release: OpenCV 3.4.13

Merge pull request #19180 from alalek:fixup_19161
js(build): fix generator with Python3
2026-07-22 03:43:03 +04:00 · 2020-12-20 22:15:49 +00:00 · 2020-12-20 18:39:01 +00:00 · 2020-12-20 17:42:47 +00:00 · 2020-12-20 16:40:58 +00:00 · 2020-12-20 16:40:28 +00:00
1245 changed files with 66551 additions and 20983 deletions
@@ -22,3 +22,5 @@ bin/
 *.log
 *.tlog
 build
+node_modules
+CMakeSettings.json
@@ -20,11 +20,14 @@ if(CMAKE_COMPILER_IS_GNUCC)
    # - matchTemplate about 5-10%
    # - goodFeaturesToTrack 10-20%
    # - cornerHarris 30% for some cases
-
-    set_source_files_properties(${carotene_sources} COMPILE_FLAGS "--param ipcp-unit-growth=100000 --param inline-unit-growth=100000 --param large-stack-frame-growth=5000")
+    if(CMAKE_CXX_COMPILER_VERSION VERSION_LESS "10.0.0")
+        set_source_files_properties(${carotene_sources} COMPILE_FLAGS "--param ipcp-unit-growth=100000 --param inline-unit-growth=100000 --param large-stack-frame-growth=5000")
+    else()
+        set_source_files_properties(${carotene_sources} COMPILE_FLAGS "--param ipa-cp-unit-growth=100000 --param inline-unit-growth=100000 --param large-stack-frame-growth=5000")
+    endif()
 endif()

-add_library(carotene_objs OBJECT
+add_library(carotene_objs OBJECT EXCLUDE_FROM_ALL
  ${carotene_headers}
  ${carotene_sources}
 )
@@ -37,4 +40,5 @@ if(WITH_NEON)
    target_compile_definitions(carotene_objs PRIVATE "-DWITH_NEON")
 endif()

-add_library(carotene STATIC EXCLUDE_FROM_ALL "$<TARGET_OBJECTS:carotene_objs>")
+# we add dummy file to fix XCode build
+add_library(carotene STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} "$<TARGET_OBJECTS:carotene_objs>" "${CAROTENE_SOURCE_DIR}/dummy.cpp")
@@ -58,22 +58,6 @@ function(compile_carotene)
  endif()

  add_subdirectory("${CAROTENE_DIR}" "${CMAKE_CURRENT_BINARY_DIR}/carotene")
-
-  if(ARM OR AARCH64)
-    if(CMAKE_BUILD_TYPE)
-      set(CMAKE_TRY_COMPILE_CONFIGURATION ${CMAKE_BUILD_TYPE})
-    endif()
-    check_cxx_compiler_flag("-mfpu=neon" CXX_HAS_MFPU_NEON)
-    check_c_compiler_flag("-mfpu=neon" C_HAS_MFPU_NEON)
-    if(${CXX_HAS_MFPU_NEON} AND ${C_HAS_MFPU_NEON} AND NOT "${CMAKE_CXX_FLAGS} " MATCHES "-mfpu=neon[^ ]*")
-      get_target_property(old_flags "carotene_objs" COMPILE_FLAGS)
-      if(old_flags)
-        set_target_properties("carotene_objs" PROPERTIES COMPILE_FLAGS "${old_flags} -mfpu=neon")
-      else()
-        set_target_properties("carotene_objs" PROPERTIES COMPILE_FLAGS "-mfpu=neon")
-      endif()
-    endif()
-  endif()
 endfunction()

 compile_carotene()
@@ -88,11 +72,16 @@ set_property(DIRECTORY APPEND PROPERTY COMPILE_DEFINITIONS ${carotene_defs})
    #   matchTemplate about 5-10%
    #   goodFeaturesToTrack 10-20%
    #   cornerHarris 30% for some cases
-    set_source_files_properties(impl.cpp $<TARGET_OBJECTS:carotene_objs> COMPILE_FLAGS "--param ipcp-unit-growth=100000 --param inline-unit-growth=100000 --param large-stack-frame-growth=5000")
+    if(CMAKE_CXX_COMPILER_VERSION VERSION_LESS "10.0.0")
+      set_source_files_properties(impl.cpp $<TARGET_OBJECTS:carotene_objs> COMPILE_FLAGS "--param ipcp-unit-growth=100000 --param inline-unit-growth=100000 --param large-stack-frame-growth=5000")
+    else()
+      set_source_files_properties(impl.cpp $<TARGET_OBJECTS:carotene_objs> COMPILE_FLAGS "--param ipa-cp-unit-growth=100000 --param inline-unit-growth=100000 --param large-stack-frame-growth=5000")
+    endif()
 #    set_source_files_properties(impl.cpp $<TARGET_OBJECTS:carotene_objs> COMPILE_FLAGS "--param ipcp-unit-growth=100000 --param inline-unit-growth=100000 --param large-stack-frame-growth=5000")
  endif()

-add_library(tegra_hal STATIC $<TARGET_OBJECTS:carotene_objs>)
+# we add dummy file to fix XCode build
+add_library(tegra_hal STATIC $<TARGET_OBJECTS:carotene_objs> "dummy.cpp")
 set_target_properties(tegra_hal PROPERTIES ARCHIVE_OUTPUT_DIRECTORY ${3P_LIBRARY_OUTPUT_PATH})
 set(OPENCV_SRC_DIR "${CMAKE_SOURCE_DIR}")
 if(NOT BUILD_SHARED_LIBS)
@@ -0,0 +1,2 @@
+// This file is needed for compilation on some platforms e.g. with XCode generator
+// Related issue: https://gitlab.kitware.com/cmake/cmake/-/issues/17457
@@ -0,0 +1,2 @@
+// This file is needed for compilation on some platforms e.g. with XCode generator
+// Related issue: https://gitlab.kitware.com/cmake/cmake/-/issues/17457
@@ -14,7 +14,7 @@ if(NOT DEFINED CPUFEATURES_SOURCES)
 endif()

 include_directories(${CPUFEATURES_INCLUDE_DIRS})
-add_library(${OPENCV_CPUFEATURES_TARGET_NAME} STATIC ${CPUFEATURES_SOURCES})
+add_library(${OPENCV_CPUFEATURES_TARGET_NAME} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${CPUFEATURES_SOURCES})

 set_target_properties(${OPENCV_CPUFEATURES_TARGET_NAME}
  PROPERTIES OUTPUT_NAME cpufeatures
@@ -29,7 +29,7 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${OPENCV_CPUFEATURES_TARGET_NAME} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${OPENCV_CPUFEATURES_TARGET_NAME} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

 ocv_install_3rdparty_licenses(cpufeatures LICENSE README.md)
@@ -1,8 +1,8 @@
-# Binaries branch name: ffmpeg/3.4_20190905
-# Binaries were created for OpenCV: fafada28ebc0f2e5423a7d8ece425574ef01ff60
-ocv_update(FFMPEG_BINARIES_COMMIT "bf1730f4c4ba1996bed1fe268b52e4e942151cd6")
-ocv_update(FFMPEG_FILE_HASH_BIN32 "6899624f88cb4cbb1604edc5e12b18ab")
-ocv_update(FFMPEG_FILE_HASH_BIN64 "4729c052cb0c53ab56e9a0eed91559df")
+# Binaries branch name: ffmpeg/3.4_20200907
+# Binaries were created for OpenCV: 03bee14372f5537daa56c62e771ec16181ca1f98
+ocv_update(FFMPEG_BINARIES_COMMIT "2a96257b743695a47f8012aab1ffb995a1dee8b4")
+ocv_update(FFMPEG_FILE_HASH_BIN32 "5e68a3ff82f43ac6524e50e448a34c9c")
+ocv_update(FFMPEG_FILE_HASH_BIN64 "205db629d893e7d4865fd1459807ff47")
 ocv_update(FFMPEG_FILE_HASH_CMAKE "3b90f67f4b429e77d3da36698cef700c")

 function(download_win_ffmpeg script_var)
@@ -17,7 +17,7 @@ file(GLOB lib_hdrs ${IPP_IW_PATH}/include/*.h ${IPP_IW_PATH}/include/iw/*.h ${IP
 #         Define the library target:
 # ----------------------------------------------------------------------------------

-add_library(${IPP_IW_LIBRARY} STATIC ${lib_srcs} ${lib_hdrs})
+add_library(${IPP_IW_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${lib_srcs} ${lib_hdrs})

 if(UNIX)
  if(CV_GCC OR CV_CLANG OR CV_ICC)
@@ -41,5 +41,5 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${IPP_IW_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${IPP_IW_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()
@@ -2,37 +2,32 @@ function(download_ippicv root_var)
  set(${root_var} "" PARENT_SCOPE)

  # Commit SHA in the opencv_3rdparty repo
-  set(IPPICV_COMMIT "32e315a5b106a7b89dbed51c28f8120a48b368b4")
+  set(IPPICV_COMMIT "a56b6ac6f030c312b2dce17430eef13aed9af274")
  # Define actual ICV versions
  if(APPLE)
    set(OPENCV_ICV_PLATFORM "macosx")
    set(OPENCV_ICV_PACKAGE_SUBDIR "ippicv_mac")
-    if(X86_64)
-      set(OPENCV_ICV_NAME "ippicv_2019_mac_intel64_general_20180723.tgz")
-      set(OPENCV_ICV_HASH "fe6b2bb75ae0e3f19ad3ae1a31dfa4a2")
-    else()
-      set(OPENCV_ICV_NAME "ippicv_2019_mac_ia32_general_20180723.tgz")
-      set(OPENCV_ICV_HASH "b5dfa78c87eb75c64470cbe5ec876f4f")
-    endif()
+    set(OPENCV_ICV_NAME "ippicv_2020_mac_intel64_20191018_general.tgz")
+    set(OPENCV_ICV_HASH "1c3d675c2a2395d094d523024896e01b")
  elseif((UNIX AND NOT ANDROID) OR (UNIX AND ANDROID_ABI MATCHES "x86"))
    set(OPENCV_ICV_PLATFORM "linux")
    set(OPENCV_ICV_PACKAGE_SUBDIR "ippicv_lnx")
    if(X86_64)
-      set(OPENCV_ICV_NAME "ippicv_2019_lnx_intel64_general_20180723.tgz")
-      set(OPENCV_ICV_HASH "c0bd78adb4156bbf552c1dfe90599607")
+      set(OPENCV_ICV_NAME "ippicv_2020_lnx_intel64_20191018_general.tgz")
+      set(OPENCV_ICV_HASH "7421de0095c7a39162ae13a6098782f9")
    else()
-      set(OPENCV_ICV_NAME "ippicv_2019_lnx_ia32_general_20180723.tgz")
-      set(OPENCV_ICV_HASH "4f38432c30bfd6423164b7a24bbc98a0")
+      set(OPENCV_ICV_NAME "ippicv_2020_lnx_ia32_20191018_general.tgz")
+      set(OPENCV_ICV_HASH "ad189a940fb60eb71f291321322fe3e8")
    endif()
  elseif(WIN32 AND NOT ARM)
    set(OPENCV_ICV_PLATFORM "windows")
    set(OPENCV_ICV_PACKAGE_SUBDIR "ippicv_win")
    if(X86_64)
-      set(OPENCV_ICV_NAME "ippicv_2019_win_intel64_20180723_general.zip")
-      set(OPENCV_ICV_HASH "1d222685246896fe089f88b8858e4b2f")
+      set(OPENCV_ICV_NAME "ippicv_2020_win_intel64_20191018_general.zip")
+      set(OPENCV_ICV_HASH "879741a7946b814455eee6c6ffde2984")
    else()
-      set(OPENCV_ICV_NAME "ippicv_2019_win_ia32_20180723_general.zip")
-      set(OPENCV_ICV_HASH "0157251a2eb9cd63a3ebc7eed0f3e59e")
+      set(OPENCV_ICV_NAME "ippicv_2020_win_ia32_20191018_general.zip")
+      set(OPENCV_ICV_HASH "cd39bdf0c2e1cac9a61101dad7a2413e")
    endif()
  else()
    return()
@@ -37,7 +37,7 @@ set(ITT_SRCS
    src/ittnotify/jitprofiling.c
 )

-add_library(${ITT_LIBRARY} STATIC ${ITT_SRCS} ${ITT_PUBLIC_HDRS} ${ITT_PRIVATE_HDRS})
+add_library(${ITT_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${ITT_SRCS} ${ITT_PUBLIC_HDRS} ${ITT_PRIVATE_HDRS})

 if(NOT WIN32)
  if(HAVE_DL_LIBRARY)
@@ -60,7 +60,7 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${ITT_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${ITT_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

 ocv_install_3rdparty_licenses(ittnotify src/ittnotify/LICENSE.BSD src/ittnotify/LICENSE.GPL)
@@ -17,7 +17,7 @@ file(GLOB lib_ext_hdrs jasper/*.h)
 #         Define the library target:
 # ----------------------------------------------------------------------------------

-add_library(${JASPER_LIBRARY} STATIC ${lib_srcs} ${lib_hdrs} ${lib_ext_hdrs})
+add_library(${JASPER_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${lib_srcs} ${lib_hdrs} ${lib_ext_hdrs})

 if(WIN32 AND NOT MINGW)
  add_definitions(-DJAS_WIN_MSVC_BUILD)
@@ -46,7 +46,7 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${JASPER_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${JASPER_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

 ocv_install_3rdparty_licenses(jasper LICENSE README copyright)
@@ -4,9 +4,9 @@ ocv_warnings_disable(CMAKE_C_FLAGS -Wunused-parameter -Wsign-compare -Wshorten-6

 set(VERSION_MAJOR 2)
 set(VERSION_MINOR 0)
-set(VERSION_REVISION 2)
+set(VERSION_REVISION 6)
 set(VERSION ${VERSION_MAJOR}.${VERSION_MINOR}.${VERSION_REVISION})
-set(LIBJPEG_TURBO_VERSION_NUMBER 2000002)
+set(LIBJPEG_TURBO_VERSION_NUMBER 2000006)

 string(TIMESTAMP BUILD "opencv-${OPENCV_VERSION}-libjpeg-turbo")
 if(CMAKE_BUILD_TYPE STREQUAL "Debug")
@@ -65,6 +65,8 @@ set(JPEG_LIB_VERSION 62)
 # OpenCV
 set(JPEG_LIB_VERSION "${VERSION}-${JPEG_LIB_VERSION}" PARENT_SCOPE)

+set(THREAD_LOCAL "")  # WITH_TURBOJPEG is not used
+
 if(MSVC)
  add_definitions(-W3 -wd4996 -wd4018)
 endif()
@@ -104,7 +106,7 @@ set(JPEG_SOURCES ${JPEG_SOURCES} jsimd_none.c)

 ocv_list_add_prefix(JPEG_SOURCES src/)

-add_library(${JPEG_LIBRARY} STATIC ${JPEG_SOURCES} ${SIMD_OBJS})
+add_library(${JPEG_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${JPEG_SOURCES} ${SIMD_OBJS})

 set_target_properties(${JPEG_LIBRARY}
  PROPERTIES OUTPUT_NAME ${JPEG_LIBRARY}
@@ -119,7 +121,7 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${JPEG_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${JPEG_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

 ocv_install_3rdparty_licenses(libjpeg-turbo README.md LICENSE.md README.ijg)
@@ -91,7 +91,7 @@ best of our understanding.
 The Modified (3-clause) BSD License
 ===================================

-Copyright (C)2009-2019 D. R. Commander.  All Rights Reserved.
+Copyright (C)2009-2020 D. R. Commander.  All Rights Reserved.
 Copyright (C)2015 Viktor Szathmáry.  All Rights Reserved.

 Redistribution and use in source and binary forms, with or without
@@ -223,12 +223,12 @@ https://www.iso.org/standard/54989.html and http://www.itu.int/rec/T-REC-T.871.
 A PDF file of the older JFIF 1.02 specification is available at
 http://www.w3.org/Graphics/JPEG/jfif3.pdf.

-The TIFF 6.0 file format specification can be obtained by FTP from
-ftp://ftp.sgi.com/graphics/tiff/TIFF6.ps.gz.  The JPEG incorporation scheme
-found in the TIFF 6.0 spec of 3-June-92 has a number of serious problems.
-IJG does not recommend use of the TIFF 6.0 design (TIFF Compression tag 6).
-Instead, we recommend the JPEG design proposed by TIFF Technical Note #2
-(Compression tag 7).  Copies of this Note can be obtained from
+The TIFF 6.0 file format specification can be obtained from
+http://mirrors.ctan.org/graphics/tiff/TIFF6.ps.gz.  The JPEG incorporation
+scheme found in the TIFF 6.0 spec of 3-June-92 has a number of serious
+problems.  IJG does not recommend use of the TIFF 6.0 design (TIFF Compression
+tag 6).  Instead, we recommend the JPEG design proposed by TIFF Technical Note
+#2 (Compression tag 7).  Copies of this Note can be obtained from
 http://www.ijg.org/files/.  It is expected that the next revision
 of the TIFF spec will replace the 6.0 JPEG design with the Note's design.
 Although IJG's own code does not support TIFF/JPEG, the free libtiff library
@@ -243,14 +243,8 @@ The most recent released version can always be found there in
 directory "files".

 The JPEG FAQ (Frequently Asked Questions) article is a source of some
-general information about JPEG.
-It is available on the World Wide Web at http://www.faqs.org/faqs/jpeg-faq/
-and other news.answers archive sites, including the official news.answers
-archive at rtfm.mit.edu: ftp://rtfm.mit.edu/pub/usenet/news.answers/jpeg-faq/.
-If you don't have Web or FTP access, send e-mail to mail-server@rtfm.mit.edu
-with body
-        send usenet/news.answers/jpeg-faq/part1
-        send usenet/news.answers/jpeg-faq/part2
+general information about JPEG.  It is available at
+http://www.faqs.org/faqs/jpeg-faq.


 FILE FORMAT COMPATIBILITY
@@ -1,14 +1,14 @@
 Background
 ==========

-libjpeg-turbo is a JPEG image codec that uses SIMD instructions (MMX, SSE2,
-AVX2, NEON, AltiVec) to accelerate baseline JPEG compression and decompression
-on x86, x86-64, ARM, and PowerPC systems, as well as progressive JPEG
-compression on x86 and x86-64 systems.  On such systems, libjpeg-turbo is
-generally 2-6x as fast as libjpeg, all else being equal.  On other types of
-systems, libjpeg-turbo can still outperform libjpeg by a significant amount, by
-virtue of its highly-optimized Huffman coding routines.  In many cases, the
-performance of libjpeg-turbo rivals that of proprietary high-speed JPEG codecs.
+libjpeg-turbo is a JPEG image codec that uses SIMD instructions to accelerate
+baseline JPEG compression and decompression on x86, x86-64, Arm, PowerPC, and
+MIPS systems, as well as progressive JPEG compression on x86 and x86-64
+systems.  On such systems, libjpeg-turbo is generally 2-6x as fast as libjpeg,
+all else being equal.  On other types of systems, libjpeg-turbo can still
+outperform libjpeg by a significant amount, by virtue of its highly-optimized
+Huffman coding routines.  In many cases, the performance of libjpeg-turbo
+rivals that of proprietary high-speed JPEG codecs.

 libjpeg-turbo implements both the traditional libjpeg API as well as the less
 powerful but more straightforward TurboJPEG API.  libjpeg-turbo also features
@@ -135,25 +135,24 @@ without recompiling.  libjpeg-turbo does not claim to support all of the
 libjpeg v7+ features, nor to produce identical output to libjpeg v7+ in all
 cases (see below.)

-By passing an argument of `--with-jpeg7` or `--with-jpeg8` to `configure`, or
-an argument of `-DWITH_JPEG7=1` or `-DWITH_JPEG8=1` to `cmake`, you can build a
-version of libjpeg-turbo that emulates the libjpeg v7 or v8 ABI, so that
-programs that are built against libjpeg v7 or v8 can be run with libjpeg-turbo.
-The following section describes which libjpeg v7+ features are supported and
-which aren't.
+By passing an argument of `-DWITH_JPEG7=1` or `-DWITH_JPEG8=1` to `cmake`, you
+can build a version of libjpeg-turbo that emulates the libjpeg v7 or v8 ABI, so
+that programs that are built against libjpeg v7 or v8 can be run with
+libjpeg-turbo.  The following section describes which libjpeg v7+ features are
+supported and which aren't.

 ### Support for libjpeg v7 and v8 Features

 #### Fully supported

- **libjpeg: IDCT scaling extensions in decompressor**<br>
+- **libjpeg API: IDCT scaling extensions in decompressor**<br>
  libjpeg-turbo supports IDCT scaling with scaling factors of 1/8, 1/4, 3/8,
  1/2, 5/8, 3/4, 7/8, 9/8, 5/4, 11/8, 3/2, 13/8, 7/4, 15/8, and 2/1 (only 1/4
  and 1/2 are SIMD-accelerated.)

- **libjpeg: Arithmetic coding**
+- **libjpeg API: Arithmetic coding**

- **libjpeg: In-memory source and destination managers**<br>
+- **libjpeg API: In-memory source and destination managers**<br>
  See notes below.

 - **cjpeg: Separate quality settings for luminance and chrominance**<br>
@@ -180,19 +179,19 @@ which aren't.

 NOTE:  As of this writing, extensive research has been conducted into the
 usefulness of DCT scaling as a means of data reduction and SmartScale as a
-means of quality improvement.  The reader is invited to peruse the research at
-<http://www.libjpeg-turbo.org/About/SmartScale> and draw his/her own conclusions,
+means of quality improvement.  Readers are invited to peruse the research at
+<http://www.libjpeg-turbo.org/About/SmartScale> and draw their own conclusions,
 but it is the general belief of our project that these features have not
 demonstrated sufficient usefulness to justify inclusion in libjpeg-turbo.

- **libjpeg: DCT scaling in compressor**<br>
+- **libjpeg API: DCT scaling in compressor**<br>
  `cinfo.scale_num` and `cinfo.scale_denom` are silently ignored.
  There is no technical reason why DCT scaling could not be supported when
  emulating the libjpeg v7+ API/ABI, but without the SmartScale extension (see
  below), only scaling factors of 1/2, 8/15, 4/7, 8/13, 2/3, 8/11, 4/5, and
  8/9 would be available, which is of limited usefulness.

- **libjpeg: SmartScale**<br>
+- **libjpeg API: SmartScale**<br>
  `cinfo.block_size` is silently ignored.
  SmartScale is an extension to the JPEG format that allows for DCT block
  sizes other than 8x8.  Providing support for this new format would be
@@ -205,7 +204,7 @@ demonstrated sufficient usefulness to justify inclusion in libjpeg-turbo.
  interest in providing this feature would be as a means of supporting
  additional DCT scaling factors.

- **libjpeg: Fancy downsampling in compressor**<br>
+- **libjpeg API: Fancy downsampling in compressor**<br>
  `cinfo.do_fancy_downsampling` is silently ignored.
  This requires the DCT scaling feature, which is not supported.

@@ -247,15 +246,14 @@ don't, and it allows those functions to be provided in the "official"
 libjpeg-turbo binaries.

 Those who are concerned about maintaining strict conformance with the libjpeg
-v6b or v7 API can pass an argument of `--without-mem-srcdst` to `configure` or
-an argument of `-DWITH_MEM_SRCDST=0` to `cmake` prior to building
-libjpeg-turbo.  This will restore the pre-1.3 behavior, in which
+v6b or v7 API can pass an argument of `-DWITH_MEM_SRCDST=0` to `cmake` prior to
+building libjpeg-turbo.  This will restore the pre-1.3 behavior, in which
 `jpeg_mem_src()` and `jpeg_mem_dest()` are only included when emulating the
 libjpeg v8 API/ABI.

 On Un*x systems, including the in-memory source/destination managers changes
-the dynamic library version from 62.1.0 to 62.2.0 if using libjpeg v6b API/ABI
-emulation and from 7.1.0 to 7.2.0 if using libjpeg v7 API/ABI emulation.
+the dynamic library version from 62.2.0 to 62.3.0 if using libjpeg v6b API/ABI
+emulation and from 7.2.0 to 7.3.0 if using libjpeg v7 API/ABI emulation.

 Note that, on most Un*x systems, the dynamic linker will not look for a
 function in a library until that function is actually used.  Thus, if a program
@@ -289,12 +287,13 @@ following reasons:
  (and slightly faster) floating point IDCT algorithm introduced in libjpeg
  v8a as opposed to the algorithm used in libjpeg v6b.  It should be noted,
  however, that this algorithm basically brings the accuracy of the floating
-  point IDCT in line with the accuracy of the slow integer IDCT.  The floating
-  point DCT/IDCT algorithms are mainly a legacy feature, and they do not
-  produce significantly more accuracy than the slow integer algorithms (to put
-  numbers on this, the typical difference in PNSR between the two algorithms
-  is less than 0.10 dB, whereas changing the quality level by 1 in the upper
-  range of the quality scale is typically more like a 1.0 dB difference.)
+  point IDCT in line with the accuracy of the accurate integer IDCT.  The
+  floating point DCT/IDCT algorithms are mainly a legacy feature, and they do
+  not produce significantly more accuracy than the accurate integer algorithms
+  (to put numbers on this, the typical difference in PNSR between the two
+  algorithms is less than 0.10 dB, whereas changing the quality level by 1 in
+  the upper range of the quality scale is typically more like a 1.0 dB
+  difference.)

 - If the floating point algorithms in libjpeg-turbo are not implemented using
  SIMD instructions on a particular platform, then the accuracy of the
@@ -331,7 +330,7 @@ in a way that makes the rest of the libjpeg infrastructure happy, so it is
 necessary to use the slow Huffman decoder when decompressing a JPEG image that
 has restart markers.  This can cause the decompression performance to drop by
 as much as 20%, but the performance will still be much greater than that of
-libjpeg.  Many consumer packages, such as PhotoShop, use restart markers when
+libjpeg.  Many consumer packages, such as Photoshop, use restart markers when
 generating JPEG images, so images generated by those programs will experience
 this issue.

@@ -342,5 +341,17 @@ The algorithm used by the SIMD-accelerated quantization function cannot produce
 correct results whenever the fast integer forward DCT is used along with a JPEG
 quality of 98-100.  Thus, libjpeg-turbo must use the non-SIMD quantization
 function in those cases.  This causes performance to drop by as much as 40%.
-It is therefore strongly advised that you use the slow integer forward DCT
+It is therefore strongly advised that you use the accurate integer forward DCT
 whenever encoding images with a JPEG quality of 98 or higher.
+
+
+Memory Debugger Pitfalls
+========================
+
+Valgrind and Memory Sanitizer (MSan) can generate false positives
+(specifically, incorrect reports of uninitialized memory accesses) when used
+with libjpeg-turbo's SIMD extensions.  It is generally recommended that the
+SIMD extensions be disabled, either by passing an argument of `-DWITH_SIMD=0`
+to `cmake` when configuring the build or by setting the environment variable
+`JSIMD_FORCENONE` to `1` at run time, when testing libjpeg-turbo with Valgrind,
+MSan, or other memory debuggers.
@@ -15,6 +15,9 @@
 #endif
 #endif

+/* How to obtain thread-local storage */
+#define THREAD_LOCAL  @THREAD_LOCAL@
+
 /* Define to the full name of this package. */
 #define PACKAGE_NAME  "@CMAKE_PROJECT_NAME@"

@@ -4,7 +4,7 @@
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1991-1997, Thomas G. Lane.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2009-2011, 2014-2016, 2018, D. R. Commander.
+ * Copyright (C) 2009-2011, 2014-2016, 2018-2019, D. R. Commander.
 * Copyright (C) 2015, Matthieu Darbois.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
@@ -34,17 +34,17 @@
 * memory footprint by 64k, which is important for some mobile applications
 * that create many isolated instances of libjpeg-turbo (web browsers, for
 * instance.)  This may improve performance on some mobile platforms as well.
- * This feature is enabled by default only on ARM processors, because some x86
+ * This feature is enabled by default only on Arm processors, because some x86
 * chips have a slow implementation of bsr, and the use of clz/bsr cannot be
 * shown to have a significant performance impact even on the x86 chips that
- * have a fast implementation of it.  When building for ARMv6, you can
+ * have a fast implementation of it.  When building for Armv6, you can
 * explicitly disable the use of clz/bsr by adding -mthumb to the compiler
 * flags (this defines __thumb__).
 */

 /* NOTE: Both GCC and Clang define __GNUC__ */
-#if defined __GNUC__ && (defined __arm__ || defined __aarch64__)
-#if !defined __thumb__ || defined __thumb2__
+#if defined(__GNUC__) && (defined(__arm__) || defined(__aarch64__))
+#if !defined(__thumb__) || defined(__thumb2__)
 #define USE_CLZ_INTRINSIC
 #endif
 #endif
@@ -356,6 +356,8 @@ dump_buffer(working_state *state)
  put_buffer = (put_buffer << size) | code; \
 }

+#if SIZEOF_SIZE_T != 8 && !defined(_WIN64)
+
 #define CHECKBUF15() { \
  if (put_bits > 15) { \
    EMIT_BYTE() \
@@ -363,6 +365,8 @@ dump_buffer(working_state *state)
  } \
 }

+#endif
+
 #define CHECKBUF31() { \
  if (put_bits > 31) { \
    EMIT_BYTE() \
@@ -428,7 +432,7 @@ dump_buffer(working_state *state)
 * scanning order-- 1, 8, 16, etc.), then this will produce an encoded block
 * larger than 200 bytes.
 */
-#define BUFSIZE  (DCTSIZE2 * 4)
+#define BUFSIZE  (DCTSIZE2 * 8)

 #define LOAD_BUFFER() { \
  if (state->free_in_buffer < BUFSIZE) { \
@@ -1,8 +1,10 @@
 /*
 * jcinit.c
 *
+ * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
+ * libjpeg-turbo Modifications:
+ * Copyright (C) 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -19,6 +21,7 @@
 #define JPEG_INTERNALS
 #include "jinclude.h"
 #include "jpeglib.h"
+#include "jpegcomp.h"


 /*
@@ -492,8 +492,8 @@ prepare_for_pass(j_compress_ptr cinfo)
     */
    master->pass_type = output_pass;
    master->pass_number++;
-    /*FALLTHROUGH*/
 #endif
+    /*FALLTHROUGH*/
  case output_pass:
    /* Do a data-output pass. */
    /* We need not repeat per-scan setup if prior optimization pass did it. */
@@ -43,17 +43,17 @@
 * memory footprint by 64k, which is important for some mobile applications
 * that create many isolated instances of libjpeg-turbo (web browsers, for
 * instance.)  This may improve performance on some mobile platforms as well.
- * This feature is enabled by default only on ARM processors, because some x86
+ * This feature is enabled by default only on Arm processors, because some x86
 * chips have a slow implementation of bsr, and the use of clz/bsr cannot be
 * shown to have a significant performance impact even on the x86 chips that
- * have a fast implementation of it.  When building for ARMv6, you can
+ * have a fast implementation of it.  When building for Armv6, you can
 * explicitly disable the use of clz/bsr by adding -mthumb to the compiler
 * flags (this defines __thumb__).
 */

 /* NOTE: Both GCC and Clang define __GNUC__ */
-#if defined __GNUC__ && (defined __arm__ || defined __aarch64__)
-#if !defined __thumb__ || defined __thumb2__
+#if defined(__GNUC__) && (defined(__arm__) || defined(__aarch64__))
+#if !defined(__thumb__) || defined(__thumb2__)
 #define USE_CLZ_INTRINSIC
 #endif
 #endif
@@ -4,8 +4,8 @@
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1995-1998, Thomas G. Lane.
 * Modified 2000-2009 by Guido Vollbeding.
- * It was modified by The libjpeg-turbo Project to include only code relevant
- * to libjpeg-turbo.
+ * libjpeg-turbo Modifications:
+ * Copyright (C) 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -17,6 +17,7 @@
 #define JPEG_INTERNALS
 #include "jinclude.h"
 #include "jpeglib.h"
+#include "jpegcomp.h"


 /* Forward declarations */
@@ -4,7 +4,7 @@
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1994-1996, Thomas G. Lane.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2010, 2015-2018, D. R. Commander.
+ * Copyright (C) 2010, 2015-2018, 2020, D. R. Commander.
 * Copyright (C) 2015, Google, Inc.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
@@ -21,6 +21,8 @@
 #include "jinclude.h"
 #include "jdmainct.h"
 #include "jdcoefct.h"
+#include "jdmaster.h"
+#include "jdmerge.h"
 #include "jdsample.h"
 #include "jmemsys.h"

@@ -316,6 +318,8 @@ LOCAL(void)
 read_and_discard_scanlines(j_decompress_ptr cinfo, JDIMENSION num_lines)
 {
  JDIMENSION n;
+  my_master_ptr master = (my_master_ptr)cinfo->master;
+  JSAMPARRAY scanlines = NULL;
  void (*color_convert) (j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
                         JDIMENSION input_row, JSAMPARRAY output_buf,
                         int num_rows) = NULL;
@@ -332,8 +336,13 @@ read_and_discard_scanlines(j_decompress_ptr cinfo, JDIMENSION num_lines)
    cinfo->cquantize->color_quantize = noop_quantize;
  }

