pa-fazil
/
android_bionic
mirror of https://github.com/fazilsheik96/android_bionic.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
android_bionic/libc/arch-arm/cortex-a9/bionic/strcpy.S

462 lines
12 KiB
ArmAsm
Raw Normal View History

Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
/*
* Copyright (C) 2013 The Android Open Source Project
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
* OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
* AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
* OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
/*
* Copyright (c) 2013 ARM Ltd
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. The name of the company may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS'' AND ANY EXPRESS OR IMPLIED
* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
* TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
* NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
Unify our assembler macros. Our <machine/asm.h> files were modified from upstream, to the extent that no architecture was actually using the upstream ENTRY or END macros, assuming that architecture even had such a macro upstream. This patch moves everyone to the same macros, with just a few tweaks remaining in the <machine/asm.h> files, which no one should now use directly. I've removed most of the unused cruft from the <machine/asm.h> files, though there's still rather a lot in the mips/mips64 ones. Bug: 12229603 Change-Id: I2fff287dc571ac1087abe9070362fb9420d85d6d
2014-02-20 00:53:20 +00:00
#include <private/bionic_asm.h>
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
.syntax unified
.thumb
.thumb_func
.macro m_push
push {r0, r4, r5, lr}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.cfi_def_cfa_offset 16
.cfi_rel_offset r0, 0
.cfi_rel_offset r4, 4
.cfi_rel_offset r5, 8
.cfi_rel_offset lr, 12
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
.endm // m_push
.macro m_ret inst
\inst {r0, r4, r5, pc}
.endm // m_ret
.macro m_copy_byte reg, cmd, label
ldrb \reg, [r1], #1
strb \reg, [r0], #1
\cmd \reg, \label
.endm // m_copy_byte
ENTRY(strcpy)
// Unroll the first 8 bytes that will be copied.
m_push
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
m_copy_byte reg=r2, cmd=cbz, label=.Lstrcpy_finish
m_copy_byte reg=r3, cmd=cbz, label=.Lstrcpy_finish
m_copy_byte reg=r4, cmd=cbz, label=.Lstrcpy_finish
m_copy_byte reg=r5, cmd=cbz, label=.Lstrcpy_finish
m_copy_byte reg=r2, cmd=cbz, label=.Lstrcpy_finish
m_copy_byte reg=r3, cmd=cbz, label=.Lstrcpy_finish
m_copy_byte reg=r4, cmd=cbz, label=.Lstrcpy_finish
m_copy_byte reg=r5, cmd=cbnz, label=.Lstrcpy_continue
.Lstrcpy_finish:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_continue:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
pld [r1, #0]
ands r3, r0, #7
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_align_dst
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_check_src_align:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
// At this point dst is aligned to a double word, check if src
// is also aligned to a double word.
ands r3, r1, #7
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_unaligned_copy
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
.p2align 2
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_mainloop:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldmia r1!, {r2, r3}
pld [r1, #64]
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_first_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_second_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
b .Lstrcpy_mainloop
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_zero_in_first_register:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
lsls lr, ip, #17
itt ne
strbne r2, [r0]
m_ret inst=popne
itt cs
strhcs r2, [r0]
m_ret inst=popcs
lsls ip, ip, #1
itt eq
streq r2, [r0]
m_ret inst=popeq
strh r2, [r0], #2
lsr r3, r2, #16
strb r3, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_zero_in_second_register:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
lsls lr, ip, #17
ittt ne
stmiane r0!, {r2}
strbne r3, [r0]
m_ret inst=popne
ittt cs
strcs r2, [r0], #4
strhcs r3, [r0]
m_ret inst=popcs
lsls ip, ip, #1
itt eq
stmiaeq r0, {r2, r3}
m_ret inst=popeq
stmia r0!, {r2}
strh r3, [r0], #2
lsr r4, r3, #16
strb r4, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_align_dst:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
// Align to a double word (64 bits).
