@bitCast from integer NaN representation to float NaN resulted in
changed bits in float. This only happened with signaled NaN.
- added test for signaled NaN
- added tests for quiet NaN (for completeness)
closes#14198
Using zig cc with CMake on Windows was failing during compiler
detection. -nostdinc was causing the crt not to be linked, and Coff/lld.zig
assumed that wWinMainCRTStartup would be present in this case.
-nostdlib did not prevent the default behaviour of linking libc++ when
zig c++ was used. This caused libc++ to be built when CMake ran
ABI detection using zig c++, which fails as libcxxabi cannot compile
under MSVC.
- Change the behaviour of COFF -nostdinc to set /entry to the function that the
default CRT method for the specified subsystem would have called.
- Fix -ENTRY being passed twice if it was specified explicitly and -nostdlib was present.
- Add support for /pdb, /version, /implib, and /subsystem as linker args (passed by CMake)
- Remove -Ddisable-zstd, no longer needed
- Add -Ddisable-libcpp for use when bootstrapping on msvc
- add support for passing through .def files to the linker,
required for building libLTO.dll in LLVM
- fixup libcpp linking conditionals
- add option to skip linking zstd for use in bootstrapping (when
building against an LLVM with LLVM_ENABLE_ZSTD=OFF)
- Add an assert that an exclusive lock is help to writeManifest
- Only call writeManifest in updateCObject if an exclusive lock is held
- cache: fixup test to verify hits don't take an exclusive lock, instead of writing the manifest
Before, --color on would affect colored compile error printing but not
affect terminal progress bar printing. It was intended for this option
to affect both; now it does.
This causes a failure when building the language reference, which
contains code for parsing terminal output and rendering HTML. Now it
must be expanded to handle 'K' and 'D' codes to simulate a terminal
cursor moving, and the CI will fail until that capability is added in a
later commit of this branch.
I extracted this change from #13560 so that the idea is not lost but we
can solve this issue separately.
When outputting the names section, we should output the actual symbol
name rather than the import name. This makes sure that symbols with
an explicit name set have the correct name but retain the import name
too.
We also now correctly mangle the name of an extern function with an
explicit library name. This ensures that functions that have a
different library name, but the same import/function name, can be
resolved correctly with other modules and don't resolve to the
same symbol.
Rather than checking for function pointers during the writing phase,
we now create a synethtic symbol when a new link job has started.
This means the symbol can correctly be resolved during link time
with the indirect function table from other object files, ensuring
we are properly performing relocations and our binary writer is now
unaware of any of its logic and simply emits the table according to
the symbol such as any other symbols.
Adds support for both the `-rdynamic` and the `--export=<value>`
flags. Support is added to both the incremental linker as well as
the traditional linker (zld).
- Add cpuid / getXCR0 functions for the cbe to use instead of asm blocks
- Don't cast between 128 bit types during truncation
- Fixup truncation to use functions for shifts / adds
- Fixup float casts for undefined values
- Add test for 128 bit integer truncation
- Fix zig_clz_u128 not respecting the bits argument. This was crashing the compile-rt addxf3 tests with the cbe
- Instead of redering a negation for negative 128 bit int literals, render the literal as twos complement. This allows
rendering int representations of floats correctly (specifically f80).