+  if (master->using_merged_upsample && cinfo->max_v_samp_factor == 2) {
+    my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
+    scanlines = &upsample->spare_row;
+  }
+
  for (n = 0; n < num_lines; n++)
-    jpeg_read_scanlines(cinfo, NULL, 1);
+    jpeg_read_scanlines(cinfo, scanlines, 1);

  if (color_convert)
    cinfo->cconvert->color_convert = color_convert;
@@ -353,6 +362,12 @@ increment_simple_rowgroup_ctr(j_decompress_ptr cinfo, JDIMENSION rows)
 {
  JDIMENSION rows_left;
  my_main_ptr main_ptr = (my_main_ptr)cinfo->main;
+  my_master_ptr master = (my_master_ptr)cinfo->master;
+
+  if (master->using_merged_upsample && cinfo->max_v_samp_factor == 2) {
+    read_and_discard_scanlines(cinfo, rows);
+    return;
+  }

  /* Increment the counter to the next row group after the skipped rows. */
  main_ptr->rowgroup_ctr += rows / cinfo->max_v_samp_factor;
@@ -382,21 +397,27 @@ jpeg_skip_scanlines(j_decompress_ptr cinfo, JDIMENSION num_lines)
 {
  my_main_ptr main_ptr = (my_main_ptr)cinfo->main;
  my_coef_ptr coef = (my_coef_ptr)cinfo->coef;
+  my_master_ptr master = (my_master_ptr)cinfo->master;
  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
  JDIMENSION i, x;
  int y;
  JDIMENSION lines_per_iMCU_row, lines_left_in_iMCU_row, lines_after_iMCU_row;
  JDIMENSION lines_to_skip, lines_to_read;

+  /* Two-pass color quantization is not supported. */
+  if (cinfo->quantize_colors && cinfo->two_pass_quantize)
+    ERREXIT(cinfo, JERR_NOTIMPL);
+
  if (cinfo->global_state != DSTATE_SCANNING)
    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);

  /* Do not skip past the bottom of the image. */
  if (cinfo->output_scanline + num_lines >= cinfo->output_height) {
+    num_lines = cinfo->output_height - cinfo->output_scanline;
    cinfo->output_scanline = cinfo->output_height;
    (*cinfo->inputctl->finish_input_pass) (cinfo);
    cinfo->inputctl->eoi_reached = TRUE;
-    return cinfo->output_height - cinfo->output_scanline;
+    return num_lines;
  }

  if (num_lines == 0)
@@ -445,8 +466,10 @@ jpeg_skip_scanlines(j_decompress_ptr cinfo, JDIMENSION num_lines)
    main_ptr->buffer_full = FALSE;
    main_ptr->rowgroup_ctr = 0;
    main_ptr->context_state = CTX_PREPARE_FOR_IMCU;
-    upsample->next_row_out = cinfo->max_v_samp_factor;
-    upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;
+    if (!master->using_merged_upsample) {
+      upsample->next_row_out = cinfo->max_v_samp_factor;
+      upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;
+    }
  }

  /* Skipping is much simpler when context rows are not required. */
@@ -458,8 +481,10 @@ jpeg_skip_scanlines(j_decompress_ptr cinfo, JDIMENSION num_lines)
      cinfo->output_scanline += lines_left_in_iMCU_row;
      main_ptr->buffer_full = FALSE;
      main_ptr->rowgroup_ctr = 0;
-      upsample->next_row_out = cinfo->max_v_samp_factor;
-      upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;
+      if (!master->using_merged_upsample) {
+        upsample->next_row_out = cinfo->max_v_samp_factor;
+        upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;
+      }
    }
  }

@@ -494,7 +519,8 @@ jpeg_skip_scanlines(j_decompress_ptr cinfo, JDIMENSION num_lines)
      cinfo->output_iMCU_row += lines_to_skip / lines_per_iMCU_row;
      increment_simple_rowgroup_ctr(cinfo, lines_to_read);
    }
-    upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;
+    if (!master->using_merged_upsample)
+      upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;
    return num_lines;
  }

@@ -535,7 +561,8 @@ jpeg_skip_scanlines(j_decompress_ptr cinfo, JDIMENSION num_lines)
   * bit odd, since "rows_to_go" seems to be redundantly keeping track of
   * output_scanline.
   */
-  upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;
+  if (!master->using_merged_upsample)
+    upsample->rows_to_go = cinfo->output_height - cinfo->output_scanline;

  /* Always skip the requested number of lines. */
  return num_lines;
@@ -143,8 +143,7 @@ empty_mem_output_buffer(j_compress_ptr cinfo)

  MEMCOPY(nextbuffer, dest->buffer, dest->bufsize);

-  if (dest->newbuffer != NULL)
-    free(dest->newbuffer);
+  free(dest->newbuffer);

  dest->newbuffer = nextbuffer;

@@ -6,7 +6,7 @@
 * libjpeg-turbo Modifications:
 * Copyright 2009 Pierre Ossman <ossman@cendio.se> for Cendio AB
 * Copyright (C) 2010, 2015-2016, D. R. Commander.
- * Copyright (C) 2015, Google, Inc.
+ * Copyright (C) 2015, 2020, Google, Inc.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -495,11 +495,13 @@ decompress_smooth_data(j_decompress_ptr cinfo, JSAMPIMAGE output_buf)
      if (first_row && block_row == 0)
        prev_block_row = buffer_ptr;
      else
-        prev_block_row = buffer[block_row - 1];
+        prev_block_row = buffer[block_row - 1] +
+                         cinfo->master->first_MCU_col[ci];
      if (last_row && block_row == block_rows - 1)
        next_block_row = buffer_ptr;
      else
-        next_block_row = buffer[block_row + 1];
+        next_block_row = buffer[block_row + 1] +
+                         cinfo->master->first_MCU_col[ci];
      /* We fetch the surrounding DC values using a sliding-register approach.
       * Initialize all nine here so as to do the right thing on narrow pics.
       */
@@ -571,11 +571,10 @@ ycck_cmyk_convert(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
 * RGB565 conversion
 */

-#define PACK_SHORT_565_LE(r, g, b)  ((((r) << 8) & 0xF800) | \
-                                     (((g) << 3) & 0x7E0) | ((b) >> 3))
-#define PACK_SHORT_565_BE(r, g, b)  (((r) & 0xF8) | ((g) >> 5) | \
-                                     (((g) << 11) & 0xE000) | \
-                                     (((b) << 5) & 0x1F00))
+#define PACK_SHORT_565_LE(r, g, b) \
+  ((((r) << 8) & 0xF800) | (((g) << 3) & 0x7E0) | ((b) >> 3))
+#define PACK_SHORT_565_BE(r, g, b) \
+  (((r) & 0xF8) | ((g) >> 5) | (((g) << 11) & 0xE000) | (((b) << 5) & 0x1F00))

 #define PACK_TWO_PIXELS_LE(l, r)    ((r << 16) | l)
 #define PACK_TWO_PIXELS_BE(l, r)    ((l << 16) | r)
@@ -4,7 +4,7 @@
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1991-1997, Thomas G. Lane.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2009-2011, 2016, 2018, D. R. Commander.
+ * Copyright (C) 2009-2011, 2016, 2018-2019, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -589,7 +589,11 @@ decode_mcu_slow(j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
    if (entropy->dc_needed[blkn]) {
      /* Convert DC difference to actual value, update last_dc_val */
      int ci = cinfo->MCU_membership[blkn];
-      s += state.last_dc_val[ci];
+      /* This is really just
+       *   s += state.last_dc_val[ci];
+       * It is written this way in order to shut up UBSan.
+       */
+      s = (int)((unsigned int)s + (unsigned int)state.last_dc_val[ci]);
      state.last_dc_val[ci] = s;
      if (block) {
        /* Output the DC coefficient (assumes jpeg_natural_order[0] = 0) */
@@ -684,7 +688,7 @@ decode_mcu_fast(j_decompress_ptr cinfo, JBLOCKROW *MCU_data)

    if (entropy->dc_needed[blkn]) {
      int ci = cinfo->MCU_membership[blkn];
-      s += state.last_dc_val[ci];
+      s = (int)((unsigned int)s + (unsigned int)state.last_dc_val[ci]);
      state.last_dc_val[ci] = s;
      if (block)
        (*block)[0] = (JCOEF)s;
@@ -5,7 +5,7 @@
 * Copyright (C) 1994-1996, Thomas G. Lane.
 * libjpeg-turbo Modifications:
 * Copyright 2009 Pierre Ossman <ossman@cendio.se> for Cendio AB
- * Copyright (C) 2009, 2011, 2014-2015, D. R. Commander.
+ * Copyright (C) 2009, 2011, 2014-2015, 2020, D. R. Commander.
 * Copyright (C) 2013, Linaro Limited.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
@@ -40,41 +40,13 @@
 #define JPEG_INTERNALS
 #include "jinclude.h"
 #include "jpeglib.h"
+#include "jdmerge.h"
 #include "jsimd.h"
 #include "jconfigint.h"

 #ifdef UPSAMPLE_MERGING_SUPPORTED


-/* Private subobject */
-
-typedef struct {
-  struct jpeg_upsampler pub;    /* public fields */
-
-  /* Pointer to routine to do actual upsampling/conversion of one row group */
-  void (*upmethod) (j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
-                    JDIMENSION in_row_group_ctr, JSAMPARRAY output_buf);
-
-  /* Private state for YCC->RGB conversion */
-  int *Cr_r_tab;                /* => table for Cr to R conversion */
-  int *Cb_b_tab;                /* => table for Cb to B conversion */
-  JLONG *Cr_g_tab;              /* => table for Cr to G conversion */
-  JLONG *Cb_g_tab;              /* => table for Cb to G conversion */
-
-  /* For 2:1 vertical sampling, we produce two output rows at a time.
-   * We need a "spare" row buffer to hold the second output row if the
-   * application provides just a one-row buffer; we also use the spare
-   * to discard the dummy last row if the image height is odd.
-   */
-  JSAMPROW spare_row;
-  boolean spare_full;           /* T if spare buffer is occupied */
-
-  JDIMENSION out_row_width;     /* samples per output row */
-  JDIMENSION rows_to_go;        /* counts rows remaining in image */
-} my_upsampler;
-
-typedef my_upsampler *my_upsample_ptr;
-
 #define SCALEBITS       16      /* speediest right-shift on some machines */
 #define ONE_HALF        ((JLONG)1 << (SCALEBITS - 1))
 #define FIX(x)          ((JLONG)((x) * (1L << SCALEBITS) + 0.5))
@@ -189,7 +161,7 @@ typedef my_upsampler *my_upsample_ptr;
 LOCAL(void)
 build_ycc_rgb_table(j_decompress_ptr cinfo)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  int i;
  JLONG x;
  SHIFT_TEMPS
@@ -232,7 +204,7 @@ build_ycc_rgb_table(j_decompress_ptr cinfo)
 METHODDEF(void)
 start_pass_merged_upsample(j_decompress_ptr cinfo)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;

  /* Mark the spare buffer empty */
  upsample->spare_full = FALSE;
@@ -254,7 +226,7 @@ merged_2v_upsample(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
                   JDIMENSION *out_row_ctr, JDIMENSION out_rows_avail)
 /* 2:1 vertical sampling case: may need a spare row. */
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  JSAMPROW work_ptrs[2];
  JDIMENSION num_rows;          /* number of rows returned to caller */

@@ -305,7 +277,7 @@ merged_1v_upsample(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
                   JDIMENSION *out_row_ctr, JDIMENSION out_rows_avail)
 /* 1:1 vertical sampling case: much easier, never need a spare row. */
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;

  /* Just do the upsampling. */
  (*upsample->upmethod) (cinfo, input_buf, *in_row_group_ctr,
@@ -420,17 +392,14 @@ h2v2_merged_upsample(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
 * RGB565 conversion
 */

-#define PACK_SHORT_565_LE(r, g, b)  ((((r) << 8) & 0xF800) | \
-                                     (((g) << 3) & 0x7E0) | ((b) >> 3))
-#define PACK_SHORT_565_BE(r, g, b)  (((r) & 0xF8) | ((g) >> 5) | \
-                                     (((g) << 11) & 0xE000) | \
-                                     (((b) << 5) & 0x1F00))
+#define PACK_SHORT_565_LE(r, g, b) \
+  ((((r) << 8) & 0xF800) | (((g) << 3) & 0x7E0) | ((b) >> 3))
+#define PACK_SHORT_565_BE(r, g, b) \
+  (((r) & 0xF8) | ((g) >> 5) | (((g) << 11) & 0xE000) | (((b) << 5) & 0x1F00))

 #define PACK_TWO_PIXELS_LE(l, r)    ((r << 16) | l)
 #define PACK_TWO_PIXELS_BE(l, r)    ((l << 16) | r)

-#define PACK_NEED_ALIGNMENT(ptr)    (((size_t)(ptr)) & 3)
-
 #define WRITE_TWO_PIXELS_LE(addr, pixels) { \
  ((INT16 *)(addr))[0] = (INT16)(pixels); \
  ((INT16 *)(addr))[1] = (INT16)((pixels) >> 16); \
@@ -568,11 +537,11 @@ h2v2_merged_upsample_565D(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
 GLOBAL(void)
 jinit_merged_upsampler(j_decompress_ptr cinfo)
 {
-  my_upsample_ptr upsample;
+  my_merged_upsample_ptr upsample;

-  upsample = (my_upsample_ptr)
+  upsample = (my_merged_upsample_ptr)
    (*cinfo->mem->alloc_small) ((j_common_ptr)cinfo, JPOOL_IMAGE,
-                                sizeof(my_upsampler));
+                                sizeof(my_merged_upsampler));
  cinfo->upsample = (struct jpeg_upsampler *)upsample;
  upsample->pub.start_pass = start_pass_merged_upsample;
  upsample->pub.need_context_rows = FALSE;
@@ -0,0 +1,47 @@
+/*
+ * jdmerge.h
+ *
+ * This file was part of the Independent JPEG Group's software:
+ * Copyright (C) 1994-1996, Thomas G. Lane.
+ * libjpeg-turbo Modifications:
+ * Copyright (C) 2020, D. R. Commander.
+ * For conditions of distribution and use, see the accompanying README.ijg
+ * file.
+ */
+
+#define JPEG_INTERNALS
+#include "jpeglib.h"
+
+#ifdef UPSAMPLE_MERGING_SUPPORTED
+
+
+/* Private subobject */
+
+typedef struct {
+  struct jpeg_upsampler pub;    /* public fields */
+
+  /* Pointer to routine to do actual upsampling/conversion of one row group */
+  void (*upmethod) (j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
+                    JDIMENSION in_row_group_ctr, JSAMPARRAY output_buf);
+
+  /* Private state for YCC->RGB conversion */
+  int *Cr_r_tab;                /* => table for Cr to R conversion */
+  int *Cb_b_tab;                /* => table for Cb to B conversion */
+  JLONG *Cr_g_tab;              /* => table for Cr to G conversion */
+  JLONG *Cb_g_tab;              /* => table for Cb to G conversion */
+
+  /* For 2:1 vertical sampling, we produce two output rows at a time.
+   * We need a "spare" row buffer to hold the second output row if the
+   * application provides just a one-row buffer; we also use the spare
+   * to discard the dummy last row if the image height is odd.
+   */
+  JSAMPROW spare_row;
+  boolean spare_full;           /* T if spare buffer is occupied */
+
+  JDIMENSION out_row_width;     /* samples per output row */
+  JDIMENSION rows_to_go;        /* counts rows remaining in image */
+} my_merged_upsampler;
+
+typedef my_merged_upsampler *my_merged_upsample_ptr;
+
+#endif /* UPSAMPLE_MERGING_SUPPORTED */
@@ -5,7 +5,7 @@
 * Copyright (C) 1994-1996, Thomas G. Lane.
 * libjpeg-turbo Modifications:
 * Copyright (C) 2013, Linaro Limited.
- * Copyright (C) 2014-2015, 2018, D. R. Commander.
+ * Copyright (C) 2014-2015, 2018, 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -19,7 +19,7 @@ h2v1_merged_upsample_565_internal(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
                                  JDIMENSION in_row_group_ctr,
                                  JSAMPARRAY output_buf)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  register int y, cred, cgreen, cblue;
  int cb, cr;
  register JSAMPROW outptr;
@@ -90,7 +90,7 @@ h2v1_merged_upsample_565D_internal(j_decompress_ptr cinfo,
                                   JDIMENSION in_row_group_ctr,
                                   JSAMPARRAY output_buf)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  register int y, cred, cgreen, cblue;
  int cb, cr;
  register JSAMPROW outptr;
@@ -163,7 +163,7 @@ h2v2_merged_upsample_565_internal(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
                                  JDIMENSION in_row_group_ctr,
                                  JSAMPARRAY output_buf)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  register int y, cred, cgreen, cblue;
  int cb, cr;
  register JSAMPROW outptr0, outptr1;
@@ -259,7 +259,7 @@ h2v2_merged_upsample_565D_internal(j_decompress_ptr cinfo,
                                   JDIMENSION in_row_group_ctr,
                                   JSAMPARRAY output_buf)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  register int y, cred, cgreen, cblue;
  int cb, cr;
  register JSAMPROW outptr0, outptr1;
@@ -4,7 +4,7 @@
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1994-1996, Thomas G. Lane.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2011, 2015, D. R. Commander.
+ * Copyright (C) 2011, 2015, 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -25,7 +25,7 @@ h2v1_merged_upsample_internal(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
                              JDIMENSION in_row_group_ctr,
                              JSAMPARRAY output_buf)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  register int y, cred, cgreen, cblue;
  int cb, cr;
  register JSAMPROW outptr;
@@ -97,7 +97,7 @@ h2v2_merged_upsample_internal(j_decompress_ptr cinfo, JSAMPIMAGE input_buf,
                              JDIMENSION in_row_group_ctr,
                              JSAMPARRAY output_buf)
 {
-  my_upsample_ptr upsample = (my_upsample_ptr)cinfo->upsample;
+  my_merged_upsample_ptr upsample = (my_merged_upsample_ptr)cinfo->upsample;
  register int y, cred, cgreen, cblue;
  int cb, cr;
  register JSAMPROW outptr0, outptr1;
@@ -8,6 +8,7 @@
 * Copyright (C) 2010, 2015-2016, D. R. Commander.
 * Copyright (C) 2014, MIPS Technologies, Inc., California.
 * Copyright (C) 2015, Google, Inc.
+ * Copyright (C) 2019, Arm Limited.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -315,9 +316,9 @@ h1v2_fancy_upsample(j_decompress_ptr cinfo, jpeg_component_info *compptr,
  JSAMPARRAY output_data = *output_data_ptr;
  JSAMPROW inptr0, inptr1, outptr;
 #if BITS_IN_JSAMPLE == 8
-  int thiscolsum;
+  int thiscolsum, bias;
 #else
-  JLONG thiscolsum;
+  JLONG thiscolsum, bias;
 #endif
  JDIMENSION colctr;
  int inrow, outrow, v;
@@ -327,15 +328,18 @@ h1v2_fancy_upsample(j_decompress_ptr cinfo, jpeg_component_info *compptr,
    for (v = 0; v < 2; v++) {
      /* inptr0 points to nearest input row, inptr1 points to next nearest */
      inptr0 = input_data[inrow];
-      if (v == 0)               /* next nearest is row above */
+      if (v == 0) {             /* next nearest is row above */
        inptr1 = input_data[inrow - 1];
-      else                      /* next nearest is row below */
+        bias = 1;
+      } else {                  /* next nearest is row below */
        inptr1 = input_data[inrow + 1];
+        bias = 2;
+      }
      outptr = output_data[outrow++];

      for (colctr = 0; colctr < compptr->downsampled_width; colctr++) {
        thiscolsum = GETJSAMPLE(*inptr0++) * 3 + GETJSAMPLE(*inptr1++);
-        *outptr++ = (JSAMPLE)((thiscolsum + 1) >> 2);
+        *outptr++ = (JSAMPLE)((thiscolsum + bias) >> 2);
      }
    }
    inrow++;
@@ -3,8 +3,8 @@
 *
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1995-1997, Thomas G. Lane.
- * It was modified by The libjpeg-turbo Project to include only code relevant
- * to libjpeg-turbo.
+ * libjpeg-turbo Modifications:
+ * Copyright (C) 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -16,6 +16,7 @@
 #define JPEG_INTERNALS
 #include "jinclude.h"
 #include "jpeglib.h"
+#include "jpegcomp.h"


 /* Forward declarations */
@@ -1,14 +1,14 @@
 /*
 * jfdctint.c
 *
- * This file was part of the Independent JPEG Group's software.
+ * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1991-1996, Thomas G. Lane.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2015, D. R. Commander.
+ * Copyright (C) 2015, 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
- * This file contains a slow-but-accurate integer implementation of the
+ * This file contains a slower but more accurate integer implementation of the
 * forward DCT (Discrete Cosine Transform).
 *
 * A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
@@ -1,15 +1,15 @@
 /*
 * jidctint.c
 *
- * This file was part of the Independent JPEG Group's software.
+ * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1991-1998, Thomas G. Lane.
 * Modification developed 2002-2009 by Guido Vollbeding.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2015, D. R. Commander.
+ * Copyright (C) 2015, 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
- * This file contains a slow-but-accurate integer implementation of the
+ * This file contains a slower but more accurate integer implementation of the
 * inverse DCT (Discrete Cosine Transform).  In the IJG code, this routine
 * must also perform dequantization of the input coefficients.
 *
@@ -1,7 +1,7 @@
 /*
 * jidctred.c
 *
- * This file was part of the Independent JPEG Group's software.
+ * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1994-1998, Thomas G. Lane.
 * libjpeg-turbo Modifications:
 * Copyright (C) 2015, D. R. Commander.
@@ -5,7 +5,7 @@
 * Copyright (C) 1991-1997, Thomas G. Lane.
 * Modified 1997-2009 by Guido Vollbeding.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2009, 2011, 2014-2015, 2018, D. R. Commander.
+ * Copyright (C) 2009, 2011, 2014-2015, 2018, 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -273,9 +273,9 @@ typedef int boolean;

 /* Capability options common to encoder and decoder: */

-#define DCT_ISLOW_SUPPORTED     /* slow but accurate integer algorithm */
-#define DCT_IFAST_SUPPORTED     /* faster, less accurate integer method */
-#define DCT_FLOAT_SUPPORTED     /* floating-point: accurate, fast on fast HW */
+#define DCT_ISLOW_SUPPORTED     /* accurate integer method */
+#define DCT_IFAST_SUPPORTED     /* less accurate int method [legacy feature] */
+#define DCT_FLOAT_SUPPORTED     /* floating-point method [legacy feature] */

 /* Encoder capability options: */

@@ -1,7 +1,7 @@
 /*
 * jpegcomp.h
 *
- * Copyright (C) 2010, D. R. Commander.
+ * Copyright (C) 2010, 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -19,6 +19,7 @@
 #define _min_DCT_v_scaled_size  min_DCT_v_scaled_size
 #define _jpeg_width  jpeg_width
 #define _jpeg_height  jpeg_height
+#define JERR_ARITH_NOTIMPL  JERR_NOT_COMPILED
 #else
 #define _DCT_scaled_size  DCT_scaled_size
 #define _DCT_h_scaled_size  DCT_scaled_size
@@ -5,7 +5,7 @@
 * Copyright (C) 1991-1998, Thomas G. Lane.
 * Modified 2002-2009 by Guido Vollbeding.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2009-2011, 2013-2014, 2016-2017, D. R. Commander.
+ * Copyright (C) 2009-2011, 2013-2014, 2016-2017, 2020, D. R. Commander.
 * Copyright (C) 2015, Google, Inc.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
@@ -244,9 +244,9 @@ typedef enum {
 /* DCT/IDCT algorithm options. */

 typedef enum {
-  JDCT_ISLOW,             /* slow but accurate integer algorithm */
-  JDCT_IFAST,             /* faster, less accurate integer method */
-  JDCT_FLOAT              /* floating-point: accurate, fast on fast HW */
+  JDCT_ISLOW,             /* accurate integer method */
+  JDCT_IFAST,             /* less accurate integer method [legacy feature] */
+  JDCT_FLOAT              /* floating-point method [legacy feature] */
 } J_DCT_METHOD;

 #ifndef JDCT_DEFAULT            /* may be overridden in jconfig.h */
@@ -4,7 +4,7 @@
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1991-1996, Thomas G. Lane.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2009, 2014-2015, D. R. Commander.
+ * Copyright (C) 2009, 2014-2015, 2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -1145,7 +1145,7 @@ start_pass_2_quant(j_decompress_ptr cinfo, boolean is_pre_scan)
  int i;

  /* Only F-S dithering or no dithering is supported. */
-  /* If user asks for ordered dither, give him F-S. */
+  /* If user asks for ordered dither, give them F-S. */
  if (cinfo->dither_mode != JDITHER_NONE)
    cinfo->dither_mode = JDITHER_FS;

@@ -1263,7 +1263,7 @@ jinit_2pass_quantizer(j_decompress_ptr cinfo)
    cquantize->sv_colormap = NULL;

  /* Only F-S dithering or no dithering is supported. */
-  /* If user asks for ordered dither, give him F-S. */
+  /* If user asks for ordered dither, give them F-S. */
  if (cinfo->dither_mode != JDITHER_NONE)
    cinfo->dither_mode = JDITHER_FS;

@@ -4,7 +4,7 @@
 * This file was part of the Independent JPEG Group's software:
 * Copyright (C) 1991-2012, Thomas G. Lane, Guido Vollbeding.
 * libjpeg-turbo Modifications:
- * Copyright (C) 2010, 2012-2019, D. R. Commander.
+ * Copyright (C) 2010, 2012-2020, D. R. Commander.
 * For conditions of distribution and use, see the accompanying README.ijg
 * file.
 *
@@ -30,23 +30,25 @@
 * NOTE: It is our convention to place the authors in the following order:
 * - libjpeg-turbo authors (2009-) in descending order of the date of their
 *   most recent contribution to the project, then in ascending order of the
- *   date of their first contribution to the project
+ *   date of their first contribution to the project, then in alphabetical
+ *   order
 * - Upstream authors in descending order of the date of the first inclusion of
 *   their code
 */

 #define JCOPYRIGHT \
-  "Copyright (C) 2009-2019 D. R. Commander\n" \
-  "Copyright (C) 2011-2016 Siarhei Siamashka\n" \
+  "Copyright (C) 2009-2020 D. R. Commander\n" \
+  "Copyright (C) 2015, 2020 Google, Inc.\n" \
+  "Copyright (C) 2019 Arm Limited\n" \
  "Copyright (C) 2015-2016, 2018 Matthieu Darbois\n" \
+  "Copyright (C) 2011-2016 Siarhei Siamashka\n" \
  "Copyright (C) 2015 Intel Corporation\n" \
-  "Copyright (C) 2015 Google, Inc.\n" \
+  "Copyright (C) 2013-2014 Linaro Limited\n" \
  "Copyright (C) 2013-2014 MIPS Technologies, Inc.\n" \
-  "Copyright (C) 2013 Linaro Limited\n" \
+  "Copyright (C) 2009, 2012 Pierre Ossman for Cendio AB\n" \
  "Copyright (C) 2009-2011 Nokia Corporation and/or its subsidiary(-ies)\n" \
-  "Copyright (C) 2009 Pierre Ossman for Cendio AB\n" \
  "Copyright (C) 1999-2006 MIYASAKA Masaru\n" \
-  "Copyright (C) 1991-2016 Thomas G. Lane, Guido Vollbeding"
+  "Copyright (C) 1991-2017 Thomas G. Lane, Guido Vollbeding"

 #define JCOPYRIGHT_SHORT \
-  "Copyright (C) 1991-2019 The libjpeg-turbo Project and many others"
+  "Copyright (C) 1991-2020 The libjpeg-turbo Project and many others"
@@ -19,7 +19,7 @@ endif()
 #         Define the library target:
 # ----------------------------------------------------------------------------------

-add_library(${JPEG_LIBRARY} STATIC ${lib_srcs} ${lib_hdrs})
+add_library(${JPEG_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${lib_srcs} ${lib_hdrs})

 if(CV_GCC OR CV_CLANG)
  set_source_files_properties(jcdctmgr.c PROPERTIES COMPILE_FLAGS "-O1")
@@ -42,7 +42,7 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${JPEG_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${JPEG_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

 ocv_install_3rdparty_licenses(libjpeg README)
@@ -1,7 +1,7 @@
 The Independent JPEG Group's JPEG software
 ==========================================

-README for release 9c of 14-Jan-2018
+README for release 9d of 12-Jan-2020
 ====================================

 This distribution contains the ninth public release of the Independent JPEG
@@ -10,8 +10,8 @@ to use it for any purpose, subject to the conditions under LEGAL ISSUES, below.

 This software is the work of Tom Lane, Guido Vollbeding, Philip Gladstone,
 Bill Allombert, Jim Boucher, Lee Crocker, Bob Friesenhahn, Ben Jackson,
-Julian Minguillon, Luis Ortiz, George Phillips, Davide Rossi, Ge' Weijers,
-and other members of the Independent JPEG Group.
+John Korejwa, Julian Minguillon, Luis Ortiz, George Phillips, Davide Rossi,
+Ge' Weijers, and other members of the Independent JPEG Group.

 IJG is not affiliated with the ISO/IEC JTC1/SC29/WG1 standards committee
 (previously known as JPEG, together with ITU-T SG16).
@@ -115,7 +115,7 @@ with respect to this software, its quality, accuracy, merchantability, or
 fitness for a particular purpose.  This software is provided "AS IS", and you,
 its user, assume the entire risk as to its quality and accuracy.

-This software is copyright (C) 1991-2018, Thomas G. Lane, Guido Vollbeding.
+This software is copyright (C) 1991-2020, Thomas G. Lane, Guido Vollbeding.
 All Rights Reserved except as specified below.

 Permission is hereby granted to use, copy, modify, and distribute this
@@ -152,13 +152,6 @@ The same holds for its supporting scripts (config.guess, config.sub,
 ltmain.sh).  Another support script, install-sh, is copyright by X Consortium
 but is also freely distributable.

-The IJG distribution formerly included code to read and write GIF files.
-To avoid entanglement with the Unisys LZW patent (now expired), GIF reading
-support has been removed altogether, and the GIF writer has been simplified
-to produce "uncompressed GIFs".  This technique does not use the LZW
-algorithm; the resulting GIF files are larger than usual, but are readable
-by all standard GIF decoders.
-

 REFERENCES
 ==========
@@ -246,8 +239,8 @@ ARCHIVE LOCATIONS
 The "official" archive site for this software is www.ijg.org.
 The most recent released version can always be found there in
 directory "files".  This particular version will be archived as
-http://www.ijg.org/files/jpegsrc.v9c.tar.gz, and in Windows-compatible
-"zip" archive format as http://www.ijg.org/files/jpegsr9c.zip.
+http://www.ijg.org/files/jpegsrc.v9d.tar.gz, and in Windows-compatible
+"zip" archive format as http://www.ijg.org/files/jpegsr9d.zip.

