1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
|
/*
* Copyright (c) 1999-2008 Mark D. Hill and David A. Wood
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met: redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer;
* redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution;
* neither the name of the copyright holders nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "arch/x86/ldstflags.hh"
#include "base/misc.hh"
#include "base/str.hh"
#include "cpu/testers/rubytest/RubyTester.hh"
#include "debug/MemoryAccess.hh"
#include "debug/ProtocolTrace.hh"
#include "debug/RubySequencer.hh"
#include "debug/RubyStats.hh"
#include "mem/protocol/PrefetchBit.hh"
#include "mem/protocol/RubyAccessMode.hh"
#include "mem/ruby/common/Global.hh"
#include "mem/ruby/profiler/Profiler.hh"
#include "mem/ruby/slicc_interface/RubyRequest.hh"
#include "mem/ruby/system/Sequencer.hh"
#include "mem/ruby/system/System.hh"
#include "mem/packet.hh"
#include "sim/system.hh"
using namespace std;
Sequencer *
RubySequencerParams::create()
{
return new Sequencer(this);
}
Sequencer::Sequencer(const Params *p)
: RubyPort(p), m_IncompleteTimes(MachineType_NUM), deadlockCheckEvent(this)
{
m_outstanding_count = 0;
m_instCache_ptr = p->icache;
m_dataCache_ptr = p->dcache;
m_max_outstanding_requests = p->max_outstanding_requests;
m_deadlock_threshold = p->deadlock_threshold;
assert(m_max_outstanding_requests > 0);
assert(m_deadlock_threshold > 0);
assert(m_instCache_ptr != NULL);
assert(m_dataCache_ptr != NULL);
m_usingNetworkTester = p->using_network_tester;
}
Sequencer::~Sequencer()
{
}
void
Sequencer::wakeup()
{
assert(getDrainState() != Drainable::Draining);
// Check for deadlock of any of the requests
Cycles current_time = curCycle();
// Check across all outstanding requests
int total_outstanding = 0;
RequestTable::iterator read = m_readRequestTable.begin();
RequestTable::iterator read_end = m_readRequestTable.end();
for (; read != read_end; ++read) {
SequencerRequest* request = read->second;
if (current_time - request->issue_time < m_deadlock_threshold)
continue;
panic("Possible Deadlock detected. Aborting!\n"
"version: %d request.paddr: 0x%x m_readRequestTable: %d "
"current time: %u issue_time: %d difference: %d\n", m_version,
Address(request->pkt->getAddr()), m_readRequestTable.size(),
current_time * clockPeriod(), request->issue_time * clockPeriod(),
(current_time * clockPeriod()) - (request->issue_time * clockPeriod()));
}
RequestTable::iterator write = m_writeRequestTable.begin();
RequestTable::iterator write_end = m_writeRequestTable.end();
for (; write != write_end; ++write) {
SequencerRequest* request = write->second;
if (current_time - request->issue_time < m_deadlock_threshold)
continue;
panic("Possible Deadlock detected. Aborting!\n"
"version: %d request.paddr: 0x%x m_writeRequestTable: %d "
"current time: %u issue_time: %d difference: %d\n", m_version,
Address(request->pkt->getAddr()), m_writeRequestTable.size(),
current_time * clockPeriod(), request->issue_time * clockPeriod(),
(current_time * clockPeriod()) - (request->issue_time * clockPeriod()));
}
total_outstanding += m_writeRequestTable.size();
total_outstanding += m_readRequestTable.size();
assert(m_outstanding_count == total_outstanding);
if (m_outstanding_count > 0) {
// If there are still outstanding requests, keep checking
schedule(deadlockCheckEvent, clockEdge(m_deadlock_threshold));
}
}
