mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-11-15 03:53:41 +00:00
978 lines
22 KiB
ArmAsm
978 lines
22 KiB
ArmAsm
// This file is generated from a similarly-named Perl script in the BoringSSL
|
|
// source tree. Do not edit by hand.
|
|
|
|
#if !defined(__has_feature)
|
|
#define __has_feature(x) 0
|
|
#endif
|
|
#if __has_feature(memory_sanitizer) && !defined(OPENSSL_NO_ASM)
|
|
#define OPENSSL_NO_ASM
|
|
#endif
|
|
|
|
#if !defined(OPENSSL_NO_ASM)
|
|
#if defined(__arm__)
|
|
#if defined(BORINGSSL_PREFIX)
|
|
#include <boringssl_prefix_symbols_asm.h>
|
|
#endif
|
|
#include <openssl/arm_arch.h>
|
|
|
|
@ Silence ARMv8 deprecated IT instruction warnings. This file is used by both
|
|
@ ARMv7 and ARMv8 processors and does not use ARMv8 instructions.
|
|
.arch armv7-a
|
|
|
|
.text
|
|
#if defined(__thumb2__)
|
|
.syntax unified
|
|
.thumb
|
|
#else
|
|
.code 32
|
|
#endif
|
|
|
|
#if __ARM_MAX_ARCH__>=7
|
|
.align 5
|
|
.LOPENSSL_armcap:
|
|
.word OPENSSL_armcap_P-.Lbn_mul_mont
|
|
#endif
|
|
|
|
.globl bn_mul_mont
|
|
.hidden bn_mul_mont
|
|
.type bn_mul_mont,%function
|
|
|
|
.align 5
|
|
bn_mul_mont:
|
|
.Lbn_mul_mont:
|
|
ldr ip,[sp,#4] @ load num
|
|
stmdb sp!,{r0,r2} @ sp points at argument block
|
|
#if __ARM_MAX_ARCH__>=7
|
|
tst ip,#7
|
|
bne .Lialu
|
|
adr r0,.Lbn_mul_mont
|
|
ldr r2,.LOPENSSL_armcap
|
|
ldr r0,[r0,r2]
|
|
#ifdef __APPLE__
|
|
ldr r0,[r0]
|
|
#endif
|
|
tst r0,#ARMV7_NEON @ NEON available?
|
|
ldmia sp, {r0,r2}
|
|
beq .Lialu
|
|
add sp,sp,#8
|
|
b bn_mul8x_mont_neon
|
|
.align 4
|
|
.Lialu:
|
|
#endif
|
|
cmp ip,#2
|
|
mov r0,ip @ load num
|
|
#ifdef __thumb2__
|
|
ittt lt
|
|
#endif
|
|
movlt r0,#0
|
|
addlt sp,sp,#2*4
|
|
blt .Labrt
|
|
|
|
stmdb sp!,{r4,r5,r6,r7,r8,r9,r10,r11,r12,lr} @ save 10 registers
|
|
|
|
mov r0,r0,lsl#2 @ rescale r0 for byte count
|
|
sub sp,sp,r0 @ alloca(4*num)
|
|
sub sp,sp,#4 @ +extra dword
|
|
sub r0,r0,#4 @ "num=num-1"
|
|
add r4,r2,r0 @ &bp[num-1]
|
|
|
|
add r0,sp,r0 @ r0 to point at &tp[num-1]
|
|
ldr r8,[r0,#14*4] @ &n0
|
|
ldr r2,[r2] @ bp[0]
|
|
ldr r5,[r1],#4 @ ap[0],ap++
|
|
ldr r6,[r3],#4 @ np[0],np++
|
|
ldr r8,[r8] @ *n0
|
|
str r4,[r0,#15*4] @ save &bp[num]
|
|
|
|
umull r10,r11,r5,r2 @ ap[0]*bp[0]
|
|
str r8,[r0,#14*4] @ save n0 value
|
|
mul r8,r10,r8 @ "tp[0]"*n0
|
|
mov r12,#0
|
|
umlal r10,r12,r6,r8 @ np[0]*n0+"t[0]"
|
|
mov r4,sp
|
|
|
|
.L1st:
|
|
ldr r5,[r1],#4 @ ap[j],ap++
|
|
mov r10,r11
|
|
ldr r6,[r3],#4 @ np[j],np++
|
|
mov r11,#0
|
|
umlal r10,r11,r5,r2 @ ap[j]*bp[0]
|
|
mov r14,#0
|
|
umlal r12,r14,r6,r8 @ np[j]*n0
|
|
adds r12,r12,r10
|
|
str r12,[r4],#4 @ tp[j-1]=,tp++
|
|
adc r12,r14,#0
|
|
cmp r4,r0
|
|
bne .L1st
|
|
|
|
adds r12,r12,r11
|
|
ldr r4,[r0,#13*4] @ restore bp
|
|
mov r14,#0
|
|
ldr r8,[r0,#14*4] @ restore n0
|
|
adc r14,r14,#0
|
|
str r12,[r0] @ tp[num-1]=
|
|
mov r7,sp
|
|
str r14,[r0,#4] @ tp[num]=
|
|
|
|
.Louter:
|
|
sub r7,r0,r7 @ "original" r0-1 value
|
|
sub r1,r1,r7 @ "rewind" ap to &ap[1]
|
|
ldr r2,[r4,#4]! @ *(++bp)
|
|
sub r3,r3,r7 @ "rewind" np to &np[1]
|
|
ldr r5,[r1,#-4] @ ap[0]
|
|