 The JPEG FAQ (Frequently Asked Questions) article is a source of some
 general information about JPEG.
@@ -1,6 +1,55 @@
 CHANGE LOG for Independent JPEG Group's JPEG software


+Version 9d  12-Jan-2020
+-----------------------
+
+Optimize the optimal Huffman code table generation to produce
+slightly smaller files.  Thank to John Korejwa for suggestion.
+Note: Requires rebuild of testimgp.jpg.
+
+Decoding Huffman: Use default tables if tables are not defined.
+Thank to Simone Azzalin for report (Motion JPEG),
+and to Martin Strunz for hint.
+
+Add sanity check in optimal Huffman code table generation.
+Thank to Adam Farley for suggestion.
+
+rdtarga.c: use read_byte(), with EOF check, instead of getc()
+in read_*_pixel().
+Thank to Chijin Zhou for cjpeg potential vulnerability report.
+
+jmemnobs.c: respect the max_memory_to_use setting in
+jpeg_mem_available() computation.  Thank to Sheng Shu and
+Dongdong She for djpeg potential vulnerability report.
+
+jdarith.c, jdhuff.c: avoid left shift of negative value
+compiler warning in decode_mcu_AC_refine().
+Thank to Indu Bhagat for suggestion.
+
+Add x64 (64-bit) platform support, avoid compiler warnings.
+Thank to Jonathan Potter, Feiyun Wang, and Sheng Shu for suggestion.
+
+Adjust libjpeg version specification for pkg-config file.
+Thank to Chen Chen for suggestion.
+
+Restore GIF read and write support from libjpeg version 6a.
+Thank to Wolfgang Werner (W.W.) Heinz for suggestion.
+
+Improve consistency in raw (downsampled) image data processing mode.
+Thank to Zhongyuan Zhou for hint.
+
+Avoid out of bounds array read (AC derived table pointers)
+in start pass in jdhuff.c.  Thank to Peng Li for report.
+
+Improve code sanity (jdhuff.c).
+Thank to Reza Mirzazade farkhani for reports.
+
+Add jpegtran -drop option; add options to the crop extension and wipe
+to fill the extra area with content from the source image region,
+instead of gray out.
+
+
 Version 9c  14-Jan-2018
 -----------------------

@@ -1,7 +1,7 @@
 /*
 * jcarith.c
 *
- * Developed 1997-2013 by Guido Vollbeding.
+ * Developed 1997-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -181,11 +181,11 @@ finish_pass (j_compress_ptr cinfo)
    if (e->zc)  /* output final pending zero bytes */
      do emit_byte(0x00, cinfo);
      while (--e->zc);
-    emit_byte((e->c >> 19) & 0xFF, cinfo);
+    emit_byte((int) ((e->c >> 19) & 0xFF), cinfo);
    if (((e->c >> 19) & 0xFF) == 0xFF)
      emit_byte(0x00, cinfo);
    if (e->c & 0x7F800L) {
-      emit_byte((e->c >> 11) & 0xFF, cinfo);
+      emit_byte((int) ((e->c >> 11) & 0xFF), cinfo);
      if (((e->c >> 11) & 0xFF) == 0xFF)
 	emit_byte(0x00, cinfo);
    }
@@ -280,7 +280,8 @@ arith_encode (j_compress_ptr cinfo, unsigned char *st, int val)
 	/* Note: The 3 spacer bits in the C register guarantee
 	 * that the new buffer byte can't be 0xFF here
 	 * (see page 160 in the P&M JPEG book). */
-	e->buffer = temp & 0xFF;  /* new output byte, might overflow later */
+	/* New output byte, might overflow later */
+	e->buffer = (int) (temp & 0xFF);
      } else if (temp == 0xFF) {
 	++e->sc;  /* stack 0xFF byte (which might overflow later) */
      } else {
@@ -302,7 +303,8 @@ arith_encode (j_compress_ptr cinfo, unsigned char *st, int val)
 	    emit_byte(0x00, cinfo);
 	  } while (--e->sc);
 	}
-	e->buffer = temp & 0xFF;  /* new output byte (can still overflow) */
+	/* New output byte (can still overflow) */
+	e->buffer = (int) (temp & 0xFF);
      }
      e->c &= 0x7FFFFL;
      e->ct += 8;
@@ -926,9 +928,8 @@ jinit_arith_encoder (j_compress_ptr cinfo)
  arith_entropy_ptr entropy;
  int i;

-  entropy = (arith_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(arith_entropy_encoder));
+  entropy = (arith_entropy_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(arith_entropy_encoder));
  cinfo->entropy = &entropy->pub;
  entropy->pub.start_pass = start_pass;
  entropy->pub.finish_pass = finish_pass;
@@ -2,7 +2,7 @@
 * jccolor.c
 *
 * Copyright (C) 1991-1996, Thomas G. Lane.
- * Modified 2011-2013 by Guido Vollbeding.
+ * Modified 2011-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -105,14 +105,14 @@ rgb_ycc_start (j_compress_ptr cinfo)
  /* Allocate and fill in the conversion tables. */
  cconvert->rgb_ycc_tab = rgb_ycc_tab = (INT32 *)
    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(TABLE_SIZE * SIZEOF(INT32)));
+				TABLE_SIZE * SIZEOF(INT32));

  for (i = 0; i <= MAXJSAMPLE; i++) {
    rgb_ycc_tab[i+R_Y_OFF] = FIX(0.299) * i;
    rgb_ycc_tab[i+G_Y_OFF] = FIX(0.587) * i;
    rgb_ycc_tab[i+B_Y_OFF] = FIX(0.114) * i   + ONE_HALF;
-    rgb_ycc_tab[i+R_CB_OFF] = (-FIX(0.168735892)) * i;
-    rgb_ycc_tab[i+G_CB_OFF] = (-FIX(0.331264108)) * i;
+    rgb_ycc_tab[i+R_CB_OFF] = (- FIX(0.168735892)) * i;
+    rgb_ycc_tab[i+G_CB_OFF] = (- FIX(0.331264108)) * i;
    /* We use a rounding fudge-factor of 0.5-epsilon for Cb and Cr.
     * This ensures that the maximum output will round to MAXJSAMPLE
     * not MAXJSAMPLE+1, and thus that we don't have to range-limit.
@@ -121,8 +121,8 @@ rgb_ycc_start (j_compress_ptr cinfo)
 /*  B=>Cb and R=>Cr tables are the same
    rgb_ycc_tab[i+R_CR_OFF] = FIX(0.5) * i    + CBCR_OFFSET + ONE_HALF-1;
 */
-    rgb_ycc_tab[i+G_CR_OFF] = (-FIX(0.418687589)) * i;
-    rgb_ycc_tab[i+B_CR_OFF] = (-FIX(0.081312411)) * i;
+    rgb_ycc_tab[i+G_CR_OFF] = (- FIX(0.418687589)) * i;
+    rgb_ycc_tab[i+B_CR_OFF] = (- FIX(0.081312411)) * i;
  }
 }

@@ -131,12 +131,12 @@ rgb_ycc_start (j_compress_ptr cinfo)
 * Convert some rows of samples to the JPEG colorspace.
 *
 * Note that we change from the application's interleaved-pixel format
- * to our internal noninterleaved, one-plane-per-component format.
- * The input buffer is therefore three times as wide as the output buffer.
+ * to our internal noninterleaved, one-plane-per-component format.  The
+ * input buffer is therefore three times as wide as the output buffer.
 *
- * A starting row offset is provided only for the output buffer.  The caller
- * can easily adjust the passed input_buf value to accommodate any row
- * offset required on that side.
+ * A starting row offset is provided only for the output buffer.  The
+ * caller can easily adjust the passed input_buf value to accommodate
+ * any row offset required on that side.
 */

 METHODDEF(void)
@@ -145,8 +145,8 @@ rgb_ycc_convert (j_compress_ptr cinfo,
 		 JDIMENSION output_row, int num_rows)
 {
  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register INT32 * ctab = cconvert->rgb_ycc_tab;
  register int r, g, b;
+  register INT32 * ctab = cconvert->rgb_ycc_tab;
  register JSAMPROW inptr;
  register JSAMPROW outptr0, outptr1, outptr2;
  register JDIMENSION col;
@@ -162,6 +162,7 @@ rgb_ycc_convert (j_compress_ptr cinfo,
      r = GETJSAMPLE(inptr[RGB_RED]);
      g = GETJSAMPLE(inptr[RGB_GREEN]);
      b = GETJSAMPLE(inptr[RGB_BLUE]);
+      inptr += RGB_PIXELSIZE;
      /* If the inputs are 0..MAXJSAMPLE, the outputs of these equations
       * must be too; we do not need an explicit range-limiting operation.
       * Hence the value being shifted is never negative, and we don't
@@ -179,7 +180,6 @@ rgb_ycc_convert (j_compress_ptr cinfo,
      outptr2[col] = (JSAMPLE)
 		((ctab[r+R_CR_OFF] + ctab[g+G_CR_OFF] + ctab[b+B_CR_OFF])
 		 >> SCALEBITS);
-      inptr += RGB_PIXELSIZE;
    }
  }
 }
@@ -201,8 +201,8 @@ rgb_gray_convert (j_compress_ptr cinfo,
 		  JDIMENSION output_row, int num_rows)
 {
  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register INT32 * ctab = cconvert->rgb_ycc_tab;
  register int r, g, b;
+  register INT32 * ctab = cconvert->rgb_ycc_tab;
  register JSAMPROW inptr;
  register JSAMPROW outptr;
  register JDIMENSION col;
@@ -215,11 +215,11 @@ rgb_gray_convert (j_compress_ptr cinfo,
      r = GETJSAMPLE(inptr[RGB_RED]);
      g = GETJSAMPLE(inptr[RGB_GREEN]);
      b = GETJSAMPLE(inptr[RGB_BLUE]);
+      inptr += RGB_PIXELSIZE;
      /* Y */
      outptr[col] = (JSAMPLE)
 		((ctab[r+R_Y_OFF] + ctab[g+G_Y_OFF] + ctab[b+B_Y_OFF])
 		 >> SCALEBITS);
-      inptr += RGB_PIXELSIZE;
    }
  }
 }
@@ -228,8 +228,8 @@ rgb_gray_convert (j_compress_ptr cinfo,
 /*
 * Convert some rows of samples to the JPEG colorspace.
 * This version handles Adobe-style CMYK->YCCK conversion,
- * where we convert R=1-C, G=1-M, and B=1-Y to YCbCr using the same
- * conversion as above, while passing K (black) unchanged.
+ * where we convert R=1-C, G=1-M, and B=1-Y to YCbCr using the
+ * same conversion as above, while passing K (black) unchanged.
 * We assume rgb_ycc_start has been called.
 */

@@ -239,8 +239,8 @@ cmyk_ycck_convert (j_compress_ptr cinfo,
 		   JDIMENSION output_row, int num_rows)
 {
  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register INT32 * ctab = cconvert->rgb_ycc_tab;
  register int r, g, b;
+  register INT32 * ctab = cconvert->rgb_ycc_tab;
  register JSAMPROW inptr;
  register JSAMPROW outptr0, outptr1, outptr2, outptr3;
  register JDIMENSION col;
@@ -259,6 +259,7 @@ cmyk_ycck_convert (j_compress_ptr cinfo,
      b = MAXJSAMPLE - GETJSAMPLE(inptr[2]);
      /* K passes through as-is */
      outptr3[col] = inptr[3];	/* don't need GETJSAMPLE here */
+      inptr += 4;
      /* If the inputs are 0..MAXJSAMPLE, the outputs of these equations
       * must be too; we do not need an explicit range-limiting operation.
       * Hence the value being shifted is never negative, and we don't
@@ -276,7 +277,6 @@ cmyk_ycck_convert (j_compress_ptr cinfo,
      outptr2[col] = (JSAMPLE)
 		((ctab[r+R_CR_OFF] + ctab[g+G_CR_OFF] + ctab[b+B_CR_OFF])
 		 >> SCALEBITS);
-      inptr += 4;
    }
  }
 }
@@ -312,13 +312,13 @@ rgb_rgb1_convert (j_compress_ptr cinfo,
      r = GETJSAMPLE(inptr[RGB_RED]);
      g = GETJSAMPLE(inptr[RGB_GREEN]);
      b = GETJSAMPLE(inptr[RGB_BLUE]);
+      inptr += RGB_PIXELSIZE;
      /* Assume that MAXJSAMPLE+1 is a power of 2, so that the MOD
       * (modulo) operator is equivalent to the bitmask operator AND.
       */
      outptr0[col] = (JSAMPLE) ((r - g + CENTERJSAMPLE) & MAXJSAMPLE);
      outptr1[col] = (JSAMPLE) g;
      outptr2[col] = (JSAMPLE) ((b - g + CENTERJSAMPLE) & MAXJSAMPLE);
-      inptr += RGB_PIXELSIZE;
    }
  }
 }
@@ -335,17 +335,17 @@ grayscale_convert (j_compress_ptr cinfo,
 		   JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
 		   JDIMENSION output_row, int num_rows)
 {
-  int instride = cinfo->input_components;
  register JSAMPROW inptr;
  register JSAMPROW outptr;
-  register JDIMENSION col;
+  register JDIMENSION count;
+  register int instride = cinfo->input_components;
  JDIMENSION num_cols = cinfo->image_width;

  while (--num_rows >= 0) {
    inptr = *input_buf++;
    outptr = output_buf[0][output_row++];
-    for (col = 0; col < num_cols; col++) {
-      outptr[col] = inptr[0];	/* don't need GETJSAMPLE() here */
+    for (count = num_cols; count > 0; count--) {
+      *outptr++ = *inptr;	/* don't need GETJSAMPLE() here */
      inptr += instride;
    }
  }
@@ -396,21 +396,21 @@ null_convert (j_compress_ptr cinfo,
 	      JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
 	      JDIMENSION output_row, int num_rows)
 {
-  int ci;
-  register int nc = cinfo->num_components;
  register JSAMPROW inptr;
  register JSAMPROW outptr;
-  register JDIMENSION col;
+  register JDIMENSION count;
+  register int num_comps = cinfo->num_components;
  JDIMENSION num_cols = cinfo->image_width;
+  int ci;

  while (--num_rows >= 0) {
    /* It seems fastest to make a separate pass for each component. */
-    for (ci = 0; ci < nc; ci++) {
+    for (ci = 0; ci < num_comps; ci++) {
      inptr = input_buf[0] + ci;
      outptr = output_buf[ci][output_row];
-      for (col = 0; col < num_cols; col++) {
+      for (count = num_cols; count > 0; count--) {
 	*outptr++ = *inptr;	/* don't need GETJSAMPLE() here */
-	inptr += nc;
+	inptr += num_comps;
      }
    }
    input_buf++;
@@ -439,9 +439,8 @@ jinit_color_converter (j_compress_ptr cinfo)
 {
  my_cconvert_ptr cconvert;

-  cconvert = (my_cconvert_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_color_converter));
+  cconvert = (my_cconvert_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(my_color_converter));
  cinfo->cconvert = &cconvert->pub;
  /* set start_pass to null method until we find out differently */
  cconvert->pub.start_pass = null_method;
@@ -455,9 +454,11 @@ jinit_color_converter (j_compress_ptr cinfo)

  case JCS_RGB:
  case JCS_BG_RGB:
+#if RGB_PIXELSIZE != 3
    if (cinfo->input_components != RGB_PIXELSIZE)
      ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
    break;
+#endif /* else share code with YCbCr */

  case JCS_YCbCr:
  case JCS_BG_YCC:
@@ -474,7 +475,6 @@ jinit_color_converter (j_compress_ptr cinfo)
  default:			/* JCS_UNKNOWN can be anything */
    if (cinfo->input_components < 1)
      ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
-    break;
  }

  /* Support color transform only for RGB colorspaces */
@@ -507,19 +507,18 @@ jinit_color_converter (j_compress_ptr cinfo)
  case JCS_BG_RGB:
    if (cinfo->num_components != 3)
      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    if (cinfo->in_color_space == cinfo->jpeg_color_space) {
-      switch (cinfo->color_transform) {
-      case JCT_NONE:
-	cconvert->pub.color_convert = rgb_convert;
-	break;
-      case JCT_SUBTRACT_GREEN:
-	cconvert->pub.color_convert = rgb_rgb1_convert;
-	break;
-      default:
-	ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-      }
-    } else
+    if (cinfo->in_color_space != cinfo->jpeg_color_space)
      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
+    switch (cinfo->color_transform) {
+    case JCT_NONE:
+      cconvert->pub.color_convert = rgb_convert;
+      break;
+    case JCT_SUBTRACT_GREEN:
+      cconvert->pub.color_convert = rgb_rgb1_convert;
+      break;
+    default:
+      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
+    }
    break;

  case JCS_YCbCr:
@@ -572,10 +571,9 @@ jinit_color_converter (j_compress_ptr cinfo)
  case JCS_CMYK:
    if (cinfo->num_components != 4)
      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    if (cinfo->in_color_space == JCS_CMYK)
-      cconvert->pub.color_convert = null_convert;
-    else
+    if (cinfo->in_color_space != JCS_CMYK)
      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
+    cconvert->pub.color_convert = null_convert;
    break;

  case JCS_YCCK:
@@ -599,6 +597,5 @@ jinit_color_converter (j_compress_ptr cinfo)
 	cinfo->num_components != cinfo->input_components)
      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
    cconvert->pub.color_convert = null_convert;
-    break;
  }
 }
@@ -2,7 +2,7 @@
 * jchuff.c
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * Modified 2006-2013 by Guido Vollbeding.
+ * Modified 2006-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -178,13 +178,12 @@ jpeg_make_c_derived_tbl (j_compress_ptr cinfo, boolean isDC, int tblno,
  htbl =
    isDC ? cinfo->dc_huff_tbl_ptrs[tblno] : cinfo->ac_huff_tbl_ptrs[tblno];
  if (htbl == NULL)
-    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
+    htbl = jpeg_std_huff_table((j_common_ptr) cinfo, isDC, tblno);

  /* Allocate a workspace if we haven't already done so. */
  if (*pdtbl == NULL)
-    *pdtbl = (c_derived_tbl *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(c_derived_tbl));
+    *pdtbl = (c_derived_tbl *) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(c_derived_tbl));
  dtbl = *pdtbl;
  
  /* Figure C.1: make table of Huffman code length for each symbol */
@@ -1256,22 +1255,88 @@ jpeg_gen_optimal_table (j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[])
  UINT8 bits[MAX_CLEN+1];	/* bits[k] = # of symbols with code length k */
  int codesize[257];		/* codesize[k] = code length of symbol k */
  int others[257];		/* next symbol in current branch of tree */
-  int c1, c2;
-  int p, i, j;
+  int c1, c2, i, j;
+  UINT8 *p;
  long v;

+  freq[256] = 1;		/* make sure 256 has a nonzero count */
+  /* Including the pseudo-symbol 256 in the Huffman procedure guarantees
+   * that no real symbol is given code-value of all ones, because 256
+   * will be placed last in the largest codeword category.
+   * In the symbol list build procedure this element serves as sentinel
+   * for the zero run loop.
+   */
+
+#ifndef DONT_USE_FANCY_HUFF_OPT
+
+  /* Build list of symbols sorted in order of descending frequency */
+  /* This approach has several benefits (thank to John Korejwa for the idea):
+   *     1.
+   * If a codelength category is split during the length limiting procedure
+   * below, the feature that more frequent symbols are assigned shorter
+   * codewords remains valid for the adjusted code.
+   *     2.
+   * To reduce consecutive ones in a Huffman data stream (thus reducing the
+   * number of stuff bytes in JPEG) it is preferable to follow 0 branches
+   * (and avoid 1 branches) as much as possible.  This is easily done by
+   * assigning symbols to leaves of the Huffman tree in order of decreasing
+   * frequency, with no secondary sort based on codelengths.
+   *     3.
+   * The symbol list can be built independently from the assignment of code
+   * lengths by the Huffman procedure below.
+   * Note: The symbol list build procedure must be performed first, because
+   * the Huffman procedure assigning the codelengths clobbers the frequency
+   * counts!
+   */
+
+  /* Here we use the others array as a linked list of nonzero frequencies
+   * to be sorted.  Already sorted elements are removed from the list.
+   */
+
+  /* Building list */
+
+  /* This item does not correspond to a valid symbol frequency and is used
+   * as starting index.
+   */
+  j = 256;
+
+  for (i = 0;; i++) {
+    if (freq[i] == 0)		/* skip zero frequencies */
+      continue;
+    if (i > 255)
+      break;
+    others[j] = i;		/* this symbol value */
+    j = i;			/* previous symbol value */
+  }
+  others[j] = -1;		/* mark end of list */
+
+  /* Sorting list */
+
+  p = htbl->huffval;
+  while ((c1 = others[256]) >= 0) {
+    v = freq[c1];
+    i = c1;			/* first symbol value */
+    j = 256;			/* pseudo symbol value for starting index */
+    while ((c2 = others[c1]) >= 0) {
+      if (freq[c2] > v) {
+	v = freq[c2];
+	i = c2;			/* this symbol value */
+	j = c1;			/* previous symbol value */
+      }
+      c1 = c2;
+    }
+    others[j] = others[i];	/* remove this symbol i from list */
+    *p++ = (UINT8) i;
+  }
+
+#endif /* DONT_USE_FANCY_HUFF_OPT */
+
  /* This algorithm is explained in section K.2 of the JPEG standard */

  MEMZERO(bits, SIZEOF(bits));
  MEMZERO(codesize, SIZEOF(codesize));
  for (i = 0; i < 257; i++)
    others[i] = -1;		/* init links to empty */
-  
-  freq[256] = 1;		/* make sure 256 has a nonzero count */
-  /* Including the pseudo-symbol 256 in the Huffman procedure guarantees
-   * that no real symbol is given code-value of all ones, because 256
-   * will be placed last in the largest codeword category.
-   */

  /* Huffman's basic algorithm to assign optimal code lengths to symbols */

@@ -1301,7 +1366,7 @@ jpeg_gen_optimal_table (j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[])
    /* Done if we've merged everything into one frequency */
    if (c2 < 0)
      break;
-    
+
    /* Else merge the two counts/trees */
    freq[c1] += freq[c2];
    freq[c2] = 0;
@@ -1312,9 +1377,9 @@ jpeg_gen_optimal_table (j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[])
      c1 = others[c1];
      codesize[c1]++;
    }
-    
+
    others[c1] = c2;		/* chain c2 onto c1's tree branch */
-    
+
    /* Increment the codesize of everything in c2's tree branch */
    codesize[c2]++;
    while (others[c2] >= 0) {
@@ -1329,7 +1394,7 @@ jpeg_gen_optimal_table (j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[])
      /* The JPEG standard seems to think that this can't happen, */
      /* but I'm paranoid... */
      if (codesize[i] > MAX_CLEN)
-	ERREXIT(cinfo, JERR_HUFF_CLEN_OVERFLOW);
+	ERREXIT(cinfo, JERR_HUFF_CLEN_OUTOFBOUNDS);

      bits[codesize[i]]++;
    }
@@ -1345,13 +1410,16 @@ jpeg_gen_optimal_table (j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[])
   * shortest nonzero BITS entry is converted into a prefix for two code words
   * one bit longer.
   */
-  
+
  for (i = MAX_CLEN; i > 16; i--) {
    while (bits[i] > 0) {
      j = i - 2;		/* find length of new prefix to be used */
-      while (bits[j] == 0)
+      while (bits[j] == 0) {
+	if (j == 0)
+	  ERREXIT(cinfo, JERR_HUFF_CLEN_OUTOFBOUNDS);
 	j--;
-      
+      }
+
      bits[i] -= 2;		/* remove two symbols */
      bits[i-1]++;		/* one goes in this length */
      bits[j+1] += 2;		/* two new symbols in this length */
@@ -1363,24 +1431,27 @@ jpeg_gen_optimal_table (j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[])
  while (bits[i] == 0)		/* find largest codelength still in use */
    i--;
  bits[i]--;
-  
+
  /* Return final symbol counts (only for lengths 0..16) */
  MEMCOPY(htbl->bits, bits, SIZEOF(htbl->bits));
-  
+
+#ifdef DONT_USE_FANCY_HUFF_OPT
+
  /* Return a list of the symbols sorted by code length */
-  /* It's not real clear to me why we don't need to consider the codelength
-   * changes made above, but the JPEG spec seems to think this works.
+  /* Note: Due to the codelength changes made above, it can happen
+   * that more frequent symbols are assigned longer codewords.
   */
-  p = 0;
+  p = htbl->huffval;
  for (i = 1; i <= MAX_CLEN; i++) {
    for (j = 0; j <= 255; j++) {
      if (codesize[j] == i) {
-	htbl->huffval[p] = (UINT8) j;
-	p++;
+	*p++ = (UINT8) j;
      }
    }
  }

+#endif /* DONT_USE_FANCY_HUFF_OPT */
+
  /* Set sent_table FALSE so updated table will be written to JPEG file. */
  htbl->sent_table = FALSE;
 }
@@ -1400,13 +1471,13 @@ finish_pass_gather (j_compress_ptr cinfo)
  boolean did_dc[NUM_HUFF_TBLS];
  boolean did_ac[NUM_HUFF_TBLS];

-  /* It's important not to apply jpeg_gen_optimal_table more than once
-   * per table, because it clobbers the input frequency counts!
-   */
  if (cinfo->progressive_mode)
    /* Flush out buffered data (all we care about is counting the EOB symbol) */
    emit_eobrun(entropy);

+  /* It's important not to apply jpeg_gen_optimal_table more than once
+   * per table, because it clobbers the input frequency counts!
+   */
  MEMZERO(did_dc, SIZEOF(did_dc));
  MEMZERO(did_ac, SIZEOF(did_ac));

@@ -1475,9 +1546,8 @@ start_pass_huff (j_compress_ptr cinfo, boolean gather_statistics)
 	entropy->pub.encode_mcu = encode_mcu_AC_refine;
 	/* AC refinement needs a correction bit buffer */
 	if (entropy->bit_buffer == NULL)
-	  entropy->bit_buffer = (char *)
-	    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-					MAX_CORR_BITS * SIZEOF(char));
+	  entropy->bit_buffer = (char *) (*cinfo->mem->alloc_small)
+	    ((j_common_ptr) cinfo, JPOOL_IMAGE, MAX_CORR_BITS * SIZEOF(char));
      }
    }

@@ -1505,9 +1575,8 @@ start_pass_huff (j_compress_ptr cinfo, boolean gather_statistics)
 	/* Allocate and zero the statistics tables */
 	/* Note that jpeg_gen_optimal_table expects 257 entries in each table! */
 	if (entropy->dc_count_ptrs[tbl] == NULL)
-	  entropy->dc_count_ptrs[tbl] = (long *)
-	    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-					257 * SIZEOF(long));
+	  entropy->dc_count_ptrs[tbl] = (long *) (*cinfo->mem->alloc_small)
+	    ((j_common_ptr) cinfo, JPOOL_IMAGE, 257 * SIZEOF(long));
 	MEMZERO(entropy->dc_count_ptrs[tbl], 257 * SIZEOF(long));
      } else {
 	/* Compute derived values for Huffman tables */
@@ -1525,9 +1594,8 @@ start_pass_huff (j_compress_ptr cinfo, boolean gather_statistics)
 	if (tbl < 0 || tbl >= NUM_HUFF_TBLS)
 	  ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tbl);
 	if (entropy->ac_count_ptrs[tbl] == NULL)
-	  entropy->ac_count_ptrs[tbl] = (long *)
-	    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-					257 * SIZEOF(long));
+	  entropy->ac_count_ptrs[tbl] = (long *) (*cinfo->mem->alloc_small)
+	    ((j_common_ptr) cinfo, JPOOL_IMAGE, 257 * SIZEOF(long));
 	MEMZERO(entropy->ac_count_ptrs[tbl], 257 * SIZEOF(long));
      } else {
 	jpeg_make_c_derived_tbl(cinfo, FALSE, tbl,
@@ -1556,9 +1624,8 @@ jinit_huff_encoder (j_compress_ptr cinfo)
  huff_entropy_ptr entropy;
  int i;

-  entropy = (huff_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(huff_entropy_encoder));
+  entropy = (huff_entropy_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(huff_entropy_encoder));
  cinfo->entropy = &entropy->pub;
  entropy->pub.start_pass = start_pass_huff;

@@ -2,7 +2,7 @@
 * jcmarker.c
 *
 * Copyright (C) 1991-1998, Thomas G. Lane.
- * Modified 2003-2013 by Guido Vollbeding.
+ * Modified 2003-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -471,7 +471,6 @@ emit_adobe_app14 (j_compress_ptr cinfo)
    break;
  default:
    emit_byte(cinfo, 0);	/* Color transform = 0 */
-    break;
  }
 }

@@ -702,9 +701,8 @@ jinit_marker_writer (j_compress_ptr cinfo)
  my_marker_ptr marker;

  /* Create the subobject */
-  marker = (my_marker_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_marker_writer));
+  marker = (my_marker_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(my_marker_writer));
  cinfo->marker = &marker->pub;
  /* Initialize method pointers */
  marker->pub.write_file_header = write_file_header;
@@ -2,7 +2,7 @@
 * jcmaster.c
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * Modified 2003-2017 by Guido Vollbeding.
+ * Modified 2003-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -62,7 +62,7 @@ initial_setup (j_compress_ptr cinfo)
  case 5: cinfo->natural_order = jpeg_natural_order5; break;
  case 6: cinfo->natural_order = jpeg_natural_order6; break;
  case 7: cinfo->natural_order = jpeg_natural_order7; break;
-  default: cinfo->natural_order = jpeg_natural_order; break;
+  default: cinfo->natural_order = jpeg_natural_order;
  }

  /* Derive lim_Se from block_size */
@@ -114,20 +114,24 @@ initial_setup (j_compress_ptr cinfo)
     */
    ssize = 1;
 #ifdef DCT_SCALING_SUPPORTED
-    while (cinfo->min_DCT_h_scaled_size * ssize <=
-	   (cinfo->do_fancy_downsampling ? DCTSIZE : DCTSIZE / 2) &&
-	   (cinfo->max_h_samp_factor % (compptr->h_samp_factor * ssize * 2)) == 0) {
-      ssize = ssize * 2;
-    }
+    if (! cinfo->raw_data_in)
+      while (cinfo->min_DCT_h_scaled_size * ssize <=
+	     (cinfo->do_fancy_downsampling ? DCTSIZE : DCTSIZE / 2) &&
+	     (cinfo->max_h_samp_factor % (compptr->h_samp_factor * ssize * 2)) ==
+	     0) {
+	ssize = ssize * 2;
+      }
 #endif
    compptr->DCT_h_scaled_size = cinfo->min_DCT_h_scaled_size * ssize;
    ssize = 1;
 #ifdef DCT_SCALING_SUPPORTED
-    while (cinfo->min_DCT_v_scaled_size * ssize <=
-	   (cinfo->do_fancy_downsampling ? DCTSIZE : DCTSIZE / 2) &&
-	   (cinfo->max_v_samp_factor % (compptr->v_samp_factor * ssize * 2)) == 0) {
-      ssize = ssize * 2;
-    }
+    if (! cinfo->raw_data_in)
+      while (cinfo->min_DCT_v_scaled_size * ssize <=
+	     (cinfo->do_fancy_downsampling ? DCTSIZE : DCTSIZE / 2) &&
+	     (cinfo->max_v_samp_factor % (compptr->v_samp_factor * ssize * 2)) ==
+	     0) {
+	ssize = ssize * 2;
+      }
 #endif
    compptr->DCT_v_scaled_size = cinfo->min_DCT_v_scaled_size * ssize;

@@ -620,9 +624,8 @@ jinit_c_master_control (j_compress_ptr cinfo, boolean transcode_only)
 {
  my_master_ptr master;

