From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753424AbbAMV0Z (ORCPT ); Tue, 13 Jan 2015 16:26:25 -0500 Received: from devils.ext.ti.com ([198.47.26.153]:49646 "EHLO devils.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751202AbbAMV0X (ORCPT ); Tue, 13 Jan 2015 16:26:23 -0500 Message-ID: <54B58D5D.5070508@ti.com> Date: Tue, 13 Jan 2015 15:25:49 -0600 From: Suman Anna User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.3.0 MIME-Version: 1.0 To: Rob Herring CC: Grant Likely , Rob Herring , Greg Kroah-Hartman , Pawel Moll , Pantelis Antoniou , Felipe Balbi , "devicetree@vger.kernel.org" , "linux-kernel@vger.kernel.org" , "linux-arm-kernel@lists.infradead.org" , linux-omap Subject: Re: [RFC PATCH 1/3] of/device: manage resources similar to platform_device_add References: <1420651854-17768-1-git-send-email-s-anna@ti.com> <1420651854-17768-2-git-send-email-s-anna@ti.com> In-Reply-To: Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Rob, On 01/13/2015 02:38 PM, Rob Herring wrote: > On Wed, Jan 7, 2015 at 11:30 AM, Suman Anna wrote: >> Drivers can use of_platform_populate() to create platform devices >> for children of the device main node, and a complementary API >> of_platform_depopulate() is provided to delete these child platform >> devices. The of_platform_depopulate() leverages the platform API >> for performing the cleanup of these devices. >> >> The platform device resources are managed differently between >> of_device_add and platform_device_add, and this asymmetry causes >> a kernel oops in platform_device_del during removal of the resources. >> Manage the platform device resources similar to platform_device_add >> to fix this kernel oops. > > This is a known issue and has been attempted to be fixed before (I > believe there is a revert in mainline). The problem is there are known > devicetrees which have overlapping resources and they will break with > your change. Are you referring to 02bbde7849e6 (Revert "of: use platform_device_add")? That one seems to be in registration path, and this crash is in the unregistration path. If so, to fix the crash, should we be skipping the release_resource() for now in platform_device_del for DT nodes, or replace platform_device_unregister with of_device_unregister in of_platform_device_destroy()? This is a common crash and we cannot use of_platform_depopulate() today in drivers to complement of_platform_populate(). Also, the platform_data crash is independent of this, I could reproduce that one even with using of_device_unregister in a loop in driver remove. regards Suman > > Rob > >> >> Signed-off-by: Suman Anna >> --- >> drivers/of/device.c | 38 +++++++++++++++++++++++++++++++++++++- >> 1 file changed, 37 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/of/device.c b/drivers/of/device.c >> index 46d6c75c1404..fa27c1c71f29 100644 >> --- a/drivers/of/device.c >> +++ b/drivers/of/device.c >> @@ -50,6 +50,8 @@ EXPORT_SYMBOL(of_dev_put); >> >> int of_device_add(struct platform_device *ofdev) >> { >> + int i, ret; >> + >> BUG_ON(ofdev->dev.of_node == NULL); >> >> /* name and id have to be set so that the platform bus doesn't get >> @@ -63,7 +65,41 @@ int of_device_add(struct platform_device *ofdev) >> if (!ofdev->dev.parent) >> set_dev_node(&ofdev->dev, of_node_to_nid(ofdev->dev.of_node)); >> >> - return device_add(&ofdev->dev); >> + for (i = 0; i < ofdev->num_resources; i++) { >> + struct resource *p, *r = &ofdev->resource[i]; >> + >> + if (!r->name) >> + r->name = dev_name(&ofdev->dev); >> + >> + p = r->parent; >> + if (!p) { >> + if (resource_type(r) == IORESOURCE_MEM) >> + p = &iomem_resource; >> + else if (resource_type(r) == IORESOURCE_IO) >> + p = &ioport_resource; >> + } >> + >> + if (p && insert_resource(p, r)) { >> + dev_err(&ofdev->dev, "failed to claim resource %d\n", >> + i); >> + ret = -EBUSY; >> + goto failed; >> + } >> + } >> + >> + ret = device_add(&ofdev->dev); >> + if (ret == 0) >> + return ret; >> + >> +failed: >> + while (--i >= 