From patchwork Wed Aug 30 01:34:26 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Shawn Lin X-Patchwork-Id: 9928629 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 6F81E602B9 for ; Wed, 30 Aug 2017 01:35:22 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 61C74205FD for ; Wed, 30 Aug 2017 01:35:22 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 5667F20683; Wed, 30 Aug 2017 01:35:22 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-1.6 required=2.0 tests=BAYES_00, RCVD_IN_BL_SPAMCOP_NET, RCVD_IN_DNSWL_HI, RCVD_IN_SBL, RCVD_IN_SBL_CSS, RCVD_IN_SORBS_SPAM autolearn=no version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id E1AD8205FD for ; Wed, 30 Aug 2017 01:35:21 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751872AbdH3BfK (ORCPT ); Tue, 29 Aug 2017 21:35:10 -0400 Received: from lucky1.263xmail.com ([211.157.147.132]:41615 "EHLO lucky1.263xmail.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751318AbdH3BfC (ORCPT ); Tue, 29 Aug 2017 21:35:02 -0400 Received: from shawn.lin?rock-chips.com (unknown [192.168.167.193]) by lucky1.263xmail.com (Postfix) with ESMTP id EAD5F64511; Wed, 30 Aug 2017 09:34:59 +0800 (CST) X-263anti-spam: KSV:0; X-MAIL-GRAY: 1 X-MAIL-DELIVERY: 0 X-KSVirus-check: 0 X-ABS-CHECKED: 4 Received: from localhost.localdomain (localhost [127.0.0.1]) by smtp.263.net (Postfix) with ESMTPA id 369F6390; Wed, 30 Aug 2017 09:34:58 +0800 (CST) X-RL-SENDER: shawn.lin@rock-chips.com X-FST-TO: ulf.hansson@linaro.org X-SENDER-IP: 58.22.7.114 X-LOGIN-NAME: shawn.lin@rock-chips.com X-UNIQUE-TAG: <2784debca3acabef8410d7ccbb81b318> X-ATTACHMENT-NUM: 0 X-SENDER: lintao@rock-chips.com X-DNS-TYPE: 0 Received: from unknown (unknown [58.22.7.114]) by smtp.263.net (Postfix) whith SMTP id 31217O8NVN3; Wed, 30 Aug 2017 09:34:59 +0800 (CST) From: Shawn Lin To: Ulf Hansson , Greg Kroah-Hartman Cc: Jaehoon Chung , "Rafael J. Wysocki" , linux-mmc@vger.kernel.org, linux-kernel@vger.kernel.org, Shawn Lin Subject: [PATCH v3 1/2] driver core: detach device's pm_domain after devres_release_all Date: Wed, 30 Aug 2017 09:34:26 +0800 Message-Id: <1504056867-199188-2-git-send-email-shawn.lin@rock-chips.com> X-Mailer: git-send-email 1.9.1 In-Reply-To: <1504056867-199188-1-git-send-email-shawn.lin@rock-chips.com> References: <1504056867-199188-1-git-send-email-shawn.lin@rock-chips.com> Sender: linux-mmc-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-mmc@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP The intention of this patch is to move dev_pm_domain_detach after devres_release_all to avoid possible accessing device's registers with genpd been powered off. Many common IP drivers use devm_request_irq to request irq for either shared irq or non-shared irq. So we rely on devres_release_all to free irq automatically. However we could see a situation that if the driver use devm_request_irq to register a shared irq and unbind the driver later, the irq could be triggerd cocurrently just between finishing dev_pm_domain_detach and calling devm_irq_release, so that driver's ISR should be called and try to access device's register, which may hang up the system. The reason is that some SoCs, including Rockchips' SoCs, couldn't support accessing controllers' registers w/o clk and power domain enabled. Also we have CONFIG_DEBUG_SHIRQ to theoretically expose this kind of problem as devm_free_irq -> __free_irq says: "It's a shared IRQ -- the driver ought to be prepared for an IRQ event to happen even now it's being freed". That will simulate