GetBackwardReferences: fail on alloc error

previously failures in the call to
VP8LBackwardReferencesTraceBackwards() would be ignored which, though it
wouldn't result in a crash, would produce non-deterministic output

Change-Id: Id9890a60883c3270ec75e968506d46eea32b76d4
(cherry picked from commit e3cfafaf71)
(cherry picked from commit 20ef03ee35)
(cherry picked from commit 89e226a3c7)

AUTHORS

@@ -1,4 +1,5 @@
 Contributors:
+- Alan Browning (browning at google dot com)
 - Charles Munger (clm at google dot com)
 - Christian Duvivier (cduvivier at google dot com)
 - Djordje Pesut (djordje dot pesut at imgtec dot com)
@@ -9,6 +10,7 @@ Contributors:
 - Johann (johann dot koenig at duck dot com)
 - Jovan Zelincevic (jovan dot zelincevic at imgtec dot com)
 - Jyrki Alakuijala (jyrki at google dot com)
+- Konstantin Ivlev (tomskside at gmail dot com)
 - Lode Vandevenne (lode at google dot com)
 - Lou Quillio (louquillio at google dot com)
 - Mans Rullgard (mans at mansr dot com)
@@ -37,3 +39,4 @@ Contributors:
 - Vincent Rabaud (vrabaud at google dot com)
 - Vlad Tsyrklevich (vtsyrklevich at chromium dot org)
 - Yang Zhang (yang dot zhang at arm dot com)
+- Yannis Guyon (yguyon at google dot com)

Android.mk

@@ -99,7 +99,6 @@ dsp_enc_srcs := \
     src/dsp/cost_mips_dsp_r2.c \
     src/dsp/cost_sse2.c \
     src/dsp/enc.c \
-    src/dsp/enc_avx2.c \
     src/dsp/enc_mips32.c \
     src/dsp/enc_mips_dsp_r2.c \
     src/dsp/enc_msa.c \
@@ -174,7 +173,7 @@ LOCAL_SRC_FILES := \
     $(utils_dec_srcs) \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES += $(LOCAL_PATH)/src
+LOCAL_EXPORT_C_INCLUDES += $(LOCAL_PATH)/src
 
 # prefer arm over thumb mode for performance gains
 LOCAL_ARM_MODE := arm
@@ -208,7 +207,7 @@ LOCAL_SRC_FILES := \
     $(utils_enc_srcs) \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES += $(LOCAL_PATH)/src
+LOCAL_EXPORT_C_INCLUDES += $(LOCAL_PATH)/src
 
 # prefer arm over thumb mode for performance gains
 LOCAL_ARM_MODE := arm
@@ -231,7 +230,7 @@ include $(CLEAR_VARS)
 LOCAL_SRC_FILES := $(demux_srcs)
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES += $(LOCAL_PATH)/src
+LOCAL_EXPORT_C_INCLUDES += $(LOCAL_PATH)/src
 
 # prefer arm over thumb mode for performance gains
 LOCAL_ARM_MODE := arm
@@ -254,7 +253,7 @@ include $(CLEAR_VARS)
 LOCAL_SRC_FILES := $(mux_srcs)
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES += $(LOCAL_PATH)/src
+LOCAL_EXPORT_C_INCLUDES += $(LOCAL_PATH)/src
 
 # prefer arm over thumb mode for performance gains
 LOCAL_ARM_MODE := arm

CMakeLists.txt

@@ -1,79 +1,118 @@
-cmake_minimum_required(VERSION 2.8.7)
+cmake_minimum_required(VERSION 3.5)
 
-project(libwebp C)
+project(WebP C)
 
 # Options for coder / decoder executables.
 option(WEBP_ENABLE_SIMD "Enable any SIMD optimization." ON)
-option(WEBP_BUILD_CWEBP "Build the cwebp command line tool." OFF)
-option(WEBP_BUILD_DWEBP "Build the dwebp command line tool." OFF)
-option(WEBP_BUILD_GIF2WEBP "Build the gif2webp conversion tool." OFF)
-option(WEBP_BUILD_IMG2WEBP "Build the img2webp animation tool." OFF)
-option(WEBP_BUILD_WEBPINFO "Build the webpinfo command line tool." OFF)
+option(WEBP_BUILD_ANIM_UTILS "Build animation utilities." ON)
+option(WEBP_BUILD_CWEBP "Build the cwebp command line tool." ON)
+option(WEBP_BUILD_DWEBP "Build the dwebp command line tool." ON)
+option(WEBP_BUILD_GIF2WEBP "Build the gif2webp conversion tool." ON)
+option(WEBP_BUILD_IMG2WEBP "Build the img2webp animation tool." ON)
+option(WEBP_BUILD_VWEBP "Build the vwebp viewer tool." ON)
+option(WEBP_BUILD_WEBPINFO "Build the webpinfo command line tool." ON)
+option(WEBP_BUILD_WEBPMUX "Build the webpmux command line tool." ON)
+option(WEBP_BUILD_EXTRAS "Build extras." ON)
 option(WEBP_BUILD_WEBP_JS "Emscripten build of webp.js." OFF)
-option(WEBP_ENABLE_NEAR_LOSSLESS "Enable near-lossless encoding" ON)
-option(WEBP_ENABLE_SWAP_16BIT_CSP "Enable byte swap for 16 bit colorspaces." OFF)
+option(WEBP_NEAR_LOSSLESS "Enable near-lossless encoding" ON)
+option(WEBP_ENABLE_SWAP_16BIT_CSP "Enable byte swap for 16 bit colorspaces."
+       OFF)
 
 if(WEBP_BUILD_WEBP_JS)
   set(WEBP_ENABLE_SIMD OFF)
+  set(WEBP_BUILD_ANIM_UTILS OFF)
+  set(WEBP_BUILD_CWEBP OFF)
+  set(WEBP_BUILD_DWEBP OFF)
+  set(WEBP_BUILD_GIF2WEBP OFF)
+  set(WEBP_BUILD_IMG2WEBP OFF)
+  set(WEBP_BUILD_VWEBP OFF)
+  set(WEBP_BUILD_WEBPINFO OFF)
+  set(WEBP_BUILD_WEBPMUX OFF)
+  set(WEBP_BUILD_EXTRAS OFF)
 endif()
 
 set(WEBP_DEP_LIBRARIES)
 set(WEBP_DEP_INCLUDE_DIRS)
 
 if(NOT CMAKE_BUILD_TYPE)
-  set(CMAKE_BUILD_TYPE "Release" CACHE
-    "Build type: Release, Debug or RelWithDebInfo" STRING FORCE
-  )
+  set(CMAKE_BUILD_TYPE "Release"
+      CACHE "Build type: Release, Debug, MinSizeRel or RelWithDebInfo" STRING
+      FORCE)
 endif()
 
 # Include dependencies.
 include(cmake/deps.cmake)
+include(GNUInstallDirs)
 
-################################################################################
+# ##############################################################################
 # Options.
 if(WEBP_ENABLE_SWAP_16BIT_CSP)
   add_definitions(-DWEBP_SWAP_16BIT_CSP=1)
 endif()
 
-################################################################################
+set(prefix ${CMAKE_INSTALL_PREFIX})
+set(exec_prefix "\$\{prefix\}")
+set(libdir "\$\{prefix\}/lib")
+set(includedir "\$\{prefix\}/include")
+set(PTHREAD_LIBS ${CMAKE_THREAD_LIBS_INIT})
+
+# ##############################################################################
 # Android only.
 if(ANDROID)
   include_directories(${ANDROID_NDK}/sources/android/cpufeatures)
   add_library(cpufeatures STATIC
-    ${ANDROID_NDK}/sources/android/cpufeatures/cpu-features.c
-  )
+              ${ANDROID_NDK}/sources/android/cpufeatures/cpu-features.c)
   target_link_libraries(cpufeatures dl)
   set(WEBP_DEP_LIBRARIES ${WEBP_DEP_LIBRARIES} cpufeatures)
   set(WEBP_DEP_INCLUDE_DIRS ${WEBP_DEP_INCLUDE_DIRS}
-    ${ANDROID_NDK}/sources/android/cpufeatures
-  )
+      ${ANDROID_NDK}/sources/android/cpufeatures)
   add_definitions(-DHAVE_CPU_FEATURES_H=1)
   set(HAVE_CPU_FEATURES_H 1)
 else()
   set(HAVE_CPU_FEATURES_H 0)
 endif()
 
-################################################################################
-# WebP source files.
-# Read the Makefile.am to get the source files.
+function(configure_pkg_config FILE)
+  configure_file("${CMAKE_CURRENT_SOURCE_DIR}/${FILE}.in"
+                 "${CMAKE_CURRENT_BINARY_DIR}/${FILE}")
 
-# We expect the Makefiles to define the sources as defined in
-# the first regex. E.g.:
-# libimagedec_la_SOURCES  = image_dec.c image_dec.h
+  if(HAVE_MATH_LIBRARY)
+    # MSVC doesn't have libm
+    file(READ ${CMAKE_CURRENT_BINARY_DIR}/${FILE} data)
+    string(REPLACE "-lm" "" data ${data})
+    file(WRITE ${CMAKE_CURRENT_BINARY_DIR}/${FILE} ${data})
+  endif()
+
+  install(
+    FILES "${CMAKE_CURRENT_BINARY_DIR}/${FILE}"
+    DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig
+  )
+endfunction()
+
+# ##############################################################################
+# WebP source files. Read the Makefile.am to get the source files.
+
+# We expect the Makefiles to define the sources as defined in the first regex.
+# E.g.: libimagedec_la_SOURCES  = image_dec.c image_dec.h
 function(parse_Makefile_am FOLDER VAR SRC_REGEX)
   file(READ ${FOLDER}/Makefile.am MAKEFILE_AM)
-  string(REGEX MATCHALL "${SRC_REGEX}_SOURCES[ ]*\\+?=[ ]+[0-9a-z\\._ ]*"
-    FILES_PER_LINE ${MAKEFILE_AM}
-  )
+  string(REGEX MATCHALL
+               "${SRC_REGEX}_SOURCES[ ]*\\+?=[ ]+[0-9a-z\\._ ]*"
+               FILES_PER_LINE
+               ${MAKEFILE_AM})
   set(SRCS ${${VAR}})
   foreach(FILES ${FILES_PER_LINE})
     string(FIND ${FILES} "=" OFFSET)
     math(EXPR OFFSET "${OFFSET} + 2")
-    string(SUBSTRING ${FILES} ${OFFSET} -1 FILES)
+    string(SUBSTRING ${FILES}
+                     ${OFFSET}
+                     -1
+                     FILES)
     if(FILES)
-      string(REGEX MATCHALL "[0-9a-z\\._]+"
-        FILES ${FILES}
-      )
+      string(REGEX MATCHALL
+                   "[0-9a-z\\._]+"
+                   FILES
+                   ${FILES})
       foreach(FILE ${FILES})
         list(APPEND SRCS ${FOLDER}/${FILE})
       endforeach()
@@ -83,16 +122,16 @@ function(parse_Makefile_am FOLDER VAR SRC_REGEX)
 endfunction()
 
 set(WEBP_SRC_DIR ${CMAKE_CURRENT_SOURCE_DIR}/src)
-parse_Makefile_am(${WEBP_SRC_DIR}/dec "WEBP_DEC_SRCS" "")
-parse_Makefile_am(${WEBP_SRC_DIR}/demux "WEBP_DEMUX_SRCS" "")
-parse_Makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_COMMON_SRCS" "COMMON")
-parse_Makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_ENC_SRCS" "ENC")
-parse_Makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_ENC_SRCS" "dsp_[^ ]*")
-parse_Makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_DEC_SRCS" "decode_[^ ]*")
-parse_Makefile_am(${WEBP_SRC_DIR}/enc "WEBP_ENC_SRCS" "")
-parse_Makefile_am(${WEBP_SRC_DIR}/utils "WEBP_UTILS_COMMON_SRCS" "COMMON")
-parse_Makefile_am(${WEBP_SRC_DIR}/utils "WEBP_UTILS_ENC_SRCS" "ENC")
-parse_Makefile_am(${WEBP_SRC_DIR}/utils "WEBP_UTILS_DEC_SRCS" "decode_[^ ]*")
+parse_makefile_am(${WEBP_SRC_DIR}/dec "WEBP_DEC_SRCS" "")
+parse_makefile_am(${WEBP_SRC_DIR}/demux "WEBP_DEMUX_SRCS" "")
+parse_makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_COMMON_SRCS" "COMMON")
+parse_makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_ENC_SRCS" "ENC")
+parse_makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_ENC_SRCS" "dsp_[^ ]*")
+parse_makefile_am(${WEBP_SRC_DIR}/dsp "WEBP_DSP_DEC_SRCS" "decode_[^ ]*")
+parse_makefile_am(${WEBP_SRC_DIR}/enc "WEBP_ENC_SRCS" "")
+parse_makefile_am(${WEBP_SRC_DIR}/utils "WEBP_UTILS_COMMON_SRCS" "COMMON")
+parse_makefile_am(${WEBP_SRC_DIR}/utils "WEBP_UTILS_ENC_SRCS" "ENC")
+parse_makefile_am(${WEBP_SRC_DIR}/utils "WEBP_UTILS_DEC_SRCS" "decode_[^ ]*")
 
 # Remove the files specific to SIMD we don't use.
 foreach(FILE ${WEBP_SIMD_FILES_NOT_TO_INCLUDE})
@@ -100,7 +139,12 @@ foreach(FILE ${WEBP_SIMD_FILES_NOT_TO_INCLUDE})
   list(REMOVE_ITEM WEBP_DSP_DEC_SRCS ${FILE})
 endforeach()
 
-### Define the mandatory libraries.
+# Generate the config.h file.
+configure_file(${CMAKE_CURRENT_LIST_DIR}/cmake/config.h.in
+               ${CMAKE_CURRENT_BINARY_DIR}/src/webp/config.h)
+add_definitions(-DHAVE_CONFIG_H)
+
+# ##############################################################################
 # Build the webpdecoder library.
 if(MSVC)
   # avoid security warnings for e.g., fopen() used in the examples.
@@ -108,254 +152,494 @@ if(MSVC)
 else()
   add_definitions(-Wall)
 endif()
-include_directories(${CMAKE_CURRENT_SOURCE_DIR} ${WEBP_DEP_INCLUDE_DIRS})
+include_directories(${WEBP_DEP_INCLUDE_DIRS})
 add_library(webpdecode OBJECT ${WEBP_DEC_SRCS})
+target_include_directories(webpdecode
+                           PRIVATE ${CMAKE_CURRENT_BINARY_DIR}
+                                   ${CMAKE_CURRENT_SOURCE_DIR})
 add_library(webpdspdecode OBJECT ${WEBP_DSP_COMMON_SRCS} ${WEBP_DSP_DEC_SRCS})
-add_library(webputilsdecode OBJECT ${WEBP_UTILS_COMMON_SRCS}
-  ${WEBP_UTILS_DEC_SRCS})
-add_library(webpdecoder $<TARGET_OBJECTS:webpdecode>
-  $<TARGET_OBJECTS:webpdspdecode> $<TARGET_OBJECTS:webputilsdecode>)
+target_include_directories(webpdspdecode
+                           PRIVATE ${CMAKE_CURRENT_BINARY_DIR}
+                                   ${CMAKE_CURRENT_SOURCE_DIR})
+add_library(webputilsdecode
+            OBJECT
+            ${WEBP_UTILS_COMMON_SRCS}
+            ${WEBP_UTILS_DEC_SRCS})
+target_include_directories(webputilsdecode
+                           PRIVATE ${CMAKE_CURRENT_BINARY_DIR}
+                                   ${CMAKE_CURRENT_SOURCE_DIR})
+add_library(webpdecoder
+            $<TARGET_OBJECTS:webpdecode>
+            $<TARGET_OBJECTS:webpdspdecode>
+            $<TARGET_OBJECTS:webputilsdecode>)
 target_link_libraries(webpdecoder ${WEBP_DEP_LIBRARIES})
+target_include_directories(
+  webpdecoder
+  PRIVATE ${CMAKE_CURRENT_BINARY_DIR} ${CMAKE_CURRENT_SOURCE_DIR}
+  INTERFACE $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}>
+            $<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>)
+set_target_properties(
+  webpdecoder
+  PROPERTIES PUBLIC_HEADER "${CMAKE_CURRENT_SOURCE_DIR}/src/webp/decode.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/types.h")
+
+configure_pkg_config("src/libwebpdecoder.pc")
 
 # Build the webp library.
 add_library(webpencode OBJECT ${WEBP_ENC_SRCS})
-add_library(webpdsp OBJECT ${WEBP_DSP_COMMON_SRCS} ${WEBP_DSP_DEC_SRCS}
-  ${WEBP_DSP_ENC_SRCS})
-add_library(webputils OBJECT ${WEBP_UTILS_COMMON_SRCS} ${WEBP_UTILS_DEC_SRCS}
-  ${WEBP_UTILS_ENC_SRCS})
-add_library(webp $<TARGET_OBJECTS:webpdecode> $<TARGET_OBJECTS:webpdsp>
-  $<TARGET_OBJECTS:webpencode> $<TARGET_OBJECTS:webputils>)
+target_include_directories(webpencode
+                           PRIVATE ${CMAKE_CURRENT_BINARY_DIR}
+                                   ${CMAKE_CURRENT_SOURCE_DIR})
+add_library(webpdsp
+            OBJECT
+            ${WEBP_DSP_COMMON_SRCS}
+            ${WEBP_DSP_DEC_SRCS}
+            ${WEBP_DSP_ENC_SRCS})
+target_include_directories(webpdsp
+                           PRIVATE ${CMAKE_CURRENT_BINARY_DIR}
+                                   ${CMAKE_CURRENT_SOURCE_DIR})
+add_library(webputils
+            OBJECT
+            ${WEBP_UTILS_COMMON_SRCS}
+            ${WEBP_UTILS_DEC_SRCS}
+            ${WEBP_UTILS_ENC_SRCS})
+target_include_directories(webputils
+                           PRIVATE ${CMAKE_CURRENT_BINARY_DIR}
+                                   ${CMAKE_CURRENT_SOURCE_DIR})
+add_library(webp
+            $<TARGET_OBJECTS:webpdecode>
+            $<TARGET_OBJECTS:webpdsp>
+            $<TARGET_OBJECTS:webpencode>
+            $<TARGET_OBJECTS:webputils>)
 target_link_libraries(webp ${WEBP_DEP_LIBRARIES})
+target_include_directories(webp
+                           PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}
+                                   ${CMAKE_CURRENT_BINARY_DIR}
+                           PUBLIC $<INSTALL_INTERFACE:include>)
+set_target_properties(
+  webp
+  PROPERTIES PUBLIC_HEADER "${CMAKE_CURRENT_SOURCE_DIR}/src/webp/decode.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/encode.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/types.h")
 
-# Make sure the OBJECT libraries are built with position independent code
-# (it is not ON by default).
-set_target_properties(webpdecode webpdspdecode webputilsdecode
-  webpencode webpdsp webputils PROPERTIES POSITION_INDEPENDENT_CODE ON)
+# Make sure the OBJECT libraries are built with position independent code (it is
+# not ON by default).
+set_target_properties(webpdecode
+                      webpdspdecode
+                      webputilsdecode
+                      webpencode
+                      webpdsp
+                      webputils
+                      PROPERTIES POSITION_INDEPENDENT_CODE ON)
+configure_pkg_config("src/libwebp.pc")
 
 # Build the webp demux library.
 add_library(webpdemux ${WEBP_DEMUX_SRCS})
 target_link_libraries(webpdemux webp)
+target_include_directories(webpdemux
+                           PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}
+                                   ${CMAKE_CURRENT_BINARY_DIR}
+                           PUBLIC $<INSTALL_INTERFACE:include>)
+set_target_properties(
+  webpdemux
+  PROPERTIES PUBLIC_HEADER "${CMAKE_CURRENT_SOURCE_DIR}/src/webp/decode.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/demux.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/mux_types.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/types.h")
+
+configure_pkg_config("src/demux/libwebpdemux.pc")
 
 # Set the version numbers.
 function(parse_version FILE NAME VAR)
   file(READ ${CMAKE_CURRENT_SOURCE_DIR}/src/${FILE} SOURCE_FILE)
-  string(REGEX MATCH "${NAME}_la_LDFLAGS[^\n]* -version-info [0-9:]+" TMP
-    ${SOURCE_FILE})
-  string(REGEX MATCH "[0-9:]+" TMP ${TMP})
-  string(REGEX REPLACE ":" "." VERSION ${TMP})
+  string(REGEX MATCH
+               "${NAME}_la_LDFLAGS[^\n]* -version-info [0-9:]+"
+               TMP
+               ${SOURCE_FILE})
+  string(REGEX MATCH
+               "[0-9:]+"
+               TMP
+               ${TMP})
+  string(REGEX
+         REPLACE ":"
+                 "."
+                 VERSION
+                 ${TMP})
   set(${VAR} "${VERSION}" PARENT_SCOPE)
 endfunction()
 parse_version(Makefile.am webp WEBP_WEBP_SOVERSION)
-set_target_properties(webp PROPERTIES VERSION ${PACKAGE_VERSION}
-  SOVERSION ${WEBP_WEBP_SOVERSION})
+set_target_properties(webp
+                      PROPERTIES VERSION
+                                 ${PACKAGE_VERSION}
+                                 SOVERSION
+                                 ${WEBP_WEBP_SOVERSION})
 parse_version(Makefile.am webpdecoder WEBP_DECODER_SOVERSION)
-set_target_properties(webpdecoder PROPERTIES VERSION ${PACKAGE_VERSION}
-  SOVERSION ${WEBP_DECODER_SOVERSION})
+set_target_properties(webpdecoder
+                      PROPERTIES VERSION
+                                 ${PACKAGE_VERSION}
+                                 SOVERSION
+                                 ${WEBP_DECODER_SOVERSION})
 parse_version(demux/Makefile.am webpdemux WEBP_DEMUX_SOVERSION)
-set_target_properties(webpdemux PROPERTIES VERSION ${PACKAGE_VERSION}
-  SOVERSION ${WEBP_DEMUX_SOVERSION})
+set_target_properties(webpdemux
+                      PROPERTIES VERSION
+                                 ${PACKAGE_VERSION}
+                                 SOVERSION
+                                 ${WEBP_DEMUX_SOVERSION})
+file(READ ${CMAKE_CURRENT_SOURCE_DIR}/configure.ac CONFIGURE_FILE)
+string(REGEX MATCH
+             "AC_INIT\\([^\n]*\\[[0-9\\.]+\\]"
+             TMP
+             ${CONFIGURE_FILE})
+string(REGEX MATCH
+             "[0-9\\.]+"
+             PROJECT_VERSION
+             ${TMP})
 
 # Define the libraries to install.
 set(INSTALLED_LIBRARIES webpdecoder webp webpdemux)
 
-### Deal with SIMD.
-# Change the compile flags for SIMD files we use.
+# Deal with SIMD. Change the compile flags for SIMD files we use.
 list(LENGTH WEBP_SIMD_FILES_TO_INCLUDE WEBP_SIMD_FILES_TO_INCLUDE_LENGTH)
 math(EXPR WEBP_SIMD_FILES_TO_INCLUDE_RANGE
-  "${WEBP_SIMD_FILES_TO_INCLUDE_LENGTH}-1"
-)
+     "${WEBP_SIMD_FILES_TO_INCLUDE_LENGTH}-1")
 
 foreach(I_FILE RANGE ${WEBP_SIMD_FILES_TO_INCLUDE_RANGE})
   list(GET WEBP_SIMD_FILES_TO_INCLUDE ${I_FILE} FILE)
   list(GET WEBP_SIMD_FLAGS_TO_INCLUDE ${I_FILE} SIMD_COMPILE_FLAG)
-  set_source_files_properties(${FILE} PROPERTIES
-    COMPILE_FLAGS ${SIMD_COMPILE_FLAG}
-  )
+  set_source_files_properties(${FILE}
+                              PROPERTIES
+                              COMPILE_FLAGS
+                              ${SIMD_COMPILE_FLAG})
 endforeach()
 
-# Build the executables if asked for.
-if(WEBP_BUILD_CWEBP OR WEBP_BUILD_DWEBP OR
-   WEBP_BUILD_GIF2WEBP OR WEBP_BUILD_IMG2WEBP OR WEBP_BUILD_WEBP_JS)
-  # Example utility library.
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "EXAMPLEUTIL_SRCS"
-    "example_util_[^ ]*")
-  list(APPEND EXAMPLEUTIL_SRCS
-    ${CMAKE_CURRENT_SOURCE_DIR}/examples/stopwatch.h)
-  add_library(exampleutil ${EXAMPLEUTIL_SRCS})
+if(WEBP_BUILD_GIF2WEBP AND NOT GIF_FOUND)
+  set(WEBP_BUILD_GIF2WEBP OFF)
+endif()
 
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEIOUTILS_SRCS"
-    "imageio_util_[^ ]*")
+if(WEBP_BUILD_ANIM_UTILS AND NOT GIF_FOUND)
+  set(WEBP_BUILD_ANIM_UTILS OFF)
+endif()
+
+# Build the executables if asked for.
+if(WEBP_BUILD_ANIM_UTILS
+   OR WEBP_BUILD_CWEBP
+   OR WEBP_BUILD_DWEBP
+   OR WEBP_BUILD_GIF2WEBP
+   OR WEBP_BUILD_IMG2WEBP
+   OR WEBP_BUILD_VWEBP)
+  # Example utility library.
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "EXAMPLEUTIL_SRCS"
+                    "example_util_[^ ]*")
+  list(APPEND EXAMPLEUTIL_SRCS ${CMAKE_CURRENT_SOURCE_DIR}/examples/stopwatch.h)
+  add_library(exampleutil ${EXAMPLEUTIL_SRCS})
+  target_include_directories(
+    exampleutil
+    PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src>)
+
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEIOUTILS_SRCS"
+                    "imageio_util_[^ ]*")
   add_library(imageioutil ${IMAGEIOUTILS_SRCS})
   target_link_libraries(imageioutil webp)
 
   # Image-decoding utility library.
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEDEC_SRCS"
-    "imagedec_[^ ]*")
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEDEC_SRCS"
+                    "imagedec_[^ ]*")
   add_library(imagedec ${IMAGEDEC_SRCS})
-  target_link_libraries(imagedec imageioutil webpdemux webp
-    ${WEBP_DEP_IMG_LIBRARIES})
+  target_link_libraries(imagedec
+                        imageioutil
+                        webpdemux
+                        webp
+                        ${WEBP_DEP_IMG_LIBRARIES})
 
   # Image-encoding utility library.
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEENC_SRCS"
-    "imageenc_[^ ]*")
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEENC_SRCS"
+                    "imageenc_[^ ]*")
   add_library(imageenc ${IMAGEENC_SRCS})
   target_link_libraries(imageenc webp)
 
-  set_property(TARGET exampleutil imageioutil imagedec imageenc
-    PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
+  set_property(TARGET exampleutil
+                      imageioutil
+                      imagedec
+                      imageenc
+               PROPERTY INCLUDE_DIRECTORIES ${CMAKE_CURRENT_SOURCE_DIR}/src
+                        ${CMAKE_CURRENT_BINARY_DIR}/src)
 endif()
 
 if(WEBP_BUILD_DWEBP)
   # dwebp
-  include_directories(${WEBP_DEP_IMG_INCLUDE_DIRS})
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "DWEBP_SRCS"
-    "dwebp")
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "DWEBP_SRCS" "dwebp")
   add_executable(dwebp ${DWEBP_SRCS})
-  target_link_libraries(dwebp exampleutil imagedec imageenc webpdecoder)
-  install(TARGETS dwebp RUNTIME DESTINATION bin)
-  set_property(TARGET dwebp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
+  target_link_libraries(dwebp exampleutil imagedec imageenc)
+  target_include_directories(dwebp PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+  install(TARGETS dwebp RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
 endif()
 
 if(WEBP_BUILD_CWEBP)
   # cwebp
-  include_directories(${WEBP_DEP_IMG_INCLUDE_DIRS})
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "CWEBP_SRCS"
-    "cwebp")
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "CWEBP_SRCS" "cwebp")
   add_executable(cwebp ${CWEBP_SRCS})
   target_link_libraries(cwebp exampleutil imagedec webp)
-  install(TARGETS cwebp RUNTIME DESTINATION bin)
-  set_property(TARGET cwebp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
-endif()
-
-if(WEBP_BUILD_GIF2WEBP AND NOT GIF_FOUND)
-  unset(WEBP_BUILD_GIF2WEBP CACHE)
+  target_include_directories(cwebp PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+  install(TARGETS cwebp RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
 endif()
 
 if(WEBP_BUILD_GIF2WEBP OR WEBP_BUILD_IMG2WEBP)
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/src/mux "WEBP_MUX_SRCS"
-    "")
-  add_library(webpmux ${WEBP_MUX_SRCS})
-  target_link_libraries(webpmux webp)
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/src/mux "WEBP_MUX_SRCS" "")
+  add_library(libwebpmux ${WEBP_MUX_SRCS})
+  target_link_libraries(libwebpmux webp)
+  target_include_directories(libwebpmux
+                             PRIVATE ${CMAKE_CURRENT_BINARY_DIR}
+                                     ${CMAKE_CURRENT_SOURCE_DIR})
   parse_version(mux/Makefile.am webpmux WEBP_MUX_SOVERSION)
-  set_target_properties(webpmux PROPERTIES VERSION ${PACKAGE_VERSION}
-    SOVERSION ${WEBP_MUX_SOVERSION})
-  list(APPEND INSTALLED_LIBRARIES webpmux)
+  set_target_properties(libwebpmux
+                        PROPERTIES VERSION
+                                   ${PACKAGE_VERSION}
+                                   SOVERSION
+                                   ${WEBP_MUX_SOVERSION})
+  set_target_properties(libwebpmux
+                        PROPERTIES PUBLIC_HEADER
+                                   "${CMAKE_CURRENT_SOURCE_DIR}/src/webp/mux.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/mux_types.h;\
+${CMAKE_CURRENT_SOURCE_DIR}/src/webp/types.h;")
+  set_target_properties(libwebpmux PROPERTIES OUTPUT_NAME webpmux)
+  list(APPEND INSTALLED_LIBRARIES libwebpmux)
+  configure_pkg_config("src/mux/libwebpmux.pc")
 endif()
 
 if(WEBP_BUILD_GIF2WEBP)
   # gif2webp
   include_directories(${WEBP_DEP_GIF_INCLUDE_DIRS})
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "GIF2WEBP_SRCS"
-    "gif2webp")
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "GIF2WEBP_SRCS"
+                    "gif2webp")
   add_executable(gif2webp ${GIF2WEBP_SRCS})
-  target_link_libraries(gif2webp exampleutil imageioutil webp webpmux
-    ${WEBP_DEP_GIF_LIBRARIES})
-  install(TARGETS gif2webp RUNTIME DESTINATION bin)
-  set_property(TARGET gif2webp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
+  target_link_libraries(gif2webp
+                        exampleutil
+                        imageioutil
+                        webp
+                        libwebpmux
+                        ${WEBP_DEP_GIF_LIBRARIES})
+  target_include_directories(gif2webp PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+  install(TARGETS gif2webp RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
 endif()
 
 if(WEBP_BUILD_IMG2WEBP)
   # img2webp
   include_directories(${WEBP_DEP_IMG_INCLUDE_DIRS})
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "IMG2WEBP_SRCS"
-    "img2webp")
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "IMG2WEBP_SRCS"
+                    "img2webp")
   add_executable(img2webp ${IMG2WEBP_SRCS})
-  target_link_libraries(img2webp exampleutil imagedec imageioutil webp webpmux)
-  install(TARGETS img2webp RUNTIME DESTINATION bin)
-  set_property(TARGET img2webp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
+  target_link_libraries(img2webp
+                        exampleutil
+                        imagedec
+                        imageioutil
+                        webp
+                        libwebpmux)
+  target_include_directories(img2webp PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+  install(TARGETS img2webp RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
 endif()
 
