diff options
| author | LDj3SNuD <35856442+LDj3SNuD@users.noreply.github.com> | 2020-06-16 20:28:02 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2020-06-16 20:28:02 +0200 |
| commit | 5e724cf24e3d696b95be859c055a617e5d37bf80 (patch) | |
| tree | bda92e3f1618cfad804f31ea3b00d7569501bc96 /ARMeilleure/CodeGen | |
| parent | fa286d353595aa91ad0bda6671d8d87e0700846f (diff) | |
Add Profiled Persistent Translation Cache. (#769)
* Delete DelegateTypes.cs
* Delete DelegateCache.cs
* Add files via upload
* Update Horizon.cs
* Update Program.cs
* Update MainWindow.cs
* Update Aot.cs
* Update RelocEntry.cs
* Update Translator.cs
* Update MemoryManager.cs
* Update InstEmitMemoryHelper.cs
* Update Delegates.cs
* Nit.
* Nit.
* Nit.
* 10 fewer MSIL bytes for us
* Add comment. Nits.
* Update Translator.cs
* Update Aot.cs
* Nits.
* Opt..
* Opt..
* Opt..
* Opt..
* Allow to change compression level.
* Update MemoryManager.cs
* Update Translator.cs
* Manage corner cases during the save phase. Nits.
* Update Aot.cs
* Translator response tweak for Aot disabled. Nit.
* Nit.
* Nits.
* Create DelegateHelpers.cs
* Update Delegates.cs
* Nit.
* Nit.
* Nits.
* Fix due to #784.
* Fixes due to #757 & #841.
* Fix due to #846.
* Fix due to #847.
* Use MethodInfo for managed method calls.
Use IR methods instead of managed methods about Max/Min (S/U).
Follow-ups & Nits.
* Add missing exception messages.
Reintroduce slow path for Fmov_Vi.
Implement slow path for Fmov_Si.
* Switch to the new folder structure.
Nits.
* Impl. index-based relocation information. Impl. cache file version field.
* Nit.
* Address gdkchan comments.
Mainly:
- fixed cache file corruption issue on exit; - exposed a way to disable AOT on the GUI.
* Address AcK77 comment.
* Address Thealexbarney, jduncanator & emmauss comments.
Header magic, CpuId (FI) & Aot -> Ptc.
* Adaptation to the new application reloading system.
Improvements to the call system of managed methods.
Follow-ups.
Nits.
* Get the same boot times as on master when PTC is disabled.
* Profiled Aot.
* A32 support (#897).
* #975 support (1 of 2).
* #975 support (2 of 2).
* Rebase fix & nits.
* Some fixes and nits (still one bug left).
* One fix & nits.
* Tests fix (by gdk) & nits.
* Support translations not only in high quality and rejit.
Nits.
* Added possibility to skip translations and continue execution, using `ESC` key.
* Update SettingsWindow.cs
* Update GLRenderer.cs
* Update Ptc.cs
* Disabled Profiled PTC by default as requested in the past by gdk.
* Fix rejit bug. Increased number of parallel translations. Add stack unwinding stuffs support (1 of 2).
Nits.
* Add stack unwinding stuffs support (2 of 2). Tuned number of parallel translations.
* Restored the ability to assemble jumps with 8-bit offset when Profiled PTC is disabled or during profiling.
Modifications due to rebase.
Nits.
* Limited profiling of the functions to be translated to the addresses belonging to the range of static objects only.
* Nits.
* Nits.
* Update Delegates.cs
* Nit.
* Update InstEmitSimdArithmetic.cs
* Address riperiperi comments.
* Fixed the issue of unjustifiably longer boot times at the second boot than at the first boot, measured at the same time or reference point and with the same number of translated functions.
* Implemented a simple redundant load/save mechanism.
Halved the value of Decoder.MaxInstsPerFunction more appropriate for the current performance of the Translator.
Replaced by Logger.PrintError to Logger.PrintDebug in TexturePool.cs about the supposed invalid texture format to avoid the spawn of the log.
Nits.
* Nit.
Improved Logger.PrintError in TexturePool.cs to avoid log spawn.