void Sequencer::resetStats()
{
m_latencyHist.reset();
m_hitLatencyHist.reset();
m_missLatencyHist.reset();
for (int i = 0; i < RubyRequestType_NUM; i++) {
m_typeLatencyHist[i]->reset();
m_hitTypeLatencyHist[i]->reset();
m_missTypeLatencyHist[i]->reset();
for (int j = 0; j < MachineType_NUM; j++) {
m_hitTypeMachLatencyHist[i][j]->reset();
m_missTypeMachLatencyHist[i][j]->reset();
}
}
for (int i = 0; i < MachineType_NUM; i++) {
m_missMachLatencyHist[i]->reset();
m_hitMachLatencyHist[i]->reset();
m_IssueToInitialDelayHist[i]->reset();
m_InitialToForwardDelayHist[i]->reset();
m_ForwardToFirstResponseDelayHist[i]->reset();
m_FirstResponseToCompletionDelayHist[i]->reset();
m_IncompleteTimes[i] = 0;
}
}
void
Sequencer::printProgress(ostream& out) const
{
#if 0
int total_demand = 0;
out << "Sequencer Stats Version " << m_version << endl;
out << "Current time = " << g_system_ptr->getTime() << endl;
out << "---------------" << endl;
out << "outstanding requests" << endl;
out << "proc " << m_Read
<< " version Requests = " << m_readRequestTable.size() << endl;
// print the request table
RequestTable::iterator read = m_readRequestTable.begin();
RequestTable::iterator read_end = m_readRequestTable.end();
for (; read != read_end; ++read) {
SequencerRequest* request = read->second;
out << "\tRequest[ " << i << " ] = " << request->type
<< " Address " << rkeys[i]
<< " Posted " << request->issue_time
<< " PF " << PrefetchBit_No << endl;
total_demand++;
}
out << "proc " << m_version
<< " Write Requests = " << m_writeRequestTable.size << endl;
// print the request table
RequestTable::iterator write = m_writeRequestTable.begin();
RequestTable::iterator write_end = m_writeRequestTable.end();
for (; write != write_end; ++write) {
SequencerRequest* request = write->second;
out << "\tRequest[ " << i << " ] = " << request.getType()
<< " Address " << wkeys[i]
<< " Posted " << request.getTime()
<< " PF " << request.getPrefetch() << endl;
if (request.getPrefetch() == PrefetchBit_No) {
total_demand++;
}
}
out << endl;
out << "Total Number Outstanding: " << m_outstanding_count << endl
<< "Total Number Demand : " << total_demand << endl
<< "Total Number Prefetches : " << m_outstanding_count - total_demand
<< endl << endl << endl;
#endif
}
// Insert the request on the correct request table. Return true if
// the entry was already present.
RequestStatus
Sequencer::insertRequest(PacketPtr pkt, RubyRequestType request_type)
{
assert(m_outstanding_count ==
(m_writeRequestTable.size() + m_readRequestTable.size()));
// See if we should schedule a deadlock check
if (!deadlockCheckEvent.scheduled() &&
getDrainState() != Drainable::Draining) {
schedule(deadlockCheckEvent, clockEdge(m_deadlock_threshold));
}
Address line_addr(pkt->getAddr());
line_addr.makeLineAddress();
// Create a default entry, mapping the address to NULL, the cast is
// there to make gcc 4.4 happy
RequestTable::value_type default_entry(line_addr,
(SequencerRequest*) NULL);
if ((request_type == RubyRequestType_ST) ||
(request_type == RubyRequestType_RMW_Read) ||
(request_type == RubyRequestType_RMW_Write) ||
(request_type == RubyRequestType_Load_Linked) ||
(request_type == RubyRequestType_Store_Conditional) ||
(request_type == RubyRequestType_Locked_RMW_Read) ||
(request_type == RubyRequestType_Locked_RMW_Write) ||
(request_type == RubyRequestType_FLUSH)) {
// Check if there is any outstanding read request for the same
// cache line.