-if (WEBP_BUILD_WEBPINFO)
+if(WEBP_BUILD_VWEBP)
+  # vwebp
+  find_package(GLUT)
+  if(GLUT_FOUND)
+    include_directories(${WEBP_DEP_IMG_INCLUDE_DIRS})
+    parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "VWEBP_SRCS" "vwebp")
+    add_executable(vwebp ${VWEBP_SRCS})
+    target_link_libraries(vwebp
+                          ${OPENGL_LIBRARIES}
+                          exampleutil
+                          GLUT::GLUT
+                          imageioutil
+                          webp
+                          webpdemux)
+    target_include_directories(vwebp
+                               PRIVATE GLUT::GLUT
+                                       ${CMAKE_CURRENT_BINARY_DIR}/src
+                                       ${OPENGL_INCLUDE_DIR})
+    install(TARGETS vwebp RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+    if(${CMAKE_SYSTEM_NAME} STREQUAL "Darwin")
+      check_c_compiler_flag("-Wno-deprecated-declarations" HAS_NO_DEPRECATED)
+      if(HAS_NO_DEPRECATED)
+        target_compile_options(vwebp PRIVATE "-Wno-deprecated-declarations")
+      endif()
+    endif()
+  endif()
+endif()
+
+if(WEBP_BUILD_WEBPINFO)
   # webpinfo
   include_directories(${WEBP_DEP_IMG_INCLUDE_DIRS})
-  parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "WEBPINFO_SRCS"
-    "webpinfo")
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "WEBPINFO_SRCS"
+                    "webpinfo")
   add_executable(webpinfo ${WEBPINFO_SRCS})
   target_link_libraries(webpinfo exampleutil imageioutil)
-  install(TARGETS webpinfo RUNTIME DESTINATION bin)
-  set_property(TARGET webpinfo PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
+  target_include_directories(webpinfo PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+  install(TARGETS webpinfo RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+endif()
+
+if(WEBP_BUILD_WEBPMUX)
+  # webpmux
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "WEBPMUX_SRCS"
+                    "webpmux")
+  add_executable(webpmux ${WEBPMUX_SRCS})
+  target_link_libraries(webpmux exampleutil imageioutil libwebpmux webp)
+  target_include_directories(webpmux PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+  install(TARGETS webpmux RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+endif()
+
+if(WEBP_BUILD_EXTRAS)
+  set(EXTRAS_MAKEFILE "${CMAKE_CURRENT_SOURCE_DIR}/extras")
+  parse_makefile_am(${EXTRAS_MAKEFILE} "WEBP_EXTRAS_SRCS" "libwebpextras_la")
+  parse_makefile_am(${EXTRAS_MAKEFILE} "GET_DISTO_SRCS" "get_disto")
+  parse_makefile_am(${EXTRAS_MAKEFILE} "WEBP_QUALITY_SRCS" "webp_quality")
+  parse_makefile_am(${EXTRAS_MAKEFILE} "VWEBP_SDL_SRCS" "vwebp_sdl")
+
+  # get_disto
+  add_executable(get_disto ${GET_DISTO_SRCS})
+  target_link_libraries(get_disto imagedec)
+  target_include_directories(get_disto
+                             PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}
+                                     ${CMAKE_CURRENT_SOURCE_DIR}/src
+                                     ${CMAKE_CURRENT_BINARY_DIR}/src)
+  install(TARGETS get_disto RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+
+  # webp_quality
+  add_executable(webp_quality ${WEBP_QUALITY_SRCS} ${WEBP_EXTRAS_SRCS})
+  target_link_libraries(webp_quality exampleutil imagedec)
+  target_include_directories(webp_quality PRIVATE ${CMAKE_CURRENT_SOURCE_DIR})
+  install(TARGETS webp_quality RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+
+  # vwebp_sdl
+  find_package(SDL)
+  if(SDL_FOUND)
+    add_executable(vwebp_sdl ${VWEBP_SDL_SRCS})
+    target_link_libraries(vwebp_sdl ${SDL_LIBRARY} imageioutil webp)
+    target_include_directories(vwebp_sdl
+                               PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}
+                                       ${CMAKE_CURRENT_SOURCE_DIR}/src
+                                       ${CMAKE_CURRENT_BINARY_DIR}
+                                       ${CMAKE_CURRENT_BINARY_DIR}/src
+                                       ${SDL_INCLUDE_DIR})
+    set(WEBP_HAVE_SDL 1)
+    target_compile_definitions(vwebp_sdl PUBLIC WEBP_HAVE_SDL)
+    install(TARGETS vwebp_sdl RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+  endif()
 endif()
 
 if(WEBP_BUILD_WEBP_JS)
   # JavaScript version
-  add_executable(webp_js
-                 ${CMAKE_CURRENT_SOURCE_DIR}/extras/webp_to_sdl.c)
+  add_executable(webp_js ${CMAKE_CURRENT_SOURCE_DIR}/extras/webp_to_sdl.c)
   target_link_libraries(webp_js webpdecoder SDL)
+  target_include_directories(webp_js PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
   set(WEBP_HAVE_SDL 1)
-  set_target_properties(webp_js PROPERTIES LINK_FLAGS
-      "-s EXPORTED_FUNCTIONS='[\"_WebpToSDL\"]' -s INVOKE_RUN=0 \
+  set_target_properties(
+    webp_js
+    PROPERTIES LINK_FLAGS
+               "-s EXPORTED_FUNCTIONS='[\"_WebpToSDL\"]' -s INVOKE_RUN=0 \
        -s EXTRA_EXPORTED_RUNTIME_METHODS='[\"cwrap\"]'")
   set_target_properties(webp_js PROPERTIES OUTPUT_NAME webp)
   target_compile_definitions(webp_js PUBLIC EMSCRIPTEN WEBP_HAVE_SDL)
 
   # WASM version
-  add_executable(webp_wasm
-                 ${CMAKE_CURRENT_SOURCE_DIR}/extras/webp_to_sdl.c)
+  add_executable(webp_wasm ${CMAKE_CURRENT_SOURCE_DIR}/extras/webp_to_sdl.c)
   target_link_libraries(webp_wasm webpdecoder SDL)
-  set_target_properties(webp_wasm PROPERTIES LINK_FLAGS
-      "-s WASM=1 -s 'BINARYEN_METHOD=\"native-wasm\"' \
+  target_include_directories(webp_wasm PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
+  set_target_properties(
+    webp_wasm
+    PROPERTIES LINK_FLAGS "-s WASM=1 -s 'BINARYEN_METHOD=\"native-wasm\"' \
        -s EXPORTED_FUNCTIONS='[\"_WebpToSDL\"]' -s INVOKE_RUN=0 \
        -s EXTRA_EXPORTED_RUNTIME_METHODS='[\"cwrap\"]'")
   target_compile_definitions(webp_wasm PUBLIC EMSCRIPTEN WEBP_HAVE_SDL)
 
-  target_compile_definitions(webpdecoder PUBLIC EMSCRIPTEN)
+  target_compile_definitions(webpdspdecode PUBLIC EMSCRIPTEN)
 endif()
 
-# Generate the config.h file.
-configure_file(${CMAKE_CURRENT_LIST_DIR}/cmake/config.h.in
-  ${CMAKE_CURRENT_BINARY_DIR}/src/webp/config.h)
-add_definitions(-DHAVE_CONFIG_H)
-# The webp folder is included as we reference config.h as
-# ../webp/config.h or webp/config.h
-include_directories(${CMAKE_CURRENT_BINARY_DIR})
+if(WEBP_BUILD_ANIM_UTILS)
+  # anim_diff
+  include_directories(${WEBP_DEP_IMG_INCLUDE_DIRS} ${WEBP_DEP_GIF_INCLUDE_DIRS})
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "ANIM_DIFF_SRCS"
+                    "anim_diff")
+  add_executable(anim_diff ${ANIM_DIFF_SRCS})
+  target_link_libraries(anim_diff
+                        exampleutil
+                        imagedec
+                        imageenc
+                        imageioutil
+                        webp
+                        libwebpmux
+                        webpdemux
+                        ${WEBP_DEP_GIF_LIBRARIES})
+  target_include_directories(anim_diff PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+
+  # anim_dump
+  include_directories(${WEBP_DEP_IMG_INCLUDE_DIRS} ${WEBP_DEP_GIF_INCLUDE_DIRS})
+  parse_makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/examples "ANIM_DUMP_SRCS"
+                    "anim_dump")
+  add_executable(anim_dump ${ANIM_DUMP_SRCS})
+  target_link_libraries(anim_dump
+                        exampleutil
+                        imagedec
+                        imageenc
+                        imageioutil
+                        webp
+                        libwebpmux
+                        webpdemux
+                        ${WEBP_DEP_GIF_LIBRARIES})
+  target_include_directories(anim_dump PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/src)
+endif()
 
 # Install the different headers and libraries.
-install(FILES ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/decode.h
-              ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/demux.h
-              ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/encode.h
-              ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/mux.h
-              ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/mux_types.h
-              ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/types.h
-        DESTINATION include/webp)
+include(GNUInstallDirs)
 install(TARGETS ${INSTALLED_LIBRARIES}
-        LIBRARY DESTINATION lib
-        ARCHIVE DESTINATION lib)
+        EXPORT ${PROJECT_NAME}Targets
+        PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/webp
+        INCLUDES
+        DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
+        ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+set(ConfigPackageLocation ${CMAKE_INSTALL_DATADIR}/${PROJECT_NAME}/cmake/)
+install(EXPORT ${PROJECT_NAME}Targets
+        NAMESPACE ${PROJECT_NAME}::
+        DESTINATION ${ConfigPackageLocation})
 
 # Create the CMake version file.
 include(CMakePackageConfigHelpers)
 write_basic_package_version_file(
   "${CMAKE_CURRENT_BINARY_DIR}/WebPConfigVersion.cmake"
   VERSION ${PACKAGE_VERSION}
-  COMPATIBILITY AnyNewerVersion
-)
+  COMPATIBILITY AnyNewerVersion)
 
 # Create the Config file.
 include(CMakePackageConfigHelpers)
-set(ConfigPackageLocation share/WebP/cmake/)
 configure_package_config_file(
   ${CMAKE_CURRENT_SOURCE_DIR}/cmake/WebPConfig.cmake.in
   ${CMAKE_CURRENT_BINARY_DIR}/WebPConfig.cmake
-  INSTALL_DESTINATION ${ConfigPackageLocation}
-)
+  INSTALL_DESTINATION
+  ${ConfigPackageLocation})
 
 # Install the generated CMake files.
-install(
-  FILES "${CMAKE_CURRENT_BINARY_DIR}/WebPConfigVersion.cmake"
-        "${CMAKE_CURRENT_BINARY_DIR}/WebPConfig.cmake"
-  DESTINATION ${ConfigPackageLocation}
-)
+install(FILES "${CMAKE_CURRENT_BINARY_DIR}/WebPConfigVersion.cmake"
+              "${CMAKE_CURRENT_BINARY_DIR}/WebPConfig.cmake"
+        DESTINATION ${ConfigPackageLocation})
 
 # Install the man pages.
-set(MAN_PAGES cwebp.1 dwebp.1 gif2webp.1 img2webp.1 vwebp.1 webpmux.1
-  webpinfo.1)
-set(EXEC_BUILDS "CWEBP" "DWEBP" "GIF2WEBP" "IMG2WEBP" "VWEBP" "WEBPMUX"
-  "WEBPINFO")
+set(MAN_PAGES
+    cwebp.1
+    dwebp.1
+    gif2webp.1
+    img2webp.1
+    vwebp.1
+    webpmux.1
+    webpinfo.1)
+set(EXEC_BUILDS
+    "CWEBP"
+    "DWEBP"
+    "GIF2WEBP"
+    "IMG2WEBP"
+    "VWEBP"
+    "WEBPMUX"
+    "WEBPINFO")
 list(LENGTH MAN_PAGES MAN_PAGES_LENGTH)
 math(EXPR MAN_PAGES_RANGE "${MAN_PAGES_LENGTH} - 1")
 
@@ -364,8 +648,7 @@ foreach(I_MAN RANGE ${MAN_PAGES_RANGE})
   if(WEBP_BUILD_${EXEC_BUILD})
     list(GET MAN_PAGES ${I_MAN} MAN_PAGE)
     install(FILES ${CMAKE_CURRENT_SOURCE_DIR}/man/${MAN_PAGE}
-      DESTINATION ${CMAKE_INSTALL_PREFIX}/share/man/man1
-      COMPONENT doc
-    )
+            DESTINATION ${CMAKE_INSTALL_MANDIR}/man1
+            COMPONENT doc)
   endif()
 endforeach()

ChangeLog

@@ -1,13 +1,116 @@
+fa8210e4 Fix pair update in stochastic entropy merging.
+825389ac README.mux: add a reference to the AnimDecoder API
+3be698c3 CMake: fix webp_js compilation
+4cbb4caf update NEWS
+f5a5918d bump version to 1.0.1
+d61385db Speed-up: Make sure we only initialize histograms when needed.
+0c570316 update AUTHORS
+301a2dda img2webp: add help note about arguments from a file
+f0abab92 Speedups for empty histograms.
+f2dfd925 Split HistogramAdd to only have the high level logic in C.
+06b7bc7d Fix compilation on windows and clang-cl+ninja.
+98179495 remove some minor TODOs
+cbf82cc0 Remove AVX2 files.
+5030e902 Merge "TIFF decoder: remove unused KINV definition"
+ac543311 Remove a few more useless #defines
+123d3306 TIFF decoder: remove unused KINV definition
+ef1094b0 Merge "- install pkg-config files during the CMake build"
+b911fbc9 libwebp: Remove duplicate GIFDisplayError in anim_util
+eee00b66 - install pkg-config files during the CMake build
+ac3ec8c9 Merge "Clean-up the common sources in dsp."
+3e13da7b Clean-up the common sources in dsp.
+5c395f1d libwebp: cmake-format all
+e7a69729 libwebp: Add extras targets in CMakeLists.txt
+e52485d6 libwebp: Rename macros in webpmux.c
+92dc0f09 clean-up MakeInputImageCopy()
+39952de2 VP8IteratorImport: add missing 'const'
+382af7a2 clean-up WebPBlendAlpha
+14d020f6 libwebp: Use ExUtilGet*() in anim_diff
+0d92ff25 libwebp: remove useless variable in gif2webp
+556cb1b4 Merge "CMake: Set WEBP_BUILD_GIF2WEBP to off"
+da26ee49 CMake: Set WEBP_BUILD_GIF2WEBP to off
+b2a867c0 cwebp: Don't premultiply during -resize if -exact
+637141bc pngdec: fix build w/libpng < 1.4.x
+bc5092b1 pngdec: set memory functions
+50d8345a Fix CMake math library.
+6aa3e8aa Fix math library on Visual Studio.
+d71df4e2 Fix math library finding in CMake.
+de08d727 cosmetics: normalize include guard comment
+009562b4 vwebp: Fix bug when Dispose then NoBlend frames
+423f2579 Fix up CMake to create targets.
+907208f9 Wait for all threads to be done in DecodeRemaining.
+4649b3c4 vwebp: Add background color display option
+78ad57a3 Fix bad glClearColor parameters
+da96d8d9 Allow for a non-initialized alpha decompressor in DoRemap.
+2563db47 fix rescaling rounding inaccuracy
+211f37ee fix endian problems in pattern copy
+5f0f5c07 Make sure partition #0 is read before VP8 data in IDecode.
+de98732b fix GetColorf() bug
+4338cd36 misc fixes in libwebpmux
+e00af13e fix signatures after a9ceda7ff1
+a9ceda7f Speed-up chunk list operations.
+2281bbf6 Merge "Better handling of bogus Huffman codes."
+39cb9aad Better handling of bogus Huffman codes.
+89cc9d37 Merge "fix read-overflow while parsing VP8X chunk"
+95fd6507 fix read-overflow while parsing VP8X chunk
+9e729fe1 Fix VP8IoTeardownHook being called twice on worker sync failure
+29fb8562 Merge "muxread,anmf: fail on multiple image chunks"
+eb82ce76 muxread,anmf: fail on multiple image chunks
+1344a2e9 fix alpha-filtering crash when image width is larger than radius
+be738c6d muxread,ChunkVerifyAndAssign: validate chunk_size
+2c70ad76 muxread,CreateInternal: fix riff size checks
+569001f1 Fix for thread race heap-use-after-free
+c56a02d9 Android.mk: use LOCAL_EXPORT_C_INCLUDES w/public libs
+15795596 CMakeLists.txt,cosmetics: normalize if() formatting
+1a44c233 Merge "cmake: add support for webpmux"
+e9569ad7 Merge "configure,*am,cosmetics: s/WANT_/BUILD_/"
+35c7de6f cmake: add support for webpmux
+0f25e61c WebpToSDL(): fix the return value in case of error
+5d8985de configure,*am,cosmetics: s/WANT_/BUILD_/
+895fd28f Merge "man/Makefile.am: add img2webp.1"
+5cf3e2af man/Makefile.am: add img2webp.1
+2a9de5b9 Add build rules for anim_diff & anim_dump utils.
+71ed73cf fix invalid check for buffer size
+af0e4fbb gif2webp: fix transcode of loop count=65535
+dce5d764 Limit memory allocation when reading invalid Huffman codes.
+f9df0081 Merge "cmake: quiet glut deprecation warnings on OS X"
+dc39b16f webpmux.1: correct grammar
+c7aa1264 cwebp.c: fix a missing \n
+53aa51e9 Merge tag 'v1.0.0'
+698b8844 update ChangeLog (tag: v1.0.0)
+8d510751 webp-container-spec: correct frame duration=0 note
+e6b2164e vwebp: Copy Chrome's behavior w/frame duration == 0
+094b3b28 cmake: quiet glut deprecation warnings on OS X
+71c39a06 webp-container-spec: correct frame duration=0 note
+fd3d5756 vwebp: Copy Chrome's behavior w/frame duration == 0
+b0c966fb Build vwebp from CMake.
+d20b7707 update ChangeLog (tag: v1.0.0-rc3)
+0d5fad46 add WEBP_DSP_INIT / WEBP_DSP_INIT_FUNC
+d77bf512 add WEBP_DSP_INIT / WEBP_DSP_INIT_FUNC
+c1cb86af fix 16b overflow in SSE2
+e577feb7 makefile.unix: add DEBUG flag for compiling w/ debug-symbol
+99be34b3 cwebp,get_disto: fix bpp output
+e122e511 cwebp,get_disto: fix bpp output
+f5565ca8 cmake: Make sure we use near-lossless by default.
+d898dc14 fix bug in WebPImport565: alpha value was not set
+1c8f358d Fix CMake with WASM.
+a0215fb7 webp_js: fix webp_js demo html
+882784b0 update ChangeLog (tag: v1.0.0-rc2)
 2f930e08 Revert "Use proper targets for CMake."
 8165e8fb Use proper targets for CMake.
 3f157dd5 Remove some very hard TODOs.
+abb47760 Merge "Use proper targets for CMake."
 cd758a17 {de,}mux/Makefile.am: add missing headers
+e155dda0 Use proper targets for CMake.
 b892b8ba makefile.unix,dist: use ascii for text output
 64a57d05 add -version option to anim_dump,anim_diff and img2webp
+994be82d Merge "Remove some very hard TODOs."
+4033e1d7 Remove some very hard TODOs.
 fc1b8e3a webp_js: fix webp_js demo html
-15aa48d9 update ChangeLog
+15aa48d9 update ChangeLog (tag: v1.0.0-rc1)
 e607dabc update AUTHORS
 38410c08 [CFI] Remove function pointer casts
+978eec25 [CFI] Remove function pointer casts
 c57b2736 bump version to 1.0.0
 cba28853 update NEWS
 c909d531 Merge "remove some deprecation warning on MacOSX"

Makefile.am

@@ -2,7 +2,7 @@ ACLOCAL_AMFLAGS = -I m4
 SUBDIRS = src imageio man
 EXTRA_DIST = COPYING autogen.sh
 
-if WANT_EXTRAS
+if BUILD_EXTRAS
   SUBDIRS += extras
 endif
 

Makefile.vc

@@ -53,11 +53,6 @@ OUTDIR = ..\obj\
 OUTDIR = $(OBJDIR)
 !ENDIF
 
-!IF "$(HAVE_AVX2)" == "1"
-CFLAGS = $(CFLAGS) /DWEBP_HAVE_AVX2
-AVX2_FLAGS = /arch:AVX2
-!ENDIF
-
 ##############################################################
 # Runtime library configuration
 !IF "$(RTLIBCFG)" == "static"
@@ -241,7 +236,6 @@ DSP_ENC_OBJS = \
     $(DIROBJ)\dsp\cost_mips_dsp_r2.obj \
     $(DIROBJ)\dsp\cost_sse2.obj \
     $(DIROBJ)\dsp\enc.obj \
-    $(DIROBJ)\dsp\enc_avx2.obj \
     $(DIROBJ)\dsp\enc_mips32.obj \
     $(DIROBJ)\dsp\enc_mips_dsp_r2.obj \
     $(DIROBJ)\dsp\enc_msa.obj \
@@ -454,9 +448,6 @@ $(DIROBJ)\$(DLLINC):
 .SUFFIXES: .c .obj .res .exe
 # File-specific flag builds. Note batch rules take precedence over wildcards,
 # so for now name each file individually.
-$(DIROBJ)\dsp\enc_avx2.obj: src\dsp\enc_avx2.c
-	$(CC) $(CFLAGS) $(AVX2_FLAGS) /Fd$(LIBWEBP_PDBNAME) /Fo$(DIROBJ)\dsp\ \
-	  src\dsp\$(@B).c
 $(DIROBJ)\examples\anim_diff.obj: examples\anim_diff.c
 	$(CC) $(CFLAGS) /DWEBP_HAVE_GIF /Fd$(LIBWEBP_PDBNAME) \
 	  /Fo$(DIROBJ)\examples\ examples\$(@B).c

NEWS

@@ -1,3 +1,15 @@
+- 11/2/2018: version 1.0.1
+  This is a binary compatible release.
+  * lossless encoder speedups
+  * big-endian fix for alpha decoding (issue #393)
+  * gif2webp fix for loop count=65535 transcode (issue #382)
+  * further security related hardening in libwebp & libwebpmux
+    (issues #383, #385, #386, #387, #388, #391)
+    (oss-fuzz #9099, #9100, #9105, #9106, #9111, #9112, #9119, #9123, #9170,
+              #9178, #9179, #9183, #9186, #9191, #9364, #9417, #9496, #10349,
+              #10423, #10634, #10700, #10838, #10922, #11021, #11088, #11152)
+  * miscellaneous bug & build fixes (issues #381, #394, #396, #397, #400)
+
 - 4/2/2018: version 1.0.0
   This is a binary compatible release.
   * lossy encoder improvements to avoid chroma shifts in various circumstances

README

@@ -4,7 +4,7 @@
           \__\__/\____/\_____/__/ ____  ___
                 / _/ /    \    \ /  _ \/ _/
                /  \_/   / /   \ \   __/  \__
-               \____/____/\_____/_____/____/v1.0.0
+               \____/____/\_____/_____/____/v1.0.1
 
 Description:
 ============
@@ -402,12 +402,14 @@ Options are:
   -nofilter .... disable in-loop filtering
   -dither <int>  dithering strength (0..100), default=50
   -noalphadither disable alpha plane dithering
+  -usebgcolor .. display background color
   -mt .......... use multi-threading
   -info ........ print info
   -h ........... this help message
 
 Keyboard shortcuts:
   'c' ................ toggle use of color profile
+  'b' ................ toggle background color display
   'i' ................ overlay file information
   'd' ................ disable blending & disposal (debug)
   'q' / 'Q' / ESC .... quit
@@ -470,6 +472,9 @@ Per-frame options (only used for subsequent images input):
 example: img2webp -loop 2 in0.png -lossy in1.jpg
                   -d 80 in2.tiff -o out.webp
 
+Note: if a single file name is passed as the argument, the arguments will be
+tokenized from this file. The file name must not start with the character '-'.
+
 Animated GIF conversion:
 ========================
 Animated GIF files can be converted to WebP files with animation using the

README.mux

@@ -1,7 +1,7 @@
           __   __  ____  ____  ____  __ __  _     __ __
          /  \\/  \/  _ \/  _ \/  _ \/  \  \/ \___/_ / _\
          \       /   __/  _  \   __/      /  /  (_/  /__
-          \__\__/\_____/_____/__/  \__//_/\_____/__/___/v1.0.0
+          \__\__/\_____/_____/__/  \__//_/\_____/__/___/v1.0.1
 
 
 Description:
@@ -211,6 +211,35 @@ Code example:
 For a detailed AnimEncoder API reference, please refer to the header file
 (src/webp/mux.h).
 
+AnimDecoder API:
+================
+This AnimDecoder API allows decoding (possibly) animated WebP images.
+
+Code Example:
+
+  WebPAnimDecoderOptions dec_options;
+  WebPAnimDecoderOptionsInit(&dec_options);
+  // Tune 'dec_options' as needed.
+  WebPAnimDecoder* dec = WebPAnimDecoderNew(webp_data, &dec_options);
+  WebPAnimInfo anim_info;
+  WebPAnimDecoderGetInfo(dec, &anim_info);
+  for (uint32_t i = 0; i < anim_info.loop_count; ++i) {
+    while (WebPAnimDecoderHasMoreFrames(dec)) {
+      uint8_t* buf;
+      int timestamp;
+      WebPAnimDecoderGetNext(dec, &buf, &timestamp);
+      // ... (Render 'buf' based on 'timestamp').
+      // ... (Do NOT free 'buf', as it is owned by 'dec').
+    }
+    WebPAnimDecoderReset(dec);
+  }
+  const WebPDemuxer* demuxer = WebPAnimDecoderGetDemuxer(dec);
+  // ... (Do something using 'demuxer'; e.g. get EXIF/XMP/ICC data).
+  WebPAnimDecoderDelete(dec);
+
+For a detailed AnimDecoder API reference, please refer to the header file
+(src/webp/demux.h).
+
 
 Bugs:
 =====

build.gradle

@@ -175,7 +175,6 @@ model {
             include "cost_mips_dsp_r2.c"
             include "cost_sse2.c"
             include "enc.c"
-            include "enc_avx2.c"
             include "enc_mips32.c"
             include "enc_mips_dsp_r2.c"
             include "enc_msa.c"

cmake/WebPConfig.cmake.in

@@ -1,5 +1,10 @@
+set(WebP_VERSION @PROJECT_VERSION@)
+set(WEBP_VERSION ${WebP_VERSION})
+
 @PACKAGE_INIT@
 
+include ("${CMAKE_CURRENT_LIST_DIR}/@PROJECT_NAME@Targets.cmake")
+
 set(WebP_INCLUDE_DIRS "webp")
 set(WEBP_INCLUDE_DIRS ${WebP_INCLUDE_DIRS})
 set(WebP_LIBRARIES "@INSTALLED_LIBRARIES@")

cmake/config.h.in

@@ -103,9 +103,6 @@
 /* Version number of package */
 #cmakedefine VERSION "@VERSION@"
 
-/* Set to 1 if AVX2 is supported */
-#cmakedefine WEBP_HAVE_AVX2 1
-
 /* Set to 1 if GIF library is installed */
 #cmakedefine WEBP_HAVE_GIF 1
 

cmake/cpu.cmake

@@ -1,4 +1,4 @@
-## Check for SIMD extensions.
+# Check for SIMD extensions.
 include(CMakePushCheckState)
 
 function(webp_check_compiler_flag WEBP_SIMD_FLAG ENABLE_SIMD)
@@ -18,8 +18,7 @@ function(webp_check_compiler_flag WEBP_SIMD_FLAG ENABLE_SIMD)
         #endif
         return 0;
       }
-    " WEBP_HAVE_FLAG_${WEBP_SIMD_FLAG}
-  )
+    " WEBP_HAVE_FLAG_${WEBP_SIMD_FLAG})
   cmake_pop_check_state()
   if(WEBP_HAVE_FLAG_${WEBP_SIMD_FLAG})
     set(WEBP_HAVE_${WEBP_SIMD_FLAG} 1 PARENT_SCOPE)
@@ -29,16 +28,18 @@ function(webp_check_compiler_flag WEBP_SIMD_FLAG ENABLE_SIMD)
 endfunction()
 
 # those are included in the names of WEBP_USE_* in c++ code.
-set(WEBP_SIMD_FLAGS "SSE2;SSE41;AVX2;MIPS32;MIPS_DSP_R2;NEON;MSA")
-set(WEBP_SIMD_FILE_EXTENSIONS "_sse2.c;_sse41.c;_avx2.c;_mips32.c;_mips_dsp_r2.c;_neon.c;_msa.c")
+set(WEBP_SIMD_FLAGS "SSE41;SSE2;MIPS32;MIPS_DSP_R2;NEON;MSA")
+set(WEBP_SIMD_FILE_EXTENSIONS
+    "_sse41.c;_sse2.c;_mips32.c;_mips_dsp_r2.c;_neon.c;_msa.c")
 if(MSVC)
-  # MSVC does not have a SSE4 flag but AVX2 support implies
-  # SSE4 support.
-  set(SIMD_ENABLE_FLAGS "/arch:SSE2;/arch:AVX2;/arch:AVX2;;;;")
+  # MSVC does not have a SSE4 flag but AVX support implies SSE4 support.
+  set(SIMD_ENABLE_FLAGS "/arch:AVX;/arch:SSE2;;;;")
   set(SIMD_DISABLE_FLAGS)
 else()
-  set(SIMD_ENABLE_FLAGS "-msse2;-msse4.1;-mavx2;-mips32;-mdspr2;-mfpu=neon;-mmsa")
-  set(SIMD_DISABLE_FLAGS "-mno-sse2;-mno-sse4.1;-mno-avx2;;-mno-dspr2;;-mno-msa")
+  set(SIMD_ENABLE_FLAGS
+      "-msse4.1;-msse2;-mips32;-mdspr2;-mfpu=neon;-mmsa")
+  set(SIMD_DISABLE_FLAGS
+      "-mno-sse4.1;-mno-sse2;;-mno-dspr2;;-mno-msa")
 endif()
 
 set(WEBP_SIMD_FILES_TO_NOT_INCLUDE)
@@ -47,16 +48,16 @@ set(WEBP_SIMD_FLAGS_TO_INCLUDE)
 
 if(${ANDROID})
   if(${ANDROID_ABI} STREQUAL "armeabi-v7a")
-    # This is because Android studio uses the configuration
-    # "-march=armv7-a -mfloat-abi=softfp -mfpu=vfpv3-d16"
-    # that does not trigger neon optimizations but should
-    # (as this configuration does not exist anymore).
+    # This is because Android studio uses the configuration "-march=armv7-a
+    # -mfloat-abi=softfp -mfpu=vfpv3-d16" that does not trigger neon
+    # optimizations but should (as this configuration does not exist anymore).
     set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mfpu=neon ")
   endif()
 endif()
 
 list(LENGTH WEBP_SIMD_FLAGS WEBP_SIMD_FLAGS_LENGTH)
 math(EXPR WEBP_SIMD_FLAGS_RANGE "${WEBP_SIMD_FLAGS_LENGTH} - 1")
+unset(HIGHEST_SSE_FLAG)
 
 foreach(I_SIMD RANGE ${WEBP_SIMD_FLAGS_RANGE})
   list(GET WEBP_SIMD_FLAGS ${I_SIMD} WEBP_SIMD_FLAG)
@@ -72,18 +73,28 @@ foreach(I_SIMD RANGE ${WEBP_SIMD_FLAGS_RANGE})
     set(CMAKE_REQUIRED_FLAGS ${SIMD_COMPILE_FLAG})
     webp_check_compiler_flag(${WEBP_SIMD_FLAG} ${WEBP_ENABLE_SIMD})
   else()
-    set(SIMD_COMPILE_FLAG " ")
+    if(MSVC)
+      list(GET SIMD_ENABLE_FLAGS ${I_SIMD} SIMD_COMPILE_FLAG)
+    else()
+      set(SIMD_COMPILE_FLAG " ")
+    endif()
   endif()
   # Check which files we should include or not.
   list(GET WEBP_SIMD_FILE_EXTENSIONS ${I_SIMD} WEBP_SIMD_FILE_EXTENSION)
   file(GLOB SIMD_FILES "${CMAKE_CURRENT_LIST_DIR}/../"
-    "src/dsp/*${WEBP_SIMD_FILE_EXTENSION}"
-  )
+            "src/dsp/*${WEBP_SIMD_FILE_EXTENSION}")
   if(WEBP_HAVE_${WEBP_SIMD_FLAG})
+    if(${I_SIMD} LESS 2 AND NOT HIGHEST_SSE_FLAG)
+      set(HIGHEST_SSE_FLAG ${SIMD_COMPILE_FLAG})
+    endif()
     # Memorize the file and flags.
     foreach(FILE ${SIMD_FILES})
       list(APPEND WEBP_SIMD_FILES_TO_INCLUDE ${FILE})
-      list(APPEND WEBP_SIMD_FLAGS_TO_INCLUDE ${SIMD_COMPILE_FLAG})
+      if(${I_SIMD} LESS 2)
+        list(APPEND WEBP_SIMD_FLAGS_TO_INCLUDE ${HIGHEST_SSE_FLAG})
+      else()
+        list(APPEND WEBP_SIMD_FLAGS_TO_INCLUDE ${SIMD_COMPILE_FLAG})
+      endif()
     endforeach()
   else()
     # Remove the file from the list.
@@ -107,10 +118,10 @@ foreach(I_SIMD RANGE ${WEBP_SIMD_FLAGS_RANGE})
           endif()
           set(CMAKE_REQUIRED_DEFINITIONS ${SIMD_COMPILE_FLAG})
           check_c_source_compiles("int main(void) {return 0;}"
-            FLAG_${SIMD_COMPILE_FLAG}
-            FAIL_REGEX "warning: argument unused during compilation:"
-            ${COMMON_PATTERNS}
-          )
+                                  FLAG_${SIMD_COMPILE_FLAG}
+                                  FAIL_REGEX
+                                  "warning: argument unused during compilation:"
+                                  ${COMMON_PATTERNS})
           if(NOT FLAG_${SIMD_COMPILE_FLAG})
             unset(HAS_COMPILE_FLAG CACHE)
           endif()

cmake/deps.cmake

@@ -1,33 +1,27 @@
 # Generate the config.h to compile with specific intrinsics / libs.
 
