ryujinx/Ryujinx.Tests/Cpu/CpuTest32.cs

533 lines
19 KiB
C#
Raw Normal View History

Add most of the A32 instruction set to ARMeilleure (#897) * Implement TEQ and MOV (Imm16) * Initial work on A32 instructions + SVC. No tests yet, hangs in rtld. * Implement CLZ, fix BFI and BFC Now stops on SIMD initialization. * Exclusive access instructions, fix to mul, system instructions. Now gets to a break after SignalProcessWideKey64. * Better impl of UBFX, add UDIV and SDIV Now boots way further - now stuck on VMOV instruction. * Many more instructions, start on SIMD and testing framework. * Fix build issues * svc: Rework 32 bit codepath Fixing once and for all argument ordering issues. * Fix 32 bits stacktrace * hle debug: Add 32 bits dynamic section parsing * Fix highCq mode, add many tests, fix some instruction bugs Still suffers from critical malloc failure :weary: * Fix incorrect opcode decoders and a few more instructions. * Add a few instructions and fix others. re-disable highCq for now. Disabled the svc memory clear since i'm not sure about it. * Fix build * Fix typo in ordered/exclusive stores. * Implement some more instructions, fix others. Uxtab16/Sxtab16 are untested. * Begin impl of pairwise, some other instructions. * Add a few more instructions, a quick hack to fix svcs for now. * Add tests and fix issues with VTRN, VZIP, VUZP * Add a few more instructions, fix Vmul_1 encoding. * Fix way too many instruction bugs, add tests for some of the more important ones. * Fix HighCq, enable FastFP paths for some floating point instructions (not entirely sure why these were disabled, so important to note this commit exists) Branching has been removed in A32 shifts until I figure out if it's worth it * Cleanup Part 1 There should be no functional change between these next few commits. Should is the key word. (except for removing break handler) * Implement 32 bits syscalls Co-authored-by: riperiperi <rhy3756547@hotmail.com> Implement all 32 bits counterparts of the 64 bits syscalls we currently have. * Refactor part 2: Move index/subindex logic to Operand May have inadvertently fixed one (1) bug * Add FlushProcessDataCache32 * Address jd's comments * Remove 16 bit encodings from OpCodeTable Still need to catch some edge cases (operands that use the "F" flag) and make Q encodings with non-even indexes undefined. * Correct Fpscr handling for FP vector slow paths WIP * Add StandardFPSCRValue behaviour for all Arithmetic instructions * Add StandardFPSCRValue behaviour to compare instructions. * Force passing of fpcr to FPProcessException and FPUnpack. Reduces potential for code error significantly * OpCode cleanup * Remove urgency from DMB comment in MRRC DMB is currently a no-op via the instruction, so it should likely still be a no-op here. * Test Cleanup * Fix FPDefaultNaN on Ryzen CPUs * Improve some tests, fix some shift instructions, add slow path for Vadd * Fix Typo * More test cleanup * Flip order of Fx and index, to indicate that the operand's is the "base" * Remove Simd32 register type, use Int32 and Int64 for scalars like A64 does. * Reintroduce alignment to DecoderHelper (removed by accident) * One more realign as reading diffs is hard * Use I32 registers in A32 (part 2) Swap default integer register type based on current execution mode. * FPSCR flags as Registers (part 1) Still need to change NativeContext and ExecutionContext to allow getting/setting with the flag values. * Use I32 registers in A32 (part 1) * FPSCR flags as registers (part 2) Only CMP flags are on the registers right now. It could be useful to use more of the space in non-fast-float when implementing A32 flags accurately in the fast path. * Address Feedback * Correct FP->Int behaviour (should saturate) * Make branches made by writing to PC eligible for Rejit Greatly improves performance in most games. * Remove unused branching for Vtbl * RejitRequest as a class rather than a tuple Makes a lot more sense than storing tuples on a dictionary. * Add VMOVN, VSHR (imm), VSHRN (imm) and related tests * Re-order InstEmitSystem32 Alphabetical sorting. * Address Feedback Feedback from Ac_K, remove and sort usings. * Address Feedback 2 * Address Feedback from LDj3SNuD Opcode table reordered to have alphabetical sorting within groups, Vmaxnm and Vminnm have split names to be less ambiguous, SoftFloat nits, Test nits and Test simplification with ValueSource. * Add Debug Asserts to A32 helpers Mainly to prevent the shift ones from being used on I64 operands, as they expect I32 input for most operations (eg. carry flag setting), and expect I32 input for shift and boolean amounts. Most other helper functions don't take Operands, throw on out of range values, and take specific types of OpCode, so didn't need any asserts. * Use ConstF rather than creating an operand. (useful for pooling in future) * Move exclusive load to helper, reference call flag rather than literal 1. * Address LDj feedback (minus table flatten) one final look before it's all gone. the world is so beautiful. * Flatten OpCodeTable oh no * Address more table ordering * Call Flag as int on A32 Co-authored-by: Natalie C. <cyuubiapps@gmail.com> Co-authored-by: Thog <thog@protonmail.com>
2020-02-23 21:20:40 +00:00
using ARMeilleure.Memory;
using ARMeilleure.State;
using ARMeilleure.Translation;
using NUnit.Framework;
using Ryujinx.Tests.Unicorn;
using System;
using System.Runtime.InteropServices;
namespace Ryujinx.Tests.Cpu
{
[TestFixture]
public class CpuTest32
{
private uint _currAddress;
private long _size;
private uint _entryPoint;
private IntPtr _ramPointer;
private MemoryManager _memory;
private ExecutionContext _context;
private Translator _translator;
private static bool _unicornAvailable;
private UnicornAArch32 _unicornEmu;
private bool usingMemory;
static CpuTest32()
{
_unicornAvailable = UnicornAArch32.IsAvailable();
if (!_unicornAvailable)
{
Console.WriteLine("WARNING: Could not find Unicorn.");
}
}
[SetUp]
public void Setup()
{
_currAddress = 0x1000;
_size = 0x1000;
_entryPoint = _currAddress;
_ramPointer = Marshal.AllocHGlobal(new IntPtr(_size * 2));
_memory = new MemoryManager(_ramPointer, addressSpaceBits: 16, useFlatPageTable: true);
