[firearm.git] / Execute.v

module Execute(
	input clk,
	input Nrst,	/* XXX not used yet */
	
	input stall_2a,
	input flush_2a,
	
	input bubble_2a,
	input [31:0] pc_2a,
	input [31:0] insn_2a,
	input [31:0] cpsr_2a,
	input [31:0] spsr_2a,
	input [31:0] op0_2a,
	input [31:0] op1_2a,
	input [31:0] op2_2a,
	input carry_2a,
	
	output reg outstall_2a = 0,
	output reg bubble_3a = 1,
	output reg [31:0] cpsr_3a = 0,
	output reg [31:0] spsr_3a = 0,
	output reg cpsrup_3a = 0,
	output reg write_reg_3a = 1'bx,
	output reg [3:0] write_num_3a = 4'bxxxx,
	output reg [31:0] write_data_3a = 32'hxxxxxxxx,
	output reg [31:0] jmppc_2a,
	output reg jmp_2a,
	output reg [31:0] pc_3a,
	output reg [31:0] insn_3a,
	output reg [31:0] op0_3a, op1_3a, op2_3a
	);
	
	reg mult_start;
	reg [31:0] mult_acc0, mult_in0, mult_in1;
	wire mult_done;
	wire [31:0] mult_result;
	
	reg [31:0] alu_in0_2a, alu_in1_2a;
	reg [3:0] alu_op_2a;
	reg alu_setflags_2a;
	wire [31:0] alu_result_2a, alu_outcpsr_2a;
	wire alu_setres_2a;
	
	reg next_bubble_3a;
	reg [31:0] next_cpsr_3a, next_spsr_3a;
	reg next_cpsrup_3a;
	
	reg next_write_reg_3a;
	reg [3:0] next_write_num_3a;
	reg [31:0] next_write_data_3a;

	Multiplier multiplier(
		.clk(clk), .Nrst(Nrst),
		.start(mult_start), .acc0(mult_acc0), .in0(mult_in0),
		.in1(mult_in1), .done(mult_done), .result(mult_result));
	
	ALU alu(
		.clk(clk), .Nrst(Nrst),
		.in0(alu_in0_2a), .in1(alu_in1_2a), .cpsr(cpsr_2a), .op(alu_op_2a),
		.setflags(alu_setflags_2a), .shifter_carry(carry_2a),
		.result(alu_result_2a), .cpsr_out(alu_outcpsr_2a), .setres(alu_setres_2a));

	always @(posedge clk)
	begin
		if (!stall_2a)
		begin
			bubble_3a <= next_bubble_3a;
			cpsr_3a <= next_cpsr_3a;
			spsr_3a <= next_spsr_3a;
			cpsrup_3a <= next_cpsrup_3a;
			write_reg_3a <= next_write_reg_3a;
			write_num_3a <= next_write_num_3a;
			write_data_3a <= next_write_data_3a;
			pc_3a <= pc_2a;
			insn_3a <= insn_2a;
			op0_3a <= op0_2a;
			op1_3a <= op1_2a;
			op2_3a <= op2_2a;
		end
	end
	
	reg delayedflush_2a = 0;
	always @(posedge clk)
		if (flush_2a && outstall_2a /* halp! I can't do it now, maybe later? */)
			delayedflush_2a <= 1;
		else if (!outstall_2a /* anything has been handled this time around */)
			delayedflush_2a <= 0;

	reg outstall_3a = 0;
	always @(posedge clk)
		outstall_3a <= outstall_2a;
	
	always @(*)
	begin
		outstall_2a = stall_2a;
		
		casez (insn_2a)
		`DECODE_ALU_MULT:	/* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
			outstall_2a = outstall_2a | ((!outstall_3a | !mult_done) && !bubble_2a);
		endcase
	end
	
	/* ALU inputs */
	always @(*)
	begin
		alu_in0_2a = op0_2a;
		alu_in1_2a = op1_2a;
		alu_op_2a = insn_2a[24:21];
		alu_setflags_2a = insn_2a[20] /* S */;
	end
	
