netdev
[Top] [All Lists]

Re: [patch 4/10] s390: network driver.

To: Tommy Christensen <tommy.christensen@xxxxxxxxx>
Subject: Re: [patch 4/10] s390: network driver.
From: Paul Jakma <paul@xxxxxxxx>
Date: Mon, 20 Dec 2004 14:16:47 +0000 (UTC)
Cc: hadi@xxxxxxxxxx, Thomas Spatzier <thomas.spatzier@xxxxxxxxxx>, "David S. Miller" <davem@xxxxxxxxxxxxx>, Hasso Tepper <hasso@xxxxxxxxx>, Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>, jgarzik@xxxxxxxxx, netdev@xxxxxxxxxxx
In-reply-to: <41C612BC.5070909@tpack.net>
Mail-followup-to: paul@xxxxxxxxxxxxxxxxxx
References: <OF28701C56.81E1D26E-ONC1256F6B.00513EDD-C1256F6B.0052AF84@de.ibm.com> <1103484552.1046.155.camel@jzny.localdomain> <41C600D7.70005@tpack.net> <1103497516.1046.231.camel@jzny.localdomain> <41C612BC.5070909@tpack.net>
Sender: netdev-bounce@xxxxxxxxxxx
On Mon, 20 Dec 2004, Tommy Christensen wrote:

For a routing protocol that actually is notified that the link went down, it should probably flush those socket buffer at that point.

Or why not return an error, as soon as possible on the socket, eg ENOBUFS, and discard anything in the queue before that. Make it configurable via a sockopt if you think it'd harm ordinary apps (though, anything that cant deal with ENOBUFS is broken already, really..) or make it apply only to nonblock sockets.


  responsibility of the application to flush the socket on
  link-down events (by down'ing the interface?).

That seems more complex than needs be, for userspace at least.

regards,
--
Paul Jakma      paul@xxxxxxxx   paul@xxxxxxxxx  Key ID: 64A2FF6A
Fortune:
"Paul Lynde to block..."
-- a contestant on "Hollywood Squares"

<Prev in Thread] Current Thread [Next in Thread>