[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: io_close proposal

From: Marcus Brinkmann
Subject: Re: io_close proposal
Date: Thu, 16 May 2002 08:46:37 -0400
User-agent: Mutt/1.3.25i

On Wed, May 15, 2002 at 11:30:52PM -0700, Thomas Bushnell, BSG wrote:
> The server for io_close would look something like:
>   set up synchronization watch, lock against creation of new send rights
>   see if we have the only outstanding send right now
>   if we do:
>     deallocate the received send right
>     wait for synchronization
>     return

On Thu, May 16, 2002 at 12:08:27AM -0400, Roland McGrath wrote:
> operation.  If the RPC uses a move-send for the receiver right, then the
> only solution I see is special-case stubs on both sides so that the EINTR
> reply instead is a special format that returns the send right.  close will
> have to put that new right (which probably has a different name) into the
> dtable so that the fd can be used in the signal handler if the close didn't
> complete.

These two don't match (and now I know what confused me about this).
If the server starts to deallocate the port (after the **** above),
I don't see an easy way to interrupt this, make a new send right
for this port and return it to the user so he can safely use it
in subsequent io operations.

But before jumping to conclusions again, I will ponder it silently.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]