-  master = (my_master_ptr)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(my_comp_master));
+  master = (my_master_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(my_comp_master));
  cinfo->master = &master->pub;
  master->pub.prepare_for_pass = prepare_for_pass;
  master->pub.pass_startup = pass_startup;
@@ -2,6 +2,7 @@
 * jcomapi.c
 *
 * Copyright (C) 1994-1997, Thomas G. Lane.
+ * Modified 2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -104,3 +105,140 @@ jpeg_alloc_huff_table (j_common_ptr cinfo)
  tbl->sent_table = FALSE;	/* make sure this is false in any new table */
  return tbl;
 }
+
+
+/*
+ * Set up the standard Huffman tables (cf. JPEG standard section K.3).
+ * IMPORTANT: these are only valid for 8-bit data precision!
+ * (Would jutils.c be a more reasonable place to put this?)
+ */
+
+GLOBAL(JHUFF_TBL *)
+jpeg_std_huff_table (j_common_ptr cinfo, boolean isDC, int tblno)
+{
+  JHUFF_TBL **htblptr, *htbl;
+  const UINT8 *bits, *val;
+  int nsymbols, len;
+
+  static const UINT8 bits_dc_luminance[17] =
+    { /* 0-base */ 0, 0, 1, 5, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0 };
+  static const UINT8 val_dc_luminance[] =
+    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };
+
+  static const UINT8 bits_dc_chrominance[17] =
+    { /* 0-base */ 0, 0, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 };
+  static const UINT8 val_dc_chrominance[] =
+    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };
+
+  static const UINT8 bits_ac_luminance[17] =
+    { /* 0-base */ 0, 0, 2, 1, 3, 3, 2, 4, 3, 5, 5, 4, 4, 0, 0, 1, 0x7d };
+  static const UINT8 val_ac_luminance[] =
+    { 0x01, 0x02, 0x03, 0x00, 0x04, 0x11, 0x05, 0x12,
+      0x21, 0x31, 0x41, 0x06, 0x13, 0x51, 0x61, 0x07,
+      0x22, 0x71, 0x14, 0x32, 0x81, 0x91, 0xa1, 0x08,
+      0x23, 0x42, 0xb1, 0xc1, 0x15, 0x52, 0xd1, 0xf0,
+      0x24, 0x33, 0x62, 0x72, 0x82, 0x09, 0x0a, 0x16,
+      0x17, 0x18, 0x19, 0x1a, 0x25, 0x26, 0x27, 0x28,
+      0x29, 0x2a, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39,
+      0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49,
+      0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59,
+      0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69,
+      0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79,
+      0x7a, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89,
+      0x8a, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98,
+      0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
+      0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6,
+      0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3, 0xc4, 0xc5,
+      0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2, 0xd3, 0xd4,
+      0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xe1, 0xe2,
+      0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea,
+      0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
+      0xf9, 0xfa };
+
+  static const UINT8 bits_ac_chrominance[17] =
+    { /* 0-base */ 0, 0, 2, 1, 2, 4, 4, 3, 4, 7, 5, 4, 4, 0, 1, 2, 0x77 };
+  static const UINT8 val_ac_chrominance[] =
+    { 0x00, 0x01, 0x02, 0x03, 0x11, 0x04, 0x05, 0x21,
+      0x31, 0x06, 0x12, 0x41, 0x51, 0x07, 0x61, 0x71,
+      0x13, 0x22, 0x32, 0x81, 0x08, 0x14, 0x42, 0x91,
+      0xa1, 0xb1, 0xc1, 0x09, 0x23, 0x33, 0x52, 0xf0,
+      0x15, 0x62, 0x72, 0xd1, 0x0a, 0x16, 0x24, 0x34,
+      0xe1, 0x25, 0xf1, 0x17, 0x18, 0x19, 0x1a, 0x26,
+      0x27, 0x28, 0x29, 0x2a, 0x35, 0x36, 0x37, 0x38,
+      0x39, 0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48,
+      0x49, 0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58,
+      0x59, 0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
+      0x69, 0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78,
+      0x79, 0x7a, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
+      0x88, 0x89, 0x8a, 0x92, 0x93, 0x94, 0x95, 0x96,
+      0x97, 0x98, 0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5,
+      0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4,
+      0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3,
+      0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2,
+      0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda,
+      0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9,
+      0xea, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
+      0xf9, 0xfa };
+
+  if (cinfo->is_decompressor) {
+    if (isDC)
+      htblptr = ((j_decompress_ptr) cinfo)->dc_huff_tbl_ptrs;
+    else
+      htblptr = ((j_decompress_ptr) cinfo)->ac_huff_tbl_ptrs;
+  } else {
+    if (isDC)
+      htblptr = ((j_compress_ptr) cinfo)->dc_huff_tbl_ptrs;
+    else
+      htblptr = ((j_compress_ptr) cinfo)->ac_huff_tbl_ptrs;
+  }
+
+  switch (tblno) {
+  case 0:
+    if (isDC) {
+      bits = bits_dc_luminance;
+      val = val_dc_luminance;
+    } else {
+      bits = bits_ac_luminance;
+      val = val_ac_luminance;
+    }
+    break;
+  case 1:
+    if (isDC) {
+      bits = bits_dc_chrominance;
+      val = val_dc_chrominance;
+    } else {
+      bits = bits_ac_chrominance;
+      val = val_ac_chrominance;
+    }
+    break;
+  default:
+    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
+    return NULL; /* avoid compiler warnings for uninitialized variables */
+  }
+
+  if (htblptr[tblno] == NULL)
+    htblptr[tblno] = jpeg_alloc_huff_table(cinfo);
+
+  htbl = htblptr[tblno];
+
+  /* Copy the number-of-symbols-of-each-code-length counts */
+  MEMCOPY(htbl->bits, bits, SIZEOF(htbl->bits));
+
+  /* Validate the counts.  We do this here mainly so we can copy the right
+   * number of symbols from the val[] array, without risking marching off
+   * the end of memory.  jxhuff.c will do a more thorough test later.
+   */
+  nsymbols = 0;
+  for (len = 1; len <= 16; len++)
+    nsymbols += bits[len];
+  if (nsymbols > 256)
+    ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
+
+  if (nsymbols > 0)
+    MEMCOPY(htbl->huffval, val, nsymbols * SIZEOF(UINT8));
+
+  /* Initialize sent_table FALSE so table will be written to JPEG file. */
+  htbl->sent_table = FALSE;
+
+  return htbl;
+}
@@ -2,7 +2,7 @@
 * jcparam.c
 *
 * Copyright (C) 1991-1998, Thomas G. Lane.
- * Modified 2003-2013 by Guido Vollbeding.
+ * Modified 2003-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -162,112 +162,23 @@ jpeg_set_quality (j_compress_ptr cinfo, int quality, boolean force_baseline)


 /*
- * Huffman table setup routines
+ * Reset standard Huffman tables
 */

-LOCAL(void)
-add_huff_table (j_compress_ptr cinfo,
-		JHUFF_TBL **htblptr, const UINT8 *bits, const UINT8 *val)
-/* Define a Huffman table */
-{
-  int nsymbols, len;
-
-  if (*htblptr == NULL)
-    *htblptr = jpeg_alloc_huff_table((j_common_ptr) cinfo);
-
-  /* Copy the number-of-symbols-of-each-code-length counts */
-  MEMCOPY((*htblptr)->bits, bits, SIZEOF((*htblptr)->bits));
-
-  /* Validate the counts.  We do this here mainly so we can copy the right
-   * number of symbols from the val[] array, without risking marching off
-   * the end of memory.  jchuff.c will do a more thorough test later.
-   */
-  nsymbols = 0;
-  for (len = 1; len <= 16; len++)
-    nsymbols += bits[len];
-  if (nsymbols < 1 || nsymbols > 256)
-    ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-
-  MEMCOPY((*htblptr)->huffval, val, nsymbols * SIZEOF(UINT8));
-
-  /* Initialize sent_table FALSE so table will be written to JPEG file. */
-  (*htblptr)->sent_table = FALSE;
-}
-
-
 LOCAL(void)
 std_huff_tables (j_compress_ptr cinfo)
-/* Set up the standard Huffman tables (cf. JPEG standard section K.3) */
-/* IMPORTANT: these are only valid for 8-bit data precision! */
 {
-  static const UINT8 bits_dc_luminance[17] =
-    { /* 0-base */ 0, 0, 1, 5, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0 };
-  static const UINT8 val_dc_luminance[] =
-    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };
-  
-  static const UINT8 bits_dc_chrominance[17] =
-    { /* 0-base */ 0, 0, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 };
-  static const UINT8 val_dc_chrominance[] =
-    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };
-  
-  static const UINT8 bits_ac_luminance[17] =
-    { /* 0-base */ 0, 0, 2, 1, 3, 3, 2, 4, 3, 5, 5, 4, 4, 0, 0, 1, 0x7d };
-  static const UINT8 val_ac_luminance[] =
-    { 0x01, 0x02, 0x03, 0x00, 0x04, 0x11, 0x05, 0x12,
-      0x21, 0x31, 0x41, 0x06, 0x13, 0x51, 0x61, 0x07,
-      0x22, 0x71, 0x14, 0x32, 0x81, 0x91, 0xa1, 0x08,
-      0x23, 0x42, 0xb1, 0xc1, 0x15, 0x52, 0xd1, 0xf0,
-      0x24, 0x33, 0x62, 0x72, 0x82, 0x09, 0x0a, 0x16,
-      0x17, 0x18, 0x19, 0x1a, 0x25, 0x26, 0x27, 0x28,
-      0x29, 0x2a, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39,
-      0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49,
-      0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59,
-      0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69,
-      0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79,
-      0x7a, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89,
-      0x8a, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98,
-      0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
-      0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6,
-      0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3, 0xc4, 0xc5,
-      0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2, 0xd3, 0xd4,
-      0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xe1, 0xe2,
-      0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea,
-      0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
-      0xf9, 0xfa };
-  
-  static const UINT8 bits_ac_chrominance[17] =
-    { /* 0-base */ 0, 0, 2, 1, 2, 4, 4, 3, 4, 7, 5, 4, 4, 0, 1, 2, 0x77 };
-  static const UINT8 val_ac_chrominance[] =
-    { 0x00, 0x01, 0x02, 0x03, 0x11, 0x04, 0x05, 0x21,
-      0x31, 0x06, 0x12, 0x41, 0x51, 0x07, 0x61, 0x71,
-      0x13, 0x22, 0x32, 0x81, 0x08, 0x14, 0x42, 0x91,
-      0xa1, 0xb1, 0xc1, 0x09, 0x23, 0x33, 0x52, 0xf0,
-      0x15, 0x62, 0x72, 0xd1, 0x0a, 0x16, 0x24, 0x34,
-      0xe1, 0x25, 0xf1, 0x17, 0x18, 0x19, 0x1a, 0x26,
-      0x27, 0x28, 0x29, 0x2a, 0x35, 0x36, 0x37, 0x38,
-      0x39, 0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48,
-      0x49, 0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58,
-      0x59, 0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
-      0x69, 0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78,
-      0x79, 0x7a, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
-      0x88, 0x89, 0x8a, 0x92, 0x93, 0x94, 0x95, 0x96,
-      0x97, 0x98, 0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5,
-      0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4,
-      0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3,
-      0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2,
-      0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda,
-      0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9,
-      0xea, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
-      0xf9, 0xfa };
-  
-  add_huff_table(cinfo, &cinfo->dc_huff_tbl_ptrs[0],
-		 bits_dc_luminance, val_dc_luminance);
-  add_huff_table(cinfo, &cinfo->ac_huff_tbl_ptrs[0],
-		 bits_ac_luminance, val_ac_luminance);
-  add_huff_table(cinfo, &cinfo->dc_huff_tbl_ptrs[1],
-		 bits_dc_chrominance, val_dc_chrominance);
-  add_huff_table(cinfo, &cinfo->ac_huff_tbl_ptrs[1],
-		 bits_ac_chrominance, val_ac_chrominance);
+  if (cinfo->dc_huff_tbl_ptrs[0] != NULL)
+    (void) jpeg_std_huff_table((j_common_ptr) cinfo, TRUE, 0);
+
+  if (cinfo->ac_huff_tbl_ptrs[0] != NULL)
+    (void) jpeg_std_huff_table((j_common_ptr) cinfo, FALSE, 0);
+
+  if (cinfo->dc_huff_tbl_ptrs[1] != NULL)
+    (void) jpeg_std_huff_table((j_common_ptr) cinfo, TRUE, 1);
+
+  if (cinfo->ac_huff_tbl_ptrs[1] != NULL)
+    (void) jpeg_std_huff_table((j_common_ptr) cinfo, FALSE, 1);
 }


@@ -306,7 +217,7 @@ jpeg_set_defaults (j_compress_ptr cinfo)
  cinfo->data_precision = BITS_IN_JSAMPLE;
  /* Set up two quantization tables using default quality of 75 */
  jpeg_set_quality(cinfo, 75, TRUE);
-  /* Set up two Huffman tables */
+  /* Reset standard Huffman tables */
  std_huff_tables(cinfo);

  /* Initialize default arithmetic coding conditioning */
@@ -1,7 +1,7 @@
 /*
 * jdarith.c
 *
- * Developed 1997-2015 by Guido Vollbeding.
+ * Developed 1997-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -280,7 +280,7 @@ decode_mcu_DC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
      if ((m = arith_decode(cinfo, st)) != 0) {
 	st = entropy->dc_stats[tbl] + 20;	/* Table F.4: X1 = 20 */
 	while (arith_decode(cinfo, st)) {
-	  if ((m <<= 1) == 0x8000) {
+	  if ((m <<= 1) == (int) 0x8000U) {
 	    WARNMS(cinfo, JWRN_ARITH_BAD_CODE);
 	    entropy->ct = -1;			/* magnitude overflow */
 	    return TRUE;
@@ -370,7 +370,7 @@ decode_mcu_AC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
 	st = entropy->ac_stats[tbl] +
 	     (k <= cinfo->arith_ac_K[tbl] ? 189 : 217);
 	while (arith_decode(cinfo, st)) {
-	  if ((m <<= 1) == 0x8000) {
+	  if ((m <<= 1) == (int) 0x8000U) {
 	    WARNMS(cinfo, JWRN_ARITH_BAD_CODE);
 	    entropy->ct = -1;			/* magnitude overflow */
 	    return TRUE;
@@ -404,7 +404,8 @@ decode_mcu_DC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
 {
  arith_entropy_ptr entropy = (arith_entropy_ptr) cinfo->entropy;
  unsigned char *st;
-  int p1, blkn;
+  JCOEF p1;
+  int blkn;

  /* Process restart marker if needed */
  if (cinfo->restart_interval) {
@@ -440,7 +441,7 @@ decode_mcu_AC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
  JCOEFPTR thiscoef;
  unsigned char *st;
  int tbl, k, kex;
-  int p1, m1;
+  JCOEF p1, m1;
  const int * natural_order;

  /* Process restart marker if needed */
@@ -459,7 +460,7 @@ decode_mcu_AC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
  tbl = cinfo->cur_comp_info[0]->ac_tbl_no;

  p1 = 1 << cinfo->Al;		/* 1 in the bit position being coded */
-  m1 = (-1) << cinfo->Al;	/* -1 in the bit position being coded */
+  m1 = -p1;			/* -1 in the bit position being coded */

  /* Establish EOBx (previous stage end-of-block) index */
  kex = cinfo->Se;
@@ -555,7 +556,7 @@ decode_mcu (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
      if ((m = arith_decode(cinfo, st)) != 0) {
 	st = entropy->dc_stats[tbl] + 20;	/* Table F.4: X1 = 20 */
 	while (arith_decode(cinfo, st)) {
-	  if ((m <<= 1) == 0x8000) {
+	  if ((m <<= 1) == (int) 0x8000U) {
 	    WARNMS(cinfo, JWRN_ARITH_BAD_CODE);
 	    entropy->ct = -1;			/* magnitude overflow */
 	    return TRUE;
@@ -612,7 +613,7 @@ decode_mcu (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
 	  st = entropy->ac_stats[tbl] +
 	       (k <= cinfo->arith_ac_K[tbl] ? 189 : 217);
 	  while (arith_decode(cinfo, st)) {
-	    if ((m <<= 1) == 0x8000) {
+	    if ((m <<= 1) == (int) 0x8000U) {
 	      WARNMS(cinfo, JWRN_ARITH_BAD_CODE);
 	      entropy->ct = -1;			/* magnitude overflow */
 	      return TRUE;
@@ -766,9 +767,8 @@ jinit_arith_decoder (j_decompress_ptr cinfo)
  arith_entropy_ptr entropy;
  int i;

-  entropy = (arith_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(arith_entropy_decoder));
+  entropy = (arith_entropy_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(arith_entropy_decoder));
  cinfo->entropy = &entropy->pub;
  entropy->pub.start_pass = start_pass;
  entropy->pub.finish_pass = finish_pass;
@@ -785,9 +785,9 @@ jinit_arith_decoder (j_decompress_ptr cinfo)
  if (cinfo->progressive_mode) {
    /* Create progression status table */
    int *coef_bit_ptr, ci;
-    cinfo->coef_bits = (int (*)[DCTSIZE2])
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  cinfo->num_components*DCTSIZE2*SIZEOF(int));
+    cinfo->coef_bits = (int (*)[DCTSIZE2]) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_IMAGE,
+       cinfo->num_components * DCTSIZE2 * SIZEOF(int));
    coef_bit_ptr = & cinfo->coef_bits[0][0];
    for (ci = 0; ci < cinfo->num_components; ci++) 
      for (i = 0; i < DCTSIZE2; i++)
@@ -2,7 +2,7 @@
 * jdatadst.c
 *
 * Copyright (C) 1994-1996, Thomas G. Lane.
- * Modified 2009-2017 by Guido Vollbeding.
+ * Modified 2009-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -46,7 +46,7 @@ typedef struct {
  struct jpeg_destination_mgr pub; /* public fields */

  unsigned char ** outbuffer;	/* target buffer */
-  unsigned long * outsize;
+  size_t * outsize;
  unsigned char * newbuffer;	/* newly allocated buffer */
  JOCTET * buffer;		/* start of buffer */
  size_t bufsize;
@@ -66,9 +66,8 @@ init_destination (j_compress_ptr cinfo)
  my_dest_ptr dest = (my_dest_ptr) cinfo->dest;

  /* Allocate the output buffer --- it will be released when done with image */
-  dest->buffer = (JOCTET *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  OUTPUT_BUF_SIZE * SIZEOF(JOCTET));
+  dest->buffer = (JOCTET *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, OUTPUT_BUF_SIZE * SIZEOF(JOCTET));

  dest->pub.next_output_byte = dest->buffer;
  dest->pub.free_in_buffer = OUTPUT_BUF_SIZE;
@@ -131,7 +130,7 @@ empty_mem_output_buffer (j_compress_ptr cinfo)
  nextbuffer = (JOCTET *) malloc(nextsize);

  if (nextbuffer == NULL)
-    ERREXIT1(cinfo, JERR_OUT_OF_MEMORY, 10);
+    ERREXIT1(cinfo, JERR_OUT_OF_MEMORY, 11);

  MEMCOPY(nextbuffer, dest->buffer, dest->bufsize);

@@ -204,9 +203,8 @@ jpeg_stdio_dest (j_compress_ptr cinfo, FILE * outfile)
   * sizes may be different.  Caveat programmer.
   */
  if (cinfo->dest == NULL) {	/* first time for this JPEG object? */
-    cinfo->dest = (struct jpeg_destination_mgr *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  SIZEOF(my_destination_mgr));
+    cinfo->dest = (struct jpeg_destination_mgr *) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_PERMANENT, SIZEOF(my_destination_mgr));
  }

  dest = (my_dest_ptr) cinfo->dest;
@@ -233,7 +231,7 @@ jpeg_stdio_dest (j_compress_ptr cinfo, FILE * outfile)

 GLOBAL(void)
 jpeg_mem_dest (j_compress_ptr cinfo,
-	       unsigned char ** outbuffer, unsigned long * outsize)
+	       unsigned char ** outbuffer, size_t * outsize)
 {
  my_mem_dest_ptr dest;

@@ -244,9 +242,8 @@ jpeg_mem_dest (j_compress_ptr cinfo,
   * can be written to the same buffer without re-executing jpeg_mem_dest.
   */
  if (cinfo->dest == NULL) {	/* first time for this JPEG object? */
-    cinfo->dest = (struct jpeg_destination_mgr *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  SIZEOF(my_mem_destination_mgr));
+    cinfo->dest = (struct jpeg_destination_mgr *) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_PERMANENT, SIZEOF(my_mem_destination_mgr));
  }

  dest = (my_mem_dest_ptr) cinfo->dest;
@@ -2,7 +2,7 @@
 * jdatasrc.c
 *
 * Copyright (C) 1994-1996, Thomas G. Lane.
- * Modified 2009-2015 by Guido Vollbeding.
+ * Modified 2009-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -156,21 +156,23 @@ METHODDEF(void)
 skip_input_data (j_decompress_ptr cinfo, long num_bytes)
 {
  struct jpeg_source_mgr * src = cinfo->src;
+  size_t nbytes;

  /* Just a dumb implementation for now.  Could use fseek() except
   * it doesn't work on pipes.  Not clear that being smart is worth
   * any trouble anyway --- large skips are infrequent.
   */
  if (num_bytes > 0) {
-    while (num_bytes > (long) src->bytes_in_buffer) {
-      num_bytes -= (long) src->bytes_in_buffer;
+    nbytes = (size_t) num_bytes;
+    while (nbytes > src->bytes_in_buffer) {
+      nbytes -= src->bytes_in_buffer;
      (void) (*src->fill_input_buffer) (cinfo);
      /* note we assume that fill_input_buffer will never return FALSE,
       * so suspension need not be handled.
       */
    }
-    src->next_input_byte += (size_t) num_bytes;
-    src->bytes_in_buffer -= (size_t) num_bytes;
+    src->next_input_byte += nbytes;
+    src->bytes_in_buffer -= nbytes;
  }
 }

@@ -219,13 +221,11 @@ jpeg_stdio_src (j_decompress_ptr cinfo, FILE * infile)
   * manager serially with the same JPEG object.  Caveat programmer.
   */
  if (cinfo->src == NULL) {	/* first time for this JPEG object? */
-    cinfo->src = (struct jpeg_source_mgr *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  SIZEOF(my_source_mgr));
+    cinfo->src = (struct jpeg_source_mgr *) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_PERMANENT, SIZEOF(my_source_mgr));
    src = (my_src_ptr) cinfo->src;
-    src->buffer = (JOCTET *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  INPUT_BUF_SIZE * SIZEOF(JOCTET));
+    src->buffer = (JOCTET *) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_PERMANENT, INPUT_BUF_SIZE * SIZEOF(JOCTET));
  }

  src = (my_src_ptr) cinfo->src;
@@ -247,7 +247,7 @@ jpeg_stdio_src (j_decompress_ptr cinfo, FILE * infile)

 GLOBAL(void)
 jpeg_mem_src (j_decompress_ptr cinfo,
-	      const unsigned char * inbuffer, unsigned long insize)
+	      const unsigned char * inbuffer, size_t insize)
 {
  struct jpeg_source_mgr * src;

@@ -259,9 +259,8 @@ jpeg_mem_src (j_decompress_ptr cinfo,
   * the first one.
   */
  if (cinfo->src == NULL) {	/* first time for this JPEG object? */
-    cinfo->src = (struct jpeg_source_mgr *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  SIZEOF(struct jpeg_source_mgr));
+    cinfo->src = (struct jpeg_source_mgr *) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_PERMANENT, SIZEOF(struct jpeg_source_mgr));
  }

  src = cinfo->src;
@@ -270,6 +269,6 @@ jpeg_mem_src (j_decompress_ptr cinfo,
  src->skip_input_data = skip_input_data;
  src->resync_to_restart = jpeg_resync_to_restart; /* use default method */
  src->term_source = term_source;
-  src->bytes_in_buffer = (size_t) insize;
+  src->bytes_in_buffer = insize;
  src->next_input_byte = (const JOCTET *) inbuffer;
 }
@@ -2,7 +2,7 @@
 * jdcolor.c
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * Modified 2011-2017 by Guido Vollbeding.
+ * Modified 2011-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -124,28 +124,22 @@ build_ycc_rgb_table (j_decompress_ptr cinfo)
  INT32 x;
  SHIFT_TEMPS

-  cconvert->Cr_r_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  cconvert->Cb_b_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  cconvert->Cr_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
-  cconvert->Cb_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
+  cconvert->Cr_r_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  cconvert->Cb_b_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  cconvert->Cr_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));
+  cconvert->Cb_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));

  for (i = 0, x = -CENTERJSAMPLE; i <= MAXJSAMPLE; i++, x++) {
    /* i is the actual input pixel value, in the range 0..MAXJSAMPLE */
    /* The Cb or Cr value we are thinking of is x = i - CENTERJSAMPLE */
    /* Cr=>R value is nearest int to 1.402 * x */
-    cconvert->Cr_r_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(1.402) * x + ONE_HALF, SCALEBITS);
+    cconvert->Cr_r_tab[i] = (int) DESCALE(FIX(1.402) * x, SCALEBITS);
    /* Cb=>B value is nearest int to 1.772 * x */
-    cconvert->Cb_b_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(1.772) * x + ONE_HALF, SCALEBITS);
+    cconvert->Cb_b_tab[i] = (int) DESCALE(FIX(1.772) * x, SCALEBITS);
    /* Cr=>G value is scaled-up -0.714136286 * x */
    cconvert->Cr_g_tab[i] = (- FIX(0.714136286)) * x;
    /* Cb=>G value is scaled-up -0.344136286 * x */
@@ -164,28 +158,22 @@ build_bg_ycc_rgb_table (j_decompress_ptr cinfo)
  INT32 x;
  SHIFT_TEMPS

-  cconvert->Cr_r_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  cconvert->Cb_b_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  cconvert->Cr_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
-  cconvert->Cb_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
+  cconvert->Cr_r_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  cconvert->Cb_b_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  cconvert->Cr_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));
+  cconvert->Cb_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));

  for (i = 0, x = -CENTERJSAMPLE; i <= MAXJSAMPLE; i++, x++) {
    /* i is the actual input pixel value, in the range 0..MAXJSAMPLE */
    /* The Cb or Cr value we are thinking of is x = i - CENTERJSAMPLE */
    /* Cr=>R value is nearest int to 2.804 * x */
-    cconvert->Cr_r_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(2.804) * x + ONE_HALF, SCALEBITS);
+    cconvert->Cr_r_tab[i] = (int) DESCALE(FIX(2.804) * x, SCALEBITS);
    /* Cb=>B value is nearest int to 3.544 * x */
-    cconvert->Cb_b_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(3.544) * x + ONE_HALF, SCALEBITS);
+    cconvert->Cb_b_tab[i] = (int) DESCALE(FIX(3.544) * x, SCALEBITS);
    /* Cr=>G value is scaled-up -1.428272572 * x */
    cconvert->Cr_g_tab[i] = (- FIX(1.428272572)) * x;
    /* Cb=>G value is scaled-up -0.688272572 * x */
@@ -201,6 +189,7 @@ build_bg_ycc_rgb_table (j_decompress_ptr cinfo)
 * Note that we change from noninterleaved, one-plane-per-component format
 * to interleaved-pixel format.  The output buffer is therefore three times
 * as wide as the input buffer.
+ *
 * A starting row offset is provided only for the input buffer.  The caller
 * can easily adjust the passed output_buf value to accommodate any row
 * offset required on that side.
@@ -264,9 +253,8 @@ build_rgb_y_table (j_decompress_ptr cinfo)
  INT32 i;

  /* Allocate and fill in the conversion tables. */
-  cconvert->rgb_y_tab = rgb_y_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(TABLE_SIZE * SIZEOF(INT32)));
+  cconvert->rgb_y_tab = rgb_y_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, TABLE_SIZE * SIZEOF(INT32));

  for (i = 0; i <= MAXJSAMPLE; i++) {
    rgb_y_tab[i+R_Y_OFF] = FIX(0.299) * i;
@@ -286,8 +274,8 @@ rgb_gray_convert (j_decompress_ptr cinfo,
 		  JSAMPARRAY output_buf, int num_rows)
 {
  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register INT32 * ctab = cconvert->rgb_y_tab;
  register int r, g, b;
+  register INT32 * ctab = cconvert->rgb_y_tab;
  register JSAMPROW outptr;
  register JSAMPROW inptr0, inptr1, inptr2;
  register JDIMENSION col;
@@ -313,6 +301,7 @@ rgb_gray_convert (j_decompress_ptr cinfo,


 /*
+ * Convert some rows of samples to the output colorspace.
 * [R-G,G,B-G] to [R,G,B] conversion with modulo calculation
 * (inverse color transform).
 * This can be seen as an adaption of the general YCbCr->RGB
@@ -364,8 +353,8 @@ rgb1_gray_convert (j_decompress_ptr cinfo,
 		   JSAMPARRAY output_buf, int num_rows)
 {
  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register INT32 * ctab = cconvert->rgb_y_tab;
  register int r, g, b;
+  register INT32 * ctab = cconvert->rgb_y_tab;
  register JSAMPROW outptr;
  register JSAMPROW inptr0, inptr1, inptr2;
  register JDIMENSION col;
@@ -396,6 +385,7 @@ rgb1_gray_convert (j_decompress_ptr cinfo,


 /*
+ * Convert some rows of samples to the output colorspace.
 * No colorspace change, but conversion from separate-planes
 * to interleaved representation.
 */
@@ -430,6 +420,7 @@ rgb_convert (j_decompress_ptr cinfo,
 /*
 * Color conversion for no colorspace change: just copy the data,
 * converting from separate-planes to interleaved representation.
+ * We assume out_color_components == num_components.
 */

 METHODDEF(void)
@@ -437,20 +428,21 @@ null_convert (j_decompress_ptr cinfo,
 	      JSAMPIMAGE input_buf, JDIMENSION input_row,
 	      JSAMPARRAY output_buf, int num_rows)
 {
-  int ci;
-  register int nc = cinfo->num_components;
  register JSAMPROW outptr;
  register JSAMPROW inptr;
-  register JDIMENSION col;
+  register JDIMENSION count;
+  register int num_comps = cinfo->num_components;
  JDIMENSION num_cols = cinfo->output_width;
+  int ci;

  while (--num_rows >= 0) {
-    for (ci = 0; ci < nc; ci++) {
+    /* It seems fastest to make a separate pass for each component. */
+    for (ci = 0; ci < num_comps; ci++) {
      inptr = input_buf[ci][input_row];
      outptr = output_buf[0] + ci;
-      for (col = 0; col < num_cols; col++) {
-	*outptr = *inptr++;	/* needn't bother with GETJSAMPLE() here */
-	outptr += nc;
+      for (count = num_cols; count > 0; count--) {
+	*outptr = *inptr++;	/* don't need GETJSAMPLE() here */
+	outptr += num_comps;
      }
    }
    input_row++;
@@ -504,9 +496,10 @@ gray_rgb_convert (j_decompress_ptr cinfo,


 /*
- * Adobe-style YCCK->CMYK conversion.
- * We convert YCbCr to R=1-C, G=1-M, and B=1-Y using the same
- * conversion as above, while passing K (black) unchanged.
+ * Convert some rows of samples to the output colorspace.
+ * This version handles Adobe-style YCCK->CMYK conversion,
+ * where we convert YCbCr to R=1-C, G=1-M, and B=1-Y using the
+ * same conversion as above, while passing K (black) unchanged.
 * We assume build_ycc_rgb_table has been called.
 */

@@ -577,9 +570,8 @@ jinit_color_deconverter (j_decompress_ptr cinfo)
  my_cconvert_ptr cconvert;
  int ci;

-  cconvert = (my_cconvert_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_color_deconverter));
+  cconvert = (my_cconvert_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(my_color_deconverter));
  cinfo->cconvert = &cconvert->pub;
  cconvert->pub.start_pass = start_pass_dcolor;

@@ -607,7 +599,6 @@ jinit_color_deconverter (j_decompress_ptr cinfo)
  default:			/* JCS_UNKNOWN can be anything */
    if (cinfo->num_components < 1)
      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    break;
  }

  /* Support color transform only for RGB colorspaces */
@@ -684,19 +675,18 @@ jinit_color_deconverter (j_decompress_ptr cinfo)

  case JCS_BG_RGB:
    cinfo->out_color_components = RGB_PIXELSIZE;
-    if (cinfo->jpeg_color_space == JCS_BG_RGB) {
-      switch (cinfo->color_transform) {
-      case JCT_NONE:
-	cconvert->pub.color_convert = rgb_convert;
-	break;
-      case JCT_SUBTRACT_GREEN:
-	cconvert->pub.color_convert = rgb1_rgb_convert;
-	break;
-      default:
-	ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-      }
-    } else
+    if (cinfo->jpeg_color_space != JCS_BG_RGB)
      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
+    switch (cinfo->color_transform) {
+    case JCT_NONE:
+      cconvert->pub.color_convert = rgb_convert;
+      break;
+    case JCT_SUBTRACT_GREEN:
+      cconvert->pub.color_convert = rgb1_rgb_convert;
+      break;
+    default:
+      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
+    }
    break;

  case JCS_CMYK:
@@ -714,14 +704,12 @@ jinit_color_deconverter (j_decompress_ptr cinfo)
    }
    break;

-  default:
-    /* Permit null conversion to same output space */
-    if (cinfo->out_color_space == cinfo->jpeg_color_space) {
-      cinfo->out_color_components = cinfo->num_components;
-      cconvert->pub.color_convert = null_convert;
-    } else			/* unsupported non-null conversion */
+  default:		/* permit null conversion to same output space */
+    if (cinfo->out_color_space != cinfo->jpeg_color_space)
+      /* unsupported non-null conversion */
      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
+    cinfo->out_color_components = cinfo->num_components;
+    cconvert->pub.color_convert = null_convert;
  }

  if (cinfo->quantize_colors)
@@ -2,7 +2,7 @@
 * jdct.h
 *
 * Copyright (C) 1994-1996, Thomas G. Lane.
- * Modified 2002-2017 by Guido Vollbeding.
+ * Modified 2002-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -358,13 +358,6 @@ EXTERN(void) jpeg_idct_1x2

 #define FIX(x)	((INT32) ((x) * CONST_SCALE + 0.5))

-/* Descale and correctly round an INT32 value that's scaled by N bits.
- * We assume RIGHT_SHIFT rounds towards minus infinity, so adding
- * the fudge factor is correct for either sign of X.
- */
-
-#define DESCALE(x,n)  RIGHT_SHIFT((x) + (ONE << ((n)-1)), n)
-
 /* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
 * This macro is used only when the two inputs will actually be no more than
 * 16 bits wide, so that a 16x16->32 bit multiply can be used instead of a
@@ -2,7 +2,7 @@
 * jdhuff.c
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * Modified 2006-2016 by Guido Vollbeding.
+ * Modified 2006-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -341,13 +341,12 @@ jpeg_make_d_derived_tbl (j_decompress_ptr cinfo, boolean isDC, int tblno,
  htbl =
    isDC ? cinfo->dc_huff_tbl_ptrs[tblno] : cinfo->ac_huff_tbl_ptrs[tblno];
  if (htbl == NULL)
-    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
+    htbl = jpeg_std_huff_table((j_common_ptr) cinfo, isDC, tblno);

  /* Allocate a workspace if we haven't already done so. */
  if (*pdtbl == NULL)
-    *pdtbl = (d_derived_tbl *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(d_derived_tbl));
+    *pdtbl = (d_derived_tbl *) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(d_derived_tbl));
  dtbl = *pdtbl;
  dtbl->pub = htbl;		/* fill in back link */
  
@@ -706,7 +705,7 @@ process_restart (j_decompress_ptr cinfo)

 METHODDEF(boolean)
 decode_mcu_DC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
+{
  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
  int Al = cinfo->Al;
  register int s, r;
@@ -730,7 +729,7 @@ decode_mcu_DC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
  if (! entropy->insufficient_data) {

    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
+    BITREAD_LOAD_STATE(cinfo, entropy->bitstate);
    ASSIGN_STATE(state, entropy->saved);

    /* Outer loop handles each block in the MCU */
@@ -759,12 +758,13 @@ decode_mcu_DC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
    }

