/*
 * copy_page, __copy_user_page, __copy_user implementation of SuperH
 *
 * Copyright (C) 2001  Niibe Yutaka & Kaz Kojima
 * Copyright (C) 2002  Toshinobu Sugioka
 * Copyright (C) 2006  Paul Mundt
 */
#include <linux/linkage.h>
#include <asm/page.h>

/*
 * copy_page
 * @to: P1 address
 * @from: P1 address
 *
 * void copy_page(void *to, void *from)
 */

/*
 * r0, r1, r2, r3, r4, r5, r6, r7 --- scratch 
 * r8 --- from + PAGE_SIZE
 * r9 --- not used
 * r10 --- to
 * r11 --- from
 */
ENTRY(copy_page)
	mov.l	r8,@-r15
	mov.l	r10,@-r15
	mov.l	r11,@-r15
	mov	r4,r10
	mov	r5,r11
	mov	r5,r8
	mov.l	.Lpsz,r0
	add	r0,r8
	!
1:	mov.l	@r11+,r0
	mov.l	@r11+,r1
	mov.l	@r11+,r2
	mov.l	@r11+,r3
	mov.l	@r11+,r4
	mov.l	@r11+,r5
	mov.l	@r11+,r6
	mov.l	@r11+,r7
#if defined(CONFIG_CPU_SH3)
	mov.l	r0,@r10
#elif defined(CONFIG_CPU_SH4)
	movca.l	r0,@r10
	mov	r10,r0
#endif
	add	#32,r10
	mov.l	r7,@-r10
	mov.l	r6,@-r10
	mov.l	r5,@-r10
	mov.l	r4,@-r10
	mov.l	r3,@-r10
	mov.l	r2,@-r10
	mov.l	r1,@-r10
#if defined(CONFIG_CPU_SH4)
	ocbwb	@r0
#endif
	cmp/eq	r11,r8
	bf/s	1b
	 add	#28,r10
	!
	mov.l	@r15+,r11
	mov.l	@r15+,r10
	mov.l	@r15+,r8
	rts
	 nop

	.balign 4
.Lpsz:	.long	PAGE_SIZE

/*
 * __kernel_size_t __copy_user(void *to, const void *from, __kernel_size_t n);
 * Return the number of bytes NOT copied
 */
#define EX(...)			\
	9999: __VA_ARGS__ ;		\
	.section __ex_table, "a";	\
	.long 9999b, 6000f	;	\
	.previous
ENTRY(__copy_user)
	! Check if small number of bytes
	mov	#11,r0
	mov	r4,r3
	cmp/gt	r0,r6		! r6 (len) > r0 (11)
	bf/s	.L_cleanup_loop_no_pop
	 add	r6,r3		! last destination address

	! Calculate bytes needed to align to src
	mov.l	r11,@-r15
	neg	r5,r0
	mov.l	r10,@-r15
	add	#4,r0
	mov.l	r9,@-r15
	and	#3,r0
	mov.l	r8,@-r15
	tst	r0,r0
	bt	2f

1:
	! Copy bytes to long word align src
EX(	mov.b	@r5+,r1		)
	dt	r0
	add	#-1,r6
EX(	mov.b	r1,@r4		)
	bf/s	1b
	 add	#1,r4

	! Jump to appropriate routine depending on dest
2:	mov	#3,r1
	mov	r6, r2
	and	r4,r1
	shlr2	r2
	shll2	r1
	mova	.L_jump_tbl,r0
	mov.l	@(r0,r1),r1
	jmp	@r1
	 nop

	.align 2
.L_jump_tbl:
	.long	.L_dest00
	.long	.L_dest01
	.long	.L_dest10
	.long	.L_dest11

/*
 * Come here if there are less than 12 bytes to copy
 *
 * Keep the branch target close, so the bf/s callee doesn't overflow
 * and result in a more expensive branch being inserted. This is the
 * fast-path for small copies, the jump via the jump table will hit the
 * default slow-path cleanup. -PFM.
 */
.L_cleanup_loop_no_pop:
	tst	r6,r6		! Check explicitly for zero
	bt	1f

2:
EX(	mov.b	@r5+,r0		)
	dt	r6
EX(	mov.b	r0,@r4		)
	bf/s	2b
	 add	#1,r4

1:	mov	#0,r0		! normal return
5000:

# Exception handler:
.section .fixup, "ax"
6000:
	mov.l	8000f,r1
	mov	r3,r0
	jmp	@r1
	 sub	r4,r0
	.align	2
8000:	.long	5000b

.previous
	rts
	 nop

! Destination = 00

.L_dest00:
	! Skip the large copy for small transfers
	mov	#(32+32-4), r0
	cmp/gt	r6, r0		! r0 (60) > r6 (len)
	bt	1f

	! Align dest to a 32 byte boundary
	neg	r4,r0
	add	#0x20, r0
	and	#0x1f, r0
	tst	r0, r0
	bt	2f

	sub	r0, r6
	shlr2	r0
3:
EX(	mov.l	@r5+,r1		)
	dt	r0
EX(	mov.l	r1,@r4		)
	bf/s	3b
	 add	#4,r4

2:
EX(	mov.l	@r5+,r0		)
EX(	mov.l	@r5+,r1		)
EX(	mov.l	@r5+,r2		)
EX(	mov.l	@r5+,r7		)
EX(	mov.l	@r5+,r8		)
EX(	mov.l	@r5+,r9		)
EX(	mov.l	@r5+,r10	)
EX(	mov.l	@r5+,r11	)
#ifdef CONFIG_CPU_SH4
EX(	movca.l	r0,@r4		)
#else
EX(	mov.l	r0,@r4		)
#endif
	add	#-32, r6
EX(	mov.l	r1,@(4,r4)	)
	mov	#32, r0
EX(	mov.l	r2,@(8,r4)	)
	cmp/gt	r6, r0		! r0 (32) > r6 (len)
EX(	mov.l	r7,@(12,r4)	)
EX(	mov.l	r8,@(16,r4)	)
EX(	mov.l	r9,@(20,r4)	)
EX(	mov.l	r10,@(24,r4)	)
EX(	mov.l	r11,@(28,r4)	)
	bf/s	2b
	 add	#32,r4

