From 3d8736f9d710375ede5100d5ba7dc9028e729ffe Mon Sep 17 00:00:00 2001 From: "Admiral H. Curtiss" Date: Sat, 12 Dec 2020 16:57:24 +0100 Subject: [PATCH] Update bundled libpng to version 1.6.37 --- Externals/libpng/CMakeLists.txt | 181 +- Externals/libpng/LICENSE | 134 ++ Externals/libpng/arm/arm_init.c | 136 ++ Externals/libpng/arm/filter_neon.S | 253 +++ Externals/libpng/arm/filter_neon_intrinsics.c | 402 ++++ .../libpng/arm/palette_neon_intrinsics.c | 149 ++ .../libpng/intel/filter_sse2_intrinsics.c | 391 ++++ Externals/libpng/intel/intel_init.c | 52 + Externals/libpng/mips/filter_msa_intrinsics.c | 808 +++++++ Externals/libpng/mips/mips_init.c | 130 ++ Externals/libpng/png.c | 227 +- Externals/libpng/png.h | 378 ++-- Externals/libpng/pngconf.h | 45 +- Externals/libpng/pngdebug.h | 6 +- Externals/libpng/pngerror.c | 18 +- Externals/libpng/pngget.c | 50 +- Externals/libpng/pnginfo.h | 26 +- Externals/libpng/pnglibconf.h | 12 +- Externals/libpng/pngmem.c | 6 +- Externals/libpng/pngpread.c | 53 +- Externals/libpng/pngpriv.h | 145 +- Externals/libpng/pngread.c | 82 +- Externals/libpng/pngrio.c | 16 +- Externals/libpng/pngrtran.c | 1985 +++++++++-------- Externals/libpng/pngrutil.c | 305 ++- Externals/libpng/pngset.c | 83 +- Externals/libpng/pngstruct.h | 32 +- Externals/libpng/pngtest.c | 135 +- Externals/libpng/pngtrans.c | 64 +- Externals/libpng/pngwio.c | 14 +- Externals/libpng/pngwrite.c | 60 +- Externals/libpng/pngwtran.c | 19 +- Externals/libpng/pngwutil.c | 223 +- .../libpng/powerpc/filter_vsx_intrinsics.c | 768 +++++++ Externals/libpng/powerpc/powerpc_init.c | 126 ++ 35 files changed, 5783 insertions(+), 1731 deletions(-) create mode 100644 Externals/libpng/LICENSE create mode 100644 Externals/libpng/arm/arm_init.c create mode 100644 Externals/libpng/arm/filter_neon.S create mode 100644 Externals/libpng/arm/filter_neon_intrinsics.c create mode 100644 Externals/libpng/arm/palette_neon_intrinsics.c create mode 100644 Externals/libpng/intel/filter_sse2_intrinsics.c create mode 100644 Externals/libpng/intel/intel_init.c create mode 100644 Externals/libpng/mips/filter_msa_intrinsics.c create mode 100644 Externals/libpng/mips/mips_init.c create mode 100644 Externals/libpng/powerpc/filter_vsx_intrinsics.c create mode 100644 Externals/libpng/powerpc/powerpc_init.c diff --git a/Externals/libpng/CMakeLists.txt b/Externals/libpng/CMakeLists.txt index 824492e285..d0da8580a6 100644 --- a/Externals/libpng/CMakeLists.txt +++ b/Externals/libpng/CMakeLists.txt @@ -1,6 +1,27 @@ +# selectively extracted and adapted from the libpng CMakeLists.txt, which has the following copyright notice: + + +# Copyright (C) 2018 Cosmin Truta +# Copyright (C) 2007,2009-2018 Glenn Randers-Pehrson +# Written by Christian Ehrlicher, 2007 +# Revised by Roger Lowman, 2009-2010 +# Revised by Clifford Yapp, 2011-2012,2017 +# Revised by Roger Leigh, 2016 +# Revised by Andreas Franek, 2016 +# Revised by Sam Serrels, 2017 +# Revised by Vadim Barkov, 2017 +# Revised by Vicky Pfau, 2018 +# Revised by Cameron Cawley, 2018 +# Revised by Cosmin Truta, 2018 +# Revised by Kyle Bentley, 2018 + +# This code is released under the libpng license. +# For conditions of distribution and use, see the disclaimer +# and license in png.h + + + add_library(png STATIC - png.h - pngconf.h png.c pngerror.c pngget.c @@ -18,16 +39,164 @@ add_library(png STATIC pngwutil.c ) -target_include_directories(png -PUBLIC +option(PNG_HARDWARE_OPTIMIZATIONS "Enable hardware optimizations for libpng" OFF) + +if(PNG_HARDWARE_OPTIMIZATIONS) + + message(STATUS "CMAKE_SYSTEM_PROCESSOR: ${CMAKE_SYSTEM_PROCESSOR}") + + # set definitions and sources for arm + if(CMAKE_SYSTEM_PROCESSOR MATCHES "^arm" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "^aarch64") + set(PNG_ARM_NEON_POSSIBLE_VALUES check on off) + set(PNG_ARM_NEON "check" CACHE STRING "Enable ARM NEON optimizations: + check: (default) use internal checking code; + off: disable the optimizations; + on: turn on unconditionally.") + set_property(CACHE PNG_ARM_NEON PROPERTY STRINGS + ${PNG_ARM_NEON_POSSIBLE_VALUES}) + list(FIND PNG_ARM_NEON_POSSIBLE_VALUES ${PNG_ARM_NEON} index) + if(index EQUAL -1) + message(FATAL_ERROR + "PNG_ARM_NEON must be one of [${PNG_ARM_NEON_POSSIBLE_VALUES}]") + elseif(NOT ${PNG_ARM_NEON} STREQUAL "off") + target_sources(png PRIVATE + arm/arm_init.c + arm/filter_neon.S + arm/filter_neon_intrinsics.c + arm/palette_neon_intrinsics.c) + + if(${PNG_ARM_NEON} STREQUAL "on") + target_compile_definitions(png PUBLIC -DPNG_ARM_NEON_OPT=2) + elseif(${PNG_ARM_NEON} STREQUAL "check") + target_compile_definitions(png PUBLIC -DPNG_ARM_NEON_CHECK_SUPPORTED) + endif() + else() + target_compile_definitions(png PUBLIC -DPNG_ARM_NEON_OPT=0) + endif() + endif() + + # set definitions and sources for powerpc + if(CMAKE_SYSTEM_PROCESSOR MATCHES "^powerpc*" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "^ppc64*") + set(PNG_POWERPC_VSX_POSSIBLE_VALUES on off) + set(PNG_POWERPC_VSX "on" CACHE STRING "Enable POWERPC VSX optimizations: + off: disable the optimizations.") + set_property(CACHE PNG_POWERPC_VSX PROPERTY STRINGS + ${PNG_POWERPC_VSX_POSSIBLE_VALUES}) + list(FIND PNG_POWERPC_VSX_POSSIBLE_VALUES ${PNG_POWERPC_VSX} index) + if(index EQUAL -1) + message(FATAL_ERROR + "PNG_POWERPC_VSX must be one of [${PNG_POWERPC_VSX_POSSIBLE_VALUES}]") + elseif(NOT ${PNG_POWERPC_VSX} STREQUAL "off") + target_sources(png PRIVATE + powerpc/powerpc_init.c + powerpc/filter_vsx_intrinsics.c) + if(${PNG_POWERPC_VSX} STREQUAL "on") + target_compile_definitions(png PUBLIC -DPNG_POWERPC_VSX_OPT=2) + endif() + else() + target_compile_definitions(png PUBLIC -DPNG_POWERPC_VSX_OPT=0) + endif() + endif() + + # set definitions and sources for intel + if(CMAKE_SYSTEM_PROCESSOR MATCHES "^i?86" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "^x86_64*") + set(PNG_INTEL_SSE_POSSIBLE_VALUES on off) + set(PNG_INTEL_SSE "on" CACHE STRING "Enable INTEL_SSE optimizations: + off: disable the optimizations") + set_property(CACHE PNG_INTEL_SSE PROPERTY STRINGS + ${PNG_INTEL_SSE_POSSIBLE_VALUES}) + list(FIND PNG_INTEL_SSE_POSSIBLE_VALUES ${PNG_INTEL_SSE} index) + if(index EQUAL -1) + message(FATAL_ERROR + "PNG_INTEL_SSE must be one of [${PNG_INTEL_SSE_POSSIBLE_VALUES}]") + elseif(NOT ${PNG_INTEL_SSE} STREQUAL "off") + target_sources(png PRIVATE + intel/intel_init.c + intel/filter_sse2_intrinsics.c) + if(${PNG_INTEL_SSE} STREQUAL "on") + target_compile_definitions(png PUBLIC -DPNG_INTEL_SSE_OPT=1) + endif() + else() + target_compile_definitions(png PUBLIC -DPNG_INTEL_SSE_OPT=0) + endif() + endif() + + # set definitions and sources for MIPS + if(CMAKE_SYSTEM_PROCESSOR MATCHES "mipsel*" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "mips64el*") + set(PNG_MIPS_MSA_POSSIBLE_VALUES on off) + set(PNG_MIPS_MSA "on" CACHE STRING "Enable MIPS_MSA optimizations: + off: disable the optimizations") + set_property(CACHE PNG_MIPS_MSA PROPERTY STRINGS + ${PNG_MIPS_MSA_POSSIBLE_VALUES}) + list(FIND PNG_MIPS_MSA_POSSIBLE_VALUES ${PNG_MIPS_MSA} index) + if(index EQUAL -1) + message(FATAL_ERROR + "PNG_MIPS_MSA must be one of [${PNG_MIPS_MSA_POSSIBLE_VALUES}]") + elseif(NOT ${PNG_MIPS_MSA} STREQUAL "off") + target_sources(png PRIVATE + mips/mips_init.c + mips/filter_msa_intrinsics.c) + if(${PNG_MIPS_MSA} STREQUAL "on") + target_compile_definitions(png PUBLIC -DPNG_MIPS_MSA_OPT=2) + endif() + else() + target_compile_definitions(png PUBLIC -DPNG_MIPS_MSA_OPT=0) + endif() + endif() + +else(PNG_HARDWARE_OPTIMIZATIONS) + + # set definitions and sources for arm + if(CMAKE_SYSTEM_PROCESSOR MATCHES "^arm" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "^aarch64") + target_compile_definitions(png PUBLIC -DPNG_ARM_NEON_OPT=0) + endif() + + # set definitions and sources for powerpc + if(CMAKE_SYSTEM_PROCESSOR MATCHES "^powerpc*" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "^ppc64*") + target_compile_definitions(png PUBLIC -DPNG_POWERPC_VSX_OPT=0) + endif() + + # set definitions and sources for intel + if(CMAKE_SYSTEM_PROCESSOR MATCHES "^i?86" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "^x86_64*") + target_compile_definitions(png PUBLIC -DPNG_INTEL_SSE_OPT=0) + endif() + + # set definitions and sources for MIPS + if(CMAKE_SYSTEM_PROCESSOR MATCHES "mipsel*" OR + CMAKE_SYSTEM_PROCESSOR MATCHES "mips64el*") + target_compile_definitions(png PUBLIC -DPNG_MIPS_MSA_OPT=0) + endif() + +endif(PNG_HARDWARE_OPTIMIZATIONS) + +target_sources(png PRIVATE + # public headers + png.h + pngconf.h + pnglibconf.h + + # private headers + pngpriv.h + pngdebug.h + pnginfo.h + pngstruct.h +) + +target_include_directories(png PUBLIC ${CMAKE_CURRENT_SOURCE_DIR} ) target_link_libraries(png PUBLIC ZLIB::ZLIB) if(NOT MSVC) - target_compile_options(png - PRIVATE + target_compile_options(png PRIVATE -Wno-self-assign ) endif() diff --git a/Externals/libpng/LICENSE b/Externals/libpng/LICENSE new file mode 100644 index 0000000000..e0c5b531cf --- /dev/null +++ b/Externals/libpng/LICENSE @@ -0,0 +1,134 @@ +COPYRIGHT NOTICE, DISCLAIMER, and LICENSE +========================================= + +PNG Reference Library License version 2 +--------------------------------------- + + * Copyright (c) 1995-2019 The PNG Reference Library Authors. + * Copyright (c) 2018-2019 Cosmin Truta. + * Copyright (c) 2000-2002, 2004, 2006-2018 Glenn Randers-Pehrson. + * Copyright (c) 1996-1997 Andreas Dilger. + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. + +The software is supplied "as is", without warranty of any kind, +express or implied, including, without limitation, the warranties +of merchantability, fitness for a particular purpose, title, and +non-infringement. In no event shall the Copyright owners, or +anyone distributing the software, be liable for any damages or +other liability, whether in contract, tort or otherwise, arising +from, out of, or in connection with the software, or the use or +other dealings in the software, even if advised of the possibility +of such damage. + +Permission is hereby granted to use, copy, modify, and distribute +this software, or portions hereof, for any purpose, without fee, +subject to the following restrictions: + + 1. The origin of this software must not be misrepresented; you + must not claim that you wrote the original software. If you + use this software in a product, an acknowledgment in the product + documentation would be appreciated, but is not required. + + 2. Altered source versions must be plainly marked as such, and must + not be misrepresented as being the original software. + + 3. This Copyright notice may not be removed or altered from any + source or altered source distribution. + + +PNG Reference Library License version 1 (for libpng 0.5 through 1.6.35) +----------------------------------------------------------------------- + +libpng versions 1.0.7, July 1, 2000, through 1.6.35, July 15, 2018 are +Copyright (c) 2000-2002, 2004, 2006-2018 Glenn Randers-Pehrson, are +derived from libpng-1.0.6, and are distributed according to the same +disclaimer and license as libpng-1.0.6 with the following individuals +added to the list of Contributing Authors: + + Simon-Pierre Cadieux + Eric S. Raymond + Mans Rullgard + Cosmin Truta + Gilles Vollant + James Yu + Mandar Sahastrabuddhe + Google Inc. + Vadim Barkov + +and with the following additions to the disclaimer: + + There is no warranty against interference with your enjoyment of + the library or against infringement. There is no warranty that our + efforts or the library will fulfill any of your particular purposes + or needs. This library is provided with all faults, and the entire + risk of satisfactory quality, performance, accuracy, and effort is + with the user. + +Some files in the "contrib" directory and some configure-generated +files that are distributed with libpng have other copyright owners, and +are released under other open source licenses. + +libpng versions 0.97, January 1998, through 1.0.6, March 20, 2000, are +Copyright (c) 1998-2000 Glenn Randers-Pehrson, are derived from +libpng-0.96, and are distributed according to the same disclaimer and +license as libpng-0.96, with the following individuals added to the +list of Contributing Authors: + + Tom Lane + Glenn Randers-Pehrson + Willem van Schaik + +libpng versions 0.89, June 1996, through 0.96, May 1997, are +Copyright (c) 1996-1997 Andreas Dilger, are derived from libpng-0.88, +and are distributed according to the same disclaimer and license as +libpng-0.88, with the following individuals added to the list of +Contributing Authors: + + John Bowler + Kevin Bracey + Sam Bushell + Magnus Holmgren + Greg Roelofs + Tom Tanner + +Some files in the "scripts" directory have other copyright owners, +but are released under this license. + +libpng versions 0.5, May 1995, through 0.88, January 1996, are +Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. + +For the purposes of this copyright and license, "Contributing Authors" +is defined as the following set of individuals: + + Andreas Dilger + Dave Martindale + Guy Eric Schalnat + Paul Schmidt + Tim Wegner + +The PNG Reference Library is supplied "AS IS". The Contributing +Authors and Group 42, Inc. disclaim all warranties, expressed or +implied, including, without limitation, the warranties of +merchantability and of fitness for any purpose. The Contributing +Authors and Group 42, Inc. assume no liability for direct, indirect, +incidental, special, exemplary, or consequential damages, which may +result from the use of the PNG Reference Library, even if advised of +the possibility of such damage. + +Permission is hereby granted to use, copy, modify, and distribute this +source code, or portions hereof, for any purpose, without fee, subject +to the following restrictions: + + 1. The origin of this source code must not be misrepresented. + + 2. Altered versions must be plainly marked as such and must not + be misrepresented as being the original source. + + 3. This Copyright notice may not be removed or altered from any + source or altered source distribution. + +The Contributing Authors and Group 42, Inc. specifically permit, +without fee, and encourage the use of this source code as a component +to supporting the PNG file format in commercial products. If you use +this source code in a product, acknowledgment is not required but would +be appreciated. diff --git a/Externals/libpng/arm/arm_init.c b/Externals/libpng/arm/arm_init.c new file mode 100644 index 0000000000..a34ecdbef7 --- /dev/null +++ b/Externals/libpng/arm/arm_init.c @@ -0,0 +1,136 @@ + +/* arm_init.c - NEON optimised filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2014,2016 Glenn Randers-Pehrson + * Written by Mans Rullgard, 2011. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +/* Below, after checking __linux__, various non-C90 POSIX 1003.1 functions are + * called. + */ +#define _POSIX_SOURCE 1 + +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED + +#if PNG_ARM_NEON_OPT > 0 +#ifdef PNG_ARM_NEON_CHECK_SUPPORTED /* Do run-time checks */ +/* WARNING: it is strongly recommended that you do not build libpng with + * run-time checks for CPU features if at all possible. In the case of the ARM + * NEON instructions there is no processor-specific way of detecting the + * presence of the required support, therefore run-time detection is extremely + * OS specific. + * + * You may set the macro PNG_ARM_NEON_FILE to the file name of file containing + * a fragment of C source code which defines the png_have_neon function. There + * are a number of implementations in contrib/arm-neon, but the only one that + * has partial support is contrib/arm-neon/linux.c - a generic Linux + * implementation which reads /proc/cpufino. + */ +#ifndef PNG_ARM_NEON_FILE +# ifdef __linux__ +# define PNG_ARM_NEON_FILE "contrib/arm-neon/linux.c" +# endif +#endif + +#ifdef PNG_ARM_NEON_FILE + +#include /* for sig_atomic_t */ +static int png_have_neon(png_structp png_ptr); +#include PNG_ARM_NEON_FILE + +#else /* PNG_ARM_NEON_FILE */ +# error "PNG_ARM_NEON_FILE undefined: no support for run-time ARM NEON checks" +#endif /* PNG_ARM_NEON_FILE */ +#endif /* PNG_ARM_NEON_CHECK_SUPPORTED */ + +#ifndef PNG_ALIGNED_MEMORY_SUPPORTED +# error "ALIGNED_MEMORY is required; set: -DPNG_ALIGNED_MEMORY_SUPPORTED" +#endif + +void +png_init_filter_functions_neon(png_structp pp, unsigned int bpp) +{ + /* The switch statement is compiled in for ARM_NEON_API, the call to + * png_have_neon is compiled in for ARM_NEON_CHECK. If both are defined + * the check is only performed if the API has not set the NEON option on + * or off explicitly. In this case the check controls what happens. + * + * If the CHECK is not compiled in and the option is UNSET the behavior prior + * to 1.6.7 was to use the NEON code - this was a bug caused by having the + * wrong order of the 'ON' and 'default' cases. UNSET now defaults to OFF, + * as documented in png.h + */ + png_debug(1, "in png_init_filter_functions_neon"); +#ifdef PNG_ARM_NEON_API_SUPPORTED + switch ((pp->options >> PNG_ARM_NEON) & 3) + { + case PNG_OPTION_UNSET: + /* Allow the run-time check to execute if it has been enabled - + * thus both API and CHECK can be turned on. If it isn't supported + * this case will fall through to the 'default' below, which just + * returns. + */ +#endif /* PNG_ARM_NEON_API_SUPPORTED */ +#ifdef PNG_ARM_NEON_CHECK_SUPPORTED + { + static volatile sig_atomic_t no_neon = -1; /* not checked */ + + if (no_neon < 0) + no_neon = !png_have_neon(pp); + + if (no_neon) + return; + } +#ifdef PNG_ARM_NEON_API_SUPPORTED + break; +#endif +#endif /* PNG_ARM_NEON_CHECK_SUPPORTED */ + +#ifdef PNG_ARM_NEON_API_SUPPORTED + default: /* OFF or INVALID */ + return; + + case PNG_OPTION_ON: + /* Option turned on */ + break; + } +#endif + + /* IMPORTANT: any new external functions used here must be declared using + * PNG_INTERNAL_FUNCTION in ../pngpriv.h. This is required so that the + * 'prefix' option to configure works: + * + * ./configure --with-libpng-prefix=foobar_ + * + * Verify you have got this right by running the above command, doing a build + * and examining pngprefix.h; it must contain a #define for every external + * function you add. (Notice that this happens automatically for the + * initialization function.) + */ + pp->read_filter[PNG_FILTER_VALUE_UP-1] = png_read_filter_row_up_neon; + + if (bpp == 3) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub3_neon; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg3_neon; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = + png_read_filter_row_paeth3_neon; + } + + else if (bpp == 4) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub4_neon; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg4_neon; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = + png_read_filter_row_paeth4_neon; + } +} +#endif /* PNG_ARM_NEON_OPT > 0 */ +#endif /* READ */ diff --git a/Externals/libpng/arm/filter_neon.S b/Externals/libpng/arm/filter_neon.S new file mode 100644 index 0000000000..2308aad13e --- /dev/null +++ b/Externals/libpng/arm/filter_neon.S @@ -0,0 +1,253 @@ + +/* filter_neon.S - NEON optimised filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2014,2017 Glenn Randers-Pehrson + * Written by Mans Rullgard, 2011. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +/* This is required to get the symbol renames, which are #defines, and the + * definitions (or not) of PNG_ARM_NEON_OPT and PNG_ARM_NEON_IMPLEMENTATION. + */ +#define PNG_VERSION_INFO_ONLY +#include "../pngpriv.h" + +#if (defined(__linux__) || defined(__FreeBSD__)) && defined(__ELF__) +.section .note.GNU-stack,"",%progbits /* mark stack as non-executable */ +#endif + +#ifdef PNG_READ_SUPPORTED + +/* Assembler NEON support - only works for 32-bit ARM (i.e. it does not work for + * ARM64). The code in arm/filter_neon_intrinsics.c supports ARM64, however it + * only works if -mfpu=neon is specified on the GCC command line. See pngpriv.h + * for the logic which sets PNG_USE_ARM_NEON_ASM: + */ +#if PNG_ARM_NEON_IMPLEMENTATION == 2 /* hand-coded assembler */ + +#if PNG_ARM_NEON_OPT > 0 + +#ifdef __ELF__ +# define ELF +#else +# define ELF @ +#endif + + .arch armv7-a + .fpu neon + +.macro func name, export=0 + .macro endfunc +ELF .size \name, . - \name + .endfunc + .purgem endfunc + .endm + .text + + /* Explicitly specifying alignment here because some versions of + * GAS don't align code correctly. This is harmless in correctly + * written versions of GAS. + */ + .align 2 + + .if \export + .global \name + .endif +ELF .type \name, STT_FUNC + .func \name +\name: +.endm + +func png_read_filter_row_sub4_neon, export=1 + ldr r3, [r0, #4] @ rowbytes + vmov.i8 d3, #0 +1: + vld4.32 {d4[],d5[],d6[],d7[]}, [r1,:128] + vadd.u8 d0, d3, d4 + vadd.u8 d1, d0, d5 + vadd.u8 d2, d1, d6 + vadd.u8 d3, d2, d7 + vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1,:128]! + subs r3, r3, #16 + bgt 1b + + bx lr +endfunc + +func png_read_filter_row_sub3_neon, export=1 + ldr r3, [r0, #4] @ rowbytes + vmov.i8 d3, #0 + mov r0, r1 + mov r2, #3 + mov r12, #12 + vld1.8 {q11}, [r0], r12 +1: + vext.8 d5, d22, d23, #3 + vadd.u8 d0, d3, d22 + vext.8 d6, d22, d23, #6 + vadd.u8 d1, d0, d5 + vext.8 d7, d23, d23, #1 + vld1.8 {q11}, [r0], r12 + vst1.32 {d0[0]}, [r1,:32], r2 + vadd.u8 d2, d1, d6 + vst1.32 {d1[0]}, [r1], r2 + vadd.u8 d3, d2, d7 + vst1.32 {d2[0]}, [r1], r2 + vst1.32 {d3[0]}, [r1], r2 + subs r3, r3, #12 + bgt 1b + + bx lr +endfunc + +func png_read_filter_row_up_neon, export=1 + ldr r3, [r0, #4] @ rowbytes +1: + vld1.8 {q0}, [r1,:128] + vld1.8 {q1}, [r2,:128]! + vadd.u8 q0, q0, q1 + vst1.8 {q0}, [r1,:128]! + subs r3, r3, #16 + bgt 1b + + bx lr +endfunc + +func png_read_filter_row_avg4_neon, export=1 + ldr r12, [r0, #4] @ rowbytes + vmov.i8 d3, #0 +1: + vld4.32 {d4[],d5[],d6[],d7[]}, [r1,:128] + vld4.32 {d16[],d17[],d18[],d19[]},[r2,:128]! + vhadd.u8 d0, d3, d16 + vadd.u8 d0, d0, d4 + vhadd.u8 d1, d0, d17 + vadd.u8 d1, d1, d5 + vhadd.u8 d2, d1, d18 + vadd.u8 d2, d2, d6 + vhadd.u8 d3, d2, d19 + vadd.u8 d3, d3, d7 + vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1,:128]! + subs r12, r12, #16 + bgt 1b + + bx lr +endfunc + +func png_read_filter_row_avg3_neon, export=1 + push {r4,lr} + ldr r12, [r0, #4] @ rowbytes + vmov.i8 d3, #0 + mov r0, r1 + mov r4, #3 + mov lr, #12 + vld1.8 {q11}, [r0], lr +1: + vld1.8 {q10}, [r2], lr + vext.8 d5, d22, d23, #3 + vhadd.u8 d0, d3, d20 + vext.8 d17, d20, d21, #3 + vadd.u8 d0, d0, d22 + vext.8 d6, d22, d23, #6 + vhadd.u8 d1, d0, d17 + vext.8 d18, d20, d21, #6 + vadd.u8 d1, d1, d5 + vext.8 d7, d23, d23, #1 + vld1.8 {q11}, [r0], lr + vst1.32 {d0[0]}, [r1,:32], r4 + vhadd.u8 d2, d1, d18 + vst1.32 {d1[0]}, [r1], r4 + vext.8 d19, d21, d21, #1 + vadd.u8 d2, d2, d6 + vhadd.u8 d3, d2, d19 + vst1.32 {d2[0]}, [r1], r4 + vadd.u8 d3, d3, d7 + vst1.32 {d3[0]}, [r1], r4 + subs r12, r12, #12 + bgt 1b + + pop {r4,pc} +endfunc + +.macro paeth rx, ra, rb, rc + vaddl.u8 q12, \ra, \rb @ a + b + vaddl.u8 q15, \rc, \rc @ 2*c + vabdl.u8 q13, \rb, \rc @ pa + vabdl.u8 q14, \ra, \rc @ pb + vabd.u16 q15, q12, q15 @ pc + vcle.u16 q12, q13, q14 @ pa <= pb + vcle.u16 q13, q13, q15 @ pa <= pc + vcle.u16 q14, q14, q15 @ pb <= pc + vand q12, q12, q13 @ pa <= pb && pa <= pc + vmovn.u16 d28, q14 + vmovn.u16 \rx, q12 + vbsl d28, \rb, \rc + vbsl \rx, \ra, d28 +.endm + +func png_read_filter_row_paeth4_neon, export=1 + ldr r12, [r0, #4] @ rowbytes + vmov.i8 d3, #0 + vmov.i8 d20, #0 +1: + vld4.32 {d4[],d5[],d6[],d7[]}, [r1,:128] + vld4.32 {d16[],d17[],d18[],d19[]},[r2,:128]! + paeth d0, d3, d16, d20 + vadd.u8 d0, d0, d4 + paeth d1, d0, d17, d16 + vadd.u8 d1, d1, d5 + paeth d2, d1, d18, d17 + vadd.u8 d2, d2, d6 + paeth d3, d2, d19, d18 + vmov d20, d19 + vadd.u8 d3, d3, d7 + vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1,:128]! + subs r12, r12, #16 + bgt 1b + + bx lr +endfunc + +func png_read_filter_row_paeth3_neon, export=1 + push {r4,lr} + ldr r12, [r0, #4] @ rowbytes + vmov.i8 d3, #0 + vmov.i8 d4, #0 + mov r0, r1 + mov r4, #3 + mov lr, #12 + vld1.8 {q11}, [r0], lr +1: + vld1.8 {q10}, [r2], lr + paeth d0, d3, d20, d4 + vext.8 d5, d22, d23, #3 + vadd.u8 d0, d0, d22 + vext.8 d17, d20, d21, #3 + paeth d1, d0, d17, d20 + vst1.32 {d0[0]}, [r1,:32], r4 + vext.8 d6, d22, d23, #6 + vadd.u8 d1, d1, d5 + vext.8 d18, d20, d21, #6 + paeth d2, d1, d18, d17 + vext.8 d7, d23, d23, #1 + vld1.8 {q11}, [r0], lr + vst1.32 {d1[0]}, [r1], r4 + vadd.u8 d2, d2, d6 + vext.8 d19, d21, d21, #1 + paeth d3, d2, d19, d18 + vst1.32 {d2[0]}, [r1], r4 + vmov d4, d19 + vadd.u8 d3, d3, d7 + vst1.32 {d3[0]}, [r1], r4 + subs r12, r12, #12 + bgt 1b + + pop {r4,pc} +endfunc +#endif /* PNG_ARM_NEON_OPT > 0 */ +#endif /* PNG_ARM_NEON_IMPLEMENTATION == 2 (assembler) */ +#endif /* READ */ diff --git a/Externals/libpng/arm/filter_neon_intrinsics.c b/Externals/libpng/arm/filter_neon_intrinsics.c new file mode 100644 index 0000000000..553c0be21c --- /dev/null +++ b/Externals/libpng/arm/filter_neon_intrinsics.c @@ -0,0 +1,402 @@ + +/* filter_neon_intrinsics.c - NEON optimised filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2014,2016 Glenn Randers-Pehrson + * Written by James Yu , October 2013. + * Based on filter_neon.S, written by Mans Rullgard, 2011. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED + +/* This code requires -mfpu=neon on the command line: */ +#if PNG_ARM_NEON_IMPLEMENTATION == 1 /* intrinsics code from pngpriv.h */ + +#if defined(_MSC_VER) && defined(_M_ARM64) +# include +#else +# include +#endif + +/* libpng row pointers are not necessarily aligned to any particular boundary, + * however this code will only work with appropriate alignment. arm/arm_init.c + * checks for this (and will not compile unless it is done). This code uses + * variants of png_aligncast to avoid compiler warnings. + */ +#define png_ptr(type,pointer) png_aligncast(type *,pointer) +#define png_ptrc(type,pointer) png_aligncastconst(const type *,pointer) + +/* The following relies on a variable 'temp_pointer' being declared with type + * 'type'. This is written this way just to hide the GCC strict aliasing + * warning; note that the code is safe because there never is an alias between + * the input and output pointers. + * + * When compiling with MSVC ARM64, the png_ldr macro can't be passed directly + * to vst4_lane_u32, because of an internal compiler error inside MSVC. + * To avoid this compiler bug, we use a temporary variable (vdest_val) to store + * the result of png_ldr. + */ +#define png_ldr(type,pointer)\ + (temp_pointer = png_ptr(type,pointer), *temp_pointer) + +#if PNG_ARM_NEON_OPT > 0 + +void +png_read_filter_row_up_neon(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_bytep rp = row; + png_bytep rp_stop = row + row_info->rowbytes; + png_const_bytep pp = prev_row; + + png_debug(1, "in png_read_filter_row_up_neon"); + + for (; rp < rp_stop; rp += 16, pp += 16) + { + uint8x16_t qrp, qpp; + + qrp = vld1q_u8(rp); + qpp = vld1q_u8(pp); + qrp = vaddq_u8(qrp, qpp); + vst1q_u8(rp, qrp); + } +} + +void +png_read_filter_row_sub3_neon(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_bytep rp = row; + png_bytep rp_stop = row + row_info->rowbytes; + + uint8x16_t vtmp = vld1q_u8(rp); + uint8x8x2_t *vrpt = png_ptr(uint8x8x2_t, &vtmp); + uint8x8x2_t vrp = *vrpt; + + uint8x8x4_t vdest; + vdest.val[3] = vdup_n_u8(0); + + png_debug(1, "in png_read_filter_row_sub3_neon"); + + for (; rp < rp_stop;) + { + uint8x8_t vtmp1, vtmp2; + uint32x2_t *temp_pointer; + + vtmp1 = vext_u8(vrp.val[0], vrp.val[1], 3); + vdest.val[0] = vadd_u8(vdest.val[3], vrp.val[0]); + vtmp2 = vext_u8(vrp.val[0], vrp.val[1], 6); + vdest.val[1] = vadd_u8(vdest.val[0], vtmp1); + + vtmp1 = vext_u8(vrp.val[1], vrp.val[1], 1); + vdest.val[2] = vadd_u8(vdest.val[1], vtmp2); + vdest.val[3] = vadd_u8(vdest.val[2], vtmp1); + + vtmp = vld1q_u8(rp + 12); + vrpt = png_ptr(uint8x8x2_t, &vtmp); + vrp = *vrpt; + + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[0]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[1]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[2]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[3]), 0); + rp += 3; + } + + PNG_UNUSED(prev_row) +} + +void +png_read_filter_row_sub4_neon(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_bytep rp = row; + png_bytep rp_stop = row + row_info->rowbytes; + + uint8x8x4_t vdest; + vdest.val[3] = vdup_n_u8(0); + + png_debug(1, "in png_read_filter_row_sub4_neon"); + + for (; rp < rp_stop; rp += 16) + { + uint32x2x4_t vtmp = vld4_u32(png_ptr(uint32_t,rp)); + uint8x8x4_t *vrpt = png_ptr(uint8x8x4_t,&vtmp); + uint8x8x4_t vrp = *vrpt; + uint32x2x4_t *temp_pointer; + uint32x2x4_t vdest_val; + + vdest.val[0] = vadd_u8(vdest.val[3], vrp.val[0]); + vdest.val[1] = vadd_u8(vdest.val[0], vrp.val[1]); + vdest.val[2] = vadd_u8(vdest.val[1], vrp.val[2]); + vdest.val[3] = vadd_u8(vdest.val[2], vrp.val[3]); + + vdest_val = png_ldr(uint32x2x4_t, &vdest); + vst4_lane_u32(png_ptr(uint32_t,rp), vdest_val, 0); + } + + PNG_UNUSED(prev_row) +} + +void +png_read_filter_row_avg3_neon(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_bytep rp = row; + png_const_bytep pp = prev_row; + png_bytep rp_stop = row + row_info->rowbytes; + + uint8x16_t vtmp; + uint8x8x2_t *vrpt; + uint8x8x2_t vrp; + uint8x8x4_t vdest; + vdest.val[3] = vdup_n_u8(0); + + vtmp = vld1q_u8(rp); + vrpt = png_ptr(uint8x8x2_t,&vtmp); + vrp = *vrpt; + + png_debug(1, "in png_read_filter_row_avg3_neon"); + + for (; rp < rp_stop; pp += 12) + { + uint8x8_t vtmp1, vtmp2, vtmp3; + + uint8x8x2_t *vppt; + uint8x8x2_t vpp; + + uint32x2_t *temp_pointer; + + vtmp = vld1q_u8(pp); + vppt = png_ptr(uint8x8x2_t,&vtmp); + vpp = *vppt; + + vtmp1 = vext_u8(vrp.val[0], vrp.val[1], 3); + vdest.val[0] = vhadd_u8(vdest.val[3], vpp.val[0]); + vdest.val[0] = vadd_u8(vdest.val[0], vrp.val[0]); + + vtmp2 = vext_u8(vpp.val[0], vpp.val[1], 3); + vtmp3 = vext_u8(vrp.val[0], vrp.val[1], 6); + vdest.val[1] = vhadd_u8(vdest.val[0], vtmp2); + vdest.val[1] = vadd_u8(vdest.val[1], vtmp1); + + vtmp2 = vext_u8(vpp.val[0], vpp.val[1], 6); + vtmp1 = vext_u8(vrp.val[1], vrp.val[1], 1); + + vtmp = vld1q_u8(rp + 12); + vrpt = png_ptr(uint8x8x2_t,&vtmp); + vrp = *vrpt; + + vdest.val[2] = vhadd_u8(vdest.val[1], vtmp2); + vdest.val[2] = vadd_u8(vdest.val[2], vtmp3); + + vtmp2 = vext_u8(vpp.val[1], vpp.val[1], 1); + + vdest.val[3] = vhadd_u8(vdest.val[2], vtmp2); + vdest.val[3] = vadd_u8(vdest.val[3], vtmp1); + + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[0]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[1]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[2]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[3]), 0); + rp += 3; + } +} + +void +png_read_filter_row_avg4_neon(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_bytep rp = row; + png_bytep rp_stop = row + row_info->rowbytes; + png_const_bytep pp = prev_row; + + uint8x8x4_t vdest; + vdest.val[3] = vdup_n_u8(0); + + png_debug(1, "in png_read_filter_row_avg4_neon"); + + for (; rp < rp_stop; rp += 16, pp += 16) + { + uint32x2x4_t vtmp; + uint8x8x4_t *vrpt, *vppt; + uint8x8x4_t vrp, vpp; + uint32x2x4_t *temp_pointer; + uint32x2x4_t vdest_val; + + vtmp = vld4_u32(png_ptr(uint32_t,rp)); + vrpt = png_ptr(uint8x8x4_t,&vtmp); + vrp = *vrpt; + vtmp = vld4_u32(png_ptrc(uint32_t,pp)); + vppt = png_ptr(uint8x8x4_t,&vtmp); + vpp = *vppt; + + vdest.val[0] = vhadd_u8(vdest.val[3], vpp.val[0]); + vdest.val[0] = vadd_u8(vdest.val[0], vrp.val[0]); + vdest.val[1] = vhadd_u8(vdest.val[0], vpp.val[1]); + vdest.val[1] = vadd_u8(vdest.val[1], vrp.val[1]); + vdest.val[2] = vhadd_u8(vdest.val[1], vpp.val[2]); + vdest.val[2] = vadd_u8(vdest.val[2], vrp.val[2]); + vdest.val[3] = vhadd_u8(vdest.val[2], vpp.val[3]); + vdest.val[3] = vadd_u8(vdest.val[3], vrp.val[3]); + + vdest_val = png_ldr(uint32x2x4_t, &vdest); + vst4_lane_u32(png_ptr(uint32_t,rp), vdest_val, 0); + } +} + +static uint8x8_t +paeth(uint8x8_t a, uint8x8_t b, uint8x8_t c) +{ + uint8x8_t d, e; + uint16x8_t p1, pa, pb, pc; + + p1 = vaddl_u8(a, b); /* a + b */ + pc = vaddl_u8(c, c); /* c * 2 */ + pa = vabdl_u8(b, c); /* pa */ + pb = vabdl_u8(a, c); /* pb */ + pc = vabdq_u16(p1, pc); /* pc */ + + p1 = vcleq_u16(pa, pb); /* pa <= pb */ + pa = vcleq_u16(pa, pc); /* pa <= pc */ + pb = vcleq_u16(pb, pc); /* pb <= pc */ + + p1 = vandq_u16(p1, pa); /* pa <= pb && pa <= pc */ + + d = vmovn_u16(pb); + e = vmovn_u16(p1); + + d = vbsl_u8(d, b, c); + e = vbsl_u8(e, a, d); + + return e; +} + +void +png_read_filter_row_paeth3_neon(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_bytep rp = row; + png_const_bytep pp = prev_row; + png_bytep rp_stop = row + row_info->rowbytes; + + uint8x16_t vtmp; + uint8x8x2_t *vrpt; + uint8x8x2_t vrp; + uint8x8_t vlast = vdup_n_u8(0); + uint8x8x4_t vdest; + vdest.val[3] = vdup_n_u8(0); + + vtmp = vld1q_u8(rp); + vrpt = png_ptr(uint8x8x2_t,&vtmp); + vrp = *vrpt; + + png_debug(1, "in png_read_filter_row_paeth3_neon"); + + for (; rp < rp_stop; pp += 12) + { + uint8x8x2_t *vppt; + uint8x8x2_t vpp; + uint8x8_t vtmp1, vtmp2, vtmp3; + uint32x2_t *temp_pointer; + + vtmp = vld1q_u8(pp); + vppt = png_ptr(uint8x8x2_t,&vtmp); + vpp = *vppt; + + vdest.val[0] = paeth(vdest.val[3], vpp.val[0], vlast); + vdest.val[0] = vadd_u8(vdest.val[0], vrp.val[0]); + + vtmp1 = vext_u8(vrp.val[0], vrp.val[1], 3); + vtmp2 = vext_u8(vpp.val[0], vpp.val[1], 3); + vdest.val[1] = paeth(vdest.val[0], vtmp2, vpp.val[0]); + vdest.val[1] = vadd_u8(vdest.val[1], vtmp1); + + vtmp1 = vext_u8(vrp.val[0], vrp.val[1], 6); + vtmp3 = vext_u8(vpp.val[0], vpp.val[1], 6); + vdest.val[2] = paeth(vdest.val[1], vtmp3, vtmp2); + vdest.val[2] = vadd_u8(vdest.val[2], vtmp1); + + vtmp1 = vext_u8(vrp.val[1], vrp.val[1], 1); + vtmp2 = vext_u8(vpp.val[1], vpp.val[1], 1); + + vtmp = vld1q_u8(rp + 12); + vrpt = png_ptr(uint8x8x2_t,&vtmp); + vrp = *vrpt; + + vdest.val[3] = paeth(vdest.val[2], vtmp2, vtmp3); + vdest.val[3] = vadd_u8(vdest.val[3], vtmp1); + + vlast = vtmp2; + + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[0]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[1]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[2]), 0); + rp += 3; + vst1_lane_u32(png_ptr(uint32_t,rp), png_ldr(uint32x2_t,&vdest.val[3]), 0); + rp += 3; + } +} + +void +png_read_filter_row_paeth4_neon(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_bytep rp = row; + png_bytep rp_stop = row + row_info->rowbytes; + png_const_bytep pp = prev_row; + + uint8x8_t vlast = vdup_n_u8(0); + uint8x8x4_t vdest; + vdest.val[3] = vdup_n_u8(0); + + png_debug(1, "in png_read_filter_row_paeth4_neon"); + + for (; rp < rp_stop; rp += 16, pp += 16) + { + uint32x2x4_t vtmp; + uint8x8x4_t *vrpt, *vppt; + uint8x8x4_t vrp, vpp; + uint32x2x4_t *temp_pointer; + uint32x2x4_t vdest_val; + + vtmp = vld4_u32(png_ptr(uint32_t,rp)); + vrpt = png_ptr(uint8x8x4_t,&vtmp); + vrp = *vrpt; + vtmp = vld4_u32(png_ptrc(uint32_t,pp)); + vppt = png_ptr(uint8x8x4_t,&vtmp); + vpp = *vppt; + + vdest.val[0] = paeth(vdest.val[3], vpp.val[0], vlast); + vdest.val[0] = vadd_u8(vdest.val[0], vrp.val[0]); + vdest.val[1] = paeth(vdest.val[0], vpp.val[1], vpp.val[0]); + vdest.val[1] = vadd_u8(vdest.val[1], vrp.val[1]); + vdest.val[2] = paeth(vdest.val[1], vpp.val[2], vpp.val[1]); + vdest.val[2] = vadd_u8(vdest.val[2], vrp.val[2]); + vdest.val[3] = paeth(vdest.val[2], vpp.val[3], vpp.val[2]); + vdest.val[3] = vadd_u8(vdest.val[3], vrp.val[3]); + + vlast = vpp.val[3]; + + vdest_val = png_ldr(uint32x2x4_t, &vdest); + vst4_lane_u32(png_ptr(uint32_t,rp), vdest_val, 0); + } +} + +#endif /* PNG_ARM_NEON_OPT > 0 */ +#endif /* PNG_ARM_NEON_IMPLEMENTATION == 1 (intrinsics) */ +#endif /* READ */ diff --git a/Externals/libpng/arm/palette_neon_intrinsics.c b/Externals/libpng/arm/palette_neon_intrinsics.c new file mode 100644 index 0000000000..b4d1fd2abf --- /dev/null +++ b/Externals/libpng/arm/palette_neon_intrinsics.c @@ -0,0 +1,149 @@ + +/* palette_neon_intrinsics.c - NEON optimised palette expansion functions + * + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 2017-2018 Arm Holdings. All rights reserved. + * Written by Richard Townsend , February 2017. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +#include "../pngpriv.h" + +#if PNG_ARM_NEON_IMPLEMENTATION == 1 + +#if defined(_MSC_VER) && defined(_M_ARM64) +# include +#else +# include +#endif + +/* Build an RGBA8 palette from the separate RGB and alpha palettes. */ +void +png_riffle_palette_neon(png_structrp png_ptr) +{ + png_const_colorp palette = png_ptr->palette; + png_bytep riffled_palette = png_ptr->riffled_palette; + png_const_bytep trans_alpha = png_ptr->trans_alpha; + int num_trans = png_ptr->num_trans; + int i; + + png_debug(1, "in png_riffle_palette_neon"); + + /* Initially black, opaque. */ + uint8x16x4_t w = {{ + vdupq_n_u8(0x00), + vdupq_n_u8(0x00), + vdupq_n_u8(0x00), + vdupq_n_u8(0xff), + }}; + + /* First, riffle the RGB colours into an RGBA8 palette. + * The alpha component is set to opaque for now. + */ + for (i = 0; i < 256; i += 16) + { + uint8x16x3_t v = vld3q_u8((png_const_bytep)(palette + i)); + w.val[0] = v.val[0]; + w.val[1] = v.val[1]; + w.val[2] = v.val[2]; + vst4q_u8(riffled_palette + (i << 2), w); + } + + /* Fix up the missing transparency values. */ + for (i = 0; i < num_trans; i++) + riffled_palette[(i << 2) + 3] = trans_alpha[i]; +} + +/* Expands a palettized row into RGBA8. */ +int +png_do_expand_palette_rgba8_neon(png_structrp png_ptr, png_row_infop row_info, + png_const_bytep row, png_bytepp ssp, png_bytepp ddp) +{ + png_uint_32 row_width = row_info->width; + const png_uint_32 *riffled_palette = + (const png_uint_32 *)png_ptr->riffled_palette; + const png_int_32 pixels_per_chunk = 4; + int i; + + png_debug(1, "in png_do_expand_palette_rgba8_neon"); + + if (row_width < pixels_per_chunk) + return 0; + + /* This function originally gets the last byte of the output row. + * The NEON part writes forward from a given position, so we have + * to seek this back by 4 pixels x 4 bytes. + */ + *ddp = *ddp - ((pixels_per_chunk * sizeof(png_uint_32)) - 1); + + for (i = 0; i < row_width; i += pixels_per_chunk) + { + uint32x4_t cur; + png_bytep sp = *ssp - i, dp = *ddp - (i << 2); + cur = vld1q_dup_u32 (riffled_palette + *(sp - 3)); + cur = vld1q_lane_u32(riffled_palette + *(sp - 2), cur, 1); + cur = vld1q_lane_u32(riffled_palette + *(sp - 1), cur, 2); + cur = vld1q_lane_u32(riffled_palette + *(sp - 0), cur, 3); + vst1q_u32((void *)dp, cur); + } + if (i != row_width) + { + /* Remove the amount that wasn't processed. */ + i -= pixels_per_chunk; + } + + /* Decrement output pointers. */ + *ssp = *ssp - i; + *ddp = *ddp - (i << 2); + return i; +} + +/* Expands a palettized row into RGB8. */ +int +png_do_expand_palette_rgb8_neon(png_structrp png_ptr, png_row_infop row_info, + png_const_bytep row, png_bytepp ssp, png_bytepp ddp) +{ + png_uint_32 row_width = row_info->width; + png_const_bytep palette = (png_const_bytep)png_ptr->palette; + const png_uint_32 pixels_per_chunk = 8; + int i; + + png_debug(1, "in png_do_expand_palette_rgb8_neon"); + + if (row_width <= pixels_per_chunk) + return 0; + + /* Seeking this back by 8 pixels x 3 bytes. */ + *ddp = *ddp - ((pixels_per_chunk * sizeof(png_color)) - 1); + + for (i = 0; i < row_width; i += pixels_per_chunk) + { + uint8x8x3_t cur; + png_bytep sp = *ssp - i, dp = *ddp - ((i << 1) + i); + cur = vld3_dup_u8(palette + sizeof(png_color) * (*(sp - 7))); + cur = vld3_lane_u8(palette + sizeof(png_color) * (*(sp - 6)), cur, 1); + cur = vld3_lane_u8(palette + sizeof(png_color) * (*(sp - 5)), cur, 2); + cur = vld3_lane_u8(palette + sizeof(png_color) * (*(sp - 4)), cur, 3); + cur = vld3_lane_u8(palette + sizeof(png_color) * (*(sp - 3)), cur, 4); + cur = vld3_lane_u8(palette + sizeof(png_color) * (*(sp - 2)), cur, 5); + cur = vld3_lane_u8(palette + sizeof(png_color) * (*(sp - 1)), cur, 6); + cur = vld3_lane_u8(palette + sizeof(png_color) * (*(sp - 0)), cur, 7); + vst3_u8((void *)dp, cur); + } + + if (i != row_width) + { + /* Remove the amount that wasn't processed. */ + i -= pixels_per_chunk; + } + + /* Decrement output pointers. */ + *ssp = *ssp - i; + *ddp = *ddp - ((i << 1) + i); + return i; +} + +#endif /* PNG_ARM_NEON_IMPLEMENTATION */ diff --git a/Externals/libpng/intel/filter_sse2_intrinsics.c b/Externals/libpng/intel/filter_sse2_intrinsics.c new file mode 100644 index 0000000000..f52aaa800a --- /dev/null +++ b/Externals/libpng/intel/filter_sse2_intrinsics.c @@ -0,0 +1,391 @@ + +/* filter_sse2_intrinsics.c - SSE2 optimized filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2016-2017 Glenn Randers-Pehrson + * Written by Mike Klein and Matt Sarett + * Derived from arm/filter_neon_intrinsics.c + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED + +#if PNG_INTEL_SSE_IMPLEMENTATION > 0 + +#include + +/* Functions in this file look at most 3 pixels (a,b,c) to predict the 4th (d). + * They're positioned like this: + * prev: c b + * row: a d + * The Sub filter predicts d=a, Avg d=(a+b)/2, and Paeth predicts d to be + * whichever of a, b, or c is closest to p=a+b-c. + */ + +static __m128i load4(const void* p) { + int tmp; + memcpy(&tmp, p, sizeof(tmp)); + return _mm_cvtsi32_si128(tmp); +} + +static void store4(void* p, __m128i v) { + int tmp = _mm_cvtsi128_si32(v); + memcpy(p, &tmp, sizeof(int)); +} + +static __m128i load3(const void* p) { + png_uint_32 tmp = 0; + memcpy(&tmp, p, 3); + return _mm_cvtsi32_si128(tmp); +} + +static void store3(void* p, __m128i v) { + int tmp = _mm_cvtsi128_si32(v); + memcpy(p, &tmp, 3); +} + +void png_read_filter_row_sub3_sse2(png_row_infop row_info, png_bytep row, + png_const_bytep prev) +{ + /* The Sub filter predicts each pixel as the previous pixel, a. + * There is no pixel to the left of the first pixel. It's encoded directly. + * That works with our main loop if we just say that left pixel was zero. + */ + size_t rb; + + __m128i a, d = _mm_setzero_si128(); + + png_debug(1, "in png_read_filter_row_sub3_sse2"); + + rb = row_info->rowbytes; + while (rb >= 4) { + a = d; d = load4(row); + d = _mm_add_epi8(d, a); + store3(row, d); + + row += 3; + rb -= 3; + } + if (rb > 0) { + a = d; d = load3(row); + d = _mm_add_epi8(d, a); + store3(row, d); + + row += 3; + rb -= 3; + } + PNG_UNUSED(prev) +} + +void png_read_filter_row_sub4_sse2(png_row_infop row_info, png_bytep row, + png_const_bytep prev) +{ + /* The Sub filter predicts each pixel as the previous pixel, a. + * There is no pixel to the left of the first pixel. It's encoded directly. + * That works with our main loop if we just say that left pixel was zero. + */ + size_t rb; + + __m128i a, d = _mm_setzero_si128(); + + png_debug(1, "in png_read_filter_row_sub4_sse2"); + + rb = row_info->rowbytes+4; + while (rb > 4) { + a = d; d = load4(row); + d = _mm_add_epi8(d, a); + store4(row, d); + + row += 4; + rb -= 4; + } + PNG_UNUSED(prev) +} + +void png_read_filter_row_avg3_sse2(png_row_infop row_info, png_bytep row, + png_const_bytep prev) +{ + /* The Avg filter predicts each pixel as the (truncated) average of a and b. + * There's no pixel to the left of the first pixel. Luckily, it's + * predicted to be half of the pixel above it. So again, this works + * perfectly with our loop if we make sure a starts at zero. + */ + + size_t rb; + + const __m128i zero = _mm_setzero_si128(); + + __m128i b; + __m128i a, d = zero; + + png_debug(1, "in png_read_filter_row_avg3_sse2"); + rb = row_info->rowbytes; + while (rb >= 4) { + __m128i avg; + b = load4(prev); + a = d; d = load4(row ); + + /* PNG requires a truncating average, so we can't just use _mm_avg_epu8 */ + avg = _mm_avg_epu8(a,b); + /* ...but we can fix it up by subtracting off 1 if it rounded up. */ + avg = _mm_sub_epi8(avg, _mm_and_si128(_mm_xor_si128(a,b), + _mm_set1_epi8(1))); + d = _mm_add_epi8(d, avg); + store3(row, d); + + prev += 3; + row += 3; + rb -= 3; + } + if (rb > 0) { + __m128i avg; + b = load3(prev); + a = d; d = load3(row ); + + /* PNG requires a truncating average, so we can't just use _mm_avg_epu8 */ + avg = _mm_avg_epu8(a,b); + /* ...but we can fix it up by subtracting off 1 if it rounded up. */ + avg = _mm_sub_epi8(avg, _mm_and_si128(_mm_xor_si128(a,b), + _mm_set1_epi8(1))); + + d = _mm_add_epi8(d, avg); + store3(row, d); + + prev += 3; + row += 3; + rb -= 3; + } +} + +void png_read_filter_row_avg4_sse2(png_row_infop row_info, png_bytep row, + png_const_bytep prev) +{ + /* The Avg filter predicts each pixel as the (truncated) average of a and b. + * There's no pixel to the left of the first pixel. Luckily, it's + * predicted to be half of the pixel above it. So again, this works + * perfectly with our loop if we make sure a starts at zero. + */ + size_t rb; + const __m128i zero = _mm_setzero_si128(); + __m128i b; + __m128i a, d = zero; + + png_debug(1, "in png_read_filter_row_avg4_sse2"); + + rb = row_info->rowbytes+4; + while (rb > 4) { + __m128i avg; + b = load4(prev); + a = d; d = load4(row ); + + /* PNG requires a truncating average, so we can't just use _mm_avg_epu8 */ + avg = _mm_avg_epu8(a,b); + /* ...but we can fix it up by subtracting off 1 if it rounded up. */ + avg = _mm_sub_epi8(avg, _mm_and_si128(_mm_xor_si128(a,b), + _mm_set1_epi8(1))); + + d = _mm_add_epi8(d, avg); + store4(row, d); + + prev += 4; + row += 4; + rb -= 4; + } +} + +/* Returns |x| for 16-bit lanes. */ +static __m128i abs_i16(__m128i x) { +#if PNG_INTEL_SSE_IMPLEMENTATION >= 2 + return _mm_abs_epi16(x); +#else + /* Read this all as, return x<0 ? -x : x. + * To negate two's complement, you flip all the bits then add 1. + */ + __m128i is_negative = _mm_cmplt_epi16(x, _mm_setzero_si128()); + + /* Flip negative lanes. */ + x = _mm_xor_si128(x, is_negative); + + /* +1 to negative lanes, else +0. */ + x = _mm_sub_epi16(x, is_negative); + return x; +#endif +} + +/* Bytewise c ? t : e. */ +static __m128i if_then_else(__m128i c, __m128i t, __m128i e) { +#if PNG_INTEL_SSE_IMPLEMENTATION >= 3 + return _mm_blendv_epi8(e,t,c); +#else + return _mm_or_si128(_mm_and_si128(c, t), _mm_andnot_si128(c, e)); +#endif +} + +void png_read_filter_row_paeth3_sse2(png_row_infop row_info, png_bytep row, + png_const_bytep prev) +{ + /* Paeth tries to predict pixel d using the pixel to the left of it, a, + * and two pixels from the previous row, b and c: + * prev: c b + * row: a d + * The Paeth function predicts d to be whichever of a, b, or c is nearest to + * p=a+b-c. + * + * The first pixel has no left context, and so uses an Up filter, p = b. + * This works naturally with our main loop's p = a+b-c if we force a and c + * to zero. + * Here we zero b and d, which become c and a respectively at the start of + * the loop. + */ + size_t rb; + const __m128i zero = _mm_setzero_si128(); + __m128i c, b = zero, + a, d = zero; + + png_debug(1, "in png_read_filter_row_paeth3_sse2"); + + rb = row_info->rowbytes; + while (rb >= 4) { + /* It's easiest to do this math (particularly, deal with pc) with 16-bit + * intermediates. + */ + __m128i pa,pb,pc,smallest,nearest; + c = b; b = _mm_unpacklo_epi8(load4(prev), zero); + a = d; d = _mm_unpacklo_epi8(load4(row ), zero); + + /* (p-a) == (a+b-c - a) == (b-c) */ + + pa = _mm_sub_epi16(b,c); + + /* (p-b) == (a+b-c - b) == (a-c) */ + pb = _mm_sub_epi16(a,c); + + /* (p-c) == (a+b-c - c) == (a+b-c-c) == (b-c)+(a-c) */ + pc = _mm_add_epi16(pa,pb); + + pa = abs_i16(pa); /* |p-a| */ + pb = abs_i16(pb); /* |p-b| */ + pc = abs_i16(pc); /* |p-c| */ + + smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb)); + + /* Paeth breaks ties favoring a over b over c. */ + nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a, + if_then_else(_mm_cmpeq_epi16(smallest, pb), b, + c)); + + /* Note `_epi8`: we need addition to wrap modulo 255. */ + d = _mm_add_epi8(d, nearest); + store3(row, _mm_packus_epi16(d,d)); + + prev += 3; + row += 3; + rb -= 3; + } + if (rb > 0) { + /* It's easiest to do this math (particularly, deal with pc) with 16-bit + * intermediates. + */ + __m128i pa,pb,pc,smallest,nearest; + c = b; b = _mm_unpacklo_epi8(load3(prev), zero); + a = d; d = _mm_unpacklo_epi8(load3(row ), zero); + + /* (p-a) == (a+b-c - a) == (b-c) */ + pa = _mm_sub_epi16(b,c); + + /* (p-b) == (a+b-c - b) == (a-c) */ + pb = _mm_sub_epi16(a,c); + + /* (p-c) == (a+b-c - c) == (a+b-c-c) == (b-c)+(a-c) */ + pc = _mm_add_epi16(pa,pb); + + pa = abs_i16(pa); /* |p-a| */ + pb = abs_i16(pb); /* |p-b| */ + pc = abs_i16(pc); /* |p-c| */ + + smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb)); + + /* Paeth breaks ties favoring a over b over c. */ + nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a, + if_then_else(_mm_cmpeq_epi16(smallest, pb), b, + c)); + + /* Note `_epi8`: we need addition to wrap modulo 255. */ + d = _mm_add_epi8(d, nearest); + store3(row, _mm_packus_epi16(d,d)); + + prev += 3; + row += 3; + rb -= 3; + } +} + +void png_read_filter_row_paeth4_sse2(png_row_infop row_info, png_bytep row, + png_const_bytep prev) +{ + /* Paeth tries to predict pixel d using the pixel to the left of it, a, + * and two pixels from the previous row, b and c: + * prev: c b + * row: a d + * The Paeth function predicts d to be whichever of a, b, or c is nearest to + * p=a+b-c. + * + * The first pixel has no left context, and so uses an Up filter, p = b. + * This works naturally with our main loop's p = a+b-c if we force a and c + * to zero. + * Here we zero b and d, which become c and a respectively at the start of + * the loop. + */ + size_t rb; + const __m128i zero = _mm_setzero_si128(); + __m128i pa,pb,pc,smallest,nearest; + __m128i c, b = zero, + a, d = zero; + + png_debug(1, "in png_read_filter_row_paeth4_sse2"); + + rb = row_info->rowbytes+4; + while (rb > 4) { + /* It's easiest to do this math (particularly, deal with pc) with 16-bit + * intermediates. + */ + c = b; b = _mm_unpacklo_epi8(load4(prev), zero); + a = d; d = _mm_unpacklo_epi8(load4(row ), zero); + + /* (p-a) == (a+b-c - a) == (b-c) */ + pa = _mm_sub_epi16(b,c); + + /* (p-b) == (a+b-c - b) == (a-c) */ + pb = _mm_sub_epi16(a,c); + + /* (p-c) == (a+b-c - c) == (a+b-c-c) == (b-c)+(a-c) */ + pc = _mm_add_epi16(pa,pb); + + pa = abs_i16(pa); /* |p-a| */ + pb = abs_i16(pb); /* |p-b| */ + pc = abs_i16(pc); /* |p-c| */ + + smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb)); + + /* Paeth breaks ties favoring a over b over c. */ + nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a, + if_then_else(_mm_cmpeq_epi16(smallest, pb), b, + c)); + + /* Note `_epi8`: we need addition to wrap modulo 255. */ + d = _mm_add_epi8(d, nearest); + store4(row, _mm_packus_epi16(d,d)); + + prev += 4; + row += 4; + rb -= 4; + } +} + +#endif /* PNG_INTEL_SSE_IMPLEMENTATION > 0 */ +#endif /* READ */ diff --git a/Externals/libpng/intel/intel_init.c b/Externals/libpng/intel/intel_init.c new file mode 100644 index 0000000000..2f8168b7c4 --- /dev/null +++ b/Externals/libpng/intel/intel_init.c @@ -0,0 +1,52 @@ + +/* intel_init.c - SSE2 optimized filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2016-2017 Glenn Randers-Pehrson + * Written by Mike Klein and Matt Sarett, Google, Inc. + * Derived from arm/arm_init.c + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED +#if PNG_INTEL_SSE_IMPLEMENTATION > 0 + +void +png_init_filter_functions_sse2(png_structp pp, unsigned int bpp) +{ + /* The techniques used to implement each of these filters in SSE operate on + * one pixel at a time. + * So they generally speed up 3bpp images about 3x, 4bpp images about 4x. + * They can scale up to 6 and 8 bpp images and down to 2 bpp images, + * but they'd not likely have any benefit for 1bpp images. + * Most of these can be implemented using only MMX and 64-bit registers, + * but they end up a bit slower than using the equally-ubiquitous SSE2. + */ + png_debug(1, "in png_init_filter_functions_sse2"); + if (bpp == 3) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub3_sse2; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg3_sse2; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = + png_read_filter_row_paeth3_sse2; + } + else if (bpp == 4) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub4_sse2; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg4_sse2; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = + png_read_filter_row_paeth4_sse2; + } + + /* No need optimize PNG_FILTER_VALUE_UP. The compiler should + * autovectorize. + */ +} + +#endif /* PNG_INTEL_SSE_IMPLEMENTATION > 0 */ +#endif /* PNG_READ_SUPPORTED */ diff --git a/Externals/libpng/mips/filter_msa_intrinsics.c b/Externals/libpng/mips/filter_msa_intrinsics.c new file mode 100644 index 0000000000..a579179421 --- /dev/null +++ b/Externals/libpng/mips/filter_msa_intrinsics.c @@ -0,0 +1,808 @@ + +/* filter_msa_intrinsics.c - MSA optimised filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2016 Glenn Randers-Pehrson + * Written by Mandar Sahastrabuddhe, August 2016. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +#include +#include +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED + +/* This code requires -mfpu=msa on the command line: */ +#if PNG_MIPS_MSA_IMPLEMENTATION == 1 /* intrinsics code from pngpriv.h */ + +#include + +/* libpng row pointers are not necessarily aligned to any particular boundary, + * however this code will only work with appropriate alignment. mips/mips_init.c + * checks for this (and will not compile unless it is done). This code uses + * variants of png_aligncast to avoid compiler warnings. + */ +#define png_ptr(type,pointer) png_aligncast(type *,pointer) +#define png_ptrc(type,pointer) png_aligncastconst(const type *,pointer) + +/* The following relies on a variable 'temp_pointer' being declared with type + * 'type'. This is written this way just to hide the GCC strict aliasing + * warning; note that the code is safe because there never is an alias between + * the input and output pointers. + */ +#define png_ldr(type,pointer)\ + (temp_pointer = png_ptr(type,pointer), *temp_pointer) + +#if PNG_MIPS_MSA_OPT > 0 + +#ifdef CLANG_BUILD + #define MSA_SRLI_B(a, b) __msa_srli_b((v16i8) a, b) + + #define LW(psrc) \ + ( { \ + uint8_t *psrc_lw_m = (uint8_t *) (psrc); \ + uint32_t val_m; \ + \ + asm volatile ( \ + "lw %[val_m], %[psrc_lw_m] \n\t" \ + \ + : [val_m] "=r" (val_m) \ + : [psrc_lw_m] "m" (*psrc_lw_m) \ + ); \ + \ + val_m; \ + } ) + + #define SH(val, pdst) \ + { \ + uint8_t *pdst_sh_m = (uint8_t *) (pdst); \ + uint16_t val_m = (val); \ + \ + asm volatile ( \ + "sh %[val_m], %[pdst_sh_m] \n\t" \ + \ + : [pdst_sh_m] "=m" (*pdst_sh_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + + #define SW(val, pdst) \ + { \ + uint8_t *pdst_sw_m = (uint8_t *) (pdst); \ + uint32_t val_m = (val); \ + \ + asm volatile ( \ + "sw %[val_m], %[pdst_sw_m] \n\t" \ + \ + : [pdst_sw_m] "=m" (*pdst_sw_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + + #if (__mips == 64) + #define SD(val, pdst) \ + { \ + uint8_t *pdst_sd_m = (uint8_t *) (pdst); \ + uint64_t val_m = (val); \ + \ + asm volatile ( \ + "sd %[val_m], %[pdst_sd_m] \n\t" \ + \ + : [pdst_sd_m] "=m" (*pdst_sd_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + #else + #define SD(val, pdst) \ + { \ + uint8_t *pdst_sd_m = (uint8_t *) (pdst); \ + uint32_t val0_m, val1_m; \ + \ + val0_m = (uint32_t) ((val) & 0x00000000FFFFFFFF); \ + val1_m = (uint32_t) (((val) >> 32) & 0x00000000FFFFFFFF); \ + \ + SW(val0_m, pdst_sd_m); \ + SW(val1_m, pdst_sd_m + 4); \ + } + #endif +#else + #define MSA_SRLI_B(a, b) (a >> b) + +#if (__mips_isa_rev >= 6) + #define LW(psrc) \ + ( { \ + uint8_t *psrc_lw_m = (uint8_t *) (psrc); \ + uint32_t val_m; \ + \ + asm volatile ( \ + "lw %[val_m], %[psrc_lw_m] \n\t" \ + \ + : [val_m] "=r" (val_m) \ + : [psrc_lw_m] "m" (*psrc_lw_m) \ + ); \ + \ + val_m; \ + } ) + + #define SH(val, pdst) \ + { \ + uint8_t *pdst_sh_m = (uint8_t *) (pdst); \ + uint16_t val_m = (val); \ + \ + asm volatile ( \ + "sh %[val_m], %[pdst_sh_m] \n\t" \ + \ + : [pdst_sh_m] "=m" (*pdst_sh_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + + #define SW(val, pdst) \ + { \ + uint8_t *pdst_sw_m = (uint8_t *) (pdst); \ + uint32_t val_m = (val); \ + \ + asm volatile ( \ + "sw %[val_m], %[pdst_sw_m] \n\t" \ + \ + : [pdst_sw_m] "=m" (*pdst_sw_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + + #if (__mips == 64) + #define SD(val, pdst) \ + { \ + uint8_t *pdst_sd_m = (uint8_t *) (pdst); \ + uint64_t val_m = (val); \ + \ + asm volatile ( \ + "sd %[val_m], %[pdst_sd_m] \n\t" \ + \ + : [pdst_sd_m] "=m" (*pdst_sd_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + #else + #define SD(val, pdst) \ + { \ + uint8_t *pdst_sd_m = (uint8_t *) (pdst); \ + uint32_t val0_m, val1_m; \ + \ + val0_m = (uint32_t) ((val) & 0x00000000FFFFFFFF); \ + val1_m = (uint32_t) (((val) >> 32) & 0x00000000FFFFFFFF); \ + \ + SW(val0_m, pdst_sd_m); \ + SW(val1_m, pdst_sd_m + 4); \ + } + #endif +#else // !(__mips_isa_rev >= 6) + #define LW(psrc) \ + ( { \ + uint8_t *psrc_lw_m = (uint8_t *) (psrc); \ + uint32_t val_m; \ + \ + asm volatile ( \ + "ulw %[val_m], %[psrc_lw_m] \n\t" \ + \ + : [val_m] "=r" (val_m) \ + : [psrc_lw_m] "m" (*psrc_lw_m) \ + ); \ + \ + val_m; \ + } ) + + #define SH(val, pdst) \ + { \ + uint8_t *pdst_sh_m = (uint8_t *) (pdst); \ + uint16_t val_m = (val); \ + \ + asm volatile ( \ + "ush %[val_m], %[pdst_sh_m] \n\t" \ + \ + : [pdst_sh_m] "=m" (*pdst_sh_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + + #define SW(val, pdst) \ + { \ + uint8_t *pdst_sw_m = (uint8_t *) (pdst); \ + uint32_t val_m = (val); \ + \ + asm volatile ( \ + "usw %[val_m], %[pdst_sw_m] \n\t" \ + \ + : [pdst_sw_m] "=m" (*pdst_sw_m) \ + : [val_m] "r" (val_m) \ + ); \ + } + + #define SD(val, pdst) \ + { \ + uint8_t *pdst_sd_m = (uint8_t *) (pdst); \ + uint32_t val0_m, val1_m; \ + \ + val0_m = (uint32_t) ((val) & 0x00000000FFFFFFFF); \ + val1_m = (uint32_t) (((val) >> 32) & 0x00000000FFFFFFFF); \ + \ + SW(val0_m, pdst_sd_m); \ + SW(val1_m, pdst_sd_m + 4); \ + } + + #define SW_ZERO(pdst) \ + { \ + uint8_t *pdst_m = (uint8_t *) (pdst); \ + \ + asm volatile ( \ + "usw $0, %[pdst_m] \n\t" \ + \ + : [pdst_m] "=m" (*pdst_m) \ + : \ + ); \ + } +#endif // (__mips_isa_rev >= 6) +#endif + +#define LD_B(RTYPE, psrc) *((RTYPE *) (psrc)) +#define LD_UB(...) LD_B(v16u8, __VA_ARGS__) +#define LD_B2(RTYPE, psrc, stride, out0, out1) \ +{ \ + out0 = LD_B(RTYPE, (psrc)); \ + out1 = LD_B(RTYPE, (psrc) + stride); \ +} +#define LD_UB2(...) LD_B2(v16u8, __VA_ARGS__) +#define LD_B4(RTYPE, psrc, stride, out0, out1, out2, out3) \ +{ \ + LD_B2(RTYPE, (psrc), stride, out0, out1); \ + LD_B2(RTYPE, (psrc) + 2 * stride , stride, out2, out3); \ +} +#define LD_UB4(...) LD_B4(v16u8, __VA_ARGS__) + +#define ST_B(RTYPE, in, pdst) *((RTYPE *) (pdst)) = (in) +#define ST_UB(...) ST_B(v16u8, __VA_ARGS__) +#define ST_B2(RTYPE, in0, in1, pdst, stride) \ +{ \ + ST_B(RTYPE, in0, (pdst)); \ + ST_B(RTYPE, in1, (pdst) + stride); \ +} +#define ST_UB2(...) ST_B2(v16u8, __VA_ARGS__) +#define ST_B4(RTYPE, in0, in1, in2, in3, pdst, stride) \ +{ \ + ST_B2(RTYPE, in0, in1, (pdst), stride); \ + ST_B2(RTYPE, in2, in3, (pdst) + 2 * stride, stride); \ +} +#define ST_UB4(...) ST_B4(v16u8, __VA_ARGS__) + +#define ADD2(in0, in1, in2, in3, out0, out1) \ +{ \ + out0 = in0 + in1; \ + out1 = in2 + in3; \ +} +#define ADD3(in0, in1, in2, in3, in4, in5, \ + out0, out1, out2) \ +{ \ + ADD2(in0, in1, in2, in3, out0, out1); \ + out2 = in4 + in5; \ +} +#define ADD4(in0, in1, in2, in3, in4, in5, in6, in7, \ + out0, out1, out2, out3) \ +{ \ + ADD2(in0, in1, in2, in3, out0, out1); \ + ADD2(in4, in5, in6, in7, out2, out3); \ +} + +#define ILVR_B2(RTYPE, in0, in1, in2, in3, out0, out1) \ +{ \ + out0 = (RTYPE) __msa_ilvr_b((v16i8) in0, (v16i8) in1); \ + out1 = (RTYPE) __msa_ilvr_b((v16i8) in2, (v16i8) in3); \ +} +#define ILVR_B2_SH(...) ILVR_B2(v8i16, __VA_ARGS__) + +#define HSUB_UB2(RTYPE, in0, in1, out0, out1) \ +{ \ + out0 = (RTYPE) __msa_hsub_u_h((v16u8) in0, (v16u8) in0); \ + out1 = (RTYPE) __msa_hsub_u_h((v16u8) in1, (v16u8) in1); \ +} +#define HSUB_UB2_SH(...) HSUB_UB2(v8i16, __VA_ARGS__) + +#define SLDI_B2_0(RTYPE, in0, in1, out0, out1, slide_val) \ +{ \ + v16i8 zero_m = { 0 }; \ + out0 = (RTYPE) __msa_sldi_b((v16i8) zero_m, (v16i8) in0, slide_val); \ + out1 = (RTYPE) __msa_sldi_b((v16i8) zero_m, (v16i8) in1, slide_val); \ +} +#define SLDI_B2_0_UB(...) SLDI_B2_0(v16u8, __VA_ARGS__) + +#define SLDI_B3_0(RTYPE, in0, in1, in2, out0, out1, out2, slide_val) \ +{ \ + v16i8 zero_m = { 0 }; \ + SLDI_B2_0(RTYPE, in0, in1, out0, out1, slide_val); \ + out2 = (RTYPE) __msa_sldi_b((v16i8) zero_m, (v16i8) in2, slide_val); \ +} +#define SLDI_B3_0_UB(...) SLDI_B3_0(v16u8, __VA_ARGS__) + +#define ILVEV_W2(RTYPE, in0, in1, in2, in3, out0, out1) \ +{ \ + out0 = (RTYPE) __msa_ilvev_w((v4i32) in1, (v4i32) in0); \ + out1 = (RTYPE) __msa_ilvev_w((v4i32) in3, (v4i32) in2); \ +} +#define ILVEV_W2_UB(...) ILVEV_W2(v16u8, __VA_ARGS__) + +#define ADD_ABS_H3(RTYPE, in0, in1, in2, out0, out1, out2) \ +{ \ + RTYPE zero = {0}; \ + \ + out0 = __msa_add_a_h((v8i16) zero, in0); \ + out1 = __msa_add_a_h((v8i16) zero, in1); \ + out2 = __msa_add_a_h((v8i16) zero, in2); \ +} +#define ADD_ABS_H3_SH(...) ADD_ABS_H3(v8i16, __VA_ARGS__) + +#define VSHF_B2(RTYPE, in0, in1, in2, in3, mask0, mask1, out0, out1) \ +{ \ + out0 = (RTYPE) __msa_vshf_b((v16i8) mask0, (v16i8) in1, (v16i8) in0); \ + out1 = (RTYPE) __msa_vshf_b((v16i8) mask1, (v16i8) in3, (v16i8) in2); \ +} +#define VSHF_B2_UB(...) VSHF_B2(v16u8, __VA_ARGS__) + +#define CMP_AND_SELECT(inp0, inp1, inp2, inp3, inp4, inp5, out0) \ +{ \ + v8i16 _sel_h0, _sel_h1; \ + v16u8 _sel_b0, _sel_b1; \ + _sel_h0 = (v8i16) __msa_clt_u_h((v8u16) inp1, (v8u16) inp0); \ + _sel_b0 = (v16u8) __msa_pckev_b((v16i8) _sel_h0, (v16i8) _sel_h0); \ + inp0 = (v8i16) __msa_bmnz_v((v16u8) inp0, (v16u8) inp1, (v16u8) _sel_h0); \ + inp4 = (v16u8) __msa_bmnz_v(inp3, inp4, _sel_b0); \ + _sel_h1 = (v8i16) __msa_clt_u_h((v8u16) inp2, (v8u16) inp0); \ + _sel_b1 = (v16u8) __msa_pckev_b((v16i8) _sel_h1, (v16i8) _sel_h1); \ + inp4 = (v16u8) __msa_bmnz_v(inp4, inp5, _sel_b1); \ + out0 += inp4; \ +} + +void png_read_filter_row_up_msa(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + size_t i, cnt, cnt16, cnt32; + size_t istop = row_info->rowbytes; + png_bytep rp = row; + png_const_bytep pp = prev_row; + v16u8 src0, src1, src2, src3, src4, src5, src6, src7; + + for (i = 0; i < (istop >> 6); i++) + { + LD_UB4(rp, 16, src0, src1, src2, src3); + LD_UB4(pp, 16, src4, src5, src6, src7); + pp += 64; + + ADD4(src0, src4, src1, src5, src2, src6, src3, src7, + src0, src1, src2, src3); + + ST_UB4(src0, src1, src2, src3, rp, 16); + rp += 64; + } + + if (istop & 0x3F) + { + cnt32 = istop & 0x20; + cnt16 = istop & 0x10; + cnt = istop & 0xF; + + if(cnt32) + { + if (cnt16 && cnt) + { + LD_UB4(rp, 16, src0, src1, src2, src3); + LD_UB4(pp, 16, src4, src5, src6, src7); + + ADD4(src0, src4, src1, src5, src2, src6, src3, src7, + src0, src1, src2, src3); + + ST_UB4(src0, src1, src2, src3, rp, 16); + rp += 64; + } + else if (cnt16 || cnt) + { + LD_UB2(rp, 16, src0, src1); + LD_UB2(pp, 16, src4, src5); + pp += 32; + src2 = LD_UB(rp + 32); + src6 = LD_UB(pp); + + ADD3(src0, src4, src1, src5, src2, src6, src0, src1, src2); + + ST_UB2(src0, src1, rp, 16); + rp += 32; + ST_UB(src2, rp); + rp += 16; + } + else + { + LD_UB2(rp, 16, src0, src1); + LD_UB2(pp, 16, src4, src5); + + ADD2(src0, src4, src1, src5, src0, src1); + + ST_UB2(src0, src1, rp, 16); + rp += 32; + } + } + else if (cnt16 && cnt) + { + LD_UB2(rp, 16, src0, src1); + LD_UB2(pp, 16, src4, src5); + + ADD2(src0, src4, src1, src5, src0, src1); + + ST_UB2(src0, src1, rp, 16); + rp += 32; + } + else if (cnt16 || cnt) + { + src0 = LD_UB(rp); + src4 = LD_UB(pp); + pp += 16; + + src0 += src4; + + ST_UB(src0, rp); + rp += 16; + } + } +} + +void png_read_filter_row_sub4_msa(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + size_t count; + size_t istop = row_info->rowbytes; + png_bytep src = row; + png_bytep nxt = row + 4; + int32_t inp0; + v16u8 src0, src1, src2, src3, src4; + v16u8 dst0, dst1; + v16u8 zero = { 0 }; + + istop -= 4; + + inp0 = LW(src); + src += 4; + src0 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp0); + + for (count = 0; count < istop; count += 16) + { + src1 = LD_UB(src); + src += 16; + + src2 = (v16u8) __msa_sldi_b((v16i8) zero, (v16i8) src1, 4); + src3 = (v16u8) __msa_sldi_b((v16i8) zero, (v16i8) src1, 8); + src4 = (v16u8) __msa_sldi_b((v16i8) zero, (v16i8) src1, 12); + src1 += src0; + src2 += src1; + src3 += src2; + src4 += src3; + src0 = src4; + ILVEV_W2_UB(src1, src2, src3, src4, dst0, dst1); + dst0 = (v16u8) __msa_pckev_d((v2i64) dst1, (v2i64) dst0); + + ST_UB(dst0, nxt); + nxt += 16; + } +} + +void png_read_filter_row_sub3_msa(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + size_t count; + size_t istop = row_info->rowbytes; + png_bytep src = row; + png_bytep nxt = row + 3; + int64_t out0; + int32_t inp0, out1; + v16u8 src0, src1, src2, src3, src4, dst0, dst1; + v16u8 zero = { 0 }; + v16i8 mask0 = { 0, 1, 2, 16, 17, 18, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + v16i8 mask1 = { 0, 1, 2, 3, 4, 5, 16, 17, 18, 19, 20, 21, 0, 0, 0, 0 }; + + istop -= 3; + + inp0 = LW(src); + src += 3; + src0 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp0); + + for (count = 0; count < istop; count += 12) + { + src1 = LD_UB(src); + src += 12; + + src2 = (v16u8) __msa_sldi_b((v16i8) zero, (v16i8) src1, 3); + src3 = (v16u8) __msa_sldi_b((v16i8) zero, (v16i8) src1, 6); + src4 = (v16u8) __msa_sldi_b((v16i8) zero, (v16i8) src1, 9); + src1 += src0; + src2 += src1; + src3 += src2; + src4 += src3; + src0 = src4; + VSHF_B2_UB(src1, src2, src3, src4, mask0, mask0, dst0, dst1); + dst0 = (v16u8) __msa_vshf_b(mask1, (v16i8) dst1, (v16i8) dst0); + out0 = __msa_copy_s_d((v2i64) dst0, 0); + out1 = __msa_copy_s_w((v4i32) dst0, 2); + + SD(out0, nxt); + nxt += 8; + SW(out1, nxt); + nxt += 4; + } +} + +void png_read_filter_row_avg4_msa(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + size_t i; + png_bytep src = row; + png_bytep nxt = row; + png_const_bytep pp = prev_row; + size_t istop = row_info->rowbytes - 4; + int32_t inp0, inp1, out0; + v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, dst0, dst1; + v16u8 zero = { 0 }; + + inp0 = LW(pp); + pp += 4; + inp1 = LW(src); + src += 4; + src0 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp0); + src1 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp1); + src0 = (v16u8) MSA_SRLI_B(src0, 1); + src1 += src0; + out0 = __msa_copy_s_w((v4i32) src1, 0); + SW(out0, nxt); + nxt += 4; + + for (i = 0; i < istop; i += 16) + { + src2 = LD_UB(pp); + pp += 16; + src6 = LD_UB(src); + src += 16; + + SLDI_B2_0_UB(src2, src6, src3, src7, 4); + SLDI_B2_0_UB(src2, src6, src4, src8, 8); + SLDI_B2_0_UB(src2, src6, src5, src9, 12); + src2 = __msa_ave_u_b(src2, src1); + src6 += src2; + src3 = __msa_ave_u_b(src3, src6); + src7 += src3; + src4 = __msa_ave_u_b(src4, src7); + src8 += src4; + src5 = __msa_ave_u_b(src5, src8); + src9 += src5; + src1 = src9; + ILVEV_W2_UB(src6, src7, src8, src9, dst0, dst1); + dst0 = (v16u8) __msa_pckev_d((v2i64) dst1, (v2i64) dst0); + + ST_UB(dst0, nxt); + nxt += 16; + } +} + +void png_read_filter_row_avg3_msa(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + size_t i; + png_bytep src = row; + png_bytep nxt = row; + png_const_bytep pp = prev_row; + size_t istop = row_info->rowbytes - 3; + int64_t out0; + int32_t inp0, inp1, out1; + int16_t out2; + v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, dst0, dst1; + v16u8 zero = { 0 }; + v16i8 mask0 = { 0, 1, 2, 16, 17, 18, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + v16i8 mask1 = { 0, 1, 2, 3, 4, 5, 16, 17, 18, 19, 20, 21, 0, 0, 0, 0 }; + + inp0 = LW(pp); + pp += 3; + inp1 = LW(src); + src += 3; + src0 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp0); + src1 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp1); + src0 = (v16u8) MSA_SRLI_B(src0, 1); + src1 += src0; + out2 = __msa_copy_s_h((v8i16) src1, 0); + SH(out2, nxt); + nxt += 2; + nxt[0] = src1[2]; + nxt++; + + for (i = 0; i < istop; i += 12) + { + src2 = LD_UB(pp); + pp += 12; + src6 = LD_UB(src); + src += 12; + + SLDI_B2_0_UB(src2, src6, src3, src7, 3); + SLDI_B2_0_UB(src2, src6, src4, src8, 6); + SLDI_B2_0_UB(src2, src6, src5, src9, 9); + src2 = __msa_ave_u_b(src2, src1); + src6 += src2; + src3 = __msa_ave_u_b(src3, src6); + src7 += src3; + src4 = __msa_ave_u_b(src4, src7); + src8 += src4; + src5 = __msa_ave_u_b(src5, src8); + src9 += src5; + src1 = src9; + VSHF_B2_UB(src6, src7, src8, src9, mask0, mask0, dst0, dst1); + dst0 = (v16u8) __msa_vshf_b(mask1, (v16i8) dst1, (v16i8) dst0); + out0 = __msa_copy_s_d((v2i64) dst0, 0); + out1 = __msa_copy_s_w((v4i32) dst0, 2); + + SD(out0, nxt); + nxt += 8; + SW(out1, nxt); + nxt += 4; + } +} + +void png_read_filter_row_paeth4_msa(png_row_infop row_info, + png_bytep row, + png_const_bytep prev_row) +{ + int32_t count, rp_end; + png_bytep nxt; + png_const_bytep prev_nxt; + int32_t inp0, inp1, res0; + v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9; + v16u8 src10, src11, src12, src13, dst0, dst1; + v8i16 vec0, vec1, vec2; + v16u8 zero = { 0 }; + + nxt = row; + prev_nxt = prev_row; + + inp0 = LW(nxt); + inp1 = LW(prev_nxt); + prev_nxt += 4; + src0 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp0); + src1 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp1); + + src1 += src0; + res0 = __msa_copy_s_w((v4i32) src1, 0); + + SW(res0, nxt); + nxt += 4; + + /* Remainder */ + rp_end = row_info->rowbytes - 4; + + for (count = 0; count < rp_end; count += 16) + { + src2 = LD_UB(prev_nxt); + prev_nxt += 16; + src6 = LD_UB(prev_row); + prev_row += 16; + src10 = LD_UB(nxt); + + SLDI_B3_0_UB(src2, src6, src10, src3, src7, src11, 4); + SLDI_B3_0_UB(src2, src6, src10, src4, src8, src12, 8); + SLDI_B3_0_UB(src2, src6, src10, src5, src9, src13, 12); + ILVR_B2_SH(src2, src6, src1, src6, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src1, src2, src6, src10); + ILVR_B2_SH(src3, src7, src10, src7, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src10, src3, src7, src11); + ILVR_B2_SH(src4, src8, src11, src8, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src11, src4, src8, src12); + ILVR_B2_SH(src5, src9, src12, src9, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src12, src5, src9, src13); + src1 = src13; + ILVEV_W2_UB(src10, src11, src12, src1, dst0, dst1); + dst0 = (v16u8) __msa_pckev_d((v2i64) dst1, (v2i64) dst0); + + ST_UB(dst0, nxt); + nxt += 16; + } +} + +void png_read_filter_row_paeth3_msa(png_row_infop row_info, + png_bytep row, + png_const_bytep prev_row) +{ + int32_t count, rp_end; + png_bytep nxt; + png_const_bytep prev_nxt; + int64_t out0; + int32_t inp0, inp1, out1; + int16_t out2; + v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, dst0, dst1; + v16u8 src10, src11, src12, src13; + v8i16 vec0, vec1, vec2; + v16u8 zero = { 0 }; + v16i8 mask0 = { 0, 1, 2, 16, 17, 18, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + v16i8 mask1 = { 0, 1, 2, 3, 4, 5, 16, 17, 18, 19, 20, 21, 0, 0, 0, 0 }; + + nxt = row; + prev_nxt = prev_row; + + inp0 = LW(nxt); + inp1 = LW(prev_nxt); + prev_nxt += 3; + src0 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp0); + src1 = (v16u8) __msa_insert_w((v4i32) zero, 0, inp1); + + src1 += src0; + out2 = __msa_copy_s_h((v8i16) src1, 0); + + SH(out2, nxt); + nxt += 2; + nxt[0] = src1[2]; + nxt++; + + /* Remainder */ + rp_end = row_info->rowbytes - 3; + + for (count = 0; count < rp_end; count += 12) + { + src2 = LD_UB(prev_nxt); + prev_nxt += 12; + src6 = LD_UB(prev_row); + prev_row += 12; + src10 = LD_UB(nxt); + + SLDI_B3_0_UB(src2, src6, src10, src3, src7, src11, 3); + SLDI_B3_0_UB(src2, src6, src10, src4, src8, src12, 6); + SLDI_B3_0_UB(src2, src6, src10, src5, src9, src13, 9); + ILVR_B2_SH(src2, src6, src1, src6, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src1, src2, src6, src10); + ILVR_B2_SH(src3, src7, src10, src7, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src10, src3, src7, src11); + ILVR_B2_SH(src4, src8, src11, src8, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src11, src4, src8, src12); + ILVR_B2_SH(src5, src9, src12, src9, vec0, vec1); + HSUB_UB2_SH(vec0, vec1, vec0, vec1); + vec2 = vec0 + vec1; + ADD_ABS_H3_SH(vec0, vec1, vec2, vec0, vec1, vec2); + CMP_AND_SELECT(vec0, vec1, vec2, src12, src5, src9, src13); + src1 = src13; + VSHF_B2_UB(src10, src11, src12, src13, mask0, mask0, dst0, dst1); + dst0 = (v16u8) __msa_vshf_b(mask1, (v16i8) dst1, (v16i8) dst0); + out0 = __msa_copy_s_d((v2i64) dst0, 0); + out1 = __msa_copy_s_w((v4i32) dst0, 2); + + SD(out0, nxt); + nxt += 8; + SW(out1, nxt); + nxt += 4; + } +} + +#endif /* PNG_MIPS_MSA_OPT > 0 */ +#endif /* PNG_MIPS_MSA_IMPLEMENTATION == 1 (intrinsics) */ +#endif /* READ */ diff --git a/Externals/libpng/mips/mips_init.c b/Externals/libpng/mips/mips_init.c new file mode 100644 index 0000000000..8dd283deef --- /dev/null +++ b/Externals/libpng/mips/mips_init.c @@ -0,0 +1,130 @@ + +/* mips_init.c - MSA optimised filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2016 Glenn Randers-Pehrson + * Written by Mandar Sahastrabuddhe, 2016. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +/* Below, after checking __linux__, various non-C90 POSIX 1003.1 functions are + * called. + */ +#define _POSIX_SOURCE 1 + +#include +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED + +#if PNG_MIPS_MSA_OPT > 0 +#ifdef PNG_MIPS_MSA_CHECK_SUPPORTED /* Do run-time checks */ +/* WARNING: it is strongly recommended that you do not build libpng with + * run-time checks for CPU features if at all possible. In the case of the MIPS + * MSA instructions there is no processor-specific way of detecting the + * presence of the required support, therefore run-time detection is extremely + * OS specific. + * + * You may set the macro PNG_MIPS_MSA_FILE to the file name of file containing + * a fragment of C source code which defines the png_have_msa function. There + * are a number of implementations in contrib/mips-msa, but the only one that + * has partial support is contrib/mips-msa/linux.c - a generic Linux + * implementation which reads /proc/cpufino. + */ +#ifndef PNG_MIPS_MSA_FILE +# ifdef __linux__ +# define PNG_MIPS_MSA_FILE "contrib/mips-msa/linux.c" +# endif +#endif + +#ifdef PNG_MIPS_MSA_FILE + +#include /* for sig_atomic_t */ +static int png_have_msa(png_structp png_ptr); +#include PNG_MIPS_MSA_FILE + +#else /* PNG_MIPS_MSA_FILE */ +# error "PNG_MIPS_MSA_FILE undefined: no support for run-time MIPS MSA checks" +#endif /* PNG_MIPS_MSA_FILE */ +#endif /* PNG_MIPS_MSA_CHECK_SUPPORTED */ + +#ifndef PNG_ALIGNED_MEMORY_SUPPORTED +# error "ALIGNED_MEMORY is required; set: -DPNG_ALIGNED_MEMORY_SUPPORTED" +#endif + +void +png_init_filter_functions_msa(png_structp pp, unsigned int bpp) +{ + /* The switch statement is compiled in for MIPS_MSA_API, the call to + * png_have_msa is compiled in for MIPS_MSA_CHECK. If both are defined + * the check is only performed if the API has not set the MSA option on + * or off explicitly. In this case the check controls what happens. + */ + +#ifdef PNG_MIPS_MSA_API_SUPPORTED + switch ((pp->options >> PNG_MIPS_MSA) & 3) + { + case PNG_OPTION_UNSET: + /* Allow the run-time check to execute if it has been enabled - + * thus both API and CHECK can be turned on. If it isn't supported + * this case will fall through to the 'default' below, which just + * returns. + */ +#endif /* PNG_MIPS_MSA_API_SUPPORTED */ +#ifdef PNG_MIPS_MSA_CHECK_SUPPORTED + { + static volatile sig_atomic_t no_msa = -1; /* not checked */ + + if (no_msa < 0) + no_msa = !png_have_msa(pp); + + if (no_msa) + return; + } +#ifdef PNG_MIPS_MSA_API_SUPPORTED + break; +#endif +#endif /* PNG_MIPS_MSA_CHECK_SUPPORTED */ + +#ifdef PNG_MIPS_MSA_API_SUPPORTED + default: /* OFF or INVALID */ + return; + + case PNG_OPTION_ON: + /* Option turned on */ + break; + } +#endif + + /* IMPORTANT: any new external functions used here must be declared using + * PNG_INTERNAL_FUNCTION in ../pngpriv.h. This is required so that the + * 'prefix' option to configure works: + * + * ./configure --with-libpng-prefix=foobar_ + * + * Verify you have got this right by running the above command, doing a build + * and examining pngprefix.h; it must contain a #define for every external + * function you add. (Notice that this happens automatically for the + * initialization function.) + */ + pp->read_filter[PNG_FILTER_VALUE_UP-1] = png_read_filter_row_up_msa; + + if (bpp == 3) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub3_msa; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg3_msa; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = png_read_filter_row_paeth3_msa; + } + + else if (bpp == 4) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub4_msa; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg4_msa; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = png_read_filter_row_paeth4_msa; + } +} +#endif /* PNG_MIPS_MSA_OPT > 0 */ +#endif /* READ */ diff --git a/Externals/libpng/png.c b/Externals/libpng/png.c index 2c9fea2123..757c755f97 100644 --- a/Externals/libpng/png.c +++ b/Externals/libpng/png.c @@ -1,10 +1,10 @@ /* png.c - location for general purpose libpng functions * - * Last changed in libpng 1.6.29 [March 16, 2017] - * Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -14,7 +14,27 @@ #include "pngpriv.h" /* Generate a compiler error if there is an old png.h in the search path. */ -typedef png_libpng_version_1_6_29 Your_png_h_is_not_version_1_6_29; +typedef png_libpng_version_1_6_37 Your_png_h_is_not_version_1_6_37; + +#ifdef __GNUC__ +/* The version tests may need to be added to, but the problem warning has + * consistently been fixed in GCC versions which obtain wide-spread release. + * The problem is that many versions of GCC rearrange comparison expressions in + * the optimizer in such a way that the results of the comparison will change + * if signed integer overflow occurs. Such comparisons are not permitted in + * ANSI C90, however GCC isn't clever enough to work out that that do not occur + * below in png_ascii_from_fp and png_muldiv, so it produces a warning with + * -Wextra. Unfortunately this is highly dependent on the optimizer and the + * machine architecture so the warning comes and goes unpredictably and is + * impossible to "fix", even were that a good idea. + */ +#if __GNUC__ == 7 && __GNUC_MINOR__ == 1 +#define GCC_STRICT_OVERFLOW 1 +#endif /* GNU 7.1.x */ +#endif /* GNU */ +#ifndef GCC_STRICT_OVERFLOW +#define GCC_STRICT_OVERFLOW 0 +#endif /* Tells libpng that we have already handled the first "num_bytes" bytes * of the PNG file signature. If the PNG data is embedded into another @@ -51,7 +71,7 @@ png_set_sig_bytes(png_structrp png_ptr, int num_bytes) * PNG signature (this is the same behavior as strcmp, memcmp, etc). */ int PNGAPI -png_sig_cmp(png_const_bytep sig, png_size_t start, png_size_t num_to_check) +png_sig_cmp(png_const_bytep sig, size_t start, size_t num_to_check) { png_byte png_signature[8] = {137, 80, 78, 71, 13, 10, 26, 10}; @@ -116,7 +136,7 @@ png_reset_crc(png_structrp png_ptr) * trouble of calculating it. */ void /* PRIVATE */ -png_calculate_crc(png_structrp png_ptr, png_const_bytep ptr, png_size_t length) +png_calculate_crc(png_structrp png_ptr, png_const_bytep ptr, size_t length) { int need_crc = 1; @@ -401,7 +421,7 @@ png_destroy_info_struct(png_const_structrp png_ptr, png_infopp info_ptr_ptr) * those cases where it does anything other than a memset. */ PNG_FUNCTION(void,PNGAPI -png_info_init_3,(png_infopp ptr_ptr, png_size_t png_info_struct_size), +png_info_init_3,(png_infopp ptr_ptr, size_t png_info_struct_size), PNG_DEPRECATED) { png_inforp info_ptr = *ptr_ptr; @@ -595,6 +615,26 @@ png_free_data(png_const_structrp png_ptr, png_inforp info_ptr, png_uint_32 mask, } #endif +#ifdef PNG_eXIf_SUPPORTED + /* Free any eXIf entry */ + if (((mask & PNG_FREE_EXIF) & info_ptr->free_me) != 0) + { +# ifdef PNG_READ_eXIf_SUPPORTED + if (info_ptr->eXIf_buf) + { + png_free(png_ptr, info_ptr->eXIf_buf); + info_ptr->eXIf_buf = NULL; + } +# endif + if (info_ptr->exif) + { + png_free(png_ptr, info_ptr->exif); + info_ptr->exif = NULL; + } + info_ptr->valid &= ~PNG_INFO_eXIf; + } +#endif + #ifdef PNG_hIST_SUPPORTED /* Free any hIST entry */ if (((mask & PNG_FREE_HIST) & info_ptr->free_me) != 0) @@ -696,7 +736,7 @@ png_save_int_32(png_bytep buf, png_int_32 i) int PNGAPI png_convert_to_rfc1123_buffer(char out[29], png_const_timep ptime) { - static PNG_CONST char short_months[12][4] = + static const char short_months[12][4] = {"Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"}; @@ -774,20 +814,14 @@ png_get_copyright(png_const_structrp png_ptr) #ifdef PNG_STRING_COPYRIGHT return PNG_STRING_COPYRIGHT #else -# ifdef __STDC__ return PNG_STRING_NEWLINE \ - "libpng version 1.6.29 - March 16, 2017" PNG_STRING_NEWLINE \ - "Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson" \ + "libpng version 1.6.37" PNG_STRING_NEWLINE \ + "Copyright (c) 2018-2019 Cosmin Truta" PNG_STRING_NEWLINE \ + "Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson" \ PNG_STRING_NEWLINE \ "Copyright (c) 1996-1997 Andreas Dilger" PNG_STRING_NEWLINE \ "Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc." \ PNG_STRING_NEWLINE; -# else - return "libpng version 1.6.29 - March 16, 2017\ - Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson\ - Copyright (c) 1996-1997 Andreas Dilger\ - Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc."; -# endif #endif } @@ -902,7 +936,7 @@ png_handle_as_unknown(png_const_structrp png_ptr, png_const_bytep chunk_name) /* The code is the fifth byte after each four byte string. Historically this * code was always searched from the end of the list, this is no longer - * necessary because the 'set' routine handles duplicate entries correcty. + * necessary because the 'set' routine handles duplicate