GPU: Use a bitmap to track buffer modified flags. (#3775)

* Initial implementation

* Some improvements.

* Fix incorrect cast

* Performance improvement and improved correctness

* Add very fast path when all handles are checked.

* Slightly faster

* Add comment

* De-virtualize region handle

All region handles are now bitmap backed.

* Remove non-bitmap tracking

* Remove unused methods

* Add docs, remove unused methods

* Address Feedback

* Rename file

.github/dependabot.yml

@@ -0,0 +1,24 @@
+version: 2
+updates:
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: weekly
+    labels:
+      - "infra"
+    reviewers:
+      - marysaka
+    commit-message:
+      prefix: "ci"
+
+  - package-ecosystem: nuget
+    directory: /
+    open-pull-requests-limit: 5
+    schedule:
+      interval: daily
+    labels:
+      - "infra"
+    reviewers:
+      - marysaka
+    commit-message:
+      prefix: nuget

.github/workflows/build.yml

@@ -48,21 +48,22 @@ jobs:
       DOTNET_CLI_TELEMETRY_OPTOUT: 1
       RYUJINX_BASE_VERSION: "1.1.0"
     steps:
-      - uses: actions/checkout@v2
-      - uses: actions/setup-dotnet@v1
+      - uses: actions/checkout@v3
+      - uses: actions/setup-dotnet@v3
         with:
           dotnet-version: 6.0.x
       - name: Ensure NuGet Source
         uses: fabriciomurta/ensure-nuget-source@v1
       - name: Get git short hash
         id: git_short_hash
-        run: echo "::set-output name=result::$(git rev-parse --short "${{ github.sha }}")"
+        run: echo "result=$(git rev-parse --short "${{ github.sha }}")" >> $GITHUB_OUTPUT
+        shell: bash
       - name: Clear
         run: dotnet clean && dotnet nuget locals all --clear
       - name: Build
         run: dotnet build -c "${{ matrix.configuration }}" /p:Version="${{ env.RYUJINX_BASE_VERSION }}" /p:SourceRevisionId="${{ steps.git_short_hash.outputs.result }}" /p:ExtraDefineConstants=DISABLE_UPDATER
       - name: Test
-        run: dotnet test -c "${{ matrix.configuration }}"
+        run: dotnet test --no-build -c "${{ matrix.configuration }}"
       - name: Publish Ryujinx
         run: dotnet publish -c "${{ matrix.configuration }}" -r "${{ matrix.DOTNET_RUNTIME_IDENTIFIER }}" -o ./publish /p:Version="${{ env.RYUJINX_BASE_VERSION }}" /p:DebugType=embedded /p:SourceRevisionId="${{ steps.git_short_hash.outputs.result }}" /p:ExtraDefineConstants=DISABLE_UPDATER Ryujinx --self-contained
         if: github.event_name == 'pull_request'
@@ -73,19 +74,19 @@ jobs:
         run: dotnet publish -c "${{ matrix.configuration }}" -r "${{ matrix.DOTNET_RUNTIME_IDENTIFIER }}" -o ./publish_ava /p:Version="1.0.0" /p:DebugType=embedded /p:SourceRevisionId="${{ steps.git_short_hash.outputs.result }}" /p:ExtraDefineConstants=DISABLE_UPDATER Ryujinx.Ava
         if: github.event_name == 'pull_request'
       - name: Upload Ryujinx artifact
-        uses: actions/upload-artifact@v2
+        uses: actions/upload-artifact@v3
         with:
           name: ryujinx-${{ matrix.configuration }}-${{ env.RYUJINX_BASE_VERSION }}+${{ steps.git_short_hash.outputs.result }}-${{ matrix.RELEASE_ZIP_OS_NAME }}
           path: publish
         if: github.event_name == 'pull_request'
       - name: Upload Ryujinx.Headless.SDL2 artifact
-        uses: actions/upload-artifact@v2
+        uses: actions/upload-artifact@v3
         with:
           name: sdl2-ryujinx-headless-${{ matrix.configuration }}-${{ env.RYUJINX_BASE_VERSION }}+${{ steps.git_short_hash.outputs.result }}-${{ matrix.RELEASE_ZIP_OS_NAME }}
           path: publish_sdl2_headless
         if: github.event_name == 'pull_request'
       - name: Upload Ryujinx.Ava artifact
-        uses: actions/upload-artifact@v2
+        uses: actions/upload-artifact@v3
         with:
           name: ava-ryujinx-${{ matrix.configuration }}-${{ env.RYUJINX_BASE_VERSION }}+${{ steps.git_short_hash.outputs.result }}-${{ matrix.RELEASE_ZIP_OS_NAME }}
           path: publish_ava

.github/workflows/nightly_pr_comment.yml

@@ -8,7 +8,7 @@ jobs:
     if: github.event.workflow_run.event == 'pull_request' && github.event.workflow_run.conclusion == 'success'
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/github-script@v3
+      - uses: actions/github-script@v6
         with:
           script: |
             const {owner, repo} = context.repo;
@@ -16,7 +16,7 @@ jobs:
             const pull_head_sha = '${{github.event.workflow_run.head_sha}}';
 
             const issue_number = await (async () => {
-              const pulls = await github.pulls.list({owner, repo});
+              const pulls = await github.rest.pulls.list({owner, repo});
               for await (const {data} of github.paginate.iterator(pulls)) {
                 for (const pull of data) {
                   if (pull.head.sha === pull_head_sha) {
@@ -31,7 +31,7 @@ jobs:
               return core.error(`No matching pull request found`);
             }
 
-            const {data: {artifacts}} = await github.actions.listWorkflowRunArtifacts({owner, repo, run_id});
+            const {data: {artifacts}} = await github.rest.actions.listWorkflowRunArtifacts({owner, repo, run_id});
             if (!artifacts.length) {
               return core.error(`No artifacts found`);
             }
@@ -57,12 +57,12 @@ jobs:
             body += hidden_headless_artifacts;
             body += hidden_debug_artifacts;
 
-            const {data: comments} = await github.issues.listComments({repo, owner, issue_number});
+            const {data: comments} = await github.rest.issues.listComments({repo, owner, issue_number});
             const existing_comment = comments.find((c) => c.user.login === 'github-actions[bot]');
             if (existing_comment) {
               core.info(`Updating comment ${existing_comment.id}`);
-              await github.issues.updateComment({repo, owner, comment_id: existing_comment.id, body});
+              await github.rest.issues.updateComment({repo, owner, comment_id: existing_comment.id, body});
             } else {
               core.info(`Creating a comment`);
-              await github.issues.createComment({repo, owner, issue_number, body});
+              await github.rest.issues.createComment({repo, owner, issue_number, body});
             }

.github/workflows/release.yml

@@ -25,8 +25,8 @@ jobs:
       RYUJINX_TARGET_RELEASE_CHANNEL_REPO: "release-channel-master"
 
     steps:
-      - uses: actions/checkout@v2
-      - uses: actions/setup-dotnet@v1
+      - uses: actions/checkout@v3
+      - uses: actions/setup-dotnet@v3
         with:
           dotnet-version: 6.0.x
       - name: Ensure NuGet Source
@@ -36,8 +36,8 @@ jobs:
       - name: Get version info
         id: version_info
         run: |
-          echo "::set-output name=build_version::${{ env.RYUJINX_BASE_VERSION }}.${{ github.run_number }}"
-          echo "::set-output name=git_short_hash::$(git rev-parse --short "${{ github.sha }}")"
+          echo "build_version=${{ env.RYUJINX_BASE_VERSION }}.${{ github.run_number }}" >> $GITHUB_OUTPUT
+          echo "git_short_hash=$(git rev-parse --short "${{ github.sha }}")" >> $GITHUB_OUTPUT
         shell: bash
       - name: Configure for release
         run: |

ARMeilleure/Decoders/OpCode32SimdCvtTB.cs

@@ -0,0 +1,44 @@
+namespace ARMeilleure.Decoders
+{
+    class OpCode32SimdCvtTB : OpCode32, IOpCode32Simd
+    {
+        public int Vd { get; }
+        public int Vm { get; }
+        public bool Op { get; } // Convert to Half / Convert from Half
+        public bool T { get; } // Top / Bottom
+        public int Size { get; } // Double / Single
+
+        public new static OpCode Create(InstDescriptor inst, ulong address, int opCode) => new OpCode32SimdCvtTB(inst, address, opCode, false);
+        public static OpCode CreateT32(InstDescriptor inst, ulong address, int opCode) => new OpCode32SimdCvtTB(inst, address, opCode, true);
+
+        public OpCode32SimdCvtTB(InstDescriptor inst, ulong address, int opCode, bool isThumb) : base(inst, address, opCode)
+        {
+            IsThumb = isThumb;
+
+            Op   = ((opCode >> 16) & 0x1) != 0;
+            T    = ((opCode >> 7) & 0x1) != 0;
+            Size = ((opCode >> 8) & 0x1);
+
+            RegisterSize = Size == 1 ? RegisterSize.Int64 : RegisterSize.Int32;
+
+            if (Size == 1)
+            {
+                if (Op)
+                {
+                    Vm = ((opCode >> 1) & 0x10) | ((opCode >> 0) & 0xf);
+                    Vd = ((opCode >> 22) & 0x1) | ((opCode >> 11) & 0x1e);
+                }
+                else
+                {
+                    Vm = ((opCode >> 5) & 0x1) | ((opCode << 1) & 0x1e);
+                    Vd = ((opCode >> 18) & 0x10) | ((opCode >> 12) & 0xf);
+                }
+            }
+            else
+            {
+                Vm = ((opCode >> 5) & 0x1) | ((opCode << 1) & 0x1e);
+                Vd = ((opCode >> 22) & 0x1) | ((opCode >> 11) & 0x1e);
+            }
+        }
+    }
+}

ARMeilleure/Decoders/OpCodeTable.cs

@@ -828,6 +828,7 @@ namespace ARMeilleure.Decoders
             SetVfp("<<<<11101x11110xxxxx101x11x0xxxx", InstName.Vcvt,   InstEmit32.Vcvt_FI,  OpCode32SimdCvtFI.Create,       OpCode32SimdCvtFI.CreateT32); // FP32 to int.
             SetVfp("<<<<11101x111000xxxx101xx1x0xxxx", InstName.Vcvt,   InstEmit32.Vcvt_FI,  OpCode32SimdCvtFI.Create,       OpCode32SimdCvtFI.CreateT32); // Int to FP32.
             SetVfp("111111101x1111xxxxxx101xx1x0xxxx", InstName.Vcvt,   InstEmit32.Vcvt_RM,  OpCode32SimdCvtFI.Create,       OpCode32SimdCvtFI.CreateT32); // The many FP32 to int encodings (fp).
+            SetVfp("<<<<11101x11001xxxxx101xx1x0xxxx", InstName.Vcvt,   InstEmit32.Vcvt_TB,  OpCode32SimdCvtTB.Create,       OpCode32SimdCvtTB.CreateT32);
             SetVfp("<<<<11101x00xxxxxxxx101xx0x0xxxx", InstName.Vdiv,   InstEmit32.Vdiv_S,   OpCode32SimdRegS.Create,        OpCode32SimdRegS.CreateT32);
             SetVfp("<<<<11101xx0xxxxxxxx1011x0x10000", InstName.Vdup,   InstEmit32.Vdup,     OpCode32SimdDupGP.Create,       OpCode32SimdDupGP.CreateT32);
             SetVfp("<<<<11101x10xxxxxxxx101xx0x0xxxx", InstName.Vfma,   InstEmit32.Vfma_S,   OpCode32SimdRegS.Create,        OpCode32SimdRegS.CreateT32);

ARMeilleure/Instructions/InstEmitSimdArithmetic.cs

@@ -1617,18 +1617,32 @@ namespace ARMeilleure.Instructions
 
         public static void Frinta_S(ArmEmitterContext context)
         {
-            EmitScalarUnaryOpF(context, (op1) =>
+            if (Optimizations.UseSse41)
             {
-                return EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1);
-            });
+                EmitSse41ScalarRoundOpF(context, FPRoundingMode.ToNearestAway);
+            }
+            else
+            {
+                EmitScalarUnaryOpF(context, (op1) =>
+                {
+                    return EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1);
+                });
+            }
         }
 
         public static void Frinta_V(ArmEmitterContext context)
         {
-            EmitVectorUnaryOpF(context, (op1) =>
+            if (Optimizations.UseSse41)
             {
-                return EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1);
-            });
+                EmitSse41VectorRoundOpF(context, FPRoundingMode.ToNearestAway);
+            }
+            else
+            {
+                EmitVectorUnaryOpF(context, (op1) =>
+                {
+                    return EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1);
+                });
+            }
         }
 
         public static void Frinti_S(ArmEmitterContext context)
@@ -3516,9 +3530,18 @@ namespace ARMeilleure.Instructions
 
             Operand n = GetVec(op.Rn);
 
-            Intrinsic inst = (op.Size & 1) != 0 ? Intrinsic.X86Roundsd : Intrinsic.X86Roundss;
+            Operand res;
 
-            Operand res = context.AddIntrinsic(inst, n, Const(X86GetRoundControl(roundMode)));
+            if (roundMode != FPRoundingMode.ToNearestAway)
+            {
+                Intrinsic inst = (op.Size & 1) != 0 ? Intrinsic.X86Roundsd : Intrinsic.X86Roundss;
+
+                res = context.AddIntrinsic(inst, n, Const(X86GetRoundControl(roundMode)));
+            }
+            else
+            {
+                res = EmitSse41RoundToNearestWithTiesToAwayOpF(context, n, scalar: true);
+            }
 
             if ((op.Size & 1) != 0)
             {
@@ -3538,9 +3561,18 @@ namespace ARMeilleure.Instructions
 
             Operand n = GetVec(op.Rn);
 
-            Intrinsic inst = (op.Size & 1) != 0 ? Intrinsic.X86Roundpd : Intrinsic.X86Roundps;
+            Operand res;
 
-            Operand res = context.AddIntrinsic(inst, n, Const(X86GetRoundControl(roundMode)));
+            if (roundMode != FPRoundingMode.ToNearestAway)
+            {
+                Intrinsic inst = (op.Size & 1) != 0 ? Intrinsic.X86Roundpd : Intrinsic.X86Roundps;
+
+                res = context.AddIntrinsic(inst, n, Const(X86GetRoundControl(roundMode)));
+            }
+            else
+            {
+                res = EmitSse41RoundToNearestWithTiesToAwayOpF(context, n, scalar: false);
+            }
 
             if (op.RegisterSize == RegisterSize.Simd64)
             {

ARMeilleure/Instructions/InstEmitSimdCvt.cs

@@ -164,32 +164,74 @@ namespace ARMeilleure.Instructions
 
         public static void Fcvtas_Gp(ArmEmitterContext context)
         {
-            EmitFcvt_s_Gp(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1));
+            if (Optimizations.UseSse41)
+            {
+                EmitSse41Fcvts_Gp(context, FPRoundingMode.ToNearestAway, isFixed: false);
+            }
+            else
+            {
+                EmitFcvt_s_Gp(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1));
+            }
         }
 
         public static void Fcvtas_S(ArmEmitterContext context)
         {
-            EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: true, scalar: true);
+            if (Optimizations.UseSse41)
+            {
+                EmitSse41FcvtsOpF(context, FPRoundingMode.ToNearestAway, scalar: true);
+            }
+            else
+            {
+                EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: true, scalar: true);
+            }
         }
 
         public static void Fcvtas_V(ArmEmitterContext context)
         {
-            EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: true, scalar: false);
+            if (Optimizations.UseSse41)
+            {
+                EmitSse41FcvtsOpF(context, FPRoundingMode.ToNearestAway, scalar: false);
+            }
+            else
+            {
+                EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: true, scalar: false);
+            }
         }
 
         public static void Fcvtau_Gp(ArmEmitterContext context)
         {
-            EmitFcvt_u_Gp(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1));
+            if (Optimizations.UseSse41)
+            {
+                EmitSse41Fcvtu_Gp(context, FPRoundingMode.ToNearestAway, isFixed: false);
+            }
+            else
+            {
+                EmitFcvt_u_Gp(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1));
+            }
         }
 
         public static void Fcvtau_S(ArmEmitterContext context)
         {
-            EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: false, scalar: true);
+            if (Optimizations.UseSse41)
+            {
+                EmitSse41FcvtuOpF(context, FPRoundingMode.ToNearestAway, scalar: true);
+            }
+            else
+            {
+                EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: false, scalar: true);
+            }
         }
 
         public static void Fcvtau_V(ArmEmitterContext context)
         {
-            EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: false, scalar: false);
+            if (Optimizations.UseSse41)
+            {
+                EmitSse41FcvtuOpF(context, FPRoundingMode.ToNearestAway, scalar: false);
+            }
+            else
+            {
+                EmitFcvt(context, (op1) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, op1), signed: false, scalar: false);
+            }
         }
 
         public static void Fcvtl_V(ArmEmitterContext context)
@@ -1223,7 +1265,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulps, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundps, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundps, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar);
+                }
 
                 Operand nInt = context.AddIntrinsic(Intrinsic.X86Cvtps2dq, nRes);
 
@@ -1265,7 +1314,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulpd, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundpd, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundpd, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar);
+                }
 
                 Operand nLong = EmitSse2CvtDoubleToInt64OpF(context, nRes, scalar);
 
@@ -1314,7 +1370,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulps, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundps, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundps, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar);
+                }
 
                 Operand zero = context.VectorZero();
 
@@ -1369,7 +1432,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulpd, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundpd, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundpd, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar);
+                }
 
                 Operand zero = context.VectorZero();
 
@@ -1424,7 +1494,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulss, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundss, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundss, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar: true);
+                }
 
                 Operand nIntOrLong = op.RegisterSize == RegisterSize.Int32
                     ? context.AddIntrinsicInt (Intrinsic.X86Cvtss2si, nRes)
@@ -1464,7 +1541,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulsd, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundsd, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundsd, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar: true);
+                }
 
                 Operand nIntOrLong = op.RegisterSize == RegisterSize.Int32
                     ? context.AddIntrinsicInt (Intrinsic.X86Cvtsd2si, nRes)
@@ -1512,7 +1596,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulss, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundss, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundss, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar: true);
+                }
 
                 Operand zero = context.VectorZero();
 
@@ -1567,7 +1658,14 @@ namespace ARMeilleure.Instructions
                     nRes = context.AddIntrinsic(Intrinsic.X86Mulsd, nRes, fpScaledMask);
                 }
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundsd, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundsd, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar: true);
+                }
 
                 Operand zero = context.VectorZero();
 

ARMeilleure/Instructions/InstEmitSimdCvt32.cs

@@ -203,6 +203,9 @@ namespace ARMeilleure.Instructions
             FPRoundingMode roundMode;
             switch (rm)
             {
+                case 0b00:
+                    roundMode = FPRoundingMode.ToNearestAway;
+                    break;
                 case 0b01:
                     roundMode = FPRoundingMode.ToNearest;
                     break;
@@ -228,7 +231,7 @@ namespace ARMeilleure.Instructions
             bool unsigned = op.Opc == 0;
             int rm = op.Opc2 & 3;
 