    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
+    BITREAD_SAVE_STATE(cinfo, entropy->bitstate);
    ASSIGN_STATE(entropy->saved, state);
  }

-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
+  /* Account for restart interval if using restarts */
+  if (cinfo->restart_interval)
+    entropy->restarts_to_go--;

  return TRUE;
 }
@@ -777,7 +777,7 @@ decode_mcu_DC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)

 METHODDEF(boolean)
 decode_mcu_AC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
+{
  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
  register int s, k, r;
  unsigned int EOBRUN;
@@ -809,7 +809,7 @@ decode_mcu_AC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
    if (EOBRUN)			/* if it's a band of zeroes... */
      EOBRUN--;			/* ...process it now (we do nothing) */
    else {
-      BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
+      BITREAD_LOAD_STATE(cinfo, entropy->bitstate);
      Se = cinfo->Se;
      Al = cinfo->Al;
      natural_order = cinfo->natural_order;
@@ -842,15 +842,16 @@ decode_mcu_AC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
 	}
      }

-      BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
+      BITREAD_SAVE_STATE(cinfo, entropy->bitstate);
    }

    /* Completed MCU, so update state */
    entropy->saved.EOBRUN = EOBRUN;	/* only part of saved state we need */
  }

-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
+  /* Account for restart interval if using restarts */
+  if (cinfo->restart_interval)
+    entropy->restarts_to_go--;

  return TRUE;
 }
@@ -864,9 +865,10 @@ decode_mcu_AC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)

 METHODDEF(boolean)
 decode_mcu_DC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
+{
  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int p1, blkn;
+  JCOEF p1;
+  int blkn;
  BITREAD_STATE_VARS;

  /* Process restart marker if needed; may have to suspend */
@@ -881,7 +883,7 @@ decode_mcu_DC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
   */

  /* Load up working state */
-  BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
+  BITREAD_LOAD_STATE(cinfo, entropy->bitstate);

  p1 = 1 << cinfo->Al;		/* 1 in the bit position being coded */

@@ -896,10 +898,11 @@ decode_mcu_DC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
  }

  /* Completed MCU, so update state */
-  BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
+  BITREAD_SAVE_STATE(cinfo, entropy->bitstate);

-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
+  /* Account for restart interval if using restarts */
+  if (cinfo->restart_interval)
+    entropy->restarts_to_go--;

  return TRUE;
 }
@@ -911,11 +914,12 @@ decode_mcu_DC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)

 METHODDEF(boolean)
 decode_mcu_AC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
+{
  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
  register int s, k, r;
  unsigned int EOBRUN;
-  int Se, p1, m1;
+  int Se;
+  JCOEF p1, m1;
  const int * natural_order;
  JBLOCKROW block;
  JCOEFPTR thiscoef;
@@ -937,11 +941,11 @@ decode_mcu_AC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)

    Se = cinfo->Se;
    p1 = 1 << cinfo->Al;	/* 1 in the bit position being coded */
-    m1 = (-1) << cinfo->Al;	/* -1 in the bit position being coded */
+    m1 = -p1;			/* -1 in the bit position being coded */
    natural_order = cinfo->natural_order;

    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
+    BITREAD_LOAD_STATE(cinfo, entropy->bitstate);
    EOBRUN = entropy->saved.EOBRUN; /* only part of saved state we need */

    /* There is always only one block per MCU */
@@ -1043,12 +1047,13 @@ decode_mcu_AC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
    }

    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
+    BITREAD_SAVE_STATE(cinfo, entropy->bitstate);
    entropy->saved.EOBRUN = EOBRUN; /* only part of saved state we need */
  }

-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
+  /* Account for restart interval if using restarts */
+  if (cinfo->restart_interval)
+    entropy->restarts_to_go--;

  return TRUE;

@@ -1091,7 +1096,7 @@ decode_mcu_sub (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
    Se = cinfo->lim_Se;

    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
+    BITREAD_LOAD_STATE(cinfo, entropy->bitstate);
    ASSIGN_STATE(state, entropy->saved);

    /* Outer loop handles each block in the MCU */
@@ -1178,12 +1183,13 @@ decode_mcu_sub (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
    }

    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
+    BITREAD_SAVE_STATE(cinfo, entropy->bitstate);
    ASSIGN_STATE(entropy->saved, state);
  }

-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
+  /* Account for restart interval if using restarts */
+  if (cinfo->restart_interval)
+    entropy->restarts_to_go--;

  return TRUE;
 }
@@ -1215,7 +1221,7 @@ decode_mcu (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
  if (! entropy->insufficient_data) {

    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
+    BITREAD_LOAD_STATE(cinfo, entropy->bitstate);
    ASSIGN_STATE(state, entropy->saved);

    /* Outer loop handles each block in the MCU */
@@ -1302,12 +1308,13 @@ decode_mcu (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
    }

    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
+    BITREAD_SAVE_STATE(cinfo, entropy->bitstate);
    ASSIGN_STATE(entropy->saved, state);
  }

-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
+  /* Account for restart interval if using restarts */
+  if (cinfo->restart_interval)
+    entropy->restarts_to_go--;

  return TRUE;
 }
@@ -1343,11 +1350,11 @@ start_pass_huff_decoder (j_decompress_ptr cinfo)
 	goto bad;
    }
    if (cinfo->Al > 13) {	/* need not check for < 0 */
-      /* Arguably the maximum Al value should be less than 13 for 8-bit precision,
-       * but the spec doesn't say so, and we try to be liberal about what we
-       * accept.  Note: large Al values could result in out-of-range DC
-       * coefficients during early scans, leading to bizarre displays due to
-       * overflows in the IDCT math.  But we won't crash.
+      /* Arguably the maximum Al value should be less than 13 for 8-bit
+       * precision, but the spec doesn't say so, and we try to be liberal
+       * about what we accept.  Note: large Al values could result in
+       * out-of-range DC coefficients during early scans, leading to bizarre
+       * displays due to overflows in the IDCT math.  But we won't crash.
       */
      bad:
      ERREXIT4(cinfo, JERR_BAD_PROGRESSION,
@@ -1451,7 +1458,8 @@ start_pass_huff_decoder (j_decompress_ptr cinfo)
      compptr = cinfo->cur_comp_info[ci];
      /* Precalculate which table to use for each block */
      entropy->dc_cur_tbls[blkn] = entropy->dc_derived_tbls[compptr->dc_tbl_no];
-      entropy->ac_cur_tbls[blkn] = entropy->ac_derived_tbls[compptr->ac_tbl_no];
+      entropy->ac_cur_tbls[blkn] =	/* AC needs no table when not present */
+	cinfo->lim_Se ? entropy->ac_derived_tbls[compptr->ac_tbl_no] : NULL;
      /* Decide whether we really care about the coefficient values */
      if (compptr->component_needed) {
 	ci = compptr->DCT_v_scaled_size;
@@ -1494,7 +1502,6 @@ start_pass_huff_decoder (j_decompress_ptr cinfo)
 	  if (ci <= 0 || ci > 8) ci = 8;
 	  if (i <= 0 || i > 8) i = 8;
 	  entropy->coef_limit[blkn] = 1 + jpeg_zigzag_order[ci - 1][i - 1];
-	  break;
 	}
      } else {
 	entropy->coef_limit[blkn] = 0;
@@ -1522,9 +1529,8 @@ jinit_huff_decoder (j_decompress_ptr cinfo)
  huff_entropy_ptr entropy;
  int i;

-  entropy = (huff_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(huff_entropy_decoder));
+  entropy = (huff_entropy_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(huff_entropy_decoder));
  cinfo->entropy = &entropy->pub;
  entropy->pub.start_pass = start_pass_huff_decoder;
  entropy->pub.finish_pass = finish_pass_huff;
@@ -1532,9 +1538,9 @@ jinit_huff_decoder (j_decompress_ptr cinfo)
  if (cinfo->progressive_mode) {
    /* Create progression status table */
    int *coef_bit_ptr, ci;
-    cinfo->coef_bits = (int (*)[DCTSIZE2])
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  cinfo->num_components*DCTSIZE2*SIZEOF(int));
+    cinfo->coef_bits = (int (*)[DCTSIZE2]) (*cinfo->mem->alloc_small)
+      ((j_common_ptr) cinfo, JPOOL_IMAGE,
+       cinfo->num_components * DCTSIZE2 * SIZEOF(int));
    coef_bit_ptr = & cinfo->coef_bits[0][0];
    for (ci = 0; ci < cinfo->num_components; ci++)
      for (i = 0; i < DCTSIZE2; i++)
@@ -1545,7 +1551,7 @@ jinit_huff_decoder (j_decompress_ptr cinfo)
      entropy->derived_tbls[i] = NULL;
    }
  } else {
-    /* Mark tables unallocated */
+    /* Mark derived tables unallocated */
    for (i = 0; i < NUM_HUFF_TBLS; i++) {
      entropy->dc_derived_tbls[i] = entropy->ac_derived_tbls[i] = NULL;
    }
@@ -2,7 +2,7 @@
 * jdmarker.c
 *
 * Copyright (C) 1991-1998, Thomas G. Lane.
- * Modified 2009-2013 by Guido Vollbeding.
+ * Modified 2009-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -496,8 +496,6 @@ get_dht (j_decompress_ptr cinfo)
    if (count > 256 || ((INT32) count) > length)
      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);

-    MEMZERO(huffval, SIZEOF(huffval)); /* pre-zero array for later copy */
-
    for (i = 0; i < count; i++)
      INPUT_BYTE(cinfo, huffval[i], return FALSE);

@@ -517,7 +515,8 @@ get_dht (j_decompress_ptr cinfo)
      *htblptr = jpeg_alloc_huff_table((j_common_ptr) cinfo);
  
    MEMCOPY((*htblptr)->bits, bits, SIZEOF((*htblptr)->bits));
-    MEMCOPY((*htblptr)->huffval, huffval, SIZEOF((*htblptr)->huffval));
+    if (count > 0)
+      MEMCOPY((*htblptr)->huffval, huffval, count * SIZEOF(UINT8));
  }

  if (length != 0)
@@ -577,14 +576,14 @@ get_dqt (j_decompress_ptr cinfo)
 	count = DCTSIZE2;
    }

-    switch (count) {
+    switch ((int) count) {
    case (2*2): natural_order = jpeg_natural_order2; break;
    case (3*3): natural_order = jpeg_natural_order3; break;
    case (4*4): natural_order = jpeg_natural_order4; break;
    case (5*5): natural_order = jpeg_natural_order5; break;
    case (6*6): natural_order = jpeg_natural_order6; break;
    case (7*7): natural_order = jpeg_natural_order7; break;
-    default:    natural_order = jpeg_natural_order;  break;
+    default:    natural_order = jpeg_natural_order;
    }

    for (i = 0; i < count; i++) {
@@ -784,7 +783,6 @@ examine_app0 (j_decompress_ptr cinfo, JOCTET FAR * data,
    default:
      TRACEMS2(cinfo, 1, JTRC_JFIF_EXTENSION,
 	       GETJOCTET(data[5]), (int) totallen);
-      break;
    }
  } else {
    /* Start of APP0 does not match "JFIF" or "JFXX", or too short */
@@ -858,7 +856,6 @@ get_interesting_appn (j_decompress_ptr cinfo)
  default:
    /* can't get here unless jpeg_save_markers chooses wrong processor */
    ERREXIT1(cinfo, JERR_UNKNOWN_MARKER, cinfo->unread_marker);
-    break;
  }

  /* skip any remaining data -- could be lots */
@@ -964,7 +961,6 @@ save_marker (j_decompress_ptr cinfo)
  default:
    TRACEMS2(cinfo, 1, JTRC_MISC_MARKER, cinfo->unread_marker,
 	     (int) (data_length + length));
-    break;
  }

  /* skip any remaining data -- could be lots */
@@ -1240,7 +1236,6 @@ read_markers (j_decompress_ptr cinfo)
       * ought to change!
       */
      ERREXIT1(cinfo, JERR_UNKNOWN_MARKER, cinfo->unread_marker);
-      break;
    }
    /* Successfully processed marker, so reset state variable */
    cinfo->unread_marker = 0;
@@ -1416,9 +1411,8 @@ jinit_marker_reader (j_decompress_ptr cinfo)
  int i;

  /* Create subobject in permanent pool */
-  marker = (my_marker_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				SIZEOF(my_marker_reader));
+  marker = (my_marker_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_PERMANENT, SIZEOF(my_marker_reader));
  cinfo->marker = &marker->pub;
  /* Initialize public method pointers */
  marker->pub.reset_marker_reader = reset_marker_reader;
@@ -2,7 +2,7 @@
 * jdmaster.c
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * Modified 2002-2017 by Guido Vollbeding.
+ * Modified 2002-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -104,7 +104,7 @@ jpeg_calc_output_dimensions (j_decompress_ptr cinfo)
 */
 {
 #ifdef IDCT_SCALING_SUPPORTED
-  int ci;
+  int ci, ssize;
  jpeg_component_info *compptr;
 #endif

@@ -124,19 +124,23 @@ jpeg_calc_output_dimensions (j_decompress_ptr cinfo)
   */
  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
       ci++, compptr++) {
-    int ssize = 1;
-    while (cinfo->min_DCT_h_scaled_size * ssize <=
-	   (cinfo->do_fancy_upsampling ? DCTSIZE : DCTSIZE / 2) &&
-	   (cinfo->max_h_samp_factor % (compptr->h_samp_factor * ssize * 2)) == 0) {
-      ssize = ssize * 2;
-    }
+    ssize = 1;
+    if (! cinfo->raw_data_out)
+      while (cinfo->min_DCT_h_scaled_size * ssize <=
+	     (cinfo->do_fancy_upsampling ? DCTSIZE : DCTSIZE / 2) &&
+	     (cinfo->max_h_samp_factor % (compptr->h_samp_factor * ssize * 2)) ==
+	     0) {
+	ssize = ssize * 2;
+      }
    compptr->DCT_h_scaled_size = cinfo->min_DCT_h_scaled_size * ssize;
    ssize = 1;
-    while (cinfo->min_DCT_v_scaled_size * ssize <=
-	   (cinfo->do_fancy_upsampling ? DCTSIZE : DCTSIZE / 2) &&
-	   (cinfo->max_v_samp_factor % (compptr->v_samp_factor * ssize * 2)) == 0) {
-      ssize = ssize * 2;
-    }
+    if (! cinfo->raw_data_out)
+      while (cinfo->min_DCT_v_scaled_size * ssize <=
+	     (cinfo->do_fancy_upsampling ? DCTSIZE : DCTSIZE / 2) &&
+	     (cinfo->max_v_samp_factor % (compptr->v_samp_factor * ssize * 2)) ==
+	     0) {
+	ssize = ssize * 2;
+      }
    compptr->DCT_v_scaled_size = cinfo->min_DCT_v_scaled_size * ssize;

    /* We don't support IDCT ratios larger than 2. */
@@ -144,13 +148,10 @@ jpeg_calc_output_dimensions (j_decompress_ptr cinfo)
 	compptr->DCT_h_scaled_size = compptr->DCT_v_scaled_size * 2;
    else if (compptr->DCT_v_scaled_size > compptr->DCT_h_scaled_size * 2)
 	compptr->DCT_v_scaled_size = compptr->DCT_h_scaled_size * 2;
-  }

-  /* Recompute downsampled dimensions of components;
-   * application needs to know these if using raw downsampled data.
-   */
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
+    /* Recompute downsampled dimensions of components;
+     * application needs to know these if using raw downsampled data.
+     */
    /* Size in samples, after IDCT scaling */
    compptr->downsampled_width = (JDIMENSION)
      jdiv_round_up((long) cinfo->image_width *
@@ -172,8 +173,10 @@ jpeg_calc_output_dimensions (j_decompress_ptr cinfo)
    break;
  case JCS_RGB:
  case JCS_BG_RGB:
+#if RGB_PIXELSIZE != 3
    cinfo->out_color_components = RGB_PIXELSIZE;
    break;
+#endif /* else share code with YCbCr */
  case JCS_YCbCr:
  case JCS_BG_YCC:
    cinfo->out_color_components = 3;
@@ -184,7 +187,6 @@ jpeg_calc_output_dimensions (j_decompress_ptr cinfo)
    break;
  default:			/* else must be same colorspace as in file */
    cinfo->out_color_components = cinfo->num_components;
-    break;
  }
  cinfo->output_components = (cinfo->quantize_colors ? 1 :
 			      cinfo->out_color_components);
@@ -525,9 +527,8 @@ jinit_master_decompress (j_decompress_ptr cinfo)
 {
  my_master_ptr master;

-  master = (my_master_ptr)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(my_decomp_master));
+  master = (my_master_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(my_decomp_master));
  cinfo->master = &master->pub;
  master->pub.prepare_for_output_pass = prepare_for_output_pass;
  master->pub.finish_output_pass = finish_output_pass;
@@ -2,7 +2,7 @@
 * jdmerge.c
 *
 * Copyright (C) 1994-1996, Thomas G. Lane.
- * Modified 2013-2017 by Guido Vollbeding.
+ * Modified 2013-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -95,28 +95,22 @@ build_ycc_rgb_table (j_decompress_ptr cinfo)
  INT32 x;
  SHIFT_TEMPS

-  upsample->Cr_r_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  upsample->Cb_b_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  upsample->Cr_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
-  upsample->Cb_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
+  upsample->Cr_r_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  upsample->Cb_b_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  upsample->Cr_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));
+  upsample->Cb_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));

  for (i = 0, x = -CENTERJSAMPLE; i <= MAXJSAMPLE; i++, x++) {
    /* i is the actual input pixel value, in the range 0..MAXJSAMPLE */
    /* The Cb or Cr value we are thinking of is x = i - CENTERJSAMPLE */
    /* Cr=>R value is nearest int to 1.402 * x */
-    upsample->Cr_r_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(1.402) * x + ONE_HALF, SCALEBITS);
+    upsample->Cr_r_tab[i] = (int) DESCALE(FIX(1.402) * x, SCALEBITS);
    /* Cb=>B value is nearest int to 1.772 * x */
-    upsample->Cb_b_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(1.772) * x + ONE_HALF, SCALEBITS);
+    upsample->Cb_b_tab[i] = (int) DESCALE(FIX(1.772) * x, SCALEBITS);
    /* Cr=>G value is scaled-up -0.714136286 * x */
    upsample->Cr_g_tab[i] = (- FIX(0.714136286)) * x;
    /* Cb=>G value is scaled-up -0.344136286 * x */
@@ -135,28 +129,22 @@ build_bg_ycc_rgb_table (j_decompress_ptr cinfo)
  INT32 x;
  SHIFT_TEMPS

-  upsample->Cr_r_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  upsample->Cb_b_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  upsample->Cr_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
-  upsample->Cb_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
+  upsample->Cr_r_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  upsample->Cb_b_tab = (int *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(int));
+  upsample->Cr_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));
+  upsample->Cb_g_tab = (INT32 *) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, (MAXJSAMPLE+1) * SIZEOF(INT32));

  for (i = 0, x = -CENTERJSAMPLE; i <= MAXJSAMPLE; i++, x++) {
    /* i is the actual input pixel value, in the range 0..MAXJSAMPLE */
    /* The Cb or Cr value we are thinking of is x = i - CENTERJSAMPLE */
    /* Cr=>R value is nearest int to 2.804 * x */
-    upsample->Cr_r_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(2.804) * x + ONE_HALF, SCALEBITS);
+    upsample->Cr_r_tab[i] = (int) DESCALE(FIX(2.804) * x, SCALEBITS);
    /* Cb=>B value is nearest int to 3.544 * x */
-    upsample->Cb_b_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(3.544) * x + ONE_HALF, SCALEBITS);
+    upsample->Cb_b_tab[i] = (int) DESCALE(FIX(3.544) * x, SCALEBITS);
    /* Cr=>G value is scaled-up -1.428272572 * x */
    upsample->Cr_g_tab[i] = (- FIX(1.428272572)) * x;
    /* Cb=>G value is scaled-up -0.688272572 * x */
@@ -419,9 +407,8 @@ jinit_merged_upsampler (j_decompress_ptr cinfo)
 {
  my_upsample_ptr upsample;

-  upsample = (my_upsample_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_upsampler));
+  upsample = (my_upsample_ptr) (*cinfo->mem->alloc_small)
+    ((j_common_ptr) cinfo, JPOOL_IMAGE, SIZEOF(my_upsampler));
  cinfo->upsample = &upsample->pub;
  upsample->pub.start_pass = start_pass_merged_upsample;
  upsample->pub.need_context_rows = FALSE;
@@ -432,9 +419,9 @@ jinit_merged_upsampler (j_decompress_ptr cinfo)
    upsample->pub.upsample = merged_2v_upsample;
    upsample->upmethod = h2v2_merged_upsample;
    /* Allocate a spare row buffer */
-    upsample->spare_row = (JSAMPROW)
-      (*cinfo->mem->alloc_large) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-		(size_t) (upsample->out_row_width * SIZEOF(JSAMPLE)));
+    upsample->spare_row = (JSAMPROW) (*cinfo->mem->alloc_large)
+      ((j_common_ptr) cinfo, JPOOL_IMAGE,
+       (size_t) upsample->out_row_width * SIZEOF(JSAMPLE));
  } else {
    upsample->pub.upsample = merged_1v_upsample;
    upsample->upmethod = h2v1_merged_upsample;
@@ -2,7 +2,7 @@
 * jerror.h
 *
 * Copyright (C) 1994-1997, Thomas G. Lane.
- * Modified 1997-2012 by Guido Vollbeding.
+ * Modified 1997-2018 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -84,7 +84,7 @@ JMESSAGE(JERR_EOI_EXPECTED, "Didn't expect more than one scan")
 JMESSAGE(JERR_FILE_READ, "Input file read error")
 JMESSAGE(JERR_FILE_WRITE, "Output file write error --- out of disk space?")
 JMESSAGE(JERR_FRACT_SAMPLE_NOTIMPL, "Fractional sampling not implemented yet")
-JMESSAGE(JERR_HUFF_CLEN_OVERFLOW, "Huffman code size table overflow")
+JMESSAGE(JERR_HUFF_CLEN_OUTOFBOUNDS, "Huffman code size table out of bounds")
 JMESSAGE(JERR_HUFF_MISSING_CODE, "Missing Huffman code table entry")
 JMESSAGE(JERR_IMAGE_TOO_BIG, "Maximum supported image dimension is %u pixels")
 JMESSAGE(JERR_INPUT_EMPTY, "Empty input file")
@@ -2,7 +2,7 @@
 * jfdctint.c
 *
 * Copyright (C) 1991-1996, Thomas G. Lane.
- * Modification developed 2003-2015 by Guido Vollbeding.
+ * Modification developed 2003-2018 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -3261,78 +3261,84 @@ jpeg_fdct_6x3 (DCTELEM * data, JSAMPARRAY sample_data, JDIMENSION start_col)
 GLOBAL(void)
 jpeg_fdct_4x2 (DCTELEM * data, JSAMPARRAY sample_data, JDIMENSION start_col)
 {
-  INT32 tmp0, tmp1;
-  INT32 tmp10, tmp11;
-  DCTELEM *dataptr;
+  DCTELEM tmp0, tmp2, tmp10, tmp12, tmp4, tmp5;
+  INT32 tmp1, tmp3, tmp11, tmp13;
+  INT32 z1, z2, z3;
  JSAMPROW elemptr;
-  int ctr;
  SHIFT_TEMPS

  /* Pre-zero output coefficient block. */
  MEMZERO(data, SIZEOF(DCTELEM) * DCTSIZE2);

  /* Pass 1: process rows.
-   * Note results are scaled up by sqrt(8) compared to a true DCT;
-   * furthermore, we scale the results by 2**PASS1_BITS.
-   * We must also scale the output by (8/4)*(8/2) = 2**3, which we add here.
+   * Note results are scaled up by sqrt(8) compared to a true DCT.
   * 4-point FDCT kernel,
   * cK represents sqrt(2) * cos(K*pi/16) [refers to 8-point FDCT].
   */

-  dataptr = data;
-  for (ctr = 0; ctr < 2; ctr++) {
-    elemptr = sample_data[ctr] + start_col;
+  /* Row 0 */
+  elemptr = sample_data[0] + start_col;

-    /* Even part */
+  /* Even part */

-    tmp0 = GETJSAMPLE(elemptr[0]) + GETJSAMPLE(elemptr[3]);
-    tmp1 = GETJSAMPLE(elemptr[1]) + GETJSAMPLE(elemptr[2]);
+  tmp4 = GETJSAMPLE(elemptr[0]) + GETJSAMPLE(elemptr[3]);
+  tmp5 = GETJSAMPLE(elemptr[1]) + GETJSAMPLE(elemptr[2]);

-    tmp10 = GETJSAMPLE(elemptr[0]) - GETJSAMPLE(elemptr[3]);
-    tmp11 = GETJSAMPLE(elemptr[1]) - GETJSAMPLE(elemptr[2]);
+  tmp0 = tmp4 + tmp5;
+  tmp2 = tmp4 - tmp5;

-    /* Apply unsigned->signed conversion. */
-    dataptr[0] = (DCTELEM)
-      ((tmp0 + tmp1 - 4 * CENTERJSAMPLE) << (PASS1_BITS+3));
-    dataptr[2] = (DCTELEM) ((tmp0 - tmp1) << (PASS1_BITS+3));
+  /* Odd part */

-    /* Odd part */
+  z2 = GETJSAMPLE(elemptr[0]) - GETJSAMPLE(elemptr[3]);
+  z3 = GETJSAMPLE(elemptr[1]) - GETJSAMPLE(elemptr[2]);

-    tmp0 = MULTIPLY(tmp10 + tmp11, FIX_0_541196100);       /* c6 */
-    /* Add fudge factor here for final descale. */
-    tmp0 += ONE << (CONST_BITS-PASS1_BITS-4);
+  z1 = MULTIPLY(z2 + z3, FIX_0_541196100);    /* c6 */
+  /* Add fudge factor here for final descale. */
+  z1 += ONE << (CONST_BITS-3-1);
+  tmp1 = z1 + MULTIPLY(z2, FIX_0_765366865); /* c2-c6 */
+  tmp3 = z1 - MULTIPLY(z3, FIX_1_847759065); /* c2+c6 */

-    dataptr[1] = (DCTELEM)
-      RIGHT_SHIFT(tmp0 + MULTIPLY(tmp10, FIX_0_765366865), /* c2-c6 */
-		  CONST_BITS-PASS1_BITS-3);
-    dataptr[3] = (DCTELEM)
-      RIGHT_SHIFT(tmp0 - MULTIPLY(tmp11, FIX_1_847759065), /* c2+c6 */
-		  CONST_BITS-PASS1_BITS-3);
+  /* Row 1 */
+  elemptr = sample_data[1] + start_col;

-    dataptr += DCTSIZE;		/* advance pointer to next row */
-  }
+  /* Even part */
+
+  tmp4 = GETJSAMPLE(elemptr[0]) + GETJSAMPLE(elemptr[3]);
+  tmp5 = GETJSAMPLE(elemptr[1]) + GETJSAMPLE(elemptr[2]);
+
+  tmp10 = tmp4 + tmp5;
+  tmp12 = tmp4 - tmp5;
+
+  /* Odd part */
+
+  z2 = GETJSAMPLE(elemptr[0]) - GETJSAMPLE(elemptr[3]);
+  z3 = GETJSAMPLE(elemptr[1]) - GETJSAMPLE(elemptr[2]);
+
+  z1 = MULTIPLY(z2 + z3, FIX_0_541196100);    /* c6 */
+  tmp11 = z1 + MULTIPLY(z2, FIX_0_765366865); /* c2-c6 */
+  tmp13 = z1 - MULTIPLY(z3, FIX_1_847759065); /* c2+c6 */

  /* Pass 2: process columns.
-   * We remove the PASS1_BITS scaling, but leave the results scaled up
-   * by an overall factor of 8.
+   * We leave the results scaled up by an overall factor of 8.
+   * We must also scale the output by (8/4)*(8/2) = 2**3.
   */

-  dataptr = data;
-  for (ctr = 0; ctr < 4; ctr++) {
-    /* Even part */
+  /* Column 0 */
+  /* Apply unsigned->signed conversion. */
+  data[DCTSIZE*0] = (tmp0 + tmp10 - 8 * CENTERJSAMPLE) << 3;
+  data[DCTSIZE*1] = (tmp0 - tmp10) << 3;

-    /* Add fudge factor here for final descale. */
-    tmp0 = dataptr[DCTSIZE*0] + (ONE << (PASS1_BITS-1));
-    tmp1 = dataptr[DCTSIZE*1];
+  /* Column 1 */
+  data[DCTSIZE*0+1] = (DCTELEM) RIGHT_SHIFT(tmp1 + tmp11, CONST_BITS-3);
+  data[DCTSIZE*1+1] = (DCTELEM) RIGHT_SHIFT(tmp1 - tmp11, CONST_BITS-3);

-    dataptr[DCTSIZE*0] = (DCTELEM) RIGHT_SHIFT(tmp0 + tmp1, PASS1_BITS);
+  /* Column 2 */
+  data[DCTSIZE*0+2] = (tmp2 + tmp12) << 3;
+  data[DCTSIZE*1+2] = (tmp2 - tmp12) << 3;

-    /* Odd part */
-
-    dataptr[DCTSIZE*1] = (DCTELEM) RIGHT_SHIFT(tmp0 - tmp1, PASS1_BITS);
-
-    dataptr++;			/* advance pointer to next column */
-  }
+  /* Column 3 */
+  data[DCTSIZE*0+3] = (DCTELEM) RIGHT_SHIFT(tmp3 + tmp13, CONST_BITS-3);
+  data[DCTSIZE*1+3] = (DCTELEM) RIGHT_SHIFT(tmp3 - tmp13, CONST_BITS-3);
 }


@@ -4312,7 +4318,6 @@ jpeg_fdct_2x4 (DCTELEM * data, JSAMPARRAY sample_data, JDIMENSION start_col)

  /* Pass 1: process rows.
   * Note results are scaled up by sqrt(8) compared to a true DCT.
-   * We must also scale the output by (8/2)*(8/4) = 2**3, which we add here.
   */

  dataptr = data;
@@ -4325,17 +4330,18 @@ jpeg_fdct_2x4 (DCTELEM * data, JSAMPARRAY sample_data, JDIMENSION start_col)
    tmp1 = GETJSAMPLE(elemptr[1]);

    /* Apply unsigned->signed conversion. */
-    dataptr[0] = (DCTELEM) ((tmp0 + tmp1 - 2 * CENTERJSAMPLE) << 3);
+    dataptr[0] = (DCTELEM) (tmp0 + tmp1 - 2 * CENTERJSAMPLE);

    /* Odd part */

-    dataptr[1] = (DCTELEM) ((tmp0 - tmp1) << 3);
+    dataptr[1] = (DCTELEM) (tmp0 - tmp1);

    dataptr += DCTSIZE;		/* advance pointer to next row */
  }

  /* Pass 2: process columns.
   * We leave the results scaled up by an overall factor of 8.
+   * We must also scale the output by (8/2)*(8/4) = 2**3.
   * 4-point FDCT kernel,
   * cK represents sqrt(2) * cos(K*pi/16) [refers to 8-point FDCT].
   */
@@ -4350,21 +4356,21 @@ jpeg_fdct_2x4 (DCTELEM * data, JSAMPARRAY sample_data, JDIMENSION start_col)
    tmp10 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*3];
    tmp11 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*2];

-    dataptr[DCTSIZE*0] = (DCTELEM) (tmp0 + tmp1);
-    dataptr[DCTSIZE*2] = (DCTELEM) (tmp0 - tmp1);
+    dataptr[DCTSIZE*0] = (DCTELEM) ((tmp0 + tmp1) << 3);
+    dataptr[DCTSIZE*2] = (DCTELEM) ((tmp0 - tmp1) << 3);

    /* Odd part */

    tmp0 = MULTIPLY(tmp10 + tmp11, FIX_0_541196100);       /* c6 */
    /* Add fudge factor here for final descale. */
-    tmp0 += ONE << (CONST_BITS-1);
+    tmp0 += ONE << (CONST_BITS-3-1);

    dataptr[DCTSIZE*1] = (DCTELEM)
      RIGHT_SHIFT(tmp0 + MULTIPLY(tmp10, FIX_0_765366865), /* c2-c6 */
-		  CONST_BITS);
+		  CONST_BITS-3);
    dataptr[DCTSIZE*3] = (DCTELEM)
      RIGHT_SHIFT(tmp0 - MULTIPLY(tmp11, FIX_1_847759065), /* c2+c6 */
-		  CONST_BITS);
+		  CONST_BITS-3);

    dataptr++;			/* advance pointer to next column */
  }
@@ -2,7 +2,7 @@
 * jidctint.c
 *
 * Copyright (C) 1991-1998, Thomas G. Lane.
- * Modification developed 2002-2016 by Guido Vollbeding.
+ * Modification developed 2002-2018 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -1474,7 +1474,7 @@ jpeg_idct_10x10 (j_decompress_ptr cinfo, jpeg_component_info * compptr,

