Re: [PATCH 1/1] x86: Fix mcheck_init_device() to handlemisc_register() correctly

[Johan Wessfeldt]

On Fri, Mar 4, 2011 at 6:16 PM, Borislav Petkov <bp@xxxxxxxxx> wrote:
> On Fri, Mar 04, 2011 at 01:29:27PM +0000, J.P. Lacerda wrote:
>> The return value for misc_register() was not being taken into account.
>> Furthermore, if misc_register() fails, we must rollback any changes made by
>> mcheck_init_device()
>
> If you're going to fix all error paths here, you still need to handle
> unrolling the setup done by mce_create_device() if we fail somewhere in
> between.
>
>> Signed-off-by: J.P. Lacerda <jp.lacerda@xxxxxxxxxxxxxxx>
>> ---
>>  arch/x86/kernel/cpu/mcheck/mce.c |   13 ++++++++++++-
>>  1 files changed, 12 insertions(+), 1 deletions(-)
>>
>> diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
>> index d916183..20c2c44 100644
>> --- a/arch/x86/kernel/cpu/mcheck/mce.c
>> +++ b/arch/x86/kernel/cpu/mcheck/mce.c
>> @@ -2140,7 +2140,18 @@ static __init int mcheck_init_device(void)
>>       }
>>
>>       register_hotcpu_notifier(&mce_cpu_notifier);
>> -     misc_register(&mce_log_device);
>> +     err = misc_register(&mce_log_device);
>> +
>
> no newline here.
>
>> +     if (err) {
>> +             unregister_hotcpu_notifier(&mce_cpu_notifier);
>> +
>> +             for_each_online_cpu(i) {
>> +                     mce_remove_device(i);
>> +             }
>
> no need for braces around a single loop body statement.
>
>> +
>> +             sysdev_class_unregister(&mce_sysclass);
>> +             free_cpumask_var(mce_dev_initialized);
>> +     }
>>
>>       return err;
>
> Anyway, while this is makes sense from correctness POV, if we hit an
> error path here this early then something else is going terribly wrong
> which would've screamed very loudly already. Are you hitting this on a
> real workload or you caught this by code staring?
The orginal patch was made with the intention of auditing the code
according to the kernel-janitors TODO list:
http://kernelnewbies.org/KernelJanitors/Todo/ReturnCodes .

To clear things up. I originally posted a minor patch, which basically passed
the return value from misc_register up the stack. See
http://marc.info/?l=linux-kernel&m=129889198732342&w=2

The patch was rejected with comments about how to clean everything up
if misc_register() fails.

Here's my second attempt:
http://marc.info/?l=linux-kernel&m=129926353413345&w=2

(Dan Carpenter just explained to me how to go about resubmitting the
patches thanks)

Sorry for the confusion.
>
> Because if it is the second case, the merit of fixing those error
> paths vs adding code which is almost never executed is significantly
> diminished.
>
> Thanks.
>
> --
> Regards/Gruss,
> Boris.
>
> Advanced Micro Devices GmbH
> Einsteinring 24, 85609 Dornach
> General Managers: Alberto Bozzo, Andrew Bowd
> Registration: Dornach, Gemeinde Aschheim, Landkreis Muenchen
> Registergericht Muenchen, HRB Nr. 43632
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/