netdev
[Top] [All Lists]

Re: [PATCH] (1/3) tcp - choose congestion algorithm at initialization

To: "David S. Miller" <davem@xxxxxxxxxx>
Subject: Re: [PATCH] (1/3) tcp - choose congestion algorithm at initialization
From: "Angelo Dell'Aera" <buffer@xxxxxxxxxxxxx>
Date: Mon, 4 Oct 2004 14:42:32 +0200
Cc: shemminger@xxxxxxxx, netdev@xxxxxxxxxxx
In-reply-to: <20040927121610.68f942a4.davem@xxxxxxxxxx>
Organization: Metro Olografix
References: <20040927111834.48c7baab@xxxxxxxxxxxxxxxxx> <20040927121610.68f942a4.davem@xxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Mon, 27 Sep 2004 12:16:10 -0700
"David S. Miller" <davem@xxxxxxxxxx> wrote:

>> The choice of congestion algorithm needs to be made when connection
>> is setup to avoid problems when the sysctl values change later and the
>> necessary data hasn't been collected.

>Looks great, applied.

I completely agree with you. Really a cool work. Just a proposal now. If 
the choice is made when the connection is established we do not need a 
struct for every supported congestion control algorithm in the struct 
tcp_opt any more. We could use for example a generic void * and then refer 
it properly after Stephen's controls such as tcp_is_vegas(), tcp_is_bic(),
etc. This should slim the struct tcp_opt because if a connection is using
Westwood its tcpbic and vegas structs will never be used. Moreover, I think
it would make the code more readable. 

Regards.

- --

Angelo Dell'Aera 'buffer' 
Antifork Research, Inc.         http://buffer.antifork.org
Metro Olografix

PGP information in e-mail header


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFBYUU4pONIzxnBXKIRArT5AJ0Ztak7xFH9DIEd24kRxCvMMxHrBACghKmE
e0CHh6gMpJaW02Y5HVA6u78=
=av9u
-----END PGP SIGNATURE-----

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