Add chip enable correctness for CellularRAM.
[firearm.git] / Execute.v
1 module Execute(
2         input clk,
3         input Nrst,     /* XXX not used yet */
4         
5         input stall_2a,
6         input flush_2a,
7         
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,
17         
18         output reg outstall_2a = 0,
19         output reg bubble_3a = 1,
20         output reg [31:0] cpsr_3a = 0,
21         output reg [31:0] spsr_3a = 0,
22         output reg cpsrup_3a = 0,
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
31         );
32         
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         
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;
43         
44         reg next_bubble_3a;
45         reg [31:0] next_cpsr_3a, next_spsr_3a;
46         reg next_cpsrup_3a;
47         
48         reg next_write_reg_3a;
49         reg [3:0] next_write_num_3a;
50         reg [31:0] next_write_data_3a;
51
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));
56         
57         ALU alu(
58                 .clk(clk), .Nrst(Nrst),
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));
62
63         always @(posedge clk)
64         begin
65                 if (!stall_2a)
66                 begin
67                         bubble_3a <= next_bubble_3a;
68                         cpsr_3a <= next_cpsr_3a;
69                         spsr_3a <= next_spsr_3a;
70                         cpsrup_3a <= next_cpsrup_3a;
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;
79                 end
80         end
81         
82         reg delayedflush_2a = 0;
83         always @(posedge clk)
84                 if (flush_2a && outstall_2a /* halp! I can't do it now, maybe later? */)
85                         delayedflush_2a <= 1;
86                 else if (!outstall_2a /* anything has been handled this time around */)
87                         delayedflush_2a <= 0;
88
89         reg outstall_3a = 0;
90         always @(posedge clk)
91                 outstall_3a <= outstall_2a;
92         
93         always @(*)
94         begin
95                 outstall_2a = stall_2a;
96                 
97                 casez (insn_2a)
98                 `DECODE_ALU_MULT:       /* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
99                         outstall_2a = outstall_2a | ((!outstall_3a | !mult_done) && !bubble_2a);
100                 endcase
101         end
102         
103         /* ALU inputs */
104         always @(*)
105         begin
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 */;
110         end
111         
112         /* Register outputs */
113         always @(*)
114         begin
115                 next_cpsr_3a = cpsr_2a;
116                 next_spsr_3a = spsr_2a;
117                 next_cpsrup_3a = 0;
118                 next_write_reg_3a = 0;
119                 next_write_num_3a = 4'hx;
120                 next_write_data_3a = 32'hxxxxxxxx;
121                 
122                 casez(insn_2a)
123                 `DECODE_ALU_MULT:       /* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
124                 begin
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;
126                         next_cpsrup_3a = insn_2a[20] /* S */;
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;
130                 end
131                 `DECODE_ALU_MRS:        /* MRS (Transfer PSR to register) */
132                 begin
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;
137                         else
138                                 next_write_data_3a = cpsr_2a;
139                 end
140                 `DECODE_ALU_MSR,        /* MSR (Transfer register to PSR) */
141                 `DECODE_ALU_MSR_FLAGS:  /* MSR (Transfer register or immediate to PSR, flag bits only) */
142                 begin
143                         if ((cpsr_2a[4:0] == `MODE_USR) || (insn_2a[16] /* that random bit */ == 1'b0)) /* flags only */
144                         begin
145                                 if (insn_2a[22] /* Ps */)
146                                         next_spsr_3a = {op0_2a[31:29], spsr_2a[28:0]};
147                                 else
148                                         next_cpsr_3a = {op0_2a[31:29], cpsr_2a[28:0]};
149                         end else begin
150                                 if (insn_2a[22] /* Ps */)
151                                         next_spsr_3a = op0_2a;
152                                 else
153                                         next_cpsr_3a = op0_2a;
154                         end
155                         next_cpsrup_3a = 1;
156                 end
157                 `DECODE_ALU_SWP,        /* Atomic swap */
158                 `DECODE_ALU_BX,         /* Branch */
159                 `DECODE_ALU_HDATA_REG,  /* Halfword transfer - register offset */
160                 `DECODE_ALU_HDATA_IMM:  /* Halfword transfer - immediate offset */
161                 begin end
162                 `DECODE_ALU:            /* ALU */
163                 begin
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;
168                         end
169                         
170                         if (insn_2a[20] /* S */) begin
171                                 next_cpsrup_3a = 1;
172                                 next_cpsr_3a = ((insn_2a[15:12] == 4'b1111) && insn_2a[20]) ? spsr_2a : alu_outcpsr_2a;
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
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;
185                         end
186                 end
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                 
198                 casez(insn_2a)
199                 `DECODE_ALU_MULT:
200                 begin
201                         if (!outstall_3a /* i.e., this is a new one */ && !bubble_2a /* i.e., this is a real one */)
202                         begin
203                                 mult_start = 1;
