nx2: decode ASTC textures with the Switch 2 GOB layout

The GPU ASTC decoder unswizzles block-linear data itself and never applied
the native (Switch 2) GOB swizzle that the other unswizzle paths use, so
ASTC textures came out with 16-byte blocks swapped wherever the image
varies. Title 4's UI is ASTC (10x5, 4x4, 8x8): portraits, buttons and the
squid cursor had hatched edges and clean interiors. Pass the native tiling
flag to the decoder and swap offset bits 5 and 6 as
block_linear_unswizzle_2d.comp does.

src/video_core/host_shaders/astc_decoder.comp

@@ -33,6 +33,7 @@ UNIFORM(3) uint block_size;
 UNIFORM(4) uint x_shift;
 UNIFORM(5) uint block_height;
 UNIFORM(6) uint block_height_mask;
+UNIFORM(7) uint native_tiling;
 END_PUSH_CONSTANTS
 
 struct EncodingData {
@@ -1404,7 +1405,12 @@ uint SwizzleOffset(uvec2 pos) {
 void main() {
     uvec3 pos = gl_GlobalInvocationID;
     pos.x <<= BYTES_PER_BLOCK_LOG2;
-    const uint swizzle = SwizzleOffset(pos.xy);
+    uint swizzle = SwizzleOffset(pos.xy);
+    if (native_tiling != 0u) {
+        // Switch 2 GOBs swap offset bits 5 and 6 (see block_linear_unswizzle_2d.comp).
+        const uint difference = ((swizzle >> 5u) ^ (swizzle >> 6u)) & 1u;
+        swizzle ^= (difference << 5u) | (difference << 6u);
+    }
     const uint block_y = pos.y >> GOB_SIZE_Y_SHIFT;
 
     uint offset = 0;

src/video_core/renderer_opengl/util_shaders.cpp

@@ -96,6 +96,7 @@ void UtilShaders::ASTCDecode(Image& image, const StagingBufferMap& map,
         glUniform1ui(4, params.x_shift);
         glUniform1ui(5, params.block_height);
         glUniform1ui(6, params.block_height_mask);
+        glUniform1ui(7, params.native_tiling);
 
         // ASTC texture data
         glBindBufferRange(GL_SHADER_STORAGE_BUFFER, BINDING_INPUT_BUFFER, map.buffer, input_offset,

src/video_core/renderer_vulkan/vk_compute_pass.cpp

@@ -183,6 +183,7 @@ struct AstcPushConstants {
     u32 x_shift;
     u32 block_height;
     u32 block_height_mask;
+    u32 native_tiling;
 };
 
 struct QueriesPrefixScanPushConstants {
@@ -583,6 +584,7 @@ void ASTCDecoderPass::Assemble(Image& image, const StagingBufferRef& map,
                 .x_shift = params.x_shift,
                 .block_height = params.block_height,
                 .block_height_mask = params.block_height_mask,
+                .native_tiling = params.native_tiling,
             };
             const VkDescriptorSet set = descriptor_allocator.Commit();
             device.GetLogical().UpdateDescriptorSet(set, *descriptor_template, descriptor_data);