-
Notifications
You must be signed in to change notification settings - Fork 2.2k
/
JitSafeMem.cpp
548 lines (465 loc) · 14.3 KB
/
JitSafeMem.cpp
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
// Copyright (c) 2012- PPSSPP Project.
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, version 2.0 or later versions.
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License 2.0 for more details.
// A copy of the GPL 2.0 should have been included with the program.
// If not, see http://www.gnu.org/licenses/
// Official git repository and contact information can be found at
// https://github.com/hrydgard/ppsspp and http://www.ppsspp.org/.
#include "ppsspp_config.h"
#if PPSSPP_ARCH(X86) || PPSSPP_ARCH(AMD64)
#include "Core/Config.h"
#include "Core/Debugger/Breakpoints.h"
#include "Core/MemMap.h"
#include "Core/MIPS/JitCommon/JitCommon.h"
#include "Core/MIPS/x86/Jit.h"
#include "Core/MIPS/x86/JitSafeMem.h"
#include "Core/System.h"
namespace MIPSComp
{
using namespace Gen;
using namespace X64JitConstants;
void JitMemCheck(u32 addr, int size, int isWrite)
{
// Should we skip this breakpoint?
if (CBreakPoints::CheckSkipFirst() == currentMIPS->pc)
return;
// Did we already hit one?
if (coreState != CORE_RUNNING && coreState != CORE_NEXTFRAME)
return;
CBreakPoints::ExecMemCheckJitBefore(addr, isWrite == 1, size, currentMIPS->pc);
}
void JitMemCheckCleanup()
{
CBreakPoints::ExecMemCheckJitCleanup();
}
JitSafeMem::JitSafeMem(Jit *jit, MIPSGPReg raddr, s32 offset, u32 alignMask)
: jit_(jit), raddr_(raddr), offset_(offset), needsCheck_(false), needsSkip_(false), alignMask_(alignMask)
{
// This makes it more instructions, so let's play it safe and say we need a far jump.
far_ = !g_Config.bIgnoreBadMemAccess || !CBreakPoints::GetMemChecks().empty();
// Mask out the kernel RAM bit, because we'll end up with a negative offset to MEMBASEREG.
if (jit_->gpr.IsImm(raddr_))
iaddr_ = (jit_->gpr.GetImm(raddr_) + offset_) & 0x7FFFFFFF;
else
iaddr_ = (u32) -1;
fast_ = g_Config.bFastMemory || raddr == MIPS_REG_SP;
// If raddr_ is going to get loaded soon, load it now for more optimal code.
// We assume that it was already locked.
const int LOOKAHEAD_OPS = 3;
if (!jit_->gpr.R(raddr_).IsImm() && MIPSAnalyst::IsRegisterUsed(raddr_, jit_->GetCompilerPC() + 4, LOOKAHEAD_OPS))
jit_->gpr.MapReg(raddr_, true, false);
}
void JitSafeMem::SetFar()
{
_dbg_assert_msg_(JIT, !needsSkip_, "Sorry, you need to call SetFar() earlier.");
far_ = true;
}
bool JitSafeMem::PrepareWrite(OpArg &dest, int size)
{
size_ = size;
// If it's an immediate, we can do the write if valid.
if (iaddr_ != (u32) -1)
{
if (ImmValid())
{
MemCheckImm(MEM_WRITE);
u32 addr = (iaddr_ & alignMask_);
#ifdef MASKED_PSP_MEMORY
addr &= Memory::MEMVIEW32_MASK;
#endif
#if PPSSPP_ARCH(32BIT)
dest = M(Memory::base + addr);
#else
dest = MDisp(MEMBASEREG, addr);
#endif
return true;
}
else
return false;
}
// Otherwise, we always can do the write (conditionally.)
