Re: [PATCH v2 0/4] MR47: kernel32/tests: Add tests for CancelSynchronousIo.

[Jinoh Kang (＠iamahuman)]

> the test would hang on wine before the implementation is in place.

Thanks!  I missed that the current test cannot verify the Windows behaviour before the commit that introduces the Wine implementation.
> the CloseHandle before the WaitForSingleObject is a workaround to get the thread to exit

I'm afraid that closing a handle in use is racy by nature; it's effectively pulling the rug out from underneath.  A better approach would be to use `CancelIoEx` on the `OVERLAPPED` structure, or `NtCancelIoFileEx` on the `IO_STATUS_BLOCK` structure to cancel the pending operation.
> moving the CloseHandle was meant to show that workaround is no longer needed and that it now performs like windows
>
> i can remove that if preferred