-            if (Optimizations.UseSse41 && rm != 0b00)
+            if (Optimizations.UseSse41)
             {
                 EmitSse41ConvertInt32(context, RMToRoundMode(rm), !unsigned);
             }
@@ -258,6 +261,74 @@ namespace ARMeilleure.Instructions
             }
         }
 
+        public static void Vcvt_TB(ArmEmitterContext context)
+        {
+            OpCode32SimdCvtTB op = (OpCode32SimdCvtTB)context.CurrOp;
+
+            if (Optimizations.UseF16c)
+            {
+                Debug.Assert(!Optimizations.ForceLegacySse);
+
+                if (op.Op)
+                {
+                    Operand res = ExtractScalar(context, op.Size == 1 ? OperandType.FP64 : OperandType.FP32, op.Vm);
+                    if (op.Size == 1)
+                    {
+                        res = context.AddIntrinsic(Intrinsic.X86Cvtsd2ss, context.VectorZero(), res);
+                    }
+                    res = context.AddIntrinsic(Intrinsic.X86Vcvtps2ph, res, Const(X86GetRoundControl(FPRoundingMode.ToNearest)));
+                    res = context.VectorExtract16(res, 0);
+                    InsertScalar16(context, op.Vd, op.T, res);
+                }
+                else
+                {
+                    Operand res = context.VectorCreateScalar(ExtractScalar16(context, op.Vm, op.T));
+                    res = context.AddIntrinsic(Intrinsic.X86Vcvtph2ps, res);
+                    if (op.Size == 1)
+                    {
+                        res = context.AddIntrinsic(Intrinsic.X86Cvtss2sd, context.VectorZero(), res);
+                    }
+                    res = context.VectorExtract(op.Size == 1 ? OperandType.I64 : OperandType.I32, res, 0);
+                    InsertScalar(context, op.Vd, res);
+                }
+            }
+            else
+            {
+                if (op.Op)
+                {
+                    // Convert to half.
+
+                    Operand src = ExtractScalar(context, op.Size == 1 ? OperandType.FP64 : OperandType.FP32, op.Vm);
+
+                    MethodInfo method = op.Size == 1
+                        ? typeof(SoftFloat64_16).GetMethod(nameof(SoftFloat64_16.FPConvert))
+                        : typeof(SoftFloat32_16).GetMethod(nameof(SoftFloat32_16.FPConvert));
+
+                    context.StoreToContext();
+                    Operand res = context.Call(method, src);
+                    context.LoadFromContext();
+
+                    InsertScalar16(context, op.Vd, op.T, res);
+                }
+                else
+                {
+                    // Convert from half.
+
+                    Operand src = ExtractScalar16(context, op.Vm, op.T);
+
+                    MethodInfo method = op.Size == 1
+                        ? typeof(SoftFloat16_64).GetMethod(nameof(SoftFloat16_64.FPConvert))
+                        : typeof(SoftFloat16_32).GetMethod(nameof(SoftFloat16_32.FPConvert));
+
+                    context.StoreToContext();
+                    Operand res = context.Call(method, src);
+                    context.LoadFromContext();
+
+                    InsertScalar(context, op.Vd, res);
+                }
+            }
+        }
+
         // VRINTA/M/N/P (floating-point).
         public static void Vrint_RM(ArmEmitterContext context)
         {
@@ -267,15 +338,21 @@ namespace ARMeilleure.Instructions
 
             int rm = op.Opc2 & 3;
 
-            if (Optimizations.UseSse2 && rm != 0b00)
+            if (Optimizations.UseSse41)
             {
                 EmitScalarUnaryOpSimd32(context, (m) =>
                 {
-                    Intrinsic inst = (op.Size & 1) == 0 ? Intrinsic.X86Roundss : Intrinsic.X86Roundsd;
-
                     FPRoundingMode roundMode = RMToRoundMode(rm);
 
-                    return context.AddIntrinsic(inst, m, Const(X86GetRoundControl(roundMode)));
+                    if (roundMode != FPRoundingMode.ToNearestAway)
+                    {
+                        Intrinsic inst = (op.Size & 1) == 0 ? Intrinsic.X86Roundss : Intrinsic.X86Roundsd;
+                        return context.AddIntrinsic(inst, m, Const(X86GetRoundControl(roundMode)));
+                    }
+                    else
+                    {
+                        return EmitSse41RoundToNearestWithTiesToAwayOpF(context, m, scalar: true);
+                    }
                 });
             }
             else
@@ -305,7 +382,17 @@ namespace ARMeilleure.Instructions
         // VRINTA (vector).
         public static void Vrinta_V(ArmEmitterContext context)
         {
-            EmitVectorUnaryOpF32(context, (m) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, m));
+            if (Optimizations.UseSse41)
+            {
+                EmitVectorUnaryOpSimd32(context, (m) =>
+                {
+                    return EmitSse41RoundToNearestWithTiesToAwayOpF(context, m, scalar: false);
+                });
+            }
+            else
+            {
+                EmitVectorUnaryOpF32(context, (m) => EmitRoundMathCall(context, MidpointRounding.AwayFromZero, m));
+            }
         }
 
         // VRINTM (vector).
@@ -413,7 +500,14 @@ namespace ARMeilleure.Instructions
                 Operand nRes = context.AddIntrinsic(Intrinsic.X86Cmpss, n, n, Const((int)CmpCondition.OrderedQ));
                 nRes = context.AddIntrinsic(Intrinsic.X86Pand, nRes, n);
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundss, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundss, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar: true);
+                }
 
                 Operand zero = context.VectorZero();
 
@@ -464,7 +558,14 @@ namespace ARMeilleure.Instructions
                 Operand nRes = context.AddIntrinsic(Intrinsic.X86Cmpsd, n, n, Const((int)CmpCondition.OrderedQ));
                 nRes = context.AddIntrinsic(Intrinsic.X86Pand, nRes, n);
 
-                nRes = context.AddIntrinsic(Intrinsic.X86Roundsd, nRes, Const(X86GetRoundControl(roundMode)));
+                if (roundMode != FPRoundingMode.ToNearestAway)
+                {
+                    nRes = context.AddIntrinsic(Intrinsic.X86Roundsd, nRes, Const(X86GetRoundControl(roundMode)));
+                }
+                else
+                {
+                    nRes = EmitSse41RoundToNearestWithTiesToAwayOpF(context, nRes, scalar: true);
+                }
 
                 Operand zero = context.VectorZero();
 

ARMeilleure/Instructions/InstEmitSimdHelper.cs

@@ -33,6 +33,14 @@ namespace ARMeilleure.Instructions
         };
 
         public static readonly long ZeroMask = 128L << 56 | 128L << 48 | 128L << 40 | 128L << 32 | 128L << 24 | 128L << 16 | 128L << 8 | 128L << 0;
+
+        public static ulong X86GetGf2p8LogicalShiftLeft(int shift)
+        {
+            ulong identity = (0b00000001UL << 56) | (0b00000010UL << 48) | (0b00000100UL << 40) | (0b00001000UL << 32) |
+                             (0b00010000UL << 24) | (0b00100000UL << 16) | (0b01000000UL <<  8) | (0b10000000UL <<  0);
+
+            return shift >= 0 ? identity >> (shift * 8) : identity << (-shift * 8);
+        }
 #endregion
 
 #region "X86 SSE Intrinsics"
@@ -243,19 +251,44 @@ namespace ARMeilleure.Instructions
             throw new ArgumentException($"Invalid rounding mode \"{roundMode}\".");
         }
 
-        public static ulong X86GetGf2p8LogicalShiftLeft(int shift)
+        public static Operand EmitSse41RoundToNearestWithTiesToAwayOpF(ArmEmitterContext context, Operand n, bool scalar)
         {
-            ulong identity = 
-                (0b00000001UL << 56) |
-                (0b00000010UL << 48) |
-                (0b00000100UL << 40) |
-                (0b00001000UL << 32) |
-                (0b00010000UL << 24) |
-                (0b00100000UL << 16) |
-                (0b01000000UL <<  8) |
-                (0b10000000UL <<  0);
+            Debug.Assert(n.Type == OperandType.V128);
 
-            return shift >= 0 ? identity >> (shift * 8) : identity << (-shift * 8);
+            Operand nCopy = context.Copy(n);
+
+            Operand rC = Const(X86GetRoundControl(FPRoundingMode.TowardsZero));
+
+            IOpCodeSimd op = (IOpCodeSimd)context.CurrOp;
+
+            if ((op.Size & 1) == 0)
+            {
+                Operand signMask = scalar ? X86GetScalar(context, int.MinValue) : X86GetAllElements(context, int.MinValue);
+                        signMask = context.AddIntrinsic(Intrinsic.X86Pand, signMask, nCopy);
+
+                // 0x3EFFFFFF == BitConverter.SingleToInt32Bits(0.5f) - 1
+                Operand valueMask = scalar ? X86GetScalar(context, 0x3EFFFFFF) : X86GetAllElements(context, 0x3EFFFFFF);
+                        valueMask = context.AddIntrinsic(Intrinsic.X86Por, valueMask, signMask);
+
+                nCopy = context.AddIntrinsic(scalar ? Intrinsic.X86Addss : Intrinsic.X86Addps, nCopy, valueMask);
+
+                nCopy = context.AddIntrinsic(scalar ? Intrinsic.X86Roundss : Intrinsic.X86Roundps, nCopy, rC);
+            }
+            else
+            {
+                Operand signMask = scalar ? X86GetScalar(context, long.MinValue) : X86GetAllElements(context, long.MinValue);
+                        signMask = context.AddIntrinsic(Intrinsic.X86Pand, signMask, nCopy);
+
+                // 0x3FDFFFFFFFFFFFFFL == BitConverter.DoubleToInt64Bits(0.5d) - 1L
+                Operand valueMask = scalar ? X86GetScalar(context, 0x3FDFFFFFFFFFFFFFL) : X86GetAllElements(context, 0x3FDFFFFFFFFFFFFFL);
+                        valueMask = context.AddIntrinsic(Intrinsic.X86Por, valueMask, signMask);
+
+                nCopy = context.AddIntrinsic(scalar ? Intrinsic.X86Addsd : Intrinsic.X86Addpd, nCopy, valueMask);
+
+                nCopy = context.AddIntrinsic(scalar ? Intrinsic.X86Roundsd : Intrinsic.X86Roundpd, nCopy, rC);
+            }
+
+            return nCopy;
         }
 
         public static Operand EmitCountSetBits8(ArmEmitterContext context, Operand op) // "size" is 8 (SIMD&FP Inst.).

ARMeilleure/Instructions/InstEmitSimdHelper32.cs

@@ -70,6 +70,22 @@ namespace ARMeilleure.Instructions
             context.Copy(vec, insert);
         }
 
+        public static Operand ExtractScalar16(ArmEmitterContext context, int reg, bool top)
+        {
+            return context.VectorExtract16(GetVecA32(reg >> 2), ((reg & 3) << 1) | (top ? 1 : 0));
+        }
+
+        public static void InsertScalar16(ArmEmitterContext context, int reg, bool top, Operand value)
+        {
+            Debug.Assert(value.Type == OperandType.FP32 || value.Type == OperandType.I32);
+
+            Operand vec, insert;
+            vec = GetVecA32(reg >> 2);
+            insert = context.VectorInsert16(vec, value, ((reg & 3) << 1) | (top ? 1 : 0));
+
+            context.Copy(vec, insert);
+        }
+
         public static Operand ExtractElement(ArmEmitterContext context, int reg, int size, bool signed)
         {
             return EmitVectorExtract32(context, reg >> (4 - size), reg & ((16 >> size) - 1), size, signed);

ARMeilleure/State/FPRoundingMode.cs

@@ -2,9 +2,10 @@ namespace ARMeilleure.State
 {
     public enum FPRoundingMode
     {
-        ToNearest            = 0,
+        ToNearest            = 0, // With ties to even.
         TowardsPlusInfinity  = 1,
         TowardsMinusInfinity = 2,
-        TowardsZero          = 3
+        TowardsZero          = 3,
+        ToNearestAway        = 4 // With ties to away.
     }
 }

ARMeilleure/Translation/PTC/Ptc.cs

@@ -27,7 +27,7 @@ namespace ARMeilleure.Translation.PTC
         private const string OuterHeaderMagicString = "PTCohd\0\0";
         private const string InnerHeaderMagicString = "PTCihd\0\0";
 
-        private const uint InternalVersion = 3710; //! To be incremented manually for each change to the ARMeilleure project.
+        private const uint InternalVersion = 3713; //! To be incremented manually for each change to the ARMeilleure project.
 
         private const string ActualDir = "0";
         private const string BackupDir = "1";

ARMeilleure/Translation/Translator.cs

@@ -455,13 +455,16 @@ namespace ARMeilleure.Translation
 
         public void InvalidateJitCacheRegion(ulong address, ulong size)
         {
-            // If rejit is running, stop it as it may be trying to rejit a function on the invalidated region.
-            ClearRejitQueue(allowRequeue: true);
-
             ulong[] overlapAddresses = Array.Empty<ulong>();
 
             int overlapsCount = Functions.GetOverlaps(address, size, ref overlapAddresses);
 
+            if (overlapsCount != 0)
+            {
+                // If rejit is running, stop it as it may be trying to rejit a function on the invalidated region.
+                ClearRejitQueue(allowRequeue: true);
+            }
+
             for (int index = 0; index < overlapsCount; index++)
             {
                 ulong overlapAddress = overlapAddresses[index];

README.md

@@ -36,8 +36,8 @@
 
 ## Compatibility
 
-As of September 2022, Ryujinx has been tested on approximately 3,600 titles; over 3,400 boot past menus and into gameplay, with roughly 2,700 of those being considered playable. You can check out the compatibility list [here](https://github.com/Ryujinx/Ryujinx-Games-List/issues).  
-Anyone is free to submit a new game test or update an existing game test entry; simply follow the new issue template and testing guidelines, or post as a reply to the applicable game issue. Use the search function to see if a game has been tested already!
+As of October 2022, Ryujinx has been tested on approximately 3,700 titles; over 3,500 boot past menus and into gameplay, with roughly 3,000 of those being considered playable.
+You can check out the compatibility list [here](https://github.com/Ryujinx/Ryujinx-Games-List/issues). Anyone is free to submit a new game test or update an existing game test entry; simply follow the new issue template and testing guidelines, or post as a reply to the applicable game issue. Use the search function to see if a game has been tested already!
 
 ## Usage
 
@@ -90,7 +90,7 @@ Ryujinx system files are stored in the `Ryujinx` folder. This folder is located
 
 - **GPU**
 
-  The GPU emulator emulates the Switch's Maxwell GPU using the OpenGL API (version 4.5 minimum) through a custom build of OpenTK. There are currently four graphics enhancements available to the end user in Ryujinx: disk shader caching, resolution scaling, aspect ratio adjustment and anisotropic filtering. These enhancements can be adjusted or toggled as desired in the GUI.
+  The GPU emulator emulates the Switch's Maxwell GPU using either the OpenGL (version 4.5 minimum) or Vulkan APIs through a custom build of OpenTK or Silk.NET respectively. There are currently four graphics enhancements available to the end user in Ryujinx: Disk Shader Caching, Resolution Scaling, Aspect Ratio Adjustment, and Anisotropic Filtering. These enhancements can be adjusted or toggled as desired in the GUI.
 