if (m_readRequestTable.count(line_addr) > 0) {
m_store_waiting_on_load++;
return RequestStatus_Aliased;
}
pair<RequestTable::iterator, bool> r =
m_writeRequestTable.insert(default_entry);
if (r.second) {
RequestTable::iterator i = r.first;
i->second = new SequencerRequest(pkt, request_type, curCycle());
m_outstanding_count++;
} else {
// There is an outstanding write request for the cache line
m_store_waiting_on_store++;
return RequestStatus_Aliased;
}
} else {
// Check if there is any outstanding write request for the same
// cache line.
if (m_writeRequestTable.count(line_addr) > 0) {
m_load_waiting_on_store++;
return RequestStatus_Aliased;
}
pair<RequestTable::iterator, bool> r =
m_readRequestTable.insert(default_entry);
if (r.second) {
RequestTable::iterator i = r.first;
i->second = new SequencerRequest(pkt, request_type, curCycle());
m_outstanding_count++;
} else {
// There is an outstanding read request for the cache line
m_load_waiting_on_load++;
return RequestStatus_Aliased;
}
}
m_outstandReqHist.sample(m_outstanding_count);
assert(m_outstanding_count ==
(m_writeRequestTable.size() + m_readRequestTable.size()));
return RequestStatus_Ready;
}
void
Sequencer::markRemoved()
{
m_outstanding_count--;
assert(m_outstanding_count ==
m_writeRequestTable.size() + m_readRequestTable.size());
}
void
Sequencer::removeRequest(SequencerRequest* srequest)
{
assert(m_outstanding_count ==
m_writeRequestTable.size() + m_readRequestTable.size());
Address line_addr(srequest->pkt->getAddr());
line_addr.makeLineAddress();
if ((srequest->m_type == RubyRequestType_ST) ||
(srequest->m_type == RubyRequestType_RMW_Read) ||
(srequest->m_type == RubyRequestType_RMW_Write) ||
(srequest->m_type == RubyRequestType_Load_Linked) ||
(srequest->m_type == RubyRequestType_Store_Conditional) ||
(srequest->m_type == RubyRequestType_Locked_RMW_Read) ||
(srequest->m_type == RubyRequestType_Locked_RMW_Write)) {
m_writeRequestTable.erase(line_addr);
} else {
m_readRequestTable.erase(line_addr);
}
markRemoved();
}
void
Sequencer::invalidateSC(const Address& address)
{
RequestTable::iterator i = m_writeRequestTable.find(address);
if (i != m_writeRequestTable.end()) {
SequencerRequest* request = i->second;
// The controller has lost the coherence permissions, hence the lock
// on the cache line maintained by the cache should be cleared.
if (request->m_type == RubyRequestType_Store_Conditional) {
m_dataCache_ptr->clearLocked(address);
}
}
}
bool
Sequencer::handleLlsc(const Address& address, SequencerRequest* request)
{
//
// The success flag indicates whether the LLSC operation was successful.
// LL ops will always succeed, but SC may fail if the cache line is no
// longer locked.
//
bool success = true;
if (request->m_type == RubyRequestType_Store_Conditional) {
if (!m_dataCache_ptr->isLocked(address, m_version)) {
//
// For failed SC requests, indicate the failure to the cpu by
// setting the extra data to zero.
//
request->pkt->req->setExtraData(0);
success = false;
} else {
//
// For successful SC requests, indicate the success to the cpu by
// setting the extra data to one.
//
request->pkt->req->setExtraData(1);
}
//
// Independent of success, all SC operations must clear the lock
//
m_dataCache_ptr->clearLocked(address);
} else if (request->m_type == RubyRequestType_Load_Linked) {
//
// Note: To fully follow Alpha LLSC semantics, should the LL clear any
// previously locked cache lines?
