From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758073Ab2CNLiL (ORCPT ); Wed, 14 Mar 2012 07:38:11 -0400 Received: from e28smtp02.in.ibm.com ([122.248.162.2]:55607 "EHLO e28smtp02.in.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755834Ab2CNLiI (ORCPT ); Wed, 14 Mar 2012 07:38:08 -0400 Message-ID: <4F608251.9030307@linux.vnet.ibm.com> Date: Wed, 14 Mar 2012 17:04:41 +0530 From: "Srivatsa S. Bhat" User-Agent: Mozilla/5.0 (X11; Linux i686; rv:10.0.1) Gecko/20120209 Thunderbird/10.0.1 MIME-Version: 1.0 To: Julia Lawall CC: Len Brown , kernel-janitors@vger.kernel.org, linux-acpi@vger.kernel.org, linux-kernel@vger.kernel.org, Julia Lawall , Deepthi Dharwar , Venkatesh Pallipadi , trenn@suse.de, bhelgaas@google.com Subject: Re: [PATCH] drivers/acpi/processor_driver.c: add missing kfree References: <1331399145-6508-1-git-send-email-Julia.Lawall@lip6.fr> In-Reply-To: <1331399145-6508-1-git-send-email-Julia.Lawall@lip6.fr> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit x-cbid: 12031411-5816-0000-0000-000001BB5007 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 03/10/2012 10:35 PM, Julia Lawall wrote: > From: Julia Lawall > > The function acpi_processor_add is stored in the ops.add field of a > acpi_driver structure. This function is then called in > acpi_bus_driver_init. On failure, this function clears the field > device->driver_data, but does not free its contents. Thus the free has to > be done by the add function. In acpi_processor_add, the corresponding > value is pr. This value is currently freed on failure before storing it in > device->driver_data, but not after. This free is added in the error > handling code at the end of the function. The static global variable "static global variable"?? never heard that one before ;-) Maybe you meant "per_cpu variable processors".. > processors is also cleared so that it does not refer to a dangling pointer. > processor_device_array is cleared as well. > > Signed-off-by: Julia Lawall > > --- > This is only compile tested. In particular, I don't know if it is correct > to add per_cpu(processor_device_array, pr->id) = NULL;. No, you shouldn't set it to NULL. processor_device_array was added to check for buggy BIOSes and return gracefully. Check commit cd8e2b48d (and also the bugzilla link in that commit). To have a robust check for buggy BIOSes, we must let it be as it is, even if it is a stale pointer. That way we can still catch subsequent calls to this function with the same acpi id (because of a buggy BIOS) and take appropriate actions. Other than that, the patch looks good to me. Regards, Srivatsa S. Bhat IBM Linux Technology Center > It has nothing to > do with pr, but it looks like stale information in the case of a failure. > > drivers/acpi/processor_driver.c | 12 ++++++++---- > 1 file changed, 8 insertions(+), 4 deletions(-) > > diff --git a/drivers/acpi/processor_driver.c b/drivers/acpi/processor_driver.c > index 2801b41..9bb0017 100644 > --- a/drivers/acpi/processor_driver.c > +++ b/drivers/acpi/processor_driver.c > @@ -536,8 +536,8 @@ static int __cpuinit acpi_processor_add(struct acpi_device *device) > return -ENOMEM; > > if (!zalloc_cpumask_var(&pr->throttling.shared_cpu_map, GFP_KERNEL)) { > - kfree(pr); > - return -ENOMEM; > + result = -ENOMEM; > + goto err_free_pr; > } > > pr->handle = device->handle; > @@ -577,7 +577,7 @@ static int __cpuinit acpi_processor_add(struct acpi_device *device) > dev = get_cpu_device(pr->id); > if (sysfs_create_link(&device->dev.kobj, &dev->kobj, "sysdev")) { > result = -EFAULT; > - goto err_free_cpumask; > + goto err_clear_processors; > } > > /* > @@ -595,9 +595,13 @@ static int __cpuinit acpi_processor_add(struct acpi_device *device) > > err_remove_sysfs: > sysfs_remove_link(&device->dev.kobj, "sysdev"); > +err_clear_processors: > + per_cpu(processors, pr->id) = NULL; > + per_cpu(processor_device_array, pr->id) = NULL; > err_free_cpumask: > free_cpumask_var(pr->throttling.shared_cpu_map); > - > +err_free_pr: > + kfree(pr); > return result; > } > >