 /*
 * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a 11x11 output block.
+ * producing an 11x11 output block.
 *
 * Optimized algorithm with 24 multiplications in the 1-D kernel.
 * cK represents sqrt(2) * cos(K*pi/22).
@@ -3675,7 +3675,7 @@ jpeg_idct_10x5 (j_decompress_ptr cinfo, jpeg_component_info * compptr,

 /*
 * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a 8x4 output block.
+ * producing an 8x4 output block.
 *
 * 4-point IDCT in pass 1 (columns), 8-point in pass 2 (rows).
 */
@@ -3835,7 +3835,7 @@ jpeg_idct_8x4 (j_decompress_ptr cinfo, jpeg_component_info * compptr,

 /*
 * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a reduced-size 6x3 output block.
+ * producing a 6x3 output block.
 *
 * 3-point IDCT in pass 1 (columns), 6-point in pass 2 (rows).
 */
@@ -4082,7 +4082,7 @@ jpeg_idct_2x1 (j_decompress_ptr cinfo, jpeg_component_info * compptr,

 /*
 * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a 8x16 output block.
+ * producing an 8x16 output block.
 *
 * 16-point IDCT in pass 1 (columns), 8-point in pass 2 (rows).
 */
@@ -5004,7 +5004,7 @@ jpeg_idct_4x8 (j_decompress_ptr cinfo, jpeg_component_info * compptr,

 /*
 * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a reduced-size 3x6 output block.
+ * producing a 3x6 output block.
 *
 * 6-point IDCT in pass 1 (columns), 3-point in pass 2 (rows).
 */
@@ -2,7 +2,7 @@
 * jmemmgr.c
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * Modified 2011-2012 by Guido Vollbeding.
+ * Modified 2011-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -130,7 +130,7 @@ typedef struct {
  jvirt_barray_ptr virt_barray_list;

  /* This counts total space obtained from jpeg_get_small/large */
-  long total_space_allocated;
+  size_t total_space_allocated;

  /* alloc_sarray and alloc_barray set this value for use by virtual
   * array routines.
@@ -195,7 +195,7 @@ print_mem_stats (j_common_ptr cinfo, int pool_id)
   * This is helpful because message parm array can't handle longs.
   */
  fprintf(stderr, "Freeing pool %d, total space = %ld\n",
-	  pool_id, mem->total_space_allocated);
+	  pool_id, (long) mem->total_space_allocated);

  for (lhdr_ptr = mem->large_list[pool_id]; lhdr_ptr != NULL;
       lhdr_ptr = lhdr_ptr->hdr.next) {
@@ -260,11 +260,11 @@ alloc_small (j_common_ptr cinfo, int pool_id, size_t sizeofobject)
 {
  my_mem_ptr mem = (my_mem_ptr) cinfo->mem;
  small_pool_ptr hdr_ptr, prev_hdr_ptr;
-  char * data_ptr;
  size_t odd_bytes, min_request, slop;
+  char * data_ptr;

  /* Check for unsatisfiable request (do now to ensure no overflow below) */
-  if (sizeofobject > (size_t) (MAX_ALLOC_CHUNK-SIZEOF(small_pool_hdr)))
+  if (sizeofobject > (size_t) MAX_ALLOC_CHUNK - SIZEOF(small_pool_hdr))
    out_of_memory(cinfo, 1);	/* request exceeds malloc's ability */

  /* Round up the requested size to a multiple of SIZEOF(ALIGN_TYPE) */
@@ -293,8 +293,8 @@ alloc_small (j_common_ptr cinfo, int pool_id, size_t sizeofobject)
    else
      slop = extra_pool_slop[pool_id];
    /* Don't ask for more than MAX_ALLOC_CHUNK */
-    if (slop > (size_t) (MAX_ALLOC_CHUNK-min_request))
-      slop = (size_t) (MAX_ALLOC_CHUNK-min_request);
+    if (slop > (size_t) MAX_ALLOC_CHUNK - min_request)
+      slop = (size_t) MAX_ALLOC_CHUNK - min_request;
    /* Try to get space, if fail reduce slop and try again */
    for (;;) {
      hdr_ptr = (small_pool_ptr) jpeg_get_small(cinfo, min_request + slop);
@@ -348,7 +348,7 @@ alloc_large (j_common_ptr cinfo, int pool_id, size_t sizeofobject)
  size_t odd_bytes;

  /* Check for unsatisfiable request (do now to ensure no overflow below) */
-  if (sizeofobject > (size_t) (MAX_ALLOC_CHUNK-SIZEOF(large_pool_hdr)))
+  if (sizeofobject > (size_t) MAX_ALLOC_CHUNK - SIZEOF(large_pool_hdr))
    out_of_memory(cinfo, 3);	/* request exceeds malloc's ability */

  /* Round up the requested size to a multiple of SIZEOF(ALIGN_TYPE) */
@@ -404,7 +404,7 @@ alloc_sarray (j_common_ptr cinfo, int pool_id,
  long ltemp;

  /* Calculate max # of rows allowed in one allocation chunk */
-  ltemp = (MAX_ALLOC_CHUNK-SIZEOF(large_pool_hdr)) /
+  ltemp = (MAX_ALLOC_CHUNK - SIZEOF(large_pool_hdr)) /
 	  ((long) samplesperrow * SIZEOF(JSAMPLE));
  if (ltemp <= 0)
    ERREXIT(cinfo, JERR_WIDTH_OVERFLOW);
@@ -416,15 +416,14 @@ alloc_sarray (j_common_ptr cinfo, int pool_id,

  /* Get space for row pointers (small object) */
  result = (JSAMPARRAY) alloc_small(cinfo, pool_id,
-				    (size_t) (numrows * SIZEOF(JSAMPROW)));
+				    (size_t) numrows * SIZEOF(JSAMPROW));

  /* Get the rows themselves (large objects) */
  currow = 0;
  while (currow < numrows) {
    rowsperchunk = MIN(rowsperchunk, numrows - currow);
    workspace = (JSAMPROW) alloc_large(cinfo, pool_id,
-	(size_t) ((size_t) rowsperchunk * (size_t) samplesperrow
-		  * SIZEOF(JSAMPLE)));
+      (size_t) rowsperchunk * (size_t) samplesperrow * SIZEOF(JSAMPLE));
    for (i = rowsperchunk; i > 0; i--) {
      result[currow++] = workspace;
      workspace += samplesperrow;
@@ -452,7 +451,7 @@ alloc_barray (j_common_ptr cinfo, int pool_id,
  long ltemp;

  /* Calculate max # of rows allowed in one allocation chunk */
-  ltemp = (MAX_ALLOC_CHUNK-SIZEOF(large_pool_hdr)) /
+  ltemp = (MAX_ALLOC_CHUNK - SIZEOF(large_pool_hdr)) /
 	  ((long) blocksperrow * SIZEOF(JBLOCK));
  if (ltemp <= 0)
    ERREXIT(cinfo, JERR_WIDTH_OVERFLOW);
@@ -464,15 +463,14 @@ alloc_barray (j_common_ptr cinfo, int pool_id,

  /* Get space for row pointers (small object) */
  result = (JBLOCKARRAY) alloc_small(cinfo, pool_id,
-				     (size_t) (numrows * SIZEOF(JBLOCKROW)));
+				     (size_t) numrows * SIZEOF(JBLOCKROW));

  /* Get the rows themselves (large objects) */
  currow = 0;
  while (currow < numrows) {
    rowsperchunk = MIN(rowsperchunk, numrows - currow);
    workspace = (JBLOCKROW) alloc_large(cinfo, pool_id,
-	(size_t) ((size_t) rowsperchunk * (size_t) blocksperrow
-		  * SIZEOF(JBLOCK)));
+      (size_t) rowsperchunk * (size_t) blocksperrow * SIZEOF(JBLOCK));
    for (i = rowsperchunk; i > 0; i--) {
      result[currow++] = workspace;
      workspace += blocksperrow;
@@ -585,8 +583,8 @@ realize_virt_arrays (j_common_ptr cinfo)
 /* Allocate the in-memory buffers for any unrealized virtual arrays */
 {
  my_mem_ptr mem = (my_mem_ptr) cinfo->mem;
-  long space_per_minheight, maximum_space, avail_mem;
-  long minheights, max_minheights;
+  long bytesperrow, space_per_minheight, maximum_space;
+  long avail_mem, minheights, max_minheights;
  jvirt_sarray_ptr sptr;
  jvirt_barray_ptr bptr;

@@ -598,18 +596,16 @@ realize_virt_arrays (j_common_ptr cinfo)
  maximum_space = 0;
  for (sptr = mem->virt_sarray_list; sptr != NULL; sptr = sptr->next) {
    if (sptr->mem_buffer == NULL) { /* if not realized yet */
-      space_per_minheight += (long) sptr->maxaccess *
-			     (long) sptr->samplesperrow * SIZEOF(JSAMPLE);
-      maximum_space += (long) sptr->rows_in_array *
-		       (long) sptr->samplesperrow * SIZEOF(JSAMPLE);
+      bytesperrow = (long) sptr->samplesperrow * SIZEOF(JSAMPLE);
+      space_per_minheight += (long) sptr->maxaccess * bytesperrow;
+      maximum_space += (long) sptr->rows_in_array * bytesperrow;
    }
  }
  for (bptr = mem->virt_barray_list; bptr != NULL; bptr = bptr->next) {
    if (bptr->mem_buffer == NULL) { /* if not realized yet */
-      space_per_minheight += (long) bptr->maxaccess *
-			     (long) bptr->blocksperrow * SIZEOF(JBLOCK);
-      maximum_space += (long) bptr->rows_in_array *
-		       (long) bptr->blocksperrow * SIZEOF(JBLOCK);
+      bytesperrow = (long) bptr->blocksperrow * SIZEOF(JBLOCK);
+      space_per_minheight += (long) bptr->maxaccess * bytesperrow;
+      maximum_space += (long) bptr->rows_in_array * bytesperrow;
    }
  }

@@ -618,7 +614,7 @@ realize_virt_arrays (j_common_ptr cinfo)

  /* Determine amount of memory to actually use; this is system-dependent. */
  avail_mem = jpeg_mem_available(cinfo, space_per_minheight, maximum_space,
-				 mem->total_space_allocated);
+				 (long) mem->total_space_allocated);

  /* If the maximum space needed is available, make all the buffers full
   * height; otherwise parcel it out with the same number of minheights
@@ -694,7 +690,7 @@ do_sarray_io (j_common_ptr cinfo, jvirt_sarray_ptr ptr, boolean writing)
  long bytesperrow, file_offset, byte_count, rows, thisrow, i;

  bytesperrow = (long) ptr->samplesperrow * SIZEOF(JSAMPLE);
-  file_offset = ptr->cur_start_row * bytesperrow;
+  file_offset = (long) ptr->cur_start_row * bytesperrow;
  /* Loop to read or write each allocation chunk in mem_buffer */
  for (i = 0; i < (long) ptr->rows_in_mem; i += ptr->rowsperchunk) {
    /* One chunk, but check for short chunk at end of buffer */
@@ -727,7 +723,7 @@ do_barray_io (j_common_ptr cinfo, jvirt_barray_ptr ptr, boolean writing)
  long bytesperrow, file_offset, byte_count, rows, thisrow, i;

  bytesperrow = (long) ptr->blocksperrow * SIZEOF(JBLOCK);
-  file_offset = ptr->cur_start_row * bytesperrow;
+  file_offset = (long) ptr->cur_start_row * bytesperrow;
  /* Loop to read or write each allocation chunk in mem_buffer */
  for (i = 0; i < (long) ptr->rows_in_mem; i += ptr->rowsperchunk) {
    /* One chunk, but check for short chunk at end of buffer */
@@ -771,7 +767,7 @@ access_virt_sarray (j_common_ptr cinfo, jvirt_sarray_ptr ptr,

  /* Make the desired part of the virtual array accessible */
  if (start_row < ptr->cur_start_row ||
-      end_row > ptr->cur_start_row+ptr->rows_in_mem) {
+      end_row > ptr->cur_start_row + ptr->rows_in_mem) {
    if (! ptr->b_s_open)
      ERREXIT(cinfo, JERR_VIRTUAL_BUG);
    /* Flush old buffer contents if necessary */
@@ -856,7 +852,7 @@ access_virt_barray (j_common_ptr cinfo, jvirt_barray_ptr ptr,

  /* Make the desired part of the virtual array accessible */
  if (start_row < ptr->cur_start_row ||
-      end_row > ptr->cur_start_row+ptr->rows_in_mem) {
+      end_row > ptr->cur_start_row + ptr->rows_in_mem) {
    if (! ptr->b_s_open)
      ERREXIT(cinfo, JERR_VIRTUAL_BUG);
    /* Flush old buffer contents if necessary */
@@ -1093,7 +1089,7 @@ jinit_memory_mgr (j_common_ptr cinfo)
  mem->total_space_allocated = SIZEOF(my_memory_mgr);

  /* Declare ourselves open for business */
-  cinfo->mem = & mem->pub;
+  cinfo->mem = &mem->pub;

  /* Check for an environment variable JPEGMEM; if found, override the
   * default max_memory setting from jpeg_mem_init.  Note that the
@@ -2,6 +2,7 @@
 * jmemnobs.c
 *
 * Copyright (C) 1992-1996, Thomas G. Lane.
+ * Modified 2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -12,7 +13,7 @@
 * This is very portable in the sense that it'll compile on almost anything,
 * but you'd better have lots of main memory (or virtual memory) if you want
 * to process big images.
- * Note that the max_memory_to_use option is ignored by this implementation.
+ * Note that the max_memory_to_use option is respected by this implementation.
 */

 #define JPEG_INTERNALS
@@ -66,13 +67,16 @@ jpeg_free_large (j_common_ptr cinfo, void FAR * object, size_t sizeofobject)

 /*
 * This routine computes the total memory space available for allocation.
- * Here we always say, "we got all you want bud!"
 */

 GLOBAL(long)
 jpeg_mem_available (j_common_ptr cinfo, long min_bytes_needed,
 		    long max_bytes_needed, long already_allocated)
 {
+  if (cinfo->mem->max_memory_to_use)
+    return cinfo->mem->max_memory_to_use - already_allocated;
+
+  /* Here we say, "we got all you want bud!" */
  return max_bytes_needed;
 }

@@ -2,7 +2,7 @@
 * jpegint.h
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
- * Modified 1997-2017 by Guido Vollbeding.
+ * Modified 1997-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -302,6 +302,13 @@ struct jpeg_color_quantizer {
 #define RIGHT_SHIFT(x,shft)	((x) >> (shft))
 #endif

+/* Descale and correctly round an INT32 value that's scaled by N bits.
+ * We assume RIGHT_SHIFT rounds towards minus infinity, so adding
+ * the fudge factor is correct for either sign of X.
+ */
+
+#define DESCALE(x,n)	RIGHT_SHIFT((x) + ((INT32) 1 << ((n)-1)), n)
+

 /* Short forms of external names for systems with brain-damaged linkers. */

@@ -2,7 +2,7 @@
 * jpeglib.h
 *
 * Copyright (C) 1991-1998, Thomas G. Lane.
- * Modified 2002-2017 by Guido Vollbeding.
+ * Modified 2002-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -39,7 +39,7 @@ extern "C" {

 #define JPEG_LIB_VERSION        90	/* Compatibility version 9.0 */
 #define JPEG_LIB_VERSION_MAJOR  9
-#define JPEG_LIB_VERSION_MINOR  3
+#define JPEG_LIB_VERSION_MINOR  4


 /* Various constants determining the sizes of things.
@@ -909,6 +909,7 @@ typedef JMETHOD(boolean, jpeg_marker_parser_method, (j_decompress_ptr cinfo));
 #define jpeg_suppress_tables	jSuppressTables
 #define jpeg_alloc_quant_table	jAlcQTable
 #define jpeg_alloc_huff_table	jAlcHTable
+#define jpeg_std_huff_table	jStdHTable
 #define jpeg_start_compress	jStrtCompress
 #define jpeg_write_scanlines	jWrtScanlines
 #define jpeg_finish_compress	jFinCompress
@@ -977,10 +978,10 @@ EXTERN(void) jpeg_stdio_src JPP((j_decompress_ptr cinfo, FILE * infile));
 /* Data source and destination managers: memory buffers. */
 EXTERN(void) jpeg_mem_dest JPP((j_compress_ptr cinfo,
 			       unsigned char ** outbuffer,
-			       unsigned long * outsize));
+			       size_t * outsize));
 EXTERN(void) jpeg_mem_src JPP((j_decompress_ptr cinfo,
 			      const unsigned char * inbuffer,
-			      unsigned long insize));
+			      size_t insize));

 /* Default parameter setup for compression */
 EXTERN(void) jpeg_set_defaults JPP((j_compress_ptr cinfo));
@@ -1005,6 +1006,8 @@ EXTERN(void) jpeg_suppress_tables JPP((j_compress_ptr cinfo,
 				       boolean suppress));
 EXTERN(JQUANT_TBL *) jpeg_alloc_quant_table JPP((j_common_ptr cinfo));
 EXTERN(JHUFF_TBL *) jpeg_alloc_huff_table JPP((j_common_ptr cinfo));
+EXTERN(JHUFF_TBL *) jpeg_std_huff_table JPP((j_common_ptr cinfo,
+					     boolean isDC, int tblno));

 /* Main entry points for compression */
 EXTERN(void) jpeg_start_compress JPP((j_compress_ptr cinfo,
@@ -2,7 +2,7 @@
 * jutils.c
 *
 * Copyright (C) 1991-1996, Thomas G. Lane.
- * Modified 2009-2011 by Guido Vollbeding.
+ * Modified 2009-2019 by Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -185,7 +185,7 @@ jcopy_sample_rows (JSAMPARRAY input_array, int source_row,
 {
  register JSAMPROW inptr, outptr;
 #ifdef FMEMCOPY
-  register size_t count = (size_t) (num_cols * SIZEOF(JSAMPLE));
+  register size_t count = (size_t) num_cols * SIZEOF(JSAMPLE);
 #else
  register JDIMENSION count;
 #endif
@@ -213,7 +213,7 @@ jcopy_block_row (JBLOCKROW input_row, JBLOCKROW output_row,
 /* Copy a row of coefficient blocks from one place to another. */
 {
 #ifdef FMEMCOPY
-  FMEMCOPY(output_row, input_row, num_blocks * (DCTSIZE2 * SIZEOF(JCOEF)));
+  FMEMCOPY(output_row, input_row, (size_t) num_blocks * (DCTSIZE2 * SIZEOF(JCOEF)));
 #else
  register JCOEFPTR inptr, outptr;
  register long count;
@@ -1,7 +1,7 @@
 /*
 * jversion.h
 *
- * Copyright (C) 1991-2018, Thomas G. Lane, Guido Vollbeding.
+ * Copyright (C) 1991-2020, Thomas G. Lane, Guido Vollbeding.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
@@ -9,6 +9,6 @@
 */


-#define JVERSION	"9c  14-Jan-2018"
+#define JVERSION	"9d  12-Jan-2020"

-#define JCOPYRIGHT	"Copyright (C) 2018, Thomas G. Lane, Guido Vollbeding"
+#define JCOPYRIGHT	"Copyright (C) 2020, Thomas G. Lane, Guido Vollbeding"
@@ -74,7 +74,7 @@ if(MSVC)
  add_definitions(-D_CRT_SECURE_NO_DEPRECATE)
 endif(MSVC)

-add_library(${PNG_LIBRARY} STATIC ${lib_srcs} ${lib_hdrs})
+add_library(${PNG_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${lib_srcs} ${lib_hdrs})
 target_link_libraries(${PNG_LIBRARY} ${ZLIB_LIBRARIES})

 ocv_warnings_disable(CMAKE_C_FLAGS -Wundef -Wcast-align -Wimplicit-fallthrough -Wunused-parameter -Wsign-compare)
@@ -92,7 +92,7 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${PNG_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${PNG_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

 ocv_install_3rdparty_licenses(libpng LICENSE README)
@@ -0,0 +1,89 @@
+# COPYRIGHT
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# License); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# AS IS BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+#
+# Copyright (c) 2020, OPEN AI LAB
+# Author: qtang@openailab.com or https://github.com/BUG1989
+#         qli@openailab.com
+#         sqfu@openailab.com
+#
+
+SET(TENGINE_COMMIT_VERSION "8a4c58e0e05cd850f4bb0936a330edc86dc0e28c")
+SET(OCV_TENGINE_DIR "${OpenCV_BINARY_DIR}/3rdparty/libtengine")
+SET(OCV_TENGINE_SOURCE_PATH "${OCV_TENGINE_DIR}/Tengine-${TENGINE_COMMIT_VERSION}")
+
+IF(EXISTS "${OCV_TENGINE_SOURCE_PATH}")
+	MESSAGE(STATUS "Tengine is exist already at: ${OCV_TENGINE_SOURCE_PATH}")
+
+	SET(Tengine_FOUND ON)
+	SET(BUILD_TENGINE ON)
+ELSE()
+	SET(OCV_TENGINE_FILENAME "${TENGINE_COMMIT_VERSION}.zip")#name2
+	SET(OCV_TENGINE_URL "https://github.com/OAID/Tengine/archive/") #url2
+	SET(tengine_md5sum f51ca8f3963faeeff3f019a6f6edc206) #md5sum2
+
+	#MESSAGE(STATUS "**** TENGINE DOWNLOAD BEGIN ****")
+	ocv_download(FILENAME ${OCV_TENGINE_FILENAME}
+						HASH ${tengine_md5sum}
+						URL
+						"${OPENCV_TENGINE_URL}"
+						"$ENV{OPENCV_TENGINE_URL}"
+						"${OCV_TENGINE_URL}"
+						DESTINATION_DIR "${OCV_TENGINE_DIR}"
+						ID TENGINE
+						STATUS res
+						UNPACK RELATIVE_URL)
+
+	if (NOT res)
+		MESSAGE(STATUS "TENGINE DOWNLOAD FAILED. Turning Tengine_FOUND off.")
+		SET(Tengine_FOUND OFF)
+	else ()
+		MESSAGE(STATUS "TENGINE DOWNLOAD success . ")
+
+		SET(Tengine_FOUND ON)
+		SET(BUILD_TENGINE ON)
+	endif()
+ENDIF()
+
+if(BUILD_TENGINE)
+	SET(HAVE_TENGINE 1)
+
+	# android system
+	if(ANDROID)
+	   if(${ANDROID_ABI} STREQUAL "armeabi-v7a")
+			   SET(CONFIG_ARCH_ARM32 ON)
+	   elseif(${ANDROID_ABI} STREQUAL "arm64-v8a")
+			   SET(CONFIG_ARCH_ARM64 ON)
+	   endif()
+	else()
+		# linux system
+		if(CMAKE_SYSTEM_PROCESSOR STREQUAL arm)
+			   SET(CONFIG_ARCH_ARM32 ON)
+		elseif(CMAKE_SYSTEM_PROCESSOR STREQUAL aarch64) ## AARCH64
+			   SET(CONFIG_ARCH_ARM64 ON)
+		endif()
+	endif()
+
+	SET(BUILT_IN_OPENCV ON) ## set for tengine compile discern .
+	SET(Tengine_INCLUDE_DIR  "${OCV_TENGINE_SOURCE_PATH}/core/include" CACHE INTERNAL "")
+	if(EXISTS "${OCV_TENGINE_SOURCE_PATH}/CMakeLists.txt")
+		add_subdirectory("${OCV_TENGINE_SOURCE_PATH}" "${OCV_TENGINE_DIR}/build")
+	else()
+		message(WARNING "TENGINE: Missing 'CMakeLists.txt' in source code package: ${OCV_TENGINE_SOURCE_PATH}")
+	endif()
+	SET(Tengine_LIB "tengine" CACHE INTERNAL "")
+endif()
@@ -126,30 +126,30 @@ endif()
 if(SIZEOF_UNSIGNED_INT EQUAL SIZEOF_SIZE_T)
  set(TIFF_SIZE_T "unsigned int")
  set(TIFF_SIZE_FORMAT "%u")
+  set(TIFF_SSIZE_T "signed int")
+  set(TIFF_SSIZE_FORMAT "%d")
 elseif(SIZEOF_UNSIGNED_LONG EQUAL SIZEOF_SIZE_T)
  set(TIFF_SIZE_T "unsigned long")
  set(TIFF_SIZE_FORMAT "%lu")
+  set(TIFF_SSIZE_T "signed long")
+  set(TIFF_SSIZE_FORMAT "%ld")
 elseif(SIZEOF_UNSIGNED_LONG_LONG EQUAL SIZEOF_SIZE_T)
  set(TIFF_SIZE_T "unsigned long")
  if(MINGW)
    set(TIFF_SIZE_FORMAT "%I64u")
+    set(TIFF_SSIZE_FORMAT "%I64d")
  else()
    set(TIFF_SIZE_FORMAT "%llu")
+    set(TIFF_SSIZE_FORMAT "%lld")
  endif()
 endif()

 if(SIZEOF_SIGNED_INT EQUAL SIZEOF_UNSIGNED_CHAR_P)
-  set(TIFF_SSIZE_T "signed int")
-  set(TIFF_SSIZE_FORMAT "%d")
 elseif(SIZEOF_SIGNED_LONG EQUAL SIZEOF_UNSIGNED_CHAR_P)
-  set(TIFF_SSIZE_T "signed long")
-  set(TIFF_SSIZE_FORMAT "%ld")
 elseif(SIZEOF_SIGNED_LONG_LONG EQUAL SIZEOF_UNSIGNED_CHAR_P)
  set(TIFF_SSIZE_T "signed long long")
  if(MINGW)
-    set(TIFF_SSIZE_FORMAT "%I64d")
  else()
-    set(TIFF_SSIZE_FORMAT "%lld")
  endif()
 endif()

@@ -216,6 +216,8 @@ endif()
 set(fillorder FILLORDER_MSB2LSB)
 if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "i.*86.*" OR
   CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "amd64.*" OR
+   # AMD64 on Windows
+   CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "AMD64" OR
   CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "x86_64.*")
  set(fillorder FILLORDER_LSB2MSB)
 endif()
@@ -460,7 +462,7 @@ ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4456 /wd4457 /wd4312) # vs2015

 ocv_warnings_disable(CMAKE_C_FLAGS /wd4267 /wd4244 /wd4018 /wd4311 /wd4312)

-add_library(${TIFF_LIBRARY} STATIC ${lib_srcs})
+add_library(${TIFF_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${lib_srcs})
 target_link_libraries(${TIFF_LIBRARY} ${ZLIB_LIBRARIES})

 set_target_properties(${TIFF_LIBRARY}
@@ -477,7 +479,7 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${TIFF_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${TIFF_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

 ocv_install_3rdparty_licenses(libtiff COPYRIGHT)
@@ -16,7 +16,11 @@ int _TIFF_vsnprintf_f(char* str, size_t size, const char* format, va_list ap)
  int count = -1;

  if (size != 0)
+#if _MSC_VER <=	1310
+    count = _vsnprintf(str, size, format, ap);
+#else
    count = _vsnprintf_s(str, size, _TRUNCATE, format, ap);
+#endif
  if (count == -1)
    count = _vscprintf(format, ap);

@@ -30,6 +30,7 @@
 #include "tiffiop.h"
 #include "tif_predict.h"
 #include <math.h>
+#include <float.h>

 uint32
 _TIFFMultiply32(TIFF* tif, uint32 first, uint32 second, const char* where)
@@ -357,6 +358,15 @@ _TIFFUInt64ToDouble(uint64 ui64)
 	}
 }

+float _TIFFClampDoubleToFloat( double val )
+{
+    if( val > FLT_MAX )
+        return FLT_MAX;
+    if( val < -FLT_MAX )
+        return -FLT_MAX;
+    return (float)val;
+}
+
 int _TIFFSeekOK(TIFF* tif, toff_t off)
 {
    /* Huge offsets, especially -1 / UINT64_MAX, can cause issues */
@@ -216,7 +216,7 @@
 #endif

 /* Number of bits in a file offset, on hosts where this is settable. */
-//disabled for OpenCV CMakeLists.txt: #define _FILE_OFFSET_BITS @FILE_OFFSET_BITS@
+#define _FILE_OFFSET_BITS @FILE_OFFSET_BITS@

 /* Define to `__inline__' or `__inline' if that's what the C compiler
   calls it, or to nothing if 'inline' is not supported under any name.  */
@@ -29,7 +29,6 @@
 * (and also some miscellaneous stuff)
 */
 #include "tiffiop.h"
-#include <float.h>

 /*
 * These are used in the backwards compatibility code...
@@ -88,13 +87,15 @@ setDoubleArrayOneValue(double** vpp, double value, size_t nmemb)
 * Install extra samples information.
 */
 static int
-setExtraSamples(TIFFDirectory* td, va_list ap, uint32* v)
+setExtraSamples(TIFF* tif, va_list ap, uint32* v)
 {
 /* XXX: Unassociated alpha data == 999 is a known Corel Draw bug, see below */
 #define EXTRASAMPLE_COREL_UNASSALPHA 999 

 	uint16* va;
 	uint32 i;
+        TIFFDirectory* td = &tif->tif_dir;
+        static const char module[] = "setExtraSamples";

 	*v = (uint16) va_arg(ap, uint16_vap);
 	if ((uint16) *v > td->td_samplesperpixel)
@@ -116,6 +117,18 @@ setExtraSamples(TIFFDirectory* td, va_list ap, uint32* v)
 				return 0;
 		}
 	}
+
+        if ( td->td_transferfunction[0] != NULL && (td->td_samplesperpixel - *v > 1) &&
+                !(td->td_samplesperpixel - td->td_extrasamples > 1))
+        {
+                TIFFWarningExt(tif->tif_clientdata,module,
+                    "ExtraSamples tag value is changing, "
+                    "but TransferFunction was read with a different value. Cancelling it");
+                TIFFClrFieldBit(tif,FIELD_TRANSFERFUNCTION);
+                _TIFFfree(td->td_transferfunction[0]);
+                td->td_transferfunction[0] = NULL;
+        }
+
 	td->td_extrasamples = (uint16) *v;
 	_TIFFsetShortArray(&td->td_sampleinfo, va, td->td_extrasamples);
 	return 1;
@@ -153,15 +166,6 @@ bad:
 	return (0);
 }

-static float TIFFClampDoubleToFloat( double val )
-{
-    if( val > FLT_MAX )
-        return FLT_MAX;
-    if( val < -FLT_MAX )
-        return -FLT_MAX;
-    return (float)val;
-}
-
 static int
 _TIFFVSetField(TIFF* tif, uint32 tag, va_list ap)
 {
@@ -285,6 +289,18 @@ _TIFFVSetField(TIFF* tif, uint32 tag, va_list ap)
                _TIFFfree(td->td_smaxsamplevalue);
                td->td_smaxsamplevalue = NULL;
            }
+            /* Test if 3 transfer functions instead of just one are now needed
+               See http://bugzilla.maptools.org/show_bug.cgi?id=2820 */
+            if( td->td_transferfunction[0] != NULL && (v - td->td_extrasamples > 1) &&
+                !(td->td_samplesperpixel - td->td_extrasamples > 1))
+            {
+                    TIFFWarningExt(tif->tif_clientdata,module,
+                        "SamplesPerPixel tag value is changing, "
+                        "but TransferFunction was read with a different value. Cancelling it");
+                    TIFFClrFieldBit(tif,FIELD_TRANSFERFUNCTION);
+                    _TIFFfree(td->td_transferfunction[0]);
+                    td->td_transferfunction[0] = NULL;
+            }
        }
 		td->td_samplesperpixel = (uint16) v;
 		break;
@@ -320,13 +336,13 @@ _TIFFVSetField(TIFF* tif, uint32 tag, va_list ap)
        dblval = va_arg(ap, double);
        if( dblval < 0 )
            goto badvaluedouble;
-		td->td_xresolution = TIFFClampDoubleToFloat( dblval );
+		td->td_xresolution = _TIFFClampDoubleToFloat( dblval );
 		break;
 	case TIFFTAG_YRESOLUTION:
        dblval = va_arg(ap, double);
        if( dblval < 0 )
            goto badvaluedouble;
-		td->td_yresolution = TIFFClampDoubleToFloat( dblval );
+		td->td_yresolution = _TIFFClampDoubleToFloat( dblval );
 		break;
 	case TIFFTAG_PLANARCONFIG:
 		v = (uint16) va_arg(ap, uint16_vap);
@@ -335,10 +351,10 @@ _TIFFVSetField(TIFF* tif, uint32 tag, va_list ap)
 		td->td_planarconfig = (uint16) v;
 		break;
 	case TIFFTAG_XPOSITION:
-		td->td_xposition = TIFFClampDoubleToFloat( va_arg(ap, double) );
+		td->td_xposition = _TIFFClampDoubleToFloat( va_arg(ap, double) );
 		break;
 	case TIFFTAG_YPOSITION:
-		td->td_yposition = TIFFClampDoubleToFloat( va_arg(ap, double) );
+		td->td_yposition = _TIFFClampDoubleToFloat( va_arg(ap, double) );
 		break;
 	case TIFFTAG_RESOLUTIONUNIT:
 		v = (uint16) va_arg(ap, uint16_vap);
@@ -361,7 +377,7 @@ _TIFFVSetField(TIFF* tif, uint32 tag, va_list ap)
 		_TIFFsetShortArray(&td->td_colormap[2], va_arg(ap, uint16*), v32);
 		break;
 	case TIFFTAG_EXTRASAMPLES:
-		if (!setExtraSamples(td, ap, &v))
+		if (!setExtraSamples(tif, ap, &v))
 			goto badvalue;
 		break;
 	case TIFFTAG_MATTEING:
@@ -684,7 +700,7 @@ _TIFFVSetField(TIFF* tif, uint32 tag, va_list ap)
 				case TIFF_SRATIONAL:
 				case TIFF_FLOAT:
 					{
-						float v2 = TIFFClampDoubleToFloat(va_arg(ap, double));
+						float v2 = _TIFFClampDoubleToFloat(va_arg(ap, double));
 						_TIFFmemcpy(val, &v2, tv_size);
 					}
 					break;
@@ -164,6 +164,7 @@ static int TIFFFetchNormalTag(TIFF*, TIFFDirEntry*, int recover);
 static int TIFFFetchStripThing(TIFF* tif, TIFFDirEntry* dir, uint32 nstrips, uint64** lpp);
 static int TIFFFetchSubjectDistance(TIFF*, TIFFDirEntry*);
 static void ChopUpSingleUncompressedStrip(TIFF*);
+static void TryChopUpUncompressedBigTiff(TIFF*);
 static uint64 TIFFReadUInt64(const uint8 *value);
 static int _TIFFGetMaxColorChannels(uint16 photometric);