	/* Register outputs */
	always @(*)
	begin
		next_cpsr_3a = cpsr_2a;
		next_spsr_3a = spsr_2a;
		next_cpsrup_3a = 0;
		next_write_reg_3a = 0;
		next_write_num_3a = 4'hx;
		next_write_data_3a = 32'hxxxxxxxx;
		
		casez(insn_2a)
		`DECODE_ALU_MULT:	/* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
		begin
			next_cpsr_3a = insn_2a[20] /* S */ ? {mult_result[31] /* N */, mult_result == 0 /* Z */, 1'b0 /* C */, cpsr_2a[28] /* V */, cpsr_2a[27:0]} : cpsr_2a;
			next_cpsrup_3a = insn_2a[20] /* S */;
			next_write_reg_3a = 1;
			next_write_num_3a = insn_2a[19:16] /* Rd -- why the fuck isn't this the same place as ALU */;
			next_write_data_3a = mult_result;
		end
		`DECODE_ALU_MRS:	/* MRS (Transfer PSR to register) */
		begin
			next_write_reg_3a = 1;
			next_write_num_3a = insn_2a[15:12];
			if (insn_2a[22] /* Ps */)
				next_write_data_3a = spsr_2a;
			else
				next_write_data_3a = cpsr_2a;
		end
		`DECODE_ALU_MSR,	/* MSR (Transfer register to PSR) */
		`DECODE_ALU_MSR_FLAGS:	/* MSR (Transfer register or immediate to PSR, flag bits only) */
		begin
			if ((cpsr_2a[4:0] == `MODE_USR) || (insn_2a[16] /* that random bit */ == 1'b0))	/* flags only */
			begin
				if (insn_2a[22] /* Ps */)
					next_spsr_3a = {op0_2a[31:29], spsr_2a[28:0]};
				else
					next_cpsr_3a = {op0_2a[31:29], cpsr_2a[28:0]};
			end else begin
				if (insn_2a[22] /* Ps */)
					next_spsr_3a = op0_2a;
				else
					next_cpsr_3a = op0_2a;
			end
			next_cpsrup_3a = 1;
		end
		`DECODE_ALU_SWP,	/* Atomic swap */
		`DECODE_ALU_BX,		/* Branch */
		`DECODE_ALU_HDATA_REG,	/* Halfword transfer - register offset */
		`DECODE_ALU_HDATA_IMM:	/* Halfword transfer - immediate offset */
		begin end
		`DECODE_ALU:		/* ALU */
		begin
			if (alu_setres_2a) begin
				next_write_reg_3a = 1;
				next_write_num_3a = insn_2a[15:12] /* Rd */;
				next_write_data_3a = alu_result_2a;
			end
			
			if (insn_2a[20] /* S */) begin
				next_cpsrup_3a = 1;
				next_cpsr_3a = ((insn_2a[15:12] == 4'b1111) && insn_2a[20]) ? spsr_2a : alu_outcpsr_2a;
			end
		end
		`DECODE_LDRSTR_UNDEFINED,	/* Undefined. I hate ARM */
		`DECODE_LDRSTR,		/* Single data transfer */
		`DECODE_LDMSTM:		/* Block data transfer */
		begin end
		`DECODE_BRANCH:		/* Branch */
		begin
			if(insn_2a[24] /* L */) begin
				next_write_reg_3a = 1;
				next_write_num_3a = 4'hE; /* link register */
				next_write_data_3a = pc_2a + 32'h4;
			end
		end
		endcase
	end
	
	/* Multiplier inputs */
	always @(*)
	begin
		mult_start = 0;
		mult_acc0 = 32'hxxxxxxxx;
		mult_in0 = 32'hxxxxxxxx;
		mult_in1 = 32'hxxxxxxxx;
		
		casez(insn_2a)
		`DECODE_ALU_MULT:
		begin
			if (!outstall_3a /* i.e., this is a new one */ && !bubble_2a /* i.e., this is a real one */)
			begin
				mult_start = 1;
				mult_acc0 = insn_2a[21] /* A */ ? op0_2a /* Rn */ : 32'h0;
				mult_in0 = op1_2a /* Rm */;
				mult_in1 = op2_2a /* Rs */;
				$display("New MUL instruction");
			end
		end
		endcase
	end

