Ryujinx/Ryujinx.HLE/HOS/Kernel/Threading/KThread.cs

1212 lines
35 KiB
C#
Raw Normal View History

Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
using ARMeilleure.Memory;
using Ryujinx.Common.Logging;
using Ryujinx.HLE.HOS.Kernel.Common;
using Ryujinx.HLE.HOS.Kernel.Process;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
using System.Threading;
namespace Ryujinx.HLE.HOS.Kernel.Threading
2018-02-05 00:08:20 +01:00
{
class KThread : KSynchronizationObject, IKFutureSchedulerObject
2018-02-05 00:08:20 +01:00
{
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
private int _hostThreadRunning;
public Thread HostThread { get; private set; }
public ARMeilleure.State.ExecutionContext Context { get; private set; }
2018-02-05 00:08:20 +01:00
public long AffinityMask { get; set; }
public long ThreadUid { get; private set; }
public long TotalTimeRunning { get; set; }
public KSynchronizationObject SignaledObj { get; set; }
public ulong CondVarAddress { get; set; }
private ulong _entrypoint;
public ulong MutexAddress { get; set; }
public KProcess Owner { get; private set; }
private ulong _tlsAddress;
public ulong TlsAddress => _tlsAddress;
public ulong TlsDramAddress { get; private set; }
public long LastScheduledTime { get; set; }
public LinkedListNode<KThread>[] SiblingsPerCore { get; private set; }
public LinkedList<KThread> Withholder { get; set; }
public LinkedListNode<KThread> WithholderNode { get; set; }
public LinkedListNode<KThread> ProcessListNode { get; set; }
private LinkedList<KThread> _mutexWaiters;
private LinkedListNode<KThread> _mutexWaiterNode;
public KThread MutexOwner { get; private set; }
2018-06-26 06:09:32 +02:00
public int ThreadHandleForUserMutex { get; set; }
private ThreadSchedState _forcePauseFlags;
public KernelResult ObjSyncResult { get; set; }
public int DynamicPriority { get; set; }
public int CurrentCore { get; set; }
public int BasePriority { get; set; }
public int PreferredCore { get; set; }
private long _affinityMaskOverride;
private int _preferredCoreOverride;
private int _affinityOverrideCount;
public ThreadSchedState SchedFlags { get; private set; }
private int _shallBeTerminated;
public bool ShallBeTerminated { get => _shallBeTerminated != 0; set => _shallBeTerminated = value ? 1 : 0; }
public bool SyncCancelled { get; set; }
public bool WaitingSync { get; set; }
private bool _hasExited;
private bool _hasBeenInitialized;
private bool _hasBeenReleased;
public bool WaitingInArbitration { get; set; }
private KScheduler _scheduler;
private KSchedulingData _schedulingData;
public long LastPc { get; set; }
public KThread(Horizon system) : base(system)
2018-02-05 00:08:20 +01:00
{
_scheduler = system.Scheduler;
_schedulingData = system.Scheduler.SchedulingData;
SiblingsPerCore = new LinkedListNode<KThread>[KScheduler.CpuCoresCount];
_mutexWaiters = new LinkedList<KThread>();
}
public KernelResult Initialize(
ulong entrypoint,
ulong argsPtr,
ulong stackTop,
int priority,
int defaultCpuCore,
KProcess owner,
ThreadType type = ThreadType.User,
ThreadStart customHostThreadStart = null)
{
if ((uint)type > 3)
{
throw new ArgumentException($"Invalid thread type \"{type}\".");
}
PreferredCore = defaultCpuCore;
