commit c269a16 Author: Jeff Muizelaar Date: Fri Sep 14 15:54:55 2012 -0400 Bug 791305. Use libjpeg's color conversion code instead of our own. r=joe,r=khuey libjpeg-turbo supports converting directly to a format compatible with cairo's FORMAT_RGB24. Use that instead of our own handcoded function. This also gives us SSE2 and NEON version of this function. --HG-- extra : rebase_source : 18f48925f023a33ec2a097d4f4e5cc2ab40be1e9 --- configure.in | 6 +- image/decoders/nsJPEGDecoder.cpp | 311 ++------------------------------------- 2 files changed, 20 insertions(+), 297 deletions(-) diff --git configure.in configure.in index f0aeb5d..1c01010 100644 --- configure.in +++ configure.in @@ -4027,11 +4027,7 @@ if test "$MOZ_NATIVE_JPEG" = 1; then #include ], [ #if JPEG_LIB_VERSION < $MOZJPEG #error "Insufficient JPEG library version ($MOZJPEG required)." - #endif - #ifndef JCS_EXTENSIONS - #error "libjpeg-turbo JCS_EXTENSIONS required" - #endif - ], + #endif ], MOZ_NATIVE_JPEG=1, AC_MSG_ERROR([Insufficient JPEG library version for --with-system-jpeg])) fi diff --git image/decoders/nsJPEGDecoder.cpp image/decoders/nsJPEGDecoder.cpp index 8fa8200..1d2a259 100644 --- image/decoders/nsJPEGDecoder.cpp +++ image/decoders/nsJPEGDecoder.cpp @@ -21,13 +21,28 @@ extern "C" { #include "iccjpeg.h" -} +#ifdef JCS_EXTENSIONS #if defined(IS_BIG_ENDIAN) #define MOZ_JCS_EXT_NATIVE_ENDIAN_XRGB JCS_EXT_XRGB #else #define MOZ_JCS_EXT_NATIVE_ENDIAN_XRGB JCS_EXT_BGRX #endif +#else +/* Colorspace conversion (copied from jpegint.h) */ +struct jpeg_color_deconverter { + JMETHOD(void, start_pass, (j_decompress_ptr cinfo)); + JMETHOD(void, color_convert, (j_decompress_ptr cinfo, + JSAMPIMAGE input_buf, JDIMENSION input_row, + JSAMPARRAY output_buf, int num_rows)); +}; + +METHODDEF(void) +ycc_rgb_convert_argb (j_decompress_ptr cinfo, + JSAMPIMAGE input_buf, JDIMENSION input_row, + JSAMPARRAY output_buf, int num_rows); +#endif +} static void cmyk_convert_rgb(JSAMPROW row, JDIMENSION width); @@ -324,6 +340,7 @@ nsJPEGDecoder::WriteInternal(const char *aBuffer, uint32_t aCount) case JCS_GRAYSCALE: case JCS_RGB: case JCS_YCbCr: +#ifdef JCS_EXTENSIONS // if we're not color managing we can decode directly to // MOZ_JCS_EXT_NATIVE_ENDIAN_XRGB if (mCMSMode != eCMSMode_All) { @@ -332,6 +349,9 @@ nsJPEGDecoder::WriteInternal(const char *aBuffer, uint32_t aCount) } else { mInfo.out_color_space = JCS_RGB; } +#else + mInfo.out_color_space = JCS_RGB; +#endif break; case JCS_CMYK: case JCS_YCCK: @@ -399,6 +419,15 @@ nsJPEGDecoder::WriteInternal(const char *aBuffer, uint32_t aCount) return; /* I/O suspension */ } +#ifndef JCS_EXTENSIONS + /* Force to use our YCbCr to Packed RGB converter when possible */ + if (!mTransform && (mCMSMode != eCMSMode_All) && + mInfo.jpeg_color_space == JCS_YCbCr && mInfo.out_color_space == JCS_RGB) { + /* Special case for the most common case: transform from YCbCr direct into packed ARGB */ + mInfo.out_color_components = 4; /* Packed ARGB pixels are always 4 bytes...*/ + mInfo.cconvert->color_convert = ycc_rgb_convert_argb; + } +#endif /* If this is a progressive JPEG ... */ mState = mInfo.buffered_image ? JPEG_DECOMPRESS_PROGRESSIVE : JPEG_DECOMPRESS_SEQUENTIAL; @@ -544,7 +573,11 @@ nsJPEGDecoder::OutputScanlines(bool* suspend) uint32_t *imageRow = ((uint32_t*)mImageData) + (mInfo.output_scanline * mInfo.output_width); +#ifdef JCS_EXTENSIONS if (mInfo.out_color_space == MOZ_JCS_EXT_NATIVE_ENDIAN_XRGB) { +#else + if (mInfo.cconvert->color_convert == ycc_rgb_convert_argb) { +#endif /* Special case: scanline will be directly converted into packed ARGB */ if (jpeg_read_scanlines(&mInfo, (JSAMPARRAY)&imageRow, 1) != 1) { *suspend = true; /* suspend */ @@ -854,6 +887,282 @@ term_source (j_decompress_ptr jd) } // namespace mozilla +#ifndef JCS_EXTENSIONS +/**************** YCbCr -> Cairo's RGB24/ARGB32 conversion: most common case **************/ + +/* + * YCbCr is defined per CCIR 601-1, except that Cb and Cr are + * normalized to the range 0..MAXJSAMPLE rather than -0.5 .. 0.5. + * The conversion equations to be implemented are therefore + * R = Y + 1.40200 * Cr + * G = Y - 0.34414 * Cb - 0.71414 * Cr + * B = Y + 1.77200 * Cb + * where Cb and Cr represent the incoming values less CENTERJSAMPLE. + * (These numbers are derived from TIFF 6.0 section 21, dated 3-June-92.) + * + * To avoid floating-point arithmetic, we represent the fractional constants + * as integers scaled up by 2^16 (about 4 digits precision); we have to divide + * the products by 2^16, with appropriate rounding, to get the correct answer. + * Notice that Y, being an integral input, does not contribute any fraction + * so it need not participate in the rounding. + * + * For even more speed, we avoid doing any multiplications in the inner loop + * by precalculating the