the aforementioned situation as it fires extra irq action to make sure driver/system is robust enough to deal with this kind of problem. So now we face a two possible choices to fix this by (1) either using request_irq and free_irq directly (2) or moving dev_pm_domain_detach after devres_release_all which makes sure we free the irq before powering off power domain. However, choice (1) implies that devm_request_irq shouldn't exist or at least shouldn't be used for shared irq case. Meanwhile we don't know how many drivers have this kind of issue and need to fix. So choice (2) makes more sense to me, and that is the reason for why we need to fix it like what this patch does. Signed-off-by: Shawn Lin --- Changes in v3: - fix the code path for consolidating the attach for both of driver and bus driver, and then move detach to the error path - rework the changelog drivers/base/dd.c | 16 ++++++++++++++++ drivers/base/platform.c | 18 ++---------------- 2 files changed, 18 insertions(+), 16 deletions(-) diff --git a/drivers/base/dd.c b/drivers/base/dd.c index ad44b40..aea0bb1 100644 --- a/drivers/base/dd.c +++ b/drivers/base/dd.c @@ -25,7 +25,9 @@ #include #include #include +#include #include +#include #include #include "base.h" @@ -356,6 +358,7 @@ static int really_probe(struct device *dev, struct device_driver *drv) int local_trigger_count = atomic_read(&deferred_trigger_count); bool test_remove = IS_ENABLED(CONFIG_DEBUG_TEST_DRIVER_REMOVE) && !drv->suppress_bind_attrs; + struct platform_driver *pdrv; if (defer_all_probes) { /* @@ -409,6 +412,16 @@ static int really_probe(struct device *dev, struct device_driver *drv) */ devices_kset_move_last(dev); + ret = dev_pm_domain_attach(dev, true); + pdrv = to_platform_driver(dev->driver); + /* don't fail if just dev_pm_domain_attach failed */ + if (pdrv && pdrv->prevent_deferred_probe && + ret == -EPROBE_DEFER) { + dev_warn(dev, "probe deferral not supported\n"); + ret = -ENXIO; + goto probe_failed; + } + if (dev->bus->probe) { ret = dev->bus->probe(dev); if (ret) @@ -428,6 +441,7 @@ static int really_probe(struct device *dev, struct device_driver *drv) drv->remove(dev); devres_release_all(dev); + dev_pm_domain_detach(dev, true); driver_sysfs_remove(dev); dev->driver = NULL; dev_set_drvdata(dev, NULL); @@ -458,6 +472,7 @@ static int really_probe(struct device *dev, struct device_driver *drv) pinctrl_bind_failed: device_links_no_driver(dev); devres_release_all(dev); + dev_pm_domain_detach(dev, true); driver_sysfs_remove(dev); dev->driver = NULL; dev_set_drvdata(dev, NULL); @@ -864,6 +879,7 @@ static void __device_release_driver(struct device *dev, struct device *parent) dma_deconfigure(dev); devres_release_all(dev); + dev_pm_domain_detach(dev, true); dev->driver = NULL; dev_set_drvdata(dev, NULL); if (dev->pm_domain && dev->pm_domain->dismiss) diff --git a/drivers/base/platform.c b/drivers/base/platform.c index d1bd992..8fa688d 100644 --- a/drivers/base/platform.c +++ b/drivers/base/platform.c @@ -572,22 +572,8 @@ static int platform_drv_probe(struct device *_dev) if (ret < 0) return ret; - ret = dev_pm_domain_attach(_dev, true); - if (ret != -EPROBE_DEFER) { - if (drv->probe) { - ret = drv->probe(dev); - if (ret) - dev_pm_domain_detach(_dev, true); - } else { - /* don't fail if just dev_pm_domain_attach failed */ - ret = 0; - } - } - - if (drv->prevent_deferred_probe && ret == -EPROBE_DEFER) { - dev_warn(_dev, "probe deferral not supported\n"); - ret = -ENXIO; - } + if (drv->probe) + ret = drv->probe(dev); return ret; }