Upstream libuv commits:
* node-v0.6.0~28 (linux: use ioctl(FIONBIO) to set O_NONBLOCK, saves a syscall, 2011-10-29)
* v1.11.0~11 (unix: use async-signal safe functions between fork and exec, 2016-12-12)
left an asymmetry in the way O_NONBLOCK is set and cleared. Normally
ioctl/FIONBIO is used for both. However, uv_spawn uses fcntl/O_NONBLOCK
to clear O_NONBLOCK between fork and exec. This fails on Linux/sparc64
where setting via ioctl/FIONBIO actually sets two bits (in O_NDELAY)
that cause clearing via fcntl/O_NONBLOCK to be silently ignored (though
clearing via ioctl/FIONBIO or fcntl/O_NDELAY works).
Since CMake commit v3.11.0-rc1~117^2 (CTest: Re-implement test process
handling using libuv, 2017-12-10), CTest uses libuv. On Linux/sparc64
child processes have been started with non-blocking output pipes. This
can lead to write errors or lost output from children not prepared to
deal with `EAGAIN` on stdout or stderr.
Fix this for now by switching libuv back to fcntl/O_NONBLOCK to set and
clear O_NONBLOCK when compiled on Linux/sparc64.
LibUV-Issue: 1830
Fixes: #17941
Extend our hack section of mappings to work around IWYU incorrectly
requiring:
* bits/stdint-uintn.h for stdint.h
* bits/stdint-intn.h for stdint.h
* bits/types/mbstate_t.h for wchar.h
b5e21d7d CTest: Re-implement test process handling using libuv
fcebff75 cmProcess: Use explicit enum for process exit exception
3dd2edf4 cmProcess: Use explicit enum for process state
5238e6db cmProcess: Remove unused ReportStatus method
c13b68e6 cmCTestRunTest: Modernize constructor and destructor decls
4d6b0903 cmCTestRunTest: Drop unused members
05da65bc cmCTestMultiProcessHandler: Factor out duplicate test finish logic
dd945345 cmCTestMultiProcessHandler: Add helper to make libuv use SA_RESTART
...
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !1455
BSD `signal(2)` semantics make some system calls (e.g. for `write`)
restartable when interrupted by a signal handler. Use `SA_RESTART` to
enable these semantics everywhere that supports them.
This is required by C++ stream libraries that interpret `EINTR` as any
other error, set `badbit`, and stop writing. I've observed this with
`libstdc++` during a `std::cout.flush()` call interrupted by `SIGCHLD`.
This commit continues the refactoring of CTest to adopt std::chrono.
After the last sets of changes that introduced std::chrono::steady_clock
and std::chrono::system_clock respectively, it makes sense to have all
the timeouts be stored as std::chrono::duration.
No functional change intended.
IWYU incorrectly classifies this internal STL type as not internal, and
suggests including `<type_traits>` for it. Work around the problem by
mapping the offending names to a file that we always include.
See include-what-you-use issue 434.
IWYU incorrectly classifies this internal STL type as not internal, and
suggests including `<type_traits>` for it. Work around the problem by
mapping the offending names to a file that we always include.
See include-what-you-use issue 434.
Update the prebuilt binary build configurations for machines where
we build our own OpenSSL library to use a newer version.
Now that we require pthreads for libuv anyway, we can use OpenSSL 1.1
along with its dependency on pthreads.
Switch to the MSVC static runtime library (and static UCRT).
Link Qt statically, including the Windows platform plugin.
Disable the Qt tests because they are incompatible with a
Qt using a static runtime library.
Some are user facing.
Found using
codespell -q 3 --skip="./Utilities" -I .cmake-whitelist.txt`
whereby the whitelist contained:
ans
dum
helpfull
emmited
emmitted
buil
iff
isnt
nto
ot
pathes
substract
te
todays
upto
whitespaces
- sed on Mac does not support embedded sed statements, hence eliminating
the non-matches via grep
- xargs on Mac does not support -d, but does support -0, hence replacing
The SDK for macOS 10.13 adds `futimens` and `utimensat` so our checks
for these symbols may pass. However, the symbols are not available at
runtime on older macOS versions. Instead on macOS we can check for
availability based on the deployment target version.
Issue: #17101
On UNIX, build only the parts of libuv we need for the filesystem,
process, and poll abstractions using the POSIX poll() backend. This
avoids many platform-specific conditions. On Windows, build all of
libuv; there are no conditional alternatives anyway.
The scanbuild analysis cannot see that decodeUnicodeEscapeSequence
either initializes the `unicode` argument or returns `false` such
that the code following it runs only if `unicode` is initialized.
Add an explicit initialization to pacify it.
Even though we disable warnings when building jsoncpp itself, including
its headers from other CMake sources may still cause warnings.
Additional work will be needed to port CMake to newer jsoncpp interfaces
while still supporting jsoncpp 1.0.0. For now, just suppress the
markup.
note: this depends on
- a suitable flex version in PATH (currently 2.6.4)
- a suitable version of 'sed' in PATH (e.g. GNU sed)
Signed-off-by: Matthias Maennich <matthias@maennich.net>
