From: roma1390 <roma1390@gmail.com>
To: Denys Vlasenko <vda.linux@googlemail.com>
Cc: linux-kernel@vger.kernel.org,
Andrew Morton <akpm@linux-foundation.org>,
Douglas W Jones <jones@cs.uiowa.edu>,
Michal Nazarewicz <mnazarewicz@google.com>
Subject: Re: [PATCH 0/1] vsprintf: optimize decimal conversion (again)
Date: Wed, 28 Mar 2012 13:24:27 +0300 [thread overview]
Message-ID: <4F72E6DB.1090207@gmail.com> (raw)
In-Reply-To: <201203281213.07856.vda.linux@googlemail.com>
[-- Attachment #1: Type: text/plain, Size: 189 bytes --]
On 2012.03.28 13:13, Denys Vlasenko wrote:
> Second: run
> arm-linux-gnueabi-gcc -O2 -Wall test_{org,new}.c -S
> and email me resulting test_{org,new}.s files.
test_{org,new}.s attached.
[-- Attachment #2: test_new.s --]
[-- Type: text/plain, Size: 13521 bytes --]
.cpu arm9tdmi
.fpu softvfp
.eabi_attribute 20, 1
.eabi_attribute 21, 1
.eabi_attribute 23, 3
.eabi_attribute 24, 1
.eabi_attribute 25, 1
.eabi_attribute 26, 2
.eabi_attribute 30, 2
.eabi_attribute 18, 4
.file "test_new.c"
.text
.align 2
.type put_dec_trunc8, %function
put_dec_trunc8:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
@ link register save eliminated.
ldr r3, .L5
stmfd sp!, {r4, r5, fp}
umull fp, ip, r1, r3
mov r2, ip
add r1, r1, #48
mov ip, r0
add r0, r2, r2, asl #2
sub r1, r1, r0, asl #1
cmp r2, #0
mov r0, ip
strb r1, [r0], #1
beq .L2
umull r4, r5, r2, r3
add r2, r2, #48
add fp, r5, r5, asl #2
sub r2, r2, fp, asl #1
cmp r5, #0
mov r1, r5
strb r2, [ip, #1]
add r0, r0, #1
beq .L2
umull r4, r5, r1, r3
add r1, r1, #48
add ip, r5, r5, asl #2
sub r1, r1, ip, asl #1
cmp r5, #0
mov r2, r5
strb r1, [r0], #1
beq .L2
umull r4, r5, r2, r3
add r2, r2, #48
add r1, r5, r5, asl #2
sub r2, r2, r1, asl #1
cmp r5, #0
strb r2, [r0], #1
beq .L2
add r2, r5, r5, asl #1
add r2, r5, r2, asl #2
rsb r2, r2, r2, asl #6
add r2, r5, r2, asl #2
mov r2, r2, asl #1
mov r2, r2, lsr #16
add r3, r5, #48
add r1, r2, r2, asl #2
sub r3, r3, r1, asl #1
cmp r2, #0
strb r3, [r0], #1
beq .L2
add r1, r2, r1, asl #3
add r1, r1, r1, asl #2
mov r3, r1, lsr #11
add r2, r2, #48
add r1, r3, r3, asl #2
sub r2, r2, r1, asl #1
cmp r3, #0
strb r2, [r0], #1
beq .L2
add r1, r3, r1, asl #3
add r1, r1, r1, asl #2
mov r2, r1, lsr #11
add r1, r2, r2, asl #2
add r3, r3, #48
cmp r2, #0
sub r3, r3, r1, asl #1
strb r3, [r0], #1
addne r2, r2, #48
strneb r2, [r0], #1
.L2:
ldmfd sp!, {r4, r5, fp}
bx lr
.L6:
.align 2
.L5:
.word 429496730
.size put_dec_trunc8, .-put_dec_trunc8
.align 2
.type put_dec_full4, %function
put_dec_full4:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
@ link register save eliminated.
add r3, r1, r1, asl #1
add r3, r3, r3, asl #4
add r3, r3, r3, asl #8
add r3, r1, r3, asl #2
mov r3, r3, lsr #19
add r2, r3, r3, asl #1
add r2, r3, r2, asl #2
rsb r2, r2, r2, asl #6
add r2, r3, r2, asl #2
mov r2, r2, asl #1
mov r2, r2, lsr #16
add ip, r2, r2, asl #2
stmfd sp!, {r4, r5}
add r4, r2, ip, asl #3
add r5, r3, r3, asl #2
add r1, r1, #48
add r4, r4, r4, asl #2
sub r1, r1, r5, asl #1
mov r4, r4, lsr #11
mov r5, r0
strb r1, [r5], #1
add r3, r3, #48
add r1, r4, r4, asl #2
add r2, r2, #48
sub r2, r2, r1, asl #1
sub ip, r3, ip, asl #1
add r1, r5, #1
add r4, r4, #48
strb ip, [r0, #1]
strb r2, [r5, #1]
add r0, r1, #2
strb r4, [r1, #1]
ldmfd sp!, {r4, r5}
bx lr
.size put_dec_full4, .-put_dec_full4
.global __aeabi_uidivmod
.global __aeabi_uidiv
.align 2
.type number, %function
number:
@ Function supports interworking.