	/* Miscellaneous cleanup. */
	always @(*)
	begin
		next_bubble_3a = bubble_2a | flush_2a | delayedflush_2a;

		jmp_2a = 1'b0;
		jmppc_2a = 32'h00000000;

		casez (insn_2a)
		`DECODE_ALU_MULT:	/* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
			next_bubble_3a = next_bubble_3a | !mult_done | !outstall_3a;
		`DECODE_ALU_MRS,	/* MRS (Transfer PSR to register) */
		`DECODE_ALU_MSR,	/* MSR (Transfer register to PSR) */
		`DECODE_ALU_MSR_FLAGS,	/* MSR (Transfer register or immediate to PSR, flag bits only) */
		`DECODE_ALU_SWP,	/* Atomic swap */
		`DECODE_ALU_BX,		/* Branch */
		`DECODE_ALU_HDATA_REG,	/* Halfword transfer - register offset */
		`DECODE_ALU_HDATA_IMM,	/* Halfword transfer - immediate offset */
		`DECODE_ALU,		/* ALU */
		`DECODE_LDRSTR_UNDEFINED,	/* Undefined. I hate ARM */
		`DECODE_LDRSTR,		/* Single data transfer */
		`DECODE_LDMSTM:		/* Block data transfer */
		begin end
		`DECODE_BRANCH:
		begin
			if(!bubble_2a && !flush_2a && !delayedflush_2a && !outstall_2a /* Let someone else take precedence. */) begin
				jmppc_2a = pc_2a + op0_2a + 32'h8;
				jmp_2a = 1'b1;
			end
		end                     /* Branch */
		`DECODE_LDCSTC,		/* Coprocessor data transfer */
		`DECODE_CDP,		/* Coprocessor data op */
		`DECODE_MRCMCR,		/* Coprocessor register transfer */
		`DECODE_SWI:		/* SWI */
		begin end
		default:		/* X everything else out */
		begin end
		endcase
	end
endmodule

module Multiplier(
	input clk,
	input Nrst,	/* XXX not used yet */
	
	input start,
	input [31:0] acc0,
	input [31:0] in0,
	input [31:0] in1,
	
	output reg done = 0,
	output reg [31:0] result);
	
	reg [31:0] bitfield;
	reg [31:0] multiplicand;
	reg [31:0] acc;
	
	always @(posedge clk)
	begin
		if (start) begin
			bitfield <= in0;
			multiplicand <= in1;
			acc <= acc0;
			done <= 0;
		end else begin
			bitfield <= {2'b00, bitfield[31:2]};
			multiplicand <= {multiplicand[29:0], 2'b00};
			acc <= acc +
				(bitfield[0] ? multiplicand : 0) +
				(bitfield[1] ? {multiplicand[30:0], 1'b0} : 0);
			if (bitfield == 0) begin
				result <= acc;
				done <= 1;
			end
		end
	end
endmodule

module ALU(
	input clk,
	input Nrst,	/* XXX not used yet */

	input [31:0] in0,
	input [31:0] in1,
	input [31:0] cpsr,
	input [3:0] op,
	input setflags,
	input shifter_carry,

	output reg [31:0] result,
	output reg [31:0] cpsr_out,
	output reg setres
);
	reg [31:0] res;
	reg flag_n, flag_z, flag_c, flag_v;
	wire [32:0] sum, diff, rdiff;
	wire sum_v, diff_v, rdiff_v;

	assign sum = {1'b0, in0} + {1'b0, in1};
	assign diff = {1'b0, in0} - {1'b0, in1};
	assign rdiff = {1'b0, in1} - {1'b0, in0};
	assign sum_v = (in0[31] ^~ in1[31]) & (sum[31] ^ in0[31]);
	assign diff_v = (in0[31] ^ in1[31]) & (diff[31] ^ in0[31]);
	assign rdiff_v = (in0[31] ^ in1[31]) & (rdiff[31] ^ in1[31]);

