...in a non-Cygwin child process. Backported from MSYS2.
Downstream commit message follows.
In https://github.com/msys2/msys2-runtime/pull/18, we discussed a change
that would allow default Windows error handling of spawned processes to
kick in (such as registered JIT debuggers). We even agreed that it would
make sense to hide this functionality behind a flag, `winjitdebug`.
However, when this got upstreamed as 21ec498d7f (cygwin: use
CREATE_DEFAULT_ERROR_MODE in spawn, 2020-12-09), that flag was deemed
unnecessary.
But it would appear that it _is_ necessary: As reported in
https://github.com/msys2/MSYS2-packages/pull/2414#issuecomment-810841296
this new behavior is pretty disruptive e.g. in CI scenarios.
So let's introduce that `winjitdebug` flag (settable via the environment
variable `MSYS`) at long last.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Currently, if_nametoindex() and if_indextoname() handle interface names
such as "ethernet_32777", while if_nameindex() returns the names such
as "{5AF7ACD0-D52E-4DFC-A4D0-54D3E6D6B2AC}". This patch unifies the
interface names to the latter.
Fixes: c356901f0d ("Rename if_indextoname to cygwin_if_indextoname (analag for if_nametoindex)")
Reviewed-by: Corinna Vinschen <corinna@vinschen.de>
Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
If disable_master_thread flag is set between the code checking that
flag not be set and the code acquiring input_mutex, input record is
processed once after setting disable_master_thread flag. This patch
prevents that.
Fixes: d4aacd50e6 ("Cygwin: console: Add missing input_mutex guard.")
Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
If non-cygwin process is executed in console, the exit code is not
set correctly. This is because the stub process for non-cygwin app
crashes in fhandler_console::set_disable_master_thread() due to NULL
pointer dereference. This bug was introduced by the commit:
3721a756b0 ("Cygwin: console: Make the console accessible from
other terminals."), that the pointer cons is accessed before fixing
when it is NULL. This patch fixes the issue.
Fixes: 3721a756b0 ("Cygwin: console: Make the console accessible from other terminals.")
Reported-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
cmdline has to be initialized, otherwise the return value is
undefined in case generating the command line fails.
Fixes: 732afede93 ("Cygwin: redefine CW_CMDLINE to CW_CMDLINE_ALLOC")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
IsBadStringPtr is deprecated and just gives a wrong sense
of memory safety. Replace with check for NULL pointer.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
On Linux, __progname and program_invocation_short_name are just
different exported names of the same string. Do the same in Cygwin.
This requires to tweak the mkglobals_h so as not to touch the
EXPORT_ALIAS expression. Also, use the base variable
program_invocation_short_name throughout. __progname is just
the export for getopt.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Since it's possible to change the __argv array, we should not rely
on __argv[0] actually representing the windows executable path in
any way. Use the real path stored in global_progname instead.
Fixes: 521953a83a ("* common.din: Export GetCommandLine{A,W}.")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
So far ps(1) always prints the full path of the executable under
the COMMAND heading.
With -f, print the command line instead.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Make sure to
- append a trailing \0 as with Windows multistrings, so the end of
the string can be recognized by the caller, and
- allocate cmdline on the user heap so the caller can free the
multistring after usage.
Fixes: 831d6fa520 ("* external.cc (cygwin_internal): Implement CW_CMDLINE.")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
- Basically maintain a hidden DIR* inside fhandlers.
- lseek has to be tweaked to allow basic seeking on the directory
descriptor.
- the current implementation does not keep the dir positions
between duplicated descriptor in sync. In fact, every descriptor
keeps its own copy of the DIR* and after dup/fork/exec, the
directory position is reset to 0, i. e., to the start of the
directory, as if rewinddir() has been called on the new descriptors.
While this behaviour isn't yet covered by the Issue 8 draft,
a bug report along these lines exists and will probably be
picked up for TC1.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
The content of DIR should never have been exposed into userspace.
Move struct __DIR to local dirent.h and only keep forward declaration
in exported dirent.h.
This allows to tweak the structure in future, for instance, to
add thread-safety.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
This change is in preparation of adding posix_getdents() from
the upcoming POSIX Base Specification Issue 8.
- Add d_reclen
- Add GLibC compatible test macros for dirent members
- Bump dirent version
- Set d_reclen to the fixed size of the dirent struct
We can do that because the size is a multiple of 8, so it fits
snugly in the buffer filled by posix_getdents and keep the
alignement.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
- drop appending .exe.lnk to files
- drop exe_suffixes, it's the same as stat_suffixes for a long time
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Remove a stray __END_DECLS. It didn't hurt in the only
(plain C) file including this header, but still...