 - **Input**
 

Ryujinx.Ava/Assets/Locales/it_IT.json

@@ -49,8 +49,8 @@
     "GameListContextMenuManageTitleUpdatesToolTip": "Apre la finestra di gestione aggiornamenti del gioco",
     "GameListContextMenuManageDlc": "Gestici DLC",
     "GameListContextMenuManageDlcToolTip": "Apre la finestra di gestione DLC",
-    "GameListContextMenuOpenModsDirectory": "Apri cartella delle mods",
-    "GameListContextMenuOpenModsDirectoryToolTip": "Apre la cartella che contiene le mods dell'applicazione",
+    "GameListContextMenuOpenModsDirectory": "Apri cartella delle mod",
+    "GameListContextMenuOpenModsDirectoryToolTip": "Apre la cartella che contiene le mod dell'applicazione",
     "GameListContextMenuCacheManagement": "Gestione della cache",
     "GameListContextMenuCacheManagementPurgePptc": "Pulisci PPTC cache",
     "GameListContextMenuCacheManagementPurgePptcToolTip": "Elimina la PPTC cache dell'applicazione",
@@ -81,7 +81,7 @@
     "SettingsTabGeneralRemove": "Rimuovi",
     "SettingsTabSystem": "Sistema",
     "SettingsTabSystemCore": "Core",
-    "SettingsTabSystemSystemRegion": "Regione di sistema:",
+    "SettingsTabSystemSystemRegion": "Regione del sistema:",
     "SettingsTabSystemSystemRegionJapan": "Giappone",
     "SettingsTabSystemSystemRegionUSA": "Stati Uniti d'America",
     "SettingsTabSystemSystemRegionEurope": "Europa",
@@ -89,7 +89,7 @@
     "SettingsTabSystemSystemRegionChina": "Cina",
     "SettingsTabSystemSystemRegionKorea": "Corea",
     "SettingsTabSystemSystemRegionTaiwan": "Taiwan",
-    "SettingsTabSystemSystemLanguage": "Lingua di sistema:",
+    "SettingsTabSystemSystemLanguage": "Lingua del sistema:",
     "SettingsTabSystemSystemLanguageJapanese": "Giapponese",
     "SettingsTabSystemSystemLanguageAmericanEnglish": "Inglese americano",
     "SettingsTabSystemSystemLanguageFrench": "Francese",
@@ -107,8 +107,8 @@
     "SettingsTabSystemSystemLanguageLatinAmericanSpanish": "Spagnolo latino americano",
     "SettingsTabSystemSystemLanguageSimplifiedChinese": "Cinese semplificato",
     "SettingsTabSystemSystemLanguageTraditionalChinese": "Cinese tradizionale",
-    "SettingsTabSystemSystemTimeZone": "Fuso orario di sistema:",
-    "SettingsTabSystemSystemTime": "Data e ora di sistema:",
+    "SettingsTabSystemSystemTimeZone": "Fuso orario del sistema:",
+    "SettingsTabSystemSystemTime": "Data e ora del sistema:",
     "SettingsTabSystemEnableVsync": "Attiva VSync",
     "SettingsTabSystemEnablePptc": "Attiva PPTC (Profiled Persistent Translation Cache)",
     "SettingsTabSystemEnableFsIntegrityChecks": "Attiva controlli d'integrità FS",
@@ -155,7 +155,7 @@
     "SettingsTabLoggingEnableTraceLogs": "Attiva Trace Logs",
     "SettingsTabLoggingEnableGuestLogs": "Attiva Guest Logs",
     "SettingsTabLoggingEnableFsAccessLogs": "Attiva Fs Access Logs",
-    "SettingsTabLoggingFsGlobalAccessLogMode": "Modalità di log accesso globale Fs:",
+    "SettingsTabLoggingFsGlobalAccessLogMode": "Modalità log accesso globale Fs:",
     "SettingsTabLoggingDeveloperOptions": "Opzioni da sviluppatore (AVVISO: Ridurrà le prestazioni)",
     "SettingsTabLoggingOpenglLogLevel": "Livello di log OpenGL:",
     "SettingsTabLoggingOpenglLogLevelNone": "Nessuno",
@@ -240,7 +240,7 @@
     "ControllerSettingsRightSR": "SR",
     "ControllerSettingsExtraButtonsLeft": "Tasto sinitro",
     "ControllerSettingsExtraButtonsRight": "Tasto destro",
-    "ControllerSettingsMisc": "Miscellanee",
+    "ControllerSettingsMisc": "Varie",
     "ControllerSettingsTriggerThreshold": "Sensibilità dei grilletti:",
     "ControllerSettingsMotion": "Movimento",
     "ControllerSettingsMotionUseCemuhookCompatibleMotion": "Usa sensore compatibile con CemuHook",
@@ -261,7 +261,7 @@
     "UserProfilesClose": "Chiudi",
     "ProfileImageSelectionTitle": "Selezione dell'immagine profilo",
     "ProfileImageSelectionHeader": "Scegli un'immagine profilo",
-    "ProfileImageSelectionNote": "Puoi importare un'immagine profilo personalizzata o selezionare un avatar dal firmware di sistema",
+    "ProfileImageSelectionNote": "Puoi importare un'immagine profilo personalizzata o selezionare un avatar dal firmware del sistema",
     "ProfileImageSelectionImportImage": "Importa file immagine",
     "ProfileImageSelectionSelectAvatar": "Seleziona avatar dal firmware",
     "InputDialogTitle": "Input Dialog",
@@ -293,7 +293,7 @@
     "ControllerSettingsRumbleStrongMultiplier": "Moltiplicatore vibrazione forte",
     "ControllerSettingsRumbleWeakMultiplier": "Moltiplicatore vibrazione debole",
     "DialogMessageSaveNotAvailableMessage": "Non ci sono dati di salvataggio per {0} [{1:x16}]",
-    "DialogMessageSaveNotAvailableCreateSaveMessage": "Vuoi creare dei dat di salvataggio per questo gioco?",
+    "DialogMessageSaveNotAvailableCreateSaveMessage": "Vuoi creare dei dati di salvataggio per questo gioco?",
     "DialogConfirmationTitle": "Ryujinx - Conferma",
     "DialogUpdaterTitle": "Ryujinx - Updater",
     "DialogErrorTitle": "Ryujinx - Errore",
@@ -318,7 +318,7 @@
     "DialogUpdaterDownloadingMessage": "Download dell'aggiornamento...",
     "DialogUpdaterExtractionMessage": "Estrazione dell'aggiornamento...",
     "DialogUpdaterRenamingMessage": "Rinominazione dell'aggiornamento...",
-    "DialogUpdaterAddingFilesMessage": "Aggiunta nuovo aggiornamento...",
+    "DialogUpdaterAddingFilesMessage": "Aggiunta del nuovo aggiornamento...",
     "DialogUpdaterCompleteMessage": "Aggiornamento completato!",
     "DialogUpdaterRestartMessage": "Vuoi riavviare Ryujinx adesso?",
     "DialogUpdaterArchNotSupportedMessage": "Non stai usando un'architettura di sistema supportata!",
@@ -331,7 +331,7 @@
     "DialogThemeRestartMessage": "Il tema è stato salvato. E' richiesto un riavvio per applicare un tema.",
     "DialogThemeRestartSubMessage": "Vuoi riavviare?",
     "DialogFirmwareInstallEmbeddedMessage": "Vuoi installare il firmware incorporato in questo gioco? (Firmware {0})",
-    "DialogFirmwareInstallEmbeddedSuccessMessage": "Non è stato trovato alcun firmware installato, ma Ryujinx è riuscito di installare il firmware {0} dal gioco fornito.\nL'emulatore si avvierà adesso.",
+    "DialogFirmwareInstallEmbeddedSuccessMessage": "Non è stato trovato alcun firmware installato, ma Ryujinx è riuscito ad installare il firmware {0} dal gioco fornito.\nL'emulatore si avvierà adesso.",
     "DialogFirmwareNoFirmwareInstalledMessage": "Nessun firmware installato",
     "DialogFirmwareInstalledMessage": "Il firmware {0} è stato installato",
     "DialogOpenSettingsWindowLabel": "Apri finestra delle impostazioni",
@@ -355,14 +355,14 @@
     "DialogShaderDeletionMessage": "Stai per eliminare la Shader cache per :\n\n{0}\n\nSei sicuro di voler proseguire?",
     "DialogShaderDeletionErrorMessage": "Errore nell'eliminazione della Shader cache a {0}: {1}",
     "DialogRyujinxErrorMessage": "Ryujinx ha incontrato un errore",
-    "DialogInvalidTitleIdErrorMessage": "Errore UI: Il gioco selezionato non ha un title ID valido",
-    "DialogFirmwareInstallerFirmwareNotFoundErrorMessage": "Un firmware di sistema valido non è stato trovato in {0}.",
+    "DialogInvalidTitleIdErrorMessage": "Errore UI: Il gioco selezionato non ha un ID titolo valido",
+    "DialogFirmwareInstallerFirmwareNotFoundErrorMessage": "Un firmware del sistema valido non è stato trovato in {0}.",
     "DialogFirmwareInstallerFirmwareInstallTitle": "Installa firmware {0}",
-    "DialogFirmwareInstallerFirmwareInstallMessage": "La versione di sistema {0} sarà installata.",
+    "DialogFirmwareInstallerFirmwareInstallMessage": "La versione del sistema {0} sarà installata.",
     "DialogFirmwareInstallerFirmwareInstallSubMessage": "\n\nQuesta sostituirà l'attuale versione di sistema {0}.",
     "DialogFirmwareInstallerFirmwareInstallConfirmMessage": "\n\nVuoi continuare?",
     "DialogFirmwareInstallerFirmwareInstallWaitMessage": "Installazione del firmware...",
-    "DialogFirmwareInstallerFirmwareInstallSuccessMessage": "La versione di sistema {0} è stata installata.",
+    "DialogFirmwareInstallerFirmwareInstallSuccessMessage": "La versione del sistema {0} è stata installata.",
     "DialogUserProfileDeletionWarningMessage": "Non ci sarebbero altri profili da aprire se il profilo selezionato viene cancellato",
     "DialogUserProfileDeletionConfirmMessage": "Vuoi eliminare il profilo selezionato?",
     "DialogControllerSettingsModifiedConfirmMessage": "Le attuali impostazioni del controller sono state aggiornate.",
@@ -557,4 +557,40 @@
     "SettingsXamlThemeFile" : "File del tema xaml",
     "SettingsTabHotkeysResScaleUpHotkey": "Aumentare la risoluzione:",
     "SettingsTabHotkeysResScaleDownHotkey": "Diminuire la risoluzione:"
+    "AvatarWindowTitle": "Gestisci account - Avatar"
+    "Amiibo": "Amiibo",
+    "Unknown": "Sconosciuto",
+    "Usage": "Utilizzo",
+    "Writable": "Scrivibile",
+    "SelectDlcDialogTitle": "Seleziona file dei DLC",
+    "SelectUpdateDialogTitle": "Seleziona file di aggiornamento",
+    "UserProfileWindowTitle": "Gestisci profili degli utenti",
+    "CheatWindowTitle": "Gestisci cheat dei giochi",
+    "DlcWindowTitle": "Gestisci DLC dei giochi",
+    "UpdateWindowTitle": "Gestisci aggiornamenti dei giochi",
+    "CheatWindowHeading": "Cheat disponibiili per {0} [{1}]",
+    "DlcWindowHeading": "DLC disponibili per {0} [{1}]",
+    "UserProfilesEditProfile": "Modifica selezionati",
+    "Cancel": "Annulla",
+    "Save": "Salva",
+    "Discard": "Scarta",
+    "UserProfilesSetProfileImage": "Imposta immagine profilo",
+    "UserProfileEmptyNameError": "È richiesto un nome",
+    "UserProfileNoImageError": "Dev'essere impostata un'immagine profilo",
+    "GameUpdateWindowHeading": "Aggiornamenti disponibili per {0} [{1}]",
+    "UserProfilesName": "Name:",
+    "UserProfilesUserId": "User Id:",
+    "SettingsTabGraphicsBackend": "Backend grafica",
+    "SettingsTabGraphicsBackendTooltip": "Backend grafica da usare",
+    "SettingsEnableTextureRecompression": "Abilita Ricompressione Texture",
+    "SettingsEnableTextureRecompressionTooltip": "Comprime alcune texture per ridurre l'utilizzo della VRAM.\n\nL'utilizzo è consigliato con GPU con meno di 4GB di VRAM.\n\nLascia su OFF se non sei sicuro.",
+    "SettingsTabGraphicsPreferredGpu": "GPU preferita",
+    "SettingsTabGraphicsPreferredGpuTooltip": "Seleziona la scheda grafica che verrà usata con la backend grafica Vulkan.\n\nNon influenza la GPU che userà OpenGL.\n\nImposta la GPU contrassegnata come \"dGPU\" se non sei sicuro. Se non ce n'è una, lascia intatta quest'impostazione.",
+    "SettingsAppRequiredRestartMessage": "È richiesto un riavvio di Ryujinx",
+    "SettingsGpuBackendRestartMessage": "Le impostazioni della backend grafica o della GPU sono state modificate. Questo richiederà un riavvio perché le modifiche siano applicate",
+    "SettingsGpuBackendRestartSubMessage": "Vuoi riavviare ora?",
+    "RyujinxUpdaterMessage": "Vuoi aggiornare Ryujinx all'ultima versione?",
+    "SettingsTabHotkeysVolumeUpHotkey": "Aumentare il volume:",
+    "SettingsTabHotkeysVolumeDownHotkey": "Diminuire il volume:",
+    "VolumeShort": "Vol"
 }

Ryujinx.Ava/Assets/Locales/pl_PL.json

@@ -588,5 +588,9 @@
   "SettingsTabGraphicsPreferredGpuTooltip": "Wybierz kartę graficzną, która będzie używana z backendem graficznym Vulkan.\n\nNie wpływa na GPU używane przez OpenGL.\n\nW razie wątpliwości ustaw flagę GPU jako \"dGPU\". Jeśli żadnej nie ma, pozostaw nietknięte.",
   "SettingsAppRequiredRestartMessage": "Wymagane Zrestartowanie Ryujinx",
   "SettingsGpuBackendRestartMessage": "Zmieniono ustawienia Backendu Graficznego lub GPU. Będzie to wymagało ponownego uruchomienia",
-  "SettingsGpuBackendRestartSubMessage": "Czy chcesz zrestartować teraz?"
+  "SettingsGpuBackendRestartSubMessage": "Czy chcesz zrestartować teraz?",
+  "RyujinxUpdaterMessage": "Czy chcesz zaktualizować Ryujinx do najnowszej wersji?",
+  "SettingsTabHotkeysVolumeUpHotkey": "Zwiększ Głośność:",
+  "SettingsTabHotkeysVolumeDownHotkey": "Zmniejsz Głośność:",
+  "VolumeShort": "Głoś"
 }

Ryujinx.Ava/Modules/Updater/Updater.cs

@@ -278,7 +278,7 @@ namespace Ryujinx.Modules
                 {
                     string ryuName = Path.GetFileName(Environment.ProcessPath);
                     string ryuExe = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, ryuName);
-                    string ryuArg = string.Join(" ", Environment.GetCommandLineArgs().Skip(1).ToArray());
+                    var ryuArg = Environment.GetCommandLineArgs().Skip(1);
 
                     if (!OperatingSystem.IsWindows())
                     {

Ryujinx.Ava/Ryujinx.Ava.csproj

@@ -34,7 +34,7 @@
     <PackageReference Include="Silk.NET.Vulkan" Version="2.10.1" />
     <PackageReference Include="Silk.NET.Vulkan.Extensions.EXT" Version="2.10.1" />
     <PackageReference Include="Silk.NET.Vulkan.Extensions.KHR" Version="2.10.1" />
-    <PackageReference Include="SPB" Version="0.0.4-build24" />
+    <PackageReference Include="SPB" Version="0.0.4-build27" />
     <PackageReference Include="SharpZipLib" Version="1.3.3" />
     <PackageReference Include="SixLabors.ImageSharp" Version="1.0.4" />
   </ItemGroup>

Ryujinx.Ava/Ui/Controls/ContentDialogHelper.cs

@@ -4,6 +4,7 @@ using Avalonia.Controls.ApplicationLifetimes;
 using Avalonia.Controls.Primitives;
 using Avalonia.Media;
 using Avalonia.Threading;
+using FluentAvalonia.Core;
 using FluentAvalonia.UI.Controls;
 using Ryujinx.Ava.Common.Locale;
 using Ryujinx.Ava.Ui.Models;
@@ -27,7 +28,10 @@ namespace Ryujinx.Ava.Ui.Controls
             string secondaryButton,
             string closeButton,
             int iconSymbol,
-            UserResult primaryButtonResult = UserResult.Ok)
+            UserResult primaryButtonResult = UserResult.Ok,
+            ManualResetEvent deferResetEvent = null,
+            Func<Window, Task> doWhileDeferred = null,
+            TypedEventHandler<ContentDialog, ContentDialogButtonClickEventArgs> deferCloseAction = null)
         {
             UserResult result = UserResult.None;
 
@@ -110,12 +114,19 @@ namespace Ryujinx.Ava.Ui.Controls
                 contentDialog.SecondaryButtonCommand = MiniCommand.Create(() =>
                 {
                     result = UserResult.No;
+                    contentDialog.PrimaryButtonClick -= deferCloseAction;
                 });
                 contentDialog.CloseButtonCommand = MiniCommand.Create(() =>
                 {
                     result = UserResult.Cancel;
+                    contentDialog.PrimaryButtonClick -= deferCloseAction;
                 });
 
+                if (deferResetEvent != null)
+                {
+                    contentDialog.PrimaryButtonClick += deferCloseAction;
+                }
+
                 await contentDialog.ShowAsync(ContentDialogPlacement.Popup);
 
                 overlay?.Close();
@@ -143,35 +154,20 @@ namespace Ryujinx.Ava.Ui.Controls
             Func<Window, Task> doWhileDeferred = null)
         {
             bool startedDeferring = false;
-
             UserResult result = UserResult.None;
 
-            ContentDialog contentDialog = new ContentDialog
-            {
-                Title = title,
-                PrimaryButtonText = primaryButton,
-                SecondaryButtonText = secondaryButton,
-                CloseButtonText = closeButton,
-                Content = CreateDialogTextContent(primaryText, secondaryText, iconSymbol),
-                PrimaryButtonCommand = MiniCommand.Create(() =>
-                {
-                    result = primaryButton == LocaleManager.Instance["InputDialogYes"] ? UserResult.Yes : UserResult.Ok;
-                }),
-            };
-            contentDialog.SecondaryButtonCommand = MiniCommand.Create(() =>
-            {
-                contentDialog.PrimaryButtonClick -= DeferClose;
-                result = UserResult.No;
-            });
-            contentDialog.CloseButtonCommand = MiniCommand.Create(() =>
-            {
-                contentDialog.PrimaryButtonClick -= DeferClose;
-                result = UserResult.Cancel;
-            });
-            contentDialog.PrimaryButtonClick += DeferClose;
-            await contentDialog.ShowAsync(ContentDialogPlacement.Popup);
-
-            return result;
+            return await ShowContentDialog(
+                title,
+                primaryText,
+                secondaryText,
+                primaryButton,
+                secondaryButton,
+                closeButton,
+                iconSymbol,
+                primaryButton == LocaleManager.Instance["InputDialogYes"] ? UserResult.Yes : UserResult.Ok,
+                deferResetEvent,
+                doWhileDeferred,
+                DeferClose);
 
             async void DeferClose(ContentDialog sender, ContentDialogButtonClickEventArgs args)
             {
@@ -180,7 +176,7 @@ namespace Ryujinx.Ava.Ui.Controls
                     return;
                 }
 
-                contentDialog.PrimaryButtonClick -= DeferClose;
+                sender.PrimaryButtonClick -= DeferClose;
 
                 startedDeferring = true;
 
@@ -188,7 +184,7 @@ namespace Ryujinx.Ava.Ui.Controls
 
                 result = primaryButton == LocaleManager.Instance["InputDialogYes"] ? UserResult.Yes : UserResult.Ok;
 
-                contentDialog.PrimaryButtonClick -= DeferClose;
+                sender.PrimaryButtonClick -= DeferClose;
 
 #pragma warning disable CS4014 // Because this call is not awaited, execution of the current method continues before the call is completed
                 Task.Run(() =>

Ryujinx.Ava/Ui/ViewModels/SettingsViewModel.cs

@@ -295,8 +295,6 @@ namespace Ryujinx.Ava.Ui.ViewModels
             if (_validTzRegions.Contains(location))
             {
                 TimeZone = location;
-
-                OnPropertyChanged(nameof(TimeZone));
             }
         }
 

Ryujinx.Ava/Ui/Windows/MainWindow.axaml

@@ -164,7 +164,7 @@
                                 <MenuItem
                                     Command="{ReflectionBinding ChangeLanguage}"
                                     CommandParameter="pl_PL"
-                                    Header="Polish" />
+                                    Header="Polski" />
                                 <MenuItem
                                     Command="{ReflectionBinding ChangeLanguage}"
                                     CommandParameter="ru_RU"

Ryujinx.Ava/Ui/Windows/SettingsWindow.axaml.cs

@@ -37,7 +37,7 @@ namespace Ryujinx.Ava.Ui.Windows
             InitializeComponent();
             Load();
 
-            FuncMultiValueConverter<string, string> converter = new(parts => string.Format("{0}  {1}   {2}", parts.ToArray()));
+            FuncMultiValueConverter<string, string> converter = new(parts => string.Format("{0}  {1}   {2}", parts.ToArray()).Trim());
             MultiBinding tzMultiBinding = new() { Converter = converter };
             tzMultiBinding.Bindings.Add(new Binding("UtcDifference"));
             tzMultiBinding.Bindings.Add(new Binding("Location"));

Ryujinx.Common/Utilities/EmbeddedResources.cs

@@ -129,7 +129,7 @@ namespace Ryujinx.Common
 
         private static (Assembly, string) ResolveManifestPath(string filename)
         {
-            var segments = filename.Split(new[] { '/' }, 2, StringSplitOptions.RemoveEmptyEntries);
+            var segments = filename.Split('/', 2, StringSplitOptions.RemoveEmptyEntries);
 
             if (segments.Length >= 2)
             {

Ryujinx.Graphics.Gpu/Shader/DiskCache/DiskCacheHostStorage.cs

@@ -22,7 +22,7 @@ namespace Ryujinx.Graphics.Gpu.Shader.DiskCache
         private const ushort FileFormatVersionMajor = 1;
         private const ushort FileFormatVersionMinor = 2;
         private const uint FileFormatVersionPacked = ((uint)FileFormatVersionMajor << 16) | FileFormatVersionMinor;
-        private const uint CodeGenVersion = 3732;
+        private const uint CodeGenVersion = 3807;
 
         private const string SharedTocFileName = "shared.toc";
         private const string SharedDataFileName = "shared.data";

Ryujinx.Graphics.Shader/CodeGen/Spirv/CodeGenContext.cs

@@ -262,6 +262,13 @@ namespace Ryujinx.Graphics.Shader.CodeGen.Spirv
 
             Instruction ioVariable, elemIndex;
 