-## Check for compiler options.
+# Check for compiler options.
 include(CheckCSourceCompiles)
 check_c_source_compiles("
     int main(void) {
       (void)__builtin_bswap16(0);
       return 0;
     }
-  "
-  HAVE_BUILTIN_BSWAP16
-)
+  " HAVE_BUILTIN_BSWAP16)
 check_c_source_compiles("
     int main(void) {
       (void)__builtin_bswap32(0);
       return 0;
     }
-  "
-  HAVE_BUILTIN_BSWAP32
-)
+  " HAVE_BUILTIN_BSWAP32)
 check_c_source_compiles("
     int main(void) {
       (void)__builtin_bswap64(0);
       return 0;
     }
-  "
-  HAVE_BUILTIN_BSWAP64
-)
+  " HAVE_BUILTIN_BSWAP64)
 
-## Check for libraries.
+# Check for libraries.
 find_package(Threads)
 if(Threads_FOUND)
   if(CMAKE_USE_PTHREADS_INIT)
@@ -40,8 +34,7 @@ if(Threads_FOUND)
           int attr = ${PTHREAD_TEST};
           return attr;
         }
-      " ${PTHREAD_TEST}
-    )
+      " ${PTHREAD_TEST})
   endforeach()
   list(APPEND WEBP_DEP_LIBRARIES ${CMAKE_THREAD_LIBS_INIT})
 endif()
@@ -51,15 +44,20 @@ set(WEBP_USE_THREAD ${Threads_FOUND})
 set(LT_OBJDIR ".libs/")
 
 # Only useful for vwebp, so useless for now.
-# find_package(OpenGL)
-# set(WEBP_HAVE_GL ${OPENGL_FOUND})
-# set(WEBP_DEP_INCLUDE_DIRS ${WEBP_DEP_INCLUDE_DIRS} ${OPENGL_INCLUDE_DIRS})
-# set(WEBP_DEP_LIBRARIES ${WEBP_DEP_LIBRARIES} ${OPENGL_LIBRARIES})
+find_package(OpenGL)
+set(WEBP_HAVE_GL ${OPENGL_FOUND})
 
-# Find the standard C math library.
-find_library(MATH_LIBRARY NAMES m)
-if(MATH_LIBRARY)
-  list(APPEND WEBP_DEP_LIBRARIES ${MATH_LIBRARY})
+# Check if we need to link to the C math library. We do not look for it as it is
+# not found when cross-compiling, while it is here.
+check_c_source_compiles("
+    #include <math.h>
+    int main(int argc, char** argv) {
+      return (int)pow(argc, 2.5);
+    }
+  " HAVE_MATH_LIBRARY)
+if(NOT HAVE_MATH_LIBRARY)
+  message(STATUS "Adding -lm flag.")
+  list(APPEND WEBP_DEP_LIBRARIES m)
 endif()
 
 # Find the standard image libraries.
@@ -70,8 +68,8 @@ foreach(I_LIB PNG JPEG TIFF)
   set(WEBP_HAVE_${I_LIB} ${${I_LIB}_FOUND})
   if(${I_LIB}_FOUND)
     list(APPEND WEBP_DEP_IMG_LIBRARIES ${${I_LIB}_LIBRARIES})
-    list(APPEND WEBP_DEP_IMG_INCLUDE_DIRS
-         ${${I_LIB}_INCLUDE_DIR} ${${I_LIB}_INCLUDE_DIRS})
+    list(APPEND WEBP_DEP_IMG_INCLUDE_DIRS ${${I_LIB}_INCLUDE_DIR}
+                ${${I_LIB}_INCLUDE_DIRS})
   endif()
 endforeach()
 if(WEBP_DEP_IMG_INCLUDE_DIRS)
@@ -87,8 +85,8 @@ set(WEBP_HAVE_GIF ${GIF_FOUND})
 if(GIF_FOUND)
   # GIF find_package only locates the header and library, it doesn't fail
   # compile tests when detecting the version, but falls back to 3 (as of at
-  # least cmake 3.7.2). Make sure the library links to avoid incorrect
-  # detection when cross compiling.
+  # least cmake 3.7.2). Make sure the library links to avoid incorrect detection
+  # when cross compiling.
   cmake_push_check_state()
   set(CMAKE_REQUIRED_LIBRARIES ${GIF_LIBRARIES})
   set(CMAKE_REQUIRED_INCLUDES ${GIF_INCLUDE_DIR})
@@ -98,8 +96,7 @@ if(GIF_FOUND)
         (void)DGifOpenFileHandle;
         return 0;
       }
-      " GIF_COMPILES
-  )
+      " GIF_COMPILES)
   cmake_pop_check_state()
   if(GIF_COMPILES)
     list(APPEND WEBP_DEP_GIF_LIBRARIES ${GIF_LIBRARIES})
@@ -109,7 +106,7 @@ if(GIF_FOUND)
   endif()
 endif()
 
-## Check for specific headers.
+# Check for specific headers.
 include(CheckIncludeFiles)
 check_include_files("stdlib.h;stdarg.h;string.h;float.h" STDC_HEADERS)
 check_include_files(dlfcn.h HAVE_DLFCN_H)
@@ -131,25 +128,32 @@ check_include_files(windows.h HAVE_WINDOWS_H)
 
 # Windows specifics
 if(HAVE_WINCODEC_H)
-  list(APPEND WEBP_DEP_LIBRARIES shlwapi ole32 windowscodecs)
+  list(APPEND WEBP_DEP_LIBRARIES
+              shlwapi
+              ole32
+              windowscodecs)
 endif()
 
-## Check for SIMD extensions.
+# Check for SIMD extensions.
 include(${CMAKE_CURRENT_LIST_DIR}/cpu.cmake)
 
-## Define extra info.
+# Define extra info.
 set(PACKAGE ${PROJECT_NAME})
 set(PACKAGE_NAME ${PROJECT_NAME})
 
 # Read from configure.ac.
 file(READ ${CMAKE_CURRENT_SOURCE_DIR}/configure.ac CONFIGURE_AC)
-string(REGEX MATCHALL "\\[([0-9a-z\\.:/]*)\\]"
-  CONFIGURE_AC_PACKAGE_INFO ${CONFIGURE_AC}
-)
+string(REGEX MATCHALL
+             "\\[([0-9a-z\\.:/]*)\\]"
+             CONFIGURE_AC_PACKAGE_INFO
+             ${CONFIGURE_AC})
 function(strip_bracket VAR)
   string(LENGTH ${${VAR}} TMP_LEN)
   math(EXPR TMP_LEN ${TMP_LEN}-2)
-  string(SUBSTRING ${${VAR}} 1 ${TMP_LEN} TMP_SUB)
+  string(SUBSTRING ${${VAR}}
+                   1
+                   ${TMP_LEN}
+                   TMP_SUB)
   set(${VAR} ${TMP_SUB} PARENT_SCOPE)
 endfunction()
 

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT([libwebp], [1.0.0],
+AC_INIT([libwebp], [1.0.1],
         [https://bugs.chromium.org/p/webp],,
         [http://developers.google.com/speed/webp])
 AC_CANONICAL_HOST
@@ -122,31 +122,6 @@ AS_IF([test "$GCC" = "yes" ], [
 AC_SUBST([AM_CFLAGS])
 
 dnl === Check for machine specific flags
-AC_ARG_ENABLE([avx2],
-              AS_HELP_STRING([--disable-avx2],
-                             [Disable detection of AVX2 support
-                              @<:@default=auto@:>@]))
-
-AS_IF([test "x$enable_avx2" != "xno" -a "x$enable_sse4_1" != "xno" \
-         -a "x$enable_sse2" != "xno"], [
-  AVX2_CFLAGS="$INTRINSICS_CFLAGS $AVX2_FLAGS"
-  TEST_AND_ADD_CFLAGS([AVX2_FLAGS], [-mavx2])
-  AS_IF([test -n "$AVX2_FLAGS"], [
-    SAVED_CFLAGS=$CFLAGS
-    CFLAGS="$CFLAGS $AVX2_FLAGS"
-    AC_CHECK_HEADER([immintrin.h],
-                    [AC_DEFINE(WEBP_HAVE_AVX2, [1],
-                     [Set to 1 if AVX2 is supported])],
-                    [AVX2_FLAGS=""],
-                    dnl it's illegal to directly include avx2intrin.h, but it's
-                    dnl included conditionally in immintrin.h, tricky!
-                    [#ifndef __AVX2__
-                     #error avx2 is not enabled
-                     #endif
-                    ])
-    CFLAGS=$SAVED_CFLAGS])
-  AC_SUBST([AVX2_FLAGS])])
-
 AC_ARG_ENABLE([sse4.1],
               AS_HELP_STRING([--disable-sse4.1],
                              [Disable detection of SSE4.1 support
@@ -722,7 +697,7 @@ AC_ARG_ENABLE([libwebpmux],
               AS_HELP_STRING([--enable-libwebpmux],
                              [Build libwebpmux @<:@default=no@:>@]))
 AC_MSG_RESULT(${enable_libwebpmux-no})
-AM_CONDITIONAL([WANT_MUX], [test "$enable_libwebpmux" = "yes"])
+AM_CONDITIONAL([BUILD_MUX], [test "$enable_libwebpmux" = "yes"])
 
 dnl === Check whether libwebpdemux should be built
 AC_MSG_CHECKING(whether libwebpdemux is to be built)
@@ -731,7 +706,7 @@ AC_ARG_ENABLE([libwebpdemux],
                              [Disable libwebpdemux @<:@default=no@:>@]),
               [], [enable_libwebpdemux=yes])
 AC_MSG_RESULT(${enable_libwebpdemux-no})
-AM_CONDITIONAL([WANT_DEMUX], [test "$enable_libwebpdemux" = "yes"])
+AM_CONDITIONAL([BUILD_DEMUX], [test "$enable_libwebpdemux" = "yes"])
 
 dnl === Check whether decoder library should be built.
 AC_MSG_CHECKING(whether decoder library is to be built)
@@ -747,7 +722,7 @@ AC_ARG_ENABLE([libwebpextras],
               AS_HELP_STRING([--enable-libwebpextras],
                              [Build libwebpextras @<:@default=no@:>@]))
 AC_MSG_RESULT(${enable_libwebpextras-no})
-AM_CONDITIONAL([WANT_EXTRAS], [test "$enable_libwebpextras" = "yes"])
+AM_CONDITIONAL([BUILD_EXTRAS], [test "$enable_libwebpextras" = "yes"])
 
 dnl =========================
 

doc/webp-container-spec.txt

@@ -446,8 +446,9 @@ Frame Height Minus One: 24 bits (_uint24_)
 Frame Duration: 24 bits (_uint24_)
 
 : The time to wait before displaying the next frame, in 1 millisecond units.
-  In particular, frame duration of 0 is useful when one wants to update
-  multiple areas of the canvas at once during the animation.
+  Note the interpretation of frame duration of 0 (and often <= 10) is
+  implementation defined. Many tools and browsers assign a minimum duration
+  similar to GIF.
 
 Reserved: 6 bits
 

examples/Android.mk

@@ -26,7 +26,6 @@ LOCAL_SRC_FILES := \
     cwebp.c \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
 LOCAL_STATIC_LIBRARIES := example_util imageio_util imagedec webpdemux webp
 
 LOCAL_MODULE := cwebp
@@ -42,7 +41,6 @@ LOCAL_SRC_FILES := \
     dwebp.c \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
 LOCAL_STATIC_LIBRARIES := example_util imagedec imageenc webpdemux webp
 LOCAL_MODULE := dwebp
 
@@ -57,7 +55,6 @@ LOCAL_SRC_FILES := \
     webpmux.c \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
 LOCAL_STATIC_LIBRARIES := example_util imageio_util webpmux webp
 
 LOCAL_MODULE := webpmux_example
@@ -73,7 +70,6 @@ LOCAL_SRC_FILES := \
     img2webp.c \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
 LOCAL_STATIC_LIBRARIES := example_util imageio_util imagedec webpmux webpdemux \
                           webp
 
@@ -90,7 +86,6 @@ LOCAL_SRC_FILES := \
     webpinfo.c \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
-LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
 LOCAL_STATIC_LIBRARIES := example_util imageio_util webp
 
 LOCAL_MODULE := webpinfo_example

examples/Makefile.am

@@ -1,7 +1,7 @@
 AM_CPPFLAGS += -I$(top_builddir)/src -I$(top_srcdir)/src
 
 bin_PROGRAMS =
-if WANT_DEMUX
+if BUILD_DEMUX
   bin_PROGRAMS += dwebp cwebp
 endif
 if BUILD_ANIMDIFF
@@ -13,7 +13,7 @@ endif
 if BUILD_IMG2WEBP
   bin_PROGRAMS += img2webp
 endif
-if WANT_MUX
+if BUILD_MUX
   bin_PROGRAMS += webpmux
 endif
 if BUILD_VWEBP
@@ -28,7 +28,7 @@ noinst_LTLIBRARIES = libexample_util.la
 libexample_util_la_SOURCES = example_util.c example_util.h
 libexample_util_la_LIBADD = ../src/libwebp.la
 
-anim_diff_SOURCES = anim_diff.c anim_util.c anim_util.h
+anim_diff_SOURCES = anim_diff.c anim_util.c anim_util.h gifdec.c gifdec.h
 anim_diff_CPPFLAGS = $(AM_CPPFLAGS) $(GIF_INCLUDES)
 anim_diff_LDADD  =
 anim_diff_LDADD += ../src/demux/libwebpdemux.la
@@ -36,7 +36,7 @@ anim_diff_LDADD += libexample_util.la
 anim_diff_LDADD += ../imageio/libimageio_util.la
 anim_diff_LDADD += $(GIF_LIBS) -lm
 
-anim_dump_SOURCES = anim_dump.c anim_util.c anim_util.h
+anim_dump_SOURCES = anim_dump.c anim_util.c anim_util.h gifdec.c gifdec.h
 anim_dump_CPPFLAGS = $(AM_CPPFLAGS) $(PNG_INCLUDES)
 anim_dump_CPPFLAGS += $(GIF_INCLUDES)
 anim_dump_LDADD  =

examples/anim_diff.c

@@ -20,6 +20,7 @@
 #include <string.h>  // for 'strcmp'.
 
 #include "./anim_util.h"
+#include "./example_util.h"
 
 #if defined(_MSC_VER) && _MSC_VER < 1900
 #define snprintf _snprintf
@@ -143,8 +144,18 @@ static int CompareAnimatedImagePair(const AnimatedImage* const img1,
   if (!ok) return 0;  // These are fatal failures, can't proceed.
 
   if (is_multi_frame_image) {  // Checks relevant for multi-frame images only.
-    ok = CompareValues(img1->loop_count, img2->loop_count,
-                       "Loop count mismatch") && ok;
+    int max_loop_count_workaround = 0;
+    // Transcodes to webp increase the gif loop count by 1 for compatibility.
+    // When the gif has the maximum value the webp value will be off by one.
+    if ((img1->format == ANIM_GIF && img1->loop_count == 65536 &&
+         img2->format == ANIM_WEBP && img2->loop_count == 65535) ||
+        (img1->format == ANIM_WEBP && img1->loop_count == 65535 &&
+         img2->format == ANIM_GIF && img2->loop_count == 65536)) {
+      max_loop_count_workaround = 1;
+    }
+    ok = (max_loop_count_workaround ||
+          CompareValues(img1->loop_count, img2->loop_count,
+                        "Loop count mismatch")) && ok;
     ok = CompareBackgroundColor(img1->bgcolor, img2->bgcolor,
                                 premultiply) && ok;
   }
@@ -218,14 +229,7 @@ int main(int argc, const char* argv[]) {
       }
     } else if (!strcmp(argv[c], "-min_psnr")) {
       if (c < argc - 1) {
-        const char* const v = argv[++c];
-        char* end = NULL;
-        const double d = strtod(v, &end);
-        if (end == v) {
-          parse_error = 1;
-          fprintf(stderr, "Error! '%s' is not a floating point number.\n", v);
-        }
-        min_psnr = d;
+        min_psnr = ExUtilGetFloat(argv[++c], &parse_error);
       } else {
         parse_error = 1;
       }
@@ -233,14 +237,7 @@ int main(int argc, const char* argv[]) {
       premultiply = 0;
     } else if (!strcmp(argv[c], "-max_diff")) {
       if (c < argc - 1) {
-        const char* const v = argv[++c];
-        char* end = NULL;
-        const int n = (int)strtol(v, &end, 10);
-        if (end == v) {
-          parse_error = 1;
-          fprintf(stderr, "Error! '%s' is not an integer.\n", v);
-        }
-        max_diff = n;
+        max_diff = ExUtilGetInt(argv[++c], 0, &parse_error);
       } else {
         parse_error = 1;
       }

examples/anim_util.c

@@ -23,6 +23,7 @@
 #include "webp/decode.h"
 #include "webp/demux.h"
 #include "../imageio/imageio_util.h"
+#include "./gifdec.h"
 
 #if defined(_MSC_VER) && _MSC_VER < 1900
 #define snprintf _snprintf
@@ -275,6 +276,7 @@ static int ReadAnimatedWebP(const char filename[],
     prev_frame_timestamp = timestamp;
   }
   ok = dump_ok;
+  if (ok) image->format = ANIM_WEBP;
 
  End:
   WebPAnimDecoderDelete(dec);
@@ -369,26 +371,6 @@ static int DGifSavedExtensionToGCB(GifFileType* GifFile, int ImageIndex,
 #define DGifCloseFile(a, b) DGifCloseFile(a)
 #endif
 
-static void GIFDisplayError(const GifFileType* const gif, int gif_error) {
-  // libgif 4.2.0 has retired PrintGifError() and added GifErrorString().
-#if LOCAL_GIF_PREREQ(4, 2)
-#if LOCAL_GIF_PREREQ(5, 0)
-  const char* error_str =
-      GifErrorString((gif == NULL) ? gif_error : gif->Error);
-#else
-  const char* error_str = GifErrorString();
-  (void)gif;
-#endif
-  if (error_str == NULL) error_str = "Unknown error";
-  fprintf(stderr, "GIFLib Error %d: %s\n", gif_error, error_str);
-#else
-  (void)gif;
-  fprintf(stderr, "GIFLib Error %d: ", gif_error);
-  PrintGifError();
-  fprintf(stderr, "\n");
-#endif
-}
-
 static int IsKeyFrameGIF(const GifImageDesc* prev_desc, int prev_dispose,
                          const DecodedFrame* const prev_frame,
                          int canvas_width, int canvas_height) {
@@ -687,6 +669,7 @@ static int ReadAnimatedGIF(const char filename[], AnimatedImage* const image,
       }
     }
   }
+  image->format = ANIM_GIF;
   DGifCloseFile(gif, NULL);
   return 1;
 }

examples/anim_util.h

@@ -22,6 +22,11 @@
 extern "C" {
 #endif
 
+typedef enum {
+  ANIM_GIF,
+  ANIM_WEBP
+} AnimatedFileFormat;
+
 typedef struct {
   uint8_t* rgba;         // Decoded and reconstructed full frame.
   int duration;          // Frame duration in milliseconds.
@@ -29,6 +34,7 @@ typedef struct {
 } DecodedFrame;
 
 typedef struct {
+  AnimatedFileFormat format;
   uint32_t canvas_width;
   uint32_t canvas_height;
   uint32_t bgcolor;

examples/cwebp.c

@@ -188,7 +188,7 @@ static void PrintExtraInfoLossless(const WebPPicture* const pic,
     fprintf(stderr, "File:      %s\n", file_name);
     fprintf(stderr, "Dimension: %d x %d\n", pic->width, pic->height);
     fprintf(stderr, "Output:    %d bytes (%.2f bpp)\n", stats->coded_size,
-            1.f * stats->coded_size / pic->width / pic->height);
+            8.f * stats->coded_size / pic->width / pic->height);
     PrintFullLosslessInfo(stats, "ARGB");
   }
 }
@@ -213,7 +213,7 @@ static void PrintExtraInfoLossy(const WebPPicture* const pic, int short_output,
             "           (%.2f bpp)\n",
             stats->coded_size,
             stats->PSNR[0], stats->PSNR[1], stats->PSNR[2], stats->PSNR[3],
-            1.f * stats->coded_size / pic->width / pic->height);
+            8.f * stats->coded_size / pic->width / pic->height);
     if (total > 0) {
       int totals[4] = { 0, 0, 0, 0 };
       fprintf(stderr, "block count:  intra4:     %6d  (%.2f%%)\n"
@@ -1012,10 +1012,53 @@ int main(int argc, const char *argv[]) {
     }
   }
   if ((resize_w | resize_h) > 0) {
+    WebPPicture picture_no_alpha;
+    if (config.exact) {
+      // If -exact, we can't premultiply RGB by A otherwise RGB is lost if A=0.
+      // We rescale an opaque copy and assemble scaled A and non-premultiplied
+      // RGB channels. This is slower but it's a very uncommon use case. Color
+      // leak at sharp alpha edges is possible.
+      if (!WebPPictureCopy(&picture, &picture_no_alpha)) {
+        fprintf(stderr, "Error! Cannot copy temporary picture\n");
+        goto Error;
+      }
+
+      // We enforced picture.use_argb = 1 above. Now, remove the alpha values.
+      {
+        int x, y;
+        uint32_t* argb_no_alpha = picture_no_alpha.argb;
+        for (y = 0; y < picture_no_alpha.height; ++y) {
+          for (x = 0; x < picture_no_alpha.width; ++x) {
+            argb_no_alpha[x] |= 0xff000000;  // Opaque copy.
+          }
+          argb_no_alpha += picture_no_alpha.argb_stride;
+        }
+      }
+
+      if (!WebPPictureRescale(&picture_no_alpha, resize_w, resize_h)) {
+        fprintf(stderr, "Error! Cannot resize temporary picture\n");
+        goto Error;
+      }
+    }
+
     if (!WebPPictureRescale(&picture, resize_w, resize_h)) {
       fprintf(stderr, "Error! Cannot resize picture\n");
       goto Error;
     }
+
+    if (config.exact) {  // Put back the alpha information.
+      int x, y;
+      uint32_t* argb_no_alpha = picture_no_alpha.argb;
+      uint32_t* argb = picture.argb;
+      for (y = 0; y < picture_no_alpha.height; ++y) {
+        for (x = 0; x < picture_no_alpha.width; ++x) {
+          argb[x] = (argb[x] & 0xff000000) | (argb_no_alpha[x] & 0x00ffffff);
+        }
+        argb_no_alpha += picture_no_alpha.argb_stride;
+        argb += picture.argb_stride;
+      }
+      WebPPictureFree(&picture_no_alpha);
+    }
   }
   if (verbose && (crop != 0 || (resize_w | resize_h) > 0)) {
     const double preproc_time = StopwatchReadAndReset(&stop_watch);
@@ -1047,7 +1090,8 @@ int main(int argc, const char *argv[]) {
   // Write info
   if (dump_file) {
     if (picture.use_argb) {
-      fprintf(stderr, "Warning: can't dump file (-d option) in lossless mode.");
+      fprintf(stderr, "Warning: can't dump file (-d option) "
+                      "in lossless mode.\n");
     } else if (!DumpPicture(&picture, dump_file)) {
       fprintf(stderr, "Warning, couldn't dump picture %s\n", dump_file);
     }

examples/gif2webp.c

@@ -100,7 +100,6 @@ int main(int argc, const char *argv[]) {
   WebPMuxError err = WEBP_MUX_OK;
   int ok = 0;
   const char *in_file = NULL, *out_file = NULL;
-  FILE* out = NULL;
   GifFileType* gif = NULL;
   int frame_duration = 0;
   int frame_timestamp = 0;
@@ -478,7 +477,7 @@ int main(int argc, const char *argv[]) {
         stored_loop_count = 1;
         loop_count = 1;
       }
-    } else if (loop_count > 0) {
+    } else if (loop_count > 0 && loop_count < 65535) {
       // adapt GIF's semantic to WebP's (except in the infinite-loop case)
       loop_count += 1;
     }
@@ -578,7 +577,6 @@ int main(int argc, const char *argv[]) {
   WebPPictureFree(&curr_canvas);
   WebPPictureFree(&prev_canvas);
   WebPAnimEncoderDelete(enc);
-  if (out != NULL && out_file != NULL) fclose(out);
 
   if (gif_error != GIF_OK) {
     GIFDisplayError(gif, gif_error);

examples/img2webp.c

@@ -61,6 +61,10 @@ static void Help(void) {
   printf("\n");
   printf("example: img2webp -loop 2 in0.png -lossy in1.jpg\n"
          "                  -d 80 in2.tiff -o out.webp\n");
+  printf("\nNote: if a single file name is passed as the argument, the "
+         "arguments will be\n");
+  printf("tokenized from this file. The file name must not start with "
+         "the character '-'.\n");
 }
 
 //------------------------------------------------------------------------------

examples/stopwatch.h

@@ -60,4 +60,4 @@ static WEBP_INLINE double StopwatchReadAndReset(Stopwatch* watch) {
 
 #endif   /* _WIN32 */
 
-#endif  /* WEBP_EXAMPLES_STOPWATCH_H_ */
+#endif  // WEBP_EXAMPLES_STOPWATCH_H_

examples/vwebp.c

@@ -56,6 +56,7 @@ static struct {
   int print_info;
   int only_deltas;
   int use_color_profile;
+  int draw_anim_background_color;
 
   int canvas_width, canvas_height;
   int loop_count;
@@ -205,6 +206,11 @@ static void decode_callback(int what) {
         }
       }
       duration = curr->duration;
+      // Behavior copied from Chrome, cf:
+      // https://cs.chromium.org/chromium/src/third_party/WebKit/Source/
+      // platform/graphics/DeferredImageDecoder.cpp?
+      // rcl=b4c33049f096cd283f32be9a58b9a9e768227c26&l=246
+      if (duration <= 10) duration = 100;
     }
     if (!Decode()) {
       kParams.decoding_error = 1;
@@ -220,6 +226,9 @@ static void decode_callback(int what) {
 // Callbacks
 
 static void HandleKey(unsigned char key, int pos_x, int pos_y) {
+  // Note: rescaling the window or toggling some features during an animation
+  // generates visual artifacts. This is not fixed because refreshing the frame
+  // may require rendering the whole animation from start till current frame.
   (void)pos_x;
   (void)pos_y;
   if (key == 'q' || key == 'Q' || key == 27 /* Esc */) {
@@ -247,9 +256,11 @@ static void HandleKey(unsigned char key, int pos_x, int pos_y) {
         glutPostRedisplay();
       }
     }
+  } else if (key == 'b') {
+    kParams.draw_anim_background_color = 1 - kParams.draw_anim_background_color;
+    if (!kParams.has_animation) ClearPreviousFrame();
+    glutPostRedisplay();
   } else if (key == 'i') {
-    // Note: doesn't handle refresh of animation's last-frame (it's quite
-    // more involved to do, since you need to save the previous frame).
     kParams.print_info = 1 - kParams.print_info;
     if (!kParams.has_animation) ClearPreviousFrame();
     glutPostRedisplay();
@@ -281,7 +292,7 @@ static void PrintString(const char* const text) {
 }
 
 static float GetColorf(uint32_t color, int shift) {
-  return (color >> shift) / 255.f;
+  return ((color >> shift) & 0xff) / 255.f;
 }
 
 static void DrawCheckerBoard(void) {
@@ -304,6 +315,43 @@ static void DrawCheckerBoard(void) {
   glPopMatrix();
 }
 
+static void DrawBackground(void) {
+  // Whole window cleared with clear color, checkerboard rendered on top of it.
+  glClear(GL_COLOR_BUFFER_BIT);
+  DrawCheckerBoard();
+
+  // ANIM background color rendered (blend) on top. Default is white for still
+  // images (without ANIM chunk). glClear() can't be used for that (no blend).
+  if (kParams.draw_anim_background_color) {
+    glPushMatrix();
+    glLoadIdentity();
+    glColor4f(GetColorf(kParams.bg_color, 16),  // BGRA from spec
+              GetColorf(kParams.bg_color, 8),
+              GetColorf(kParams.bg_color, 0),
+              GetColorf(kParams.bg_color, 24));
+    glRecti(-1, -1, +1, +1);
+    glPopMatrix();
+  }
+}
+
+// Draw background in a scissored rectangle.
+static void DrawBackgroundScissored(int window_x, int window_y, int frame_w,
+                                    int frame_h) {
+  // Only update the requested area, not the whole canvas.
+  window_x = window_x * kParams.viewport_width / kParams.canvas_width;
+  window_y = window_y * kParams.viewport_height / kParams.canvas_height;
+  frame_w = frame_w * kParams.viewport_width / kParams.canvas_width;
+  frame_h = frame_h * kParams.viewport_height / kParams.canvas_height;
+
+  // glScissor() takes window coordinates (0,0 at bottom left).
+  window_y = kParams.viewport_height - window_y - frame_h;
+
+  glEnable(GL_SCISSOR_TEST);
+  glScissor(window_x, window_y, frame_w, frame_h);
+  DrawBackground();
+  glDisable(GL_SCISSOR_TEST);
+}
+
 static void HandleDisplay(void) {
   const WebPDecBuffer* const pic = kParams.pic;
   const WebPIterator* const curr = &kParams.curr_frame;
@@ -320,38 +368,21 @@ static void HandleDisplay(void) {
   glPixelStorei(GL_UNPACK_ROW_LENGTH, pic->u.RGBA.stride / 4);
 
   if (kParams.only_deltas) {
-    DrawCheckerBoard();
-  } else if (prev->dispose_method == WEBP_MUX_DISPOSE_BACKGROUND ||
-      curr->blend_method == WEBP_MUX_NO_BLEND) {
-    // glScissor() takes window coordinates (0,0 at bottom left).
-    int window_x, window_y;
-    int frame_w, frame_h;
+    DrawBackground();
+  } else {
+    // The rectangle of the previous frame might be different than the current
+    // frame, so we may need to DrawBackgroundScissored for both.
     if (prev->dispose_method == WEBP_MUX_DISPOSE_BACKGROUND) {
       // Clear the previous frame rectangle.
-      window_x = prev->x_offset;
-      window_y = kParams.canvas_height - prev->y_offset - prev->height;
-      frame_w = prev->width;
-      frame_h = prev->height;
-    } else {  // curr->blend_method == WEBP_MUX_NO_BLEND.
-      // We simulate no-blending behavior by first clearing the current frame
-      // rectangle (to a checker-board) and then alpha-blending against it.
-      window_x = curr->x_offset;
-      window_y = kParams.canvas_height - curr->y_offset - curr->height;
-      frame_w = curr->width;
-      frame_h = curr->height;
+      DrawBackgroundScissored(prev->x_offset, prev->y_offset, prev->width,
+                              prev->height);
+    }
+    if (curr->blend_method == WEBP_MUX_NO_BLEND) {
+      // We simulate no-blending behavior by first clearing the current frame
+      // rectangle and then alpha-blending against it.
+      DrawBackgroundScissored(curr->x_offset, curr->y_offset, curr->width,
+                              curr->height);
     }
-    glEnable(GL_SCISSOR_TEST);
-    // Only update the requested area, not the whole canvas.
-    window_x = window_x * kParams.viewport_width / kParams.canvas_width;
-    window_y = window_y * kParams.viewport_height / kParams.canvas_height;
-    frame_w = frame_w * kParams.viewport_width / kParams.canvas_width;
-    frame_h = frame_h * kParams.viewport_height / kParams.canvas_height;
-    glScissor(window_x, window_y, frame_w, frame_h);
-
-    glClear(GL_COLOR_BUFFER_BIT);  // use clear color
-    DrawCheckerBoard();
-
-    glDisable(GL_SCISSOR_TEST);
   }
 
   *prev = *curr;
@@ -403,37 +434,35 @@ static void StartDisplay(void) {
   glutKeyboardFunc(HandleKey);
   glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
   glEnable(GL_BLEND);
-  glClearColor(GetColorf(kParams.bg_color, 0),
-               GetColorf(kParams.bg_color, 8),
-               GetColorf(kParams.bg_color, 16),
-               GetColorf(kParams.bg_color, 24));
-  glClear(GL_COLOR_BUFFER_BIT);
-  DrawCheckerBoard();
+  glClearColor(0, 0, 0, 0);  // window will be cleared to black (no blend)
+  DrawBackground();
 }
 