Fixes: 030a762535 ("Cygwin: fix arc4random after fork(2)")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
After using fork(), arc4random does not reseed itself, which
causes the results to become predictable. Activate droppingfork-recognition
Fixes: e0fc33322d ("Delete Cygwin's arc4random in favor of new Newlib implementation")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
So far the global variable in_forkee only indicated if the
process is the child process during fork(2) itself.
However, we need an indicator accessible from plain C code
in newlib, allowing to check for a process being a forked
process all the time, after fork(2) succeeded.
Redefine bool in_forkee to int __in_forkee to allow exposing
it to newlib. Redefine how it indicates fork state (not
forked, forking, forked).
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
If pthread_once() is called with pthread_once_t initialized using
PTREAD_ONCE_INIT, pthread_once does not release pthread_mutex used
internally. This patch fixes that by calling pthread_mutex_destroy()
in the thread which has called init_routine.
Reviewed-by: Corinna Vinschen <corinna@vinschen.de>
Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
A process which is exiting due to a core dumping signal doesn't
propagate the correct exist status after dumping core, because 'dumper'
itself forcibly terminates the process.
Use 'dumper -n' to avoid killing the dumped process, so we continue to
the end of signal_exit(), to exit with the 128+signal exit status.
Busy-wait in exec_prepared_command() in an attempt to reliably notice
the dumper attaching, so we don't get stuck there.
Also: document these important facts for custom uses of error_start.
Commit 3f3bd10104 ("* Throughout, use __try/__except/__endtry blocks [...]")
introduced setting EINVAL, marked as "Diagnosis". The reason
for this is lost in time and space, but looks very much like
a debug helper which was supposed to be removed before release.
It's rather pointless, so remove it.
Fixes: 3f3bd10104 ("* Throughout, use __try/__except/__endtry blocks [...]")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Unfortunately fgrep is now deprecated in a very pushy way.
Make sure to use grep -F instead all around, even in docs
and comments/
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Provide the same debugging opportunities for api_fatal() as we do for a
core-dumping signal:
1) Break into any attached debugger
2) Start JIT debugger (if configured) (keeping these under DEBUGGING doesn't seem helpful)
3) Write a coredump (if rlim_core > 1MB)
4) Write a stackdump (if that failed, or 0 < rlim_core <= 1MB)
Pre-format a command to be executed on a fatal error to run 'dumper'
(using an absolute path).
Factor out executing a pre-formatted command, so we can use that for
invoking the JIT debugger in try_to_debug() (if error_start is present
in the CYGWIN env var) and to invoke dumper when a fatal error occurs.
On a fatal error, if the core file size limit is greater than 1MB,
invoke dumper to write a core dump. Otherwise, if that limit is greater
than 0, write a .stackdump file, as previously.
Adjust and clarify the associated documentation.
Also: Fix so that the error_start JIT debugger is now invoked, even when
ulimit -c is zero.
Also: Fix uses of console_printf() inside exec_prepared_command(). It's
output is written via the Windows console device, so needs to use
Windows-style line endings.
Also: consistently return non-zero from try_to_debug() if we debugged.
Future work: Truncate or remove the file written, if it exceeds the
maximum size set by the ulimit.
Future work: Using the words "fatal error" could probably be improved
on. This means exiting on one of the "certain signals whose default
action is to cause the process to terminate and produce a core dump
file".
This function closes or sets the close-on-exec flag for a specified
range of file descriptors. It is available on FreeBSD and Linux.
Signed-off-by: Christian Franke <christian.franke@t-online.de>
- len must not be <= 0
- offset + len must not exceed off_t (max. file size)
Fixes: e01c50c7b0 ("Cygwin: introduce fallocate(2)")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
- len must not be <= 0
- offset + len must not exceed off_t (max. file size)
Fixes: 7636b58590 ("* autoload.cc (NtSetInformationFile): Define.")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
The out pointer is only used if data_chunk_count is > 0,
so there's no reason to set it to NULL in the error case.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Given that SSDs don't have a seek penalty, we can enable
automatic sparsifying of files on SSDs, even if the "sparse"
mount option is not set.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
In preparation of using the FileFsSectorSizeInformation info class,
add a couple of missing definitions.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Corinna Vinschen
Johannes Schindelin
Takashi Yano
Jon Turney
Christian Franke