1:	mov	r6, r0
	shlr2	r0
	tst	r0, r0
	bt	.L_cleanup
1:
EX(	mov.l	@r5+,r1		)
	dt	r0
EX(	mov.l	r1,@r4		)
	bf/s	1b
	 add	#4,r4

	bra	.L_cleanup
	 nop

! Destination = 10

.L_dest10:
	mov	r2,r7
	shlr2	r7
	shlr	r7
	tst	r7,r7
	mov	#7,r0
	bt/s	1f
	 and	r0,r2
2:
	dt	r7
#ifdef CONFIG_CPU_LITTLE_ENDIAN
EX(	mov.l	@r5+,r0		)
EX(	mov.l	@r5+,r1		)
EX(	mov.l	@r5+,r8		)
EX(	mov.l	@r5+,r9		)
EX(	mov.l	@r5+,r10	)
EX(	mov.w	r0,@r4		)
	add	#2,r4
	xtrct	r1,r0
	xtrct	r8,r1
	xtrct	r9,r8
	xtrct	r10,r9

EX(	mov.l	r0,@r4		)
EX(	mov.l	r1,@(4,r4)	)
EX(	mov.l	r8,@(8,r4)	)
EX(	mov.l	r9,@(12,r4)	)

EX(	mov.l	@r5+,r1		)
EX(	mov.l	@r5+,r8		)
EX(	mov.l	@r5+,r0		)
	xtrct	r1,r10
	xtrct	r8,r1
	xtrct	r0,r8
	shlr16	r0
EX(	mov.l	r10,@(16,r4)	)
EX(	mov.l	r1,@(20,r4)	)
EX(	mov.l	r8,@(24,r4)	)
EX(	mov.w	r0,@(28,r4)	)
	bf/s	2b
	 add	#30,r4
#else
EX(	mov.l	@(28,r5),r0	)
EX(	mov.l	@(24,r5),r8	)
EX(	mov.l	@(20,r5),r9	)
EX(	mov.l	@(16,r5),r10	)
EX(	mov.w	r0,@(30,r4)	)
	add	#-2,r4
	xtrct	r8,r0
	xtrct	r9,r8
	xtrct	r10,r9
EX(	mov.l	r0,@(28,r4)	)
EX(	mov.l	r8,@(24,r4)	)
EX(	mov.l	r9,@(20,r4)	)

EX(	mov.l	@(12,r5),r0	)
EX(	mov.l	@(8,r5),r8	)
	xtrct	r0,r10
EX(	mov.l	@(4,r5),r9	)
	mov.l	r10,@(16,r4)
EX(	mov.l	@r5,r10		)
	xtrct	r8,r0
	xtrct	r9,r8
	xtrct	r10,r9
EX(	mov.l	r0,@(12,r4)	)
EX(	mov.l	r8,@(8,r4)	)
	swap.w	r10,r0
EX(	mov.l	r9,@(4,r4)	)
EX(	mov.w	r0,@(2,r4)	)

	add	#32,r5
	bf/s	2b
	 add	#34,r4
#endif
	tst	r2,r2
	bt	.L_cleanup

1:	! Read longword, write two words per iteration
EX(	mov.l	@r5+,r0		)
	dt	r2
#ifdef CONFIG_CPU_LITTLE_ENDIAN
EX(	mov.w	r0,@r4		)
	shlr16	r0
EX(	mov.w 	r0,@(2,r4)	)
#else
EX(	mov.w	r0,@(2,r4)	)
	shlr16	r0
EX(	mov.w	r0,@r4		)
#endif
	bf/s	1b
	 add	#4,r4

	bra	.L_cleanup
	 nop

! Destination = 01 or 11

.L_dest01:
.L_dest11:
	! Read longword, write byte, word, byte per iteration
EX(	mov.l	@r5+,r0		)
	dt	r2
#ifdef CONFIG_CPU_LITTLE_ENDIAN
EX(	mov.b	r0,@r4		)
	shlr8	r0
	add	#1,r4
EX(	mov.w	r0,@r4		)
	shlr16	r0
EX(	mov.b	r0,@(2,r4)	)
	bf/s	.L_dest01
	 add	#3,r4
#else
EX(	mov.b	r0,@(3,r4)	)
	shlr8	r0
	swap.w	r0,r7
EX(	mov.b	r7,@r4		)
	add	#1,r4
EX(	mov.w	r0,@r4		)
	bf/s	.L_dest01
	 add	#3,r4
#endif

! Cleanup last few bytes
.L_cleanup:
	mov	r6,r0
	and	#3,r0
	tst	r0,r0
	bt	.L_exit
	mov	r0,r6

.L_cleanup_loop:
EX(	mov.b	@r5+,r0		)
	dt	r6
EX(	mov.b	r0,@r4		)
	bf/s	.L_cleanup_loop
	 add	#1,r4

.L_exit:
	mov	#0,r0		! normal return

5000:

# Exception handler:
.section .fixup, "ax"
6000:
	mov.l	8000f,r1
	mov	r3,r0
	jmp	@r1
	 sub	r4,r0
	.align	2
8000:	.long	5000b

.previous
	mov.l	@r15+,r8
	mov.l	@r15+,r9
	mov.l	@r15+,r10
	rts
	 mov.l	@r15+,r11