Browse Source

Fix whitespace damage in AVR32 string assembler optimized functions.

Hans-Christian Egtvedt 16 years ago
parent
commit
1995bf637f

+ 2 - 2
libc/string/avr32/Makefile

@@ -16,8 +16,8 @@
 # along with this program; if not, write to the Free Software Foundation, Inc.,
 # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
 
-top_srcdir     := ../../../
-top_builddir   := ../../../
+top_srcdir	:= ../../../
+top_builddir	:= ../../../
 
 all: objs
 

+ 10 - 10
libc/string/avr32/bcopy.S

@@ -10,17 +10,17 @@
 
 #ifdef __UCLIBC_SUSV3_LEGACY__
 
-       .text
-       .global bcopy
-       .type   bcopy, @function
-       .align  1
+	.text
+	.global bcopy
+	.type	bcopy, @function
+	.align	1
 bcopy:
-       /* Swap the first two arguments */
-       eor     r11, r12
-       eor     r12, r11
-       eor     r11, r12
-       rjmp    HIDDEN_JUMPTARGET(memmove)
+	/* Swap the first two arguments */
+	eor	r11, r12
+	eor	r12, r11
+	eor	r11, r12
+	rjmp	HIDDEN_JUMPTARGET(memmove)
 
-       .size   bcopy, . - bcopy
+	.size	bcopy, . - bcopy
 
 #endif /* __UCLIBC_SUSV3_LEGACY__ */

+ 8 - 8
libc/string/avr32/bzero.S

@@ -10,15 +10,15 @@
 
 #ifdef __UCLIBC_SUSV3_LEGACY__
 
-       .text
-       .global bzero
-       .type   bzero, @function
-       .align  1
+	.text
+	.global bzero
+	.type	bzero, @function
+	.align	1
 bzero:
-       mov     r10, r11
-       mov     r11, 0
-       rjmp    HIDDEN_JUMPTARGET(memset)
+	mov	r10, r11
+	mov	r11, 0
+	rjmp	HIDDEN_JUMPTARGET(memset)
 
-       .size   bzero, . - bzero
+	.size	bzero, . - bzero
 
 #endif /* __UCLIBC_SUSV3_LEGACY__ */

+ 34 - 34
libc/string/avr32/memcmp.S

@@ -12,48 +12,48 @@
 #define s2 r11
 #define len r10
 
-       .text
-       .global memcmp
-       .type   memcmp, @function
-       .align  1
+	.text
+	.global memcmp
+	.type	memcmp, @function
+	.align	1
 memcmp:
-       sub     len, 4
-       brlt    .Lless_than_4
+	sub	len, 4
+	brlt	.Lless_than_4
 
-1:     ld.w    r8, s1++
-       ld.w    r9, s2++
-       cp.w    r8, r9
-       brne    .Lfound_word
-       sub     len, 4
-       brge    1b
+1:	ld.w	r8, s1++
+	ld.w	r9, s2++
+	cp.w	r8, r9
+	brne	.Lfound_word
+	sub	len, 4
+	brge	1b
 
 .Lless_than_4:
-       sub     len, -4
-       reteq   0
+	sub	len, -4
+	reteq	0
 
-1:     ld.ub   r8, s1++
-       ld.ub   r9, s2++
-       sub     r8, r9
-       retne   r8
-       sub     len, 1
-       brgt    1b
+1:	ld.ub	r8, s1++
+	ld.ub	r9, s2++
+	sub	r8, r9
+	retne	r8
+	sub	len, 1
+	brgt	1b
 