@ args = 8, pretend = 0, frame = 120
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
sub sp, sp, #124
ldrb ip, [sp, #161] @ zero_extendqisi2
mov r4, r1
ldrh r5, [sp, #166]
ldrb r1, [sp, #162] @ zero_extendqisi2
str r0, [sp, #8]
ands r0, ip, #64
str r1, [sp, #16]
str r5, [sp, #32]
ldrh r1, [sp, #164]
str ip, [sp, #12]
beq .L69
ldr r0, [sp, #16]
subs r0, r0, #10
movne r0, #1
.L69:
ands r5, ip, #16
str r0, [sp, #24]
and r0, ip, #32
andne ip, ip, #254
strne ip, [sp, #12]
str r5, [sp, #36]
ldr r5, [sp, #12]
str r0, [sp, #4]
tst r5, #2
beq .L14
cmp r3, #0
blt .L58
tst r5, #4
bne .L71
ldr r5, [sp, #12]
tst r5, #8
beq .L14
sub r1, r1, #1
mov r1, r1, asl #16
mov r1, r1, lsr #16
mov r0, #32
str r1, [sp, #20]
str r0, [sp, #28]
b .L17
.L14:
str r1, [sp, #20]
mov r1, #0
str r1, [sp, #28]
.L17:
ldr r5, [sp, #24]
cmp r5, #0
beq .L19
ldr r0, [sp, #20]
ldr r5, [sp, #16]
sub r1, r0, #1
mov r1, r1, asl #16
mov r1, r1, lsr #16
cmp r5, #16
str r1, [sp, #20]
subeq r1, r1, #1
moveq r1, r1, asl #16
moveq r1, r1, lsr #16
streq r1, [sp, #20]
.L19:
cmp r3, #0
bne .L20
cmp r2, #7
bls .L72
.L20:
ldr r0, [sp, #16]
cmp r0, #10
beq .L23
cmp r0, #16
movne r6, #3
moveq r6, #4
add r1, sp, #52
ldr r9, .L76
sub sl, r0, #1
mov r5, #0
str r1, [sp, #40]
sub r7, r6, #32
rsb fp, r6, #32
str r4, [sp, #44]
mov r8, r1
.L26:
mov ip, r2, lsr r6
cmp r7, #0
orr ip, ip, r3, asl fp
movge ip, r3, lsr r7
mov r4, r3, lsr r6
and r2, r2, #255
mov r0, ip
and r2, r2, sl
ldrb ip, [r9, r2] @ zero_extendqisi2
mov r2, r0
ldr r0, [sp, #4]
orrs r1, r2, r4
orr ip, r0, ip
strb ip, [r8, r5]
mov r3, r4
add r5, r5, #1
bne .L26
ldr r4, [sp, #44]
sub ip, r5, #1
.L22:
ldr r2, [sp, #32]
ldr r3, [sp, #20]
mov r0, r2, asl #16
cmp r5, r0, asr #16
movgt r0, r5, asl #16
mov r0, r0, lsr #16
ldr r1, [sp, #12]
rsb r7, r0, r3
mov r7, r7, asl #16
mov r7, r7, lsr #16
tst r1, #17
mov r1, r7
bne .L34
sub r1, r7, #1
mov r1, r1, asl #16
cmp r1, #0
mov r1, r1, lsr #16
blt .L34
ldr r3, [sp, #8]
mov r8, r1
add r2, r3, #1
add r2, r2, r1
mov r6, #32
.L36:
cmp r4, r3
strhib r6, [r3, #0]
add r3, r3, #1
cmp r3, r2
bne .L36
rsb r7, r7, #1
ldr r2, [sp, #8]
add r1, r1, r7
mov r1, r1, asl #16
add r8, r8, #1
sub r1, r1, #65536
add r2, r2, r8
str r2, [sp, #8]
mov r1, r1, lsr #16
.L34:
ldr r3, [sp, #28]
cmp r3, #0
beq .L37
ldr r2, [sp, #8]
cmp r2, r4
strccb r3, [r2, #0]
ldr r3, [sp, #8]
add r3, r3, #1
str r3, [sp, #8]
.L37:
ldr r2, [sp, #24]
cmp r2, #0
beq .L39
ldr r3, [sp, #8]
cmp r3, r4
ldrcc r2, [sp, #8]
movcc r3, #48
strccb r3, [r2, #0]
ldr r2, [sp, #8]
ldr r3, [sp, #16]
add r2, r2, #1
cmp r3, #16
str r2, [sp, #8]
beq .L73
.L39:
ldr r2, [sp, #36]
cmp r2, #0
movne r6, r1
movne r7, r6, asl #16
bne .L43
sub r6, r1, #1
ldr r3, [sp, #12]
mov r6, r6, asl #16
tst r3, #1
mov r6, r6, lsr #16
movne r8, #48
moveq r8, #32
movs r7, r6, asl #16
bmi .L43
sub r2, r1, #1
ldr r3, [sp, #8]
mov r2, r2, asl #16
add r2, r3, r2, lsr #16
add r2, r2, #1
.L47:
cmp r4, r3
strhib r8, [r3, #0]
add r3, r3, #1
cmp r3, r2
bne .L47
rsb r6, r1, r6
mov r6, r6, asl #16
mov r6, r6, lsr #16
str r3, [sp, #8]
mov r7, r6, asl #16
.L43:
sub r3, r0, #1
mov r3, r3, asl #16
cmp r5, r3, asr #16
bgt .L48
sub r1, r0, #2
ldr r3, [sp, #8]
mov r1, r1, asl #16
add r1, r3, r1, asr #16
add r1, r1, #1
mov r0, #48
.L50:
cmp r4, r3
strhib r0, [r3, #0]
add r3, r3, #1
rsb r2, r3, r1
cmp r5, r2
ble .L50
str r3, [sp, #8]
.L48:
cmp ip, #0
blt .L51
add r2, sp, #52
ldr r3, [sp, #8]
sub r1, r2, #1
add r2, r2, ip
.L53:
cmp r4, r3
ldrhib r0, [r2, #0] @ zero_extendqisi2
sub r2, r2, #1
strhib r0, [r3, #0]
cmp r2, r1
add r3, r3, #1
bne .L53
ldr r5, [sp, #8]
add ip, ip, #1