entries correctly. */ do /* num_chunk_list > 0, so at least one */ { @@ -1081,7 +1115,7 @@ png_colorspace_set_gamma(png_const_structrp png_ptr, png_colorspacerp colorspace, png_fixed_point gAMA) { /* Changed in libpng-1.5.4 to limit the values to ensure overflow can't - * occur. Since the fixed point representation is asymetrical it is + * occur. Since the fixed point representation is asymmetrical it is * possible for 1/gamma to overflow the limit of 21474 and this means the * gamma value must be at least 5/100000 and hence at most 20000.0. For * safety the limits here are a little narrower. The values are 0.00016 to @@ -1873,12 +1907,12 @@ png_colorspace_set_sRGB(png_const_structrp png_ptr, png_colorspacerp colorspace, */ if (intent < 0 || intent >= PNG_sRGB_INTENT_LAST) return png_icc_profile_error(png_ptr, colorspace, "sRGB", - (unsigned)intent, "invalid sRGB rendering intent"); + (png_alloc_size_t)intent, "invalid sRGB rendering intent"); if ((colorspace->flags & PNG_COLORSPACE_HAVE_INTENT) != 0 && colorspace->rendering_intent != intent) return png_icc_profile_error(png_ptr, colorspace, "sRGB", - (unsigned)intent, "inconsistent rendering intents"); + (png_alloc_size_t)intent, "inconsistent rendering intents"); if ((colorspace->flags & PNG_COLORSPACE_FROM_sRGB) != 0) { @@ -1939,7 +1973,6 @@ icc_check_length(png_const_structrp png_ptr, png_colorspacerp colorspace, if (profile_length < 132) return png_icc_profile_error(png_ptr, colorspace, name, profile_length, "too short"); - return 1; } @@ -2028,7 +2061,7 @@ png_icc_check_header(png_const_structrp png_ptr, png_colorspacerp colorspace, */ /* Data checks (could be skipped). These checks must be independent of the - * version number; however, the version number doesn't accomodate changes in + * version number; however, the version number doesn't accommodate changes in * the header fields (just the known tags and the interpretation of the * data.) */ @@ -2184,15 +2217,6 @@ png_icc_check_tag_table(png_const_structrp png_ptr, png_colorspacerp colorspace, * being in range. All defined tag types have an 8 byte header - a 4 byte * type signature then 0. */ - if ((tag_start & 3) != 0) - { - /* CNHP730S.icc shipped with Microsoft Windows 64 violates this, it is - * only a warning here because libpng does not care about the - * alignment. - */ - (void)png_icc_profile_error(png_ptr, NULL, name, tag_id, - "ICC profile tag start not a multiple of 4"); - } /* This is a hard error; potentially it can cause read outside the * profile. @@ -2200,6 +2224,16 @@ png_icc_check_tag_table(png_const_structrp png_ptr, png_colorspacerp colorspace, if (tag_start > profile_length || tag_length > profile_length - tag_start) return png_icc_profile_error(png_ptr, colorspace, name, tag_id, "ICC profile tag outside profile"); + + if ((tag_start & 3) != 0) + { + /* CNHP730S.icc shipped with Microsoft Windows 64 violates this; it is + * only a warning here because libpng does not care about the + * alignment. + */ + (void)png_icc_profile_error(png_ptr, NULL, name, tag_id, + "ICC profile tag start not a multiple of 4"); + } } return 1; /* success, maybe with warnings */ @@ -2667,7 +2701,7 @@ png_check_IHDR(png_const_structrp png_ptr, #if defined(PNG_sCAL_SUPPORTED) || defined(PNG_pCAL_SUPPORTED) /* ASCII to fp functions */ -/* Check an ASCII formated floating point value, see the more detailed +/* Check an ASCII formatted floating point value, see the more detailed * comments in pngpriv.h */ /* The following is used internally to preserve the sticky flags */ @@ -2675,11 +2709,11 @@ png_check_IHDR(png_const_structrp png_ptr, #define png_fp_set(state, value) ((state) = (value) | ((state) & PNG_FP_STICKY)) int /* PRIVATE */ -png_check_fp_number(png_const_charp string, png_size_t size, int *statep, +png_check_fp_number(png_const_charp string, size_t size, int *statep, png_size_tp whereami) { int state = *statep; - png_size_t i = *whereami; + size_t i = *whereami; while (i < size) { @@ -2802,10 +2836,10 @@ PNG_FP_End: /* The same but for a complete string. */ int -png_check_fp_string(png_const_charp string, png_size_t size) +png_check_fp_string(png_const_charp string, size_t size) { int state=0; - png_size_t char_index=0; + size_t char_index=0; if (png_check_fp_number(string, size, &state, &char_index) != 0 && (char_index == size || string[char_index] == 0)) @@ -2832,7 +2866,7 @@ png_pow10(int power) if (power < 0) { if (power < DBL_MIN_10_EXP) return 0; - recip = 1, power = -power; + recip = 1; power = -power; } if (power > 0) @@ -2857,8 +2891,16 @@ png_pow10(int power) /* Function to format a floating point value in ASCII with a given * precision. */ +#if GCC_STRICT_OVERFLOW +#pragma GCC diagnostic push +/* The problem arises below with exp_b10, which can never overflow because it + * comes, originally, from frexp and is therefore limited to a range which is + * typically +/-710 (log2(DBL_MAX)/log2(DBL_MIN)). + */ +#pragma GCC diagnostic warning "-Wstrict-overflow=2" +#endif /* GCC_STRICT_OVERFLOW */ void /* PRIVATE */ -png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, +png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, size_t size, double fp, unsigned int precision) { /* We use standard functions from math.h, but not printf because @@ -2910,7 +2952,9 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, double test = png_pow10(exp_b10+1); if (test <= DBL_MAX) - ++exp_b10, base = test; + { + ++exp_b10; base = test; + } else break; @@ -2924,7 +2968,10 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, * test on DBL_MAX above. */ fp /= base; - while (fp >= 1) fp /= 10, ++exp_b10; + while (fp >= 1) + { + fp /= 10; ++exp_b10; + } /* Because of the code above fp may, at this point, be * less than .1, this is ok because the code below can @@ -2941,7 +2988,7 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, */ if (exp_b10 < 0 && exp_b10 > -3) /* PLUS 3 TOTAL 4 */ { - czero = (unsigned int)(-exp_b10); /* PLUS 2 digits: TOTAL 3 */ + czero = 0U-exp_b10; /* PLUS 2 digits: TOTAL 3 */ exp_b10 = 0; /* Dot added below before first output. */ } else @@ -2975,7 +3022,7 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, /* Rounding up to 10, handle that here. */ if (czero > 0) { - --czero, d = 1; + --czero; d = 1; if (cdigits == 0) --clead; } else @@ -2989,7 +3036,7 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, else if (ch == 46) { - ch = *--ascii, ++size; + ch = *--ascii; ++size; /* Advance exp_b10 to '1', so that the * decimal point happens after the * previous digit. @@ -3016,7 +3063,9 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, int ch = *--ascii; if (ch == 46) - ++size, exp_b10 = 1; + { + ++size; exp_b10 = 1; + } /* Else lost a leading zero, so 'exp_b10' is * still ok at (-1) @@ -3052,21 +3101,26 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, */ if (exp_b10 != (-1)) { - if (exp_b10 == 0) *ascii++ = 46, --size; + if (exp_b10 == 0) + { + *ascii++ = 46; --size; + } /* PLUS 1: TOTAL 4 */ --exp_b10; } - *ascii++ = 48, --czero; + *ascii++ = 48; --czero; } if (exp_b10 != (-1)) { if (exp_b10 == 0) - *ascii++ = 46, --size; /* counted above */ + { + *ascii++ = 46; --size; /* counted above */ + } --exp_b10; } - *ascii++ = (char)(48 + (int)d), ++cdigits; + *ascii++ = (char)(48 + (int)d); ++cdigits; } } while (cdigits+czero < precision+clead && fp > DBL_MIN); @@ -3074,11 +3128,11 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, /* The total output count (max) is now 4+precision */ /* Check for an exponent, if we don't need one we are - * done and just need to terminate the string. At - * this point exp_b10==(-1) is effectively if flag - it got - * to '-1' because of the decrement after outputting - * the decimal point above (the exponent required is - * *not* -1!) + * done and just need to terminate the string. At this + * point, exp_b10==(-1) is effectively a flag: it got + * to '-1' because of the decrement, after outputting + * the decimal point above. (The exponent required is + * *not* -1.) */ if (exp_b10 >= (-1) && exp_b10 <= 2) { @@ -3089,7 +3143,7 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, * zeros were *not* output, so this doesn't increase * the output count. */ - while (--exp_b10 >= 0) *ascii++ = 48; + while (exp_b10-- > 0) *ascii++ = 48; *ascii = 0; @@ -3107,7 +3161,7 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, */ size -= cdigits; - *ascii++ = 69, --size; /* 'E': PLUS 1 TOTAL 2+precision */ + *ascii++ = 69; --size; /* 'E': PLUS 1 TOTAL 2+precision */ /* The following use of an unsigned temporary avoids ambiguities in * the signed arithmetic on exp_b10 and permits GCC at least to do @@ -3118,12 +3172,12 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, if (exp_b10 < 0) { - *ascii++ = 45, --size; /* '-': PLUS 1 TOTAL 3+precision */ - uexp_b10 = (unsigned int)(-exp_b10); + *ascii++ = 45; --size; /* '-': PLUS 1 TOTAL 3+precision */ + uexp_b10 = 0U-exp_b10; } else - uexp_b10 = (unsigned int)exp_b10; + uexp_b10 = 0U+exp_b10; cdigits = 0; @@ -3166,6 +3220,9 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, /* Here on buffer too small. */ png_error(png_ptr, "ASCII conversion buffer too small"); } +#if GCC_STRICT_OVERFLOW +#pragma GCC diagnostic pop +#endif /* GCC_STRICT_OVERFLOW */ # endif /* FLOATING_POINT */ @@ -3174,7 +3231,7 @@ png_ascii_from_fp(png_const_structrp png_ptr, png_charp ascii, png_size_t size, */ void /* PRIVATE */ png_ascii_from_fixed(png_const_structrp png_ptr, png_charp ascii, - png_size_t size, png_fixed_point fp) + size_t size, png_fixed_point fp) { /* Require space for 10 decimal digits, a decimal point, a minus sign and a * trailing \0, 13 characters: @@ -3185,7 +3242,9 @@ png_ascii_from_fixed(png_const_structrp png_ptr, png_charp ascii, /* Avoid overflow here on the minimum integer. */ if (fp < 0) - *ascii++ = 45, num = (png_uint_32)(-fp); + { + *ascii++ = 45; num = (png_uint_32)(-fp); + } else num = (png_uint_32)fp; @@ -3223,7 +3282,10 @@ png_ascii_from_fixed(png_const_structrp png_ptr, png_charp ascii, * then ndigits digits to first: */ i = 5; - while (ndigits < i) *ascii++ = 48, --i; + while (ndigits < i) + { + *ascii++ = 48; --i; + } while (ndigits >= first) *ascii++ = digits[--ndigits]; /* Don't output the trailing zeros! */ } @@ -3274,6 +3336,15 @@ png_fixed(png_const_structrp png_ptr, double fp, png_const_charp text) * the nearest .00001). Overflow and divide by zero are signalled in * the result, a boolean - true on success, false on overflow. */ +#if GCC_STRICT_OVERFLOW /* from above */ +/* It is not obvious which comparison below gets optimized in such a way that + * signed overflow would change the result; looking through the code does not + * reveal any tests which have the form GCC complains about, so presumably the + * optimizer is moving an add or subtract into the 'if' somewhere. + */ +#pragma GCC diagnostic push +#pragma GCC diagnostic warning "-Wstrict-overflow=2" +#endif /* GCC_STRICT_OVERFLOW */ int png_muldiv(png_fixed_point_p res, png_fixed_point a, png_int_32 times, png_int_32 divisor) @@ -3388,6 +3459,9 @@ png_muldiv(png_fixed_point_p res, png_fixed_point a, png_int_32 times, return 0; } +#if GCC_STRICT_OVERFLOW +#pragma GCC diagnostic pop +#endif /* GCC_STRICT_OVERFLOW */ #endif /* READ_GAMMA || INCH_CONVERSIONS */ #if defined(PNG_READ_GAMMA_SUPPORTED) || defined(PNG_INCH_CONVERSIONS_SUPPORTED) @@ -3681,7 +3755,7 @@ png_log16bit(png_uint_32 x) * of getting this accuracy in practice. * * To deal with this the following exp() function works out the exponent of the - * frational part of the logarithm by using an accurate 32-bit value from the + * fractional part of the logarithm by using an accurate 32-bit value from the * top four fractional bits then multiplying in the remaining bits. */ static const png_uint_32 @@ -3896,18 +3970,18 @@ png_gamma_correct(png_structrp png_ptr, unsigned int value, */ static void png_build_16bit_table(png_structrp png_ptr, png_uint_16pp *ptable, - PNG_CONST unsigned int shift, PNG_CONST png_fixed_point gamma_val) + unsigned int shift, png_fixed_point gamma_val) { /* Various values derived from 'shift': */ - PNG_CONST unsigned int num = 1U << (8U - shift); + unsigned int num = 1U << (8U - shift); #ifdef PNG_FLOATING_ARITHMETIC_SUPPORTED /* CSE the division and work round wacky GCC warnings (see the comments * in png_gamma_8bit_correct for where these come from.) */ - PNG_CONST double fmax = 1./(((png_int_32)1 << (16U - shift))-1); + double fmax = 1.0 / (((png_int_32)1 << (16U - shift)) - 1); #endif - PNG_CONST unsigned int max = (1U << (16U - shift))-1U; - PNG_CONST unsigned int max_by_2 = 1U << (15U-shift); + unsigned int max = (1U << (16U - shift)) - 1U; + unsigned int max_by_2 = 1U << (15U - shift); unsigned int i; png_uint_16pp table = *ptable = @@ -3973,10 +4047,10 @@ png_build_16bit_table(png_structrp png_ptr, png_uint_16pp *ptable, */ static void png_build_16to8_table(png_structrp png_ptr, png_uint_16pp *ptable, - PNG_CONST unsigned int shift, PNG_CONST png_fixed_point gamma_val) + unsigned int shift, png_fixed_point gamma_val) { - PNG_CONST unsigned int num = 1U << (8U - shift); - PNG_CONST unsigned int max = (1U << (16U - shift))-1U; + unsigned int num = 1U << (8U - shift); + unsigned int max = (1U << (16U - shift))-1U; unsigned int i; png_uint_32 last; @@ -4041,7 +4115,7 @@ png_build_16to8_table(png_structrp png_ptr, png_uint_16pp *ptable, */ static void png_build_8bit_table(png_structrp png_ptr, png_bytepp ptable, - PNG_CONST png_fixed_point gamma_val) + png_fixed_point gamma_val) { unsigned int i; png_bytep table = *ptable = (png_bytep)png_malloc(png_ptr, 256); @@ -4264,7 +4338,7 @@ png_set_option(png_structrp png_ptr, int option, int onoff) png_uint_32 setting = (2U + (onoff != 0)) << option; png_uint_32 current = png_ptr->options; - png_ptr->options = (png_uint_32)(((current & ~mask) | setting) & 0xff); + png_ptr->options = (png_uint_32)((current & ~mask) | setting); return (int)(current & mask) >> option; } @@ -4278,7 +4352,7 @@ png_set_option(png_structrp png_ptr, int option, int onoff) defined(PNG_SIMPLIFIED_WRITE_SUPPORTED) /* sRGB conversion tables; these are machine generated with the code in * contrib/tools/makesRGB.c. The actual sRGB transfer curve defined in the - * specification (see the article at http://en.wikipedia.org/wiki/SRGB) + * specification (see the article at https://en.wikipedia.org/wiki/SRGB) * is used, not the gamma=1/2.2 approximation use elsewhere in libpng. * The sRGB to linear table is exact (to the nearest 16-bit linear fraction). * The inverse (linear to sRGB) table has accuracies as follows: @@ -4514,8 +4588,7 @@ png_image_free(png_imagep image) if (image != NULL && image->opaque != NULL && image->opaque->error_buf == NULL) { - /* Ignore errors here: */ - (void)png_safe_execute(image, png_image_free_function, image); + png_image_free_function(image); image->opaque = NULL; } } diff --git a/Externals/libpng/png.h b/Externals/libpng/png.h index 2431e0dfc0..139eb0dc0f 100644 --- a/Externals/libpng/png.h +++ b/Externals/libpng/png.h @@ -1,68 +1,105 @@ /* png.h - header file for PNG reference library * - * libpng version 1.6.29, March 16, 2017 + * libpng version 1.6.37 - April 14, 2019 * - * Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * - * This code is released under the libpng license (See LICENSE, below) + * This code is released under the libpng license. (See LICENSE, below.) * * Authors and maintainers: * libpng versions 0.71, May 1995, through 0.88, January 1996: Guy Schalnat * libpng versions 0.89, June 1996, through 0.96, May 1997: Andreas Dilger - * libpng versions 0.97, January 1998, through 1.6.29, March 16, 2017: - * Glenn Randers-Pehrson. + * libpng versions 0.97, January 1998, through 1.6.35, July 2018: + * Glenn Randers-Pehrson + * libpng versions 1.6.36, December 2018, through 1.6.37, April 2019: + * Cosmin Truta * See also "Contributing Authors", below. */ /* - * COPYRIGHT NOTICE, DISCLAIMER, and LICENSE: + * COPYRIGHT NOTICE, DISCLAIMER, and LICENSE + * ========================================= * - * If you modify libpng you may insert additional notices immediately following - * this sentence. + * PNG Reference Library License version 2 + * --------------------------------------- * - * This code is released under the libpng license. + * * Copyright (c) 1995-2019 The PNG Reference Library Authors. + * * Copyright (c) 2018-2019 Cosmin Truta. + * * Copyright (c) 2000-2002, 2004, 2006-2018 Glenn Randers-Pehrson. + * * Copyright (c) 1996-1997 Andreas Dilger. + * * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * - * libpng versions 1.0.7, July 1, 2000 through 1.6.29, March 16, 2017 are - * Copyright (c) 2000-2002, 2004, 2006-2017 Glenn Randers-Pehrson, are + * The software is supplied "as is", without warranty of any kind, + * express or implied, including, without limitation, the warranties + * of merchantability, fitness for a particular purpose, title, and + * non-infringement. In no event shall the Copyright owners, or + * anyone distributing the software, be liable for any damages or + * other liability, whether in contract, tort or otherwise, arising + * from, out of, or in connection with the software, or the use or + * other dealings in the software, even if advised of the possibility + * of such damage. + * + * Permission is hereby granted to use, copy, modify, and distribute + * this software, or portions hereof, for any purpose, without fee, + * subject to the following restrictions: + * + * 1. The origin of this software must not be misrepresented; you + * must not claim that you wrote the original software. If you + * use this software in a product, an acknowledgment in the product + * documentation would be appreciated, but is not required. + * + * 2. Altered source versions must be plainly marked as such, and must + * not be misrepresented as being the original software. + * + * 3. This Copyright notice may not be removed or altered from any + * source or altered source distribution. + * + * + * PNG Reference Library License version 1 (for libpng 0.5 through 1.6.35) + * ----------------------------------------------------------------------- + * + * libpng versions 1.0.7, July 1, 2000, through 1.6.35, July 15, 2018 are + * Copyright (c) 2000-2002, 2004, 2006-2018 Glenn Randers-Pehrson, are * derived from libpng-1.0.6, and are distributed according to the same * disclaimer and license as libpng-1.0.6 with the following individuals * added to the list of Contributing Authors: * - * Simon-Pierre Cadieux - * Eric S. Raymond - * Mans Rullgard - * Cosmin Truta - * Gilles Vollant - * James Yu - * Mandar Sahastrabuddhe - * Google Inc. - * Vadim Barkov + * Simon-Pierre Cadieux + * Eric S. Raymond + * Mans Rullgard + * Cosmin Truta + * Gilles Vollant + * James Yu + * Mandar Sahastrabuddhe + * Google Inc. + * Vadim Barkov * * and with the following additions to the disclaimer: * - * There is no warranty against interference with your enjoyment of the - * library or against infringement. There is no warranty that our - * efforts or the library will fulfill any of your particular purposes - * or needs. This library is provided with all faults, and the entire - * risk of satisfactory quality, performance, accuracy, and effort is with - * the user. + * There is no warranty against interference with your enjoyment of + * the library or against infringement. There is no warranty that our + * efforts or the library will fulfill any of your particular purposes + * or needs. This library is provided with all faults, and the entire + * risk of satisfactory quality, performance, accuracy, and effort is + * with the user. * * Some files in the "contrib" directory and some configure-generated - * files that are distributed with libpng have other copyright owners and + * files that are distributed with libpng have other copyright owners, and * are released under other open source licenses. * * libpng versions 0.97, January 1998, through 1.0.6, March 20, 2000, are * Copyright (c) 1998-2000 Glenn Randers-Pehrson, are derived from * libpng-0.96, and are distributed according to the same disclaimer and - * license as libpng-0.96, with the following individuals added to the list - * of Contributing Authors: + * license as libpng-0.96, with the following individuals added to the + * list of Contributing Authors: * - * Tom Lane - * Glenn Randers-Pehrson - * Willem van Schaik + * Tom Lane + * Glenn Randers-Pehrson + * Willem van Schaik * * libpng versions 0.89, June 1996, through 0.96, May 1997, are * Copyright (c) 1996-1997 Andreas Dilger, are derived from libpng-0.88, @@ -70,14 +107,14 @@ * libpng-0.88, with the following individuals added to the list of * Contributing Authors: * - * John Bowler - * Kevin Bracey - * Sam Bushell - * Magnus Holmgren - * Greg Roelofs - * Tom Tanner + * John Bowler + * Kevin Bracey + * Sam Bushell + * Magnus Holmgren + * Greg Roelofs + * Tom Tanner * - * Some files in the "scripts" directory have other copyright owners + * Some files in the "scripts" directory have other copyright owners, * but are released under this license. * * libpng versions 0.5, May 1995, through 0.88, January 1996, are @@ -86,62 +123,49 @@ * For the purposes of this copyright and license, "Contributing Authors" * is defined as the following set of individuals: * - * Andreas Dilger - * Dave Martindale - * Guy Eric Schalnat - * Paul Schmidt - * Tim Wegner + * Andreas Dilger + * Dave Martindale + * Guy Eric Schalnat + * Paul Schmidt + * Tim Wegner * - * The PNG Reference Library is supplied "AS IS". The Contributing Authors - * and Group 42, Inc. disclaim all warranties, expressed or implied, - * including, without limitation, the warranties of merchantability and of - * fitness for any purpose. The Contributing Authors and Group 42, Inc. - * assume no liability for direct, indirect, incidental, special, exemplary, - * or consequential damages, which may result from the use of the PNG - * Reference Library, even if advised of the possibility of such damage. + * The PNG Reference Library is supplied "AS IS". The Contributing + * Authors and Group 42, Inc. disclaim all warranties, expressed or + * implied, including, without limitation, the warranties of + * merchantability and of fitness for any purpose. The Contributing + * Authors and Group 42, Inc. assume no liability for direct, indirect, + * incidental, special, exemplary, or consequential damages, which may + * result from the use of the PNG Reference Library, even if advised of + * the possibility of such damage. * * Permission is hereby granted to use, copy, modify, and distribute this * source code, or portions hereof, for any purpose, without fee, subject * to the following restrictions: * - * 1. The origin of this source code must not be misrepresented. + * 1. The origin of this source code must not be misrepresented. * - * 2. Altered versions must be plainly marked as such and must not - * be misrepresented as being the original source. + * 2. Altered versions must be plainly marked as such and must not + * be misrepresented as being the original source. * - * 3. This Copyright notice may not be removed or altered from any - * source or altered source distribution. + * 3. This Copyright notice may not be removed or altered from any + * source or altered source distribution. * - * The Contributing Authors and Group 42, Inc. specifically permit, without - * fee, and encourage the use of this source code as a component to - * supporting the PNG file format in commercial products. If you use this - * source code in a product, acknowledgment is not required but would be - * appreciated. + * The Contributing Authors and Group 42, Inc. specifically permit, + * without fee, and encourage the use of this source code as a component + * to supporting the PNG file format in commercial products. If you use + * this source code in a product, acknowledgment is not required but would + * be appreciated. * * END OF COPYRIGHT NOTICE, DISCLAIMER, and LICENSE. * - * TRADEMARK: + * TRADEMARK + * ========= * - * The name "libpng" has not been registered by the Copyright owner + * The name "libpng" has not been registered by the Copyright owners * as a trademark in any jurisdiction. However, because libpng has * been distributed and maintained world-wide, continually since 1995, - * the Copyright owner claims "common-law trademark protection" in any + * the Copyright owners claim "common-law trademark protection" in any * jurisdiction where common-law trademark is recognized. - * - * OSI CERTIFICATION: - * - * Libpng is OSI Certified Open Source Software. OSI Certified Open Source is - * a certification mark of the Open Source Initiative. OSI has not addressed - * the additional disclaimers inserted at version 1.0.7. - * - * EXPORT CONTROL: - * - * The Copyright owner believes that the Export Control Classification - * Number (ECCN) for libpng is EAR99, which means not subject to export - * controls or International Traffic in Arms Regulations (ITAR) because - * it is open source, publicly available software, that does not contain - * any encryption software. See the EAR, paragraphs 734.3(b)(3) and - * 734.7(b). */ /* @@ -207,23 +231,25 @@ * 1.0.7rc1-2 1 10007 2.1.0.7rc1-2 (binary compatible) * 1.0.7 1 10007 (still compatible) * ... - * 1.0.19 10 10019 10.so.0.19[.0] + * 1.0.69 10 10069 10.so.0.69[.0] * ... - * 1.2.57 13 10257 12.so.0.57[.0] + * 1.2.59 13 10259 12.so.0.59[.0] * ... - * 1.5.28 15 10527 15.so.15.28[.0] + * 1.4.20 14 10420 14.so.0.20[.0] * ... - * 1.6.29 16 10629 16.so.16.29[.0] + * 1.5.30 15 10530 15.so.15.30[.0] + * ... + * 1.6.37 16 10637 16.so.16.37[.0] * - * Henceforth the source version will match the shared-library major - * and minor numbers; the shared-library major version number will be - * used for changes in backward compatibility, as it is intended. The - * PNG_LIBPNG_VER macro, which is not used within libpng but is available - * for applications, is an unsigned integer of the form xyyzz corresponding - * to the source version x.y.z (leading zeros in y and z). Beta versions - * were given the previous public release number plus a letter, until - * version 1.0.6j; from then on they were given the upcoming public - * release number plus "betaNN" or "rcNN". + * Henceforth the source version will match the shared-library major and + * minor numbers; the shared-library major version number will be used for + * changes in backward compatibility, as it is intended. + * The PNG_LIBPNG_VER macro, which is not used within libpng but is + * available for applications, is an unsigned integer of the form XYYZZ + * corresponding to the source version X.Y.Z (leading zeros in Y and Z). + * Beta versions were given the previous public release number plus a + * letter, until version 1.0.6j; from then on they were given the upcoming + * public release number plus "betaNN" or "rcNN". * * Binary incompatibility exists only when applications make direct access * to the info_ptr or png_ptr members through png.h, and the compiled @@ -233,65 +259,8 @@ * in binary compatibility (e.g., when a new feature is added). * * See libpng.txt or libpng.3 for more information. The PNG specification - * is available as a W3C Recommendation and as an ISO Specification, - * */ #ifndef PNG_H @@ -309,8 +278,8 @@ */ /* Version information for png.h - this should match the version in png.c */ -#define PNG_LIBPNG_VER_STRING "1.6.29" -#define PNG_HEADER_VERSION_STRING " libpng version 1.6.29 - March 16, 2017\n" +#define PNG_LIBPNG_VER_STRING "1.6.37" +#define PNG_HEADER_VERSION_STRING " libpng version 1.6.37 - April 14, 2019\n" #define PNG_LIBPNG_VER_SONUM 16 #define PNG_LIBPNG_VER_DLLNUM 16 @@ -318,12 +287,11 @@ /* These should match the first 3 components of PNG_LIBPNG_VER_STRING: */ #define PNG_LIBPNG_VER_MAJOR 1 #define PNG_LIBPNG_VER_MINOR 6 -#define PNG_LIBPNG_VER_RELEASE 29 +#define PNG_LIBPNG_VER_RELEASE 37 -/* This should match the numeric part of the final component of - * PNG_LIBPNG_VER_STRING, omitting any leading zero: +/* This should be zero for a public release, or non-zero for a + * development version. [Deprecated] */ - #define PNG_LIBPNG_VER_BUILD 0 /* Release Status */ @@ -343,13 +311,14 @@ #define PNG_LIBPNG_BUILD_BASE_TYPE PNG_LIBPNG_BUILD_STABLE -/* Careful here. At one time, Guy wanted to use 082, but that would be octal. - * We must not include leading zeros. - * Versions 0.7 through 1.0.0 were in the range 0 to 100 here (only - * version 1.0.0 was mis-numbered 100 instead of 10000). From - * version 1.0.1 it's xxyyzz, where x=major, y=minor, z=release +/* Careful here. At one time, Guy wanted to use 082, but that + * would be octal. We must not include leading zeros. + * Versions 0.7 through 1.0.0 were in the range 0 to 100 here + * (only version 1.0.0 was mis-numbered 100 instead of 10000). + * From version 1.0.1 it is: + * XXYYZZ, where XX=major, YY=minor, ZZ=release */ -#define PNG_LIBPNG_VER 10629 /* 1.6.29 */ +#define PNG_LIBPNG_VER 10637 /* 1.6.37 */ /* Library configuration: these options cannot be changed after * the library has been built. @@ -459,7 +428,7 @@ extern "C" { /* This triggers a compiler error in png.c, if png.c and png.h * do not agree upon the version number. */ -typedef char* png_libpng_version_1_6_29; +typedef char* png_libpng_version_1_6_37; /* Basic control structions. Read libpng-manual.txt or libpng.3 for more info. * @@ -600,8 +569,8 @@ typedef struct png_text_struct png_charp key; /* keyword, 1-79 character description of "text" */ png_charp text; /* comment, may be an empty string (ie "") or a NULL pointer */ - png_size_t text_length; /* length of the text string */ - png_size_t itxt_length; /* length of the itxt string */ + size_t text_length; /* length of the text string */ + size_t itxt_length; /* length of the itxt string */ png_charp lang; /* language code, 0-79 characters or a NULL pointer */ png_charp lang_key; /* keyword translated UTF-8 string, 0 or more @@ -654,7 +623,7 @@ typedef struct png_unknown_chunk_t { png_byte name[5]; /* Textual chunk name with '\0' terminator */ png_byte *data; /* Data, should not be modified on read! */ - png_size_t size; + size_t size; /* On write 'location' must be set using the flag values listed below. * Notice that on read it is set by libpng however the values stored have @@ -679,7 +648,7 @@ typedef png_unknown_chunk * * png_unknown_chunkpp; /* Maximum positive integer used in PNG is (2^31)-1 */ #define PNG_UINT_31_MAX ((png_uint_32)0x7fffffffL) #define PNG_UINT_32_MAX ((png_uint_32)(-1)) -#define PNG_SIZE_MAX ((png_size_t)(-1)) +#define PNG_SIZE_MAX ((size_t)(-1)) /* These are constants for fixed point values encoded in the * PNG specification manner (x100000) @@ -776,6 +745,7 @@ typedef png_unknown_chunk * * png_unknown_chunkpp; #define PNG_INFO_sPLT 0x2000U /* ESR, 1.0.6 */ #define PNG_INFO_sCAL 0x4000U /* ESR, 1.0.6 */ #define PNG_INFO_IDAT 0x8000U /* ESR, 1.0.6 */ +#define PNG_INFO_eXIf 0x10000U /* GR-P, 1.6.31 */ /* This is used for the transformation routines, as some of them * change these values for the row. It also should enable using @@ -784,7 +754,7 @@ typedef png_unknown_chunk * * png_unknown_chunkpp; typedef struct png_row_info_struct { png_uint_32 width; /* width of row */ - png_size_t rowbytes; /* number of bytes in row */ + size_t rowbytes; /* number of bytes in row */ png_byte color_type; /* color type of row */ png_byte bit_depth; /* bit depth of row */ png_byte channels; /* number of channels (1, 2, 3, or 4) */ @@ -803,7 +773,7 @@ typedef png_row_info * * png_row_infopp; * expected to return the read data in the buffer. */ typedef PNG_CALLBACK(void, *png_error_ptr, (png_structp, png_const_charp)); -typedef PNG_CALLBACK(void, *png_rw_ptr, (png_structp, png_bytep, png_size_t)); +typedef PNG_CALLBACK(void, *png_rw_ptr, (png_structp, png_bytep, size_t)); typedef PNG_CALLBACK(void, *png_flush_ptr, (png_structp)); typedef PNG_CALLBACK(void, *png_read_status_ptr, (png_structp, png_uint_32, int)); @@ -940,8 +910,8 @@ PNG_EXPORT(2, void, png_set_sig_bytes, (png_structrp png_ptr, int num_bytes)); * signature, and non-zero otherwise. Having num_to_check == 0 or * start > 7 will always fail (ie return non-zero). */ -PNG_EXPORT(3, int, png_sig_cmp, (png_const_bytep sig, png_size_t start, - png_size_t num_to_check)); +PNG_EXPORT(3, int, png_sig_cmp, (png_const_bytep sig, size_t start, + size_t num_to_check)); /* Simple signature checking function. This is the same as calling * png_check_sig(sig, n) := !png_sig_cmp(sig, 0, n). @@ -960,11 +930,11 @@ PNG_EXPORTA(5, png_structp, png_create_write_struct, png_error_ptr warn_fn), PNG_ALLOCATED); -PNG_EXPORT(6, png_size_t, png_get_compression_buffer_size, +PNG_EXPORT(6, size_t, png_get_compression_buffer_size, (png_const_structrp png_ptr)); PNG_EXPORT(7, void, png_set_compression_buffer_size, (png_structrp png_ptr, - png_size_t size)); + size_t size)); /* Moved from pngconf.h in 1.4.0 and modified to ensure setjmp/longjmp * match up. @@ -1017,7 +987,7 @@ PNG_EXPORT(13, void, png_write_sig, (png_structrp png_ptr)); /* Write a PNG chunk - size, type, (optional) data, CRC. */ PNG_EXPORT(14, void, png_write_chunk, (png_structrp png_ptr, png_const_bytep - chunk_name, png_const_bytep data, png_size_t length)); + chunk_name, png_const_bytep data, size_t length)); /* Write the start of a PNG chunk - length and chunk name. */ PNG_EXPORT(15, void, png_write_chunk_start, (png_structrp png_ptr, @@ -1025,7 +995,7 @@ PNG_EXPORT(15, void, png_write_chunk_start, (png_structrp png_ptr, /* Write the data of a PNG chunk started with png_write_chunk_start(). */ PNG_EXPORT(16, void, png_write_chunk_data, (png_structrp png_ptr, - png_const_bytep data, png_size_t length)); + png_const_bytep data, size_t length)); /* Finish a chunk started with png_write_chunk_start() (includes CRC). */ PNG_EXPORT(17, void, png_write_chunk_end, (png_structrp png_ptr)); @@ -1039,7 +1009,7 @@ PNG_EXPORTA(18, png_infop, png_create_info_struct, (png_const_structrp png_ptr), * the API will be removed in the future. */ PNG_EXPORTA(19, void, png_info_init_3, (png_infopp info_ptr, - png_size_t png_info_struct_size), PNG_DEPRECATED); + size_t png_info_struct_size), PNG_DEPRECATED); /* Writes all the PNG information before the image. */ PNG_EXPORT(20, void, png_write_info_before_PLTE, @@ -1136,7 +1106,7 @@ PNG_EXPORT(35, void, png_build_grayscale_palette, (int bit_depth, * corresponding composited pixel, and the color channels are unassociated * (not premultiplied). The gamma encoded color channels must be scaled * according to the contribution and to do this it is necessary to undo - * the encoding, scale the color values, perform the composition and reencode + * the encoding, scale the color values, perform the composition and re-encode * the values. This is the 'PNG' mode. * * The alternative is to 'associate' the alpha with the color information by @@ -1192,7 +1162,7 @@ PNG_FIXED_EXPORT(228, void, png_set_alpha_mode_fixed, (png_structrp png_ptr, * * png_set_alpha_mode(pp, PNG_ALPHA_PNG, PNG_GAMMA_MAC); * In this case the output is assumed to be something like an sRGB conformant - * display preceeded by a power-law lookup table of power 1.45. This is how + * display preceded by a power-law lookup table of power 1.45. This is how * early Mac systems behaved. * * png_set_alpha_mode(pp, PNG_ALPHA_STANDARD, PNG_GAMMA_LINEAR); @@ -1239,7 +1209,7 @@ PNG_FIXED_EXPORT(228, void, png_set_alpha_mode_fixed, (png_structrp png_ptr, * * When the default gamma of PNG files doesn't match the output gamma. * If you have PNG files with no gamma information png_set_alpha_mode allows - * you to provide a default gamma, but it also sets the ouput gamma to the + * you to provide a default gamma, but it also sets the output gamma to the * matching value. If you know your PNG files have a gamma that doesn't * match the output you can take advantage of the fact that * png_set_alpha_mode always sets the output gamma but only sets the PNG @@ -1690,7 +1660,7 @@ PNG_EXPORT(218, png_byte, png_get_current_pass_number, (png_const_structrp)); * chunk will cause an error at this point unless it is to be saved. * positive: The chunk was handled, libpng will ignore/discard it. * - * See "INTERACTION WTIH USER CHUNK CALLBACKS" below for important notes about + * See "INTERACTION WITH USER CHUNK CALLBACKS" below for important notes about * how this behavior will change in libpng 1.7 */ PNG_EXPORT(88, void, png_set_read_user_chunk_fn, (png_structrp png_ptr, @@ -1715,7 +1685,7 @@ PNG_EXPORT(91, png_voidp, png_get_progressive_ptr, /* Function to be called when data becomes available */ PNG_EXPORT(92, void, png_process_data, (png_structrp png_ptr, - png_inforp info_ptr, png_bytep buffer, png_size_t buffer_size)); + png_inforp info_ptr, png_bytep buffer, size_t buffer_size)); /* A function which may be called *only* within png_process_data to stop the * processing of any more data. The function returns the number of bytes @@ -1724,7 +1694,7 @@ PNG_EXPORT(92, void, png_process_data, (png_structrp png_ptr, * 'save' is set to true the routine will first save all the pending data and * will always return 0. */ -PNG_EXPORT(219, png_size_t, png_process_data_pause, (png_structrp, int save)); +PNG_EXPORT(219, size_t, png_process_data_pause, (png_structrp, int save)); /* A function which may be called *only* outside (after) a call to * png_process_data. It returns the number of bytes of data to skip in the @@ -1788,7 +1758,8 @@ PNG_EXPORT(99, void, png_data_freer, (png_const_structrp png_ptr, #define PNG_FREE_PLTE 0x1000U #define PNG_FREE_TRNS 0x2000U #define PNG_FREE_TEXT 0x4000U -#define PNG_FREE_ALL 0x7fffU +#define PNG_FREE_EXIF 0x8000U /* Added at libpng-1.6.31 */ +#define PNG_FREE_ALL 0xffffU #define PNG_FREE_MUL 0x4220U /* PNG_FREE_SPLT|PNG_FREE_TEXT|PNG_FREE_UNKN */ #ifdef PNG_USER_MEM_SUPPORTED @@ -1868,7 +1839,7 @@ PNG_EXPORT(110, png_uint_32, png_get_valid, (png_const_structrp png_ptr, png_const_inforp info_ptr, png_uint_32 flag)); /* Returns number of bytes needed to hold a transformed row. */ -PNG_EXPORT(111, png_size_t, png_get_rowbytes, (png_const_structrp png_ptr, +PNG_EXPORT(111, size_t, png_get_rowbytes, (png_const_structrp png_ptr, png_const_inforp info_ptr)); #ifdef PNG_INFO_IMAGE_SUPPORTED @@ -2007,6 +1978,18 @@ PNG_FIXED_EXPORT(233, void, png_set_cHRM_XYZ_fixed, (png_const_structrp png_ptr, png_fixed_point int_blue_Z)) #endif +#ifdef PNG_eXIf_SUPPORTED +PNG_EXPORT(246, png_uint_32, png_get_eXIf, (png_const_structrp png_ptr, + png_inforp info_ptr, png_bytep *exif)); +PNG_EXPORT(247, void, png_set_eXIf, (png_const_structrp png_ptr, + png_inforp info_ptr, png_bytep exif)); + +PNG_EXPORT(248, png_uint_32, png_get_eXIf_1, (png_const_structrp png_ptr, + png_const_inforp info_ptr, png_uint_32 *num_exif, png_bytep *exif)); +PNG_EXPORT(249, void, png_set_eXIf_1, (png_const_structrp png_ptr, + png_inforp info_ptr, png_uint_32 num_exif, png_bytep exif)); +#endif + #ifdef PNG_gAMA_SUPPORTED PNG_FP_EXPORT(137, png_uint_32, png_get_gAMA, (png_const_structrp png_ptr, png_const_inforp info_ptr, double *file_gamma)) @@ -2025,9 +2008,6 @@ PNG_FIXED_EXPORT(140, void, png_set_gAMA_fixed, (png_const_structrp png_ptr, #ifdef PNG_hIST_SUPPORTED PNG_EXPORT(141, png_uint_32, png_get_hIST, (png_const_structrp png_ptr, png_inforp info_ptr, png_uint_16p *hist)); -#endif - -#ifdef PNG_hIST_SUPPORTED PNG_EXPORT(142, void, png_set_hIST, (png_const_structrp png_ptr, png_inforp info_ptr, png_const_uint_16p hist)); #endif @@ -2228,7 +2208,7 @@ PNG_EXPORT(171, void, png_set_sCAL_s, (png_const_structrp png_ptr, * to specifying "NEVER", however when "AS_DEFAULT" is used for specific chunks * it simply resets the behavior to the libpng default. * - * INTERACTION WTIH USER CHUNK CALLBACKS: + * INTERACTION WITH USER CHUNK CALLBACKS: * The per-chunk handling is always used when there is a png_user_chunk_ptr * callback and the callback returns 0; the chunk is then always stored *unless* * it is critical and the per-chunk setting is other than ALWAYS. Notice that @@ -2647,7 +2627,7 @@ PNG_EXPORT(243, int, png_get_palette_max, (png_const_structp png_ptr, * The simplified API hides the details of both libpng and the PNG file format * itself. It allows PNG files to be read into a very limited number of * in-memory bitmap formats or to be written from the same formats. If these - * formats do not accomodate your needs then you can, and should, use the more + * formats do not accommodate your needs then you can, and should, use the more * sophisticated APIs above - these support a wide variety of in-memory formats * and a wide variety of sophisticated transformations to those formats as well * as a wide variety of APIs to manipulate ancillary information. @@ -2753,7 +2733,7 @@ typedef struct * * When the simplified API needs to convert between sRGB and linear colorspaces, * the