_memory.Map((long)_currAddress, 0, _size*2);
_context = new ExecutionContext();
_context.IsAarch32 = true;
_translator = new Translator(_memory);
if (_unicornAvailable)
{
_unicornEmu = new UnicornAArch32();
_unicornEmu.MemoryMap(_currAddress, (ulong)_size, MemoryPermission.READ | MemoryPermission.EXEC);
_unicornEmu.MemoryMap((ulong)(_currAddress + _size), (ulong)_size, MemoryPermission.READ | MemoryPermission.WRITE);
_unicornEmu.PC = _entryPoint;
}
}
[TearDown]
public void Teardown()
{
Marshal.FreeHGlobal(_ramPointer);
_memory = null;
_context = null;
_translator = null;
_unicornEmu = null;
}
protected void Reset()
{
Teardown();
Setup();
}
protected void Opcode(uint opcode)
{
_memory.WriteUInt32((long)_currAddress, opcode);
if (_unicornAvailable)
{
_unicornEmu.MemoryWrite32((ulong)_currAddress, opcode);
}
_currAddress += 4;
}
protected ExecutionContext GetContext() => _context;
protected void SetContext(uint r0 = 0,
uint r1 = 0,
uint r2 = 0,
uint r3 = 0,
uint sp = 0,
V128 v0 = default,
V128 v1 = default,
V128 v2 = default,
V128 v3 = default,
V128 v4 = default,
V128 v5 = default,
V128 v14 = default,
V128 v15 = default,
bool overflow = false,
bool carry = false,
bool zero = false,
bool negative = false,
int fpscr = 0)
{
_context.SetX(0, r0);
_context.SetX(1, r1);
_context.SetX(2, r2);
_context.SetX(3, r3);
_context.SetX(0xd, sp);
_context.SetV(0, v0);
_context.SetV(1, v1);
_context.SetV(2, v2);
_context.SetV(3, v3);
_context.SetV(4, v4);
_context.SetV(5, v5);
_context.SetV(14, v14);
_context.SetV(15, v15);
_context.SetPstateFlag(PState.VFlag, overflow);
_context.SetPstateFlag(PState.CFlag, carry);
_context.SetPstateFlag(PState.ZFlag, zero);
_context.SetPstateFlag(PState.NFlag, negative);
_context.Fpsr = FPSR.A32Mask & (FPSR)fpscr;
_context.Fpcr = FPCR.A32Mask & (FPCR)fpscr;
if (_unicornAvailable)
{
_unicornEmu.R[0] = r0;
_unicornEmu.R[1] = r1;
_unicornEmu.R[2] = r2;
_unicornEmu.R[3] = r3;
_unicornEmu.SP = sp;
_unicornEmu.Q[0] = V128ToSimdValue(v0);
_unicornEmu.Q[1] = V128ToSimdValue(v1);
_unicornEmu.Q[2] = V128ToSimdValue(v2);
_unicornEmu.Q[3] = V128ToSimdValue(v3);
_unicornEmu.Q[4] = V128ToSimdValue(v4);
_unicornEmu.Q[5] = V128ToSimdValue(v5);
_unicornEmu.Q[14] = V128ToSimdValue(v14);
_unicornEmu.Q[15] = V128ToSimdValue(v15);
_unicornEmu.OverflowFlag = overflow;
_unicornEmu.CarryFlag = carry;
_unicornEmu.ZeroFlag = zero;
_unicornEmu.NegativeFlag = negative;
_unicornEmu.Fpscr = fpscr;
}
}
protected void ExecuteOpcodes(bool runUnicorn = true)
Add most of the A32 instruction set to ARMeilleure (#897) * Implement TEQ and MOV (Imm16) * Initial work on A32 instructions + SVC. No tests yet, hangs in rtld. * Implement CLZ, fix BFI and BFC Now stops on SIMD initialization. * Exclusive access instructions, fix to mul, system instructions. Now gets to a break after SignalProcessWideKey64. * Better impl of UBFX, add UDIV and SDIV Now boots way further - now stuck on VMOV instruction. * Many more instructions, start on SIMD and testing framework. * Fix build issues * svc: Rework 32 bit codepath Fixing once and for all argument ordering issues. * Fix 32 bits stacktrace * hle debug: Add 32 bits dynamic section parsing * Fix highCq mode, add many tests, fix some instruction bugs Still suffers from critical malloc failure :weary: * Fix incorrect opcode decoders and a few more instructions. * Add a few instructions and fix others. re-disable highCq for now. Disabled the svc memory clear since i'm not sure about it. * Fix build * Fix typo in ordered/exclusive stores. * Implement some more instructions, fix others. Uxtab16/Sxtab16 are untested. * Begin impl of pairwise, some other instructions. * Add a few more instructions, a quick hack to fix svcs for now. * Add tests and fix issues with VTRN, VZIP, VUZP * Add a few more instructions, fix Vmul_1 encoding. * Fix way too many instruction bugs, add tests for some of the more important ones. * Fix HighCq, enable FastFP paths for some floating point instructions (not entirely sure why these were disabled, so important to note this commit exists) Branching has been removed in A32 shifts until I figure out if it's worth it * Cleanup Part 1 There should be no functional change between these next few commits. Should is the key word. (except for removing break handler) * Implement 32 bits syscalls Co-authored-by: riperiperi <rhy3756547@hotmail.com> Implement all 32 bits counterparts of the 64 bits syscalls we currently have. * Refactor part 2: Move index/subindex logic to Operand May have inadvertently fixed one (1) bug * Add FlushProcessDataCache32 * Address jd's comments * Remove 16 bit encodings from OpCodeTable Still need to catch some edge cases (operands that use the "F" flag) and make Q encodings with non-even indexes undefined. * Correct Fpscr handling for FP vector slow paths WIP * Add StandardFPSCRValue behaviour for all Arithmetic instructions * Add StandardFPSCRValue behaviour to compare instructions. * Force passing of fpcr to FPProcessException and FPUnpack. Reduces potential for code error significantly * OpCode cleanup * Remove urgency from DMB comment in MRRC DMB is currently a no-op via the instruction, so it should likely still be a no-op here. * Test Cleanup * Fix FPDefaultNaN on Ryzen CPUs * Improve some tests, fix some shift instructions, add slow path for Vadd * Fix Typo * More test cleanup * Flip order of Fx and index, to indicate that the operand's is the "base" * Remove Simd32 register type, use Int32 and Int64 for scalars like A64 does. * Reintroduce alignment to DecoderHelper (removed by accident) * One more realign as reading diffs is hard * Use I32 registers in A32 (part 2) Swap default integer register type based on current execution mode. * FPSCR flags as Registers (part 1) Still need to change NativeContext and ExecutionContext to allow getting/setting with the flag values. * Use I32 registers in A32 (part 1) * FPSCR flags as registers (part 2) Only CMP flags are on the registers right now. It could be useful to use more of the space in non-fast-float when implementing A32 flags accurately in the fast path. * Address Feedback * Correct FP->Int behaviour (should