Re: Reauthentication implementation flaw due to EINTR

[Carl Fredrik Hammar]

On Wed, Dec 30, 2009 at 02:01:12PM +0100, Samuel Thibault wrote:
> Carl Fredrik Hammar, le Wed 30 Dec 2009 12:40:02 +0100, a écrit :
> >
> > The notification has been requested by
> > auth_server_authenticate which is waiting for auth_user_authenticate
> > to arive.
> 
> Only when the user isn't arrived already.

Right, but we know it hasn't since the server has posted a pending
operation, otherwise we'd be in the else part of the if.

> > In the then part of the if statement, the user has found that the
> > server is waiting for the condition to be signaled, which is right
> > time to cancel the notification.
> 
> Err, isn't the notification bound to an RPC?
> (really, I don't know anything about these and the Mach documentation
> doesn't help me so much).

The low-level interface is bound to the port we want notifications about.
The rendezvous port in this case, which is the same for auth_user and
auth_server.

libports associates it with the RPC and thread too though, so to clean
it up properly I guess we'd need to pass such information through the
pending structure.

Regards,
  Fredrik