actual sRGB transfer curve defined in the sRGB specification (see the - * article at http://en.wikipedia.org/wiki/SRGB) is used, not the gamma=1/2.2 + * article at ) is used, not the gamma=1/2.2 * approximation used elsewhere in libpng. * * When an alpha channel is present it is expected to denote pixel coverage @@ -2808,6 +2788,8 @@ typedef struct # define PNG_FORMAT_FLAG_AFIRST 0x20U /* alpha channel comes first */ #endif +#define PNG_FORMAT_FLAG_ASSOCIATED_ALPHA 0x40U /* alpha channel is associated */ + /* Commonly used formats have predefined macros. * * First the single byte (sRGB) formats: @@ -2954,7 +2936,7 @@ typedef struct * 'flags' field of png_image. */ #define PNG_IMAGE_FLAG_COLORSPACE_NOT_sRGB 0x01 - /* This indicates the the RGB values of the in-memory bitmap do not + /* This indicates that the RGB values of the in-memory bitmap do not * correspond to the red, green and blue end-points defined by sRGB. */ @@ -3007,7 +2989,7 @@ PNG_EXPORT(235, int, png_image_begin_read_from_stdio, (png_imagep image, #endif /* STDIO */ PNG_EXPORT(236, int, png_image_begin_read_from_memory, (png_imagep image, - png_const_voidp memory, png_size_t size)); + png_const_voidp memory, size_t size)); /* The PNG header is read from the given memory buffer. */ PNG_EXPORT(237, int, png_image_finish_read, (png_imagep image, @@ -3120,7 +3102,7 @@ PNG_EXPORT(245, int, png_image_write_to_memory, (png_imagep image, void *memory, * than or equal to the original value. * * If the function returns false and *memory_bytes was not changed an error - * occured during write. If *memory_bytes was changed, or is not 0 if + * occurred during write. If *memory_bytes was changed, or is not 0 if * 'memory' was NULL, the write would have succeeded but for the memory * buffer being too small. *memory_bytes contains the required number of * bytes and will be bigger that the original value. @@ -3204,7 +3186,7 @@ PNG_EXPORT(245, int, png_image_write_to_memory, (png_imagep image, void *memory, * option and 'onoff' is 0 (off) or non-0 (on). The value returned is given * by the PNG_OPTION_ defines below. * - * HARDWARE: normally hardware capabilites, such as the Intel SSE instructions, + * HARDWARE: normally hardware capabilities, such as the Intel SSE instructions, * are detected at run time, however sometimes it may be impossible * to do this in user mode, in which case it is necessary to discover * the capabilities in an OS specific way. Such capabilities are @@ -3253,7 +3235,7 @@ PNG_EXPORT(244, int, png_set_option, (png_structrp png_ptr, int option, * one to use is one more than this.) */ #ifdef PNG_EXPORT_LAST_ORDINAL - PNG_EXPORT_LAST_ORDINAL(245); + PNG_EXPORT_LAST_ORDINAL(249); #endif #ifdef __cplusplus diff --git a/Externals/libpng/pngconf.h b/Externals/libpng/pngconf.h index 78d8b92b0e..927a769dbe 100644 --- a/Externals/libpng/pngconf.h +++ b/Externals/libpng/pngconf.h @@ -1,11 +1,12 @@ -/* pngconf.h - machine configurable file for libpng +/* pngconf.h - machine-configurable file for libpng * - * libpng version 1.6.29, March 16, 2017 + * libpng version 1.6.37 * - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2016,2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -57,14 +58,13 @@ #endif /* PNG_BUILDING_SYMBOL_TABLE */ -/* Prior to 1.6.0 it was possible to turn off 'const' in declarations using - * PNG_NO_CONST; this is no longer supported except for data declarations which - * apparently still cause problems in 2011 on some compilers. +/* Prior to 1.6.0, it was possible to turn off 'const' in declarations, + * using PNG_NO_CONST. This is no longer supported. */ #define PNG_CONST const /* backward compatibility only */ -/* This controls optimization of the reading of 16-bit and 32-bit values - * from PNG files. It can be set on a per-app-file basis - it +/* This controls optimization of the reading of 16-bit and 32-bit + * values from PNG files. It can be set on a per-app-file basis: it * just changes whether a macro is used when the function is called. * The library builder sets the default; if read functions are not * built into the library the macro implementation is forced on. @@ -127,7 +127,7 @@ * * These cases only differ if the operating system does not use the C * calling convention, at present this just means the above cases - * (x86 DOS/Windows sytems) and, even then, this does not apply to + * (x86 DOS/Windows systems) and, even then, this does not apply to * Cygwin running on those systems. * * Note that the value must be defined in pnglibconf.h so that what @@ -515,8 +515,10 @@ # error "libpng requires an unsigned 32-bit (or more) type" #endif -/* Prior to 1.6.0 it was possible to disable the use of size_t, 1.6.0, however, - * requires an ISOC90 compiler and relies on consistent behavior of sizeof. +/* Prior to 1.6.0, it was possible to disable the use of size_t and ptrdiff_t. + * From 1.6.0 onwards, an ISO C90 compiler, as well as a standard-compliant + * behavior of sizeof and ptrdiff_t are required. + * The legacy typedefs are provided here for backwards compatibility. */ typedef size_t png_size_t; typedef ptrdiff_t png_ptrdiff_t; @@ -537,13 +539,12 @@ typedef ptrdiff_t png_ptrdiff_t; # endif #endif -/* png_alloc_size_t is guaranteed to be no smaller than png_size_t, and no - * smaller than png_uint_32. Casts from png_size_t or png_uint_32 to - * png_alloc_size_t are not necessary; in fact, it is recommended not to use - * them at all so that the compiler can complain when something turns out to be - * problematic. +/* png_alloc_size_t is guaranteed to be no smaller than size_t, and no smaller + * than png_uint_32. Casts from size_t or png_uint_32 to png_alloc_size_t are + * not necessary; in fact, it is recommended not to use them at all, so that + * the compiler can complain when something turns out to be problematic. * - * Casts in the other direction (from png_alloc_size_t to png_size_t or + * Casts in the other direction (from png_alloc_size_t to size_t or * png_uint_32) should be explicitly applied; however, we do not expect to * encounter practical situations that require such conversions. * @@ -553,7 +554,7 @@ typedef ptrdiff_t png_ptrdiff_t; #ifdef PNG_SMALL_SIZE_T typedef png_uint_32 png_alloc_size_t; #else - typedef png_size_t png_alloc_size_t; + typedef size_t png_alloc_size_t; #endif /* Prior to 1.6.0 libpng offered limited support for Microsoft C compiler @@ -589,8 +590,8 @@ typedef char * png_charp; typedef const char * png_const_charp; typedef png_fixed_point * png_fixed_point_p; typedef const png_fixed_point * png_const_fixed_point_p; -typedef png_size_t * png_size_tp; -typedef const png_size_t * png_const_size_tp; +typedef size_t * png_size_tp; +typedef const size_t * png_const_size_tp; #ifdef PNG_STDIO_SUPPORTED typedef FILE * png_FILE_p; diff --git a/Externals/libpng/pngdebug.h b/Externals/libpng/pngdebug.h index 15a7ed0c95..00d5a4569e 100644 --- a/Externals/libpng/pngdebug.h +++ b/Externals/libpng/pngdebug.h @@ -1,10 +1,10 @@ /* pngdebug.h - Debugging macros for libpng, also used in pngtest.c * - * Last changed in libpng 1.6.8 [December 19, 2013] + * Copyright (c) 2018 Cosmin Truta * Copyright (c) 1998-2002,2004,2006-2013 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer diff --git a/Externals/libpng/pngerror.c b/Externals/libpng/pngerror.c index 00d76f7c05..ec3a709b9d 100644 --- a/Externals/libpng/pngerror.c +++ b/Externals/libpng/pngerror.c @@ -1,10 +1,10 @@ /* pngerror.c - stub functions for i/o and memory allocation * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -163,7 +163,7 @@ png_format_number(png_const_charp start, png_charp end, int format, case PNG_NUMBER_FORMAT_02u: /* Expects at least 2 digits. */ mincount = 2; - /* FALL THROUGH */ + /* FALLTHROUGH */ case PNG_NUMBER_FORMAT_u: *--end = digits[number % 10]; @@ -173,7 +173,7 @@ png_format_number(png_const_charp start, png_charp end, int format, case PNG_NUMBER_FORMAT_02x: /* This format expects at least two digits */ mincount = 2; - /* FALL THROUGH */ + /* FALLTHROUGH */ case PNG_NUMBER_FORMAT_x: *--end = digits[number & 0xf]; @@ -425,7 +425,7 @@ png_app_error(png_const_structrp png_ptr, png_const_charp error_message) * if the character is invalid. */ #define isnonalpha(c) ((c) < 65 || (c) > 122 || ((c) > 90 && (c) < 97)) -static PNG_CONST char png_digit[16] = { +static const char png_digit[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; @@ -885,7 +885,7 @@ PNG_FUNCTION(void /* PRIVATE */, (PNGCBAPI png_safe_error),(png_structp png_nonconst_ptr, png_const_charp error_message), PNG_NORETURN) { - const png_const_structrp png_ptr = png_nonconst_ptr; + png_const_structrp png_ptr = png_nonconst_ptr; png_imagep image = png_voidcast(png_imagep, png_ptr->error_ptr); /* An error is always logged here, overwriting anything (typically a warning) @@ -920,7 +920,7 @@ png_safe_error),(png_structp png_nonconst_ptr, png_const_charp error_message), void /* PRIVATE */ PNGCBAPI png_safe_warning(png_structp png_nonconst_ptr, png_const_charp warning_message) { - const png_const_structrp png_ptr = png_nonconst_ptr; + png_const_structrp png_ptr = png_nonconst_ptr; png_imagep image = png_voidcast(png_imagep, png_ptr->error_ptr); /* A warning is only logged if there is no prior warning or error. */ diff --git a/Externals/libpng/pngget.c b/Externals/libpng/pngget.c index 141c393330..5abf1efd9f 100644 --- a/Externals/libpng/pngget.c +++ b/Externals/libpng/pngget.c @@ -1,10 +1,10 @@ /* pngget.c - retrieval of values from info struct * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -26,7 +26,7 @@ png_get_valid(png_const_structrp png_ptr, png_const_inforp info_ptr, return(0); } -png_size_t PNGAPI +size_t PNGAPI png_get_rowbytes(png_const_structrp png_ptr, png_const_inforp info_ptr) { if (png_ptr != NULL && info_ptr != NULL) @@ -367,7 +367,7 @@ png_get_y_pixels_per_inch(png_const_structrp png_ptr, png_const_inforp info_ptr) static png_fixed_point png_fixed_inches_from_microns(png_const_structrp png_ptr, png_int_32 microns) { - /* Convert from metres * 1,000,000 to inches * 100,000, meters to + /* Convert from meters * 1,000,000 to inches * 100,000, meters to * inches is simply *(100/2.54), so we want *(10/2.54) == 500/127. * Notice that this can overflow - a warning is output and 0 is * returned. @@ -741,8 +741,7 @@ png_get_iCCP(png_const_structrp png_ptr, png_inforp info_ptr, if (png_ptr != NULL && info_ptr != NULL && (info_ptr->valid & PNG_INFO_iCCP) != 0 && - name != NULL && compression_type != NULL && profile != NULL && - proflen != NULL) + name != NULL && profile != NULL && proflen != NULL) { *name = info_ptr->iccp_name; *profile = info_ptr->iccp_profile; @@ -750,11 +749,13 @@ png_get_iCCP(png_const_structrp png_ptr, png_inforp info_ptr, /* This is somewhat irrelevant since the profile data returned has * actually been uncompressed. */ - *compression_type = PNG_COMPRESSION_TYPE_BASE; + if (compression_type != NULL) + *compression_type = PNG_COMPRESSION_TYPE_BASE; return (PNG_INFO_iCCP); } return (0); + } #endif @@ -773,6 +774,35 @@ png_get_sPLT(png_const_structrp png_ptr, png_inforp info_ptr, } #endif +#ifdef PNG_eXIf_SUPPORTED +png_uint_32 PNGAPI +png_get_eXIf(png_const_structrp png_ptr, png_inforp info_ptr, + png_bytep *exif) +{ + png_warning(png_ptr, "png_get_eXIf does not work; use png_get_eXIf_1"); + PNG_UNUSED(info_ptr) + PNG_UNUSED(exif) + return 0; +} + +png_uint_32 PNGAPI +png_get_eXIf_1(png_const_structrp png_ptr, png_const_inforp info_ptr, + png_uint_32 *num_exif, png_bytep *exif) +{ + png_debug1(1, "in %s retrieval function", "eXIf"); + + if (png_ptr != NULL && info_ptr != NULL && + (info_ptr->valid & PNG_INFO_eXIf) != 0 && exif != NULL) + { + *num_exif = info_ptr->num_exif; + *exif = info_ptr->exif; + return (PNG_INFO_eXIf); + } + + return (0); +} +#endif + #ifdef PNG_hIST_SUPPORTED png_uint_32 PNGAPI png_get_hIST(png_const_structrp png_ptr, png_inforp info_ptr, @@ -1135,7 +1165,7 @@ png_get_user_chunk_ptr(png_const_structrp png_ptr) } #endif -png_size_t PNGAPI +size_t PNGAPI png_get_compression_buffer_size(png_const_structrp png_ptr) { if (png_ptr == NULL) diff --git a/Externals/libpng/pnginfo.h b/Externals/libpng/pnginfo.h index 361ed8be70..1f98dedc42 100644 --- a/Externals/libpng/pnginfo.h +++ b/Externals/libpng/pnginfo.h @@ -1,10 +1,10 @@ /* pnginfo.h - header file for PNG reference library * - * Last changed in libpng 1.6.1 [March 28, 2013] - * Copyright (c) 1998-2002,2004,2006-2013 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2013,2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -55,10 +55,10 @@ struct png_info_def { /* The following are necessary for every PNG file */ - png_uint_32 width; /* width of image in pixels (from IHDR) */ - png_uint_32 height; /* height of image in pixels (from IHDR) */ - png_uint_32 valid; /* valid chunk data (see PNG_INFO_ below) */ - png_size_t rowbytes; /* bytes needed to hold an untransformed row */ + png_uint_32 width; /* width of image in pixels (from IHDR) */ + png_uint_32 height; /* height of image in pixels (from IHDR) */ + png_uint_32 valid; /* valid chunk data (see PNG_INFO_ below) */ + size_t rowbytes; /* bytes needed to hold an untransformed row */ png_colorp palette; /* array of color values (valid & PNG_INFO_PLTE) */ png_uint_16 num_palette; /* number of color entries in "palette" (PLTE) */ png_uint_16 num_trans; /* number of transparent palette color (tRNS) */ @@ -185,6 +185,14 @@ defined(PNG_READ_BACKGROUND_SUPPORTED) png_byte phys_unit_type; /* resolution type (see PNG_RESOLUTION_ below) */ #endif +#ifdef PNG_eXIf_SUPPORTED + int num_exif; /* Added at libpng-1.6.31 */ + png_bytep exif; +# ifdef PNG_READ_eXIf_SUPPORTED + png_bytep eXIf_buf; /* Added at libpng-1.6.32 */ +# endif +#endif + #ifdef PNG_hIST_SUPPORTED /* The hIST chunk contains the relative frequency or importance of the * various palette entries, so that a viewer can intelligently select a @@ -239,7 +247,7 @@ defined(PNG_READ_BACKGROUND_SUPPORTED) /* The sCAL chunk describes the actual physical dimensions of the * subject matter of the graphic. The chunk contains a unit specification * a byte value, and two ASCII strings representing floating-point - * values. The values are width and height corresponsing to one pixel + * values. The values are width and height corresponding to one pixel * in the image. Data values are valid if (valid & PNG_INFO_sCAL) is * non-zero. */ diff --git a/Externals/libpng/pnglibconf.h b/Externals/libpng/pnglibconf.h index 5393d19159..e1e27e957e 100644 --- a/Externals/libpng/pnglibconf.h +++ b/Externals/libpng/pnglibconf.h @@ -1,10 +1,9 @@ -/* libpng 1.6.29 STANDARD API DEFINITION */ - /* pnglibconf.h - library build configuration */ -/* Libpng version 1.6.29 - March 16, 2017 */ +/* libpng version 1.6.37 */ -/* Copyright (c) 1998-2015 Glenn Randers-Pehrson */ +/* Copyright (c) 2018-2019 Cosmin Truta */ +/* Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson */ /* This code is released under the libpng license. */ /* For conditions of distribution and use, see the disclaimer */ @@ -44,6 +43,8 @@ #define PNG_IO_STATE_SUPPORTED #define PNG_MNG_FEATURES_SUPPORTED #define PNG_POINTER_INDEXING_SUPPORTED +/*#undef PNG_POWERPC_VSX_API_SUPPORTED*/ +/*#undef PNG_POWERPC_VSX_CHECK_SUPPORTED*/ #define PNG_PROGRESSIVE_READ_SUPPORTED #define PNG_READ_16BIT_SUPPORTED #define PNG_READ_ALPHA_MODE_SUPPORTED @@ -82,6 +83,7 @@ #define PNG_READ_USER_TRANSFORM_SUPPORTED #define PNG_READ_bKGD_SUPPORTED #define PNG_READ_cHRM_SUPPORTED +#define PNG_READ_eXIf_SUPPORTED #define PNG_READ_gAMA_SUPPORTED #define PNG_READ_hIST_SUPPORTED #define PNG_READ_iCCP_SUPPORTED @@ -151,6 +153,7 @@ #define PNG_WRITE_WEIGHTED_FILTER_SUPPORTED #define PNG_WRITE_bKGD_SUPPORTED #define PNG_WRITE_cHRM_SUPPORTED +#define PNG_WRITE_eXIf_SUPPORTED #define PNG_WRITE_gAMA_SUPPORTED #define PNG_WRITE_hIST_SUPPORTED #define PNG_WRITE_iCCP_SUPPORTED @@ -168,6 +171,7 @@ #define PNG_WRITE_zTXt_SUPPORTED #define PNG_bKGD_SUPPORTED #define PNG_cHRM_SUPPORTED +#define PNG_eXIf_SUPPORTED #define PNG_gAMA_SUPPORTED #define PNG_hIST_SUPPORTED #define PNG_iCCP_SUPPORTED diff --git a/Externals/libpng/pngmem.c b/Externals/libpng/pngmem.c index ff3ef7e88c..09ed9c1c99 100644 --- a/Externals/libpng/pngmem.c +++ b/Externals/libpng/pngmem.c @@ -1,10 +1,10 @@ /* pngmem.c - stub functions for memory allocation * - * Last changed in libpng 1.6.26 [October 20, 2016] + * Copyright (c) 2018 Cosmin Truta * Copyright (c) 1998-2002,2004,2006-2014,2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer diff --git a/Externals/libpng/pngpread.c b/Externals/libpng/pngpread.c index 650ba1e232..e283627b77 100644 --- a/Externals/libpng/pngpread.c +++ b/Externals/libpng/pngpread.c @@ -1,10 +1,10 @@ /* pngpread.c - read a png file in push mode * - * Last changed in libpng 1.6.24 [August 4, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -34,7 +34,7 @@ if (png_ptr->buffer_size < N) \ void PNGAPI png_process_data(png_structrp png_ptr, png_inforp info_ptr, - png_bytep buffer, png_size_t buffer_size) + png_bytep buffer, size_t buffer_size) { if (png_ptr == NULL || info_ptr == NULL) return; @@ -47,7 +47,7 @@ png_process_data(png_structrp png_ptr, png_inforp info_ptr, } } -png_size_t PNGAPI +size_t PNGAPI png_process_data_pause(png_structrp png_ptr, int save) { if (png_ptr != NULL) @@ -60,7 +60,7 @@ png_process_data_pause(png_structrp png_ptr, int save) else { /* This includes any pending saved bytes: */ - png_size_t remaining = png_ptr->buffer_size; + size_t remaining = png_ptr->buffer_size; png_ptr->buffer_size = 0; /* So subtract the saved buffer size, unless all the data @@ -133,8 +133,8 @@ png_process_some_data(png_structrp png_ptr, png_inforp info_ptr) void /* PRIVATE */ png_push_read_sig(png_structrp png_ptr, png_inforp info_ptr) { - png_size_t num_checked = png_ptr->sig_bytes, /* SAFE, does not exceed 8 */ - num_to_check = 8 - num_checked; + size_t num_checked = png_ptr->sig_bytes; /* SAFE, does not exceed 8 */ + size_t num_to_check = 8 - num_checked; if (png_ptr->buffer_size < num_to_check) { @@ -189,6 +189,7 @@ png_push_read_chunk(png_structrp png_ptr, png_inforp info_ptr) png_crc_read(png_ptr, chunk_tag, 4); png_ptr->chunk_name = PNG_CHUNK_FROM_STRING(chunk_tag); png_check_chunk_name(png_ptr, png_ptr->chunk_name); + png_check_chunk_length(png_ptr, png_ptr->push_length); png_ptr->mode |= PNG_HAVE_CHUNK_HEADER; } @@ -417,7 +418,7 @@ png_push_read_chunk(png_structrp png_ptr, png_inforp info_ptr) } void PNGCBAPI -png_push_fill_buffer(png_structp png_ptr, png_bytep buffer, png_size_t length) +png_push_fill_buffer(png_structp png_ptr, png_bytep buffer, size_t length) { png_bytep ptr; @@ -427,7 +428,7 @@ png_push_fill_buffer(png_structp png_ptr, png_bytep buffer, png_size_t length) ptr = buffer; if (png_ptr->save_buffer_size != 0) { - png_size_t save_size; + size_t save_size; if (length < png_ptr->save_buffer_size) save_size = length; @@ -444,7 +445,7 @@ png_push_fill_buffer(png_structp png_ptr, png_bytep buffer, png_size_t length) } if (length != 0 && png_ptr->current_buffer_size != 0) { - png_size_t save_size; + size_t save_size; if (length < png_ptr->current_buffer_size) save_size = length; @@ -466,7 +467,7 @@ png_push_save_buffer(png_structrp png_ptr) { if (png_ptr->save_buffer_ptr != png_ptr->save_buffer) { - png_size_t i, istop; + size_t i, istop; png_bytep sp; png_bytep dp; @@ -481,7 +482,7 @@ png_push_save_buffer(png_structrp png_ptr) if (png_ptr->save_buffer_size + png_ptr->current_buffer_size > png_ptr->save_buffer_max) { - png_size_t new_max; + size_t new_max; png_bytep old_buffer; if (png_ptr->save_buffer_size > PNG_SIZE_MAX - @@ -493,7 +494,7 @@ png_push_save_buffer(png_structrp png_ptr) new_max = png_ptr->save_buffer_size + png_ptr->current_buffer_size + 256; old_buffer = png_ptr->save_buffer; png_ptr->save_buffer = (png_bytep)png_malloc_warn(png_ptr, - (png_size_t)new_max); + (size_t)new_max); if (png_ptr->save_buffer == NULL) { @@ -521,7 +522,7 @@ png_push_save_buffer(png_structrp png_ptr) void /* PRIVATE */ png_push_restore_buffer(png_structrp png_ptr, png_bytep buffer, - png_size_t buffer_length) + size_t buffer_length) { png_ptr->current_buffer = buffer; png_ptr->current_buffer_size = buffer_length; @@ -561,7 +562,7 @@ png_push_read_IDAT(png_structrp png_ptr) if (png_ptr->idat_size != 0 && png_ptr->save_buffer_size != 0) { - png_size_t save_size = png_ptr->save_buffer_size; + size_t save_size = png_ptr->save_buffer_size; png_uint_32 idat_size = png_ptr->idat_size; /* We want the smaller of 'idat_size' and 'current_buffer_size', but they @@ -571,7 +572,7 @@ png_push_read_IDAT(png_structrp png_ptr) * will break on either 16-bit or 64-bit platforms. */ if (idat_size < save_size) - save_size = (png_size_t)idat_size; + save_size = (size_t)idat_size; else idat_size = (png_uint_32)save_size; @@ -588,7 +589,7 @@ png_push_read_IDAT(png_structrp png_ptr) if (png_ptr->idat_size != 0 && png_ptr->current_buffer_size != 0) { - png_size_t save_size = png_ptr->current_buffer_size; + size_t save_size = png_ptr->current_buffer_size; png_uint_32 idat_size = png_ptr->idat_size; /* We want the smaller of 'idat_size' and 'current_buffer_size', but they @@ -597,7 +598,7 @@ png_push_read_IDAT(png_structrp png_ptr) * larger - this cannot overflow. */ if (idat_size < save_size) - save_size = (png_size_t)idat_size; + save_size = (size_t)idat_size; else idat_size = (png_uint_32)save_size; @@ -624,7 +625,7 @@ png_push_read_IDAT(png_structrp png_ptr) void /* PRIVATE */ png_process_IDAT_data(png_structrp png_ptr, png_bytep buffer, - png_size_t buffer_length) + size_t buffer_length) { /* The caller checks for a non-zero buffer length. */ if (!(buffer_length > 0) || buffer == NULL) @@ -971,20 +972,20 @@ png_read_push_finish_row(png_structrp png_ptr) /* Arrays to facilitate easy interlacing - use pass (0 - 6) as index */ /* Start of interlace block */ - static PNG_CONST png_byte png_pass_start[] = {0, 4, 0, 2, 0, 1, 0}; + static const png_byte png_pass_start[] = {0, 4, 0, 2, 0, 1, 0}; /* Offset to next interlace block */ - static PNG_CONST png_byte png_pass_inc[] = {8, 8, 4, 4, 2, 2, 1}; + static const png_byte png_pass_inc[] = {8, 8, 4, 4, 2, 2, 1}; /* Start of interlace block in the y direction */ - static PNG_CONST png_byte png_pass_ystart[] = {0, 0, 4, 0, 2, 0, 1}; + static const png_byte png_pass_ystart[] = {0, 0, 4, 0, 2, 0, 1}; /* Offset to next interlace block in the y direction */ - static PNG_CONST png_byte png_pass_yinc[] = {8, 8, 8, 4, 4, 2, 2}; + static const png_byte png_pass_yinc[] = {8, 8, 8, 4, 4, 2, 2}; /* Height of interlace block. This is not currently used - if you need * it, uncomment it here and in png.h - static PNG_CONST png_byte png_pass_height[] = {8, 8, 4, 4, 2, 2, 1}; + static const png_byte png_pass_height[] = {8, 8, 4, 4, 2, 2, 1}; */ #endif diff --git a/Externals/libpng/pngpriv.h b/Externals/libpng/pngpriv.h index 01fc380094..583c26f9bd 100644 --- a/Externals/libpng/pngpriv.h +++ b/Externals/libpng/pngpriv.h @@ -1,10 +1,10 @@ /* pngpriv.h - private declarations for use inside libpng * - * Last changed in libpng 1.6.29 [March 16, 2017] - * Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -35,7 +35,9 @@ * Windows/Visual Studio) there is no effect; the OS specific tests below are * still required (as of 2011-05-02.) */ -#define _POSIX_SOURCE 1 /* Just the POSIX 1003.1 and C89 APIs */ +#ifndef _POSIX_SOURCE +# define _POSIX_SOURCE 1 /* Just the POSIX 1003.1 and C89 APIs */ +#endif #ifndef PNG_VERSION_INFO_ONLY /* Standard library headers not required by png.h: */ @@ -105,7 +107,6 @@ * this in $(CC), e.g. "CC=gcc -mfpu=neon", but people who build libpng rarely * do this. */ -#define PNG_ARM_NEON_OPT 0 #ifndef PNG_ARM_NEON_OPT /* ARM NEON optimizations are being controlled by the compiler settings, * typically the target FPU. If the FPU has been set to NEON (-mfpu=neon @@ -173,7 +174,10 @@ # else /* !defined __ARM_NEON__ */ /* The 'intrinsics' code simply won't compile without this -mfpu=neon: */ -# define PNG_ARM_NEON_IMPLEMENTATION 2 +# if !defined(__aarch64__) + /* The assembler code currently does not work on ARM64 */ +# define PNG_ARM_NEON_IMPLEMENTATION 2 +# endif /* __aarch64__ */ # endif /* __ARM_NEON__ */ # endif /* !PNG_ARM_NEON_IMPLEMENTATION */ @@ -209,7 +213,11 @@ defined(__SSE2__) || defined(_M_X64) || defined(_M_AMD64) || \ (defined(_M_IX86_FP) && _M_IX86_FP >= 2) # define PNG_INTEL_SSE_OPT 1 +# else +# define PNG_INTEL_SSE_OPT 0 # endif +# else +# define PNG_INTEL_SSE_OPT 0 # endif #endif @@ -233,6 +241,8 @@ # if PNG_INTEL_SSE_IMPLEMENTATION > 0 # define PNG_FILTER_OPTIMIZATIONS png_init_filter_functions_sse2 # endif +#else +# define PNG_INTEL_SSE_IMPLEMENTATION 0 #endif #if PNG_MIPS_MSA_OPT > 0 @@ -453,25 +463,6 @@ # define png_fixed_error(s1,s2) png_err(s1) #endif -/* C allows up-casts from (void*) to any pointer and (const void*) to any - * pointer to a const object. C++ regards this as a type error and requires an - * explicit, static, cast and provides the static_cast<> rune to ensure that - * const is not cast away. - */ -#ifdef __cplusplus -# define png_voidcast(type, value) static_cast(value) -# define png_constcast(type, value) const_cast(value) -# define png_aligncast(type, value) \ - static_cast(static_cast(value)) -# define png_aligncastconst(type, value) \ - static_cast(static_cast(value)) -#else -# define png_voidcast(type, value) (value) -# define png_constcast(type, value) ((type)(value)) -# define png_aligncast(type, value) ((void*)(value)) -# define png_aligncastconst(type, value) ((const void*)(value)) -#endif /* __cplusplus */ - /* Some fixed point APIs are still required even if not exported because * they get used by the corresponding floating point APIs. This magic * deals with this: @@ -486,6 +477,35 @@ /* Other defines specific to compilers can go here. Try to keep * them inside an appropriate ifdef/endif pair for portability. */ + +/* C allows up-casts from (void*) to any pointer and (const void*) to any + * pointer to a const object. C++ regards this as a type error and requires an + * explicit, static, cast and provides the static_cast<> rune to ensure that + * const is not cast away. + */ +#ifdef __cplusplus +# define png_voidcast(type, value) static_cast(value) +# define png_constcast(type, value) const_cast(value) +# define png_aligncast(type, value) \ + static_cast(static_cast(value)) +# define png_aligncastconst(type, value) \ + static_cast(static_cast(value)) +#else +# define png_voidcast(type, value) (value) +# ifdef _WIN64 +# ifdef __GNUC__ + typedef unsigned long long png_ptruint; +# else + typedef unsigned __int64 png_ptruint; +# endif +# else + typedef unsigned long png_ptruint; +# endif +# define png_constcast(type, value) ((type)(png_ptruint)(const void*)(value)) +# define png_aligncast(type, value) ((void*)(value)) +# define png_aligncastconst(type, value) ((const void*)(value)) +#endif /* __cplusplus */ + #if defined(PNG_FLOATING_POINT_SUPPORTED) ||\ defined(PNG_FLOATING_ARITHMETIC_SUPPORTED) /* png.c requires the following ANSI-C constants if the conversion of @@ -717,8 +737,8 @@ /* Added to libpng-1.2.6 JB */ #define PNG_ROWBYTES(pixel_bits, width) \ ((pixel_bits) >= 8 ? \ - ((png_size_t)(width) * (((png_size_t)(pixel_bits)) >> 3)) : \ - (( ((png_size_t)(width) * ((png_size_t)(pixel_bits))) + 7) >> 3) ) + ((size_t)(width) * (((size_t)(pixel_bits)) >> 3)) : \ + (( ((size_t)(width) * ((size_t)(pixel_bits))) + 7) >> 3) ) /* This returns the number of trailing bits in the last byte of a row, 0 if the * last byte is completely full of pixels. It is, in principle, (pixel_bits x @@ -831,6 +851,7 @@ #define png_PLTE PNG_U32( 80, 76, 84, 69) #define png_bKGD PNG_U32( 98, 75, 71, 68) #define png_cHRM PNG_U32( 99, 72, 82, 77) +#define png_eXIf PNG_U32(101, 88, 73, 102) /* registered July 2017 */ #define png_fRAc PNG_U32(102, 82, 65, 99) /* registered, not defined */ #define png_gAMA PNG_U32(103, 65, 77, 65) #define png_gIFg PNG_U32(103, 73, 70, 103) @@ -905,7 +926,7 @@ * PNG files the -I directives must match. * * The most likely explanation is that you passed a -I in CFLAGS. This will - * not work; all the preprocessor directories and in particular all the -I + * not work; all the preprocessor directives and in particular all the -I * directives must be in CPPFLAGS. */ #endif @@ -1034,15 +1055,15 @@ PNG_INTERNAL_FUNCTION(void,png_zfree,(voidpf png_ptr, voidpf ptr),PNG_EMPTY); */ PNG_INTERNAL_FUNCTION(void PNGCBAPI,png_default_read_data,(png_structp png_ptr, - png_bytep data, png_size_t length),PNG_EMPTY); + png_bytep data, size_t length),PNG_EMPTY); #ifdef PNG_PROGRESSIVE_READ_SUPPORTED PNG_INTERNAL_FUNCTION(void PNGCBAPI,png_push_fill_buffer,(png_structp png_ptr, - png_bytep buffer, png_size_t length),PNG_EMPTY); + png_bytep buffer, size_t length),PNG_EMPTY); #endif PNG_INTERNAL_FUNCTION(void PNGCBAPI,png_default_write_data,(png_structp png_ptr, - png_bytep data, png_size_t length),PNG_EMPTY); + png_bytep data, size_t length),PNG_EMPTY); #ifdef PNG_WRITE_FLUSH_SUPPORTED # ifdef PNG_STDIO_SUPPORTED @@ -1056,7 +1077,7 @@ PNG_INTERNAL_FUNCTION(void,png_reset_crc,(png_structrp png_ptr),PNG_EMPTY); /* Write the "data" buffer to whatever output you are using */ PNG_INTERNAL_FUNCTION(void,png_write_data,(png_structrp png_ptr, - png_const_bytep data, png_size_t length),PNG_EMPTY); + png_const_bytep data, size_t length),PNG_EMPTY); /* Read and check the PNG file signature */ PNG_INTERNAL_FUNCTION(void,png_read_sig,(png_structrp png_ptr, @@ -1068,7 +1089,7 @@ PNG_INTERNAL_FUNCTION(png_uint_32,png_read_chunk_header,(png_structrp png_ptr), /* Read data from whatever input you are using into the "data" buffer */ PNG_INTERNAL_FUNCTION(void,png_read_data,(png_structrp png_ptr, png_bytep data, - png_size_t length),PNG_EMPTY); + size_t length),PNG_EMPTY); /* Read bytes into buf, and update png_ptr->crc */ PNG_INTERNAL_FUNCTION(void,png_crc_read,(png_structrp png_ptr, png_bytep buf, @@ -1086,7 +1107,7 @@ PNG_INTERNAL_FUNCTION(int,png_crc_error,(png_structrp png_ptr),PNG_EMPTY); * since this is the maximum buffer size we can specify. */ PNG_INTERNAL_FUNCTION(void,png_calculate_crc,(png_structrp png_ptr, - png_const_bytep ptr, png_size_t length),PNG_EMPTY); + png_const_bytep ptr, size_t length),PNG_EMPTY); #ifdef PNG_WRITE_FLUSH_SUPPORTED PNG_INTERNAL_FUNCTION(void,png_flush,(png_structrp png_ptr),PNG_EMPTY); @@ -1131,6 +1152,11 @@ PNG_INTERNAL_FUNCTION(void,png_write_sRGB,(png_structrp png_ptr, int intent),PNG_EMPTY); #endif +#ifdef PNG_WRITE_eXIf_SUPPORTED +PNG_INTERNAL_FUNCTION(void,png_write_eXIf,(png_structrp png_ptr, + png_bytep exif, int num_exif),PNG_EMPTY); +#endif + #ifdef PNG_WRITE_iCCP_SUPPORTED PNG_INTERNAL_FUNCTION(void,png_write_iCCP,(png_structrp png_ptr, png_const_charp name, png_const_bytep profile), PNG_EMPTY); @@ -1164,7 +1190,7 @@ PNG_INTERNAL_FUNCTION(void,png_write_hIST,(png_structrp png_ptr, /* Chunks that have keywords */ #ifdef PNG_WRITE_tEXt_SUPPORTED PNG_INTERNAL_FUNCTION(void,png_write_tEXt,(png_structrp png_ptr, - png_const_charp key, png_const_charp text, png_size_t text_len),PNG_EMPTY); + png_const_charp key, png_const_charp text, size_t text_len),PNG_EMPTY); #endif #ifdef PNG_WRITE_zTXt_SUPPORTED @@ -1430,6 +1456,11 @@ PNG_INTERNAL_FUNCTION(void,png_handle_cHRM,(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length),PNG_EMPTY); #endif +#ifdef PNG_READ_eXIf_SUPPORTED +PNG_INTERNAL_FUNCTION(void,png_handle_eXIf,(png_structrp png_ptr, + png_inforp info_ptr, png_uint_32 length),PNG_EMPTY); +#endif + #ifdef PNG_READ_gAMA_SUPPORTED PNG_INTERNAL_FUNCTION(void,png_handle_gAMA,(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length),PNG_EMPTY); @@ -1505,9 +1536,12 @@ PNG_INTERNAL_FUNCTION(void,png_handle_zTXt,(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length),PNG_EMPTY); #endif -PNG_INTERNAL_FUNCTION(void,png_check_chunk_name,(png_structrp png_ptr, +PNG_INTERNAL_FUNCTION(void,png_check_chunk_name,(png_const_structrp png_ptr, png_uint_32 chunk_name),PNG_EMPTY); +PNG_INTERNAL_FUNCTION(void,png_check_chunk_length,(png_const_structrp png_ptr, + png_uint_32 chunk_length),PNG_EMPTY); + PNG_INTERNAL_FUNCTION(void,png_handle_unknown,(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length, int keep),PNG_EMPTY); /* This is the function that gets called for unknown chunks. The 'keep' @@ -1549,10 +1583,10 @@ PNG_INTERNAL_FUNCTION(void,png_push_check_crc,(png_structrp png_ptr),PNG_EMPTY); PNG_INTERNAL_FUNCTION(void,png_push_save_buffer,(png_structrp png_ptr), PNG_EMPTY); PNG_INTERNAL_FUNCTION(void,png_push_restore_buffer,(png_structrp png_ptr, - png_bytep buffer, png_size_t buffer_length),PNG_EMPTY); + png_bytep buffer, size_t buffer_length),PNG_EMPTY); PNG_INTERNAL_FUNCTION(void,png_push_read_IDAT,(png_structrp png_ptr),PNG_EMPTY); PNG_INTERNAL_FUNCTION(void,png_process_IDAT_data,(png_structrp png_ptr, - png_bytep buffer, png_size_t buffer_length),PNG_EMPTY); + png_bytep buffer, size_t buffer_length),PNG_EMPTY); PNG_INTERNAL_FUNCTION(void,png_push_process_row,(png_structrp png_ptr), PNG_EMPTY); PNG_INTERNAL_FUNCTION(void,png_push_handle_unknown,(png_structrp png_ptr, @@ -1822,13 +1856,13 @@ PNG_INTERNAL_FUNCTION(void,png_chunk_report,(png_const_structrp png_ptr, #ifdef PNG_FLOATING_POINT_SUPPORTED PNG_INTERNAL_FUNCTION(void,png_ascii_from_fp,(png_const_structrp png_ptr, - png_charp ascii, png_size_t size, double fp, unsigned int precision), + png_charp ascii, size_t size, double fp, unsigned int precision), PNG_EMPTY); #endif /* FLOATING_POINT */ #ifdef PNG_FIXED_POINT_SUPPORTED PNG_INTERNAL_FUNCTION(void,png_ascii_from_fixed,(png_const_structrp png_ptr, - png_charp ascii, png_size_t size, png_fixed_point fp),PNG_EMPTY); + png_charp ascii, size_t size, png_fixed_point fp),PNG_EMPTY); #endif /* FIXED_POINT */ #endif /* sCAL */ @@ -1921,7 +1955,7 @@ PNG_INTERNAL_FUNCTION(void,png_ascii_from_fixed,(png_const_structrp png_ptr, * the problem character.) This has not been tested within libpng. */ PNG_INTERNAL_FUNCTION(int,png_check_fp_number,(png_const_charp string, - png_size_t size, int *statep, png_size_tp whereami),PNG_EMPTY); + size_t size, int *statep, png_size_tp whereami),PNG_EMPTY); /* This is the same but it checks a complete string and returns true * only if it just contains a floating point number. As of 1.5.4 this @@ -1930,7 +1964,7 @@ PNG_INTERNAL_FUNCTION(int,png_check_fp_number,(png_const_charp string, * for negative or zero values using the sticky flag. */ PNG_INTERNAL_FUNCTION(int,png_check_fp_string,(png_const_charp string, - png_size_t size),PNG_EMPTY); + size_t size),PNG_EMPTY); #endif /* pCAL || sCAL */ #if defined(PNG_GAMMA_SUPPORTED) ||\ @@ -2005,7 +2039,7 @@ typedef struct png_control png_voidp error_buf; /* Always a jmp_buf at present. */ png_const_bytep memory; /* Memory buffer. */ - png_size_t size; /* Size of the memory buffer. */ + size_t size; /* Size of the memory buffer. */ unsigned int for_write :1; /* Otherwise it is a read structure */ unsigned int owned_file :1; /* We own the file in io_ptr */ @@ -2083,6 +2117,29 @@ PNG_INTERNAL_FUNCTION(void, png_init_filter_functions_sse2, PNG_INTERNAL_FUNCTION(png_uint_32, png_check_keyword, (png_structrp png_ptr, png_const_charp key, png_bytep new_key), PNG_EMPTY); +#if PNG_ARM_NEON_IMPLEMENTATION == 1 +PNG_INTERNAL_FUNCTION(void, + png_riffle_palette_neon, + (png_structrp), + PNG_EMPTY); +PNG_INTERNAL_FUNCTION(int, + png_do_expand_palette_rgba8_neon, + (png_structrp, + png_row_infop, + png_const_bytep, + const png_bytepp, + const png_bytepp), + PNG_EMPTY); +PNG_INTERNAL_FUNCTION(int, + png_do_expand_palette_rgb8_neon, + (png_structrp, + png_row_infop, + png_const_bytep, + const png_bytepp, + const png_bytepp), + PNG_EMPTY); +#endif + /* Maintainer: Put new private prototypes here ^ */ #include "pngdebug.h" diff --git a/Externals/libpng/pngread.c b/Externals/libpng/pngread.c index 106a80588c..8fa7d9f162 100644 --- a/Externals/libpng/pngread.c +++ b/Externals/libpng/pngread.c @@ -1,10 +1,10 @@ /* pngread.c - read a PNG file * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -175,6 +175,11 @@ png_read_info(png_structrp png_ptr, png_inforp info_ptr) png_handle_cHRM(png_ptr, info_ptr, length); #endif +#ifdef PNG_READ_eXIf_SUPPORTED + else if (chunk_name == png_eXIf) + png_handle_eXIf(png_ptr, info_ptr, length); +#endif + #ifdef PNG_READ_gAMA_SUPPORTED else if (chunk_name == png_gAMA) png_handle_gAMA(png_ptr, info_ptr, length); @@ -534,6 +539,7 @@ png_read_row(png_structrp png_ptr, png_bytep row, png_bytep dsp_row) png_error(png_ptr, "Invalid attempt to read row data"); /* Fill the row with IDAT data: */ + png_ptr->row_buf[0]=255; /* to force error if no data was found */ png_read_IDAT_data(png_ptr, png_ptr->row_buf, row_info.rowbytes + 1); if (png_ptr->row_buf[0] > PNG_FILTER_VALUE_NONE) @@ -842,6 +848,11 @@ png_read_end(png_structrp png_ptr, png_inforp info_ptr) png_handle_cHRM(png_ptr, info_ptr, length); #endif +#ifdef PNG_READ_eXIf_SUPPORTED + else if (chunk_name == png_eXIf) + png_handle_eXIf(png_ptr, info_ptr, length); +#endif + #ifdef PNG_READ_gAMA_SUPPORTED else if (chunk_name == png_gAMA) png_handle_gAMA(png_ptr, info_ptr, length); @@ -983,6 +994,12 @@ png_read_destroy(png_structrp png_ptr) png_ptr->chunk_list = NULL; #endif +#if defined(PNG_READ_EXPAND_SUPPORTED) && \ + defined(PNG_ARM_NEON_IMPLEMENTATION) + png_free(png_ptr, png_ptr->riffled_palette); + png_ptr->riffled_palette = NULL; +#endif + /* NOTE: the 'setjmp' buffer may still be allocated and the memory and error * callbacks are still set at this point. They are required to complete the * destruction of the png_struct itself. @@ -1521,7 +1538,7 @@ png_image_begin_read_from_file(png_imagep image, const char *file_name) #endif /* STDIO */ static void PNGCBAPI -png_image_memory_read(png_structp png_ptr, png_bytep out, png_size_t need) +png_image_memory_read(png_structp png_ptr, png_bytep out, size_t need) { if (png_ptr != NULL) { @@ -1532,7 +1549,7 @@ png_image_memory_read(png_structp png_ptr, png_bytep out, png_size_t need) if (cp != NULL) { png_const_bytep memory = cp->memory; - png_size_t size = cp->size; + size_t size = cp->size; if (memory != NULL && size >= need) { @@ -1551,7 +1568,7 @@ png_image_memory_read(png_structp png_ptr, png_bytep out, png_size_t need) } int PNGAPI png_image_begin_read_from_memory(png_imagep image, - png_const_voidp memory, png_size_t size) + png_const_voidp memory, size_t size) { if (image != NULL && image->version == PNG_IMAGE_VERSION) { @@ -1610,7 +1627,7 @@ png_image_skip_unused_chunks(png_structrp png_ptr) * errors (which are unfortunately quite common.) */ { - static PNG_CONST png_byte chunks_to_process[] = { + static const png_byte chunks_to_process[] = { 98, 75, 71, 68, '\0', /* bKGD */ 99, 72, 82, 77, '\0', /* cHRM */ 103, 65, 77, 65, '\0', /* gAMA */ @@ -1747,9 +1764,9 @@ png_create_colormap_entry(png_image_read_control *display, png_uint_32 alpha, int encoding) { png_imagep image = display->image; - const int output_encoding = (image->format & PNG_FORMAT_FLAG_LINEAR) != 0 ? + int output_encoding = (image->format & PNG_FORMAT_FLAG_LINEAR) != 0 ? P_LINEAR : P_sRGB; - const int convert_to_Y = (image->format & PNG_FORMAT_FLAG_COLOR) == 0 && + int convert_to_Y = (image->format & PNG_FORMAT_FLAG_COLOR) == 0 && (red != green || green != blue); if (ip > 255) @@ -1858,13 +1875,13 @@ png_create_colormap_entry(png_image_read_control *display, /* Store the value. */ { # ifdef PNG_FORMAT_AFIRST_SUPPORTED - const int afirst = (image->format & PNG_FORMAT_FLAG_AFIRST) != 0 && + int afirst = (image->format & PNG_FORMAT_FLAG_AFIRST) != 0 && (image->format & PNG_FORMAT_FLAG_ALPHA) != 0; # else # define afirst 0 # endif # ifdef PNG_FORMAT_BGR_SUPPORTED - const int bgr = (image->format & PNG_FORMAT_FLAG_BGR) != 0 ? 2 : 0; + int bgr = (image->format & PNG_FORMAT_FLAG_BGR) != 0 ? 2 : 0; # else # define bgr 0 # endif @@ -1883,7 +1900,7 @@ png_create_colormap_entry(png_image_read_control *display, { case 4: entry[afirst ? 0 : 3] = (png_uint_16)alpha; - /* FALL THROUGH */ + /* FALLTHROUGH */ case 3: if (alpha < 65535) @@ -1905,7 +1922,7 @@ png_create_colormap_entry(png_image_read_control *display, case 2: entry[1 ^ afirst] = (png_uint_16)alpha; - /* FALL THROUGH */ + /* FALLTHROUGH */ case 1: if (alpha < 65535) @@ -1934,6 +1951,7 @@ png_create_colormap_entry(png_image_read_control *display, { case 4: entry[afirst ? 