Added missing code for FZ handling (in output) for fp max/min instructions (slow paths).
* Add configuration migration for PTC
Co-authored-by: Thog <me@thog.eu>
Diffstat (limited to 'ARMeilleure/CodeGen')
| -rw-r--r-- | ARMeilleure/CodeGen/Optimizations/ConstantFolding.cs | 8 | ||||
| -rw-r--r-- | ARMeilleure/CodeGen/Unwinding/UnwindPseudoOp.cs | 11 | ||||
| -rw-r--r-- | ARMeilleure/CodeGen/Unwinding/UnwindPseudoOperation.cs | 11 | ||||
| -rw-r--r-- | ARMeilleure/CodeGen/X86/Assembler.cs | 28 | ||||
| -rw-r--r-- | ARMeilleure/CodeGen/X86/CodeGenContext.cs | 132 | ||||
| -rw-r--r-- | ARMeilleure/CodeGen/X86/CodeGenerator.cs | 7 |
6 files changed, 125 insertions, 72 deletions
diff --git a/ARMeilleure/CodeGen/Optimizations/ConstantFolding.cs b/ARMeilleure/CodeGen/Optimizations/ConstantFolding.cs index 84eedee0..eff53217 100644 --- a/ARMeilleure/CodeGen/Optimizations/ConstantFolding.cs +++ b/ARMeilleure/CodeGen/Optimizations/ConstantFolding.cs @@ -14,7 +14,7 @@ namespace ARMeilleure.CodeGen.Optimizations return; } - if (!AreAllSourcesConstant(operation)) + if (!AreAllSourcesConstantAndCFEnabled(operation)) { return; } @@ -212,11 +212,13 @@ namespace ARMeilleure.CodeGen.Optimizations } } - private static bool AreAllSourcesConstant(Operation operation) + private static bool AreAllSourcesConstantAndCFEnabled(Operation operation) { for (int index = 0; index < operation.SourcesCount; index++) { - if (operation.GetSource(index).Kind != OperandKind.Constant) + Operand srcOp = operation.GetSource(index); + + if (srcOp.Kind != OperandKind.Constant || srcOp.DisableCF) { return false; } diff --git a/ARMeilleure/CodeGen/Unwinding/UnwindPseudoOp.cs b/ARMeilleure/CodeGen/Unwinding/UnwindPseudoOp.cs new file mode 100644 index 00000000..4a8288a2 --- /dev/null +++ b/ARMeilleure/CodeGen/Unwinding/UnwindPseudoOp.cs @@ -0,0 +1,11 @@ +namespace ARMeilleure.CodeGen.Unwinding +{ + enum UnwindPseudoOp + { + PushReg = 0, + SetFrame = 1, + AllocStack = 2, + SaveReg = 3, + SaveXmm128 = 4 + } +}
\ No newline at end of file diff --git a/ARMeilleure/CodeGen/Unwinding/UnwindPseudoOperation.cs b/ARMeilleure/CodeGen/Unwinding/UnwindPseudoOperation.cs deleted file mode 100644 index 44ed23f5..00000000 --- a/ARMeilleure/CodeGen/Unwinding/UnwindPseudoOperation.cs +++ /dev/null @@ -1,11 +0,0 @@ -namespace ARMeilleure.CodeGen.Unwinding -{ - enum UnwindPseudoOp - { - PushReg, - SetFrame, - AllocStack, - SaveReg, - SaveXmm128 - } -}
\ No newline at end of file diff --git a/ARMeilleure/CodeGen/X86/Assembler.cs b/ARMeilleure/CodeGen/X86/Assembler.cs index 5ad54289..99df3cb5 100644 --- a/ARMeilleure/CodeGen/X86/Assembler.cs +++ b/ARMeilleure/CodeGen/X86/Assembler.cs @@ -1,4 +1,5 @@ using ARMeilleure.IntermediateRepresentation; +using ARMeilleure.Translation.PTC; using System; using System.Diagnostics; using System.IO; @@ -64,6 +65,9 @@ namespace ARMeilleure.CodeGen.X86 private Stream _stream; + private PtcInfo _ptcInfo; + private bool _ptcDisabled; + static Assembler() { _instTable = new InstructionInfo[(int)X86Instruction.Count]; @@ -273,9 +277,12 @@ namespace ARMeilleure.CodeGen.X86 _instTable[(int)inst] = info; } - public Assembler(Stream stream) + public Assembler(Stream stream, PtcInfo ptcInfo = null) { _stream = stream; + + _ptcInfo = ptcInfo; + _ptcDisabled = ptcInfo == null; } public void Add(Operand dest, Operand source, OperandType type) @@ -456,7 +463,7 @@ namespace ARMeilleure.CodeGen.X86 public void Jcc(X86Condition condition, long offset) { - if (ConstFitsOnS8(offset)) + if (_ptcDisabled && ConstFitsOnS8(offset)) { WriteByte((byte)(0x70 | (int)condition)); @@ -477,7 +484,7 @@ namespace ARMeilleure.CodeGen.X86 public void Jmp(long offset) { - if (ConstFitsOnS8(offset)) + if (_ptcDisabled && ConstFitsOnS8(offset)) { WriteByte(0xeb); @@ -915,6 +922,8 @@ namespace ARMeilleure.CodeGen.X86 } else if (dest != null && dest.Kind == OperandKind.Register && info.OpRImm64 != BadOp) { + int? index = source.PtcIndex; + int rexPrefix = GetRexPrefix(dest, source, type, rrm: false); if (rexPrefix != 0) @@ -924,6 +933,11 @@ namespace ARMeilleure.CodeGen.X86 WriteByte((byte)(info.OpRImm64 + (dest.GetRegister().Index & 0b111))); + if (_ptcInfo != null && index != null) + { + _ptcInfo.WriteRelocEntry(new RelocEntry((int)_stream.Position, (int)index)); + } + WriteUInt64(imm); } else @@ -1316,9 +1330,9 @@ namespace ARMeilleure.CodeGen.X86 return ConstFitsOnS32(value); } - public static int GetJccLength(long offset) + public static int GetJccLength(long offset, bool ptcDisabled = true) { - if (ConstFitsOnS8(offset < 0 ? offset - 2 : offset)) + if (ptcDisabled && ConstFitsOnS8(offset < 0 ? offset - 2 : offset)) { return 2; } @@ -1332,9 +1346,9 @@ namespace ARMeilleure.CodeGen.X86 } } - public static int GetJmpLength(long offset) + public static int GetJmpLength(long offset, bool ptcDisabled = true) { - if (ConstFitsOnS8(offset < 0 ? offset - 2 : offset)) + if (ptcDisabled && ConstFitsOnS8(offset < 0 ? offset - 2 : offset)) { return 2; } diff --git a/ARMeilleure/CodeGen/X86/CodeGenContext.cs b/ARMeilleure/CodeGen/X86/CodeGenContext.cs index d719b516..da147cca 100644 --- a/ARMeilleure/CodeGen/X86/CodeGenContext.cs +++ b/ARMeilleure/CodeGen/X86/CodeGenContext.cs @@ -1,6 +1,7 @@ using ARMeilleure.CodeGen.RegisterAllocators; using ARMeilleure.Common; using ARMeilleure.IntermediateRepresentation; +using ARMeilleure.Translation.PTC; using System.Collections.Generic; using System.Diagnostics; using System.IO; @@ -13,6 +14,9 @@ namespace ARMeilleure.CodeGen.X86 private Stream _stream; + private PtcInfo _ptcInfo; + private bool _ptcDisabled; + public int StreamOffset => (int)_stream.Length; public AllocationResult AllocResult { get; } @@ -40,7 +44,7 @@ namespace ARMeilleure.CodeGen.X86 public int InstSize { get; set; } - public Jump(BasicBlock target, long jumpPosition) + public Jump(BasicBlock target, long jumpPosition, int instSize = 0) { IsConditional = false; Condition = 0; @@ -49,10 +53,10 @@ namespace ARMeilleure.CodeGen.X86 RelativeOffset = 0; - InstSize = 0; + InstSize = instSize; } - public Jump(X86Condition condition, BasicBlock target, long jumpPosition) + public Jump(X86Condition condition, BasicBlock target, long jumpPosition, int instSize = 0) { IsConditional = true; Condition = condition; @@ -61,7 +65,7 @@ namespace ARMeilleure.CodeGen.X86 RelativeOffset = 0; - InstSize = 0; + InstSize = instSize; } } @@ -72,13 +76,13 @@ namespace ARMeilleure.CodeGen.X86 private long _jNearPosition; private int _jNearLength; - public CodeGenContext(Stream stream, AllocationResult allocResult, int maxCallArgs, int