GPU: Avoid using garbage size for non-cb0 storage buffers (#4999)

* GPU: Avoid using garbage size for non-cb0 storage buffers

In the depths area, Tears of the Kingdom uses a global memory access with address on constant buffer slot 6. This isn't standard and thus doesn't actually have a size 8 bytes after it, so we were reading back a garbage size that ended up very large (at least in version 1.1.0), and would synchronize a lot of data per frame.

This PR makes storage buffers created from addresses outside constant buffer slot 0 get their size as the number of bytes remaining in the GPU mapping starting at the given virtual address. This should bound the buffer to a reasonable size, and ideally stop it crossing into other memory.

* Limit max size

* Add TODO

* Feedback

src/Ryujinx.Ava/Common/ApplicationHelper.cs

@@ -193,7 +193,7 @@ namespace Ryujinx.Ava.Common
                             if (nca.Header.ContentType == NcaContentType.Program)
                             {
                                 int dataIndex = Nca.GetSectionIndexFromType(NcaSectionType.Data, NcaContentType.Program);
-                                if (nca.Header.GetFsHeader(dataIndex).IsPatchSection())
+                                if (nca.SectionExists(NcaSectionType.Data) && nca.Header.GetFsHeader(dataIndex).IsPatchSection())
                                 {
                                     patchNca = nca;
                                 }

src/Ryujinx.Graphics.Gpu/Engine/Threed/StateUpdater.cs

@@ -23,6 +23,8 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
         public const int PrimitiveRestartStateIndex = 12;
         public const int RenderTargetStateIndex = 27;
 
+        private const ulong MaxUnknownStorageSize = 0x100000;
+
         private readonly GpuContext _context;
         private readonly GpuChannel _channel;
         private readonly DeviceStateWithShadow<ThreedClassState> _state;
@@ -356,7 +358,19 @@ namespace Ryujinx.Graphics.Gpu.Engine.Threed
 
                     SbDescriptor sbDescriptor = _channel.MemoryManager.Physical.Read<SbDescriptor>(sbDescAddress);
 
-                    _channel.BufferManager.SetGraphicsStorageBuffer(stage, sb.Slot, sbDescriptor.PackAddress(), (uint)sbDescriptor.Size, sb.Flags);
+                    uint size;
+                    if (sb.SbCbSlot == 0)
+                    {
+                        // Only trust the SbDescriptor size if it comes from slot 0.
+                        size = (uint)sbDescriptor.Size;
+                    }
+                    else
+                    {
+                        // TODO: Use full mapped size and somehow speed up buffer sync.
+                        size = (uint)_channel.MemoryManager.GetMappedSize(sbDescriptor.PackAddress(), MaxUnknownStorageSize);
+                    }
+
+                    _channel.BufferManager.SetGraphicsStorageBuffer(stage, sb.Slot, sbDescriptor.PackAddress(), size, sb.Flags);
                 }
             }
         }

src/Ryujinx.Graphics.Gpu/GpuContext.cs

@@ -99,6 +99,7 @@ namespace Ryujinx.Graphics.Gpu
         private bool _pendingSync;
 
         private long _modifiedSequence;
+        private ulong _firstTimestamp;
 
         /// <summary>
         /// Creates a new instance of the GPU emulation context.
@@ -123,6 +124,8 @@ namespace Ryujinx.Graphics.Gpu
             DeferredActions = new Queue<Action>();
 
             PhysicalMemoryRegistry = new ConcurrentDictionary<ulong, PhysicalMemory>();
+
+            _firstTimestamp = ConvertNanosecondsToTicks((ulong)PerformanceCounter.ElapsedNanoseconds);
         }
 
         /// <summary>
@@ -217,7 +220,8 @@ namespace Ryujinx.Graphics.Gpu
         /// <returns>The current GPU timestamp</returns>
         public ulong GetTimestamp()
         {
-            ulong ticks = ConvertNanosecondsToTicks((ulong)PerformanceCounter.ElapsedNanoseconds);
+            // Guest timestamp will start at 0, instead of host value. 
+            ulong ticks = ConvertNanosecondsToTicks((ulong)PerformanceCounter.ElapsedNanoseconds) - _firstTimestamp;
 
             if (GraphicsConfig.FastGpuTime)
             {

src/Ryujinx.Graphics.Gpu/Memory/MemoryManager.cs

@@ -637,6 +637,33 @@ namespace Ryujinx.Graphics.Gpu.Memory
             return UnpackPaFromPte(pte) + (va & PageMask);
         }
 
+        /// <summary>
+        /// Translates a GPU virtual address and returns the number of bytes that are mapped after it.
+        /// </summary>
+        /// <param name="va">GPU virtual address to be translated</param>
+        /// <param name="maxSize">Maximum size in bytes to scan</param>
+        /// <returns>Number of bytes, 0 if unmapped</returns>
+        public ulong GetMappedSize(ulong va, ulong maxSize)
+        {
+            if (!ValidateAddress(va))
+            {
+                return 0;
+            }
+
+            ulong startVa = va;
+            ulong endVa = va + maxSize;
+
+            ulong pte = GetPte(va);
+
+            while (pte != PteUnmapped && va < endVa)
+            {
+                va += PageSize - (va & PageMask);
+                pte = GetPte(va);
+            }
+
+            return Math.Min(maxSize, va - startVa);
+        }
+
         /// <summary>
         /// Gets the kind of a given memory page.
         /// This might indicate the type of resource that can be allocated on the page, and also texture tiling.