#include "textflag.h"

// func archCeil(x float64) float64
TEXT ·archCeil(SB),NOSPLIT,$0
	FMOVD   x+0(FP), F0  // F0=x
	FSTCW   -2(SP)       // save old Control Word
	MOVW    -2(SP), AX
	ANDW    $0xf3ff, AX
	ORW     $0x0800, AX  // Rounding Control set to +Inf
	MOVW    AX, -4(SP)   // store new Control Word
	FLDCW   -4(SP)       // load new Control Word
	FRNDINT              // F0=Ceil(x)
	FLDCW   -2(SP)       // load old Control Word

	OR	$65536, R1            // c01d00010000
	OR	$-2, R1               // c0a1fffffffeb981001a
	ORW	R1, R2                // 1621
	ORW	R1, R2, R3            // b9f62031
	ORW	$1, R1                // a51b0001
	ORW	$131071, R1           // c01d0001ffff
	ORW	$65536, R1            // a51a0001
	ORW	$-2, R1               // c01dfffffffe
	XOR	R1, R2                // b9820021
	XOR	R1, R2, R3            // b9e72031

	FMOVD   F0, x-8(SP)
	WORD    $0x5B20F008     //s %r2, 8(%r15)
	RISBGZ	$57, $60, $51, R2, R3
	ANDW    $0xFFFF0000, R2
	RISBGN	$0, $31, $32, R2, R1
	ADDW    $0x4000000, R2
	BLEU    L17
L8:
	SRW     $8, R2, R2
	ORW     $0x45000000, R2
L4:
	FMOVD   log10rodataL19<>+120(SB), F2
	LDGR    R1, F4
	WFMADB  V4, V0, V2, V0
	FMOVD   log10rodataL19<>+112(SB), F4
	FMOVD   log10rodataL19<>+104(SB), F6
	WFMADB  V0, V6, V4, V6

	MOVH	$0x0, R1
	SUBW	R5, R3
	FMOVD	$0, F10
	RISBGZ	$32, $47, $0, R3, R4
	RISBGZ	$57, $60, $51, R3, R3
	BYTE	$0x18	//lr	%r2,%r4
	BYTE	$0x24
	RISBGN	$0, $31, $32, R4, R1
	SUBW	$0x100000, R2
	SRAW	$8, R2, R2
	ORW	$0x45000000, R2
L5:
	LDGR	R1, F0
	FMOVD	104(R9), F2
	FMADD	F8, F0, F2
	FMOVD	96(R9), F4
	WFMADB	V10, V0, V2, V0
	FMOVD	88(R9), F6
	FMOVD	80(R9), F2
	WFMADB	V0, V6, V4, V6
	FMOVD	72(R9), F1

	BYTE	$0x00
	BYTE	$0x90
	SUBW	$0x1A0000, R5
	SLD	$3, R0, R3
	MOVD	$·powtm<>+0(SB), R4
	MOVH	$0x0, R8
	ANDW	$0x7FF00000, R2
	ORW	R5, R1
	WORD	$0x5A234000	//a	%r2,0(%r3,%r4)
	MOVD	$0x3FF0000000000000, R5
	RISBGZ	$40, $63, $56, R2, R3
	RISBGN	$0, $31, $32, R2, R8
	ORW	$0x45000000, R3
	MOVW	R1, R6
	CMPBLT	R6, $0, L42
	FMOVD	F0, F4
L2:
	VLVGF	$0, R3, V1
	MOVD	$·pow_xa<>+0(SB), R2

	SRAD	$32, R5
	MOVH	$0x0, R2
	SUBW	R5, R3
	FMOVD	$0, F8
	RISBGZ	$32, $47, $0, R3, R4
	BYTE	$0x18	//lr	%r1,%r4
	BYTE	$0x14
	RISBGN	$0, $31, $32, R4, R2
	SUBW	$0x100000, R1
	SRAW	$8, R1, R1
	ORW	$0x45000000, R1
	BR	L6
L2:
	MOVD	$0x30000000, R2
	CMPW	R1, R2
	BGT	L16
	FMOVD	200(R9), F2
	FMADD	F2, F0, F0
L1:
	FMOVD	F0, ret+8(FP)
	RET
L14:
	LTDBR	F0, F0
	BLTU	L17

	FMOVD   48(R5), F5
	WFMADB  V4, V5, V3, V4
	FMOVD   40(R5), F3
	FMADD   F1, F6, F4
	FMOVD   32(R5), F1
	FMADD   F3, F2, F1
	ANDW    $0xFFFFFF00, R1
	WFMADB  V6, V4, V1, V6
	FMOVD   24(R5), F3
	ORW $0x45000000, R1
	WFMADB  V2, V6, V3, V6
	VLVGF   $0, R1, V4
	LDEBR   F4, F4
	RISBGZ  $57, $60, $51, R2, R2
	MOVD    $·atanhtab2076<>+0(SB), R1
	FMOVD   16(R5), F3
	WORD    $0x68521000 //ld    %f5,0(%r2,%r1)

	"MOVW",
	"MOVWZ",
	"MOVD",
	"MOVDBR",
	"MOVDEQ",
	"MOVDGE",
	"MOVDGT",
	"MOVDLE",
	"MOVDLT",
	"MOVDNE",
	"LOCR",
	"LOCGR",
	"FLOGR",
	"POPCNT",
	"AND",
	"ANDW",
	"OR",
	"ORW",
	"XOR",
	"XORW",
	"SLW",
	"SLD",
	"SRW",
	"SRAW",
	"SRD",
	"SRAD",
	"RLL",
	"RLLG",
	"RNSBG",
	"RXSBG",
	"ROSBG",
	"RNSBGT",
	"RXSBGT",
	"ROSBGT",
	"RISBG",
	"RISBGN",

	MOVBZ	val+8(FP), R4
	// We don't have atomic operations that work on individual bytes so we
	// need to align addr down to a word boundary and create a mask
	// containing v to AND with the entire word atomically.
	ORW	$~0xff, R4           // R4 = uint32(v) | 0xffffff00
	MOVD	$(3<<3), R5
	RXSBG	$59, $60, $3, R3, R5 // R5 = 24 - ((addr % 4) * 8) = ((addr & 3) << 3) ^ (3 << 3)
	ANDW	$~3, R3              // R3 = floor(addr, 4) = addr &^ 3

		{name: "OR", argLength: 2, reg: gp21, asm: "OR", commutative: true, clobberFlags: true},                                                                    // arg0 | arg1
		{name: "ORW", argLength: 2, reg: gp21, asm: "ORW", commutative: true, clobberFlags: true},                                                                  // arg0 | arg1