 //------------------------------------------------------------------------------
 // Main
 
 static void Help(void) {
-  printf("Usage: vwebp in_file [options]\n\n"
-         "Decodes the WebP image file and visualize it using OpenGL\n"
-         "Options are:\n"
-         "  -version ..... print version number and exit\n"
-         "  -noicc ....... don't use the icc profile if present\n"
-         "  -nofancy ..... don't use the fancy YUV420 upscaler\n"
-         "  -nofilter .... disable in-loop filtering\n"
-         "  -dither <int>  dithering strength (0..100), default=50\n"
-         "  -noalphadither disable alpha plane dithering\n"
-         "  -mt .......... use multi-threading\n"
-         "  -info ........ print info\n"
-         "  -h ........... this help message\n"
-         "\n"
-         "Keyboard shortcuts:\n"
-         "  'c' ................ toggle use of color profile\n"
-         "  'i' ................ overlay file information\n"
-         "  'd' ................ disable blending & disposal (debug)\n"
-         "  'q' / 'Q' / ESC .... quit\n"
-        );
+  printf(
+      "Usage: vwebp in_file [options]\n\n"
+      "Decodes the WebP image file and visualize it using OpenGL\n"
+      "Options are:\n"
+      "  -version ..... print version number and exit\n"
+      "  -noicc ....... don't use the icc profile if present\n"
+      "  -nofancy ..... don't use the fancy YUV420 upscaler\n"
+      "  -nofilter .... disable in-loop filtering\n"
+      "  -dither <int>  dithering strength (0..100), default=50\n"
+      "  -noalphadither disable alpha plane dithering\n"
+      "  -usebgcolor .. display background color\n"
+      "  -mt .......... use multi-threading\n"
+      "  -info ........ print info\n"
+      "  -h ........... this help message\n"
+      "\n"
+      "Keyboard shortcuts:\n"
+      "  'c' ................ toggle use of color profile\n"
+      "  'b' ................ toggle background color display\n"
+      "  'i' ................ overlay file information\n"
+      "  'd' ................ disable blending & disposal (debug)\n"
+      "  'q' / 'Q' / ESC .... quit\n");
 }
 
 int main(int argc, char *argv[]) {
@@ -448,6 +477,8 @@ int main(int argc, char *argv[]) {
   config->options.dithering_strength = 50;
   config->options.alpha_dithering_strength = 100;
   kParams.use_color_profile = 1;
+  // Background color hidden by default to see transparent areas.
+  kParams.draw_anim_background_color = 0;
 
   for (c = 1; c < argc; ++c) {
     int parse_error = 0;
@@ -462,6 +493,8 @@ int main(int argc, char *argv[]) {
       config->options.bypass_filtering = 1;
     } else if (!strcmp(argv[c], "-noalphadither")) {
       config->options.alpha_dithering_strength = 0;
+    } else if (!strcmp(argv[c], "-usebgcolor")) {
+      kParams.draw_anim_background_color = 1;
     } else if (!strcmp(argv[c], "-dither") && c + 1 < argc) {
       config->options.dithering_strength =
           ExUtilGetInt(argv[++c], 0, &parse_error);

examples/webpmux.c

@@ -595,18 +595,22 @@ static int ValidateCommandLine(const CommandLineArguments* const cmd_args,
 
 #define FEATURETYPE_IS_NIL (config->type_ == NIL_FEATURE)
 
-#define CHECK_NUM_ARGS_LESS(NUM, LABEL)                                  \
+#define CHECK_NUM_ARGS_AT_LEAST(NUM, LABEL)                              \
   if (argc < i + (NUM)) {                                                \
     fprintf(stderr, "ERROR: Too few arguments for '%s'.\n", argv[i]);    \
     goto LABEL;                                                          \
   }
 
-#define CHECK_NUM_ARGS_NOT_EQUAL(NUM, LABEL)                             \
-  if (argc != i + (NUM)) {                                               \
+#define CHECK_NUM_ARGS_AT_MOST(NUM, LABEL)                               \
+  if (argc > i + (NUM)) {                                                \
     fprintf(stderr, "ERROR: Too many arguments for '%s'.\n", argv[i]);   \
     goto LABEL;                                                          \
   }
 
+#define CHECK_NUM_ARGS_EXACTLY(NUM, LABEL)                               \
+  CHECK_NUM_ARGS_AT_LEAST(NUM, LABEL);                                   \
+  CHECK_NUM_ARGS_AT_MOST(NUM, LABEL);
+
 // Parses command-line arguments to fill up config object. Also performs some
 // semantic checks.
 static int ParseCommandLine(Config* config) {
@@ -627,7 +631,7 @@ static int ParseCommandLine(Config* config) {
         }
         ++i;
       } else if (!strcmp(argv[i], "-duration")) {
-        CHECK_NUM_ARGS_LESS(2, ErrParse);
+        CHECK_NUM_ARGS_AT_LEAST(2, ErrParse);
         if (ACTION_IS_NIL || config->action_type_ == ACTION_DURATION) {
           config->action_type_ = ACTION_DURATION;
         } else {
@@ -657,7 +661,7 @@ static int ParseCommandLine(Config* config) {
         }
         ++i;
       } else if (!strcmp(argv[i], "-frame")) {
-        CHECK_NUM_ARGS_LESS(3, ErrParse);
+        CHECK_NUM_ARGS_AT_LEAST(3, ErrParse);
         if (ACTION_IS_NIL || config->action_type_ == ACTION_SET) {
           config->action_type_ = ACTION_SET;
         } else {
@@ -674,7 +678,7 @@ static int ParseCommandLine(Config* config) {
         ++feature_arg_index;
         i += 3;
       } else if (!strcmp(argv[i], "-loop") || !strcmp(argv[i], "-bgcolor")) {
-        CHECK_NUM_ARGS_LESS(2, ErrParse);
+        CHECK_NUM_ARGS_AT_LEAST(2, ErrParse);
         if (ACTION_IS_NIL || config->action_type_ == ACTION_SET) {
           config->action_type_ = ACTION_SET;
         } else {
@@ -691,11 +695,11 @@ static int ParseCommandLine(Config* config) {
         ++feature_arg_index;
         i += 2;
       } else if (!strcmp(argv[i], "-o")) {
-        CHECK_NUM_ARGS_LESS(2, ErrParse);
+        CHECK_NUM_ARGS_AT_LEAST(2, ErrParse);
         config->output_ = argv[i + 1];
         i += 2;
       } else if (!strcmp(argv[i], "-info")) {
-        CHECK_NUM_ARGS_NOT_EQUAL(2, ErrParse);
+        CHECK_NUM_ARGS_EXACTLY(2, ErrParse);
         if (config->action_type_ != NIL_ACTION) {
           ERROR_GOTO1("ERROR: Multiple actions specified.\n", ErrParse);
         } else {
@@ -742,7 +746,7 @@ static int ParseCommandLine(Config* config) {
           ERROR_GOTO1("ERROR: Multiple features specified.\n", ErrParse);
         }
         if (config->action_type_ == ACTION_SET) {
-          CHECK_NUM_ARGS_LESS(2, ErrParse);
+          CHECK_NUM_ARGS_AT_LEAST(2, ErrParse);
           arg->filename_ = argv[i + 1];
           ++feature_arg_index;
           i += 2;
@@ -751,7 +755,7 @@ static int ParseCommandLine(Config* config) {
         }
       } else if (!strcmp(argv[i], "frame") &&
                  (config->action_type_ == ACTION_GET)) {
-        CHECK_NUM_ARGS_LESS(2, ErrParse);
+        CHECK_NUM_ARGS_AT_LEAST(2, ErrParse);
         config->type_ = FEATURE_ANMF;
         arg->params_ = argv[i + 1];
         ++feature_arg_index;
@@ -836,8 +840,9 @@ static int InitializeConfig(int argc, const char* argv[],
 
 #undef ACTION_IS_NIL
 #undef FEATURETYPE_IS_NIL
-#undef CHECK_NUM_ARGS_LESS
-#undef CHECK_NUM_ARGS_MORE
+#undef CHECK_NUM_ARGS_AT_LEAST
+#undef CHECK_NUM_ARGS_AT_MOST
+#undef CHECK_NUM_ARGS_EXACTLY
 
 //------------------------------------------------------------------------------
 // Processing.

extras/Makefile.am

@@ -14,7 +14,7 @@ libwebpextras_la_LIBADD = ../src/libwebp.la
 
 noinst_PROGRAMS =
 noinst_PROGRAMS += webp_quality
-if WANT_DEMUX
+if BUILD_DEMUX
   noinst_PROGRAMS += get_disto
 endif
 if BUILD_VWEBP_SDL

extras/extras.c

@@ -18,7 +18,7 @@
 
 #define XTRA_MAJ_VERSION 1
 #define XTRA_MIN_VERSION 0
-#define XTRA_REV_VERSION 0
+#define XTRA_REV_VERSION 1
 
 //------------------------------------------------------------------------------
 
@@ -48,13 +48,14 @@ int WebPImportGray(const uint8_t* gray_data, WebPPicture* pic) {
 
 int WebPImportRGB565(const uint8_t* rgb565, WebPPicture* pic) {
   int x, y;
+  uint32_t* dst;
   if (pic == NULL || rgb565 == NULL) return 0;
   pic->colorspace = WEBP_YUV420;
   pic->use_argb = 1;
   if (!WebPPictureAlloc(pic)) return 0;
+  dst = pic->argb;
   for (y = 0; y < pic->height; ++y) {
     const int width = pic->width;
-    uint32_t* dst = pic->argb + y * pic->argb_stride;
     for (x = 0; x < width; ++x) {
 #ifdef WEBP_SWAP_16BIT_CSP
       const uint32_t rg = rgb565[2 * x + 1];
@@ -70,22 +71,24 @@ int WebPImportRGB565(const uint8_t* rgb565, WebPPicture* pic) {
       r = r | (r >> 5);
       g = g | (g >> 6);
       b = b | (b >> 5);
-      dst[x] = (r << 16) | (g << 8) | b;
+      dst[x] = (0xffu << 24) | (r << 16) | (g << 8) | b;
     }
     rgb565 += 2 * width;
+    dst += pic->argb_stride;
   }
   return 1;
 }
 
 int WebPImportRGB4444(const uint8_t* rgb4444, WebPPicture* pic) {
   int x, y;
+  uint32_t* dst;
   if (pic == NULL || rgb4444 == NULL) return 0;
   pic->colorspace = WEBP_YUV420;
   pic->use_argb = 1;
   if (!WebPPictureAlloc(pic)) return 0;
+  dst = pic->argb;
   for (y = 0; y < pic->height; ++y) {
     const int width = pic->width;
-    uint32_t* dst = pic->argb + y * pic->argb_stride;
     for (x = 0; x < width; ++x) {
 #ifdef WEBP_SWAP_16BIT_CSP
       const uint32_t rg = rgb4444[2 * x + 1];
@@ -106,6 +109,7 @@ int WebPImportRGB4444(const uint8_t* rgb4444, WebPPicture* pic) {
       dst[x] = (a << 24) | (r << 16) | (g << 8) | b;
     }
     rgb4444 += 2 * width;
+    dst += pic->argb_stride;
   }
   return 1;
 }

extras/extras.h

@@ -67,4 +67,4 @@ WEBP_EXTERN int VP8EstimateQuality(const uint8_t* const data, size_t size);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_EXTRAS_EXTRAS_H_ */
+#endif  // WEBP_EXTRAS_EXTRAS_H_

extras/get_disto.c

@@ -293,7 +293,7 @@ int main(int argc, const char *argv[]) {
   printf("%u %.2f    %.2f %.2f %.2f %.2f [ %.2f bpp ]\n",
          (unsigned int)size1,
          disto[4], disto[0], disto[1], disto[2], disto[3],
-         1.f * size1 / pic1.width / pic1.height);
+         8.f * size1 / pic1.width / pic1.height);
 
   if (output != NULL) {
     uint8_t* data = NULL;

extras/webp_to_sdl.c

@@ -40,7 +40,7 @@ int WebpToSDL(const char* data, unsigned int data_size) {
 
   if (!WebPInitDecoderConfig(&config)) {
     fprintf(stderr, "Library version mismatch!\n");
-    return 1;
+    return 0;
   }
 
   if (!init_ok) {

imageio/Makefile.am

@@ -1,7 +1,7 @@
 AM_CPPFLAGS += -I$(top_builddir)/src -I$(top_srcdir)/src
 noinst_LTLIBRARIES =
 noinst_LTLIBRARIES += libimageio_util.la
-if WANT_DEMUX
+if BUILD_DEMUX
   noinst_LTLIBRARIES += libimagedec.la
 endif
 noinst_LTLIBRARIES += libimageenc.la

imageio/pngdec.c

@@ -18,6 +18,9 @@
 #include <stdio.h>
 
 #ifdef WEBP_HAVE_PNG
+#ifndef PNG_USER_MEM_SUPPORTED
+#define PNG_USER_MEM_SUPPORTED  // for png_create_read_struct_2
+#endif
 #include <png.h>
 #include <setjmp.h>   // note: this must be included *after* png.h
 #include <stdlib.h>
@@ -27,11 +30,33 @@
 #include "./imageio_util.h"
 #include "./metadata.h"
 
+#define LOCAL_PNG_VERSION ((PNG_LIBPNG_VER_MAJOR << 8) | PNG_LIBPNG_VER_MINOR)
+#define LOCAL_PNG_PREREQ(maj, min) \
+   (LOCAL_PNG_VERSION >= (((maj) << 8) | (min)))
+
 static void PNGAPI error_function(png_structp png, png_const_charp error) {
   if (error != NULL) fprintf(stderr, "libpng error: %s\n", error);
   longjmp(png_jmpbuf(png), 1);
 }
 
+#if LOCAL_PNG_PREREQ(1,4)
+typedef png_alloc_size_t LocalPngAllocSize;
+#else
+typedef png_size_t LocalPngAllocSize;
+#endif
+
+static png_voidp MallocFunc(png_structp png_ptr, LocalPngAllocSize size) {
+  (void)png_ptr;
+  if (size != (size_t)size) return NULL;
+  if (!ImgIoUtilCheckSizeArgumentsOverflow(size, 1)) return NULL;
+  return (png_voidp)malloc((size_t)size);
+}
+
+static void FreeFunc(png_structp png_ptr, png_voidp ptr) {
+  (void)png_ptr;
+  free(ptr);
+}
+
 // Converts the NULL terminated 'hexstring' which contains 2-byte character
 // representations of hex values to raw data.
 // 'hexstring' may contain values consisting of [A-F][a-f][0-9] in pairs,
@@ -171,11 +196,10 @@ static int ExtractMetadataFromPNG(png_structp png,
     {
       png_charp name;
       int comp_type;
-#if ((PNG_LIBPNG_VER_MAJOR << 8) | PNG_LIBPNG_VER_MINOR << 0) < \
-    ((1 << 8) | (5 << 0))
-      png_charp profile;
-#else  // >= libpng 1.5.0
+#if LOCAL_PNG_PREREQ(1,5)
       png_bytep profile;
+#else
+      png_charp profile;
 #endif
       png_uint_32 len;
 
@@ -224,7 +248,8 @@ int ReadPNG(const uint8_t* const data, size_t data_size,
   context.data = data;
   context.data_size = data_size;
 
-  png = png_create_read_struct(PNG_LIBPNG_VER_STRING, 0, 0, 0);
+  png = png_create_read_struct_2(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL,
+                                 NULL, MallocFunc, FreeFunc);
   if (png == NULL) goto End;
 
   png_set_error_fn(png, 0, error_function, NULL);

imageio/tiffdec.c

@@ -121,7 +121,6 @@ static tsize_t MyRead(thandle_t opaque, void* dst, tsize_t size) {
 // (we don't want to force a dependency to a libdspdec library).
 #define MFIX 24    // 24bit fixed-point arithmetic
 #define HALF ((1u << MFIX) >> 1)
-#define KINV_255 ((1u << MFIX) / 255u)
 
 static uint32_t Unmult(uint8_t x, uint32_t mult) {
   const uint32_t v = (x * mult + HALF) >> MFIX;
@@ -132,6 +131,9 @@ static WEBP_INLINE uint32_t GetScale(uint32_t a) {
   return (255u << MFIX) / a;
 }
 
+#undef MFIX
+#undef HALF
+
 static void MultARGBRow(uint8_t* ptr, int width) {
   int x;
   for (x = 0; x < width; ++x, ptr += 4) {

makefile.unix

@@ -89,12 +89,6 @@ EXTRA_FLAGS += -DWEBP_HAVE_SSE41
 src/dsp/%_sse41.o: EXTRA_FLAGS += -msse4.1
 endif
 
-# AVX2-specific flags:
-ifeq ($(HAVE_AVX2), 1)
-EXTRA_FLAGS += -DWEBP_HAVE_AVX2
-src/dsp/%_avx2.o: EXTRA_FLAGS += -mavx2
-endif
-
 # NEON-specific flags:
 # EXTRA_FLAGS += -march=armv7-a -mfloat-abi=hard -mfpu=neon -mtune=cortex-a8
 # -> seems to make the overall lib slower: -fno-split-wide-types
@@ -112,7 +106,12 @@ endif
 AR = ar
 ARFLAGS = r
 CPPFLAGS = -I. -Isrc/ -Wall
-CFLAGS = -O3 -DNDEBUG $(EXTRA_FLAGS)
+ifeq ($(DEBUG), 1)
+  CFLAGS = -g
+else
+  CFLAGS = -O3 -DNDEBUG
+endif
+CFLAGS += $(EXTRA_FLAGS)
 CC = gcc
 INSTALL = install
 GROFF = /usr/bin/groff
@@ -188,7 +187,6 @@ DSP_ENC_OBJS = \
     src/dsp/cost_mips_dsp_r2.o \
     src/dsp/cost_sse2.o \
     src/dsp/enc.o \
-    src/dsp/enc_avx2.o \
     src/dsp/enc_mips32.o \
     src/dsp/enc_mips_dsp_r2.o \
     src/dsp/enc_msa.o \
@@ -387,7 +385,7 @@ src/demux/libwebpdemux.a: $(LIBWEBPDEMUX_OBJS)
 	$(AR) $(ARFLAGS) $@ $^
 
 examples/anim_diff: examples/anim_diff.o $(ANIM_UTIL_OBJS) $(GIFDEC_OBJS)
-examples/anim_dump: examples/anim_dump.o $(ANIM_UTIL_OBJS)
+examples/anim_dump: examples/anim_dump.o $(ANIM_UTIL_OBJS) $(GIFDEC_OBJS)
 examples/cwebp: examples/cwebp.o
 examples/dwebp: examples/dwebp.o
 examples/gif2webp: examples/gif2webp.o $(GIFDEC_OBJS)
@@ -401,7 +399,7 @@ examples/anim_diff: src/demux/libwebpdemux.a examples/libexample_util.a
 examples/anim_diff: imageio/libimageio_util.a src/libwebp.a
 examples/anim_diff: EXTRA_LIBS += $(GIF_LIBS)
 examples/anim_diff: EXTRA_FLAGS += -DWEBP_HAVE_GIF
-examples/anim_dump: examples/libanim_util.a
+examples/anim_dump: examples/libanim_util.a examples/libgifdec.a
 examples/anim_dump: src/demux/libwebpdemux.a
 examples/anim_dump: examples/libexample_util.a
 examples/anim_dump: imageio/libimageio_util.a

man/Makefile.am

@@ -1,10 +1,13 @@
 man_MANS = cwebp.1 dwebp.1
-if WANT_MUX
+if BUILD_MUX
   man_MANS += webpmux.1
 endif
 if BUILD_GIF2WEBP
   man_MANS += gif2webp.1
 endif
+if BUILD_IMG2WEBP
+  man_MANS += img2webp.1
+endif
 if BUILD_VWEBP
   man_MANS += vwebp.1
 endif

man/vwebp.1

@@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH VWEBP 1 "November 25, 2016"
+.TH VWEBP 1 "July 20, 2018"
 .SH NAME
 vwebp \- decompress a WebP file and display it in a window
 .SH SYNOPSIS
@@ -38,6 +38,10 @@ It helps by smoothing gradients and avoiding banding artifacts. Default: 50.
 By default, quantized transparency planes are dithered during decompression,
 to smooth the gradients. This flag will prevent this dithering.
 .TP
+.B \-usebgcolor
+Fill transparent areas with the bitstream's own background color instead of
+checkerboard only. Default is white for non-animated images.
+.TP
 .B \-mt
 Use multi-threading for decoding, if possible.
 .TP
@@ -56,6 +60,9 @@ the data will be read from \fIstdin\fP instead of a file.
 .B 'c'
 Toggle use of color profile.
 .TP
+.B 'b'
+Toggle display of background color.
+.TP
 .B 'i'
 Overlay file information.
 .TP

man/webpmux.1

@@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH WEBPMUX 1 "December 1, 2017"
+.TH WEBPMUX 1 "April 23, 2018"
 .SH NAME
 webpmux \- create animated WebP files from non\-animated WebP images, extract
 frames from animated WebP images, and manage XMP/EXIF metadata and ICC profile.
@@ -58,7 +58,7 @@ command.
 \fBwebpmux\fP can be used to create/extract from animated WebP files, as well as
 to add/extract/strip XMP/EXIF metadata and ICC profile.
 If a single file name (not starting with the character '\-') is supplied as
-the argument, the command line argument are actually tokenized from this file.
+the argument, the command line arguments are actually tokenized from this file.
 This allows for easy scripting or using large number of arguments.
 .SH OPTIONS
 .SS GET_OPTIONS (\-get):

src/Makefile.am

@@ -1,10 +1,10 @@
 # The mux and demux libraries depend on libwebp, thus the '.' to force
 # the build order so it's available to them.
 SUBDIRS = dec enc dsp utils .
-if WANT_MUX
+if BUILD_MUX
   SUBDIRS += mux
 endif
-if WANT_DEMUX
+if BUILD_DEMUX
   SUBDIRS += demux
 endif
 
@@ -36,7 +36,7 @@ libwebp_la_LIBADD += utils/libwebputils.la
 # other than the ones listed on the command line, i.e., after linking, it will
 # not have unresolved symbols. Some platforms (Windows among them) require all
 # symbols in shared libraries to be resolved at library creation.
-libwebp_la_LDFLAGS = -no-undefined -version-info 7:2:0
+libwebp_la_LDFLAGS = -no-undefined -version-info 7:3:0
 libwebpincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebp.pc
 
@@ -48,7 +48,7 @@ if BUILD_LIBWEBPDECODER
   libwebpdecoder_la_LIBADD += dsp/libwebpdspdecode.la
   libwebpdecoder_la_LIBADD += utils/libwebputilsdecode.la
 
-  libwebpdecoder_la_LDFLAGS = -no-undefined -version-info 3:2:0
+  libwebpdecoder_la_LDFLAGS = -no-undefined -version-info 3:3:0
   pkgconfig_DATA += libwebpdecoder.pc
 endif
 

src/dec/alphai_dec.h

@@ -51,4 +51,4 @@ void WebPDeallocateAlphaMemory(VP8Decoder* const dec);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_ALPHAI_DEC_H_ */
+#endif  // WEBP_DEC_ALPHAI_DEC_H_

src/dec/buffer_dec.c

@@ -74,7 +74,8 @@ static VP8StatusCode CheckDecBuffer(const WebPDecBuffer* const buffer) {
   } else {    // RGB checks
     const WebPRGBABuffer* const buf = &buffer->u.RGBA;
     const int stride = abs(buf->stride);
-    const uint64_t size = MIN_BUFFER_SIZE(width, height, stride);
+    const uint64_t size =
+        MIN_BUFFER_SIZE(width * kModeBpp[mode], height, stride);
     ok &= (size <= buf->size);
     ok &= (stride >= width * kModeBpp[mode]);
     ok &= (buf->rgba != NULL);

src/dec/common_dec.h

@@ -51,4 +51,4 @@ enum { MB_FEATURE_TREE_PROBS = 3,
        NUM_PROBAS = 11
      };
 
-#endif    // WEBP_DEC_COMMON_DEC_H_
+#endif  // WEBP_DEC_COMMON_DEC_H_

src/dec/frame_dec.c

@@ -338,7 +338,6 @@ void VP8InitDithering(const WebPDecoderOptions* const options,
       for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
         VP8QuantMatrix* const dqm = &dec->dqm_[s];
         if (dqm->uv_quant_ < DITHER_AMP_TAB_SIZE) {
-          // TODO(skal): should we specially dither more for uv_quant_ < 0?
           const int idx = (dqm->uv_quant_ < 0) ? 0 : dqm->uv_quant_;
           dqm->dither_ = (f * kQuantToDitherAmp[idx]) >> 3;
         }
@@ -669,15 +668,9 @@ int VP8GetThreadMethod(const WebPDecoderOptions* const options,
   (void)height;
   assert(headers == NULL || !headers->is_lossless);
 #if defined(WEBP_USE_THREAD)
-  if (width < MIN_WIDTH_FOR_THREADS) return 0;
-  // TODO(skal): tune the heuristic further
-#if 0
-  if (height < 2 * width) return 2;
+  if (width >= MIN_WIDTH_FOR_THREADS) return 2;
 #endif
-  return 2;
-#else   // !WEBP_USE_THREAD
   return 0;
-#endif
 }
 
 #undef MT_CACHE_LINES

src/dec/idec_dec.c

@@ -140,10 +140,9 @@ static void DoRemap(WebPIDecoder* const idec, ptrdiff_t offset) {
       if (NeedCompressedAlpha(idec)) {
         ALPHDecoder* const alph_dec = dec->alph_dec_;
         dec->alpha_data_ += offset;
-        if (alph_dec != NULL) {
+        if (alph_dec != NULL && alph_dec->vp8l_dec_ != NULL) {
           if (alph_dec->method_ == ALPHA_LOSSLESS_COMPRESSION) {
             VP8LDecoder* const alph_vp8l_dec = alph_dec->vp8l_dec_;
-            assert(alph_vp8l_dec != NULL);
             assert(dec->alpha_data_size_ >= ALPHA_HEADER_LEN);
             VP8LBitReaderSetBuffer(&alph_vp8l_dec->br_,
                                    dec->alpha_data_ + ALPHA_HEADER_LEN,
@@ -283,10 +282,8 @@ static void RestoreContext(const MBContext* context, VP8Decoder* const dec,
 
 static VP8StatusCode IDecError(WebPIDecoder* const idec, VP8StatusCode error) {
   if (idec->state_ == STATE_VP8_DATA) {
-    VP8Io* const io = &idec->io_;
-    if (io->teardown != NULL) {
-      io->teardown(io);
-    }
+    // Synchronize the thread, clean-up and check for errors.
+    VP8ExitCritical((VP8Decoder*)idec->dec_, &idec->io_);
   }
   idec->state_ = STATE_ERROR;
   return error;
@@ -451,7 +448,10 @@ static VP8StatusCode DecodeRemaining(WebPIDecoder* const idec) {
   VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
   VP8Io* const io = &idec->io_;
 
-  assert(dec->ready_);
+  // Make sure partition #0 has been read before, to set dec to ready_.
+  if (!dec->ready_) {
+    return IDecError(idec, VP8_STATUS_BITSTREAM_ERROR);
+  }
   for (; dec->mb_y_ < dec->mb_h_; ++dec->mb_y_) {
     if (idec->last_mb_y_ != dec->mb_y_) {
       if (!VP8ParseIntraModeRow(&dec->br_, dec)) {
@@ -473,6 +473,12 @@ static VP8StatusCode DecodeRemaining(WebPIDecoder* const idec) {
             MemDataSize(&idec->mem_) > MAX_MB_SIZE) {
           return IDecError(idec, VP8_STATUS_BITSTREAM_ERROR);
         }
+        // Synchronize the threads.
+        if (dec->mt_method_ > 0) {
+          if (!WebPGetWorkerInterface()->Sync(&dec->worker_)) {
+            return IDecError(idec, VP8_STATUS_BITSTREAM_ERROR);
+          }
+        }
         RestoreContext(&context, dec, token_br);
         return VP8_STATUS_SUSPENDED;
       }
@@ -491,6 +497,7 @@ static VP8StatusCode DecodeRemaining(WebPIDecoder* const idec) {
   }
   // Synchronize the thread and check for errors.
   if (!VP8ExitCritical(dec, io)) {
+    idec->state_ = STATE_ERROR;  // prevent re-entry in IDecError
     return IDecError(idec, VP8_STATUS_USER_ABORT);
   }
   dec->ready_ = 0;
@@ -571,6 +578,10 @@ static VP8StatusCode IDecode(WebPIDecoder* idec) {
     status = DecodePartition0(idec);
   }
   if (idec->state_ == STATE_VP8_DATA) {
+    const VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
+    if (dec == NULL) {
+      return VP8_STATUS_SUSPENDED;  // can't continue if we have no decoder.
+    }
     status = DecodeRemaining(idec);
   }
   if (idec->state_ == STATE_VP8L_HEADER) {

src/dec/vp8_dec.c

@@ -491,7 +491,7 @@ static int GetCoeffsAlt(VP8BitReader* const br,
   return 16;
 }
 
-WEBP_TSAN_IGNORE_FUNCTION static void InitGetCoeffs(void) {
+static WEBP_TSAN_IGNORE_FUNCTION void InitGetCoeffs(void) {
   if (GetCoeffs == NULL) {
     if (VP8GetCPUInfo != NULL && VP8GetCPUInfo(kSlowSSSE3)) {
       GetCoeffs = GetCoeffsAlt;

src/dec/vp8_dec.h

@@ -182,4 +182,4 @@ WEBP_EXTERN int VP8LGetInfo(
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_VP8_DEC_H_ */
+#endif  // WEBP_DEC_VP8_DEC_H_

src/dec/vp8i_dec.h

@@ -32,7 +32,7 @@ extern "C" {
 // version numbers
 #define DEC_MAJ_VERSION 1
 #define DEC_MIN_VERSION 0
-#define DEC_REV_VERSION 0
+#define DEC_REV_VERSION 1
 
 // YUV-cache parameters. Cache is 32-bytes wide (= one cacheline).
 // Constraints are: We need to store one 16x16 block of luma samples (y),
@@ -316,4 +316,4 @@ const uint8_t* VP8DecompressAlphaRows(VP8Decoder* const dec,
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_VP8I_DEC_H_ */
+#endif  // WEBP_DEC_VP8I_DEC_H_

src/dec/vp8l_dec.c

@@ -362,12 +362,19 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
   VP8LMetadata* const hdr = &dec->hdr_;
   uint32_t* huffman_image = NULL;
   HTreeGroup* htree_groups = NULL;
+  // When reading htrees, some might be unused, as the format allows it.
+  // We will still read them but put them in this htree_group_bogus.
+  HTreeGroup htree_group_bogus;
   HuffmanCode* huffman_tables = NULL;
+  HuffmanCode* huffman_tables_bogus = NULL;
   HuffmanCode* next = NULL;
   int num_htree_groups = 1;
+  int num_htree_groups_max = 1;
   int max_alphabet_size = 0;
   int* code_lengths = NULL;
   const int table_size = kTableSize[color_cache_bits];
+  int* mapping = NULL;
+  int ok = 0;
 
   if (allow_recursion && VP8LReadBits(br, 1)) {
     // use meta Huffman codes.
@@ -384,10 +391,42 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
       // The huffman data is stored in red and green bytes.
       const int group = (huffman_image[i] >> 8) & 0xffff;
       huffman_image[i] = group;
-      if (group >= num_htree_groups) {
-        num_htree_groups = group + 1;
+      if (group >= num_htree_groups_max) {
+        num_htree_groups_max = group + 1;
       }
     }
+    // Check the validity of num_htree_groups_max. If it seems too big, use a
+    // smaller value for later. This will prevent big memory allocations to end
+    // up with a bad bitstream anyway.
+    // The value of 1000 is totally arbitrary. We know that num_htree_groups_max
+    // is smaller than (1 << 16) and should be smaller than the number of pixels
+    // (though the format allows it to be bigger).
+    if (num_htree_groups_max > 1000 || num_htree_groups_max > xsize * ysize) {
+      // Create a mapping from the used indices to the minimal set of used
+      // values [0, num_htree_groups)
+      mapping = (int*)WebPSafeMalloc(num_htree_groups_max, sizeof(*mapping));
+      if (mapping == NULL) {
+        dec->status_ = VP8_STATUS_OUT_OF_MEMORY;
+        goto Error;
+      }
+      // -1 means a value is unmapped, and therefore unused in the Huffman
+      // image.
+      memset(mapping, 0xff, num_htree_groups_max * sizeof(*mapping));
+      for (num_htree_groups = 0, i = 0; i < huffman_pixs; ++i) {
+        // Get the current mapping for the group and remap the Huffman image.
+        int* const mapped_group = &mapping[huffman_image[i]];
+        if (*mapped_group == -1) *mapped_group = num_htree_groups++;
+        huffman_image[i] = *mapped_group;
+      }
+      huffman_tables_bogus = (HuffmanCode*)WebPSafeMalloc(
+          table_size, sizeof(*huffman_tables_bogus));
+      if (huffman_tables_bogus == NULL) {
+        dec->status_ = VP8_STATUS_OUT_OF_MEMORY;
+        goto Error;
+      }
+    } else {
+      num_htree_groups = num_htree_groups_max;
+    }
   }
 
   if (br->eos_) goto Error;
@@ -403,11 +442,11 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
     }
   }
 
