.macro m_scan_byte
ldrb r3, [r0]
- cbz r3, strcat_r0_scan_done
+ cbz r3, .L_strcat_r0_scan_done
add r0, #1
.endm // m_scan_byte
// Quick check to see if src is empty.
ldrb r2, [r1]
pld [r1, #0]
- cbnz r2, strcat_continue
+ cbnz r2, .L_strcat_continue
bx lr
-strcat_continue:
+.L_strcat_continue:
// To speed up really small dst strings, unroll checking the first 4 bytes.
m_push
m_scan_byte
m_scan_byte
ands r3, r0, #7
- beq strcat_mainloop
+ beq .L_strcat_mainloop
// Align to a double word (64 bits).
rsb r3, r3, #8
lsls ip, r3, #31
- beq strcat_align_to_32
+ beq .L_strcat_align_to_32
ldrb r5, [r0]
- cbz r5, strcat_r0_scan_done
+ cbz r5, .L_strcat_r0_scan_done
add r0, r0, #1
-strcat_align_to_32:
- bcc strcat_align_to_64
+.L_strcat_align_to_32:
+ bcc .L_strcat_align_to_64
ldrb r2, [r0]
- cbz r2, strcat_r0_scan_done
+ cbz r2, .L_strcat_r0_scan_done
add r0, r0, #1
ldrb r4, [r0]
- cbz r4, strcat_r0_scan_done
+ cbz r4, .L_strcat_r0_scan_done
add r0, r0, #1
-strcat_align_to_64:
+.L_strcat_align_to_64:
tst r3, #4
- beq strcat_mainloop
+ beq .L_strcat_mainloop
ldr r3, [r0], #4
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcat_zero_in_second_register
- b strcat_mainloop
+ bne .L_strcat_zero_in_second_register
+ b .L_strcat_mainloop
-strcat_r0_scan_done:
+.L_strcat_r0_scan_done:
// For short copies, hard-code checking the first 8 bytes since this
// new code doesn't win until after about 8 bytes.
- m_copy_byte reg=r2, cmd=cbz, label=strcpy_finish
- m_copy_byte reg=r3, cmd=cbz, label=strcpy_finish
- m_copy_byte reg=r4, cmd=cbz, label=strcpy_finish
- m_copy_byte reg=r5, cmd=cbz, label=strcpy_finish
- m_copy_byte reg=r2, cmd=cbz, label=strcpy_finish
- m_copy_byte reg=r3, cmd=cbz, label=strcpy_finish
- m_copy_byte reg=r4, cmd=cbz, label=strcpy_finish
- m_copy_byte reg=r5, cmd=cbnz, label=strcpy_continue
-
-strcpy_finish:
+ m_copy_byte reg=r2, cmd=cbz, label=.L_strcpy_finish
+ m_copy_byte reg=r3, cmd=cbz, label=.L_strcpy_finish
+ m_copy_byte reg=r4, cmd=cbz, label=.L_strcpy_finish
+ m_copy_byte reg=r5, cmd=cbz, label=.L_strcpy_finish
+ m_copy_byte reg=r2, cmd=cbz, label=.L_strcpy_finish
+ m_copy_byte reg=r3, cmd=cbz, label=.L_strcpy_finish
+ m_copy_byte reg=r4, cmd=cbz, label=.L_strcpy_finish
+ m_copy_byte reg=r5, cmd=cbnz, label=.L_strcpy_continue
+
+.L_strcpy_finish:
m_pop
-strcpy_continue:
+.L_strcpy_continue:
ands r3, r0, #7
- beq strcpy_check_src_align
+ beq .L_strcpy_check_src_align
// Align to a double word (64 bits).
rsb r3, r3, #8
lsls ip, r3, #31
- beq strcpy_align_to_32
+ beq .L_strcpy_align_to_32
ldrb r2, [r1], #1
strb r2, [r0], #1
- cbz r2, strcpy_complete
+ cbz r2, .L_strcpy_complete
-strcpy_align_to_32:
- bcc strcpy_align_to_64
+.L_strcpy_align_to_32:
+ bcc .L_strcpy_align_to_64
ldrb r2, [r1], #1
strb r2, [r0], #1
- cbz r2, strcpy_complete
+ cbz r2, .L_strcpy_complete
ldrb r2, [r1], #1
strb r2, [r0], #1
- cbz r2, strcpy_complete
+ cbz r2, .L_strcpy_complete
-strcpy_align_to_64:
+.L_strcpy_align_to_64:
tst r3, #4
- beq strcpy_check_src_align
+ beq .L_strcpy_check_src_align
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcpy_zero_in_first_register
+ bne .L_strcpy_zero_in_first_register
str r2, [r0], #4
-strcpy_check_src_align:
+.L_strcpy_check_src_align:
// At this point dst is aligned to a double word, check if src
// is also aligned to a double word.
