arm64: include/linux/compiler_types.h:542:38: error: call to '__compiletime_assert_1050' declared with attribute error: clamp() low limit min greater than high limit max_avail

[Dan Carpenter]

On Wed, Dec 11, 2024 at 01:46:11PM +0100, Bartosz Golaszewski wrote:
> On Fri, Dec 6, 2024 at 3:20 AM David Laight <David.Laight at aculab.com> wrote:
> >
> > From: Naresh Kamboju
> > > Sent: 05 December 2024 18:42
> > >
> > > On Thu, 5 Dec 2024 at 20:46, Dan Carpenter <dan.carpenter at linaro.org> wrote:
> > > >
> > > > Add David to the CC list.
> > >
> > > Anders bisected this reported issue and found the first bad commit as,
> > >
> > > # first bad commit:
> > >   [ef32b92ac605ba1b7692827330b9c60259f0af49]
> > >   minmax.h: use BUILD_BUG_ON_MSG() for the lo < hi test in clamp()
> >
> > That 'just' changed the test to use __builtin_constant_p() and
> > thus gets checked after the optimiser has run.
> >
> > I can paraphrase the code as:
> > unsigned int fn(unsigned int x)
> > {
> >         return clamp(10, 5, x == 0 ? 0 : x - 1);
> > }
> > which is never actually called with x <= 5.
> > The compiler converts it to:
> >         return x < 0 ? clamp(10, 5, 0) : clamp(10, 5, x);
> > (Probably because it can see that clamp(10, 5, 0) is constant.)
> > And then the compile-time sanity check in clamp() fires.
> >
> > The order of the arguments to clamp is just wrong!
> >
> >         David
> >
> 
> The build is still failing with today's next, should the offending
> commit be reverted?
> 

It's a simple fix.  I've sent a patch.

regards,
dan carpenter