On Tue, Oct 17, 2023 at 11:40 AM Jakub Kicinski [off-list ref] wrote:

On Tue, 17 Oct 2023 11:27:36 -0400 Jamal Hadi Salim wrote:

quoted

quoted

patch-by-patch W=1 C=1 should be good enough to catch the problems.

Thanks - this helps. We didnt pay good attention to
https://www.kernel.org/doc/Documentation/process/maintainer-netdev.rst
Only thing that is missing now is the mention of C=1 in the doc. Patch
to the doc acceptable?
Also a note about false positives in sparse output (there were a few
in the warnings from the bot) would be apropos.

Um. Maybe.. Sparse generates more false positives than good warnings
lately :( We'd have to add some extra info like "Note that sparse
is known to generate false-positive warnings, if you think that the
warning generated with C=1 is bogus, ignore it and note that fact
in the commit message".

I don't like documenting things which aren't clear-cut :(

Upto you - couldnt sum up from above if you want a patch or not. I
think it makes sense to document C=1 somewhere since it helps your
overhead.
But the comment Similar in spirit to the checkpatch comment if - "But
do not be mindlessly robotic in doing so..."

I'm pretty sure you have pure W=1 warnings here, too.

True - I am not sure how we missed one with function not used.

cheers,
jamal