AffinityMask |= 1L << defaultCpuCore;
SchedFlags = type == ThreadType.Dummy
? ThreadSchedState.Running
: ThreadSchedState.None;
CurrentCore = PreferredCore;
DynamicPriority = priority;
BasePriority = priority;
ObjSyncResult = KernelResult.ThreadNotStarted;
_entrypoint = entrypoint;
if (type == ThreadType.User)
{
if (owner.AllocateThreadLocalStorage(out _tlsAddress) != KernelResult.Success)
{
return KernelResult.OutOfMemory;
}
TlsDramAddress = owner.MemoryManager.GetDramAddressFromVa(_tlsAddress);
MemoryHelper.FillWithZeros(owner.CpuMemory, (long)_tlsAddress, KTlsPageInfo.TlsEntrySize);
}
bool is64Bits;
if (owner != null)
{
Owner = owner;
owner.IncrementReferenceCount();
owner.IncrementThreadCount();
is64Bits = (owner.MmuFlags & 1) != 0;
}
else
{
is64Bits = true;
}
HostThread = new Thread(customHostThreadStart ?? (() => ThreadStart(entrypoint)));
Context = new ARMeilleure.State.ExecutionContext();
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
bool isAarch32 = (Owner.MmuFlags & 1) == 0;
Context.IsAarch32 = isAarch32;
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
Context.SetX(0, argsPtr);
if (isAarch32)
{
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
Context.SetX(13, (uint)stackTop);
}
else
{
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
Context.SetX(31, stackTop);
}
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
Context.CntfrqEl0 = 19200000;
Context.Tpidr = (long)_tlsAddress;
owner.SubscribeThreadEventHandlers(Context);
ThreadUid = System.GetThreadUid();
HostThread.Name = $"HLE.HostThread.{ThreadUid}";
_hasBeenInitialized = true;
if (owner != null)
{
owner.AddThread(this);
if (owner.IsPaused)
{
System.CriticalSection.Enter();
if (ShallBeTerminated || SchedFlags == ThreadSchedState.TerminationPending)
{
System.CriticalSection.Leave();
return KernelResult.Success;
}
_forcePauseFlags |= ThreadSchedState.ProcessPauseFlag;
CombineForcePauseFlags();
System.CriticalSection.Leave();
}
}
return KernelResult.Success;
}
public KernelResult Start()
{
if (!System.KernelInitialized)
{
System.CriticalSection.Enter();
if (!ShallBeTerminated && SchedFlags != ThreadSchedState.TerminationPending)
{
_forcePauseFlags |= ThreadSchedState.KernelInitPauseFlag;
CombineForcePauseFlags();
}
System.CriticalSection.Leave();
}
KernelResult result = KernelResult.ThreadTerminating;
System.CriticalSection.Enter();
if (!ShallBeTerminated)
{
KThread currentThread = System.Scheduler.GetCurrentThread();
while (SchedFlags != ThreadSchedState.TerminationPending &&
currentThread.SchedFlags != ThreadSchedState.TerminationPending &&
!currentThread.ShallBeTerminated)
{
if ((SchedFlags & ThreadSchedState.LowMask) != ThreadSchedState.None)
{
result = KernelResult.InvalidState;
break;
}
if (currentThread._forcePauseFlags == ThreadSchedState.None)
{
if (Owner != null && _forcePauseFlags != ThreadSchedState.None)
{
CombineForcePauseFlags();
}
SetNewSchedFlags(ThreadSchedState.Running);
result = KernelResult.Success;
break;
}
else
{
currentThread.CombineForcePauseFlags();
System.CriticalSection.Leave();
System.CriticalSection.Enter();
if (currentThread.ShallBeTerminated)
{
break;
}
}
}
}
System.CriticalSection.Leave();
return result;
}
public void Exit()
{
// TODO: Debug event.