add r5, r5, ip
str r5, [sp, #8]
.L51:
cmp r7, #0
ble .L54
ldr r0, [sp, #8]
sub r2, r6, #1
mov r2, r2, asl #16
add r2, r0, r2, lsr #16
add r2, r2, #1
mov r3, r0
mov r1, #32
.L56:
cmp r4, r3
strhib r1, [r3, #0]
add r3, r3, #1
cmp r3, r2
bne .L56
str r3, [sp, #8]
.L54:
ldr r0, [sp, #8]
add sp, sp, #124
ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
bx lr
.L71:
sub r1, r1, #1
mov r1, r1, asl #16
mov r1, r1, lsr #16
str r1, [sp, #20]
mov r1, #43
str r1, [sp, #28]
b .L17
.L58:
sub r1, r1, #1
mov r1, r1, asl #16
mov r1, r1, lsr #16
mov r0, #45
rsbs r2, r2, #0
rsc r3, r3, #0
str r1, [sp, #20]
str r0, [sp, #28]
b .L17
.L72:
add r1, r2, #48
strb r1, [sp, #52]
mov ip, r3
mov r5, #1
b .L22
.L23:
cmp r3, #0
beq .L74
.L27:
mov r1, r2, asl #16
ldr r0, .L76+4
mov r8, r2, lsr #16
mov r1, r1, lsr #16
mla ip, r0, r8, r1
mov r7, r3, lsr #16
mov r6, r3
mov r3, #656
mla r2, r3, r7, ip
mov r6, r6, asl #16
mov r6, r6, lsr #16
mov r3, #7296
mla r5, r3, r6, r2
add r0, sp, #52
str r0, [sp, #40]
ldr r1, .L76+8
mov r0, r5
bl __aeabi_uidivmod
ldr r0, [sp, #40]
bl put_dec_full4
ldr r3, .L76+12
mov sl, r0
mul r2, r3, r6
sub r3, r3, #1824
sub r3, r3, #1
mla ip, r3, r7, r2
mov r0, r5
mov r3, #6
ldr r1, .L76+8
mla r5, r3, r8, ip
bl __aeabi_uidiv
add r5, r5, r0
mov r0, r5
ldr r1, .L76+8
bl __aeabi_uidivmod
mov r0, sl
bl put_dec_full4
ldr r3, .L76+16
mov r8, r0
mul r2, r3, r7
mov r0, r5
mov r3, #42
ldr r1, .L76+8
mla r5, r3, r6, r2
bl __aeabi_uidiv
add r5, r5, r0
mov r0, r5
ldr r1, .L76+8
bl __aeabi_uidivmod
mov r0, r8
bl put_dec_full4
ldr r3, .L76+20
mov r6, r0
ldr r1, .L76+8
mov r0, r5
mul r5, r3, r7
bl __aeabi_uidiv
adds r5, r0, r5
bne .L75
.L30:
mov r3, r6
.L31:
mov r0, r3
ldrb r2, [r3, #-1]! @ zero_extendqisi2
cmp r2, #48
beq .L31
.L29:
ldr r1, [sp, #40]
rsb r5, r1, r0
sub ip, r5, #1
b .L22
.L73:
cmp r4, r2
ldrhi r2, [sp, #4]
orrhi r3, r2, #88
ldrhi r2, [sp, #8]
strhib r3, [r2, #0]
ldr r3, [sp, #8]
add r3, r3, #1
str r3, [sp, #8]
b .L39
.L74:
ldr r1, .L76+24
cmp r2, r1
bhi .L27
add r5, sp, #52
mov r1, r2
mov r0, r5
str r5, [sp, #40]
bl put_dec_trunc8
b .L29
.L75:
mov r0, r5
ldr r1, .L76+8
bl __aeabi_uidivmod
mov r0, r6
bl put_dec_full4
ldr r1, .L76+8
mov r6, r0
mov r0, r5
bl __aeabi_uidiv
subs r1, r0, #0
beq .L30
mov r0, r6
bl put_dec_full4
mov r6, r0
b .L30
.L77:
.align 2
.L76:
.word .LANCHOR0
.word 5536
.word 10000
.word 9496
.word 4749
.word 281
.word 99999999
.size number, .-number
.align 2
.type measure_number, %function
measure_number:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 72
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
mov fp, r3
sub sp, sp, #84
ldr r3, [r0, #0]
str r0, [sp, #12]
mov r0, #0
str r3, [sp, #8]
mov sl, r2
bl time
ldr r3, [sp, #8]
mov r9, #0
add r6, sp, #16
cmp r3, r0
mov r7, r9
ldr r5, .L85
add r8, r6, #63
bne .L84
.L81:
ldr r4, .L85+4
.L80:
mov ip, sp
ldmia r5, {r0, r1}
mov r2, sl
stmia ip, {r0, r1}
mov r3, fp
mov r0, r6
mov r1, r8
bl number
sub r4, r4, #1
cmn r4, #1
strb r7, [r0, #0]
bne .L80
mov r0, #0
bl time
ldr r3, [sp, #8]
add r9, r9, #4000
cmp r3, r0
beq .L81
.L84:
ldr ip, [sp, #12]
str r0, [ip, #0]
mov r0, r9
add sp, sp, #84
ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
bx lr
.L86:
.align 2
.L85:
.word .LANCHOR0+16
.word 3999
.size measure_number, .-measure_number
.align 2
.type measure, %function
measure:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 8
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, lr}
mov r0, #0
sub sp, sp, #24
bl time
str r0, [sp, #20]
.L88:
mov r0, #0
bl time
ldr r3, [sp, #20]
cmp r0, r3
beq .L88
add r8, sp, #24
str r0, [r8, #-4]!
