wait_and_close_child_process() simplifies waiting for the child process
and closing its handles. And because it uses wait_child_process() this
ensures that any error happening in the child process is detected.
reload_child_dump() wraps a cryptic WritePrivateProfileStringA() call
and avoids having to add a comment every time.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
The buffer content is not used before being overwritten again.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
The process must still be running for AssignProcessToJobObject() to
succeed.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Don't include 'sync ' in the semaphore name because it will end up as
a separate argument in the child process, causing it to not find the
semaphore. Switch the child to OpenSemaphoreA() to reduce the risk of
accidentally create a new semaphore instead of opening the parent's
one. Use wait_child_process() instead of a raw
WaitForSingleObject(). The timeout is longer but the process is
expected to exit immediately anyway and this allows proper handling of
child failures (such as if there is a bug with the semaphore
handling).
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Wine intentionally treats these specially, so there is no point in testing them.
Signed-off-by: Zebediah Figura <zfigura@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Since Windows 8.1, these functions have been deprecated and run in a sort
of compatibility mode, reporting Windows 8 unless the application supplies a
manifest that specifies compatibility with newer Windows versions explicitly
(by listing their GUIDs).
Some applications have bad non-forward-compatible checks based
on GetVersionEx, and depend on this behavior (they do not supply a
manifest). Currently, they break on Wine if we use a Windows 10 prefix for
example, since we always report the real version. One example is the game
Rock of Ages.
Signed-off-by: Gabriel Ivăncescu <gabrielopcode@gmail.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Report the line number where the test failed to wait for the child so
one can identify which child process did not behave as expected.
Also wait_child_process() is meant for the general case so report
all non-crash error cases as test failures so they are accounted for.
Omit the "winetest_" prefix to match the other Wine test functions and
so the underlying winetest_wait_child_process() function can be wrapped
with the usual line-capturing macros.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=48651
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
The argv[0] tests mostly care about the value of argvA0 in the child
process. So remove most of the extra arguments and use the remaining
ones to check other aspects.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Such a short pause cannot serve any purpose and this shaves off 1 to 2
seconds of run time.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Synchronize with the child process to ensure it does not exit before
being added to the test job.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=48642
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
On Windows, processes are not reassigned to pid 1 when their parent
dies. This means many processes could have the same ppid despite being
started by unrelated processes. It also means test_Toolhelp() cannot
reliably look up the nested process by its ppid.
So look up for the nested process by its pid and only then check its
ppid, executable file, etc.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=48664
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Our test framework depends on CreateProcess() initializing the
hProcess field in case of a failure. The process tests depend on it
not modifying the STARTUPINFO fields.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Waiting for and closing the job object does not kill the child process
which then may print its 'tests executed' line at the same time as the
parent traces, sometimes mangling failure messages.
Signed-off-by: Francois Gouget <fgouget@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Rémi Bernon