+  code_lengths = (int*)WebPSafeCalloc((uint64_t)max_alphabet_size,
+                                      sizeof(*code_lengths));
   huffman_tables = (HuffmanCode*)WebPSafeMalloc(num_htree_groups * table_size,
                                                 sizeof(*huffman_tables));
   htree_groups = VP8LHtreeGroupsNew(num_htree_groups);
-  code_lengths = (int*)WebPSafeCalloc((uint64_t)max_alphabet_size,
-                                      sizeof(*code_lengths));
 
   if (htree_groups == NULL || code_lengths == NULL || huffman_tables == NULL) {
     dec->status_ = VP8_STATUS_OUT_OF_MEMORY;
@@ -415,28 +454,35 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
   }
 
   next = huffman_tables;
-  for (i = 0; i < num_htree_groups; ++i) {
-    HTreeGroup* const htree_group = &htree_groups[i];
+  for (i = 0; i < num_htree_groups_max; ++i) {
+    // If the index "i" is unused in the Huffman image, read the coefficients
+    // but store them to a bogus htree_group.
+    const int is_bogus = (mapping != NULL && mapping[i] == -1);
+    HTreeGroup* const htree_group =
+        is_bogus ? &htree_group_bogus :
+        &htree_groups[(mapping == NULL) ? i : mapping[i]];
     HuffmanCode** const htrees = htree_group->htrees;
+    HuffmanCode* huffman_tables_i = is_bogus ? huffman_tables_bogus : next;
     int size;
     int total_size = 0;
     int is_trivial_literal = 1;
     int max_bits = 0;
     for (j = 0; j < HUFFMAN_CODES_PER_META_CODE; ++j) {
       int alphabet_size = kAlphabetSize[j];
-      htrees[j] = next;
+      htrees[j] = huffman_tables_i;
       if (j == 0 && color_cache_bits > 0) {
         alphabet_size += 1 << color_cache_bits;
       }
-      size = ReadHuffmanCode(alphabet_size, dec, code_lengths, next);
+      size =
+          ReadHuffmanCode(alphabet_size, dec, code_lengths, huffman_tables_i);
       if (size == 0) {
         goto Error;
       }
       if (is_trivial_literal && kLiteralMap[j] == 1) {
-        is_trivial_literal = (next->bits == 0);
+        is_trivial_literal = (huffman_tables_i->bits == 0);
       }
-      total_size += next->bits;
-      next += size;
+      total_size += huffman_tables_i->bits;
+      huffman_tables_i += size;
       if (j <= ALPHA) {
         int local_max_bits = code_lengths[0];
         int k;
@@ -448,38 +494,41 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
         max_bits += local_max_bits;
       }
     }
+    if (!is_bogus) next = huffman_tables_i;
     htree_group->is_trivial_literal = is_trivial_literal;
     htree_group->is_trivial_code = 0;
     if (is_trivial_literal) {
       const int red = htrees[RED][0].value;
       const int blue = htrees[BLUE][0].value;
       const int alpha = htrees[ALPHA][0].value;
-      htree_group->literal_arb =
-          ((uint32_t)alpha << 24) | (red << 16) | blue;
+      htree_group->literal_arb = ((uint32_t)alpha << 24) | (red << 16) | blue;
       if (total_size == 0 && htrees[GREEN][0].value < NUM_LITERAL_CODES) {
         htree_group->is_trivial_code = 1;
         htree_group->literal_arb |= htrees[GREEN][0].value << 8;
       }
     }
-    htree_group->use_packed_table = !htree_group->is_trivial_code &&
-                                    (max_bits < HUFFMAN_PACKED_BITS);
+    htree_group->use_packed_table =
+        !htree_group->is_trivial_code && (max_bits < HUFFMAN_PACKED_BITS);
     if (htree_group->use_packed_table) BuildPackedTable(htree_group);
   }
-  WebPSafeFree(code_lengths);
+  ok = 1;
 
-  // All OK. Finalize pointers and return.
+  // All OK. Finalize pointers.
   hdr->huffman_image_ = huffman_image;
   hdr->num_htree_groups_ = num_htree_groups;
   hdr->htree_groups_ = htree_groups;
   hdr->huffman_tables_ = huffman_tables;
-  return 1;
 
  Error:
   WebPSafeFree(code_lengths);
-  WebPSafeFree(huffman_image);
-  WebPSafeFree(huffman_tables);
-  VP8LHtreeGroupsFree(htree_groups);
-  return 0;
+  WebPSafeFree(huffman_tables_bogus);
+  WebPSafeFree(mapping);
+  if (!ok) {
+    WebPSafeFree(huffman_image);
+    WebPSafeFree(huffman_tables);
+    VP8LHtreeGroupsFree(htree_groups);
+  }
+  return ok;
 }
 
 //------------------------------------------------------------------------------
@@ -884,7 +933,11 @@ static WEBP_INLINE void CopyBlock8b(uint8_t* const dst, int dist, int length) {
 #endif
         break;
       case 2:
+#if !defined(WORDS_BIGENDIAN)
         memcpy(&pattern, src, sizeof(uint16_t));
+#else
+        pattern = ((uint32_t)src[0] << 8) | src[1];
+#endif
 #if defined(__arm__) || defined(_M_ARM)
         pattern |= pattern << 16;
 #elif defined(WEBP_USE_MIPS_DSP_R2)
@@ -1523,7 +1576,6 @@ int VP8LDecodeAlphaHeader(ALPHDecoder* const alph_dec,
   if (dec == NULL) return 0;
 
   assert(alph_dec != NULL);
-  alph_dec->vp8l_dec_ = dec;
 
   dec->width_ = alph_dec->width_;
   dec->height_ = alph_dec->height_;
@@ -1555,11 +1607,12 @@ int VP8LDecodeAlphaHeader(ALPHDecoder* const alph_dec,
 
   if (!ok) goto Err;
 
+  // Only set here, once we are sure it is valid (to avoid thread races).
+  alph_dec->vp8l_dec_ = dec;
   return 1;
 
  Err:
-  VP8LDelete(alph_dec->vp8l_dec_);
-  alph_dec->vp8l_dec_ = NULL;
+  VP8LDelete(dec);
   return 0;
 }
 

src/dec/vp8li_dec.h

@@ -132,4 +132,4 @@ void VP8LDelete(VP8LDecoder* const dec);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_VP8LI_DEC_H_ */
+#endif  // WEBP_DEC_VP8LI_DEC_H_

src/dec/webpi_dec.h

@@ -130,4 +130,4 @@ int WebPAvoidSlowMemory(const WebPDecBuffer* const output,
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_WEBPI_DEC_H_ */
+#endif  // WEBP_DEC_WEBPI_DEC_H_

src/demux/Makefile.am

@@ -13,6 +13,6 @@ noinst_HEADERS =
 noinst_HEADERS += ../webp/format_constants.h
 
 libwebpdemux_la_LIBADD = ../libwebp.la
-libwebpdemux_la_LDFLAGS = -no-undefined -version-info 2:4:0
+libwebpdemux_la_LDFLAGS = -no-undefined -version-info 2:5:0
 libwebpdemuxincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebpdemux.pc

src/demux/demux.c

@@ -25,7 +25,7 @@
 
 #define DMUX_MAJ_VERSION 1
 #define DMUX_MIN_VERSION 0
-#define DMUX_REV_VERSION 0
+#define DMUX_REV_VERSION 1
 
 typedef struct {
   size_t start_;        // start location of the data

src/demux/libwebpdemux.rc

@@ -6,8 +6,8 @@
 LANGUAGE LANG_ENGLISH, SUBLANG_ENGLISH_US
 
 VS_VERSION_INFO VERSIONINFO
- FILEVERSION 1,0,0,0
- PRODUCTVERSION 1,0,0,0
+ FILEVERSION 1,0,0,1
+ PRODUCTVERSION 1,0,0,1
  FILEFLAGSMASK 0x3fL
 #ifdef _DEBUG
  FILEFLAGS 0x1L
@@ -24,12 +24,12 @@ BEGIN
         BEGIN
             VALUE "CompanyName", "Google, Inc."
             VALUE "FileDescription", "libwebpdemux DLL"
-            VALUE "FileVersion", "1.0.0"
+            VALUE "FileVersion", "1.0.1"
             VALUE "InternalName", "libwebpdemux.dll"
             VALUE "LegalCopyright", "Copyright (C) 2018"
             VALUE "OriginalFilename", "libwebpdemux.dll"
             VALUE "ProductName", "WebP Image Demuxer"
-            VALUE "ProductVersion", "1.0.0"
+            VALUE "ProductVersion", "1.0.1"
         END
     END
     BLOCK "VarFileInfo"

src/dsp/Makefile.am

@@ -1,7 +1,6 @@
 AM_CPPFLAGS += -I$(top_builddir) -I$(top_srcdir)
 noinst_LTLIBRARIES =
 noinst_LTLIBRARIES += libwebpdsp.la
-noinst_LTLIBRARIES += libwebpdsp_avx2.la
 noinst_LTLIBRARIES += libwebpdsp_sse2.la
 noinst_LTLIBRARIES += libwebpdspdecode_sse2.la
 noinst_LTLIBRARIES += libwebpdsp_sse41.la
@@ -10,6 +9,10 @@ noinst_LTLIBRARIES += libwebpdsp_neon.la
 noinst_LTLIBRARIES += libwebpdspdecode_neon.la
 noinst_LTLIBRARIES += libwebpdsp_msa.la
 noinst_LTLIBRARIES += libwebpdspdecode_msa.la
+noinst_LTLIBRARIES += libwebpdsp_mips32.la
+noinst_LTLIBRARIES += libwebpdspdecode_mips32.la
+noinst_LTLIBRARIES += libwebpdsp_mips_dsp_r2.la
+noinst_LTLIBRARIES += libwebpdspdecode_mips_dsp_r2.la
 
 if BUILD_LIBWEBPDECODER
   noinst_LTLIBRARIES += libwebpdspdecode.la
@@ -20,48 +23,25 @@ commondir = $(includedir)/webp
 
 COMMON_SOURCES =
 COMMON_SOURCES += alpha_processing.c
-COMMON_SOURCES += alpha_processing_mips_dsp_r2.c
-COMMON_SOURCES += common_sse2.h
 COMMON_SOURCES += cpu.c
 COMMON_SOURCES += dec.c
 COMMON_SOURCES += dec_clip_tables.c
-COMMON_SOURCES += dec_mips32.c
-COMMON_SOURCES += dec_mips_dsp_r2.c
 COMMON_SOURCES += dsp.h
 COMMON_SOURCES += filters.c
-COMMON_SOURCES += filters_mips_dsp_r2.c
 COMMON_SOURCES += lossless.c
 COMMON_SOURCES += lossless.h
 COMMON_SOURCES += lossless_common.h
-COMMON_SOURCES += lossless_mips_dsp_r2.c
-COMMON_SOURCES += mips_macro.h
 COMMON_SOURCES += rescaler.c
-COMMON_SOURCES += rescaler_mips32.c
-COMMON_SOURCES += rescaler_mips_dsp_r2.c
 COMMON_SOURCES += upsampling.c
-COMMON_SOURCES += upsampling_mips_dsp_r2.c
 COMMON_SOURCES += yuv.c
 COMMON_SOURCES += yuv.h
-COMMON_SOURCES += yuv_mips32.c
-COMMON_SOURCES += yuv_mips_dsp_r2.c
 
 ENC_SOURCES =
 ENC_SOURCES += cost.c
-ENC_SOURCES += cost_mips32.c
-ENC_SOURCES += cost_mips_dsp_r2.c
 ENC_SOURCES += enc.c
-ENC_SOURCES += enc_mips32.c
-ENC_SOURCES += enc_mips_dsp_r2.c
 ENC_SOURCES += lossless_enc.c
-ENC_SOURCES += lossless_enc_mips32.c
-ENC_SOURCES += lossless_enc_mips_dsp_r2.c
 ENC_SOURCES += ssim.c
 
-libwebpdsp_avx2_la_SOURCES =
-libwebpdsp_avx2_la_SOURCES += enc_avx2.c
-libwebpdsp_avx2_la_CPPFLAGS = $(libwebpdsp_la_CPPFLAGS)
-libwebpdsp_avx2_la_CFLAGS = $(AM_CFLAGS) $(AVX2_FLAGS)
-
 libwebpdspdecode_sse41_la_SOURCES =
 libwebpdspdecode_sse41_la_SOURCES += alpha_processing_sse41.c
 libwebpdspdecode_sse41_la_SOURCES += dec_sse41.c
@@ -72,6 +52,7 @@ libwebpdspdecode_sse41_la_CFLAGS = $(AM_CFLAGS) $(SSE41_FLAGS)
 
 libwebpdspdecode_sse2_la_SOURCES =
 libwebpdspdecode_sse2_la_SOURCES += alpha_processing_sse2.c
+libwebpdspdecode_sse2_la_SOURCES += common_sse2.h
 libwebpdspdecode_sse2_la_SOURCES += dec_sse2.c
 libwebpdspdecode_sse2_la_SOURCES += filters_sse2.c
 libwebpdspdecode_sse2_la_SOURCES += lossless_sse2.c
@@ -103,6 +84,26 @@ libwebpdspdecode_msa_la_SOURCES += upsampling_msa.c
 libwebpdspdecode_msa_la_CPPFLAGS = $(libwebpdsp_msa_la_CPPFLAGS)
 libwebpdspdecode_msa_la_CFLAGS = $(libwebpdsp_msa_la_CFLAGS)
 
+libwebpdspdecode_mips32_la_SOURCES =
+libwebpdspdecode_mips32_la_SOURCES += dec_mips32.c
+libwebpdspdecode_mips32_la_SOURCES += mips_macro.h
+libwebpdspdecode_mips32_la_SOURCES += rescaler_mips32.c
+libwebpdspdecode_mips32_la_SOURCES += yuv_mips32.c
+libwebpdspdecode_mips32_la_CPPFLAGS = $(libwebpdsp_mips32_la_CPPFLAGS)
+libwebpdspdecode_mips32_la_CFLAGS = $(libwebpdsp_mips32_la_CFLAGS)
+
+libwebpdspdecode_mips_dsp_r2_la_SOURCES =
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += alpha_processing_mips_dsp_r2.c
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += dec_mips_dsp_r2.c
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += filters_mips_dsp_r2.c
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += lossless_mips_dsp_r2.c
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += mips_macro.h
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += rescaler_mips_dsp_r2.c
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += upsampling_mips_dsp_r2.c
+libwebpdspdecode_mips_dsp_r2_la_SOURCES += yuv_mips_dsp_r2.c
+libwebpdspdecode_mips_dsp_r2_la_CPPFLAGS = $(libwebpdsp_mips_dsp_r2_la_CPPFLAGS)
+libwebpdspdecode_mips_dsp_r2_la_CFLAGS = $(libwebpdsp_mips_dsp_r2_la_CFLAGS)
+
 libwebpdsp_sse2_la_SOURCES =
 libwebpdsp_sse2_la_SOURCES += cost_sse2.c
 libwebpdsp_sse2_la_SOURCES += enc_sse2.c
@@ -133,6 +134,22 @@ libwebpdsp_msa_la_CPPFLAGS = $(libwebpdsp_la_CPPFLAGS)
 libwebpdsp_msa_la_CFLAGS = $(AM_CFLAGS)
 libwebpdsp_msa_la_LIBADD = libwebpdspdecode_msa.la
 
+libwebpdsp_mips32_la_SOURCES =
+libwebpdsp_mips32_la_SOURCES += cost_mips32.c
+libwebpdsp_mips32_la_SOURCES += enc_mips32.c
+libwebpdsp_mips32_la_SOURCES += lossless_enc_mips32.c
+libwebpdsp_mips32_la_CPPFLAGS = $(libwebpdsp_la_CPPFLAGS)
+libwebpdsp_mips32_la_CFLAGS = $(AM_CFLAGS)
+libwebpdsp_mips32_la_LIBADD = libwebpdspdecode_mips32.la
+
+libwebpdsp_mips_dsp_r2_la_SOURCES =
+libwebpdsp_mips_dsp_r2_la_SOURCES += cost_mips_dsp_r2.c
+libwebpdsp_mips_dsp_r2_la_SOURCES += enc_mips_dsp_r2.c
+libwebpdsp_mips_dsp_r2_la_SOURCES += lossless_enc_mips_dsp_r2.c
+libwebpdsp_mips_dsp_r2_la_CPPFLAGS = $(libwebpdsp_la_CPPFLAGS)
+libwebpdsp_mips_dsp_r2_la_CFLAGS = $(AM_CFLAGS)
+libwebpdsp_mips_dsp_r2_la_LIBADD = libwebpdspdecode_mips_dsp_r2.la
+
 libwebpdsp_la_SOURCES = $(COMMON_SOURCES) $(ENC_SOURCES)
 
 noinst_HEADERS =
@@ -144,11 +161,12 @@ libwebpdsp_la_CPPFLAGS += $(AM_CPPFLAGS)
 libwebpdsp_la_CPPFLAGS += $(USE_SWAP_16BIT_CSP)
 libwebpdsp_la_LDFLAGS = -lm
 libwebpdsp_la_LIBADD =
-libwebpdsp_la_LIBADD += libwebpdsp_avx2.la
 libwebpdsp_la_LIBADD += libwebpdsp_sse2.la
 libwebpdsp_la_LIBADD += libwebpdsp_sse41.la
 libwebpdsp_la_LIBADD += libwebpdsp_neon.la
 libwebpdsp_la_LIBADD += libwebpdsp_msa.la
+libwebpdsp_la_LIBADD += libwebpdsp_mips32.la
+libwebpdsp_la_LIBADD += libwebpdsp_mips_dsp_r2.la
 
 if BUILD_LIBWEBPDECODER
   libwebpdspdecode_la_SOURCES = $(COMMON_SOURCES)
@@ -160,4 +178,6 @@ if BUILD_LIBWEBPDECODER
   libwebpdspdecode_la_LIBADD += libwebpdspdecode_sse41.la
   libwebpdspdecode_la_LIBADD += libwebpdspdecode_neon.la
   libwebpdspdecode_la_LIBADD += libwebpdspdecode_msa.la
+  libwebpdspdecode_la_LIBADD += libwebpdspdecode_mips32.la
+  libwebpdspdecode_la_LIBADD += libwebpdspdecode_mips_dsp_r2.la
 endif

src/dsp/alpha_processing.c

@@ -409,12 +409,7 @@ extern void WebPInitAlphaProcessingSSE2(void);
 extern void WebPInitAlphaProcessingSSE41(void);
 extern void WebPInitAlphaProcessingNEON(void);
 
-static volatile VP8CPUInfo alpha_processing_last_cpuinfo_used =
-    (VP8CPUInfo)&alpha_processing_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessing(void) {
-  if (alpha_processing_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(WebPInitAlphaProcessing) {
   WebPMultARGBRow = WebPMultARGBRow_C;
   WebPMultRow = WebPMultRow_C;
   WebPApplyAlphaMultiply4444 = ApplyAlphaMultiply_16b_C;
@@ -474,6 +469,4 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessing(void) {
   assert(WebPPackRGB != NULL);
   assert(WebPHasAlpha8b != NULL);
   assert(WebPHasAlpha32b != NULL);
-
-  alpha_processing_last_cpuinfo_used = VP8GetCPUInfo;
 }

src/dsp/alpha_processing_neon.c

@@ -83,7 +83,7 @@ static void ApplyAlphaMultiply_NEON(uint8_t* rgba, int alpha_first,
 static int DispatchAlpha_NEON(const uint8_t* alpha, int alpha_stride,
                               int width, int height,
                               uint8_t* dst, int dst_stride) {
-  uint32_t alpha_mask = 0xffffffffu;
+  uint32_t alpha_mask = 0xffu;
   uint8x8_t mask8 = vdup_n_u8(0xff);
   uint32_t tmp[2];
   int i, j;
@@ -107,6 +107,7 @@ static int DispatchAlpha_NEON(const uint8_t* alpha, int alpha_stride,
     dst += dst_stride;
   }
   vst1_u8((uint8_t*)tmp, mask8);
+  alpha_mask *= 0x01010101;
   alpha_mask &= tmp[0];
   alpha_mask &= tmp[1];
   return (alpha_mask != 0xffffffffu);
@@ -134,7 +135,7 @@ static void DispatchAlphaToGreen_NEON(const uint8_t* alpha, int alpha_stride,
 static int ExtractAlpha_NEON(const uint8_t* argb, int argb_stride,
                              int width, int height,
                              uint8_t* alpha, int alpha_stride) {
-  uint32_t alpha_mask = 0xffffffffu;
+  uint32_t alpha_mask = 0xffu;
   uint8x8_t mask8 = vdup_n_u8(0xff);
   uint32_t tmp[2];
   int i, j;
@@ -156,6 +157,7 @@ static int ExtractAlpha_NEON(const uint8_t* argb, int argb_stride,
     alpha += alpha_stride;
   }
   vst1_u8((uint8_t*)tmp, mask8);
+  alpha_mask *= 0x01010101;
   alpha_mask &= tmp[0];
   alpha_mask &= tmp[1];
   return (alpha_mask == 0xffffffffu);

src/dsp/cost.c

@@ -378,12 +378,7 @@ extern void VP8EncDspCostInitMIPS32(void);
 extern void VP8EncDspCostInitMIPSdspR2(void);
 extern void VP8EncDspCostInitSSE2(void);
 
-static volatile VP8CPUInfo cost_last_cpuinfo_used =
-    (VP8CPUInfo)&cost_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspCostInit(void) {
-  if (cost_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(VP8EncDspCostInit) {
   VP8GetResidualCost = GetResidualCost_C;
   VP8SetResidualCoeffs = SetResidualCoeffs_C;
 
@@ -405,8 +400,6 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspCostInit(void) {
     }
 #endif
   }
-
-  cost_last_cpuinfo_used = VP8GetCPUInfo;
 }
 
 //------------------------------------------------------------------------------

src/dsp/dec.c

@@ -741,12 +741,7 @@ extern void VP8DspInitMIPS32(void);
 extern void VP8DspInitMIPSdspR2(void);
 extern void VP8DspInitMSA(void);
 
-static volatile VP8CPUInfo dec_last_cpuinfo_used =
-    (VP8CPUInfo)&dec_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void VP8DspInit(void) {
-  if (dec_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(VP8DspInit) {
   VP8InitClipTables();
 
 #if !WEBP_NEON_OMIT_C_CODE
@@ -889,6 +884,4 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8DspInit(void) {
   assert(VP8PredChroma8[5] != NULL);
   assert(VP8PredChroma8[6] != NULL);
   assert(VP8DitherCombine8x8 != NULL);
-
-  dec_last_cpuinfo_used = VP8GetCPUInfo;
 }

src/dsp/dsp.h

@@ -76,10 +76,6 @@ extern "C" {
 #define WEBP_USE_SSE41
 #endif
 
-#if defined(__AVX2__) || defined(WEBP_HAVE_AVX2)
-#define WEBP_USE_AVX2
-#endif
-
 // The intrinsics currently cause compiler errors with arm-nacl-gcc and the
 // inline assembly would need to be modified for use with Native Client.
 #if (defined(__ARM_NEON__) || \
@@ -141,6 +137,42 @@ extern "C" {
 #endif
 #endif
 
+#if defined(WEBP_USE_THREAD) && !defined(_WIN32)
+#include <pthread.h>  // NOLINT
+
+#define WEBP_DSP_INIT(func) do {                                    \
+  static volatile VP8CPUInfo func ## _last_cpuinfo_used =           \
+      (VP8CPUInfo)&func ## _last_cpuinfo_used;                      \
+  static pthread_mutex_t func ## _lock = PTHREAD_MUTEX_INITIALIZER; \
+  if (pthread_mutex_lock(&func ## _lock)) break;                    \
+  if (func ## _last_cpuinfo_used != VP8GetCPUInfo) func();          \
+  func ## _last_cpuinfo_used = VP8GetCPUInfo;                       \
+  (void)pthread_mutex_unlock(&func ## _lock);                       \
+} while (0)
+#else  // !(defined(WEBP_USE_THREAD) && !defined(_WIN32))
+#define WEBP_DSP_INIT(func) do {                                    \
+  static volatile VP8CPUInfo func ## _last_cpuinfo_used =           \
+      (VP8CPUInfo)&func ## _last_cpuinfo_used;                      \
+  if (func ## _last_cpuinfo_used == VP8GetCPUInfo) break;           \
+  func();                                                           \
+  func ## _last_cpuinfo_used = VP8GetCPUInfo;                       \
+} while (0)
+#endif  // defined(WEBP_USE_THREAD) && !defined(_WIN32)
+
+// Defines an Init + helper function that control multiple initialization of
+// function pointers / tables.
+/* Usage:
+   WEBP_DSP_INIT_FUNC(InitFunc) {
+     ...function body
+   }
+*/
+#define WEBP_DSP_INIT_FUNC(name)                             \
+  static WEBP_TSAN_IGNORE_FUNCTION void name ## _body(void); \
+  WEBP_TSAN_IGNORE_FUNCTION void name(void) {                \
+    WEBP_DSP_INIT(name ## _body);                            \
+  }                                                          \
+  static WEBP_TSAN_IGNORE_FUNCTION void name ## _body(void)
+
 #define WEBP_UBSAN_IGNORE_UNDEF
 #define WEBP_UBSAN_IGNORE_UNSIGNED_OVERFLOW
 #if defined(__clang__) && defined(__has_attribute)
@@ -196,7 +228,7 @@ WEBP_EXTERN VP8CPUInfo VP8GetCPUInfo;
 // avoiding a compiler warning.
 #define WEBP_DSP_INIT_STUB(func) \
   extern void func(void); \
-  WEBP_TSAN_IGNORE_FUNCTION void func(void) {}
+  void func(void) {}
 
 //------------------------------------------------------------------------------
 // Encoding
@@ -643,4 +675,4 @@ void VP8FiltersInit(void);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DSP_DSP_H_ */
+#endif  // WEBP_DSP_DSP_H_

src/dsp/enc.c

@@ -734,18 +734,12 @@ VP8BlockCopy VP8Copy16x8;
 
 extern void VP8EncDspInitSSE2(void);
 extern void VP8EncDspInitSSE41(void);
-extern void VP8EncDspInitAVX2(void);
 extern void VP8EncDspInitNEON(void);
 extern void VP8EncDspInitMIPS32(void);
 extern void VP8EncDspInitMIPSdspR2(void);
 extern void VP8EncDspInitMSA(void);
 
-static volatile VP8CPUInfo enc_last_cpuinfo_used =
-    (VP8CPUInfo)&enc_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspInit(void) {
-  if (enc_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(VP8EncDspInit) {
   VP8DspInit();  // common inverse transforms
   InitTables();
 
@@ -789,11 +783,6 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspInit(void) {
 #endif
     }
 #endif
-#if defined(WEBP_USE_AVX2)
-    if (VP8GetCPUInfo(kAVX2)) {
-      VP8EncDspInitAVX2();
-    }
-#endif
 #if defined(WEBP_USE_MIPS32)
     if (VP8GetCPUInfo(kMIPS32)) {
       VP8EncDspInitMIPS32();
@@ -838,6 +827,4 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspInit(void) {
   assert(VP8EncQuantizeBlockWHT != NULL);
   assert(VP8Copy4x4 != NULL);
   assert(VP8Copy16x8 != NULL);
-
-  enc_last_cpuinfo_used = VP8GetCPUInfo;
 }

src/dsp/enc_avx2.c

@@ -1,21 +0,0 @@
-// Copyright 2014 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// AVX2 version of speed-critical encoding functions.
-
-#include "src/dsp/dsp.h"
-
-#if defined(WEBP_USE_AVX2)
-
-#endif  // WEBP_USE_AVX2
-
-//------------------------------------------------------------------------------
-// Entry point
-
-WEBP_DSP_INIT_STUB(VP8EncDspInitAVX2)

src/dsp/filters.c

@@ -238,12 +238,7 @@ extern void VP8FiltersInitMSA(void);
 extern void VP8FiltersInitNEON(void);
 extern void VP8FiltersInitSSE2(void);
 
-static volatile VP8CPUInfo filters_last_cpuinfo_used =
-    (VP8CPUInfo)&filters_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void VP8FiltersInit(void) {
-  if (filters_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(VP8FiltersInit) {
   WebPUnfilters[WEBP_FILTER_NONE] = NULL;
 #if !WEBP_NEON_OMIT_C_CODE
   WebPUnfilters[WEBP_FILTER_HORIZONTAL] = HorizontalUnfilter_C;
@@ -289,6 +284,4 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8FiltersInit(void) {
   assert(WebPFilters[WEBP_FILTER_HORIZONTAL] != NULL);
   assert(WebPFilters[WEBP_FILTER_VERTICAL] != NULL);
   assert(WebPFilters[WEBP_FILTER_GRADIENT] != NULL);
-
-  filters_last_cpuinfo_used = VP8GetCPUInfo;
 }

src/dsp/lossless.c

@@ -23,8 +23,6 @@
 #include "src/dsp/lossless.h"
 #include "src/dsp/lossless_common.h"
 
-#define MAX_DIFF_COST (1e30f)
-
 //------------------------------------------------------------------------------
 // Image transforms.
 