saturate) * Make branches made by writing to PC eligible for Rejit Greatly improves performance in most games. * Remove unused branching for Vtbl * RejitRequest as a class rather than a tuple Makes a lot more sense than storing tuples on a dictionary. * Add VMOVN, VSHR (imm), VSHRN (imm) and related tests * Re-order InstEmitSystem32 Alphabetical sorting. * Address Feedback Feedback from Ac_K, remove and sort usings. * Address Feedback 2 * Address Feedback from LDj3SNuD Opcode table reordered to have alphabetical sorting within groups, Vmaxnm and Vminnm have split names to be less ambiguous, SoftFloat nits, Test nits and Test simplification with ValueSource. * Add Debug Asserts to A32 helpers Mainly to prevent the shift ones from being used on I64 operands, as they expect I32 input for most operations (eg. carry flag setting), and expect I32 input for shift and boolean amounts. Most other helper functions don't take Operands, throw on out of range values, and take specific types of OpCode, so didn't need any asserts. * Use ConstF rather than creating an operand. (useful for pooling in future) * Move exclusive load to helper, reference call flag rather than literal 1. * Address LDj feedback (minus table flatten) one final look before it's all gone. the world is so beautiful. * Flatten OpCodeTable oh no * Address more table ordering * Call Flag as int on A32 Co-authored-by: Natalie C. <cyuubiapps@gmail.com> Co-authored-by: Thog <thog@protonmail.com>
2020-02-23 21:20:40 +00:00
{
_translator.Execute(_context, _entryPoint);
if (_unicornAvailable && runUnicorn)
Add most of the A32 instruction set to ARMeilleure (#897) * Implement TEQ and MOV (Imm16) * Initial work on A32 instructions + SVC. No tests yet, hangs in rtld. * Implement CLZ, fix BFI and BFC Now stops on SIMD initialization. * Exclusive access instructions, fix to mul, system instructions. Now gets to a break after SignalProcessWideKey64. * Better impl of UBFX, add UDIV and SDIV Now boots way further - now stuck on VMOV instruction. * Many more instructions, start on SIMD and testing framework. * Fix build issues * svc: Rework 32 bit codepath Fixing once and for all argument ordering issues. * Fix 32 bits stacktrace * hle debug: Add 32 bits dynamic section parsing * Fix highCq mode, add many tests, fix some instruction bugs Still suffers from critical malloc failure :weary: * Fix incorrect opcode decoders and a few more instructions. * Add a few instructions and fix others. re-disable highCq for now. Disabled the svc memory clear since i'm not sure about it. * Fix build * Fix typo in ordered/exclusive stores. * Implement some more instructions, fix others. Uxtab16/Sxtab16 are untested. * Begin impl of pairwise, some other instructions. * Add a few more instructions, a quick hack to fix svcs for now. * Add tests and fix issues with VTRN, VZIP, VUZP * Add a few more instructions, fix Vmul_1 encoding. * Fix way too many instruction bugs, add tests for some of the more important ones. * Fix HighCq, enable FastFP paths for some floating point instructions (not entirely sure why these were disabled, so important to note this commit exists) Branching has been removed in A32 shifts until I figure out if it's worth it * Cleanup Part 1 There should be no functional change between these next few commits. Should is the key word. (except for removing break handler) * Implement 32 bits syscalls Co-authored-by: riperiperi <rhy3756547@hotmail.com> Implement all 32 bits counterparts of the 64 bits syscalls we currently have. * Refactor part 2: Move index/subindex logic to Operand May have inadvertently fixed one (1) bug * Add FlushProcessDataCache32 * Address jd's comments * Remove 16 bit encodings from OpCodeTable Still need to catch some edge cases (operands that use the "F" flag) and make Q encodings with non-even indexes undefined. * Correct Fpscr handling for FP vector slow paths WIP * Add StandardFPSCRValue behaviour for all Arithmetic instructions * Add StandardFPSCRValue behaviour to compare instructions. * Force passing of fpcr to FPProcessException and FPUnpack. Reduces potential for code error significantly * OpCode cleanup * Remove urgency from DMB comment in MRRC DMB is currently a no-op via the instruction, so it should likely still be a no-op here. * Test Cleanup * Fix FPDefaultNaN on Ryzen CPUs * Improve some tests, fix some shift instructions, add slow path for Vadd * Fix Typo * More test cleanup * Flip order of Fx and index, to indicate that the operand's is the "base" * Remove Simd32 register type, use Int32 and Int64 for scalars like A64 does. * Reintroduce alignment to DecoderHelper (removed by accident) * One more realign as reading diffs is hard * Use I32 registers in A32 (part 2) Swap default integer register type based on current execution mode. * FPSCR flags as Registers (part 1) Still need to change NativeContext and ExecutionContext to allow getting/setting with the flag values. * Use I32 registers in A32 (part 1) * FPSCR flags as registers (part 2) Only CMP flags are on the registers right now. It could be useful to use more of the space in non-fast-float when implementing A32 flags accurately in the fast path. * Address Feedback * Correct FP->Int behaviour (should saturate) * Make branches made by writing to PC eligible for Rejit Greatly improves performance in most games. * Remove unused branching for Vtbl * RejitRequest as a class rather than a tuple Makes a lot more sense than storing tuples on a dictionary. * Add VMOVN, VSHR (imm), VSHRN (imm) and related tests * Re-order InstEmitSystem32 Alphabetical sorting. * Address Feedback Feedback from Ac_K, remove and sort usings. * Address Feedback 2 * Address Feedback from LDj3SNuD Opcode table reordered to have alphabetical sorting within groups, Vmaxnm and Vminnm have split names to be less ambiguous, SoftFloat nits, Test nits and Test simplification with ValueSource. * Add Debug Asserts to A32 helpers Mainly to prevent the shift ones from being used on I64 operands, as they expect I32 input for most operations (eg. carry flag setting), and expect I32 input for shift and boolean amounts. Most other helper functions don't take Operands, throw on out of range values, and take specific types of OpCode, so didn't need any asserts. * Use ConstF rather than creating an operand. (useful for pooling in future) * Move exclusive load to helper, reference call flag rather than literal 1. * Address LDj feedback (minus table flatten) one final look before it's all gone. the world is so beautiful. * Flatten OpCodeTable oh no * Address more table ordering * Call Flag as int on A32 Co-authored-by: Natalie C. <cyuubiapps@gmail.com> Co-authored-by: Thog <thog@protonmail.com>