constants times Cb and Cr for all possible values. + * For 8-bit JSAMPLEs this is very reasonable (only 256 entries per table); + * for 12-bit samples it is still acceptable. It's not very reasonable for + * 16-bit samples, but if you want lossless storage you shouldn't be changing + * colorspace anyway. + * The Cr=>R and Cb=>B values can be rounded to integers in advance; the + * values for the G calculation are left scaled up, since we must add them + * together before rounding. + */ + +#define SCALEBITS 16 /* speediest right-shift on some machines */ + +/* Use static tables for color processing. */ +/* Four tables, each 256 entries of 4 bytes totals 4K which is not bad... */ + +const int Cr_r_tab[(MAXJSAMPLE+1) * sizeof(int)] ={ + -0xb3, -0xb2, -0xb1, -0xaf, -0xae, -0xac, + -0xab, -0xaa, -0xa8, -0xa7, -0xa5, -0xa4, + -0xa3, -0xa1, -0xa0, -0x9e, -0x9d, -0x9c, + -0x9a, -0x99, -0x97, -0x96, -0x95, -0x93, + -0x92, -0x90, -0x8f, -0x8e, -0x8c, -0x8b, + -0x89, -0x88, -0x87, -0x85, -0x84, -0x82, + -0x81, -0x80, -0x7e, -0x7d, -0x7b, -0x7a, + -0x79, -0x77, -0x76, -0x74, -0x73, -0x72, + -0x70, -0x6f, -0x6d, -0x6c, -0x6b, -0x69, + -0x68, -0x66, -0x65, -0x64, -0x62, -0x61, + -0x5f, -0x5e, -0x5d, -0x5b, -0x5a, -0x58, + -0x57, -0x56, -0x54, -0x53, -0x51, -0x50, + -0x4f, -0x4d, -0x4c, -0x4a, -0x49, -0x48, + -0x46, -0x45, -0x43, -0x42, -0x40, -0x3f, + -0x3e, -0x3c, -0x3b, -0x39, -0x38, -0x37, + -0x35, -0x34, -0x32, -0x31, -0x30, -0x2e, + -0x2d, -0x2b, -0x2a, -0x29, -0x27, -0x26, + -0x24, -0x23, -0x22, -0x20, -0x1f, -0x1d, + -0x1c, -0x1b, -0x19, -0x18, -0x16, -0x15, + -0x14, -0x12, -0x11, -0x0f, -0x0e, -0x0d, + -0x0b, -0x0a, -0x08, -0x07, -0x06, -0x04, + -0x03, -0x01, 0x00, 0x01, 0x03, 0x04, + 0x06, 0x07, 0x08, 0x0a, 0x0b, 0x0d, + 0x0e, 0x0f, 0x11, 0x12, 0x14, 0x15, + 0x16, 0x18, 0x19, 0x1b, 0x1c, 0x1d, + 0x1f, 0x20, 0x22, 0x23, 0x24, 0x26, + 0x27, 0x29, 0x2a, 0x2b, 0x2d, 0x2e, + 0x30, 0x31, 0x32, 0x34, 0x35, 0x37, + 0x38, 0x39, 0x3b, 0x3c, 0x3e, 0x3f, + 0x40, 0x42, 0x43, 0x45, 0x46, 0x48, + 0x49, 0x4a, 0x4c, 0x4d, 0x4f, 0x50, + 0x51, 0x53, 0x54, 0x56, 0x57, 0x58, + 0x5a, 0x5b, 0x5d, 0x5e, 0x5f, 0x61, + 0x62, 0x64, 0x65, 0x66, 0x68, 0x69, + 0x6b, 0x6c, 0x6d, 0x6f, 0x70, 0x72, + 0x73, 0x74, 0x76, 0x77, 0x79, 0x7a, + 0x7b, 0x7d, 0x7e, 0x80, 0x81, 0x82, + 0x84, 0x85, 0x87, 0x88, 0x89, 0x8b, + 0x8c, 0x8e, 0x8f, 0x90, 0x92, 0x93, + 0x95, 0x96, 0x97, 0x99, 0x9a, 0x9c, + 0x9d, 0x9e, 