1/*
2 * Copyright (c) 2002-2005 The Regents of The University of Michigan
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions are
7 * met: redistributions of source code must retain the above copyright
8 * notice, this list of conditions and the following disclaimer;
9 * redistributions in binary form must reproduce the above copyright
10 * notice, this list of conditions and the following disclaimer in the
11 * documentation and/or other materials provided with the distribution;
12 * neither the name of the copyright holders nor the names of its
13 * contributors may be used to endorse or promote products derived from
14 * this software without specific prior written permission.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
17 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
18 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
19 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
20 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
21 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
22 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
26 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 *
28 * Authors: Steve Reinhardt
29 */
30
31#include "arch/utility.hh"
32#include "arch/faults.hh"
33#include "base/cprintf.hh"
34#include "base/inifile.hh"
35#include "base/loader/symtab.hh"
36#include "base/misc.hh"
37#include "base/pollevent.hh"
38#include "base/range.hh"
39#include "base/stats/events.hh"
40#include "base/trace.hh"
41#include "cpu/base.hh"
42#include "cpu/exetrace.hh"
43#include "cpu/profile.hh"
44#include "cpu/simple/base.hh"
45#include "cpu/simple_thread.hh"
46#include "cpu/smt.hh"
47#include "cpu/static_inst.hh"
48#include "cpu/thread_context.hh"
49#include "mem/packet.hh"
50#include "sim/builder.hh"
51#include "sim/byteswap.hh"
52#include "sim/debug.hh"
53#include "sim/host.hh"
54#include "sim/sim_events.hh"
55#include "sim/sim_object.hh"
56#include "sim/stats.hh"
57#include "sim/system.hh"
58
59#if FULL_SYSTEM
60#include "arch/kernel_stats.hh"
61#include "arch/stacktrace.hh"
62#include "arch/tlb.hh"
63#include "arch/vtophys.hh"
64#include "base/remote_gdb.hh"
65#else // !FULL_SYSTEM
66#include "mem/mem_object.hh"
67#endif // FULL_SYSTEM
68
69using namespace std;
70using namespace TheISA;
71
72BaseSimpleCPU::BaseSimpleCPU(Params *p)
73 : BaseCPU(p), thread(NULL), predecoder(NULL)
74{
75#if FULL_SYSTEM
76 thread = new SimpleThread(this, 0, p->system, p->itb, p->dtb);
77#else
78 thread = new SimpleThread(this, /* thread_num */ 0, p->process,
79 /* asid */ 0);
80#endif // !FULL_SYSTEM
81
82 thread->setStatus(ThreadContext::Suspended);
83
84 tc = thread->getTC();
85
86 numInst = 0;
87 startNumInst = 0;
88 numLoad = 0;
89 startNumLoad = 0;
90 lastIcacheStall = 0;
91 lastDcacheStall = 0;
92
93 threadContexts.push_back(tc);
94
95 fetchOffset = 0;
96 stayAtPC = false;
97}
98
99BaseSimpleCPU::~BaseSimpleCPU()
100{
101}
102
103void
104BaseSimpleCPU::deallocateContext(int thread_num)
105{
106 // for now, these are equivalent
107 suspendContext(thread_num);
108}
109
110
111void
112BaseSimpleCPU::haltContext(int thread_num)
113{
114 // for now, these are equivalent
115 suspendContext(thread_num);
116}
117
118
119void
120BaseSimpleCPU::regStats()
121{
122 using namespace Stats;
123
124 BaseCPU::regStats();
125
126 numInsts
127 .name(name() + ".num_insts")
128 .desc("Number of instructions executed")
129 ;
130
131 numMemRefs
