cpu: `Minor' in-order CPU model
[gem5.git] / src / cpu / minor / pipeline.cc
1 /*
2 * Copyright (c) 2013-2014 ARM Limited
3 * All rights reserved
4 *
5 * The license below extends only to copyright in the software and shall
6 * not be construed as granting a license to any other intellectual
7 * property including but not limited to intellectual property relating
8 * to a hardware implementation of the functionality of the software
9 * licensed hereunder. You may use the software subject to the license
10 * terms below provided that you ensure that this notice is replicated
11 * unmodified and in its entirety in all distributions of the software,
12 * modified or unmodified, in source code or in binary form.
13 *
14 * Redistribution and use in source and binary forms, with or without
15 * modification, are permitted provided that the following conditions are
16 * met: redistributions of source code must retain the above copyright
17 * notice, this list of conditions and the following disclaimer;
18 * redistributions in binary form must reproduce the above copyright
19 * notice, this list of conditions and the following disclaimer in the
20 * documentation and/or other materials provided with the distribution;
21 * neither the name of the copyright holders nor the names of its
22 * contributors may be used to endorse or promote products derived from
23 * this software without specific prior written permission.
24 *
25 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
26 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
27 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
28 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
29 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
30 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
31 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
32 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
33 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
34 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
35 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
36 *
37 * Authors: Andrew Bardsley
38 */
39
40 #include <algorithm>
41
42 #include "cpu/minor/decode.hh"
43 #include "cpu/minor/execute.hh"
44 #include "cpu/minor/fetch1.hh"
45 #include "cpu/minor/fetch2.hh"
46 #include "cpu/minor/pipeline.hh"
47 #include "debug/Drain.hh"
48 #include "debug/MinorCPU.hh"
49 #include "debug/MinorTrace.hh"
50 #include "debug/Quiesce.hh"
51
52 namespace Minor
53 {
54
55 Pipeline::Pipeline(MinorCPU &cpu_, MinorCPUParams &params) :
56 Ticked(cpu_, &(cpu_.BaseCPU::numCycles)),
57 cpu(cpu_),
58 allow_idling(params.enableIdling),
59 f1ToF2(cpu.name() + ".f1ToF2", "lines",
60 params.fetch1ToFetch2ForwardDelay),
61 f2ToF1(cpu.name() + ".f2ToF1", "prediction",
62 params.fetch1ToFetch2BackwardDelay, true),
63 f2ToD(cpu.name() + ".f2ToD", "insts",
64 params.fetch2ToDecodeForwardDelay),
65 dToE(cpu.name() + ".dToE", "insts",
66 params.decodeToExecuteForwardDelay),
67 eToF1(cpu.name() + ".eToF1", "branch",
68 params.executeBranchDelay),
69 execute(cpu.name() + ".execute", cpu, params,
70 dToE.output(), eToF1.input()),
71 decode(cpu.name() + ".decode", cpu, params,
72 f2ToD.output(), dToE.input(), execute.inputBuffer),
73 fetch2(cpu.name() + ".fetch2", cpu, params,
74 f1ToF2.output(), eToF1.output(), f2ToF1.input(), f2ToD.input(),
75 decode.inputBuffer),
76 fetch1(cpu.name() + ".fetch1", cpu, params,
77 eToF1.output(), f1ToF2.input(), f2ToF1.output(), fetch2.inputBuffer),
78 activityRecorder(cpu.name() + ".activity", Num_StageId,
79 /* The max depth of inter-stage FIFOs */
80 std::max(params.fetch1ToFetch2ForwardDelay,
81 std::max(params.fetch2ToDecodeForwardDelay,
82 std::max(params.decodeToExecuteForwardDelay,
83 params.executeBranchDelay)))),
84 needToSignalDrained(false)
85 {
86 if (params.fetch1ToFetch2ForwardDelay < 1) {
87 fatal("%s: fetch1ToFetch2ForwardDelay must be >= 1 (%d)\n",
88 cpu.name(), params.fetch1ToFetch2ForwardDelay);
89 }
90
91 if (params.fetch2ToDecodeForwardDelay < 1) {
92 fatal("%s: fetch2ToDecodeForwardDelay must be >= 1 (%d)\n",
93 cpu.name(), params.fetch2ToDecodeForwardDelay);
94 }
95
96 if (params.decodeToExecuteForwardDelay < 1) {