else
dest = PrepareMemoryOpArg(MEM_WRITE);
return true;
}
bool JitSafeMem::PrepareRead(OpArg &src, int size)
{
size_ = size;
if (iaddr_ != (u32) -1)
{
if (ImmValid())
{
MemCheckImm(MEM_READ);
u32 addr = (iaddr_ & alignMask_);
#ifdef MASKED_PSP_MEMORY
addr &= Memory::MEMVIEW32_MASK;
#endif
#if PPSSPP_ARCH(32BIT)
src = M(Memory::base + addr);
#else
src = MDisp(MEMBASEREG, addr);
#endif
return true;
}
else
return false;
}
else
src = PrepareMemoryOpArg(MEM_READ);
return true;
}
OpArg JitSafeMem::NextFastAddress(int suboffset)
{
if (iaddr_ != (u32) -1)
{
u32 addr = (iaddr_ + suboffset) & alignMask_;
#ifdef MASKED_PSP_MEMORY
addr &= Memory::MEMVIEW32_MASK;
#endif
#if PPSSPP_ARCH(32BIT)
return M(Memory::base + addr);
#else
return MDisp(MEMBASEREG, addr);
#endif
}
_dbg_assert_msg_(JIT, (suboffset & alignMask_) == suboffset, "suboffset must be aligned");
#if PPSSPP_ARCH(32BIT)
return MDisp(xaddr_, (u32) Memory::base + offset_ + suboffset);
#else
return MComplex(MEMBASEREG, xaddr_, SCALE_1, offset_ + suboffset);
#endif
}
OpArg JitSafeMem::PrepareMemoryOpArg(MemoryOpType type)
{
// We may not even need to move into EAX as a temporary.
bool needTemp = alignMask_ != 0xFFFFFFFF;
#ifdef MASKED_PSP_MEMORY
bool needMask = true; // raddr_ != MIPS_REG_SP; // Commented out this speedhack due to low impact
// We always mask on 32 bit in fast memory mode.
needTemp = needTemp || (fast_ && needMask);
#endif
if (jit_->gpr.R(raddr_).IsSimpleReg() && !needTemp)
{
jit_->gpr.MapReg(raddr_, true, false);
xaddr_ = jit_->gpr.RX(raddr_);
}
else
{
jit_->MOV(32, R(EAX), jit_->gpr.R(raddr_));
xaddr_ = EAX;
}
MemCheckAsm(type);
if (!fast_)
{
// Is it in physical ram?
jit_->CMP(32, R(xaddr_), Imm32(PSP_GetKernelMemoryBase() - offset_));
tooLow_ = jit_->J_CC(CC_B);
jit_->CMP(32, R(xaddr_), Imm32(PSP_GetUserMemoryEnd() - offset_ - (size_ - 1)));
tooHigh_ = jit_->J_CC(CC_AE);
// We may need to jump back up here.
safe_ = jit_->GetCodePtr();
}
else
{
#ifdef MASKED_PSP_MEMORY
if (needMask) {
jit_->AND(32, R(EAX), Imm32(Memory::MEMVIEW32_MASK));
}
#endif
}
// TODO: This could be more optimal, but the common case is that we want xaddr_ not to include offset_.
// Since we need to align them after add, we add and subtract.
if (alignMask_ != 0xFFFFFFFF)
{
jit_->ADD(32, R(xaddr_), Imm32(offset_));
jit_->AND(32, R(xaddr_), Imm32(alignMask_));
jit_->SUB(32, R(xaddr_), Imm32(offset_));
}
#if PPSSPP_ARCH(32BIT)
return MDisp(xaddr_, (u32) Memory::base + offset_);
#else
return MComplex(MEMBASEREG, xaddr_, SCALE_1, offset_);
#endif
}
void JitSafeMem::PrepareSlowAccess()
{
// Skip the fast path (which the caller wrote just now.)
skip_ = jit_->J(far_);
needsSkip_ = true;
jit_->SetJumpTarget(tooLow_);
jit_->SetJumpTarget(tooHigh_);
// Might also be the scratchpad.
jit_->CMP(32, R(xaddr_), Imm32(PSP_GetScratchpadMemoryBase() - offset_));
FixupBranch tooLow = jit_->J_CC(CC_B);
jit_->CMP(32, R(xaddr_), Imm32(PSP_GetScratchpadMemoryEnd() - offset_ - (size_ - 1)));
jit_->J_CC(CC_B, safe_);
jit_->SetJumpTarget(tooLow);
}
bool JitSafeMem::PrepareSlowWrite()
{
// If it's immediate, we only need a slow write on invalid.