if (Owner != null)
{
Owner.ResourceLimit?.Release(LimitableResource.Thread, 0, 1);
_hasBeenReleased = true;
}
System.CriticalSection.Enter();
_forcePauseFlags &= ~ThreadSchedState.ForcePauseMask;
ExitImpl();
System.CriticalSection.Leave();
DecrementReferenceCount();
}
public ThreadSchedState PrepareForTermination()
{
System.CriticalSection.Enter();
ThreadSchedState result;
if (Interlocked.CompareExchange(ref _shallBeTerminated, 1, 0) == 0)
{
if ((SchedFlags & ThreadSchedState.LowMask) == ThreadSchedState.None)
{
SchedFlags = ThreadSchedState.TerminationPending;
}
else
{
if (_forcePauseFlags != ThreadSchedState.None)
{
_forcePauseFlags &= ~ThreadSchedState.ThreadPauseFlag;
ThreadSchedState oldSchedFlags = SchedFlags;
SchedFlags &= ThreadSchedState.LowMask;
AdjustScheduling(oldSchedFlags);
}
if (BasePriority >= 0x10)
{
SetPriority(0xF);
}
if ((SchedFlags & ThreadSchedState.LowMask) == ThreadSchedState.Running)
{
// TODO: GIC distributor stuffs (sgir changes ect)
}
SignaledObj = null;
ObjSyncResult = KernelResult.ThreadTerminating;
ReleaseAndResume();
}
}
result = SchedFlags;
System.CriticalSection.Leave();
return result & ThreadSchedState.LowMask;
}
public void Terminate()
{
ThreadSchedState state = PrepareForTermination();
if (state != ThreadSchedState.TerminationPending)
{
System.Synchronization.WaitFor(new KSynchronizationObject[] { this }, -1, out _);
}
}
public void HandlePostSyscall()
{
ThreadSchedState state;
do
{
if (ShallBeTerminated || SchedFlags == ThreadSchedState.TerminationPending)
{
System.Scheduler.ExitThread(this);
Exit();
// As the death of the thread is handled by the CPU emulator, we differ from the official kernel and return here.
break;
}
System.CriticalSection.Enter();
if (ShallBeTerminated || SchedFlags == ThreadSchedState.TerminationPending)
{
state = ThreadSchedState.TerminationPending;
}
else
{
if (_forcePauseFlags != ThreadSchedState.None)
{
CombineForcePauseFlags();
}
state = ThreadSchedState.Running;
}
System.CriticalSection.Leave();
} while (state == ThreadSchedState.TerminationPending);
}
private void ExitImpl()
{
System.CriticalSection.Enter();
SetNewSchedFlags(ThreadSchedState.TerminationPending);
_hasExited = true;
Signal();
System.CriticalSection.Leave();
}
public KernelResult Sleep(long timeout)
{
System.CriticalSection.Enter();
if (ShallBeTerminated || SchedFlags == ThreadSchedState.TerminationPending)
{
System.CriticalSection.Leave();
return KernelResult.ThreadTerminating;
}
SetNewSchedFlags(ThreadSchedState.Paused);
if (timeout > 0)
{
System.TimeManager.ScheduleFutureInvocation(this, timeout);
}
System.CriticalSection.Leave();
if (timeout > 0)
{
System.TimeManager.UnscheduleFutureInvocation(this);
}
return 0;
}
public void Yield()
{
System.CriticalSection.Enter();
if (SchedFlags != ThreadSchedState.Running)
{
System.CriticalSection.Leave();
System.Scheduler.ContextSwitch();
return;
}
if (DynamicPriority < KScheduler.PrioritiesCount)
{
// Move current thread to the end of the queue.
_schedulingData.Reschedule(DynamicPriority, CurrentCore, this);
}
_scheduler.ThreadReselectionRequested = true;
System.CriticalSection.Leave();
System.Scheduler.ContextSwitch();
}
public void YieldWithLoadBalancing()
{
System.CriticalSection.Enter();
if (SchedFlags != ThreadSchedState.Running)
{
System.CriticalSection.Leave();
System.Scheduler.ContextSwitch();
return;
}
int prio = DynamicPriority;
int core = CurrentCore;
KThread nextThreadOnCurrentQueue = null;
if (DynamicPriority < KScheduler.PrioritiesCount)
{
// Move current thread to the end of the queue.
_schedulingData.Reschedule(prio, core, this);
Func<KThread, bool> predicate = x => x.DynamicPriority == prio;
nextThreadOnCurrentQueue = _schedulingData.ScheduledThreads(core).FirstOrDefault(predicate);
}
IEnumerable<KThread> SuitableCandidates()
{
foreach (KThread thread in _schedulingData.SuggestedThreads(core))
{
int srcCore = thread.CurrentCore;
if (srcCore >= 0)
{
KThread selectedSrcCore = _scheduler.CoreContexts[srcCore].SelectedThread;
if (selectedSrcCore == thread || ((selectedSrcCore?.DynamicPriority ?? 2) < 2))
{
continue;
}
}
// If the candidate was scheduled after the current thread, then it's not worth it,
// unless the priority is higher than the current one.