+            Instruction invocationId = null;
+
+            if (Config.Stage == ShaderStage.TessellationControl && isOutAttr)
+            {
+                invocationId = Load(TypeS32(), Inputs[AttributeConsts.InvocationId]);
+            }
+
             bool isUserAttr = attr >= AttributeConsts.UserAttributeBase && attr < AttributeConsts.UserAttributeEnd;
 
             if (isUserAttr &&
@@ -273,7 +280,17 @@ namespace Ryujinx.Graphics.Shader.CodeGen.Spirv
                 elemIndex = Constant(TypeU32(), attrInfo.GetInnermostIndex());
                 var vecIndex = Constant(TypeU32(), (attr - AttributeConsts.UserAttributeBase) >> 4);
 
-                if (AttributeInfo.IsArrayAttributeSpirv(Config.Stage, isOutAttr))
+                bool isArray = AttributeInfo.IsArrayAttributeSpirv(Config.Stage, isOutAttr);
+
+                if (invocationId != null && isArray)
+                {
+                    return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId, index, vecIndex, elemIndex);
+                }
+                else if (invocationId != null)
+                {
+                    return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId, vecIndex, elemIndex);
+                }
+                else if (isArray)
                 {
                     return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, index, vecIndex, elemIndex);
                 }
@@ -308,12 +325,29 @@ namespace Ryujinx.Graphics.Shader.CodeGen.Spirv
 
             if ((type & (AggregateType.Array | AggregateType.Vector)) == 0)
             {
-                return isIndexed ? AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, index) : ioVariable;
+                if (invocationId != null)
+                {
+                    return isIndexed
+                        ? AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId, index)
+                        : AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId);
+                }
+                else
+                {
+                    return isIndexed ? AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, index) : ioVariable;
+                }
             }
 
             elemIndex = Constant(TypeU32(), attrInfo.GetInnermostIndex());
 
-            if (isIndexed)
+            if (invocationId != null && isIndexed)
+            {
+                return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId, index, elemIndex);
+            }
+            else if (invocationId != null)
+            {
+                return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId, elemIndex);
+            }
+            else if (isIndexed)
             {
                 return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, index, elemIndex);
             }
@@ -327,12 +361,29 @@ namespace Ryujinx.Graphics.Shader.CodeGen.Spirv
         {
             var storageClass = isOutAttr ? StorageClass.Output : StorageClass.Input;
 
+            Instruction invocationId = null;
+
+            if (Config.Stage == ShaderStage.TessellationControl && isOutAttr)
+            {
+                invocationId = Load(TypeS32(), Inputs[AttributeConsts.InvocationId]);
+            }
+
             elemType = AggregateType.FP32;
             var ioVariable = isOutAttr ? OutputsArray : InputsArray;
             var vecIndex = ShiftRightLogical(TypeS32(), attrIndex, Constant(TypeS32(), 2));
             var elemIndex = BitwiseAnd(TypeS32(), attrIndex, Constant(TypeS32(), 3));
 
-            if (AttributeInfo.IsArrayAttributeSpirv(Config.Stage, isOutAttr))
+            bool isArray = AttributeInfo.IsArrayAttributeSpirv(Config.Stage, isOutAttr);
+
+            if (invocationId != null && isArray)
+            {
+                return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId, index, vecIndex, elemIndex);
+            }
+            else if (invocationId != null)
+            {
+                return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, invocationId, vecIndex, elemIndex);
+            }
+            else if (isArray)
             {
                 return AccessChain(TypePointer(storageClass, GetType(elemType)), ioVariable, index, vecIndex, elemIndex);
             }

Ryujinx.Graphics.Shader/CodeGen/Spirv/Declarations.cs

@@ -473,6 +473,11 @@ namespace Ryujinx.Graphics.Shader.CodeGen.Spirv
                         var attrType = context.TypeVector(context.TypeFP32(), (LiteralInteger)4);
                         attrType = context.TypeArray(attrType, context.Constant(context.TypeU32(), (LiteralInteger)MaxAttributes));
 
+                        if (context.Config.Stage == ShaderStage.TessellationControl)
+                        {
+                            attrType = context.TypeArray(attrType, context.Constant(context.TypeU32(), context.Config.ThreadsPerInputPrimitive));
+                        }
+
                         var spvType = context.TypePointer(StorageClass.Output, attrType);
                         var spvVar = context.Variable(spvType, StorageClass.Output);
 
@@ -543,6 +548,11 @@ namespace Ryujinx.Graphics.Shader.CodeGen.Spirv
                 }
             }
 
+            if (context.Config.Stage == ShaderStage.TessellationControl && isOutAttr && !perPatch)
+            {
+                attrType = context.TypeArray(attrType, context.Constant(context.TypeU32(), context.Config.ThreadsPerInputPrimitive));
+            }
+
             var spvType = context.TypePointer(storageClass, attrType);
             var spvVar = context.Variable(spvType, storageClass);
 
@@ -634,6 +644,11 @@ namespace Ryujinx.Graphics.Shader.CodeGen.Spirv
                 attrType = context.TypeArray(attrType, context.Constant(context.TypeU32(), (LiteralInteger)arraySize));
             }
 
+            if (context.Config.Stage == ShaderStage.TessellationControl && isOutAttr)
+            {
+                attrType = context.TypeArray(attrType, context.Constant(context.TypeU32(), context.Config.ThreadsPerInputPrimitive));
+            }
+
             var spvType = context.TypePointer(storageClass, attrType);
             var spvVar = context.Variable(spvType, storageClass);
 

Ryujinx.Graphics.Shader/Decoders/Decoder.cs

@@ -377,6 +377,8 @@ namespace Ryujinx.Graphics.Shader.Decoders
 
                 if (lastOp.Name == InstName.Brx && block.Successors.Count == (hasNext ? 1 : 0))
                 {
+                    HashSet<ulong> visited = new HashSet<ulong>();
+
                     InstBrx opBrx = new InstBrx(lastOp.RawOpCode);
                     ulong baseOffset = lastOp.GetAbsoluteAddress();
 
@@ -392,9 +394,14 @@ namespace Ryujinx.Graphics.Shader.Decoders
                     for (int i = 0; i < cbOffsetsCount; i++)
                     {
                         uint targetOffset = config.ConstantBuffer1Read(cbBaseOffset + i * 4);
-                        Block target = getBlock(baseOffset + targetOffset);
-                        target.Predecessors.Add(block);
-                        block.Successors.Add(target);
+                        ulong targetAddress = baseOffset + targetOffset;
+
+                        if (visited.Add(targetAddress))
+                        {
+                            Block target = getBlock(targetAddress);
+                            target.Predecessors.Add(block);
+                            block.Successors.Add(target);
+                        }
                     }
                 }
             }

Ryujinx.Graphics.Shader/Instructions/InstEmitFlowControl.cs

@@ -41,24 +41,82 @@ namespace Ryujinx.Graphics.Shader.Instructions
 
             Operand address = context.IAdd(Register(op.SrcA, RegisterType.Gpr), Const(offset));
 
-            // Sorting the target addresses in descending order improves the code,
-            // since it will always check the most distant targets first, then the
-            // near ones. This can be easily transformed into if/else statements.
-            var sortedTargets = context.CurrBlock.Successors.Skip(startIndex).OrderByDescending(x => x.Address);
+            var targets = context.CurrBlock.Successors.Skip(startIndex);
 
-            Block lastTarget = sortedTargets.LastOrDefault();
+            bool allTargetsSinglePred = true;
+            int total = context.CurrBlock.Successors.Count - startIndex;
+            int count = 0;
 
-            foreach (Block possibleTarget in sortedTargets)
+            foreach (var target in targets.OrderBy(x => x.Address))
             {
-                Operand label = context.GetLabel(possibleTarget.Address);
-
-                if (possibleTarget != lastTarget)
+                if (++count < total && (target.Predecessors.Count > 1 || target.Address <= context.CurrBlock.Address))
                 {
-                    context.BranchIfTrue(label, context.ICompareEqual(address, Const((int)possibleTarget.Address)));
+                    allTargetsSinglePred = false;
+                    break;
                 }
-                else
+            }
+
+            if (allTargetsSinglePred)
+            {
+                // Chain blocks, each target block will check if the BRX target address
+                // matches its own address, if not, it jumps to the next target which will do the same check,
+                // until it reaches the last possible target, which executed unconditionally.
+                // We can only do this if the BRX block is the only predecessor of all target blocks.
+                // Additionally, this is not supported for blocks located before the current block,
+                // since it will be too late to insert a label, but this is something that can be improved
+                // in the future if necessary.
+
+                var sortedTargets = targets.OrderBy(x => x.Address);
+
+                Block currentTarget = null;
+                ulong firstTargetAddress = 0;
+
+                foreach (Block nextTarget in sortedTargets)
                 {
-                    context.Branch(label);
+                    if (currentTarget != null)
+                    {
+                        if (currentTarget.Address != nextTarget.Address)
+                        {
+                            context.SetBrxTarget(currentTarget.Address, address, (int)currentTarget.Address, nextTarget.Address);
+                        }
+                    }
+                    else
+                    {
+                        firstTargetAddress = nextTarget.Address;
+                    }
+
+                    currentTarget = nextTarget;
+                }
+
+                context.Branch(context.GetLabel(firstTargetAddress));
+            }
+            else
+            {
+                // Emit the branches sequentially.
+                // This generates slightly worse code, but should work for all cases.
+
+                var sortedTargets = targets.OrderByDescending(x => x.Address);
+                ulong lastTargetAddress = ulong.MaxValue;
+
+                count = 0;
+
+                foreach (Block target in sortedTargets)
+                {
+                    Operand label = context.GetLabel(target.Address);
+
+                    if (++count < total)
+                    {
+                        if (target.Address != lastTargetAddress)
+                        {
+                            context.BranchIfTrue(label, context.ICompareEqual(address, Const((int)target.Address)));
+                        }
+
+                        lastTargetAddress = target.Address;
+                    }
+                    else
+                    {
+                        context.Branch(label);
+                    }
                 }
             }
         }

Ryujinx.Graphics.Shader/StructuredIr/StructuredProgramContext.cs

@@ -37,7 +37,12 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
 
             Config = config;
 
-            if (config.GpPassthrough)
+            if (config.Stage == ShaderStage.TessellationControl)
+            {
+                // Required to index outputs.
+                Info.Inputs.Add(AttributeConsts.InvocationId);
+            }
+            else if (config.GpPassthrough)
             {
                 int passthroughAttributes = config.PassthroughAttributes;
                 while (passthroughAttributes != 0)

Ryujinx.Graphics.Shader/Translation/EmitterContext.cs

@@ -21,8 +21,33 @@ namespace Ryujinx.Graphics.Shader.Translation
 
         public int OperationsCount => _operations.Count;
 
+        private struct BrxTarget
+        {
+            public readonly Operand Selector;
+            public readonly int ExpectedValue;
+            public readonly ulong NextTargetAddress;
+
+            public BrxTarget(Operand selector, int expectedValue, ulong nextTargetAddress)
+            {
+                Selector = selector;
+                ExpectedValue = expectedValue;
+                NextTargetAddress = nextTargetAddress;
+            }
+        }
+
+        private class BlockLabel
+        {
+            public readonly Operand Label;
+            public BrxTarget BrxTarget;
+
+            public BlockLabel(Operand label)
+            {
+                Label = label;
+            }
+        }
+
         private readonly List<Operation> _operations;
-        private readonly Dictionary<ulong, Operand> _labels;
+        private readonly Dictionary<ulong, BlockLabel> _labels;
 
         public EmitterContext(DecodedProgram program, ShaderConfig config, bool isNonMain)
         {
@@ -30,7 +55,7 @@ namespace Ryujinx.Graphics.Shader.Translation
             Config = config;
             IsNonMain = isNonMain;
             _operations = new List<Operation>();
-            _labels = new Dictionary<ulong, Operand>();
+            _labels = new Dictionary<ulong, BlockLabel>();
 
             EmitStart();
         }
@@ -158,14 +183,40 @@ namespace Ryujinx.Graphics.Shader.Translation
 
         public Operand GetLabel(ulong address)
         {
-            if (!_labels.TryGetValue(address, out Operand label))
-            {
-                label = Label();
+            return EnsureBlockLabel(address).Label;
+        }
 
-                _labels.Add(address, label);
+        public void SetBrxTarget(ulong address, Operand selector, int targetValue, ulong nextTargetAddress)
+        {
+            BlockLabel blockLabel = EnsureBlockLabel(address);
+            Debug.Assert(blockLabel.BrxTarget.Selector == null);
+            blockLabel.BrxTarget = new BrxTarget(selector, targetValue, nextTargetAddress);
+        }
+
+        public void EnterBlock(ulong address)
+        {
+            BlockLabel blockLabel = EnsureBlockLabel(address);
+
+            MarkLabel(blockLabel.Label);
+
+            BrxTarget brxTarget = blockLabel.BrxTarget;
+
+            if (brxTarget.Selector != null)
+            {
+                this.BranchIfFalse(GetLabel(brxTarget.NextTargetAddress), this.ICompareEqual(brxTarget.Selector, Const(brxTarget.ExpectedValue)));
+            }
+        }
+
+        private BlockLabel EnsureBlockLabel(ulong address)
+        {
+            if (!_labels.TryGetValue(address, out BlockLabel blockLabel))
+            {
+                blockLabel = new BlockLabel(Label());
+
+                _labels.Add(address, blockLabel);
             }
 
-            return label;
+            return blockLabel;
         }
 
         public void PrepareForVertexReturn()

Ryujinx.Graphics.Shader/Translation/ShaderConfig.cs

@@ -306,7 +306,10 @@ namespace Ryujinx.Graphics.Shader.Translation
                 config._perPatchAttributeLocations = locationsMap;
             }
 
-            if (config.Stage != ShaderStage.Fragment)
+            // We don't consider geometry shaders using the geometry shader passthrough feature
+            // as being the last because when this feature is used, it can't actually modify any of the outputs,
+            // so the stage that comes before it is the last one that can do modifications.
+            if (config.Stage != ShaderStage.Fragment && (config.Stage != ShaderStage.Geometry || !config.GpPassthrough))
             {
                 LastInVertexPipeline = false;
             }

Ryujinx.Graphics.Shader/Translation/Translator.cs

@@ -162,7 +162,7 @@ namespace Ryujinx.Graphics.Shader.Translation
                 {
                     context.CurrBlock = block;
 
-                    context.MarkLabel(context.GetLabel(block.Address));
+                    context.EnterBlock(block.Address);
 
                     EmitOps(context, block);
                 }

Ryujinx.Graphics.Vulkan/BufferHolder.cs

@@ -12,13 +12,12 @@ namespace Ryujinx.Graphics.Vulkan
         private const int MaxUpdateBufferSize = 0x10000;
 
         public const AccessFlags DefaultAccessFlags =
+            AccessFlags.AccessIndirectCommandReadBit |
             AccessFlags.AccessShaderReadBit |
             AccessFlags.AccessShaderWriteBit |
             AccessFlags.AccessTransferReadBit |
             AccessFlags.AccessTransferWriteBit |
-            AccessFlags.AccessUniformReadBit |
-            AccessFlags.AccessShaderReadBit |
-            AccessFlags.AccessShaderWriteBit;
+            AccessFlags.AccessUniformReadBit;
 
         private readonly VulkanRenderer _gd;
         private readonly Device _device;
@@ -386,8 +385,25 @@ namespace Ryujinx.Graphics.Vulkan
             _waitable.WaitForFences(_gd.Api, _device, offset, size);
         }
 
+        private bool BoundToRange(int offset, ref int size)
+        {
+            if (offset >= Size)
+            {
+                return false;
+            }
+
+            size = Math.Min(Size - offset, size);
+
+            return true;
+        }
+
         public Auto<DisposableBuffer> GetBufferI8ToI16(CommandBufferScoped cbs, int offset, int size)
         {
+            if (!BoundToRange(offset, ref size))
+            {
+                return null;
+            }
+
             var key = new I8ToI16CacheKey(_gd);
 
             if (!_cachedConvertedBuffers.TryGetValue(offset, size, key, out var holder))
@@ -407,6 +423,11 @@ namespace Ryujinx.Graphics.Vulkan
 
         public Auto<DisposableBuffer> GetAlignedVertexBuffer(CommandBufferScoped cbs, int offset, int size, int stride, int alignment)
         {
+            if (!BoundToRange(offset, ref size))
+            {
+                return null;
+            }
+
             var key = new AlignedVertexBufferCacheKey(_gd, stride, alignment);
 
             if (!_cachedConvertedBuffers.TryGetValue(offset, size, key, out var holder))
@@ -428,6 +449,11 @@ namespace Ryujinx.Graphics.Vulkan
 
         public Auto<DisposableBuffer> GetBufferTopologyConversion(CommandBufferScoped cbs, int offset, int size, IndexBufferPattern pattern, int indexSize)
         {
+            if (!BoundToRange(offset, ref size))
+            {
+                return null;
+            }
+
             var key = new TopologyConversionCacheKey(_gd, pattern, indexSize);
 
             if (!_cachedConvertedBuffers.TryGetValue(offset, size, key, out var holder))

Ryujinx.Graphics.Vulkan/IndexBufferPattern.cs

@@ -38,7 +38,7 @@ namespace Ryujinx.Graphics.Vulkan
 
         public int GetPrimitiveCount(int vertexCount)
         {
-            return Math.Max(0, ((vertexCount - BaseIndex) + IndexStride - 1) / IndexStride);
+            return Math.Max(0, (vertexCount - BaseIndex) / IndexStride);
         }
 
         public int GetConvertedCount(int indexCount)

Ryujinx.Graphics.Vulkan/IndexBufferState.cs

@@ -49,7 +49,12 @@ namespace Ryujinx.Graphics.Vulkan
             }
             else
             {
-                autoBuffer = gd.BufferManager.GetBuffer(cbs.CommandBuffer, _handle, false, out int _);
+                autoBuffer = gd.BufferManager.GetBuffer(cbs.CommandBuffer, _handle, false, out int bufferSize);
+
+                if (_offset >= bufferSize)
+                {
+                    autoBuffer = null;
+                }
 
                 offset = _offset;
                 size = _size;

Ryujinx.Graphics.Vulkan/PipelineBase.cs

@@ -95,7 +95,7 @@ namespace Ryujinx.Graphics.Vulkan
 
             using var emptyVb = gd.BufferManager.Create(gd, EmptyVbSize);
             emptyVb.SetData(0, new byte[EmptyVbSize]);
-            _vertexBuffers[0] = new VertexBufferState(emptyVb.GetBuffer(), 0, EmptyVbSize, 0);
+            _vertexBuffers[0] = new VertexBufferState(emptyVb.GetBuffer(), 0, 0, EmptyVbSize, 0);
             _vertexBuffersDirty = ulong.MaxValue >> (64 - _vertexBuffers.Length);
 
             ClearScissor = new Rectangle<int>(0, 0, 0xffff, 0xffff);
@@ -228,10 +228,26 @@ namespace Ryujinx.Graphics.Vulkan
             Gd.Api.CmdClearAttachments(CommandBuffer, 1, &attachment, 1, &clearRect);
         }
 
-        public void CommandBufferBarrier()
+        public unsafe void CommandBufferBarrier()
         {
-            // TODO: More specific barrier?
-            Barrier();
+            MemoryBarrier memoryBarrier = new MemoryBarrier()
+            {
+                SType = StructureType.MemoryBarrier,
+                SrcAccessMask = BufferHolder.DefaultAccessFlags,
+                DstAccessMask = AccessFlags.AccessIndirectCommandReadBit
+            };
+
+            Gd.Api.CmdPipelineBarrier(
+                CommandBuffer,
+                PipelineStageFlags.PipelineStageAllCommandsBit,
+                PipelineStageFlags.PipelineStageDrawIndirectBit,
+                0,
+                1,
+                memoryBarrier,
+                0,
+                null,
+                0,
+                null);
         }
 
         public void CopyBuffer(BufferHandle source, BufferHandle destination, int srcOffset, int dstOffset, int size)
@@ -535,10 +551,11 @@ namespace Ryujinx.Graphics.Vulkan
                 vkBlend = new PipelineColorBlendAttachmentState();
             }
 
-            _newState.BlendConstantR = blend.BlendConstant.Red;
-            _newState.BlendConstantG = blend.BlendConstant.Green;
-            _newState.BlendConstantB = blend.BlendConstant.Blue;
-            _newState.BlendConstantA = blend.BlendConstant.Alpha;
+            DynamicState.SetBlendConstants(
+                blend.BlendConstant.Red,
+                blend.BlendConstant.Green,
+                blend.BlendConstant.Blue,
+                blend.BlendConstant.Alpha);
 
             SignalStateChange();
         }
@@ -823,7 +840,7 @@ namespace Ryujinx.Graphics.Vulkan
 
                 if (range.Handle != BufferHandle.Null)
                 {
-                    _transformFeedbackBuffers[i] = 
+                    _transformFeedbackBuffers[i] =
                         new BufferState(Gd.BufferManager.GetBuffer(CommandBuffer, range.Handle, range.Offset, range.Size, true), range.Offset, range.Size);
                     _transformFeedbackBuffers[i].BindTransformFeedbackBuffer(Gd, Cbs, (uint)i);
                 }
@@ -1243,7 +1260,7 @@ namespace Ryujinx.Graphics.Vulkan
 
                     _vertexBuffers[i].BindVertexBuffer(Gd, Cbs, (uint)i, ref _newState);
 
-                    _vertexBuffersDirty &= ~(1u << i);
+                    _vertexBuffersDirty &= ~(1UL << i);
                 }
             }
 

Ryujinx.Graphics.Vulkan/PipelineConverter.cs

@@ -135,11 +135,6 @@ namespace Ryujinx.Graphics.Vulkan
 
             // It is assumed that Dynamic State is enabled when this conversion is used.
 
