[Linux-kernel-mentees] [PATCH] checkpatch: add fix option for COMMIT_LOG_LONG_LINE

Aditya yashsri421 at gmail.com
Mon Nov 2 13:31:08 UTC 2020


On 2/11/20 12:55 pm, Lukas Bulwahn wrote:
> 
> 
> On Sat, 31 Oct 2020, Aditya Srivastava wrote:
> 
>> Currently, if a line exceeds 75 characters in a commit message,
>> checkpatch.pl gives a warning corresponding to the same.
>>
>> E.g., running checkpatch on commit a1af2afbd244 ("drm/nouveau/volt:Fix
>> for some cards having 0 maximum voltage") reports this warning:
>>
>> WARNING: Possible unwrapped commit description (prefer a maximum 75
>> chars per line)
>>
>> Some, mostly Fermi, vbioses appear to have zero max voltage. That causes Nouveau to not parse voltage entries, thus users not being able to set higher clocks.
>>
>> Fix this warning by truncating the line after 75 chars, and push the
>> remaining content to the next line
>>
>> Signed-off-by: Aditya Srivastava <yashsri421 at gmail.com>
> 
> This is wrong on multiple levels:
> 
> 1. We must first ensure that the COMMIT_LOG_LONG_LINE are actually 
> largely true positives, i.e., it really makes sense to break those lines. 
> In my evaluations, I found many cases where it is clear that lines should 
> not be broken because it is tool output etc.
> 
> 2. You cannot just break at character 75. You would need to identify word 
> boundaries and break there.
> 
> 3. The proper fix is to identify the whole paragraph and then word-wrap 
> the whole paragraph with a limit of 75. Just adding some content in a new 
> line will make the commit message even more unreadable than before. So, 
> that is not a fix.
> 
> 
> So, I suggest to come up with a good strategy to really identify 
> COMMIT_LOG_LONG_LINE that sensibly can be word-wrapped, i.e.,
> distinguish between free-floating commit description and tool output,
> and then implement that word wrap properly.
> 
> Or to prioritize another issue and possiblity for fix rules.
> 
> You can see that there are potentially some good options for fixes around 
> the tags and sign-offs Dwaipayan has investigated.
> 

I agree. I'll take it up again after sometime.

Thanks
Aditya


More information about the Linux-kernel-mentees mailing list