//
m_dataCache_ptr->setLocked(address, m_version);
} else if ((m_dataCache_ptr->isTagPresent(address)) &&
(m_dataCache_ptr->isLocked(address, m_version))) {
//
// Normal writes should clear the locked address
//
m_dataCache_ptr->clearLocked(address);
}
return success;
}
void
Sequencer::recordMissLatency(const Cycles cycles, const RubyRequestType type,
const MachineType respondingMach,
bool isExternalHit, Cycles issuedTime,
Cycles initialRequestTime,
Cycles forwardRequestTime,
Cycles firstResponseTime, Cycles completionTime)
{
m_latencyHist.sample(cycles);
m_typeLatencyHist[type]->sample(cycles);
if (isExternalHit) {
m_missLatencyHist.sample(cycles);
m_missTypeLatencyHist[type]->sample(cycles);
if (respondingMach != MachineType_NUM) {
m_missMachLatencyHist[respondingMach]->sample(cycles);
m_missTypeMachLatencyHist[type][respondingMach]->sample(cycles);
if ((issuedTime <= initialRequestTime) &&
(initialRequestTime <= forwardRequestTime) &&
(forwardRequestTime <= firstResponseTime) &&
(firstResponseTime <= completionTime)) {
m_IssueToInitialDelayHist[respondingMach]->sample(
initialRequestTime - issuedTime);
m_InitialToForwardDelayHist[respondingMach]->sample(
forwardRequestTime - initialRequestTime);
m_ForwardToFirstResponseDelayHist[respondingMach]->sample(
firstResponseTime - forwardRequestTime);
m_FirstResponseToCompletionDelayHist[respondingMach]->sample(
completionTime - firstResponseTime);
} else {
m_IncompleteTimes[respondingMach]++;
}
}
} else {
m_hitLatencyHist.sample(cycles);
m_hitTypeLatencyHist[type]->sample(cycles);
if (respondingMach != MachineType_NUM) {
m_hitMachLatencyHist[respondingMach]->sample(cycles);
m_hitTypeMachLatencyHist[type][respondingMach]->sample(cycles);
}
}
}
void
Sequencer::writeCallback(const Address& address, DataBlock& data,
const bool externalHit, const MachineType mach,
const Cycles initialRequestTime,
const Cycles forwardRequestTime,
const Cycles firstResponseTime)
{
assert(address == line_address(address));
assert(m_writeRequestTable.count(line_address(address)));
RequestTable::iterator i = m_writeRequestTable.find(address);
assert(i != m_writeRequestTable.end());
SequencerRequest* request = i->second;
m_writeRequestTable.erase(i);
markRemoved();
assert((request->m_type == RubyRequestType_ST) ||
(request->m_type == RubyRequestType_ATOMIC) ||
(request->m_type == RubyRequestType_RMW_Read) ||
(request->m_type == RubyRequestType_RMW_Write) ||
(request->m_type == RubyRequestType_Load_Linked) ||
(request->m_type == RubyRequestType_Store_Conditional) ||
(request->m_type == RubyRequestType_Locked_RMW_Read) ||
(request->m_type == RubyRequestType_Locked_RMW_Write) ||
(request->m_type == RubyRequestType_FLUSH));
//
// For Alpha, properly handle LL, SC, and write requests with respect to
// locked cache blocks.
