netdev
[Top] [All Lists]

Re: [PATCH] Improve behaviour of Netlink Sockets

To: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
Subject: Re: [PATCH] Improve behaviour of Netlink Sockets
From: jamal <hadi@xxxxxxxxxx>
Date: 19 Sep 2004 18:39:29 -0400
Cc: Pablo Neira <pablo@xxxxxxxxxxx>, "David S. Miller" <davem@xxxxxxxxxx>, netdev@xxxxxxxxxxx
In-reply-to: <20040919215915.GB9573@xxxxxxxxxxxxxxxxxxx>
Organization: jamalopolous
References: <414D0CCD.90209@xxxxxxxxxxx> <E1C8way-0000aH-00@xxxxxxxxxxxxxxxxxxxxxxxx> <20040919120249.GA5963@xxxxxxxxxxxxxxxxxxx> <414DF11C.1080505@xxxxxxxxxxx> <20040919215915.GB9573@xxxxxxxxxxxxxxxxxxx>
Reply-to: hadi@xxxxxxxxxx
Sender: netdev-bounce@xxxxxxxxxxx
Theres one valuable piece in his patch. Ability to congestion control
the kernel so user space can clear the socket queue. The alternative is
an overun. You have shot a hole in the workqueue approach. May i also
recommend we revert the patch until we have a better resolution, Dave?

Pablo you really need to provide some good testcases to prove this.
I have my own netlink stress testing that i havent gotten around to
testing this issue. 

cheers,
jamal







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