blocksCount) + public CodeGenContext(Stream stream, AllocationResult allocResult, int maxCallArgs, int blocksCount, PtcInfo ptcInfo = null) { _stream = stream; AllocResult = allocResult; - Assembler = new Assembler(stream); + Assembler = new Assembler(stream, ptcInfo); CallArgsRegionSize = GetCallArgsRegionSize(allocResult, maxCallArgs, out int xmmSaveRegionSize); XmmSaveRegionSize = xmmSaveRegionSize; @@ -86,6 +90,9 @@ namespace ARMeilleure.CodeGen.X86 _blockOffsets = new long[blocksCount]; _jumps = new List<Jump>(); + + _ptcInfo = ptcInfo; + _ptcDisabled = ptcInfo == null; } private int GetCallArgsRegionSize(AllocationResult allocResult, int maxCallArgs, out int xmmSaveRegionSize) @@ -136,23 +143,41 @@ namespace ARMeilleure.CodeGen.X86 public void JumpTo(BasicBlock target) { - _jumps.Add(new Jump(target, _stream.Position)); + if (_ptcDisabled) + { + _jumps.Add(new Jump(target, _stream.Position)); - WritePadding(ReservedBytesForJump); + WritePadding(ReservedBytesForJump); + } + else + { + _jumps.Add(new Jump(target, _stream.Position, 5)); + + WritePadding(5); + } } public void JumpTo(X86Condition condition, BasicBlock target) { - _jumps.Add(new Jump(condition, target, _stream.Position)); + if (_ptcDisabled) + { + _jumps.Add(new Jump(condition, target, _stream.Position)); + + WritePadding(ReservedBytesForJump); + } + else + { + _jumps.Add(new Jump(condition, target, _stream.Position, 6)); - WritePadding(ReservedBytesForJump); + WritePadding(6); + } } public void JumpToNear(X86Condition condition) { _jNearCondition = condition; _jNearPosition = _stream.Position; - _jNearLength = Assembler.GetJccLength(0); + _jNearLength = Assembler.GetJccLength(0, _ptcDisabled); _stream.Seek(_jNearLength, SeekOrigin.Current); } @@ -165,7 +190,7 @@ namespace ARMeilleure.CodeGen.X86 long offset = currentPosition - (_jNearPosition + _jNearLength); - Debug.Assert(_jNearLength == Assembler.GetJccLength(offset), "Relative offset doesn't fit on near jump."); + Debug.Assert(_jNearLength == Assembler.GetJccLength(offset, _ptcDisabled), "Relative offset doesn't fit on near jump."); Assembler.Jcc(_jNearCondition, offset); @@ -197,56 +222,63 @@ namespace ARMeilleure.CodeGen.X86 long offset = jumpTarget - jump.JumpPosition; - if (offset < 0) + if (_ptcDisabled) { - for (int index2 = index - 1; index2 >= 0; index2--) + if (offset < 0) { - Jump jump2 = _jumps[index2]; - - if (jump2.JumpPosition < jumpTarget) + for (int index2 = index - 1; index2 >= 0; index2--) { - break; - } + Jump jump2 = _jumps[index2]; - offset -= jump2.InstSize - ReservedBytesForJump; + if (jump2.JumpPosition < jumpTarget) + { + break; + } + + offset -= jump2.InstSize - ReservedBytesForJump; + } } - } - else - { - for (int index2 = index + 1; index2 < _jumps.Count; index2++) + else { - Jump jump2 = _jumps[index2]; - - if (jump2.JumpPosition >= jumpTarget) + for (int index2 = index + 1; index2 < _jumps.Count; index2++) { - break; + Jump jump2 = _jumps[index2]; + + if (jump2.JumpPosition >= jumpTarget) + { + break; + } + + offset += jump2.InstSize - ReservedBytesForJump; } - offset += jump2.InstSize - ReservedBytesForJump; + offset -= ReservedBytesForJump; } - offset -= ReservedBytesForJump; - } + if (jump.IsConditional) + { + jump.InstSize = Assembler.GetJccLength(offset); + } + else + { + jump.InstSize = Assembler.GetJmpLength(offset); + } - if (jump.IsConditional) - { - jump.InstSize = Assembler.GetJccLength(offset); + // The jump is relative to the next instruction, not the current one. + // Since we didn't know the next instruction address when calculating + // the offset (as the size of the current jump instruction was not known), + // we now need to compensate the offset with the jump instruction size. + // It's also worth noting that: + // - This is only needed for backward jumps. + // - The GetJmpLength and GetJccLength also compensates the offset + // internally when computing the jump instruction size. + if (offset < 0) + { + offset -= jump.InstSize; + } } else { - jump.InstSize = Assembler.GetJmpLength(offset); - } - - // The jump is relative to the next instruction, not the current one. - // Since we didn't know the next instruction address when calculating - // the offset (as the size of the current jump instruction was not know), - // we now need to compensate the offset with the jump instruction size. - // It's also worth to note that: - // - This is only needed for backward jumps. - // - The GetJmpLength and GetJccLength also compensates the offset - // internally when computing the jump instruction size. - if (offset < 0) - { offset -= jump.InstSize; } @@ -267,7 +299,7 @@ namespace ARMeilleure.CodeGen.X86 using (MemoryStream codeStream = new MemoryStream()) { - Assembler assembler = new Assembler(codeStream); + Assembler assembler = new Assembler(codeStream, _ptcInfo); byte[] buffer; @@ -278,7 +310,7 @@ namespace ARMeilleure.CodeGen.X86 buffer = new byte[jump.JumpPosition - _stream.Position]; _stream.Read(buffer, 0, buffer.Length); - _stream.Seek(ReservedBytesForJump, SeekOrigin.Current); + _stream.Seek(_ptcDisabled ? ReservedBytesForJump : jump.InstSize, SeekOrigin.Current); codeStream.Write(buffer); @@ -298,6 +330,8 @@ namespace ARMeilleure.CodeGen.X86 codeStream.Write(buffer); + _ptcInfo?.WriteCode(codeStream); + return codeStream.ToArray(); } } diff --git a/ARMeilleure/CodeGen/X86/CodeGenerator.cs b/ARMeilleure/CodeGen/X86/CodeGenerator.cs index 0faba6dd..e7e7553e 100644 --- a/ARMeilleure/CodeGen/X86/CodeGenerator.cs +++ b/ARMeilleure/CodeGen/X86/CodeGenerator.cs @@ -5,6 +5,7 @@ using ARMeilleure.Common; using ARMeilleure.Diagnostics; using ARMeilleure.IntermediateRepresentation; using ARMeilleure.Translation; +using ARMeilleure.Translation.PTC; using System; using System.Collections.Generic; using System.Diagnostics; @@ -100,7 +101,7 @@ namespace ARMeilleure.CodeGen.X86 _instTable[(int)inst] = func; } - public static CompiledFunction Generate(CompilerContext cctx) + public static CompiledFunction Generate(CompilerContext cctx, PtcInfo ptcInfo = null) { ControlFlowGraph cfg = cctx.Cfg; @@ -158,10 +159,12 @@ namespace ARMeilleure.CodeGen.X86 using (MemoryStream stream = new MemoryStream()) { - CodeGenContext context = new CodeGenContext(stream, allocResult, maxCallArgs, cfg.Blocks.Count); + CodeGenContext context = new CodeGenContext(stream, allocResult, maxCallArgs, cfg.Blocks.Count, ptcInfo); UnwindInfo unwindInfo = WritePrologue(context); + ptcInfo?.WriteUnwindInfo(unwindInfo); + for (BasicBlock block = cfg.Blocks.First; block != null; block = block.ListNext) { context.EnterBlock(block); |