL	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
 	li	r3,MSR_FP|MSR_FE0|MSR_FE1
 	andc	r4,r4,r3		/* disable FP for previous task */
-	stw	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
+	STL	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
 1:
 #ifndef CONFIG_SMP
 	li	r5,0
-	lis	r4,last_task_used_math@ha
+	LOADBASE(r4,last_task_used_math)
-	stw	r5,last_task_used_math@l(r4)
+	STL	r5,OFF(last_task_used_math)(r4)
 #endif /* CONFIG_SMP */
 	blr

arch/powerpc/kernel/head_64.S

@@ -80,7 +80,7 @@ _stext:
 _GLOBAL(__start)
 	/* NOP this out unconditionally */
 BEGIN_FTR_SECTION
-	b .__start_initialization_multiplatform
+	b	.__start_initialization_multiplatform
 END_FTR_SECTION(0, 1)
 #endif /* CONFIG_PPC_MULTIPLATFORM */
 
@@ -857,62 +857,6 @@ fp_unavailable_common:
 	bl	.kernel_fp_unavailable_exception
 	BUG_OPCODE
 
-/*
- * load_up_fpu(unused, unused, tsk)
- * Disable FP for the task which had the FPU previously,
- * and save its floating-point registers in its thread_struct.
- * Enables the FPU for use in the kernel on return.
- * On SMP we know the fpu is free, since we give it up every
- * switch (ie, no lazy save of the FP registers).
- * On entry: r13 == 'current' && last_task_used_math != 'current'
- */
-_STATIC(load_up_fpu)
-	mfmsr	r5			/* grab the current MSR */
-	ori	r5,r5,MSR_FP
-	mtmsrd	r5			/* enable use of fpu now */
-	isync
-/*
- * For SMP, we don't do lazy FPU switching because it just gets too
- * horrendously complex, especially when a task switches from one CPU
- * to another.  Instead we call giveup_fpu in switch_to.
- *
- */
-#ifndef CONFIG_SMP
-	ld	r3,last_task_used_math@got(r2)
-	ld	r4,0(r3)
-	cmpdi	0,r4,0
-	beq	1f
-	/* Save FP state to last_task_used_math's THREAD struct */
-	addi	r4,r4,THREAD
-	SAVE_32FPRS(0, r4)
-	mffs	fr0
-	stfd	fr0,THREAD_FPSCR(r4)
-	/* Disable FP for last_task_used_math */
-	ld	r5,PT_REGS(r4)
-	ld	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
-	li	r6,MSR_FP|MSR_FE0|MSR_FE1
-	andc	r4,r4,r6
-	std	r4,_MSR-STACK_FRAME_OVERHEAD(r5)
-1:
-#endif /* CONFIG_SMP */
-	/* enable use of FP after return */
-	ld	r4,PACACURRENT(r13)
-	addi	r5,r4,THREAD		/* Get THREAD */
-	ld	r4,THREAD_FPEXC_MODE(r5)
-	ori	r12,r12,MSR_FP
-	or	r12,r12,r4
-	std	r12,_MSR(r1)
-	lfd	fr0,THREAD_FPSCR(r5)
-	mtfsf	0xff,fr0
-	REST_32FPRS(0, r5)
-#ifndef CONFIG_SMP
-	/* Update last_task_used_math to 'current' */
-	subi	r4,r5,THREAD		/* Back to 'current' */
-	std	r4,0(r3)
-#endif /* CONFIG_SMP */
-	/* restore registers and return */
-	b	fast_exception_return
-
 	.align	7
 	.globl altivec_unavailable_common
 altivec_unavailable_common:

include/asm-powerpc/ppc_asm.h

@@ -103,12 +103,13 @@
 	oris	rn,rn,name##@h;		\
 	ori	rn,rn,name##@l
 
-#define LOADBASE(rn,name) \
+#define LOADBASE(rn,name)		\
-	lis	rn,name@highest;	\
+	.section .toc,"aw";		\
-	ori	rn,rn,name@higher;	\
+1:	.tc	name[TC],name;		\
-	rldicr	rn,rn,32,31;		\
+	.previous;			\
-	oris	rn,rn,name@ha
+	ld	rn,1b@toc(r2)
 
+#define OFF(name)	0
 
 #define SET_REG_TO_CONST(reg, value)	         	\
 	lis     reg,(((value)>>48)&0xFFFF);             \
@@ -123,6 +124,23 @@
 	rldicr  reg,reg,32,31;                          \
 	oris    reg,reg,(label)@h;                      \
 	ori     reg,reg,(label)@l;
+
+/* operations for longs and pointers */
+#define LDL	ld
+#define STL	std
+#define CMPI	cmpdi
+
+#else /* 32-bit */
+#define LOADBASE(rn,name)	\
+	lis	rn,name@ha
+
+#define OFF(name)	name@l
+
+/* operations for longs and pointers */
+#define LDL	lwz
+#define STL	stw
+#define CMPI	cmpwi
+
 #endif
 
 /* various errata or part fixups */