0 : 3] = (png_byte)alpha; + /* FALLTHROUGH */ case 3: entry[afirst + (2 ^ bgr)] = (png_byte)blue; entry[afirst + 1] = (png_byte)green; @@ -1942,6 +1960,7 @@ png_create_colormap_entry(png_image_read_control *display, case 2: entry[1 ^ afirst] = (png_byte)alpha; + /* FALLTHROUGH */ case 1: entry[afirst] = (png_byte)green; break; @@ -2072,11 +2091,11 @@ png_image_read_colormap(png_voidp argument) { png_image_read_control *display = png_voidcast(png_image_read_control*, argument); - const png_imagep image = display->image; + png_imagep image = display->image; - const png_structrp png_ptr = image->opaque->png_ptr; - const png_uint_32 output_format = image->format; - const int output_encoding = (output_format & PNG_FORMAT_FLAG_LINEAR) != 0 ? + png_structrp png_ptr = image->opaque->png_ptr; + png_uint_32 output_format = image->format; + int output_encoding = (output_format & PNG_FORMAT_FLAG_LINEAR) != 0 ? P_LINEAR : P_sRGB; unsigned int cmap_entries; @@ -2789,7 +2808,7 @@ png_image_read_colormap(png_voidp argument) unsigned int num_trans = png_ptr->num_trans; png_const_bytep trans = num_trans > 0 ? png_ptr->trans_alpha : NULL; png_const_colorp colormap = png_ptr->palette; - const int do_background = trans != NULL && + int do_background = trans != NULL && (output_format & PNG_FORMAT_FLAG_ALPHA) == 0; unsigned int i; @@ -2861,7 +2880,7 @@ png_image_read_colormap(png_voidp argument) case P_sRGB: /* Change to 8-bit sRGB */ png_set_alpha_mode_fixed(png_ptr, PNG_ALPHA_PNG, PNG_GAMMA_sRGB); - /* FALL THROUGH */ + /* FALLTHROUGH */ case P_FILE: if (png_ptr->bit_depth > 8) @@ -3179,8 +3198,7 @@ png_image_read_colormapped(png_voidp argument) image->colormap_entries == 244 /* 216 + 1 + 27 */) break; - /* goto bad_output; */ - /* FALL THROUGH */ + goto bad_output; default: bad_output: @@ -3747,7 +3765,13 @@ png_image_read_direct(png_voidp argument) mode = PNG_ALPHA_PNG; output_gamma = PNG_DEFAULT_sRGB; } - + + if ((change & PNG_FORMAT_FLAG_ASSOCIATED_ALPHA) != 0) + { + mode = PNG_ALPHA_OPTIMIZED; + change &= ~PNG_FORMAT_FLAG_ASSOCIATED_ALPHA; + } + /* If 'do_local_background' is set check for the presence of gamma * correction; this is part of the work-round for the libpng bug * described above. @@ -3928,7 +3952,7 @@ png_image_read_direct(png_voidp argument) */ if (linear != 0) { - PNG_CONST png_uint_16 le = 0x0001; + png_uint_16 le = 0x0001; if ((*(png_const_bytep) & le) != 0) png_set_swap(png_ptr); @@ -3973,6 +3997,10 @@ png_image_read_direct(png_voidp argument) else if (do_local_compose != 0) /* internal error */ png_error(png_ptr, "png_image_read: alpha channel lost"); + if ((format & PNG_FORMAT_FLAG_ASSOCIATED_ALPHA) != 0) { + info_format |= PNG_FORMAT_FLAG_ASSOCIATED_ALPHA; + } + if (info_ptr->bit_depth == 16) info_format |= PNG_FORMAT_FLAG_LINEAR; @@ -4086,7 +4114,7 @@ png_image_finish_read(png_imagep image, png_const_colorp background, * original PNG format because it may not occur in the output PNG format * and libpng deals with the issues of reading the original. */ - const unsigned int channels = PNG_IMAGE_PIXEL_CHANNELS(image->format); + unsigned int channels = PNG_IMAGE_PIXEL_CHANNELS(image->format); /* The following checks just the 'row_stride' calculation to ensure it * fits in a signed 32-bit value. Because channels/components can be @@ -4097,7 +4125,7 @@ png_image_finish_read(png_imagep image, png_const_colorp background, if (image->width <= 0x7fffffffU/channels) /* no overflow */ { png_uint_32 check; - const png_uint_32 png_row_stride = image->width * channels; + png_uint_32 png_row_stride = image->width * channels; if (row_stride == 0) row_stride = (png_int_32)/*SAFE*/png_row_stride; @@ -4128,7 +4156,7 @@ png_image_finish_read(png_imagep image, png_const_colorp background, * * NOTE: this will be changed in 1.7 because PNG_IMAGE_BUFFER_SIZE * will be changed to use png_alloc_size_t; bigger images can be - * accomodated on 64-bit systems. + * accommodated on 64-bit systems. */ if (image->height <= 0xffffffffU/PNG_IMAGE_PIXEL_COMPONENT_SIZE(image->format)/check) diff --git a/Externals/libpng/pngrio.c b/Externals/libpng/pngrio.c index 7e26e855ca..7946358101 100644 --- a/Externals/libpng/pngrio.c +++ b/Externals/libpng/pngrio.c @@ -1,10 +1,10 @@ /* pngrio.c - functions for data input * - * Last changed in libpng 1.6.24 [August 4, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2016,2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -29,7 +29,7 @@ * to read more than 64K on a 16-bit machine. */ void /* PRIVATE */ -png_read_data(png_structrp png_ptr, png_bytep data, png_size_t length) +png_read_data(png_structrp png_ptr, png_bytep data, size_t length) { png_debug1(4, "reading %d bytes", (int)length); @@ -47,14 +47,14 @@ png_read_data(png_structrp png_ptr, png_bytep data, png_size_t length) * than changing the library. */ void PNGCBAPI -png_default_read_data(png_structp png_ptr, png_bytep data, png_size_t length) +png_default_read_data(png_structp png_ptr, png_bytep data, size_t length) { - png_size_t check; + size_t check; if (png_ptr == NULL) return; - /* fread() returns 0 on error, so it is OK to store this in a png_size_t + /* fread() returns 0 on error, so it is OK to store this in a size_t * instead of an int, which is what fread() actually returns. */ check = fread(data, 1, length, png_voidcast(png_FILE_p, png_ptr->io_ptr)); diff --git a/Externals/libpng/pngrtran.c b/Externals/libpng/pngrtran.c index 16c1ed6086..9a8fad9f4a 100644 --- a/Externals/libpng/pngrtran.c +++ b/Externals/libpng/pngrtran.c @@ -1,10 +1,10 @@ /* pngrtran.c - transforms the data in a row for PNG readers * - * Last changed in libpng 1.6.29 [March 16, 2017] - * Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -18,6 +18,17 @@ #include "pngpriv.h" +#ifdef PNG_ARM_NEON_IMPLEMENTATION +# if PNG_ARM_NEON_IMPLEMENTATION == 1 +# define PNG_ARM_NEON_INTRINSICS_AVAILABLE +# if defined(_MSC_VER) && defined(_M_ARM64) +# include +# else +# include +# endif +# endif +#endif + #ifdef PNG_READ_SUPPORTED /* Set the action on getting a CRC error for an ancillary or critical chunk. */ @@ -49,6 +60,7 @@ png_set_crc_action(png_structrp png_ptr, int crit_action, int ancil_action) case PNG_CRC_WARN_DISCARD: /* Not a valid action for critical data */ png_warning(png_ptr, "Can't discard critical data on CRC error"); + /* FALLTHROUGH */ case PNG_CRC_ERROR_QUIT: /* Error/quit */ case PNG_CRC_DEFAULT: @@ -291,7 +303,7 @@ png_set_alpha_mode_fixed(png_structrp png_ptr, int mode, * who use the inverse of the gamma value accidentally! Since some of these * values are reasonable this may have to be changed: * - * 1.6.x: changed from 0.07..3 to 0.01..100 (to accomodate the optimal 16-bit + * 1.6.x: changed from 0.07..3 to 0.01..100 (to accommodate the optimal 16-bit * gamma of 36, and its reciprocal.) */ if (output_gamma < 1000 || output_gamma > 10000000) @@ -429,7 +441,7 @@ png_set_quantize(png_structrp png_ptr, png_colorp palette, int i; png_ptr->quantize_index = (png_bytep)png_malloc(png_ptr, - (png_uint_32)((png_uint_32)num_palette * (sizeof (png_byte)))); + (png_alloc_size_t)((png_uint_32)num_palette * (sizeof (png_byte)))); for (i = 0; i < num_palette; i++) png_ptr->quantize_index[i] = (png_byte)i; } @@ -446,7 +458,7 @@ png_set_quantize(png_structrp png_ptr, png_colorp palette, /* Initialize an array to sort colors */ png_ptr->quantize_sort = (png_bytep)png_malloc(png_ptr, - (png_uint_32)((png_uint_32)num_palette * (sizeof (png_byte)))); + (png_alloc_size_t)((png_uint_32)num_palette * (sizeof (png_byte)))); /* Initialize the quantize_sort array */ for (i = 0; i < num_palette; i++) @@ -580,9 +592,11 @@ png_set_quantize(png_structrp png_ptr, png_colorp palette, /* Initialize palette index arrays */ png_ptr->index_to_palette = (png_bytep)png_malloc(png_ptr, - (png_uint_32)((png_uint_32)num_palette * (sizeof (png_byte)))); + (png_alloc_size_t)((png_uint_32)num_palette * + (sizeof (png_byte)))); png_ptr->palette_to_index = (png_bytep)png_malloc(png_ptr, - (png_uint_32)((png_uint_32)num_palette * (sizeof (png_byte)))); + (png_alloc_size_t)((png_uint_32)num_palette * + (sizeof (png_byte)))); /* Initialize the sort array */ for (i = 0; i < num_palette; i++) @@ -591,7 +605,7 @@ png_set_quantize(png_structrp png_ptr, png_colorp palette, png_ptr->palette_to_index[i] = (png_byte)i; } - hash = (png_dsortpp)png_calloc(png_ptr, (png_uint_32)(769 * + hash = (png_dsortpp)png_calloc(png_ptr, (png_alloc_size_t)(769 * (sizeof (png_dsortp)))); num_new_palette = num_palette; @@ -622,7 +636,7 @@ png_set_quantize(png_structrp png_ptr, png_colorp palette, { t = (png_dsortp)png_malloc_warn(png_ptr, - (png_uint_32)(sizeof (png_dsort))); + (png_alloc_size_t)(sizeof (png_dsort))); if (t == NULL) break; @@ -744,12 +758,12 @@ png_set_quantize(png_structrp png_ptr, png_colorp palette, int num_red = (1 << PNG_QUANTIZE_RED_BITS); int num_green = (1 << PNG_QUANTIZE_GREEN_BITS); int num_blue = (1 << PNG_QUANTIZE_BLUE_BITS); - png_size_t num_entries = ((png_size_t)1 << total_bits); + size_t num_entries = ((size_t)1 << total_bits); png_ptr->palette_lookup = (png_bytep)png_calloc(png_ptr, - (png_uint_32)(num_entries * (sizeof (png_byte)))); + (png_alloc_size_t)(num_entries * (sizeof (png_byte)))); - distance = (png_bytep)png_malloc(png_ptr, (png_uint_32)(num_entries * + distance = (png_bytep)png_malloc(png_ptr, (png_alloc_size_t)(num_entries * (sizeof (png_byte)))); memset(distance, 0xff, num_entries * (sizeof (png_byte))); @@ -1168,20 +1182,20 @@ png_init_palette_transformations(png_structrp png_ptr) png_ptr->palette[png_ptr->background.index].blue; #ifdef PNG_READ_INVERT_ALPHA_SUPPORTED - if ((png_ptr->transformations & PNG_INVERT_ALPHA) != 0) - { - if ((png_ptr->transformations & PNG_EXPAND_tRNS) == 0) - { - /* Invert the alpha channel (in tRNS) unless the pixels are - * going to be expanded, in which case leave it for later - */ - int i, istop = png_ptr->num_trans; + if ((png_ptr->transformations & PNG_INVERT_ALPHA) != 0) + { + if ((png_ptr->transformations & PNG_EXPAND_tRNS) == 0) + { + /* Invert the alpha channel (in tRNS) unless the pixels are + * going to be expanded, in which case leave it for later + */ + int i, istop = png_ptr->num_trans; - for (i=0; itrans_alpha[i] = (png_byte)(255 - - png_ptr->trans_alpha[i]); - } - } + for (i = 0; i < istop; i++) + png_ptr->trans_alpha[i] = + (png_byte)(255 - png_ptr->trans_alpha[i]); + } + } #endif /* READ_INVERT_ALPHA */ } } /* background expand and (therefore) no alpha association. */ @@ -1253,7 +1267,7 @@ png_init_rgb_transformations(png_structrp png_ptr) default: case 8: - /* FALL THROUGH (Already 8 bits) */ + /* FALLTHROUGH */ /* (Already 8 bits) */ case 16: /* Already a full 16 bits */ @@ -1314,7 +1328,7 @@ png_init_read_transformations(png_structrp png_ptr) else if (png_ptr->screen_gamma != 0) /* The converse - assume the file matches the screen, note that this - * perhaps undesireable default can (from 1.5.4) be changed by calling + * perhaps undesirable default can (from 1.5.4) be changed by calling * png_set_alpha_mode (even if the alpha handling mode isn't required * or isn't changed from the default.) */ @@ -1882,7 +1896,7 @@ png_init_read_transformations(png_structrp png_ptr) png_ptr->transformations &= ~PNG_SHIFT; - /* significant bits can be in the range 1 to 7 for a meaninful result, if + /* significant bits can be in the range 1 to 7 for a meaningful result, if * the number of significant bits is 0 then no shift is done (this is an * error condition which is silently ignored.) */ @@ -2148,8 +2162,8 @@ png_do_unpack(png_row_infop row_info, png_bytep row) { case 1: { - png_bytep sp = row + (png_size_t)((row_width - 1) >> 3); - png_bytep dp = row + (png_size_t)row_width - 1; + png_bytep sp = row + (size_t)((row_width - 1) >> 3); + png_bytep dp = row + (size_t)row_width - 1; png_uint_32 shift = 7U - ((row_width + 7U) & 0x07); for (i = 0; i < row_width; i++) { @@ -2172,8 +2186,8 @@ png_do_unpack(png_row_infop row_info, png_bytep row) case 2: { - png_bytep sp = row + (png_size_t)((row_width - 1) >> 2); - png_bytep dp = row + (png_size_t)row_width - 1; + png_bytep sp = row + (size_t)((row_width - 1) >> 2); + png_bytep dp = row + (size_t)row_width - 1; png_uint_32 shift = ((3U - ((row_width + 3U) & 0x03)) << 1); for (i = 0; i < row_width; i++) { @@ -2195,8 +2209,8 @@ png_do_unpack(png_row_infop row_info, png_bytep row) case 4: { - png_bytep sp = row + (png_size_t)((row_width - 1) >> 1); - png_bytep dp = row + (png_size_t)row_width - 1; + png_bytep sp = row + (size_t)((row_width - 1) >> 1); + png_bytep dp = row + (size_t)row_width - 1; png_uint_32 shift = ((1U - ((row_width + 1U) & 0x01)) << 2); for (i = 0; i < row_width; i++) { @@ -2460,95 +2474,94 @@ png_do_chop(png_row_infop row_info, png_bytep row) static void png_do_read_swap_alpha(png_row_infop row_info, png_bytep row) { + png_uint_32 row_width = row_info->width; + png_debug(1, "in png_do_read_swap_alpha"); + if (row_info->color_type == PNG_COLOR_TYPE_RGB_ALPHA) { - png_uint_32 row_width = row_info->width; - if (row_info->color_type == PNG_COLOR_TYPE_RGB_ALPHA) + /* This converts from RGBA to ARGB */ + if (row_info->bit_depth == 8) { - /* This converts from RGBA to ARGB */ - if (row_info->bit_depth == 8) + png_bytep sp = row + row_info->rowbytes; + png_bytep dp = sp; + png_byte save; + png_uint_32 i; + + for (i = 0; i < row_width; i++) { - png_bytep sp = row + row_info->rowbytes; - png_bytep dp = sp; - png_byte save; - png_uint_32 i; - - for (i = 0; i < row_width; i++) - { - save = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = save; - } + save = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = save; } - -#ifdef PNG_READ_16BIT_SUPPORTED - /* This converts from RRGGBBAA to AARRGGBB */ - else - { - png_bytep sp = row + row_info->rowbytes; - png_bytep dp = sp; - png_byte save[2]; - png_uint_32 i; - - for (i = 0; i < row_width; i++) - { - save[0] = *(--sp); - save[1] = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = save[0]; - *(--dp) = save[1]; - } - } -#endif } - else if (row_info->color_type == PNG_COLOR_TYPE_GRAY_ALPHA) +#ifdef PNG_READ_16BIT_SUPPORTED + /* This converts from RRGGBBAA to AARRGGBB */ + else { - /* This converts from GA to AG */ - if (row_info->bit_depth == 8) - { - png_bytep sp = row + row_info->rowbytes; - png_bytep dp = sp; - png_byte save; - png_uint_32 i; + png_bytep sp = row + row_info->rowbytes; + png_bytep dp = sp; + png_byte save[2]; + png_uint_32 i; - for (i = 0; i < row_width; i++) - { - save = *(--sp); - *(--dp) = *(--sp); - *(--dp) = save; - } + for (i = 0; i < row_width; i++) + { + save[0] = *(--sp); + save[1] = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = save[0]; + *(--dp) = save[1]; } + } +#endif + } + + else if (row_info->color_type == PNG_COLOR_TYPE_GRAY_ALPHA) + { + /* This converts from GA to AG */ + if (row_info->bit_depth == 8) + { + png_bytep sp = row + row_info->rowbytes; + png_bytep dp = sp; + png_byte save; + png_uint_32 i; + + for (i = 0; i < row_width; i++) + { + save = *(--sp); + *(--dp) = *(--sp); + *(--dp) = save; + } + } #ifdef PNG_READ_16BIT_SUPPORTED - /* This converts from GGAA to AAGG */ - else - { - png_bytep sp = row + row_info->rowbytes; - png_bytep dp = sp; - png_byte save[2]; - png_uint_32 i; + /* This converts from GGAA to AAGG */ + else + { + png_bytep sp = row + row_info->rowbytes; + png_bytep dp = sp; + png_byte save[2]; + png_uint_32 i; - for (i = 0; i < row_width; i++) - { - save[0] = *(--sp); - save[1] = *(--sp); - *(--dp) = *(--sp); - *(--dp) = *(--sp); - *(--dp) = save[0]; - *(--dp) = save[1]; - } + for (i = 0; i < row_width; i++) + { + save[0] = *(--sp); + save[1] = *(--sp); + *(--dp) = *(--sp); + *(--dp) = *(--sp); + *(--dp) = save[0]; + *(--dp) = save[1]; } -#endif } +#endif } } #endif @@ -2678,8 +2691,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, if ((flags & PNG_FLAG_FILLER_AFTER) != 0) { /* This changes the data from G to GX */ - png_bytep sp = row + (png_size_t)row_width; - png_bytep dp = sp + (png_size_t)row_width; + png_bytep sp = row + (size_t)row_width; + png_bytep dp = sp + (size_t)row_width; for (i = 1; i < row_width; i++) { *(--dp) = lo_filler; @@ -2694,8 +2707,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, else { /* This changes the data from G to XG */ - png_bytep sp = row + (png_size_t)row_width; - png_bytep dp = sp + (png_size_t)row_width; + png_bytep sp = row + (size_t)row_width; + png_bytep dp = sp + (size_t)row_width; for (i = 0; i < row_width; i++) { *(--dp) = *(--sp); @@ -2713,8 +2726,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, if ((flags & PNG_FLAG_FILLER_AFTER) != 0) { /* This changes the data from GG to GGXX */ - png_bytep sp = row + (png_size_t)row_width * 2; - png_bytep dp = sp + (png_size_t)row_width * 2; + png_bytep sp = row + (size_t)row_width * 2; + png_bytep dp = sp + (size_t)row_width * 2; for (i = 1; i < row_width; i++) { *(--dp) = lo_filler; @@ -2732,8 +2745,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, else { /* This changes the data from GG to XXGG */ - png_bytep sp = row + (png_size_t)row_width * 2; - png_bytep dp = sp + (png_size_t)row_width * 2; + png_bytep sp = row + (size_t)row_width * 2; + png_bytep dp = sp + (size_t)row_width * 2; for (i = 0; i < row_width; i++) { *(--dp) = *(--sp); @@ -2755,8 +2768,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, if ((flags & PNG_FLAG_FILLER_AFTER) != 0) { /* This changes the data from RGB to RGBX */ - png_bytep sp = row + (png_size_t)row_width * 3; - png_bytep dp = sp + (png_size_t)row_width; + png_bytep sp = row + (size_t)row_width * 3; + png_bytep dp = sp + (size_t)row_width; for (i = 1; i < row_width; i++) { *(--dp) = lo_filler; @@ -2773,8 +2786,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, else { /* This changes the data from RGB to XRGB */ - png_bytep sp = row + (png_size_t)row_width * 3; - png_bytep dp = sp + (png_size_t)row_width; + png_bytep sp = row + (size_t)row_width * 3; + png_bytep dp = sp + (size_t)row_width; for (i = 0; i < row_width; i++) { *(--dp) = *(--sp); @@ -2794,8 +2807,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, if ((flags & PNG_FLAG_FILLER_AFTER) != 0) { /* This changes the data from RRGGBB to RRGGBBXX */ - png_bytep sp = row + (png_size_t)row_width * 6; - png_bytep dp = sp + (png_size_t)row_width * 2; + png_bytep sp = row + (size_t)row_width * 6; + png_bytep dp = sp + (size_t)row_width * 2; for (i = 1; i < row_width; i++) { *(--dp) = lo_filler; @@ -2817,8 +2830,8 @@ png_do_read_filler(png_row_infop row_info, png_bytep row, else { /* This changes the data from RRGGBB to XXRRGGBB */ - png_bytep sp = row + (png_size_t)row_width * 6; - png_bytep dp = sp + (png_size_t)row_width * 2; + png_bytep sp = row + (size_t)row_width * 6; + png_bytep dp = sp + (size_t)row_width * 2; for (i = 0; i < row_width; i++) { *(--dp) = *(--sp); @@ -2859,8 +2872,8 @@ png_do_gray_to_rgb(png_row_infop row_info, png_bytep row) if (row_info->bit_depth == 8) { /* This changes G to RGB */ - png_bytep sp = row + (png_size_t)row_width - 1; - png_bytep dp = sp + (png_size_t)row_width * 2; + png_bytep sp = row + (size_t)row_width - 1; + png_bytep dp = sp + (size_t)row_width * 2; for (i = 0; i < row_width; i++) { *(dp--) = *sp; @@ -2872,8 +2885,8 @@ png_do_gray_to_rgb(png_row_infop row_info, png_bytep row) else { /* This changes GG to RRGGBB */ - png_bytep sp = row + (png_size_t)row_width * 2 - 1; - png_bytep dp = sp + (png_size_t)row_width * 4; + png_bytep sp = row + (size_t)row_width * 2 - 1; + png_bytep dp = sp + (size_t)row_width * 4; for (i = 0; i < row_width; i++) { *(dp--) = *sp; @@ -2891,8 +2904,8 @@ png_do_gray_to_rgb(png_row_infop row_info, png_bytep row) if (row_info->bit_depth == 8) { /* This changes GA to RGBA */ - png_bytep sp = row + (png_size_t)row_width * 2 - 1; - png_bytep dp = sp + (png_size_t)row_width * 2; + png_bytep sp = row + (size_t)row_width * 2 - 1; + png_bytep dp = sp + (size_t)row_width * 2; for (i = 0; i < row_width; i++) { *(dp--) = *(sp--); @@ -2905,8 +2918,8 @@ png_do_gray_to_rgb(png_row_infop row_info, png_bytep row) else { /* This changes GGAA to RRGGBBAA */ - png_bytep sp = row + (png_size_t)row_width * 4 - 1; - png_bytep dp = sp + (png_size_t)row_width * 4; + png_bytep sp = row + (size_t)row_width * 4 - 1; + png_bytep dp = sp + (size_t)row_width * 4; for (i = 0; i < row_width; i++) { *(dp--) = *(sp--); @@ -2934,7 +2947,7 @@ png_do_gray_to_rgb(png_row_infop row_info, png_bytep row) * using the equation given in Poynton's ColorFAQ of 1998-01-04 at * (THIS LINK IS DEAD June 2008 but * versions dated 1998 through November 2002 have been archived at - * http://web.archive.org/web/20000816232553/http://www.inforamp.net/ + * https://web.archive.org/web/20000816232553/www.inforamp.net/ * ~poynton/notes/colour_and_gamma/ColorFAQ.txt ) * Charles Poynton poynton at poynton.com * @@ -2977,14 +2990,13 @@ png_do_gray_to_rgb(png_row_infop row_info, png_bytep row) * values this results in an implicit assumption that the original PNG RGB * values were linear. * - * Other integer coefficents can be used via png_set_rgb_to_gray(). Because + * Other integer coefficients can be used via png_set_rgb_to_gray(). Because * the API takes just red and green coefficients the blue coefficient is * calculated to make the sum 32768. This will result in different rounding * to that used above. */ static int png_do_rgb_to_gray(png_structrp png_ptr, png_row_infop row_info, png_bytep row) - { int rgb_error = 0; @@ -2993,12 +3005,11 @@ png_do_rgb_to_gray(png_structrp png_ptr, png_row_infop row_info, png_bytep row) if ((row_info->color_type & PNG_COLOR_MASK_PALETTE) == 0 && (row_info->color_type & PNG_COLOR_MASK_COLOR) != 0) { - PNG_CONST png_uint_32 rc = png_ptr->rgb_to_gray_red_coeff; - PNG_CONST png_uint_32 gc = png_ptr->rgb_to_gray_green_coeff; - PNG_CONST png_uint_32 bc = 32768 - rc - gc; - PNG_CONST png_uint_32 row_width = row_info->width; - PNG_CONST int have_alpha = - (row_info->color_type & PNG_COLOR_MASK_ALPHA) != 0; + png_uint_32 rc = png_ptr->rgb_to_gray_red_coeff; + png_uint_32 gc = png_ptr->rgb_to_gray_green_coeff; + png_uint_32 bc = 32768 - rc - gc; + png_uint_32 row_width = row_info->width; + int have_alpha = (row_info->color_type & PNG_COLOR_MASK_ALPHA) != 0; if (row_info->bit_depth == 8) { @@ -3206,419 +3217,229 @@ png_do_compose(png_row_infop row_info, png_bytep row, png_structrp png_ptr) png_debug(1, "in png_do_compose"); + switch (row_info->color_type) { - switch (row_info->color_type) + case PNG_COLOR_TYPE_GRAY: { - case PNG_COLOR_TYPE_GRAY: + switch (row_info->bit_depth) { - switch (row_info->bit_depth) + case 1: { - case 1: + sp = row; + shift = 7; + for (i = 0; i < row_width; i++) { - sp = row; - shift = 7; - for (i = 0; i < row_width; i++) + if ((png_uint_16)((*sp >> shift) & 0x01) + == png_ptr->trans_color.gray) { - if ((png_uint_16)((*sp >> shift) & 0x01) - == png_ptr->trans_color.gray) - { - unsigned int tmp = *sp & (0x7f7f >> (7 - shift)); - tmp |= - (unsigned int)(png_ptr->background.gray << shift); - *sp = (png_byte)(tmp & 0xff); - } - - if (shift == 0) - { - shift = 7; - sp++; - } - - else - shift--; + unsigned int tmp = *sp & (0x7f7f >> (7 - shift)); + tmp |= + (unsigned int)(png_ptr->background.gray << shift); + *sp = (png_byte)(tmp & 0xff); } - break; - } - case 2: - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_table != NULL) + if (shift == 0) { - sp = row; - shift = 6; - for (i = 0; i < row_width; i++) - { - if ((png_uint_16)((*sp >> shift) & 0x03) - == png_ptr->trans_color.gray) - { - unsigned int tmp = *sp & (0x3f3f >> (6 - shift)); - tmp |= - (unsigned int)png_ptr->background.gray << shift; - *sp = (png_byte)(tmp & 0xff); - } - - else - { - unsigned int p = (*sp >> shift) & 0x03; - unsigned int g = (gamma_table [p | (p << 2) | - (p << 4) | (p << 6)] >> 6) & 0x03; - unsigned int tmp = *sp & (0x3f3f >> (6 - shift)); - tmp |= (unsigned int)(g << shift); - *sp = (png_byte)(tmp & 0xff); - } - - if (shift == 0) - { - shift = 6; - sp++; - } - - else - shift -= 2; - } + shift = 7; + sp++; } else -#endif - { - sp = row; - shift = 6; - for (i = 0; i < row_width; i++) - { - if ((png_uint_16)((*sp >> shift) & 0x03) - == png_ptr->trans_color.gray) - { - unsigned int tmp = *sp & (0x3f3f >> (6 - shift)); - tmp |= - (unsigned int)png_ptr->background.gray << shift; - *sp = (png_byte)(tmp & 0xff); - } - - if (shift == 0) - { - shift = 6; - sp++; - } - - else - shift -= 2; - } - } - break; + shift--; } - - case 4: - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_table != NULL) - { - sp = row; - shift = 4; - for (i = 0; i < row_width; i++) - { - if ((png_uint_16)((*sp >> shift) & 0x0f) - == png_ptr->trans_color.gray) - { - unsigned int tmp = *sp & (0x0f0f >> (4 - shift)); - tmp |= - (unsigned int)(png_ptr->background.gray << shift); - *sp = (png_byte)(tmp & 0xff); - } - - else - { - unsigned int p = (*sp >> shift) & 0x0f; - unsigned int g = (gamma_table[p | (p << 4)] >> 4) & - 0x0f; - unsigned int tmp = *sp & (0x0f0f >> (4 - shift)); - tmp |= (unsigned int)(g << shift); - *sp = (png_byte)(tmp & 0xff); - } - - if (shift == 0) - { - shift = 4; - sp++; - } - - else - shift -= 4; - } - } - - else -#endif - { - sp = row; - shift = 4; - for (i = 0; i < row_width; i++) - { - if ((png_uint_16)((*sp >> shift) & 0x0f) - == png_ptr->trans_color.gray) - { - unsigned int tmp = *sp & (0x0f0f >> (4 - shift)); - tmp |= - (unsigned int)(png_ptr->background.gray << shift); - *sp = (png_byte)(tmp & 0xff); - } - - if (shift == 0) - { - shift = 4; - sp++; - } - - else - shift -= 4; - } - } - break; - } - - case 8: - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_table != NULL) - { - sp = row; - for (i = 0; i < row_width; i++, sp++) - { - if (*sp == png_ptr->trans_color.gray) - *sp = (png_byte)png_ptr->background.gray; - - else - *sp = gamma_table[*sp]; - } - } - else -#endif - { - sp = row; - for (i = 0; i < row_width; i++, sp++) - { - if (*sp == png_ptr->trans_color.gray) - *sp = (png_byte)png_ptr->background.gray; - } - } - break; - } - - case 16: - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_16 != NULL) - { - sp = row; - for (i = 0; i < row_width; i++, sp += 2) - { - png_uint_16 v; - - v = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - - if (v == png_ptr->trans_color.gray) - { - /* Background is already in screen gamma */ - *sp = (png_byte)((png_ptr->background.gray >> 8) - & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.gray - & 0xff); - } - - else - { - v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; - *sp = (png_byte)((v >> 8) & 0xff); - *(sp + 1) = (png_byte)(v & 0xff); - } - } - } - else -#endif - { - sp = row; - for (i = 0; i < row_width; i++, sp += 2) - { - png_uint_16 v; - - v = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - - if (v == png_ptr->trans_color.gray) - { - *sp = (png_byte)((png_ptr->background.gray >> 8) - & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.gray - & 0xff); - } - } - } - break; - } - - default: - break; + break; } - break; - } - case PNG_COLOR_TYPE_RGB: - { - if (row_info->bit_depth == 8) + case 2: { #ifdef PNG_READ_GAMMA_SUPPORTED if (gamma_table != NULL) { sp = row; - for (i = 0; i < row_width; i++, sp += 3) + shift = 6; + for (i = 0; i < row_width; i++) { - if (*sp == png_ptr->trans_color.red && - *(sp + 1) == png_ptr->trans_color.green && - *(sp + 2) == png_ptr->trans_color.blue) + if ((png_uint_16)((*sp >> shift) & 0x03) + == png_ptr->trans_color.gray) { - *sp = (png_byte)png_ptr->background.red; - *(sp + 1) = (png_byte)png_ptr->background.green; - *(sp + 2) = (png_byte)png_ptr->background.blue; + unsigned int tmp = *sp & (0x3f3f >> (6 - shift)); + tmp |= + (unsigned int)png_ptr->background.gray << shift; + *sp = (png_byte)(tmp & 0xff); } else { - *sp = gamma_table[*sp]; - *(sp + 1) = gamma_table[*(sp + 1)]; - *(sp + 2) = gamma_table[*(sp + 2)]; + unsigned int p = (*sp >> shift) & 0x03; + unsigned int g = (gamma_table [p | (p << 2) | + (p << 4) | (p << 6)] >> 6) & 0x03; + unsigned int tmp = *sp & (0x3f3f >> (6 - shift)); + tmp |= (unsigned int)(g << shift); + *sp = (png_byte)(tmp & 0xff); } + + if (shift == 0) + { + shift = 6; + sp++; + } + + else + shift -= 2; + } + } + + else +#endif + { + sp = row; + shift = 6; + for (i = 0; i < row_width; i++) + { + if ((png_uint_16)((*sp >> shift) & 0x03) + == png_ptr->trans_color.gray) + { + unsigned int tmp = *sp & (0x3f3f >> (6 - shift)); + tmp |= + (unsigned int)png_ptr->background.gray << shift; + *sp = (png_byte)(tmp & 0xff); + } + + if (shift == 0) + { + shift = 6; + sp++; + } + + else + shift -= 2; + } + } + break; + } + + case 4: + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_table != NULL) + { + sp = row; + shift = 4; + for (i = 0; i < row_width; i++) + { + if ((png_uint_16)((*sp >> shift) & 0x0f) + == png_ptr->trans_color.gray) + { + unsigned int tmp = *sp & (0x0f0f >> (4 - shift)); + tmp |= + (unsigned int)(png_ptr->background.gray << shift); + *sp = (png_byte)(tmp & 0xff); + } + + else + { + unsigned int p = (*sp >> shift) & 0x0f; + unsigned int g = (gamma_table[p | (p << 4)] >> 4) & + 0x0f; + unsigned int tmp = *sp & (0x0f0f >> (4 - shift)); + tmp |= (unsigned int)(g << shift); + *sp = (png_byte)(tmp & 0xff); + } + + if (shift == 0) + { + shift = 4; + sp++; + } + + else + shift -= 4; + } + } + + else +#endif + { + sp = row; + shift = 4; + for (i = 0; i < row_width; i++) + { + if ((png_uint_16)((*sp >> shift) & 0x0f) + == png_ptr->trans_color.gray) + { + unsigned int tmp = *sp & (0x0f0f >> (4 - shift)); + tmp |= + (unsigned int)(png_ptr->background.gray << shift); + *sp = (png_byte)(tmp & 0xff); + } + + if (shift == 0) + { + shift = 4; + sp++; + } + + else + shift -= 4; + } + } + break; + } + + case 8: + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_table != NULL) + { + sp = row; + for (i = 0; i < row_width; i++, sp++) + { + if (*sp == png_ptr->trans_color.gray) + *sp = (png_byte)png_ptr->background.gray; + + else + *sp = gamma_table[*sp]; } } else #endif { sp = row; - for (i = 0; i < row_width; i++, sp += 3) + for (i = 0; i < row_width; i++, sp++) { - if (*sp == png_ptr->trans_color.red && - *(sp + 1) == png_ptr->trans_color.green && - *(sp + 2) == png_ptr->trans_color.blue) - { - *sp = (png_byte)png_ptr->background.red; - *(sp + 1) = (png_byte)png_ptr->background.green; - *(sp + 2) = (png_byte)png_ptr->background.blue; - } + if (*sp == png_ptr->trans_color.gray) + *sp = (png_byte)png_ptr->background.gray; } } + break; } - else /* if (row_info->bit_depth == 16) */ + + case 16: { #ifdef PNG_READ_GAMMA_SUPPORTED if (gamma_16 != NULL) - { - sp = row; - for (i = 0; i < row_width; i++, sp += 6) - { - png_uint_16 r = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - - png_uint_16 g = (png_uint_16)(((*(sp + 2)) << 8) - + *(sp + 3)); - - png_uint_16 b = (png_uint_16)(((*(sp + 4)) << 8) - + *(sp + 5)); - - if (r == png_ptr->trans_color.red && - g == png_ptr->trans_color.green && - b == png_ptr->trans_color.blue) - { - /* Background is already in screen gamma */ - *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); - *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) - & 0xff); - *(sp + 3) = (png_byte)(png_ptr->background.green - & 0xff); - *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) - & 0xff); - *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); - } - - else - { - png_uint_16 v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; - *sp = (png_byte)((v >> 8) & 0xff); - *(sp + 1) = (png_byte)(v & 0xff); - - v = gamma_16[*(sp + 3) >> gamma_shift][*(sp + 2)]; - *(sp + 2) = (png_byte)((v >> 8) & 0xff); - *(sp + 3) = (png_byte)(v & 0xff); - - v = gamma_16[*(sp + 5) >> gamma_shift][*(sp + 4)]; - *(sp + 4) = (png_byte)((v >> 8) & 0xff); - *(sp + 5) = (png_byte)(v & 0xff); - } - } - } - - else -#endif - { - sp = row; - for (i = 0; i < row_width; i++, sp += 6) - { - png_uint_16 r = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - - png_uint_16 g = (png_uint_16)(((*(sp + 2)) << 8) - + *(sp + 3)); - - png_uint_16 b = (png_uint_16)(((*(sp + 4)) << 8) - + *(sp + 5)); - - if (r == png_ptr->trans_color.red && - g == png_ptr->trans_color.green && - b == png_ptr->trans_color.blue) - { - *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); - *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) - & 0xff); - *(sp + 3) = (png_byte)(png_ptr->background.green - & 0xff); - *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) - & 0xff); - *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); - } - } - } - } - break; - } - - case PNG_COLOR_TYPE_GRAY_ALPHA: - { - if (row_info->bit_depth == 8) - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_to_1 != NULL && gamma_from_1 != NULL && - gamma_table != NULL) { sp = row; for (i = 0; i < row_width; i++, sp += 2) { - png_uint_16 a = *(sp + 1); + png_uint_16 v; - if (a == 0xff) - *sp = gamma_table[*sp]; + v = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - else if (a == 0) + if (v == png_ptr->trans_color.gray) { /* Background is already in screen gamma */ - *sp = (png_byte)png_ptr->background.gray; + *sp = (png_byte)((png_ptr->background.gray >> 8) + & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.gray + & 0xff); } else { - png_byte v, w; - - v = gamma_to_1[*sp]; - png_composite(w, v, a, png_ptr->background_1.gray); - if (optimize == 0) - w = gamma_from_1[w]; - *sp = w; + v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; + *sp = (png_byte)((v >> 8) & 0xff); + *(sp + 1) = (png_byte)(v & 0xff); } } } @@ -3628,298 +3449,486 @@ png_do_compose(png_row_infop row_info, png_bytep row, png_structrp png_ptr) sp = row; for (i = 0; i < row_width; i++, sp += 2) { - png_byte a = *(sp + 1); + png_uint_16 v; - if (a == 0) - *sp = (png_byte)png_ptr->background.gray; + v = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - else if (a < 0xff) - png_composite(*sp, *sp, a, png_ptr->background.gray); - } - } - } - else /* if (png_ptr->bit_depth == 16) */ - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_16 != NULL && gamma_16_from_1 != NULL && - gamma_16_to_1 != NULL) - { - sp = row; - for (i = 0; i < row_width; i++, sp += 4) - { - png_uint_16 a = (png_uint_16)(((*(sp + 2)) << 8) - + *(sp + 3)); - - if (a == (png_uint_16)0xffff) - { - png_uint_16 v; - - v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; - *sp = (png_byte)((v >> 8) & 0xff); - *(sp + 1) = (png_byte)(v & 0xff); - } - - else if (a == 0) - { - /* Background is already in screen gamma */ - *sp = (png_byte)((png_ptr->background.gray >> 8) - & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.gray & 0xff); - } - - else - { - png_uint_16 g, v, w; - - g = gamma_16_to_1[*(sp + 1) >> gamma_shift][*sp]; - png_composite_16(v, g, a, png_ptr->background_1.gray); - if (optimize != 0) - w = v; - else - w = gamma_16_from_1[(v & 0xff) >> - gamma_shift][v >> 8]; - *sp = (png_byte)((w >> 8) & 0xff); - *(sp + 1) = (png_byte)(w & 0xff); - } - } - } - else -#endif - { - sp = row; - for (i = 0; i < row_width; i++, sp += 4) - { - png_uint_16 a = (png_uint_16)(((*(sp + 2)) << 8) - + *(sp + 3)); - - if (a == 0) + if (v == png_ptr->trans_color.gray) { *sp = (png_byte)((png_ptr->background.gray >> 8) - & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.gray & 0xff); - } - - else if (a < 0xffff) - { - png_uint_16 g, v; - - g = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - png_composite_16(v, g, a, png_ptr->background.gray); - *sp = (png_byte)((v >> 8) & 0xff); - *(sp + 1) = (png_byte)(v & 0xff); + & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.gray + & 0xff); } } } + break; } - break; + + default: + break; } - - case PNG_COLOR_TYPE_RGB_ALPHA: - { - if (row_info->bit_depth == 8) - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_to_1 != NULL && gamma_from_1 != NULL && - gamma_table != NULL) - { - sp = row; - for (i = 0; i < row_width; i++, sp += 4) - { - png_byte a = *(sp + 3); - - if (a == 0xff) - { - *sp = gamma_table[*sp]; - *(sp + 1) = gamma_table[*(sp + 1)]; - *(sp + 2) = gamma_table[*(sp + 2)]; - } - - else if (a == 0) - { - /* Background is already in screen gamma */ - *sp = (png_byte)png_ptr->background.red; - *(sp + 1) = (png_byte)png_ptr->background.green; - *(sp + 2) = (png_byte)png_ptr->background.blue; - } - - else - { - png_byte v, w; - - v = gamma_to_1[*sp]; - png_composite(w, v, a, png_ptr->background_1.red); - if (optimize == 0) w = gamma_from_1[w]; - *sp = w; - - v = gamma_to_1[*(sp + 1)]; - png_composite(w, v, a, png_ptr->background_1.green); - if (optimize == 0) w = gamma_from_1[w]; - *(sp + 1) = w; - - v = gamma_to_1[*(sp + 2)]; - png_composite(w, v, a, png_ptr->background_1.blue); - if (optimize == 0) w = gamma_from_1[w]; - *(sp + 2) = w; - } - } - } - else -#endif - { - sp = row; - for (i = 0; i < row_width; i++, sp += 4) - { - png_byte a = *(sp + 3); - - if (a == 0) - { - *sp = (png_byte)png_ptr->background.red; - *(sp + 1) = (png_byte)png_ptr->background.green; - *(sp + 2) = (png_byte)png_ptr->background.blue; - } - - else if (a < 0xff) - { - png_composite(*sp, *sp, a, png_ptr->background.red); - - png_composite(*(sp + 1), *(sp + 1), a, - png_ptr->background.green); - - png_composite(*(sp + 2), *(sp + 2), a, - png_ptr->background.blue); - } - } - } - } - else /* if (row_info->bit_depth == 16) */ - { -#ifdef PNG_READ_GAMMA_SUPPORTED - if (gamma_16 != NULL && gamma_16_from_1 != NULL && - gamma_16_to_1 != NULL) - { - sp = row; - for (i = 0; i < row_width; i++, sp += 8) - { - png_uint_16 a = (png_uint_16)(((png_uint_16)(*(sp + 6)) - << 8) + (png_uint_16)(*(sp + 7))); - - if (a == (png_uint_16)0xffff) - { - png_uint_16 v; - - v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; - *sp = (png_byte)((v >> 8) & 0xff); - *(sp + 1) = (png_byte)(v & 0xff); - - v = gamma_16[*(sp + 3) >> gamma_shift][*(sp + 2)]; - *(sp + 2) = (png_byte)((v >> 8) & 0xff); - *(sp + 3) = (png_byte)(v & 0xff); - - v = gamma_16[*(sp + 5) >> gamma_shift][*(sp + 4)]; - *(sp + 4) = (png_byte)((v >> 8) & 0xff); - *(sp + 5) = (png_byte)(v & 0xff); - } - - else if (a == 0) - { - /* Background is already in screen gamma */ - *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); - *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) - & 0xff); - *(sp + 3) = (png_byte)(png_ptr->background.green - & 0xff); - *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) - & 0xff); - *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); - } - - else - { - png_uint_16 v, w; - - v = gamma_16_to_1[*(sp + 1) >> gamma_shift][*sp]; - png_composite_16(w, v, a, png_ptr->background_1.red); - if (optimize == 0) - w = gamma_16_from_1[((w & 0xff) >> gamma_shift)][w >> - 8]; - *sp = (png_byte)((w >> 8) & 0xff); - *(sp + 1) = (png_byte)(w & 0xff); - - v = gamma_16_to_1[*(sp + 3) >> gamma_shift][*(sp + 2)]; - png_composite_16(w, v, a, png_ptr->background_1.green); - if (optimize == 0) - w = gamma_16_from_1[((w & 0xff) >> gamma_shift)][w >> - 8]; - - *(sp + 2) = (png_byte)((w >> 8) & 0xff); - *(sp + 3) = (png_byte)(w & 0xff); - - v = gamma_16_to_1[*(sp + 5) >> gamma_shift][*(sp + 4)]; - png_composite_16(w, v, a, png_ptr->background_1.blue); - if (optimize == 0) - w = gamma_16_from_1[((w & 0xff) >> gamma_shift)][w >> - 8]; - - *(sp + 4) = (png_byte)((w >> 8) & 0xff); - *(sp + 5) = (png_byte)(w & 0xff); - } - } - } - - else -#endif - { - sp = row; - for (i = 0; i < row_width; i++, sp += 8) - { - png_uint_16 a = (png_uint_16)(((png_uint_16)(*(sp + 6)) - << 8) + (png_uint_16)(*(sp + 7))); - - if (a == 0) - { - *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); - *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); - *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) - & 0xff); - *(sp + 3) = (png_byte)(png_ptr->background.green - & 0xff); - *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) - & 0xff); - *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); - } - - else if (a < 0xffff) - { - png_uint_16 v; - - png_uint_16 r = (png_uint_16)(((*sp) << 8) + *(sp + 1)); - png_uint_16 g = (png_uint_16)(((*(sp + 2)) << 8) - + *(sp + 3)); - png_uint_16 b = (png_uint_16)(((*(sp + 4)) << 8) - + *(sp + 5)); - - png_composite_16(v, r, a, png_ptr->background.red); - *sp = (png_byte)((v >> 8) & 0xff); - *(sp + 1) = (png_byte)(v & 0xff); - - png_composite_16(v, g, a, png_ptr->background.green); - *(sp + 2) = (png_byte)((v >> 8) & 0xff); - *(sp + 3) = (png_byte)(v & 0xff); - - png_composite_16(v, b, a, png_ptr->background.blue); - *(sp + 4) = (png_byte)((v >> 8) & 0xff); - *(sp + 5) = (png_byte)(v & 0xff); - } - } - } - } - break; - } - - default: - break; + break; } + + case PNG_COLOR_TYPE_RGB: + { + if (row_info->bit_depth == 8) + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_table != NULL) + { + sp = row; + for (i = 0; i < row_width; i++, sp += 3) + { + if (*sp == png_ptr->trans_color.red && + *(sp + 1) == png_ptr->trans_color.green && + *(sp + 2) == png_ptr->trans_color.blue) + { + *sp = (png_byte)png_ptr->background.red; + *(sp + 1) = (png_byte)png_ptr->background.green; + *(sp + 2) = (png_byte)png_ptr->background.blue; + } + + else + { + *sp = gamma_table[*sp]; + *(sp + 1) = gamma_table[*(sp + 1)]; + *(sp + 2) = gamma_table[*(sp + 2)]; + } + } + } + else +#endif + { + sp = row; + for (i = 0; i < row_width; i++, sp += 3) + { + if (*sp == png_ptr->trans_color.red && + *(sp + 1) == png_ptr->trans_color.green && + *(sp + 2) == png_ptr->trans_color.blue) + { + *sp = (png_byte)png_ptr->background.red; + *(sp + 1) = (png_byte)png_ptr->background.green; + *(sp + 2) = (png_byte)png_ptr->background.blue; + } + } + } + } + else /* if (row_info->bit_depth == 16) */ + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_16 != NULL) + { + sp = row; + for (i = 0; i < row_width; i++, sp += 6) + { + png_uint_16 r = (png_uint_16)(((*sp) << 8) + *(sp + 1)); + + png_uint_16 g = (png_uint_16)(((*(sp + 2)) << 8) + + *(sp + 3)); + + png_uint_16 b = (png_uint_16)(((*(sp + 4)) << 8) + + *(sp + 5)); + + if (r == png_ptr->trans_color.red && + g == png_ptr->trans_color.green && + b == png_ptr->trans_color.blue) + { + /* Background is already in screen gamma */ + *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); + *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) + & 0xff); + *(sp + 3) = (png_byte)(png_ptr->background.green + & 0xff); + *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) + & 0xff); + *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); + } + + else + { + png_uint_16 v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; + *sp = (png_byte)((v >> 8) & 0xff); + *(sp + 1) = (png_byte)(v & 0xff); + + v = gamma_16[*(sp + 3) >> gamma_shift][*(sp + 2)]; + *(sp + 2) = (png_byte)((v >> 8) & 0xff); + *(sp + 3) = (png_byte)(v & 0xff); + + v = gamma_16[*(sp + 5) >> gamma_shift][*(sp + 4)]; + *(sp + 4) = (png_byte)((v >> 8) & 0xff); + *(sp + 5) = (png_byte)(v & 0xff); + } + } + } + + else +#endif + { + sp = row; + for (i = 0; i < row_width; i++, sp += 6) + { + png_uint_16 r = (png_uint_16)(((*sp) << 8) + *(sp + 1)); + + png_uint_16 g = (png_uint_16)(((*(sp + 2)) << 8) + + *(sp + 3)); + + png_uint_16 b = (png_uint_16)(((*(sp + 4)) << 8) + + *(sp + 5)); + + if (r == png_ptr->trans_color.red && + g == png_ptr->trans_color.green && + b == png_ptr->trans_color.blue) + { + *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); + *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) + & 0xff); + *(sp + 3) = (png_byte)(png_ptr->background.green + & 0xff); + *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) + & 0xff); + *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); + } + } + } + } + break; + } + + case PNG_COLOR_TYPE_GRAY_ALPHA: + { + if (row_info->bit_depth == 8) + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_to_1 != NULL && gamma_from_1 != NULL && + gamma_table != NULL) + { + sp = row; + for (i = 0; i < row_width; i++, sp += 2) + { + png_uint_16 a = *(sp + 1); + + if (a == 0xff) + *sp = gamma_table[*sp]; + + else if (a == 0) + { + /* Background is already in screen gamma */ + *sp = (png_byte)png_ptr->background.gray; + } + + else + { + png_byte v, w; + + v = gamma_to_1[*sp]; + png_composite(w, v, a, png_ptr->background_1.gray); + if (optimize == 0) + w = gamma_from_1[w]; + *sp = w; + } + } + } + else +#endif + { + sp = row; + for (i = 0; i < row_width; i++, sp += 2) + { + png_byte a = *(sp + 1); + + if (a == 0) + *sp = (png_byte)png_ptr->background.gray; + + else if (a < 0xff) + png_composite(*sp, *sp, a, png_ptr->background.gray); + } + } + } + else /* if (png_ptr->bit_depth == 16) */ + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_16 != NULL && gamma_16_from_1 != NULL && + gamma_16_to_1 != NULL) + { + sp = row; + for (i = 0; i < row_width; i++, sp += 4) + { + png_uint_16 a = (png_uint_16)(((*(sp + 2)) << 8) + + *(sp + 3)); + + if (a == (png_uint_16)0xffff) + { + png_uint_16 v; + + v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; + *sp = (png_byte)((v >> 8) & 0xff); + *(sp + 1) = (png_byte)(v & 0xff); + } + + else if (a == 0) + { + /* Background is already in screen gamma */ + *sp = (png_byte)((png_ptr->background.gray >> 8) + & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.gray & 0xff); + } + + else + { + png_uint_16 g, v, w; + + g = gamma_16_to_1[*(sp + 1) >> gamma_shift][*sp]; + png_composite_16(v, g, a, png_ptr->background_1.gray); + if (optimize != 0) + w = v; + else + w = gamma_16_from_1[(v & 0xff) >> + gamma_shift][v >> 8]; + *sp = (png_byte)((w >> 8) & 0xff); + *(sp + 1) = (png_byte)(w & 0xff); + } + } + } + else +#endif + { + sp = row; + for (i = 0; i < row_width; i++, sp += 4) + { + png_uint_16 a = (png_uint_16)(((*(sp + 2)) << 8) + + *(sp + 3)); + + if (a == 0) + { + *sp = (png_byte)((png_ptr->background.gray >> 8) + & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.gray & 0xff); + } + + else if (a < 0xffff) + { + png_uint_16 g, v; + + g = (png_uint_16)(((*sp) << 8) + *(sp + 1)); + png_composite_16(v, g, a, png_ptr->background.gray); + *sp = (png_byte)((v >> 8) & 0xff); + *(sp + 1) = (png_byte)(v & 0xff); + } + } + } + } + break; + } + + case PNG_COLOR_TYPE_RGB_ALPHA: + { + if (row_info->bit_depth == 8) + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_to_1 != NULL && gamma_from_1 != NULL && + gamma_table != NULL) + { + sp = row; + for (i = 0; i < row_width; i++, sp += 4) + { + png_byte a = *(sp + 3); + + if (a == 0xff) + { + *sp = gamma_table[*sp]; + *(sp + 1) = gamma_table[*(sp + 1)]; + *(sp + 2) = gamma_table[*(sp + 2)]; + } + + else if (a == 0) + { + /* Background is already in screen gamma */ + *sp = (png_byte)png_ptr->background.red; + *(sp + 1) = (png_byte)png_ptr->background.green; + *(sp + 2) = (png_byte)png_ptr->background.blue; + } + + else + { + png_byte v, w; + + v = gamma_to_1[*sp]; + png_composite(w, v, a, png_ptr->background_1.red); + if (optimize == 0) w = gamma_from_1[w]; + *sp = w; + + v = gamma_to_1[*(sp + 1)]; + png_composite(w, v, a, png_ptr->background_1.green); + if (optimize == 0) w = gamma_from_1[w]; + *(sp + 1) = w; + + v = gamma_to_1[*(sp + 2)]; + png_composite(w, v, a, png_ptr->background_1.blue); + if (optimize == 0) w = gamma_from_1[w]; + *(sp + 2) = w; + } + } + } + else +#endif + { + sp = row; + for (i = 0; i < row_width; i++, sp += 4) + { + png_byte a = *(sp + 3); + + if (a == 0) + { + *sp = (png_byte)png_ptr->background.red; + *(sp + 1) = (png_byte)png_ptr->background.green; + *(sp + 2) = (png_byte)png_ptr->background.blue; + } + + else if (a < 0xff) + { + png_composite(*sp, *sp, a, png_ptr->background.red); + + png_composite(*(sp + 1), *(sp + 1), a, + png_ptr->background.green); + + png_composite(*(sp + 2), *(sp + 2), a, + png_ptr->background.blue); + } + } + } + } + else /* if (row_info->bit_depth == 16) */ + { +#ifdef PNG_READ_GAMMA_SUPPORTED + if (gamma_16 != NULL && gamma_16_from_1 != NULL && + gamma_16_to_1 != NULL) + { + sp = row; + for (i = 0; i < row_width; i++, sp += 8) + { + png_uint_16 a = (png_uint_16)(((png_uint_16)(*(sp + 6)) + << 8) + (png_uint_16)(*(sp + 7))); + + if (a == (png_uint_16)0xffff) + { + png_uint_16 v; + + v = gamma_16[*(sp + 1) >> gamma_shift][*sp]; + *sp = (png_byte)((v >> 8) & 0xff); + *(sp + 1) = (png_byte)(v & 0xff); + + v = gamma_16[*(sp + 3) >> gamma_shift][*(sp + 2)]; + *(sp + 2) = (png_byte)((v >> 8) & 0xff); + *(sp + 3) = (png_byte)(v & 0xff); + + v = gamma_16[*(sp + 5) >> gamma_shift][*(sp + 4)]; + *(sp + 4) = (png_byte)((v >> 8) & 0xff); + *(sp + 5) = (png_byte)(v & 0xff); + } + + else if (a == 0) + { + /* Background is already in screen gamma */ + *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); + *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) + & 0xff); + *(sp + 3) = (png_byte)(png_ptr->background.green + & 0xff); + *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) + & 0xff); + *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); + } + + else + { + png_uint_16 v, w; + + v = gamma_16_to_1[*(sp + 1) >> gamma_shift][*sp]; + png_composite_16(w, v, a, png_ptr->background_1.red); + if (optimize == 0) + w = gamma_16_from_1[((w & 0xff) >> gamma_shift)][w >> + 8]; + *sp = (png_byte)((w >> 8) & 0xff); + *(sp + 1) = (png_byte)(w & 0xff); + + v = gamma_16_to_1[*(sp + 3) >> gamma_shift][*(sp + 2)]; + png_composite_16(w, v, a, png_ptr->background_1.green); + if (optimize == 0) + w = gamma_16_from_1[((w & 0xff) >> gamma_shift)][w >> + 8]; + + *(sp + 2) = (png_byte)((w >> 8) & 0xff); + *(sp + 3) = (png_byte)(w & 0xff); + + v = gamma_16_to_1[*(sp + 5) >> gamma_shift][*(sp + 4)]; + png_composite_16(w, v, a, png_ptr->background_1.blue); + if (optimize == 0) + w = gamma_16_from_1[((w & 0xff) >> gamma_shift)][w >> + 8]; + + *(sp + 4) = (png_byte)((w >> 8) & 0xff); + *(sp + 5) = (png_byte)(w & 0xff); + } + } + } + + else +#endif + { + sp = row; + for (i = 0; i < row_width; i++, sp += 8) + { + png_uint_16 a = (png_uint_16)(((png_uint_16)(*(sp + 6)) + << 8) + (png_uint_16)(*(sp + 7))); + + if (a == 0) + { + *sp = (png_byte)((png_ptr->background.red >> 8) & 0xff); + *(sp + 1) = (png_byte)(png_ptr->background.red & 0xff); + *(sp + 2) = (png_byte)((png_ptr->background.green >> 8) + & 0xff); + *(sp + 3) = (png_byte)(png_ptr->background.green + & 0xff); + *(sp + 4) = (png_byte)((png_ptr->background.blue >> 8) + & 0xff); + *(sp + 5) = (png_byte)(png_ptr->background.blue & 0xff); + } + + else if (a < 0xffff) + { + png_uint_16 v; + + png_uint_16 r = (png_uint_16)(((*sp) << 8) + *(sp + 1)); + png_uint_16 g = (png_uint_16)(((*(sp + 2)) << 8) + + *(sp + 3)); + png_uint_16 b = (png_uint_16)(((*(sp + 4)) << 8) + + *(sp + 5)); + + png_composite_16(v, r, a, png_ptr->background.red); + *sp = (png_byte)((v >> 8) & 0xff); + *(sp + 1) = (png_byte)(v & 0xff); + + png_composite_16(v, g, a, png_ptr->background.green); + *(sp + 2) = (png_byte)((v >> 8) & 0xff); + *(sp + 3) = (png_byte)(v & 0xff); + + png_composite_16(v, b, a, png_ptr->background.blue); + *(sp + 4) = (png_byte)((v >> 8) & 0xff); + *(sp + 5) = (png_byte)(v & 0xff); + } + } + } + } + break; + } + + default: + break; } } #endif /* READ_BACKGROUND || READ_ALPHA_MODE */ @@ -4143,12 +4152,11 @@ png_do_encode_alpha(png_row_infop row_info, png_bytep row, png_structrp png_ptr) { if (row_info->bit_depth == 8) { - PNG_CONST png_bytep table = png_ptr->gamma_from_1; + png_bytep table = png_ptr->gamma_from_1; if (table != NULL) { - PNG_CONST int step = - (row_info->color_type & PNG_COLOR_MASK_COLOR) ? 4 : 2; + int step = (row_info->color_type & PNG_COLOR_MASK_COLOR) ? 4 : 2; /* The alpha channel is the last component: */ row += step - 1; @@ -4162,13 +4170,12 @@ png_do_encode_alpha(png_row_infop row_info, png_bytep row, png_structrp png_ptr) else if (row_info->bit_depth == 16) { - PNG_CONST png_uint_16pp table = png_ptr->gamma_16_from_1; - PNG_CONST int gamma_shift = png_ptr->gamma_shift; + png_uint_16pp table = png_ptr->gamma_16_from_1; + int gamma_shift = png_ptr->gamma_shift; if (table != NULL) { - PNG_CONST int step = - (row_info->color_type & PNG_COLOR_MASK_COLOR) ? 8 : 4; + int step = (row_info->color_type & PNG_COLOR_MASK_COLOR) ? 8 : 4; /* The alpha channel is the last component: */ row += step - 2; @@ -4199,8 +4206,9 @@ png_do_encode_alpha(png_row_infop row_info, png_bytep row, png_structrp png_ptr) * upon whether you supply trans and num_trans. */ static void -png_do_expand_palette(png_row_infop row_info, png_bytep row, - png_const_colorp palette, png_const_bytep trans_alpha, int num_trans) +png_do_expand_palette(png_structrp png_ptr, png_row_infop row_info, + png_bytep row, png_const_colorp palette, png_const_bytep trans_alpha, + int num_trans) { int shift, value; png_bytep sp, dp; @@ -4217,8 +4225,8 @@ png_do_expand_palette(png_row_infop row_info, png_bytep row, { case 1: { - sp = row + (png_size_t)((row_width - 1) >> 3); - dp = row + (png_size_t)row_width - 1; + sp = row + (size_t)((row_width - 1) >> 3); + dp = row + (size_t)row_width - 1; shift = 7 - (int)((row_width + 7) & 0x07); for (i = 0; i < row_width; i++) { @@ -4244,8 +4252,8 @@ png_do_expand_palette(png_row_infop row_info, png_bytep row, case 2: { - sp = row + (png_size_t)((row_width - 1) >> 2); - dp = row + (png_size_t)row_width - 1; + sp = row + (size_t)((row_width - 1) >> 2); + dp = row + (size_t)row_width - 1; shift = (int)((3 - ((row_width + 3) & 0x03)) << 1); for (i = 0; i < row_width; i++) { @@ -4267,8 +4275,8 @@ png_do_expand_palette(png_row_infop row_info, png_bytep row, case 4: { - sp = row + (png_size_t)((row_width - 1) >> 1); - dp = row + (png_size_t)row_width - 1; + sp = row + (size_t)((row_width - 1) >> 1); + dp = row + (size_t)row_width - 1; shift = (int)((row_width & 0x01) << 2); for (i = 0; i < row_width; i++) { @@ -4301,17 +4309,30 @@ png_do_expand_palette(png_row_infop row_info, png_bytep row, { if (num_trans > 0) { - sp = row + (png_size_t)row_width - 1; - dp = row + ((png_size_t)row_width << 2) - 1; + sp = row + (size_t)row_width - 1; + dp = row + ((size_t)row_width << 2) - 1; - for (i = 0; i < row_width; i++) + i = 0; +#ifdef PNG_ARM_NEON_INTRINSICS_AVAILABLE + if (png_ptr->riffled_palette != NULL) + { + /* The RGBA optimization works with png_ptr->bit_depth == 8 + * but sometimes row_info->bit_depth has been changed to 8. + * In these cases, the palette hasn't been riffled. + */ + i = png_do_expand_palette_rgba8_neon(png_ptr, row_info, row, + &sp, &dp); + } +#else + PNG_UNUSED(png_ptr) +#endif + + for (; i < row_width; i++) { if ((int)(*sp) >= num_trans) *dp-- = 0xff; - else *dp-- = trans_alpha[*sp]; - *dp-- = palette[*sp].blue; *dp-- = palette[*sp].green; *dp-- = palette[*sp].red; @@ -4326,10 +4347,17 @@ png_do_expand_palette(png_row_infop row_info, png_bytep row, else { - sp = row + (png_size_t)row_width - 1; - dp = row + (png_size_t)(row_width * 3) - 1; + sp = row + (size_t)row_width - 1; + dp = row + (size_t)(row_width * 3) - 1; + i = 0; +#ifdef PNG_ARM_NEON_INTRINSICS_AVAILABLE + i = png_do_expand_palette_rgb8_neon(png_ptr, row_info, row, + &sp, &dp); +#else + PNG_UNUSED(png_ptr) +#endif - for (i = 0; i < row_width; i++) + for (; i < row_width; i++) { *dp-- = palette[*sp].blue; *dp-- = palette[*sp].green; @@ -4362,195 +4390,130 @@ png_do_expand(png_row_infop row_info, png_bytep row, png_debug(1, "in png_do_expand"); + if (row_info->color_type == PNG_COLOR_TYPE_GRAY) { - if (row_info->color_type == PNG_COLOR_TYPE_GRAY) + unsigned int gray = trans_color != NULL ? trans_color->gray : 0; + + if (row_info->bit_depth < 8) { - unsigned int gray = trans_color != NULL ? trans_color->gray : 0; - - if (row_info->bit_depth < 8) + switch (row_info->bit_depth) { - switch (row_info->bit_depth) + case 1: { - case 1: - { - gray = (gray & 0x01) * 0xff; - sp = row + (png_size_t)((row_width - 1) >> 3); - dp = row + (png_size_t)row_width - 1; - shift = 7 - (int)((row_width + 7) & 0x07); - for (i = 0; i < row_width; i++) - { - if ((*sp >> shift) & 0x01) - *dp = 0xff; - - else - *dp = 0; - - if (shift == 7) - { - shift = 0; - sp--; - } - - else - shift++; - - dp--; - } - break; - } - - case 2: - { - gray = (gray & 0x03) * 0x55; - sp = row + (png_size_t)((row_width - 1) >> 2); - dp = row + (png_size_t)row_width - 1; - shift = (int)((3 - ((row_width + 3) & 0x03)) << 1); - for (i = 0; i < row_width; i++) - { - value = (*sp >> shift) & 0x03; - *dp = (png_byte)(value | (value << 2) | (value << 4) | - (value << 6)); - if (shift == 6) - { - shift = 0; - sp--; - } - - else - shift += 2; - - dp--; - } - break; - } - - case 4: - { - gray = (gray & 0x0f) * 0x11; - sp = row + (png_size_t)((row_width - 1) >> 1); - dp = row + (png_size_t)row_width - 1; - shift = (int)((1 - ((row_width + 1) & 0x01)) << 2); - for (i = 0; i < row_width; i++) - { - value = (*sp >> shift) & 0x0f; - *dp = (png_byte)(value | (value << 4)); - if (shift == 4) - { - shift = 0; - sp--; - } - - else - shift = 4; - - dp--; - } - break; - } - - default: - break; - } - - row_info->bit_depth = 8; - row_info->pixel_depth = 8; - row_info->rowbytes = row_width; - } - - if (trans_color != NULL) - { - if (row_info->bit_depth == 8) - { - gray = gray & 0xff; - sp = row + (png_size_t)row_width - 1; - dp = row + ((png_size_t)row_width << 1) - 1; - + gray = (gray & 0x01) * 0xff; + sp = row + (size_t)((row_width - 1) >> 3); + dp = row + (size_t)row_width - 1; + shift = 7 - (int)((row_width + 7) & 0x07); for (i = 0; i < row_width; i++) { - if ((*sp & 0xffU) == gray) - *dp-- = 0; + if ((*sp >> shift) & 0x01) + *dp = 0xff; else - *dp-- = 0xff; + *dp = 0; - *dp-- = *sp--; - } - } - - else if (row_info->bit_depth == 16) - { - unsigned int gray_high = (gray >> 8) & 0xff; - unsigned int gray_low = gray & 0xff; - sp = row + row_info->rowbytes - 1; - dp = row + (row_info->rowbytes << 1) - 1; - for (i = 0; i < row_width; i++) - { - if ((*(sp - 1) & 0xffU) == gray_high && - (*(sp) & 0xffU) == gray_low) + if (shift == 7) { - *dp-- = 0; - *dp-- = 0; + shift = 0; + sp--; } else - { - *dp-- = 0xff; - *dp-- = 0xff; - } + shift++; - *dp-- = *sp--; - *dp-- = *sp--; + dp--; } + break; } - row_info->color_type = PNG_COLOR_TYPE_GRAY_ALPHA; - row_info->channels = 2; - row_info->pixel_depth = (png_byte)(row_info->bit_depth << 1); - row_info->rowbytes = PNG_ROWBYTES(row_info->pixel_depth, - row_width); + case 2: + { + gray = (gray & 0x03) * 0x55; + sp = row + (size_t)((row_width - 1) >> 2); + dp = row + (size_t)row_width - 1; + shift = (int)((3 - ((row_width + 3) & 0x03)) << 1); + for (i = 0; i < row_width; i++) + { + value = (*sp >> shift) & 0x03; + *dp = (png_byte)(value | (value << 2) | (value << 4) | + (value << 6)); + if (shift == 6) + { + shift = 0; + sp--; + } + + else + shift += 2; + + dp--; + } + break; + } + + case 4: + { + gray = (gray & 0x0f) * 0x11; + sp = row + (size_t)((row_width - 1) >> 1); + dp = row + (size_t)row_width - 1; + shift = (int)((1 - ((row_width + 1) & 0x01)) << 2); + for (i = 0; i < row_width; i++) + { + value = (*sp >> shift) & 0x0f; + *dp = (png_byte)(value | (value << 4)); + if (shift == 4) + { + shift = 0; + sp--; + } + + else + shift = 4; + + dp--; + } + break; + } + + default: + break; } + + row_info->bit_depth = 8; + row_info->pixel_depth = 8; + row_info->rowbytes = row_width; } - else if (row_info->color_type == PNG_COLOR_TYPE_RGB && - trans_color != NULL) + + if (trans_color != NULL) { if (row_info->bit_depth == 8) { - png_byte red = (png_byte)(trans_color->red & 0xff); - png_byte green = (png_byte)(trans_color->green & 0xff); - png_byte blue = (png_byte)(trans_color->blue & 0xff); - sp = row + (png_size_t)row_info->rowbytes - 1; - dp = row + ((png_size_t)row_width << 2) - 1; + gray = gray & 0xff; + sp = row + (size_t)row_width - 1; + dp = row + ((size_t)row_width << 1) - 1; + for (i = 0; i < row_width; i++) { - if (*(sp - 2) == red && *(sp - 1) == green && *(sp) == blue) + if ((*sp & 0xffU) == gray) *dp-- = 0; else *dp-- = 0xff; *dp-- = *sp--; - *dp-- = *sp--; - *dp-- = *sp--; } } + else if (row_info->bit_depth == 16) { - png_byte red_high = (png_byte)((trans_color->red >> 8) & 0xff); - png_byte green_high = (png_byte)((trans_color->green >> 8) & 0xff); - png_byte blue_high = (png_byte)((trans_color->blue >> 8) & 0xff); - png_byte red_low = (png_byte)(trans_color->red & 0xff); - png_byte green_low = (png_byte)(trans_color->green & 0xff); - png_byte blue_low = (png_byte)(trans_color->blue & 0xff); + unsigned int gray_high = (gray >> 8) & 0xff; + unsigned int gray_low = gray & 0xff; sp = row + row_info->rowbytes - 1; - dp = row + ((png_size_t)row_width << 3) - 1; + dp = row + (row_info->rowbytes << 1) - 1; for (i = 0; i < row_width; i++) { - if (*(sp - 5) == red_high && - *(sp - 4) == red_low && - *(sp - 3) == green_high && - *(sp - 2) == green_low && - *(sp - 1) == blue_high && - *(sp ) == blue_low) + if ((*(sp - 1) & 0xffU) == gray_high && + (*(sp) & 0xffU) == gray_low) { *dp-- = 0; *dp-- = 0; @@ -4564,18 +4527,81 @@ png_do_expand(png_row_infop row_info, png_bytep row, *dp-- = *sp--; *dp-- = *sp--; - *dp-- = *sp--; - *dp-- = *sp--; - *dp-- = *sp--; - *dp-- = *sp--; } } - row_info->color_type = PNG_COLOR_TYPE_RGB_ALPHA; - row_info->channels = 4; - row_info->pixel_depth = (png_byte)(row_info->bit_depth << 2); - row_info->rowbytes = PNG_ROWBYTES(row_info->pixel_depth, row_width); + + row_info->color_type = PNG_COLOR_TYPE_GRAY_ALPHA; + row_info->channels = 2; + row_info->pixel_depth = (png_byte)(row_info->bit_depth << 1); + row_info->rowbytes = PNG_ROWBYTES(row_info->pixel_depth, + row_width); } } + else if (row_info->color_type == PNG_COLOR_TYPE_RGB && + trans_color != NULL) + { + if (row_info->bit_depth == 8) + { + png_byte red = (png_byte)(trans_color->red & 0xff); + png_byte green = (png_byte)(trans_color->green & 0xff); + png_byte blue = (png_byte)(trans_color->blue & 0xff); + sp = row + (size_t)row_info->rowbytes - 1; + dp = row + ((size_t)row_width << 2) - 1; + for (i = 0; i < row_width; i++) + { + if (*(sp - 2) == red && *(sp - 1) == green && *(sp) == blue) + *dp-- = 0; + + else + *dp-- = 0xff; + + *dp-- = *sp--; + *dp-- = *sp--; + *dp-- = *sp--; + } + } + else if (row_info->bit_depth == 16) + { + png_byte red_high = (png_byte)((trans_color->red >> 8) & 0xff); + png_byte green_high = (png_byte)((trans_color->green >> 8) & 0xff); + png_byte blue_high = (png_byte)((trans_color->blue >> 8) & 0xff); + png_byte red_low = (png_byte)(trans_color->red & 0xff); + png_byte green_low = (png_byte)(trans_color->green & 0xff); + png_byte blue_low = (png_byte)(trans_color->blue & 0xff); + sp = row + row_info->rowbytes - 1; + dp = row + ((size_t)row_width << 3) - 1; + for (i = 0; i < row_width; i++) + { + if (*(sp - 5) == red_high && + *(sp - 4) == red_low && + *(sp - 3) == green_high && + *(sp - 2) == green_low && + *(sp - 1) == blue_high && + *(sp ) == blue_low) + { + *dp-- = 0; + *dp-- = 0; + } + + else + { + *dp-- = 0xff; + *dp-- = 0xff; + } + + *dp-- = *sp--; + *dp-- = *sp--; + *dp-- = *sp--; + *dp-- = *sp--; + *dp-- = *sp--; + *dp-- = *sp--; + } + } + row_info->color_type = PNG_COLOR_TYPE_RGB_ALPHA; + row_info->channels = 4; + row_info->pixel_depth = (png_byte)(row_info->bit_depth << 2); + row_info->rowbytes = PNG_ROWBYTES(row_info->pixel_depth, row_width); + } } #endif @@ -4601,7 +4627,9 @@ png_do_expand_16(png_row_infop row_info, png_bytep row) png_byte *sp = row + row_info->rowbytes; /* source, last byte + 1 */ png_byte *dp = sp + row_info->rowbytes; /* destination, end + 1 */ while (dp > sp) - dp[-2] = dp[-1] = *--sp, dp -= 2; + { + dp[-2] = dp[-1] = *--sp; dp -= 2; + } row_info->rowbytes *= 2; row_info->bit_depth = 16; @@ -4743,7 +4771,19 @@ png_do_read_transformations(png_structrp png_ptr, png_row_infop row_info) { if (row_info->color_type == PNG_COLOR_TYPE_PALETTE) { - png_do_expand_palette(row_info, png_ptr->row_buf + 1, +#ifdef PNG_ARM_NEON_INTRINSICS_AVAILABLE + if ((png_ptr->num_trans > 0) && (png_ptr->bit_depth == 8)) + { + if (png_ptr->riffled_palette == NULL) + { + /* Initialize the accelerated palette expansion. */ + png_ptr->riffled_palette = + (png_bytep)png_malloc(png_ptr, 256 * 4); + png_riffle_palette_neon(png_ptr); + } + } +#endif + png_do_expand_palette(png_ptr, row_info, png_ptr->row_buf + 1, png_ptr->palette, png_ptr->trans_alpha, png_ptr->num_trans); } @@ -4755,8 +4795,7 @@ png_do_read_transformations(png_structrp png_ptr, png_row_infop row_info) &(png_ptr->trans_color)); else - png_do_expand(row_info, png_ptr->row_buf + 1, - NULL); + png_do_expand(row_info, png_ptr->row_buf + 1, NULL); } } #endif @@ -4980,7 +5019,7 @@ png_do_read_transformations(png_structrp png_ptr, png_row_infop row_info) (png_ptr, /* png_ptr */ row_info, /* row_info: */ /* png_uint_32 width; width of row */ - /* png_size_t rowbytes; number of bytes in row */ + /* size_t rowbytes; number of bytes in row */ /* png_byte color_type; color type of pixels */ /* png_byte bit_depth; bit depth of samples */ /* png_byte channels; number of channels (1-4) */ diff --git a/Externals/libpng/pngrutil.c b/Externals/libpng/pngrutil.c index 76f079a69f..d5fa08c397 100644 --- a/Externals/libpng/pngrutil.c +++ b/Externals/libpng/pngrutil.c @@ -1,10 +1,10 @@ /* pngrutil.c - utilities to read a PNG file * - * Last changed in libpng 1.6.29 [March 16, 2017] - * Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -102,7 +102,7 @@ png_get_int_32)(png_const_bytep buf) png_uint_16 (PNGAPI png_get_uint_16)(png_const_bytep buf) { - /* ANSI-C requires an int value to accomodate at least 16 bits so this + /* ANSI-C requires an int value to accommodate at least 16 bits so this * works and allows the compiler not to worry about possible narrowing * on 32-bit systems. (Pre-ANSI systems did not make integers smaller * than 16 bits either.) @@ -120,7 +120,7 @@ png_get_uint_16)(png_const_bytep buf) void /* PRIVATE */ png_read_sig(png_structrp png_ptr, png_inforp info_ptr) { - png_size_t num_checked, num_to_check; + size_t num_checked, num_to_check; /* Exit if the user application does not expect a signature. */ if (png_ptr->sig_bytes >= 8) @@ -181,6 +181,9 @@ png_read_chunk_header(png_structrp png_ptr) /* Check to see if chunk name is valid. */ png_check_chunk_name(png_ptr, png_ptr->chunk_name); + /* Check for too-large chunk length */ + png_check_chunk_length(png_ptr, length); + #ifdef PNG_IO_STATE_SUPPORTED png_ptr->io_state = PNG_IO_READING | PNG_IO_CHUNK_DATA; #endif @@ -311,6 +314,7 @@ png_read_buffer(png_structrp png_ptr, png_alloc_size_t new_size, int warn) if (buffer != NULL) { + memset(buffer, 0, new_size); /* just in case */ png_ptr->read_buffer = buffer; png_ptr->read_buffer_size = new_size; } @@ -670,6 +674,8 @@ png_decompress_chunk(png_structrp png_ptr, if (text != NULL) { + memset(text, 0, buffer_size); + ret = png_inflate(png_ptr, png_ptr->chunk_name, 1/*finish*/, png_ptr->read_buffer + prefix_size, &lzsize, text + prefix_size, newlength); @@ -733,9 +739,7 @@ png_decompress_chunk(png_structrp png_ptr, { /* inflateReset failed, store the error message */ png_zstream_error(png_ptr, ret); - - if (ret == Z_STREAM_END) - ret = PNG_UNEXPECTED_ZLIB_RETURN; + ret = PNG_UNEXPECTED_ZLIB_RETURN; } } @@ -1377,11 +1381,13 @@ png_handle_iCCP(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) * chunk is just ignored, so does not invalidate the color space. An * alternative is to set the 'invalid' flags at the start of this routine * and only clear them in they were not set before and all the tests pass. - * The minimum 'deflate' stream is assumed to be just the 2 byte header and - * 4 byte checksum. The keyword must be at least one character and there is - * a terminator (0) byte and the compression method. */ - if (length < 9) + + /* The keyword must be at least one character and there is a + * terminator (0) byte and the compression method byte, and the + * 'zlib' datastream is at least 11 bytes. + */ + if (length < 14) { png_crc_finish(png_ptr, length); png_chunk_benign_error(png_ptr, "too short"); @@ -1413,6 +1419,16 @@ png_handle_iCCP(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) png_crc_read(png_ptr, (png_bytep)keyword, read_length); length -= read_length; + /* The minimum 'zlib' stream is assumed to be just the 2 byte header, + * 5 bytes minimum 'deflate' stream, and the 4 byte checksum. + */ + if (length < 11) + { + png_crc_finish(png_ptr, length); + png_chunk_benign_error(png_ptr, "too short"); + return; + } + keyword_length = 0; while (keyword_length < 80 && keyword_length < read_length && keyword[keyword_length] != 0) @@ -1431,7 +1447,7 @@ png_handle_iCCP(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) if (png_inflate_claim(png_ptr, png_iCCP) == Z_OK) { - Byte profile_header[132]; + Byte profile_header[132]={0}; Byte local_buffer[PNG_INFLATE_BUF_SIZE]; png_alloc_size_t size = (sizeof profile_header); @@ -1445,8 +1461,7 @@ png_handle_iCCP(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) { /* We have the ICC profile header; do the basic header checks. */ - const png_uint_32 profile_length = - png_get_uint_32(profile_header); + png_uint_32 profile_length = png_get_uint_32(profile_header); if (png_icc_check_length(png_ptr, &png_ptr->colorspace, keyword, profile_length) != 0) @@ -1461,10 +1476,10 @@ png_handle_iCCP(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) /* Now read the tag table; a variable size buffer is * needed at this point, allocate one for the whole * profile. The header check has already validated - * that none of these stuff will overflow. + * that none of this stuff will overflow. */ - const png_uint_32 tag_count = png_get_uint_32( - profile_header+128); + png_uint_32 tag_count = + png_get_uint_32(profile_header + 128); png_bytep profile = png_read_buffer(png_ptr, profile_length, 2/*silent*/); @@ -1568,19 +1583,11 @@ png_handle_iCCP(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) return; } } - - else if (size > 0) - errmsg = "truncated"; - -#ifndef __COVERITY__ - else + if (errmsg == NULL) errmsg = png_ptr->zstream.msg; -#endif } - /* else png_icc_check_tag_table output an error */ } - else /* profile truncated */ errmsg = png_ptr->zstream.msg; } @@ -1640,7 +1647,7 @@ png_handle_sPLT(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) int entry_size, i; png_uint_32 skip = 0; png_uint_32 dl; - png_size_t max_dl; + size_t max_dl; png_debug(1, "in png_handle_sPLT"); @@ -1989,6 +1996,15 @@ png_handle_bKGD(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) else if ((png_ptr->color_type & PNG_COLOR_MASK_COLOR) == 0) /* GRAY */ { + if (png_ptr->bit_depth <= 8) + { + if (buf[0] != 0 || buf[1] >= (unsigned int)(1 << png_ptr->bit_depth)) + { + png_chunk_benign_error(png_ptr, "invalid gray level"); + return; + } + } + background.index = 0; background.red = background.green = @@ -1998,6 +2014,15 @@ png_handle_bKGD(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) else { + if (png_ptr->bit_depth <= 8) + { + if (buf[0] != 0 || buf[2] != 0 || buf[4] != 0) + { + png_chunk_benign_error(png_ptr, "invalid color"); + return; + } + } + background.index = 0; background.red = png_get_uint_16(buf); background.green = png_get_uint_16(buf + 2); @@ -2009,6 +2034,69 @@ png_handle_bKGD(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) } #endif +#ifdef PNG_READ_eXIf_SUPPORTED +void /* PRIVATE */ +png_handle_eXIf(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) +{ + unsigned int i; + + png_debug(1, "in png_handle_eXIf"); + + if ((png_ptr->mode & PNG_HAVE_IHDR) == 0) + png_chunk_error(png_ptr, "missing IHDR"); + + if (length < 2) + { + png_crc_finish(png_ptr, length); + png_chunk_benign_error(png_ptr, "too short"); + return; + } + + else if (info_ptr == NULL || (info_ptr->valid & PNG_INFO_eXIf) != 0) + { + png_crc_finish(png_ptr, length); + png_chunk_benign_error(png_ptr, "duplicate"); + return; + } + + info_ptr->free_me |= PNG_FREE_EXIF; + + info_ptr->eXIf_buf = png_voidcast(png_bytep, + png_malloc_warn(png_ptr, length)); + + if (info_ptr->eXIf_buf == NULL) + { + png_crc_finish(png_ptr, length); + png_chunk_benign_error(png_ptr, "out of memory"); + return; + } + + for (i = 0; i < length; i++) + { + png_byte buf[1]; + png_crc_read(png_ptr, buf, 1); + info_ptr->eXIf_buf[i] = buf[0]; + if (i == 1 && buf[0] != 'M' && buf[0] != 'I' + && info_ptr->eXIf_buf[0] != buf[0]) + { + png_crc_finish(png_ptr, length); + png_chunk_benign_error(png_ptr, "incorrect byte-order specifier"); + png_free(png_ptr, info_ptr->eXIf_buf); + info_ptr->eXIf_buf = NULL; + return; + } + } + + if (png_crc_finish(png_ptr, 0) != 0) + return; + + png_set_eXIf_1(png_ptr, info_ptr, length, info_ptr->eXIf_buf); + + png_free(png_ptr, info_ptr->eXIf_buf); + info_ptr->eXIf_buf = NULL; +} +#endif + #ifdef PNG_READ_hIST_SUPPORTED void /* PRIVATE */ png_handle_hIST(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) @@ -2288,7 +2376,7 @@ void /* PRIVATE */ png_handle_sCAL(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) { png_bytep buffer; - png_size_t i; + size_t i; int state; png_debug(1, "in png_handle_sCAL"); @@ -2358,7 +2446,7 @@ png_handle_sCAL(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) else { - png_size_t heighti = i; + size_t heighti = i; state = 0; if (png_check_fp_number((png_const_charp)buffer, length, @@ -2537,6 +2625,9 @@ png_handle_zTXt(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) if ((png_ptr->mode & PNG_HAVE_IDAT) != 0) png_ptr->mode |= PNG_AFTER_IDAT; + /* Note, "length" is sufficient here; we won't be adding + * a null terminator later. + */ buffer = png_read_buffer(png_ptr, length, 2/*silent*/); if (buffer == NULL) @@ -2583,23 +2674,28 @@ png_handle_zTXt(png_structrp png_ptr, png_inforp info_ptr, png_uint_32 length) { png_text text; - /* It worked; png_ptr->read_buffer now looks like a tEXt chunk except - * for the extra compression type byte and the fact that it isn't - * necessarily '\0' terminated. - */ - buffer = png_ptr->read_buffer; - buffer[uncompressed_length+(keyword_length+2)] = 0; + if (png_ptr->read_buffer == NULL) + errmsg="Read failure in png_handle_zTXt"; + else + { + /* It worked; png_ptr->read_buffer now looks like a tEXt chunk + * except for the extra compression type byte and the fact that + * it isn't necessarily '\0' terminated. + */ + buffer = png_ptr->read_buffer; + buffer[uncompressed_length+(keyword_length+2)] = 0; - text.compression = PNG_TEXT_COMPRESSION_zTXt; - text.key = (png_charp)buffer; - text.text = (png_charp)(buffer + keyword_length+2); - text.text_length = uncompressed_length; - text.itxt_length = 0; - text.lang = NULL; - text.lang_key = NULL; + text.compression = PNG_TEXT_COMPRESSION_zTXt; + text.key = (png_charp)buffer; + text.text = (png_charp)(buffer + keyword_length+2); + text.text_length = uncompressed_length; + text.itxt_length = 0; + text.lang = NULL; + text.lang_key = NULL; - if (png_set_text_2(png_ptr, info_ptr, &text, 1) != 0) - errmsg = "insufficient memory"; + if (png_set_text_2(png_ptr, info_ptr, &text, 1) != 0) + errmsg = "insufficient memory"; + } } else @@ -2788,7 +2884,7 @@ png_cache_unknown_chunk(png_structrp png_ptr, png_uint_32 length) { PNG_CSTRING_FROM_CHUNK(png_ptr->unknown_chunk.name, png_ptr->chunk_name); /* The following is safe because of the PNG_SIZE_MAX init above */ - png_ptr->unknown_chunk.size = (png_size_t)length/*SAFE*/; + png_ptr->unknown_chunk.size = (size_t)length/*SAFE*/; /* 'mode' is a flag array, only the bottom four bits matter here */ png_ptr->unknown_chunk.location = (png_byte)png_ptr->mode/*SAFE*/; @@ -2975,7 +3071,7 @@ png_handle_unknown(png_structrp png_ptr, png_inforp info_ptr, case 2: png_ptr->user_chunk_cache_max = 1; png_chunk_benign_error(png_ptr, "no space in chunk cache"); - /* FALL THROUGH */ + /* FALLTHROUGH */ case 1: /* NOTE: prior to 1.6.0 this case resulted in an unknown critical * chunk being skipped, now there will be a hard error below. @@ -2984,7 +3080,7 @@ png_handle_unknown(png_structrp png_ptr, png_inforp info_ptr, default: /* not at limit */ --(png_ptr->user_chunk_cache_max); - /* FALL THROUGH */ + /* FALLTHROUGH */ case 0: /* no limit */ # endif /* USER_LIMITS */ /* Here when the limit isn't reached or when limits are compiled @@ -3035,20 +3131,61 @@ png_handle_unknown(png_structrp png_ptr, png_inforp info_ptr, */ void /* PRIVATE */ -png_check_chunk_name(png_structrp png_ptr, png_uint_32 chunk_name) +png_check_chunk_name(png_const_structrp png_ptr, png_uint_32 chunk_name) { int i; + png_uint_32 cn=chunk_name; png_debug(1, "in png_check_chunk_name"); for (i=1; i<=4; ++i) { - int c = chunk_name & 0xff; + int c = cn & 0xff; if (c < 65 || c > 122 || (c > 90 && c < 97)) png_chunk_error(png_ptr, "invalid chunk type"); - chunk_name >>= 8; + cn >>= 8; + } +} + +void /* PRIVATE */ +png_check_chunk_length(png_const_structrp png_ptr, png_uint_32 length) +{ + png_alloc_size_t limit = PNG_UINT_31_MAX; + +# ifdef PNG_SET_USER_LIMITS_SUPPORTED + if (png_ptr->user_chunk_malloc_max > 0 && + png_ptr->user_chunk_malloc_max < limit) + limit = png_ptr->user_chunk_malloc_max; +# elif PNG_USER_CHUNK_MALLOC_MAX > 0 + if (PNG_USER_CHUNK_MALLOC_MAX < limit) + limit = PNG_USER_CHUNK_MALLOC_MAX; +# endif + if (png_ptr->chunk_name == png_IDAT) + { + png_alloc_size_t idat_limit = PNG_UINT_31_MAX; + size_t row_factor = + (size_t)png_ptr->width + * (size_t)png_ptr->channels + * (png_ptr->bit_depth > 8? 2: 1) + + 1 + + (png_ptr->interlaced? 6: 0); + if (png_ptr->height > PNG_UINT_32_MAX/row_factor) + idat_limit = PNG_UINT_31_MAX; + else + idat_limit = png_ptr->height * row_factor; + row_factor = row_factor > 32566? 