-       retal   0
+	retal	0
 
 .Lfound_word:
-       mov     len, 4
-
-2:     bfextu  r11, r9, 24, 8
-       bfextu  r12, r8, 24, 8
-       sub     r12, r11
-       retne   r12
-       lsl     r8, 8
-       lsl     r9, 8
-       sub     len, 1
-       brne    2b
-       retal   r12
-
-       .size   memcmp, . - memcmp
+	mov	len, 4
+
+2:	bfextu	r11, r9, 24, 8
+	bfextu	r12, r8, 24, 8
+	sub	r12, r11
+	retne	r12
+	lsl	r8, 8
+	lsl	r9, 8
+	sub	len, 1
+	brne	2b
+	retal	r12
+
+	.size	memcmp, . - memcmp
 
 libc_hidden_def(memcmp)
 #ifdef __UCLIBC_SUSV3_LEGACY__

+ 78 - 78
libc/string/avr32/memcpy.S

@@ -11,101 +11,101 @@
 #define src r11
 #define len r10
 
-       .text
-       .global memcpy
-       .type   memcpy, @function
+	.text
+	.global	memcpy
+	.type	memcpy, @function
 memcpy:
-       pref    src[0]
-       mov     dst, r12
+	pref	src[0]
+	mov	dst, r12
 
-       /* If we have less than 32 bytes, don't do anything fancy */
-       cp.w    len, 32
-       brge    .Lmore_than_31
+	/* If we have less than 32 bytes, don't do anything fancy */
+	cp.w	len, 32
+	brge	.Lmore_than_31
 
-       sub     len, 1
-       retlt   r12
-1:     ld.ub   r8, src++
-       st.b    dst++, r8
-       sub     len, 1
-       brge    1b
-       retal   r12
+	sub	len, 1
+	retlt	r12
+1:	ld.ub	r8, src++
+	st.b	dst++, r8
+	sub	len, 1
+	brge	1b
+	retal	r12
 
 .Lmore_than_31:
-       pushm   r0-r7, lr
+	pushm	r0-r7, lr
 
-       /* Check alignment */
-       mov     r8, src
-       andl    r8, 31, COH
-       brne    .Lunaligned_src
-       mov     r8, dst
-       andl    r8, 3, COH
-       brne    .Lunaligned_dst
+	/* Check alignment */
+	mov	r8, src
+	andl	r8, 31, COH
+	brne	.Lunaligned_src
+	mov	r8, dst
+	andl	r8, 3, COH
+	brne	.Lunaligned_dst
 
 .Laligned_copy:
-       sub     len, 32
-       brlt    .Lless_than_32
+	sub	len, 32
+	brlt	.Lless_than_32
 
-1:     /* Copy 32 bytes at a time */
-       ldm     src, r0-r7
-       sub     src, -32
-       stm     dst, r0-r7
-       sub     dst, -32
-       sub     len, 32
-       brge    1b
+1:	/* Copy 32 bytes at a time */
+	ldm	src, r0-r7
+	sub	src, -32
+	stm	dst, r0-r7
+	sub	dst, -32
+	sub	len, 32
+	brge	1b
 
 .Lless_than_32:
-       /* Copy 16 more bytes if possible */
-       sub     len, -16
-       brlt    .Lless_than_16
-       ldm     src, r0-r3
-       sub     src, -16
-       sub     len, 16
-       stm     dst, r0-r3
-       sub     dst, -16
+	/* Copy 16 more bytes if possible */
+	sub	len, -16
+	brlt	.Lless_than_16
+	ldm	src, r0-r3
+	sub	src, -16
+	sub	len, 16
+	stm	dst, r0-r3
+	sub	dst, -16
 
 .Lless_than_16:
-       /* Do the remaining as byte copies */
-       neg     len
-       add     pc, pc, len << 2
-       .rept   15
-       ld.ub   r0, src++
-       st.b    dst++, r0
-       .endr
+	/* Do the remaining as byte copies */
+	neg	len
+	add	pc, pc, len << 2
+	.rept	15
+	ld.ub	r0, src++
+	st.b	dst++, r0
+	.endr
 
-       popm    r0-r7, pc
+	popm	r0-r7, pc
 
 .Lunaligned_src:
-       /* Make src cacheline-aligned. r8 = (src & 31) */
-       rsub    r8, r8, 32
-       sub     len, r8
-1:     ld.ub   r0, src++
-       st.b    dst++, r0
-       sub     r8, 1
-       brne    1b
-
-       /* If dst is word-aligned, we're ready to go */
-       pref    src[0]
-       mov     r8, 3
-       tst     dst, r8
-       breq    .Laligned_copy
+	/* Make src cacheline-aligned. r8 = (src & 31) */
+	rsub	r8, r8, 32
+	sub	len, r8
+1:	ld.ub	r0, src++
+	st.b	dst++, r0
+	sub	r8, 1
+	brne	1b
+
+	/* If dst is word-aligned, we're ready to go */
+	pref	src[0]
+	mov	r8, 3
+	tst	dst, r8
+	breq	.Laligned_copy
 