if (nextThreadOnCurrentQueue.LastScheduledTime >= thread.LastScheduledTime ||
nextThreadOnCurrentQueue.DynamicPriority < thread.DynamicPriority)
{
yield return thread;
}
}
}
KThread dst = SuitableCandidates().FirstOrDefault(x => x.DynamicPriority <= prio);
if (dst != null)
{
_schedulingData.TransferToCore(dst.DynamicPriority, core, dst);
_scheduler.ThreadReselectionRequested = true;
}
if (this != nextThreadOnCurrentQueue)
{
_scheduler.ThreadReselectionRequested = true;
}
System.CriticalSection.Leave();
System.Scheduler.ContextSwitch();
}
public void YieldAndWaitForLoadBalancing()
{
System.CriticalSection.Enter();
if (SchedFlags != ThreadSchedState.Running)
{
System.CriticalSection.Leave();
System.Scheduler.ContextSwitch();
return;
}
int core = CurrentCore;
_schedulingData.TransferToCore(DynamicPriority, -1, this);
KThread selectedThread = null;
if (!_schedulingData.ScheduledThreads(core).Any())
{
foreach (KThread thread in _schedulingData.SuggestedThreads(core))
{
if (thread.CurrentCore < 0)
{
continue;
}
KThread firstCandidate = _schedulingData.ScheduledThreads(thread.CurrentCore).FirstOrDefault();
if (firstCandidate == thread)
{
continue;
}
if (firstCandidate == null || firstCandidate.DynamicPriority >= 2)
{
_schedulingData.TransferToCore(thread.DynamicPriority, core, thread);
selectedThread = thread;
}
break;
}
}
if (selectedThread != this)
{
_scheduler.ThreadReselectionRequested = true;
}
System.CriticalSection.Leave();
System.Scheduler.ContextSwitch();
}
public void SetPriority(int priority)
{
System.CriticalSection.Enter();
BasePriority = priority;
UpdatePriorityInheritance();
System.CriticalSection.Leave();
}
public KernelResult SetActivity(bool pause)
{
KernelResult result = KernelResult.Success;
System.CriticalSection.Enter();
ThreadSchedState lowNibble = SchedFlags & ThreadSchedState.LowMask;
if (lowNibble != ThreadSchedState.Paused && lowNibble != ThreadSchedState.Running)
{
System.CriticalSection.Leave();
return KernelResult.InvalidState;
}
System.CriticalSection.Enter();
if (!ShallBeTerminated && SchedFlags != ThreadSchedState.TerminationPending)
{
if (pause)
{
// Pause, the force pause flag should be clear (thread is NOT paused).
if ((_forcePauseFlags & ThreadSchedState.ThreadPauseFlag) == 0)
{
_forcePauseFlags |= ThreadSchedState.ThreadPauseFlag;
CombineForcePauseFlags();
}
else
{
result = KernelResult.InvalidState;
}
}
else
{
// Unpause, the force pause flag should be set (thread is paused).
if ((_forcePauseFlags & ThreadSchedState.ThreadPauseFlag) != 0)
{
ThreadSchedState oldForcePauseFlags = _forcePauseFlags;
_forcePauseFlags &= ~ThreadSchedState.ThreadPauseFlag;
if ((oldForcePauseFlags & ~ThreadSchedState.ThreadPauseFlag) == ThreadSchedState.None)
{
ThreadSchedState oldSchedFlags = SchedFlags;
SchedFlags &= ThreadSchedState.LowMask;
AdjustScheduling(oldSchedFlags);
}
}
else
{
result = KernelResult.InvalidState;
}
}
}
System.CriticalSection.Leave();
System.CriticalSection.Leave();
return result;
}
public void CancelSynchronization()
{
System.CriticalSection.Enter();
if ((SchedFlags & ThreadSchedState.LowMask) != ThreadSchedState.Paused || !WaitingSync)
{
SyncCancelled = true;
}
else if (Withholder != null)
{
Withholder.Remove(WithholderNode);
SetNewSchedFlags(ThreadSchedState.Running);
Withholder = null;
SyncCancelled = true;
}
else
{
SignaledObj = null;
ObjSyncResult = KernelResult.Cancelled;
SetNewSchedFlags(ThreadSchedState.Running);
SyncCancelled = false;
}
System.CriticalSection.Leave();
}
public KernelResult SetCoreAndAffinityMask(int newCore, long newAffinityMask)
{
System.CriticalSection.Enter();
bool useOverride = _affinityOverrideCount != 0;
// The value -3 is "do not change the preferred core".
