[Cocci] Checking run times for transformation of Linux source code with SmPL

[Julia Lawall]

In terms of the running time, I get a running time of 11 seconds with the
command line of 48 files and I get a running time of 22 seconds if I just
run spatch on the entire kernel.

* Can such a comparison result be amazing?

No, I don't see anything suprising about it.

I suggest to take another look at these performance numbers from your
powerful test environment. All Linux source files were checked to some degree
according to another specific search pattern.
The half of this execution time was taken by only a tiny fraction
of update candidates.

The 48 files are all parsed. Coccinelle doesn't parse every file in the
kernel. It only parses the files that contain the name of the function
being modified.

julia

More work takes more time.

This usual in principle. It seems that I got a bit of understanding difficulties
for this simple test report.
How reasonable is it that a lot of more work could be performed somehow
in the double time range?
Regards,
Markus

[Julia Lawall]

Coccinelle doesn't parse every file in the kernel.

I find this information strange.

It only parses the files that contain the name of the function being modified.

Do you interpret the required source code parsing efforts in a special way?

Coccinelle searches the semantic patch for the important words. Then it
implements something like grep to find files that contain those imprtant
words. Or maybe it just uses git grep by default.

julia