0xa0, 0xa1, 0xa3, 0xa4, + 0xa5, 0xa7, 0xa8, 0xaa, 0xab, 0xac, + 0xae, 0xaf, 0xb1, 0xb2, + }; + +const int Cb_b_tab[(MAXJSAMPLE+1) * sizeof(int)] ={ + -0xe3, -0xe1, -0xdf, -0xde, -0xdc, -0xda, + -0xd8, -0xd6, -0xd5, -0xd3, -0xd1, -0xcf, + -0xce, -0xcc, -0xca, -0xc8, -0xc6, -0xc5, + -0xc3, -0xc1, -0xbf, -0xbe, -0xbc, -0xba, + -0xb8, -0xb7, -0xb5, -0xb3, -0xb1, -0xaf, + -0xae, -0xac, -0xaa, -0xa8, -0xa7, -0xa5, + -0xa3, -0xa1, -0x9f, -0x9e, -0x9c, -0x9a, + -0x98, -0x97, -0x95, -0x93, -0x91, -0x90, + -0x8e, -0x8c, -0x8a, -0x88, -0x87, -0x85, + -0x83, -0x81, -0x80, -0x7e, -0x7c, -0x7a, + -0x78, -0x77, -0x75, -0x73, -0x71, -0x70, + -0x6e, -0x6c, -0x6a, -0x69, -0x67, -0x65, + -0x63, -0x61, -0x60, -0x5e, -0x5c, -0x5a, + -0x59, -0x57, -0x55, -0x53, -0x52, -0x50, + -0x4e, -0x4c, -0x4a, -0x49, -0x47, -0x45, + -0x43, -0x42, -0x40, -0x3e, -0x3c, -0x3a, + -0x39, -0x37, -0x35, -0x33, -0x32, -0x30, + -0x2e, -0x2c, -0x2b, -0x29, -0x27, -0x25, + -0x23, -0x22, -0x20, -0x1e, -0x1c, -0x1b, + -0x19, -0x17, -0x15, -0x13, -0x12, -0x10, + -0x0e, -0x0c, -0x0b, -0x09, -0x07, -0x05, + -0x04, -0x02, 0x00, 0x02, 0x04, 0x05, + 0x07, 0x09, 0x0b, 0x0c, 0x0e, 0x10, + 0x12, 0x13, 0x15, 0x17, 0x19, 0x1b, + 0x1c, 0x1e, 0x20, 0x22, 0x23, 0x25, + 0x27, 0x29, 0x2b, 0x2c, 0x2e, 0x30, + 0x32, 0x33, 0x35, 0x37, 0x39, 0x3a, + 0x3c, 0x3e, 0x40, 0x42, 0x43, 0x45, + 0x47, 0x49, 0x4a, 0x4c, 0x4e, 0x50, + 0x52, 0x53, 0x55, 0x57, 0x59, 0x5a, + 0x5c, 0x5e, 0x60, 0x61, 0x63, 0x65, + 0x67, 0x69, 0x6a, 0x6c, 0x6e, 0x70, + 0x71, 0x73, 0x75, 0x77, 0x78, 0x7a, + 0x7c, 0x7e, 0x80, 0x81, 0x83, 0x85, + 0x87, 0x88, 0x8a, 0x8c, 0x8e, 0x90, + 0x91, 0x93, 0x95, 0x97, 0x98, 0x9a, + 0x9c, 0x9e, 0x9f, 0xa1, 0xa3, 0xa5, + 0xa7, 0xa8, 0xaa, 0xac, 0xae, 0xaf, + 0xb1, 0xb3, 0xb5, 0xb7, 0xb8, 0xba, + 0xbc, 0xbe, 0xbf, 0xc1, 0xc3, 0xc5, + 0xc6, 0xc8, 0xca, 0xcc, 0xce, 0xcf, + 0xd1, 0xd3, 0xd5, 0xd6, 0xd8, 0xda, + 0xdc, 0xde, 0xdf, 0xe1, + }; + +const int Cr_g_tab[(MAXJSAMPLE+1) * sizeof(int)] ={ + 0x5b6900, 0x5ab22e, 0x59fb5c, 0x59448a, 0x588db8, 0x57d6e6, + 0x572014, 0x566942, 0x55b270, 0x54fb9e, 0x5444cc, 0x538dfa, + 0x52d728, 0x522056, 0x516984, 0x50b2b2, 0x4ffbe0, 0x4f450e, + 0x4e8e3c, 0x4dd76a, 0x4d2098, 0x4c69c6, 0x4bb2f4, 0x4afc22, + 0x4a4550, 0x498e7e, 0x48d7ac, 0x4820da, 0x476a08, 0x46b336, + 0x45fc64, 0x454592, 0x448ec0, 0x43d7ee, 0x43211c, 0x426a4a, + 0x41b378, 0x40fca6, 0x4045d4, 0x3f8f02, 0x3ed830, 0x3e215e, + 