Archived
1
0
Fork 0
forked from Mirror/Ryujinx
This repository has been archived on 2024-10-11. You can view files and clone it, but cannot push or open issues or pull requests.
jinx/ARMeilleure/Memory/MemoryManager.cs

835 lines
22 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 18:56:22 +00:00
using ARMeilleure.State;
using System;
using System.Runtime.InteropServices;
using System.Threading;
using static ARMeilleure.Memory.MemoryManagement;
namespace ARMeilleure.Memory
{
public unsafe class MemoryManager : IMemoryManager
{
public const int PageBits = 12;
public const int PageSize = 1 << PageBits;
public const int PageMask = PageSize - 1;
private const long PteFlagNotModified = 1;
internal const long PteFlagsMask = 7;
public IntPtr Ram { get; private set; }
private byte* _ramPtr;
private IntPtr _pageTable;
internal IntPtr PageTable => _pageTable;
internal int PtLevelBits { get; }
internal int PtLevelSize { get; }
internal int PtLevelMask { get; }
public bool HasWriteWatchSupport => MemoryManagement.HasWriteWatchSupport;
public int AddressSpaceBits { get; }
public long AddressSpaceSize { get; }
public MemoryManager(
IntPtr ram,
int addressSpaceBits = 48,
bool useFlatPageTable = false)
{
Ram = ram;
_ramPtr = (byte*)ram;
AddressSpaceBits = addressSpaceBits;
AddressSpaceSize = 1L << addressSpaceBits;
// When flat page table is requested, we use a single
// array for the mappings of the entire address space.
// This has better performance, but also high memory usage.
// The multi level page table uses 9 bits per level, so
// the memory usage is lower, but the performance is also
// lower, since each address translation requires multiple reads.
if (useFlatPageTable)
{
PtLevelBits = addressSpaceBits - PageBits;
}
else
{
PtLevelBits = 9;
}
PtLevelSize = 1 << PtLevelBits;
PtLevelMask = PtLevelSize - 1;
_pageTable = Allocate((ulong)(PtLevelSize * IntPtr.Size));
}
public void Map(long va, long pa, long size)
{
SetPtEntries(va, _ramPtr + pa, size);
}
public void Unmap(long position, long size)
{
SetPtEntries(position, null, size);
}
public bool IsMapped(long position)
{
return Translate(position) != IntPtr.Zero;
}
public long GetPhysicalAddress(long virtualAddress)
{
byte* ptr = (byte*)Translate(virtualAddress);
return (long)(ptr - _ramPtr);
}
private IntPtr Translate(long position)
{
if (!IsValidPosition(position))
{
return IntPtr.Zero;
}
byte* ptr = GetPtEntry(position);
ulong ptrUlong = (ulong)ptr;
if ((ptrUlong & PteFlagsMask) != 0)
{
ptrUlong &= ~(ulong)PteFlagsMask;
ptr = (byte*)ptrUlong;
}
return new IntPtr(ptr + (position & PageMask));
}
private IntPtr TranslateWrite(long position)
{
if (!IsValidPosition(position))
{
return IntPtr.Zero;
}
byte* ptr = GetPtEntry(position);
ulong ptrUlong = (ulong)ptr;
if ((ptrUlong & PteFlagsMask) != 0)
{
if ((ptrUlong & PteFlagNotModified) != 0)
{
ClearPtEntryFlag(position, PteFlagNotModified);
}
ptrUlong &= ~(ulong)PteFlagsMask;
ptr = (byte*)ptrUlong;
}
return new IntPtr(ptr + (position & PageMask));
}
private byte* GetPtEntry(long position)
{
return *(byte**)GetPtPtr(position);
}
private void SetPtEntries(long va, byte* ptr, long size)
{
long endPosition = (va + size + PageMask) & ~PageMask;
while ((ulong)va < (ulong)endPosition)
{
SetPtEntry(va, ptr);
va += PageSize;
if (ptr != null)
{
ptr += PageSize;
}
}
}
private void SetPtEntry(long position, byte* ptr)
{
*(byte**)GetPtPtr(position) = ptr;
}
private void SetPtEntryFlag(long position, long flag)
{
ModifyPtEntryFlag(position, flag, setFlag: true);
}
private void ClearPtEntryFlag(long position, long flag)
{
ModifyPtEntryFlag(position, flag, setFlag: false);
}
private void ModifyPtEntryFlag(long position, long flag, bool setFlag)
{
IntPtr* pt = (IntPtr*)_pageTable;
while (true)
{
IntPtr* ptPtr = GetPtPtr(position);
IntPtr old = *ptPtr;
long modified = old.ToInt64();
if (setFlag)
{
modified |= flag;
}
else
{
modified &= ~flag;
}
IntPtr origValue = Interlocked.CompareExchange(ref *ptPtr, new IntPtr(modified), old);
if (origValue == old)
{
break;
}
}
}
private IntPtr* GetPtPtr(long position)
{
if (!IsValidPosition(position))
{
throw new ArgumentOutOfRangeException(nameof(position));
}
IntPtr nextPtr = _pageTable;
IntPtr* ptePtr = null;
int bit = PageBits;
while (true)
{
long index = (position >> bit) & PtLevelMask;
ptePtr = &((IntPtr*)nextPtr)[index];
bit += PtLevelBits;
if (bit >= AddressSpaceBits)
{
break;
}
nextPtr = *ptePtr;
if (nextPtr == IntPtr.Zero)
{
// Entry does not yet exist, allocate a new one.