 .Lunaligned_dst:
-       /* src is aligned, but dst is not. Expect bad performance */
-       sub     len, 4
-       brlt    2f
-1:     ld.w    r0, src++
-       st.w    dst++, r0
-       sub     len, 4
-       brge    1b
-
-2:     neg     len
-       add     pc, pc, len << 2
-       .rept   3
-       ld.ub   r0, src++
-       st.b    dst++, r0
-       .endr
-
-       popm    r0-r7, pc
-       .size   memcpy, . - memcpy
+	/* src is aligned, but dst is not. Expect bad performance */
+	sub	len, 4
+	brlt	2f
+1:	ld.w	r0, src++
+	st.w	dst++, r0
+	sub	len, 4
+	brge	1b
+
+2:	neg	len
+	add	pc, pc, len << 2
+	.rept	3
+	ld.ub	r0, src++
+	st.b	dst++, r0
+	.endr
+
+	popm	r0-r7, pc
+	.size	memcpy, . - memcpy
 
 libc_hidden_def(memcpy)

+ 86 - 86
libc/string/avr32/memmove.S

@@ -10,107 +10,107 @@
 #define src r11
 #define len r10
 
-       .text
-       .global memmove
-       .type   memmove, @function
+	.text
+	.global memmove
+	.type	memmove, @function
 memmove:
-       cp.w    src, dst
-       brge    HIDDEN_JUMPTARGET(memcpy)
-
-       add     dst, len
-       add     src, len
-       pref    src[-1]
-
-       /*
-        * The rest is basically the same as in memcpy.S except that
-        * the direction is reversed.
-        */
-       cp.w    len, 32
-       brge    .Lmore_than_31
-
-       sub     len, 1
-       retlt   r12
-1:     ld.ub   r8, --src
-       st.b    --dst, r8
-       sub     len, 1
-       brge    1b
-       retal   r12
+	cp.w	src, dst
+	brge	HIDDEN_JUMPTARGET(memcpy)
+
+	add	dst, len
+	add	src, len
+	pref	src[-1]
+
+	/*
+	 * The rest is basically the same as in memcpy.S except that
+	 * the direction is reversed.
+	 */
+	cp.w	len, 32
+	brge	.Lmore_than_31
+
+	sub	len, 1
+	retlt	r12
+1:	ld.ub	r8, --src
+	st.b	--dst, r8
+	sub	len, 1
+	brge	1b
+	retal	r12
 
 .Lmore_than_31:
-       pushm   r0-r7, lr
+	pushm	r0-r7, lr
 
-       /* Check alignment */
-       mov     r8, src
-       andl    r8, 31, COH
-       brne    .Lunaligned_src
-       mov     r8, r12
-       andl    r8, 3, COH
-       brne    .Lunaligned_dst
+/* Check alignment */
+	mov	r8, src
+	andl	r8, 31, COH
+	brne	.Lunaligned_src
+	mov	r8, r12
+	andl	r8, 3, COH
+	brne	.Lunaligned_dst
 
 .Laligned_copy:
-       sub     len, 32
-       brlt    .Lless_than_32
+	sub	len, 32
+	brlt	.Lless_than_32
 
-1:     /* Copy 32 bytes at a time */
-       sub     src, 32
-       ldm     src, r0-r7
-       sub     dst, 32
-       sub     len, 32
-       stm     dst, r0-r7
-       brge    1b
+1:	/* Copy 32 bytes at a time */
+	sub	src, 32
+	ldm	src, r0-r7
+	sub	dst, 32
+	sub	len, 32
+	stm	dst, r0-r7
+	brge	1b
 