@@ -577,9 +575,6 @@ extern void VP8LDspInitNEON(void);
 extern void VP8LDspInitMIPSdspR2(void);
 extern void VP8LDspInitMSA(void);
 
-static volatile VP8CPUInfo lossless_last_cpuinfo_used =
-    (VP8CPUInfo)&lossless_last_cpuinfo_used;
-
 #define COPY_PREDICTOR_ARRAY(IN, OUT) do {                \
   (OUT)[0] = IN##0_C;                                     \
   (OUT)[1] = IN##1_C;                                     \
@@ -599,9 +594,7 @@ static volatile VP8CPUInfo lossless_last_cpuinfo_used =
   (OUT)[15] = IN##0_C;                                    \
 } while (0);
 
-WEBP_TSAN_IGNORE_FUNCTION void VP8LDspInit(void) {
-  if (lossless_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(VP8LDspInit) {
   COPY_PREDICTOR_ARRAY(Predictor, VP8LPredictors)
   COPY_PREDICTOR_ARRAY(Predictor, VP8LPredictors_C)
   COPY_PREDICTOR_ARRAY(PredictorAdd, VP8LPredictorsAdd)
@@ -658,8 +651,6 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8LDspInit(void) {
   assert(VP8LConvertBGRAToRGB565 != NULL);
   assert(VP8LMapColor32b != NULL);
   assert(VP8LMapColor8b != NULL);
-
-  lossless_last_cpuinfo_used = VP8GetCPUInfo;
 }
 #undef COPY_PREDICTOR_ARRAY
 

src/dsp/lossless.h

@@ -163,7 +163,7 @@ extern VP8LCostCombinedFunc VP8LExtraCostCombined;
 extern VP8LCombinedShannonEntropyFunc VP8LCombinedShannonEntropy;
 
 typedef struct {        // small struct to hold counters
-  int counts[2];        // index: 0=zero steak, 1=non-zero streak
+  int counts[2];        // index: 0=zero streak, 1=non-zero streak
   int streaks[2][2];    // [zero/non-zero][streak<3 / streak>=3]
 } VP8LStreaks;
 
@@ -194,10 +194,14 @@ extern VP8LGetEntropyUnrefinedFunc VP8LGetEntropyUnrefined;
 void VP8LBitsEntropyUnrefined(const uint32_t* const array, int n,
                               VP8LBitEntropy* const entropy);
 
-typedef void (*VP8LHistogramAddFunc)(const VP8LHistogram* const a,
-                                     const VP8LHistogram* const b,
-                                     VP8LHistogram* const out);
-extern VP8LHistogramAddFunc VP8LHistogramAdd;
+typedef void (*VP8LAddVectorFunc)(const uint32_t* a, const uint32_t* b,
+                                  uint32_t* out, int size);
+extern VP8LAddVectorFunc VP8LAddVector;
+typedef void (*VP8LAddVectorEqFunc)(const uint32_t* a, uint32_t* out, int size);
+extern VP8LAddVectorEqFunc VP8LAddVectorEq;
+void VP8LHistogramAdd(const VP8LHistogram* const a,
+                      const VP8LHistogram* const b,
+                      VP8LHistogram* const out);
 
 // -----------------------------------------------------------------------------
 // PrefixEncode()

src/dsp/lossless_enc.c

@@ -632,38 +632,67 @@ static double ExtraCostCombined_C(const uint32_t* X, const uint32_t* Y,
 
 //------------------------------------------------------------------------------
 
-static void HistogramAdd_C(const VP8LHistogram* const a,
-                           const VP8LHistogram* const b,
-                           VP8LHistogram* const out) {
+static void AddVector_C(const uint32_t* a, const uint32_t* b, uint32_t* out,
+                        int size) {
+  int i;
+  for (i = 0; i < size; ++i) out[i] = a[i] + b[i];
+}
+
+static void AddVectorEq_C(const uint32_t* a, uint32_t* out, int size) {
+  int i;
+  for (i = 0; i < size; ++i) out[i] += a[i];
+}
+
+#define ADD(X, ARG, LEN) do {                                                  \
+  if (a->is_used_[X]) {                                                        \
+    if (b->is_used_[X]) {                                                      \
+      VP8LAddVector(a->ARG, b->ARG, out->ARG, (LEN));                          \
+    } else {                                                                   \
+      memcpy(&out->ARG[0], &a->ARG[0], (LEN) * sizeof(out->ARG[0]));           \
+    }                                                                          \
+  } else if (b->is_used_[X]) {                                                 \
+    memcpy(&out->ARG[0], &b->ARG[0], (LEN) * sizeof(out->ARG[0]));             \
+  } else {                                                                     \
+    memset(&out->ARG[0], 0, (LEN) * sizeof(out->ARG[0]));                      \
+  }                                                                            \
+} while (0)
+
+#define ADD_EQ(X, ARG, LEN) do {                                               \
+  if (a->is_used_[X]) {                                                        \
+    if (out->is_used_[X]) {                                                    \
+      VP8LAddVectorEq(a->ARG, out->ARG, (LEN));                                \
+    } else {                                                                   \
+      memcpy(&out->ARG[0], &a->ARG[0], (LEN) * sizeof(out->ARG[0]));           \
+    }                                                                          \
+  }                                                                            \
+} while (0)
+
+void VP8LHistogramAdd(const VP8LHistogram* const a,
+                      const VP8LHistogram* const b, VP8LHistogram* const out) {
   int i;
   const int literal_size = VP8LHistogramNumCodes(a->palette_code_bits_);
   assert(a->palette_code_bits_ == b->palette_code_bits_);
+
   if (b != out) {
-    for (i = 0; i < literal_size; ++i) {
-      out->literal_[i] = a->literal_[i] + b->literal_[i];
-    }
-    for (i = 0; i < NUM_DISTANCE_CODES; ++i) {
-      out->distance_[i] = a->distance_[i] + b->distance_[i];
-    }
-    for (i = 0; i < NUM_LITERAL_CODES; ++i) {
-      out->red_[i] = a->red_[i] + b->red_[i];
-      out->blue_[i] = a->blue_[i] + b->blue_[i];
-      out->alpha_[i] = a->alpha_[i] + b->alpha_[i];
+    ADD(0, literal_, literal_size);
+    ADD(1, red_, NUM_LITERAL_CODES);
+    ADD(2, blue_, NUM_LITERAL_CODES);
+    ADD(3, alpha_, NUM_LITERAL_CODES);
+    ADD(4, distance_, NUM_DISTANCE_CODES);
+    for (i = 0; i < 5; ++i) {
+      out->is_used_[i] = (a->is_used_[i] | b->is_used_[i]);
     }
   } else {
-    for (i = 0; i < literal_size; ++i) {
-      out->literal_[i] += a->literal_[i];
-    }
-    for (i = 0; i < NUM_DISTANCE_CODES; ++i) {
-      out->distance_[i] += a->distance_[i];
-    }
-    for (i = 0; i < NUM_LITERAL_CODES; ++i) {
-      out->red_[i] += a->red_[i];
-      out->blue_[i] += a->blue_[i];
-      out->alpha_[i] += a->alpha_[i];
-    }
+    ADD_EQ(0, literal_, literal_size);
+    ADD_EQ(1, red_, NUM_LITERAL_CODES);
+    ADD_EQ(2, blue_, NUM_LITERAL_CODES);
+    ADD_EQ(3, alpha_, NUM_LITERAL_CODES);
+    ADD_EQ(4, distance_, NUM_DISTANCE_CODES);
+    for (i = 0; i < 5; ++i) out->is_used_[i] |= a->is_used_[i];
   }
 }
+#undef ADD
+#undef ADD_EQ
 
 //------------------------------------------------------------------------------
 // Image transforms.
@@ -848,7 +877,8 @@ VP8LCombinedShannonEntropyFunc VP8LCombinedShannonEntropy;
 VP8LGetEntropyUnrefinedFunc VP8LGetEntropyUnrefined;
 VP8LGetCombinedEntropyUnrefinedFunc VP8LGetCombinedEntropyUnrefined;
 
-VP8LHistogramAddFunc VP8LHistogramAdd;
+VP8LAddVectorFunc VP8LAddVector;
+VP8LAddVectorEqFunc VP8LAddVectorEq;
 
 VP8LVectorMismatchFunc VP8LVectorMismatch;
 VP8LBundleColorMapFunc VP8LBundleColorMap;
@@ -863,12 +893,7 @@ extern void VP8LEncDspInitMIPS32(void);
 extern void VP8LEncDspInitMIPSdspR2(void);
 extern void VP8LEncDspInitMSA(void);
 
-static volatile VP8CPUInfo lossless_enc_last_cpuinfo_used =
-    (VP8CPUInfo)&lossless_enc_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void VP8LEncDspInit(void) {
-  if (lossless_enc_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(VP8LEncDspInit) {
   VP8LDspInit();
 
 #if !WEBP_NEON_OMIT_C_CODE
@@ -890,7 +915,8 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8LEncDspInit(void) {
   VP8LGetEntropyUnrefined = GetEntropyUnrefined_C;
   VP8LGetCombinedEntropyUnrefined = GetCombinedEntropyUnrefined_C;
 
-  VP8LHistogramAdd = HistogramAdd_C;
+  VP8LAddVector = AddVector_C;
+  VP8LAddVectorEq = AddVectorEq_C;
 
   VP8LVectorMismatch = VectorMismatch_C;
   VP8LBundleColorMap = VP8LBundleColorMap_C;
@@ -976,7 +1002,8 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8LEncDspInit(void) {
   assert(VP8LCombinedShannonEntropy != NULL);
   assert(VP8LGetEntropyUnrefined != NULL);
   assert(VP8LGetCombinedEntropyUnrefined != NULL);
-  assert(VP8LHistogramAdd != NULL);
+  assert(VP8LAddVector != NULL);
+  assert(VP8LAddVectorEq != NULL);
   assert(VP8LVectorMismatch != NULL);
   assert(VP8LBundleColorMap != NULL);
   assert(VP8LPredictorsSub[0] != NULL);
@@ -1011,8 +1038,6 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8LEncDspInit(void) {
   assert(VP8LPredictorsSub_C[13] != NULL);
   assert(VP8LPredictorsSub_C[14] != NULL);
   assert(VP8LPredictorsSub_C[15] != NULL);
-
-  lossless_enc_last_cpuinfo_used = VP8GetCPUInfo;
 }
 
 //------------------------------------------------------------------------------

src/dsp/lossless_enc_mips32.c

@@ -344,65 +344,29 @@ static void GetCombinedEntropyUnrefined_MIPS32(const uint32_t X[],
     ASM_END_COMMON_0                                    \
     ASM_END_COMMON_1
 
-#define ADD_VECTOR(A, B, OUT, SIZE, EXTRA_SIZE)  do {   \
-  const uint32_t* pa = (const uint32_t*)(A);            \
-  const uint32_t* pb = (const uint32_t*)(B);            \
-  uint32_t* pout = (uint32_t*)(OUT);                    \
-  const uint32_t* const LoopEnd = pa + (SIZE);          \
-  assert((SIZE) % 4 == 0);                              \
-  ASM_START                                             \
-  ADD_TO_OUT(0, 4, 8, 12, 1, pa, pb, pout)              \
-  ASM_END_0                                             \
-  if ((EXTRA_SIZE) > 0) {                               \
-    const int last = (EXTRA_SIZE);                      \
-    int i;                                              \
-    for (i = 0; i < last; ++i) pout[i] = pa[i] + pb[i]; \
-  }                                                     \
-} while (0)
-
-#define ADD_VECTOR_EQ(A, OUT, SIZE, EXTRA_SIZE)  do {   \
-  const uint32_t* pa = (const uint32_t*)(A);            \
-  uint32_t* pout = (uint32_t*)(OUT);                    \
-  const uint32_t* const LoopEnd = pa + (SIZE);          \
-  assert((SIZE) % 4 == 0);                              \
-  ASM_START                                             \
-  ADD_TO_OUT(0, 4, 8, 12, 0, pa, pout, pout)            \
-  ASM_END_1                                             \
-  if ((EXTRA_SIZE) > 0) {                               \
-    const int last = (EXTRA_SIZE);                      \
-    int i;                                              \
-    for (i = 0; i < last; ++i) pout[i] += pa[i];        \
-  }                                                     \
-} while (0)
-
-static void HistogramAdd_MIPS32(const VP8LHistogram* const a,
-                                const VP8LHistogram* const b,
-                                VP8LHistogram* const out) {
+static void AddVector_MIPS32(const uint32_t* pa, const uint32_t* pb,
+                             uint32_t* pout, int size) {
   uint32_t temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
-  const int extra_cache_size = VP8LHistogramNumCodes(a->palette_code_bits_)
-                             - (NUM_LITERAL_CODES + NUM_LENGTH_CODES);
-  assert(a->palette_code_bits_ == b->palette_code_bits_);
-
-  if (b != out) {
-    ADD_VECTOR(a->literal_, b->literal_, out->literal_,
-               NUM_LITERAL_CODES + NUM_LENGTH_CODES, extra_cache_size);
-    ADD_VECTOR(a->distance_, b->distance_, out->distance_,
-               NUM_DISTANCE_CODES, 0);
-    ADD_VECTOR(a->red_, b->red_, out->red_, NUM_LITERAL_CODES, 0);
-    ADD_VECTOR(a->blue_, b->blue_, out->blue_, NUM_LITERAL_CODES, 0);
-    ADD_VECTOR(a->alpha_, b->alpha_, out->alpha_, NUM_LITERAL_CODES, 0);
-  } else {
-    ADD_VECTOR_EQ(a->literal_, out->literal_,
-                  NUM_LITERAL_CODES + NUM_LENGTH_CODES, extra_cache_size);
-    ADD_VECTOR_EQ(a->distance_, out->distance_, NUM_DISTANCE_CODES, 0);
-    ADD_VECTOR_EQ(a->red_, out->red_, NUM_LITERAL_CODES, 0);
-    ADD_VECTOR_EQ(a->blue_, out->blue_, NUM_LITERAL_CODES, 0);
-    ADD_VECTOR_EQ(a->alpha_, out->alpha_, NUM_LITERAL_CODES, 0);
-  }
+  const int end = ((size) / 4) * 4;
+  const uint32_t* const LoopEnd = pa + end;
+  int i;
+  ASM_START
+  ADD_TO_OUT(0, 4, 8, 12, 1, pa, pb, pout)
+  ASM_END_0
+  for (i = 0; i < size - end; ++i) pout[i] = pa[i] + pb[i];
+}
+
+static void AddVectorEq_MIPS32(const uint32_t* pa, uint32_t* pout, int size) {
+  uint32_t temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
+  const int end = ((size) / 4) * 4;
+  const uint32_t* const LoopEnd = pa + end;
+  int i;
+  ASM_START
+  ADD_TO_OUT(0, 4, 8, 12, 0, pa, pout, pout)
+  ASM_END_1
+  for (i = 0; i < size - end; ++i) pout[i] += pa[i];
 }
 
-#undef ADD_VECTOR_EQ
-#undef ADD_VECTOR
 #undef ASM_END_1
 #undef ASM_END_0
 #undef ASM_END_COMMON_1
@@ -422,7 +386,8 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8LEncDspInitMIPS32(void) {
   VP8LExtraCostCombined = ExtraCostCombined_MIPS32;
   VP8LGetEntropyUnrefined = GetEntropyUnrefined_MIPS32;
   VP8LGetCombinedEntropyUnrefined = GetCombinedEntropyUnrefined_MIPS32;
-  VP8LHistogramAdd = HistogramAdd_MIPS32;
+  VP8LAddVector = AddVector_MIPS32;
+  VP8LAddVectorEq = AddVectorEq_MIPS32;
 }
 
 #else  // !WEBP_USE_MIPS32

src/dsp/lossless_enc_sse2.c

@@ -170,12 +170,13 @@ static void CollectColorRedTransforms_SSE2(const uint32_t* argb, int stride,
 
 //------------------------------------------------------------------------------
 
+// Note we are adding uint32_t's as *signed* int32's (using _mm_add_epi32). But
+// that's ok since the histogram values are less than 1<<28 (max picture size).
 #define LINE_SIZE 16    // 8 or 16
 static void AddVector_SSE2(const uint32_t* a, const uint32_t* b, uint32_t* out,
                            int size) {
   int i;
-  assert(size % LINE_SIZE == 0);
-  for (i = 0; i < size; i += LINE_SIZE) {
+  for (i = 0; i + LINE_SIZE <= size; i += LINE_SIZE) {
     const __m128i a0 = _mm_loadu_si128((const __m128i*)&a[i +  0]);
     const __m128i a1 = _mm_loadu_si128((const __m128i*)&a[i +  4]);
 #if (LINE_SIZE == 16)
@@ -195,12 +196,14 @@ static void AddVector_SSE2(const uint32_t* a, const uint32_t* b, uint32_t* out,
     _mm_storeu_si128((__m128i*)&out[i + 12], _mm_add_epi32(a3, b3));
 #endif
   }
+  for (; i < size; ++i) {
+    out[i] = a[i] + b[i];
+  }
 }
 
 static void AddVectorEq_SSE2(const uint32_t* a, uint32_t* out, int size) {
   int i;
-  assert(size % LINE_SIZE == 0);
-  for (i = 0; i < size; i += LINE_SIZE) {
+  for (i = 0; i + LINE_SIZE <= size; i += LINE_SIZE) {
     const __m128i a0 = _mm_loadu_si128((const __m128i*)&a[i +  0]);
     const __m128i a1 = _mm_loadu_si128((const __m128i*)&a[i +  4]);
 #if (LINE_SIZE == 16)
@@ -220,36 +223,12 @@ static void AddVectorEq_SSE2(const uint32_t* a, uint32_t* out, int size) {
     _mm_storeu_si128((__m128i*)&out[i + 12], _mm_add_epi32(a3, b3));
 #endif
   }
+  for (; i < size; ++i) {
+    out[i] += a[i];
+  }
 }
 #undef LINE_SIZE
 
-// Note we are adding uint32_t's as *signed* int32's (using _mm_add_epi32). But
-// that's ok since the histogram values are less than 1<<28 (max picture size).
-static void HistogramAdd_SSE2(const VP8LHistogram* const a,
-                              const VP8LHistogram* const b,
-                              VP8LHistogram* const out) {
-  int i;
-  const int literal_size = VP8LHistogramNumCodes(a->palette_code_bits_);
-  assert(a->palette_code_bits_ == b->palette_code_bits_);
-  if (b != out) {
-    AddVector_SSE2(a->literal_, b->literal_, out->literal_, NUM_LITERAL_CODES);
-    AddVector_SSE2(a->red_, b->red_, out->red_, NUM_LITERAL_CODES);
-    AddVector_SSE2(a->blue_, b->blue_, out->blue_, NUM_LITERAL_CODES);
-    AddVector_SSE2(a->alpha_, b->alpha_, out->alpha_, NUM_LITERAL_CODES);
-  } else {
-    AddVectorEq_SSE2(a->literal_, out->literal_, NUM_LITERAL_CODES);
-    AddVectorEq_SSE2(a->red_, out->red_, NUM_LITERAL_CODES);
-    AddVectorEq_SSE2(a->blue_, out->blue_, NUM_LITERAL_CODES);
-    AddVectorEq_SSE2(a->alpha_, out->alpha_, NUM_LITERAL_CODES);
-  }
-  for (i = NUM_LITERAL_CODES; i < literal_size; ++i) {
-    out->literal_[i] = a->literal_[i] + b->literal_[i];
-  }
-  for (i = 0; i < NUM_DISTANCE_CODES; ++i) {
-    out->distance_[i] = a->distance_[i] + b->distance_[i];
-  }
-}
-
 //------------------------------------------------------------------------------
 // Entropy
 
@@ -675,7 +654,8 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8LEncDspInitSSE2(void) {
   VP8LTransformColor = TransformColor_SSE2;
   VP8LCollectColorBlueTransforms = CollectColorBlueTransforms_SSE2;
   VP8LCollectColorRedTransforms = CollectColorRedTransforms_SSE2;
-  VP8LHistogramAdd = HistogramAdd_SSE2;
+  VP8LAddVector = AddVector_SSE2;
+  VP8LAddVectorEq = AddVectorEq_SSE2;
   VP8LCombinedShannonEntropy = CombinedShannonEntropy_SSE2;
   VP8LVectorMismatch = VectorMismatch_SSE2;
   VP8LBundleColorMap = BundleColorMap_SSE2;

src/dsp/msa_macro.h

@@ -1389,4 +1389,4 @@ static WEBP_INLINE uint32_t func_hadd_uh_u32(v8u16 in) {
 } while (0)
 #define AVER_UB2_UB(...) AVER_UB2(v16u8, __VA_ARGS__)
 
-#endif  /* WEBP_DSP_MSA_MACRO_H_ */
+#endif  // WEBP_DSP_MSA_MACRO_H_

src/dsp/rescaler.c

@@ -21,6 +21,7 @@
 
 #define ROUNDER (WEBP_RESCALER_ONE >> 1)
 #define MULT_FIX(x, y) (((uint64_t)(x) * (y) + ROUNDER) >> WEBP_RESCALER_RFIX)
+#define MULT_FIX_FLOOR(x, y) (((uint64_t)(x) * (y)) >> WEBP_RESCALER_RFIX)
 
 //------------------------------------------------------------------------------
 // Row import
@@ -108,8 +109,7 @@ void WebPRescalerExportRowExpand_C(WebPRescaler* const wrk) {
     for (x_out = 0; x_out < x_out_max; ++x_out) {
       const uint32_t J = frow[x_out];
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   } else {
     const uint32_t B = WEBP_RESCALER_FRAC(-wrk->y_accum, wrk->y_sub);
@@ -119,8 +119,7 @@ void WebPRescalerExportRowExpand_C(WebPRescaler* const wrk) {
                        + (uint64_t)B * irow[x_out];
       const uint32_t J = (uint32_t)((I + ROUNDER) >> WEBP_RESCALER_RFIX);
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   }
 }
@@ -137,22 +136,21 @@ void WebPRescalerExportRowShrink_C(WebPRescaler* const wrk) {
   assert(!wrk->y_expand);
   if (yscale) {
     for (x_out = 0; x_out < x_out_max; ++x_out) {
-      const uint32_t frac = (uint32_t)MULT_FIX(frow[x_out], yscale);
+      const uint32_t frac = (uint32_t)MULT_FIX_FLOOR(frow[x_out], yscale);
       const int v = (int)MULT_FIX(irow[x_out] - frac, wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = frac;   // new fractional start
     }
   } else {
     for (x_out = 0; x_out < x_out_max; ++x_out) {
       const int v = (int)MULT_FIX(irow[x_out], wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = 0;
     }
   }
 }
 
+#undef MULT_FIX_FLOOR
 #undef MULT_FIX
 #undef ROUNDER
 
@@ -204,11 +202,7 @@ extern void WebPRescalerDspInitMIPSdspR2(void);
 extern void WebPRescalerDspInitMSA(void);
 extern void WebPRescalerDspInitNEON(void);
 
-static volatile VP8CPUInfo rescaler_last_cpuinfo_used =
-    (VP8CPUInfo)&rescaler_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void WebPRescalerDspInit(void) {
-  if (rescaler_last_cpuinfo_used == VP8GetCPUInfo) return;
+WEBP_DSP_INIT_FUNC(WebPRescalerDspInit) {
 #if !defined(WEBP_REDUCE_SIZE)
 #if !WEBP_NEON_OMIT_C_CODE
   WebPRescalerExportRowExpand = WebPRescalerExportRowExpand_C;
@@ -253,5 +247,4 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPRescalerDspInit(void) {
   assert(WebPRescalerImportRowExpand != NULL);
   assert(WebPRescalerImportRowShrink != NULL);
 #endif   // WEBP_REDUCE_SIZE
-  rescaler_last_cpuinfo_used = VP8GetCPUInfo;
 }

src/dsp/rescaler_mips32.c

@@ -209,6 +209,7 @@ static void ExportRowExpand_MIPS32(WebPRescaler* const wrk) {
   }
 }
 
+#if 0  // disabled for now. TODO(skal): make match the C-code
 static void ExportRowShrink_MIPS32(WebPRescaler* const wrk) {
   const int x_out_max = wrk->dst_width * wrk->num_channels;
   uint8_t* dst = wrk->dst;
@@ -273,6 +274,7 @@ static void ExportRowShrink_MIPS32(WebPRescaler* const wrk) {
     );
   }
 }
+#endif  // 0
 
 //------------------------------------------------------------------------------
 // Entry point
@@ -283,7 +285,7 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPRescalerDspInitMIPS32(void) {
   WebPRescalerImportRowExpand = ImportRowExpand_MIPS32;
   WebPRescalerImportRowShrink = ImportRowShrink_MIPS32;
   WebPRescalerExportRowExpand = ExportRowExpand_MIPS32;
-  WebPRescalerExportRowShrink = ExportRowShrink_MIPS32;
+//  WebPRescalerExportRowShrink = ExportRowShrink_MIPS32;
 }
 
 #else  // !WEBP_USE_MIPS32

src/dsp/rescaler_mips_dsp_r2.c

@@ -20,10 +20,12 @@
 
 #define ROUNDER (WEBP_RESCALER_ONE >> 1)
 #define MULT_FIX(x, y) (((uint64_t)(x) * (y) + ROUNDER) >> WEBP_RESCALER_RFIX)
+#define MULT_FIX_FLOOR(x, y) (((uint64_t)(x) * (y)) >> WEBP_RESCALER_RFIX)
 
 //------------------------------------------------------------------------------
 // Row export
 
+#if 0  // disabled for now. TODO(skal): make match the C-code
 static void ExportRowShrink_MIPSdspR2(WebPRescaler* const wrk) {
   int i;
   const int x_out_max = wrk->dst_width * wrk->num_channels;
@@ -105,10 +107,9 @@ static void ExportRowShrink_MIPSdspR2(WebPRescaler* const wrk) {
       );
     }
     for (i = 0; i < (x_out_max & 0x3); ++i) {
-      const uint32_t frac = (uint32_t)MULT_FIX(*frow++, yscale);
+      const uint32_t frac = (uint32_t)MULT_FIX_FLOOR(*frow++, yscale);
       const int v = (int)MULT_FIX(*irow - frac, wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      *dst++ = v;
+      *dst++ = (v > 255) ? 255u : (uint8_t)v;
       *irow++ = frac;   // new fractional start
     }
   } else {
@@ -154,13 +155,13 @@ static void ExportRowShrink_MIPSdspR2(WebPRescaler* const wrk) {
       );
     }
     for (i = 0; i < (x_out_max & 0x3); ++i) {
-      const int v = (int)MULT_FIX(*irow, wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      *dst++ = v;
+      const int v = (int)MULT_FIX_FLOOR(*irow, wrk->fxy_scale);
+      *dst++ = (v > 255) ? 255u : (uint8_t)v;
       *irow++ = 0;
     }
   }
 }
+#endif  // 0
 
 static void ExportRowExpand_MIPSdspR2(WebPRescaler* const wrk) {
   int i;
@@ -216,8 +217,7 @@ static void ExportRowExpand_MIPSdspR2(WebPRescaler* const wrk) {
     for (i = 0; i < (x_out_max & 0x3); ++i) {
       const uint32_t J = *frow++;
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      *dst++ = v;
+      *dst++ = (v > 255) ? 255u : (uint8_t)v;
     }
   } else {
     const uint32_t B = WEBP_RESCALER_FRAC(-wrk->y_accum, wrk->y_sub);
@@ -288,12 +288,12 @@ static void ExportRowExpand_MIPSdspR2(WebPRescaler* const wrk) {
                        + (uint64_t)B * *irow++;
       const uint32_t J = (uint32_t)((I + ROUNDER) >> WEBP_RESCALER_RFIX);
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      *dst++ = v;
+      *dst++ = (v > 255) ? 255u : (uint8_t)v;
     }
   }
 }
 
+#undef MULT_FIX_FLOOR
 #undef MULT_FIX
 #undef ROUNDER
 
@@ -304,7 +304,7 @@ extern void WebPRescalerDspInitMIPSdspR2(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void WebPRescalerDspInitMIPSdspR2(void) {
   WebPRescalerExportRowExpand = ExportRowExpand_MIPSdspR2;
-  WebPRescalerExportRowShrink = ExportRowShrink_MIPSdspR2;
+//  WebPRescalerExportRowShrink = ExportRowShrink_MIPSdspR2;
 }
 
 #else  // !WEBP_USE_MIPS_DSP_R2

src/dsp/rescaler_msa.c

@@ -22,6 +22,7 @@
 
 #define ROUNDER (WEBP_RESCALER_ONE >> 1)
 #define MULT_FIX(x, y) (((uint64_t)(x) * (y) + ROUNDER) >> WEBP_RESCALER_RFIX)
+#define MULT_FIX_FLOOR(x, y) (((uint64_t)(x) * (y)) >> WEBP_RESCALER_RFIX)
 
 #define CALC_MULT_FIX_16(in0, in1, in2, in3, scale, shift, dst) do {  \
   v4u32 tmp0, tmp1, tmp2, tmp3;                                       \
@@ -165,8 +166,7 @@ static WEBP_INLINE void ExportRowExpand_0(const uint32_t* frow, uint8_t* dst,
     for (x_out = 0; x_out < length; ++x_out) {
       const uint32_t J = frow[x_out];
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   }
 }
@@ -240,8 +240,7 @@ static WEBP_INLINE void ExportRowExpand_1(const uint32_t* frow, uint32_t* irow,
                        + (uint64_t)B * irow[x_out];
       const uint32_t J = (uint32_t)((I + ROUNDER) >> WEBP_RESCALER_RFIX);
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   }
 }
@@ -262,6 +261,7 @@ static void RescalerExportRowExpand_MIPSdspR2(WebPRescaler* const wrk) {
   }
 }
 
+#if 0  // disabled for now. TODO(skal): make match the C-code
 static WEBP_INLINE void ExportRowShrink_0(const uint32_t* frow, uint32_t* irow,
                                           uint8_t* dst, int length,
                                           const uint32_t yscale,
@@ -340,10 +340,9 @@ static WEBP_INLINE void ExportRowShrink_0(const uint32_t* frow, uint32_t* irow,
       length -= 4;
     }
     for (x_out = 0; x_out < length; ++x_out) {
-      const uint32_t frac = (uint32_t)MULT_FIX(frow[x_out], yscale);
+      const uint32_t frac = (uint32_t)MULT_FIX_FLOOR(frow[x_out], yscale);
       const int v = (int)MULT_FIX(irow[x_out] - frac, wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = frac;
     }
   }
@@ -404,8 +403,7 @@ static WEBP_INLINE void ExportRowShrink_1(uint32_t* irow, uint8_t* dst,
     }
     for (x_out = 0; x_out < length; ++x_out) {
       const int v = (int)MULT_FIX(irow[x_out], wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = 0;
     }
   }
@@ -426,6 +424,7 @@ static void RescalerExportRowShrink_MIPSdspR2(WebPRescaler* const wrk) {
     ExportRowShrink_1(irow, dst, x_out_max, wrk);
   }
 }
+#endif  // 0
 
 //------------------------------------------------------------------------------
 // Entry point
@@ -434,7 +433,7 @@ extern void WebPRescalerDspInitMSA(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void WebPRescalerDspInitMSA(void) {
   WebPRescalerExportRowExpand = RescalerExportRowExpand_MIPSdspR2;
-  WebPRescalerExportRowShrink = RescalerExportRowShrink_MIPSdspR2;
+//  WebPRescalerExportRowShrink = RescalerExportRowShrink_MIPSdspR2;
 }
 
 #else     // !WEBP_USE_MSA

src/dsp/rescaler_neon.c

@@ -22,6 +22,7 @@
 
 #define ROUNDER (WEBP_RESCALER_ONE >> 1)
 #define MULT_FIX_C(x, y) (((uint64_t)(x) * (y) + ROUNDER) >> WEBP_RESCALER_RFIX)
+#define MULT_FIX_FLOOR_C(x, y) (((uint64_t)(x) * (y)) >> WEBP_RESCALER_RFIX)
 
 #define LOAD_32x4(SRC, DST) const uint32x4_t DST = vld1q_u32((SRC))
 #define LOAD_32x8(SRC, DST0, DST1)                                    \
@@ -35,8 +36,11 @@
 
 #if (WEBP_RESCALER_RFIX == 32)
 #define MAKE_HALF_CST(C) vdupq_n_s32((int32_t)((C) >> 1))
-#define MULT_FIX(A, B) /* note: B is actualy scale>>1. See MAKE_HALF_CST */ \
+// note: B is actualy scale>>1. See MAKE_HALF_CST
+#define MULT_FIX(A, B) \
     vreinterpretq_u32_s32(vqrdmulhq_s32(vreinterpretq_s32_u32((A)), (B)))
+#define MULT_FIX_FLOOR(A, B) \
+    vreinterpretq_u32_s32(vqdmulhq_s32(vreinterpretq_s32_u32((A)), (B)))
 #else
 #error "MULT_FIX/WEBP_RESCALER_RFIX need some more work"
 #endif
@@ -77,14 +81,13 @@ static void RescalerExportRowExpand_NEON(WebPRescaler* const wrk) {
       const uint32x4_t B1 = MULT_FIX(A1, fy_scale_half);
       const uint16x4_t C0 = vmovn_u32(B0);
       const uint16x4_t C1 = vmovn_u32(B1);
-      const uint8x8_t D = vmovn_u16(vcombine_u16(C0, C1));
+      const uint8x8_t D = vqmovn_u16(vcombine_u16(C0, C1));
       vst1_u8(dst + x_out, D);
     }
     for (; x_out < x_out_max; ++x_out) {
       const uint32_t J = frow[x_out];
       const int v = (int)MULT_FIX_C(J, fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   } else {
     const uint32_t B = WEBP_RESCALER_FRAC(-wrk->y_accum, wrk->y_sub);
@@ -98,7 +101,7 @@ static void RescalerExportRowExpand_NEON(WebPRescaler* const wrk) {
       const uint32x4_t D1 = MULT_FIX(C1, fy_scale_half);
       const uint16x4_t E0 = vmovn_u32(D0);
       const uint16x4_t E1 = vmovn_u32(D1);
-      const uint8x8_t F = vmovn_u16(vcombine_u16(E0, E1));
+      const uint8x8_t F = vqmovn_u16(vcombine_u16(E0, E1));
       vst1_u8(dst + x_out, F);
     }
     for (; x_out < x_out_max; ++x_out) {
@@ -106,8 +109,7 @@ static void RescalerExportRowExpand_NEON(WebPRescaler* const wrk) {
                        + (uint64_t)B * irow[x_out];
       const uint32_t J = (uint32_t)((I + ROUNDER) >> WEBP_RESCALER_RFIX);
       const int v = (int)MULT_FIX_C(J, fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   }
 }
@@ -131,23 +133,22 @@ static void RescalerExportRowShrink_NEON(WebPRescaler* const wrk) {
     for (x_out = 0; x_out < max_span; x_out += 8) {
       LOAD_32x8(frow + x_out, in0, in1);
       LOAD_32x8(irow + x_out, in2, in3);
-      const uint32x4_t A0 = MULT_FIX(in0, yscale_half);
-      const uint32x4_t A1 = MULT_FIX(in1, yscale_half);
+      const uint32x4_t A0 = MULT_FIX_FLOOR(in0, yscale_half);
+      const uint32x4_t A1 = MULT_FIX_FLOOR(in1, yscale_half);
       const uint32x4_t B0 = vqsubq_u32(in2, A0);
       const uint32x4_t B1 = vqsubq_u32(in3, A1);
       const uint32x4_t C0 = MULT_FIX(B0, fxy_scale_half);
       const uint32x4_t C1 = MULT_FIX(B1, fxy_scale_half);
       const uint16x4_t D0 = vmovn_u32(C0);
       const uint16x4_t D1 = vmovn_u32(C1);
-      const uint8x8_t E = vmovn_u16(vcombine_u16(D0, D1));
+      const uint8x8_t E = vqmovn_u16(vcombine_u16(D0, D1));
       vst1_u8(dst + x_out, E);
       STORE_32x8(A0, A1, irow + x_out);
     }
     for (; x_out < x_out_max; ++x_out) {
-      const uint32_t frac = (uint32_t)MULT_FIX_C(frow[x_out], yscale);
-      const int v = (int)MULT_FIX_C(irow[x_out] - frac, wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      const uint32_t frac = (uint32_t)MULT_FIX_FLOOR_C(frow[x_out], yscale);
+      const int v = (int)MULT_FIX_C(irow[x_out] - frac, fxy_scale);
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = frac;   // new fractional start
     }
   } else {
@@ -157,19 +158,24 @@ static void RescalerExportRowShrink_NEON(WebPRescaler* const wrk) {
       const uint32x4_t A1 = MULT_FIX(in1, fxy_scale_half);
       const uint16x4_t B0 = vmovn_u32(A0);
       const uint16x4_t B1 = vmovn_u32(A1);
-      const uint8x8_t C = vmovn_u16(vcombine_u16(B0, B1));
+      const uint8x8_t C = vqmovn_u16(vcombine_u16(B0, B1));
       vst1_u8(dst + x_out, C);
       STORE_32x8(zero, zero, irow + x_out);
     }
     for (; x_out < x_out_max; ++x_out) {
       const int v = (int)MULT_FIX_C(irow[x_out], fxy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = 0;
     }
   }
 }
 