ldr r10,[sp] @ tp[0]
|
|
ldr r6,[r3,#-4] @ np[0]
|
|
ldr r7,[sp,#4] @ tp[1]
|
|
|
|
mov r11,#0
|
|
umlal r10,r11,r5,r2 @ ap[0]*bp[i]+tp[0]
|
|
str r4,[r0,#13*4] @ save bp
|
|
mul r8,r10,r8
|
|
mov r12,#0
|
|
umlal r10,r12,r6,r8 @ np[0]*n0+"tp[0]"
|
|
mov r4,sp
|
|
|
|
.Linner:
|
|
ldr r5,[r1],#4 @ ap[j],ap++
|
|
adds r10,r11,r7 @ +=tp[j]
|
|
ldr r6,[r3],#4 @ np[j],np++
|
|
mov r11,#0
|
|
umlal r10,r11,r5,r2 @ ap[j]*bp[i]
|
|
mov r14,#0
|
|
umlal r12,r14,r6,r8 @ np[j]*n0
|
|
adc r11,r11,#0
|
|
ldr r7,[r4,#8] @ tp[j+1]
|
|
adds r12,r12,r10
|
|
str r12,[r4],#4 @ tp[j-1]=,tp++
|
|
adc r12,r14,#0
|
|
cmp r4,r0
|
|
bne .Linner
|
|
|
|
adds r12,r12,r11
|
|
mov r14,#0
|
|
ldr r4,[r0,#13*4] @ restore bp
|
|
adc r14,r14,#0
|
|
ldr r8,[r0,#14*4] @ restore n0
|
|
adds r12,r12,r7
|
|
ldr r7,[r0,#15*4] @ restore &bp[num]
|
|
adc r14,r14,#0
|
|
str r12,[r0] @ tp[num-1]=
|
|
str r14,[r0,#4] @ tp[num]=
|
|
|
|
cmp r4,r7
|
|
#ifdef __thumb2__
|
|
itt ne
|
|
#endif
|
|
movne r7,sp
|
|
bne .Louter
|
|
|
|
ldr r2,[r0,#12*4] @ pull rp
|
|
mov r5,sp
|
|
add r0,r0,#4 @ r0 to point at &tp[num]
|
|
sub r5,r0,r5 @ "original" num value
|
|
mov r4,sp @ "rewind" r4
|
|
mov r1,r4 @ "borrow" r1
|
|
sub r3,r3,r5 @ "rewind" r3 to &np[0]
|
|
|
|
subs r7,r7,r7 @ "clear" carry flag
|
|
.Lsub: ldr r7,[r4],#4
|
|
ldr r6,[r3],#4
|
|
sbcs r7,r7,r6 @ tp[j]-np[j]
|
|
str r7,[r2],#4 @ rp[j]=
|
|
teq r4,r0 @ preserve carry
|
|
bne .Lsub
|
|
sbcs r14,r14,#0 @ upmost carry
|
|
mov r4,sp @ "rewind" r4
|
|
sub r2,r2,r5 @ "rewind" r2
|
|
|
|
.Lcopy: ldr r7,[r4] @ conditional copy
|
|
ldr r5,[r2]
|
|
str sp,[r4],#4 @ zap tp
|
|
#ifdef __thumb2__
|
|
it cc
|
|
#endif
|
|
movcc r5,r7
|
|
str r5,[r2],#4
|
|
teq r4,r0 @ preserve carry
|
|
bne .Lcopy
|
|
|
|
mov sp,r0
|
|
add sp,sp,#4 @ skip over tp[num+1]
|
|
ldmia sp!,{r4,r5,r6,r7,r8,r9,r10,r11,r12,lr} @ restore registers
|
|
add sp,sp,#2*4 @ skip over {r0,r2}
|
|
mov r0,#1
|
|
.Labrt:
|
|
#if __ARM_ARCH__>=5
|
|
bx lr @ bx lr
|
|
#else
|
|
tst lr,#1
|
|
moveq pc,lr @ be binary compatible with V4, yet
|
|
.word 0xe12fff1e @ interoperable with Thumb ISA:-)
|
|
#endif
|
|
.size bn_mul_mont,.-bn_mul_mont
|
|
#if __ARM_MAX_ARCH__>=7
|
|
.arch armv7-a
|
|
.fpu neon
|
|
|
|
.type bn_mul8x_mont_neon,%function
|
|
.align 5
|
|
bn_mul8x_mont_neon:
|
|
mov ip,sp
|
|
stmdb sp!,{r4,r5,r6,r7,r8,r9,r10,r11}
|
|
vstmdb sp!,{d8,d9,d10,d11,d12,d13,d14,d15} @ ABI specification says so
|
|
ldmia ip,{r4,r5} @ load rest of parameter block
|
|
mov ip,sp
|
|
|
|
cmp r5,#8
|
|
bhi .LNEON_8n
|
|
|
|
@ special case for r5==8, everything is in register bank...
|
|
|
|
vld1.32 {d28[0]}, [r2,:32]!
|
|
veor d8,d8,d8
|
|
sub r7,sp,r5,lsl#4
|
|
vld1.32 {d0,d1,d2,d3}, [r1]! @ can't specify :32 :-(
|
|
and r7,r7,#-64
|
|
vld1.32 {d30[0]}, [r4,:32]
|
|
mov sp,r7 @ alloca
|
|
vzip.16 d28,d8
|
|
|
|
vmull.u32 q6,d28,d0[0]
|
|
vmull.u32 q7,d28,d0[1]
|
|
vmull.u32 q8,d28,d1[0]
|
|
vshl.i64 d29,d13,#16
|
|
vmull.u32 q9,d28,d1[1]
|
|
|
|
vadd.u64 d29,d29,d12
|
|
veor d8,d8,d8
|
|
vmul.u32 d29,d29,d30
|
|
|
|
vmull.u32 q10,d28,d2[0]
|
|
vld1.32 {d4,d5,d6,d7}, [r3]!