32566 : row_factor; + idat_limit += 6 + 5*(idat_limit/row_factor+1); /* zlib+deflate overhead */ + idat_limit=idat_limit < PNG_UINT_31_MAX? idat_limit : PNG_UINT_31_MAX; + limit = limit < idat_limit? idat_limit : limit; + } + + if (length > limit) + { + png_debug2(0," length = %lu, limit = %lu", + (unsigned long)length,(unsigned long)limit); + png_chunk_error(png_ptr, "chunk data is too large"); } } @@ -3225,7 +3362,7 @@ png_combine_row(png_const_structrp png_ptr, png_bytep dp, int display) /* Hence the pre-compiled masks indexed by PACKSWAP (or not), depth and * then pass: */ - static PNG_CONST png_uint_32 row_mask[2/*PACKSWAP*/][3/*depth*/][6] = + static const png_uint_32 row_mask[2/*PACKSWAP*/][3/*depth*/][6] = { /* Little-endian byte masks for PACKSWAP */ { S_MASKS(1,0), S_MASKS(2,0), S_MASKS(4,0) }, @@ -3236,7 +3373,7 @@ png_combine_row(png_const_structrp png_ptr, png_bytep dp, int display) /* display_mask has only three entries for the odd passes, so index by * pass>>1. */ - static PNG_CONST png_uint_32 display_mask[2][3][3] = + static const png_uint_32 display_mask[2][3][3] = { /* Little-endian byte masks for PACKSWAP */ { B_MASKS(1,0), B_MASKS(2,0), B_MASKS(4,0) }, @@ -3377,7 +3514,7 @@ png_combine_row(png_const_structrp png_ptr, png_bytep dp, int display) */ do { - dp[0] = sp[0], dp[1] = sp[1]; + dp[0] = sp[0]; dp[1] = sp[1]; if (row_width <= bytes_to_jump) return; @@ -3398,7 +3535,7 @@ png_combine_row(png_const_structrp png_ptr, png_bytep dp, int display) */ for (;;) { - dp[0] = sp[0], dp[1] = sp[1], dp[2] = sp[2]; + dp[0] = sp[0]; dp[1] = sp[1]; dp[2] = sp[2]; if (row_width <= bytes_to_jump) return; @@ -3549,7 +3686,7 @@ png_do_read_interlace(png_row_infop row_info, png_bytep row, int pass, { /* Arrays to facilitate easy interlacing - use pass (0 - 6) as index */ /* Offset to next interlace block */ - static PNG_CONST unsigned int png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; + static const unsigned int png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; png_debug(1, "in png_do_read_interlace"); if (row != NULL && row_info != NULL) @@ -3562,8 +3699,8 @@ png_do_read_interlace(png_row_infop row_info, png_bytep row, int pass, { case 1: { - png_bytep sp = row + (png_size_t)((row_info->width - 1) >> 3); - png_bytep dp = row + (png_size_t)((final_width - 1) >> 3); + png_bytep sp = row + (size_t)((row_info->width - 1) >> 3); + png_bytep dp = row + (size_t)((final_width - 1) >> 3); unsigned int sshift, dshift; unsigned int s_start, s_end; int s_inc; @@ -3689,8 +3826,8 @@ png_do_read_interlace(png_row_infop row_info, png_bytep row, int pass, case 4: { - png_bytep sp = row + (png_size_t)((row_info->width - 1) >> 1); - png_bytep dp = row + (png_size_t)((final_width - 1) >> 1); + png_bytep sp = row + (size_t)((row_info->width - 1) >> 1); + png_bytep dp = row + (size_t)((final_width - 1) >> 1); unsigned int sshift, dshift; unsigned int s_start, s_end; int s_inc; @@ -3752,12 +3889,12 @@ png_do_read_interlace(png_row_infop row_info, png_bytep row, int pass, default: { - png_size_t pixel_bytes = (row_info->pixel_depth >> 3); + size_t pixel_bytes = (row_info->pixel_depth >> 3); - png_bytep sp = row + (png_size_t)(row_info->width - 1) + png_bytep sp = row + (size_t)(row_info->width - 1) * pixel_bytes; - png_bytep dp = row + (png_size_t)(final_width - 1) * pixel_bytes; + png_bytep dp = row + (size_t)(final_width - 1) * pixel_bytes; int jstop = (int)png_pass_inc[pass]; png_uint_32 i; @@ -3794,8 +3931,8 @@ static void png_read_filter_row_sub(png_row_infop row_info, png_bytep row, png_const_bytep prev_row) { - png_size_t i; - png_size_t istop = row_info->rowbytes; + size_t i; + size_t istop = row_info->rowbytes; unsigned int bpp = (row_info->pixel_depth + 7) >> 3; png_bytep rp = row + bpp; @@ -3812,8 +3949,8 @@ static void png_read_filter_row_up(png_row_infop row_info, png_bytep row, png_const_bytep prev_row) { - png_size_t i; - png_size_t istop = row_info->rowbytes; + size_t i; + size_t istop = row_info->rowbytes; png_bytep rp = row; png_const_bytep pp = prev_row; @@ -3828,11 +3965,11 @@ static void png_read_filter_row_avg(png_row_infop row_info, png_bytep row, png_const_bytep prev_row) { - png_size_t i; + size_t i; png_bytep rp = row; png_const_bytep pp = prev_row; unsigned int bpp = (row_info->pixel_depth + 7) >> 3; - png_size_t istop = row_info->rowbytes - bpp; + size_t istop = row_info->rowbytes - bpp; for (i = 0; i < bpp; i++) { @@ -3887,7 +4024,10 @@ png_read_filter_row_paeth_1byte_pixel(png_row_infop row_info, png_bytep row, /* Find the best predictor, the least of pa, pb, pc favoring the earlier * ones in the case of a tie. */ - if (pb < pa) pa = pb, a = b; + if (pb < pa) + { + pa = pb; a = b; + } if (pc < pa) a = c; /* Calculate the current pixel in a, and move the previous row pixel to c @@ -3939,7 +4079,10 @@ png_read_filter_row_paeth_multibyte_pixel(png_row_infop row_info, png_bytep row, pc = (p + pc) < 0 ? -(p + pc) : p + pc; #endif - if (pb < pa) pa = pb, a = b; + if (pb < pa) + { + pa = pb; a = b; + } if (pc < pa) a = c; a += *row; @@ -4185,16 +4328,16 @@ png_read_finish_row(png_structrp png_ptr) /* Arrays to facilitate easy interlacing - use pass (0 - 6) as index */ /* Start of interlace block */ - static PNG_CONST png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; + static const png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; /* Offset to next interlace block */ - static PNG_CONST png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; + static const png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; /* Start of interlace block in the y direction */ - static PNG_CONST png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; + static const png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; /* Offset to next interlace block in the y direction */ - static PNG_CONST png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; + static const png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; png_debug(1, "in png_read_finish_row"); png_ptr->row_number++; @@ -4250,19 +4393,19 @@ png_read_start_row(png_structrp png_ptr) /* Arrays to facilitate easy interlacing - use pass (0 - 6) as index */ /* Start of interlace block */ - static PNG_CONST png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; + static const png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; /* Offset to next interlace block */ - static PNG_CONST png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; + static const png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; /* Start of interlace block in the y direction */ - static PNG_CONST png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; + static const png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; /* Offset to next interlace block in the y direction */ - static PNG_CONST png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; + static const png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; unsigned int max_pixel_depth; - png_size_t row_bytes; + size_t row_bytes; png_debug(1, "in png_read_start_row"); diff --git a/Externals/libpng/pngset.c b/Externals/libpng/pngset.c index 28ff3a064d..ec75dbe369 100644 --- a/Externals/libpng/pngset.c +++ b/Externals/libpng/pngset.c @@ -1,10 +1,10 @@ /* pngset.c - storage of image information into info struct * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -134,6 +134,53 @@ png_set_cHRM_XYZ(png_const_structrp png_ptr, png_inforp info_ptr, double red_X, #endif /* cHRM */ +#ifdef PNG_eXIf_SUPPORTED +void PNGAPI +png_set_eXIf(png_const_structrp png_ptr, png_inforp info_ptr, + png_bytep eXIf_buf) +{ + png_warning(png_ptr, "png_set_eXIf does not work; use png_set_eXIf_1"); + PNG_UNUSED(info_ptr) + PNG_UNUSED(eXIf_buf) +} + +void PNGAPI +png_set_eXIf_1(png_const_structrp png_ptr, png_inforp info_ptr, + png_uint_32 num_exif, png_bytep eXIf_buf) +{ + int i; + + png_debug1(1, "in %s storage function", "eXIf"); + + if (png_ptr == NULL || info_ptr == NULL) + return; + + if (info_ptr->exif) + { + png_free(png_ptr, info_ptr->exif); + info_ptr->exif = NULL; + } + + info_ptr->num_exif = num_exif; + + info_ptr->exif = png_voidcast(png_bytep, png_malloc_warn(png_ptr, + info_ptr->num_exif)); + + if (info_ptr->exif == NULL) + { + png_warning(png_ptr, "Insufficient memory for eXIf chunk data"); + return; + } + + info_ptr->free_me |= PNG_FREE_EXIF; + + for (i = 0; i < (int) info_ptr->num_exif; i++) + info_ptr->exif[i] = eXIf_buf[i]; + + info_ptr->valid |= PNG_INFO_eXIf; +} +#endif /* eXIf */ + #ifdef PNG_gAMA_SUPPORTED void PNGFAPI png_set_gAMA_fixed(png_const_structrp png_ptr, png_inforp info_ptr, @@ -266,7 +313,7 @@ png_set_pCAL(png_const_structrp png_ptr, png_inforp info_ptr, png_const_charp purpose, png_int_32 X0, png_int_32 X1, int type, int nparams, png_const_charp units, png_charpp params) { - png_size_t length; + size_t length; int i; png_debug1(1, "in %s storage function", "pCAL"); @@ -343,7 +390,7 @@ png_set_pCAL(png_const_structrp png_ptr, png_inforp info_ptr, memcpy(info_ptr->pcal_units, units, length); info_ptr->pcal_params = png_voidcast(png_charpp, png_malloc_warn(png_ptr, - (png_size_t)(((unsigned int)nparams + 1) * (sizeof (png_charp))))); + (size_t)(((unsigned int)nparams + 1) * (sizeof (png_charp))))); if (info_ptr->pcal_params == NULL) { @@ -383,7 +430,7 @@ void PNGAPI png_set_sCAL_s(png_const_structrp png_ptr, png_inforp info_ptr, int unit, png_const_charp swidth, png_const_charp sheight) { - png_size_t lengthw = 0, lengthh = 0; + size_t lengthw = 0, lengthh = 0; png_debug1(1, "in %s storage function", "sCAL"); @@ -644,7 +691,7 @@ png_set_iCCP(png_const_structrp png_ptr, png_inforp info_ptr, { png_charp new_iccp_name; png_bytep new_iccp_profile; - png_size_t length; + size_t length; png_debug1(1, "in %s storage function", "iCCP"); @@ -971,7 +1018,7 @@ png_set_tRNS(png_structrp png_ptr, png_inforp info_ptr, /* Changed from num_trans to PNG_MAX_PALETTE_LENGTH in version 1.2.1 */ info_ptr->trans_alpha = png_voidcast(png_bytep, png_malloc(png_ptr, PNG_MAX_PALETTE_LENGTH)); - memcpy(info_ptr->trans_alpha, trans_alpha, (png_size_t)num_trans); + memcpy(info_ptr->trans_alpha, trans_alpha, (size_t)num_trans); } png_ptr->trans_alpha = info_ptr->trans_alpha; } @@ -1051,7 +1098,7 @@ png_set_sPLT(png_const_structrp png_ptr, do { - png_size_t length; + size_t length; /* Skip invalid input entries */ if (entries->name == NULL || entries->entries == NULL) @@ -1102,8 +1149,9 @@ png_set_sPLT(png_const_structrp png_ptr, info_ptr->valid |= PNG_INFO_sPLT; ++(info_ptr->splt_palettes_num); ++np; + ++entries; } - while (++entries, --nentries); + while (--nentries); if (nentries > 0) png_chunk_report(png_ptr, "sPLT out of memory", PNG_CHUNK_WRITE_ERROR); @@ -1351,9 +1399,10 @@ png_set_keep_unknown_chunks(png_structrp png_ptr, int keep, /* Ignore all unknown chunks and all chunks recognized by * libpng except for IHDR, PLTE, tRNS, IDAT, and IEND */ - static PNG_CONST png_byte chunks_to_ignore[] = { + static const png_byte chunks_to_ignore[] = { 98, 75, 71, 68, '\0', /* bKGD */ 99, 72, 82, 77, '\0', /* cHRM */ + 101, 88, 73, 102, '\0', /* eXIf */ 103, 65, 77, 65, '\0', /* gAMA */ 104, 73, 83, 84, '\0', /* hIST */ 105, 67, 67, 80, '\0', /* iCCP */ @@ -1514,7 +1563,7 @@ png_set_rows(png_const_structrp png_ptr, png_inforp info_ptr, #endif void PNGAPI -png_set_compression_buffer_size(png_structrp png_ptr, png_size_t size) +png_set_compression_buffer_size(png_structrp png_ptr, size_t size) { if (png_ptr == NULL) return; @@ -1696,14 +1745,16 @@ png_check_keyword(png_structrp png_ptr, png_const_charp key, png_bytep new_key) png_byte ch = (png_byte)*key++; if ((ch > 32 && ch <= 126) || (ch >= 161 /*&& ch <= 255*/)) - *new_key++ = ch, ++key_len, space = 0; + { + *new_key++ = ch; ++key_len; space = 0; + } else if (space == 0) { /* A space or an invalid character when one wasn't seen immediately * before; output just a space. */ - *new_key++ = 32, ++key_len, space = 1; + *new_key++ = 32; ++key_len; space = 1; /* If the character was not a space then it is invalid. */ if (ch != 32) @@ -1716,7 +1767,7 @@ png_check_keyword(png_structrp png_ptr, png_const_charp key, png_bytep new_key) if (key_len > 0 && space != 0) /* trailing space */ { - --key_len, --new_key; + --key_len; --new_key; if (bad_character == 0) bad_character = 32; } diff --git a/Externals/libpng/pngstruct.h b/Externals/libpng/pngstruct.h index 749d7e35b1..8bdc7ce46d 100644 --- a/Externals/libpng/pngstruct.h +++ b/Externals/libpng/pngstruct.h @@ -1,10 +1,10 @@ /* pngstruct.h - header file for PNG reference library * - * Last changed in libpng 1.6.28 [January 5, 2017] - * Copyright (c) 1998-2002,2004,2006-2017 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -47,7 +47,7 @@ /* zlib.h declares a magic type 'uInt' that limits the amount of data that zlib * can handle at once. This type need be no larger than 16 bits (so maximum of * 65535), this define allows us to discover how big it is, but limited by the - * maximuum for png_size_t. The value can be overriden in a library build + * maximum for size_t. The value can be overridden in a library build * (pngusr.h, or set it in CPPFLAGS) and it works to set it to a considerably * lower value (e.g. 255 works). A lower value may help memory usage (slightly) * and may even improve performance on some systems (and degrade it on others.) @@ -214,7 +214,7 @@ struct png_struct_def png_uint_32 height; /* height of image in pixels */ png_uint_32 num_rows; /* number of rows in current pass */ png_uint_32 usr_width; /* width of row at start of write */ - png_size_t rowbytes; /* size of row in bytes */ + size_t rowbytes; /* size of row in bytes */ png_uint_32 iwidth; /* width of current interlaced row in pixels */ png_uint_32 row_number; /* current row in interlace pass */ png_uint_32 chunk_name; /* PNG_CHUNK() id of current chunk */ @@ -232,7 +232,7 @@ struct png_struct_def png_bytep try_row; /* buffer to save trial row when filtering */ png_bytep tst_row; /* buffer to save best trial row when filtering */ #endif - png_size_t info_rowbytes; /* Added in 1.5.4: cache of updated row bytes */ + size_t info_rowbytes; /* Added in 1.5.4: cache of updated row bytes */ png_uint_32 idat_size; /* current IDAT size for read */ png_uint_32 crc; /* current chunk CRC value */ @@ -307,7 +307,7 @@ struct png_struct_def #endif #if defined(PNG_READ_SHIFT_SUPPORTED) || defined(PNG_WRITE_SHIFT_SUPPORTED) - png_color_8 shift; /* shift for significant bit tranformation */ + png_color_8 shift; /* shift for significant bit transformation */ #endif #if defined(PNG_tRNS_SUPPORTED) || defined(PNG_READ_BACKGROUND_SUPPORTED) \ @@ -328,10 +328,10 @@ struct png_struct_def png_bytep current_buffer; /* buffer for recently used data */ png_uint_32 push_length; /* size of current input chunk */ png_uint_32 skip_length; /* bytes to skip in input data */ - png_size_t save_buffer_size; /* amount of data now in save_buffer */ - png_size_t save_buffer_max; /* total size of save_buffer */ - png_size_t buffer_size; /* total amount of available input data */ - png_size_t current_buffer_size; /* amount of data now in current_buffer */ + size_t save_buffer_size; /* amount of data now in save_buffer */ + size_t save_buffer_max; /* total size of save_buffer */ + size_t buffer_size; /* total amount of available input data */ + size_t current_buffer_size; /* amount of data now in current_buffer */ int process_mode; /* what push library is currently doing */ int cur_palette; /* current push library palette index */ @@ -392,6 +392,12 @@ struct png_struct_def /* deleted in 1.5.5: rgb_to_gray_blue_coeff; */ #endif +/* New member added in libpng-1.6.36 */ +#if defined(PNG_READ_EXPAND_SUPPORTED) && \ + defined(PNG_ARM_NEON_IMPLEMENTATION) + png_bytep riffled_palette; /* buffer for accelerated palette expansion */ +#endif + /* New member added in libpng-1.0.4 (renamed in 1.0.9) */ #if defined(PNG_MNG_FEATURES_SUPPORTED) /* Changed from png_byte to png_uint_32 at version 1.2.0 */ @@ -451,7 +457,7 @@ struct png_struct_def #endif /* New member added in libpng-1.2.26 */ - png_size_t old_big_row_buf_size; + size_t old_big_row_buf_size; #ifdef PNG_READ_SUPPORTED /* New member added in libpng-1.2.30 */ diff --git a/Externals/libpng/pngtest.c b/Externals/libpng/pngtest.c index 255fd14734..a715ae1123 100644 --- a/Externals/libpng/pngtest.c +++ b/Externals/libpng/pngtest.c @@ -1,10 +1,10 @@ /* pngtest.c - a simple test program to test libpng * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -144,6 +144,7 @@ tIME_to_str(png_structp png_ptr, png_charp ts, png_const_timep t) static int verbose = 0; static int strict = 0; static int relaxed = 0; +static int xfail = 0; static int unsupported_chunks = 0; /* chunk unsupported by libpng in input */ static int error_count = 0; /* count calls to png_error */ static int warning_count = 0; /* count calls to png_warning */ @@ -344,10 +345,10 @@ count_zero_samples(png_structp png_ptr, png_row_infop row_info, png_bytep data) #ifdef PNG_IO_STATE_SUPPORTED void -pngtest_check_io_state(png_structp png_ptr, png_size_t data_length, +pngtest_check_io_state(png_structp png_ptr, size_t data_length, png_uint_32 io_op); void -pngtest_check_io_state(png_structp png_ptr, png_size_t data_length, +pngtest_check_io_state(png_structp png_ptr, size_t data_length, png_uint_32 io_op) { png_uint_32 io_state = png_get_io_state(png_ptr); @@ -385,12 +386,12 @@ pngtest_check_io_state(png_structp png_ptr, png_size_t data_length, #endif static void PNGCBAPI -pngtest_read_data(png_structp png_ptr, png_bytep data, png_size_t length) +pngtest_read_data(png_structp png_ptr, png_bytep data, size_t length) { - png_size_t check = 0; + size_t check = 0; png_voidp io_ptr; - /* fread() returns 0 on error, so it is OK to store this in a png_size_t + /* fread() returns 0 on error, so it is OK to store this in a size_t * instead of an int, which is what fread() actually returns. */ io_ptr = png_get_io_ptr(png_ptr); @@ -424,9 +425,9 @@ pngtest_flush(png_structp png_ptr) * than changing the library. */ static void PNGCBAPI -pngtest_write_data(png_structp png_ptr, png_bytep data, png_size_t length) +pngtest_write_data(png_structp png_ptr, png_bytep data, size_t length) { - png_size_t check; + size_t check; check = fwrite(data, 1, length, (png_FILE_p)png_get_io_ptr(png_ptr)); @@ -448,13 +449,13 @@ pngtest_write_data(png_structp png_ptr, png_bytep data, png_size_t length) */ typedef struct { - PNG_CONST char *file_name; + const char *file_name; } pngtest_error_parameters; static void PNGCBAPI pngtest_warning(png_structp png_ptr, png_const_charp message) { - PNG_CONST char *name = "UNKNOWN (ERROR!)"; + const char *name = "UNKNOWN (ERROR!)"; pngtest_error_parameters *test = (pngtest_error_parameters*)png_get_error_ptr(png_ptr); @@ -463,7 +464,7 @@ pngtest_warning(png_structp png_ptr, png_const_charp message) if (test != NULL && test->file_name != NULL) name = test->file_name; - fprintf(STDERR, "%s: libpng warning: %s\n", name, message); + fprintf(STDERR, "\n%s: libpng warning: %s\n", name, message); } /* This is the default error handling function. Note that replacements for @@ -704,7 +705,7 @@ read_user_chunk_callback(png_struct *png_ptr, png_unknown_chunkp chunk) * The unknown chunk structure contains the chunk data: * png_byte name[5]; * png_byte *data; - * png_size_t size; + * size_t size; * * Note that libpng has already taken care of the CRC handling. */ @@ -849,7 +850,7 @@ pngtest_check_text_support(png_structp png_ptr, png_textp text_ptr, /* Test one file */ static int -test_one_file(PNG_CONST char *inname, PNG_CONST char *outname) +test_one_file(const char *inname, const char *outname) { static png_FILE_p fpin; static png_FILE_p fpout; /* "static" prevents setjmp corruption */ @@ -936,8 +937,12 @@ test_one_file(PNG_CONST char *inname, PNG_CONST char *outname) fprintf(STDERR, "%s -> %s: libpng read error\n", inname, outname); png_free(read_ptr, row_buf); row_buf = NULL; + if (verbose != 0) + fprintf(STDERR, " destroy read structs\n"); png_destroy_read_struct(&read_ptr, &read_info_ptr, &end_info_ptr); #ifdef PNG_WRITE_SUPPORTED + if (verbose != 0) + fprintf(STDERR, " destroy write structs\n"); png_destroy_info_struct(write_ptr, &write_end_info_ptr); png_destroy_write_struct(&write_ptr, &write_info_ptr); #endif @@ -952,11 +957,15 @@ test_one_file(PNG_CONST char *inname, PNG_CONST char *outname) if (setjmp(png_jmpbuf(write_ptr))) { fprintf(STDERR, "%s -> %s: libpng write error\n", inname, outname); + png_free(read_ptr, row_buf); + row_buf = NULL; + if (verbose != 0) + fprintf(STDERR, " destroying read structs\n"); png_destroy_read_struct(&read_ptr, &read_info_ptr, &end_info_ptr); + if (verbose != 0) + fprintf(STDERR, " destroying write structs\n"); png_destroy_info_struct(write_ptr, &write_end_info_ptr); -#ifdef PNG_WRITE_SUPPORTED png_destroy_write_struct(&write_ptr, &write_info_ptr); -#endif FCLOSE(fpin); FCLOSE(fpout); return (1); @@ -1192,6 +1201,22 @@ test_one_file(PNG_CONST char *inname, PNG_CONST char *outname) } } #endif +#ifdef PNG_READ_eXIf_SUPPORTED + { + png_bytep exif=NULL; + png_uint_32 exif_length; + + if (png_get_eXIf_1(read_ptr, read_info_ptr, &exif_length, &exif) != 0) + { + if (exif_length > 1) + fprintf(STDERR," eXIf type %c%c, %lu bytes\n",exif[0],exif[1], + (unsigned long)exif_length); +# ifdef PNG_WRITE_eXIf_SUPPORTED + png_set_eXIf_1(write_ptr, write_info_ptr, exif_length, exif); +# endif + } + } +#endif #ifdef PNG_hIST_SUPPORTED { png_uint_16p hist; @@ -1302,10 +1327,10 @@ test_one_file(PNG_CONST char *inname, PNG_CONST char *outname) { int i; - printf("\n"); + fprintf(STDERR,"\n"); for (i=0; i 1) + fprintf(STDERR," eXIf type %c%c, %lu bytes\n",exif[0],exif[1], + (unsigned long)exif_length); +# ifdef PNG_WRITE_eXIf_SUPPORTED + png_set_eXIf_1(write_ptr, write_end_info_ptr, exif_length, exif); +# endif + } + } +#endif #ifdef PNG_tIME_SUPPORTED { png_timep mod_time; @@ -1683,7 +1728,7 @@ test_one_file(PNG_CONST char *inname, PNG_CONST char *outname) for (;;) { static int wrote_question = 0; - png_size_t num_in, num_out; + size_t num_in, num_out; char inbuf[256], outbuf[256]; num_in = fread(inbuf, 1, sizeof inbuf, fpin); @@ -1766,11 +1811,11 @@ test_one_file(PNG_CONST char *inname, PNG_CONST char *outname) /* Input and output filenames */ #ifdef RISCOS -static PNG_CONST char *inname = "pngtest/png"; -static PNG_CONST char *outname = "pngout/png"; +static const char *inname = "pngtest/png"; +static const char *outname = "pngout/png"; #else -static PNG_CONST char *inname = "pngtest.png"; -static PNG_CONST char *outname = "pngout.png"; +static const char *inname = "pngtest.png"; +static const char *outname = "pngout.png"; #endif int @@ -1845,6 +1890,7 @@ main(int argc, char *argv[]) inname = argv[2]; strict++; relaxed = 0; + multiple=1; } else if (strcmp(argv[1], "--relaxed") == 0) @@ -1854,6 +1900,17 @@ main(int argc, char *argv[]) inname = argv[2]; strict = 0; relaxed++; + multiple=1; + } + else if (strcmp(argv[1], "--xfail") == 0) + { + status_dots_requested = 0; + verbose = 1; + inname = argv[2]; + strict = 0; + xfail++; + relaxed++; + multiple=1; } else @@ -1911,8 +1968,13 @@ main(int argc, char *argv[]) else { - fprintf(STDERR, " FAIL\n"); - ierror += kerror; + if (xfail) + fprintf(STDERR, " XFAIL\n"); + else + { + fprintf(STDERR, " FAIL\n"); + ierror += kerror; + } } #if defined(PNG_USER_MEM_SUPPORTED) && PNG_DEBUG if (allocation_now != current_allocation) @@ -2000,8 +2062,13 @@ main(int argc, char *argv[]) #endif } - fprintf(STDERR, " FAIL\n"); - ierror += kerror; + if (xfail) + fprintf(STDERR, " XFAIL\n"); + else + { + fprintf(STDERR, " FAIL\n"); + ierror += kerror; + } } #if defined(PNG_USER_MEM_SUPPORTED) && PNG_DEBUG if (allocation_now != current_allocation) @@ -2088,4 +2155,4 @@ main(void) #endif /* Generate a compiler error if there is an old png.h in the search path. */ -typedef png_libpng_version_1_6_29 Your_png_h_is_not_version_1_6_29; +typedef png_libpng_version_1_6_37 Your_png_h_is_not_version_1_6_37; diff --git a/Externals/libpng/pngtrans.c b/Externals/libpng/pngtrans.c index da7413fb80..1100f46ebe 100644 --- a/Externals/libpng/pngtrans.c +++ b/Externals/libpng/pngtrans.c @@ -1,10 +1,10 @@ /* pngtrans.c - transforms the data in a row (used by both readers and writers) * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -269,8 +269,8 @@ png_do_invert(png_row_infop row_info, png_bytep row) if (row_info->color_type == PNG_COLOR_TYPE_GRAY) { png_bytep rp = row; - png_size_t i; - png_size_t istop = row_info->rowbytes; + size_t i; + size_t istop = row_info->rowbytes; for (i = 0; i < istop; i++) { @@ -283,8 +283,8 @@ png_do_invert(png_row_infop row_info, png_bytep row) row_info->bit_depth == 8) { png_bytep rp = row; - png_size_t i; - png_size_t istop = row_info->rowbytes; + size_t i; + size_t istop = row_info->rowbytes; for (i = 0; i < istop; i += 2) { @@ -298,8 +298,8 @@ png_do_invert(png_row_infop row_info, png_bytep row) row_info->bit_depth == 16) { png_bytep rp = row; - png_size_t i; - png_size_t istop = row_info->rowbytes; + size_t i; + size_t istop = row_info->rowbytes; for (i = 0; i < istop; i += 4) { @@ -345,7 +345,7 @@ png_do_swap(png_row_infop row_info, png_bytep row) #endif #if defined(PNG_READ_PACKSWAP_SUPPORTED)||defined(PNG_WRITE_PACKSWAP_SUPPORTED) -static PNG_CONST png_byte onebppswaptable[256] = { +static const png_byte onebppswaptable[256] = { 0x00, 0x80, 0x40, 0xC0, 0x20, 0xA0, 0x60, 0xE0, 0x10, 0x90, 0x50, 0xD0, 0x30, 0xB0, 0x70, 0xF0, 0x08, 0x88, 0x48, 0xC8, 0x28, 0xA8, 0x68, 0xE8, @@ -380,7 +380,7 @@ static PNG_CONST png_byte onebppswaptable[256] = { 0x1F, 0x9F, 0x5F, 0xDF, 0x3F, 0xBF, 0x7F, 0xFF }; -static PNG_CONST png_byte twobppswaptable[256] = { +static const png_byte twobppswaptable[256] = { 0x00, 0x40, 0x80, 0xC0, 0x10, 0x50, 0x90, 0xD0, 0x20, 0x60, 0xA0, 0xE0, 0x30, 0x70, 0xB0, 0xF0, 0x04, 0x44, 0x84, 0xC4, 0x14, 0x54, 0x94, 0xD4, @@ -415,7 +415,7 @@ static PNG_CONST png_byte twobppswaptable[256] = { 0x2F, 0x6F, 0xAF, 0xEF, 0x3F, 0x7F, 0xBF, 0xFF }; -static PNG_CONST png_byte fourbppswaptable[256] = { +static const png_byte fourbppswaptable[256] = { 0x00, 0x10, 0x20, 0x30, 0x40, 0x50, 0x60, 0x70, 0x80, 0x90, 0xA0, 0xB0, 0xC0, 0xD0, 0xE0, 0xF0, 0x01, 0x11, 0x21, 0x31, 0x41, 0x51, 0x61, 0x71, @@ -514,11 +514,15 @@ png_do_strip_channel(png_row_infop row_info, png_bytep row, int at_start) if (at_start != 0) /* Skip initial filler */ ++sp; else /* Skip initial channel and, for sp, the filler */ - sp += 2, ++dp; + { + sp += 2; ++dp; + } /* For a 1 pixel wide image there is nothing to do */ while (sp < ep) - *dp++ = *sp, sp += 2; + { + *dp++ = *sp; sp += 2; + } row_info->pixel_depth = 8; } @@ -528,10 +532,14 @@ png_do_strip_channel(png_row_infop row_info, png_bytep row, int at_start) if (at_start != 0) /* Skip initial filler */ sp += 2; else /* Skip initial channel and, for sp, the filler */ - sp += 4, dp += 2; + { + sp += 4; dp += 2; + } while (sp < ep) - *dp++ = *sp++, *dp++ = *sp, sp += 3; + { + *dp++ = *sp++; *dp++ = *sp; sp += 3; + } row_info->pixel_depth = 16; } @@ -554,11 +562,15 @@ png_do_strip_channel(png_row_infop row_info, png_bytep row, int at_start) if (at_start != 0) /* Skip initial filler */ ++sp; else /* Skip initial channels and, for sp, the filler */ - sp += 4, dp += 3; + { + sp += 4; dp += 3; + } /* Note that the loop adds 3 to dp and 4 to sp each time. */ while (sp < ep) - *dp++ = *sp++, *dp++ = *sp++, *dp++ = *sp, sp += 2; + { + *dp++ = *sp++; *dp++ = *sp++; *dp++ = *sp; sp += 2; + } row_info->pixel_depth = 24; } @@ -568,14 +580,16 @@ png_do_strip_channel(png_row_infop row_info, png_bytep row, int at_start) if (at_start != 0) /* Skip initial filler */ sp += 2; else /* Skip initial channels and, for sp, the filler */ - sp += 8, dp += 6; + { + sp += 8; dp += 6; + } while (sp < ep) { /* Copy 6 bytes, skip 2 */ - *dp++ = *sp++, *dp++ = *sp++; - *dp++ = *sp++, *dp++ = *sp++; - *dp++ = *sp++, *dp++ = *sp, sp += 3; + *dp++ = *sp++; *dp++ = *sp++; + *dp++ = *sp++; *dp++ = *sp++; + *dp++ = *sp++; *dp++ = *sp; sp += 3; } row_info->pixel_depth = 48; @@ -595,7 +609,7 @@ png_do_strip_channel(png_row_infop row_info, png_bytep row, int at_start) return; /* The filler channel has gone already */ /* Fix the rowbytes value. */ - row_info->rowbytes = (unsigned int)(dp-row); + row_info->rowbytes = (size_t)(dp-row); } #endif @@ -694,7 +708,7 @@ png_do_check_palette_indexes(png_structrp png_ptr, png_row_infop row_info) * forms produced on either GCC or MSVC. */ int padding = PNG_PADBITS(row_info->pixel_depth, row_info->width); - png_bytep rp = png_ptr->row_buf + row_info->rowbytes; + png_bytep rp = png_ptr->row_buf + row_info->rowbytes - 1; switch (row_info->bit_depth) { diff --git a/Externals/libpng/pngwio.c b/Externals/libpng/pngwio.c index 37c7c3a7f0..10e919dd03 100644 --- a/Externals/libpng/pngwio.c +++ b/Externals/libpng/pngwio.c @@ -1,10 +1,10 @@ /* pngwio.c - functions for data output * - * Last changed in libpng 1.6.24 [August 4, 2016] - * Copyright (c) 1998-2002,2004,2006-2014,2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2014,2016,2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -30,7 +30,7 @@ */ void /* PRIVATE */ -png_write_data(png_structrp png_ptr, png_const_bytep data, png_size_t length) +png_write_data(png_structrp png_ptr, png_const_bytep data, size_t length) { /* NOTE: write_data_fn must not change the buffer! */ if (png_ptr->write_data_fn != NULL ) @@ -48,9 +48,9 @@ png_write_data(png_structrp png_ptr, png_const_bytep data, png_size_t length) * than changing the library. */ void PNGCBAPI -png_default_write_data(png_structp png_ptr, png_bytep data, png_size_t length) +png_default_write_data(png_structp png_ptr, png_bytep data, size_t length) { - png_size_t check; + size_t check; if (png_ptr == NULL) return; diff --git a/Externals/libpng/pngwrite.c b/Externals/libpng/pngwrite.c index 07088ee75e..59377a4dde 100644 --- a/Externals/libpng/pngwrite.c +++ b/Externals/libpng/pngwrite.c @@ -1,10 +1,10 @@ /* pngwrite.c - general routines to write a PNG file * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018-2019 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -237,6 +237,11 @@ png_write_info(png_structrp png_ptr, png_const_inforp info_ptr) png_write_bKGD(png_ptr, &(info_ptr->background), info_ptr->color_type); #endif +#ifdef PNG_WRITE_eXIf_SUPPORTED + if ((info_ptr->valid & PNG_INFO_eXIf) != 0) + png_write_eXIf(png_ptr, info_ptr->exif, info_ptr->num_exif); +#endif + #ifdef PNG_WRITE_hIST_SUPPORTED if ((info_ptr->valid & PNG_INFO_hIST) != 0) png_write_hIST(png_ptr, info_ptr->hist, info_ptr->num_palette); @@ -432,6 +437,12 @@ png_write_end(png_structrp png_ptr, png_inforp info_ptr) } } #endif + +#ifdef PNG_WRITE_eXIf_SUPPORTED + if ((info_ptr->valid & PNG_INFO_eXIf) != 0) + png_write_eXIf(png_ptr, info_ptr->exif, info_ptr->num_exif); +#endif + #ifdef PNG_WRITE_UNKNOWN_CHUNKS_SUPPORTED write_unknown_chunks(png_ptr, info_ptr, PNG_AFTER_IDAT); #endif @@ -458,7 +469,7 @@ png_write_end(png_structrp png_ptr, png_inforp info_ptr) #ifdef PNG_CONVERT_tIME_SUPPORTED void PNGAPI -png_convert_from_struct_tm(png_timep ptime, PNG_CONST struct tm * ttime) +png_convert_from_struct_tm(png_timep ptime, const struct tm * ttime) { png_debug(1, "in png_convert_from_struct_tm"); @@ -1007,8 +1018,8 @@ png_set_filter(png_structrp png_ptr, int method, int filters) case 5: case 6: case 7: png_app_error(png_ptr, "Unknown row filter for method 0"); - /* FALL THROUGH */ #endif /* WRITE_FILTER */ + /* FALLTHROUGH */ case PNG_FILTER_VALUE_NONE: png_ptr->do_filter = PNG_FILTER_NONE; break; @@ -1525,7 +1536,7 @@ png_write_image_16bit(png_voidp argument) display->first_row); png_uint_16p output_row = png_voidcast(png_uint_16p, display->local_row); png_uint_16p row_end; - const unsigned int channels = (image->format & PNG_FORMAT_FLAG_COLOR) != 0 ? + unsigned int channels = (image->format & PNG_FORMAT_FLAG_COLOR) != 0 ? 3 : 1; int aindex = 0; png_uint_32 y = image->height; @@ -1562,7 +1573,7 @@ png_write_image_16bit(png_voidp argument) while (out_ptr < row_end) { - const png_uint_16 alpha = in_ptr[aindex]; + png_uint_16 alpha = in_ptr[aindex]; png_uint_32 reciprocal = 0; int c; @@ -1625,7 +1636,7 @@ png_write_image_16bit(png_voidp argument) * calculation can be done to 15 bits of accuracy; however, the output needs to * be scaled in the range 0..255*65535, so include that scaling here. */ -# define UNP_RECIPROCAL(alpha) ((((0xffff*0xff)<<7)+(alpha>>1))/alpha) +# define UNP_RECIPROCAL(alpha) ((((0xffff*0xff)<<7)+((alpha)>>1))/(alpha)) static png_byte png_unpremultiply(png_uint_32 component, png_uint_32 alpha, @@ -1684,7 +1695,7 @@ png_write_image_8bit(png_voidp argument) display->first_row); png_bytep output_row = png_voidcast(png_bytep, display->local_row); png_uint_32 y = image->height; - const unsigned int channels = (image->format & PNG_FORMAT_FLAG_COLOR) != 0 ? + unsigned int channels = (image->format & PNG_FORMAT_FLAG_COLOR) != 0 ? 3 : 1; if ((image->format & PNG_FORMAT_FLAG_ALPHA) != 0) @@ -1772,25 +1783,25 @@ png_write_image_8bit(png_voidp argument) static void png_image_set_PLTE(png_image_write_control *display) { - const png_imagep image = display->image; + png_imagep image = display->image; const void *cmap = display->colormap; - const int entries = image->colormap_entries > 256 ? 256 : + int entries = image->colormap_entries > 256 ? 256 : (int)image->colormap_entries; /* NOTE: the caller must check for cmap != NULL and entries != 0 */ - const png_uint_32 format = image->format; - const unsigned int channels = PNG_IMAGE_SAMPLE_CHANNELS(format); + png_uint_32 format = image->format; + unsigned int channels = PNG_IMAGE_SAMPLE_CHANNELS(format); # if defined(PNG_FORMAT_BGR_SUPPORTED) &&\ defined(PNG_SIMPLIFIED_WRITE_AFIRST_SUPPORTED) - const int afirst = (format & PNG_FORMAT_FLAG_AFIRST) != 0 && + int afirst = (format & PNG_FORMAT_FLAG_AFIRST) != 0 && (format & PNG_FORMAT_FLAG_ALPHA) != 0; # else # define afirst 0 # endif # ifdef PNG_FORMAT_BGR_SUPPORTED - const int bgr = (format & PNG_FORMAT_FLAG_BGR) != 0 ? 2 : 0; + int bgr = (format & PNG_FORMAT_FLAG_BGR) != 0 ? 2 : 0; # else # define bgr 0 # endif @@ -1875,7 +1886,7 @@ png_image_set_PLTE(png_image_write_control *display) tRNS[i] = entry[afirst ? 0 : 3]; if (tRNS[i] < 255) num_trans = i+1; - /* FALL THROUGH */ + /* FALLTHROUGH */ case 3: palette[i].blue = entry[afirst + (2 ^ bgr)]; palette[i].green = entry[afirst + 1]; @@ -1886,7 +1897,7 @@ png_image_set_PLTE(png_image_write_control *display) tRNS[i] = entry[1 ^ afirst]; if (tRNS[i] < 255) num_trans = i+1; - /* FALL THROUGH */ + /* FALLTHROUGH */ case 1: palette[i].blue = palette[i].red = palette[i].green = entry[afirst]; @@ -1929,7 +1940,7 @@ png_image_write_main(png_voidp argument) int colormap = (format & PNG_FORMAT_FLAG_COLORMAP); int linear = !colormap && (format & PNG_FORMAT_FLAG_LINEAR); /* input */ int alpha = !colormap && (format & PNG_FORMAT_FLAG_ALPHA); - int write_16bit = linear && !colormap && (display->convert_to_8bit == 0); + int write_16bit = linear && (display->convert_to_8bit == 0); # ifdef PNG_BENIGN_ERRORS_SUPPORTED /* Make sure we error out on any bad situation */ @@ -1940,12 +1951,12 @@ png_image_write_main(png_voidp argument) * and total image size to ensure that they are within the system limits. */ { - const unsigned int channels = PNG_IMAGE_PIXEL_CHANNELS(image->format); + unsigned int channels = PNG_IMAGE_PIXEL_CHANNELS(image->format); if (image->width <= 0x7fffffffU/channels) /* no overflow */ { png_uint_32 check; - const png_uint_32 png_row_stride = image->width * channels; + png_uint_32 png_row_stride = image->width * channels; if (display->row_stride == 0) display->row_stride = (png_int_32)/*SAFE*/png_row_stride; @@ -2041,7 +2052,7 @@ png_image_write_main(png_voidp argument) */ if (write_16bit != 0) { - PNG_CONST png_uint_16 le = 0x0001; + png_uint_16 le = 0x0001; if ((*(png_const_bytep) & le) != 0) png_set_swap(png_ptr); @@ -2151,12 +2162,11 @@ png_image_write_main(png_voidp argument) static void (PNGCBAPI -image_memory_write)(png_structp png_ptr, png_bytep/*const*/ data, - png_size_t size) +image_memory_write)(png_structp png_ptr, png_bytep/*const*/ data, size_t size) { png_image_write_control *display = png_voidcast(png_image_write_control*, png_ptr->io_ptr/*backdoor: png_get_io_ptr(png_ptr)*/); - const png_alloc_size_t ob = display->output_bytes; + png_alloc_size_t ob = display->output_bytes; /* Check for overflow; this should never happen: */ if (size <= ((png_alloc_size_t)-1) - ob) diff --git a/Externals/libpng/pngwtran.c b/Externals/libpng/pngwtran.c index 377b43e5ca..49a13c1e98 100644 --- a/Externals/libpng/pngwtran.c +++ b/Externals/libpng/pngwtran.c @@ -1,10 +1,10 @@ /* pngwtran.c - transforms the data in a row for PNG writers * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2016,2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -212,9 +212,9 @@ png_do_shift(png_row_infop row_info, png_bytep row, if (row_info->bit_depth < 8) { png_bytep bp = row; - png_size_t i; + size_t i; unsigned int mask; - png_size_t row_bytes = row_info->rowbytes; + size_t row_bytes = row_info->rowbytes; if (bit_depth->gray == 1 && row_info->bit_depth == 2) mask = 0x55; @@ -254,8 +254,7 @@ png_do_shift(png_row_infop row_info, png_bytep row, for (i = 0; i < istop; i++, bp++) { - - const unsigned int c = i%channels; + unsigned int c = i%channels; int j; unsigned int v, out; @@ -283,7 +282,7 @@ png_do_shift(png_row_infop row_info, png_bytep row, for (bp = row, i = 0; i < istop; i++) { - const unsigned int c = i%channels; + unsigned int c = i%channels; int j; unsigned int value, v; @@ -514,7 +513,7 @@ png_do_write_transformations(png_structrp png_ptr, png_row_infop row_info) (png_ptr, /* png_ptr */ row_info, /* row_info: */ /* png_uint_32 width; width of row */ - /* png_size_t rowbytes; number of bytes in row */ + /* size_t rowbytes; number of bytes in row */ /* png_byte color_type; color type of pixels */ /* png_byte bit_depth; bit depth of samples */ /* png_byte channels; number of channels (1-4) */ diff --git a/Externals/libpng/pngwutil.c b/Externals/libpng/pngwutil.c index 0f98d582da..16345e4c0b 100644 --- a/Externals/libpng/pngwutil.c +++ b/Externals/libpng/pngwutil.c @@ -1,10 +1,10 @@ /* pngwutil.c - utilities to write a PNG file * - * Last changed in libpng 1.6.26 [October 20, 2016] - * Copyright (c) 1998-2002,2004,2006-2016 Glenn Randers-Pehrson - * (Version 0.96 Copyright (c) 1996, 1997 Andreas Dilger) - * (Version 0.88 Copyright (c) 1995, 1996 Guy Eric Schalnat, Group 42, Inc.) + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 1998-2002,2004,2006-2018 Glenn Randers-Pehrson + * Copyright (c) 1996-1997 Andreas Dilger + * Copyright (c) 1995-1996 Guy Eric Schalnat, Group 42, Inc. * * This code is released under the libpng license. * For conditions of distribution and use, see the disclaimer @@ -59,7 +59,7 @@ png_write_sig(png_structrp png_ptr) /* Write the rest of the 8 byte signature */ png_write_data(png_ptr, &png_signature[png_ptr->sig_bytes], - (png_size_t)(8 - png_ptr->sig_bytes)); + (size_t)(8 - png_ptr->sig_bytes)); if (png_ptr->sig_bytes < 3) png_ptr->mode |= PNG_HAVE_PNG_SIGNATURE; @@ -124,8 +124,7 @@ png_write_chunk_start(png_structrp png_ptr, png_const_bytep chunk_string, * given to png_write_chunk_header(). */ void PNGAPI -png_write_chunk_data(png_structrp png_ptr, png_const_bytep data, - png_size_t length) +png_write_chunk_data(png_structrp png_ptr, png_const_bytep data, size_t length) { /* Write the data, and run the CRC over it */ if (png_ptr == NULL) @@ -160,7 +159,7 @@ png_write_chunk_end(png_structrp png_ptr) /* Write the crc in a single operation */ png_save_uint_32(buf, png_ptr->crc); - png_write_data(png_ptr, buf, (png_size_t)4); + png_write_data(png_ptr, buf, 4); } /* Write a PNG chunk all at once. The type is an array of ASCII characters @@ -174,7 +173,7 @@ png_write_chunk_end(png_structrp png_ptr) */ static void png_write_complete_chunk(png_structrp png_ptr, png_uint_32 chunk_name, - png_const_bytep data, png_size_t length) + png_const_bytep data, size_t length) { if (png_ptr == NULL) return; @@ -191,7 +190,7 @@ png_write_complete_chunk(png_structrp png_ptr, png_uint_32 chunk_name, /* This is the API that calls the internal function above. */ void PNGAPI png_write_chunk(png_structrp png_ptr, png_const_bytep chunk_string, - png_const_bytep data, png_size_t length) + png_const_bytep data, size_t length) { png_write_complete_chunk(png_ptr, PNG_CHUNK_FROM_STRING(chunk_string), data, length); @@ -820,7 +819,7 @@ png_write_IHDR(png_structrp png_ptr, png_uint_32 width, png_uint_32 height, buf[12] = (png_byte)interlace_type; /* Write the chunk */ - png_write_complete_chunk(png_ptr, png_IHDR, buf, (png_size_t)13); + png_write_complete_chunk(png_ptr, png_IHDR, buf, 13); if ((png_ptr->do_filter) == PNG_NO_FILTERS) { @@ -889,7 +888,7 @@ png_write_PLTE(png_structrp png_ptr, png_const_colorp palette, buf[0] = pal_ptr->red; buf[1] = pal_ptr->green; buf[2] = pal_ptr->blue; - png_write_chunk_data(png_ptr, buf, (png_size_t)3); + png_write_chunk_data(png_ptr, buf, 3); } #else @@ -903,7 +902,7 @@ png_write_PLTE(png_structrp png_ptr, png_const_colorp palette, buf[0] = pal_ptr[i].red; buf[1] = pal_ptr[i].green; buf[2] = pal_ptr[i].blue; - png_write_chunk_data(png_ptr, buf, (png_size_t)3); + png_write_chunk_data(png_ptr, buf, 3); } #endif @@ -1003,7 +1002,8 @@ png_compress_IDAT(png_structrp png_ptr, png_const_bytep input, optimize_cmf(data, png_image_size(png_ptr)); #endif - png_write_complete_chunk(png_ptr, png_IDAT, data, size); + if (size > 0) + png_write_complete_chunk(png_ptr, png_IDAT, data, size); png_ptr->mode |= PNG_HAVE_IDAT; png_ptr->zstream.next_out = data; @@ -1049,7 +1049,8 @@ png_compress_IDAT(png_structrp png_ptr, png_const_bytep input, optimize_cmf(data, png_image_size(png_ptr)); #endif - png_write_complete_chunk(png_ptr, png_IDAT, data, size); + if (size > 0) + png_write_complete_chunk(png_ptr, png_IDAT, data, size); png_ptr->zstream.avail_out = 0; png_ptr->zstream.next_out = NULL; png_ptr->mode |= PNG_HAVE_IDAT | PNG_AFTER_IDAT; @@ -1073,7 +1074,7 @@ png_write_IEND(png_structrp png_ptr) { png_debug(1, "in png_write_IEND"); - png_write_complete_chunk(png_ptr, png_IEND, NULL, (png_size_t)0); + png_write_complete_chunk(png_ptr, png_IEND, NULL, 0); png_ptr->mode |= PNG_HAVE_IEND; } @@ -1088,7 +1089,7 @@ png_write_gAMA_fixed(png_structrp png_ptr, png_fixed_point file_gamma) /* file_gamma is saved in 1/100,000ths */ png_save_uint_32(buf, (png_uint_32)file_gamma); - png_write_complete_chunk(png_ptr, png_gAMA, buf, (png_size_t)4); + png_write_complete_chunk(png_ptr, png_gAMA, buf, 4); } #endif @@ -1106,7 +1107,7 @@ png_write_sRGB(png_structrp png_ptr, int srgb_intent) "Invalid sRGB rendering intent specified"); buf[0]=(png_byte)srgb_intent; - png_write_complete_chunk(png_ptr, png_sRGB, buf, (png_size_t)1); + png_write_complete_chunk(png_ptr, png_sRGB, buf, 1); } #endif @@ -1180,8 +1181,8 @@ png_write_sPLT(png_structrp png_ptr, png_const_sPLT_tp spalette) png_uint_32 name_len; png_byte new_name[80]; png_byte entrybuf[10]; - png_size_t entry_size = (spalette->depth == 8 ? 6 : 10); - png_size_t palette_size = entry_size * (png_size_t)spalette->nentries; + size_t entry_size = (spalette->depth == 8 ? 