if (iaddr_ != (u32) -1)
return !fast_ && !ImmValid();
if (!fast_)
{
PrepareSlowAccess();
return true;
}
else
return false;
}
void JitSafeMem::DoSlowWrite(const void *safeFunc, const OpArg& src, int suboffset)
{
if (iaddr_ != (u32) -1)
jit_->MOV(32, R(EAX), Imm32((iaddr_ + suboffset) & alignMask_));
else
{
jit_->LEA(32, EAX, MDisp(xaddr_, offset_ + suboffset));
if (alignMask_ != 0xFFFFFFFF)
jit_->AND(32, R(EAX), Imm32(alignMask_));
}
#if PPSSPP_ARCH(32BIT)
jit_->PUSH(EDX);
#endif
if (!src.IsSimpleReg(EDX)) {
jit_->MOV(32, R(EDX), src);
}
if (!g_Config.bIgnoreBadMemAccess) {
jit_->MOV(32, MIPSSTATE_VAR(pc), Imm32(jit_->GetCompilerPC()));
}
// This is a special jit-ABI'd function.
jit_->CALL(safeFunc);
#if PPSSPP_ARCH(32BIT)
jit_->POP(EDX);
#endif
needsCheck_ = true;
}
bool JitSafeMem::PrepareSlowRead(const void *safeFunc)
{
if (!fast_)
{
if (iaddr_ != (u32) -1)
{
// No slow read necessary.
if (ImmValid())
return false;
jit_->MOV(32, R(EAX), Imm32(iaddr_ & alignMask_));
}
else
{
PrepareSlowAccess();
jit_->LEA(32, EAX, MDisp(xaddr_, offset_));
if (alignMask_ != 0xFFFFFFFF)
jit_->AND(32, R(EAX), Imm32(alignMask_));
}
if (!g_Config.bIgnoreBadMemAccess) {
jit_->MOV(32, MIPSSTATE_VAR(pc), Imm32(jit_->GetCompilerPC()));
}
// This is a special jit-ABI'd function.
jit_->CALL(safeFunc);
needsCheck_ = true;
return true;
}
else
return false;
}
void JitSafeMem::NextSlowRead(const void *safeFunc, int suboffset)
{
_dbg_assert_msg_(JIT, !fast_, "NextSlowRead() called in fast memory mode?");
// For simplicity, do nothing for 0. We already read in PrepareSlowRead().
if (suboffset == 0)
return;
if (jit_->gpr.IsImm(raddr_))
{
_dbg_assert_msg_(JIT, !Memory::IsValidAddress(iaddr_ + suboffset), "NextSlowRead() for an invalid immediate address?");
jit_->MOV(32, R(EAX), Imm32((iaddr_ + suboffset) & alignMask_));
}
// For GPR, if xaddr_ was the dest register, this will be wrong. Don't use in GPR.
else
{
jit_->LEA(32, EAX, MDisp(xaddr_, offset_ + suboffset));
if (alignMask_ != 0xFFFFFFFF)
jit_->AND(32, R(EAX), Imm32(alignMask_));
}
if (!g_Config.bIgnoreBadMemAccess) {
jit_->MOV(32, MIPSSTATE_VAR(pc), Imm32(jit_->GetCompilerPC()));
}
// This is a special jit-ABI'd function.
jit_->CALL(safeFunc);
}
bool JitSafeMem::ImmValid()
{
return iaddr_ != (u32) -1 && Memory::IsValidAddress(iaddr_) && Memory::IsValidAddress(iaddr_ + size_ - 1);
}
void JitSafeMem::Finish()
{
// Memory::Read_U32/etc. may have tripped coreState.
