0
0
Fork 0
This repository has been archived on 2024-10-12. You can view files and clone it, but cannot push or open issues or pull requests.
ryujinx-final/Ryujinx.Tests/Cpu/CpuTestSimdMov32.cs

600 lines
22 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
#define SimdMov32
using ARMeilleure.State;
using NUnit.Framework;
namespace Ryujinx.Tests.Cpu
{
[Category("SimdMov32")]
public sealed class CpuTestSimdMov32 : CpuTest32
{
#if SimdMov32
private const int RndCntImm = 2;
[Test, Pairwise, Description("VMOV.I<size> <Dd/Qd>, #<imm>")]
public void Movi_V([Range(0u, 10u)] uint variant,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0x0u)] [Random(1u, 0xffu, RndCntImm)] uint imm,
[Values] bool q)
{
uint[] variants =
{
// I32
0b0000_0,
0b0010_0,
0b0100_0,
0b0110_0,
// I16
0b1000_0,
0b1010_0,
// DT
0b1100_0,
0b1101_0,
0b1110_0,
0b1111_0,
0b1110_1
};
uint opcode = 0xf2800010u; // VMOV.I32 D0, #0
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
uint cmodeOp = variants[variant];
if (q)
{
vd <<= 1;
}
opcode |= ((cmodeOp & 1) << 5) | ((cmodeOp & 0x1e) << 7);
opcode |= (q ? 1u : 0u) << 6;
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 |= (imm & 0xf) | ((imm & 0x70) << 12) | ((imm & 0x80) << 16);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
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
SingleOpcode(opcode);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VMOV.F<size> <Sd>, #<imm>")]
public void Movi_S([Range(2u, 3u)] uint size,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0x0u)] [Random(0u, 0xffu, RndCntImm)] uint imm)
{
uint opcode = 0xeeb00800u;
opcode |= (size & 3) << 8;
opcode |= (imm & 0xf) | ((imm & 0xf0) << 12);
if (size == 2)
{
opcode |= ((vd & 0x1) << 22);
opcode |= ((vd & 0x1e) << 11);
}
else
{
opcode |= ((vd & 0x10) << 18);
opcode |= ((vd & 0xf) << 12);
}
SingleOpcode(opcode);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VMOV <Rd>, <Sd>")]
public void Mov_GP([Values(0u, 1u, 2u, 3u)] uint vn,
[Values(0u, 1u, 2u, 3u)] uint rt,
[Random(RndCntImm)] uint valueRn,
[Random(RndCntImm)] ulong valueVn1,
[Random(RndCntImm)] ulong valueVn2,
[Values] bool op)
{
uint opcode = 0xee000a10u; // VMOV S0, R0
opcode |= (vn & 1) << 7;
opcode |= (vn & 0x1e) << 15;
opcode |= (rt & 0xf) << 12;
if (op) opcode |= 1 << 20;
SingleOpcode(opcode, r0: valueRn, r1: valueRn, r2: valueRn, r3: valueRn, v0: new V128(valueVn1, valueVn2));
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VMOV.<size> <Rt>, <Dn[x]>")]
public void Mov_GP_Elem([Range(0u, 7u)] uint vn,
[Values(0u, 1u, 2u, 3u)] uint rt,
[Range(0u, 2u)] uint size,
[Range(0u, 7u)] uint index,
[Random(1)] uint valueRn,
[Random(1)] ulong valueVn1,
[Random(1)] ulong valueVn2,
[Values] bool op,
[Values] bool u)
{
uint opcode = 0xee000b10u; // VMOV.32 D0[0], R0
uint opEncode = 0b01000;
switch (size)
{
case 0:
opEncode = (0b1000) | index & 7;
break;
case 1:
opEncode = (0b0001) | ((index & 3) << 1);
break;
case 2:
opEncode = (index & 1) << 2;
break;
}
opcode |= ((opEncode >> 2) << 21) | ((opEncode & 3) << 5);
opcode |= (vn & 0x10) << 3;
opcode |= (vn & 0xf) << 16;
opcode |= (rt & 0xf) << 12;
if (op)
{
opcode |= 1 << 20;
if (u && size != 2)
{
opcode |= 1 << 23;
}
}
SingleOpcode(opcode, r0: valueRn, r1: valueRn, r2: valueRn, r3: valueRn, v0: new V128(valueVn1, valueVn2), v1: new V128(valueVn2, valueVn1));
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("(VMOV <Rt>, <Rt2>, <Dm>), (VMOV <Dm>, <Rt>, <Rt2>)")]
public void Mov_GP_D([Values(0u, 1u, 2u, 3u)] uint vm,
[Values(0u, 1u, 2u, 3u)] uint rt,
[Values(0u, 1u, 2u, 3u)] uint rt2,
[Random(RndCntImm)] uint valueRt1,
[Random(RndCntImm)] uint valueRt2,
[Random(RndCntImm)] ulong valueVn1,
[Random(RndCntImm)] ulong valueVn2,
[Values] bool op)
{
uint opcode = 0xec400b10u; // VMOV D0, R0, R0
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