mov r2, #8
mov r3, #0
mov r0, r8
bl measure_number
mov r2, #123
mov sl, r0
mov r3, #0
mov r0, r8
bl measure_number
ldr r2, .L91
mov r7, r0
mov r3, #0
mov r0, r8
bl measure_number
ldr r2, .L91+4
mov r6, r0
mov r3, #0
mov r0, r8
bl measure_number
ldr r2, .L91+8
mov r5, r0
mov r3, #0
mov r0, r8
bl measure_number
mvn r2, #0
mov r4, r0
mov r3, #0
mov r0, r8
bl measure_number
mvn r2, #0
mov r9, r0
mvn r3, #0
mov r0, r8
bl measure_number
mov r1, sl
str r0, [sp, #12]
mov r2, r7
mov r3, r6
ldr r0, .L91+12
str r5, [sp, #0]
stmib sp, {r4, r9} @ phole stm
bl printf
add sp, sp, #24
ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, lr}
bx lr
.L92:
.align 2
.L91:
.word 123456
.word 12345678
.word 123456789
.word .LC0
.size measure, .-measure
.align 2
.type check, %function
check:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 128
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, lr}
ldr r3, .L97
sub sp, sp, #140
mov r5, r0
mov r6, r1
add r4, sp, #72
ldmia r3, {r0, r1}
mov r3, sp
stmia r3, {r0, r1}
mov r2, r5
mov r3, r6
add r1, r4, #63
mov r0, r4
bl number
add r7, sp, #8
mov r3, #0
strb r3, [r0, #0]
mov r2, r5
mov r3, r6
ldr r1, .L97+4
mov r0, r7
bl sprintf
mov r0, r4
mov r1, r7
bl strcmp
cmp r0, #0
bne .L96
add sp, sp, #140
ldmfd sp!, {r4, r5, r6, r7, lr}
bx lr
.L96:
mov r2, r5
mov r3, r6
ldr r0, .L97+8
str r4, [sp, #0]
bl printf
mov r0, #1
bl exit
.L98:
.align 2
.L97:
.word .LANCHOR0+16
.word .LC1
.word .LC2
.size check, .-check
.align 2
.global main
.type main, %function
main:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r3, r4, r5, r6, r7, r8, r9, sl, fp, lr}
bl measure
mov r4, #0
bl measure
mov r5, #0
bl measure
ldr r6, .L103
bl measure
mov r7, #0
mov sl, #1
mov fp, #0
b .L101
.L100:
adds r4, r4, sl
adc r5, r5, fp
.L101:
mov r0, r4
mov r1, r5
bl check
and r8, r4, r6
rsbs r0, r4, #0
rsc r1, r5, #0
and r9, r5, r7
bl check
orrs r8, r8, r9
bne .L100
mov r2, r4
mov r3, r5
ldr r0, .L103+4
bl printf
mov r0, r8
bl fflush
b .L100
.L104:
.align 2
.L103:
.word 262143
.word .LC3
.size main, .-main
.section .rodata
.align 2
.LANCHOR0 = . + 0
.type digits.3938, %object
.size digits.3938, 16
digits.3938:
.ascii "0123456789ABCDEF"
.type dummy_spec, %object
.size dummy_spec, 8
dummy_spec:
.byte 8
.byte 0
.byte 10
.byte 0
.short 0
.short 0
.section .rodata.str1.4,"aMS",%progbits,1
.align 2
.LC0:
.ascii "Conversions per second: 8:%d 123:%d 123456:%d 12345"
.ascii "678:%d 123456789:%d 2^32:%d 2^64:%d\012\000"
.LC1:
.ascii "%llu\000"
.space 3
.LC2:
.ascii "Error in formatting %llu:'%s'\012\000"
.space 1
.LC3:
.ascii "\015Tested %llu \000"
.ident "GCC: (Debian 4.4.5-8) 4.4.5"
.section .note.GNU-stack,"",%progbits
[-- Attachment #3: test_org.s --]
[-- Type: text/plain, Size: 13375 bytes --]
.cpu arm9tdmi
.fpu softvfp
.eabi_attribute 20, 1
.eabi_attribute 21, 1
.eabi_attribute 23, 3
.eabi_attribute 24, 1
.eabi_attribute 25, 1
.eabi_attribute 26, 2
.eabi_attribute 30, 2
.eabi_attribute 18, 4
.file "test_org.c"
.text
.align 2
.type put_dec_trunc, %function
put_dec_trunc:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
@ link register save eliminated.