if (needsCheck_ && !g_Config.bIgnoreBadMemAccess)
jit_->js.afterOp |= JitState::AFTER_CORE_STATE;
if (needsSkip_)
jit_->SetJumpTarget(skip_);
for (auto it = skipChecks_.begin(), end = skipChecks_.end(); it != end; ++it)
jit_->SetJumpTarget(*it);
}
void JitSafeMem::MemCheckImm(MemoryOpType type)
{
MemCheck *check = CBreakPoints::GetMemCheck(iaddr_, size_);
if (check)
{
if (!(check->cond & MEMCHECK_READ) && type == MEM_READ)
return;
if (!(check->cond & MEMCHECK_WRITE) && type == MEM_WRITE)
return;
jit_->MOV(32, MIPSSTATE_VAR(pc), Imm32(jit_->GetCompilerPC()));
jit_->CallProtectedFunction(&JitMemCheck, iaddr_, size_, type == MEM_WRITE ? 1 : 0);
// CORE_RUNNING is <= CORE_NEXTFRAME.
jit_->CMP(32, M(&coreState), Imm32(CORE_NEXTFRAME));
skipChecks_.push_back(jit_->J_CC(CC_G, true));
jit_->js.afterOp |= JitState::AFTER_CORE_STATE | JitState::AFTER_REWIND_PC_BAD_STATE | JitState::AFTER_MEMCHECK_CLEANUP;
}
}
void JitSafeMem::MemCheckAsm(MemoryOpType type)
{
const auto memchecks = CBreakPoints::GetMemCheckRanges();
bool possible = false;
for (auto it = memchecks.begin(), end = memchecks.end(); it != end; ++it)
{
if (!(it->cond & MEMCHECK_READ) && type == MEM_READ)
continue;
if (!(it->cond & MEMCHECK_WRITE) && type == MEM_WRITE)
continue;
possible = true;
FixupBranch skipNext, skipNextRange;
if (it->end != 0)
{
jit_->CMP(32, R(xaddr_), Imm32(it->start - offset_ - size_));
skipNext = jit_->J_CC(CC_BE);
jit_->CMP(32, R(xaddr_), Imm32(it->end - offset_));
skipNextRange = jit_->J_CC(CC_AE);
}
else
{
jit_->CMP(32, R(xaddr_), Imm32(it->start - offset_));
skipNext = jit_->J_CC(CC_NE);
}
// Keep the stack 16-byte aligned, just PUSH/POP 4 times.
for (int i = 0; i < 4; ++i)
jit_->PUSH(xaddr_);
jit_->MOV(32, MIPSSTATE_VAR(pc), Imm32(jit_->GetCompilerPC()));
jit_->ADD(32, R(xaddr_), Imm32(offset_));
jit_->CallProtectedFunction(&JitMemCheck, R(xaddr_), size_, type == MEM_WRITE ? 1 : 0);
for (int i = 0; i < 4; ++i)
jit_->POP(xaddr_);
jit_->SetJumpTarget(skipNext);
if (it->end != 0)
jit_->SetJumpTarget(skipNextRange);
}
if (possible)
{
// CORE_RUNNING is <= CORE_NEXTFRAME.
jit_->CMP(32, M(&coreState), Imm32(CORE_NEXTFRAME));
skipChecks_.push_back(jit_->J_CC(CC_G, true));
jit_->js.afterOp |= JitState::AFTER_CORE_STATE | JitState::AFTER_REWIND_PC_BAD_STATE | JitState::AFTER_MEMCHECK_CLEANUP;
}
}
static const int FUNCS_ARENA_SIZE = 512 * 1024;
void JitSafeMemFuncs::Init(ThunkManager *thunks) {
using namespace Gen;
AllocCodeSpace(FUNCS_ARENA_SIZE);
thunks_ = thunks;
readU32 = GetCodePtr();
CreateReadFunc(32, (const void *)&Memory::Read_U32);
readU16 = GetCodePtr();
CreateReadFunc(16, (const void *)&Memory::Read_U16);
readU8 = GetCodePtr();
CreateReadFunc(8, (const void *)&Memory::Read_U8);
writeU32 = GetCodePtr();
CreateWriteFunc(32, (const void *)&Memory::Write_U32);
writeU16 = GetCodePtr();
CreateWriteFunc(16, (const void *)&Memory::Write_U16);
writeU8 = GetCodePtr();
CreateWriteFunc(8, (const void *)&Memory::Write_U8);
}
void JitSafeMemFuncs::Shutdown() {
ResetCodePtr(0);
FreeCodeSpace();
}
// Mini ABI:
// Read funcs take address in EAX, return in RAX.
