diff options
Diffstat (limited to 'src/video_core/textures')
-rw-r--r-- | src/video_core/textures/astc.cpp | 41 | ||||
-rw-r--r-- | src/video_core/textures/decoders.cpp | 8 |
2 files changed, 27 insertions, 22 deletions
diff --git a/src/video_core/textures/astc.cpp b/src/video_core/textures/astc.cpp index acd5bdd78..3625b666c 100644 --- a/src/video_core/textures/astc.cpp +++ b/src/video_core/textures/astc.cpp @@ -42,21 +42,24 @@ constexpr u32 Popcnt(u32 n) { class InputBitStream { public: - constexpr explicit InputBitStream(const u8* ptr, std::size_t start_offset = 0) - : cur_byte{ptr}, next_bit{start_offset % 8} {} + constexpr explicit InputBitStream(std::span<const u8> data, size_t start_offset = 0) + : cur_byte{data.data()}, total_bits{data.size()}, next_bit{start_offset % 8} {} - constexpr std::size_t GetBitsRead() const { + constexpr size_t GetBitsRead() const { return bits_read; } constexpr bool ReadBit() { - const bool bit = (*cur_byte >> next_bit++) & 1; + if (bits_read >= total_bits * 8) { + return 0; + } + const bool bit = ((*cur_byte >> next_bit) & 1) != 0; + ++next_bit; while (next_bit >= 8) { next_bit -= 8; - cur_byte++; + ++cur_byte; } - - bits_read++; + ++bits_read; return bit; } @@ -79,8 +82,9 @@ public: private: const u8* cur_byte; - std::size_t next_bit = 0; - std::size_t bits_read = 0; + size_t total_bits = 0; + size_t next_bit = 0; + size_t bits_read = 0; }; class OutputBitStream { @@ -193,15 +197,15 @@ struct IntegerEncodedValue { }; }; using IntegerEncodedVector = boost::container::static_vector< - IntegerEncodedValue, 64, + IntegerEncodedValue, 256, boost::container::static_vector_options< boost::container::inplace_alignment<alignof(IntegerEncodedValue)>, boost::container::throw_on_overflow<false>>::type>; static void DecodeTritBlock(InputBitStream& bits, IntegerEncodedVector& result, u32 nBitsPerValue) { // Implement the algorithm in section C.2.12 - u32 m[5]; - u32 t[5]; + std::array<u32, 5> m; + std::array<u32, 5> t; u32 T; // Read the trit encoded block according to @@ -866,7 +870,7 @@ public: } }; -static void DecodeColorValues(u32* out, u8* data, const u32* modes, const u32 nPartitions, +static void DecodeColorValues(u32* out, std::span<u8> data, const u32* modes, const u32 nPartitions, const u32 nBitsForColorData) { // First figure out how many color values we have u32 nValues = 0; @@ -898,7 +902,7 @@ static void DecodeColorValues(u32* out, u8* data, const u32* modes, const u32 nP // We now have enough to decode our integer sequence. IntegerEncodedVector decodedColorValues; - InputBitStream colorStream(data); + InputBitStream colorStream(data, 0); DecodeIntegerSequence(decodedColorValues, colorStream, range, nValues); // Once we have the decoded values, we need to dequantize them to the 0-255 range @@ -1441,7 +1445,7 @@ static void ComputeEndpos32s(Pixel& ep1, Pixel& ep2, const u32*& colorValues, static void DecompressBlock(std::span<const u8, 16> inBuf, const u32 blockWidth, const u32 blockHeight, std::span<u32, 12 * 12> outBuf) { - InputBitStream strm(inBuf.data()); + InputBitStream strm(inBuf); TexelWeightParams weightParams = DecodeBlockInfo(strm); // Was there an error? @@ -1619,15 +1623,16 @@ static void DecompressBlock(std::span<const u8, 16> inBuf, const u32 blockWidth, // Make sure that higher non-texel bits are set to zero const u32 clearByteStart = (weightParams.GetPackedBitSize() >> 3) + 1; - if (clearByteStart > 0) { + if (clearByteStart > 0 && clearByteStart <= texelWeightData.size()) { texelWeightData[clearByteStart - 1] &= static_cast<u8>((1 << (weightParams.GetPackedBitSize() % 8)) - 1); + std::memset(texelWeightData.data() + clearByteStart, 0, + std::min(16U - clearByteStart, 16U)); } - std::memset(texelWeightData.data() + clearByteStart, 0, std::min(16U - clearByteStart, 16U)); IntegerEncodedVector texelWeightValues; - InputBitStream weightStream(texelWeightData.data()); + InputBitStream weightStream(texelWeightData); DecodeIntegerSequence(texelWeightValues, weightStream, weightParams.m_MaxWeight, weightParams.GetNumWeightValues()); diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp index 9f5181318..62685a183 100644 --- a/src/video_core/textures/decoders.cpp +++ b/src/video_core/textures/decoders.cpp @@ -49,7 +49,7 @@ void Swizzle(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixe // We can configure here a custom pitch // As it's not exposed 'width * bpp' will be the expected pitch. const u32 pitch = width * bytes_per_pixel; - const u32 stride = Common::AlignBits(width, stride_alignment) * bytes_per_pixel; + const u32 stride = Common::AlignUpLog2(width, stride_alignment) * bytes_per_pixel; const u32 gobs_in_x = Common::DivCeilLog2(stride, GOB_SIZE_X_SHIFT); const u32 block_size = gobs_in_x << (GOB_SIZE_SHIFT + block_height + block_depth); @@ -217,9 +217,9 @@ void SwizzleKepler(const u32 width, const u32 height, const u32 dst_x, const u32 std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height, u32 depth, u32 block_height, u32 block_depth) { if (tiled) { - const u32 aligned_width = Common::AlignBits(width * bytes_per_pixel, GOB_SIZE_X_SHIFT); - const u32 aligned_height = Common::AlignBits(height, GOB_SIZE_Y_SHIFT + block_height); - const u32 aligned_depth = Common::AlignBits(depth, GOB_SIZE_Z_SHIFT + block_depth); + const u32 aligned_width = Common::AlignUpLog2(width * bytes_per_pixel, GOB_SIZE_X_SHIFT); + const u32 aligned_height = Common::AlignUpLog2(height, GOB_SIZE_Y_SHIFT + block_height); + const u32 aligned_depth = Common::AlignUpLog2(depth, GOB_SIZE_Z_SHIFT + block_depth); return aligned_width * aligned_height * aligned_depth; } else { return width * height * depth * bytes_per_pixel; |