From: Nishanth Menon Subject: Re: [PATCH] crypto: omap-des - handle error of pm_runtime_get_sync Date: Tue, 15 Apr 2014 11:44:15 -0500 Message-ID: <534D61DF.9090203@ti.com> References: <1397575982-24866-1-git-send-email-nm@ti.com> <20140415155031.GA14623@saruman.home> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit Cc: Joel Fernandes , Herbert Xu , "David S. Miller" , , , , Joachim Eastwood To: Return-path: Received: from arroyo.ext.ti.com ([192.94.94.40]:35717 "EHLO arroyo.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751019AbaDOQoV (ORCPT ); Tue, 15 Apr 2014 12:44:21 -0400 In-Reply-To: <20140415155031.GA14623@saruman.home> Sender: linux-crypto-owner@vger.kernel.org List-ID: On 04/15/2014 10:50 AM, Felipe Balbi wrote: > Hi, > > On Tue, Apr 15, 2014 at 10:33:02AM -0500, Nishanth Menon wrote: >> pm_runtime_get_sync may not always succeed depending on SoC involved. So >> handle the error appropriately. >> >> Reported-by: Joachim Eastwood >> Signed-off-by: Nishanth Menon >> --- >> >> based on v3.15-rc1 >> >> Report-thread: http://marc.info/?t=139750330400002&r=1&w=2 >> >> omap2plus_defconfig + CONFIG_CRYPTO_DEV_OMAP_DES=y >> >> pandaboard-es-before: http://slexy.org/raw/s21rGPFnKl >> pandaboard-es-after: http://slexy.org/raw/s2A4UFQVna >> >> drivers/crypto/omap-des.c | 18 ++++++++++++++---- >> 1 file changed, 14 insertions(+), 4 deletions(-) >> >> diff --git a/drivers/crypto/omap-des.c b/drivers/crypto/omap-des.c >> index ec5f131..c2d362f 100644 >> --- a/drivers/crypto/omap-des.c >> +++ b/drivers/crypto/omap-des.c >> @@ -223,12 +223,18 @@ static void omap_des_write_n(struct omap_des_dev *dd, u32 offset, >> >> static int omap_des_hw_init(struct omap_des_dev *dd) >> { >> + int err; >> + >> /* >> * clocks are enabled when request starts and disabled when finished. >> * It may be long delays between requests. >> * Device might go to off mode to save power. >> */ >> - pm_runtime_get_sync(dd->dev); >> + err = pm_runtime_get_sync(dd->dev); >> + if (err < 0) { >> + dev_err(dd->dev, "failed to get_sync(%d)\n", err); >> + return err; > > even when get() fails, you must put(). > >> + } >> >> if (!(dd->flags & FLAGS_INIT)) { >> dd->flags |= FLAGS_INIT; >> @@ -1083,7 +1089,11 @@ static int omap_des_probe(struct platform_device *pdev) >> dd->phys_base = res->start; >> >> pm_runtime_enable(dev); >> - pm_runtime_get_sync(dev); >> + err = pm_runtime_get_sync(dev); >> + if (err < 0) { >> + dev_err(dd->dev, "failed to get_sync(%d)\n", err); >> + goto err_get; > > this leaves usage_count incremented, needs a put() here too. > Yeah, I see the code in runtime.c, and pm_runtime_put_noidle is the right invocation in the case of error. Wish we had a coccinelle script for this. -- Regards, Nishanth Menon