	always @(*) begin
		res = 32'hxxxxxxxx;
		setres = 1'bx;
		flag_c = cpsr[`CPSR_C];
		flag_v = cpsr[`CPSR_V];
		case(op)
		`ALU_AND: begin
			result = in0 & in1;
			flag_c = shifter_carry;
			setres = 1'b1;
		end
		`ALU_EOR: begin
			result = in0 ^ in1;
			flag_c = shifter_carry;
			setres = 1'b1;
		end
		`ALU_SUB: begin
			{flag_c, result} = diff;
			flag_c = !flag_c;
			flag_v = diff_v;
			setres = 1'b1;
		end
		`ALU_RSB: begin
			{flag_c, result} = rdiff;
			flag_c = !flag_c;
			flag_v = rdiff_v;
			setres = 1'b1;
		end
		`ALU_ADD: begin
			{flag_c, result} = sum;
			flag_v = sum_v;
			setres = 1'b1;
		end
		`ALU_ADC: begin
			{flag_c, result} = sum + {32'b0, cpsr[`CPSR_C]};
			flag_v = sum_v | (~sum[31] & result[31]);
			setres = 1'b1;
		end
		`ALU_SBC: begin
			{flag_c, result} = diff - {32'b0, (~cpsr[`CPSR_C])};
			flag_c = !flag_c;
			flag_v = diff_v | (diff[31] & ~result[31]);
			setres = 1'b1;
		end
		`ALU_RSC: begin
			{flag_c, result} = rdiff - {32'b0, (~cpsr[`CPSR_C])};
			flag_c = !flag_c;
			flag_v = rdiff_v | (rdiff[31] & ~result[31]);
			setres = 1'b1;
		end
		`ALU_TST: begin
			result = in0 & in1;
			flag_c = shifter_carry;
			setres = 1'b0;
		end
		`ALU_TEQ: begin
			result = in0 ^ in1;
			flag_c = shifter_carry;
			setres = 1'b0;
		end
		`ALU_CMP: begin
			{flag_c, result} = diff;
			flag_c = !flag_c;
			flag_v = diff_v;
			setres = 1'b0;
		end
		`ALU_CMN: begin
			{flag_c, result} = sum;
			flag_v = sum_v;
			setres = 1'b0;
		end
		`ALU_ORR: begin
			result = in0 | in1;
			flag_c = shifter_carry;
			setres = 1'b1;
		end
		`ALU_MOV: begin
			result = in1;
			flag_c = shifter_carry;
			setres = 1'b1;
		end
		`ALU_BIC: begin
			result = in0 & (~in1);
			flag_c = shifter_carry;
			setres = 1'b1;
		end
		`ALU_MVN: begin
			result = ~in1;
			flag_c = shifter_carry;
			setres = 1'b1;
		end
		endcase
		
		flag_z = (result == 0);
		flag_n = result[31];
		
