mirror of
https://github.com/GreemDev/Ryujinx.git
synced 2024-12-23 16:15:47 +00:00
f695a215ad
* Add Fast Paths for Crypto instructions (A32/A64) * Replace additional XOR with passing in const zero.
95 lines
2.8 KiB
C#
95 lines
2.8 KiB
C#
using ARMeilleure.Decoders;
|
|
using ARMeilleure.IntermediateRepresentation;
|
|
using ARMeilleure.Translation;
|
|
|
|
using static ARMeilleure.Instructions.InstEmitHelper;
|
|
|
|
namespace ARMeilleure.Instructions
|
|
{
|
|
partial class InstEmit32
|
|
{
|
|
public static void Aesd_V(ArmEmitterContext context)
|
|
{
|
|
OpCode32Simd op = (OpCode32Simd)context.CurrOp;
|
|
|
|
Operand d = GetVecA32(op.Qd);
|
|
Operand n = GetVecA32(op.Qm);
|
|
|
|
Operand res;
|
|
if (Optimizations.UseAesni)
|
|
{
|
|
res = context.AddIntrinsic(Intrinsic.X86Aesdeclast, context.AddIntrinsic(Intrinsic.X86Xorpd, d, n), context.VectorZero());
|
|
}
|
|
else
|
|
{
|
|
res = context.Call(new _V128_V128_V128(SoftFallback.Decrypt), d, n);
|
|
}
|
|
|
|
context.Copy(d, res);
|
|
}
|
|
|
|
public static void Aese_V(ArmEmitterContext context)
|
|
{
|
|
OpCode32Simd op = (OpCode32Simd)context.CurrOp;
|
|
|
|
Operand d = GetVecA32(op.Qd);
|
|
Operand n = GetVecA32(op.Qm);
|
|
|
|
Operand res;
|
|
if (Optimizations.UseAesni)
|
|
{
|
|
res = context.AddIntrinsic(Intrinsic.X86Aesenclast, context.AddIntrinsic(Intrinsic.X86Xorpd, d, n), context.VectorZero());
|
|
}
|
|
else
|
|
{
|
|
res = context.Call(new _V128_V128_V128(SoftFallback.Encrypt), d, n);
|
|
}
|
|
|
|
context.Copy(d, res);
|
|
}
|
|
|
|
public static void Aesimc_V(ArmEmitterContext context)
|
|
{
|
|
OpCode32Simd op = (OpCode32Simd)context.CurrOp;
|
|
|
|
Operand n = GetVecA32(op.Qm);
|
|
|
|
Operand res;
|
|
if (Optimizations.UseAesni)
|
|
{
|
|
res = context.AddIntrinsic(Intrinsic.X86Aesimc, n);
|
|
}
|
|
else
|
|
{
|
|
res = context.Call(new _V128_V128(SoftFallback.InverseMixColumns), n);
|
|
}
|
|
|
|
context.Copy(GetVecA32(op.Qd), res);
|
|
}
|
|
|
|
public static void Aesmc_V(ArmEmitterContext context)
|
|
{
|
|
OpCode32Simd op = (OpCode32Simd)context.CurrOp;
|
|
|
|
Operand n = GetVecA32(op.Qm);
|
|
|
|
Operand res;
|
|
if (Optimizations.UseAesni)
|
|
{
|
|
Operand roundKey = context.VectorZero();
|
|
|
|
// Inverse Shift Rows, Inverse Sub Bytes, xor 0 so nothing happens.
|
|
res = context.AddIntrinsic(Intrinsic.X86Aesdeclast, n, roundKey);
|
|
|
|
// Shift Rows, Sub Bytes, Mix Columns (!), xor 0 so nothing happens.
|
|
res = context.AddIntrinsic(Intrinsic.X86Aesenc, res, roundKey);
|
|
}
|
|
else
|
|
{
|
|
res = context.Call(new _V128_V128(SoftFallback.MixColumns), n);
|
|
}
|
|
|
|
context.Copy(GetVecA32(op.Qd), res);
|
|
}
|
|
}
|
|
}
|