Re: [PATCH 1/2] xen: correct error handling in xen-acpi-processor

From: Juergen Gross
Date: Fri Mar 31 2017 - 10:19:36 EST


On 31/03/17 16:10, Boris Ostrovsky wrote:
> On 03/31/2017 07:19 AM, Juergen Gross wrote:
>> Error handling in upload_pm_data() is rather questionable: possible
>> errno values from called functions are or'ed together resulting in
>> strange return values: -EINVAL and -ENOMEM would e.g. result in
>> returning -EXDEV.
>
> And it doesn't matter anyway since noone checks return value. (not that
> OR-ing errors makes much sense)

Ha, you are right!

So I can just fold the patches into one and turn some more functions
into retuning void.


Juergen

>
>>
>> Fix this by bailing out early after the first error.
>
> I am not sure about this: why should we not try to load P states if C
> states failed to load?
>
> -boris
>
>>
>> Cc: stable@xxxxxxxxxxxxxxx
>> Signed-off-by: Juergen Gross <jgross@xxxxxxxx>
>> ---
>> drivers/xen/xen-acpi-processor.c | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/xen/xen-acpi-processor.c b/drivers/xen/xen-acpi-processor.c
>> index 23e391d..3659ed3 100644
>> --- a/drivers/xen/xen-acpi-processor.c
>> +++ b/drivers/xen/xen-acpi-processor.c
>> @@ -283,8 +283,8 @@ static int upload_pm_data(struct acpi_processor *_pr)
>> if (_pr->flags.power)
>> err = push_cxx_to_hypervisor(_pr);
>>
>> - if (_pr->performance && _pr->performance->states)
>> - err |= push_pxx_to_hypervisor(_pr);
>> + if (!err && _pr->performance && _pr->performance->states)
>> + err = push_pxx_to_hypervisor(_pr);
>>
>> mutex_unlock(&acpi_ids_mutex);
>> return err;
>
>