From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752371Ab1KEOjZ (ORCPT ); Sat, 5 Nov 2011 10:39:25 -0400 Received: from imr4.ericy.com ([198.24.6.9]:46192 "EHLO imr4.ericy.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751311Ab1KEOjY (ORCPT ); Sat, 5 Nov 2011 10:39:24 -0400 Date: Sat, 5 Nov 2011 07:38:01 -0700 From: Guenter Roeck To: Jean Delvare CC: Alexander Beregalov , "linux-kernel@vger.kernel.org" , "stable@kernel.org" Subject: Re: [PATCH] hwmon: (w83627ehf): fix broken driver init Message-ID: <20111105143801.GA14535@ericsson.com> References: <1320454450-738-1-git-send-email-a.beregalov@gmail.com> <20111105045156.GA10668@ericsson.com> <20111105091351.0d436eb5@endymion.delvare> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Disposition: inline In-Reply-To: <20111105091351.0d436eb5@endymion.delvare> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sat, Nov 05, 2011 at 04:13:51AM -0400, Jean Delvare wrote: > Hi Alexander, Guenter, > > On Fri, 4 Nov 2011 21:51:56 -0700, Guenter Roeck wrote: > > On Fri, Nov 04, 2011 at 08:54:10PM -0400, Alexander Beregalov wrote: > > > Commit 2265cef2 (hwmon: (w83627ehf) Properly report PECI and AMD-SI > > > sensor types) results in kernel panic if data->temp_label was not > > > initialized. > > > The problem was found with chip W83627DHG-P. > > > > > > backtrace: > > > page_fault() > > > w83627ehf_probe() 0x8e2 > > > > > > w83627ehf_init_device: > > > 1837: for (i = 0; i < 3; i++) { > > > const char *label = data->temp_label[data->temp_src[i]]; > > > > > > /* Digital source overrides analog type */ > > > if (strncmp(label, "PECI", 4) == 0) > > > data->temp_type[i] = 6; > > > > > > movzbl 128(%r14), %edx # data->temp_src, tmp737 > > > movq 144(%r13), %rax # data_32->temp_label, data_32->temp_label > > > movq $.LC18, %rsi #, > > > movl %r14d, %r15d #, i > > > subl %r13d, %r15d # D.29277, i > > > 0x928 movq (%rax,%rdx,8), %rcx #* data_32->temp_label, label > > > movl $4, %edx #, > > > movq %rcx, %rdi # label, > > > movq %rcx, -96(%rbp) #, > > > call strncmp # > > > > > > page_fault() is called at 0x928 (0x8e2+0x46). > > Oops, my bad. I forgot that not all chips have temperature labels, > sorry :( Thanks for reporting so quickly! > > > > > > > Add check if data->temp->label was set before use. > > > > > > Cc: Jean Delvare > > > Cc: Guenter Roeck > > > Cc: stable@kernel.org > > > Signed-off-by: Alexander Beregalov > > > --- > > > drivers/hwmon/w83627ehf.c | 17 +++++++++-------- > > > 1 files changed, 9 insertions(+), 8 deletions(-) > > > > > > diff --git a/drivers/hwmon/w83627ehf.c b/drivers/hwmon/w83627ehf.c > > > index 483cb26..70f1c73 100644 > > > --- a/drivers/hwmon/w83627ehf.c > > > +++ b/drivers/hwmon/w83627ehf.c > > > @@ -1835,14 +1835,15 @@ static inline void __devinit w83627ehf_init_device(struct w83627ehf_data *data, > > > diode = 0x70; > > > } > > > for (i = 0; i < 3; i++) { > > > - const char *label = data->temp_label[data->temp_src[i]]; > > > - > > > - /* Digital source overrides analog type */ > > > - if (strncmp(label, "PECI", 4) == 0) > > > - data->temp_type[i] = 6; > > > - else if (strncmp(label, "AMD", 3) == 0) > > > - data->temp_type[i] = 5; > > > - else if ((tmp & (0x02 << i))) > > > + if (data->temp_label) { > > > + const char *label = data->temp_label[data->temp_src[i]]; > > > + > > > + /* Digital source overrides analog type */ > > > + if (strncmp(label, "PECI", 4) == 0) > > > + data->temp_type[i] = 6; > > > + else if (strncmp(label, "AMD", 3) == 0) > > > + data->temp_type[i] = 5; > > > + } else if ((tmp & (0x02 << i))) > > > > Followup - something like > > const char *label = NULL; > > > > if (data->temp_label) > > label = data->temp_label[data->temp_src[i]]; > > > > /* Digital source overrides analog type */ > > if (label && strncmp(label, "PECI", 4) == 0) > > data->temp_type[i] = 6; > > else if (label && strncmp(label, "AMD", 3) == 0) > > data->temp_type[i] = 5; > > else if ((tmp & (0x02 << i))) > > > > should do it. I am open to better ideas... > > Looks good enough to me, this is one-time initialization code, it > doesn't need to be lightning fast. Will you send a formal patch, or do > you want me to do it? > I'll send one in a minute. Guenter