2020-02-23 21:20:40 +00:00
{
_unicornEmu.RunForCount((ulong)(_currAddress - _entryPoint - 4) / 4);
}
}
protected ExecutionContext SingleOpcode(uint opcode,
uint r0 = 0,
uint r1 = 0,
uint r2 = 0,
uint r3 = 0,
uint sp = 0,
V128 v0 = default,
V128 v1 = default,
V128 v2 = default,
V128 v3 = default,
V128 v4 = default,
V128 v5 = default,
V128 v14 = default,
V128 v15 = default,
bool overflow = false,
bool carry = false,
bool zero = false,
bool negative = false,
int fpscr = 0,
bool copyFpFlags = false,
bool runUnicorn = true)
Add most of the A32 instruction set to ARMeilleure (#897) * Implement TEQ and MOV (Imm16) * Initial work on A32 instructions + SVC. No tests yet, hangs in rtld. * Implement CLZ, fix BFI and BFC Now stops on SIMD initialization. * Exclusive access instructions, fix to mul, system instructions. Now gets to a break after SignalProcessWideKey64. * Better impl of UBFX, add UDIV and SDIV Now boots way further - now stuck on VMOV instruction. * Many more instructions, start on SIMD and testing framework. * Fix build issues * svc: Rework 32 bit codepath Fixing once and for all argument ordering issues. * Fix 32 bits stacktrace * hle debug: Add 32 bits dynamic section parsing * Fix highCq mode, add many tests, fix some instruction bugs Still suffers from critical malloc failure :weary: * Fix incorrect opcode decoders and a few more instructions. * Add a few instructions and fix others. re-disable highCq for now. Disabled the svc memory clear since i'm not sure about it. * Fix build * Fix typo in ordered/exclusive stores. * Implement some more instructions, fix others. Uxtab16/Sxtab16 are untested. * Begin impl of pairwise, some other instructions. * Add a few more instructions, a quick hack to fix svcs for now. * Add tests and fix issues with VTRN, VZIP, VUZP * Add a few more instructions, fix Vmul_1 encoding. * Fix way too many instruction bugs, add tests for some of the more important ones. * Fix HighCq, enable FastFP paths for some floating point instructions (not entirely sure why these were disabled, so important to note this commit exists) Branching has been removed in A32 shifts until I figure out if it's worth it * Cleanup Part 1 There should be no functional change between these next few commits. Should is the key word. (except for removing break handler) * Implement 32 bits syscalls Co-authored-by: riperiperi <rhy3756547@hotmail.com> Implement all 32 bits counterparts of the 64 bits syscalls we currently have. * Refactor part 2: Move index/subindex logic to Operand May have inadvertently fixed one (1) bug * Add FlushProcessDataCache32 * Address jd's comments * Remove 16 bit encodings from OpCodeTable Still need to catch some edge cases (operands that use the "F" flag) and make Q encodings with non-even indexes undefined. * Correct Fpscr handling for FP vector slow paths WIP * Add StandardFPSCRValue behaviour for all Arithmetic instructions * Add StandardFPSCRValue behaviour to compare instructions. * Force passing of fpcr to FPProcessException and FPUnpack. Reduces potential for code error significantly * OpCode cleanup * Remove urgency from DMB comment in MRRC DMB is currently a no-op via the instruction, so it should likely still be a no-op here. * Test Cleanup * Fix FPDefaultNaN on Ryzen CPUs * Improve some tests, fix some shift instructions, add slow path for Vadd * Fix Typo * More test cleanup * Flip order of Fx and index, to indicate that the operand's is the "base" * Remove Simd32 register type, use Int32 and Int64 for scalars like A64 does. * Reintroduce alignment to DecoderHelper (removed by accident) * One more realign as reading diffs is hard * Use I32 registers in A32 (part 2) Swap default integer register type based on current execution mode. * FPSCR flags as Registers (part 1) Still need to change NativeContext and ExecutionContext to allow getting/setting with the flag values. * Use I32 registers in A32 (part 1) * FPSCR flags as registers (part 2) Only CMP flags are on the registers right now. It could be useful to use more of the space in non-fast-float when implementing A32 flags accurately in the fast path. * Address Feedback * Correct FP->Int behaviour (should saturate) * Make branches made by writing to PC eligible for Rejit Greatly improves performance in most games. * Remove unused branching for Vtbl * RejitRequest as a class rather than a tuple Makes a lot more sense than storing tuples on a dictionary. * Add VMOVN, VSHR (imm), VSHRN (imm) and related tests * Re-order InstEmitSystem32 Alphabetical sorting. * Address Feedback Feedback from Ac_K, remove and sort usings. * Address Feedback 2 * Address Feedback from LDj3SNuD Opcode table reordered to have alphabetical sorting within groups, Vmaxnm and Vminnm have split names to be less ambiguous, SoftFloat nits, Test nits and Test simplification with ValueSource. * Add Debug Asserts to A32 helpers Mainly to prevent the shift ones from being used on I64 operands, as they expect I32 input for most operations (eg. carry flag setting), and expect I32 input for shift and boolean amounts. Most other helper functions don't take Operands, throw on out of range values, and take specific types of OpCode, so didn't need any asserts. * Use ConstF rather than creating an operand. (useful for pooling in future) * Move exclusive load to helper, reference call flag rather than literal 1. * Address LDj feedback (minus table flatten) one final look before it's all gone. the world is so beautiful. * Flatten OpCodeTable oh no * Address more table ordering * Call Flag as int on A32 Co-authored-by: Natalie C. <cyuubiapps@gmail.com> Co-authored-by: Thog <thog@protonmail.com>