 .Lless_than_32:
-       /* Copy 16 more bytes if possible */
-       sub     len, -16
-       brlt    .Lless_than_16
-       sub     src, 16
-       ldm     src, r0-r3
-       sub     dst, 16
-       sub     len, 16
-       stm     dst, r0-r3
+	/* Copy 16 more bytes if possible */
+	sub	len, -16
+	brlt	.Lless_than_16
+	sub	src, 16
+	ldm	src, r0-r3
+	sub	dst, 16
+	sub	len, 16
+	stm	dst, r0-r3
 
 .Lless_than_16:
-       /* Do the remaining as byte copies */
-       sub     len, -16
-       breq    2f
-1:     ld.ub   r0, --src
-       st.b    --dst, r0
-       sub     len, 1
-       brne    1b
+	/* Do the remaining as byte copies */
+	sub	len, -16
+	breq	2f
+1:	ld.ub	r0, --src
+	st.b	--dst, r0
+	sub	len, 1
+	brne	1b
 
-2:     popm    r0-r7, pc
+2:	popm	r0-r7, pc
 
 .Lunaligned_src:
-       /* Make src cacheline-aligned. r8 = (src & 31) */
-       sub     len, r8
-1:     ld.ub   r0, --src
-       st.b    --dst, r0
-       sub     r8, 1
-       brne    1b
-
-       /* If dst is word-aligned, we're ready to go */
-       pref    src[-4]
-       mov     r8, 3
-       tst     dst, r8
-       breq    .Laligned_copy
+	/* Make src cacheline-aligned. r8 = (src & 31) */
+	sub	len, r8
+1:	ld.ub	r0, --src
+	st.b	--dst, r0
+	sub	r8, 1
+	brne	1b
+
+	/* If dst is word-aligned, we're ready to go */
+	pref	src[-4]
+	mov	r8, 3
+	tst	dst, r8
+	breq	.Laligned_copy
 
 .Lunaligned_dst:
-       /* src is aligned, but dst is not. Expect bad performance */
-       sub     len, 4
-       brlt    2f
-1:     ld.w    r0, --src
-       st.w    --dst, r0
-       sub     len, 4
-       brge    1b
-
-2:     neg     len
-       add     pc, pc, len << 2
-       .rept   3
-       ld.ub   r0, --src
-       st.b    --dst, r0
-       .endr
-
-       popm    r0-r7, pc
-       .size   memmove, . - memmove
+	/* src is aligned, but dst is not. Expect bad performance */
+	sub	len, 4
+	brlt	2f
+1:	ld.w	r0, --src
+	st.w	--dst, r0
+	sub	len, 4
+	brge	1b
+
+2:	neg	len
+	add	pc, pc, len << 2
+	.rept	3
+	ld.ub	r0, --src
+	st.b	--dst, r0
+	.endr
+
+	popm	r0-r7, pc
+	.size	memmove, . - memmove
 
 libc_hidden_def(memmove)

+ 38 - 38
libc/string/avr32/memset.S

@@ -12,54 +12,54 @@
 #define c r11
 #define n r10
 
-       .text
-       .global memset
-       .type   memset, @function
+	.text
+	.global memset
+	.type	memset, @function
 
-       .align  1
+	.align	1
 memset:
-       cp.w    n, 32
-       mov     r9, s
-       brge    .Llarge_memset
+	cp.w	n, 32
+	mov	r9, s
+	brge	.Llarge_memset
 
-       sub     n, 1
-       retlt   s
-1:     st.b    s++, c
-       sub     n, 1
-       brge    1b
+	sub	n, 1
+	retlt	s
+1:	st.b	s++, c
+	sub	n, 1
+	brge	1b
 
-       retal   r9
+	retal	r9
 
 .Llarge_memset:
-       mov     r8, r11
-       mov     r11, 3
-       bfins   r8, r8, 8, 8
-       bfins   r8, r8, 16, 16
-       tst     s, r11
-       breq    2f
+	mov	r8, r11
+	mov	r11, 3
+	bfins	r8, r8, 8, 8
+	bfins	r8, r8, 16, 16
+	tst	s, r11
+	breq	2f
 