//
// Not valid for Network_test protocl
//
bool success = true;
if(!m_usingNetworkTester)
success = handleLlsc(address, request);
if (request->m_type == RubyRequestType_Locked_RMW_Read) {
m_controller->blockOnQueue(address, m_mandatory_q_ptr);
} else if (request->m_type == RubyRequestType_Locked_RMW_Write) {
m_controller->unblock(address);
}
hitCallback(request, data, success, mach, externalHit,
initialRequestTime, forwardRequestTime, firstResponseTime);
}
void
Sequencer::readCallback(const Address& address, DataBlock& data,
bool externalHit, const MachineType mach,
Cycles initialRequestTime,
Cycles forwardRequestTime,
Cycles firstResponseTime)
{
assert(address == line_address(address));
assert(m_readRequestTable.count(line_address(address)));
RequestTable::iterator i = m_readRequestTable.find(address);
assert(i != m_readRequestTable.end());
SequencerRequest* request = i->second;
m_readRequestTable.erase(i);
markRemoved();
assert((request->m_type == RubyRequestType_LD) ||
(request->m_type == RubyRequestType_IFETCH));
hitCallback(request, data, true, mach, externalHit,
initialRequestTime, forwardRequestTime, firstResponseTime);
}
void
Sequencer::hitCallback(SequencerRequest* srequest, DataBlock& data,
bool llscSuccess,
const MachineType mach, const bool externalHit,
const Cycles initialRequestTime,
const Cycles forwardRequestTime,
const Cycles firstResponseTime)
{
PacketPtr pkt = srequest->pkt;
Address request_address(pkt->getAddr());
Address request_line_address(pkt->getAddr());
request_line_address.makeLineAddress();
RubyRequestType type = srequest->m_type;
Cycles issued_time = srequest->issue_time;
// Set this cache entry to the most recently used
if (type == RubyRequestType_IFETCH) {
m_instCache_ptr->setMRU(request_line_address);
} else {
m_dataCache_ptr->setMRU(request_line_address);
}
assert(curCycle() >= issued_time);
Cycles total_latency = curCycle() - issued_time;
// Profile the latency for all demand accesses.
recordMissLatency(total_latency, type, mach, externalHit, issued_time,
initialRequestTime, forwardRequestTime,
firstResponseTime, curCycle());
DPRINTFR(ProtocolTrace, "%15s %3s %10s%20s %6s>%-6s %s %d cycles\n",
curTick(), m_version, "Seq",
llscSuccess ? "Done" : "SC_Failed", "", "",
request_address, total_latency);
// update the data
if (g_system_ptr->m_warmup_enabled) {
assert(pkt->getPtr<uint8_t>(false) != NULL);
data.setData(pkt->getPtr<uint8_t>(false),
request_address.getOffset(), pkt->getSize());
} else if (pkt->getPtr<uint8_t>(true) != NULL) {
if ((type == RubyRequestType_LD) ||
(type == RubyRequestType_IFETCH) ||
(type == RubyRequestType_RMW_Read) ||
(type == RubyRequestType_Locked_RMW_Read) ||
(type == RubyRequestType_Load_Linked)) {
memcpy(pkt->getPtr<uint8_t>(true),
data.getData(request_address.getOffset(), pkt->getSize()),
pkt->getSize());
} else {
data.setData(pkt->getPtr<uint8_t>(true),
request_address.getOffset(), pkt->getSize());
}
} else {
DPRINTF(MemoryAccess,
"WARNING. Data not transfered from Ruby to M5 for type %s\n",
RubyRequestType_to_string(type));
}
// If using the RubyTester, update the RubyTester sender state's
// subBlock with the recieved data. The tester will later access
// this state.
if (m_usingRubyTester) {
RubyTester::SenderState* testerSenderState =
pkt->findNextSenderState<RubyTester::SenderState>();
assert(testerSenderState);
testerSenderState->subBlock.mergeFrom(data);
}
delete srequest;
if (g_system_ptr->m_warmup_enabled) {
assert(pkt->req);
delete pkt->req;
delete pkt;
g_system_ptr->m_cache_recorder->enqueueNextFetchRequest();
} else if (g_system_ptr->m_cooldown_enabled) {
delete pkt;
g_system_ptr->m_cache_recorder->enqueueNextFlushRequest();
} else {
ruby_hit_callback(pkt);
}
}
bool
Sequencer::empty() const
{
return m_writeRequestTable.empty() && m_readRequestTable.empty();
}
RequestStatus
Sequencer::makeRequest(PacketPtr pkt)
{
if (m_outstanding_count >= m_max_outstanding_requests) {
return RequestStatus_BufferFull;
}
RubyRequestType primary_type = RubyRequestType_NULL;
RubyRequestType secondary_type = RubyRequestType_NULL;
if (pkt->isLLSC()) {
//
// Alpha LL/SC instructions need to be handled carefully by the cache
// coherence protocol to ensure they follow the proper semantics. In
// particular, by identifying the operations as atomic, the protocol
// should understand that migratory sharing optimizations should not
// be performed (i.e. a load between the LL and SC should not steal
// away exclusive permission).