if (newCore == -3)
{
newCore = useOverride ? _preferredCoreOverride : PreferredCore;
if ((newAffinityMask & (1 << newCore)) == 0)
{
System.CriticalSection.Leave();
return KernelResult.InvalidCombination;
}
}
if (useOverride)
{
_preferredCoreOverride = newCore;
_affinityMaskOverride = newAffinityMask;
}
else
{
long oldAffinityMask = AffinityMask;
PreferredCore = newCore;
AffinityMask = newAffinityMask;
if (oldAffinityMask != newAffinityMask)
{
int oldCore = CurrentCore;
if (CurrentCore >= 0 && ((AffinityMask >> CurrentCore) & 1) == 0)
{
if (PreferredCore < 0)
{
CurrentCore = HighestSetCore(AffinityMask);
}
else
{
CurrentCore = PreferredCore;
}
}
AdjustSchedulingForNewAffinity(oldAffinityMask, oldCore);
}
}
System.CriticalSection.Leave();
return KernelResult.Success;
}
private static int HighestSetCore(long mask)
{
for (int core = KScheduler.CpuCoresCount - 1; core >= 0; core--)
{
if (((mask >> core) & 1) != 0)
{
return core;
}
}
return -1;
}
private void CombineForcePauseFlags()
{
ThreadSchedState oldFlags = SchedFlags;
ThreadSchedState lowNibble = SchedFlags & ThreadSchedState.LowMask;
SchedFlags = lowNibble | _forcePauseFlags;
AdjustScheduling(oldFlags);
}
private void SetNewSchedFlags(ThreadSchedState newFlags)
{
System.CriticalSection.Enter();
ThreadSchedState oldFlags = SchedFlags;
SchedFlags = (oldFlags & ThreadSchedState.HighMask) | newFlags;
if ((oldFlags & ThreadSchedState.LowMask) != newFlags)
{
AdjustScheduling(oldFlags);
}
System.CriticalSection.Leave();
}
public void ReleaseAndResume()
{
System.CriticalSection.Enter();
if ((SchedFlags & ThreadSchedState.LowMask) == ThreadSchedState.Paused)
{
if (Withholder != null)
{
Withholder.Remove(WithholderNode);
SetNewSchedFlags(ThreadSchedState.Running);
Withholder = null;
}
else
{
SetNewSchedFlags(ThreadSchedState.Running);
}
}
System.CriticalSection.Leave();
}
public void Reschedule(ThreadSchedState newFlags)
{
System.CriticalSection.Enter();
ThreadSchedState oldFlags = SchedFlags;
SchedFlags = (oldFlags & ThreadSchedState.HighMask) |
(newFlags & ThreadSchedState.LowMask);
AdjustScheduling(oldFlags);
System.CriticalSection.Leave();
}
public void AddMutexWaiter(KThread requester)
{
AddToMutexWaitersList(requester);
requester.MutexOwner = this;
UpdatePriorityInheritance();
}
public void RemoveMutexWaiter(KThread thread)
{
if (thread._mutexWaiterNode?.List != null)
{
_mutexWaiters.Remove(thread._mutexWaiterNode);
}
thread.MutexOwner = null;
UpdatePriorityInheritance();
}
public KThread RelinquishMutex(ulong mutexAddress, out int count)
{
count = 0;
if (_mutexWaiters.First == null)
{
return null;
}
KThread newMutexOwner = null;
LinkedListNode<KThread> currentNode = _mutexWaiters.First;
do
{
// Skip all threads that are not waiting for this mutex.