|
|
vmull.u32 q11,d28,d2[1]
|
|
vmull.u32 q12,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmull.u32 q13,d28,d3[1]
|
|
|
|
vmlal.u32 q6,d29,d4[0]
|
|
sub r9,r5,#1
|
|
vmlal.u32 q7,d29,d4[1]
|
|
vmlal.u32 q8,d29,d5[0]
|
|
vmlal.u32 q9,d29,d5[1]
|
|
|
|
vmlal.u32 q10,d29,d6[0]
|
|
vmov q5,q6
|
|
vmlal.u32 q11,d29,d6[1]
|
|
vmov q6,q7
|
|
vmlal.u32 q12,d29,d7[0]
|
|
vmov q7,q8
|
|
vmlal.u32 q13,d29,d7[1]
|
|
vmov q8,q9
|
|
vmov q9,q10
|
|
vshr.u64 d10,d10,#16
|
|
vmov q10,q11
|
|
vmov q11,q12
|
|
vadd.u64 d10,d10,d11
|
|
vmov q12,q13
|
|
veor q13,q13
|
|
vshr.u64 d10,d10,#16
|
|
|
|
b .LNEON_outer8
|
|
|
|
.align 4
|
|
.LNEON_outer8:
|
|
vld1.32 {d28[0]}, [r2,:32]!
|
|
veor d8,d8,d8
|
|
vzip.16 d28,d8
|
|
vadd.u64 d12,d12,d10
|
|
|
|
vmlal.u32 q6,d28,d0[0]
|
|
vmlal.u32 q7,d28,d0[1]
|
|
vmlal.u32 q8,d28,d1[0]
|
|
vshl.i64 d29,d13,#16
|
|
vmlal.u32 q9,d28,d1[1]
|
|
|
|
vadd.u64 d29,d29,d12
|
|
veor d8,d8,d8
|
|
subs r9,r9,#1
|
|
vmul.u32 d29,d29,d30
|
|
|
|
vmlal.u32 q10,d28,d2[0]
|
|
vmlal.u32 q11,d28,d2[1]
|
|
vmlal.u32 q12,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q13,d28,d3[1]
|
|
|
|
vmlal.u32 q6,d29,d4[0]
|
|
vmlal.u32 q7,d29,d4[1]
|
|
vmlal.u32 q8,d29,d5[0]
|
|
vmlal.u32 q9,d29,d5[1]
|
|
|
|
vmlal.u32 q10,d29,d6[0]
|
|
vmov q5,q6
|
|
vmlal.u32 q11,d29,d6[1]
|
|
vmov q6,q7
|
|
vmlal.u32 q12,d29,d7[0]
|
|
vmov q7,q8
|
|
vmlal.u32 q13,d29,d7[1]
|
|
vmov q8,q9
|
|
vmov q9,q10
|
|
vshr.u64 d10,d10,#16
|
|
vmov q10,q11
|
|
vmov q11,q12
|
|
vadd.u64 d10,d10,d11
|
|
vmov q12,q13
|
|
veor q13,q13
|
|
vshr.u64 d10,d10,#16
|
|
|
|
bne .LNEON_outer8
|
|
|
|
vadd.u64 d12,d12,d10
|
|
mov r7,sp
|
|
vshr.u64 d10,d12,#16
|
|
mov r8,r5
|
|
vadd.u64 d13,d13,d10
|
|
add r6,sp,#96
|
|
vshr.u64 d10,d13,#16
|
|
vzip.16 d12,d13
|
|
|
|
b .LNEON_tail_entry
|
|
|
|
.align 4
|
|
.LNEON_8n:
|
|
veor q6,q6,q6
|
|
sub r7,sp,#128
|
|
veor q7,q7,q7
|
|
sub r7,r7,r5,lsl#4
|
|
veor q8,q8,q8
|
|
and r7,r7,#-64
|
|
veor q9,q9,q9
|
|
mov sp,r7 @ alloca
|
|
veor q10,q10,q10
|
|
add r7,r7,#256
|
|
veor q11,q11,q11
|
|
sub r8,r5,#8
|
|
veor q12,q12,q12
|
|
veor q13,q13,q13
|
|
|
|
.LNEON_8n_init:
|
|
vst1.64 {q6,q7},[r7,:256]!
|
|
subs r8,r8,#8
|
|
vst1.64 {q8,q9},[r7,:256]!
|
|
vst1.64 {q10,q11},[r7,:256]!
|
|
vst1.64 {q12,q13},[r7,:256]!
|
|
bne .LNEON_8n_init
|
|
|
|
add r6,sp,#256
|
|
vld1.32 {d0,d1,d2,d3},[r1]!
|
|
add r10,sp,#8
|
|
vld1.32 {d30[0]},[r4,:32]
|
|
mov r9,r5
|
|
b .LNEON_8n_outer
|
|
|
|
.align 4
|
|
.LNEON_8n_outer:
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
veor d8,d8,d8
|
|
vzip.16 d28,d8
|
|
add r7,sp,#128
|
|
vld1.32 {d4,d5,d6,d7},[r3]!
|
|
|
|
vmlal.u32 q6,d28,d0[0]
|
|
vmlal.u32 q7,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q8,d28,d1[0]
|
|
vshl.i64 d29,d13,#16
|
|
vmlal.u32 q9,d28,d1[1]
|
|
vadd.u64 d29,d29,d12
|
|
vmlal.u32 q10,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q11,d28,d2[1]
|
|
vst1.32 {d28},[sp,:64] @ put aside smashed b[8*i+0]
|
|
vmlal.u32 q12,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q13,d28,d3[1]
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
vmlal.u32 q6,d29,d4[0]
|
|
veor d10,d10,d10
|
|
vmlal.u32 q7,d29,d4[1]
|
|
vzip.16 d28,d10
|
|
vmlal.u32 q8,d29,d5[0]
|
|
vshr.u64 d12,d12,#16
|
|
vmlal.u32 q9,d29,d5[1]
|
|
vmlal.u32 q10,d29,d6[0]
|
|
vadd.u64 d12,d12,d13
|
|
vmlal.u32 q11,d29,d6[1]
|
|
vshr.u64 d12,d12,#16
|
|
vmlal.u32 q12,d29,d7[0]
|
|
vmlal.u32 q13,d29,d7[1]
|
|
vadd.u64 d14,d14,d12
|
|
vst1.32 {d29},[r10,:64]! @ put aside smashed m[8*i+0]
|
|
vmlal.u32 q7,d28,d0[0]
|
|
vld1.64 {q6},[r6,:128]!