132 .name(name() + ".num_refs")
133 .desc("Number of memory references")
134 ;
135
136 notIdleFraction
137 .name(name() + ".not_idle_fraction")
138 .desc("Percentage of non-idle cycles")
139 ;
140
141 idleFraction
142 .name(name() + ".idle_fraction")
143 .desc("Percentage of idle cycles")
144 ;
145
146 icacheStallCycles
147 .name(name() + ".icache_stall_cycles")
148 .desc("ICache total stall cycles")
149 .prereq(icacheStallCycles)
150 ;
151
152 dcacheStallCycles
153 .name(name() + ".dcache_stall_cycles")
154 .desc("DCache total stall cycles")
155 .prereq(dcacheStallCycles)
156 ;
157
158 icacheRetryCycles
159 .name(name() + ".icache_retry_cycles")
160 .desc("ICache total retry cycles")
161 .prereq(icacheRetryCycles)
162 ;
163
164 dcacheRetryCycles
165 .name(name() + ".dcache_retry_cycles")
166 .desc("DCache total retry cycles")
167 .prereq(dcacheRetryCycles)
168 ;
169
170 idleFraction = constant(1.0) - notIdleFraction;
171}
172
173void
174BaseSimpleCPU::resetStats()
175{
176// startNumInst = numInst;
177 // notIdleFraction = (_status != Idle);
178}
179
180void
181BaseSimpleCPU::serialize(ostream &os)
182{
183 BaseCPU::serialize(os);
184// SERIALIZE_SCALAR(inst);
185 nameOut(os, csprintf("%s.xc.0", name()));
186 thread->serialize(os);
187}
188
189void
190BaseSimpleCPU::unserialize(Checkpoint *cp, const string &section)
191{
192 BaseCPU::unserialize(cp, section);
193// UNSERIALIZE_SCALAR(inst);
194 thread->unserialize(cp, csprintf("%s.xc.0", section));
195}
196
197void
198change_thread_state(int thread_number, int activate, int priority)
199{
200}
201
202Fault
203BaseSimpleCPU::copySrcTranslate(Addr src)
204{
205#if 0
206 static bool no_warn = true;
207 int blk_size = (dcacheInterface) ? dcacheInterface->getBlockSize() : 64;
208 // Only support block sizes of 64 atm.
209 assert(blk_size == 64);
210 int offset = src & (blk_size - 1);
211
212 // Make sure block doesn't span page
213 if (no_warn &&
214 (src & PageMask) != ((src + blk_size) & PageMask) &&
215 (src >> 40) != 0xfffffc) {
216 warn("Copied block source spans pages %x.", src);
217 no_warn = false;
218 }
219
220 memReq->reset(src & ~(blk_size - 1), blk_size);
221
222 // translate to physical address
223 Fault fault = thread->translateDataReadReq(req);
224
225 if (fault == NoFault) {
226 thread->copySrcAddr = src;
227 thread->copySrcPhysAddr = memReq->paddr + offset;
228 } else {
229 assert(!fault->isAlignmentFault());
230
231 thread->copySrcAddr = 0;
232 thread->copySrcPhysAddr = 0;
233 }
234 return fault;
235#else
236 return NoFault;
237#endif
238}
239
240Fault
241BaseSimpleCPU::copy(Addr dest)
242{
243#if 0
244 static bool no_warn = true;
245 int blk_size = (dcacheInterface) ? dcacheInterface->getBlockSize() : 64;
246 // Only support block sizes of 64 atm.
247 assert(blk_size == 64);
248 uint8_t data[blk_size];
249 //assert(thread->copySrcAddr);
250 int offset = dest & (blk_size - 1);
251
252 // Make sure block doesn't span page
253 if (no_warn &&
254 (dest & PageMask) != ((dest + blk_size) & PageMask) &&
255 (dest >> 40) != 0xfffffc) {
256 no_warn = false;
257 warn("Copied block destination spans pages %x. ", dest);
258 }
259
260 memReq->reset(dest & ~(blk_size -1), blk_size);
261 // translate to physical address
262 Fault fault = thread->translateDataWriteReq(req);
263
264 if (fault == NoFault) {
265 Addr dest_addr = memReq->paddr + offset;
266 // Need to read straight from memory since we have more than 8 bytes.
