From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alan Cox Subject: Re: [PATCH] serial: handle pci_enable_device() failure upon resume Date: Sat, 14 Oct 2006 00:15:43 +0100 Message-ID: <1160781343.25218.89.camel@localhost.localdomain> References: <20061012014720.GA12935@havoc.gtf.org> <20061013075953.GC28654@flint.arm.linux.org.uk> <20061013131516.227e99ee.akpm@osdl.org> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from outpipe-village-512-1.bc.nu ([81.2.110.250]:44165 "EHLO lxorguk.ukuu.org.uk") by vger.kernel.org with ESMTP id S1751959AbWJMWuS (ORCPT ); Fri, 13 Oct 2006 18:50:18 -0400 In-Reply-To: <20061013131516.227e99ee.akpm@osdl.org> Sender: linux-serial-owner@vger.kernel.org List-Id: linux-serial@vger.kernel.org To: Andrew Morton Cc: Russell King , Jeff Garzik , linux-serial@vger.kernel.org, LKML Ar Gwe, 2006-10-13 am 13:15 -0700, ysgrifennodd Andrew Morton: > Right now, we get silent failure *and* a compile-time warning. It's hard > to see how that situation could be made worse. I support Russells NAK but for another reason. You can't use printk in the paths of a serial driver that may be the console. There is already a whole ordering issue here and I'm not convinced it is handled correctly reviewing the diff. Printks will make it worse in any driver (and waste memory on handling impossible untestable situations that don't appen) The printk case and serial first ordering want fixing correctly because serial console is about the only sane way to debug resume in the first place. Alan