stmfd sp!, {r4, r5, r6}
mov ip, r1, lsr #8
mov r4, r1, lsr #4
and r4, r4, #15
and ip, ip, #15
mov r3, r1, lsr #12
add r2, r4, ip
add r2, r2, r3
add r2, r2, r2, asl #1
and r1, r1, #15
add r1, r1, r2, asl #1
add r2, r1, r1, asl #2
add r2, r1, r2, asl #3
add r2, r2, r2, asl #2
mov r2, r2, lsr #11
add r5, r3, r3, asl #3
add r4, r5, r4
add r6, r2, r2, asl #2
add r5, ip, ip, asl #2
add r4, r4, r5
sub r1, r1, r6, asl #1
add r1, r1, #48
adds r2, r4, r2
mov r5, r0
strb r1, [r0], #1
beq .L2
add r1, r2, r2, asl #2
add r1, r2, r1, asl #3
add r1, r1, r1, asl #2
mov r1, r1, lsr #11
add r4, r1, r1, asl #2
sub r2, r2, r4, asl #1
add ip, r1, ip, asl #1
add r2, r2, #48
orrs r1, ip, r3
strb r2, [r5, #1]
add r0, r0, #1
beq .L2
add r2, ip, ip, asl #1
add r2, ip, r2, asl #2
mov r2, r2, lsr #7
add r1, r2, r2, asl #2
sub ip, ip, r1, asl #1
add ip, ip, #48
adds r3, r2, r3, asl #2
strb ip, [r0], #1
beq .L2
add r1, r3, r3, asl #2
add r1, r3, r1, asl #3
add r1, r1, r1, asl #2
mov r1, r1, lsr #11
add r2, r1, r1, asl #2
sub r3, r3, r2, asl #1
cmp r1, #0
add r3, r3, #48
strb r3, [r0], #1
addne r1, r1, #48
strneb r1, [r0], #1
.L2:
ldmfd sp!, {r4, r5, r6}
bx lr
.size put_dec_trunc, .-put_dec_trunc
.align 2
.type put_dec_full, %function
put_dec_full:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
@ link register save eliminated.
mov ip, r1, lsr #4
mov r2, r1, lsr #8
and r2, r2, #15
and ip, ip, #15
stmfd sp!, {r4, r5, r6, r7, r8}
mov r3, r1, lsr #12
add r4, ip, r2
add r4, r4, r3
add r4, r4, r4, asl #1
and r1, r1, #15
add r1, r1, r4, asl #1
add r6, r1, r1, asl #2
add r4, r3, r3, asl #3
add r6, r1, r6, asl #3
add ip, r4, ip
add r6, r6, r6, asl #2
add r4, r2, r2, asl #2
add ip, ip, r4
mov r6, r6, lsr #11
add ip, ip, r6
add r5, ip, ip, asl #2
add r5, ip, r5, asl #3
add r5, r5, r5, asl #2
mov r5, r5, lsr #11
add r2, r5, r2, asl #1
add r4, r2, r2, asl #1
add r4, r2, r4, asl #2
mov r4, r4, lsr #7
add r3, r4, r3, asl #2
add r7, r3, r3, asl #2
add r7, r3, r7, asl #3
add r6, r6, r6, asl #2
sub r1, r1, r6, asl #1
add r7, r7, r7, asl #2
mov r6, r0
mov r7, r7, lsr #11
add r1, r1, #48
strb r1, [r6], #1
add r5, r5, r5, asl #2
add r1, r7, r7, asl #2
add r4, r4, r4, asl #2
add r8, r6, #1
sub ip, ip, r5, asl #1
sub r2, r2, r4, asl #1
sub r3, r3, r1, asl #1
add ip, ip, #48
add r1, r8, #1
add r2, r2, #48
add r3, r3, #48
add r7, r7, #48
strb ip, [r0, #1]
strb r2, [r6, #1]
add r0, r1, #2
strb r3, [r8, #1]
strb r7, [r1, #1]
ldmfd sp!, {r4, r5, r6, r7, r8}
bx lr
.size put_dec_full, .-put_dec_full
.align 2
.type put_dec, %function
put_dec:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
cmp r3, #0
stmfd sp!, {r4, r5, r6, r7, r8, sl, fp, lr}
beq .L18
.L15:
ldr sl, .L19
ldr r5, .L19+4
ldr r6, .L19+8
mov r8, #0
.L17:
#APP
@ 101 "test_org.c" 1
umull fp, ip, r5, r2
cmn fp, r5
adcs ip, ip, r6
adc fp, r8, #0
@ 0 "" 2
mov r1, r8
#APP
@ 101 "test_org.c" 1
umlal ip, fp, r6, r2
umlal ip, r1, r5, r3
mov ip, #0
adds fp, r1, fp
adc ip, ip, #0
umlal fp, ip, r6, r3
@ 0 "" 2
mov r4, fp, lsr #16
orr r4, r4, ip, asl #16
add r1, r4, r4, asl #1
add r1, r1, r1, asl #6
add r1, r4, r1, asl #2
add r1, r4, r1, asl #2
mov r7, ip, lsr #16
sub r1, r2, r1, asl #5
bl put_dec_full
cmp r7, #0
mov r2, r4
mov r3, r7
bne .L17
cmp r4, sl
bhi .L17
.L10:
mov r1, r4
ldmfd sp!, {r4, r5, r6, r7, r8, sl, fp, lr}
b put_dec_trunc
.L18:
ldr r1, .L19
cmp r2, r1
bhi .L15
mov r4, r2
b .L10
.L20:
.align 2
.L19:
.word 99999
.word 457671715
.word -1480217529
.size put_dec, .-put_dec
.align 2
.type number, %function
number:
@ Function supports interworking.
