GPU: Fixed the bit 25 in the display transfer flags.

It is used to downscale the input image horizontally and vertically, previously we were only downscaling it vertically so this caused a hard-to-debug memory corruption problem.

src/core/hw/gpu.cpp

@@ -116,8 +116,8 @@ inline void Write(u32 addr, const T data) {
             u8* src_pointer = Memory::GetPointer(Memory::PhysicalToVirtualAddress(config.GetPhysicalInputAddress()));
             u8* dst_pointer = Memory::GetPointer(Memory::PhysicalToVirtualAddress(config.GetPhysicalOutputAddress()));
 
-            unsigned horizontal_scale = (config.scale_horizontally != 0) ? 2 : 1;
-            unsigned vertical_scale = (config.scale_vertically != 0) ? 2 : 1;
+            unsigned horizontal_scale = (config.scale_x != 0 || config.scale_xy != 0) ? 2 : 1;
+            unsigned vertical_scale = (config.scale_xy != 0) ? 2 : 1;
             
             u32 output_width = config.output_width / horizontal_scale;
             u32 output_height = config.output_height / vertical_scale;

src/core/hw/gpu.h

@@ -197,8 +197,8 @@ struct Regs {
             BitField< 8, 3, PixelFormat> input_format;
             BitField<12, 3, PixelFormat> output_format;
 
-            BitField<24, 1, u32> scale_horizontally;
-            BitField<25, 1, u32> scale_vertically;
+            BitField<24, 1, u32> scale_x;  // Shrinks the image in half horizontally, blending the extra pixels
+            BitField<25, 1, u32> scale_xy; // Shrinks the image horizontally and vertically, blending the extra pixels
         };
 
         INSERT_PADDING_WORDS(0x1);