Re: [PATCH] credit: track residual from divisions done during accounting

[David Vrabel <david.vrabel@citrix.com>]

On 26/02/13 11:52, Tim Deegan wrote:
> At 11:46 +0000 on 26 Feb (1361879193), Jan Beulich wrote:
>>>>> On 26.02.13 at 12:26, George Dunlap <george.dunlap@eu.citrix.com> wrote:
>>> On 02/25/2013 11:30 AM, Jan Beulich wrote:
>>>>>>> On 25.02.13 at 12:12, David Vrabel <david.vrabel@citrix.com> wrote:
>>>>> On 25/02/13 09:29, Jan Beulich wrote:
>>>>>>>>> On 22.02.13 at 18:26, Dario Faggioli <dario.faggioli@citrix.com> wrote:
>>>>>>> On Mon, 2013-02-18 at 12:37 +0000, Jan Beulich wrote:
>>>>>>>> +    ASSERT(credits == val);
>>>>>>>
>>>>>>> I may be missing something, but how can the assert ever be false, given
>>>>>>> the assignment right before it?
>>>>>>
>>>>>> val being wider than credit, this checks that there was no truncation.
>>>>>
>>>>> ASSERT(val <= UINT_MAX);
>>>>>
>>>>> Would be clearer.
>>>>
>>>> A matter of taste perhaps...
>>>
>>> I have a taste for coders having to keep as little state in their head 
>>> as possible. :-)  Comparing to UINT_MAX prompts the coder specifically 
>>> to think about the size of the variables.
>>
>> Okay, assuming this is the only thing you dislike, I'll change it then
>> and re-submit.
>>
>> But for the record - using UINT_MAX here will get things out of
>> sync the moment the type of "credits" changes, whereas with
>> the way I had coded it this would be taken care of implicitly.
> 
> How about ASSERT(((typeof credits) val) == val) before the assignment?

FWIW, this works for me.

David