|
|
vmlal.u32 q8,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q9,d28,d1[0]
|
|
vshl.i64 d29,d15,#16
|
|
vmlal.u32 q10,d28,d1[1]
|
|
vadd.u64 d29,d29,d14
|
|
vmlal.u32 q11,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q12,d28,d2[1]
|
|
vst1.32 {d28},[r10,:64]! @ put aside smashed b[8*i+1]
|
|
vmlal.u32 q13,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q6,d28,d3[1]
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
vmlal.u32 q7,d29,d4[0]
|
|
veor d10,d10,d10
|
|
vmlal.u32 q8,d29,d4[1]
|
|
vzip.16 d28,d10
|
|
vmlal.u32 q9,d29,d5[0]
|
|
vshr.u64 d14,d14,#16
|
|
vmlal.u32 q10,d29,d5[1]
|
|
vmlal.u32 q11,d29,d6[0]
|
|
vadd.u64 d14,d14,d15
|
|
vmlal.u32 q12,d29,d6[1]
|
|
vshr.u64 d14,d14,#16
|
|
vmlal.u32 q13,d29,d7[0]
|
|
vmlal.u32 q6,d29,d7[1]
|
|
vadd.u64 d16,d16,d14
|
|
vst1.32 {d29},[r10,:64]! @ put aside smashed m[8*i+1]
|
|
vmlal.u32 q8,d28,d0[0]
|
|
vld1.64 {q7},[r6,:128]!
|
|
vmlal.u32 q9,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q10,d28,d1[0]
|
|
vshl.i64 d29,d17,#16
|
|
vmlal.u32 q11,d28,d1[1]
|
|
vadd.u64 d29,d29,d16
|
|
vmlal.u32 q12,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q13,d28,d2[1]
|
|
vst1.32 {d28},[r10,:64]! @ put aside smashed b[8*i+2]
|
|
vmlal.u32 q6,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q7,d28,d3[1]
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
vmlal.u32 q8,d29,d4[0]
|
|
veor d10,d10,d10
|
|
vmlal.u32 q9,d29,d4[1]
|
|
vzip.16 d28,d10
|
|
vmlal.u32 q10,d29,d5[0]
|
|
vshr.u64 d16,d16,#16
|
|
vmlal.u32 q11,d29,d5[1]
|
|
vmlal.u32 q12,d29,d6[0]
|
|
vadd.u64 d16,d16,d17
|
|
vmlal.u32 q13,d29,d6[1]
|
|
vshr.u64 d16,d16,#16
|
|
vmlal.u32 q6,d29,d7[0]
|
|
vmlal.u32 q7,d29,d7[1]
|
|
vadd.u64 d18,d18,d16
|
|
vst1.32 {d29},[r10,:64]! @ put aside smashed m[8*i+2]
|
|
vmlal.u32 q9,d28,d0[0]
|
|
vld1.64 {q8},[r6,:128]!
|
|
vmlal.u32 q10,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q11,d28,d1[0]
|
|
vshl.i64 d29,d19,#16
|
|
vmlal.u32 q12,d28,d1[1]
|
|
vadd.u64 d29,d29,d18
|
|
vmlal.u32 q13,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q6,d28,d2[1]
|
|
vst1.32 {d28},[r10,:64]! @ put aside smashed b[8*i+3]
|
|
vmlal.u32 q7,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q8,d28,d3[1]
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
vmlal.u32 q9,d29,d4[0]
|
|
veor d10,d10,d10
|
|
vmlal.u32 q10,d29,d4[1]
|
|
vzip.16 d28,d10
|
|
vmlal.u32 q11,d29,d5[0]
|
|
vshr.u64 d18,d18,#16
|
|
vmlal.u32 q12,d29,d5[1]
|
|
vmlal.u32 q13,d29,d6[0]
|
|
vadd.u64 d18,d18,d19
|
|
vmlal.u32 q6,d29,d6[1]
|
|
vshr.u64 d18,d18,#16
|
|
vmlal.u32 q7,d29,d7[0]
|
|
vmlal.u32 q8,d29,d7[1]
|
|
vadd.u64 d20,d20,d18
|
|
vst1.32 {d29},[r10,:64]! @ put aside smashed m[8*i+3]
|
|
vmlal.u32 q10,d28,d0[0]
|
|
vld1.64 {q9},[r6,:128]!
|
|
vmlal.u32 q11,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q12,d28,d1[0]
|
|
vshl.i64 d29,d21,#16
|
|
vmlal.u32 q13,d28,d1[1]
|
|
vadd.u64 d29,d29,d20
|
|
vmlal.u32 q6,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q7,d28,d2[1]
|
|
vst1.32 {d28},[r10,:64]! @ put aside smashed b[8*i+4]
|
|
vmlal.u32 q8,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q9,d28,d3[1]
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
vmlal.u32 q10,d29,d4[0]
|
|
veor d10,d10,d10
|
|
vmlal.u32 q11,d29,d4[1]
|
|
vzip.16 d28,d10
|
|
vmlal.u32 q12,d29,d5[0]
|
|
vshr.u64 d20,d20,#16
|
|
vmlal.u32 q13,d29,d5[1]
|
|
vmlal.u32 q6,d29,d6[0]
|
|
vadd.u64 d20,d20,d21
|
|
vmlal.u32 q7,d29,d6[1]
|
|
vshr.u64 d20,d20,#16
|
|
vmlal.u32 q8,d29,d7[0]
|
|
vmlal.u32 q9,d29,d7[1]
|
|
vadd.u64 d22,d22,d20
|
|
vst1.32 {d29},[r10,:64]! @ put aside smashed m[8*i+4]
|
|
vmlal.u32 q11,d28,d0[0]
|
|
vld1.64 {q10},[r6,:128]!