		cpsr_out = setflags ? {flag_n, flag_z, flag_c, flag_v, cpsr[27:0]} : cpsr;
	end
endmodule
Commit	Line	Data
5b3daee2 JW	1	module Execute(
	2	input clk,
	3	input Nrst, /* XXX not used yet */
	4
8c1ba494	5	input stall_2a,
131123bc	6	input flush_2a,
5b3daee2	7
8c1ba494 JW	8	input bubble_2a,
	9	input [31:0] pc_2a,
	10	input [31:0] insn_2a,
	11	input [31:0] cpsr_2a,
	12	input [31:0] spsr_2a,
	13	input [31:0] op0_2a,
	14	input [31:0] op1_2a,
	15	input [31:0] op2_2a,
	16	input carry_2a,
5b3daee2	17
8c1ba494 JW	18	output reg outstall_2a = 0,
8c1ba494 JW	19	output reg bubble_3a = 1,
131123bc JW	20	output reg [31:0] cpsr_3a = 0,
	21	output reg [31:0] spsr_3a = 0,
	22	output reg cpsrup_3a = 0,
8c1ba494 JW	23	output reg write_reg_3a = 1'bx,
	24	output reg [3:0] write_num_3a = 4'bxxxx,
	25	output reg [31:0] write_data_3a = 32'hxxxxxxxx,
	26	output reg [31:0] jmppc_2a,
	27	output reg jmp_2a,
	28	output reg [31:0] pc_3a,
	29	output reg [31:0] insn_3a,
	30	output reg [31:0] op0_3a, op1_3a, op2_3a
5b3daee2	31	);
5b3daee2	32
bc572c5f JW	33	reg mult_start;
	34	reg [31:0] mult_acc0, mult_in0, mult_in1;
	35	wire mult_done;
	36	wire [31:0] mult_result;
	37
8c1ba494 JW	38	reg [31:0] alu_in0_2a, alu_in1_2a;
	39	reg [3:0] alu_op_2a;
	40	reg alu_setflags_2a;
	41	wire [31:0] alu_result_2a, alu_outcpsr_2a;
	42	wire alu_setres_2a;
732b7730	43
8c1ba494	44	reg next_bubble_3a;
131123bc JW	45	reg [31:0] next_cpsr_3a, next_spsr_3a;
131123bc JW	46	reg next_cpsrup_3a;
8c1ba494 JW	47
	48	reg next_write_reg_3a;
	49	reg [3:0] next_write_num_3a;
	50	reg [31:0] next_write_data_3a;
149bcd1a	51
bc572c5f JW	52	Multiplier multiplier(
	53	.clk(clk), .Nrst(Nrst),
	54	.start(mult_start), .acc0(mult_acc0), .in0(mult_in0),
	55	.in1(mult_in1), .done(mult_done), .result(mult_result));
6e3dfd79 JW	56
	57	ALU alu(
	58	.clk(clk), .Nrst(Nrst),
8c1ba494 JW	59	.in0(alu_in0_2a), .in1(alu_in1_2a), .cpsr(cpsr_2a), .op(alu_op_2a),
	60	.setflags(alu_setflags_2a), .shifter_carry(carry_2a),
	61	.result(alu_result_2a), .cpsr_out(alu_outcpsr_2a), .setres(alu_setres_2a));
b770ec9a	62
732b7730 JW	63	always @(posedge clk)
732b7730 JW	64	begin
8c1ba494	65	if (!stall_2a)
732b7730	66	begin
8c1ba494	67	bubble_3a <= next_bubble_3a;
131123bc JW	68	cpsr_3a <= next_cpsr_3a;
	69	spsr_3a <= next_spsr_3a;
	70	cpsrup_3a <= next_cpsrup_3a;
8c1ba494 JW	71	write_reg_3a <= next_write_reg_3a;
	72	write_num_3a <= next_write_num_3a;
	73	write_data_3a <= next_write_data_3a;
	74	pc_3a <= pc_2a;
	75	insn_3a <= insn_2a;
	76	op0_3a <= op0_2a;
	77	op1_3a <= op1_2a;
	78	op2_3a <= op2_2a;
732b7730 JW	79	end
732b7730 JW	80	end
b770ec9a	81
131123bc	82	reg delayedflush_2a = 0;
b770ec9a	83	always @(posedge clk)
131123bc JW	84	if (flush_2a && outstall_2a /* halp! I can't do it now, maybe later? */)
131123bc JW	85	delayedflush_2a <= 1;
8c1ba494	86	else if (!outstall_2a /* anything has been handled this time around */)
131123bc	87	delayedflush_2a <= 0;
2b091cd4	88
8c1ba494	89	reg outstall_3a = 0;
af096d96	90	always @(posedge clk)
8c1ba494	91	outstall_3a <= outstall_2a;
3ccca009	92
2b091cd4	93	always @(*)
732b7730	94	begin
8c1ba494	95	outstall_2a = stall_2a;
3ccca009	96
8c1ba494	97	casez (insn_2a)
3ccca009	98	`DECODE_ALU_MULT: /* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
8c1ba494	99	outstall_2a = outstall_2a \| ((!outstall_3a \| !mult_done) && !bubble_2a);
