7 files changed, 41 insertions, 29 deletions

diff --git a/src/video_core/engines/engine_upload.cpp b/src/video_core/engines/engine_upload.cpp
index c776b9a56..d44ad0cd8 100644
--- a/src/video_core/engines/engine_upload.cpp
+++ b/src/video_core/engines/engine_upload.cpp
@@ -39,7 +39,7 @@ void State::ProcessData(const u32 data, const bool is_last_call) {
         UNIMPLEMENTED_IF(regs.dest.BlockWidth() != 0);
         UNIMPLEMENTED_IF(regs.dest.BlockDepth() != 0);
         const std::size_t dst_size = Tegra::Texture::CalculateSize(
-            true, 1, regs.dest.width, regs.dest.height, 1, regs.dest.BlockHeight(), 1);
+            true, 1, regs.dest.width, regs.dest.height, 1, regs.dest.BlockHeight(), 0);
         tmp_buffer.resize(dst_size);
         memory_manager.ReadBlock(address, tmp_buffer.data(), dst_size);
         Tegra::Texture::SwizzleKepler(regs.dest.width, regs.dest.height, regs.dest.x, regs.dest.y,
diff --git a/src/video_core/engines/engine_upload.h b/src/video_core/engines/engine_upload.h
index cb294aec3..462da419e 100644
--- a/src/video_core/engines/engine_upload.h
+++ b/src/video_core/engines/engine_upload.h
@@ -39,15 +39,15 @@ struct Registers {
         }
 
         u32 BlockWidth() const {
-            return block_width;
+            return block_width.Value();
         }
 
         u32 BlockHeight() const {
-            return block_height;
+            return block_height.Value();
         }
 
         u32 BlockDepth() const {
-            return block_depth;
+            return block_depth.Value();
         }
     } dest;
 };
diff --git a/src/video_core/engines/fermi_2d.h b/src/video_core/engines/fermi_2d.h
index 0a4c7c5ad..05421d185 100644
--- a/src/video_core/engines/fermi_2d.h
+++ b/src/video_core/engines/fermi_2d.h
@@ -84,15 +84,15 @@ public:
             }
 
             u32 BlockWidth() const {
-                return block_width;
+                return block_width.Value();
             }
 
             u32 BlockHeight() const {
-                return block_height;
+                return block_height.Value();
             }
 
             u32 BlockDepth() const {
-                return block_depth;
+                return block_depth.Value();
             }
         };
         static_assert(sizeof(Surface) == 0x28, "Surface has incorrect size");
diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp
index afb9578d0..3a5dfef0c 100644
--- a/src/video_core/engines/maxwell_dma.cpp
+++ b/src/video_core/engines/maxwell_dma.cpp
@@ -111,7 +111,7 @@ void MaxwellDMA::HandleCopy() {
 
         memory_manager.WriteBlock(dest, write_buffer.data(), dst_size);
     } else {
-        ASSERT(regs.dst_params.BlockDepth() == 0);
+        ASSERT(regs.dst_params.BlockDepth() == 1);
 
         const u32 src_bytes_per_pixel = regs.src_pitch / regs.x_count;
 
diff --git a/src/video_core/engines/maxwell_dma.h b/src/video_core/engines/maxwell_dma.h
index 522fa97dc..17b015ca7 100644
--- a/src/video_core/engines/maxwell_dma.h
+++ b/src/video_core/engines/maxwell_dma.h
@@ -59,11 +59,11 @@ public:
             };
 
             u32 BlockHeight() const {
-                return block_height;
+                return block_height.Value();
             }
 
             u32 BlockDepth() const {
-                return block_depth;
+                return block_depth.Value();
             }
         };
 
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index 201c4d42e..7a9b4c27d 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -335,6 +335,9 @@ private:
         if (untopological == MatchTopologyResult::CompressUnmatch) {
             return RecycleStrategy::Flush;
         }
+        if (untopological == MatchTopologyResult::FullMatch && !params.is_tiled) {
+            return RecycleStrategy::Flush;
+        }
         return RecycleStrategy::Ignore;
     }
 
@@ -372,6 +375,11 @@ private:
             }
             return InitializeSurface(gpu_addr, params, preserve_contents);
         }
+        case RecycleStrategy::BufferCopy: {
+            auto new_surface = GetUncachedSurface(gpu_addr, params);
+            BufferCopy(overlaps[0], new_surface);
+            return {new_surface, new_surface->GetMainView()};
+        }
         default: {
             UNIMPLEMENTED_MSG("Unimplemented Texture Cache Recycling Strategy!");
             return InitializeSurface(gpu_addr, params, do_load);
@@ -520,6 +528,10 @@ private:
         const auto host_ptr{memory_manager->GetPointer(gpu_addr)};
         const auto cache_addr{ToCacheAddr(host_ptr)};
 
+        if (gpu_addr == 0x00000001682F0000ULL) {
+            LOG_CRITICAL(HW_GPU, "Here's the texture!");
+        }
+
         // Step 0: guarantee a valid surface
         if (!cache_addr) {
             // Return a null surface if it's invalid
@@ -566,6 +578,10 @@ private:
             return InitializeSurface(gpu_addr, params, preserve_contents);
         }
 
