On Thu, 2020-05-07 at 15:45 -0400, Steven Rostedt wrote:
> On Thu, 07 May 2020 12:06:56 -0700
> Joe Perches <j...@perches.com> wrote:
>
> > People describe changes as a "fix" all the time for stuff
> > that isn't an actual fix for a logic defect but is instead
> > an update to a particular style preference.
> >
> > Then the "fix" word causes the patch to be rather uselessly
> > applied to stable trees by AUTOSEL.
> >
> > It's especially bad when the 'Fixes: <sha1> ("description")'
> > tag is also added.
> >
> > It's a difficult thing to regulate and I don't believe a
> > good mechanism would be possible to add to checkpatch or
> > coccinelle to help isolate these things.
> >
> > git diff -w sometimes helps, but that's not really a thing
> > that checkpatch could do.
> >
> > Any suggestions?
>
> I'm unfamiliar with how the coccinelle script is used, but I thought there
> was some discussion some time back to have checkpatch not produces the same
> kinds of warnings to code as it does to patches.
>
> A lot of useless updates were being submitted when people were running
> checkpatch on existing kernel code and producing warnings that are not
> worth "fixing", but something that new code should try to avoid.
checkpatch already has several blocks that look like
if (input_is_a_patch)
warn(...)
else if (input_is_a_file)
check(...)
where by default, check() is not output.
I've also suggested variations discouraging checkpatch
use on files outside of drivers/staging/ multiple times
https://lore.kernel.org/lkml/0753cae7829b98998ac3f5f9fcb52ba1f2475ee1.ca...@perches.com/