6 : 10); + size_t palette_size = entry_size * (size_t)spalette->nentries; png_sPLT_entryp ep; #ifndef PNG_POINTER_INDEXING_SUPPORTED int i; @@ -1198,10 +1199,9 @@ png_write_sPLT(png_structrp png_ptr, png_const_sPLT_tp spalette) png_write_chunk_header(png_ptr, png_sPLT, (png_uint_32)(name_len + 2 + palette_size)); - png_write_chunk_data(png_ptr, (png_bytep)new_name, - (png_size_t)(name_len + 1)); + png_write_chunk_data(png_ptr, (png_bytep)new_name, (size_t)(name_len + 1)); - png_write_chunk_data(png_ptr, &spalette->depth, (png_size_t)1); + png_write_chunk_data(png_ptr, &spalette->depth, 1); /* Loop through each palette entry, writing appropriately */ #ifdef PNG_POINTER_INDEXING_SUPPORTED @@ -1263,7 +1263,7 @@ void /* PRIVATE */ png_write_sBIT(png_structrp png_ptr, png_const_color_8p sbit, int color_type) { png_byte buf[4]; - png_size_t size; + size_t size; png_debug(1, "in png_write_sBIT"); @@ -1363,7 +1363,7 @@ png_write_tRNS(png_structrp png_ptr, png_const_bytep trans_alpha, /* Write the chunk out as it is */ png_write_complete_chunk(png_ptr, png_tRNS, trans_alpha, - (png_size_t)num_trans); + (size_t)num_trans); } else if (color_type == PNG_COLOR_TYPE_GRAY) @@ -1378,7 +1378,7 @@ png_write_tRNS(png_structrp png_ptr, png_const_bytep trans_alpha, } png_save_uint_16(buf, tran->gray); - png_write_complete_chunk(png_ptr, png_tRNS, buf, (png_size_t)2); + png_write_complete_chunk(png_ptr, png_tRNS, buf, 2); } else if (color_type == PNG_COLOR_TYPE_RGB) @@ -1398,7 +1398,7 @@ png_write_tRNS(png_structrp png_ptr, png_const_bytep trans_alpha, return; } - png_write_complete_chunk(png_ptr, png_tRNS, buf, (png_size_t)6); + png_write_complete_chunk(png_ptr, png_tRNS, buf, 6); } else @@ -1431,7 +1431,7 @@ png_write_bKGD(png_structrp png_ptr, png_const_color_16p back, int color_type) } buf[0] = back->index; - png_write_complete_chunk(png_ptr, png_bKGD, buf, (png_size_t)1); + png_write_complete_chunk(png_ptr, png_bKGD, buf, 1); } else if ((color_type & PNG_COLOR_MASK_COLOR) != 0) @@ -1452,7 +1452,7 @@ png_write_bKGD(png_structrp png_ptr, png_const_color_16p back, int color_type) return; } - png_write_complete_chunk(png_ptr, png_bKGD, buf, (png_size_t)6); + png_write_complete_chunk(png_ptr, png_bKGD, buf, 6); } else @@ -1466,11 +1466,33 @@ png_write_bKGD(png_structrp png_ptr, png_const_color_16p back, int color_type) } png_save_uint_16(buf, back->gray); - png_write_complete_chunk(png_ptr, png_bKGD, buf, (png_size_t)2); + png_write_complete_chunk(png_ptr, png_bKGD, buf, 2); } } #endif +#ifdef PNG_WRITE_eXIf_SUPPORTED +/* Write the Exif data */ +void /* PRIVATE */ +png_write_eXIf(png_structrp png_ptr, png_bytep exif, int num_exif) +{ + int i; + png_byte buf[1]; + + png_debug(1, "in png_write_eXIf"); + + png_write_chunk_header(png_ptr, png_eXIf, (png_uint_32)(num_exif)); + + for (i = 0; i < num_exif; i++) + { + buf[0] = exif[i]; + png_write_chunk_data(png_ptr, buf, 1); + } + + png_write_chunk_end(png_ptr); +} +#endif + #ifdef PNG_WRITE_hIST_SUPPORTED /* Write the histogram */ void /* PRIVATE */ @@ -1495,7 +1517,7 @@ png_write_hIST(png_structrp png_ptr, png_const_uint_16p hist, int num_hist) for (i = 0; i < num_hist; i++) { png_save_uint_16(buf, hist[i]); - png_write_chunk_data(png_ptr, buf, (png_size_t)2); + png_write_chunk_data(png_ptr, buf, 2); } png_write_chunk_end(png_ptr); @@ -1506,7 +1528,7 @@ png_write_hIST(png_structrp png_ptr, png_const_uint_16p hist, int num_hist) /* Write a tEXt chunk */ void /* PRIVATE */ png_write_tEXt(png_structrp png_ptr, png_const_charp key, png_const_charp text, - png_size_t text_len) + size_t text_len) { png_uint_32 key_len; png_byte new_key[80]; @@ -1603,7 +1625,7 @@ png_write_iTXt(png_structrp png_ptr, int compression, png_const_charp key, png_const_charp lang, png_const_charp lang_key, png_const_charp text) { png_uint_32 key_len, prefix_len; - png_size_t lang_len, lang_key_len; + size_t lang_len, lang_key_len; png_byte new_key[82]; compression_state comp; @@ -1713,7 +1735,7 @@ png_write_oFFs(png_structrp png_ptr, png_int_32 x_offset, png_int_32 y_offset, png_save_int_32(buf + 4, y_offset); buf[8] = (png_byte)unit_type; - png_write_complete_chunk(png_ptr, png_oFFs, buf, (png_size_t)9); + png_write_complete_chunk(png_ptr, png_oFFs, buf, 9); } #endif #ifdef PNG_WRITE_pCAL_SUPPORTED @@ -1724,7 +1746,7 @@ png_write_pCAL(png_structrp png_ptr, png_charp purpose, png_int_32 X0, png_charpp params) { png_uint_32 purpose_len; - png_size_t units_len, total_len; + size_t units_len, total_len; png_size_tp params_len; png_byte buf[10]; png_byte new_purpose[80]; @@ -1748,7 +1770,7 @@ png_write_pCAL(png_structrp png_ptr, png_charp purpose, png_int_32 X0, total_len = purpose_len + units_len + 10; params_len = (png_size_tp)png_malloc(png_ptr, - (png_alloc_size_t)((png_alloc_size_t)nparams * (sizeof (png_size_t)))); + (png_alloc_size_t)((png_alloc_size_t)nparams * (sizeof (size_t)))); /* Find the length of each parameter, making sure we don't count the * null terminator for the last parameter. @@ -1768,8 +1790,8 @@ png_write_pCAL(png_structrp png_ptr, png_charp purpose, png_int_32 X0, png_save_int_32(buf + 4, X1); buf[8] = (png_byte)type; buf[9] = (png_byte)nparams; - png_write_chunk_data(png_ptr, buf, (png_size_t)10); - png_write_chunk_data(png_ptr, (png_const_bytep)units, (png_size_t)units_len); + png_write_chunk_data(png_ptr, buf, 10); + png_write_chunk_data(png_ptr, (png_const_bytep)units, (size_t)units_len); for (i = 0; i < nparams; i++) { @@ -1788,7 +1810,7 @@ png_write_sCAL_s(png_structrp png_ptr, int unit, png_const_charp width, png_const_charp height) { png_byte buf[64]; - png_size_t wlen, hlen, total_len; + size_t wlen, hlen, total_len; png_debug(1, "in png_write_sCAL_s"); @@ -1829,7 +1851,7 @@ png_write_pHYs(png_structrp png_ptr, png_uint_32 x_pixels_per_unit, png_save_uint_32(buf + 4, y_pixels_per_unit); buf[8] = (png_byte)unit_type; - png_write_complete_chunk(png_ptr, png_pHYs, buf, (png_size_t)9); + png_write_complete_chunk(png_ptr, png_pHYs, buf, 9); } #endif @@ -1859,7 +1881,7 @@ png_write_tIME(png_structrp png_ptr, png_const_timep mod_time) buf[5] = mod_time->minute; buf[6] = mod_time->second; - png_write_complete_chunk(png_ptr, png_tIME, buf, (png_size_t)7); + png_write_complete_chunk(png_ptr, png_tIME, buf, 7); } #endif @@ -1871,16 +1893,16 @@ png_write_start_row(png_structrp png_ptr) /* Arrays to facilitate easy interlacing - use pass (0 - 6) as index */ /* Start of interlace block */ - static PNG_CONST png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; + static const png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; /* Offset to next interlace block */ - static PNG_CONST png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; + static const png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; /* Start of interlace block in the y direction */ - static PNG_CONST png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; + static const png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; /* Offset to next interlace block in the y direction */ - static PNG_CONST png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; + static const png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; #endif png_alloc_size_t buf_size; @@ -1986,16 +2008,16 @@ png_write_finish_row(png_structrp png_ptr) /* Arrays to facilitate easy interlacing - use pass (0 - 6) as index */ /* Start of interlace block */ - static PNG_CONST png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; + static const png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; /* Offset to next interlace block */ - static PNG_CONST png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; + static const png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; /* Start of interlace block in the y direction */ - static PNG_CONST png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; + static const png_byte png_pass_ystart[7] = {0, 0, 4, 0, 2, 0, 1}; /* Offset to next interlace block in the y direction */ - static PNG_CONST png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; + static const png_byte png_pass_yinc[7] = {8, 8, 8, 4, 4, 2, 2}; #endif png_debug(1, "in png_write_finish_row"); @@ -2049,8 +2071,8 @@ png_write_finish_row(png_structrp png_ptr) { if (png_ptr->prev_row != NULL) memset(png_ptr->prev_row, 0, - (png_size_t)(PNG_ROWBYTES(png_ptr->usr_channels* - png_ptr->usr_bit_depth, png_ptr->width)) + 1); + PNG_ROWBYTES(png_ptr->usr_channels * + png_ptr->usr_bit_depth, png_ptr->width) + 1); return; } @@ -2076,10 +2098,10 @@ png_do_write_interlace(png_row_infop row_info, png_bytep row, int pass) /* Arrays to facilitate easy interlacing - use pass (0 - 6) as index */ /* Start of interlace block */ - static PNG_CONST png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; + static const png_byte png_pass_start[7] = {0, 4, 0, 2, 0, 1, 0}; /* Offset to next interlace block */ - static PNG_CONST png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; + static const png_byte png_pass_inc[7] = {8, 8, 4, 4, 2, 2, 1}; png_debug(1, "in png_do_write_interlace"); @@ -2106,7 +2128,7 @@ png_do_write_interlace(png_row_infop row_info, png_bytep row, int pass) for (i = png_pass_start[pass]; i < row_width; i += png_pass_inc[pass]) { - sp = row + (png_size_t)(i >> 3); + sp = row + (size_t)(i >> 3); value = (int)(*sp >> (7 - (int)(i & 0x07))) & 0x01; d |= (value << shift); @@ -2144,7 +2166,7 @@ png_do_write_interlace(png_row_infop row_info, png_bytep row, int pass) for (i = png_pass_start[pass]; i < row_width; i += png_pass_inc[pass]) { - sp = row + (png_size_t)(i >> 2); + sp = row + (size_t)(i >> 2); value = (*sp >> ((3 - (int)(i & 0x03)) << 1)) & 0x03; d |= (value << shift); @@ -2180,7 +2202,7 @@ png_do_write_interlace(png_row_infop row_info, png_bytep row, int pass) for (i = png_pass_start[pass]; i < row_width; i += png_pass_inc[pass]) { - sp = row + (png_size_t)(i >> 1); + sp = row + (size_t)(i >> 1); value = (*sp >> ((1 - (int)(i & 0x01)) << 2)) & 0x0f; d |= (value << shift); @@ -2206,7 +2228,7 @@ png_do_write_interlace(png_row_infop row_info, png_bytep row, int pass) png_bytep dp; png_uint_32 i; png_uint_32 row_width = row_info->width; - png_size_t pixel_bytes; + size_t pixel_bytes; /* Start at the beginning */ dp = row; @@ -2219,7 +2241,7 @@ png_do_write_interlace(png_row_infop row_info, png_bytep row, int pass) i += png_pass_inc[pass]) { /* Find out where the original pixel is */ - sp = row + (png_size_t)i * pixel_bytes; + sp = row + (size_t)i * pixel_bytes; /* Move the pixel */ if (dp != sp) @@ -2250,16 +2272,16 @@ png_do_write_interlace(png_row_infop row_info, png_bytep row, int pass) */ static void /* PRIVATE */ png_write_filtered_row(png_structrp png_ptr, png_bytep filtered_row, - png_size_t row_bytes); + size_t row_bytes); #ifdef PNG_WRITE_FILTER_SUPPORTED -static png_size_t /* PRIVATE */ -png_setup_sub_row(png_structrp png_ptr, const png_uint_32 bpp, - const png_size_t row_bytes, const png_size_t lmins) +static size_t /* PRIVATE */ +png_setup_sub_row(png_structrp png_ptr, png_uint_32 bpp, + size_t row_bytes, size_t lmins) { png_bytep rp, dp, lp; - png_size_t i; - png_size_t sum = 0; + size_t i; + size_t sum = 0; unsigned int v; png_ptr->try_row[0] = PNG_FILTER_VALUE_SUB; @@ -2293,11 +2315,11 @@ png_setup_sub_row(png_structrp png_ptr, const png_uint_32 bpp, } static void /* PRIVATE */ -png_setup_sub_row_only(png_structrp png_ptr, const png_uint_32 bpp, - const png_size_t row_bytes) +png_setup_sub_row_only(png_structrp png_ptr, png_uint_32 bpp, + size_t row_bytes) { png_bytep rp, dp, lp; - png_size_t i; + size_t i; png_ptr->try_row[0] = PNG_FILTER_VALUE_SUB; @@ -2314,13 +2336,12 @@ png_setup_sub_row_only(png_structrp png_ptr, const png_uint_32 bpp, } } -static png_size_t /* PRIVATE */ -png_setup_up_row(png_structrp png_ptr, const png_size_t row_bytes, - const png_size_t lmins) +static size_t /* PRIVATE */ +png_setup_up_row(png_structrp png_ptr, size_t row_bytes, size_t lmins) { png_bytep rp, dp, pp; - png_size_t i; - png_size_t sum = 0; + size_t i; + size_t sum = 0; unsigned int v; png_ptr->try_row[0] = PNG_FILTER_VALUE_UP; @@ -2343,10 +2364,10 @@ png_setup_up_row(png_structrp png_ptr, const png_size_t row_bytes, return (sum); } static void /* PRIVATE */ -png_setup_up_row_only(png_structrp png_ptr, const png_size_t row_bytes) +png_setup_up_row_only(png_structrp png_ptr, size_t row_bytes) { png_bytep rp, dp, pp; - png_size_t i; + size_t i; png_ptr->try_row[0] = PNG_FILTER_VALUE_UP; @@ -2358,13 +2379,13 @@ png_setup_up_row_only(png_structrp png_ptr, const png_size_t row_bytes) } } -static png_size_t /* PRIVATE */ -png_setup_avg_row(png_structrp png_ptr, const png_uint_32 bpp, - const png_size_t row_bytes, const png_size_t lmins) +static size_t /* PRIVATE */ +png_setup_avg_row(png_structrp png_ptr, png_uint_32 bpp, + size_t row_bytes, size_t lmins) { png_bytep rp, dp, pp, lp; png_uint_32 i; - png_size_t sum = 0; + size_t sum = 0; unsigned int v; png_ptr->try_row[0] = PNG_FILTER_VALUE_AVG; @@ -2399,8 +2420,8 @@ png_setup_avg_row(png_structrp png_ptr, const png_uint_32 bpp, return (sum); } static void /* PRIVATE */ -png_setup_avg_row_only(png_structrp png_ptr, const png_uint_32 bpp, - const png_size_t row_bytes) +png_setup_avg_row_only(png_structrp png_ptr, png_uint_32 bpp, + size_t row_bytes) { png_bytep rp, dp, pp, lp; png_uint_32 i; @@ -2420,13 +2441,13 @@ png_setup_avg_row_only(png_structrp png_ptr, const png_uint_32 bpp, } } -static png_size_t /* PRIVATE */ -png_setup_paeth_row(png_structrp png_ptr, const png_uint_32 bpp, - const png_size_t row_bytes, const png_size_t lmins) +static size_t /* PRIVATE */ +png_setup_paeth_row(png_structrp png_ptr, png_uint_32 bpp, + size_t row_bytes, size_t lmins) { png_bytep rp, dp, pp, cp, lp; - png_size_t i; - png_size_t sum = 0; + size_t i; + size_t sum = 0; unsigned int v; png_ptr->try_row[0] = PNG_FILTER_VALUE_PAETH; @@ -2482,11 +2503,11 @@ png_setup_paeth_row(png_structrp png_ptr, const png_uint_32 bpp, return (sum); } static void /* PRIVATE */ -png_setup_paeth_row_only(png_structrp png_ptr, const png_uint_32 bpp, - const png_size_t row_bytes) +png_setup_paeth_row_only(png_structrp png_ptr, png_uint_32 bpp, + size_t row_bytes) { png_bytep rp, dp, pp, cp, lp; - png_size_t i; + size_t i; png_ptr->try_row[0] = PNG_FILTER_VALUE_PAETH; @@ -2535,8 +2556,8 @@ png_write_find_filter(png_structrp png_ptr, png_row_infop row_info) png_bytep row_buf; png_bytep best_row; png_uint_32 bpp; - png_size_t mins; - png_size_t row_bytes = row_info->rowbytes; + size_t mins; + size_t row_bytes = row_info->rowbytes; png_debug(1, "in png_write_find_filter"); @@ -2591,8 +2612,8 @@ png_write_find_filter(png_structrp png_ptr, png_row_infop row_info) * 'none' filter. */ png_bytep rp; - png_size_t sum = 0; - png_size_t i; + size_t sum = 0; + size_t i; unsigned int v; { @@ -2620,8 +2641,8 @@ png_write_find_filter(png_structrp png_ptr, png_row_infop row_info) else if ((filter_to_do & PNG_FILTER_SUB) != 0) { - png_size_t sum; - png_size_t lmins = mins; + size_t sum; + size_t lmins = mins; sum = png_setup_sub_row(png_ptr, bpp, row_bytes, lmins); @@ -2646,8 +2667,8 @@ png_write_find_filter(png_structrp png_ptr, png_row_infop row_info) else if ((filter_to_do & PNG_FILTER_UP) != 0) { - png_size_t sum; - png_size_t lmins = mins; + size_t sum; + size_t lmins = mins; sum = png_setup_up_row(png_ptr, row_bytes, lmins); @@ -2672,8 +2693,8 @@ png_write_find_filter(png_structrp png_ptr, png_row_infop row_info) else if ((filter_to_do & PNG_FILTER_AVG) != 0) { - png_size_t sum; - png_size_t lmins = mins; + size_t sum; + size_t lmins = mins; sum= png_setup_avg_row(png_ptr, bpp, row_bytes, lmins); @@ -2698,8 +2719,8 @@ png_write_find_filter(png_structrp png_ptr, png_row_infop row_info) else if ((filter_to_do & PNG_FILTER_PAETH) != 0) { - png_size_t sum; - png_size_t lmins = mins; + size_t sum; + size_t lmins = mins; sum = png_setup_paeth_row(png_ptr, bpp, row_bytes, lmins); @@ -2724,7 +2745,7 @@ png_write_find_filter(png_structrp png_ptr, png_row_infop row_info) /* Do the actual writing of a previously filtered row. */ static void png_write_filtered_row(png_structrp png_ptr, png_bytep filtered_row, - png_size_t full_row_length/*includes filter byte*/) + size_t full_row_length/*includes filter byte*/) { png_debug(1, "in png_write_filtered_row"); diff --git a/Externals/libpng/powerpc/filter_vsx_intrinsics.c b/Externals/libpng/powerpc/filter_vsx_intrinsics.c new file mode 100644 index 0000000000..01cf8800dc --- /dev/null +++ b/Externals/libpng/powerpc/filter_vsx_intrinsics.c @@ -0,0 +1,768 @@ +/* filter_vsx_intrinsics.c - PowerPC optimised filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2017 Glenn Randers-Pehrson + * Written by Vadim Barkov, 2017. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +#include +#include +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED + +/* This code requires -maltivec and -mvsx on the command line: */ +#if PNG_POWERPC_VSX_IMPLEMENTATION == 1 /* intrinsics code from pngpriv.h */ + +#include + +#if PNG_POWERPC_VSX_OPT > 0 + +#ifndef __VSX__ +# error "This code requires VSX support (POWER7 and later). Please provide -mvsx compiler flag." +#endif + +#define vec_ld_unaligned(vec,data) vec = vec_vsx_ld(0,data) +#define vec_st_unaligned(vec,data) vec_vsx_st(vec,0,data) + + +/* Functions in this file look at most 3 pixels (a,b,c) to predict the 4th (d). + * They're positioned like this: + * prev: c b + * row: a d + * The Sub filter predicts d=a, Avg d=(a+b)/2, and Paeth predicts d to be + * whichever of a, b, or c is closest to p=a+b-c. + * ( this is taken from ../intel/filter_sse2_intrinsics.c ) + */ + +#define vsx_declare_common_vars(row_info,row,prev_row,offset) \ + png_byte i;\ + png_bytep rp = row + offset;\ + png_const_bytep pp = prev_row;\ + size_t unaligned_top = 16 - (((size_t)rp % 16));\ + size_t istop;\ + if(unaligned_top == 16)\ + unaligned_top = 0;\ + istop = row_info->rowbytes;\ + if((unaligned_top < istop))\ + istop -= unaligned_top;\ + else{\ + unaligned_top = istop;\ + istop = 0;\ + } + +void png_read_filter_row_up_vsx(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + vector unsigned char rp_vec; + vector unsigned char pp_vec; + vsx_declare_common_vars(row_info,row,prev_row,0) + + /* Altivec operations require 16-byte aligned data + * but input can be unaligned. So we calculate + * unaligned part as usual. + */ + for (i = 0; i < unaligned_top; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*pp++)) & 0xff); + rp++; + } + + /* Using SIMD while we can */ + while( istop >= 16 ) + { + rp_vec = vec_ld(0,rp); + vec_ld_unaligned(pp_vec,pp); + + rp_vec = vec_add(rp_vec,pp_vec); + + vec_st(rp_vec,0,rp); + + pp += 16; + rp += 16; + istop -= 16; + } + + if(istop > 0) + { + /* If byte count of row is not divisible by 16 + * we will process remaining part as usual + */ + for (i = 0; i < istop; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*pp++)) & 0xff); + rp++; + } +} + +} + +static const vector unsigned char VSX_LEFTSHIFTED1_4 = {16,16,16,16, 0, 1, 2, 3,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_LEFTSHIFTED2_4 = {16,16,16,16,16,16,16,16, 4, 5, 6, 7,16,16,16,16}; +static const vector unsigned char VSX_LEFTSHIFTED3_4 = {16,16,16,16,16,16,16,16,16,16,16,16, 8, 9,10,11}; + +static const vector unsigned char VSX_LEFTSHIFTED1_3 = {16,16,16, 0, 1, 2,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_LEFTSHIFTED2_3 = {16,16,16,16,16,16, 3, 4, 5,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_LEFTSHIFTED3_3 = {16,16,16,16,16,16,16,16,16, 6, 7, 8,16,16,16,16}; +static const vector unsigned char VSX_LEFTSHIFTED4_3 = {16,16,16,16,16,16,16,16,16,16,16,16, 9,10,11,16}; + +static const vector unsigned char VSX_NOT_SHIFTED1_4 = {16,16,16,16, 4, 5, 6, 7,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_NOT_SHIFTED2_4 = {16,16,16,16,16,16,16,16, 8, 9,10,11,16,16,16,16}; +static const vector unsigned char VSX_NOT_SHIFTED3_4 = {16,16,16,16,16,16,16,16,16,16,16,16,12,13,14,15}; + +static const vector unsigned char VSX_NOT_SHIFTED1_3 = {16,16,16, 3, 4, 5,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_NOT_SHIFTED2_3 = {16,16,16,16,16,16, 6, 7, 8,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_NOT_SHIFTED3_3 = {16,16,16,16,16,16,16,16,16, 9,10,11,16,16,16,16}; +static const vector unsigned char VSX_NOT_SHIFTED4_3 = {16,16,16,16,16,16,16,16,16,16,16,16,12,13,14,16}; + +static const vector unsigned char VSX_CHAR_ZERO = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0}; +#ifdef __LITTLE_ENDIAN__ + +static const vector unsigned char VSX_CHAR_TO_SHORT1_4 = { 4,16, 5,16, 6,16, 7,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT2_4 = { 8,16, 9,16,10,16,11,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT3_4 = {12,16,13,16,14,16,15,16,16,16,16,16,16,16,16,16}; + +static const vector unsigned char VSX_SHORT_TO_CHAR1_4 = {16,16,16,16, 0, 2, 4, 6,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR2_4 = {16,16,16,16,16,16,16,16, 0, 2, 4, 6,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR3_4 = {16,16,16,16,16,16,16,16,16,16,16,16, 0, 2, 4, 6}; + +static const vector unsigned char VSX_CHAR_TO_SHORT1_3 = { 3,16, 4,16, 5,16,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT2_3 = { 6,16, 7,16, 8,16,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT3_3 = { 9,16,10,16,11,16,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT4_3 = {12,16,13,16,14,16,16,16,16,16,16,16,16,16,16,16}; + +static const vector unsigned char VSX_SHORT_TO_CHAR1_3 = {16,16,16, 0, 2, 4,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR2_3 = {16,16,16,16,16,16, 0, 2, 4,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR3_3 = {16,16,16,16,16,16,16,16,16, 0, 2, 4,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR4_3 = {16,16,16,16,16,16,16,16,16,16,16,16, 0, 2, 4,16}; + +#elif defined(__BIG_ENDIAN__) + +static const vector unsigned char VSX_CHAR_TO_SHORT1_4 = {16, 4,16, 5,16, 6,16, 7,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT2_4 = {16, 8,16, 9,16,10,16,11,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT3_4 = {16,12,16,13,16,14,16,15,16,16,16,16,16,16,16,16}; + +static const vector unsigned char VSX_SHORT_TO_CHAR1_4 = {16,16,16,16, 1, 3, 5, 7,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR2_4 = {16,16,16,16,16,16,16,16, 1, 3, 5, 7,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR3_4 = {16,16,16,16,16,16,16,16,16,16,16,16, 1, 3, 5, 7}; + +static const vector unsigned char VSX_CHAR_TO_SHORT1_3 = {16, 3,16, 4,16, 5,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT2_3 = {16, 6,16, 7,16, 8,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT3_3 = {16, 9,16,10,16,11,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_CHAR_TO_SHORT4_3 = {16,12,16,13,16,14,16,16,16,16,16,16,16,16,16,16}; + +static const vector unsigned char VSX_SHORT_TO_CHAR1_3 = {16,16,16, 1, 3, 5,16,16,16,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR2_3 = {16,16,16,16,16,16, 1, 3, 5,16,16,16,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR3_3 = {16,16,16,16,16,16,16,16,16, 1, 3, 5,16,16,16,16}; +static const vector unsigned char VSX_SHORT_TO_CHAR4_3 = {16,16,16,16,16,16,16,16,16,16,16,16, 1, 3, 5,16}; + +#endif + +#define vsx_char_to_short(vec,offset,bpp) (vector unsigned short)vec_perm((vec),VSX_CHAR_ZERO,VSX_CHAR_TO_SHORT##offset##_##bpp) +#define vsx_short_to_char(vec,offset,bpp) vec_perm(((vector unsigned char)(vec)),VSX_CHAR_ZERO,VSX_SHORT_TO_CHAR##offset##_##bpp) + +#ifdef PNG_USE_ABS +# define vsx_abs(number) abs(number) +#else +# define vsx_abs(number) (number > 0) ? (number) : -(number) +#endif + +void png_read_filter_row_sub4_vsx(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_byte bpp = 4; + + vector unsigned char rp_vec; + vector unsigned char part_vec; + + vsx_declare_common_vars(row_info,row,prev_row,bpp) + + PNG_UNUSED(pp) + + /* Altivec operations require 16-byte aligned data + * but input can be unaligned. So we calculate + * unaligned part as usual. + */ + for (i = 0; i < unaligned_top; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*(rp-bpp))) & 0xff); + rp++; + } + + /* Using SIMD while we can */ + while( istop >= 16 ) + { + for(i=0;i < bpp ; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*(rp-bpp))) & 0xff); + rp++; + } + rp -= bpp; + + rp_vec = vec_ld(0,rp); + part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED1_4); + rp_vec = vec_add(rp_vec,part_vec); + + part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED2_4); + rp_vec = vec_add(rp_vec,part_vec); + + part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED3_4); + rp_vec = vec_add(rp_vec,part_vec); + + vec_st(rp_vec,0,rp); + + rp += 16; + istop -= 16; + } + + if(istop > 0) + for (i = 0; i < istop % 16; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*(rp - bpp))) & 0xff); + rp++; + } + +} + +void png_read_filter_row_sub3_vsx(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_byte bpp = 3; + + vector unsigned char rp_vec; + vector unsigned char part_vec; + + vsx_declare_common_vars(row_info,row,prev_row,bpp) + + PNG_UNUSED(pp) + + /* Altivec operations require 16-byte aligned data + * but input can be unaligned. So we calculate + * unaligned part as usual. + */ + for (i = 0; i < unaligned_top; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*(rp-bpp))) & 0xff); + rp++; + } + + /* Using SIMD while we can */ + while( istop >= 16 ) + { + for(i=0;i < bpp ; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*(rp-bpp))) & 0xff); + rp++; + } + rp -= bpp; + + rp_vec = vec_ld(0,rp); + part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED1_3); + rp_vec = vec_add(rp_vec,part_vec); + + part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED2_3); + rp_vec = vec_add(rp_vec,part_vec); + + part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED3_3); + rp_vec = vec_add(rp_vec,part_vec); + + part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED4_3); + rp_vec = vec_add(rp_vec,part_vec); + + vec_st(rp_vec,0,rp); + rp += 15; + istop -= 16; + + /* Since 16 % bpp = 16 % 3 = 1, last element of array must + * be proceeded manually + */ + *rp = (png_byte)(((int)(*rp) + (int)(*(rp-bpp))) & 0xff); + rp++; + } + + if(istop > 0) + for (i = 0; i < istop % 16; i++) + { + *rp = (png_byte)(((int)(*rp) + (int)(*(rp-bpp))) & 0xff); + rp++; + } +} + +void png_read_filter_row_avg4_vsx(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_byte bpp = 4; + + vector unsigned char rp_vec; + vector unsigned char pp_vec; + vector unsigned char pp_part_vec; + vector unsigned char rp_part_vec; + vector unsigned char avg_vec; + + vsx_declare_common_vars(row_info,row,prev_row,bpp) + rp -= bpp; + if(istop >= bpp) + istop -= bpp; + + for (i = 0; i < bpp; i++) + { + *rp = (png_byte)(((int)(*rp) + + ((int)(*pp++) / 2 )) & 0xff); + + rp++; + } + + /* Altivec operations require 16-byte aligned data + * but input can be unaligned. So we calculate + * unaligned part as usual. + */ + for (i = 0; i < unaligned_top; i++) + { + *rp = (png_byte)(((int)(*rp) + + (int)(*pp++ + *(rp-bpp)) / 2 ) & 0xff); + + rp++; + } + + /* Using SIMD while we can */ + while( istop >= 16 ) + { + for(i=0;i < bpp ; i++) + { + *rp = (png_byte)(((int)(*rp) + + (int)(*pp++ + *(rp-bpp)) / 2 ) & 0xff); + + rp++; + } + rp -= bpp; + pp -= bpp; + + vec_ld_unaligned(pp_vec,pp); + rp_vec = vec_ld(0,rp); + + rp_part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED1_4); + pp_part_vec = vec_perm(pp_vec,VSX_CHAR_ZERO,VSX_NOT_SHIFTED1_4); + avg_vec = vec_avg(rp_part_vec,pp_part_vec); + avg_vec = vec_sub(avg_vec, vec_and(vec_xor(rp_part_vec,pp_part_vec),vec_splat_u8(1))); + rp_vec = vec_add(rp_vec,avg_vec); + + rp_part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED2_4); + pp_part_vec = vec_perm(pp_vec,VSX_CHAR_ZERO,VSX_NOT_SHIFTED2_4); + avg_vec = vec_avg(rp_part_vec,pp_part_vec); + avg_vec = vec_sub(avg_vec, vec_and(vec_xor(rp_part_vec,pp_part_vec),vec_splat_u8(1))); + rp_vec = vec_add(rp_vec,avg_vec); + + rp_part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED3_4); + pp_part_vec = vec_perm(pp_vec,VSX_CHAR_ZERO,VSX_NOT_SHIFTED3_4); + avg_vec = vec_avg(rp_part_vec,pp_part_vec); + avg_vec = vec_sub(avg_vec, vec_and(vec_xor(rp_part_vec,pp_part_vec),vec_splat_u8(1))); + rp_vec = vec_add(rp_vec,avg_vec); + + vec_st(rp_vec,0,rp); + + rp += 16; + pp += 16; + istop -= 16; + } + + if(istop > 0) + for (i = 0; i < istop % 16; i++) + { + *rp = (png_byte)(((int)(*rp) + + (int)(*pp++ + *(rp-bpp)) / 2 ) & 0xff); + + rp++; + } +} + +void png_read_filter_row_avg3_vsx(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_byte bpp = 3; + + vector unsigned char rp_vec; + vector unsigned char pp_vec; + vector unsigned char pp_part_vec; + vector unsigned char rp_part_vec; + vector unsigned char avg_vec; + + vsx_declare_common_vars(row_info,row,prev_row,bpp) + rp -= bpp; + if(istop >= bpp) + istop -= bpp; + + for (i = 0; i < bpp; i++) + { + *rp = (png_byte)(((int)(*rp) + + ((int)(*pp++) / 2 )) & 0xff); + + rp++; + } + + /* Altivec operations require 16-byte aligned data + * but input can be unaligned. So we calculate + * unaligned part as usual. + */ + for (i = 0; i < unaligned_top; i++) + { + *rp = (png_byte)(((int)(*rp) + + (int)(*pp++ + *(rp-bpp)) / 2 ) & 0xff); + + rp++; + } + + /* Using SIMD while we can */ + while( istop >= 16 ) + { + for(i=0;i < bpp ; i++) + { + *rp = (png_byte)(((int)(*rp) + + (int)(*pp++ + *(rp-bpp)) / 2 ) & 0xff); + + rp++; + } + rp -= bpp; + pp -= bpp; + + vec_ld_unaligned(pp_vec,pp); + rp_vec = vec_ld(0,rp); + + rp_part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED1_3); + pp_part_vec = vec_perm(pp_vec,VSX_CHAR_ZERO,VSX_NOT_SHIFTED1_3); + avg_vec = vec_avg(rp_part_vec,pp_part_vec); + avg_vec = vec_sub(avg_vec, vec_and(vec_xor(rp_part_vec,pp_part_vec),vec_splat_u8(1))); + rp_vec = vec_add(rp_vec,avg_vec); + + rp_part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED2_3); + pp_part_vec = vec_perm(pp_vec,VSX_CHAR_ZERO,VSX_NOT_SHIFTED2_3); + avg_vec = vec_avg(rp_part_vec,pp_part_vec); + avg_vec = vec_sub(avg_vec, vec_and(vec_xor(rp_part_vec,pp_part_vec),vec_splat_u8(1))); + rp_vec = vec_add(rp_vec,avg_vec); + + rp_part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED3_3); + pp_part_vec = vec_perm(pp_vec,VSX_CHAR_ZERO,VSX_NOT_SHIFTED3_3); + avg_vec = vec_avg(rp_part_vec,pp_part_vec); + avg_vec = vec_sub(avg_vec, vec_and(vec_xor(rp_part_vec,pp_part_vec),vec_splat_u8(1))); + rp_vec = vec_add(rp_vec,avg_vec); + + rp_part_vec = vec_perm(rp_vec,VSX_CHAR_ZERO,VSX_LEFTSHIFTED4_3); + pp_part_vec = vec_perm(pp_vec,VSX_CHAR_ZERO,VSX_NOT_SHIFTED4_3); + avg_vec = vec_avg(rp_part_vec,pp_part_vec); + avg_vec = vec_sub(avg_vec, vec_and(vec_xor(rp_part_vec,pp_part_vec),vec_splat_u8(1))); + rp_vec = vec_add(rp_vec,avg_vec); + + vec_st(rp_vec,0,rp); + + rp += 15; + pp += 15; + istop -= 16; + + /* Since 16 % bpp = 16 % 3 = 1, last element of array must + * be proceeded manually + */ + *rp = (png_byte)(((int)(*rp) + + (int)(*pp++ + *(rp-bpp)) / 2 ) & 0xff); + rp++; + } + + if(istop > 0) + for (i = 0; i < istop % 16; i++) + { + *rp = (png_byte)(((int)(*rp) + + (int)(*pp++ + *(rp-bpp)) / 2 ) & 0xff); + + rp++; + } +} + +/* Bytewise c ? t : e. */ +#define if_then_else(c,t,e) vec_sel(e,t,c) + +#define vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) {\ + c = *(pp - bpp);\ + a = *(rp - bpp);\ + b = *pp++;\ + p = b - c;\ + pc = a - c;\ + pa = vsx_abs(p);\ + pb = vsx_abs(pc);\ + pc = vsx_abs(p + pc);\ + if (pb < pa) pa = pb, a = b;\ + if (pc < pa) a = c;\ + a += *rp;\ + *rp++ = (png_byte)a;\ + } + +void png_read_filter_row_paeth4_vsx(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_byte bpp = 4; + + int a, b, c, pa, pb, pc, p; + vector unsigned char rp_vec; + vector unsigned char pp_vec; + vector unsigned short a_vec,b_vec,c_vec,nearest_vec; + vector signed short pa_vec,pb_vec,pc_vec,smallest_vec; + + vsx_declare_common_vars(row_info,row,prev_row,bpp) + rp -= bpp; + if(istop >= bpp) + istop -= bpp; + + /* Process the first pixel in the row completely (this is the same as 'up' + * because there is only one candidate predictor for the first row). + */ + for(i = 0; i < bpp ; i++) + { + *rp = (png_byte)( *rp + *pp); + rp++; + pp++; + } + + for(i = 0; i < unaligned_top ; i++) + { + vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) + } + + while( istop >= 16) + { + for(i = 0; i < bpp ; i++) + { + vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) + } + + rp -= bpp; + pp -= bpp; + rp_vec = vec_ld(0,rp); + vec_ld_unaligned(pp_vec,pp); + + a_vec = vsx_char_to_short(vec_perm(rp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED1_4),1,4); + b_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_NOT_SHIFTED1_4),1,4); + c_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED1_4),1,4); + pa_vec = (vector signed short) vec_sub(b_vec,c_vec); + pb_vec = (vector signed short) vec_sub(a_vec , c_vec); + pc_vec = vec_add(pa_vec,pb_vec); + pa_vec = vec_abs(pa_vec); + pb_vec = vec_abs(pb_vec); + pc_vec = vec_abs(pc_vec); + smallest_vec = vec_min(pc_vec, vec_min(pa_vec,pb_vec)); + nearest_vec = if_then_else( + vec_cmpeq(pa_vec,smallest_vec), + a_vec, + if_then_else( + vec_cmpeq(pb_vec,smallest_vec), + b_vec, + c_vec + ) + ); + rp_vec = vec_add(rp_vec,(vsx_short_to_char(nearest_vec,1,4))); + + a_vec = vsx_char_to_short(vec_perm(rp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED2_4),2,4); + b_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_NOT_SHIFTED2_4),2,4); + c_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED2_4),2,4); + pa_vec = (vector signed short) vec_sub(b_vec,c_vec); + pb_vec = (vector signed short) vec_sub(a_vec , c_vec); + pc_vec = vec_add(pa_vec,pb_vec); + pa_vec = vec_abs(pa_vec); + pb_vec = vec_abs(pb_vec); + pc_vec = vec_abs(pc_vec); + smallest_vec = vec_min(pc_vec, vec_min(pa_vec,pb_vec)); + nearest_vec = if_then_else( + vec_cmpeq(pa_vec,smallest_vec), + a_vec, + if_then_else( + vec_cmpeq(pb_vec,smallest_vec), + b_vec, + c_vec + ) + ); + rp_vec = vec_add(rp_vec,(vsx_short_to_char(nearest_vec,2,4))); + + a_vec = vsx_char_to_short(vec_perm(rp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED3_4),3,4); + b_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_NOT_SHIFTED3_4),3,4); + c_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED3_4),3,4); + pa_vec = (vector signed short) vec_sub(b_vec,c_vec); + pb_vec = (vector signed short) vec_sub(a_vec , c_vec); + pc_vec = vec_add(pa_vec,pb_vec); + pa_vec = vec_abs(pa_vec); + pb_vec = vec_abs(pb_vec); + pc_vec = vec_abs(pc_vec); + smallest_vec = vec_min(pc_vec, vec_min(pa_vec,pb_vec)); + nearest_vec = if_then_else( + vec_cmpeq(pa_vec,smallest_vec), + a_vec, + if_then_else( + vec_cmpeq(pb_vec,smallest_vec), + b_vec, + c_vec + ) + ); + rp_vec = vec_add(rp_vec,(vsx_short_to_char(nearest_vec,3,4))); + + vec_st(rp_vec,0,rp); + + rp += 16; + pp += 16; + istop -= 16; + } + + if(istop > 0) + for (i = 0; i < istop % 16; i++) + { + vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) + } +} + +void png_read_filter_row_paeth3_vsx(png_row_infop row_info, png_bytep row, + png_const_bytep prev_row) +{ + png_byte bpp = 3; + + int a, b, c, pa, pb, pc, p; + vector unsigned char rp_vec; + vector unsigned char pp_vec; + vector unsigned short a_vec,b_vec,c_vec,nearest_vec; + vector signed short pa_vec,pb_vec,pc_vec,smallest_vec; + + vsx_declare_common_vars(row_info,row,prev_row,bpp) + rp -= bpp; + if(istop >= bpp) + istop -= bpp; + + /* Process the first pixel in the row completely (this is the same as 'up' + * because there is only one candidate predictor for the first row). + */ + for(i = 0; i < bpp ; i++) + { + *rp = (png_byte)( *rp + *pp); + rp++; + pp++; + } + + for(i = 0; i < unaligned_top ; i++) + { + vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) + } + + while( istop >= 16) + { + for(i = 0; i < bpp ; i++) + { + vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) + } + + rp -= bpp; + pp -= bpp; + rp_vec = vec_ld(0,rp); + vec_ld_unaligned(pp_vec,pp); + + a_vec = vsx_char_to_short(vec_perm(rp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED1_3),1,3); + b_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_NOT_SHIFTED1_3),1,3); + c_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED1_3),1,3); + pa_vec = (vector signed short) vec_sub(b_vec,c_vec); + pb_vec = (vector signed short) vec_sub(a_vec , c_vec); + pc_vec = vec_add(pa_vec,pb_vec); + pa_vec = vec_abs(pa_vec); + pb_vec = vec_abs(pb_vec); + pc_vec = vec_abs(pc_vec); + smallest_vec = vec_min(pc_vec, vec_min(pa_vec,pb_vec)); + nearest_vec = if_then_else( + vec_cmpeq(pa_vec,smallest_vec), + a_vec, + if_then_else( + vec_cmpeq(pb_vec,smallest_vec), + b_vec, + c_vec + ) + ); + rp_vec = vec_add(rp_vec,(vsx_short_to_char(nearest_vec,1,3))); + + a_vec = vsx_char_to_short(vec_perm(rp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED2_3),2,3); + b_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_NOT_SHIFTED2_3),2,3); + c_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED2_3),2,3); + pa_vec = (vector signed short) vec_sub(b_vec,c_vec); + pb_vec = (vector signed short) vec_sub(a_vec , c_vec); + pc_vec = vec_add(pa_vec,pb_vec); + pa_vec = vec_abs(pa_vec); + pb_vec = vec_abs(pb_vec); + pc_vec = vec_abs(pc_vec); + smallest_vec = vec_min(pc_vec, vec_min(pa_vec,pb_vec)); + nearest_vec = if_then_else( + vec_cmpeq(pa_vec,smallest_vec), + a_vec, + if_then_else( + vec_cmpeq(pb_vec,smallest_vec), + b_vec, + c_vec + ) + ); + rp_vec = vec_add(rp_vec,(vsx_short_to_char(nearest_vec,2,3))); + + a_vec = vsx_char_to_short(vec_perm(rp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED3_3),3,3); + b_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_NOT_SHIFTED3_3),3,3); + c_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED3_3),3,3); + pa_vec = (vector signed short) vec_sub(b_vec,c_vec); + pb_vec = (vector signed short) vec_sub(a_vec , c_vec); + pc_vec = vec_add(pa_vec,pb_vec); + pa_vec = vec_abs(pa_vec); + pb_vec = vec_abs(pb_vec); + pc_vec = vec_abs(pc_vec); + smallest_vec = vec_min(pc_vec, vec_min(pa_vec,pb_vec)); + nearest_vec = if_then_else( + vec_cmpeq(pa_vec,smallest_vec), + a_vec, + if_then_else( + vec_cmpeq(pb_vec,smallest_vec), + b_vec, + c_vec + ) + ); + rp_vec = vec_add(rp_vec,(vsx_short_to_char(nearest_vec,3,3))); + + a_vec = vsx_char_to_short(vec_perm(rp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED4_3),4,3); + b_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_NOT_SHIFTED4_3),4,3); + c_vec = vsx_char_to_short(vec_perm(pp_vec , VSX_CHAR_ZERO , VSX_LEFTSHIFTED4_3),4,3); + pa_vec = (vector signed short) vec_sub(b_vec,c_vec); + pb_vec = (vector signed short) vec_sub(a_vec , c_vec); + pc_vec = vec_add(pa_vec,pb_vec); + pa_vec = vec_abs(pa_vec); + pb_vec = vec_abs(pb_vec); + pc_vec = vec_abs(pc_vec); + smallest_vec = vec_min(pc_vec, vec_min(pa_vec,pb_vec)); + nearest_vec = if_then_else( + vec_cmpeq(pa_vec,smallest_vec), + a_vec, + if_then_else( + vec_cmpeq(pb_vec,smallest_vec), + b_vec, + c_vec + ) + ); + rp_vec = vec_add(rp_vec,(vsx_short_to_char(nearest_vec,4,3))); + + vec_st(rp_vec,0,rp); + + rp += 15; + pp += 15; + istop -= 16; + + /* Since 16 % bpp = 16 % 3 = 1, last element of array must + * be proceeded manually + */ + vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) + } + + if(istop > 0) + for (i = 0; i < istop % 16; i++) + { + vsx_paeth_process(rp,pp,a,b,c,pa,pb,pc,bpp) + } +} + +#endif /* PNG_POWERPC_VSX_OPT > 0 */ +#endif /* PNG_POWERPC_VSX_IMPLEMENTATION == 1 (intrinsics) */ +#endif /* READ */ diff --git a/Externals/libpng/powerpc/powerpc_init.c b/Externals/libpng/powerpc/powerpc_init.c new file mode 100644 index 0000000000..54426c558e --- /dev/null +++ b/Externals/libpng/powerpc/powerpc_init.c @@ -0,0 +1,126 @@ + +/* powerpc_init.c - POWERPC optimised filter functions + * + * Copyright (c) 2018 Cosmin Truta + * Copyright (c) 2017 Glenn Randers-Pehrson + * Written by Vadim Barkov, 2017. + * + * This code is released under the libpng license. + * For conditions of distribution and use, see the disclaimer + * and license in png.h + */ + +/* Below, after checking __linux__, various non-C90 POSIX 1003.1 functions are + * called. + */ +#define _POSIX_SOURCE 1 + +#include +#include "../pngpriv.h" + +#ifdef PNG_READ_SUPPORTED + +#if PNG_POWERPC_VSX_OPT > 0 +#ifdef PNG_POWERPC_VSX_CHECK_SUPPORTED /* Do run-time checks */ +/* WARNING: it is strongly recommended that you do not build libpng with + * run-time checks for CPU features if at all possible. In the case of the PowerPC + * VSX instructions there is no processor-specific way of detecting the + * presence of the required support, therefore run-time detection is extremely + * OS specific. + * + * You may set the macro PNG_POWERPC_VSX_FILE to the file name of file containing + * a fragment of C source code which defines the png_have_vsx function. There + * are a number of implementations in contrib/powerpc-vsx, but the only one that + * has partial support is contrib/powerpc-vsx/linux.c - a generic Linux + * implementation which reads /proc/cpufino. + */ +#ifndef PNG_POWERPC_VSX_FILE +# ifdef __linux__ +# define PNG_POWERPC_VSX_FILE "contrib/powerpc-vsx/linux_aux.c" +# endif +#endif + +#ifdef PNG_POWERPC_VSX_FILE + +#include /* for sig_atomic_t */ +static int png_have_vsx(png_structp png_ptr); +#include PNG_POWERPC_VSX_FILE + +#else /* PNG_POWERPC_VSX_FILE */ +# error "PNG_POWERPC_VSX_FILE undefined: no support for run-time POWERPC VSX checks" +#endif /* PNG_POWERPC_VSX_FILE */ +#endif /* PNG_POWERPC_VSX_CHECK_SUPPORTED */ + +void +png_init_filter_functions_vsx(png_structp pp, unsigned int bpp) +{ + /* The switch statement is compiled in for POWERPC_VSX_API, the call to + * png_have_vsx is compiled in for POWERPC_VSX_CHECK. If both are defined + * the check is only performed if the API has not set the PowerPC option on + * or off explicitly. In this case the check controls what happens. + */ + +#ifdef PNG_POWERPC_VSX_API_SUPPORTED + switch ((pp->options >> PNG_POWERPC_VSX) & 3) + { + case PNG_OPTION_UNSET: + /* Allow the run-time check to execute if it has been enabled - + * thus both API and CHECK can be turned on. If it isn't supported + * this case will fall through to the 'default' below, which just + * returns. + */ +#endif /* PNG_POWERPC_VSX_API_SUPPORTED */ +#ifdef PNG_POWERPC_VSX_CHECK_SUPPORTED + { + static volatile sig_atomic_t no_vsx = -1; /* not checked */ + + if (no_vsx < 0) + no_vsx = !png_have_vsx(pp); + + if (no_vsx) + return; + } +#ifdef PNG_POWERPC_VSX_API_SUPPORTED + break; +#endif +#endif /* PNG_POWERPC_VSX_CHECK_SUPPORTED */ + +#ifdef PNG_POWERPC_VSX_API_SUPPORTED + default: /* OFF or INVALID */ + return; + + case PNG_OPTION_ON: + /* Option turned on */ + break; + } +#endif + + /* IMPORTANT: any new internal functions used here must be declared using + * PNG_INTERNAL_FUNCTION in ../pngpriv.h. This is required so that the + * 'prefix' option to configure works: + * + * ./configure --with-libpng-prefix=foobar_ + * + * Verify you have got this right by running the above command, doing a build + * and examining pngprefix.h; it must contain a #define for every external + * function you add. (Notice that this happens automatically for the + * initialization function.) + */ + pp->read_filter[PNG_FILTER_VALUE_UP-1] = png_read_filter_row_up_vsx; + + if (bpp == 3) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub3_vsx; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg3_vsx; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = png_read_filter_row_paeth3_vsx; + } + + else if (bpp == 4) + { + pp->read_filter[PNG_FILTER_VALUE_SUB-1] = png_read_filter_row_sub4_vsx; + pp->read_filter[PNG_FILTER_VALUE_AVG-1] = png_read_filter_row_avg4_vsx; + pp->read_filter[PNG_FILTER_VALUE_PAETH-1] = png_read_filter_row_paeth4_vsx; + } +} +#endif /* PNG_POWERPC_VSX_OPT > 0 */ +#endif /* READ */