267 memReq->paddr = thread->copySrcPhysAddr;
268 thread->mem->read(memReq, data);
269 memReq->paddr = dest_addr;
270 thread->mem->write(memReq, data);
271 if (dcacheInterface) {
272 memReq->cmd = Copy;
273 memReq->completionEvent = NULL;
274 memReq->paddr = thread->copySrcPhysAddr;
275 memReq->dest = dest_addr;
276 memReq->size = 64;
277 memReq->time = curTick;
278 memReq->flags &= ~INST_READ;
279 dcacheInterface->access(memReq);
280 }
281 }
282 else
283 assert(!fault->isAlignmentFault());
284
285 return fault;
286#else
287 panic("copy not implemented");
288 return NoFault;
289#endif
290}
291
292#if FULL_SYSTEM
293Addr
294BaseSimpleCPU::dbg_vtophys(Addr addr)
295{
296 return vtophys(tc, addr);
297}
298#endif // FULL_SYSTEM
299
300#if FULL_SYSTEM
301void
302BaseSimpleCPU::post_interrupt(int int_num, int index)
303{
304 BaseCPU::post_interrupt(int_num, index);
305
306 if (thread->status() == ThreadContext::Suspended) {
307 DPRINTF(Quiesce,"Suspended Processor awoke\n");
308 thread->activate();
309 }
310}
311#endif // FULL_SYSTEM
312
313void
314BaseSimpleCPU::checkForInterrupts()
315{
316#if FULL_SYSTEM
317 if (check_interrupts(tc)) {
318 Fault interrupt = interrupts.getInterrupt(tc);
319
320 if (interrupt != NoFault) {
321 interrupts.updateIntrInfo(tc);
322 interrupt->invoke(tc);
323 }
324 }
325#endif
326}
327
328
329Fault
330BaseSimpleCPU::setupFetchRequest(Request *req)
331{
332 // set up memory request for instruction fetch
333#if ISA_HAS_DELAY_SLOT
334 DPRINTF(Fetch,"Fetch: PC:%08p NPC:%08p NNPC:%08p\n",thread->readPC(),
335 thread->readNextPC(),thread->readNextNPC());
336#else
337 DPRINTF(Fetch,"Fetch: PC:%08p NPC:%08p",thread->readPC(),
338 thread->readNextPC());
339#endif
340
338 // This will generate a mask which aligns the pc on MachInst size
339 // boundaries. It won't work for non-power-of-two sized MachInsts, but
340 // it will work better than a hard coded mask.
341 const Addr PCMask = ~(sizeof(MachInst) - 1);
342 req->setVirt(0, thread->readPC() & PCMask, sizeof(MachInst), 0,
343 thread->readPC());
342 Addr fetchPC = thread->readPC() + fetchOffset;
343 req->setVirt(0, fetchPC & PCMask, sizeof(MachInst), 0, thread->readPC());
344
345 Fault fault = thread->translateInstReq(req);
346
347 return fault;
348}
349
350
351void
352BaseSimpleCPU::preExecute()
353{
354 // maintain $r0 semantics
355 thread->setIntReg(ZeroReg, 0);
356#if THE_ISA == ALPHA_ISA
357 thread->setFloatReg(ZeroReg, 0.0);
358#endif // ALPHA_ISA
359
360 // keep an instruction count
361 numInst++;
362 numInsts++;
363
364 thread->funcExeInst++;
365
366 // check for instruction-count-based events
367 comInstEventQueue[0]->serviceEvents(numInst);
368
369 // decode the instruction
370 inst = gtoh(inst);
371
372 //If we're not in the middle of a macro instruction
373 if (!curMacroStaticInst) {
374
375 StaticInstPtr instPtr = NULL;
376
377 //Predecode, ie bundle up an ExtMachInst
378 //This should go away once the constructor can be set up properly
379 predecoder.setTC(thread->getTC());
380 //If more fetch data is needed, pass it in.
381 const Addr PCMask = ~(sizeof(MachInst) - 1);
382 if(predecoder.needMoreBytes())
380 predecoder.moreBytes(thread->readPC(), 0, inst);
383 predecoder.moreBytes((thread->readPC() & PCMask) + fetchOffset,
384 0, inst);
385 else
386 predecoder.process();
383 //If an instruction is ready, decode it
384 if (predecoder.extMachInstReady())
387
388 //If an instruction is ready, decode it. Otherwise, we'll have to
389 //fetch beyond the MachInst at the current pc.