IntPtr newPtr = Allocate((ulong)(PtLevelSize * IntPtr.Size));
// Try to swap the current pointer (should be zero), with the allocated one.
nextPtr = Interlocked.CompareExchange(ref *ptePtr, newPtr, IntPtr.Zero);
// If the old pointer is not null, then another thread already has set it.
if (nextPtr != IntPtr.Zero)
{
Free(newPtr);
}
else
{
nextPtr = newPtr;
}
}
}
return ptePtr;
}
public bool IsRegionModified(long position, long size)
{
if (!HasWriteWatchSupport)
{
return IsRegionModifiedFallback(position, size);
}
IntPtr address = Translate(position);
IntPtr baseAddr = address;
IntPtr expectedAddr = address;
long pendingPages = 0;
long pages = size / PageSize;
bool modified = false;
bool IsAnyPageModified()
{
IntPtr pendingSize = new IntPtr(pendingPages * PageSize);
IntPtr[] addresses = new IntPtr[pendingPages];
bool result = GetModifiedPages(baseAddr, pendingSize, addresses, out ulong count);
if (result)
{
return count != 0;
}
else
{
return true;
}
}
while (pages-- > 0)
{
if (address != expectedAddr)
{
modified |= IsAnyPageModified();
baseAddr = address;
pendingPages = 0;
}
expectedAddr = address + PageSize;
pendingPages++;
if (pages == 0)
{
break;
}
position += PageSize;
address = Translate(position);
}
if (pendingPages != 0)
{
modified |= IsAnyPageModified();
}
return modified;
}
private unsafe bool IsRegionModifiedFallback(long position, long size)
{
long endAddr = (position + size + PageMask) & ~PageMask;
bool modified = false;
while ((ulong)position < (ulong)endAddr)
{
if (IsValidPosition(position))
{
byte* ptr = ((byte**)_pageTable)[position >> PageBits];
ulong ptrUlong = (ulong)ptr;
if ((ptrUlong & PteFlagNotModified) == 0)
{
modified = true;
SetPtEntryFlag(position, PteFlagNotModified);
}
}
else
{
modified = true;
}
position += PageSize;
}
return modified;
}
public bool TryGetHostAddress(long position, long size, out IntPtr ptr)
{
if (IsContiguous(position, size))
{
ptr = (IntPtr)Translate(position);
return true;
}
ptr = IntPtr.Zero;
return false;
}
private bool IsContiguous(long position, long size)
{
long endPos = position + size;
position &= ~PageMask;
long expectedPa = GetPhysicalAddress(position);
while ((ulong)position < (ulong)endPos)
{
long pa = GetPhysicalAddress(position);
if (pa != expectedPa)
{
return false;
}
position += PageSize;
expectedPa += PageSize;
}
return true;
}
public bool IsValidPosition(long position)
{
return (ulong)position < (ulong)AddressSpaceSize;
}
internal V128 AtomicLoadInt128(long position)
{
if ((position & 0xf) != 0)
{
AbortWithAlignmentFault(position);
}
IntPtr ptr = TranslateWrite(position);
return MemoryManagerPal.AtomicLoad128(ptr);
}
internal bool AtomicCompareExchangeByte(long position, byte expected, byte desired)
{
int* ptr = (int*)Translate(position);
int currentValue = *ptr;
int expected32 = (currentValue & ~byte.MaxValue) | expected;
int desired32 = (currentValue & ~byte.MaxValue) | desired;
return Interlocked.CompareExchange(ref *ptr, desired32, expected32) == expected32;
}
internal bool AtomicCompareExchangeInt16(long position, short expected, short desired)
{
if ((position & 1) != 0)
{
AbortWithAlignmentFault(position);
}
int* ptr = (int*)Translate(position);
int currentValue = *ptr;
int expected32 = (currentValue & ~ushort.MaxValue) | (ushort)expected;
int desired32 = (currentValue & ~ushort.MaxValue) | (ushort)desired;