ands r3, r1, #7
- bne strcpy_unaligned_copy
+ bne .L_strcpy_unaligned_copy
.p2align 2
-strcpy_mainloop:
+.L_strcpy_mainloop:
ldrd r2, r3, [r1], #8
pld [r1, #64]
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcpy_zero_in_first_register
+ bne .L_strcpy_zero_in_first_register
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcpy_zero_in_second_register
+ bne .L_strcpy_zero_in_second_register
strd r2, r3, [r0], #8
- b strcpy_mainloop
+ b .L_strcpy_mainloop
-strcpy_complete:
+.L_strcpy_complete:
m_pop
-strcpy_zero_in_first_register:
+.L_strcpy_zero_in_first_register:
lsls lr, ip, #17
- bne strcpy_copy1byte
- bcs strcpy_copy2bytes
+ bne .L_strcpy_copy1byte
+ bcs .L_strcpy_copy2bytes
lsls ip, ip, #1
- bne strcpy_copy3bytes
+ bne .L_strcpy_copy3bytes
-strcpy_copy4bytes:
+.L_strcpy_copy4bytes:
// Copy 4 bytes to the destiniation.
str r2, [r0]
m_pop
-strcpy_copy1byte:
+.L_strcpy_copy1byte:
strb r2, [r0]
m_pop
-strcpy_copy2bytes:
+.L_strcpy_copy2bytes:
strh r2, [r0]
m_pop
-strcpy_copy3bytes:
+.L_strcpy_copy3bytes:
strh r2, [r0], #2
lsr r2, #16
strb r2, [r0]
m_pop
-strcpy_zero_in_second_register:
+.L_strcpy_zero_in_second_register:
lsls lr, ip, #17
- bne strcpy_copy5bytes
- bcs strcpy_copy6bytes
+ bne .L_strcpy_copy5bytes
+ bcs .L_strcpy_copy6bytes
lsls ip, ip, #1
- bne strcpy_copy7bytes
+ bne .L_strcpy_copy7bytes
// Copy 8 bytes to the destination.
strd r2, r3, [r0]
m_pop
-strcpy_copy5bytes:
+.L_strcpy_copy5bytes:
str r2, [r0], #4
strb r3, [r0]
m_pop
-strcpy_copy6bytes:
+.L_strcpy_copy6bytes:
str r2, [r0], #4
strh r3, [r0]
m_pop
-strcpy_copy7bytes:
+.L_strcpy_copy7bytes:
str r2, [r0], #4
strh r3, [r0], #2
lsr r3, #16
strb r3, [r0]
m_pop
-strcpy_unaligned_copy:
+.L_strcpy_unaligned_copy:
// 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]
-strcpy_unaligned_branchtable:
+.L_strcpy_unaligned_branchtable:
.byte 0
- .byte ((strcpy_unalign7 - strcpy_unaligned_branchtable)/2)
- .byte ((strcpy_unalign6 - strcpy_unaligned_branchtable)/2)
- .byte ((strcpy_unalign5 - strcpy_unaligned_branchtable)/2)
- .byte ((strcpy_unalign4 - strcpy_unaligned_branchtable)/2)
- .byte ((strcpy_unalign3 - strcpy_unaligned_branchtable)/2)
- .byte ((strcpy_unalign2 - strcpy_unaligned_branchtable)/2)
- .byte ((strcpy_unalign1 - strcpy_unaligned_branchtable)/2)
+ .byte ((.L_strcpy_unalign7 - .L_strcpy_unaligned_branchtable)/2)
+ .byte ((.L_strcpy_unalign6 - .L_strcpy_unaligned_branchtable)/2)
+ .byte ((.L_strcpy_unalign5 - .L_strcpy_unaligned_branchtable)/2)
+ .byte ((.L_strcpy_unalign4 - .L_strcpy_unaligned_branchtable)/2)
+ .byte ((.L_strcpy_unalign3 - .L_strcpy_unaligned_branchtable)/2)
+ .byte ((.L_strcpy_unalign2 - .L_strcpy_unaligned_branchtable)/2)
+ .byte ((.L_strcpy_unalign1 - .L_strcpy_unaligned_branchtable)/2)
.p2align 2
// Can read 7 bytes before possibly crossing a page.