//
if (pkt->isWrite()) {
DPRINTF(RubySequencer, "Issuing SC\n");
primary_type = RubyRequestType_Store_Conditional;
} else {
DPRINTF(RubySequencer, "Issuing LL\n");
assert(pkt->isRead());
primary_type = RubyRequestType_Load_Linked;
}
secondary_type = RubyRequestType_ATOMIC;
} else if (pkt->req->isLocked()) {
//
// x86 locked instructions are translated to store cache coherence
// requests because these requests should always be treated as read
// exclusive operations and should leverage any migratory sharing
// optimization built into the protocol.
//
if (pkt->isWrite()) {
DPRINTF(RubySequencer, "Issuing Locked RMW Write\n");
primary_type = RubyRequestType_Locked_RMW_Write;
} else {
DPRINTF(RubySequencer, "Issuing Locked RMW Read\n");
assert(pkt->isRead());
primary_type = RubyRequestType_Locked_RMW_Read;
}
secondary_type = RubyRequestType_ST;
} else {
if (pkt->isRead()) {
if (pkt->req->isInstFetch()) {
primary_type = secondary_type = RubyRequestType_IFETCH;
} else {
bool storeCheck = false;
// only X86 need the store check
if (system->getArch() == Arch::X86ISA) {
uint32_t flags = pkt->req->getFlags();
storeCheck = flags &
(X86ISA::StoreCheck << X86ISA::FlagShift);
}
if (storeCheck) {
primary_type = RubyRequestType_RMW_Read;
secondary_type = RubyRequestType_ST;
} else {
primary_type = secondary_type = RubyRequestType_LD;
}
}
} else if (pkt->isWrite()) {
//
// Note: M5 packets do not differentiate ST from RMW_Write
//
primary_type = secondary_type = RubyRequestType_ST;
} else if (pkt->isFlush()) {
primary_type = secondary_type = RubyRequestType_FLUSH;
} else {
panic("Unsupported ruby packet type\n");
}
}
RequestStatus status = insertRequest(pkt, primary_type);
if (status != RequestStatus_Ready)
return status;
issueRequest(pkt, secondary_type);
// TODO: issue hardware prefetches here
return RequestStatus_Issued;
}
void
Sequencer::issueRequest(PacketPtr pkt, RubyRequestType secondary_type)
{
assert(pkt != NULL);
int proc_id = -1;
if (pkt->req->hasContextId()) {
proc_id = pkt->req->contextId();
}
// If valid, copy the pc to the ruby request
Addr pc = 0;
if (pkt->req->hasPC()) {
pc = pkt->req->getPC();
}
std::shared_ptr<RubyRequest> msg =
std::make_shared<RubyRequest>(clockEdge(), pkt->getAddr(),
pkt->getPtr<uint8_t>(true),
pkt->getSize(), pc, secondary_type,
RubyAccessMode_Supervisor, pkt,
PrefetchBit_No, proc_id);
DPRINTFR(ProtocolTrace, "%15s %3s %10s%20s %6s>%-6s %s %s\n",
curTick(), m_version, "Seq", "Begin", "", "",
msg->getPhysicalAddress(),
RubyRequestType_to_string(secondary_type));
Cycles latency(0); // initialzed to an null value
if (secondary_type == RubyRequestType_IFETCH)
latency = m_instCache_ptr->getLatency();
else
latency = m_dataCache_ptr->getLatency();
// Send the message to the cache controller
assert(latency > 0);
assert(m_mandatory_q_ptr != NULL);
m_mandatory_q_ptr->enqueue(msg, latency);
}
template <class KEY, class VALUE>
std::ostream &
operator<<(ostream &out, const m5::hash_map<KEY, VALUE> &map)
{
typename m5::hash_map<KEY, VALUE>::const_iterator i = map.begin();
typename m5::hash_map<KEY, VALUE>::const_iterator end = map.end();
out << "[";
for (; i != end; ++i)
out << " " << i->first << "=" << i->second;
