linux/arch/sparc/lib/memscan_64.S

/*
 * memscan.S: Optimized memscan for Sparc64.
 *
 * Copyright (C) 1997,1998 Jakub Jelinek (jj@ultra.linux.cz)
 * Copyright (C) 1998 David S. Miller (davem@redhat.com)
 */

#define HI_MAGIC	0x8080808080808080
#define LO_MAGIC	0x0101010101010101
#define ASI_PL		0x88

	.text
	.align	32
	.globl		__memscan_zero, __memscan_generic
	.globl		memscan

__memscan_zero:
	/* %o0 = bufp, %o1 = size */
	brlez,pn	%o1, szzero
	 andcc		%o0, 7, %g0
	be,pt		%icc, we_are_aligned
	 sethi		%hi(HI_MAGIC), %o4
	ldub		[%o0], %o5
1:	subcc		%o1, 1, %o1
	brz,pn		%o5, 10f
	 add		%o0, 1, %o0

	be,pn		%xcc, szzero
	 andcc		%o0, 7, %g0
	bne,a,pn	%icc, 1b
	 ldub		[%o0], %o5
we_are_aligned:
	ldxa		[%o0] ASI_PL, %o5
	or		%o4, %lo(HI_MAGIC), %o3
	sllx		%o3, 32, %o4
	or		%o4, %o3, %o3

	srlx		%o3, 7, %o2
msloop:
	sub		%o1, 8, %o1
	add		%o0, 8, %o0
	sub		%o5, %o2, %o4
	xor		%o4, %o5, %o4
	andcc		%o4, %o3, %g3
	bne,pn		%xcc, check_bytes
	 srlx		%o4, 32, %g3

	brgz,a,pt	%o1, msloop
	 ldxa		[%o0] ASI_PL, %o5
check_bytes:
	bne,a,pn	%icc, 2f
	 andcc		%o5, 0xff, %g0
	add		%o0, -5, %g2
	ba,pt		%xcc, 3f
	 srlx		%o5, 32, %g7

2:	srlx		%o5, 8, %g7
	be,pn		%icc, 1f
	 add		%o0, -8, %g2
	andcc		%g7, 0xff, %g0
	srlx		%g7, 8, %g7
	be,pn		%icc, 1f
	 inc		%g2
	andcc		%g7, 0xff, %g0

	srlx		%g7, 8, %g7
	be,pn		%icc, 1f
	 inc		%g2
	andcc		%g7, 0xff, %g0
	srlx		%g7, 8, %g7
	be,pn		%icc, 1f
	 inc		%g2
	andcc		%g3, %o3, %g0

	be,a,pn		%icc, 2f
	 mov		%o0, %g2
3:	andcc		%g7, 0xff, %g0
	srlx		%g7, 8, %g7
	be,pn		%icc, 1f
	 inc		%g2
	andcc		%g7, 0xff, %g0
	srlx		%g7, 8, %g7

	be,pn		%icc, 1f
	 inc		%g2
	andcc		%g7, 0xff, %g0
	srlx		%g7, 8, %g7
	be,pn		%icc, 1f
	 inc		%g2
	andcc		%g7, 0xff, %g0
	srlx		%g7, 8, %g7

	be,pn		%icc, 1f
	 inc		%g2
2:	brgz,a,pt	%o1, msloop
	 ldxa		[%o0] ASI_PL, %o5
	inc		%g2
1:	add		%o0, %o1, %o0
	cmp		%g2, %o0
	retl

	 movle		%xcc, %g2, %o0
10:	retl
	 sub		%o0, 1, %o0
szzero:	retl
	 nop

memscan:
__memscan_generic:
	/* %o0 = addr, %o1 = c, %o2 = size */
	brz,pn		%o2, 3f
	 add		%o0, %o2, %o3
	ldub		[%o0], %o5
	sub		%g0, %o2, %o4
1:
	cmp		%o5, %o1
	be,pn		%icc, 2f
	 addcc		%o4, 1, %o4
	bne,a,pt 	%xcc, 1b
	 ldub		[%o3 + %o4], %o5
	retl
	/* The delay slot is the same as the next insn, this is just to make it look more awful */
2:
	 add		%o3, %o4, %o0
	retl
	 sub		%o0, 1, %o0
3:
	retl
	 nop
sparc64: remove CVS keywords This patch removes the CVS keywords that weren't updated for a long time from comments. Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> 2008-05-20 03:52:27 +04:00			`/*`
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip! 2005-04-17 02:20:36 +04:00			`* memscan.S: Optimized memscan for Sparc64.`
			`*`
			`* Copyright (C) 1997,1998 Jakub Jelinek (jj@ultra.linux.cz)`
			`* Copyright (C) 1998 David S. Miller (davem@redhat.com)`
			`*/`

