From mboxrd@z Thu Jan  1 00:00:00 1970
From: "David S. Miller" <davem@redhat.com>
Subject: Re: Ethernet bridge performance
Date: Thu, 7 Aug 2003 12:35:47 -0700
Sender: netdev-bounce@oss.sgi.com
Message-ID: <20030807123547.1dcf2353.davem@redhat.com>
References: <3F3217E7.2080903@allot.com>
	<3F3284EA.5050406@candelatech.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Cc: felix@allot.com, netdev@oss.sgi.com
Return-path: <netdev-bounce@oss.sgi.com>
To: Ben Greear <greearb@candelatech.com>
In-Reply-To: <3F3284EA.5050406@candelatech.com>
Errors-to: netdev-bounce@oss.sgi.com
List-Id: netdev.vger.kernel.org

On Thu, 07 Aug 2003 09:57:14 -0700
Ben Greear <greearb@candelatech.com> wrote:

> Btw, I've considered saving, say, 10k skbs on a list in my module,
> allocated by GFP_KERNEL at module load time, and using them when
> GFP_ATOMIC skb_alloc fails in the IRQ handling portion of the code....
> 
> Anyone think that's a good idea? :)

Not really.

GFP_ATOMIC should not fail regularly under normal (even heavy load)
operation.  If it does, it means the amount of reserved pages
the kernel keeps around is not set correctly for your system.

In 2.6.x, play with /proc/sys/vm/min_free_kbytes