+#undef MULT_FIX_FLOOR_C
+#undef MULT_FIX_C
+#undef MULT_FIX_FLOOR
+#undef MULT_FIX
+#undef ROUNDER
+
 //------------------------------------------------------------------------------
 
 extern void WebPRescalerDspInitNEON(void);

src/dsp/rescaler_sse2.c

@@ -25,6 +25,7 @@
 
 #define ROUNDER (WEBP_RESCALER_ONE >> 1)
 #define MULT_FIX(x, y) (((uint64_t)(x) * (y) + ROUNDER) >> WEBP_RESCALER_RFIX)
+#define MULT_FIX_FLOOR(x, y) (((uint64_t)(x) * (y)) >> WEBP_RESCALER_RFIX)
 
 // input: 8 bytes ABCDEFGH -> output: A0E0B0F0C0G0D0H0
 static void LoadTwoPixels_SSE2(const uint8_t* const src, __m128i* out) {
@@ -36,7 +37,7 @@ static void LoadTwoPixels_SSE2(const uint8_t* const src, __m128i* out) {
 }
 
 // input: 8 bytes ABCDEFGH -> output: A0B0C0D0E0F0G0H0
-static void LoadHeightPixels_SSE2(const uint8_t* const src, __m128i* out) {
+static void LoadEightPixels_SSE2(const uint8_t* const src, __m128i* out) {
   const __m128i zero = _mm_setzero_si128();
   const __m128i A = _mm_loadl_epi64((const __m128i*)(src));  // ABCDEFGH
   *out = _mm_unpacklo_epi8(A, zero);
@@ -50,13 +51,15 @@ static void RescalerImportRowExpand_SSE2(WebPRescaler* const wrk,
   int accum = x_add;
   __m128i cur_pixels;
 
+  // SSE2 implementation only works with 16b signed arithmetic at max.
+  if (wrk->src_width < 8 || accum >= (1 << 15)) {
+    WebPRescalerImportRowExpand_C(wrk, src);
+    return;
+  }
+
   assert(!WebPRescalerInputDone(wrk));
   assert(wrk->x_expand);
   if (wrk->num_channels == 4) {
-    if (wrk->src_width < 2) {
-      WebPRescalerImportRowExpand_C(wrk, src);
-      return;
-    }
     LoadTwoPixels_SSE2(src, &cur_pixels);
     src += 4;
     while (1) {
@@ -75,11 +78,7 @@ static void RescalerImportRowExpand_SSE2(WebPRescaler* const wrk,
   } else {
     int left;
     const uint8_t* const src_limit = src + wrk->src_width - 8;
-    if (wrk->src_width < 8) {
-      WebPRescalerImportRowExpand_C(wrk, src);
-      return;
-    }
-    LoadHeightPixels_SSE2(src, &cur_pixels);
+    LoadEightPixels_SSE2(src, &cur_pixels);
     src += 7;
     left = 7;
     while (1) {
@@ -94,7 +93,7 @@ static void RescalerImportRowExpand_SSE2(WebPRescaler* const wrk,
         if (--left) {
           cur_pixels = _mm_srli_si128(cur_pixels, 2);
         } else if (src <= src_limit) {
-          LoadHeightPixels_SSE2(src, &cur_pixels);
+          LoadEightPixels_SSE2(src, &cur_pixels);
           src += 7;
           left = 7;
         } else {   // tail
@@ -246,8 +245,7 @@ static void RescalerExportRowExpand_SSE2(WebPRescaler* const wrk) {
     for (; x_out < x_out_max; ++x_out) {
       const uint32_t J = frow[x_out];
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   } else {
     const uint32_t B = WEBP_RESCALER_FRAC(-wrk->y_accum, wrk->y_sub);
@@ -280,8 +278,7 @@ static void RescalerExportRowExpand_SSE2(WebPRescaler* const wrk) {
                        + (uint64_t)B * irow[x_out];
       const uint32_t J = (uint32_t)((I + ROUNDER) >> WEBP_RESCALER_RFIX);
       const int v = (int)MULT_FIX(J, wrk->fy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
     }
   }
 }
@@ -300,20 +297,15 @@ static void RescalerExportRowShrink_SSE2(WebPRescaler* const wrk) {
     const int scale_xy = wrk->fxy_scale;
     const __m128i mult_xy = _mm_set_epi32(0, scale_xy, 0, scale_xy);
     const __m128i mult_y = _mm_set_epi32(0, yscale, 0, yscale);
-    const __m128i rounder = _mm_set_epi32(0, ROUNDER, 0, ROUNDER);
     for (x_out = 0; x_out + 8 <= x_out_max; x_out += 8) {
       __m128i A0, A1, A2, A3, B0, B1, B2, B3;
       LoadDispatchAndMult_SSE2(irow + x_out, NULL, &A0, &A1, &A2, &A3);
       LoadDispatchAndMult_SSE2(frow + x_out, &mult_y, &B0, &B1, &B2, &B3);
       {
-        const __m128i C0 = _mm_add_epi64(B0, rounder);
-        const __m128i C1 = _mm_add_epi64(B1, rounder);
-        const __m128i C2 = _mm_add_epi64(B2, rounder);
-        const __m128i C3 = _mm_add_epi64(B3, rounder);
-        const __m128i D0 = _mm_srli_epi64(C0, WEBP_RESCALER_RFIX);   // = frac
-        const __m128i D1 = _mm_srli_epi64(C1, WEBP_RESCALER_RFIX);
-        const __m128i D2 = _mm_srli_epi64(C2, WEBP_RESCALER_RFIX);
-        const __m128i D3 = _mm_srli_epi64(C3, WEBP_RESCALER_RFIX);
+        const __m128i D0 = _mm_srli_epi64(B0, WEBP_RESCALER_RFIX);   // = frac
+        const __m128i D1 = _mm_srli_epi64(B1, WEBP_RESCALER_RFIX);
+        const __m128i D2 = _mm_srli_epi64(B2, WEBP_RESCALER_RFIX);
+        const __m128i D3 = _mm_srli_epi64(B3, WEBP_RESCALER_RFIX);
         const __m128i E0 = _mm_sub_epi64(A0, D0);   // irow[x] - frac
         const __m128i E1 = _mm_sub_epi64(A1, D1);
         const __m128i E2 = _mm_sub_epi64(A2, D2);
@@ -328,10 +320,9 @@ static void RescalerExportRowShrink_SSE2(WebPRescaler* const wrk) {
       }
     }
     for (; x_out < x_out_max; ++x_out) {
-      const uint32_t frac = (int)MULT_FIX(frow[x_out], yscale);
+      const uint32_t frac = (int)MULT_FIX_FLOOR(frow[x_out], yscale);
       const int v = (int)MULT_FIX(irow[x_out] - frac, wrk->fxy_scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = frac;   // new fractional start
     }
   } else {
@@ -347,13 +338,13 @@ static void RescalerExportRowShrink_SSE2(WebPRescaler* const wrk) {
     }
     for (; x_out < x_out_max; ++x_out) {
       const int v = (int)MULT_FIX(irow[x_out], scale);
-      assert(v >= 0 && v <= 255);
-      dst[x_out] = v;
+      dst[x_out] = (v > 255) ? 255u : (uint8_t)v;
       irow[x_out] = 0;
     }
   }
 }
 
+#undef MULT_FIX_FLOOR
 #undef MULT_FIX
 #undef ROUNDER
 

src/dsp/ssim.c

@@ -139,12 +139,7 @@ VP8AccumulateSSEFunc VP8AccumulateSSE;
 
 extern void VP8SSIMDspInitSSE2(void);
 
-static volatile VP8CPUInfo ssim_last_cpuinfo_used =
-    (VP8CPUInfo)&ssim_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void VP8SSIMDspInit(void) {
-  if (ssim_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(VP8SSIMDspInit) {
 #if !defined(WEBP_REDUCE_SIZE)
   VP8SSIMGetClipped = SSIMGetClipped_C;
   VP8SSIMGet = SSIMGet_C;
@@ -161,6 +156,4 @@ WEBP_TSAN_IGNORE_FUNCTION void VP8SSIMDspInit(void) {
     }
 #endif
   }
-
-  ssim_last_cpuinfo_used = VP8GetCPUInfo;
 }

src/dsp/upsampling.c

@@ -219,12 +219,7 @@ extern void WebPInitYUV444ConvertersMIPSdspR2(void);
 extern void WebPInitYUV444ConvertersSSE2(void);
 extern void WebPInitYUV444ConvertersSSE41(void);
 
-static volatile VP8CPUInfo upsampling_last_cpuinfo_used1 =
-    (VP8CPUInfo)&upsampling_last_cpuinfo_used1;
-
-WEBP_TSAN_IGNORE_FUNCTION void WebPInitYUV444Converters(void) {
-  if (upsampling_last_cpuinfo_used1 == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(WebPInitYUV444Converters) {
   WebPYUV444Converters[MODE_RGBA]      = WebPYuv444ToRgba_C;
   WebPYUV444Converters[MODE_BGRA]      = WebPYuv444ToBgra_C;
   WebPYUV444Converters[MODE_RGB]       = WebPYuv444ToRgb_C;
@@ -254,7 +249,6 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPInitYUV444Converters(void) {
     }
 #endif
   }
-  upsampling_last_cpuinfo_used1 = VP8GetCPUInfo;
 }
 
 //------------------------------------------------------------------------------
@@ -266,12 +260,7 @@ extern void WebPInitUpsamplersNEON(void);
 extern void WebPInitUpsamplersMIPSdspR2(void);
 extern void WebPInitUpsamplersMSA(void);
 
-static volatile VP8CPUInfo upsampling_last_cpuinfo_used2 =
-    (VP8CPUInfo)&upsampling_last_cpuinfo_used2;
-
-WEBP_TSAN_IGNORE_FUNCTION void WebPInitUpsamplers(void) {
-  if (upsampling_last_cpuinfo_used2 == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(WebPInitUpsamplers) {
 #ifdef FANCY_UPSAMPLING
 #if !WEBP_NEON_OMIT_C_CODE
   WebPUpsamplers[MODE_RGBA]      = UpsampleRgbaLinePair_C;
@@ -333,7 +322,6 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPInitUpsamplers(void) {
 #endif
 
 #endif  // FANCY_UPSAMPLING
-  upsampling_last_cpuinfo_used2 = VP8GetCPUInfo;
 }
 
 //------------------------------------------------------------------------------

src/dsp/yuv.c

@@ -75,12 +75,7 @@ extern void WebPInitSamplersSSE41(void);
 extern void WebPInitSamplersMIPS32(void);
 extern void WebPInitSamplersMIPSdspR2(void);
 
-static volatile VP8CPUInfo yuv_last_cpuinfo_used =
-    (VP8CPUInfo)&yuv_last_cpuinfo_used;
-
-WEBP_TSAN_IGNORE_FUNCTION void WebPInitSamplers(void) {
-  if (yuv_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(WebPInitSamplers) {
   WebPSamplers[MODE_RGB]       = YuvToRgbRow;
   WebPSamplers[MODE_RGBA]      = YuvToRgbaRow;
   WebPSamplers[MODE_BGR]       = YuvToBgrRow;
@@ -116,7 +111,6 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPInitSamplers(void) {
     }
 #endif  // WEBP_USE_MIPS_DSP_R2
   }
-  yuv_last_cpuinfo_used = VP8GetCPUInfo;
 }
 
 //-----------------------------------------------------------------------------
@@ -260,18 +254,13 @@ void (*WebPSharpYUVUpdateRGB)(const int16_t* ref, const int16_t* src,
 void (*WebPSharpYUVFilterRow)(const int16_t* A, const int16_t* B, int len,
                               const uint16_t* best_y, uint16_t* out);
 
-static volatile VP8CPUInfo rgba_to_yuv_last_cpuinfo_used =
-    (VP8CPUInfo)&rgba_to_yuv_last_cpuinfo_used;
-
 extern void WebPInitConvertARGBToYUVSSE2(void);
 extern void WebPInitConvertARGBToYUVSSE41(void);
 extern void WebPInitConvertARGBToYUVNEON(void);
 extern void WebPInitSharpYUVSSE2(void);
 extern void WebPInitSharpYUVNEON(void);
 
-WEBP_TSAN_IGNORE_FUNCTION void WebPInitConvertARGBToYUV(void) {
-  if (rgba_to_yuv_last_cpuinfo_used == VP8GetCPUInfo) return;
-
+WEBP_DSP_INIT_FUNC(WebPInitConvertARGBToYUV) {
   WebPConvertARGBToY = ConvertARGBToY_C;
   WebPConvertARGBToUV = WebPConvertARGBToUV_C;
 
@@ -316,6 +305,4 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPInitConvertARGBToYUV(void) {
   assert(WebPSharpYUVUpdateY != NULL);
   assert(WebPSharpYUVUpdateRGB != NULL);
   assert(WebPSharpYUVFilterRow != NULL);
-
-  rgba_to_yuv_last_cpuinfo_used = VP8GetCPUInfo;
 }

src/dsp/yuv.h

@@ -207,4 +207,4 @@ static WEBP_INLINE int VP8RGBToV(int r, int g, int b, int rounding) {
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DSP_YUV_H_ */
+#endif  // WEBP_DSP_YUV_H_

src/enc/analysis_enc.c

@@ -458,7 +458,7 @@ static void MergeJobs(const SegmentJob* const src, SegmentJob* const dst) {
   dst->uv_alpha += src->uv_alpha;
 }
 
-// initialize the job struct with some TODOs
+// initialize the job struct with some tasks to perform
 static void InitSegmentJob(VP8Encoder* const enc, SegmentJob* const job,
                            int start_row, int end_row) {
   WebPGetWorkerInterface()->Init(&job->worker);

src/enc/backward_references_cost_enc.c

@@ -67,7 +67,7 @@ static int CostModelBuild(CostModel* const m, int xsize, int cache_bits,
 
   // The following code is similar to VP8LHistogramCreate but converts the
   // distance to plane code.
-  VP8LHistogramInit(histo, cache_bits);
+  VP8LHistogramInit(histo, cache_bits, /*init_arrays=*/ 1);
   while (VP8LRefsCursorOk(&c)) {
     VP8LHistogramAddSinglePixOrCopy(histo, c.cur_pos, VP8LDistanceToPlaneCode,
                                     xsize);
@@ -577,7 +577,7 @@ static int BackwardReferencesHashChainDistanceOnly(
       (CostModel*)WebPSafeCalloc(1ULL, cost_model_size);
   VP8LColorCache hashers;
   CostManager* cost_manager =
-      (CostManager*)WebPSafeMalloc(1ULL, sizeof(*cost_manager));
+      (CostManager*)WebPSafeCalloc(1ULL, sizeof(*cost_manager));
   int offset_prev = -1, len_prev = -1;
   double offset_cost = -1;
   int first_offset_is_constant = -1;  // initialized with 'impossible' value

src/enc/backward_references_enc.c

@@ -715,6 +715,7 @@ static int CalculateBestCacheSize(const uint32_t* argb, int quality,
   for (i = 0; i <= cache_bits_max; ++i) {
     histos[i] = VP8LAllocateHistogram(i);
     if (histos[i] == NULL) goto Error;
+    VP8LHistogramInit(histos[i], i, /*init_arrays=*/ 1);
     if (i == 0) continue;
     cc_init[i] = VP8LColorCacheInit(&hashers[i], i);
     if (!cc_init[i]) goto Error;
@@ -910,13 +911,14 @@ static VP8LBackwardRefs* GetBackwardReferences(
       quality >= 25) {
     const VP8LHashChain* const hash_chain_tmp =
         (lz77_type_best == kLZ77Standard) ? hash_chain : &hash_chain_box;
-    if (VP8LBackwardReferencesTraceBackwards(width, height, argb, *cache_bits,
-                                             hash_chain_tmp, best, worst)) {
-      double bit_cost_trace;
-      VP8LHistogramCreate(histo, worst, *cache_bits);
-      bit_cost_trace = VP8LHistogramEstimateBits(histo);
-      if (bit_cost_trace < bit_cost_best) best = worst;
+    double bit_cost_trace;
+    if (!VP8LBackwardReferencesTraceBackwards(width, height, argb, *cache_bits,
+                                              hash_chain_tmp, best, worst)) {
+      goto Error;
     }
+    VP8LHistogramCreate(histo, worst, *cache_bits);
+    bit_cost_trace = VP8LHistogramEstimateBits(histo);
+    if (bit_cost_trace < bit_cost_best) best = worst;
   }
 
   BackwardReferences2DLocality(width, best);

src/enc/cost_enc.h

@@ -79,4 +79,4 @@ extern const uint16_t VP8FixedCostsI4[NUM_BMODES][NUM_BMODES][NUM_BMODES];
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_ENC_COST_ENC_H_ */
+#endif  // WEBP_ENC_COST_ENC_H_

src/enc/histogram_enc.c

@@ -51,10 +51,12 @@ static void HistogramCopy(const VP8LHistogram* const src,
                           VP8LHistogram* const dst) {
   uint32_t* const dst_literal = dst->literal_;
   const int dst_cache_bits = dst->palette_code_bits_;
+  const int literal_size = VP8LHistogramNumCodes(dst_cache_bits);
   const int histo_size = VP8LGetHistogramSize(dst_cache_bits);
   assert(src->palette_code_bits_ == dst_cache_bits);
   memcpy(dst, src, histo_size);
   dst->literal_ = dst_literal;
+  memcpy(dst->literal_, src->literal_, literal_size * sizeof(*dst->literal_));
 }
 
 int VP8LGetHistogramSize(int cache_bits) {
@@ -91,9 +93,19 @@ void VP8LHistogramCreate(VP8LHistogram* const p,
   VP8LHistogramStoreRefs(refs, p);
 }
 
-void VP8LHistogramInit(VP8LHistogram* const p, int palette_code_bits) {
+void VP8LHistogramInit(VP8LHistogram* const p, int palette_code_bits,
+                       int init_arrays) {
   p->palette_code_bits_ = palette_code_bits;
-  HistogramClear(p);
+  if (init_arrays) {
+    HistogramClear(p);
+  } else {
+    p->trivial_symbol_ = 0;
+    p->bit_cost_ = 0.;
+    p->literal_cost_ = 0.;
+    p->red_cost_ = 0.;
+    p->blue_cost_ = 0.;
+    memset(p->is_used_, 0, sizeof(p->is_used_));
+  }
 }
 
 VP8LHistogram* VP8LAllocateHistogram(int cache_bits) {
@@ -104,37 +116,70 @@ VP8LHistogram* VP8LAllocateHistogram(int cache_bits) {
   histo = (VP8LHistogram*)memory;
   // literal_ won't necessary be aligned.
   histo->literal_ = (uint32_t*)(memory + sizeof(VP8LHistogram));
-  VP8LHistogramInit(histo, cache_bits);
+  VP8LHistogramInit(histo, cache_bits, /*init_arrays=*/ 0);
   return histo;
 }
 
+// Resets the pointers of the histograms to point to the bit buffer in the set.
+static void HistogramSetResetPointers(VP8LHistogramSet* const set,
+                                      int cache_bits) {
+  int i;
+  const int histo_size = VP8LGetHistogramSize(cache_bits);
+  uint8_t* memory = (uint8_t*) (set->histograms);
+  memory += set->max_size * sizeof(*set->histograms);
+  for (i = 0; i < set->max_size; ++i) {
+    memory = (uint8_t*) WEBP_ALIGN(memory);
+    set->histograms[i] = (VP8LHistogram*) memory;
+    // literal_ won't necessary be aligned.
+    set->histograms[i]->literal_ = (uint32_t*)(memory + sizeof(VP8LHistogram));
+    memory += histo_size;
+  }
+}
+
+// Returns the total size of the VP8LHistogramSet.
+static size_t HistogramSetTotalSize(int size, int cache_bits) {
+  const int histo_size = VP8LGetHistogramSize(cache_bits);
+  return (sizeof(VP8LHistogramSet) + size * (sizeof(VP8LHistogram*) +
+          histo_size + WEBP_ALIGN_CST));
+}
+
 VP8LHistogramSet* VP8LAllocateHistogramSet(int size, int cache_bits) {
   int i;
   VP8LHistogramSet* set;
-  const int histo_size = VP8LGetHistogramSize(cache_bits);
-  const size_t total_size =
-      sizeof(*set) + size * (sizeof(*set->histograms) +
-      histo_size + WEBP_ALIGN_CST);
+  const size_t total_size = HistogramSetTotalSize(size, cache_bits);
   uint8_t* memory = (uint8_t*)WebPSafeMalloc(total_size, sizeof(*memory));
   if (memory == NULL) return NULL;
 
   set = (VP8LHistogramSet*)memory;
   memory += sizeof(*set);
   set->histograms = (VP8LHistogram**)memory;
-  memory += size * sizeof(*set->histograms);
   set->max_size = size;
   set->size = size;
+  HistogramSetResetPointers(set, cache_bits);
   for (i = 0; i < size; ++i) {
-    memory = (uint8_t*)WEBP_ALIGN(memory);
-    set->histograms[i] = (VP8LHistogram*)memory;
-    // literal_ won't necessary be aligned.
-    set->histograms[i]->literal_ = (uint32_t*)(memory + sizeof(VP8LHistogram));
-    VP8LHistogramInit(set->histograms[i], cache_bits);
-    memory += histo_size;
+    VP8LHistogramInit(set->histograms[i], cache_bits, /*init_arrays=*/ 0);
   }
   return set;
 }
 
+void VP8LHistogramSetClear(VP8LHistogramSet* const set) {
+  int i;
+  const int cache_bits = set->histograms[0]->palette_code_bits_;
+  const int size = set->size;
+  const size_t total_size = HistogramSetTotalSize(size, cache_bits);
+  uint8_t* memory = (uint8_t*)set;
+
+  memset(memory, 0, total_size);
+  memory += sizeof(*set);
+  set->histograms = (VP8LHistogram**)memory;
+  set->max_size = size;
+  set->size = size;
+  HistogramSetResetPointers(set, cache_bits);
+  for (i = 0; i < size; ++i) {
+    set->histograms[i]->palette_code_bits_ = cache_bits;
+  }
+}
+
 // -----------------------------------------------------------------------------
 
 void VP8LHistogramAddSinglePixOrCopy(VP8LHistogram* const histo,
@@ -237,7 +282,8 @@ static double FinalHuffmanCost(const VP8LStreaks* const stats) {
 // Get the symbol entropy for the distribution 'population'.
 // Set 'trivial_sym', if there's only one symbol present in the distribution.
 static double PopulationCost(const uint32_t* const population, int length,
-                             uint32_t* const trivial_sym) {
+                             uint32_t* const trivial_sym,
+                             uint8_t* const is_used) {
   VP8LBitEntropy bit_entropy;
   VP8LStreaks stats;
   VP8LGetEntropyUnrefined(population, length, &bit_entropy, &stats);
@@ -245,6 +291,8 @@ static double PopulationCost(const uint32_t* const population, int length,
     *trivial_sym = (bit_entropy.nonzeros == 1) ? bit_entropy.nonzero_code
                                                : VP8L_NON_TRIVIAL_SYM;
   }
+  // The histogram is used if there is at least one non-zero streak.
+  *is_used = (stats.streaks[1][0] != 0 || stats.streaks[1][1] != 0);
 
   return BitsEntropyRefine(&bit_entropy) + FinalHuffmanCost(&stats);
 }
@@ -253,7 +301,9 @@ static double PopulationCost(const uint32_t* const population, int length,
 // non-zero: both the zero-th one, or both the last one.
 static WEBP_INLINE double GetCombinedEntropy(const uint32_t* const X,
                                              const uint32_t* const Y,
-                                             int length, int trivial_at_end) {
+                                             int length, int is_X_used,
+                                             int is_Y_used,
+                                             int trivial_at_end) {
   VP8LStreaks stats;
   if (trivial_at_end) {
     // This configuration is due to palettization that transforms an indexed
@@ -262,28 +312,43 @@ static WEBP_INLINE double GetCombinedEntropy(const uint32_t* const X,
     // Only FinalHuffmanCost needs to be evaluated.
     memset(&stats, 0, sizeof(stats));
     // Deal with the non-zero value at index 0 or length-1.
-    stats.streaks[1][0] += 1;
+    stats.streaks[1][0] = 1;
     // Deal with the following/previous zero streak.
-    stats.counts[0] += 1;
-    stats.streaks[0][1] += length - 1;
+    stats.counts[0] = 1;
+    stats.streaks[0][1] = length - 1;
     return FinalHuffmanCost(&stats);
   } else {
     VP8LBitEntropy bit_entropy;
-    VP8LGetCombinedEntropyUnrefined(X, Y, length, &bit_entropy, &stats);
+    if (is_X_used) {
+      if (is_Y_used) {
+        VP8LGetCombinedEntropyUnrefined(X, Y, length, &bit_entropy, &stats);
+      } else {
+        VP8LGetEntropyUnrefined(X, length, &bit_entropy, &stats);
+      }
+    } else {
+      if (is_Y_used) {
+        VP8LGetEntropyUnrefined(Y, length, &bit_entropy, &stats);
+      } else {
+        memset(&stats, 0, sizeof(stats));
+        stats.counts[0] = 1;
+        stats.streaks[0][length > 3] = length;
+        VP8LBitEntropyInit(&bit_entropy);
+      }
+    }
 
     return BitsEntropyRefine(&bit_entropy) + FinalHuffmanCost(&stats);
   }
 }
 
 // Estimates the Entropy + Huffman + other block overhead size cost.
-double VP8LHistogramEstimateBits(const VP8LHistogram* const p) {
+double VP8LHistogramEstimateBits(VP8LHistogram* const p) {
   return
-      PopulationCost(
-          p->literal_, VP8LHistogramNumCodes(p->palette_code_bits_), NULL)
-      + PopulationCost(p->red_, NUM_LITERAL_CODES, NULL)
-      + PopulationCost(p->blue_, NUM_LITERAL_CODES, NULL)
-      + PopulationCost(p->alpha_, NUM_LITERAL_CODES, NULL)
-      + PopulationCost(p->distance_, NUM_DISTANCE_CODES, NULL)
+      PopulationCost(p->literal_, VP8LHistogramNumCodes(p->palette_code_bits_),
+                     NULL, &p->is_used_[0])
+      + PopulationCost(p->red_, NUM_LITERAL_CODES, NULL, &p->is_used_[1])
+      + PopulationCost(p->blue_, NUM_LITERAL_CODES, NULL, &p->is_used_[2])
+      + PopulationCost(p->alpha_, NUM_LITERAL_CODES, NULL, &p->is_used_[3])
+      + PopulationCost(p->distance_, NUM_DISTANCE_CODES, NULL, &p->is_used_[4])
       + VP8LExtraCost(p->literal_ + NUM_LITERAL_CODES, NUM_LENGTH_CODES)
       + VP8LExtraCost(p->distance_, NUM_DISTANCE_CODES);
 }
@@ -299,7 +364,8 @@ static int GetCombinedHistogramEntropy(const VP8LHistogram* const a,
   int trivial_at_end = 0;
   assert(a->palette_code_bits_ == b->palette_code_bits_);
   *cost += GetCombinedEntropy(a->literal_, b->literal_,
-                              VP8LHistogramNumCodes(palette_code_bits), 0);
+                              VP8LHistogramNumCodes(palette_code_bits),
+                              a->is_used_[0], b->is_used_[0], 0);
   *cost += VP8LExtraCostCombined(a->literal_ + NUM_LITERAL_CODES,
                                  b->literal_ + NUM_LITERAL_CODES,
                                  NUM_LENGTH_CODES);
@@ -319,19 +385,23 @@ static int GetCombinedHistogramEntropy(const VP8LHistogram* const a,
   }
 
   *cost +=
-      GetCombinedEntropy(a->red_, b->red_, NUM_LITERAL_CODES, trivial_at_end);
+      GetCombinedEntropy(a->red_, b->red_, NUM_LITERAL_CODES, a->is_used_[1],
+                         b->is_used_[1], trivial_at_end);
   if (*cost > cost_threshold) return 0;
 
   *cost +=
-      GetCombinedEntropy(a->blue_, b->blue_, NUM_LITERAL_CODES, trivial_at_end);
-  if (*cost > cost_threshold) return 0;
-
-  *cost += GetCombinedEntropy(a->alpha_, b->alpha_, NUM_LITERAL_CODES,
-                              trivial_at_end);
+      GetCombinedEntropy(a->blue_, b->blue_, NUM_LITERAL_CODES, a->is_used_[2],
+                         b->is_used_[2], trivial_at_end);
   if (*cost > cost_threshold) return 0;
 
   *cost +=
-      GetCombinedEntropy(a->distance_, b->distance_, NUM_DISTANCE_CODES, 0);
+      GetCombinedEntropy(a->alpha_, b->alpha_, NUM_LITERAL_CODES,
+                         a->is_used_[3], b->is_used_[3], trivial_at_end);
+  if (*cost > cost_threshold) return 0;
+
+  *cost +=
+      GetCombinedEntropy(a->distance_, b->distance_, NUM_DISTANCE_CODES,
+                         a->is_used_[4], b->is_used_[4], 0);
   *cost +=
       VP8LExtraCostCombined(a->distance_, b->distance_, NUM_DISTANCE_CODES);
   if (*cost > cost_threshold) return 0;
@@ -419,16 +489,19 @@ static void UpdateDominantCostRange(
 static void UpdateHistogramCost(VP8LHistogram* const h) {
   uint32_t alpha_sym, red_sym, blue_sym;
   const double alpha_cost =
-      PopulationCost(h->alpha_, NUM_LITERAL_CODES, &alpha_sym);
+      PopulationCost(h->alpha_, NUM_LITERAL_CODES, &alpha_sym,
+                     &h->is_used_[3]);
   const double distance_cost =
-      PopulationCost(h->distance_, NUM_DISTANCE_CODES, NULL) +
+      PopulationCost(h->distance_, NUM_DISTANCE_CODES, NULL, &h->is_used_[4]) +
       VP8LExtraCost(h->distance_, NUM_DISTANCE_CODES);
   const int num_codes = VP8LHistogramNumCodes(h->palette_code_bits_);
-  h->literal_cost_ = PopulationCost(h->literal_, num_codes, NULL) +
-                     VP8LExtraCost(h->literal_ + NUM_LITERAL_CODES,
-                                   NUM_LENGTH_CODES);
-  h->red_cost_ = PopulationCost(h->red_, NUM_LITERAL_CODES, &red_sym);
-  h->blue_cost_ = PopulationCost(h->blue_, NUM_LITERAL_CODES, &blue_sym);
+  h->literal_cost_ =
+      PopulationCost(h->literal_, num_codes, NULL, &h->is_used_[0]) +
+          VP8LExtraCost(h->literal_ + NUM_LITERAL_CODES, NUM_LENGTH_CODES);
+  h->red_cost_ =
+      PopulationCost(h->red_, NUM_LITERAL_CODES, &red_sym, &h->is_used_[1]);
+  h->blue_cost_ =
+      PopulationCost(h->blue_, NUM_LITERAL_CODES, &blue_sym, &h->is_used_[2]);
   h->bit_cost_ = h->literal_cost_ + h->red_cost_ + h->blue_cost_ +
                  alpha_cost + distance_cost;
   if ((alpha_sym | red_sym | blue_sym) == VP8L_NON_TRIVIAL_SYM) {
@@ -473,6 +546,7 @@ static void HistogramBuild(
   VP8LHistogram** const histograms = image_histo->histograms;
   VP8LRefsCursor c = VP8LRefsCursorInit(backward_refs);
   assert(histo_bits > 0);
+  VP8LHistogramSetClear(image_histo);
   while (VP8LRefsCursorOk(&c)) {
     const PixOrCopy* const v = c.cur_pos;
     const int ix = (y >> histo_bits) * histo_xsize + (x >> histo_bits);
@@ -493,11 +567,19 @@ static void HistogramCopyAndAnalyze(
   const int histo_size = orig_histo->size;
   VP8LHistogram** const orig_histograms = orig_histo->histograms;
   VP8LHistogram** const histograms = image_histo->histograms;
+  image_histo->size = 0;
   for (i = 0; i < histo_size; ++i) {
     VP8LHistogram* const histo = orig_histograms[i];
     UpdateHistogramCost(histo);
+
+    // Skip the histogram if it is completely empty, which can happen for tiles
+    // with no information (when they are skipped because of LZ77).
+    if (!histo->is_used_[0] && !histo->is_used_[1] && !histo->is_used_[2]
+        && !histo->is_used_[3] && !histo->is_used_[4]) {
+      continue;
+    }
     // Copy histograms from orig_histo[] to image_histo[].
-    HistogramCopy(histo, histograms[i]);
+    HistogramCopy(histo, histograms[image_histo->size++]);
   }
 }
 