-            pipeline.BlendConstantA = state.BlendDescriptors[0].BlendConstant.Alpha;
-            pipeline.BlendConstantB = state.BlendDescriptors[0].BlendConstant.Blue;
-            pipeline.BlendConstantG = state.BlendDescriptors[0].BlendConstant.Green;
-            pipeline.BlendConstantR = state.BlendDescriptors[0].BlendConstant.Red;
-
             pipeline.CullMode = state.CullEnable ? state.CullMode.Convert() : CullModeFlags.CullModeNone;
 
             pipeline.DepthBoundsTestEnable = false; // Not implemented.

Ryujinx.Graphics.Vulkan/PipelineDynamicState.cs

@@ -19,21 +19,34 @@ namespace Ryujinx.Graphics.Vulkan
         private uint _frontWriteMask;
         private uint _frontReference;
 
+        private Array4<float> _blendConstants;
+
         public int ViewportsCount;
         public Array16<Viewport> Viewports;
 
         private enum DirtyFlags
         {
             None = 0,
-            DepthBias = 1 << 0,
-            Scissor = 1 << 1,
-            Stencil = 1 << 2,
-            Viewport = 1 << 3,
-            All = DepthBias | Scissor | Stencil | Viewport
+            Blend = 1 << 0,
+            DepthBias = 1 << 1,
+            Scissor = 1 << 2,
+            Stencil = 1 << 3,
+            Viewport = 1 << 4,
+            All = Blend | DepthBias | Scissor | Stencil | Viewport
         }
 
         private DirtyFlags _dirty;
 
+        public void SetBlendConstants(float r, float g, float b, float a)
+        {
+            _blendConstants[0] = r;
+            _blendConstants[1] = g;
+            _blendConstants[2] = b;
+            _blendConstants[3] = a;
+
+            _dirty |= DirtyFlags.Blend;
+        }
+
         public void SetDepthBias(float slopeFactor, float constantFactor, float clamp)
         {
             _depthBiasSlopeFactor = slopeFactor;
@@ -87,6 +100,11 @@ namespace Ryujinx.Graphics.Vulkan
 
         public void ReplayIfDirty(Vk api, CommandBuffer commandBuffer)
         {
+            if (_dirty.HasFlag(DirtyFlags.Blend))
+            {
+                RecordBlend(api, commandBuffer);
+            }
+
             if (_dirty.HasFlag(DirtyFlags.DepthBias))
             {
                 RecordDepthBias(api, commandBuffer);
@@ -110,6 +128,11 @@ namespace Ryujinx.Graphics.Vulkan
             _dirty = DirtyFlags.None;
         }
 
+        private void RecordBlend(Vk api, CommandBuffer commandBuffer)
+        {
+            api.CmdSetBlendConstants(commandBuffer, _blendConstants.AsSpan());
+        }
+
         private void RecordDepthBias(Vk api, CommandBuffer commandBuffer)
         {
             api.CmdSetDepthBias(commandBuffer, _depthBiasConstantFactor, _depthBiasClamp, _depthBiasSlopeFactor);

Ryujinx.Graphics.Vulkan/PipelineState.cs

@@ -499,7 +499,7 @@ namespace Ryujinx.Graphics.Vulkan
                 colorBlendState.BlendConstants[3] = BlendConstantA;
 
                 bool supportsExtDynamicState = gd.Capabilities.SupportsExtendedDynamicState;
-                int dynamicStatesCount = supportsExtDynamicState ? 8 : 7;
+                int dynamicStatesCount = supportsExtDynamicState ? 9 : 8;
 
                 DynamicState* dynamicStates = stackalloc DynamicState[dynamicStatesCount];
 
@@ -510,10 +510,11 @@ namespace Ryujinx.Graphics.Vulkan
                 dynamicStates[4] = DynamicState.StencilCompareMask;
                 dynamicStates[5] = DynamicState.StencilWriteMask;
                 dynamicStates[6] = DynamicState.StencilReference;
+                dynamicStates[7] = DynamicState.BlendConstants;
 
                 if (supportsExtDynamicState)
                 {
-                    dynamicStates[7] = DynamicState.VertexInputBindingStrideExt;
+                    dynamicStates[8] = DynamicState.VertexInputBindingStrideExt;
                 }
 
                 var pipelineDynamicStateCreateInfo = new PipelineDynamicStateCreateInfo()

Ryujinx.Graphics.Vulkan/TextureStorage.cs

@@ -480,6 +480,8 @@ namespace Ryujinx.Graphics.Vulkan
             if (--_viewsCount == 0)
             {
                 _gd.PipelineInternal?.FlushCommandsIfWeightExceeding(_imageAuto, _size);
+
+                Dispose();
             }
         }
 

Ryujinx.Graphics.Vulkan/TextureView.cs

@@ -515,10 +515,10 @@ namespace Ryujinx.Graphics.Vulkan
                 dst.Info,
                 srcRegion,
                 dstRegion,
-                src.FirstLevel,
-                dst.FirstLevel,
                 src.FirstLayer,
                 dst.FirstLayer,
+                src.FirstLevel,
+                dst.FirstLevel,
                 layers,
                 levels,
                 linearFilter,

Ryujinx.Graphics.Vulkan/VertexBufferState.cs

@@ -57,37 +57,48 @@ namespace Ryujinx.Graphics.Vulkan
                 if (gd.NeedsVertexBufferAlignment(AttributeScalarAlignment, out int alignment) && (_stride % alignment) != 0)
                 {
                     autoBuffer = gd.BufferManager.GetAlignedVertexBuffer(cbs, _handle, _offset, _size, _stride, alignment);
-                    int stride = (_stride + (alignment - 1)) & -alignment;
 
-                    var buffer = autoBuffer.Get(cbs, _offset, _size).Value;
+                    if (autoBuffer != null)
+                    {
+                        int stride = (_stride + (alignment - 1)) & -alignment;
+                        int newSize = (_size / _stride) * stride;
 
-                    if (gd.Capabilities.SupportsExtendedDynamicState)
-                    {
-                        gd.ExtendedDynamicStateApi.CmdBindVertexBuffers2(
-                            cbs.CommandBuffer,
-                            binding,
-                            1,
-                            buffer,
-                            0,
-                            (ulong)(_size / _stride) * (ulong)stride,
-                            (ulong)stride);
-                    }
-                    else
-                    {
-                        gd.Api.CmdBindVertexBuffers(cbs.CommandBuffer, binding, 1, buffer, 0);
+                        var buffer = autoBuffer.Get(cbs, 0, newSize).Value;
+
+                        if (gd.Capabilities.SupportsExtendedDynamicState)
+                        {
+                            gd.ExtendedDynamicStateApi.CmdBindVertexBuffers2(
+                                cbs.CommandBuffer,
+                                binding,
+                                1,
+                                buffer,
+                                0,
+                                (ulong)newSize,
+                                (ulong)stride);
+                        }
+                        else
+                        {
+                            gd.Api.CmdBindVertexBuffers(cbs.CommandBuffer, binding, 1, buffer, 0);
+                        }
+
+                        _buffer = autoBuffer;
                     }
 
-                    _buffer = autoBuffer;
-                    state.Internal.VertexBindingDescriptions[DescriptorIndex].Stride = (uint)stride;
+                    state.Internal.VertexBindingDescriptions[DescriptorIndex].Stride = (uint)_stride;
 
                     return;
                 }
                 else
                 {
-                    autoBuffer = gd.BufferManager.GetBuffer(cbs.CommandBuffer, _handle, false, out int _);
+                    autoBuffer = gd.BufferManager.GetBuffer(cbs.CommandBuffer, _handle, false, out int size);
 
                     // The original stride must be reapplied in case it was rewritten.
                     state.Internal.VertexBindingDescriptions[DescriptorIndex].Stride = (uint)_stride;
+
+                    if (_offset >= size)
+                    {
+                        autoBuffer = null;
+                    }
                 }
             }
 

Ryujinx.Graphics.Vulkan/VulkanInitialization.cs

@@ -53,7 +53,7 @@ namespace Ryujinx.Graphics.Vulkan
             "VUID-VkSubpassDependency-srcSubpass-00867"
         };
 
-        internal static Instance CreateInstance(Vk api, GraphicsDebugLevel logLevel, string[] requiredExtensions, out ExtDebugReport debugReport, out DebugReportCallbackEXT debugReportCallback)
+        internal static Instance CreateInstance(Vk api, GraphicsDebugLevel logLevel, string[] requiredExtensions, out ExtDebugUtils debugUtils, out DebugUtilsMessengerEXT debugUtilsMessenger)
         {
             var enabledLayers = new List<string>();
 
@@ -89,7 +89,7 @@ namespace Ryujinx.Graphics.Vulkan
                 AddAvailableLayer("VK_LAYER_KHRONOS_validation");
             }
 
-            var enabledExtensions = requiredExtensions.Append(ExtDebugReport.ExtensionName).ToArray();
+            var enabledExtensions = requiredExtensions.Append(ExtDebugUtils.ExtensionName).ToArray();
 
             var appName = Marshal.StringToHGlobalAnsi(AppName);
 
@@ -139,22 +139,18 @@ namespace Ryujinx.Graphics.Vulkan
                 Marshal.FreeHGlobal(ppEnabledLayers[i]);
             }
 
-            CreateDebugCallbacks(api, logLevel, instance, out debugReport, out debugReportCallback);
+            CreateDebugMessenger(api, logLevel, instance, out debugUtils, out debugUtilsMessenger);
 
             return instance;
         }
 
-        private unsafe static uint DebugReport(
-            uint flags,
-            DebugReportObjectTypeEXT objectType,
-            ulong @object,
-            nuint location,
-            int messageCode,
-            byte* layerPrefix,
-            byte* message,
-            void* userData)
+        private unsafe static uint DebugMessenger(
+            DebugUtilsMessageSeverityFlagsEXT messageSeverity,
+            DebugUtilsMessageTypeFlagsEXT messageTypes,
+            DebugUtilsMessengerCallbackDataEXT* pCallbackData,
+            void* pUserData)
         {
-            var msg = Marshal.PtrToStringAnsi((IntPtr)message);
+            var msg = Marshal.PtrToStringAnsi((IntPtr)pCallbackData->PMessage);
 
             foreach (string excludedMessagePart in _excludedMessages)
             {
@@ -164,26 +160,20 @@ namespace Ryujinx.Graphics.Vulkan
                 }
             }
 
-            DebugReportFlagsEXT debugFlags = (DebugReportFlagsEXT)flags;
-
-            if (debugFlags.HasFlag(DebugReportFlagsEXT.DebugReportErrorBitExt))
+            if (messageSeverity.HasFlag(DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityErrorBitExt))
             {
                 Logger.Error?.Print(LogClass.Gpu, msg);
                 //throw new Exception(msg);
             }
-            else if (debugFlags.HasFlag(DebugReportFlagsEXT.DebugReportWarningBitExt))
+            else if (messageSeverity.HasFlag(DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityWarningBitExt))
             {
                 Logger.Warning?.Print(LogClass.Gpu, msg);
             }
-            else if (debugFlags.HasFlag(DebugReportFlagsEXT.DebugReportInformationBitExt))
+            else if (messageSeverity.HasFlag(DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityInfoBitExt))
             {
                 Logger.Info?.Print(LogClass.Gpu, msg);
             }
-            else if (debugFlags.HasFlag(DebugReportFlagsEXT.DebugReportPerformanceWarningBitExt))
-            {
-                Logger.Warning?.Print(LogClass.Gpu, msg);
-            }
-            else
+            else // if (messageSeverity.HasFlag(DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityVerboseBitExt))
             {
                 Logger.Debug?.Print(LogClass.Gpu, msg);
             }
@@ -551,46 +541,59 @@ namespace Ryujinx.Graphics.Vulkan
             return new CommandBufferPool(api, device, queue, queueLock, queueFamilyIndex);
         }
 
-        internal unsafe static void CreateDebugCallbacks(
+        internal unsafe static void CreateDebugMessenger(
             Vk api,
             GraphicsDebugLevel logLevel,
             Instance instance,
-            out ExtDebugReport debugReport,
-            out DebugReportCallbackEXT debugReportCallback)
+            out ExtDebugUtils debugUtils,
+            out DebugUtilsMessengerEXT debugUtilsMessenger)
         {
-            debugReport = default;
+            debugUtils = default;
 
             if (logLevel != GraphicsDebugLevel.None)
             {
-                if (!api.TryGetInstanceExtension(instance, out debugReport))
+                if (!api.TryGetInstanceExtension(instance, out debugUtils))
                 {
-                    debugReportCallback = default;
+                    debugUtilsMessenger = default;
                     return;
                 }
 
-                var flags = logLevel switch
+                var filterLogType = logLevel switch
                 {
-                    GraphicsDebugLevel.Error => DebugReportFlagsEXT.DebugReportErrorBitExt,
-                    GraphicsDebugLevel.Slowdowns => DebugReportFlagsEXT.DebugReportErrorBitExt | DebugReportFlagsEXT.DebugReportPerformanceWarningBitExt,
-                    GraphicsDebugLevel.All => DebugReportFlagsEXT.DebugReportInformationBitExt |
-                                              DebugReportFlagsEXT.DebugReportWarningBitExt |
-                                              DebugReportFlagsEXT.DebugReportPerformanceWarningBitExt |
-                                              DebugReportFlagsEXT.DebugReportErrorBitExt |
-                                              DebugReportFlagsEXT.DebugReportDebugBitExt,
+                    GraphicsDebugLevel.Error => DebugUtilsMessageTypeFlagsEXT.DebugUtilsMessageTypeValidationBitExt,
+                    GraphicsDebugLevel.Slowdowns => DebugUtilsMessageTypeFlagsEXT.DebugUtilsMessageTypeValidationBitExt |
+                                                    DebugUtilsMessageTypeFlagsEXT.DebugUtilsMessageTypePerformanceBitExt,
+                    GraphicsDebugLevel.All => DebugUtilsMessageTypeFlagsEXT.DebugUtilsMessageTypeGeneralBitExt |
+                                              DebugUtilsMessageTypeFlagsEXT.DebugUtilsMessageTypeValidationBitExt |
+                                              DebugUtilsMessageTypeFlagsEXT.DebugUtilsMessageTypePerformanceBitExt,
                     _ => throw new ArgumentException($"Invalid log level \"{logLevel}\".")
                 };
-                var debugReportCallbackCreateInfo = new DebugReportCallbackCreateInfoEXT()
+
+                var filterLogSeverity = logLevel switch
                 {
-                    SType = StructureType.DebugReportCallbackCreateInfoExt,
-                    Flags = flags,
-                    PfnCallback = new PfnDebugReportCallbackEXT(DebugReport)
+                    GraphicsDebugLevel.Error => DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityErrorBitExt,
+                    GraphicsDebugLevel.Slowdowns => DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityErrorBitExt |
+                                                    DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityWarningBitExt,
+                    GraphicsDebugLevel.All => DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityInfoBitExt |
+                                              DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityWarningBitExt |
+                                              DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityVerboseBitExt |
+                                              DebugUtilsMessageSeverityFlagsEXT.DebugUtilsMessageSeverityErrorBitExt,
+                    _ => throw new ArgumentException($"Invalid log level \"{logLevel}\".")
                 };
 
-                debugReport.CreateDebugReportCallback(instance, in debugReportCallbackCreateInfo, null, out debugReportCallback).ThrowOnError();
+                var debugUtilsMessengerCreateInfo = new DebugUtilsMessengerCreateInfoEXT()
+                {
+                    SType = StructureType.DebugUtilsMessengerCreateInfoExt,
+                    MessageType = filterLogType,
+                    MessageSeverity = filterLogSeverity,
+                    PfnUserCallback = new PfnDebugUtilsMessengerCallbackEXT(DebugMessenger)
+                };
+
+                debugUtils.CreateDebugUtilsMessenger(instance, in debugUtilsMessengerCreateInfo, null, out debugUtilsMessenger).ThrowOnError();
             }
             else
             {
-                debugReportCallback = default;
+                debugUtilsMessenger = default;
             }
         }
     }

Ryujinx.Graphics.Vulkan/VulkanRenderer.cs

@@ -33,7 +33,7 @@ namespace Ryujinx.Graphics.Vulkan
         internal KhrPushDescriptor PushDescriptorApi { get; private set; }
         internal ExtTransformFeedback TransformFeedbackApi { get; private set; }
         internal KhrDrawIndirectCount DrawIndirectCountApi { get; private set; }
-        internal ExtDebugReport DebugReportApi { get; private set; }
+        internal ExtDebugUtils DebugUtilsApi { get; private set; }
 
         internal uint QueueFamilyIndex { get; private set; }
         internal Queue Queue { get; private set; }
@@ -57,7 +57,7 @@ namespace Ryujinx.Graphics.Vulkan
         private SyncManager _syncManager;
 
         private PipelineFull _pipeline;
-        private DebugReportCallbackEXT _debugReportCallback;
+        private DebugUtilsMessengerEXT _debugUtilsMessenger;
 
         internal HelperShader HelperShader { get; private set; }
         internal PipelineFull PipelineInternal => _pipeline;
@@ -237,9 +237,9 @@ namespace Ryujinx.Graphics.Vulkan
 
