On 01/22/2018 08:01 PM, Alexandre Julliard wrote:
Zebediah Figura z.figura12@gmail.com writes:
Signed-off-by: Zebediah Figura z.figura12@gmail.com
dlls/kernel32/tests/pipe.c | 8 ++++++++ 1 file changed, 8 insertions(+)
diff --git a/dlls/kernel32/tests/pipe.c b/dlls/kernel32/tests/pipe.c index 922eed9..1708c28 100644 --- a/dlls/kernel32/tests/pipe.c +++ b/dlls/kernel32/tests/pipe.c @@ -2962,10 +2962,15 @@ static void child_process_write_pipe(HANDLE pipe) { OVERLAPPED overlapped; char buf[10000];
HANDLE event;
memset(buf, 'x', sizeof(buf)); overlapped_write_async(pipe, buf, sizeof(buf), &overlapped);
event = OpenEventA(EVENT_MODIFY_STATE, FALSE, "wine_test_pipe_event");
SetEvent(event);
CloseHandle(event);
It would be more robust to pass the event to the child instead of using a global name. Also as long as we are synchronizing them, having the parent signal the child to exit would be cleaner than using TerminateProcess.
TerminateProcess here was intentional because it triggered a more interesting code path in server, but sure, it can be changed.
The problem with this test is a todo_wine like in output. I was thinking about fixing the problem in kernel32 shown by the test and removing todo_wine.
Thanks,
Jacek