[ldv-project] Questions on LDV Specifications and Variability

[Evgeny Novikov]

Hi Alexander

On 09/12/2013 11:51 AM, Alexander von Rhein wrote:
> Hi Evgeny,
>
> sorry, that I have not written for such a long time ;)
> I am still working on the driver-simulator generation, but it is much 
> more difficult than we anticipated.
>
I understand it ;)
> We are working on another project in parallel, for which we have some 
> questions for you.
> Perhaps you already have some experience on this.
>
> 1. Are the LDV-Specifications (32_1, 43_x, ...) appliccable only to a 
> special group of drivers or can they be used with any driver?
They are very general and can be used for all drivers and even more (for 
net protocols, file systems, etc.). But this doesn't mean that all 
drivers use those elements of the Linux kernel API that are described 
and checked in specifications.
32* specifications check that macros and functions for work with mutexes 
are used properly. You can see how many "drivers" use them: 
http://lxr.free-electrons.com/ident?i=mutex_lock.
>
> 2. We have seen that some of the drivers have only very little #ifdef 
> variability.
> Do you know of any drivers that have more variability (which also 
> affects functionality)?
>
I often see that CONFIG_PM affects drivers: 
http://lxr.free-electrons.com/ident?i=CONFIG_PM. As for other #ifdefs, 
you can try to see on code coverage. At the moment we have not a good 
account with coverage due to some internal LDV transformations, but 
corresponding tasks are on our road map. I hope that we will have a very 
pretty account this month. Then we will send it to you.
> Thank you for your help.
> --Alex


-- 
Best regards, Evgeny Novikov.