204                                 mult_acc0 = insn_2a[21] /* A */ ? op0_2a /* Rn */ : 32'h0;
205                                 mult_in0 = op1_2a /* Rm */;
206                                 mult_in1 = op2_2a /* Rs */;
207                                 $display("New MUL instruction");
208                         end
209                 end
210                 endcase
211         end
212
213         /* Miscellaneous cleanup. */
214         always @(*)
215         begin
216                 next_bubble_3a = bubble_2a | flush_2a | delayedflush_2a;
217
218                 jmp_2a = 1'b0;
219                 jmppc_2a = 32'h00000000;
220
221                 casez (insn_2a)
222                 `DECODE_ALU_MULT:       /* Multiply -- must come before ALU, because it pattern matches a specific case of ALU */
223                         next_bubble_3a = next_bubble_3a | !mult_done | !outstall_3a;
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 */
232                 `DECODE_LDRSTR_UNDEFINED,       /* Undefined. I hate ARM */
233                 `DECODE_LDRSTR,         /* Single data transfer */
234                 `DECODE_LDMSTM:         /* Block data transfer */
235                 begin end
236                 `DECODE_BRANCH:
237                 begin
238                         if(!bubble_2a && !flush_2a && !delayedflush_2a && !outstall_2a /* Let someone else take precedence. */) begin
239                                 jmppc_2a = pc_2a + op0_2a + 32'h8;
240                                 jmp_2a = 1'b1;
241                         end
242                 end                     /* Branch */
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
251         end
252 endmodule
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
290
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,
304         output reg setres
305 );
306         reg [31:0] res;
307         reg flag_n, flag_z, flag_c, flag_v;
308         wire [32:0] sum, diff, rdiff;
309         wire sum_v, diff_v, rdiff_v;
310
311         assign sum = {1'b0, in0} + {1'b0, in1};
312         assign diff = {1'b0, in0} - {1'b0, in1};
313         assign rdiff = {1'b0, in1} - {1'b0, in0};
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]);
317
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
325                         result = in0 & in1;
326                         flag_c = shifter_carry;
327                         setres = 1'b1;
328                 end
329                 `ALU_EOR: begin
330                         result = in0 ^ in1;
331                         flag_c = shifter_carry;
332                         setres = 1'b1;
333                 end
334                 `ALU_SUB: begin
335                         {flag_c, result} = diff;
336                         flag_c = !flag_c;
337                         flag_v = diff_v;
338                         setres = 1'b1;
339                 end
340                 `ALU_RSB: begin
341                         {flag_c, result} = rdiff;
342                         flag_c = !flag_c;
343                         flag_v = rdiff_v;
344                         setres = 1'b1;
345                 end
346                 `ALU_ADD: begin
347                         {flag_c, result} = sum;
348                         flag_v = sum_v;
349                         setres = 1'b1;
350                 end
351                 `ALU_ADC: begin
352                         {flag_c, result} = sum + {32'b0, cpsr[`CPSR_C]};
353                         flag_v = sum_v | (~sum[31] & result[31]);
354                         setres = 1'b1;
355                 end
356                 `ALU_SBC: begin
357                         {flag_c, result} = diff - {32'b0, (~cpsr[`CPSR_C])};
358                         flag_c = !flag_c;
359                         flag_v = diff_v | (diff[31] & ~result[31]);
360                         setres = 1'b1;
361                 end
362                 `ALU_RSC: begin
363                         {flag_c, result} = rdiff - {32'b0, (~cpsr[`CPSR_C])};
364                         flag_c = !flag_c;
365                         flag_v = rdiff_v | (rdiff[31] & ~result[31]);
366                         setres = 1'b1;
367                 end
368                 `ALU_TST: begin
369                         result = in0 & in1;
370                         flag_c = shifter_carry;
371                         setres = 1'b0;
372                 end
373                 `ALU_TEQ: begin
374                         result = in0 ^ in1;
375                         flag_c = shifter_carry;
376                         setres = 1'b0;
377                 end
378                 `ALU_CMP: begin
379                         {flag_c, result} = diff;
380                         flag_c = !flag_c;
381                         flag_v = diff_v;
382                         setres = 1'b0;
383                 end
384                 `ALU_CMN: begin
385                         {flag_c, result} = sum;
386                         flag_v = sum_v;
387                         setres = 1'b0;
388                 end
389                 `ALU_ORR: begin
390                         result = in0 | in1;
391                         flag_c = shifter_carry;
392                         setres = 1'b1;
393                 end
394                 `ALU_MOV: begin
395                         result = in1;
396                         flag_c = shifter_carry;
397                         setres = 1'b1;
398                 end
399                 `ALU_BIC: begin
400                         result = in0 & (~in1);
401                         flag_c = shifter_carry;
402                         setres = 1'b1;
403                 end
404                 `ALU_MVN: begin
405                         result = ~in1;
406                         flag_c = shifter_carry;
407                         setres = 1'b1;
408                 end
409                 endcase
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;
415         end
416 endmodule
This page took 0.03933 seconds and 4 git commands to generate.