-1:     st.b    s++, r8
-       sub     n, 1
-       tst     s, r11
-       brne    1b
+1:	st.b	s++, r8
+	sub	n, 1
+	tst	s, r11
+	brne	1b
 
-2:     mov     r11, r9
-       mov     r9, r8
-       sub     n, 8
+2:	mov	r11, r9
+	mov	r9, r8
+	sub	n, 8
 
-3:     st.d    s++, r8
-       sub     n, 8
-       brge    3b
+3:	st.d	s++, r8
+	sub	n, 8
+	brge	3b
 
-       /* If we are done, n == -8 and we'll skip all st.b insns below */
-       neg     n
-       lsl     n, 1
-       add     pc, n
-       .rept   7
-       st.b    s++, r8
-       .endr
-       retal   r11
+	/* If we are done, n == -8 and we'll skip all st.b insns below */
+	neg	n
+	lsl	n, 1
+	add	pc, n
+	.rept	7
+	st.b	s++, r8
+	.endr
+	retal	r11
 
-       .size   memset, . - memset
+	.size	memset, . - memset
 
 libc_hidden_def(memset)

+ 62 - 62
libc/string/avr32/strcmp.S

@@ -12,77 +12,77 @@
 #define s2 r11
 #define len r10
 
-       .text
-       .global strcmp
-       .type   strcmp, @function
-       .align  1
+	.text
+	.global strcmp
+	.type	strcmp, @function
+	.align	1
 strcmp:
-       mov     r8, 3
-       tst     s1, r8
-       brne    .Lunaligned_s1
-       tst     s2, r8
-       brne    .Lunaligned_s2
+	mov	r8, 3
+	tst	s1, r8
+	brne	.Lunaligned_s1
+	tst	s2, r8
+	brne	.Lunaligned_s2
 
-1:     ld.w    r8, s1++
-       ld.w    r9, s2++
-       cp.w    r8, r9
-       brne    2f
-       tnbz    r8
-       brne    1b
-       retal   0
+1:	ld.w	r8, s1++
+	ld.w	r9, s2++
+	cp.w	r8, r9
+	brne	2f
+	tnbz	r8
+	brne	1b
+	retal	0
 
-2:     bfextu  r12, r8, 24, 8
-       bfextu  r11, r9, 24, 8
-       sub     r12, r11
-       retne   r12
-       cp.w    r11, 0
-       reteq   0
-       bfextu  r12, r8, 16, 8
-       bfextu  r11, r9, 16, 8
-       sub     r12, r11
-       retne   r12
-       cp.w    r11, 0
-       reteq   0
-       bfextu  r12, r8, 8, 8
-       bfextu  r11, r9, 8, 8
-       sub     r12, r11
-       retne   r12
-       cp.w    r11, 0
-       reteq   0
-       bfextu  r12, r8, 0, 8
-       bfextu  r11, r9, 0, 8
-       sub     r12, r11
-       retal   r12
+2:	bfextu	r12, r8, 24, 8
+	bfextu	r11, r9, 24, 8
+	sub	r12, r11
+	retne	r12
+	cp.w	r11, 0
+	reteq	0
+	bfextu	r12, r8, 16, 8
+	bfextu	r11, r9, 16, 8
+	sub	r12, r11
+	retne	r12
+	cp.w	r11, 0
+	reteq	0
+	bfextu	r12, r8, 8, 8
+	bfextu	r11, r9, 8, 8
+	sub	r12, r11
+	retne	r12
+	cp.w	r11, 0
+	reteq	0
+	bfextu	r12, r8, 0, 8
+	bfextu	r11, r9, 0, 8
+	sub	r12, r11
+	retal	r12
 
 .Lunaligned_s1:
-3:     tst     s1, r8
-       breq    4f
-       ld.ub   r10, s1++
-       ld.ub   r9, s2++
-       sub     r10, r9
-       retne   r10
-       cp.w    r9, 0
-       brne    3b
-       retal   r10
+3:	tst	s1, r8
+	breq	4f
+	ld.ub	r10, s1++
+	ld.ub	r9, s2++
+	sub	r10, r9
+	retne	r10
+	cp.w	r9, 0
+	brne	3b
+	retal	r10
 