3ccca009 JW	100	endcase
	101	end
	102
	103	/* ALU inputs */
	104	always @(*)
	105	begin
8c1ba494 JW	106	alu_in0_2a = op0_2a;
	107	alu_in1_2a = op1_2a;
	108	alu_op_2a = insn_2a[24:21];
	109	alu_setflags_2a = insn_2a[20] /* S */;
3ccca009 JW	110	end
	111
	112	/* Register outputs */
	113	always @(*)
	114	begin
131123bc JW	115	next_cpsr_3a = cpsr_2a;
	116	next_spsr_3a = spsr_2a;
	117	next_cpsrup_3a = 0;
8c1ba494 JW	118	next_write_reg_3a = 0;
	119	next_write_num_3a = 4'hx;
	120	next_write_data_3a = 32'hxxxxxxxx;
3ccca009	121
8c1ba494	122	casez(insn_2a)
af096d96 JW	123	`DECODE_ALU_MULT: /* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
af096d96 JW	124	begin
131123bc JW	125	next_cpsr_3a = insn_2a[20] /* S / ? {mult_result[31] / N /, mult_result == 0 / Z /, 1'b0 / C /, cpsr_2a[28] / V */, cpsr_2a[27:0]} : cpsr_2a;
131123bc JW	126	next_cpsrup_3a = insn_2a[20] /* S */;
8c1ba494 JW	127	next_write_reg_3a = 1;
	128	next_write_num_3a = insn_2a[19:16] /* Rd -- why the fuck isn't this the same place as ALU */;
	129	next_write_data_3a = mult_result;
af096d96	130	end
cb0428b6 JW	131	`DECODE_ALU_MRS: /* MRS (Transfer PSR to register) */
cb0428b6 JW	132	begin
8c1ba494 JW	133	next_write_reg_3a = 1;
	134	next_write_num_3a = insn_2a[15:12];
	135	if (insn_2a[22] /* Ps */)
	136	next_write_data_3a = spsr_2a;
cb0428b6	137	else
8c1ba494	138	next_write_data_3a = cpsr_2a;
cb0428b6	139	end
2b091cd4	140	`DECODE_ALU_MSR, /* MSR (Transfer register to PSR) */
cb0428b6	141	`DECODE_ALU_MSR_FLAGS: /* MSR (Transfer register or immediate to PSR, flag bits only) */
fdecc897	142	begin
8c1ba494	143	if ((cpsr_2a[4:0] == `MODE_USR) \|\| (insn_2a[16] /* that random bit / == 1'b0)) / flags only */
cb0428b6	144	begin
8c1ba494	145	if (insn_2a[22] /* Ps */)
131123bc	146	next_spsr_3a = {op0_2a[31:29], spsr_2a[28:0]};
cb0428b6	147	else
131123bc	148	next_cpsr_3a = {op0_2a[31:29], cpsr_2a[28:0]};
cb0428b6	149	end else begin
8c1ba494	150	if (insn_2a[22] /* Ps */)
131123bc	151	next_spsr_3a = op0_2a;
cb0428b6	152	else
131123bc	153	next_cpsr_3a = op0_2a;
cb0428b6	154	end
131123bc	155	next_cpsrup_3a = 1;
fdecc897	156	end
2b091cd4 JW	157	`DECODE_ALU_SWP, /* Atomic swap */
	158	`DECODE_ALU_BX, /* Branch */
	159	`DECODE_ALU_HDATA_REG, /* Halfword transfer - register offset */
732b7730 JW	160	`DECODE_ALU_HDATA_IMM: /* Halfword transfer - immediate offset */
	161	begin end
	162	`DECODE_ALU: /* ALU */
	163	begin
8c1ba494 JW	164	if (alu_setres_2a) begin
	165	next_write_reg_3a = 1;
	166	next_write_num_3a = insn_2a[15:12] /* Rd */;
	167	next_write_data_3a = alu_result_2a;
732b7730 JW	168	end
732b7730 JW	169
8c1ba494	170	if (insn_2a[20] /* S */) begin
131123bc JW	171	next_cpsrup_3a = 1;
131123bc JW	172	next_cpsr_3a = ((insn_2a[15:12] == 4'b1111) && insn_2a[20]) ? spsr_2a : alu_outcpsr_2a;
3ccca009 JW	173	end
	174	end
	175	`DECODE_LDRSTR_UNDEFINED, /* Undefined. I hate ARM */
	176	`DECODE_LDRSTR, /* Single data transfer */
	177	`DECODE_LDMSTM: /* Block data transfer */
	178	begin end
	179	`DECODE_BRANCH: /* Branch */
	180	begin
8c1ba494 JW	181	if(insn_2a[24] /* L */) begin
	182	next_write_reg_3a = 1;
	183	next_write_num_3a = 4'hE; /* link register */
	184	next_write_data_3a = pc_2a + 32'h4;
3ccca009	185	end
732b7730	186	end
3ccca009 JW	187	endcase
	188	end
	189
	190	/* Multiplier inputs */
	191	always @(*)
	192	begin
	193	mult_start = 0;