@@ -674,6 +756,18 @@ static void HistoQueueUpdateHead(HistoQueue* const histo_queue,
   }
 }
 
+// Update the cost diff and combo of a pair of histograms. This needs to be
+// called when the the histograms have been merged with a third one.
+static void HistoQueueUpdatePair(const VP8LHistogram* const h1,
+                                 const VP8LHistogram* const h2,
+                                 double threshold,
+                                 HistogramPair* const pair) {
+  const double sum_cost = h1->bit_cost_ + h2->bit_cost_;
+  pair->cost_combo = 0.;
+  GetCombinedHistogramEntropy(h1, h2, sum_cost + threshold, &pair->cost_combo);
+  pair->cost_diff = pair->cost_combo - sum_cost;
+}
+
 // Create a pair from indices "idx1" and "idx2" provided its cost
 // is inferior to "threshold", a negative entropy.
 // It returns the cost of the pair, or 0. if it superior to threshold.
@@ -683,7 +777,6 @@ static double HistoQueuePush(HistoQueue* const histo_queue,
   const VP8LHistogram* h1;
   const VP8LHistogram* h2;
   HistogramPair pair;
-  double sum_cost;
 
   assert(threshold <= 0.);
   if (idx1 > idx2) {
@@ -695,10 +788,8 @@ static double HistoQueuePush(HistoQueue* const histo_queue,
   pair.idx2 = idx2;
   h1 = histograms[idx1];
   h2 = histograms[idx2];
-  sum_cost = h1->bit_cost_ + h2->bit_cost_;
-  pair.cost_combo = 0.;
-  GetCombinedHistogramEntropy(h1, h2, sum_cost + threshold, &pair.cost_combo);
-  pair.cost_diff = pair.cost_combo - sum_cost;
+
+  HistoQueueUpdatePair(h1, h2, threshold, &pair);
 
   // Do not even consider the pair if it does not improve the entropy.
   if (pair.cost_diff >= threshold) return 0.;
@@ -891,8 +982,7 @@ static int HistogramCombineStochastic(VP8LHistogramSet* const image_histo,
       }
       if (do_eval) {
         // Re-evaluate the cost of an updated pair.
-        GetCombinedHistogramEntropy(histograms[p->idx1], histograms[p->idx2], 0,
-                                    &p->cost_diff);
+        HistoQueueUpdatePair(histograms[p->idx1], histograms[p->idx2], 0., p);
         if (p->cost_diff >= 0.) {
           HistoQueuePopPair(&histo_queue, p);
           continue;
@@ -987,8 +1077,7 @@ int VP8LGetHistoImageSymbols(int xsize, int ysize,
   // histograms of small sizes (as bin_map will be very sparse) and
   // maximum quality q==100 (to preserve the compression gains at that level).
   const int entropy_combine_num_bins = low_effort ? NUM_PARTITIONS : BIN_SIZE;
-  const int entropy_combine =
-      (orig_histo->size > entropy_combine_num_bins * 2) && (quality < 100);
+  int entropy_combine;
 
   if (orig_histo == NULL) goto Error;
 
@@ -996,15 +1085,16 @@ int VP8LGetHistoImageSymbols(int xsize, int ysize,
   HistogramBuild(xsize, histo_bits, refs, orig_histo);
   // Copies the histograms and computes its bit_cost.
   HistogramCopyAndAnalyze(orig_histo, image_histo);
-
+  entropy_combine =
+      (image_histo->size > entropy_combine_num_bins * 2) && (quality < 100);
   if (entropy_combine) {
-    const int bin_map_size = orig_histo->size;
+    const int bin_map_size = image_histo->size;
     // Reuse histogram_symbols storage. By definition, it's guaranteed to be ok.
     uint16_t* const bin_map = histogram_symbols;
     const double combine_cost_factor =
         GetCombineCostFactor(image_histo_raw_size, quality);
 
-    HistogramAnalyzeEntropyBin(orig_histo, bin_map, low_effort);
+    HistogramAnalyzeEntropyBin(image_histo, bin_map, low_effort);
     // Collapse histograms with similar entropy.
     HistogramCombineEntropyBin(image_histo, tmp_histo, bin_map, bin_map_size,
                                entropy_combine_num_bins, combine_cost_factor,

src/enc/histogram_enc.h

@@ -44,6 +44,7 @@ typedef struct {
   double literal_cost_;      // Cached values of dominant entropy costs:
   double red_cost_;          // literal, red & blue.
   double blue_cost_;
+  uint8_t is_used_[5];       // 5 for literal, red, blue, alpha, distance
 } VP8LHistogram;
 
 // Collection of histograms with fixed capacity, allocated as one
@@ -67,7 +68,9 @@ void VP8LHistogramCreate(VP8LHistogram* const p,
 int VP8LGetHistogramSize(int palette_code_bits);
 
 // Set the palette_code_bits and reset the stats.
-void VP8LHistogramInit(VP8LHistogram* const p, int palette_code_bits);
+// If init_arrays is true, the arrays are also filled with 0's.
+void VP8LHistogramInit(VP8LHistogram* const p, int palette_code_bits,
+                       int init_arrays);
 
 // Collect all the references into a histogram (without reset)
 void VP8LHistogramStoreRefs(const VP8LBackwardRefs* const refs,
@@ -83,6 +86,9 @@ void VP8LFreeHistogramSet(VP8LHistogramSet* const histo);
 // using 'cache_bits'. Return NULL in case of memory error.
 VP8LHistogramSet* VP8LAllocateHistogramSet(int size, int cache_bits);
 
+// Set the histograms in set to 0.
+void VP8LHistogramSetClear(VP8LHistogramSet* const set);
+
 // Allocate and initialize histogram object with specified 'cache_bits'.
 // Returns NULL in case of memory error.
 // Special case of VP8LAllocateHistogramSet, with size equals 1.
@@ -113,7 +119,7 @@ double VP8LBitsEntropy(const uint32_t* const array, int n);
 
 // Estimate how many bits the combined entropy of literals and distance
 // approximately maps to.
-double VP8LHistogramEstimateBits(const VP8LHistogram* const p);
+double VP8LHistogramEstimateBits(VP8LHistogram* const p);
 
 #ifdef __cplusplus
 }

src/enc/iterator_enc.c

@@ -128,7 +128,7 @@ static void ImportLine(const uint8_t* src, int src_stride,
   for (; i < total_len; ++i) dst[i] = dst[len - 1];
 }
 
-void VP8IteratorImport(VP8EncIterator* const it, uint8_t* tmp_32) {
+void VP8IteratorImport(VP8EncIterator* const it, uint8_t* const tmp_32) {
   const VP8Encoder* const enc = it->enc_;
   const int x = it->x_, y = it->y_;
   const WebPPicture* const pic = enc->pic_;

src/enc/near_lossless_enc.c

@@ -146,6 +146,6 @@ int VP8ApplyNearLossless(const WebPPicture* const picture, int quality,
 
 // Define a stub to suppress compiler warnings.
 extern void VP8LNearLosslessStub(void);
-WEBP_TSAN_IGNORE_FUNCTION void VP8LNearLosslessStub(void) {}
+void VP8LNearLosslessStub(void) {}
 
 #endif  // (WEBP_NEAR_LOSSLESS == 1)

src/enc/picture_csp_enc.c

@@ -126,7 +126,7 @@ static WEBP_INLINE int LinearToGamma(uint32_t base_value, int shift) {
 
 #else
 
-static WEBP_TSAN_IGNORE_FUNCTION void InitGammaTables(void) {}
+static void InitGammaTables(void) {}
 static WEBP_INLINE uint32_t GammaToLinear(uint8_t v) { return v; }
 static WEBP_INLINE int LinearToGamma(uint32_t base_value, int shift) {
   return (int)(base_value << shift);
@@ -238,7 +238,7 @@ static WEBP_INLINE uint32_t LinearToGammaS(uint32_t value) {
 
 #else
 
-static WEBP_TSAN_IGNORE_FUNCTION void InitGammaTablesS(void) {}
+static void InitGammaTablesS(void) {}
 static WEBP_INLINE uint32_t GammaToLinearS(int v) {
   return (v << GAMMA_TO_LINEAR_BITS) / MAX_Y_T;
 }

src/enc/picture_tools_enc.c

@@ -16,10 +16,6 @@
 #include "src/enc/vp8i_enc.h"
 #include "src/dsp/yuv.h"
 
-static WEBP_INLINE uint32_t MakeARGB32(int r, int g, int b) {
-  return (0xff000000u | (r << 16) | (g << 8) | b);
-}
-
 //------------------------------------------------------------------------------
 // Helper: clean up fully transparent area to help compressibility.
 
@@ -195,6 +191,10 @@ void WebPCleanupTransparentAreaLossless(WebPPicture* const pic) {
 #define BLEND_10BIT(V0, V1, ALPHA) \
     ((((V0) * (1020 - (ALPHA)) + (V1) * (ALPHA)) * 0x101 + 1024) >> 18)
 
+static WEBP_INLINE uint32_t MakeARGB32(int r, int g, int b) {
+  return (0xff000000u | (r << 16) | (g << 8) | b);
+}
+
 void WebPBlendAlpha(WebPPicture* pic, uint32_t background_rgb) {
   const int red = (background_rgb >> 16) & 0xff;
   const int green = (background_rgb >> 8) & 0xff;
@@ -208,39 +208,44 @@ void WebPBlendAlpha(WebPPicture* pic, uint32_t background_rgb) {
     const int U0 = VP8RGBToU(4 * red, 4 * green, 4 * blue, 4 * YUV_HALF);
     const int V0 = VP8RGBToV(4 * red, 4 * green, 4 * blue, 4 * YUV_HALF);
     const int has_alpha = pic->colorspace & WEBP_CSP_ALPHA_BIT;
-    if (!has_alpha || pic->a == NULL) return;    // nothing to do
+    uint8_t* y_ptr = pic->y;
+    uint8_t* u_ptr = pic->u;
+    uint8_t* v_ptr = pic->v;
+    uint8_t* a_ptr = pic->a;
+    if (!has_alpha || a_ptr == NULL) return;    // nothing to do
     for (y = 0; y < pic->height; ++y) {
       // Luma blending
-      uint8_t* const y_ptr = pic->y + y * pic->y_stride;
-      uint8_t* const a_ptr = pic->a + y * pic->a_stride;
       for (x = 0; x < pic->width; ++x) {
-        const int alpha = a_ptr[x];
+        const uint8_t alpha = a_ptr[x];
         if (alpha < 0xff) {
-          y_ptr[x] = BLEND(Y0, y_ptr[x], a_ptr[x]);
+          y_ptr[x] = BLEND(Y0, y_ptr[x], alpha);
         }
       }
       // Chroma blending every even line
       if ((y & 1) == 0) {
-        uint8_t* const u = pic->u + (y >> 1) * pic->uv_stride;
-        uint8_t* const v = pic->v + (y >> 1) * pic->uv_stride;
         uint8_t* const a_ptr2 =
             (y + 1 == pic->height) ? a_ptr : a_ptr + pic->a_stride;
         for (x = 0; x < uv_width; ++x) {
           // Average four alpha values into a single blending weight.
           // TODO(skal): might lead to visible contouring. Can we do better?
-          const int alpha =
+          const uint32_t alpha =
               a_ptr[2 * x + 0] + a_ptr[2 * x + 1] +
               a_ptr2[2 * x + 0] + a_ptr2[2 * x + 1];
-          u[x] = BLEND_10BIT(U0, u[x], alpha);
-          v[x] = BLEND_10BIT(V0, v[x], alpha);
+          u_ptr[x] = BLEND_10BIT(U0, u_ptr[x], alpha);
+          v_ptr[x] = BLEND_10BIT(V0, v_ptr[x], alpha);
         }
         if (pic->width & 1) {   // rightmost pixel
-          const int alpha = 2 * (a_ptr[2 * x + 0] + a_ptr2[2 * x + 0]);
-          u[x] = BLEND_10BIT(U0, u[x], alpha);
-          v[x] = BLEND_10BIT(V0, v[x], alpha);
+          const uint32_t alpha = 2 * (a_ptr[2 * x + 0] + a_ptr2[2 * x + 0]);
+          u_ptr[x] = BLEND_10BIT(U0, u_ptr[x], alpha);
+          v_ptr[x] = BLEND_10BIT(V0, v_ptr[x], alpha);
         }
+      } else {
+        u_ptr += pic->uv_stride;
+        v_ptr += pic->uv_stride;
       }
-      memset(a_ptr, 0xff, pic->width);
+      memset(a_ptr, 0xff, pic->width);  // reset alpha value to opaque
+      a_ptr += pic->a_stride;
+      y_ptr += pic->y_stride;
     }
   } else {
     uint32_t* argb = pic->argb;

src/enc/vp8i_enc.h

@@ -32,7 +32,7 @@ extern "C" {
 // version numbers
 #define ENC_MAJ_VERSION 1
 #define ENC_MIN_VERSION 0
-#define ENC_REV_VERSION 0
+#define ENC_REV_VERSION 1
 
 enum { MAX_LF_LEVELS = 64,       // Maximum loop filter level
        MAX_VARIABLE_LEVEL = 67,  // last (inclusive) level with variable cost
@@ -278,7 +278,7 @@ int VP8IteratorIsDone(const VP8EncIterator* const it);
 // Import uncompressed samples from source.
 // If tmp_32 is not NULL, import boundary samples too.
 // tmp_32 is a 32-bytes scratch buffer that must be aligned in memory.
-void VP8IteratorImport(VP8EncIterator* const it, uint8_t* tmp_32);
+void VP8IteratorImport(VP8EncIterator* const it, uint8_t* const tmp_32);
 // export decimated samples
 void VP8IteratorExport(const VP8EncIterator* const it);
 // go to next macroblock. Returns false if not finished.
@@ -515,4 +515,4 @@ void WebPCleanupTransparentAreaLossless(WebPPicture* const pic);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_ENC_VP8I_ENC_H_ */
+#endif  // WEBP_ENC_VP8I_ENC_H_

src/enc/vp8l_enc.c

@@ -809,6 +809,7 @@ static WebPEncodingError EncodeImageNoHuffman(VP8LBitWriter* const bw,
     err = VP8_ENC_ERROR_OUT_OF_MEMORY;
     goto Error;
   }
+  VP8LHistogramSetClear(histogram_image);
 
   // Build histogram image and symbols from backward references.
   VP8LHistogramStoreRefs(refs, histogram_image->histograms[0]);
@@ -1248,14 +1249,20 @@ static WebPEncodingError MakeInputImageCopy(VP8LEncoder* const enc) {
   const WebPPicture* const picture = enc->pic_;
   const int width = picture->width;
   const int height = picture->height;
-  int y;
+
   err = AllocateTransformBuffer(enc, width, height);
   if (err != VP8_ENC_OK) return err;
   if (enc->argb_content_ == kEncoderARGB) return VP8_ENC_OK;
-  for (y = 0; y < height; ++y) {
-    memcpy(enc->argb_ + y * width,
-           picture->argb + y * picture->argb_stride,
-           width * sizeof(*enc->argb_));
+
+  {
+    uint32_t* dst = enc->argb_;
+    const uint32_t* src = picture->argb;
+    int y;
+    for (y = 0; y < height; ++y) {
+      memcpy(dst, src, width * sizeof(*dst));
+      dst += width;
+      src += picture->argb_stride;
+    }
   }
   enc->argb_content_ = kEncoderARGB;
   assert(enc->current_width_ == width);
@@ -1685,11 +1692,16 @@ WebPEncodingError VP8LEncodeStream(const WebPConfig* const config,
   const WebPWorkerInterface* const worker_interface = WebPGetWorkerInterface();
   int ok_main;
 
+  if (enc_main == NULL || !VP8LBitWriterInit(&bw_side, 0)) {
+    WebPEncodingSetError(picture, VP8_ENC_ERROR_OUT_OF_MEMORY);
+    VP8LEncoderDelete(enc_main);
+    return 0;
+  }
+
   // Analyze image (entropy, num_palettes etc)
-  if (enc_main == NULL ||
-      !EncoderAnalyze(enc_main, crunch_configs, &num_crunch_configs_main,
+  if (!EncoderAnalyze(enc_main, crunch_configs, &num_crunch_configs_main,
                       &red_and_blue_always_zero) ||
-      !EncoderInit(enc_main) || !VP8LBitWriterInit(&bw_side, 0)) {
+      !EncoderInit(enc_main)) {
     err = VP8_ENC_ERROR_OUT_OF_MEMORY;
     goto Error;
   }

src/enc/vp8li_enc.h

@@ -115,4 +115,4 @@ void VP8LColorSpaceTransform(int width, int height, int bits, int quality,
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_ENC_VP8LI_ENC_H_ */
+#endif  // WEBP_ENC_VP8LI_ENC_H_

src/libwebp.rc

@@ -6,8 +6,8 @@
 LANGUAGE LANG_ENGLISH, SUBLANG_ENGLISH_US
 
 VS_VERSION_INFO VERSIONINFO
- FILEVERSION 1,0,0,0
- PRODUCTVERSION 1,0,0,0
+ FILEVERSION 1,0,0,1
+ PRODUCTVERSION 1,0,0,1
  FILEFLAGSMASK 0x3fL
 #ifdef _DEBUG
  FILEFLAGS 0x1L
@@ -24,12 +24,12 @@ BEGIN
         BEGIN
             VALUE "CompanyName", "Google, Inc."
             VALUE "FileDescription", "libwebp DLL"
-            VALUE "FileVersion", "1.0.0"
+            VALUE "FileVersion", "1.0.1"
             VALUE "InternalName", "libwebp.dll"
             VALUE "LegalCopyright", "Copyright (C) 2018"
             VALUE "OriginalFilename", "libwebp.dll"
             VALUE "ProductName", "WebP Image Codec"
-            VALUE "ProductVersion", "1.0.0"
+            VALUE "ProductVersion", "1.0.1"
         END
     END
     BLOCK "VarFileInfo"

src/libwebpdecoder.rc

@@ -6,8 +6,8 @@
 LANGUAGE LANG_ENGLISH, SUBLANG_ENGLISH_US
 
 VS_VERSION_INFO VERSIONINFO
- FILEVERSION 1,0,0,0
- PRODUCTVERSION 1,0,0,0
+ FILEVERSION 1,0,0,1
+ PRODUCTVERSION 1,0,0,1
  FILEFLAGSMASK 0x3fL
 #ifdef _DEBUG
  FILEFLAGS 0x1L
@@ -24,12 +24,12 @@ BEGIN
         BEGIN
             VALUE "CompanyName", "Google, Inc."
             VALUE "FileDescription", "libwebpdecoder DLL"
-            VALUE "FileVersion", "1.0.0"
+            VALUE "FileVersion", "1.0.1"
             VALUE "InternalName", "libwebpdecoder.dll"
             VALUE "LegalCopyright", "Copyright (C) 2018"
             VALUE "OriginalFilename", "libwebpdecoder.dll"
             VALUE "ProductName", "WebP Image Decoder"
-            VALUE "ProductVersion", "1.0.0"
+            VALUE "ProductVersion", "1.0.1"
         END
     END
     BLOCK "VarFileInfo"

src/mux/Makefile.am

@@ -17,6 +17,6 @@ noinst_HEADERS =
 noinst_HEADERS += ../webp/format_constants.h
 
 libwebpmux_la_LIBADD = ../libwebp.la
-libwebpmux_la_LDFLAGS = -no-undefined -version-info 3:2:0 -lm
+libwebpmux_la_LDFLAGS = -no-undefined -version-info 3:3:0 -lm
 libwebpmuxincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebpmux.pc

src/mux/animi.h

@@ -40,4 +40,4 @@ int WebPAnimEncoderRefineRect(
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_MUX_ANIMI_H_ */
+#endif  // WEBP_MUX_ANIMI_H_

src/mux/libwebpmux.rc

@@ -6,8 +6,8 @@
 LANGUAGE LANG_ENGLISH, SUBLANG_ENGLISH_US
 
 VS_VERSION_INFO VERSIONINFO
- FILEVERSION 1,0,0,0
- PRODUCTVERSION 1,0,0,0
+ FILEVERSION 1,0,0,1
+ PRODUCTVERSION 1,0,0,1
  FILEFLAGSMASK 0x3fL
 #ifdef _DEBUG
  FILEFLAGS 0x1L
@@ -24,12 +24,12 @@ BEGIN
         BEGIN
             VALUE "CompanyName", "Google, Inc."
             VALUE "FileDescription", "libwebpmux DLL"
-            VALUE "FileVersion", "1.0.0"
+            VALUE "FileVersion", "1.0.1"
             VALUE "InternalName", "libwebpmux.dll"
             VALUE "LegalCopyright", "Copyright (C) 2018"
             VALUE "OriginalFilename", "libwebpmux.dll"
             VALUE "ProductName", "WebP Image Muxer"
-            VALUE "ProductVersion", "1.0.0"
+            VALUE "ProductVersion", "1.0.1"
         END
     END
     BLOCK "VarFileInfo"

src/mux/muxedit.c

@@ -69,12 +69,12 @@ void WebPMuxDelete(WebPMux* mux) {
   if (idx == (INDEX)) {                                                        \
     err = ChunkAssignData(&chunk, data, copy_data, tag);                       \
     if (err == WEBP_MUX_OK) {                                                  \
-      err = ChunkSetNth(&chunk, (LIST), nth);                                  \
+      err = ChunkSetHead(&chunk, (LIST));                                      \
     }                                                                          \
     return err;                                                                \
   }
 
-static WebPMuxError MuxSet(WebPMux* const mux, uint32_t tag, uint32_t nth,
+static WebPMuxError MuxSet(WebPMux* const mux, uint32_t tag,
                            const WebPData* const data, int copy_data) {
   WebPChunk chunk;
   WebPMuxError err = WEBP_MUX_NOT_FOUND;
@@ -190,7 +190,7 @@ WebPMuxError WebPMuxSetChunk(WebPMux* mux, const char fourcc[4],
   if (err != WEBP_MUX_OK && err != WEBP_MUX_NOT_FOUND) return err;
 
   // Add the given chunk.
-  return MuxSet(mux, tag, 1, chunk_data, copy_data);
+  return MuxSet(mux, tag, chunk_data, copy_data);
 }
 
 // Creates a chunk from given 'data' and sets it as 1st chunk in 'chunk_list'.
@@ -202,7 +202,7 @@ static WebPMuxError AddDataToChunkList(
   ChunkInit(&chunk);
   err = ChunkAssignData(&chunk, data, copy_data, tag);
   if (err != WEBP_MUX_OK) goto Err;
-  err = ChunkSetNth(&chunk, chunk_list, 1);
+  err = ChunkSetHead(&chunk, chunk_list);
   if (err != WEBP_MUX_OK) goto Err;
   return WEBP_MUX_OK;
  Err:
@@ -266,14 +266,14 @@ WebPMuxError WebPMuxPushFrame(WebPMux* mux, const WebPMuxFrameInfo* info,
                               int copy_data) {
   WebPMuxImage wpi;
   WebPMuxError err;
-  const WebPData* const bitstream = &info->bitstream;
 
   // Sanity checks.
   if (mux == NULL || info == NULL) return WEBP_MUX_INVALID_ARGUMENT;
 
   if (info->id != WEBP_CHUNK_ANMF) return WEBP_MUX_INVALID_ARGUMENT;
 
-  if (bitstream->bytes == NULL || bitstream->size > MAX_CHUNK_PAYLOAD) {
+  if (info->bitstream.bytes == NULL ||
+      info->bitstream.size > MAX_CHUNK_PAYLOAD) {
     return WEBP_MUX_INVALID_ARGUMENT;
   }
 
@@ -287,7 +287,7 @@ WebPMuxError WebPMuxPushFrame(WebPMux* mux, const WebPMuxFrameInfo* info,
   }
 
   MuxImageInit(&wpi);
-  err = SetAlphaAndImageChunks(bitstream, copy_data, &wpi);
+  err = SetAlphaAndImageChunks(&info->bitstream, copy_data, &wpi);
   if (err != WEBP_MUX_OK) goto Err;
   assert(wpi.img_ != NULL);  // As SetAlphaAndImageChunks() was successful.
 
@@ -342,7 +342,7 @@ WebPMuxError WebPMuxSetAnimationParams(WebPMux* mux,
   // Set the animation parameters.
   PutLE32(data, params->bgcolor);
   PutLE16(data + 4, params->loop_count);
-  return MuxSet(mux, kChunks[IDX_ANIM].tag, 1, &anim, 1);
+  return MuxSet(mux, kChunks[IDX_ANIM].tag, &anim, 1);
 }
 
 WebPMuxError WebPMuxSetCanvasSize(WebPMux* mux,
@@ -540,7 +540,7 @@ static WebPMuxError CreateVP8XChunk(WebPMux* const mux) {
   PutLE24(data + 4, width - 1);   // canvas width.
   PutLE24(data + 7, height - 1);  // canvas height.
 
-  return MuxSet(mux, kChunks[IDX_VP8X].tag, 1, &vp8x, 1);
+  return MuxSet(mux, kChunks[IDX_VP8X].tag, &vp8x, 1);
 }
 
 // Cleans up 'mux' by removing any unnecessary chunks.

src/mux/muxi.h

@@ -14,6 +14,7 @@
 #ifndef WEBP_MUX_MUXI_H_
 #define WEBP_MUX_MUXI_H_
 
+#include <assert.h>
 #include <stdlib.h>
 #include "src/dec/vp8i_dec.h"
 #include "src/dec/vp8li_dec.h"
@@ -28,7 +29,7 @@ extern "C" {
 
 #define MUX_MAJ_VERSION 1
 #define MUX_MIN_VERSION 0
-#define MUX_REV_VERSION 0
+#define MUX_REV_VERSION 1
 
 // Chunk object.
 typedef struct WebPChunk WebPChunk;
@@ -126,11 +127,14 @@ WebPChunk* ChunkSearchList(WebPChunk* first, uint32_t nth, uint32_t tag);
 WebPMuxError ChunkAssignData(WebPChunk* chunk, const WebPData* const data,
                              int copy_data, uint32_t tag);
 
-// Sets 'chunk' at nth position in the 'chunk_list'.
-// nth = 0 has the special meaning "last of the list".
+// Sets 'chunk' as the only element in 'chunk_list' if it is empty.
 // On success ownership is transferred from 'chunk' to the 'chunk_list'.
-WebPMuxError ChunkSetNth(WebPChunk* chunk, WebPChunk** chunk_list,
-                         uint32_t nth);
+WebPMuxError ChunkSetHead(WebPChunk* const chunk, WebPChunk** const chunk_list);
+// Sets 'chunk' at last position in the 'chunk_list'.
+// On success ownership is transferred from 'chunk' to the 'chunk_list'.
+// *chunk_list also points towards the last valid element of the initial
+// *chunk_list.
+WebPMuxError ChunkAppend(WebPChunk* const chunk, WebPChunk*** const chunk_list);
 
 // Releases chunk and returns chunk->next_.
 WebPChunk* ChunkRelease(WebPChunk* const chunk);
@@ -143,13 +147,13 @@ void ChunkListDelete(WebPChunk** const chunk_list);
 
 // Returns size of the chunk including chunk header and padding byte (if any).
 static WEBP_INLINE size_t SizeWithPadding(size_t chunk_size) {
+  assert(chunk_size <= MAX_CHUNK_PAYLOAD);
   return CHUNK_HEADER_SIZE + ((chunk_size + 1) & ~1U);
 }
 
 // Size of a chunk including header and padding.
 static WEBP_INLINE size_t ChunkDiskSize(const WebPChunk* chunk) {
   const size_t data_size = chunk->data_.size;
-  assert(data_size < MAX_CHUNK_PAYLOAD);
   return SizeWithPadding(data_size);
 }
 
@@ -227,4 +231,4 @@ WebPMuxError MuxValidate(const WebPMux* const mux);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_MUX_MUXI_H_ */
+#endif  // WEBP_MUX_MUXI_H_

src/mux/muxinternal.c

@@ -111,27 +111,6 @@ WebPChunk* ChunkSearchList(WebPChunk* first, uint32_t nth, uint32_t tag) {
   return ((nth > 0) && (iter > 0)) ? NULL : first;
 }
 
-// Outputs a pointer to 'prev_chunk->next_',
-//   where 'prev_chunk' is the pointer to the chunk at position (nth - 1).
-// Returns true if nth chunk was found.
-static int ChunkSearchListToSet(WebPChunk** chunk_list, uint32_t nth,
-                                WebPChunk*** const location) {
-  uint32_t count = 0;
-  assert(chunk_list != NULL);
-  *location = chunk_list;
-
-  while (*chunk_list != NULL) {
-    WebPChunk* const cur_chunk = *chunk_list;
-    ++count;
-    if (count == nth) return 1;  // Found.
-    chunk_list = &cur_chunk->next_;
-    *location = chunk_list;
-  }
-
-  // *chunk_list is ok to be NULL if adding at last location.
-  return (nth == 0 || (count == nth - 1)) ? 1 : 0;
-}
-
 //------------------------------------------------------------------------------
 // Chunk writer methods.
 
@@ -156,11 +135,12 @@ WebPMuxError ChunkAssignData(WebPChunk* chunk, const WebPData* const data,
   return WEBP_MUX_OK;
 }
 
-WebPMuxError ChunkSetNth(WebPChunk* chunk, WebPChunk** chunk_list,
-                         uint32_t nth) {
+WebPMuxError ChunkSetHead(WebPChunk* const chunk,
+                          WebPChunk** const chunk_list) {
   WebPChunk* new_chunk;
 
-  if (!ChunkSearchListToSet(chunk_list, nth, &chunk_list)) {
+  assert(chunk_list != NULL);
+  if (*chunk_list != NULL) {
     return WEBP_MUX_NOT_FOUND;
   }
 
@@ -168,11 +148,26 @@ WebPMuxError ChunkSetNth(WebPChunk* chunk, WebPChunk** chunk_list,
   if (new_chunk == NULL) return WEBP_MUX_MEMORY_ERROR;
   *new_chunk = *chunk;
   chunk->owner_ = 0;
-  new_chunk->next_ = *chunk_list;
+  new_chunk->next_ = NULL;
   *chunk_list = new_chunk;
   return WEBP_MUX_OK;
 }
 
+WebPMuxError ChunkAppend(WebPChunk* const chunk,
+                         WebPChunk*** const chunk_list) {
+  assert(chunk_list != NULL && *chunk_list != NULL);
+
+  if (**chunk_list == NULL) {
+    ChunkSetHead(chunk, *chunk_list);
+  } else {
+    WebPChunk* last_chunk = **chunk_list;
+    while (last_chunk->next_ != NULL) last_chunk = last_chunk->next_;
+    ChunkSetHead(chunk, &last_chunk->next_);
+    *chunk_list = &last_chunk->next_;
+  }
+  return WEBP_MUX_OK;
+}
+
 //------------------------------------------------------------------------------
 // Chunk deletion method(s).
 
@@ -232,9 +227,11 @@ void MuxImageInit(WebPMuxImage* const wpi) {
 WebPMuxImage* MuxImageRelease(WebPMuxImage* const wpi) {
   WebPMuxImage* next;
   if (wpi == NULL) return NULL;
-  ChunkDelete(wpi->header_);
-  ChunkDelete(wpi->alpha_);
-  ChunkDelete(wpi->img_);
+  // There should be at most one chunk of header_, alpha_, img_ but we call
+  // ChunkListDelete to be safe
+  ChunkListDelete(&wpi->header_);
+  ChunkListDelete(&wpi->alpha_);
+  ChunkListDelete(&wpi->img_);
   ChunkListDelete(&wpi->unknown_);
 
   next = wpi->next_;