Re: [PATCH] linux/bits.h: fix -Wtype-limits warnings in GENMASK_INPUT_CHECK()

[Vincent MAILHOL <mailhol.vincent@wanadoo.fr>]

On Sun. 6 Mar 2022 at 06:33, Andy Shevchenko <andy.shevchenko@gmail.com> wrote:
> On Sat, Mar 5, 2022 at 2:43 PM Vincent MAILHOL
> <mailhol.vincent@wanadoo.fr> wrote:
> > On Tue. 5 Mar 2022 at 03:46, Andy Shevchenko <andy.shevchenko@gmail.com> wrote:
> > > On Fri, Mar 4, 2022 at 7:36 PM Vincent Mailhol
> > > <mailhol.vincent@wanadoo.fr> wrote:
>
> ...
>
> > > NAK.
> >
> > Are you willing to change your decision following my comments?
>
> Have you read this discussion (read the thread in full)
> https://lore.kernel.org/lkml/cover.1590017578.git.syednwaris@gmail.com/

Thank you, this was an instructive read.

For what I understand, there was an effort to fix this when
-Wtype-limits was still a W=1 warning but the effort was stopped
after -Wtype-limits was moved to W=2 despite a v4 patch being very
close to the goal.

Back to my patch, it successfully passes the lib/test_bits.c
build test (including the TEST_GENMASK_FAILURES) and it also
fixes the last open warning from the thread you pointed me to (on
drivers/crypto/inside-secure/safexcel.o):
https://lore.kernel.org/lkml/20200709123011.GA18734@gondor.apana.org.au/

So, I am still not sure to understand what issue you see with my
patch. Is it that we should just not care about fixing W=2? Or
do you still see some issues which are not being addressed (if
so, sorry for not understanding)?

I do agree that fixing a W=2 has small value for all the files
which are still emitting some W=1. However, I think it is
beneficial to remove this W=2 spam for all the developers who
produced W=1 clean files and would like to tackle the W=2
warnings.


Yours sincerely,
Vincent Mailhol