[PATCH 1/2] msvcrt: Do not acquire fd lock in msvcrt_create_io_inherit_block. (resend)

[Thomas Faber]

Apologies, my user agent apparently messed this one up somehow.

[PATCH 2/2] msvcrt/tests: Show that spawn does not require locking of fds. (resend)

[Thomas Faber]

This comes second since the test will deadlock without the fix applied.

[PATCH 2/2] msvcrt/tests: Show that spawn does not require locking of fds. (resend)

[Piotr Caban]

Hi Thomas,

On 11/12/21 14:40, Thomas Faber wrote:
> - _write(pipefds[1], "a", 1);
> - WaitForSingleObject(thread_handle, INFINITE);
> - CloseHandle(thread_handle);

You're leaking pipefds here.

Thanks,
Piotr

[PATCH 2/2] msvcrt/tests: Show that spawn does not require locking of fds. (resend)

[Thomas Faber]

On 2021-11-12 11:33, Piotr Caban wrote:
> On 11/12/21 14:40, Thomas Faber wrote:
> > +    _write(pipefds[1], "a", 1);
> > +    WaitForSingleObject(thread_handle, INFINITE);
> > +    CloseHandle(thread_handle);
>
> You're leaking pipefds here.

Whoops! Thanks for catching that! V2 incoming with both changes.

Thanks,
Thomas

[Piotr Caban]

Hi Thomas,

On 11/12/21 14:39, Thomas Faber wrote:
> unsigned msvcrt_create_io_inherit_block(WORD *size, BYTE **block)

Could you please change the function return type to BOOL while you're at it?

Thanks,
Piotr