0x3d6a8c, 0x3cb3ba, 0x3bfce8, 0x3b4616, 0x3a8f44, 0x39d872, + 0x3921a0, 0x386ace, 0x37b3fc, 0x36fd2a, 0x364658, 0x358f86, + 0x34d8b4, 0x3421e2, 0x336b10, 0x32b43e, 0x31fd6c, 0x31469a, + 0x308fc8, 0x2fd8f6, 0x2f2224, 0x2e6b52, 0x2db480, 0x2cfdae, + 0x2c46dc, 0x2b900a, 0x2ad938, 0x2a2266, 0x296b94, 0x28b4c2, + 0x27fdf0, 0x27471e, 0x26904c, 0x25d97a, 0x2522a8, 0x246bd6, + 0x23b504, 0x22fe32, 0x224760, 0x21908e, 0x20d9bc, 0x2022ea, + 0x1f6c18, 0x1eb546, 0x1dfe74, 0x1d47a2, 0x1c90d0, 0x1bd9fe, + 0x1b232c, 0x1a6c5a, 0x19b588, 0x18feb6, 0x1847e4, 0x179112, + 0x16da40, 0x16236e, 0x156c9c, 0x14b5ca, 0x13fef8, 0x134826, + 0x129154, 0x11da82, 0x1123b0, 0x106cde, 0x0fb60c, 0x0eff3a, + 0x0e4868, 0x0d9196, 0x0cdac4, 0x0c23f2, 0x0b6d20, 0x0ab64e, + 0x09ff7c, 0x0948aa, 0x0891d8, 0x07db06, 0x072434, 0x066d62, + 0x05b690, 0x04ffbe, 0x0448ec, 0x03921a, 0x02db48, 0x022476, + 0x016da4, 0x00b6d2, 0x000000, -0x00b6d2, -0x016da4, -0x022476, + -0x02db48, -0x03921a, -0x0448ec, -0x04ffbe, -0x05b690, -0x066d62, + -0x072434, -0x07db06, -0x0891d8, -0x0948aa, -0x09ff7c, -0x0ab64e, + -0x0b6d20, -0x0c23f2, -0x0cdac4, -0x0d9196, -0x0e4868, -0x0eff3a, + -0x0fb60c, -0x106cde, -0x1123b0, -0x11da82, -0x129154, -0x134826, + -0x13fef8, -0x14b5ca, -0x156c9c, -0x16236e, -0x16da40, -0x179112, + -0x1847e4, -0x18feb6, -0x19b588, -0x1a6c5a, -0x1b232c, -0x1bd9fe, + -0x1c90d0, -0x1d47a2, -0x1dfe74, -0x1eb546, -0x1f6c18, -0x2022ea, + -0x20d9bc, -0x21908e, -0x224760, -0x22fe32, -0x23b504, -0x246bd6, + -0x2522a8, -0x25d97a, -0x26904c, -0x27471e, -0x27fdf0, -0x28b4c2, + -0x296b94, -0x2a2266, -0x2ad938, -0x2b900a, -0x2c46dc, -0x2cfdae, + -0x2db480, -0x2e6b52, -0x2f2224, -0x2fd8f6, -0x308fc8, -0x31469a, + -0x31fd6c, -0x32b43e, -0x336b10, -0x3421e2, -0x34d8b4, -0x358f86, + -0x364658, -0x36fd2a, -0x37b3fc, -0x386ace, -0x3921a0, -0x39d872, + -0x3a8f44, -0x3b4616, -0x3bfce8, -0x3cb3ba, -0x3d6a8c, -0x3e215e, + -0x3ed830, -0x3f8f02, -0x4045d4, -0x40fca6, -0x41b378, -0x426a4a, + -0x43211c, -0x43d7ee, -0x448ec0, -0x454592, -0x45fc64, -0x46b336, + -0x476a08, -0x4820da, -0x48d7ac, -0x498e7e, -0x4a4550, -0x4afc22, + -0x4bb2f4, -0x4c69c6, -0x4d2098, -0x4dd76a, -0x4e8e3c, -0x4f450e, + -0x4ffbe0, -0x50b2b2, -0x516984, -0x522056, -0x52d728, -0x538dfa, + -0x5444cc, -0x54fb9e, -0x55b270, -0x566942, -0x572014, -0x57d6e6, + -0x588db8, -0x59448a, -0x59fb5c, -0x5ab22e, + }; + +const int Cb_g_tab[(MAXJSAMPLE+1) * sizeof(int)] ={ + 