@ args = 8, pretend = 0, frame = 120
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
sub sp, sp, #124
ldrb ip, [sp, #161] @ zero_extendqisi2
ldrb r4, [sp, #162] @ zero_extendqisi2
ldrh r5, [sp, #166]
str r0, [sp, #12]
ands r0, ip, #64
str r4, [sp, #20]
str r5, [sp, #36]
ldrh r4, [sp, #164]
str ip, [sp, #16]
beq .L72
ldr r0, [sp, #20]
subs r0, r0, #10
movne r0, #1
.L72:
ands r5, ip, #16
str r0, [sp, #28]
and r0, ip, #32
andne ip, ip, #254
strne ip, [sp, #16]
str r5, [sp, #40]
ldr r5, [sp, #16]
str r0, [sp, #8]
tst r5, #2
beq .L26
cmp r3, #0
blt .L64
tst r5, #4
bne .L74
ldr r5, [sp, #16]
tst r5, #8
beq .L26
sub r0, r4, #1
mov r0, r0, asl #16
mov r0, r0, lsr #16
mov ip, #32
str r0, [sp, #24]
str ip, [sp, #32]
b .L29
.L26:
mov r0, #0
str r4, [sp, #24]
str r0, [sp, #32]
.L29:
ldr r4, [sp, #28]
cmp r4, #0
beq .L31
ldr r5, [sp, #24]
ldr ip, [sp, #20]
sub r0, r5, #1
mov r0, r0, asl #16
mov r0, r0, lsr #16
cmp ip, #16
str r0, [sp, #24]
subeq r0, r0, #1
moveq r0, r0, asl #16
moveq r0, r0, lsr #16
streq r0, [sp, #24]
.L31:
orrs ip, r2, r3
moveq r3, #48
streqb r3, [sp, #52]
moveq r4, #1
beq .L33
ldr r0, [sp, #20]
cmp r0, #10
beq .L34
cmp r0, #16
movne r5, #3
moveq r5, #4
sub r8, r0, #1
ldr sl, .L76
rsb r0, r5, #32
mov r4, #0
add r9, sp, #52
sub r6, r5, #32
str r1, [sp, #44]
mov fp, r0
.L37:
mov ip, r2, lsr r5
cmp r6, #0
orr ip, ip, r3, asl fp
movge ip, r3, lsr r6
mov r7, r3, lsr r5
and r2, r2, #255
and r2, r2, r8
mov r0, ip
ldr r1, [sp, #8]
ldrb ip, [sl, r2] @ zero_extendqisi2
mov r2, r0
orr ip, r1, ip
orrs r0, r2, r7
strb ip, [r9, r4]
mov r3, r7
add r4, r4, #1
bne .L37
ldr r1, [sp, #44]
sub ip, r4, #1
.L33:
ldr r2, [sp, #36]
ldr r3, [sp, #24]
mov r5, r2, asl #16
cmp r4, r5, asr #16
movgt r5, r4, asl #16
mov r5, r5, lsr #16
ldr r0, [sp, #16]
rsb r7, r5, r3
mov r7, r7, asl #16
mov r7, r7, lsr #16
tst r0, #17
mov r0, r7
bne .L40
sub r0, r7, #1
mov r0, r0, asl #16
cmp r0, #0
mov r0, r0, lsr #16
blt .L40
ldr r3, [sp, #12]
mov r8, r0
add r2, r3, #1
add r2, r2, r0
mov r6, #32
.L42:
cmp r1, r3
strhib r6, [r3, #0]
add r3, r3, #1
cmp r3, r2
bne .L42
rsb r7, r7, #1
ldr r2, [sp, #12]
add r0, r0, r7
mov r0, r0, asl #16
add r3, r8, #1
sub r0, r0, #65536
add r2, r2, r3
str r2, [sp, #12]
mov r0, r0, lsr #16
.L40:
ldr r3, [sp, #32]
cmp r3, #0
beq .L43
ldr r2, [sp, #12]
cmp r2, r1
strccb r3, [r2, #0]
ldr r3, [sp, #12]
add r3, r3, #1
str r3, [sp, #12]
.L43:
ldr r2, [sp, #28]
cmp r2, #0
beq .L45
ldr r3, [sp, #12]
cmp r3, r1
ldrcc r2, [sp, #12]
movcc r3, #48
strccb r3, [r2, #0]
ldr r2, [sp, #12]
ldr r3, [sp, #20]
add r2, r2, #1
cmp r3, #16
str r2, [sp, #12]
beq .L75
.L45:
ldr r2, [sp, #40]
cmp r2, #0
movne r6, r0
movne r7, r6, asl #16
bne .L49
sub r6, r0, #1
ldr r3, [sp, #16]
mov r6, r6, asl #16
tst r3, #1
mov r6, r6, lsr #16
movne r8, #48
moveq r8, #32
movs r7, r6, asl #16
bmi .L49
sub r2, r0, #1
ldr r3, [sp, #12]
mov r2, r2, asl #16
