// license that can be found in the LICENSE file.

//go:build !math_big_pure_go

#include "textflag.h"

// This file provides fast assembly versions for the elementary
// arithmetic operations on vectors implemented in arith.go.

// func addVV(z, x, y []Word) (c Word)
TEXT ·addVV(SB),NOSPLIT,$0
	ADD.S	$0, R0		// clear carry flag
	MOVW	z+0(FP), R1
	MOVW	z_len+4(FP), R4
	MOVW	x+12(FP), R2
	MOVW	y+24(FP), R3

	BLT	tail		// Jump to the small string case
	SRD	$5, R4, R20
	MOVD	R20, CTR
	MOVD	$16, R21
	XXLXOR	V4, V4, V4
	XXLXOR	V5, V5, V5

	PCALIGN	$16
cmploop:
	LXVD2X	(R0)(R3), V0	// Count 32B per loop with two vector accumulators.
	LXVD2X	(R21)(R3), V2
	VCMPEQUB V2, V1, V2
	VCMPEQUB V0, V1, V0
	VPOPCNTD V2, V2		// A match is 0xFF or 0. Count the bits into doubleword buckets.
	VPOPCNTD V0, V0

type ofb struct {
	b       Block
	cipher  []byte
	out     []byte
	outUsed int
}

// NewOFB returns a [Stream] that encrypts or decrypts using the block cipher b
// in output feedback mode. The initialization vector iv's length must be equal
// to b's block size.
func NewOFB(b Block, iv []byte) Stream {
	blockSize := b.BlockSize()
	if len(iv) != blockSize {
		panic("cipher.NewOFB: IV length must equal block size")
	}

	// Create a struct callbackArgs on our stack.
	MOVD	$cbargs-(18*8+callbackArgs__size)(SP), R13
	MOVD	R12, callbackArgs_index(R13)	// callback index
	MOVD	R14, R0
	MOVD	R0, callbackArgs_args(R13)		// address of args vector
	MOVD	$0, R0
	MOVD	R0, callbackArgs_result(R13)	// result

	// Call cgocallback, which will call callbackWrap(frame).

	// Create a struct callbackArgs on our stack.
	MOVW	R12, (16+callbackArgs_index)(R13)	// callback index
	MOVW	$(16+callbackArgs__size+4*9)(R13), R0
	MOVW	R0, (16+callbackArgs_args)(R13)		// address of args vector
	MOVW	$0, R0
	MOVW	R0, (16+callbackArgs_result)(R13)	// result

	// Prepare for entry to Go.
	BL	runtime·load_g(SB)

	// Call cgocallback, which will call callbackWrap(frame).
	MOVW	$0, R0

// license that can be found in the LICENSE file.

//go:build !math_big_pure_go

#include "textflag.h"

// This file provides fast assembly versions for the elementary
// arithmetic operations on vectors implemented in arith.go.

// The carry bit is saved with SBBQ Rx, Rx: if the carry was set, Rx is -1, otherwise it is 0.
// It is restored with ADDQ Rx, Rx: if Rx was -1 the carry is set, otherwise it is cleared.

	fmt.Fprintln(w, `// Code generated by "go run gen_const_ppc64le.go"; DO NOT EDIT.`)
	fmt.Fprintln(w)
	fmt.Fprintln(w, `#include "textflag.h"`)

	// These are the polynomials supported in vector now.
	// If adding others, include the polynomial and a name
	// to identify it.

	genCrc32ConstTable(w, 0xedb88320, "IEEE")
	genCrc32ConstTable(w, 0x82f63b78, "Cast")
	genCrc32ConstTable(w, 0xeb31d82e, "Koop")

	// Create a struct callbackArgs on our stack.
	SUBL	$(12+callbackArgs__size), SP
	MOVL	AX, (12+callbackArgs_index)(SP)		// callback index
	MOVL	CX, (12+callbackArgs_args)(SP)		// address of args vector
	MOVL	$0, (12+callbackArgs_result)(SP)	// result
	LEAL	12(SP), AX	// AX = &callbackArgs{...}

	// Call cgocallback, which will call callbackWrap(frame).
	MOVL	$0, 8(SP)	// context

	MOVD	$32,R11		// set offsets to load into vector
	MOVD	$48,R12		// set offsets to load into vector

	PCALIGN	$16
cmp64_loop:
	LXVD2X	(R5)(R0),V3	// load bytes of A at offset 0 into vector
	LXVD2X	(R6)(R0),V4	// load bytes of B at offset 0 into vector
	VCMPEQUDCC	V3,V4,V1
	BGE	CR6,different	// jump out if its different

	LXVD2X	(R5)(R10),V3	// load bytes of A at offset 16 into vector

	MOVD  plist_base+0(FP), R1  // r1 points to plist
	MOVD  bpx_offset+24(FP), R2 // r2 offset to BPX vector table
	MOVD  R14, R7               // save r14
	MOVD  R15, R8               // save r15
	MOVWZ 16(R0), R9
	MOVWZ 544(R9), R9
	MOVWZ 24(R9), R9            // call vector in r9
	ADD   R2, R9                // add offset to vector table
	MOVWZ (R9), R9              // r9 points to entry point