From patchwork Sat Feb 7 23:12:28 2009 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Meelis Roos X-Patchwork-Id: 22538 X-Patchwork-Delegate: davem@davemloft.net Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.176.167]) by ozlabs.org (Postfix) with ESMTP id 85893DDDEF for ; Sun, 8 Feb 2009 12:16:26 +1100 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753548AbZBGXMb (ORCPT ); Sat, 7 Feb 2009 18:12:31 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753746AbZBGXMb (ORCPT ); Sat, 7 Feb 2009 18:12:31 -0500 Received: from smtp2.it.da.ut.ee ([193.40.5.67]:33868 "EHLO smtp2.it.da.ut.ee" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753548AbZBGXMb (ORCPT ); Sat, 7 Feb 2009 18:12:31 -0500 Received: from math.ut.ee (math.ut.ee [193.40.36.2]) by smtp2.it.da.ut.ee (Postfix) with ESMTP id AC51273D89F; Sun, 8 Feb 2009 01:12:28 +0200 (EET) Received: by math.ut.ee (Postfix, from userid 1014) id 9878EFD7A1; Sun, 8 Feb 2009 01:12:28 +0200 (EET) Received: from localhost (localhost [127.0.0.1]) by math.ut.ee (Postfix) with ESMTP id 8BDADFD79B; Sun, 8 Feb 2009 01:12:28 +0200 (EET) Date: Sun, 8 Feb 2009 01:12:28 +0200 (EET) From: Meelis Roos To: David Miller cc: ebrower@gmail.com, sparclinux@vger.kernel.org Subject: [PATCH] fix Quattro HME irq registration on proble failures In-Reply-To: <20090207.022003.260994505.davem@davemloft.net> Message-ID: References: <20090206.234600.211802423.davem@davemloft.net> <20090207.022003.260994505.davem@davemloft.net> MIME-Version: 1.0 Sender: sparclinux-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: sparclinux@vger.kernel.org > Hmmm... one thing we could do is only register the IRQ if all 4 > slots of the per-quattro array are filled in. > > Similar checks would be needed in quattro_sbus_free_irqs(). > > This means we'll also have to add some code to the error paths > of happy_meal_sbus_probe_one() to make it NULL out the > qp->happy_meals[qfe_slot] entry on failure. Here is a preliminary patch, with some doubts: * should we really panic if request_irq fails in quattro_sbus_register_irqs()? Better propagate the error up. However, do we still need to return success if any IRQ was registered correctly (like the logic of of_register_driver one step before)? Should this be a different patch/commit? * happy_meal_open() does request_irq for PCI cards and non-quattro SBus cards, so the printk on failure may lie about SBus? Or did I misread the code? But here is the patch with description: --------------------- Currently, the sunhme driver installs SBus Quattro interrupt handler when at least one HME card was initialized correctly and at least one Quattro card is present. This breaks when a Quattro card fails initialization for whatever reason - IRQ is registered and OOPS happens when it fires. The solution, as suggested by David Miller, was to keep track which cards of the Quattro bundles have been initialized, and request/free the Quattro IRQ only when all four devices have been successfully initialized. The patch only touches SBus initialization - PCI init already resets the card pointer to NULL on init failure. The patch has been tested on Sun E3500 with SBus and PCI single HME cards and one PCI Quattro HME card in a situation where any PCI card failed init when the SBus routines tried to init them by mistake. Additionally it replaces Quattro request_irq panic with error return - if this card fails to work, at least let the others work. This change is untested. Signed-off-by: Meelis Roos diff --git a/drivers/net/sunhme.c b/drivers/net/sunhme.c index 7a72a31..8887ed9 100644 --- a/drivers/net/sunhme.c +++ b/drivers/net/sunhme.c @@ -2543,25 +2543,36 @@ static struct quattro * __devinit quattro_sbus_find(struct of_device *child) } /* After all quattro cards have been probed, we call these functions - * to register the IRQ handlers. + * to register the IRQ handlers for the cards that have been + * successfully probed and skip the cards that failed to initialize */ -static void __init quattro_sbus_register_irqs(void) +static int __init quattro_sbus_register_irqs(void) { struct quattro *qp; for (qp = qfe_sbus_list; qp != NULL; qp = qp->next) { struct of_device *op = qp->quattro_dev; int err; + int qfe_slot, skip = 0; + + for (qfe_slot = 0; qfe_slot < 4; qfe_slot++) { + if (qp->happy_meals[qfe_slot] == NULL) { + skip = 1; + } + } + if (skip) continue; err = request_irq(op->irqs[0], quattro_sbus_interrupt, IRQF_SHARED, "Quattro", qp); if (err != 0) { - printk(KERN_ERR "Quattro: Fatal IRQ registery error %d.\n", err); - panic("QFE request irq"); + printk(KERN_ERR "Quattro HME: IRQ registration error %d.\n", err); + return err; } } + + return 0; } static void quattro_sbus_free_irqs(void) @@ -2570,6 +2581,14 @@ static void quattro_sbus_free_irqs(void) for (qp = qfe_sbus_list; qp != NULL; qp = qp->next) { struct of_device *op = qp->quattro_dev; + int qfe_slot, skip = 0; + + for (qfe_slot = 0; qfe_slot < 4; qfe_slot++) { + if (qp->happy_meals[qfe_slot] == NULL) { + skip = 1; + } + } + if (skip) continue; free_irq(op->irqs[0], qp); } @@ -2824,6 +2843,10 @@ err_out_iounmap: if (hp->tcvregs) of_iounmap(&op->resource[4], hp->tcvregs, TCVR_REG_SIZE); + if (qp != NULL ) { + qp->happy_meals[qfe_slot] = NULL; + } + err_out_free_netdev: free_netdev(dev); @@ -3281,7 +3304,7 @@ static int __init happy_meal_sbus_init(void) err = of_register_driver(&hme_sbus_driver, &of_bus_type); if (!err) - quattro_sbus_register_irqs(); + err = quattro_sbus_register_irqs(); return err; }