Our development workflow tooling prevents trailing blank lines
from being added, but some such lines remain from before that
was enforced. Remove them to make it easier to rename files
without triggering enforcement.
Since commit 11da882a12 (Apple: Introduce separate system name for iOS,
tvOS, and watchOS, 2018-01-15, v3.14.0-rc1~14^2~1) our toolchain
inspection steps, like ABI detection, tell `try_compile` to use a
`STATIC_LIBRARY` instead of an `EXECUTABLE`. This was needed at the
time to avoid codesign requirements. However, commit d3a64c4e3f (Xcode:
Explicitly turn off signing in try_compile projects, 2020-07-16,
v3.19.0-rc1~483^2) introduced a more general solution to that problem.
Restore linking during toolchain inspection so that we can detect and
identify the linker.
Suggested-by: Marc Chevrier <marc.chevrier@gmail.com>
Fixes: #26443
When bootstrapping Debian for musl-linux-any, the multiarch directories
are not found due to an overly-specific regular expression.
See https://bugs.debian.org/1085507 for the original bug report.
The following platforms were missing:
* Android
* WindowsCE, WindowsPhone, WindowsStore and WindowsKernelModeDriver
And enhance various Apple platforms support (iOS, tvOS, etc...)
Fixes: #26439
On modern macOS, compiler wrappers like `/usr/bin/cc` automatically
choose a SDK to pass via `-isysroot` to an underlying compiler from
Xcode or the CommandLineTools. Other toolchains like Homebrew's `gcc-*`
come with a default SDK too. Therefore, when targeting macOS, we no
longer need to choose any SDK or pass an `-isysroot` flag by default.
Update initialization of `CMAKE_OSX_SYSROOT` to be empty by default when
targeting macOS.
Fixes: #19180
When using the WindowsKernelModeDriver, do not add the new paths
if the variable was not previously defined. Instead, append them
to the existing value.
Add a new GUID for the Experimental mode gate
Detect Windows Kernel-Mode Driver include directories and library search
paths from the WDK command-line environment. Require toolchain files to
specify the KMDF target version via a new variable.
Since this changes the behavior of the WindowsKernelModeDriver
experimental feature, update its UUID.
Since commit 0b552eb877 (MSVC: Embed manifests directly for
non-incremental vs_link_exe links, 2023-02-20, v3.27.0-rc1~438^2) we use
`/MANIFEST:EMBED`, but that is not supported by older MSVC linkers.
Restore use of the standalone manifest tool.
Update `Tests/MSManifest` to recognize `dpiAwareness` in the manifest
content as written by MSVC v100 tooling.
Fixes: #26179
`NDK_PROC_*_ABI` and `NDK_ARCH_*_ABI` were removed in the Android
NDK 27 but will be reintroduced in the Android NDK 27b:
* https://github.com/android/ndk/issues/2049
* https://android-review.googlesource.com/c/platform/ndk/+/3196345
Both are only used when `CMAKE_ANDROID_ARCH_ABI` is NOT given. But
currently the existence is also checked when `CMAKE_ANDROID_ARCH_ABI`
is given. So we move the checks to the position they are required.
A new set of files are dedicated to linker configuration.
This set of files enable a fine-tuned configuration based of the linker
type as identified during compiler detection.
Fixes: #25360
Remove `/DAMD64` and `/DARM64EC` flags added by commit 4ea3a88625 (MSVC:
Add support for targeting ARM64EC, 2020-12-30, v3.20.0-rc1~121^2) that
are not longer expected.
When using Gentoo Prefix on Debian, find_library for libm.so fails,
because it needs to locate the lib in $sysroot/usr/lib64/libm.so, i.e.
lib64/ not lib/.
To fix this, do not disable per-arch lib directories when a sysroot is
set.
Fixes: #26069
4af20bb794 NAG-Fortran: Do not repeat preprocessing with Ninja on Apple platforms
91bb8dd872 NAG-Fortran: Fix MODULE library creation on Apple platforms
e056006116 NAG-Fortran: Tell the Ninja generator how to preprocess Fortran sources
765a611956 NAG-Fortran: Added initial default compilation flags
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !9407
Wire up the flags needed to support module libraries built and used with
Swift. We need to pass `-bundle` to the linker when linking module
libraries on Darwin, and we need to pass `-export-dynamic` to the linker
when emitting an executable that exports symbols on Linux. This patch
wires up `CMAKE_SHARED_MODULE_CREATE_Swift_FLAGS` and
`CMAKE_SHARED_MODULE_LOADER_Swift_FLAG` on Darwin, and hooks up
`CMAKE_EXE_EXPORTS_Swift_FLAG` on Linux in order to support passing
things correctly.
We can't expose `CMAKE_EXE_LINKER_FLAGS` to Swift, as it contains flags
that the Swift compiler doesn't recognize, but the other
language-specific variables are safe to expose.
Thread-local variables are put in the tdata section on AIX / XCOFF [1]. Unfortunately
the ExportImportList script currently doesn't consider this section when looking for
symbols to export.
This results in linking failures in applications (such as LLVM) when building which expect these thread local variables to be exported from their shared libraries.
[1]: https://www.ibm.com/docs/en/aix/7.3?topic=formats-xcoff-object-file-format
The link flag wrapping added by commit 5d5a712303 (IntelLLVM: Pass
linker flags to the compiler when used as linker, 2022-05-27,
v3.25.0-rc1~344^2~4) is not needed with Visual Studio generators because
MSBuild invokes the linker directly. Furthermore, that commit omitted
wrapping of the per-config linker flags.
Revise the logic to add the link flag wrapping more precisely.
Fixes: #25533
Update the platform module added by commit 462fbd1eca (Add support for
ARTOS platform using GNU C with ac compiler driver, 2015-08-13,
v3.4.0-rc1~201^2) to tell `find_library` to consider library file names
starting in `lib`.
FindEXPAT needs this on ARTOS since commit 13079f382b (FindEXPAT: Fix
for debug and other modified builds, 2023-02-18, v3.27.0-rc1~463^2)
removed its explicit `libexpat` library name candidate.
Fixes: #25776
Extend commit 96a953b1ed (Add options to specify linker tool, 2023-09-27,
v3.29.0-rc1~577^2) with the `CMAKE_Swift_USING_LINKER_*` variables
needed to support `CMAKE_LINKER_TYPE` for Swift.