rsb r3, r3, #8
lsls ip, r3, #31
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_align_to_32
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r2, [r1], #1
strb r2, [r0], #1
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r2, .Lstrcpy_complete
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_align_to_32:
bcc .Lstrcpy_align_to_64
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r4, [r1], #1
strb r4, [r0], #1
cmp r4, #0
it eq
m_ret inst=popeq
ldrb r5, [r1], #1
strb r5, [r0], #1
cmp r5, #0
it eq
m_ret inst=popeq
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_align_to_64:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
tst r3, #4
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_check_src_align
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_first_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
b .Lstrcpy_check_src_align
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_complete:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unaligned_copy:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
// Dst is aligned to a double word, while src is at an unknown alignment.
// There are 7 different versions of the unaligned copy code
// to prevent overreading the src. The mainloop of every single version
// will store 64 bits per loop. The difference is how much of src can
// be read without potentially crossing a page boundary.
tbb [pc, r3]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unaligned_branchtable:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
.byte 0
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.byte ((.Lstrcpy_unalign7 - .Lstrcpy_unaligned_branchtable)/2)
.byte ((.Lstrcpy_unalign6 - .Lstrcpy_unaligned_branchtable)/2)
.byte ((.Lstrcpy_unalign5 - .Lstrcpy_unaligned_branchtable)/2)
.byte ((.Lstrcpy_unalign4 - .Lstrcpy_unaligned_branchtable)/2)
.byte ((.Lstrcpy_unalign3 - .Lstrcpy_unaligned_branchtable)/2)
.byte ((.Lstrcpy_unalign2 - .Lstrcpy_unaligned_branchtable)/2)
.byte ((.Lstrcpy_unalign1 - .Lstrcpy_unaligned_branchtable)/2)
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
.p2align 2
// Can read 7 bytes before possibly crossing a page.
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign7:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_first_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r3, [r1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r3, .Lstrcpy_unalign7_copy5bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r4, [r1, #1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r4, .Lstrcpy_unalign7_copy6bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r5, [r1, #2]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r5, .Lstrcpy_unalign7_copy7bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r3, [r1], #4
pld [r1, #64]
lsrs ip, r3, #24
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_unalign_return
b .Lstrcpy_unalign7
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign7_copy5bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2}
strb r3, [r0]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign_return:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign7_copy6bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2}
strb r3, [r0], #1
strb r4, [r0], #1
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign7_copy7bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2}
strb r3, [r0], #1
strb r4, [r0], #1
strb r5, [r0], #1
m_ret inst=pop
.p2align 2
// Can read 6 bytes before possibly crossing a page.
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign6:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_first_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r4, [r1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r4, .Lstrcpy_unalign_copy5bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r5, [r1, #1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r5, .Lstrcpy_unalign_copy6bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r3, [r1], #4
pld [r1, #64]
tst r3, #0xff0000
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_unalign6_copy7bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
lsrs ip, r3, #24
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_unalign_return
b .Lstrcpy_unalign6
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign6_copy7bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2}
strh r3, [r0], #2
lsr r3, #16
strb r3, [r0]
m_ret inst=pop
.p2align 2
// Can read 5 bytes before possibly crossing a page.
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign5:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_first_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r4, [r1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r4, .Lstrcpy_unalign_copy5bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r3, [r1], #4
pld [r1, #64]
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_second_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
b .Lstrcpy_unalign5
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign_copy5bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2}
strb r4, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign_copy6bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2}
strb r4, [r0], #1
strb r5, [r0]
m_ret inst=pop
.p2align 2
// Can read 4 bytes before possibly crossing a page.
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign4:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldmia r1!, {r2}
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_first_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldmia r1!, {r3}
pld [r1, #64]
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_second_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
b .Lstrcpy_unalign4
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
.p2align 2
// Can read 3 bytes before possibly crossing a page.