return Interlocked.CompareExchange(ref *ptr, desired32, expected32) == expected32;
}
public bool AtomicCompareExchangeInt32(long position, int expected, int desired)
{
if ((position & 3) != 0)
{
AbortWithAlignmentFault(position);
}
int* ptr = (int*)TranslateWrite(position);
return Interlocked.CompareExchange(ref *ptr, desired, expected) == expected;
}
internal bool AtomicCompareExchangeInt64(long position, long expected, long desired)
{
if ((position & 7) != 0)
{
AbortWithAlignmentFault(position);
}
long* ptr = (long*)TranslateWrite(position);
return Interlocked.CompareExchange(ref *ptr, desired, expected) == expected;
}
internal bool AtomicCompareExchangeInt128(long position, V128 expected, V128 desired)
{
if ((position & 0xf) != 0)
{
AbortWithAlignmentFault(position);
}
IntPtr ptr = TranslateWrite(position);
return MemoryManagerPal.CompareAndSwap128(ptr, expected, desired) == expected;
}
public int AtomicIncrementInt32(long position)
{
if ((position & 3) != 0)
{
AbortWithAlignmentFault(position);
}
int* ptr = (int*)TranslateWrite(position);
return Interlocked.Increment(ref *ptr);
}
public int AtomicDecrementInt32(long position)
{
if ((position & 3) != 0)
{
AbortWithAlignmentFault(position);
}
int* ptr = (int*)TranslateWrite(position);
return Interlocked.Decrement(ref *ptr);
}
private void AbortWithAlignmentFault(long position)
{
// TODO: Abort mode and exception support on the CPU.
throw new InvalidOperationException($"Tried to compare exchange a misaligned address 0x{position:X16}.");
}
public sbyte ReadSByte(long position)
{
return (sbyte)ReadByte(position);
}
public short ReadInt16(long position)
{
return (short)ReadUInt16(position);
}
public int ReadInt32(long position)
{
return (int)ReadUInt32(position);
}
public long ReadInt64(long position)
{
return (long)ReadUInt64(position);
}
public byte ReadByte(long position)
{
return *((byte*)Translate(position));
}
public ushort ReadUInt16(long position)
{
if ((position & 1) == 0)
{
return *((ushort*)Translate(position));
}
else
{
return (ushort)(ReadByte(position + 0) << 0 |
ReadByte(position + 1) << 8);
}
}
public uint ReadUInt32(long position)
{
if ((position & 3) == 0)
{
return *((uint*)Translate(position));
}
else
{
return (uint)(ReadUInt16(position + 0) << 0 |
ReadUInt16(position + 2) << 16);
}
}
public ulong ReadUInt64(long position)
{
if ((position & 7) == 0)
{
return *((ulong*)Translate(position));
}
else
{
return (ulong)ReadUInt32(position + 0) << 0 |
(ulong)ReadUInt32(position + 4) << 32;
}
}
public V128 ReadVector128(long position)
{
return new V128(ReadUInt64(position), ReadUInt64(position + 8));
}
public byte[] ReadBytes(long position, long size)
{
long endAddr = position + size;
if ((ulong)size > int.MaxValue)
{
throw new ArgumentOutOfRangeException(nameof(size));
}
if ((ulong)endAddr < (ulong)position)
{
throw new ArgumentOutOfRangeException(nameof(position));
}
byte[] data = new byte[size];
int offset = 0;
while ((ulong)position < (ulong)endAddr)
{
long pageLimit = (position + PageSize) & ~(long)PageMask;
if ((ulong)pageLimit > (ulong)endAddr)
{
pageLimit = endAddr;
}
int copySize = (int)(pageLimit - position);
Marshal.Copy(Translate(position), data, offset, copySize);
position += copySize;
offset += copySize;
}
return data;
}
public void ReadBytes(long position, byte[] data, int startIndex, int size)
{
// Note: This will be moved later.