             Api = api;
 
-            _instance = VulkanInitialization.CreateInstance(api, logLevel, _getRequiredExtensions(), out ExtDebugReport debugReport, out _debugReportCallback);
+            _instance = VulkanInitialization.CreateInstance(api, logLevel, _getRequiredExtensions(), out ExtDebugUtils debugUtils, out _debugUtilsMessenger);
 
-            DebugReportApi = debugReport;
+            DebugUtilsApi = debugUtils;
 
             if (api.TryGetInstanceExtension(_instance, out KhrSurface surfaceApi))
             {
@@ -310,7 +310,7 @@ namespace Ryujinx.Graphics.Vulkan
         internal TextureView CreateTextureView(TextureCreateInfo info, float scale)
         {
             // This should be disposed when all views are destroyed.
-            using var storage = CreateTextureStorage(info, scale);
+            var storage = CreateTextureStorage(info, scale);
             return storage.CreateView(info, 0, 0);
         }
 
@@ -584,9 +584,9 @@ namespace Ryujinx.Graphics.Vulkan
 
             MemoryAllocator.Dispose();
 
-            if (_debugReportCallback.Handle != 0)
+            if (_debugUtilsMessenger.Handle != 0)
             {
-                DebugReportApi.DestroyDebugReportCallback(_instance, _debugReportCallback, null);
+                DebugUtilsApi.DestroyDebugUtilsMessenger(_instance, _debugUtilsMessenger, null);
             }
 
             foreach (var shader in Shaders)

Ryujinx.HLE/HOS/Diagnostics/Demangler/Demangler.cs

@@ -918,7 +918,7 @@ namespace Ryujinx.HLE.HOS.Diagnostics.Demangler
                 return -1;
             }
 
-            return int.Parse(part.Substring(0, numberLength));
+            return int.Parse(part.AsSpan(0, numberLength));
         }
 
         private string ParseNumber(bool isSigned = false)

Ryujinx.HLE/HOS/Kernel/Memory/KPageTableBase.cs

@@ -2540,11 +2540,10 @@ namespace Ryujinx.HLE.HOS.Kernel.Memory
 
                 for (int attempt = 0; attempt < 8; attempt++)
                 {
-                    address = BitUtils.AlignDown(regionStart + GetRandomValue(0, aslrMaxOffset) * (ulong)alignment, alignment);
+                    ulong aslrAddress = BitUtils.AlignDown(regionStart + GetRandomValue(0, aslrMaxOffset) * (ulong)alignment, alignment);
+                    ulong aslrEndAddr = aslrAddress + totalNeededSize;
 
-                    ulong endAddr = address + totalNeededSize;
-
-                    KMemoryInfo info = _blockManager.FindBlock(address).GetInfo();
+                    KMemoryInfo info = _blockManager.FindBlock(aslrAddress).GetInfo();
 
                     if (info.State != MemoryState.Unmapped)
                     {
@@ -2554,11 +2553,12 @@ namespace Ryujinx.HLE.HOS.Kernel.Memory
                     ulong currBaseAddr = info.Address + reservedPagesCount * PageSize;
                     ulong currEndAddr = info.Address + info.Size;
 
-                    if (address >= regionStart &&
-                        address >= currBaseAddr &&
-                        endAddr - 1 <= regionEndAddr - 1 &&
-                        endAddr - 1 <= currEndAddr - 1)
+                    if (aslrAddress >= regionStart &&
+                        aslrAddress >= currBaseAddr &&
+                        aslrEndAddr - 1 <= regionEndAddr - 1 &&
+                        aslrEndAddr - 1 <= currEndAddr - 1)
                     {
+                        address = aslrAddress;
                         break;
                     }
                 }
@@ -2603,7 +2603,7 @@ namespace Ryujinx.HLE.HOS.Kernel.Memory
 
             ulong totalNeededSize = reservedSize + neededPagesCount * PageSize;
 
-            ulong regionEndAddr = regionStart + regionPagesCount * PageSize;
+            ulong regionEndAddr = (regionStart + regionPagesCount * PageSize) - 1;
 
             KMemoryBlock currBlock = _blockManager.FindBlock(regionStart);
 

Ryujinx.HLE/HOS/Services/Am/AppletAE/AllSystemAppletProxiesService/SystemAppletProxy/ISelfController.cs

@@ -32,6 +32,7 @@ namespace Ryujinx.HLE.HOS.Services.Am.AppletAE.AllSystemAppletProxiesService.Sys
         private bool _handlesRequestToDisplay            = false;
         private bool _autoSleepDisabled                  = false;
         private bool _albumImageTakenNotificationEnabled = false;
+        private bool _recordVolumeMuted = false;
 
         private uint _screenShotImageOrientation = 0;
         private uint _idleTimeDetectionExtension = 0;
@@ -389,5 +390,18 @@ namespace Ryujinx.HLE.HOS.Services.Am.AppletAE.AllSystemAppletProxiesService.Sys
 
             return ResultCode.Success;
         }
+
+        [CommandHipc(130)] // 13.0.0+
+        // SetRecordVolumeMuted(b8)
+        public ResultCode SetRecordVolumeMuted(ServiceCtx context)
+        {
+            bool recordVolumeMuted = context.RequestData.ReadBoolean();
+
+            Logger.Stub?.PrintStub(LogClass.ServiceAm, new { recordVolumeMuted });
+
+            _recordVolumeMuted = recordVolumeMuted;
+
+            return ResultCode.Success;
+        }
     }
-}
+}

Ryujinx.HLE/HOS/Services/Fs/IFileSystemProxy.cs

@@ -12,6 +12,7 @@ using LibHac.Tools.FsSystem.NcaUtils;
 using Ryujinx.Common;
 using Ryujinx.Common.Logging;
 using Ryujinx.HLE.HOS.Services.Fs.FileSystemProxy;
+using System;
 using System.IO;
 
 using static Ryujinx.HLE.Utilities.StringUtils;
@@ -787,6 +788,26 @@ namespace Ryujinx.HLE.HOS.Services.Fs
             return ResultCode.Success;
         }
 
+        [CommandHipc(205)]
+        // OpenDataStorageWithProgramIndex(u8 program_index) -> object<nn::fssrv::sf::IStorage>
+        public ResultCode OpenDataStorageWithProgramIndex(ServiceCtx context)
+        {
+            byte programIndex = context.RequestData.ReadByte();
+
+            if ((context.Device.Application.TitleId & 0xf) != programIndex)
+            {
+                throw new NotImplementedException($"Accessing storage from other programs is not supported (program index = {programIndex}).");
+            }
+
+            var storage = context.Device.FileSystem.RomFs.AsStorage(true);
+            using var sharedStorage = new SharedRef<LibHac.Fs.IStorage>(storage);
+            using var sfStorage = new SharedRef<IStorage>(new StorageInterfaceAdapter(ref sharedStorage.Ref()));
+
+            MakeObject(context, new FileSystemProxy.IStorage(ref sfStorage.Ref()));
+
+            return ResultCode.Success;
+        }
+
         [CommandHipc(400)]
         // OpenDataStorageByCurrentProcess() -> object<nn::fssrv::sf::IStorage> dataStorage
         public ResultCode OpenDeviceOperator(ServiceCtx context)

Ryujinx.HLE/HOS/Services/Hid/Irs/IIrSensorServer.cs

@@ -203,6 +203,18 @@ namespace Ryujinx.HLE.HOS.Services.Hid.Irs
             return ResultCode.Success;
         }
 
+        [CommandHipc(318)] // 4.0.0+
+        // StopImageProcessorAsync(nn::irsensor::IrCameraHandle, nn::applet::AppletResourceUserId, pid)
+        public ResultCode StopImageProcessorAsync(ServiceCtx context)
+        {
+            int  irCameraHandle       = context.RequestData.ReadInt32();
+            long appletResourceUserId = context.RequestData.ReadInt64();
+
+            Logger.Stub?.PrintStub(LogClass.ServiceIrs, new { appletResourceUserId, irCameraHandle });
+
+            return ResultCode.Success;
+        }
+
         [CommandHipc(319)] // 4.0.0+
         // ActivateIrsensorWithFunctionLevel(nn::applet::AppletResourceUserId, nn::irsensor::PackedFunctionLevel, pid)
         public ResultCode ActivateIrsensorWithFunctionLevel(ServiceCtx context)
@@ -225,4 +237,4 @@ namespace Ryujinx.HLE.HOS.Services.Hid.Irs
             return ResultCode.Success;
         }
     }
-}
+}

Ryujinx.HLE/HOS/Services/Nfc/NfcManager/INfc.cs

@@ -5,22 +5,48 @@ namespace Ryujinx.HLE.HOS.Services.Nfc.NfcManager
     class INfc : IpcService
     {
         private NfcPermissionLevel _permissionLevel;
+        private State _state;
 
         public INfc(NfcPermissionLevel permissionLevel)
         {
             _permissionLevel = permissionLevel;
+            _state = State.NonInitialized;
         }
 
         [CommandHipc(0)]
         [CommandHipc(400)] // 4.0.0+
-        // Initialize()
+        // Initialize(u64, u64, pid, buffer<unknown, 5>)
         public ResultCode Initialize(ServiceCtx context)
         {
+            _state = State.Initialized;
+
             Logger.Stub?.PrintStub(LogClass.ServiceNfc, new { _permissionLevel });
 
             return ResultCode.Success;
         }
 
+        [CommandHipc(1)]
+        [CommandHipc(401)] // 4.0.0+
+        // Finalize()
+        public ResultCode Finalize(ServiceCtx context)
+        {
+            _state = State.NonInitialized;
+
+            Logger.Stub?.PrintStub(LogClass.ServiceNfc, new { _permissionLevel });
+
+            return ResultCode.Success;
+        }
+
+        [CommandHipc(2)]
+        [CommandHipc(402)] // 4.0.0+
+        // GetState() -> u32
+        public ResultCode GetState(ServiceCtx context) 
+        {
+            context.ResponseData.Write((int)_state);
+
+            return ResultCode.Success;
+        }
+
         [CommandHipc(3)]
         [CommandHipc(403)] // 4.0.0+
         // IsNfcEnabled() -> b8

Ryujinx.HLE/HOS/Services/Nfc/NfcManager/Types/State.cs

@@ -0,0 +1,8 @@
+namespace Ryujinx.HLE.HOS.Services.Nfc.NfcManager
+{
+    enum State
+    {
+        NonInitialized,
+        Initialized
+    }
+}

Ryujinx.HLE/HOS/Services/Nfc/Nfp/NfpManager/INfp.cs

@@ -816,11 +816,11 @@ namespace Ryujinx.HLE.HOS.Services.Nfc.Nfp
                                 Reserved = new Array57<byte>()
                             };
 
-                            modelInfo.CharacterId      = BinaryPrimitives.ReverseEndianness(ushort.Parse(context.Device.System.NfpDevices[i].AmiiboId.Substring(0, 4), NumberStyles.HexNumber));
-                            modelInfo.CharacterVariant = byte.Parse(context.Device.System.NfpDevices[i].AmiiboId.Substring(4, 2), NumberStyles.HexNumber);
-                            modelInfo.Series           = byte.Parse(context.Device.System.NfpDevices[i].AmiiboId.Substring(12, 2), NumberStyles.HexNumber);
-                            modelInfo.ModelNumber      = ushort.Parse(context.Device.System.NfpDevices[i].AmiiboId.Substring(8, 4), NumberStyles.HexNumber);
-                            modelInfo.Type             = byte.Parse(context.Device.System.NfpDevices[i].AmiiboId.Substring(6, 2), NumberStyles.HexNumber);
+                            modelInfo.CharacterId      = BinaryPrimitives.ReverseEndianness(ushort.Parse(context.Device.System.NfpDevices[i].AmiiboId.AsSpan(0, 4), NumberStyles.HexNumber));
+                            modelInfo.CharacterVariant = byte.Parse(context.Device.System.NfpDevices[i].AmiiboId.AsSpan(4, 2), NumberStyles.HexNumber);
+                            modelInfo.Series           = byte.Parse(context.Device.System.NfpDevices[i].AmiiboId.AsSpan(12, 2), NumberStyles.HexNumber);
+                            modelInfo.ModelNumber      = ushort.Parse(context.Device.System.NfpDevices[i].AmiiboId.AsSpan(8, 4), NumberStyles.HexNumber);
+                            modelInfo.Type             = byte.Parse(context.Device.System.NfpDevices[i].AmiiboId.AsSpan(6, 2), NumberStyles.HexNumber);
 
                             context.Memory.Write(outputPosition, modelInfo);
 

Ryujinx.HLE/HOS/Services/Ssl/SslService/ISslConnection.cs

@@ -349,7 +349,11 @@ namespace Ryujinx.HLE.HOS.Services.Ssl.SslService
         // GetSessionCacheMode() -> nn::ssl::sf::SessionCacheMode
         public ResultCode GetSessionCacheMode(ServiceCtx context)
         {
-            throw new ServiceNotImplementedException(this, context);
+            context.ResponseData.Write((uint)_sessionCacheMode);
+
+            Logger.Stub?.PrintStub(LogClass.ServiceSsl, new { _sessionCacheMode });
+
+            return ResultCode.Success;
         }
 
         [CommandHipc(19)]

Ryujinx.HLE/HOS/Tamper/Conditions/InputMask.cs

@@ -13,7 +13,7 @@ namespace Ryujinx.HLE.HOS.Tamper.Conditions
 
         public bool Evaluate()
         {
-            return (_input.Value & _mask) != 0;
+            return (_input.Value & _mask) == _mask;
         }
     }
 }

Ryujinx.HLE/HOS/Tamper/InstructionHelper.cs

@@ -122,9 +122,8 @@ namespace Ryujinx.HLE.HOS.Tamper
                 for (int nybbleIndex = 0; nybbleIndex < wordSize; nybbleIndex++)
                 {
                     int index = wordIndex * wordSize + nybbleIndex;
-                    string byteData = word.Substring(nybbleIndex, 1);
 
-                    instruction[index] = byte.Parse(byteData, NumberStyles.HexNumber, CultureInfo.InvariantCulture);
+                    instruction[index] = byte.Parse(word.AsSpan(nybbleIndex, 1), NumberStyles.HexNumber, CultureInfo.InvariantCulture);
                 }
             }
 

Ryujinx.HLE/Loaders/Mods/IPSwitchPatcher.cs

@@ -132,7 +132,7 @@ namespace Ryujinx.HLE.Loaders.Mods
         {
             if (str[0] == '0' && (str[1] == 'x' || str[1] == 'X'))
             {
-                return int.TryParse(str.Substring(2), System.Globalization.NumberStyles.HexNumber, null, out value);
+                return int.TryParse(str.AsSpan(2), System.Globalization.NumberStyles.HexNumber, null, out value);
             }
             else
             {

Ryujinx.HLE/Ryujinx.HLE.csproj

@@ -26,7 +26,7 @@
     <PackageReference Include="MsgPack.Cli" Version="1.0.1" />
     <PackageReference Include="SixLabors.ImageSharp" Version="1.0.4" />
     <PackageReference Include="SixLabors.ImageSharp.Drawing" Version="1.0.0-beta11" />
-    <PackageReference Include="System.IdentityModel.Tokens.Jwt" Version="6.15.0" />
+    <PackageReference Include="System.IdentityModel.Tokens.Jwt" Version="6.25.0" />
   </ItemGroup>
 
   <!-- Due to Concentus. -->

Ryujinx.HLE/Utilities/StringUtils.cs

@@ -54,7 +54,7 @@ namespace Ryujinx.HLE.Utilities
 
             for (int index = 0; index < bytesInHex; index++)
             {
-                output[index] = byte.Parse(hexString.Substring(index * 2, 2), NumberStyles.HexNumber);
+                output[index] = byte.Parse(hexString.AsSpan(index * 2, 2), NumberStyles.HexNumber);
             }
 
             return output;

Ryujinx.Input/HLE/NpadManager.cs

@@ -51,7 +51,16 @@ namespace Ryujinx.Input.HLE
         {
             lock (_lock)
             {
-                _device.Hid.RefreshInputConfig(_inputConfig);
+                List<InputConfig> validInputs = new List<InputConfig>();
+                foreach (var inputConfigEntry in _inputConfig)
+                {
+                    if (_controllers[(int)inputConfigEntry.PlayerIndex] != null)
+                    {
+                        validInputs.Add(inputConfigEntry);
+                    }
+                }
+
+                _device.Hid.RefreshInputConfig(validInputs);
             }
         }
 
@@ -103,6 +112,8 @@ namespace Ryujinx.Input.HLE
                     _controllers[i] = null;
                 }
 
+                List<InputConfig> validInputs = new List<InputConfig>();
+
                 foreach (InputConfig inputConfigEntry in inputConfig)
                 {
                     NpadController controller = new NpadController(_cemuHookClient);
@@ -116,6 +127,7 @@ namespace Ryujinx.Input.HLE
                     else
                     {
                         _controllers[(int)inputConfigEntry.PlayerIndex] = controller;
+                        validInputs.Add(inputConfigEntry);
                     }
                 }
 
@@ -123,7 +135,7 @@ namespace Ryujinx.Input.HLE
                 _enableKeyboard = enableKeyboard;
                 _enableMouse    = enableMouse;
 
-                _device.Hid.RefreshInputConfig(inputConfig);
+                _device.Hid.RefreshInputConfig(validInputs);
             }
         }
 

Ryujinx.Memory.Tests/Tests.cs

@@ -92,5 +92,31 @@ namespace Ryujinx.Memory.Tests
                 }
             }
         }
+
+        [Test]
+        public void Test_AliasMapLeak()
+        {
+            if (OperatingSystem.IsMacOS())
+            {
+                // Memory aliasing tests fail on CI at the moment.
+                return;
+            }
+
+            ulong pageSize = 4096;
+            ulong size = 100000 * pageSize; // The mappings limit on Linux is usually around 65K, so let's make sure we are above that.
+
+            using MemoryBlock backing = new MemoryBlock(pageSize, MemoryAllocationFlags.Mirrorable);
+            using MemoryBlock toAlias = new MemoryBlock(size, MemoryAllocationFlags.Reserve | MemoryAllocationFlags.ViewCompatible);
+
+            for (ulong offset = 0; offset < size; offset += pageSize)
+            {
+                toAlias.MapView(backing, 0, offset, pageSize);
+
+                toAlias.Write(offset, 0xbadc0de);
+                Assert.AreEqual(0xbadc0de, backing.Read<int>(0));
+
+                toAlias.UnmapView(backing, offset, pageSize);
+            }
+        }
     }
 }