@@ -4246,6 +4247,19 @@ TIFFReadDirectory(TIFF* tif)
 		ChopUpSingleUncompressedStrip(tif);
    }

+        /* There are also uncompressed stripped files with strips larger than */
+        /* 2 GB, which make them unfriendly with a lot of code. If possible, */
+        /* try to expose smaller "virtual" strips. */
+        if( tif->tif_dir.td_planarconfig == PLANARCONFIG_CONTIG &&
+            tif->tif_dir.td_compression == COMPRESSION_NONE &&
+            (tif->tif_flags&(TIFF_STRIPCHOP|TIFF_ISTILED)) == TIFF_STRIPCHOP &&
+            TIFFStripSize64(tif) > 0x7FFFFFFFUL )
+        {
+            if ( !_TIFFFillStriles(tif) || !tif->tif_dir.td_stripbytecount )
+                return 0;
+            TryChopUpUncompressedBigTiff(tif);
+        }
+
        /*
         * Clear the dirty directory flag. 
         */
@@ -5699,6 +5713,63 @@ TIFFFetchSubjectDistance(TIFF* tif, TIFFDirEntry* dir)
 	}
 }

+static void allocChoppedUpStripArrays(TIFF* tif, uint32 nstrips,
+                                      uint64 stripbytes, uint32 rowsperstrip)
+{
+    TIFFDirectory *td = &tif->tif_dir;
+    uint64 bytecount;
+    uint64 offset;
+    uint32 i;
+    uint64 *newcounts;
+    uint64 *newoffsets;
+
+    newcounts = (uint64*) _TIFFCheckMalloc(tif, nstrips, sizeof (uint64),
+                            "for chopped \"StripByteCounts\" array");
+    newoffsets = (uint64*) _TIFFCheckMalloc(tif, nstrips, sizeof (uint64),
+                            "for chopped \"StripOffsets\" array");
+    if (newcounts == NULL || newoffsets == NULL) {
+        /*
+        * Unable to allocate new strip information, give up and use
+        * the original one strip information.
+        */
+        if (newcounts != NULL)
+            _TIFFfree(newcounts);
+        if (newoffsets != NULL)
+            _TIFFfree(newoffsets);
+        return;
+    }
+
+    /*
+     * Fill the strip information arrays with new bytecounts and offsets
+     * that reflect the broken-up format.
+     */
+    offset = td->td_stripoffset[0];
+    bytecount = td->td_stripoffset[td->td_nstrips-1] +
+                td->td_stripbytecount[td->td_nstrips-1] - offset;
+    for (i = 0; i < nstrips; i++)
+    {
+        if (stripbytes > bytecount)
+            stripbytes = bytecount;
+        newcounts[i] = stripbytes;
+        newoffsets[i] = stripbytes ? offset : 0;
+        offset += stripbytes;
+        bytecount -= stripbytes;
+    }
+
+    /*
+     * Replace old single strip info with multi-strip info.
+     */
+    td->td_stripsperimage = td->td_nstrips = nstrips;
+    TIFFSetField(tif, TIFFTAG_ROWSPERSTRIP, rowsperstrip);
+
+    _TIFFfree(td->td_stripbytecount);
+    _TIFFfree(td->td_stripoffset);
+    td->td_stripbytecount = newcounts;
+    td->td_stripoffset = newoffsets;
+    td->td_stripbytecountsorted = 1;
+}
+
+
 /*
 * Replace a single strip (tile) of uncompressed data by multiple strips
 * (tiles), each approximately STRIP_SIZE_DEFAULT bytes. This is useful for
@@ -5714,11 +5785,8 @@ ChopUpSingleUncompressedStrip(TIFF* tif)
 	uint32 rowblock;
 	uint64 rowblockbytes;
 	uint64 stripbytes;
-	uint32 strip;
 	uint32 nstrips;
 	uint32 rowsperstrip;
-	uint64* newcounts;
-	uint64* newoffsets;

 	bytecount = td->td_stripbytecount[0];
        /* On a newly created file, just re-opened to be filled, we */
@@ -5769,46 +5837,106 @@ ChopUpSingleUncompressedStrip(TIFF* tif)
            return;
        }

-	newcounts = (uint64*) _TIFFCheckMalloc(tif, nstrips, sizeof (uint64),
-				"for chopped \"StripByteCounts\" array");
-	newoffsets = (uint64*) _TIFFCheckMalloc(tif, nstrips, sizeof (uint64),
-				"for chopped \"StripOffsets\" array");
-	if (newcounts == NULL || newoffsets == NULL) {
-		/*
-		 * Unable to allocate new strip information, give up and use
-		 * the original one strip information.
-		 */
-		if (newcounts != NULL)
-			_TIFFfree(newcounts);
-		if (newoffsets != NULL)
-			_TIFFfree(newoffsets);
-		return;
-	}
-	/*
-	 * Fill the strip information arrays with new bytecounts and offsets
-	 * that reflect the broken-up format.
-	 */
-	for (strip = 0; strip < nstrips; strip++) {
-		if (stripbytes > bytecount)
-			stripbytes = bytecount;
-		newcounts[strip] = stripbytes;
-		newoffsets[strip] = stripbytes ? offset : 0;
-		offset += stripbytes;
-		bytecount -= stripbytes;
-	}
-	/*
-	 * Replace old single strip info with multi-strip info.
-	 */
-	td->td_stripsperimage = td->td_nstrips = nstrips;
-	TIFFSetField(tif, TIFFTAG_ROWSPERSTRIP, rowsperstrip);
-
-	_TIFFfree(td->td_stripbytecount);
-	_TIFFfree(td->td_stripoffset);
-	td->td_stripbytecount = newcounts;
-	td->td_stripoffset = newoffsets;
-	td->td_stripbytecountsorted = 1;
+        allocChoppedUpStripArrays(tif, nstrips, stripbytes, rowsperstrip);
 }

+
+/*
+ * Replace a file with contiguous strips > 2 GB of uncompressed data by
+ * multiple smaller strips. This is useful for
+ * dealing with large images or for dealing with machines with a limited
+ * amount memory.
+ */
+static void TryChopUpUncompressedBigTiff( TIFF* tif )
+{
+    TIFFDirectory *td = &tif->tif_dir;
+    uint32 rowblock;
+    uint64 rowblockbytes;
+    uint32 i;
+    uint64 stripsize;
+    uint32 rowblocksperstrip;
+    uint32 rowsperstrip;
+    uint64 stripbytes;
+    uint32 nstrips;
+
+    stripsize = TIFFStripSize64(tif);
+
+    assert( tif->tif_dir.td_planarconfig == PLANARCONFIG_CONTIG );
+    assert( tif->tif_dir.td_compression == COMPRESSION_NONE );
+    assert( (tif->tif_flags&(TIFF_STRIPCHOP|TIFF_ISTILED)) == TIFF_STRIPCHOP );
+    assert( stripsize > 0x7FFFFFFFUL );
+
+    /* On a newly created file, just re-opened to be filled, we */
+    /* don't want strip chop to trigger as it is going to cause issues */
+    /* later ( StripOffsets and StripByteCounts improperly filled) . */
+    if( td->td_stripbytecount[0] == 0 && tif->tif_mode != O_RDONLY )
+        return;
+
+    if ((td->td_photometric == PHOTOMETRIC_YCBCR)&&
+        (!isUpSampled(tif)))
+        rowblock = td->td_ycbcrsubsampling[1];
+    else
+        rowblock = 1;
+    rowblockbytes = TIFFVStripSize64(tif, rowblock);
+    if( rowblockbytes == 0 || rowblockbytes > 0x7FFFFFFFUL )
+    {
+        /* In case of file with gigantic width */
+        return;
+    }
+
+    /* Check that the strips are contiguous and of the expected size */
+    for( i = 0; i < td->td_nstrips; i++ )
+    {
+        if( i == td->td_nstrips - 1 )
+        {
+            if( td->td_stripbytecount[i] < TIFFVStripSize64(
+                    tif, td->td_imagelength - i * td->td_rowsperstrip ) )
+            {
+                return;
+            }
+        }
+        else
+        {
+            if( td->td_stripbytecount[i] != stripsize )
+            {
+                return;
+            }
+            if( i > 0 && td->td_stripoffset[i] !=
+                    td->td_stripoffset[i-1] + td->td_stripbytecount[i - 1] )
+            {
+                return;
+            }
+        }
+    }
+
+    /* Aim for 512 MB strips (that will still be manageable by 32 bit builds */
+    rowblocksperstrip = (uint32) (512 * 1024 * 1024 / rowblockbytes);
+    if( rowblocksperstrip == 0 )
+        rowblocksperstrip = 1;
+    rowsperstrip = rowblocksperstrip * rowblock;
+    stripbytes = rowblocksperstrip * rowblockbytes;
+    assert( stripbytes <= 0x7FFFFFFFUL );
+
+    nstrips = TIFFhowmany_32(td->td_imagelength, rowsperstrip);
+    if( nstrips == 0 )
+        return;
+
+    /* If we are going to allocate a lot of memory, make sure that the */
+    /* file is as big as needed */
+    if( tif->tif_mode == O_RDONLY &&
+        nstrips > 1000000 &&
+        (td->td_stripoffset[td->td_nstrips-1] > TIFFGetFileSize(tif) ||
+         td->td_stripoffset[td->td_nstrips-1] +
+         td->td_stripbytecount[td->td_nstrips-1] > TIFFGetFileSize(tif)) )
+    {
+        return;
+    }
+
+    allocChoppedUpStripArrays(tif, nstrips, stripbytes, rowsperstrip);
+}
+
+
+
 int _TIFFFillStriles( TIFF *tif )
 {
    return _TIFFFillStrilesInternal( tif, 1 );
@@ -28,7 +28,6 @@
 * Directory Write Support Routines.
 */
 #include "tiffiop.h"
-#include <float.h>

 #ifdef HAVE_IEEEFP
 #define TIFFCvtNativeToIEEEFloat(tif, n, fp)
@@ -946,15 +945,6 @@ bad:
 	return(0);
 }

-static float TIFFClampDoubleToFloat( double val )
-{
-    if( val > FLT_MAX )
-        return FLT_MAX;
-    if( val < -FLT_MAX )
-        return -FLT_MAX;
-    return (float)val;
-}
-
 static int8 TIFFClampDoubleToInt8( double val )
 {
    if( val > 127 )
@@ -1029,7 +1019,7 @@ TIFFWriteDirectoryTagSampleformatArray(TIFF* tif, uint32* ndir, TIFFDirEntry* di
 			if (tif->tif_dir.td_bitspersample<=32)
 			{
 				for (i = 0; i < count; ++i)
-					((float*)conv)[i] = TIFFClampDoubleToFloat(value[i]);
+					((float*)conv)[i] = _TIFFClampDoubleToFloat(value[i]);
 				ok = TIFFWriteDirectoryTagFloatArray(tif,ndir,dir,tag,count,(float*)conv);
 			}
 			else
@@ -1893,12 +1883,14 @@ TIFFWriteDirectoryTagTransferfunction(TIFF* tif, uint32* ndir, TIFFDirEntry* dir
 		n=3;
 	if (n==3)
 	{
-		if (!_TIFFmemcmp(tif->tif_dir.td_transferfunction[0],tif->tif_dir.td_transferfunction[2],m*sizeof(uint16)))
+		if (tif->tif_dir.td_transferfunction[2] == NULL ||
+		    !_TIFFmemcmp(tif->tif_dir.td_transferfunction[0],tif->tif_dir.td_transferfunction[2],m*sizeof(uint16)))
 			n=2;
 	}
 	if (n==2)
 	{
-		if (!_TIFFmemcmp(tif->tif_dir.td_transferfunction[0],tif->tif_dir.td_transferfunction[1],m*sizeof(uint16)))
+		if (tif->tif_dir.td_transferfunction[1] == NULL ||
+		    !_TIFFmemcmp(tif->tif_dir.td_transferfunction[0],tif->tif_dir.td_transferfunction[1],m*sizeof(uint16)))
 			n=1;
 	}
 	if (n==0)
@@ -742,9 +742,14 @@ LogLuvEncodeTile(TIFF* tif, uint8* bp, tmsize_t cc, uint16 s)
 #undef exp2  /* Conflict with C'99 function */
 #define exp2(x)		exp(M_LN2*(x))

-#define itrunc(x,m)	((m)==SGILOGENCODE_NODITHER ? \
-				(int)(x) : \
-				(int)((x) + rand()*(1./RAND_MAX) - .5))
+static int itrunc(double x, int m)
+{
+    if( m == SGILOGENCODE_NODITHER )
+        return (int)x;
+    /* Silence CoverityScan warning about bad crypto function */
+    /* coverity[dont_call] */
+    return (int)(x + rand()*(1./RAND_MAX) - .5);
+}

 #if !LOGLUV_PUBLIC
 static
@@ -247,6 +247,8 @@ LZWSetupDecode(TIFF* tif)
 		/*
 		 * Zero-out the unused entries
                 */
+                /* Silence false positive */
+                /* coverity[overrun-buffer-arg] */
                 _TIFFmemset(&sp->dec_codetab[CODE_CLEAR], 0,
 			     (CODE_FIRST - CODE_CLEAR) * sizeof (code_t));
 	}
@@ -640,6 +640,7 @@ PixarLogGuessDataFmt(TIFFDirectory *td)
 static tmsize_t
 multiply_ms(tmsize_t m1, tmsize_t m2)
 {
+        assert(m1 >= 0 && m2 >= 0);
        if( m1 == 0 || m2 > TIFF_TMSIZE_T_MAX / m1 )
            return 0;
        return m1 * m2;
@@ -648,6 +649,7 @@ multiply_ms(tmsize_t m1, tmsize_t m2)
 static tmsize_t
 add_ms(tmsize_t m1, tmsize_t m2)
 {
+        assert(m1 >= 0 && m2 >= 0);
 	/* if either input is zero, assume overflow already occurred */
 	if (m1 == 0 || m2 == 0)
 		return 0;
@@ -817,9 +819,7 @@ PixarLogDecode(TIFF* tif, uint8* op, tmsize_t occ, uint16 s)
 			TIFFErrorExt(tif->tif_clientdata, module,
 			    "Decoding error at scanline %lu, %s",
 			    (unsigned long) tif->tif_row, sp->stream.msg ? sp->stream.msg : "(null)");
-			if (inflateSync(&sp->stream) != Z_OK)
-				return (0);
-			continue;
+			return (0);
 		}
 		if (state != Z_OK) {
 			TIFFErrorExt(tif->tif_clientdata, module, "ZLib error: %s",
@@ -103,6 +103,11 @@ static int TIFFReadAndRealloc( TIFF* tif, tmsize_t size,
                }
                tif->tif_rawdata = new_rawdata;
            }
+            if( tif->tif_rawdata == NULL )
+            {
+                /* should not happen in practice but helps CoverityScan */
+                return 0;
+            }

            bytes_read = TIFFReadFile(tif,
                tif->tif_rawdata + rawdata_offset + already_read, to_read);
@@ -1367,7 +1372,8 @@ TIFFFillTile(TIFF* tif, uint32 tile)
                        tif->tif_rawdataoff = 0;
                        tif->tif_rawdataloaded = bytecountm;
                        
-			if (!isFillOrder(tif, td->td_fillorder) &&
+			if (tif->tif_rawdata != NULL &&
+                            !isFillOrder(tif, td->td_fillorder) &&
 			    (tif->tif_flags & TIFF_NOBITREV) == 0)
 				TIFFReverseBits(tif->tif_rawdata,
                                                tif->tif_rawdataloaded);
@@ -349,6 +349,12 @@ TWebPSetupEncode(TIFF* tif)

  sp->state |= LSTATE_INIT_ENCODE;

+  if (!WebPPictureInit(&sp->sPicture)) {
+    TIFFErrorExt(tif->tif_clientdata, module,
+        "Error initializing WebP picture.");
+    return 0;
+  }
+
  if (!WebPConfigInitInternal(&sp->sEncoderConfig, WEBP_PRESET_DEFAULT,
                              sp->quality_level,
                              WEBP_ENCODER_ABI_VERSION)) {
@@ -357,9 +363,13 @@ TWebPSetupEncode(TIFF* tif)
    return 0;
  }

-#if WEBP_ENCODER_ABI_VERSION >= 0x0100
-  sp->sEncoderConfig.lossless = sp->lossless;
-#endif
+  // WebPConfigInitInternal above sets lossless to false
+  #if WEBP_ENCODER_ABI_VERSION >= 0x0100
+    sp->sEncoderConfig.lossless = sp->lossless;
+    if (sp->lossless) {
+      sp->sPicture.use_argb = 1;
+    }
+  #endif

  if (!WebPValidateConfig(&sp->sEncoderConfig)) {
    TIFFErrorExt(tif->tif_clientdata, module,
@@ -367,12 +377,6 @@ TWebPSetupEncode(TIFF* tif)
    return 0;
  }

-  if (!WebPPictureInit(&sp->sPicture)) {
-    TIFFErrorExt(tif->tif_clientdata, module,
-        "Error initializing WebP picture.");
-    return 0;
-  }
-
  return 1;
 }

@@ -415,6 +419,12 @@ TWebPPreEncode(TIFF* tif, uint16 s)
  /* set up buffer for raw data */
  /* given above check and that nSamples <= 4, buffer_size is <= 1 GB */
  sp->buffer_size = segment_width * segment_height * sp->nSamples;
+  
+  if (sp->pBuffer != NULL) {
+      _TIFFfree(sp->pBuffer);
+      sp->pBuffer = NULL;    
+  }
+  
  sp->pBuffer = _TIFFmalloc(sp->buffer_size);
  if( !sp->pBuffer) {
      TIFFErrorExt(tif->tif_clientdata, module, "Cannot allocate buffer");
@@ -460,7 +470,7 @@ TWebPPostEncode(TIFF* tif)
                    "WebPPictureImportRGB() failed");
      return 0;
  }
-
+  
  if (!WebPEncode(&sp->sEncoderConfig, &sp->sPicture)) {

 #if WEBP_ENCODER_ABI_VERSION >= 0x0100
@@ -540,15 +550,13 @@ TWebPCleanup(TIFF* tif)
  }
  
  if (sp->pBuffer != NULL) {
-    _TIFFfree(sp->pBuffer);
-    sp->pBuffer = NULL;
+      _TIFFfree(sp->pBuffer);
+      sp->pBuffer = NULL;    
  }

-  if (tif->tif_data) {
-    _TIFFfree(tif->tif_data);
-    tif->tif_data = NULL;
-  }
-  
+  _TIFFfree(tif->tif_data);
+  tif->tif_data = NULL;
+
  _TIFFSetDefaultCompressionState(tif);
 }

@@ -570,6 +578,9 @@ TWebPVSetField(TIFF* tif, uint32 tag, va_list ap)
  case TIFFTAG_WEBP_LOSSLESS:
    #if WEBP_ENCODER_ABI_VERSION >= 0x0100
    sp->lossless = va_arg(ap, int);
+    if (sp->lossless){
+      sp->quality_level = 100.0f;      
+    }
    return 1;
    #else
      TIFFErrorExt(tif->tif_clientdata, module,
@@ -124,7 +124,6 @@ ZIPSetupDecode(TIFF* tif)
 static int
 ZIPPreDecode(TIFF* tif, uint16 s)
 {
-	static const char module[] = "ZIPPreDecode";
 	ZIPState* sp = DecoderState(tif);

 	(void) s;
@@ -138,12 +137,7 @@ ZIPPreDecode(TIFF* tif, uint16 s)
 	    we need to simplify this code to reflect a ZLib that is likely updated
 	    to deal with 8byte memory sizes, though this code will respond
 	    appropriately even before we simplify it */
-	sp->stream.avail_in = (uInt) tif->tif_rawcc;
-	if ((tmsize_t)sp->stream.avail_in != tif->tif_rawcc)
-	{
-		TIFFErrorExt(tif->tif_clientdata, module, "ZLib cannot deal with buffers this size");
-		return (0);
-	}
+	sp->stream.avail_in = (uint64)tif->tif_rawcc < 0xFFFFFFFFU ? (uInt) tif->tif_rawcc : 0xFFFFFFFFU;
 	return (inflateReset(&sp->stream) == Z_OK);
 }

@@ -158,46 +152,43 @@ ZIPDecode(TIFF* tif, uint8* op, tmsize_t occ, uint16 s)
 	assert(sp->state == ZSTATE_INIT_DECODE);

        sp->stream.next_in = tif->tif_rawcp;
-	sp->stream.avail_in = (uInt) tif->tif_rawcc;
        
 	sp->stream.next_out = op;
 	assert(sizeof(sp->stream.avail_out)==4);  /* if this assert gets raised,
 	    we need to simplify this code to reflect a ZLib that is likely updated
 	    to deal with 8byte memory sizes, though this code will respond
 	    appropriately even before we simplify it */
-	sp->stream.avail_out = (uInt) occ;
-	if ((tmsize_t)sp->stream.avail_out != occ)
-	{
-		TIFFErrorExt(tif->tif_clientdata, module, "ZLib cannot deal with buffers this size");
-		return (0);
-	}
 	do {
-		int state = inflate(&sp->stream, Z_PARTIAL_FLUSH);
+                int state;
+                uInt avail_in_before = (uint64)tif->tif_rawcc <= 0xFFFFFFFFU ? (uInt)tif->tif_rawcc : 0xFFFFFFFFU;
+                uInt avail_out_before = (uint64)occ < 0xFFFFFFFFU ? (uInt) occ : 0xFFFFFFFFU;
+                sp->stream.avail_in = avail_in_before;
+                sp->stream.avail_out = avail_out_before;
+		state = inflate(&sp->stream, Z_PARTIAL_FLUSH);
+		tif->tif_rawcc -= (avail_in_before - sp->stream.avail_in);
+                occ -= (avail_out_before - sp->stream.avail_out);
 		if (state == Z_STREAM_END)
 			break;
 		if (state == Z_DATA_ERROR) {
 			TIFFErrorExt(tif->tif_clientdata, module,
 			    "Decoding error at scanline %lu, %s",
 			     (unsigned long) tif->tif_row, SAFE_MSG(sp));
-			if (inflateSync(&sp->stream) != Z_OK)
-				return (0);
-			continue;
+			return (0);
 		}
 		if (state != Z_OK) {
 			TIFFErrorExt(tif->tif_clientdata, module, 
 				     "ZLib error: %s", SAFE_MSG(sp));
 			return (0);
 		}
-	} while (sp->stream.avail_out > 0);
-	if (sp->stream.avail_out != 0) {
+	} while (occ > 0);
+	if (occ != 0) {
 		TIFFErrorExt(tif->tif_clientdata, module,
 		    "Not enough data at scanline %lu (short " TIFF_UINT64_FORMAT " bytes)",
-		    (unsigned long) tif->tif_row, (TIFF_UINT64_T) sp->stream.avail_out);
+		    (unsigned long) tif->tif_row, (TIFF_UINT64_T) occ);
 		return (0);
 	}

        tif->tif_rawcp = sp->stream.next_in;
-        tif->tif_rawcc = sp->stream.avail_in;

 	return (1);
 }
@@ -229,7 +220,6 @@ ZIPSetupEncode(TIFF* tif)
 static int
 ZIPPreEncode(TIFF* tif, uint16 s)
 {
-	static const char module[] = "ZIPPreEncode";
 	ZIPState *sp = EncoderState(tif);

 	(void) s;
@@ -242,12 +232,7 @@ ZIPPreEncode(TIFF* tif, uint16 s)
 	    we need to simplify this code to reflect a ZLib that is likely updated
 	    to deal with 8byte memory sizes, though this code will respond
 	    appropriately even before we simplify it */
-	sp->stream.avail_out = (uInt)tif->tif_rawdatasize;
-	if ((tmsize_t)sp->stream.avail_out != tif->tif_rawdatasize)
-	{
-		TIFFErrorExt(tif->tif_clientdata, module, "ZLib cannot deal with buffers this size");
-		return (0);
-	}
+	sp->stream.avail_out = (uint64)tif->tif_rawdatasize <= 0xFFFFFFFFU ? (uInt)tif->tif_rawdatasize : 0xFFFFFFFFU;
 	return (deflateReset(&sp->stream) == Z_OK);
 }

@@ -269,13 +254,9 @@ ZIPEncode(TIFF* tif, uint8* bp, tmsize_t cc, uint16 s)
 	    we need to simplify this code to reflect a ZLib that is likely updated
 	    to deal with 8byte memory sizes, though this code will respond
 	    appropriately even before we simplify it */
-	sp->stream.avail_in = (uInt) cc;
-	if ((tmsize_t)sp->stream.avail_in != cc)
-	{
-		TIFFErrorExt(tif->tif_clientdata, module, "ZLib cannot deal with buffers this size");
-		return (0);
-	}
 	do {
+                uInt avail_in_before = (uint64)cc <= 0xFFFFFFFFU ? (uInt)cc : 0xFFFFFFFFU;
+                sp->stream.avail_in = avail_in_before;
 		if (deflate(&sp->stream, Z_NO_FLUSH) != Z_OK) {
 			TIFFErrorExt(tif->tif_clientdata, module, 
 				     "Encoder error: %s",
@@ -286,9 +267,10 @@ ZIPEncode(TIFF* tif, uint8* bp, tmsize_t cc, uint16 s)
 			tif->tif_rawcc = tif->tif_rawdatasize;
 			TIFFFlushData1(tif);
 			sp->stream.next_out = tif->tif_rawdata;
-			sp->stream.avail_out = (uInt) tif->tif_rawdatasize;  /* this is a safe typecast, as check is made already in ZIPPreEncode */
+			sp->stream.avail_out = (uint64)tif->tif_rawdatasize <= 0xFFFFFFFFU ? (uInt)tif->tif_rawdatasize : 0xFFFFFFFFU;
 		}
-	} while (sp->stream.avail_in > 0);
+		cc -= (avail_in_before - sp->stream.avail_in);
+	} while (cc > 0);
 	return (1);
 }

@@ -314,7 +296,7 @@ ZIPPostEncode(TIFF* tif)
 				tif->tif_rawcc =  tif->tif_rawdatasize - sp->stream.avail_out;
 				TIFFFlushData1(tif);
 				sp->stream.next_out = tif->tif_rawdata;
-				sp->stream.avail_out = (uInt) tif->tif_rawdatasize;  /* this is a safe typecast, as check is made already in ZIPPreEncode */
+				sp->stream.avail_out = (uint64)tif->tif_rawdatasize <= 0xFFFFFFFFU ? (uInt)tif->tif_rawdatasize : 0xFFFFFFFFU;
 			}
 			break;
 		default:
@@ -374,6 +374,8 @@ extern void* _TIFFCheckRealloc(TIFF*, void*, tmsize_t, tmsize_t, const char*);
 extern double _TIFFUInt64ToDouble(uint64);
 extern float _TIFFUInt64ToFloat(uint64);

+extern float _TIFFClampDoubleToFloat(double);
+
 extern tmsize_t
 _TIFFReadEncodedStripAndAllocBuffer(TIFF* tif, uint32 strip,
                                    void **buf, tmsize_t bufsizetoalloc,
@@ -34,7 +34,7 @@ endif()

 add_definitions(-DWEBP_USE_THREAD)

-add_library(${WEBP_LIBRARY} STATIC ${lib_srcs} ${lib_hdrs})
+add_library(${WEBP_LIBRARY} STATIC ${OPENCV_3RDPARTY_EXCLUDE_FROM_ALL} ${lib_srcs} ${lib_hdrs})
 if(ANDROID)
  target_link_libraries(${WEBP_LIBRARY} ${CPUFEATURES_LIBRARIES})
 endif()
@@ -59,6 +59,6 @@ if(ENABLE_SOLUTION_FOLDERS)
 endif()

 if(NOT BUILD_SHARED_LIBS)
-  ocv_install_target(${WEBP_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev)
+  ocv_install_target(${WEBP_LIBRARY} EXPORT OpenCVModules ARCHIVE DESTINATION ${OPENCV_3P_LIB_INSTALL_PATH} COMPONENT dev OPTIONAL)
 endif()

@@ -732,7 +732,7 @@ static int AllocateMemory(VP8Decoder* const dec) {
  mem += f_info_size;
  dec->thread_ctx_.id_ = 0;
  dec->thread_ctx_.f_info_ = dec->f_info_;
-  if (dec->mt_method_ > 0) {
+  if (dec->filter_type_ > 0 && dec->mt_method_ > 0) {
    // secondary cache line. The deblocking process need to make use of the
    // filtering strength from previous macroblock row, while the new ones
    // are being decoded in parallel. We'll just swap the pointers.
@@ -166,9 +166,11 @@ static int AppendToMemBuffer(WebPIDecoder* const idec,
  VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
  MemBuffer* const mem = &idec->mem_;
  const int need_compressed_alpha = NeedCompressedAlpha(idec);
-  const uint8_t* const old_start = mem->buf_ + mem->start_;
+  const uint8_t* const old_start =
+      (mem->buf_ == NULL) ? NULL : mem->buf_ + mem->start_;
  const uint8_t* const old_base =
      need_compressed_alpha ? dec->alpha_data_ : old_start;
+  assert(mem->buf_ != NULL || mem->start_ == 0);
  assert(mem->mode_ == MEM_MODE_APPEND);
  if (data_size > MAX_CHUNK_PAYLOAD) {
    // security safeguard: trying to allocate more than what the format
@@ -184,7 +186,7 @@ static int AppendToMemBuffer(WebPIDecoder* const idec,
    uint8_t* const new_buf =
        (uint8_t*)WebPSafeMalloc(extra_size, sizeof(*new_buf));
    if (new_buf == NULL) return 0;
-    memcpy(new_buf, old_base, current_size);
+    if (old_base != NULL) memcpy(new_buf, old_base, current_size);
    WebPSafeFree(mem->buf_);
    mem->buf_ = new_buf;
    mem->buf_size_ = (size_t)extra_size;
@@ -192,6 +194,7 @@ static int AppendToMemBuffer(WebPIDecoder* const idec,
    mem->end_ = current_size;
  }