-strcpy_unalign7:
+.L_strcpy_unalign7:
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcpy_zero_in_first_register
+ bne .L_strcpy_zero_in_first_register
ldrb r3, [r1]
- cbz r3, strcpy_unalign7_copy5bytes
+ cbz r3, .L_strcpy_unalign7_copy5bytes
ldrb r4, [r1, #1]
- cbz r4, strcpy_unalign7_copy6bytes
+ cbz r4, .L_strcpy_unalign7_copy6bytes
ldrb r5, [r1, #2]
- cbz r5, strcpy_unalign7_copy7bytes
+ cbz r5, .L_strcpy_unalign7_copy7bytes
ldr r3, [r1], #4
pld [r1, #64]
lsrs ip, r3, #24
strd r2, r3, [r0], #8
- beq strcpy_unalign_return
- b strcpy_unalign7
+ beq .L_strcpy_unalign_return
+ b .L_strcpy_unalign7
-strcpy_unalign7_copy5bytes:
+.L_strcpy_unalign7_copy5bytes:
str r2, [r0], #4
strb r3, [r0]
-strcpy_unalign_return:
+.L_strcpy_unalign_return:
m_pop
-strcpy_unalign7_copy6bytes:
+.L_strcpy_unalign7_copy6bytes:
str r2, [r0], #4
strb r3, [r0], #1
strb r4, [r0], #1
m_pop
-strcpy_unalign7_copy7bytes:
+.L_strcpy_unalign7_copy7bytes:
str r2, [r0], #4
strb r3, [r0], #1
strb r4, [r0], #1
.p2align 2
// Can read 6 bytes before possibly crossing a page.
-strcpy_unalign6:
+.L_strcpy_unalign6:
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcpy_zero_in_first_register
+ bne .L_strcpy_zero_in_first_register
ldrb r4, [r1]
- cbz r4, strcpy_unalign_copy5bytes
+ cbz r4, .L_strcpy_unalign_copy5bytes
ldrb r5, [r1, #1]
- cbz r5, strcpy_unalign_copy6bytes
+ cbz r5, .L_strcpy_unalign_copy6bytes
ldr r3, [r1], #4
pld [r1, #64]
tst r3, #0xff0000
- beq strcpy_copy7bytes
+ beq .L_strcpy_copy7bytes
lsrs ip, r3, #24
strd r2, r3, [r0], #8
- beq strcpy_unalign_return
- b strcpy_unalign6
+ beq .L_strcpy_unalign_return
+ b .L_strcpy_unalign6
.p2align 2
// Can read 5 bytes before possibly crossing a page.
-strcpy_unalign5:
+.L_strcpy_unalign5:
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcpy_zero_in_first_register
+ bne .L_strcpy_zero_in_first_register
ldrb r4, [r1]
- cbz r4, strcpy_unalign_copy5bytes
+ cbz r4, .L_strcpy_unalign_copy5bytes
ldr r3, [r1], #4
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcpy_zero_in_second_register
+ bne .L_strcpy_zero_in_second_register
strd r2, r3, [r0], #8
- b strcpy_unalign5
+ b .L_strcpy_unalign5
-strcpy_unalign_copy5bytes:
+.L_strcpy_unalign_copy5bytes:
str r2, [r0], #4
strb r4, [r0]
m_pop
-strcpy_unalign_copy6bytes:
+.L_strcpy_unalign_copy6bytes:
str r2, [r0], #4
strb r4, [r0], #1
strb r5, [r0]
.p2align 2
// Can read 4 bytes before possibly crossing a page.
-strcpy_unalign4:
+.L_strcpy_unalign4:
ldr r2, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcpy_zero_in_first_register
+ bne .L_strcpy_zero_in_first_register
ldr r3, [r1], #4
pld [r1, #64]
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcpy_zero_in_second_register
+ bne .L_strcpy_zero_in_second_register
strd r2, r3, [r0], #8
- b strcpy_unalign4
+ b .L_strcpy_unalign4
.p2align 2
// Can read 3 bytes before possibly crossing a page.
-strcpy_unalign3:
+.L_strcpy_unalign3:
ldrb r2, [r1]
- cbz r2, strcpy_unalign3_copy1byte
+ cbz r2, .L_strcpy_unalign3_copy1byte
ldrb r3, [r1, #1]
- cbz r3, strcpy_unalign3_copy2bytes
+ cbz r3, .L_strcpy_unalign3_copy2bytes
ldrb r4, [r1, #2]
- cbz r4, strcpy_unalign3_copy3bytes
+ cbz r4, .L_strcpy_unalign3_copy3bytes
ldr r2, [r1], #4
ldr r3, [r1], #4
pld [r1, #64]
lsrs lr, r2, #24
- beq strcpy_copy4bytes
+ beq .L_strcpy_copy4bytes
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcpy_zero_in_second_register
+ bne .L_strcpy_zero_in_second_register
strd r2, r3, [r0], #8
- b strcpy_unalign3
+ b .L_strcpy_unalign3
-strcpy_unalign3_copy1byte:
+.L_strcpy_unalign3_copy1byte:
strb r2, [r0]
m_pop
-strcpy_unalign3_copy2bytes:
+.L_strcpy_unalign3_copy2bytes:
strb r2, [r0], #1
strb r3, [r0]
m_pop
-strcpy_unalign3_copy3bytes:
+.L_strcpy_unalign3_copy3bytes:
strb r2, [r0], #1
strb r3, [r0], #1
strb r4, [r0]
.p2align 2
// Can read 2 bytes before possibly crossing a page.