Ryujinx.Memory/MemoryManagementUnix.cs

@@ -177,7 +177,7 @@ namespace Ryujinx.Memory
 
         public static void UnmapView(IntPtr location, ulong size)
         {
-            mmap(location, size, MmapProts.PROT_NONE, MmapFlags.MAP_FIXED, -1, 0);
+            mmap(location, size, MmapProts.PROT_NONE, MmapFlags.MAP_FIXED | MmapFlags.MAP_PRIVATE | MmapFlags.MAP_ANONYMOUS | MmapFlags.MAP_NORESERVE, -1, 0);
         }
     }
 }

Ryujinx.Memory/Tracking/BitMap.cs

@@ -0,0 +1,199 @@
+using System.Runtime.CompilerServices;
+
+namespace Ryujinx.Memory.Tracking
+{
+    /// <summary>
+    /// A bitmap that can check or set large ranges of true/false values at once.
+    /// </summary>
+    struct BitMap
+    {
+        public const int IntSize = 64;
+
+        private const int IntShift = 6;
+        private const int IntMask = IntSize - 1;
+
+        /// <summary>
+        /// Masks representing the bitmap. Least significant bit first, 64-bits per mask.
+        /// </summary>
+        public readonly long[] Masks;
+
+        /// <summary>
+        /// Create a new bitmap.
+        /// </summary>
+        /// <param name="count">The number of bits to reserve</param>
+        public BitMap(int count)
+        {
+            Masks = new long[(count + IntMask) / IntSize];
+        }
+
+        /// <summary>
+        /// Check if any bit in the bitmap is set.
+        /// </summary>
+        /// <returns>True if any bits are set, false otherwise</returns>
+        public bool AnySet()
+        {
+            for (int i = 0; i < Masks.Length; i++)
+            {
+                if (Masks[i] != 0)
+                {
+                    return true;
+                }
+            }
+
+            return false;
+        }
+
+        /// <summary>
+        /// Check if a bit in the bitmap is set.
+        /// </summary>
+        /// <param name="bit">The bit index to check</param>
+        /// <returns>True if the bit is set, false otherwise</returns>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public bool IsSet(int bit)
+        {
+            int wordIndex = bit >> IntShift;
+            int wordBit = bit & IntMask;
+
+            long wordMask = 1L << wordBit;
+
+            return (Masks[wordIndex] & wordMask) != 0;
+        }
+
+        /// <summary>
+        /// Check if any bit in a range of bits in the bitmap are set. (inclusive)
+        /// </summary>
+        /// <param name="start">The first bit index to check</param>
+        /// <param name="end">The last bit index to check</param>
+        /// <returns>True if a bit is set, false otherwise</returns>
+        public bool IsSet(int start, int end)
+        {
+            if (start == end)
+            {
+                return IsSet(start);
+            }
+
+            int startIndex = start >> IntShift;
+            int startBit = start & IntMask;
+            long startMask = -1L << startBit;
+
+            int endIndex = end >> IntShift;
+            int endBit = end & IntMask;
+            long endMask = (long)(ulong.MaxValue >> (IntMask - endBit));
+
+            if (startIndex == endIndex)
+            {
+                return (Masks[startIndex] & startMask & endMask) != 0;
+            }
+
+            if ((Masks[startIndex] & startMask) != 0)
+            {
+                return true;
+            }
+
+            for (int i = startIndex + 1; i < endIndex; i++)
+            {
+                if (Masks[i] != 0)
+                {
+                    return true;
+                }
+            }
+
+            if ((Masks[endIndex] & endMask) != 0)
+            {
+                return true;
+            }
+
+            return false;
+        }
+
+        /// <summary>
+        /// Set a bit at a specific index to 1.
+        /// </summary>
+        /// <param name="bit">The bit index to set</param>
+        /// <returns>True if the bit is set, false if it was already set</returns>
+        public bool Set(int bit)
+        {
+            int wordIndex = bit >> IntShift;
+            int wordBit = bit & IntMask;
+
+            long wordMask = 1L << wordBit;
+
+            if ((Masks[wordIndex] & wordMask) != 0)
+            {
+                return false;
+            }
+
+            Masks[wordIndex] |= wordMask;
+
+            return true;
+        }
+
+        /// <summary>
+        /// Set a range of bits in the bitmap to 1.
+        /// </summary>
+        /// <param name="start">The first bit index to set</param>
+        /// <param name="end">The last bit index to set</param>
+        public void SetRange(int start, int end)
+        {
+            if (start == end)
+            {
+                Set(start);
+                return;
+            }
+
+            int startIndex = start >> IntShift;
+            int startBit = start & IntMask;
+            long startMask = -1L << startBit;
+
+            int endIndex = end >> IntShift;
+            int endBit = end & IntMask;
+            long endMask = (long)(ulong.MaxValue >> (IntMask - endBit));
+
+            if (startIndex == endIndex)
+            {
+                Masks[startIndex] |= startMask & endMask;
+            }
+            else
+            {
+                Masks[startIndex] |= startMask;
+
+                for (int i = startIndex + 1; i < endIndex; i++)
+                {
+                    Masks[i] |= -1;
+                }
+
+                Masks[endIndex] |= endMask;
+            }
+        }
+
+        /// <summary>
+        /// Clear a bit at a specific index to 0.
+        /// </summary>
+        /// <param name="bit">The bit index to clear</param>
+        /// <returns>True if the bit was set, false if it was not</returns>
+        public bool Clear(int bit)
+        {
+            int wordIndex = bit >> IntShift;
+            int wordBit = bit & IntMask;
+
+            long wordMask = 1L << wordBit;
+
+            bool wasSet = (Masks[wordIndex] & wordMask) != 0;
+
+            Masks[wordIndex] &= ~wordMask;
+
+            return wasSet;
+        }
+
+        /// <summary>
+        /// Clear the bitmap entirely, setting all bits to 0.
+        /// </summary>
+        public void Clear()
+        {
+            for (int i = 0; i < Masks.Length; i++)
+            {
+                Masks[i] = 0;
+            }
+        }
+    }
+}

Ryujinx.Memory/Tracking/ConcurrentBitmap.cs

@@ -0,0 +1,161 @@
+using System;
+using System.Threading;
+
+namespace Ryujinx.Memory.Tracking
+{
+    /// <summary>
+    /// A bitmap that can be safely modified from multiple threads.
+    /// </summary>
+    internal class ConcurrentBitmap
+    {
+        public const int IntSize = 64;
+
+        public const int IntShift = 6;
+        public const int IntMask = IntSize - 1;
+
+        /// <summary>
+        /// Masks representing the bitmap. Least significant bit first, 64-bits per mask.
+        /// </summary>
+        public readonly long[] Masks;
+
+        /// <summary>
+        /// Create a new multithreaded bitmap.
+        /// </summary>
+        /// <param name="count">The number of bits to reserve</param>
+        /// <param name="set">Whether the bits should be initially set or not</param>
+        public ConcurrentBitmap(int count, bool set)
+        {
+            Masks = new long[(count + IntMask) / IntSize];
+
+            if (set)
+            {
+                Array.Fill(Masks, -1L);
+            }
+        }
+
+        /// <summary>
+        /// Check if any bit in the bitmap is set.
+        /// </summary>
+        /// <returns>True if any bits are set, false otherwise</returns>
+        public bool AnySet()
+        {
+            for (int i = 0; i < Masks.Length; i++)
+            {
+                if (Volatile.Read(ref Masks[i]) != 0)
+                {
+                    return true;
+                }
+            }
+
+            return false;
+        }
+
+        /// <summary>
+        /// Check if a bit in the bitmap is set.
+        /// </summary>
+        /// <param name="bit">The bit index to check</param>
+        /// <returns>True if the bit is set, false otherwise</returns>
+        public bool IsSet(int bit)
+        {
+            int wordIndex = bit >> IntShift;
+            int wordBit = bit & IntMask;
+
+            long wordMask = 1L << wordBit;
+
+            return (Volatile.Read(ref Masks[wordIndex]) & wordMask) != 0;
+        }
+
+        /// <summary>
+        /// Check if any bit in a range of bits in the bitmap are set. (inclusive)
+        /// </summary>
+        /// <param name="start">The first bit index to check</param>
+        /// <param name="end">The last bit index to check</param>
+        /// <returns>True if a bit is set, false otherwise</returns>
+        public bool IsSet(int start, int end)
+        {
+            if (start == end)
+            {
+                return IsSet(start);
+            }
+
+            int startIndex = start >> IntShift;
+            int startBit = start & IntMask;
+            long startMask = -1L << startBit;
+
+            int endIndex = end >> IntShift;
+            int endBit = end & IntMask;
+            long endMask = (long)(ulong.MaxValue >> (IntMask - endBit));
+
+            long startValue = Volatile.Read(ref Masks[startIndex]);
+
+            if (startIndex == endIndex)
+            {
+                return (startValue & startMask & endMask) != 0;
+            }
+
+            if ((startValue & startMask) != 0)
+            {
+                return true;
+            }
+
+            for (int i = startIndex + 1; i < endIndex; i++)
+            {
+                if (Volatile.Read(ref Masks[i]) != 0)
+                {
+                    return true;
+                }
+            }
+
+            long endValue = Volatile.Read(ref Masks[endIndex]);
+
+            if ((endValue & endMask) != 0)
+            {
+                return true;
+            }
+
+            return false;
+        }
+
+        /// <summary>
+        /// Set a bit at a specific index to either true or false.
+        /// </summary>
+        /// <param name="bit">The bit index to set</param>
+        /// <param name="value">Whether the bit should be set or not</param>
+        public void Set(int bit, bool value)
+        {
+            int wordIndex = bit >> IntShift;
+            int wordBit = bit & IntMask;
+
+            long wordMask = 1L << wordBit;
+
+            long existing;
+            long newValue;
+
+            do
+            {
+                existing = Volatile.Read(ref Masks[wordIndex]);
+
+                if (value)
+                {
+                    newValue = existing | wordMask;
+                }
+                else
+                {
+                    newValue = existing & ~wordMask;
+                }
+            }
+            while (Interlocked.CompareExchange(ref Masks[wordIndex], newValue, existing) != existing);
+        }
+
+        /// <summary>
+        /// Clear the bitmap entirely, setting all bits to 0.
+        /// </summary>
+        public void Clear()
+        {
+            for (int i = 0; i < Masks.Length; i++)
+            {
+                Volatile.Write(ref Masks[i], 0);
+            }
+        }
+    }
+}

Ryujinx.Memory/Tracking/MemoryTracking.cs

@@ -176,6 +176,26 @@ namespace Ryujinx.Memory.Tracking
             }
         }
 
+        /// <summary>
+        /// Obtains a memory tracking handle for the given virtual region. This should be disposed when finished with.
+        /// </summary>
+        /// <param name="address">CPU virtual address of the region</param>
+        /// <param name="size">Size of the region</param>
+        /// <param name="bitmap">The bitmap owning the dirty flag for this handle</param>
+        /// <param name="bit">The bit of this handle within the dirty flag</param>
+        /// <returns>The memory tracking handle</returns>
+        internal RegionHandle BeginTrackingBitmap(ulong address, ulong size, ConcurrentBitmap bitmap, int bit)
+        {
+            (address, size) = PageAlign(address, size);
+
+            lock (TrackingLock)
+            {
+                RegionHandle handle = new RegionHandle(this, address, size, bitmap, bit, _memoryManager.IsRangeMapped(address, size));
+
+                return handle;
+            }
+        }
+
         /// <summary>
         /// Signal that a virtual memory event happened at the given location (one byte).
         /// </summary>

Ryujinx.Memory/Tracking/MultiRegionHandle.cs

@@ -1,11 +1,15 @@
 using System;
 using System.Collections.Generic;
+using System.Numerics;
+using System.Runtime.CompilerServices;
+using System.Threading;
 
 namespace Ryujinx.Memory.Tracking
 {
     /// <summary>
     /// A region handle that tracks a large region using many smaller handles, to provide
-    /// granular tracking that can be used to track partial updates.
+    /// granular tracking that can be used to track partial updates. Backed by a bitmap
+    /// to improve performance when scanning large regions.
     /// </summary>
     public class MultiRegionHandle : IMultiRegionHandle
     {
@@ -17,6 +21,12 @@ namespace Ryujinx.Memory.Tracking
         private readonly ulong Granularity;
         private readonly ulong Size;
 
+        private ConcurrentBitmap _dirtyBitmap;
+
+        private int _sequenceNumber;
+        private BitMap _sequenceNumberBitmap;
+        private int _uncheckedHandles;
+
         public bool Dirty { get; private set; } = true;
 
         internal MultiRegionHandle(MemoryTracking tracking, ulong address, ulong size, IEnumerable<IRegionHandle> handles, ulong granularity)
@@ -24,6 +34,9 @@ namespace Ryujinx.Memory.Tracking
             _handles = new RegionHandle[size / granularity];
             Granularity = granularity;
 
+            _dirtyBitmap = new ConcurrentBitmap(_handles.Length, true);
+            _sequenceNumberBitmap = new BitMap(_handles.Length);
+
             int i = 0;
 
             if (handles != null)
@@ -40,37 +53,43 @@ namespace Ryujinx.Memory.Tracking
                     // Fill any gap left before this handle.
                     while (i < startIndex)
                     {
-                        RegionHandle fillHandle = tracking.BeginTracking(address + (ulong)i * granularity, granularity);
+                        RegionHandle fillHandle = tracking.BeginTrackingBitmap(address + (ulong)i * granularity, granularity, _dirtyBitmap, i);
                         fillHandle.Parent = this;
                         _handles[i++] = fillHandle;
                     }
 
-                    if (handle.Size == granularity)
+                    lock (tracking.TrackingLock)
                     {
-                        handle.Parent = this;
-                        _handles[i++] = handle;
-                    }
-                    else
-                    {
-                        int endIndex = (int)((handle.EndAddress - address) / granularity);
-
-                        while (i < endIndex)
+                        if (handle is RegionHandle bitHandle && handle.Size == granularity)
                         {
-                            RegionHandle splitHandle = tracking.BeginTracking(address + (ulong)i * granularity, granularity);
-                            splitHandle.Parent = this;
+                            handle.Parent = this;
 
-                            splitHandle.Reprotect(handle.Dirty);
+                            bitHandle.ReplaceBitmap(_dirtyBitmap, i);
 
-                            RegionSignal signal = handle.PreAction;
-                            if (signal != null)
+                            _handles[i++] = bitHandle;
+                        }
+                        else
+                        {
+                            int endIndex = (int)((handle.EndAddress - address) / granularity);
+
+                            while (i < endIndex)
                             {
-                                splitHandle.RegisterAction(signal);
+                                RegionHandle splitHandle = tracking.BeginTrackingBitmap(address + (ulong)i * granularity, granularity, _dirtyBitmap, i);
+                                splitHandle.Parent = this;
+
+                                splitHandle.Reprotect(handle.Dirty);
+
+                                RegionSignal signal = handle.PreAction;
+                                if (signal != null)
+                                {
+                                    splitHandle.RegisterAction(signal);
+                                }
+
+                                _handles[i++] = splitHandle;
                             }
 
-                            _handles[i++] = splitHandle;
+                            handle.Dispose();
                         }
-
-                        handle.Dispose();
                     }
                 }
             }
@@ -78,15 +97,27 @@ namespace Ryujinx.Memory.Tracking
             // Fill any remaining space with new handles.
             while (i < _handles.Length)
             {
-                RegionHandle handle = tracking.BeginTracking(address + (ulong)i * granularity, granularity);
+                RegionHandle handle = tracking.BeginTrackingBitmap(address + (ulong)i * granularity, granularity, _dirtyBitmap, i);
                 handle.Parent = this;
                 _handles[i++] = handle;
             }
 
+            _uncheckedHandles = _handles.Length;
+
             Address = address;
             Size = size;
         }
 
+        public void SignalWrite()
+        {
+            Dirty = true;
+        }
+
+        public IEnumerable<RegionHandle> GetHandles()
+        {
+            return _handles;
+        }
+
         public void ForceDirty(ulong address, ulong size)
         {
             Dirty = true;
@@ -96,21 +127,15 @@ namespace Ryujinx.Memory.Tracking
 
             for (int i = startHandle; i <= lastHandle; i++)
             {
-                _handles[i].SequenceNumber--;
+                if (_sequenceNumberBitmap.Clear(i))
+                {
+                    _uncheckedHandles++;
+                }
+
                 _handles[i].ForceDirty();
             }
         }
 
-        public IEnumerable<RegionHandle> GetHandles()
-        {
-            return _handles;
-        }
-
-        public void SignalWrite()
-        {
-            Dirty = true;
-        }
-
         public void QueryModified(Action<ulong, ulong> modifiedAction)
         {
             if (!Dirty)
@@ -123,33 +148,95 @@ namespace Ryujinx.Memory.Tracking
             QueryModified(Address, Size, modifiedAction);
         }
 
-        public void QueryModified(ulong address, ulong size, Action<ulong, ulong> modifiedAction)
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        private void ParseDirtyBits(long dirtyBits, ref int baseBit, ref int prevHandle, ref ulong rgStart, ref ulong rgSize, Action<ulong, ulong> modifiedAction)
         {
-            int startHandle = (int)((address - Address) / Granularity);
-            int lastHandle = (int)((address + (size - 1) - Address) / Granularity);
-
-            ulong rgStart = _handles[startHandle].Address;
-            ulong rgSize = 0;
-
-            for (int i = startHandle; i <= lastHandle; i++)
+            while (dirtyBits != 0)
             {
-                RegionHandle handle = _handles[i];
+                int bit = BitOperations.TrailingZeroCount(dirtyBits);
+
+                dirtyBits &= ~(1L << bit);
+
+                int handleIndex = baseBit + bit;
+
+                RegionHandle handle = _handles[handleIndex];
+
+                if (handleIndex != prevHandle + 1)
+                {
+                    // Submit handles scanned until the gap as dirty
+                    if (rgSize != 0)
+                    {
+                        modifiedAction(rgStart, rgSize);
+                        rgSize = 0;
+                    }
+                    rgStart = handle.Address;
+                }
 
                 if (handle.Dirty)
                 {
                     rgSize += handle.Size;
                     handle.Reprotect();
                 }
-                else
+
+                prevHandle = handleIndex;
+            }
+
+            baseBit += ConcurrentBitmap.IntSize;
+        }
+
+        public void QueryModified(ulong address, ulong size, Action<ulong, ulong> modifiedAction)
+        {
+            int startHandle = (int)((address - Address) / Granularity);
+            int lastHandle = (int)((address + (size - 1) - Address) / Granularity);
+
+            ulong rgStart = _handles[startHandle].Address;
+
+            if (startHandle == lastHandle)
+            {
+                RegionHandle handle = _handles[startHandle];
+
+                if (handle.Dirty)
                 {
-                    // Submit the region scanned so far as dirty
-                    if (rgSize != 0)
-                    {
-                        modifiedAction(rgStart, rgSize);
-                        rgSize = 0;
-                    }
-                    rgStart = handle.EndAddress;
+                    handle.Reprotect();
+                    modifiedAction(rgStart, handle.Size);
                 }
+
+                return;
+            }
+
+            ulong rgSize = 0;
+
+            long[] masks = _dirtyBitmap.Masks;
+
+            int startIndex = startHandle >> ConcurrentBitmap.IntShift;
+            int startBit = startHandle & ConcurrentBitmap.IntMask;
+            long startMask = -1L << startBit;
+
+            int endIndex = lastHandle >> ConcurrentBitmap.IntShift;
+            int endBit = lastHandle & ConcurrentBitmap.IntMask;
+            long endMask = (long)(ulong.MaxValue >> (ConcurrentBitmap.IntMask - endBit));
+
+            long startValue = Volatile.Read(ref masks[startIndex]);
+
+            int baseBit = startIndex << ConcurrentBitmap.IntShift;
+            int prevHandle = startHandle - 1;
+
+            if (startIndex == endIndex)
+            {
+                ParseDirtyBits(startValue & startMask & endMask, ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
+            }
+            else
+            {
+                ParseDirtyBits(startValue & startMask, ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
+
+                for (int i = startIndex + 1; i < endIndex; i++)
+                {
+                    ParseDirtyBits(Volatile.Read(ref masks[i]), ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
+                }
+
+                long endValue = Volatile.Read(ref masks[endIndex]);
+
+                ParseDirtyBits(endValue & endMask, ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
             }
 
             if (rgSize != 0)
@@ -158,35 +245,120 @@ namespace Ryujinx.Memory.Tracking
             }
         }
 
