Re: [Xen-devel] [PATCH] xen-netback: correct return value checks on xenbus_scanf()

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 07.07.16 at 12:55, <Paul.Durrant@xxxxxxxxxx> wrote:
>>  -----Original Message-----
>> From: netdev-owner@xxxxxxxxxxxxxxx [mailto:netdev-
>> owner@xxxxxxxxxxxxxxx] On Behalf Of David Vrabel
>> Sent: 07 July 2016 11:45
>> To: Wei Liu; David Vrabel
>> Cc: xen-devel@xxxxxxxxxxxxxxxxxxxx; Jan Beulich; netdev@xxxxxxxxxxxxxxx 
>> Subject: Re: [Xen-devel] [PATCH] xen-netback: correct return value checks
>> on xenbus_scanf()
>> 
>> On 07/07/16 11:35, Wei Liu wrote:
>> > On Thu, Jul 07, 2016 at 10:58:16AM +0100, David Vrabel wrote:
>> >> On 07/07/16 08:57, Jan Beulich wrote:
>> >>> Only a positive return value indicates success.
>> >>
>> >> This is not correct.
>> >>
>> >
>> > Do you mean the commit message is not correct or the code is not
>> > correct? If it is the formal, do you have any suggestion to fix it?
>> 
>> This code is correct as-is, thus the commit message is wrong or misleading.
> 
> Is that true? Jan is correct in saying that only >0 is an indicator of 
> success according to the usual semantics of sccanf().

As was correctly pointed out, xenbus_scanf(), other than scanf(),
can't return zero right now (which I think has corner cases where
this might be a problem). So if I would get the feeling that a
correction (benign or not at this point in time) would be accepted,
what about "Only a positive return value is guaranteed to indicates
success" as commit description?

> Personally I think the 
> code would be clearer if the checks for failure were < 1 rather than <= 0.

I'd be fine with that, albeit if comparing with any non-zero number
then I think it would better be == or != instead of < or <=.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel