From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751128AbVLOVnN (ORCPT ); Thu, 15 Dec 2005 16:43:13 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751129AbVLOVnN (ORCPT ); Thu, 15 Dec 2005 16:43:13 -0500 Received: from dsl027-180-168.sfo1.dsl.speakeasy.net ([216.27.180.168]:40411 "EHLO sunset.davemloft.net") by vger.kernel.org with ESMTP id S1751128AbVLOVnL (ORCPT ); Thu, 15 Dec 2005 16:43:11 -0500 Date: Thu, 15 Dec 2005 13:42:16 -0800 (PST) Message-Id: <20051215.134216.18400210.davem@davemloft.net> To: rdreier@cisco.com Cc: benh@kernel.crashing.org, linux-kernel@vger.kernel.org, linux-pci@atrey.karlin.mff.cuni.cz Subject: Re: MSI and driver APIs From: "David S. Miller" In-Reply-To: References: <1134680882.16880.37.camel@gaston> X-Mailer: Mew version 4.2.53 on Emacs 21.4 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org From: Roland Dreier Date: Thu, 15 Dec 2005 13:18:04 -0800 > Benjamin> I'm tempted to leave them enabled and only disable them > Benjamin> when request_irq() is done on the legacy INTx... Does > Benjamin> anybody see a problem with this approach ? > > You might run into trouble on hardware (think tg3 or its ilk again) > where you might have to do something beyond disabling MSI in the PCI > header to switch the chip out of MSI mode. I think because of kinds of cases and other issues, going with MSI by default is a non-starter. Perhaps a better approach is to use a flag in the pci_driver_struct or similar that says "you can have MSI enabled by default". And gradually convert drivers over which we know will handle it properly. Doing some tom foolery with request_irq() sounds like a half-baked idea at best. The biggest argument against that is that this is not a PCI interface, so expecting it to have PCI side effects is really asking for trouble.