opcode |= (rt & 0xf) << 12;
opcode |= (rt2 & 0xf) << 16;
if (op)
{
opcode |= 1 << 20;
}
SingleOpcode(opcode, r0: valueRt1, r1: valueRt2, r2: valueRt1, r3: valueRt2, v0: new V128(valueVn1, valueVn2));
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("(VMOV <Rt>, <Rt2>, <Sm>, <Sm1>), (VMOV <Sm>, <Sm1>, <Rt>, <Rt2>)")]
public void Mov_GP_2([Range(0u, 7u)] uint vm,
[Values(0u, 1u, 2u, 3u)] uint rt,
[Values(0u, 1u, 2u, 3u)] uint rt2,
[Random(RndCntImm)] uint valueRt1,
[Random(RndCntImm)] uint valueRt2,
[Random(RndCntImm)] ulong valueVn1,
[Random(RndCntImm)] ulong valueVn2,
[Values] bool op)
{
uint opcode = 0xec400a10u; // VMOV S0, S1, R0, R0
opcode |= (vm & 1) << 5;
opcode |= (vm & 0x1e) >> 1;
opcode |= (rt & 0xf) << 12;
opcode |= (rt2 & 0xf) << 16;
if (op)
{
opcode |= 1 << 20;
}
SingleOpcode(opcode, r0: valueRt1, r1: valueRt2, r2: valueRt1, r3: valueRt2, v0: new V128(valueVn1, valueVn2), v1: new V128(valueVn2, valueVn1));
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VMOVN.<size> <Dt>, <Qm>")]
public void Movn_V([Range(0u, 1u, 2u)] uint size,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 2u, 4u, 8u)] uint vm)
{
uint opcode = 0xf3b20200u; // VMOVN.I16 D0, Q0
opcode |= (size & 0x3) << 18;
opcode |= ((vm & 0x10) << 1);
opcode |= ((vm & 0xf) << 0);
opcode |= ((vd & 0x10) << 18);
opcode |= ((vd & 0xf) << 12);
V128 v0 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VMOVL.<size> <Qd>, <Dm>")]
public void Vmovl([Values(0u, 1u, 2u, 3u)] uint vm,
[Values(0u, 2u, 4u, 6u)] uint vd,
[Values(1u, 2u, 4u)] uint imm3H,
[Values] bool u)
{
// This is not VMOVL because imm3H = 0, but once
// we shift in the imm3H value it turns into VMOVL.
uint opcode = 0xf2800a10u; // VMOV.I16 D0, #0
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
opcode |= (imm3H & 0x7) << 19;
if (u)
{
opcode |= 1 << 24;
}
V128 v0 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VMVN.<size> <Vt>, <Vm>")]
public void Vmvn([Range(0u, 1u, 2u)] uint size,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 2u, 4u, 8u)] uint vm,
[Values] bool q)
{
uint opcode = 0xf3b00580u; // VMVN D0, D0
if (q)
{
opcode |= 1 << 6;
vm <<= 1;
vd <<= 1;
}
opcode |= (size & 0x3) << 18;
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf) << 0;
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
V128 v0 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VMVN.I<size> <Dd/Qd>, #<imm>")]
public void Mvni_V([Range(0u, 7u)] uint variant,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0x0u)] [Random(1u, 0xffu, RndCntImm)] uint imm,
[Values] bool q)
{
uint[] variants =
{
// I32
0b0000,
0b0010,
0b0100,
0b0110,
// I16
0b1000,
0b1010,
// I32
0b1100,
0b1101,
};
uint opcode = 0xf2800030u; // VMVN.I32 D0, #0
uint cmodeOp = variants[variant];
if (q)
{
vd <<= 1;
}
opcode |= (cmodeOp & 0xf) << 8;
opcode |= (q ? 1u : 0u) << 6;
opcode |= (imm & 0xf) | ((imm & 0x70) << 12) | ((imm & 0x80) << 16);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
SingleOpcode(opcode);
CompareAgainstUnicorn();
}
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
[Test, Pairwise, Description("VTRN.<size> <Vd>, <Vm>")]
public void Vtrn([Values(0u, 1u, 2u, 3u)] uint vm,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 1u, 2u)] uint size,
[Values] bool q)
{
uint opcode = 0xf3b20080u; // VTRN.8 D0, D0
if (vm == vd)
{
return; // Undefined.
}
if (q)
{
opcode |= 1 << 6;
vd <<= 1; vm <<= 1;
}
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
opcode |= (size & 0x3) << 18;
V128 v0 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VZIP.<size> <Vd>, <Vm>")]
public void Vzip([Values(0u, 1u, 2u, 3u)] uint vm,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 1u, 2u)] uint size,
[Values] bool q)
{
uint opcode = 0xf3b20180u; // VZIP.8 D0, D0
if (vm == vd || (size == 2 && !q))
{
return; // Undefined.