	194	mult_acc0 = 32'hxxxxxxxx;
	195	mult_in0 = 32'hxxxxxxxx;
	196	mult_in1 = 32'hxxxxxxxx;
	197
8c1ba494	198	casez(insn_2a)
3ccca009 JW	199	`DECODE_ALU_MULT:
3ccca009 JW	200	begin
8c1ba494	201	if (!outstall_3a /* i.e., this is a new one / && !bubble_2a / i.e., this is a real one */)
3ccca009 JW	202	begin
3ccca009 JW	203	mult_start = 1;
8c1ba494 JW	204	mult_acc0 = insn_2a[21] /* A / ? op0_2a / Rn */ : 32'h0;
	205	mult_in0 = op1_2a /* Rm */;
	206	mult_in1 = op2_2a /* Rs */;
3ccca009 JW	207	$display("New MUL instruction");
	208	end
	209	end
	210	endcase
	211	end
	212
	213	/* Miscellaneous cleanup. */
	214	always @(*)
	215	begin
131123bc	216	next_bubble_3a = bubble_2a \| flush_2a \| delayedflush_2a;
3ccca009	217
8c1ba494 JW	218	jmp_2a = 1'b0;
8c1ba494 JW	219	jmppc_2a = 32'h00000000;
3ccca009	220
8c1ba494	221	casez (insn_2a)
3ccca009	222	`DECODE_ALU_MULT: /* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
8c1ba494	223	next_bubble_3a = next_bubble_3a \| !mult_done \| !outstall_3a;
3ccca009 JW	224	`DECODE_ALU_MRS, /* MRS (Transfer PSR to register) */
	225	`DECODE_ALU_MSR, /* MSR (Transfer register to PSR) */
	226	`DECODE_ALU_MSR_FLAGS, /* MSR (Transfer register or immediate to PSR, flag bits only) */
	227	`DECODE_ALU_SWP, /* Atomic swap */
	228	`DECODE_ALU_BX, /* Branch */
	229	`DECODE_ALU_HDATA_REG, /* Halfword transfer - register offset */
	230	`DECODE_ALU_HDATA_IMM, /* Halfword transfer - immediate offset */
	231	`DECODE_ALU, /* ALU */
2b091cd4 JW	232	`DECODE_LDRSTR_UNDEFINED, /* Undefined. I hate ARM */
2b091cd4 JW	233	`DECODE_LDRSTR, /* Single data transfer */
314dac21 CL	234	`DECODE_LDMSTM: /* Block data transfer */
	235	begin end
	236	`DECODE_BRANCH:
	237	begin
131123bc	238	if(!bubble_2a && !flush_2a && !delayedflush_2a && !outstall_2a /* Let someone else take precedence. */) begin
8c1ba494 JW	239	jmppc_2a = pc_2a + op0_2a + 32'h8;
8c1ba494 JW	240	jmp_2a = 1'b1;
314dac21 CL	241	end
314dac21 CL	242	end /* Branch */
2b091cd4 JW	243	`DECODE_LDCSTC, /* Coprocessor data transfer */
	244	`DECODE_CDP, /* Coprocessor data op */
	245	`DECODE_MRCMCR, /* Coprocessor register transfer */
	246	`DECODE_SWI: /* SWI */
	247	begin end
	248	default: /* X everything else out */
	249	begin end
	250	endcase
732b7730	251	end
5b3daee2	252	endmodule
07fbfa80 JW	253
	254	module Multiplier(
	255	input clk,
	256	input Nrst, /* XXX not used yet */
	257
	258	input start,
	259	input [31:0] acc0,
	260	input [31:0] in0,
	261	input [31:0] in1,
	262
	263	output reg done = 0,
	264	output reg [31:0] result);
	265
	266	reg [31:0] bitfield;
	267	reg [31:0] multiplicand;
	268	reg [31:0] acc;
	269
	270	always @(posedge clk)
	271	begin
	272	if (start) begin
	273	bitfield <= in0;
	274	multiplicand <= in1;
	275	acc <= acc0;
	276	done <= 0;
	277	end else begin
	278	bitfield <= {2'b00, bitfield[31:2]};
	279	multiplicand <= {multiplicand[29:0], 2'b00};
	280	acc <= acc +
	281	(bitfield[0] ? multiplicand : 0) +
	282	(bitfield[1] ? {multiplicand[30:0], 1'b0} : 0);
	283	if (bitfield == 0) begin
	284	result <= acc;
	285	done <= 1;
	286	end
	287	end
	288	end
	289	endmodule
879a3986	290
879a3986 CL	291	module ALU(
	292	input clk,
	293	input Nrst, /* XXX not used yet */
	294
	295	input [31:0] in0,
	296	input [31:0] in1,
	297	input [31:0] cpsr,
	298	input [3:0] op,
	299	input setflags,
	300	input shifter_carry,
	301
