netdev
[Top] [All Lists]

Re: [PATCH] hashed device lookup (Does NOT meet Linus' sumission policy!

To: Ben Greear <greearb@xxxxxxxxxxxxxxx>
Subject: Re: [PATCH] hashed device lookup (Does NOT meet Linus' sumission policy!) (Benchmarks)
From: Andi Kleen <ak@xxxxxxx>
Date: Sun, 7 Jan 2001 08:15:33 +0100
Cc: Andi Kleen <ak@xxxxxxx>, linux-kernel <linux-kernel@xxxxxxxxxxxxxxx>, "netdev@xxxxxxxxxxx" <netdev@xxxxxxxxxxx>, "David S. Miller" <davem@xxxxxxxxxx>
In-reply-to: <3A58249F.86DD52BC@xxxxxxxxxxxxxxx>; from greearb@xxxxxxxxxxxxxxx on Sun, Jan 07, 2001 at 01:11:11AM -0700
References: <3A578F27.D2A9DF52@xxxxxxxxxxxxxxx> <20010107042959.A14330@xxxxxxxxxxxxxxxxxxx> <3A580B31.7998C783@xxxxxxxxxxxxxxx> <20010107062744.A15198@xxxxxxxxxxxxxxxxxxx> <3A58249F.86DD52BC@xxxxxxxxxxxxxxx>
Sender: owner-netdev@xxxxxxxxxxx
User-agent: Mutt/1.2.5i
On Sun, Jan 07, 2001 at 01:11:11AM -0700, Ben Greear wrote:
> > Packet socket binding or SO_BINDTODEVICE will search the list, but it is 
> > unlikely
> > that these paths are worth optimizing for.
> 
> The patch has been written, so even if it helps just a little more than it
> hurts, it might be worth including.  Of course, it may actually hurt more
> than help.
> 
> I'd be very interested in lucid arguments as to why adding the patch would
> actually be worse than not adding it, not just why I'm lame for considering
> it *grin* :)

It's like David said: it's not a very good idea to tune things that are not
commonly used. If some user really realistically has some workload where the
linear search is a bottleneck it can be considered. Currently it doesn't look
like it.




-Andi

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