LDP	-16(R4), (R6, R7)
	STP	(R12, R13), -16(R5)
	LDP	-32(R4), (R8, R9)
	LDP	-48(R4), (R10, R11)
	LDP.W	-64(R4), (R12, R13)
	SUB	R14, R5, R5
	SUBS	$128, R2, R2
	BLS	copy64_from_start

loop64_backward:
	STP	(R6, R7), -16(R5)
	LDP	-16(R4), (R6, R7)
	STP	(R8, R9), -32(R5)
	LDP	-32(R4), (R8, R9)
	STP	(R10, R11), -48(R5)
	LDP	-48(R4), (R10, R11)
	STP.W	(R12, R13), -64(R5)
	LDP.W	-64(R4), (R12, R13)

	MOVW trap+0(FP), R7 // syscall number
	MOVW a1+4(FP), R0 // a1
	MOVW a2+8(FP), R1 // a2
	MOVW a3+12(FP), R2 // a3
	MOVW a4+16(FP), R3 // a4
	MOVW R13, R4
	MOVW $a5+20(FP), R13 // a5 to a6 are passed on stack
	SWI $0 // syscall
	MOVW R4, R13
	MOVW $0, R2
	BCS error6
	MOVW R0, r1+28(FP) // r1
	MOVW R1, r2+32(FP) // r2
	MOVW R2, err+36(FP) // errno
	BL runtime·exitsyscall(SB)
	RET
error6:

TEXT ·useVFPv3(SB),NOSPLIT,$0
	WORD $0xeeb70b00	// vmov.f64 d0, #112
	RET

// try to run ARMv6K (or above) "ldrexd" instruction
TEXT ·useARMv6K(SB),NOSPLIT,$32
	MOVW R13, R2
	BIC  $15, R13
	WORD $0xe1bd0f9f	// ldrexd r0, r1, [sp]
	WORD $0xf57ff01f	// clrex
	MOVW R2, R13

	SHA256ROUND0(5, 0x59f111f1, R11, R12, R13, R14, R15, R8, R9, R10)
	SHA256ROUND0(6, 0x923f82a4, R10, R11, R12, R13, R14, R15, R8, R9)
	SHA256ROUND0(7, 0xab1c5ed5, R9, R10, R11, R12, R13, R14, R15, R8)
	SHA256ROUND0(8, 0xd807aa98, R8, R9, R10, R11, R12, R13, R14, R15)
	SHA256ROUND0(9, 0x12835b01, R15, R8, R9, R10, R11, R12, R13, R14)
	SHA256ROUND0(10, 0x243185be, R14, R15, R8, R9, R10, R11, R12, R13)

less_than32:
	TBZ	$4, R3, less_than16
	LDP.P	16(R1), (R11, R12)
	LDP.P	16(R2), (R13, R14)
	EOR	R11, R13, R13
	EOR	R12, R14, R14
	STP.P	(R13, R14), 16(R0)
less_than16:
	TBZ	$3, R3, less_than8
	MOVD.P	8(R1), R11
	MOVD.P	8(R2), R12
	EOR	R11, R12, R12
	MOVD.P	R12, 8(R0)
less_than8:
	TBZ	$2, R3, less_than4
	MOVWU.P	4(R1), R13
	MOVWU.P	4(R2), R14
	EORW	R13, R14, R14
	MOVWU.P	R14, 4(R0)
less_than4:
	TBZ	$1, R3, less_than2

	AND $0x7ffffffffffff, R1, R11
	AND $0x7ffffffffffff, R2, R12
	AND $0x7ffffffffffff, R3, R13
	AND $0x7ffffffffffff, R4, R14

	ADD R0>>51, R11, R11
	ADD R1>>51, R12, R12
	ADD R2>>51, R13, R13
	ADD R3>>51, R14, R14
	// R4>>51 * 19 + R10 -> R10
	LSR $51, R4, R21
	MOVD $19, R22
	MADD R22, R10, R21, R10

	STP (R10, R11), 0(R20)
	STP (R12, R13), 16(R20)
	MOVD R14, 32(R20)

// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

#include "textflag.h"

TEXT _rt0_arm_linux(SB),NOSPLIT|NOFRAME,$0
	MOVW	(R13), R0	// argc
	MOVW	$4(R13), R1		// argv
	MOVW	$_rt0_arm_linux1(SB), R4
	B		(R4)

// When building with -buildmode=c-shared, this symbol is called when the shared
// library is loaded.
TEXT _rt0_arm_linux_lib(SB),NOSPLIT,$0

	ADDU	$32, R1
	MOVW	0(R2), R8
	MOVW	4(R2), R9
	MOVW	8(R2), R10
	MOVW	12(R2), R11
	MOVW	16(R2), R12
	MOVW	20(R2), R13
	MOVW	24(R2), R14
	MOVW	28(R2), R15
	ADDU	$32, R2
	MOVW	R8, -32(R1)
	MOVW	R9, -28(R1)
	MOVW	R10, -24(R1)
	MOVW	R11, -20(R1)
	MOVW	R12, -16(R1)
	MOVW	R13, -12(R1)
	MOVW	R14, -8(R1)
	MOVW	R15, -4(R1)
	JMP	f_large

f_words:
	BEQ	R1, R6, f_tail
	ADDU	$4, R1
	MOVW	0(R2), R8

multiply:
	POLY1305_MUL(R8, R9, R10, R11, R12, BX, CX, R13, R14)
	SUBQ $16, R15
	CMPQ R15, $16
	JAE  loop

bytes_between_0_and_15:
	TESTQ R15, R15
	JZ    done
	MOVQ  $1, BX
	XORQ  CX, CX
	XORQ  R13, R13
	ADDQ  R15, SI

flush_buffer:
	SHLQ $8, BX, CX
	SHLQ $8, BX
	MOVB -1(SI), R13
	XORQ R13, BX
	DECQ SI
	DECQ R15
	JNZ  flush_buffer

	ADDQ BX, R8
	ADCQ CX, R9

	ADDV	$(1024*8), R3

	// Ask signaller to setgid
	MOVW	$1, R12
	DBAR
	MOVW	R12, ·Baton(SB)
	DBAR

	// Wait for setgid completion
loop:
	DBAR
	MOVW	·Baton(SB), R12
	OR	R13, R13, R13	// hint that we're in a spin loop
	BNE	R12, loop
	DBAR

	// Restore stack
	ADDV	$(-1024*8), R3