+  assert(mem->buf_ != NULL);
  memcpy(mem->buf_ + mem->end_, data, data_size);
  mem->end_ += data_size;
  assert(mem->end_ <= mem->buf_size_);
@@ -204,7 +207,9 @@ static int RemapMemBuffer(WebPIDecoder* const idec,
                          const uint8_t* const data, size_t data_size) {
  MemBuffer* const mem = &idec->mem_;
  const uint8_t* const old_buf = mem->buf_;
-  const uint8_t* const old_start = old_buf + mem->start_;
+  const uint8_t* const old_start =
+      (old_buf == NULL) ? NULL : old_buf + mem->start_;
+  assert(old_buf != NULL || mem->start_ == 0);
  assert(mem->mode_ == MEM_MODE_MAP);

  if (data_size < mem->buf_size_) return 0;  // can't remap to a shorter buffer!
@@ -61,12 +61,17 @@ static const uint16_t kAcTable[128] = {

 void VP8ParseQuant(VP8Decoder* const dec) {
  VP8BitReader* const br = &dec->br_;
-  const int base_q0 = VP8GetValue(br, 7);
-  const int dqy1_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dqy2_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dqy2_ac = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dquv_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dquv_ac = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
+  const int base_q0 = VP8GetValue(br, 7, "global-header");
+  const int dqy1_dc = VP8Get(br, "global-header") ?
+       VP8GetSignedValue(br, 4, "global-header") : 0;
+  const int dqy2_dc = VP8Get(br, "global-header") ?
+       VP8GetSignedValue(br, 4, "global-header") : 0;
+  const int dqy2_ac = VP8Get(br, "global-header") ?
+       VP8GetSignedValue(br, 4, "global-header") : 0;
+  const int dquv_dc = VP8Get(br, "global-header") ?
+       VP8GetSignedValue(br, 4, "global-header") : 0;
+  const int dquv_ac = VP8Get(br, "global-header") ?
+       VP8GetSignedValue(br, 4, "global-header") : 0;

  const VP8SegmentHeader* const hdr = &dec->segment_hdr_;
  int i;
@@ -296,20 +296,21 @@ static void ParseIntraMode(VP8BitReader* const br,
  // to decode more than 1 keyframe.
  if (dec->segment_hdr_.update_map_) {
    // Hardcoded tree parsing
-    block->segment_ = !VP8GetBit(br, dec->proba_.segments_[0])
-                    ? VP8GetBit(br, dec->proba_.segments_[1])
-                    : 2 + VP8GetBit(br, dec->proba_.segments_[2]);
+    block->segment_ = !VP8GetBit(br, dec->proba_.segments_[0], "segments")
+                    ?  VP8GetBit(br, dec->proba_.segments_[1], "segments")
+                    :  VP8GetBit(br, dec->proba_.segments_[2], "segments") + 2;
  } else {
    block->segment_ = 0;  // default for intra
  }
-  if (dec->use_skip_proba_) block->skip_ = VP8GetBit(br, dec->skip_p_);
+  if (dec->use_skip_proba_) block->skip_ = VP8GetBit(br, dec->skip_p_, "skip");

-  block->is_i4x4_ = !VP8GetBit(br, 145);   // decide for B_PRED first
+  block->is_i4x4_ = !VP8GetBit(br, 145, "block-size");
  if (!block->is_i4x4_) {
    // Hardcoded 16x16 intra-mode decision tree.
    const int ymode =
-        VP8GetBit(br, 156) ? (VP8GetBit(br, 128) ? TM_PRED : H_PRED)
-                           : (VP8GetBit(br, 163) ? V_PRED : DC_PRED);
+        VP8GetBit(br, 156, "pred-modes") ?
+            (VP8GetBit(br, 128, "pred-modes") ? TM_PRED : H_PRED) :
+            (VP8GetBit(br, 163, "pred-modes") ? V_PRED : DC_PRED);
    block->imodes_[0] = ymode;
    memset(top, ymode, 4 * sizeof(*top));
    memset(left, ymode, 4 * sizeof(*left));
@@ -323,22 +324,25 @@ static void ParseIntraMode(VP8BitReader* const br,
        const uint8_t* const prob = kBModesProba[top[x]][ymode];
 #if (USE_GENERIC_TREE == 1)
        // Generic tree-parsing
-        int i = kYModesIntra4[VP8GetBit(br, prob[0])];
+        int i = kYModesIntra4[VP8GetBit(br, prob[0], "pred-modes")];
        while (i > 0) {
-          i = kYModesIntra4[2 * i + VP8GetBit(br, prob[i])];
+          i = kYModesIntra4[2 * i + VP8GetBit(br, prob[i], "pred-modes")];
        }
        ymode = -i;
 #else
        // Hardcoded tree parsing
-        ymode = !VP8GetBit(br, prob[0]) ? B_DC_PRED :
-                  !VP8GetBit(br, prob[1]) ? B_TM_PRED :
-                    !VP8GetBit(br, prob[2]) ? B_VE_PRED :
-                      !VP8GetBit(br, prob[3]) ?
-                        (!VP8GetBit(br, prob[4]) ? B_HE_PRED :
-                          (!VP8GetBit(br, prob[5]) ? B_RD_PRED : B_VR_PRED)) :
-                        (!VP8GetBit(br, prob[6]) ? B_LD_PRED :
-                          (!VP8GetBit(br, prob[7]) ? B_VL_PRED :
-                            (!VP8GetBit(br, prob[8]) ? B_HD_PRED : B_HU_PRED)));
+        ymode = !VP8GetBit(br, prob[0], "pred-modes") ? B_DC_PRED :
+                  !VP8GetBit(br, prob[1], "pred-modes") ? B_TM_PRED :
+                    !VP8GetBit(br, prob[2], "pred-modes") ? B_VE_PRED :
+                      !VP8GetBit(br, prob[3], "pred-modes") ?
+                        (!VP8GetBit(br, prob[4], "pred-modes") ? B_HE_PRED :
+                          (!VP8GetBit(br, prob[5], "pred-modes") ? B_RD_PRED
+                                                                 : B_VR_PRED)) :
+                        (!VP8GetBit(br, prob[6], "pred-modes") ? B_LD_PRED :
+                          (!VP8GetBit(br, prob[7], "pred-modes") ? B_VL_PRED :
+                            (!VP8GetBit(br, prob[8], "pred-modes") ? B_HD_PRED
+                                                                   : B_HU_PRED))
+                        );
 #endif  // USE_GENERIC_TREE
        top[x] = ymode;
      }
@@ -348,9 +352,9 @@ static void ParseIntraMode(VP8BitReader* const br,
    }
  }
  // Hardcoded UVMode decision tree
-  block->uvmode_ = !VP8GetBit(br, 142) ? DC_PRED
-                 : !VP8GetBit(br, 114) ? V_PRED
-                 : VP8GetBit(br, 183) ? TM_PRED : H_PRED;
+  block->uvmode_ = !VP8GetBit(br, 142, "pred-modes-uv") ? DC_PRED
+                 : !VP8GetBit(br, 114, "pred-modes-uv") ? V_PRED
+                 : VP8GetBit(br, 183, "pred-modes-uv") ? TM_PRED : H_PRED;
 }

 int VP8ParseIntraModeRow(VP8BitReader* const br, VP8Decoder* const dec) {
@@ -514,8 +518,10 @@ void VP8ParseProba(VP8BitReader* const br, VP8Decoder* const dec) {
    for (b = 0; b < NUM_BANDS; ++b) {
      for (c = 0; c < NUM_CTX; ++c) {
        for (p = 0; p < NUM_PROBAS; ++p) {
-          const int v = VP8GetBit(br, CoeffsUpdateProba[t][b][c][p]) ?
-                        VP8GetValue(br, 8) : CoeffsProba0[t][b][c][p];
+          const int v =
+              VP8GetBit(br, CoeffsUpdateProba[t][b][c][p], "global-header") ?
+                        VP8GetValue(br, 8, "global-header") :
+                        CoeffsProba0[t][b][c][p];
          proba->bands_[t][b].probas_[c][p] = v;
        }
      }
@@ -524,9 +530,8 @@ void VP8ParseProba(VP8BitReader* const br, VP8Decoder* const dec) {
      proba->bands_ptr_[t][b] = &proba->bands_[t][kBands[b]];
    }
  }
-  dec->use_skip_proba_ = VP8Get(br);
+  dec->use_skip_proba_ = VP8Get(br, "global-header");
  if (dec->use_skip_proba_) {
-    dec->skip_p_ = VP8GetValue(br, 8);
+    dec->skip_p_ = VP8GetValue(br, 8, "global-header");
  }
 }
-
@@ -161,23 +161,26 @@ static int ParseSegmentHeader(VP8BitReader* br,
                              VP8SegmentHeader* hdr, VP8Proba* proba) {
  assert(br != NULL);
  assert(hdr != NULL);
-  hdr->use_segment_ = VP8Get(br);
+  hdr->use_segment_ = VP8Get(br, "global-header");
  if (hdr->use_segment_) {
-    hdr->update_map_ = VP8Get(br);
-    if (VP8Get(br)) {   // update data
+    hdr->update_map_ = VP8Get(br, "global-header");
+    if (VP8Get(br, "global-header")) {   // update data
      int s;
-      hdr->absolute_delta_ = VP8Get(br);
+      hdr->absolute_delta_ = VP8Get(br, "global-header");
      for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
-        hdr->quantizer_[s] = VP8Get(br) ? VP8GetSignedValue(br, 7) : 0;
+        hdr->quantizer_[s] = VP8Get(br, "global-header") ?
+            VP8GetSignedValue(br, 7, "global-header") : 0;
      }
      for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
-        hdr->filter_strength_[s] = VP8Get(br) ? VP8GetSignedValue(br, 6) : 0;
+        hdr->filter_strength_[s] = VP8Get(br, "global-header") ?
+            VP8GetSignedValue(br, 6, "global-header") : 0;
      }
    }
    if (hdr->update_map_) {
      int s;
      for (s = 0; s < MB_FEATURE_TREE_PROBS; ++s) {
-        proba->segments_[s] = VP8Get(br) ? VP8GetValue(br, 8) : 255u;
+        proba->segments_[s] = VP8Get(br, "global-header") ?
+            VP8GetValue(br, 8, "global-header") : 255u;
      }
    }
  } else {
@@ -205,7 +208,7 @@ static VP8StatusCode ParsePartitions(VP8Decoder* const dec,
  size_t last_part;
  size_t p;

-  dec->num_parts_minus_one_ = (1 << VP8GetValue(br, 2)) - 1;
+  dec->num_parts_minus_one_ = (1 << VP8GetValue(br, 2, "global-header")) - 1;
  last_part = dec->num_parts_minus_one_;
  if (size < 3 * last_part) {
    // we can't even read the sizes with sz[]! That's a failure.
@@ -229,21 +232,21 @@ static VP8StatusCode ParsePartitions(VP8Decoder* const dec,
 // Paragraph 9.4
 static int ParseFilterHeader(VP8BitReader* br, VP8Decoder* const dec) {
  VP8FilterHeader* const hdr = &dec->filter_hdr_;
-  hdr->simple_    = VP8Get(br);
-  hdr->level_     = VP8GetValue(br, 6);
-  hdr->sharpness_ = VP8GetValue(br, 3);
-  hdr->use_lf_delta_ = VP8Get(br);
+  hdr->simple_    = VP8Get(br, "global-header");
+  hdr->level_     = VP8GetValue(br, 6, "global-header");
+  hdr->sharpness_ = VP8GetValue(br, 3, "global-header");
+  hdr->use_lf_delta_ = VP8Get(br, "global-header");
  if (hdr->use_lf_delta_) {
-    if (VP8Get(br)) {   // update lf-delta?
+    if (VP8Get(br, "global-header")) {   // update lf-delta?
      int i;
      for (i = 0; i < NUM_REF_LF_DELTAS; ++i) {
-        if (VP8Get(br)) {
-          hdr->ref_lf_delta_[i] = VP8GetSignedValue(br, 6);
+        if (VP8Get(br, "global-header")) {
+          hdr->ref_lf_delta_[i] = VP8GetSignedValue(br, 6, "global-header");
        }
      }
      for (i = 0; i < NUM_MODE_LF_DELTAS; ++i) {
-        if (VP8Get(br)) {
-          hdr->mode_lf_delta_[i] = VP8GetSignedValue(br, 6);
+        if (VP8Get(br, "global-header")) {
+          hdr->mode_lf_delta_[i] = VP8GetSignedValue(br, 6, "global-header");
        }
      }
    }
@@ -352,8 +355,8 @@ int VP8GetHeaders(VP8Decoder* const dec, VP8Io* const io) {
  buf_size -= frm_hdr->partition_length_;

  if (frm_hdr->key_frame_) {
-    pic_hdr->colorspace_ = VP8Get(br);
-    pic_hdr->clamp_type_ = VP8Get(br);
+    pic_hdr->colorspace_ = VP8Get(br, "global-header");
+    pic_hdr->clamp_type_ = VP8Get(br, "global-header");
  }
  if (!ParseSegmentHeader(br, &dec->segment_hdr_, &dec->proba_)) {
    return VP8SetError(dec, VP8_STATUS_BITSTREAM_ERROR,
@@ -378,7 +381,7 @@ int VP8GetHeaders(VP8Decoder* const dec, VP8Io* const io) {
                       "Not a key frame.");
  }

-  VP8Get(br);   // ignore the value of update_proba_
+  VP8Get(br, "global-header");   // ignore the value of update_proba_

  VP8ParseProba(br, dec);

@@ -403,28 +406,28 @@ static const uint8_t kZigzag[16] = {
 // See section 13-2: http://tools.ietf.org/html/rfc6386#section-13.2
 static int GetLargeValue(VP8BitReader* const br, const uint8_t* const p) {
  int v;
-  if (!VP8GetBit(br, p[3])) {
-    if (!VP8GetBit(br, p[4])) {
+  if (!VP8GetBit(br, p[3], "coeffs")) {
+    if (!VP8GetBit(br, p[4], "coeffs")) {
      v = 2;
    } else {
-      v = 3 + VP8GetBit(br, p[5]);
+      v = 3 + VP8GetBit(br, p[5], "coeffs");
    }
  } else {
-    if (!VP8GetBit(br, p[6])) {
-      if (!VP8GetBit(br, p[7])) {
-        v = 5 + VP8GetBit(br, 159);
+    if (!VP8GetBit(br, p[6], "coeffs")) {
+      if (!VP8GetBit(br, p[7], "coeffs")) {
+        v = 5 + VP8GetBit(br, 159, "coeffs");
      } else {
-        v = 7 + 2 * VP8GetBit(br, 165);
-        v += VP8GetBit(br, 145);
+        v = 7 + 2 * VP8GetBit(br, 165, "coeffs");
+        v += VP8GetBit(br, 145, "coeffs");
      }
    } else {
      const uint8_t* tab;
-      const int bit1 = VP8GetBit(br, p[8]);
-      const int bit0 = VP8GetBit(br, p[9 + bit1]);
+      const int bit1 = VP8GetBit(br, p[8], "coeffs");
+      const int bit0 = VP8GetBit(br, p[9 + bit1], "coeffs");
      const int cat = 2 * bit1 + bit0;
      v = 0;
      for (tab = kCat3456[cat]; *tab; ++tab) {
-        v += v + VP8GetBit(br, *tab);
+        v += v + VP8GetBit(br, *tab, "coeffs");
      }
      v += 3 + (8 << cat);
    }
@@ -438,24 +441,24 @@ static int GetCoeffsFast(VP8BitReader* const br,
                         int ctx, const quant_t dq, int n, int16_t* out) {
  const uint8_t* p = prob[n]->probas_[ctx];
  for (; n < 16; ++n) {
-    if (!VP8GetBit(br, p[0])) {
+    if (!VP8GetBit(br, p[0], "coeffs")) {
      return n;  // previous coeff was last non-zero coeff
    }
-    while (!VP8GetBit(br, p[1])) {       // sequence of zero coeffs
+    while (!VP8GetBit(br, p[1], "coeffs")) {       // sequence of zero coeffs
      p = prob[++n]->probas_[0];
      if (n == 16) return 16;
    }
    {        // non zero coeff
      const VP8ProbaArray* const p_ctx = &prob[n + 1]->probas_[0];
      int v;
-      if (!VP8GetBit(br, p[2])) {
+      if (!VP8GetBit(br, p[2], "coeffs")) {
        v = 1;
        p = p_ctx[1];
      } else {
        v = GetLargeValue(br, p);
        p = p_ctx[2];
      }
-      out[kZigzag[n]] = VP8GetSigned(br, v) * dq[n > 0];
+      out[kZigzag[n]] = VP8GetSigned(br, v, "coeffs") * dq[n > 0];
    }
  }
  return 16;
@@ -468,24 +471,24 @@ static int GetCoeffsAlt(VP8BitReader* const br,
                        int ctx, const quant_t dq, int n, int16_t* out) {
  const uint8_t* p = prob[n]->probas_[ctx];
  for (; n < 16; ++n) {
-    if (!VP8GetBitAlt(br, p[0])) {
+    if (!VP8GetBitAlt(br, p[0], "coeffs")) {
      return n;  // previous coeff was last non-zero coeff
    }
-    while (!VP8GetBitAlt(br, p[1])) {       // sequence of zero coeffs
+    while (!VP8GetBitAlt(br, p[1], "coeffs")) {       // sequence of zero coeffs
      p = prob[++n]->probas_[0];
      if (n == 16) return 16;
    }
    {        // non zero coeff
      const VP8ProbaArray* const p_ctx = &prob[n + 1]->probas_[0];
      int v;
-      if (!VP8GetBitAlt(br, p[2])) {
+      if (!VP8GetBitAlt(br, p[2], "coeffs")) {
        v = 1;
        p = p_ctx[1];
      } else {
        v = GetLargeValue(br, p);
        p = p_ctx[2];
      }
-      out[kZigzag[n]] = VP8GetSigned(br, v) * dq[n > 0];
+      out[kZigzag[n]] = VP8GetSigned(br, v, "coeffs") * dq[n > 0];
    }
  }
  return 16;
@@ -31,8 +31,8 @@ extern "C" {

 // version numbers
 #define DEC_MAJ_VERSION 1
-#define DEC_MIN_VERSION 0
-#define DEC_REV_VERSION 2
+#define DEC_MIN_VERSION 1
+#define DEC_REV_VERSION 0

 // YUV-cache parameters. Cache is 32-bytes wide (= one cacheline).
 // Constraints are: We need to store one 16x16 block of luma samples (y),
@@ -362,12 +362,8 @@ 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;
+  HuffmanCode* huffman_table = NULL;
  int num_htree_groups = 1;
  int num_htree_groups_max = 1;
  int max_alphabet_size = 0;
@@ -418,12 +414,6 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
        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;
    }
@@ -453,63 +443,71 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
    goto Error;
  }

-  next = huffman_tables;
+  huffman_table = huffman_tables;
  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] = huffman_tables_i;
-      if (j == 0 && color_cache_bits > 0) {
-        alphabet_size += 1 << color_cache_bits;
-      }
-      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 = (huffman_tables_i->bits == 0);
-      }
-      total_size += huffman_tables_i->bits;
-      huffman_tables_i += size;
-      if (j <= ALPHA) {
-        int local_max_bits = code_lengths[0];
-        int k;
-        for (k = 1; k < alphabet_size; ++k) {
-          if (code_lengths[k] > local_max_bits) {
-            local_max_bits = code_lengths[k];
-          }
+    // If the index "i" is unused in the Huffman image, just make sure the
+    // coefficients are valid but do not store them.
+    if (mapping != NULL && mapping[i] == -1) {
+      for (j = 0; j < HUFFMAN_CODES_PER_META_CODE; ++j) {
+        int alphabet_size = kAlphabetSize[j];
+        if (j == 0 && color_cache_bits > 0) {
+          alphabet_size += (1 << color_cache_bits);
+        }
+        // Passing in NULL so that nothing gets filled.
+        if (!ReadHuffmanCode(alphabet_size, dec, code_lengths, NULL)) {
+          goto Error;
        }
-        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;
-      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;
+    } else {
+      HTreeGroup* const htree_group =
+          &htree_groups[(mapping == NULL) ? i : mapping[i]];
+      HuffmanCode** const htrees = htree_group->htrees;
+      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] = huffman_table;
+        if (j == 0 && color_cache_bits > 0) {
+          alphabet_size += (1 << color_cache_bits);
+        }
+        size = ReadHuffmanCode(alphabet_size, dec, code_lengths, huffman_table);
+        if (size == 0) {
+          goto Error;
+        }
+        if (is_trivial_literal && kLiteralMap[j] == 1) {
+          is_trivial_literal = (huffman_table->bits == 0);
+        }
+        total_size += huffman_table->bits;
+        huffman_table += size;
+        if (j <= ALPHA) {
+          int local_max_bits = code_lengths[0];
+          int k;
+          for (k = 1; k < alphabet_size; ++k) {
+            if (code_lengths[k] > local_max_bits) {
+              local_max_bits = code_lengths[k];
+            }
+          }
+          max_bits += local_max_bits;
+        }
      }
+      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;
+        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);
+      if (htree_group->use_packed_table) BuildPackedTable(htree_group);
    }
-    htree_group->use_packed_table =
-        !htree_group->is_trivial_code && (max_bits < HUFFMAN_PACKED_BITS);
-    if (htree_group->use_packed_table) BuildPackedTable(htree_group);
  }
  ok = 1;

@@ -521,7 +519,6 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,

 Error:
  WebPSafeFree(code_lengths);
-  WebPSafeFree(huffman_tables_bogus);
  WebPSafeFree(mapping);
  if (!ok) {
    WebPSafeFree(huffman_image);
@@ -757,11 +754,11 @@ static WEBP_INLINE HTreeGroup* GetHtreeGroupForPos(VP8LMetadata* const hdr,

 typedef void (*ProcessRowsFunc)(VP8LDecoder* const dec, int row);

-static void ApplyInverseTransforms(VP8LDecoder* const dec, int num_rows,
+static void ApplyInverseTransforms(VP8LDecoder* const dec,
+                                   int start_row, int num_rows,
                                   const uint32_t* const rows) {
  int n = dec->next_transform_;
  const int cache_pixs = dec->width_ * num_rows;
-  const int start_row = dec->last_row_;
  const int end_row = start_row + num_rows;
  const uint32_t* rows_in = rows;
  uint32_t* const rows_out = dec->argb_cache_;
@@ -792,8 +789,7 @@ static void ProcessRows(VP8LDecoder* const dec, int row) {
    VP8Io* const io = dec->io_;
    uint8_t* rows_data = (uint8_t*)dec->argb_cache_;
    const int in_stride = io->width * sizeof(uint32_t);  // in unit of RGBA
-
-    ApplyInverseTransforms(dec, num_rows, rows);
+    ApplyInverseTransforms(dec, dec->last_row_, num_rows, rows);
    if (!SetCropWindow(io, dec->last_row_, row, &rows_data, in_stride)) {
      // Nothing to output (this time).
    } else {
@@ -1196,6 +1192,7 @@ static int DecodeImageData(VP8LDecoder* const dec, uint32_t* const data,
      VP8LFillBitWindow(br);
      dist_code = GetCopyDistance(dist_symbol, br);
      dist = PlaneCodeToDistance(width, dist_code);
+
      if (VP8LIsEndOfStream(br)) break;
      if (src - data < (ptrdiff_t)dist || src_end - src < (ptrdiff_t)length) {
        goto Error;
@@ -1556,7 +1553,7 @@ static void ExtractAlphaRows(VP8LDecoder* const dec, int last_row) {
    const int cache_pixs = width * num_rows_to_process;
    uint8_t* const dst = output + width * cur_row;
    const uint32_t* const src = dec->argb_cache_;
-    ApplyInverseTransforms(dec, num_rows_to_process, in);
+    ApplyInverseTransforms(dec, cur_row, num_rows_to_process, in);
    WebPExtractGreen(src, dst, cache_pixs);
    AlphaApplyFilter(alph_dec,
                     cur_row, cur_row + num_rows_to_process, dst, width);
@@ -37,7 +37,7 @@ struct VP8LTransform {
  int                    bits_;   // subsampling bits defining transform window.
  int                    xsize_;  // transform window X index.
  int                    ysize_;  // transform window Y index.
-  uint32_t              *data_;   // transform data.
+  uint32_t*              data_;   // transform data.
 };

 typedef struct {
@@ -48,23 +48,23 @@ typedef struct {
  int             huffman_mask_;
  int             huffman_subsample_bits_;
  int             huffman_xsize_;
-  uint32_t       *huffman_image_;
+  uint32_t*       huffman_image_;
  int             num_htree_groups_;
-  HTreeGroup     *htree_groups_;
-  HuffmanCode    *huffman_tables_;
+  HTreeGroup*     htree_groups_;
+  HuffmanCode*    huffman_tables_;
 } VP8LMetadata;

 typedef struct VP8LDecoder VP8LDecoder;
 struct VP8LDecoder {
  VP8StatusCode    status_;
  VP8LDecodeState  state_;
-  VP8Io           *io_;
+  VP8Io*           io_;

-  const WebPDecBuffer *output_;    // shortcut to io->opaque->output
+  const WebPDecBuffer* output_;    // shortcut to io->opaque->output

-  uint32_t        *pixels_;        // Internal data: either uint8_t* for alpha
+  uint32_t*        pixels_;        // Internal data: either uint8_t* for alpha
                                   // or uint32_t* for BGRA.
-  uint32_t        *argb_cache_;    // Scratch buffer for temporary BGRA storage.
+  uint32_t*        argb_cache_;    // Scratch buffer for temporary BGRA storage.

  VP8LBitReader    br_;
  int              incremental_;   // if true, incremental decoding is expected
@@ -86,8 +86,8 @@ struct VP8LDecoder {
  // or'd bitset storing the transforms types.
  uint32_t         transforms_seen_;

-  uint8_t         *rescaler_memory;  // Working memory for rescaling work.
-  WebPRescaler    *rescaler;         // Common rescaler for all channels.
+  uint8_t*         rescaler_memory;  // Working memory for rescaling work.
+  WebPRescaler*    rescaler;         // Common rescaler for all channels.
 };

 //------------------------------------------------------------------------------
@@ -24,8 +24,8 @@
 #include "src/webp/format_constants.h"

 #define DMUX_MAJ_VERSION 1
-#define DMUX_MIN_VERSION 0
-#define DMUX_REV_VERSION 2
+#define DMUX_MIN_VERSION 1
+#define DMUX_REV_VERSION 0

 typedef struct {
  size_t start_;        // start location of the data
@@ -214,7 +214,7 @@ static void ApplyAlphaMultiply_SSE2(uint8_t* rgba, int alpha_first,
 // Alpha detection

 static int HasAlpha8b_SSE2(const uint8_t* src, int length) {
-  const __m128i all_0xff = _mm_set1_epi8(0xff);
+  const __m128i all_0xff = _mm_set1_epi8((char)0xff);
  int i = 0;
  for (; i + 16 <= length; i += 16) {
    const __m128i v = _mm_loadu_si128((const __m128i*)(src + i));
@@ -228,7 +228,7 @@ static int HasAlpha8b_SSE2(const uint8_t* src, int length) {

 static int HasAlpha32b_SSE2(const uint8_t* src, int length) {
  const __m128i alpha_mask = _mm_set1_epi32(0xff);
-  const __m128i all_0xff = _mm_set1_epi8(0xff);
+  const __m128i all_0xff = _mm_set1_epi8((char)0xff);
  int i = 0;
  // We don't know if we can access the last 3 bytes after the last alpha
  // value 'src[4 * length - 4]' (because we don't know if alpha is the first
@@ -173,8 +173,8 @@ static int AndroidCPUInfo(CPUFeature feature) {
  const AndroidCpuFamily cpu_family = android_getCpuFamily();
  const uint64_t cpu_features = android_getCpuFeatures();
  if (feature == kNEON) {
-    return (cpu_family == ANDROID_CPU_FAMILY_ARM &&
-            0 != (cpu_features & ANDROID_CPU_ARM_FEATURE_NEON));
+    return cpu_family == ANDROID_CPU_FAMILY_ARM &&
+           (cpu_features & ANDROID_CPU_ARM_FEATURE_NEON) != 0;
  }
  return 0;
 }
@@ -1361,7 +1361,8 @@ static void RD4_NEON(uint8_t* dst) {   // Down-right
  const uint32_t J = dst[-1 + 1 * BPS];
  const uint32_t K = dst[-1 + 2 * BPS];
  const uint32_t L = dst[-1 + 3 * BPS];
-  const uint64x1_t LKJI____ = vcreate_u64(L | (K << 8) | (J << 16) | (I << 24));
+  const uint64x1_t LKJI____ =
+      vcreate_u64((uint64_t)L | (K << 8) | (J << 16) | (I << 24));
  const uint64x1_t LKJIXABC = vorr_u64(LKJI____, ____XABC);
  const uint8x8_t KJIXABC_ = vreinterpret_u8_u64(vshr_n_u64(LKJIXABC, 8));
  const uint8x8_t JIXABC__ = vreinterpret_u8_u64(vshr_n_u64(LKJIXABC, 16));
@@ -1427,10 +1428,16 @@ static WEBP_INLINE void DC8_NEON(uint8_t* dst, int do_top, int do_left) {

  if (do_top) {
    const uint8x8_t A = vld1_u8(dst - BPS);  // top row
+#if defined(__aarch64__)
+    const uint16x8_t B = vmovl_u8(A);
+    const uint16_t p2 = vaddvq_u16(B);
+    sum_top = vdupq_n_u16(p2);
+#else
    const uint16x4_t p0 = vpaddl_u8(A);  // cascading summation of the top
    const uint16x4_t p1 = vpadd_u16(p0, p0);
    const uint16x4_t p2 = vpadd_u16(p1, p1);
    sum_top = vcombine_u16(p2, p2);
+#endif
  }

  if (do_left) {
@@ -326,7 +326,7 @@ static WEBP_INLINE void Update2Pixels_SSE2(__m128i* const pi, __m128i* const qi,
  const __m128i a1_lo = _mm_srai_epi16(*a0_lo, 7);
  const __m128i a1_hi = _mm_srai_epi16(*a0_hi, 7);
  const __m128i delta = _mm_packs_epi16(a1_lo, a1_hi);
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
+  const __m128i sign_bit = _mm_set1_epi8((char)0x80);
  *pi = _mm_adds_epi8(*pi, delta);
  *qi = _mm_subs_epi8(*qi, delta);
  FLIP_SIGN_BIT2(*pi, *qi);
@@ -338,9 +338,9 @@ static WEBP_INLINE void NeedsFilter_SSE2(const __m128i* const p1,
                                         const __m128i* const q0,
                                         const __m128i* const q1,
                                         int thresh, __m128i* const mask) {
-  const __m128i m_thresh = _mm_set1_epi8(thresh);
+  const __m128i m_thresh = _mm_set1_epi8((char)thresh);
  const __m128i t1 = MM_ABS(*p1, *q1);        // abs(p1 - q1)
-  const __m128i kFE = _mm_set1_epi8(0xFE);
+  const __m128i kFE = _mm_set1_epi8((char)0xFE);
  const __m128i t2 = _mm_and_si128(t1, kFE);  // set lsb of each byte to zero
  const __m128i t3 = _mm_srli_epi16(t2, 1);   // abs(p1 - q1) / 2

@@ -360,7 +360,7 @@ static WEBP_INLINE void DoFilter2_SSE2(__m128i* const p1, __m128i* const p0,
                                       __m128i* const q0, __m128i* const q1,
                                       int thresh) {
  __m128i a, mask;
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
+  const __m128i sign_bit = _mm_set1_epi8((char)0x80);
  // convert p1/q1 to int8_t (for GetBaseDelta_SSE2)
  const __m128i p1s = _mm_xor_si128(*p1, sign_bit);
  const __m128i q1s = _mm_xor_si128(*q1, sign_bit);
@@ -380,7 +380,7 @@ static WEBP_INLINE void DoFilter4_SSE2(__m128i* const p1, __m128i* const p0,
                                       const __m128i* const mask,
                                       int hev_thresh) {
  const __m128i zero = _mm_setzero_si128();
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
+  const __m128i sign_bit = _mm_set1_epi8((char)0x80);
  const __m128i k64 = _mm_set1_epi8(64);
  const __m128i k3 = _mm_set1_epi8(3);
  const __m128i k4 = _mm_set1_epi8(4);
@@ -427,7 +427,7 @@ static WEBP_INLINE void DoFilter6_SSE2(__m128i* const p2, __m128i* const p1,
                                       const __m128i* const mask,
                                       int hev_thresh) {
  const __m128i zero = _mm_setzero_si128();
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
+  const __m128i sign_bit = _mm_set1_epi8((char)0x80);
  __m128i a, not_hev;

  // compute hev mask
@@ -941,7 +941,7 @@ static void VR4_SSE2(uint8_t* dst) {   // Vertical-Right
  const __m128i ABCD0 = _mm_srli_si128(XABCD, 1);
  const __m128i abcd = _mm_avg_epu8(XABCD, ABCD0);
  const __m128i _XABCD = _mm_slli_si128(XABCD, 1);
-  const __m128i IXABCD = _mm_insert_epi16(_XABCD, I | (X << 8), 0);
+  const __m128i IXABCD = _mm_insert_epi16(_XABCD, (short)(I | (X << 8)), 0);
  const __m128i avg1 = _mm_avg_epu8(IXABCD, ABCD0);
  const __m128i lsb = _mm_and_si128(_mm_xor_si128(IXABCD, ABCD0), one);
  const __m128i avg2 = _mm_subs_epu8(avg1, lsb);
@@ -246,9 +246,9 @@ extern VP8Fdct VP8FTransform2;   // performs two transforms at a time
 extern VP8WHT VP8FTransformWHT;
 // Predictions
 // *dst is the destination block. *top and *left can be NULL.
-typedef void (*VP8IntraPreds)(uint8_t *dst, const uint8_t* left,
+typedef void (*VP8IntraPreds)(uint8_t* dst, const uint8_t* left,
                              const uint8_t* top);
-typedef void (*VP8Intra4Preds)(uint8_t *dst, const uint8_t* top);
+typedef void (*VP8Intra4Preds)(uint8_t* dst, const uint8_t* top);
 extern VP8Intra4Preds VP8EncPredLuma4;
 extern VP8IntraPreds VP8EncPredLuma16;
 extern VP8IntraPreds VP8EncPredChroma8;
--- a/Show More
+++ b/Show More