From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mike McCormack Subject: Re: [PATCH 2/2] sky2: Allocate initial skbs in sky2_alloc_buffers Date: Sun, 7 Feb 2010 18:56:44 +0900 Message-ID: <392fb48f1002070156r7a1d615j460506b9441e128c@mail.gmail.com> References: <4B6CC473.4090008@ring3k.org> <20100205181045.42908795@nehalam> <20100205221246.6fdb3632@s6510> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Cc: netdev@vger.kernel.org To: Stephen Hemminger Return-path: Received: from mail-gx0-f224.google.com ([209.85.217.224]:55370 "EHLO mail-gx0-f224.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751260Ab0BGJ4p (ORCPT ); Sun, 7 Feb 2010 04:56:45 -0500 Received: by gxk24 with SMTP id 24so5824458gxk.1 for ; Sun, 07 Feb 2010 01:56:44 -0800 (PST) In-Reply-To: <20100205221246.6fdb3632@s6510> Sender: netdev-owner@vger.kernel.org List-ID: On 6 February 2010 15:12, Stephen Hemminger wrote: > On Fri, 5 Feb 2010 18:10:45 -0800 > Stephen Hemminger wrote: > >> On Sat, 06 Feb 2010 10:22:59 +0900 >> Mike McCormack wrote: >> >> > Allocating everything in one place means there's a single point >> > of failure in sky2_up, and sky2_rx_start can no longer fail. >> >> If ring is never allocated, how then it must fail in up. >> Plus if the initial ring allocation is partial it should fail. > > Let me put that clearer... > When dev_open is called, the system might be very short of memory > and unable to allocate the number of receive buffers; in that case, > I would prefer that an error was returned to the application. > Yes, this is a corner case; but it is better to fail with a noisy > error than limp along with a dead device. Hi Stephen, I think you've misread my patch. I have not attempted to change the way allocation failure is handled, just to move all allocations to the same place. The end goal is to refactor sky2_up into a piece that initializes hardware following memory allocation (say sky2_start) so we can do sky2_reset() without free'ing and allocating memory, or detaching the device. thanks, Mike