add r2, r3, r2, lsr #16
add r2, r2, #1
.L53:
cmp r1, r3
strhib r8, [r3, #0]
add r3, r3, #1
cmp r3, r2
bne .L53
rsb r6, r0, r6
mov r6, r6, asl #16
mov r6, r6, lsr #16
str r3, [sp, #12]
mov r7, r6, asl #16
.L49:
sub r3, r5, #1
mov r3, r3, asl #16
cmp r4, r3, asr #16
bgt .L54
sub r0, r5, #2
ldr r3, [sp, #12]
mov r0, r0, asl #16
add r0, r3, r0, asr #16
add r0, r0, #1
mov r5, #48
.L56:
cmp r1, r3
strhib r5, [r3, #0]
add r3, r3, #1
rsb r2, r3, r0
cmp r4, r2
ble .L56
str r3, [sp, #12]
.L54:
cmp ip, #0
blt .L57
add r2, sp, #52
ldr r3, [sp, #12]
sub r0, r2, #1
add r2, r2, ip
.L59:
cmp r1, r3
ldrhib r4, [r2, #0] @ zero_extendqisi2
sub r2, r2, #1
strhib r4, [r3, #0]
cmp r2, r0
add r3, r3, #1
bne .L59
ldr r4, [sp, #12]
add ip, ip, #1
add r4, r4, ip
str r4, [sp, #12]
.L57:
cmp r7, #0
ble .L60
ldr r5, [sp, #12]
sub r2, r6, #1
mov r2, r2, asl #16
add r2, r5, r2, lsr #16
add r2, r2, #1
mov r3, r5
mov r0, #32
.L62:
cmp r1, r3
strhib r0, [r3, #0]
add r3, r3, #1
cmp r3, r2
bne .L62
str r3, [sp, #12]
.L60:
ldr r0, [sp, #12]
add sp, sp, #124
ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
bx lr
.L74:
sub r0, r4, #1
mov r0, r0, asl #16
mov r0, r0, lsr #16
str r0, [sp, #24]
mov r0, #43
str r0, [sp, #32]
b .L29
.L75:
cmp r1, r2
ldrhi r2, [sp, #8]
orrhi r3, r2, #88
ldrhi r2, [sp, #12]
strhib r3, [r2, #0]
ldr r3, [sp, #12]
add r3, r3, #1
str r3, [sp, #12]
b .L45
.L64:
sub r0, r4, #1
mov r0, r0, asl #16
mov r0, r0, lsr #16
mov ip, #45
rsbs r2, r2, #0
rsc r3, r3, #0
str r0, [sp, #24]
str ip, [sp, #32]
b .L29
.L34:
add r4, sp, #52
mov r0, r4
str r1, [sp, #4]
bl put_dec
rsb r4, r4, r0
sub ip, r4, #1
ldr r1, [sp, #4]
b .L33
.L77:
.align 2
.L76:
.word .LANCHOR0
.size number, .-number
.align 2
.type measure_number, %function
measure_number:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 72
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
mov fp, r3
sub sp, sp, #84
ldr r3, [r0, #0]
str r0, [sp, #12]
mov r0, #0
str r3, [sp, #8]
mov sl, r2
bl time
ldr r3, [sp, #8]
mov r9, #0
add r6, sp, #16
cmp r3, r0
mov r7, r9
ldr r5, .L85
add r8, r6, #63
bne .L84
.L81:
ldr r4, .L85+4
.L80:
mov ip, sp
ldmia r5, {r0, r1}
mov r2, sl
stmia ip, {r0, r1}
mov r3, fp
mov r0, r6
mov r1, r8
bl number
sub r4, r4, #1
cmn r4, #1
strb r7, [r0, #0]
bne .L80
mov r0, #0
bl time
ldr r3, [sp, #8]
add r9, r9, #4000
cmp r3, r0
beq .L81
.L84:
ldr ip, [sp, #12]
str r0, [ip, #0]
mov r0, r9
add sp, sp, #84
ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, fp, lr}
bx lr
.L86:
.align 2
.L85:
.word .LANCHOR0+16
.word 3999
.size measure_number, .-measure_number
.align 2
.type measure, %function
measure:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 8
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, r8, r9, sl, lr}
mov r0, #0
sub sp, sp, #24
bl time
str r0, [sp, #20]
.L88:
mov r0, #0
bl time
ldr r3, [sp, #20]
cmp r0, r3
beq .L88
add r8, sp, #24
str r0, [r8, #-4]!
