This makes sure that exiting by calling _exit or ExitProcess doesn't
run global objects' destructors, which happened since starting to use
__cxa_atexit since 11e04f0466.
This fixes LLD built for windows with this toolchain.
This version fixes issues with linking mingw-style SEH data (making
sure all unwind info is included), adds support for assembler
directives like .rva, which often is used in SEH contexts, and
updates clang's unwind.h to match the libgcc ABI for SEH unwinding.
This fixes potential pointer truncation on 64 bit windows.
(Existing tests have ran fine in practice; either these are codepaths
not exercised in current setups, or potential codepaths that could
be triggered with more extensive tests.)
Now that we can build compiler-rt and libunwind with a stripped
llvm install, we can build things in a few smaller steps without
ending up with a larger docker image than before.
This still inflates the docker image chain since it's done in
a separate RUN command though.
This tests that one still can (re)build
compiler-rt/libunwind/libcxxabi/libcxx after stripping down the llvm
installation.
For compiler-rt and libunwind, this makes them build successfully even
if llvm-config and the llvm cmake files aren't available.
For libcxxabi, this brings in support for using SEH exceptions.
For libcxx, this removes a few build time warnings.
When using a global C++ object, the destructor isn't handled via
the .dtors section, but the destructor is set up with atexit() once
the constructor has been called.
For x86 architectures, binutils objcopy and strip can be used as
such.
For armv7 and aarch64, binutils doesn't recognize the architectures
and refuse to touch the files. For most/everything that objcopy/strip
do, the actual architecture doesn't really matter though. Therefore,
provide a wrapper script and a minimal tool for editing PE executables,
that change the architecture from armv7 and aarch64 to i686 and x86_64
respectively before invoking binutils, and change it back afterwards.
In case the binutils tools are missing, just return a successful result
without doing anything for stripping operations (previously, we had the
<triplet>-strip tool symlinked to /bin/true, so this isn't any worse than
before) and fail for other objcopy operations.
This allows using GNU binutils strip and objcopy on executables with
dwarf debug info linked by lld.
This also improves clang's handling of detecting known libraries
that override default libraries to link.
This version includes the windows api sets version of the universal
crt, fixes a few other functions when used with the universal crt,
and fixes the widl tool when running on macos.
If installing on top of a prebuilt win32 release of llvm instead of
building from source, the nm and strings executables are missing,
and MSYS2 errors out when creating a symlink to a file that doesn't
exist (since it does a copy instead of making an actual symlink).
This fixes an issue with code generation for functions that needed
a stack probe on armv7, improves llvm-rc and adds a few more options
to the mingw lld frontend.
This fixes building resource files that refer to external files
relative to the source directory, without specifying any directory
via include flags to windres itself. This is needed since llvm-rc
doesn't see the original source file but only sees the preprocessed
one in a temporary directory.
Alternatively, the preprocessed file could be placed in the same
directory as the source file, but when building e.g. outside of the
source tree, writing files (even temporary ones) inside the source
tree isn't generally expected.
This version improves llvm-rc to support more rc script features
and features built-in filtering of CPP output in llvm-rc. It also
improves the handling of some def features in llvm-dlltool, allows
producing codeview debug information when targeting mingw and adds
handling of a few more ARM64 relocations in lld.
This avoids needing the clang++ symlink. On platforms without symlinks
(i.e. windows), clang.exe and clang++.exe are two separate but identical
executables. This allows getting rid of the latter, saving size.
When building libcxx, this is also already enabled via the cmake
option -DLIBCXX_HAS_WIN32_THREAD_API=ON.
This fixes building libcxxabi if pthreads are available.
In this version, the binary actually is named clang-7 instead of
clang-7.0. This requires extending the test a little for keeping such
binaries but removing other clang-foo tools.
gcc/libssp's configure script runs checks for flags that clang doesn't
implement. We actually just need to set a few HAVE defines and compile
the .c sources, so just check out this subdirectory using svn and build
it using a custom makefile.
As long as we pass --sysroot explicitly in clang-target-wrapper.sh,
this isn't necessary. And since clang SVN r329946. it's no longer
necessary when not specifying the sysroot (and having it is in
fact harmful).
This doesn't help reduce the docker image size since it's done
in a separate RUN step than the one that added the files. (compiler-rt
and libunwind currently fail to build after stripping the llvm
installation though.)
This version fixes issues with long double math functions on x86_64
(and some mingw-w64 float/double math functions are implemented with
a long double function internally), and adds support for the --kill-at
flag in lld.
This fixes issues with -mms-bitfields together with "long double"
on i686, and allows using the -static flag while building DLLs
(to link dependent libraries statically), and implements stack
probing for alloca calls on aarch64.
Martin Storsjö