while (currentNode != null && currentNode.Value.MutexAddress != mutexAddress)
{
currentNode = currentNode.Next;
}
if (currentNode == null)
{
break;
}
LinkedListNode<KThread> nextNode = currentNode.Next;
_mutexWaiters.Remove(currentNode);
currentNode.Value.MutexOwner = newMutexOwner;
if (newMutexOwner != null)
{
// New owner was already selected, re-insert on new owner list.
newMutexOwner.AddToMutexWaitersList(currentNode.Value);
}
else
{
// New owner not selected yet, use current thread.
newMutexOwner = currentNode.Value;
}
count++;
currentNode = nextNode;
}
while (currentNode != null);
if (newMutexOwner != null)
{
UpdatePriorityInheritance();
newMutexOwner.UpdatePriorityInheritance();
}
return newMutexOwner;
}
private void UpdatePriorityInheritance()
{
// If any of the threads waiting for the mutex has
// higher priority than the current thread, then
// the current thread inherits that priority.
int highestPriority = BasePriority;
if (_mutexWaiters.First != null)
{
int waitingDynamicPriority = _mutexWaiters.First.Value.DynamicPriority;
if (waitingDynamicPriority < highestPriority)
{
highestPriority = waitingDynamicPriority;
}
}
if (highestPriority != DynamicPriority)
{
int oldPriority = DynamicPriority;
DynamicPriority = highestPriority;
AdjustSchedulingForNewPriority(oldPriority);
if (MutexOwner != null)
{
// Remove and re-insert to ensure proper sorting based on new priority.
MutexOwner._mutexWaiters.Remove(_mutexWaiterNode);
MutexOwner.AddToMutexWaitersList(this);
MutexOwner.UpdatePriorityInheritance();
}
}
}
private void AddToMutexWaitersList(KThread thread)
{
LinkedListNode<KThread> nextPrio = _mutexWaiters.First;
int currentPriority = thread.DynamicPriority;
while (nextPrio != null && nextPrio.Value.DynamicPriority <= currentPriority)
{
nextPrio = nextPrio.Next;
}
if (nextPrio != null)
{
thread._mutexWaiterNode = _mutexWaiters.AddBefore(nextPrio, thread);
}
else
{
thread._mutexWaiterNode = _mutexWaiters.AddLast(thread);
}
}
private void AdjustScheduling(ThreadSchedState oldFlags)
{
if (oldFlags == SchedFlags)
{
return;
}
if (oldFlags == ThreadSchedState.Running)
{
// Was running, now it's stopped.
if (CurrentCore >= 0)
{
_schedulingData.Unschedule(DynamicPriority, CurrentCore, this);
}
for (int core = 0; core < KScheduler.CpuCoresCount; core++)
{
if (core != CurrentCore && ((AffinityMask >> core) & 1) != 0)
{
_schedulingData.Unsuggest(DynamicPriority, core, this);
}
}
}
else if (SchedFlags == ThreadSchedState.Running)
{
// Was stopped, now it's running.
if (CurrentCore >= 0)
{
_schedulingData.Schedule(DynamicPriority, CurrentCore, this);
}
for (int core = 0; core < KScheduler.CpuCoresCount; core++)
{
if (core != CurrentCore && ((AffinityMask >> core) & 1) != 0)
{
_schedulingData.Suggest(DynamicPriority, core, this);
}
}
}
_scheduler.ThreadReselectionRequested = true;
}
private void AdjustSchedulingForNewPriority(int oldPriority)
{
if (SchedFlags != ThreadSchedState.Running)
{
return;
}
// Remove thread from the old priority queues.
if (CurrentCore >= 0)
{
_schedulingData.Unschedule(oldPriority, CurrentCore, this);
}
for (int core = 0; core < KScheduler.CpuCoresCount; core++)
{
if (core != CurrentCore && ((AffinityMask >> core) & 1) != 0)
{
_schedulingData.Unsuggest(oldPriority, core, this);
}
}
// Add thread to the new priority queues.