|
|
vmlal.u32 q12,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q13,d28,d1[0]
|
|
vshl.i64 d29,d23,#16
|
|
vmlal.u32 q6,d28,d1[1]
|
|
vadd.u64 d29,d29,d22
|
|
vmlal.u32 q7,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q8,d28,d2[1]
|
|
vst1.32 {d28},[r10,:64]! @ put aside smashed b[8*i+5]
|
|
vmlal.u32 q9,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q10,d28,d3[1]
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
vmlal.u32 q11,d29,d4[0]
|
|
veor d10,d10,d10
|
|
vmlal.u32 q12,d29,d4[1]
|
|
vzip.16 d28,d10
|
|
vmlal.u32 q13,d29,d5[0]
|
|
vshr.u64 d22,d22,#16
|
|
vmlal.u32 q6,d29,d5[1]
|
|
vmlal.u32 q7,d29,d6[0]
|
|
vadd.u64 d22,d22,d23
|
|
vmlal.u32 q8,d29,d6[1]
|
|
vshr.u64 d22,d22,#16
|
|
vmlal.u32 q9,d29,d7[0]
|
|
vmlal.u32 q10,d29,d7[1]
|
|
vadd.u64 d24,d24,d22
|
|
vst1.32 {d29},[r10,:64]! @ put aside smashed m[8*i+5]
|
|
vmlal.u32 q12,d28,d0[0]
|
|
vld1.64 {q11},[r6,:128]!
|
|
vmlal.u32 q13,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q6,d28,d1[0]
|
|
vshl.i64 d29,d25,#16
|
|
vmlal.u32 q7,d28,d1[1]
|
|
vadd.u64 d29,d29,d24
|
|
vmlal.u32 q8,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q9,d28,d2[1]
|
|
vst1.32 {d28},[r10,:64]! @ put aside smashed b[8*i+6]
|
|
vmlal.u32 q10,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q11,d28,d3[1]
|
|
vld1.32 {d28[0]},[r2,:32]! @ *b++
|
|
vmlal.u32 q12,d29,d4[0]
|
|
veor d10,d10,d10
|
|
vmlal.u32 q13,d29,d4[1]
|
|
vzip.16 d28,d10
|
|
vmlal.u32 q6,d29,d5[0]
|
|
vshr.u64 d24,d24,#16
|
|
vmlal.u32 q7,d29,d5[1]
|
|
vmlal.u32 q8,d29,d6[0]
|
|
vadd.u64 d24,d24,d25
|
|
vmlal.u32 q9,d29,d6[1]
|
|
vshr.u64 d24,d24,#16
|
|
vmlal.u32 q10,d29,d7[0]
|
|
vmlal.u32 q11,d29,d7[1]
|
|
vadd.u64 d26,d26,d24
|
|
vst1.32 {d29},[r10,:64]! @ put aside smashed m[8*i+6]
|
|
vmlal.u32 q13,d28,d0[0]
|
|
vld1.64 {q12},[r6,:128]!
|
|
vmlal.u32 q6,d28,d0[1]
|
|
veor d8,d8,d8
|
|
vmlal.u32 q7,d28,d1[0]
|
|
vshl.i64 d29,d27,#16
|
|
vmlal.u32 q8,d28,d1[1]
|
|
vadd.u64 d29,d29,d26
|
|
vmlal.u32 q9,d28,d2[0]
|
|
vmul.u32 d29,d29,d30
|
|
vmlal.u32 q10,d28,d2[1]
|
|
vst1.32 {d28},[r10,:64]! @ put aside smashed b[8*i+7]
|
|
vmlal.u32 q11,d28,d3[0]
|
|
vzip.16 d29,d8
|
|
vmlal.u32 q12,d28,d3[1]
|
|
vld1.32 {d28},[sp,:64] @ pull smashed b[8*i+0]
|
|
vmlal.u32 q13,d29,d4[0]
|
|
vld1.32 {d0,d1,d2,d3},[r1]!
|
|
vmlal.u32 q6,d29,d4[1]
|
|
vmlal.u32 q7,d29,d5[0]
|
|
vshr.u64 d26,d26,#16
|
|
vmlal.u32 q8,d29,d5[1]
|
|
vmlal.u32 q9,d29,d6[0]
|
|
vadd.u64 d26,d26,d27
|
|
vmlal.u32 q10,d29,d6[1]
|
|
vshr.u64 d26,d26,#16
|
|
vmlal.u32 q11,d29,d7[0]
|
|
vmlal.u32 q12,d29,d7[1]
|
|
vadd.u64 d12,d12,d26
|
|
vst1.32 {d29},[r10,:64] @ put aside smashed m[8*i+7]
|
|
add r10,sp,#8 @ rewind
|
|
sub r8,r5,#8
|
|
b .LNEON_8n_inner
|
|
|
|
.align 4
|
|
.LNEON_8n_inner:
|
|
subs r8,r8,#8
|
|
vmlal.u32 q6,d28,d0[0]
|
|
vld1.64 {q13},[r6,:128]
|
|
vmlal.u32 q7,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+0]
|
|
vmlal.u32 q8,d28,d1[0]
|
|
vld1.32 {d4,d5,d6,d7},[r3]!