mov r2, #8
mov r3, #0
mov r0, r8
bl measure_number
mov r2, #123
mov sl, r0
mov r3, #0
mov r0, r8
bl measure_number
ldr r2, .L91
mov r7, r0
mov r3, #0
mov r0, r8
bl measure_number
ldr r2, .L91+4
mov r6, r0
mov r3, #0
mov r0, r8
bl measure_number
ldr r2, .L91+8
mov r5, r0
mov r3, #0
mov r0, r8
bl measure_number
mvn r2, #0
mov r4, r0
mov r3, #0
mov r0, r8
bl measure_number
mvn r2, #0
mov r9, r0
mvn r3, #0
mov r0, r8
bl measure_number
mov r1, sl
str r0, [sp, #12]
mov r2, r7
mov r3, r6
ldr r0, .L91+12
str r5, [sp, #0]
stmib sp, {r4, r9} @ phole stm
bl printf
add sp, sp, #24
ldmfd sp!, {r4, r5, r6, r7, r8, r9, sl, lr}
bx lr
.L92:
.align 2
.L91:
.word 123456
.word 12345678
.word 123456789
.word .LC0
.size measure, .-measure
.align 2
.type check, %function
check:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 128
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r6, r7, lr}
ldr r3, .L97
sub sp, sp, #140
mov r5, r0
mov r6, r1
add r4, sp, #72
ldmia r3, {r0, r1}
mov r3, sp
stmia r3, {r0, r1}
mov r2, r5
mov r3, r6
add r1, r4, #63
mov r0, r4
bl number
add r7, sp, #8
mov r3, #0
strb r3, [r0, #0]
mov r2, r5
mov r3, r6
ldr r1, .L97+4
mov r0, r7
bl sprintf
mov r0, r4
mov r1, r7
bl strcmp
cmp r0, #0
bne .L96
add sp, sp, #140
ldmfd sp!, {r4, r5, r6, r7, lr}
bx lr
.L96:
mov r2, r5
mov r3, r6
ldr r0, .L97+8
str r4, [sp, #0]
bl printf
mov r0, #1
bl exit
.L98:
.align 2
.L97:
.word .LANCHOR0+16
.word .LC1
.word .LC2
.size check, .-check
.align 2
.global main
.type main, %function
main:
@ Function supports interworking.
@ args = 0, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r3, r4, r5, r6, r7, r8, r9, sl, fp, lr}
bl measure
mov r4, #0
bl measure
mov r5, #0
bl measure
ldr r6, .L103
bl measure
mov r7, #0
mov sl, #1
mov fp, #0
b .L101
.L100:
adds r4, r4, sl
adc r5, r5, fp
.L101:
mov r0, r4
mov r1, r5
bl check
and r8, r4, r6
rsbs r0, r4, #0
rsc r1, r5, #0
and r9, r5, r7
bl check
orrs r8, r8, r9
bne .L100
mov r2, r4
mov r3, r5
ldr r0, .L103+4
bl printf
mov r0, r8
bl fflush
b .L100
.L104:
.align 2
.L103:
.word 262143
.word .LC3
.size main, .-main
.section .rodata
.align 2
.LANCHOR0 = . + 0
.type digits.4070, %object
.size digits.4070, 16
digits.4070:
.ascii "0123456789ABCDEF"
.type dummy_spec, %object
.size dummy_spec, 8
dummy_spec:
.byte 8
.byte 0
.byte 10
.byte 0
.short 0
.short 0
.section .rodata.str1.4,"aMS",%progbits,1
.align 2
.LC0:
.ascii "Conversions per second: 8:%d 123:%d 123456:%d 12345"
.ascii "678:%d 123456789:%d 2^32:%d 2^64:%d\012\000"
.LC1:
.ascii "%llu\000"
.space 3
.LC2:
.ascii "Error in formatting %llu:'%s'\012\000"
.space 1
.LC3:
.ascii "\015Tested %llu \000"
.ident "GCC: (Debian 4.4.5-8) 4.4.5"
.section .note.GNU-stack,"",%progbits
next prev parent reply other threads:[~2012-03-28 10:24 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-03-26 18:47 [PATCH 0/1] vsprintf: optimize decimal conversion (again) Denys Vlasenko
2012-03-26 18:51 ` [PATCH 1/1] " Denys Vlasenko
2012-03-26 19:51 ` Andrew Morton
2012-03-26 19:56 ` Denys Vlasenko
2012-03-26 20:13 ` Andrew Morton
2012-03-26 20:18 ` Geert Uytterhoeven
2012-03-26 23:18 ` Denys Vlasenko
2012-03-27 0:30 ` Denys Vlasenko
2012-03-27 3:49 ` H. Peter Anvin
2012-03-26 20:20 ` H. Peter Anvin
2012-03-27 17:12 ` Michal Nazarewicz
2012-03-27 17:17 ` H. Peter Anvin
2012-03-27 0:26 ` Denys Vlasenko
2012-03-27 12:08 ` [PATCH 0/1] " roma1390
2012-03-27 15:32 ` Denys Vlasenko
2012-03-27 15:42 ` Denys Vlasenko
2012-03-28 5:56 ` roma1390
2012-03-28 10:13 ` Denys Vlasenko
2012-03-28 10:24 ` roma1390 [this message]
2012-03-28 10:33 ` Denys Vlasenko
2012-03-28 10:39 ` roma1390
2012-03-28 11:20 ` Denys Vlasenko
2012-03-29 10:35 ` Denys Vlasenko
2012-03-28 10:31 ` roma1390
2012-03-28 11:23 ` Denys Vlasenko
2012-03-29 5:23 ` roma1390
2012-03-29 10:33 ` Denys Vlasenko
2012-03-27 13:49 ` roma1390
2012-03-27 15:33 ` Denys Vlasenko
2012-03-29 5:16 ` roma1390
2012-03-29 10:33 ` Denys Vlasenko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4F72E6DB.1090207@gmail.com \
--to=roma1390@gmail.com \
--cc=akpm@linux-foundation.org \
--cc=jones@cs.uiowa.edu \
--cc=linux-kernel@vger.kernel.org \
--cc=mnazarewicz@google.com \
--cc=vda.linux@googlemail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox