From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753260AbdJaLAL (ORCPT ); Tue, 31 Oct 2017 07:00:11 -0400 Received: from userp1040.oracle.com ([156.151.31.81]:45576 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751393AbdJaLAK (ORCPT ); Tue, 31 Oct 2017 07:00:10 -0400 Date: Tue, 31 Oct 2017 13:59:57 +0300 From: Dan Carpenter To: Arvind Yadav Cc: gregkh@linuxfoundation.org, robsonde@gmail.com, tvboxspy@gmail.com, mayhs11saini@gmail.com, devel@driverdev.osuosl.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH] staging: rtl8192e: set priv->irq as 0 after the irq is freed Message-ID: <20171030105532.jm5i2royvm32iqqk@mwanda> References: <57b20b92720091f1cdb63f08714ef94a41aee885.1509171955.git.arvind.yadav.cs@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <57b20b92720091f1cdb63f08714ef94a41aee885.1509171955.git.arvind.yadav.cs@gmail.com> User-Agent: NeoMutt/20170609 (1.8.3) X-Source-IP: aserv0021.oracle.com [141.146.126.233] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sat, Oct 28, 2017 at 12:03:46PM +0530, Arvind Yadav wrote: > _rtl92e_init can fail here, we must set priv->irq as 0 after free_irq. > The changelog isn't useful. It should say that we are doing this because we call free_irq() again in _rtl92e_pci_probe() so it's a double free. This fix won't work. _rtl92e_pci_probe() doesn't check if priv->irq is zero, and also it frees dev->irq which is different from priv->irq. We should really just get rid of priv->irq. Really each allocation function should have a corresponding free function so there should be a _rtl92e_undo_init() which frees the memory and the IRQ. Currently we just free the IRQ and leak the memory. Anyway, the right fix is to just do this: --- a/drivers/staging/rtl8192e/rtl8192e/rtl_core.c +++ b/drivers/staging/rtl8192e/rtl8192e/rtl_core.c @@ -2524,7 +2524,7 @@ static int _rtl92e_pci_probe(struct pci_dev *pdev, RT_TRACE(COMP_INIT, "Driver probe completed1\n"); if (_rtl92e_init(dev) != 0) { netdev_warn(dev, "Initialization failed"); - goto err_free_irq; + goto err_unmap; } netif_carrier_off(dev); _rtl92e_init() is a useless name as well... :/ It would be nice to preserve the error codes, except they are all -1 which is lazy and wrong. regards, dan carpenter