2020-02-23 21:20:40 +00:00
{
Opcode(opcode);
if (copyFpFlags)
{
Opcode(0xeef1fa10);
}
Opcode(0xe12fff1e); // BX LR
SetContext(r0, r1, r2, r3, sp, v0, v1, v2, v3, v4, v5, v14, v15, overflow, carry, zero, negative, fpscr);
ExecuteOpcodes(runUnicorn);
Add most of the A32 instruction set to ARMeilleure (#897) * Implement TEQ and MOV (Imm16) * Initial work on A32 instructions + SVC. No tests yet, hangs in rtld. * Implement CLZ, fix BFI and BFC Now stops on SIMD initialization. * Exclusive access instructions, fix to mul, system instructions. Now gets to a break after SignalProcessWideKey64. * Better impl of UBFX, add UDIV and SDIV Now boots way further - now stuck on VMOV instruction. * Many more instructions, start on SIMD and testing framework. * Fix build issues * svc: Rework 32 bit codepath Fixing once and for all argument ordering issues. * Fix 32 bits stacktrace * hle debug: Add 32 bits dynamic section parsing * Fix highCq mode, add many tests, fix some instruction bugs Still suffers from critical malloc failure :weary: * Fix incorrect opcode decoders and a few more instructions. * Add a few instructions and fix others. re-disable highCq for now. Disabled the svc memory clear since i'm not sure about it. * Fix build * Fix typo in ordered/exclusive stores. * Implement some more instructions, fix others. Uxtab16/Sxtab16 are untested. * Begin impl of pairwise, some other instructions. * Add a few more instructions, a quick hack to fix svcs for now. * Add tests and fix issues with VTRN, VZIP, VUZP * Add a few more instructions, fix Vmul_1 encoding. * Fix way too many instruction bugs, add tests for some of the more important ones. * Fix HighCq, enable FastFP paths for some floating point instructions (not entirely sure why these were disabled, so important to note this commit exists) Branching has been removed in A32 shifts until I figure out if it's worth it * Cleanup Part 1 There should be no functional change between these next few commits. Should is the key word. (except for removing break handler) * Implement 32 bits syscalls Co-authored-by: riperiperi <rhy3756547@hotmail.com> Implement all 32 bits counterparts of the 64 bits syscalls we currently have. * Refactor part 2: Move index/subindex logic to Operand May have inadvertently fixed one (1) bug * Add FlushProcessDataCache32 * Address jd's comments * Remove 16 bit encodings from OpCodeTable Still need to catch some edge cases (operands that use the "F" flag) and make Q encodings with non-even indexes undefined. * Correct Fpscr handling for FP vector slow paths WIP * Add StandardFPSCRValue behaviour for all Arithmetic instructions * Add StandardFPSCRValue behaviour to compare instructions. * Force passing of fpcr to FPProcessException and FPUnpack. Reduces potential for code error significantly * OpCode cleanup * Remove urgency from DMB comment in MRRC DMB is currently a no-op via the instruction, so it should likely still be a no-op here. * Test Cleanup * Fix FPDefaultNaN on Ryzen CPUs * Improve some tests, fix some shift instructions, add slow path for Vadd * Fix Typo * More test cleanup * Flip order of Fx and index, to indicate that the operand's is the "base" * Remove Simd32 register type, use Int32 and Int64 for scalars like A64 does. * Reintroduce alignment to DecoderHelper (removed by accident) * One more realign as reading diffs is hard * Use I32 registers in A32 (part 2) Swap default integer register type based on current execution mode. * FPSCR flags as Registers (part 1) Still need to change NativeContext and ExecutionContext to allow getting/setting with the flag values. * Use I32 registers in A32 (part 1) * FPSCR flags as registers (part 2) Only CMP flags are on the registers right now. It could be useful to use more of the space in non-fast-float when implementing A32 flags accurately in the fast path. * Address Feedback * Correct FP->Int behaviour (should saturate) * Make branches made by writing to PC eligible for Rejit Greatly improves performance in most games. * Remove unused branching for Vtbl * RejitRequest as a class rather than a tuple Makes a lot more sense than storing tuples on a dictionary. * Add VMOVN, VSHR (imm), VSHRN (imm) and related tests * Re-order InstEmitSystem32 Alphabetical sorting. * Address Feedback Feedback from Ac_K, remove and sort usings. * Address Feedback 2 * Address Feedback from LDj3SNuD Opcode table reordered to have alphabetical sorting within groups, Vmaxnm and Vminnm have split names to be less ambiguous, SoftFloat nits, Test nits and Test simplification with ValueSource. * Add Debug Asserts to A32 helpers Mainly to prevent the shift ones from being used on I64 operands, as they expect I32 input for most operations (eg. carry flag setting), and expect I32 input for shift and boolean amounts. Most other helper functions don't take Operands, throw on out of range values, and take specific types of OpCode, so didn't need any asserts. * Use ConstF rather than creating an operand. (useful for pooling in future) * Move exclusive load to helper, reference call flag rather than literal 1. * Address LDj feedback (minus table flatten) one final look before it's all gone. the world is so beautiful. * Flatten OpCodeTable oh no * Address more table ordering * Call Flag as int on A32 Co-authored-by: Natalie C. <cyuubiapps@gmail.com> Co-authored-by: Thog <thog@protonmail.com>
2020-02-23 21:20:40 +00:00
return GetContext();
}
protected void SetWorkingMemory(byte[] data)
{
_memory.WriteBytes(0x2000, data);
if (_unicornAvailable)
{
_unicornEmu.MemoryWrite((ulong)(0x2000), data);
}
usingMemory = true; // When true, CompareAgainstUnicorn checks the working memory for equality too.
}
/// <summary>Rounding Mode control field.</summary>
public enum RMode
{
/// <summary>Round to Nearest mode.</summary>
Rn,
/// <summary>Round towards Plus Infinity mode.</summary>
Rp,
/// <summary>Round towards Minus Infinity mode.</summary>
Rm,
/// <summary>Round towards Zero mode.</summary>
Rz
};
/// <summary>Floating-point Control Register.</summary>
protected enum Fpcr
{
/// <summary>Rounding Mode control field.</summary>
RMode = 22,