out << " ]";
return out;
}
void
Sequencer::print(ostream& out) const
{
out << "[Sequencer: " << m_version
<< ", outstanding requests: " << m_outstanding_count
<< ", read request table: " << m_readRequestTable
<< ", write request table: " << m_writeRequestTable
<< "]";
}
// this can be called from setState whenever coherence permissions are
// upgraded when invoked, coherence violations will be checked for the
// given block
void
Sequencer::checkCoherence(const Address& addr)
{
#ifdef CHECK_COHERENCE
g_system_ptr->checkGlobalCoherenceInvariant(addr);
#endif
}
void
Sequencer::recordRequestType(SequencerRequestType requestType) {
DPRINTF(RubyStats, "Recorded statistic: %s\n",
SequencerRequestType_to_string(requestType));
}
void
Sequencer::evictionCallback(const Address& address)
{
ruby_eviction_callback(address);
}
void
Sequencer::regStats()
{
m_store_waiting_on_load
.name(name() + ".store_waiting_on_load")
.desc("Number of times a store aliased with a pending load")
.flags(Stats::nozero);
m_store_waiting_on_store
.name(name() + ".store_waiting_on_store")
.desc("Number of times a store aliased with a pending store")
.flags(Stats::nozero);
m_load_waiting_on_load
.name(name() + ".load_waiting_on_load")
.desc("Number of times a load aliased with a pending load")
.flags(Stats::nozero);
m_load_waiting_on_store
.name(name() + ".load_waiting_on_store")
.desc("Number of times a load aliased with a pending store")
.flags(Stats::nozero);
// These statistical variables are not for display.
// The profiler will collate these across different
// sequencers and display those collated statistics.
m_outstandReqHist.init(10);
m_latencyHist.init(10);
m_hitLatencyHist.init(10);
m_missLatencyHist.init(10);
for (int i = 0; i < RubyRequestType_NUM; i++) {
m_typeLatencyHist.push_back(new Stats::Histogram());
m_typeLatencyHist[i]->init(10);
m_hitTypeLatencyHist.push_back(new Stats::Histogram());
m_hitTypeLatencyHist[i]->init(10);
m_missTypeLatencyHist.push_back(new Stats::Histogram());
m_missTypeLatencyHist[i]->init(10);
}
for (int i = 0; i < MachineType_NUM; i++) {
m_hitMachLatencyHist.push_back(new Stats::Histogram());
m_hitMachLatencyHist[i]->init(10);
m_missMachLatencyHist.push_back(new Stats::Histogram());
m_missMachLatencyHist[i]->init(10);
m_IssueToInitialDelayHist.push_back(new Stats::Histogram());
m_IssueToInitialDelayHist[i]->init(10);
m_InitialToForwardDelayHist.push_back(new Stats::Histogram());
m_InitialToForwardDelayHist[i]->init(10);
m_ForwardToFirstResponseDelayHist.push_back(new Stats::Histogram());
m_ForwardToFirstResponseDelayHist[i]->init(10);
m_FirstResponseToCompletionDelayHist.push_back(new Stats::Histogram());
m_FirstResponseToCompletionDelayHist[i]->init(10);
}
for (int i = 0; i < RubyRequestType_NUM; i++) {
m_hitTypeMachLatencyHist.push_back(std::vector<Stats::Histogram *>());
m_missTypeMachLatencyHist.push_back(std::vector<Stats::Histogram *>());
for (int j = 0; j < MachineType_NUM; j++) {
m_hitTypeMachLatencyHist[i].push_back(new Stats::Histogram());
m_hitTypeMachLatencyHist[i][j]->init(10);
m_missTypeMachLatencyHist[i].push_back(new Stats::Histogram());
m_missTypeMachLatencyHist[i][j]->init(10);
}
}
}
|