-4:     tst     s2, r8
-       breq    1b
+4:	tst	s2, r8
+	breq	1b
 
 .Lunaligned_s2:
-       /*
-        * s1 and s2 can't both be aligned, and unaligned word loads
-        * can trigger spurious exceptions if we cross a page boundary.
-        * Do it the slow way...
-        */
-1:     ld.ub   r8, s1++
-       ld.ub   r9, s2++
-       sub     r8, r9
-       retne   r8
-       cp.w    r9, 0
-       brne    1b
-       retal   0
+	/*
+	 * s1 and s2 can't both be aligned, and unaligned word loads
+	 * can trigger spurious exceptions if we cross a page boundary.
+	 * Do it the slow way...
+	 */
+1:	ld.ub	r8, s1++
+	ld.ub	r9, s2++
+	sub	r8, r9
+	retne	r8
+	cp.w	r9, 0
+	brne	1b
+	retal	0
 
-       .size   strcmp, . - strcmp
+	.size	strcmp, . - strcmp
 
 libc_hidden_def(strcmp)
 #ifndef __UCLIBC_HAS_LOCALE__

+ 45 - 45
libc/string/avr32/strlen.S

@@ -10,53 +10,53 @@
 
 #define str r12
 
-       .text
-       .global strlen
-       .type   strlen, @function
+	.text
+	.global strlen
+	.type	strlen, @function
 strlen:
-       mov     r11, r12
-
-       mov     r9, str
-       andl    r9, 3, COH
-       brne    .Lunaligned_str
-
-1:     ld.w    r8, str++
-       tnbz    r8
-       brne    1b
-
-       sub     r12, r11
-       bfextu  r9, r8, 24, 8
-       cp.w    r9, 0
-       subeq   r12, 4
-       reteq   r12
-       bfextu  r9, r8, 16, 8
-       cp.w    r9, 0
-       subeq   r12, 3
-       reteq   r12
-       bfextu  r9, r8, 8, 8
-       cp.w    r9, 0
-       subeq   r12, 2
-       reteq   r12
-       sub     r12, 1
-       retal   r12
+	mov	r11, r12
+
+	mov	r9, str
+	andl	r9, 3, COH
+	brne	.Lunaligned_str
+
+1:	ld.w	r8, str++
+	tnbz	r8
+	brne	1b
+
+	sub	r12, r11
+	bfextu	r9, r8, 24, 8
+	cp.w	r9, 0
+	subeq	r12, 4
+	reteq	r12
+	bfextu	r9, r8, 16, 8
+	cp.w	r9, 0
+	subeq	r12, 3
+	reteq	r12
+	bfextu	r9, r8, 8, 8
+	cp.w	r9, 0
+	subeq	r12, 2
+	reteq	r12
+	sub	r12, 1
+	retal	r12
 
 .Lunaligned_str:
-       add     pc, pc, r9 << 3
-       sub     r0, r0, 0       /* 4-byte nop */
-       ld.ub   r8, str++
-       sub     r8, r8, 0
-       breq    1f
-       ld.ub   r8, str++
-       sub     r8, r8, 0
-       breq    1f
-       ld.ub   r8, str++
-       sub     r8, r8, 0
-       brne    1b
-
-1:     sub     r12, 1
-       sub     r12, r11
-       retal   r12
-
-       .size   strlen, . - strlen
+	add	pc, pc, r9 << 3
+	sub	r0, r0, 0	/* 4-byte nop */
+	ld.ub	r8, str++
+	sub	r8, r8, 0
+	breq	1f
+	ld.ub	r8, str++
+	sub	r8, r8, 0
+	breq	1f
+	ld.ub	r8, str++
+	sub	r8, r8, 0
+	brne	1b
+
+1:	sub	r12, 1
+	sub	r12, r11
+	retal	r12
+
+	.size	strlen, . - strlen
 
 libc_hidden_def(strlen)