-        public void QueryModified(ulong address, ulong size, Action<ulong, ulong> modifiedAction, int sequenceNumber)
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        private void ParseDirtyBits(long dirtyBits, long mask, int index, long[] seqMasks, ref int baseBit, ref int prevHandle, ref ulong rgStart, ref ulong rgSize, Action<ulong, ulong> modifiedAction)
         {
-            int startHandle = (int)((address - Address) / Granularity);
-            int lastHandle = (int)((address + (size - 1) - Address) / Granularity);
+            long seqMask = mask & ~seqMasks[index];
+            dirtyBits &= seqMask;
 
-            ulong rgStart = _handles[startHandle].Address;
-            ulong rgSize = 0;
-
-            for (int i = startHandle; i <= lastHandle; i++)
+            while (dirtyBits != 0)
             {
-                RegionHandle handle = _handles[i];
+                int bit = BitOperations.TrailingZeroCount(dirtyBits);
 
-                if (sequenceNumber != handle.SequenceNumber && handle.DirtyOrVolatile())
+                dirtyBits &= ~(1L << bit);
+
+                int handleIndex = baseBit + bit;
+
+                RegionHandle handle = _handles[handleIndex];
+
+                if (handleIndex != prevHandle + 1)
                 {
-                    rgSize += handle.Size;
-                    handle.Reprotect();
-                }
-                else
-                {
-                    // Submit the region scanned so far as dirty
+                    // Submit handles scanned until the gap as dirty
                     if (rgSize != 0)
                     {
                         modifiedAction(rgStart, rgSize);
                         rgSize = 0;
                     }
-                    rgStart = handle.EndAddress;
+                    rgStart = handle.Address;
                 }
 
-                handle.SequenceNumber = sequenceNumber;
+                rgSize += handle.Size;
+                handle.Reprotect();
+
+                prevHandle = handleIndex;
+            }
+
+            seqMasks[index] |= mask;
+            _uncheckedHandles -= BitOperations.PopCount((ulong)seqMask);
+
+            baseBit += ConcurrentBitmap.IntSize;
+        }
+
+        public void QueryModified(ulong address, ulong size, Action<ulong, ulong> modifiedAction, int sequenceNumber)
+        {
+            int startHandle = (int)((address - Address) / Granularity);
+            int lastHandle = (int)((address + (size - 1) - Address) / Granularity);
+
+            ulong rgStart = Address + (ulong)startHandle * Granularity;
+
+            if (sequenceNumber != _sequenceNumber)
+            {
+                if (_uncheckedHandles != _handles.Length)
+                {
+                    _sequenceNumberBitmap.Clear();
+                    _uncheckedHandles = _handles.Length;
+                }
+
+                _sequenceNumber = sequenceNumber;
+            }
+
+            if (startHandle == lastHandle)
+            {
+                var handle = _handles[startHandle];
+                if (_sequenceNumberBitmap.Set(startHandle))
+                {
+                    _uncheckedHandles--;
+
+                    if (handle.DirtyOrVolatile())
+                    {
+                        handle.Reprotect();
+
+                        modifiedAction(rgStart, handle.Size);
+                    }
+                }
+
+                return;
+            }
+
+            if (_uncheckedHandles == 0)
+            {
+                return;
+            }
+
+            ulong rgSize = 0;
+
+            long[] seqMasks = _sequenceNumberBitmap.Masks;
+            long[] masks = _dirtyBitmap.Masks;
+
+            int startIndex = startHandle >> ConcurrentBitmap.IntShift;
+            int startBit = startHandle & ConcurrentBitmap.IntMask;
+            long startMask = -1L << startBit;
+
+            int endIndex = lastHandle >> ConcurrentBitmap.IntShift;
+            int endBit = lastHandle & ConcurrentBitmap.IntMask;
+            long endMask = (long)(ulong.MaxValue >> (ConcurrentBitmap.IntMask - endBit));
+
+            long startValue = Volatile.Read(ref masks[startIndex]);
+
+            int baseBit = startIndex << ConcurrentBitmap.IntShift;
+            int prevHandle = startHandle - 1;
+
+            if (startIndex == endIndex)
+            {
+                ParseDirtyBits(startValue, startMask & endMask, startIndex, seqMasks, ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
+            }
+            else
+            {
+                ParseDirtyBits(startValue, startMask, startIndex, seqMasks, ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
+
+                for (int i = startIndex + 1; i < endIndex; i++)
+                {
+                    ParseDirtyBits(Volatile.Read(ref masks[i]), -1L, i, seqMasks, ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
+                }
+
+                long endValue = Volatile.Read(ref masks[endIndex]);
+
+                ParseDirtyBits(endValue, endMask, endIndex, seqMasks, ref baseBit, ref prevHandle, ref rgStart, ref rgSize, modifiedAction);
             }
 
             if (rgSize != 0)

Ryujinx.Memory/Tracking/RegionHandle.cs

@@ -1,5 +1,4 @@
-using Ryujinx.Memory.Range;
-using System;
+using System;
 using System.Collections.Generic;
 using System.Linq;
 using System.Threading;
@@ -10,7 +9,7 @@ namespace Ryujinx.Memory.Tracking
     /// A tracking handle for a given region of virtual memory. The Dirty flag is updated whenever any changes are made,
     /// and an action can be performed when the region is read to or written from.
     /// </summary>
-    public class RegionHandle : IRegionHandle, IRange
+    public class RegionHandle : IRegionHandle
     {
         /// <summary>
         /// If more than this number of checks have been performed on a dirty flag since its last reprotect,
@@ -23,7 +22,20 @@ namespace Ryujinx.Memory.Tracking
         /// </summary>
         private static int VolatileThreshold = 5;
 
-        public bool Dirty { get; private set; }
+        public bool Dirty
+        {
+            get
+            {
+                return Bitmap.IsSet(DirtyBit);
+            }
+            protected set
+            {
+                Bitmap.Set(DirtyBit, value);
+            }
+        }
+
+        internal int SequenceNumber { get; set; }
+
         public bool Unmapped { get; private set; }
 
         public ulong Address { get; }
@@ -31,7 +43,6 @@ namespace Ryujinx.Memory.Tracking
         public ulong EndAddress { get; }
 
         internal IMultiRegionHandle Parent { get; set; }
-        internal int SequenceNumber { get; set; }
 
         private event Action _onDirty;
 
@@ -68,17 +79,26 @@ namespace Ryujinx.Memory.Tracking
 
         internal RegionSignal PreAction => _preAction;
 
+        internal ConcurrentBitmap Bitmap;
+        internal int DirtyBit;
+
         /// <summary>
-        /// Create a new region handle. The handle is registered with the given tracking object,
+        /// Create a new bitmap backed region handle. The handle is registered with the given tracking object,
         /// and will be notified of any changes to the specified region.
         /// </summary>
         /// <param name="tracking">Tracking object for the target memory block</param>
         /// <param name="address">Virtual address of the region to track</param>
         /// <param name="size">Size of the region to track</param>
+        /// <param name="bitmap">The bitmap the dirty flag for this handle is stored in</param>
+        /// <param name="bit">The bit index representing the dirty flag for this handle</param>
         /// <param name="mapped">True if the region handle starts mapped</param>
-        internal RegionHandle(MemoryTracking tracking, ulong address, ulong size, bool mapped = true)
+        internal RegionHandle(MemoryTracking tracking, ulong address, ulong size, ConcurrentBitmap bitmap, int bit, bool mapped = true)
         {
+            Bitmap = bitmap;
+            DirtyBit = bit;
+
             Dirty = mapped;
+
             Unmapped = !mapped;
             Address = address;
             Size = size;
@@ -92,6 +112,54 @@ namespace Ryujinx.Memory.Tracking
             }
         }
 
+        /// <summary>
+        /// Create a new region handle. The handle is registered with the given tracking object,
+        /// and will be notified of any changes to the specified region.
+        /// </summary>
+        /// <param name="tracking">Tracking object for the target memory block</param>
+        /// <param name="address">Virtual address of the region to track</param>
+        /// <param name="size">Size of the region to track</param>
+        /// <param name="mapped">True if the region handle starts mapped</param>
+        internal RegionHandle(MemoryTracking tracking, ulong address, ulong size, bool mapped = true)
+        {
+            Bitmap = new ConcurrentBitmap(1, mapped);
+
+            Unmapped = !mapped;
+            Address = address;
+            Size = size;
+            EndAddress = address + size;
+
+            _tracking = tracking;
+            _regions = tracking.GetVirtualRegionsForHandle(address, size);
+            foreach (var region in _regions)
+            {
+                region.Handles.Add(this);
+            }
+        }
+
+        /// <summary>
+        /// Replace the bitmap and bit index used to track dirty state.
+        /// </summary>
+        /// <remarks>
+        /// The tracking lock should be held when this is called, to ensure neither bitmap is modified.
+        /// </remarks>
+        /// <param name="bitmap">The bitmap the dirty flag for this handle is stored in</param>
+        /// <param name="bit">The bit index representing the dirty flag for this handle</param>
+        internal void ReplaceBitmap(ConcurrentBitmap bitmap, int bit)
+        {
+            // Assumes the tracking lock is held, so nothing else can signal right now.
+
+            var oldBitmap = Bitmap;
+            var oldBit = DirtyBit;
+
+            bitmap.Set(bit, Dirty);
+
+            Bitmap = bitmap;
+            DirtyBit = bit;
+
+            Dirty |= oldBitmap.IsSet(oldBit);
+        }
+
         /// <summary>
         /// Clear the volatile state of this handle.
         /// </summary>
@@ -108,7 +176,7 @@ namespace Ryujinx.Memory.Tracking
         public bool DirtyOrVolatile()
         {
             _checkCount++;
-            return Dirty || _volatile;
+            return _volatile || Dirty;
         }
 
         /// <summary>
@@ -195,6 +263,8 @@ namespace Ryujinx.Memory.Tracking
         /// </summary>
         public void ForceDirty()
         {
+            _checkCount++;
+
             Dirty = true;
         }
 

Ryujinx.Tests.Unicorn/UnicornAArch32.cs

@@ -3,9 +3,10 @@ using System;
 
 namespace Ryujinx.Tests.Unicorn
 {
-    public class UnicornAArch32
+    public class UnicornAArch32 : IDisposable
     {
         internal readonly IntPtr uc;
+        private bool _isDisposed = false;
 
         public IndexedProperty<int, uint> R
         {
@@ -107,7 +108,22 @@ namespace Ryujinx.Tests.Unicorn
 
         ~UnicornAArch32()
         {
-            Interface.Checked(Native.Interface.uc_close(uc));
+            Dispose(false);
+        }
+
+        public void Dispose()
+        {
+            Dispose(true);
+            GC.SuppressFinalize(this);
+        }
+
+        protected virtual void Dispose(bool disposing)
+        {
+            if (!_isDisposed)
+            {
+                Interface.Checked(Native.Interface.uc_close(uc));
+                _isDisposed = true;
+            }
         }
 
         public void RunForCount(ulong count)

Ryujinx.Tests.Unicorn/UnicornAArch64.cs

@@ -3,9 +3,10 @@ using System;
 
 namespace Ryujinx.Tests.Unicorn
 {
-    public class UnicornAArch64
+    public class UnicornAArch64 : IDisposable
     {
         internal readonly IntPtr uc;
+        private bool _isDisposed = false;
 
         public IndexedProperty<int, ulong> X
         {
@@ -96,7 +97,22 @@ namespace Ryujinx.Tests.Unicorn
 
         ~UnicornAArch64()
         {
-            Interface.Checked(Native.Interface.uc_close(uc));
+            Dispose(false);
+        }
+
+        public void Dispose()
+        {
+            Dispose(true);
+            GC.SuppressFinalize(this);
+        }
+
+        protected virtual void Dispose(bool disposing)
+        {
+            if (!_isDisposed)
+            {
+                Interface.Checked(Native.Interface.uc_close(uc));
+                _isDisposed = true;
+            }
         }
 
         public void RunForCount(ulong count)

Ryujinx.Tests/Cpu/CpuTest.cs

@@ -80,6 +80,12 @@ namespace Ryujinx.Tests.Cpu
         [TearDown]
         public void Teardown()
         {
+            if (_unicornAvailable)
+            {
+                _unicornEmu.Dispose();
+                _unicornEmu = null;
+            }
+
             _memory.DecrementReferenceCount();
             _context.Dispose();
             _ram.Dispose();

Ryujinx.Tests/Cpu/CpuTest32.cs

@@ -76,6 +76,12 @@ namespace Ryujinx.Tests.Cpu
         [TearDown]
         public void Teardown()
         {
+            if (_unicornAvailable)
+            {
+                _unicornEmu.Dispose();
+                _unicornEmu = null;
+            }
+
             _memory.DecrementReferenceCount();
             _context.Dispose();
             _ram.Dispose();

Ryujinx.Tests/Cpu/CpuTestSimdCvt32.cs

@@ -339,6 +339,93 @@ namespace Ryujinx.Tests.Cpu
 
             CompareAgainstUnicorn();
         }
+
+        [Explicit]
+        [Test, Pairwise, Description("VCVT<top>.F16.F32 <Sd>, <Dm>")]
+        public void Vcvt_F32_F16([Values(0u, 1u, 2u, 3u)] uint rd,
+                                 [Values(0u, 1u, 2u, 3u)] uint rm,
+                                 [ValueSource(nameof(_1S_))] [Random(RndCnt)] uint s0,
+                                 [ValueSource(nameof(_1S_))] [Random(RndCnt)] uint s1,
+                                 [ValueSource(nameof(_1S_))] [Random(RndCnt)] uint s2,
+                                 [ValueSource(nameof(_1S_))] [Random(RndCnt)] uint s3,
+                                 [Values] bool top)
+        {
+            uint opcode = 0xeeb30a40; // VCVTB.F16.F32 S0, D0
+
+            if (top)
+            {
+                opcode |= 1 << 7;
+            }
+
+            opcode |= ((rd & 0x1e) << 11) | ((rd & 0x1) << 22);
+            opcode |= ((rm & 0x1e) >> 1) | ((rm & 0x1) << 5);
+
+            V128 v0 = MakeVectorE0E1E2E3(s0, s1, s2, s3);
+
+            SingleOpcode(opcode, v0: v0);
+
+            CompareAgainstUnicorn();
+        }
+
+        [Explicit]
+        [Test, Pairwise, Description("VCVT<top>.F16.F64 <Sd>, <Dm>")]
+        public void Vcvt_F64_F16([Values(0u, 1u, 2u, 3u)] uint rd,
+                                 [Values(0u, 1u)] uint rm,
+                                 [ValueSource(nameof(_1D_F_))] ulong d0,
+                                 [ValueSource(nameof(_1D_F_))] ulong d1,
+                                 [Values] bool top)
+        {
+            uint opcode = 0xeeb30b40; // VCVTB.F16.F64 S0, D0
+
+            if (top)
+            {
+                opcode |= 1 << 7;
+            }
+
+            opcode |= ((rd & 0x1e) << 11) | ((rd & 0x1) << 22);
+            opcode |= ((rm & 0xf) << 0) | ((rm & 0x10) << 1);
+
+            V128 v0 = MakeVectorE0E1(d0, d1);
+
+            SingleOpcode(opcode, v0: v0);
+
+            CompareAgainstUnicorn();
+        }
+
+        [Explicit]
+        [Test, Pairwise, Description("VCVT<top>.F<size>.F16 <Vd>, <Sm>")]
+        public void Vcvt_F16_Fx([Values(0u, 1u, 2u, 3u)] uint rd,
+                                 [Values(0u, 1u, 2u, 3u)] uint rm,
+                                 [ValueSource(nameof(_1D_F_))] ulong d0,
+                                 [ValueSource(nameof(_1D_F_))] ulong d1,
+                                 [Values] bool top,
+                                 [Values] bool sz)
+        {
+            uint opcode = 0xeeb20a40; // VCVTB.F32.F16 S0, S0
+
+            if (top)
+            {
+                opcode |= 1 << 7;
+            }
+
+            if (sz)
+            {
+                opcode |= 1 << 8;
+                opcode |= ((rd & 0xf) << 12) | ((rd & 0x10) << 18);
+            }
+            else
+            {
+                opcode |= ((rd & 0x1e) << 11) | ((rd & 0x1) << 22);
+            }
+
+            opcode |= ((rm & 0xf) << 0) | ((rm & 0x10) << 1);
+
+            V128 v0 = MakeVectorE0E1(d0, d1);
+
+            SingleOpcode(opcode, v0: v0);
+
+            CompareAgainstUnicorn();
+        }
 #endif
     }
 }

Ryujinx/Modules/Updater/UpdateDialog.cs

@@ -48,7 +48,7 @@ namespace Ryujinx.Modules
             {
                 string ryuName = OperatingSystem.IsWindows() ? "Ryujinx.exe" : "Ryujinx";
                 string ryuExe  = System.IO.Path.Combine(AppDomain.CurrentDomain.BaseDirectory, ryuName);
-                string ryuArg  = string.Join(" ", Environment.GetCommandLineArgs().AsEnumerable().Skip(1).ToArray());
+                var ryuArg  = Environment.GetCommandLineArgs().AsEnumerable().Skip(1);
 
                 Process.Start(ryuExe, ryuArg);
 

Ryujinx/Ryujinx.csproj

@@ -23,7 +23,7 @@
     <PackageReference Include="Ryujinx.Graphics.Nvdec.Dependencies" Version="5.0.1-build10" Condition="'$(RuntimeIdentifier)' != 'linux-x64' AND '$(RuntimeIdentifier)' != 'osx-x64'" />
     <PackageReference Include="Ryujinx.Audio.OpenAL.Dependencies" Version="1.21.0.1" Condition="'$(RuntimeIdentifier)' != 'linux-x64' AND '$(RuntimeIdentifier)' != 'osx-x64'" />
     <PackageReference Include="OpenTK.Graphics" Version="4.7.2" />
-    <PackageReference Include="SPB" Version="0.0.4-build24" />
+    <PackageReference Include="SPB" Version="0.0.4-build27" />
     <PackageReference Include="SharpZipLib" Version="1.3.3" />
     <PackageReference Include="SixLabors.ImageSharp" Version="1.0.4" />
   </ItemGroup>