}
if (q)
{
opcode |= 1 << 6;
vd <<= 1; vm <<= 1;
}
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
opcode |= (size & 0x3) << 18;
V128 v0 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VUZP.<size> <Vd>, <Vm>")]
public void Vuzp([Values(0u, 1u, 2u, 3u)] uint vm,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 1u, 2u)] uint size,
[Values] bool q)
{
uint opcode = 0xf3b20100u; // VUZP.8 d0, d0
if (vm == vd || (size == 2 && !q))
{
return; // Undefined.
}
if (q)
{
opcode |= 1 << 6;
vd <<= 1; vm <<= 1;
}
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
opcode |= (size & 0x3) << 18;
V128 v0 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VTBL.8 <Dd>, {list}, <Dm>")]
public void Vtbl([Range(0u, 6u)] uint vm, // Indices, include potentially invalid.
[Range(4u, 12u)] uint vn, // Selection.
[Values(0u, 1u)] uint vd, // Destinations.
[Range(0u, 3u)] uint length,
[Values] bool x)
{
uint opcode = 0xf3b00800u; // VTBL.8 D0, {D0}, D0
if (vn + length > 31)
{
return; // Undefined.
}
if (x)
{
opcode |= 1 << 6;
}
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
opcode |= (vn & 0x10) << 3;
opcode |= (vn & 0xf) << 16;
opcode |= (length & 0x3) << 8;
var rnd = TestContext.CurrentContext.Random;
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v4 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v5 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
byte maxIndex = (byte)(length * 8 - 1);
byte[] b0 = new byte[16];
byte[] b1 = new byte[16];
for (int i=0; i<16; i++)
{
b0[i] = rnd.NextByte(maxIndex);
b1[i] = rnd.NextByte(maxIndex);
}
V128 v0 = new V128(b0);
V128 v1 = new V128(b1);
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3, v4: v4, v5: v5);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VEXT.8 {<Vd>,} <Vn>, <Vm>, #<imm>")]
public void Vext([Values(0u, 1u, 2u, 3u)] uint vm,
[Values(0u, 1u, 2u, 3u)] uint vn,
[Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 15u)] uint imm4,
[Values] bool q)
{
uint opcode = 0xf2b00000; // VEXT.32 D0, D0, D0, #0
if (q)
{
opcode |= 1 << 6;
vd <<= 1; vm <<= 1; vn <<= 1;
}
else if (imm4 > 7)
{
return; // Undefined.
}
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
opcode |= (vn & 0x10) << 3;
opcode |= (vn & 0xf) << 16;
opcode |= (imm4 & 0xf) << 8;
V128 v0 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: v0, v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VDUP.<size> <Vd>, <Rt>")]
public void Vdup_GP([Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 1u, 2u, 3u)] uint rt,
[Values(0u, 1u, 2u)] uint size,
[Random(RndCntImm)] uint valueRn,
[Random(RndCntImm)] ulong valueVn1,
[Random(RndCntImm)] ulong valueVn2,
[Values] bool q)
{
uint opcode = 0xee800b10; // VDUP.32 d0, r0
if (q)
{
opcode |= 1 << 21;
vd <<= 1;
}
opcode |= (vd & 0x10) << 3;
opcode |= (vd & 0xf) << 16;
opcode |= (rt & 0xf) << 12;
opcode |= (size & 1) << 5; // E
opcode |= (size & 2) << 21; // B
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, r0: valueRn, r1: valueRn, r2: valueRn, r3: valueRn, v0: new V128(valueVn1, valueVn2), v1: v1);
CompareAgainstUnicorn();
}
[Test, Pairwise, Description("VDUP.<size> <Vd>, <Dm[x]>")]
public void Vdup_S([Values(0u, 1u, 2u, 3u)] uint vd,
[Values(0u, 1u, 2u, 3u)] uint vm,
[Values(0u, 1u, 2u)] uint size,
[Range(0u, 7u)] uint index,
[Random(RndCntImm)] ulong valueVn1,
[Random(RndCntImm)] ulong valueVn2,
[Values] bool q)
{
uint opcode = 0xf3b00c00;
if (q)
{
opcode |= 1 << 6;
vd <<= 1;
}
opcode |= (vd & 0x10) << 18;
opcode |= (vd & 0xf) << 12;
opcode |= (vm & 0x10) << 1;
opcode |= (vm & 0xf);
uint imm4 = 0;
switch (size)
{
case 0:
imm4 |= 0b0100 | ((index & 1) << 3);
break;
case 1:
imm4 |= 0b0010 | ((index & 3) << 2);
break;
case 2:
imm4 |= 0b0001 | ((index & 7) << 1);
break;
}
opcode |= imm4 << 16;
V128 v1 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v2 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
V128 v3 = new V128(TestContext.CurrentContext.Random.NextULong(), TestContext.CurrentContext.Random.NextULong());
SingleOpcode(opcode, v0: new V128(valueVn1, valueVn2), v1: v1, v2: v2, v3: v3);
CompareAgainstUnicorn();
}
#endif
}
}