|
|
vmlal.u32 q9,d28,d1[1]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q10,d28,d2[0]
|
|
vmlal.u32 q11,d28,d2[1]
|
|
vmlal.u32 q12,d28,d3[0]
|
|
vmlal.u32 q13,d28,d3[1]
|
|
vld1.32 {d28},[r10,:64]! @ pull smashed b[8*i+1]
|
|
vmlal.u32 q6,d29,d4[0]
|
|
vmlal.u32 q7,d29,d4[1]
|
|
vmlal.u32 q8,d29,d5[0]
|
|
vmlal.u32 q9,d29,d5[1]
|
|
vmlal.u32 q10,d29,d6[0]
|
|
vmlal.u32 q11,d29,d6[1]
|
|
vmlal.u32 q12,d29,d7[0]
|
|
vmlal.u32 q13,d29,d7[1]
|
|
vst1.64 {q6},[r7,:128]!
|
|
vmlal.u32 q7,d28,d0[0]
|
|
vld1.64 {q6},[r6,:128]
|
|
vmlal.u32 q8,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+1]
|
|
vmlal.u32 q9,d28,d1[0]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q10,d28,d1[1]
|
|
vmlal.u32 q11,d28,d2[0]
|
|
vmlal.u32 q12,d28,d2[1]
|
|
vmlal.u32 q13,d28,d3[0]
|
|
vmlal.u32 q6,d28,d3[1]
|
|
vld1.32 {d28},[r10,:64]! @ pull smashed b[8*i+2]
|
|
vmlal.u32 q7,d29,d4[0]
|
|
vmlal.u32 q8,d29,d4[1]
|
|
vmlal.u32 q9,d29,d5[0]
|
|
vmlal.u32 q10,d29,d5[1]
|
|
vmlal.u32 q11,d29,d6[0]
|
|
vmlal.u32 q12,d29,d6[1]
|
|
vmlal.u32 q13,d29,d7[0]
|
|
vmlal.u32 q6,d29,d7[1]
|
|
vst1.64 {q7},[r7,:128]!
|
|
vmlal.u32 q8,d28,d0[0]
|
|
vld1.64 {q7},[r6,:128]
|
|
vmlal.u32 q9,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+2]
|
|
vmlal.u32 q10,d28,d1[0]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q11,d28,d1[1]
|
|
vmlal.u32 q12,d28,d2[0]
|
|
vmlal.u32 q13,d28,d2[1]
|
|
vmlal.u32 q6,d28,d3[0]
|
|
vmlal.u32 q7,d28,d3[1]
|
|
vld1.32 {d28},[r10,:64]! @ pull smashed b[8*i+3]
|
|
vmlal.u32 q8,d29,d4[0]
|
|
vmlal.u32 q9,d29,d4[1]
|
|
vmlal.u32 q10,d29,d5[0]
|
|
vmlal.u32 q11,d29,d5[1]
|
|
vmlal.u32 q12,d29,d6[0]
|
|
vmlal.u32 q13,d29,d6[1]
|
|
vmlal.u32 q6,d29,d7[0]
|
|
vmlal.u32 q7,d29,d7[1]
|
|
vst1.64 {q8},[r7,:128]!
|
|
vmlal.u32 q9,d28,d0[0]
|
|
vld1.64 {q8},[r6,:128]
|
|
vmlal.u32 q10,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+3]
|
|
vmlal.u32 q11,d28,d1[0]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q12,d28,d1[1]
|
|
vmlal.u32 q13,d28,d2[0]
|
|
vmlal.u32 q6,d28,d2[1]
|
|
vmlal.u32 q7,d28,d3[0]
|
|
vmlal.u32 q8,d28,d3[1]
|
|
vld1.32 {d28},[r10,:64]! @ pull smashed b[8*i+4]
|
|
vmlal.u32 q9,d29,d4[0]
|
|
vmlal.u32 q10,d29,d4[1]
|
|
vmlal.u32 q11,d29,d5[0]
|
|
vmlal.u32 q12,d29,d5[1]
|
|
vmlal.u32 q13,d29,d6[0]
|
|
vmlal.u32 q6,d29,d6[1]
|
|
vmlal.u32 q7,d29,d7[0]
|
|
vmlal.u32 q8,d29,d7[1]
|
|
vst1.64 {q9},[r7,:128]!
|
|
vmlal.u32 q10,d28,d0[0]
|
|
vld1.64 {q9},[r6,:128]
|
|
vmlal.u32 q11,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+4]
|
|
vmlal.u32 q12,d28,d1[0]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q13,d28,d1[1]
|
|
vmlal.u32 q6,d28,d2[0]
|
|
vmlal.u32 q7,d28,d2[1]
|
|
vmlal.u32 q8,d28,d3[0]
|
|
vmlal.u32 q9,d28,d3[1]
|
|
vld1.32 {d28},[r10,:64]! @ pull smashed b[8*i+5]
|
|
vmlal.u32 q10,d29,d4[0]
|
|
vmlal.u32 q11,d29,d4[1]
|
|
vmlal.u32 q12,d29,d5[0]
|
|
vmlal.u32 q13,d29,d5[1]
|
|
vmlal.u32 q6,d29,d6[0]
|
|
vmlal.u32 q7,d29,d6[1]
|
|
vmlal.u32 q8,d29,d7[0]
|
|
vmlal.u32 q9,d29,d7[1]
|
|
vst1.64 {q10},[r7,:128]!
