From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759517AbXKGSud (ORCPT ); Wed, 7 Nov 2007 13:50:33 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752973AbXKGSu0 (ORCPT ); Wed, 7 Nov 2007 13:50:26 -0500 Received: from smtp-out0.tiscali.nl ([195.241.79.175]:50659 "EHLO smtp-out0.tiscali.nl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752143AbXKGSuZ (ORCPT ); Wed, 7 Nov 2007 13:50:25 -0500 Message-ID: <473208ED.9000707@tiscali.nl> Date: Wed, 07 Nov 2007 19:50:21 +0100 From: Roel Kluin <12o3l@tiscali.nl> User-Agent: Thunderbird 2.0.0.6 (X11/20070728) MIME-Version: 1.0 To: Ray Lee CC: lkml Subject: Re: [PATCH] fix incorrect test in trident_ac97_set(); sound/oss/trident.c References: <47320533.4020400@tiscali.nl> <2c0942db0711071043p7cd31f6di9031816c1fed81ca@mail.gmail.com> In-Reply-To: <2c0942db0711071043p7cd31f6di9031816c1fed81ca@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Ray Lee wrote: > On Nov 7, 2007 10:34 AM, Roel Kluin <12o3l@tiscali.nl> wrote: >> If count reaches zero, the loop ends, but the postfix decrement subtracts it. >> so, testing for 'count == 0' will not work. >> >> Signed-off-by: Roel Kluin <12o3l@tiscali.nl> >> --- >> diff --git a/sound/oss/trident.c b/sound/oss/trident.c >> index 96adc47..94b5fb4 100644 >> --- a/sound/oss/trident.c >> +++ b/sound/oss/trident.c >> @@ -2939,7 +2939,7 @@ trident_ac97_set(struct ac97_codec *codec, u8 reg, u16 val) >> >> data |= (mask | (reg & AC97_REG_ADDR)); >> >> - if (count == 0) { >> + if (count == -1) { >> printk(KERN_ERR "trident: AC97 CODEC write timed out.\n"); >> spin_unlock_irqrestore(&card->lock, flags); >> return; >> @@ -2999,7 +2999,7 @@ trident_ac97_get(struct ac97_codec *codec, u8 reg) >> } while (count--); >> spin_unlock_irqrestore(&card->lock, flags); >> >> - if (count == 0) { >> + if (count == -1) { >> printk(KERN_ERR "trident: AC97 CODEC read timed out.\n"); >> data = 0; >> } > > You didn't test this: count is unsigned. Change the loop condition to > be --count instead. > > Ray Yep, you're right, here: -- If count reaches zero, the loop ends, but the postfix decrement still subtracts: testing for 'count == 0' will not work. Signed-off-by: Roel Kluin <12o3l@tiscali.nl> --- diff --git a/sound/oss/trident.c b/sound/oss/trident.c index 96adc47..6959ee1 100644 --- a/sound/oss/trident.c +++ b/sound/oss/trident.c @@ -2935,7 +2935,7 @@ trident_ac97_set(struct ac97_codec *codec, u8 reg, u16 val) do { if ((inw(TRID_REG(card, address)) & busy) == 0) break; - } while (count--); + } while (--count); data |= (mask | (reg & AC97_REG_ADDR)); @@ -2996,7 +2996,7 @@ trident_ac97_get(struct ac97_codec *codec, u8 reg) data = inl(TRID_REG(card, address)); if ((data & busy) == 0) break; - } while (count--); + } while (--count); spin_unlock_irqrestore(&card->lock, flags); if (count == 0) {