KThread currentThread = _scheduler.GetCurrentThread();
if (CurrentCore >= 0)
{
if (currentThread == this)
{
_schedulingData.SchedulePrepend(DynamicPriority, CurrentCore, this);
}
else
{
_schedulingData.Schedule(DynamicPriority, CurrentCore, this);
}
}
for (int core = 0; core < KScheduler.CpuCoresCount; core++)
{
if (core != CurrentCore && ((AffinityMask >> core) & 1) != 0)
{
_schedulingData.Suggest(DynamicPriority, core, this);
}
}
_scheduler.ThreadReselectionRequested = true;
}
private void AdjustSchedulingForNewAffinity(long oldAffinityMask, int oldCore)
{
if (SchedFlags != ThreadSchedState.Running || DynamicPriority >= KScheduler.PrioritiesCount)
{
return;
}
// Remove thread from the old priority queues.
for (int core = 0; core < KScheduler.CpuCoresCount; core++)
{
if (((oldAffinityMask >> core) & 1) != 0)
{
if (core == oldCore)
{
_schedulingData.Unschedule(DynamicPriority, core, this);
}
else
{
_schedulingData.Unsuggest(DynamicPriority, core, this);
}
}
}
// Add thread to the new priority queues.
for (int core = 0; core < KScheduler.CpuCoresCount; core++)
{
if (((AffinityMask >> core) & 1) != 0)
{
if (core == CurrentCore)
{
_schedulingData.Schedule(DynamicPriority, core, this);
}
else
{
_schedulingData.Suggest(DynamicPriority, core, this);
}
}
}
_scheduler.ThreadReselectionRequested = true;
}
public void SetEntryArguments(long argsPtr, int threadHandle)
{
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
Context.SetX(0, (ulong)argsPtr);
Context.SetX(1, (ulong)threadHandle);
}
public void TimeUp()
{
ReleaseAndResume();
}
public string GetGuestStackTrace()
{
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
return Owner.Debugger.GetGuestStackTrace(Context);
}
public void PrintGuestStackTrace()
{
StringBuilder trace = new StringBuilder();
trace.AppendLine("Guest stack trace:");
trace.AppendLine(GetGuestStackTrace());
Logger.PrintInfo(LogClass.Cpu, trace.ToString());
}
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
public void Execute()
{
if (Interlocked.CompareExchange(ref _hostThreadRunning, 1, 0) == 0)
{
HostThread.Start();
}
}
private void ThreadStart(ulong entrypoint)
{
Owner.Translator.Execute(Context, entrypoint);
Context.Dispose();
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
ThreadExit();
}
private void ThreadExit()
{
System.Scheduler.ExitThread(this);
System.Scheduler.RemoveThread(this);
}
Add a new JIT compiler for CPU code (#693) * Start of the ARMeilleure project * Refactoring around the old IRAdapter, now renamed to PreAllocator * Optimize the LowestBitSet method * Add CLZ support and fix CLS implementation * Add missing Equals and GetHashCode overrides on some structs, misc small tweaks * Implement the ByteSwap IR instruction, and some refactoring on the assembler * Implement the DivideUI IR instruction and fix 64-bits IDIV * Correct constant operand type on CSINC * Move division instructions implementation to InstEmitDiv * Fix destination type for the ConditionalSelect IR instruction * Implement UMULH and SMULH, with new IR instructions * Fix some issues with shift instructions * Fix constant types for BFM instructions * Fix up new tests using the new V128 struct * Update tests * Move DIV tests to a separate file * Add support for calls, and some instructions that depends on them * Start adding support for SIMD & FP types, along with some of the related ARM instructions * Fix some typos and the divide instruction with FP operands * Fix wrong method call on Clz_V * Implement ARM FP & SIMD move instructions, Saddlv_V, and misc. fixes * Implement SIMD logical instructions and more misc. fixes * Fix PSRAD x86 instruction encoding, TRN, UABD and UABDL implementations * Implement float conversion instruction, merge in LDj3SNuD fixes, and some other misc. fixes * Implement SIMD shift instruction and fix Dup_V * Add SCVTF and UCVTF (vector, fixed-point) variants to the opcode table * Fix check with tolerance on tester * Implement FP & SIMD comparison instructions, and some fixes * Update FCVT (Scalar) encoding on the table to support the Half-float