-strcpy_unalign2:
+.L_strcpy_unalign2:
ldrb r2, [r1]
- cbz r2, strcpy_unalign_copy1byte
+ cbz r2, .L_strcpy_unalign_copy1byte
ldrb r4, [r1, #1]
- cbz r4, strcpy_unalign_copy2bytes
+ cbz r4, .L_strcpy_unalign_copy2bytes
ldr r2, [r1], #4
ldr r3, [r1], #4
pld [r1, #64]
tst r2, #0xff0000
- beq strcpy_copy3bytes
+ beq .L_strcpy_copy3bytes
lsrs ip, r2, #24
- beq strcpy_copy4bytes
+ beq .L_strcpy_copy4bytes
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcpy_zero_in_second_register
+ bne .L_strcpy_zero_in_second_register
strd r2, r3, [r0], #8
- b strcpy_unalign2
+ b .L_strcpy_unalign2
.p2align 2
// Can read 1 byte before possibly crossing a page.
-strcpy_unalign1:
+.L_strcpy_unalign1:
ldrb r2, [r1]
- cbz r2, strcpy_unalign_copy1byte
+ cbz r2, .L_strcpy_unalign_copy1byte
ldr r2, [r1], #4
ldr r3, [r1], #4
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcpy_zero_in_first_register
+ bne .L_strcpy_zero_in_first_register
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcpy_zero_in_second_register
+ bne .L_strcpy_zero_in_second_register
strd r2, r3, [r0], #8
- b strcpy_unalign1
+ b .L_strcpy_unalign1
-strcpy_unalign_copy1byte:
+.L_strcpy_unalign_copy1byte:
strb r2, [r0]
m_pop
-strcpy_unalign_copy2bytes:
+.L_strcpy_unalign_copy2bytes:
strb r2, [r0], #1
strb r4, [r0]
m_pop
.p2align 2
-strcat_mainloop:
+.L_strcat_mainloop:
ldrd r2, r3, [r0], #8
pld [r0, #64]
sub ip, r2, #0x01010101
bic ip, ip, r2
ands ip, ip, #0x80808080
- bne strcat_zero_in_first_register
+ bne .L_strcat_zero_in_first_register
sub ip, r3, #0x01010101
bic ip, ip, r3
ands ip, ip, #0x80808080
- bne strcat_zero_in_second_register
- b strcat_mainloop
+ bne .L_strcat_zero_in_second_register
+ b .L_strcat_mainloop
-strcat_zero_in_first_register:
+.L_strcat_zero_in_first_register:
// Prefetch the src now, it's going to be used soon.
pld [r1, #0]
lsls lr, ip, #17
- bne strcat_sub8
- bcs strcat_sub7
+ bne .L_strcat_sub8
+ bcs .L_strcat_sub7
lsls ip, ip, #1
- bne strcat_sub6
+ bne .L_strcat_sub6
sub r0, r0, #5
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
-strcat_sub8:
+.L_strcat_sub8:
sub r0, r0, #8
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
-strcat_sub7:
+.L_strcat_sub7:
sub r0, r0, #7
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
-strcat_sub6:
+.L_strcat_sub6:
sub r0, r0, #6
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
-strcat_zero_in_second_register:
+.L_strcat_zero_in_second_register:
// Prefetch the src now, it's going to be used soon.
pld [r1, #0]
lsls lr, ip, #17
- bne strcat_sub4
- bcs strcat_sub3
+ bne .L_strcat_sub4
+ bcs .L_strcat_sub3
lsls ip, ip, #1
- bne strcat_sub2
+ bne .L_strcat_sub2
sub r0, r0, #1
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
-strcat_sub4:
+.L_strcat_sub4:
sub r0, r0, #4
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
-strcat_sub3:
+.L_strcat_sub3:
sub r0, r0, #3
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
-strcat_sub2:
+.L_strcat_sub2:
sub r0, r0, #2
- b strcat_r0_scan_done
+ b .L_strcat_r0_scan_done
END(strcat)