|
|
vmlal.u32 q11,d28,d0[0]
|
|
vld1.64 {q10},[r6,:128]
|
|
vmlal.u32 q12,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+5]
|
|
vmlal.u32 q13,d28,d1[0]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q6,d28,d1[1]
|
|
vmlal.u32 q7,d28,d2[0]
|
|
vmlal.u32 q8,d28,d2[1]
|
|
vmlal.u32 q9,d28,d3[0]
|
|
vmlal.u32 q10,d28,d3[1]
|
|
vld1.32 {d28},[r10,:64]! @ pull smashed b[8*i+6]
|
|
vmlal.u32 q11,d29,d4[0]
|
|
vmlal.u32 q12,d29,d4[1]
|
|
vmlal.u32 q13,d29,d5[0]
|
|
vmlal.u32 q6,d29,d5[1]
|
|
vmlal.u32 q7,d29,d6[0]
|
|
vmlal.u32 q8,d29,d6[1]
|
|
vmlal.u32 q9,d29,d7[0]
|
|
vmlal.u32 q10,d29,d7[1]
|
|
vst1.64 {q11},[r7,:128]!
|
|
vmlal.u32 q12,d28,d0[0]
|
|
vld1.64 {q11},[r6,:128]
|
|
vmlal.u32 q13,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+6]
|
|
vmlal.u32 q6,d28,d1[0]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q7,d28,d1[1]
|
|
vmlal.u32 q8,d28,d2[0]
|
|
vmlal.u32 q9,d28,d2[1]
|
|
vmlal.u32 q10,d28,d3[0]
|
|
vmlal.u32 q11,d28,d3[1]
|
|
vld1.32 {d28},[r10,:64]! @ pull smashed b[8*i+7]
|
|
vmlal.u32 q12,d29,d4[0]
|
|
vmlal.u32 q13,d29,d4[1]
|
|
vmlal.u32 q6,d29,d5[0]
|
|
vmlal.u32 q7,d29,d5[1]
|
|
vmlal.u32 q8,d29,d6[0]
|
|
vmlal.u32 q9,d29,d6[1]
|
|
vmlal.u32 q10,d29,d7[0]
|
|
vmlal.u32 q11,d29,d7[1]
|
|
vst1.64 {q12},[r7,:128]!
|
|
vmlal.u32 q13,d28,d0[0]
|
|
vld1.64 {q12},[r6,:128]
|
|
vmlal.u32 q6,d28,d0[1]
|
|
vld1.32 {d29},[r10,:64]! @ pull smashed m[8*i+7]
|
|
vmlal.u32 q7,d28,d1[0]
|
|
it ne
|
|
addne r6,r6,#16 @ don't advance in last iteration
|
|
vmlal.u32 q8,d28,d1[1]
|
|
vmlal.u32 q9,d28,d2[0]
|
|
vmlal.u32 q10,d28,d2[1]
|
|
vmlal.u32 q11,d28,d3[0]
|
|
vmlal.u32 q12,d28,d3[1]
|
|
it eq
|
|
subeq r1,r1,r5,lsl#2 @ rewind
|
|
vmlal.u32 q13,d29,d4[0]
|
|
vld1.32 {d28},[sp,:64] @ pull smashed b[8*i+0]
|
|
vmlal.u32 q6,d29,d4[1]
|
|
vld1.32 {d0,d1,d2,d3},[r1]!
|
|
vmlal.u32 q7,d29,d5[0]
|
|
add r10,sp,#8 @ rewind
|
|
vmlal.u32 q8,d29,d5[1]
|
|
vmlal.u32 q9,d29,d6[0]
|
|
vmlal.u32 q10,d29,d6[1]
|
|
vmlal.u32 q11,d29,d7[0]
|
|
vst1.64 {q13},[r7,:128]!
|
|
vmlal.u32 q12,d29,d7[1]
|
|
|
|
bne .LNEON_8n_inner
|
|
add r6,sp,#128
|
|
vst1.64 {q6,q7},[r7,:256]!
|
|
veor q2,q2,q2 @ d4-d5
|
|
vst1.64 {q8,q9},[r7,:256]!
|
|
veor q3,q3,q3 @ d6-d7
|
|
vst1.64 {q10,q11},[r7,:256]!
|
|
vst1.64 {q12},[r7,:128]
|
|
|
|
subs r9,r9,#8
|
|
vld1.64 {q6,q7},[r6,:256]!
|
|
vld1.64 {q8,q9},[r6,:256]!
|
|
vld1.64 {q10,q11},[r6,:256]!
|
|
vld1.64 {q12,q13},[r6,:256]!
|
|
|
|
itt ne
|
|
subne r3,r3,r5,lsl#2 @ rewind
|
|
bne .LNEON_8n_outer
|
|
|
|
add r7,sp,#128
|
|
vst1.64 {q2,q3}, [sp,:256]! @ start wiping stack frame
|
|
vshr.u64 d10,d12,#16
|
|
vst1.64 {q2,q3},[sp,:256]!
|
|
vadd.u64 d13,d13,d10
|
|
vst1.64 {q2,q3}, [sp,:256]!
|
|
vshr.u64 d10,d13,#16
|
|
vst1.64 {q2,q3}, [sp,:256]!
|
|
vzip.16 d12,d13
|
|
|
|
mov r8,r5
|
|
b .LNEON_tail_entry
|
|
|
|
.align 4
|
|
.LNEON_tail:
|
|
vadd.u64 d12,d12,d10
|
|
vshr.u64 d10,d12,#16
|
|
vld1.64 {q8,q9}, [r6, :256]!
|
|
vadd.u64 d13,d13,d10
|
|
vld1.64 {q10,q11}, [r6, :256]!
|
|
vshr.u64 d10,d13,#16
|
|
vld1.64 {q12,q13}, [r6, :256]!
|
|
vzip.16 d12,d13
|
|
|
|
.LNEON_tail_entry:
|
|
vadd.u64 d14,d14,d10
|
|
vst1.32 {d12[0]}, [r7, :32]!