// Write funcs take address in EAX, data in RDX.
// On x86-32, Write funcs also have an extra 4 bytes on the stack.
void JitSafeMemFuncs::CreateReadFunc(int bits, const void *fallbackFunc) {
CheckDirectEAX();
// Since we were CALLed, we need to align the stack before calling C++.
#if PPSSPP_ARCH(32BIT)
SUB(32, R(ESP), Imm8(16 - 4));
ABI_CallFunctionA(thunks_->ProtectFunction(fallbackFunc, 1), R(EAX));
ADD(32, R(ESP), Imm8(16 - 4));
#else
SUB(64, R(RSP), Imm8(0x28));
ABI_CallFunctionA(thunks_->ProtectFunction(fallbackFunc, 1), R(EAX));
ADD(64, R(RSP), Imm8(0x28));
#endif
RET();
StartDirectAccess();
#if PPSSPP_ARCH(32BIT)
MOVZX(32, bits, EAX, MDisp(EAX, (u32)Memory::base));
#else
MOVZX(32, bits, EAX, MRegSum(MEMBASEREG, EAX));
#endif
RET();
}
void JitSafeMemFuncs::CreateWriteFunc(int bits, const void *fallbackFunc) {
CheckDirectEAX();
// Since we were CALLed, we need to align the stack before calling C++.
#if PPSSPP_ARCH(32BIT)
// 4 for return, 4 for saved reg on stack.
SUB(32, R(ESP), Imm8(16 - 4 - 4));
ABI_CallFunctionAA(thunks_->ProtectFunction(fallbackFunc, 2), R(EDX), R(EAX));
ADD(32, R(ESP), Imm8(16 - 4 - 4));
#else
SUB(64, R(RSP), Imm8(0x28));
ABI_CallFunctionAA(thunks_->ProtectFunction(fallbackFunc, 2), R(EDX), R(EAX));
ADD(64, R(RSP), Imm8(0x28));
#endif
RET();
StartDirectAccess();
#if PPSSPP_ARCH(32BIT)
MOV(bits, MDisp(EAX, (u32)Memory::base), R(EDX));
#else
MOV(bits, MRegSum(MEMBASEREG, EAX), R(EDX));
#endif
RET();
}
void JitSafeMemFuncs::CheckDirectEAX() {
// Clear any cache/kernel bits.
AND(32, R(EAX), Imm32(0x3FFFFFFF));
CMP(32, R(EAX), Imm32(PSP_GetUserMemoryEnd()));
FixupBranch tooHighRAM = J_CC(CC_AE);
CMP(32, R(EAX), Imm32(PSP_GetKernelMemoryBase()));
skips_.push_back(J_CC(CC_AE));
CMP(32, R(EAX), Imm32(PSP_GetVidMemEnd()));
FixupBranch tooHighVid = J_CC(CC_AE);
CMP(32, R(EAX), Imm32(PSP_GetVidMemBase()));
skips_.push_back(J_CC(CC_AE));
CMP(32, R(EAX), Imm32(PSP_GetScratchpadMemoryEnd()));
FixupBranch tooHighScratch = J_CC(CC_AE);
CMP(32, R(EAX), Imm32(PSP_GetScratchpadMemoryBase()));
skips_.push_back(J_CC(CC_AE));
SetJumpTarget(tooHighRAM);
SetJumpTarget(tooHighVid);
SetJumpTarget(tooHighScratch);
}
void JitSafeMemFuncs::StartDirectAccess() {
for (auto it = skips_.begin(), end = skips_.end(); it != end; ++it) {
SetJumpTarget(*it);
}
skips_.clear();
}
};
#endif // PPSSPP_ARCH(X86) || PPSSPP_ARCH(AMD64)