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign3:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r2, [r1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r2, .Lstrcpy_unalign3_copy1byte
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r3, [r1, #1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r3, .Lstrcpy_unalign3_copy2bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r4, [r1, #2]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r4, .Lstrcpy_unalign3_copy3bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r2, [r1], #4
ldr r3, [r1], #4
pld [r1, #64]
lsrs lr, r2, #24
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_unalign_copy4bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_second_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
b .Lstrcpy_unalign3
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign3_copy1byte:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
strb r2, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign3_copy2bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
strb r2, [r0], #1
strb r3, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign3_copy3bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
strb r2, [r0], #1
strb r3, [r0], #1
strb r4, [r0]
m_ret inst=pop
.p2align 2
// Can read 2 bytes before possibly crossing a page.
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign2:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r2, [r1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r2, .Lstrcpy_unalign_copy1byte
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r3, [r1, #1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r3, .Lstrcpy_unalign_copy2bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r2, [r1], #4
ldr r3, [r1], #4
pld [r1, #64]
tst r2, #0xff0000
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_unalign_copy3bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
lsrs ip, r2, #24
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
beq .Lstrcpy_unalign_copy4bytes
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_second_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
b .Lstrcpy_unalign2
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
.p2align 2
// Can read 1 byte before possibly crossing a page.
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign1:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldrb r2, [r1]
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
cbz r2, .Lstrcpy_unalign_copy1byte
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
ldr r2, [r1], #4
ldr r3, [r1], #4
pld [r1, #64]
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_first_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
bne .Lstrcpy_zero_in_second_register
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0!, {r2, r3}
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
b .Lstrcpy_unalign1
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign_copy1byte:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
strb r2, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign_copy2bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
strb r2, [r0], #1
strb r3, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign_copy3bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
strh r2, [r0], #2
lsr r2, #16
strb r2, [r0]
m_ret inst=pop
Cleanup arm assembly. Remove the old arm directives. Change the non-local labels to .L labels. Add cfi directives to strcpy.S. Change-Id: I9bafee1ffe5d85c92d07cfa8a85338cef9759562
2014-09-29 22:34:20 +00:00
.Lstrcpy_unalign_copy4bytes:
Optimize strcat/strcpy, small tweaks to strlen. DO NOT MERGE Create one version of strcat/strcpy/strlen for cortex-a15/krait and another version for cortex-a9. Tested with the libc_test strcat/strcpy/strlen tests. Including new tests that verify that the src for strcat/strcpy do not overread across page boundaries. NOTE: The handling of unaligned strcpy (same code in strcat) could probably be optimized further such that the src is read 64 bits at a time instead of the partial reads occurring now. strlen improves slightly since it was recently optimized. Performance improvements for strcpy and strcat (using an empty dest string): cortex-a9 - Small copies vary from about 5% to 20% as the size gets above 10 bytes. - Copies >= 1024, about a 60% improvement. - Unaligned copies, from about 40% improvement. cortex-a15 - Most small copies exhibit a 100% improvement, a few copies only improve by 20%. - Copies >= 1024, about 150% improvement. - Unaligned copies, about 100% improvement. krait - Most small copies vary widely, but on average 20% improvement, then the performance gets better, hitting about a 100% improvement when copies 64 bytes of data. - Copies >= 1024, about 100% improvement. - When coping MBs of data, about 50% improvement. - Unaligned copies, about 90% improvement. As strcat destination strings get larger in size: cortex-a9 - about 40% improvement for small dst strings (>= 32). - about 250% improvement for dst strings >= 1024. cortex-a15 - about 200% improvement for small dst strings (>=32). - about 250% improvement for dst strings >= 1024. krait - about 25% improvement for small dst strings (>=32). - about 100% improvement for dst strings >=1024. Merge from internal master. (cherry-picked from d119b7b6f48fe507088cfb98bcafa99b320fd884) Change-Id: I296463b251ef9fab004ee4dded2793feca5b547a
2013-07-15 19:49:26 +00:00
stmia r0, {r2}
m_ret inst=pop
END(strcpy)