97 fatal("%s: decodeToExecuteForwardDelay must be >= 1 (%d)\n",
98 cpu.name(), params.decodeToExecuteForwardDelay);
99 }
100
101 if (params.executeBranchDelay < 1) {
102 fatal("%s: executeBranchDelay must be >= 1\n",
103 cpu.name(), params.executeBranchDelay);
104 }
105 }
106
107 void
108 Pipeline::minorTrace() const
109 {
110 fetch1.minorTrace();
111 f1ToF2.minorTrace();
112 f2ToF1.minorTrace();
113 fetch2.minorTrace();
114 f2ToD.minorTrace();
115 decode.minorTrace();
116 dToE.minorTrace();
117 execute.minorTrace();
118 eToF1.minorTrace();
119 activityRecorder.minorTrace();
120 }
121
122 void
123 Pipeline::evaluate()
124 {
125 /* Note that it's important to evaluate the stages in order to allow
126 * 'immediate', 0-time-offset TimeBuffer activity to be visible from
127 * later stages to earlier ones in the same cycle */
128 execute.evaluate();
129 decode.evaluate();
130 fetch2.evaluate();
131 fetch1.evaluate();
132
133 if (DTRACE(MinorTrace))
134 minorTrace();
135
136 /* Update the time buffers after the stages */
137 f1ToF2.evaluate();
138 f2ToF1.evaluate();
139 f2ToD.evaluate();
140 dToE.evaluate();
141 eToF1.evaluate();
142
143 /* The activity recorder must be be called after all the stages and
144 * before the idler (which acts on the advice of the activity recorder */
145 activityRecorder.evaluate();
146
147 if (allow_idling) {
148 /* Become idle if we can but are not draining */
149 if (!activityRecorder.active() && !needToSignalDrained) {
150 DPRINTF(Quiesce, "Suspending as the processor is idle\n");
151 stop();
152 }
153
154 /* Deactivate all stages. Note that the stages *could*
155 * activate and deactivate themselves but that's fraught
156 * with additional difficulty.
157 * As organised herre */
158 activityRecorder.deactivateStage(Pipeline::CPUStageId);
159 activityRecorder.deactivateStage(Pipeline::Fetch1StageId);
160 activityRecorder.deactivateStage(Pipeline::Fetch2StageId);
161 activityRecorder.deactivateStage(Pipeline::DecodeStageId);
162 activityRecorder.deactivateStage(Pipeline::ExecuteStageId);
163 }
164
165 if (needToSignalDrained) /* Must be draining */
166 {
167 DPRINTF(Drain, "Still draining\n");
168 if (isDrained()) {
169 DPRINTF(Drain, "Signalling end of draining\n");
170 cpu.signalDrainDone();
171 needToSignalDrained = false;
172 stop();
173 }
174 }
175 }
176
177 MinorCPU::MinorCPUPort &
178 Pipeline::getInstPort()
179 {
180 return fetch1.getIcachePort();
181 }
182
183 MinorCPU::MinorCPUPort &
184 Pipeline::getDataPort()
185 {
186 return execute.getDcachePort();
187 }
188
189 void
190 Pipeline::wakeupFetch()
191 {
192 execute.wakeupFetch();
193 }
194
195 unsigned int
196 Pipeline::drain(DrainManager *manager)
197 {
198 DPRINTF(MinorCPU, "Draining pipeline by halting inst fetches. "
199 " Execution should drain naturally\n");
200
201 execute.drain();
202
203 /* Make sure that needToSignalDrained isn't accidentally set if we
204 * are 'pre-drained' */
205 bool drained = isDrained();
206 needToSignalDrained = !drained;
207
208 return (drained ? 0 : 1);
209 }
210
211 void
212 Pipeline::drainResume()
213 {
214 DPRINTF(Drain, "Drain resume\n");
215 execute.drainResume();
216 }
217
218 bool
219 Pipeline::isDrained()
220 {
221 bool fetch1_drained = fetch1.isDrained();
222 bool fetch2_drained = fetch2.isDrained();
223 bool decode_drained = decode.isDrained();
224 bool execute_drained = execute.isDrained();
225
226 bool f1_to_f2_drained = f1ToF2.empty();
227 bool f2_to_f1_drained = f2ToF1.empty();
228 bool f2_to_d_drained = f2ToD.empty();
229 bool d_to_e_drained = dToE.empty();
230
231 bool ret = fetch1_drained && fetch2_drained &&
232 decode_drained && execute_drained &&
233 f1_to_f2_drained && f2_to_f1_drained &&
234 f2_to_d_drained && d_to_e_drained;
235
236 DPRINTF(MinorCPU, "Pipeline undrained stages state:%s%s%s%s%s%s%s%s\n",
237 (fetch1_drained ? "" : " Fetch1"),
238 (fetch2_drained ? "" : " Fetch2"),
239 (decode_drained ? "" : " Decode"),
240 (execute_drained ? "" : " Execute"),
241 (f1_to_f2_drained ? "" : " F1->F2"),
242 (f2_to_f1_drained ? "" : " F2->F1"),
243 (f2_to_d_drained ? "" : " F2->D"),
244 (d_to_e_drained ? "" : " D->E")
245 );
246
247 return ret;
248 }
249
250 }