[Linux-kernel-mentees] RRRe: Evaluate and Improve checkpatch: tasks

Lukas Bulwahn lukas.bulwahn at gmail.com
Wed Sep 16 12:06:56 UTC 2020



On Wed, 16 Sep 2020, Ujjwal Kumar wrote:

> On 13/09/20 4:45 pm, Lukas Bulwahn wrote:
> > 
> > 
> > On Sun, 13 Sep 2020, Ujjwal Kumar wrote:
> > 
> >> On 12/09/20 6:02 pm, Lukas Bulwahn wrote:
> >>> On Sat, 12 Sep 2020, Ujjwal Kumar wrote:
> >>>
> >>>> On 12/09/20 5:14 pm, Lukas Bulwahn wrote:
> >>>>>
> >>>>>
> >>>>> On Sat, 12 Sep 2020, Ujjwal Kumar wrote:
> >>>>>
> >>>>>> Hello Lukas,
> >>>>>>
> >>>>>> I attempted the tasks.
> >>>>>>
> >>>>>> Below is the link of GitHub repository containing my write-up and findings.
> >>>>>> https://github.com/Ujjwal0501/lkmp-checkpatch
> >>>>>>
> >>>>>
> >>>>> Thanks, it shows 404. Good luck next time.
> >>>>
> >>>> I'm really sorry for the noise. The link is active now.
> >>>
> >>> Okay, scripts and data looks sound. So, let us try those scripts.
> >>>
> >>> Please have a look at this patch:
> >>>
> >>> https://lore.kernel.org/linux-kernel-mentees/20200912094826.150170-1-ayush@disroot.org/
> >>>
> >>> The author states:
> >>>
> >>> This issue was discovered through a thorough analysis of checkpatch.pl
> >>> errors and warnings of type GIT_COMMIT_ID on commits between v5.7 and
> >>> v5.8.
> >>>
> >>> Before applying this patch, checkpatch.pl reported 342 errors of type
> >>> GIT_COMMIT_ID. After applying patch, errors reduced to 284.
> >>>
> >>>
> >>> If your scripts work, you should be able to confirm the statement.
> >>>
> >>
> >> I'm unable to reproduce the same count as above because the script is run 
> >> on different sets of commits.
> >>
> > 
> > So, you say the evaluation from Ayush was not really a good evaluation of 
> > v5.7..v.5.8 due to some misconception of Ayush how to obtain the set of 
> > proper commits from v5.7..v5.8?
> 
> That is not totally correct as he correctly generated the list of commits in 
> the first task. But the script used to generate the statistics used a 
> different way so I was trying to point that specific thing.
> 
> >>
> >>> The tasks are:
> >>>
> >>> 1. Run your scripts and create a full statistics of all error types with
> >>> their according count for v5.7..v5.8.
> >>>
> >>> 2. Apply the patch with git am.
> >>>
> >>> 3. Run your scripts again and create a new statistics.
> >>>
> >>> 4. Compare before and after
> >>>
> >>> 5. Make all results available on your github repository.
> >>>
> >>
> >> The results are available at the link 
> >> https://github.com/Ujjwal0501/lkmp-checkpatch/blob/master/tasks.md
> >>
> >> Kindly take a look at it.
> >>
> > 
> > Can you share the errors that have changed after the patch was applied?
> > 
> > - Which cases of GIT_COMMIT_ID have disappeared?
> > - Which cases of GIT_COMMIT_ID are new?
> > - Which cases of GIT_COMMIT_ID have changed?
> > - Have cases of other type of errors/warnings/checks/notes disappeared, 
> > been added or changed?
> > 
> > You state GIT_COMMIT_ID changed from 272 to 230, but the tester in
> > 
> > https://lore.kernel.org/linux-kernel-mentees/CABJPP5BFNjBhdcvBWqDYnR-o2sZ0oY77bYr9PxducXW4Y=482Q@mail.gmail.com/
> > 
> > claimed that it changed from 270 to 251.
> > 
> > Which numbers are right? Why is there a difference in both statistics?
> 
> The patch from Ayush appears to have a bug which I have detailed in my write up.
> 
> Kindly take a look at the link.
> https://github.com/Ujjwal0501/lkmp-checkpatch/blob/master/followuptask1.md
>

Ayush, please fix the bug.

Ujjwal, please do not create screenshots. That is wasteful and harmful to 
the environment.

You can share the same information by sharing plain text in nicely 
formated ways.

Ujjwal, I do not understand the diff you presented. We are really not 
acquainted with your stat file, maybe you can explain what you read out of 
that.

Which results are from which exact commit?

Ujjwal, can you judge if the findings that disappeared should have not 
been warned about in the first place?

And the ones that appear new? Should they be really warned about or not?

Are there further patterns in the GIT_COMMIT_ID findings that probably 
should not be warned about and could be improved?


Lukas


More information about the Linux-kernel-mentees mailing list