/// <summary>Flush-to-zero mode control bit.</summary>
Fz = 24,
/// <summary>Default NaN mode control bit.</summary>
Dn = 25,
/// <summary>Alternative half-precision control bit.</summary>
Ahp = 26
}
/// <summary>Floating-point Status Register.</summary>
[Flags]
protected enum Fpsr
{
None = 0,
/// <summary>Invalid Operation cumulative floating-point exception bit.</summary>
Ioc = 1 << 0,
/// <summary>Divide by Zero cumulative floating-point exception bit.</summary>
Dzc = 1 << 1,
/// <summary>Overflow cumulative floating-point exception bit.</summary>
Ofc = 1 << 2,
/// <summary>Underflow cumulative floating-point exception bit.</summary>
Ufc = 1 << 3,
/// <summary>Inexact cumulative floating-point exception bit.</summary>
Ixc = 1 << 4,
/// <summary>Input Denormal cumulative floating-point exception bit.</summary>
Idc = 1 << 7,
/// <summary>Cumulative saturation bit.</summary>
Qc = 1 << 27,
/// <summary>NZCV flags</summary>
Nzcv = (1 << 28) | (1 << 29) | (1 << 30) | (1 << 31)
}
[Flags]
protected enum FpSkips
{
None = 0,
IfNaNS = 1,
IfNaND = 2,
IfUnderflow = 4,
IfOverflow = 8
}
protected enum FpTolerances
{
None,
UpToOneUlpsS,
UpToOneUlpsD
}
protected void CompareAgainstUnicorn(
Fpsr fpsrMask = Fpsr.None,
FpSkips fpSkips = FpSkips.None,
FpTolerances fpTolerances = FpTolerances.None)
{
if (!_unicornAvailable)
{
return;
}
if (fpSkips != FpSkips.None)
{
ManageFpSkips(fpSkips);
}
Assert.That(_context.GetX(0), Is.EqualTo(_unicornEmu.R[0]));
Assert.That(_context.GetX(1), Is.EqualTo(_unicornEmu.R[1]));
Assert.That(_context.GetX(2), Is.EqualTo(_unicornEmu.R[2]));
Assert.That(_context.GetX(3), Is.EqualTo(_unicornEmu.R[3]));
Assert.That(_context.GetX(4), Is.EqualTo(_unicornEmu.R[4]));
Assert.That(_context.GetX(5), Is.EqualTo(_unicornEmu.R[5]));
Assert.That(_context.GetX(6), Is.EqualTo(_unicornEmu.R[6]));
Assert.That(_context.GetX(7), Is.EqualTo(_unicornEmu.R[7]));
Assert.That(_context.GetX(8), Is.EqualTo(_unicornEmu.R[8]));
Assert.That(_context.GetX(9), Is.EqualTo(_unicornEmu.R[9]));
Assert.That(_context.GetX(10), Is.EqualTo(_unicornEmu.R[10]));
Assert.That(_context.GetX(11), Is.EqualTo(_unicornEmu.R[11]));
Assert.That(_context.GetX(12), Is.EqualTo(_unicornEmu.R[12]));
Assert.That(_context.GetX(13), Is.EqualTo(_unicornEmu.R[13]));
Assert.That(_context.GetX(14), Is.EqualTo(_unicornEmu.R[14]));
if (fpTolerances == FpTolerances.None)
{
Assert.That(V128ToSimdValue(_context.GetV(0)), Is.EqualTo(_unicornEmu.Q[0]));
}
else
{
ManageFpTolerances(fpTolerances);
}
Assert.That(V128ToSimdValue(_context.GetV(1)), Is.EqualTo(_unicornEmu.Q[1]));
Assert.That(V128ToSimdValue(_context.GetV(2)), Is.EqualTo(_unicornEmu.Q[2]));
Assert.That(V128ToSimdValue(_context.GetV(3)), Is.EqualTo(_unicornEmu.Q[3]));
Assert.That(V128ToSimdValue(_context.GetV(4)), Is.EqualTo(_unicornEmu.Q[4]));
Assert.That(V128ToSimdValue(_context.GetV(5)), Is.EqualTo(_unicornEmu.Q[5]));
Assert.That(V128ToSimdValue(_context.GetV(6)), Is.EqualTo(_unicornEmu.Q[6]));
Assert.That(V128ToSimdValue(_context.GetV(7)), Is.EqualTo(_unicornEmu.Q[7]));
Assert.That(V128ToSimdValue(_context.GetV(8)), Is.EqualTo(_unicornEmu.Q[8]));
Assert.That(V128ToSimdValue(_context.GetV(9)), Is.EqualTo(_unicornEmu.Q[9]));
Assert.That(V128ToSimdValue(_context.GetV(10)), Is.EqualTo(_unicornEmu.Q[10]));
Assert.That(V128ToSimdValue(_context.GetV(11)), Is.EqualTo(_unicornEmu.Q[11]));
Assert.That(V128ToSimdValue(_context.GetV(12)), Is.EqualTo(_unicornEmu.Q[12]));
Assert.That(V128ToSimdValue(_context.GetV(13)), Is.EqualTo(_unicornEmu.Q[13]));
Assert.That(V128ToSimdValue(_context.GetV(14)), Is.EqualTo(_unicornEmu.Q[14]));
Assert.That(V128ToSimdValue(_context.GetV(15)), Is.EqualTo(_unicornEmu.Q[15]));
Assert.That((int)_context.Fpcr | ((int)_context.Fpsr & (int)fpsrMask), Is.EqualTo(_unicornEmu.Fpscr));
Assert.That(_context.GetPstateFlag(PState.QFlag), Is.EqualTo(_unicornEmu.QFlag));
Add most of the A32 instruction set to ARMeilleure (#897) * Implement TEQ and MOV (Imm16) * Initial work on A32 instructions + SVC. No tests yet, hangs in rtld. * Implement CLZ, fix BFI and BFC Now stops on SIMD initialization. * Exclusive access instructions, fix to mul, system instructions. Now gets to a break after SignalProcessWideKey64. * Better impl of UBFX, add UDIV and SDIV Now boots way further - now stuck on VMOV instruction. * Many more instructions, start on SIMD and testing framework. * Fix build issues * svc: Rework 32 bit codepath Fixing once and for all argument ordering issues. * Fix 32 bits stacktrace * hle debug: Add 32 bits dynamic section parsing * Fix highCq mode, add many tests, fix some instruction bugs Still suffers from critical malloc failure :weary: * Fix incorrect opcode decoders and a few more instructions. * Add a few instructions and fix others. re-disable highCq for now. Disabled the svc memory clear since i'm not sure about it. * Fix build * Fix typo in ordered/exclusive stores. * Implement some more instructions, fix others. Uxtab16/Sxtab16 are untested. * Begin impl of pairwise, some other instructions. * Add a few more instructions, a quick hack to fix svcs for now. * Add tests and fix issues with VTRN, VZIP, VUZP * Add a few more instructions, fix Vmul_1 encoding. * Fix way too many instruction bugs, add tests for some of the more important ones. * Fix HighCq, enable FastFP paths for some floating point instructions (not entirely sure why these were disabled, so important to note this commit exists) Branching has been removed in A32 shifts until I figure out if it's worth it * Cleanup Part 1 There should be no functional change between these next few commits. Should is the key word. (except for removing break handler) * Implement 32 bits syscalls Co-authored-by: riperiperi <rhy3756547@hotmail.com> Implement all 32 bits counterparts of the 64 bits syscalls we currently have. * Refactor part 2: Move index/subindex logic to Operand May have inadvertently fixed one (1) bug * Add FlushProcessDataCache32 * Address jd's comments * Remove 16 bit encodings from OpCodeTable Still need to catch some edge cases (operands that use the "F" flag) and make Q encodings with non-even indexes undefined. * Correct Fpscr handling for FP vector slow paths WIP * Add