[Bug 24101] Recursive CreateProcess() fails

[wine-bugs at winehq.org]

http://bugs.winehq.org/show_bug.cgi?id=24101

Ken Thomases <ken at codeweavers.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ken at codeweavers.com

--- Comment #36 from Ken Thomases <ken at codeweavers.com> 2010-09-22 10:05:17 CDT ---
(In reply to comment #32)
> On MacOSX the function wine_server_send_fd is not blocking, so when
> close(socketfd[1] ); is executed sometimes the server hasn't received the fd.

What do you mean that it's "not blocking"?  How did you determine this?

If wine_server_send_fd() isn't operating as expected, that seems like a problem
with possibly more serious and widespread ramifications.  Moving the close()
for this one case may not be solving the problem the right way.

Looking at the code, it certainly seems like the fd has been passed off to the
kernel via the sendmsg() call before wine_server_send_fd() has returned.  I
suppose there could be a bug in the kernel such that the file object is not
independent of the sending user process until it's been copied to the receiving
process.

-- 
Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email
Do not reply to this email, post in Bugzilla using the
above URL to reply.
------- You are receiving this mail because: -------
You are watching all bug changes.