Re: [cocci] Working with parameter/expression lists by SmPL

[Jakob Koschel <jkl820.git@gmail.com>]

> On 18. Aug 2022, at 19:42, Markus Elfring <Markus.Elfring@web.de> wrote:
> 
>> Let me show what I was trying to archive and how I solved it:
> 
> 
> How do you think about to achieve further software adjustments?

I'm trying to detect certain code patterns spanning across files with this.
They will *probably* not easily be automatically patchable since I'm trying to
find bugs with this in an ongoing research project.

> 
> 
> …
> 
>>    interesting_func@p(..., i, ...)
> …
> 
> I interpret such SmPL code in the way that the specified position variable
> could be omitted because it was not used by another subsequent SmPL rule.

You're right, the positions can be omitted here, but I'm using them with a
simple python script afterwards to report the matches I found where the positions
will be useful.

> 
> 
> …
> 
>>    e = e1
>>    + // here the argument to interesting_func was assigned
> …
>> +++ /tmp/cocci-output-test1.c
>> @@ -6,7 +6,8 @@ int interesting_func(int argc, char *str
>> }
>> 
>> int main(int argc, char *argv[]) {
>> -  int x = 0;
>> +  int x = 0
>> +    // test;
>> 
>>   func1(x, "asdf");
>> }
> 
> 
> Do you vary comment additions for your test results?

yes exactly.

Only issue I'm having now is that it doesn't work across files.

I've found this [Link], suggesting to use '--file-groups' or '--kbuild-info' to 
o analysis across multiple files. I've managed to get '--file-groups' working
on a small example but it will get challenging to do on the entire kernel.
The '--kbuild-info' option *looks* outdated at least I couldn't find an easy way to get
it running on the kernel.

- jakob

> 
> Regards,
> Markus
> 

Link: https://lore.kernel.org/all/alpine.DEB.2.20.1711300716120.2079@hadrien/t/#u