0) { >> + struct resource *r = &ofdev->resource[i]; >> + unsigned long type = resource_type(r); >> + >> + if (type == IORESOURCE_MEM || type == IORESOURCE_IO) >> + release_resource(r); >> + } >> + return ret; >> } >> >> int of_device_register(struct platform_device *pdev) >> -- >> 2.2.1 >> From mboxrd@z Thu Jan 1 00:00:00 1970 From: Suman Anna Subject: Re: [RFC PATCH 1/3] of/device: manage resources similar to platform_device_add Date: Tue, 13 Jan 2015 15:25:49 -0600 Message-ID: <54B58D5D.5070508@ti.com> References: <1420651854-17768-1-git-send-email-s-anna@ti.com> <1420651854-17768-2-git-send-email-s-anna@ti.com> Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Rob Herring Cc: Grant Likely , Rob Herring , Greg Kroah-Hartman , Pawel Moll , Pantelis Antoniou , Felipe Balbi , "devicetree@vger.kernel.org" , "linux-kernel@vger.kernel.org" , "linux-arm-kernel@lists.infradead.org" , linux-omap List-Id: devicetree@vger.kernel.org Hi Rob, On 01/13/2015 02:38 PM, Rob Herring wrote: > On Wed, Jan 7, 2015 at 11:30 AM, Suman Anna wrote: >> Drivers can use of_platform_populate() to create platform devices >> for children of the device main node, and a complementary API >> of_platform_depopulate() is provided to delete these child platform >> devices. The of_platform_depopulate() leverages the platform API >> for performing the cleanup of these devices. >> >> The platform device resources are managed differently between >> of_device_add and platform_device_add, and this asymmetry causes >> a kernel oops in platform_device_del during removal of the resources. >> Manage the platform device resources similar to platform_device_add >> to fix this kernel oops. > > This is a known issue and has been attempted to be fixed before (I > believe there is a revert in mainline). The problem is there are known > devicetrees which have overlapping resources and they will break with > your change. Are you referring to 02bbde7849e6 (Revert "of: use platform_device_add")? That one seems to be in registration path, and this crash is in the unregistration path. If so, to fix the crash, should we be skipping the release_resource() for now in platform_device_del for DT nodes, or replace platform_device_unregister with of_device_unregister in of_platform_device_destroy()? This is a common crash and we cannot use of_platform_depopulate() today in drivers to complement of_platform_populate(). Also, the platform_data crash is independent of this, I could reproduce that one even with using of_device_unregister in a loop in driver remove. regards Suman > > Rob > >> >> Signed-off-by: Suman Anna >> --- >> drivers/of/device.c | 38 +++++++++++++++++++++++++++++++++++++- >> 1 file changed, 37 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/of/device.c b/drivers/of/device.c >> index 46d6c75c1404..fa27c1c71f29 100644 >> --- a/drivers/of/device.c >> +++ b/drivers/of/device.c >> @@ -50,6 +50,8 @@ EXPORT_SYMBOL(of_dev_put); >> >> int of_device_add(struct platform_device *ofdev) >> { >> + int i, ret; >> + >> BUG_ON(ofdev->dev.of_node == NULL); >> >> /* name and id have to be set so that the platform bus doesn't get >> @@ -63,7 +65,41 @@ int of_device_add(struct platform_device *ofdev) >> if (!ofdev->dev.parent) >> set_dev_node(&ofdev->dev, of_node_to_nid(ofdev->dev.of_node)); >> >> - return device_add(&ofdev->dev); >> + for (i = 0; i < ofdev->num_resources; i++) { >> + struct resource *p, *r = &ofdev->resource[i]; >> + >> + if (!r->name) >> + r->name = dev_name(&ofdev->dev); >> + >> + p = r->parent; >> + if (!p) { >> + if (resource_type(r) == IORESOURCE_MEM) >> + p = &iomem_resource; >> + else if (resource_type(r) == IORESOURCE_IO) >> + p = &ioport_resource; >> + } >> + >> + if (p && insert_resource(p, r)) { >> + dev_err(&ofdev->dev, "failed to claim resource %d\n", >> + i); >> + ret = -EBUSY; >> + goto failed; >> + } >> + } >> + >> + ret = device_add(&ofdev->dev); >> + if (ret == 0) >> + return ret; >> + >> +failed: >> + while (--i >= 0) { >> + struct resource *r = &ofdev->resource[i]; >> + unsigned long type = resource_type(r); >> + >> + if (type == IORESOURCE_MEM || type == IORESOURCE_IO) >> + release_resource(r); >> + } >> + return ret; >> } >> >> int of_device_register(struct platform_device *pdev) >> -- >> 2.2.1 >> From mboxrd@z Thu Jan 1 00:00:00 1970 From: s-anna@ti.com (Suman Anna) Date: Tue, 13 Jan 2015 15:25:49 -0600 Subject: [RFC PATCH 1/3] of/device: manage resources similar to platform_device_add In-Reply-To: References: <1420651854-17768-1-git-send-email-s-anna@ti.com> <1420651854-17768-2-git-send-email-s-anna@ti.com> Message-ID: <54B58D5D.5070508@ti.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi Rob, On 01/13/2015 02:38 PM, Rob Herring wrote: > On Wed, Jan 7, 2015 at 11:30 AM, Suman Anna wrote: >> Drivers can use of_platform_populate() to create platform devices >> for children of the device main node, and a complementary API >> of_platform_depopulate() is provided to delete these child platform >> devices. The of_platform_depopulate() leverages the platform API >> for performing the cleanup of these devices. >> >> The platform device resources are managed differently between >> of_device_add and platform_device_add, and this asymmetry causes >> a kernel oops in platform_device_del during removal of the resources. >> Manage the platform device resources similar to platform_device_add >> to fix this kernel oops. > > This is a known issue and has been attempted to be fixed before (I > believe there is a revert in mainline). The problem is there are known > devicetrees which have overlapping resources and they will break with > your change. Are you referring to 02bbde7849e6 (Revert "of: use platform_device_add")? That one seems to be in registration path, and this crash is in the unregistration path. If so, to fix the crash, should we be skipping the release_resource() for now in platform_device_del for DT nodes, or replace platform_device_unregister with of_device_unregister in of_platform_device_destroy()? This is a common crash and we cannot use of_platform_depopulate() today in drivers to complement of_platform_populate(). Also, the platform_data crash is independent of this, I could reproduce that one even with using of_device_unregister in a loop in driver remove. regards Suman > > Rob > >> >> Signed-off-by: Suman Anna >> --- >> drivers/of/device.c | 38 +++++++++++++++++++++++++++++++++++++- >> 1 file changed, 37 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/of/device.c b/drivers/of/device.c >> index 46d6c75c1404..fa27c1c71f29 100644 >> --- a/drivers/of/device.c >> +++ b/drivers/of/device.c >> @@ -50,6 +50,8 @@ EXPORT_SYMBOL(of_dev_put); >> >> int of_device_add(struct platform_device *ofdev) >> { >> + int i, ret; >> + >> BUG_ON(ofdev->dev.of_node == NULL); >> >> /* name and id have to be set so that the platform bus doesn't get >> @@ -63,7 +65,41 @@ int of_device_add(struct platform_device *ofdev) >> if (!ofdev->dev.parent) >> set_dev_node(&ofdev->dev, of_node_to_nid(ofdev->dev.of_node)); >> >> - return device_add(&ofdev->dev); >> + for (i = 0; i < ofdev->num_resources; i++) { >> + struct resource *p, *r = &ofdev->resource[i]; >> + >> + if (!r->name) >> + r->name = dev_name(&ofdev->dev); >> + >> + p = r->parent; >> + if (!p) { >> + if (resource_type(r) == IORESOURCE_MEM) >> + p = &iomem_resource; >> + else if (resource_type(r) == IORESOURCE_IO) >> + p = &ioport_resource; >> + } >> + >> + if (p && insert_resource(p, r)) { >> + dev_err(&ofdev->dev, "failed to claim resource %d\n", >> + i); >> + ret = -EBUSY; >> + goto failed; >> + } >> + } >> + >> + ret = device_add(&ofdev->dev); >> + if (ret == 0) >> + return ret; >> + >> +failed: >> + while (--i >= 0) { >> + struct resource *r = &ofdev->resource[i]; >> + unsigned long type = resource_type(r); >> + >> + if (type == IORESOURCE_MEM || type == IORESOURCE_IO) >> + release_resource(r); >> + } >> + return ret; >> } >> >> int of_device_register(struct platform_device *pdev) >> -- >> 2.2.1 >>