			`#define HI_MAGIC 0x8080808080808080`
			`#define LO_MAGIC 0x0101010101010101`
			`#define ASI_PL 0x88`

			`.text`
			`.align 32`
			`.globl __memscan_zero, __memscan_generic`
			`.globl memscan`

			`__memscan_zero:`
			`/* %o0 = bufp, %o1 = size */`
			`brlez,pn %o1, szzero`
			`andcc %o0, 7, %g0`
			`be,pt %icc, we_are_aligned`
			`sethi %hi(HI_MAGIC), %o4`
			`ldub [%o0], %o5`
			`1: subcc %o1, 1, %o1`
			`brz,pn %o5, 10f`
			`add %o0, 1, %o0`

			`be,pn %xcc, szzero`
			`andcc %o0, 7, %g0`
			`bne,a,pn %icc, 1b`
			`ldub [%o0], %o5`
			`we_are_aligned:`
			`ldxa [%o0] ASI_PL, %o5`
			`or %o4, %lo(HI_MAGIC), %o3`
			`sllx %o3, 32, %o4`
			`or %o4, %o3, %o3`

			`srlx %o3, 7, %o2`
			`msloop:`
			`sub %o1, 8, %o1`
			`add %o0, 8, %o0`
			`sub %o5, %o2, %o4`
			`xor %o4, %o5, %o4`
			`andcc %o4, %o3, %g3`
			`bne,pn %xcc, check_bytes`
			`srlx %o4, 32, %g3`

			`brgz,a,pt %o1, msloop`
			`ldxa [%o0] ASI_PL, %o5`
			`check_bytes:`
			`bne,a,pn %icc, 2f`
			`andcc %o5, 0xff, %g0`
			`add %o0, -5, %g2`
			`ba,pt %xcc, 3f`
			`srlx %o5, 32, %g7`

			`2: srlx %o5, 8, %g7`
			`be,pn %icc, 1f`
			`add %o0, -8, %g2`
			`andcc %g7, 0xff, %g0`
			`srlx %g7, 8, %g7`
			`be,pn %icc, 1f`
			`inc %g2`
			`andcc %g7, 0xff, %g0`

			`srlx %g7, 8, %g7`
			`be,pn %icc, 1f`
			`inc %g2`
			`andcc %g7, 0xff, %g0`
			`srlx %g7, 8, %g7`
			`be,pn %icc, 1f`
			`inc %g2`
			`andcc %g3, %o3, %g0`

			`be,a,pn %icc, 2f`
			`mov %o0, %g2`
			`3: andcc %g7, 0xff, %g0`
			`srlx %g7, 8, %g7`
			`be,pn %icc, 1f`
			`inc %g2`
			`andcc %g7, 0xff, %g0`
			`srlx %g7, 8, %g7`

			`be,pn %icc, 1f`
			`inc %g2`
			`andcc %g7, 0xff, %g0`
			`srlx %g7, 8, %g7`
			`be,pn %icc, 1f`
			`inc %g2`
			`andcc %g7, 0xff, %g0`
			`srlx %g7, 8, %g7`

			`be,pn %icc, 1f`
			`inc %g2`
			`2: brgz,a,pt %o1, msloop`
			`ldxa [%o0] ASI_PL, %o5`
			`inc %g2`
			`1: add %o0, %o1, %o0`
			`cmp %g2, %o0`
			`retl`

			`movle %xcc, %g2, %o0`
			`10: retl`
			`sub %o0, 1, %o0`
			`szzero: retl`
			`nop`

			`memscan:`
			`__memscan_generic:`
			`/* %o0 = addr, %o1 = c, %o2 = size */`
			`brz,pn %o2, 3f`
			`add %o0, %o2, %o3`
			`ldub [%o0], %o5`
			`sub %g0, %o2, %o4`
			`1:`
			`cmp %o5, %o1`
			`be,pn %icc, 2f`
			`addcc %o4, 1, %o4`
			`bne,a,pt %xcc, 1b`
			`ldub [%o3 + %o4], %o5`
			`retl`
			`/* The delay slot is the same as the next insn, this is just to make it look more awful */`
			`2:`
			`add %o3, %o4, %o0`
			`retl`
			`sub %o0, 1, %o0`
			`3:`
			`retl`
			`nop`