|
|
vshr.u64 d10,d14,#16
|
|
vadd.u64 d15,d15,d10
|
|
vshr.u64 d10,d15,#16
|
|
vzip.16 d14,d15
|
|
vadd.u64 d16,d16,d10
|
|
vst1.32 {d14[0]}, [r7, :32]!
|
|
vshr.u64 d10,d16,#16
|
|
vadd.u64 d17,d17,d10
|
|
vshr.u64 d10,d17,#16
|
|
vzip.16 d16,d17
|
|
vadd.u64 d18,d18,d10
|
|
vst1.32 {d16[0]}, [r7, :32]!
|
|
vshr.u64 d10,d18,#16
|
|
vadd.u64 d19,d19,d10
|
|
vshr.u64 d10,d19,#16
|
|
vzip.16 d18,d19
|
|
vadd.u64 d20,d20,d10
|
|
vst1.32 {d18[0]}, [r7, :32]!
|
|
vshr.u64 d10,d20,#16
|
|
vadd.u64 d21,d21,d10
|
|
vshr.u64 d10,d21,#16
|
|
vzip.16 d20,d21
|
|
vadd.u64 d22,d22,d10
|
|
vst1.32 {d20[0]}, [r7, :32]!
|
|
vshr.u64 d10,d22,#16
|
|
vadd.u64 d23,d23,d10
|
|
vshr.u64 d10,d23,#16
|
|
vzip.16 d22,d23
|
|
vadd.u64 d24,d24,d10
|
|
vst1.32 {d22[0]}, [r7, :32]!
|
|
vshr.u64 d10,d24,#16
|
|
vadd.u64 d25,d25,d10
|
|
vshr.u64 d10,d25,#16
|
|
vzip.16 d24,d25
|
|
vadd.u64 d26,d26,d10
|
|
vst1.32 {d24[0]}, [r7, :32]!
|
|
vshr.u64 d10,d26,#16
|
|
vadd.u64 d27,d27,d10
|
|
vshr.u64 d10,d27,#16
|
|
vzip.16 d26,d27
|
|
vld1.64 {q6,q7}, [r6, :256]!
|
|
subs r8,r8,#8
|
|
vst1.32 {d26[0]}, [r7, :32]!
|
|
bne .LNEON_tail
|
|
|
|
vst1.32 {d10[0]}, [r7, :32] @ top-most bit
|
|
sub r3,r3,r5,lsl#2 @ rewind r3
|
|
subs r1,sp,#0 @ clear carry flag
|
|
add r2,sp,r5,lsl#2
|
|
|
|
.LNEON_sub:
|
|
ldmia r1!, {r4,r5,r6,r7}
|
|
ldmia r3!, {r8,r9,r10,r11}
|
|
sbcs r8, r4,r8
|
|
sbcs r9, r5,r9
|
|
sbcs r10,r6,r10
|
|
sbcs r11,r7,r11
|
|
teq r1,r2 @ preserves carry
|
|
stmia r0!, {r8,r9,r10,r11}
|
|
bne .LNEON_sub
|
|
|
|
ldr r10, [r1] @ load top-most bit
|
|
mov r11,sp
|
|
veor q0,q0,q0
|
|
sub r11,r2,r11 @ this is num*4
|
|
veor q1,q1,q1
|
|
mov r1,sp
|
|
sub r0,r0,r11 @ rewind r0
|
|
mov r3,r2 @ second 3/4th of frame
|
|
sbcs r10,r10,#0 @ result is carry flag
|
|
|
|
.LNEON_copy_n_zap:
|
|
ldmia r1!, {r4,r5,r6,r7}
|
|
ldmia r0, {r8,r9,r10,r11}
|
|
it cc
|
|
movcc r8, r4
|
|
vst1.64 {q0,q1}, [r3,:256]! @ wipe
|
|
itt cc
|
|
movcc r9, r5
|
|
movcc r10,r6
|
|
vst1.64 {q0,q1}, [r3,:256]! @ wipe
|
|
it cc
|
|
movcc r11,r7
|
|
ldmia r1, {r4,r5,r6,r7}
|
|
stmia r0!, {r8,r9,r10,r11}
|
|
sub r1,r1,#16
|
|
ldmia r0, {r8,r9,r10,r11}
|
|
it cc
|
|
movcc r8, r4
|
|
vst1.64 {q0,q1}, [r1,:256]! @ wipe
|
|
itt cc
|
|
movcc r9, r5
|
|
movcc r10,r6
|
|
vst1.64 {q0,q1}, [r3,:256]! @ wipe
|
|
it cc
|
|
movcc r11,r7
|
|
teq r1,r2 @ preserves carry
|
|
stmia r0!, {r8,r9,r10,r11}
|
|
bne .LNEON_copy_n_zap
|
|
|
|
mov sp,ip
|
|
vldmia sp!,{d8,d9,d10,d11,d12,d13,d14,d15}
|
|
ldmia sp!,{r4,r5,r6,r7,r8,r9,r10,r11}
|
|
bx lr @ bx lr
|
|
.size bn_mul8x_mont_neon,.-bn_mul8x_mont_neon
|
|
#endif
|
|
.byte 77,111,110,116,103,111,109,101,114,121,32,109,117,108,116,105,112,108,105,99,97,116,105,111,110,32,102,111,114,32,65,82,77,118,52,47,78,69,79,78,44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111,114,103,62,0
|
|
.align 2
|
|
.align 2
|
|
#if __ARM_MAX_ARCH__>=7
|
|
.comm OPENSSL_armcap_P,4,4
|
|
.hidden OPENSSL_armcap_P
|
|
#endif
|
|
#endif
|
|
#endif // !OPENSSL_NO_ASM
|
|
.section .note.GNU-stack,"",%progbits
|