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

#include "go_asm.h"
#include "textflag.h"

TEXT ·Count<ABIInternal>(SB),NOSPLIT,$0-40
	// X10 = b_base
	// X11 = b_len
	// X12 = b_cap (unused)
	// X13 = byte to count (want in X12)
	AND	$0xff, X13, X12
	MOV	ZERO, X14	// count
	ADD	X10, X11	// end

	PCALIGN	$16
loop:
	BEQ	X10, X11, done

	}

	bases := make([]profileSource, 0, len(s.Base))
	for _, src := range s.Base {
		bases = append(bases, profileSource{
			addr:   src,
			source: s,
		})
	}

	p, pbase, m, mbase, save, err := grabSourcesAndBases(sources, bases, o.Fetch, o.Obj, o.UI, o.HTTPTransport)
	if err != nil {
		return nil, err
	}

	if pbase != nil {
		if s.DiffBase {
			pbase.SetLabel("pprof::base", []string{"true"})
		}
		if s.Normalize {

#include "go_asm.h"
#include "textflag.h"

TEXT ·IndexByte(SB),NOSPLIT,$0-20
	MOVL	b_base+0(FP), SI
	MOVL	b_len+4(FP), CX
	MOVB	c+12(FP), AL
	MOVL	SI, DI
	CLD; REPN; SCASB
	JZ 3(PC)
	MOVL	$-1, ret+16(FP)
	RET
	SUBL	SI, DI
	SUBL	$1, DI
	MOVL	DI, ret+16(FP)
	RET

TEXT ·IndexByteString(SB),NOSPLIT,$0-16
	MOVL	s_base+0(FP), SI
	MOVL	s_len+4(FP), CX
	MOVB	c+8(FP), AL
	MOVL	SI, DI
	CLD; REPN; SCASB

	SHA256H2	V9.S4, V8, V3 \
	VMOV	V2.B16, V8.B16

// func sha256block(h []uint32, p []byte, k []uint32)
TEXT ·sha256block(SB),NOSPLIT,$0
	MOVD	h_base+0(FP), R0                           // Hash value first address
	MOVD	p_base+24(FP), R1                          // message first address
	MOVD	k_base+48(FP), R2                          // k constants first address
	MOVD	p_len+32(FP), R3                           // message length

#include "go_asm.h"
#include "textflag.h"

TEXT ·IndexByte(SB),NOSPLIT,$0-40
	MOVV	b_base+0(FP), R1
	MOVV	b_len+8(FP), R2
	MOVBU	c+24(FP), R3	// byte to find
	MOVV	R1, R4		// store base for later
	ADDV	R1, R2		// end
	ADDV	$-1, R1

loop:
	ADDV	$1, R1
	BEQ	R1, R2, notfound
	MOVBU	(R1), R5
	BNE	R3, R5, loop

	SUBV	R4, R1		// remove base
	MOVV	R1, ret+32(FP)
	RET

notfound:
	MOVV	$-1, R1
	MOVV	R1, ret+32(FP)

TEXT ·Compare<ABIInternal>(SB),NOSPLIT,$0-56
	// AX = a_base (want in SI)
	// BX = a_len  (want in BX)
	// CX = a_cap  (unused)
	// DI = b_base (want in DI)
	// SI = b_len  (want in DX)
	// R8 = b_cap  (unused)
	MOVQ	SI, DX
	MOVQ	AX, SI
	JMP	cmpbody<>(SB)

TEXT runtime·cmpstring<ABIInternal>(SB),NOSPLIT,$0-40
	// AX = a_base (want in SI)
	// BX = a_len  (want in BX)
	// CX = b_base (want in DI)
	// DI = b_len  (want in DX)

TEXT ·Index(SB),NOSPLIT|NOFRAME,$0-56
	LMG	a_base+0(FP), R1, R2  // R1=&s[0],   R2=len(s)
	LMG	b_base+24(FP), R3, R4 // R3=&sep[0], R4=len(sep)
	MOVD	$ret+48(FP), R5
	BR	indexbody<>(SB)

// Caller must confirm availability of vx facility before calling.
TEXT ·IndexString(SB),NOSPLIT|NOFRAME,$0-40
	LMG	a_base+0(FP), R1, R2  // R1=&s[0],   R2=len(s)
	LMG	b_base+16(FP), R3, R4 // R3=&sep[0], R4=len(sep)
	MOVD	$ret+32(FP), R5

TEXT ·Compare<ABIInternal>(SB),NOSPLIT|NOFRAME,$0-56
	// X10 = a_base
	// X11 = a_len
	// X12 = a_cap (unused)
	// X13 = b_base (want in X12)
	// X14 = b_len (want in X13)
	// X15 = b_cap (unused)
	MOV	X13, X12
	MOV	X14, X13
	JMP	compare<>(SB)

TEXT runtime·cmpstring<ABIInternal>(SB),NOSPLIT|NOFRAME,$0-40
	// X10 = a_base
	// X11 = a_len
	// X12 = b_base
	// X13 = b_len
	JMP	compare<>(SB)

// On entry:

#include "go_asm.h"
#include "textflag.h"

TEXT ·Index(SB),NOSPLIT,$0-56
	MOVD	a_base+0(FP), R0
	MOVD	a_len+8(FP), R1
	MOVD	b_base+24(FP), R2
	MOVD	b_len+32(FP), R3
	MOVD	$ret+48(FP), R9
	B	indexbody<>(SB)

TEXT ·IndexString(SB),NOSPLIT,$0-40
	MOVD	a_base+0(FP), R0
	MOVD	a_len+8(FP), R1
	MOVD	b_base+16(FP), R2
	MOVD	b_len+24(FP), R3
	MOVD	$ret+32(FP), R9
	B	indexbody<>(SB)

// input:

#include "go_asm.h"
#include "textflag.h"

TEXT ·Index(SB),NOSPLIT,$0-56
	MOVQ a_base+0(FP), DI
	MOVQ a_len+8(FP), DX
	MOVQ b_base+24(FP), R8
	MOVQ b_len+32(FP), AX
	MOVQ DI, R10
	LEAQ ret+48(FP), R11
	JMP  indexbody<>(SB)

TEXT ·IndexString(SB),NOSPLIT,$0-40
	MOVQ a_base+0(FP), DI
	MOVQ a_len+8(FP), DX
	MOVQ b_base+16(FP), R8
	MOVQ b_len+24(FP), AX
	MOVQ DI, R10
	LEAQ ret+32(FP), R11
	JMP  indexbody<>(SB)