StandardFPSCRValue behaviour for all Arithmetic instructions * Add StandardFPSCRValue behaviour to compare instructions. * Force passing of fpcr to FPProcessException and FPUnpack. Reduces potential for code error significantly * OpCode cleanup * Remove urgency from DMB comment in MRRC DMB is currently a no-op via the instruction, so it should likely still be a no-op here. * Test Cleanup * Fix FPDefaultNaN on Ryzen CPUs * Improve some tests, fix some shift instructions, add slow path for Vadd * Fix Typo * More test cleanup * Flip order of Fx and index, to indicate that the operand's is the "base" * Remove Simd32 register type, use Int32 and Int64 for scalars like A64 does. * Reintroduce alignment to DecoderHelper (removed by accident) * One more realign as reading diffs is hard * Use I32 registers in A32 (part 2) Swap default integer register type based on current execution mode. * FPSCR flags as Registers (part 1) Still need to change NativeContext and ExecutionContext to allow getting/setting with the flag values. * Use I32 registers in A32 (part 1) * FPSCR flags as registers (part 2) Only CMP flags are on the registers right now. It could be useful to use more of the space in non-fast-float when implementing A32 flags accurately in the fast path. * Address Feedback * Correct FP->Int behaviour (should saturate) * Make branches made by writing to PC eligible for Rejit Greatly improves performance in most games. * Remove unused branching for Vtbl * RejitRequest as a class rather than a tuple Makes a lot more sense than storing tuples on a dictionary. * Add VMOVN, VSHR (imm), VSHRN (imm) and related tests * Re-order InstEmitSystem32 Alphabetical sorting. * Address Feedback Feedback from Ac_K, remove and sort usings. * Address Feedback 2 * Address Feedback from LDj3SNuD Opcode table reordered to have alphabetical sorting within groups, Vmaxnm and Vminnm have split names to be less ambiguous, SoftFloat nits, Test nits and Test simplification with ValueSource. * Add Debug Asserts to A32 helpers Mainly to prevent the shift ones from being used on I64 operands, as they expect I32 input for most operations (eg. carry flag setting), and expect I32 input for shift and boolean amounts. Most other helper functions don't take Operands, throw on out of range values, and take specific types of OpCode, so didn't need any asserts. * Use ConstF rather than creating an operand. (useful for pooling in future) * Move exclusive load to helper, reference call flag rather than literal 1. * Address LDj feedback (minus table flatten) one final look before it's all gone. the world is so beautiful. * Flatten OpCodeTable oh no * Address more table ordering * Call Flag as int on A32 Co-authored-by: Natalie C. <cyuubiapps@gmail.com> Co-authored-by: Thog <thog@protonmail.com>
2020-02-23 21:20:40 +00:00
Assert.That(_context.GetPstateFlag(PState.VFlag), Is.EqualTo(_unicornEmu.OverflowFlag));
Assert.That(_context.GetPstateFlag(PState.CFlag), Is.EqualTo(_unicornEmu.CarryFlag));
Assert.That(_context.GetPstateFlag(PState.ZFlag), Is.EqualTo(_unicornEmu.ZeroFlag));
Assert.That(_context.GetPstateFlag(PState.NFlag), Is.EqualTo(_unicornEmu.NegativeFlag));
if (usingMemory)
{
byte[] meilleureMem = _memory.ReadBytes((long)(0x2000), _size);
byte[] unicornMem = _unicornEmu.MemoryRead((ulong)(0x2000), (ulong)_size);
for (int i = 0; i < _size; i++)
Add most of the A32 instruction set to ARMeilleure (#897) * Implement TEQ and MOV (Imm16) * Initial work on A32 instructions + SVC. No tests yet, hangs in rtld. * Implement CLZ, fix BFI and BFC Now stops on SIMD initialization. * Exclusive access instructions, fix to mul, system instructions. Now gets to a break after SignalProcessWideKey64. * Better impl of UBFX, add UDIV and SDIV Now boots way further - now stuck on VMOV instruction. * Many more instructions, start on SIMD and testing framework. * Fix build issues * svc: Rework 32 bit codepath Fixing once and for all argument ordering issues. * Fix 32 bits stacktrace * hle debug: Add 32 bits dynamic section parsing * Fix highCq mode, add many tests, fix some instruction bugs Still suffers from critical malloc failure :weary: * Fix incorrect opcode decoders and a few more instructions. * Add a few instructions and fix others. re-disable highCq for now. Disabled the svc memory clear since i'm not sure about it. * Fix build * Fix typo in ordered/exclusive stores. * Implement some more instructions, fix others. Uxtab16/Sxtab16 are untested. * Begin impl of pairwise, some other instructions. * Add a few more instructions, a quick hack to fix svcs for now. * Add tests and fix issues with VTRN, VZIP, VUZP * Add a few more instructions, fix Vmul_1 encoding. * Fix way too many instruction bugs, add tests for some of the more important ones. * Fix HighCq, enable FastFP paths for some floating point instructions (not entirely sure why these were disabled, so important to note this commit exists) Branching has been removed in A32 shifts until I figure out if it's worth it * Cleanup Part 1 There should be no functional change between these next few commits. Should is the key word. (except for removing break handler) * Implement 32 bits syscalls Co-authored-by: riperiperi <rhy3756547@hotmail.com> Implement all 32 bits counterparts of the 64 bits syscalls we currently have. * Refactor part 2: Move index/subindex logic to Operand May have inadvertently fixed one (1) bug * Add FlushProcessDataCache32 * Address jd's comments * Remove 16 bit encodings from OpCodeTable Still need to catch some edge cases (operands that use the "F" flag) and make Q encodings with non-even indexes undefined. * Correct Fpscr handling for FP vector slow paths WIP * Add StandardFPSCRValue behaviour for all Arithmetic instructions * Add StandardFPSCRValue behaviour to compare instructions. * Force passing of fpcr to FPProcessException and FPUnpack. Reduces potential for code error significantly * OpCode cleanup * Remove urgency from DMB comment in MRRC DMB is currently a no-op via the instruction, so it should likely still be a no-op here. * Test Cleanup * Fix FPDefaultNaN on Ryzen CPUs * Improve some tests, fix some shift instructions, add