variants * Support passing V128 structs, some cleanup on the register allocator, merge LDj3SNuD fixes * Use old memory access methods, made a start on SIMD memory insts support, some fixes * Fix float constant passed to functions, save and restore non-volatile XMM registers, other fixes * Fix arguments count with struct return values, other fixes * More instructions * Misc. fixes and integrate LDj3SNuD fixes * Update tests * Add a faster linear scan allocator, unwinding support on windows, and other changes * Update Ryujinx.HLE * Update Ryujinx.Graphics * Fix V128 return pointer passing, RCX is clobbered * Update Ryujinx.Tests * Update ITimeZoneService * Stop using GetFunctionPointer as that can't be called from native code, misc. fixes and tweaks * Use generic GetFunctionPointerForDelegate method and other tweaks * Some refactoring on the code generator, assert on invalid operations and use a separate enum for intrinsics * Remove some unused code on the assembler * Fix REX.W prefix regression on float conversion instructions, add some sort of profiler * Add hardware capability detection * Fix regression on Sha1h and revert Fcm** changes * Add SSE2-only paths on vector extract and insert, some refactoring on the pre-allocator * Fix silly mistake introduced on last commit on CpuId * Generate inline stack probes when the stack allocation is too large * Initial support for the System-V ABI * Support multiple destination operands * Fix SSE2 VectorInsert8 path, and other fixes * Change placement of XMM callee save and restore code to match other compilers * Rename Dest to Destination and Inst to Instruction * Fix a regression related to calls and the V128 type * Add an extra space on comments to match code style * Some refactoring * Fix vector insert FP32 SSE2 path * Port over the ARM32 instructions * Avoid memory protection races on JIT Cache * Another fix on VectorInsert FP32 (thanks to LDj3SNuD * Float operands don't need to use the same register when VEX is supported * Add a new register allocator, higher quality code for hot code (tier up), and other tweaks * Some nits, small improvements on the pre allocator * CpuThreadState is gone * Allow changing CPU emulators with a config entry * Add runtime identifiers on the ARMeilleure project * Allow switching between CPUs through a config entry (pt. 2) * Change win10-x64 to win-x64 on projects * Update the Ryujinx project to use ARMeilleure * Ensure that the selected register is valid on the hybrid allocator * Allow exiting on returns to 0 (should fix test regression) * Remove register assignments for most used variables on the hybrid allocator * Do not use fixed registers as spill temp * Add missing namespace and remove unneeded using * Address PR feedback * Fix types, etc * Enable AssumeStrictAbiCompliance by default * Ensure that Spill and Fill don't load or store any more than necessary
2019-08-08 20:56:22 +02:00
public bool IsCurrentHostThread()
{
return Thread.CurrentThread == HostThread;
}
public override bool IsSignaled()
{
return _hasExited;
}
protected override void Destroy()
{
if (_hasBeenInitialized)
{
FreeResources();
bool released = Owner != null || _hasBeenReleased;
if (Owner != null)
{
Owner.ResourceLimit?.Release(LimitableResource.Thread, 1, released ? 0 : 1);
Owner.DecrementReferenceCount();
}
else
{
System.ResourceLimit.Release(LimitableResource.Thread, 1, released ? 0 : 1);
}
}
}
private void FreeResources()
{
Owner?.RemoveThread(this);
if (_tlsAddress != 0 && Owner.FreeThreadLocalStorage(_tlsAddress) != KernelResult.Success)
{
throw new InvalidOperationException("Unexpected failure freeing thread local storage.");
}
System.CriticalSection.Enter();
// Wake up all threads that may be waiting for a mutex being held by this thread.
foreach (KThread thread in _mutexWaiters)
{
thread.MutexOwner = null;
thread._preferredCoreOverride = 0;
thread.ObjSyncResult = KernelResult.InvalidState;
thread.ReleaseAndResume();
}
System.CriticalSection.Leave();
Owner?.DecrementThreadCountAndTerminateIfZero();
2018-02-05 00:08:20 +01:00
}
}
}