+        if (!params.is_tiled) {
+            return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
+                                  MatchTopologyResult::FullMatch);
+        }
         // Step 3
         // Now we need to figure the relationship between the texture and its overlaps
         // we do a topological test to ensure we can find some relationship. If it fails
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index f45fd175a..9a2f4198a 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -256,19 +256,18 @@ std::vector<u8> UnswizzleTexture(u8* address, u32 tile_size_x, u32 tile_size_y,
 }
 
 void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
-                    u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height) {
-    const u32 block_height_size{1U << block_height};
+                    u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height_bit) {
+    const u32 block_height = 1U << block_height_bit;
     const u32 image_width_in_gobs{(swizzled_width * bytes_per_pixel + (gob_size_x - 1)) /
                                   gob_size_x};
     for (u32 line = 0; line < subrect_height; ++line) {
         const u32 gob_address_y =
-            (line / (gob_size_y * block_height_size)) * gob_size * block_height_size *
-                image_width_in_gobs +
-            ((line % (gob_size_y * block_height_size)) / gob_size_y) * gob_size;
+            (line / (gob_size_y * block_height)) * gob_size * block_height * image_width_in_gobs +
+            ((line % (gob_size_y * block_height)) / gob_size_y) * gob_size;
         const auto& table = legacy_swizzle_table[line % gob_size_y];
         for (u32 x = 0; x < subrect_width; ++x) {
             const u32 gob_address =
-                gob_address_y + (x * bytes_per_pixel / gob_size_x) * gob_size * block_height_size;
+                gob_address_y + (x * bytes_per_pixel / gob_size_x) * gob_size * block_height;
             const u32 swizzled_offset = gob_address + table[(x * bytes_per_pixel) % gob_size_x];
             u8* source_line = unswizzled_data + line * source_pitch + x * bytes_per_pixel;
             u8* dest_addr = swizzled_data + swizzled_offset;
@@ -279,19 +278,17 @@ void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32
 }
 
 void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32 swizzled_width,
-                      u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height,
+                      u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height_bit,
                       u32 offset_x, u32 offset_y) {
-    const u32 block_height_size{1U << block_height};
+    const u32 block_height = 1U << block_height_bit;
     for (u32 line = 0; line < subrect_height; ++line) {
         const u32 y2 = line + offset_y;
-        const u32 gob_address_y =
-            (y2 / (gob_size_y * block_height_size)) * gob_size * block_height_size +
-            ((y2 % (gob_size_y * block_height_size)) / gob_size_y) * gob_size;
+        const u32 gob_address_y = (y2 / (gob_size_y * block_height)) * gob_size * block_height +
+                                  ((y2 % (gob_size_y * block_height)) / gob_size_y) * gob_size;
         const auto& table = legacy_swizzle_table[y2 % gob_size_y];
         for (u32 x = 0; x < subrect_width; ++x) {
             const u32 x2 = (x + offset_x) * bytes_per_pixel;
-            const u32 gob_address =
-                gob_address_y + (x2 / gob_size_x) * gob_size * block_height_size;
+            const u32 gob_address = gob_address_y + (x2 / gob_size_x) * gob_size * block_height;
             const u32 swizzled_offset = gob_address + table[x2 % gob_size_x];
             u8* dest_line = unswizzled_data + line * dest_pitch + x * bytes_per_pixel;
             u8* source_addr = swizzled_data + swizzled_offset;
@@ -302,20 +299,19 @@ void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32
 }
 
 void SwizzleKepler(const u32 width, const u32 height, const u32 dst_x, const u32 dst_y,
-                   const u32 block_height, const std::size_t copy_size, const u8* source_data,
+                   const u32 block_height_bit, const std::size_t copy_size, const u8* source_data,
                    u8* swizzle_data) {
-    const u32 block_height_size{1U << block_height};
+    const u32 block_height = 1U << block_height_bit;
     const u32 image_width_in_gobs{(width + gob_size_x - 1) / gob_size_x};
     std::size_t count = 0;
     for (std::size_t y = dst_y; y < height && count < copy_size; ++y) {
         const std::size_t gob_address_y =
-            (y / (gob_size_y * block_height_size)) * gob_size * block_height_size *
-                image_width_in_gobs +
-            ((y % (gob_size_y * block_height_size)) / gob_size_y) * gob_size;
+            (y / (gob_size_y * block_height)) * gob_size * block_height * image_width_in_gobs +
+            ((y % (gob_size_y * block_height)) / gob_size_y) * gob_size;
         const auto& table = legacy_swizzle_table[y % gob_size_y];
         for (std::size_t x = dst_x; x < width && count < copy_size; ++x) {
             const std::size_t gob_address =
-                gob_address_y + (x / gob_size_x) * gob_size * block_height_size;
+                gob_address_y + (x / gob_size_x) * gob_size * block_height;
             const std::size_t swizzled_offset = gob_address + table[x % gob_size_x];
             const u8* source_line = source_data + count;
             u8* dest_addr = swizzle_data + swizzled_offset;