diff options
| author | tsdgeos <aacid@kde.org> | 2019-09-26 18:17:32 +0200 |
|---|---|---|
| committer | jonathanmetzman <31354670+jonathanmetzman@users.noreply.github.com> | 2019-09-26 09:17:32 -0700 |
| commit | 58df29fd3f290fc25d7d7f61d4ce9ce782b1727b (patch) | |
| tree | 8f565b32230d08f90466917370e945e501559dd7 /projects/kimageformats | |
| parent | 198345dc17be486f587c8d609bc92c6affdcf9f3 (diff) | |
| download | oss-fuzz-58df29fd3f290fc25d7d7f61d4ce9ce782b1727b.tar.gz | |
Use 5.15 branch of Qt for now (#2888)
A direct clone now gives you dev which is future Qt 6, none of the
code using Qt expects Qt6 yet
Using 5.15 now (instead 5.12 before) means we have to disable
sanitize=vptr in parts of qt in kimageformats since those
parts now compile with no-rtti
While at it make the compilation script a bit better:
* Don't need to disable compressing rcc files, only needed to pass
the CFLAGS to the QMAKE_CFLAGS
* Also fix the "make qmake faster" sed command
Diffstat (limited to 'projects/kimageformats')
| -rw-r--r-- | projects/kimageformats/Dockerfile | 2 | ||||
| -rw-r--r-- | projects/kimageformats/build.sh | 12 |
2 files changed, 7 insertions, 7 deletions
diff --git a/projects/kimageformats/Dockerfile b/projects/kimageformats/Dockerfile index 5dd3dfac7..8e92b33a2 100644 --- a/projects/kimageformats/Dockerfile +++ b/projects/kimageformats/Dockerfile @@ -20,7 +20,7 @@ RUN apt-get install --yes cmake RUN git clone --depth 1 https://github.com/madler/zlib.git RUN git clone --depth 1 https://github.com/nih-at/libzip.git RUN git clone --depth 1 git://anongit.kde.org/extra-cmake-modules -RUN git clone --depth 1 git://code.qt.io/qt/qtbase.git +RUN git clone --depth 1 --branch=5.15 git://code.qt.io/qt/qtbase.git RUN git clone --depth 1 git://anongit.kde.org/karchive RUN git clone --depth 1 git://anongit.kde.org/kimageformats COPY build.sh $SRC diff --git a/projects/kimageformats/build.sh b/projects/kimageformats/build.sh index c4842e495..2928af297 100644 --- a/projects/kimageformats/build.sh +++ b/projects/kimageformats/build.sh @@ -15,13 +15,13 @@ make install -j$(nproc) cd $SRC cd qtbase -# add the flags to Qt build too, we may as well sanitize Qt too (and also fixes memory sanitizer build) -sed -i -e "s/QMAKE_CXXFLAGS += -stdlib=libc++/QMAKE_CXXFLAGS += -stdlib=libc++ $CXXFLAGS/g" mkspecs/linux-clang-libc++/qmake.conf +# add the flags to Qt build too +sed -i -e "s/QMAKE_CXXFLAGS += -stdlib=libc++/QMAKE_CXXFLAGS += -stdlib=libc++ $CXXFLAGS\nQMAKE_CFLAGS += $CFLAGS/g" mkspecs/linux-clang-libc++/qmake.conf sed -i -e "s/QMAKE_LFLAGS += -stdlib=libc++/QMAKE_LFLAGS += -stdlib=libc++ -lpthread $CXXFLAGS/g" mkspecs/linux-clang-libc++/qmake.conf -# make qmake compile faster TODO -sed -i -e "s/MAKE\")/MAKE\" -j10)/g" configure -# Disable compressing rcc files, triggers a warning in the memory sanitizer that i'm not sure is valid. TODO investigate properly -sed -i -e "s/DEFINES += QT_RCC QT_NO_CAST_FROM_ASCII QT_NO_FOREACH/DEFINES += QT_NO_COMPRESS QT_RCC QT_NO_CAST_FROM_ASCII QT_NO_FOREACH/g" src/tools/rcc/rcc.pro +# disable sanitize=vptr for harfbuzz since it compiles without rtti +sed -i -e "s/TARGET = qtharfbuzz/TARGET = qtharfbuzz\nQMAKE_CXXFLAGS += -fno-sanitize=vptr/g" src/3rdparty/harfbuzz-ng/harfbuzz-ng.pro +# make qmake compile faster +sed -i -e "s/MAKE\")/MAKE\" -j$(nproc))/g" configure ./configure --glib=no --libpng=qt -opensource -confirm-license -static -no-opengl -no-icu -platform linux-clang-libc++ -v cd src ../bin/qmake -o Makefile src.pro |