slow path for Vadd * Fix Typo * More test cleanup * Flip order of Fx and index, to indicate that the operand's is the "base" * Remove Simd32 register type, use Int32 and Int64 for scalars like A64 does. * Reintroduce alignment to DecoderHelper (removed by accident) * One more realign as reading diffs is hard * Use I32 registers in A32 (part 2) Swap default integer register type based on current execution mode. * FPSCR flags as Registers (part 1) Still need to change NativeContext and ExecutionContext to allow getting/setting with the flag values. * Use I32 registers in A32 (part 1) * FPSCR flags as registers (part 2) Only CMP flags are on the registers right now. It could be useful to use more of the space in non-fast-float when implementing A32 flags accurately in the fast path. * Address Feedback * Correct FP->Int behaviour (should saturate) * Make branches made by writing to PC eligible for Rejit Greatly improves performance in most games. * Remove unused branching for Vtbl * RejitRequest as a class rather than a tuple Makes a lot more sense than storing tuples on a dictionary. * Add VMOVN, VSHR (imm), VSHRN (imm) and related tests * Re-order InstEmitSystem32 Alphabetical sorting. * Address Feedback Feedback from Ac_K, remove and sort usings. * Address Feedback 2 * Address Feedback from LDj3SNuD Opcode table reordered to have alphabetical sorting within groups, Vmaxnm and Vminnm have split names to be less ambiguous, SoftFloat nits, Test nits and Test simplification with ValueSource. * Add Debug Asserts to A32 helpers Mainly to prevent the shift ones from being used on I64 operands, as they expect I32 input for most operations (eg. carry flag setting), and expect I32 input for shift and boolean amounts. Most other helper functions don't take Operands, throw on out of range values, and take specific types of OpCode, so didn't need any asserts. * Use ConstF rather than creating an operand. (useful for pooling in future) * Move exclusive load to helper, reference call flag rather than literal 1. * Address LDj feedback (minus table flatten) one final look before it's all gone. the world is so beautiful. * Flatten OpCodeTable oh no * Address more table ordering * Call Flag as int on A32 Co-authored-by: Natalie C. <cyuubiapps@gmail.com> Co-authored-by: Thog <thog@protonmail.com>
2020-02-23 21:20:40 +00:00
{
Assert.AreEqual(meilleureMem[i], unicornMem[i]);
}
}
}
private void ManageFpSkips(FpSkips fpSkips)
{
if (fpSkips.HasFlag(FpSkips.IfNaNS))
{
if (float.IsNaN(_unicornEmu.Q[0].AsFloat()))
{
Assert.Ignore("NaN test.");
}
}
else if (fpSkips.HasFlag(FpSkips.IfNaND))
{
if (double.IsNaN(_unicornEmu.Q[0].AsDouble()))
{
Assert.Ignore("NaN test.");
}
}
if (fpSkips.HasFlag(FpSkips.IfUnderflow))
{
if ((_unicornEmu.Fpscr & (int)Fpsr.Ufc) != 0)
{
Assert.Ignore("Underflow test.");
}
}
if (fpSkips.HasFlag(FpSkips.IfOverflow))
{
if ((_unicornEmu.Fpscr & (int)Fpsr.Ofc) != 0)
{
Assert.Ignore("Overflow test.");
}
}
}
private void ManageFpTolerances(FpTolerances fpTolerances)
{
bool IsNormalOrSubnormalS(float f) => float.IsNormal(f) || float.IsSubnormal(f);
bool IsNormalOrSubnormalD(double d) => double.IsNormal(d) || double.IsSubnormal(d);
if (!Is.EqualTo(_unicornEmu.Q[0]).ApplyTo(V128ToSimdValue(_context.GetV(0))).IsSuccess)
{
if (fpTolerances == FpTolerances.UpToOneUlpsS)
{
if (IsNormalOrSubnormalS(_unicornEmu.Q[0].AsFloat()) &&
IsNormalOrSubnormalS(_context.GetV(0).AsFloat()))
{
Assert.That(_context.GetV(0).GetFloat(0),
Is.EqualTo(_unicornEmu.Q[0].GetFloat(0)).Within(1).Ulps);
Assert.That(_context.GetV(0).GetFloat(1),
Is.EqualTo(_unicornEmu.Q[0].GetFloat(1)).Within(1).Ulps);
Assert.That(_context.GetV(0).GetFloat(2),
Is.EqualTo(_unicornEmu.Q[0].GetFloat(2)).Within(1).Ulps);
Assert.That(_context.GetV(0).GetFloat(3),
Is.EqualTo(_unicornEmu.Q[0].GetFloat(3)).Within(1).Ulps);
Console.WriteLine(fpTolerances);
}
else
{
Assert.That(V128ToSimdValue(_context.GetV(0)), Is.EqualTo(_unicornEmu.Q[0]));
}
}
if (fpTolerances == FpTolerances.UpToOneUlpsD)
{
if (IsNormalOrSubnormalD(_unicornEmu.Q[0].AsDouble()) &&
IsNormalOrSubnormalD(_context.GetV(0).AsDouble()))
{
Assert.That(_context.GetV(0).GetDouble(0),
Is.EqualTo(_unicornEmu.Q[0].GetDouble(0)).Within(1).Ulps);
Assert.That(_context.GetV(0).GetDouble(1),
Is.EqualTo(_unicornEmu.Q[0].GetDouble(1)).Within(1).Ulps);
Console.WriteLine(fpTolerances);
}
else
{
Assert.That(V128ToSimdValue(_context.GetV(0)), Is.EqualTo(_unicornEmu.Q[0]));
}
}
}
}
private static SimdValue V128ToSimdValue(V128 value)
{
return new SimdValue(value.GetUInt64(0), value.GetUInt64(1));
}
protected static V128 MakeVectorScalar(float value) => new V128(value);
protected static V128 MakeVectorScalar(double value) => new V128(value);
protected static V128 MakeVectorE0(ulong e0) => new V128(e0, 0);
protected static V128 MakeVectorE1(ulong e1) => new V128(0, e1);
protected static V128 MakeVectorE0E1(ulong e0, ulong e1) => new V128(e0, e1);
protected static ulong GetVectorE0(V128 vector) => vector.GetUInt64(0);
protected static ulong GetVectorE1(V128 vector) => vector.GetUInt64(1);
protected static ushort GenNormalH()
{
uint rnd;
do rnd = TestContext.CurrentContext.Random.NextUShort();
while ((rnd & 0x7C00u) == 0u ||
(~rnd & 0x7C00u) == 0u);
return (ushort)rnd;
}
protected static ushort GenSubnormalH()
{
uint rnd;
do rnd = TestContext.CurrentContext.Random.NextUShort();
while ((rnd & 0x03FFu) == 0u);
return (ushort)(rnd & 0x83FFu);
}
protected static uint GenNormalS()
{
uint rnd;
do rnd = TestContext.CurrentContext.Random.NextUInt();
while ((rnd & 0x7F800000u) == 0u ||
(~rnd & 0x7F800000u) == 0u);
return rnd;
}
protected static uint GenSubnormalS()
{
uint rnd;
do rnd = TestContext.CurrentContext.Random.NextUInt();
while ((rnd & 0x007FFFFFu) == 0u);
return rnd & 0x807FFFFFu;
}
protected static ulong GenNormalD()
{
ulong rnd;
do rnd = TestContext.CurrentContext.Random.NextULong();
while ((rnd & 0x7FF0000000000000ul) == 0ul ||
(~rnd & 0x7FF0000000000000ul) == 0ul);
return rnd;
}
protected static ulong GenSubnormalD()
{
ulong rnd;
do rnd = TestContext.CurrentContext.Random.NextULong();
while ((rnd & 0x000FFFFFFFFFFFFFul) == 0ul);
return rnd & 0x800FFFFFFFFFFFFFul;
}
}
}