0x2c8d00, 0x2c34e6, 0x2bdccc, 0x2b84b2, 0x2b2c98, 0x2ad47e, + 0x2a7c64, 0x2a244a, 0x29cc30, 0x297416, 0x291bfc, 0x28c3e2, + 0x286bc8, 0x2813ae, 0x27bb94, 0x27637a, 0x270b60, 0x26b346, + 0x265b2c, 0x260312, 0x25aaf8, 0x2552de, 0x24fac4, 0x24a2aa, + 0x244a90, 0x23f276, 0x239a5c, 0x234242, 0x22ea28, 0x22920e, + 0x2239f4, 0x21e1da, 0x2189c0, 0x2131a6, 0x20d98c, 0x208172, + 0x202958, 0x1fd13e, 0x1f7924, 0x1f210a, 0x1ec8f0, 0x1e70d6, + 0x1e18bc, 0x1dc0a2, 0x1d6888, 0x1d106e, 0x1cb854, 0x1c603a, + 0x1c0820, 0x1bb006, 0x1b57ec, 0x1affd2, 0x1aa7b8, 0x1a4f9e, + 0x19f784, 0x199f6a, 0x194750, 0x18ef36, 0x18971c, 0x183f02, + 0x17e6e8, 0x178ece, 0x1736b4, 0x16de9a, 0x168680, 0x162e66, + 0x15d64c, 0x157e32, 0x152618, 0x14cdfe, 0x1475e4, 0x141dca, + 0x13c5b0, 0x136d96, 0x13157c, 0x12bd62, 0x126548, 0x120d2e, + 0x11b514, 0x115cfa, 0x1104e0, 0x10acc6, 0x1054ac, 0x0ffc92, + 0x0fa478, 0x0f4c5e, 0x0ef444, 0x0e9c2a, 0x0e4410, 0x0debf6, + 0x0d93dc, 0x0d3bc2, 0x0ce3a8, 0x0c8b8e, 0x0c3374, 0x0bdb5a, + 0x0b8340, 0x0b2b26, 0x0ad30c, 0x0a7af2, 0x0a22d8, 0x09cabe, + 0x0972a4, 0x091a8a, 0x08c270, 0x086a56, 0x08123c, 0x07ba22, + 0x076208, 0x0709ee, 0x06b1d4, 0x0659ba, 0x0601a0, 0x05a986, + 0x05516c, 0x04f952, 0x04a138, 0x04491e, 0x03f104, 0x0398ea, + 0x0340d0, 0x02e8b6, 0x02909c, 0x023882, 0x01e068, 0x01884e, + 0x013034, 0x00d81a, 0x008000, 0x0027e6, -0x003034, -0x00884e, + -0x00e068, -0x013882, -0x01909c, -0x01e8b6, -0x0240d0, -0x0298ea, + -0x02f104, -0x03491e, -0x03a138, -0x03f952, -0x04516c, -0x04a986, + -0x0501a0, -0x0559ba, -0x05b1d4, -0x0609ee, -0x066208, -0x06ba22, + -0x07123c, -0x076a56, -0x07c270, -0x081a8a, -0x0872a4, -0x08cabe, + -0x0922d8, -0x097af2, -0x09d30c, -0x0a2b26, -0x0a8340, -0x0adb5a, + -0x0b3374, -0x0b8b8e, -0x0be3a8, -0x0c3bc2, -0x0c93dc, -0x0cebf6, + -0x0d4410, -0x0d9c2a, -0x0df444, -0x0e4c5e, -0x0ea478, -0x0efc92, + -0x0f54ac, -0x0facc6, -0x1004e0, -0x105cfa, -0x10b514, -0x110d2e, + -0x116548, -0x11bd62, -0x12157c, -0x126d96, -0x12c5b0, -0x131dca, + -0x1375e4, -0x13cdfe, -0x142618, -0x147e32, -0x14d64c, -0x152e66, + -0x158680, -0x15de9a, -0x1636b4, -0x168ece, -0x16e6e8, -0x173f02, + -0x17971c, -0x17ef36, -0x184750, -0x189f6a, -0x18f784, -0x194f9e, + -0x19a7b8, -0x19ffd2, -0x1a57ec, -0x1ab006, -0x1b0820, -0x1b603a, + -0x1bb854, -0x1c106e, -0x1c6888, -0x1cc0a2, -0x1d18bc, -0x1d70d6, + -0x1dc8f0, -0x1e210a, -0x1e7924, -0x1ed13e, -0x1f2958, -0x1f8172, + -0x1fd98c, -0x