390 if (predecoder.extMachInstReady()) {
391#if THE_ISA == X86_ISA
392 thread->setNextPC(thread->readPC() + predecoder.getInstSize());
393#endif // X86_ISA
394 stayAtPC = false;
395 instPtr = StaticInst::decode(predecoder.getExtMachInst());
396 } else {
397 stayAtPC = true;
398 fetchOffset += sizeof(MachInst);
399 }
400
401 //If we decoded an instruction and it's microcoded, start pulling
402 //out micro ops
403 if (instPtr && instPtr->isMacroOp()) {
404 curMacroStaticInst = instPtr;
405 curStaticInst = curMacroStaticInst->
406 fetchMicroOp(thread->readMicroPC());
407 } else {
408 curStaticInst = instPtr;
409 }
410 } else {
411 //Read the next micro op from the macro op
412 curStaticInst = curMacroStaticInst->
413 fetchMicroOp(thread->readMicroPC());
414 }
415
416 //If we decoded an instruction this "tick", record information about it.
417 if(curStaticInst)
418 {
419 traceData = Trace::getInstRecord(curTick, tc, curStaticInst,
420 thread->readPC());
421
422 DPRINTF(Decode,"Decode: Decoded %s instruction: 0x%x\n",
423 curStaticInst->getName(), curStaticInst->machInst);
424
425#if FULL_SYSTEM
426 thread->setInst(inst);
427#endif // FULL_SYSTEM
428 }
429}
430
431void
432BaseSimpleCPU::postExecute()
433{
434#if FULL_SYSTEM
435 if (thread->profile) {
436 bool usermode = TheISA::inUserMode(tc);
437 thread->profilePC = usermode ? 1 : thread->readPC();
438 StaticInstPtr si(inst);
439 ProfileNode *node = thread->profile->consume(tc, si);
440 if (node)
441 thread->profileNode = node;
442 }
443#endif
444
445 if (curStaticInst->isMemRef()) {
446 numMemRefs++;
447 }
448
449 if (curStaticInst->isLoad()) {
450 ++numLoad;
451 comLoadEventQueue[0]->serviceEvents(numLoad);
452 }
453
454 traceFunctions(thread->readPC());
455
456 if (traceData) {
457 traceData->dump();
458 delete traceData;
459 traceData = NULL;
460 }
461}
462
463
464void
465BaseSimpleCPU::advancePC(Fault fault)
466{
467 //Since we're moving to a new pc, zero out the offset
468 fetchOffset = 0;
469 if (fault != NoFault) {
470 curMacroStaticInst = StaticInst::nullStaticInstPtr;
471 fault->invoke(tc);
472 thread->setMicroPC(0);
473 thread->setNextMicroPC(1);
458 } else if (predecoder.needMoreBytes()) {
474 } else {
475 //If we're at the last micro op for this instruction
476 if (curStaticInst && curStaticInst->isLastMicroOp()) {
477 //We should be working with a macro op
478 assert(curMacroStaticInst);
479 //Close out this macro op, and clean up the
480 //microcode state
481 curMacroStaticInst = StaticInst::nullStaticInstPtr;
482 thread->setMicroPC(0);
483 thread->setNextMicroPC(1);
484 }
485 //If we're still in a macro op
486 if (curMacroStaticInst) {
487 //Advance the micro pc
488 thread->setMicroPC(thread->readNextMicroPC());
489 //Advance the "next" micro pc. Note that there are no delay
490 //slots, and micro ops are "word" addressed.
491 thread->setNextMicroPC(thread->readNextMicroPC() + 1);
492 } else {
493 // go to the next instruction
494 thread->setPC(thread->readNextPC());
495 thread->setNextPC(thread->readNextNPC());
496 thread->setNextNPC(thread->readNextNPC() + sizeof(MachInst));
497 assert(thread->readNextPC() != thread->readNextNPC());
498 }
499 }
500
501#if FULL_SYSTEM
502 Addr oldpc;
503 do {
504 oldpc = thread->readPC();
505 system->pcEventQueue.service(tc);
506 } while (oldpc != thread->readPC());
507#endif
508}
509