long endAddr = position + size;
if ((ulong)size > int.MaxValue)
{
throw new ArgumentOutOfRangeException(nameof(size));
}
if ((ulong)endAddr < (ulong)position)
{
throw new ArgumentOutOfRangeException(nameof(position));
}
int offset = startIndex;
while ((ulong)position < (ulong)endAddr)
{
long pageLimit = (position + PageSize) & ~(long)PageMask;
if ((ulong)pageLimit > (ulong)endAddr)
{
pageLimit = endAddr;
}
int copySize = (int)(pageLimit - position);
Marshal.Copy(Translate(position), data, offset, copySize);
position += copySize;
offset += copySize;
}
}
public void WriteSByte(long position, sbyte value)
{
WriteByte(position, (byte)value);
}
public void WriteInt16(long position, short value)
{
WriteUInt16(position, (ushort)value);
}
public void WriteInt32(long position, int value)
{
WriteUInt32(position, (uint)value);
}
public void WriteInt64(long position, long value)
{
WriteUInt64(position, (ulong)value);
}
public void WriteByte(long position, byte value)
{
*((byte*)TranslateWrite(position)) = value;
}
public void WriteUInt16(long position, ushort value)
{
if ((position & 1) == 0)
{
*((ushort*)TranslateWrite(position)) = value;
}
else
{
WriteByte(position + 0, (byte)(value >> 0));
WriteByte(position + 1, (byte)(value >> 8));
}
}
public void WriteUInt32(long position, uint value)
{
if ((position & 3) == 0)
{
*((uint*)TranslateWrite(position)) = value;
}
else
{
WriteUInt16(position + 0, (ushort)(value >> 0));
WriteUInt16(position + 2, (ushort)(value >> 16));
}
}
public void WriteUInt64(long position, ulong value)
{
if ((position & 7) == 0)
{
*((ulong*)TranslateWrite(position)) = value;
}
else
{
WriteUInt32(position + 0, (uint)(value >> 0));
WriteUInt32(position + 4, (uint)(value >> 32));
}
}
public void WriteVector128(long position, V128 value)
{
WriteUInt64(position + 0, value.GetUInt64(0));
WriteUInt64(position + 8, value.GetUInt64(1));
}
public void WriteBytes(long position, byte[] data)
{
long endAddr = position + data.Length;
if ((ulong)endAddr < (ulong)position)
{
throw new ArgumentOutOfRangeException(nameof(position));
}
int offset = 0;
while ((ulong)position < (ulong)endAddr)
{
long pageLimit = (position + PageSize) & ~(long)PageMask;
if ((ulong)pageLimit > (ulong)endAddr)
{
pageLimit = endAddr;
}
int copySize = (int)(pageLimit - position);
Marshal.Copy(data, offset, TranslateWrite(position), copySize);
position += copySize;
offset += copySize;
}
}
public void WriteBytes(long position, byte[] data, int startIndex, int size)
{
// Note: This will be moved later.
long endAddr = position + size;
if ((ulong)endAddr < (ulong)position)
{
throw new ArgumentOutOfRangeException(nameof(position));
}
int offset = startIndex;
while ((ulong)position < (ulong)endAddr)
{
long pageLimit = (position + PageSize) & ~(long)PageMask;
if ((ulong)pageLimit > (ulong)endAddr)
{
pageLimit = endAddr;
}
int copySize = (int)(pageLimit - position);
Marshal.Copy(data, offset, Translate(position), copySize);
position += copySize;
offset += copySize;
}
}
public void CopyBytes(long src, long dst, long size)
{
// Note: This will be moved later.
if (IsContiguous(src, size) &&
IsContiguous(dst, size))
{
byte* srcPtr = (byte*)Translate(src);
byte* dstPtr = (byte*)Translate(dst);
Buffer.MemoryCopy(srcPtr, dstPtr, size, size);
}
else
{
WriteBytes(dst, ReadBytes(src, size));
}
}
public void Dispose()
{
Dispose(true);
}
protected virtual void Dispose(bool disposing)
{
IntPtr ptr = Interlocked.Exchange(ref _pageTable, IntPtr.Zero);
if (ptr != IntPtr.Zero)
{
FreePageTableEntry(ptr, PageBits);
}
}
private void FreePageTableEntry(IntPtr ptr, int levelBitEnd)
{
levelBitEnd += PtLevelBits;
if (levelBitEnd >= AddressSpaceBits)
{
Free(ptr);
return;
}
for (int index = 0; index < PtLevelSize; index++)
{
IntPtr ptePtr = ((IntPtr*)ptr)[index];
if (ptePtr != IntPtr.Zero)
{
FreePageTableEntry(ptePtr, levelBitEnd);
}
}
Free(ptr);
}
}
}