2031a6, -0x2089c0, -0x20e1da, -0x2139f4, -0x21920e, + -0x21ea28, -0x224242, -0x229a5c, -0x22f276, -0x234a90, -0x23a2aa, + -0x23fac4, -0x2452de, -0x24aaf8, -0x250312, -0x255b2c, -0x25b346, + -0x260b60, -0x26637a, -0x26bb94, -0x2713ae, -0x276bc8, -0x27c3e2, + -0x281bfc, -0x287416, -0x28cc30, -0x29244a, -0x297c64, -0x29d47e, + -0x2a2c98, -0x2a84b2, -0x2adccc, -0x2b34e6, + }; + + +/* We assume that right shift corresponds to signed division by 2 with + * rounding towards minus infinity. This is correct for typical "arithmetic + * shift" instructions that shift in copies of the sign bit. But some + * C compilers implement >> with an unsigned shift. For these machines you + * must define RIGHT_SHIFT_IS_UNSIGNED. + * RIGHT_SHIFT provides a proper signed right shift of an INT32 quantity. + * It is only applied with constant shift counts. SHIFT_TEMPS must be + * included in the variables of any routine using RIGHT_SHIFT. + */ + +#ifdef RIGHT_SHIFT_IS_UNSIGNED +#define SHIFT_TEMPS INT32 shift_temp; +#define RIGHT_SHIFT(x,shft) \ + ((shift_temp = (x)) < 0 ? \ + (shift_temp >> (shft)) | ((~((INT32) 0)) << (32-(shft))) : \ + (shift_temp >> (shft))) +#else +#define SHIFT_TEMPS +#define RIGHT_SHIFT(x,shft) ((x) >> (shft)) +#endif + + +METHODDEF(void) +ycc_rgb_convert_argb (j_decompress_ptr cinfo, + JSAMPIMAGE input_buf, JDIMENSION input_row, + JSAMPARRAY output_buf, int num_rows) +{ + JDIMENSION num_cols = cinfo->output_width; + JSAMPLE * range_limit = cinfo->sample_range_limit; + + SHIFT_TEMPS + + /* This is used if we don't have SSE2 */ + + while (--num_rows >= 0) { + JSAMPROW inptr0 = input_buf[0][input_row]; + JSAMPROW inptr1 = input_buf[1][input_row]; + JSAMPROW inptr2 = input_buf[2][input_row]; + input_row++; + uint32_t *outptr = (uint32_t *) *output_buf++; + for (JDIMENSION col = 0; col < num_cols; col++) { + int y = GETJSAMPLE(inptr0[col]); + int cb = GETJSAMPLE(inptr1[col]); + int cr = GETJSAMPLE(inptr2[col]); + JSAMPLE * range_limit_y = range_limit + y; + /* Range-limiting is essential due to noise introduced by DCT losses. */ + outptr[col] = 0xFF000000 | + ( range_limit_y[Cr_r_tab[cr]] << 16 ) | + ( range_limit_y[((int) RIGHT_SHIFT(Cb_g_tab[cb] + Cr_g_tab[cr], SCALEBITS))] << 8 ) | + ( range_limit_y[Cb_b_tab[cb]] ); + } + } +} +#endif + + /**************** Inverted CMYK -> RGB conversion **************/ /* * Input is (Inverted) CMYK stored as 4 bytes per pixel.