Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933195Ab3GPRNq (ORCPT ); Tue, 16 Jul 2013 13:13:46 -0400 Received: from caramon.arm.linux.org.uk ([78.32.30.218]:37138 "EHLO caramon.arm.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932381Ab3GPRNo (ORCPT ); Tue, 16 Jul 2013 13:13:44 -0400 Date: Tue, 16 Jul 2013 18:13:24 +0100 From: Russell King - ARM Linux To: Boris BREZILLON Cc: Nicolas Ferre , Ludovic Desroches , Jean-Christophe Plagniol-Villard , Chris Ball , linux-mmc@vger.kernel.org, linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org Subject: Re: [PATCH v3 2/7] mmc: atmel-mci: prepare clk before calling enable Message-ID: <20130716171324.GF24642@n2100.arm.linux.org.uk> References: <1373986995-23899-1-git-send-email-b.brezillon@overkiz.com> <1373987208-24120-1-git-send-email-b.brezillon@overkiz.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1373987208-24120-1-git-send-email-b.brezillon@overkiz.com> User-Agent: Mutt/1.5.19 (2009-01-05) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1891 Lines: 65 On Tue, Jul 16, 2013 at 05:06:48PM +0200, Boris BREZILLON wrote: > @@ -389,9 +391,13 @@ static int atmci_regs_show(struct seq_file *s, void *v) > * consistent. > */ > spin_lock_bh(&host->lock); > - clk_enable(host->mck); > + ret = clk_prepare_enable(host->mck); > + if (ret) { > + spin_unlock_bh(&host->lock); > + goto out; > + } NAK. This is buggy. clk_prepare() can sleep. Calling clk_prepare() even via clk_prepare_enable() is a blatent bug. > memcpy_fromio(buf, host->regs, ATMCI_REGS_SIZE); > - clk_disable(host->mck); > + clk_disable_unprepare(host->mck); > spin_unlock_bh(&host->lock); Now, given that the CLK API counts enables/disables, having the spin lock around the clk API calls is utterly pointless. This should be: ret = clk_prepare_enable(host->mck); if (ret) goto out; spin_lock_bh(&host->lock); memcpy_fromio(buf, host->regs, ATMCI_REGS_SIZE); spin_unlock_bh(&host->lock); clk_disable_unprepare(host->mclk); or, if you really need to have the clock enabled/disabled within the spinlock (very very very unlikely): ret = clk_prepare(host->mck); if (ret) goto out; spin_lock_bh(&host->lock); ret = clk_enable(host->mck); if (ret == 0) { memcpy_fromio(buf, host->regs, ATMCI_REGS_SIZE); clk_disable(host->mck); } spin_unlock_bh(&host->lock); clk_unprepare(host->mclk); if (ret) goto out; > @@ -1279,7 +1286,7 @@ static void atmci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios) > > spin_lock_bh(&host->lock); > if (!host->mode_reg) { > - clk_enable(host->mck); > + clk_prepare_enable(host->mck); Again, buggy - calling clk_prepare beneath a spinlock is illegal. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/