	302	output reg [31:0] result,
	303	output reg [31:0] cpsr_out,
732b7730	304	output reg setres
879a3986	305	);
7947b9c7 JW	306	reg [31:0] res;
7947b9c7 JW	307	reg flag_n, flag_z, flag_c, flag_v;
879a3986	308	wire [32:0] sum, diff, rdiff;
793482e9	309	wire sum_v, diff_v, rdiff_v;
879a3986 CL	310
	311	assign sum = {1'b0, in0} + {1'b0, in1};
	312	assign diff = {1'b0, in0} - {1'b0, in1};
a4c270c7	313	assign rdiff = {1'b0, in1} - {1'b0, in0};
793482e9 CL	314	assign sum_v = (in0[31] ^~ in1[31]) & (sum[31] ^ in0[31]);
	315	assign diff_v = (in0[31] ^ in1[31]) & (diff[31] ^ in0[31]);
	316	assign rdiff_v = (in0[31] ^ in1[31]) & (rdiff[31] ^ in1[31]);
879a3986	317
879a3986 CL	318	always @(*) begin
	319	res = 32'hxxxxxxxx;
	320	setres = 1'bx;
	321	flag_c = cpsr[`CPSR_C];
	322	flag_v = cpsr[`CPSR_V];
	323	case(op)
	324	`ALU_AND: begin
732b7730	325	result = in0 & in1;
879a3986 CL	326	flag_c = shifter_carry;
	327	setres = 1'b1;
	328	end
	329	`ALU_EOR: begin
732b7730	330	result = in0 ^ in1;
879a3986 CL	331	flag_c = shifter_carry;
	332	setres = 1'b1;
	333	end
	334	`ALU_SUB: begin
732b7730	335	{flag_c, result} = diff;
1bc060d9	336	flag_c = !flag_c;
793482e9	337	flag_v = diff_v;
879a3986 CL	338	setres = 1'b1;
	339	end
	340	`ALU_RSB: begin
732b7730	341	{flag_c, result} = rdiff;
1bc060d9	342	flag_c = !flag_c;
793482e9	343	flag_v = rdiff_v;
879a3986 CL	344	setres = 1'b1;
	345	end
	346	`ALU_ADD: begin
732b7730	347	{flag_c, result} = sum;
793482e9	348	flag_v = sum_v;
879a3986 CL	349	setres = 1'b1;
	350	end
	351	`ALU_ADC: begin
732b7730	352	{flag_c, result} = sum + {32'b0, cpsr[`CPSR_C]};
793482e9	353	flag_v = sum_v \| (~sum[31] & result[31]);
879a3986 CL	354	setres = 1'b1;
	355	end
	356	`ALU_SBC: begin
732b7730	357	{flag_c, result} = diff - {32'b0, (~cpsr[`CPSR_C])};
1bc060d9	358	flag_c = !flag_c;
793482e9	359	flag_v = diff_v \| (diff[31] & ~result[31]);
879a3986 CL	360	setres = 1'b1;
	361	end
	362	`ALU_RSC: begin
732b7730	363	{flag_c, result} = rdiff - {32'b0, (~cpsr[`CPSR_C])};
1bc060d9	364	flag_c = !flag_c;
793482e9	365	flag_v = rdiff_v \| (rdiff[31] & ~result[31]);
879a3986 CL	366	setres = 1'b1;
	367	end
	368	`ALU_TST: begin
732b7730	369	result = in0 & in1;
879a3986 CL	370	flag_c = shifter_carry;
	371	setres = 1'b0;
	372	end
	373	`ALU_TEQ: begin
732b7730	374	result = in0 ^ in1;
879a3986 CL	375	flag_c = shifter_carry;
	376	setres = 1'b0;
	377	end
	378	`ALU_CMP: begin
732b7730	379	{flag_c, result} = diff;
1bc060d9	380	flag_c = !flag_c;
793482e9	381	flag_v = diff_v;
879a3986 CL	382	setres = 1'b0;
	383	end
	384	`ALU_CMN: begin
732b7730	385	{flag_c, result} = sum;
793482e9	386	flag_v = sum_v;
879a3986 CL	387	setres = 1'b0;
	388	end
	389	`ALU_ORR: begin
732b7730	390	result = in0 \| in1;
879a3986 CL	391	flag_c = shifter_carry;
	392	setres = 1'b1;
	393	end
	394	`ALU_MOV: begin
732b7730	395	result = in1;
879a3986 CL	396	flag_c = shifter_carry;
	397	setres = 1'b1;
	398	end
	399	`ALU_BIC: begin
732b7730	400	result = in0 & (~in1);
879a3986 CL	401	flag_c = shifter_carry;
	402	setres = 1'b1;
	403	end
	404	`ALU_MVN: begin
732b7730	405	result = ~in1;
879a3986 CL	406	flag_c = shifter_carry;
	407	setres = 1'b1;
	408	end
	409	endcase
732b7730 JW	410
	411	flag_z = (result == 0);
	412	flag_n = result[31];
	413
	414	cpsr_out = setflags ? {flag_n, flag_z, flag_c, flag_v, cpsr[27:0]} : cpsr;
879a3986	415	end
879a3986	416	endmodule