Re: [PATCH] bitmap: simplify GENMASK(size - 1, 0) lines

From: Yury Norov
Date: Tue Oct 26 2021 - 15:28:55 EST


On Tue, Oct 26, 2021 at 10:21:58PM +0300, Andy Shevchenko wrote:
> On Tue, Oct 26, 2021 at 11:54:16AM -0700, Yury Norov wrote:
> > On Tue, Oct 26, 2021 at 10:41:08AM -0400, Qian Cai wrote:
> > > Since "size" is an "unsigned int", the rvalue "size - 1" will still be
> > > "unsigned int" according to the C standard (3.2.1.5 Usual arithmetic
> > > conversions). Therefore, GENMASK(size - 1, 0) will always return 0UL. Those
> > > are also caught by GCC (W=2):
> > >
> > > ./include/linux/find.h: In function 'find_first_bit':
> > > ./include/linux/bits.h:25:22: warning: comparison of unsigned expression in '< 0' is always false [-Wtype-limits]
> > > 25 | __is_constexpr((l) > (h)), (l) > (h), 0)))
> > > | ^
> > > ./include/linux/build_bug.h:16:62: note: in definition of macro 'BUILD_BUG_ON_ZERO'
> > > 16 | #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
> > > | ^
> > > ./include/linux/bits.h:25:3: note: in expansion of macro '__is_constexpr'
> > > 25 | __is_constexpr((l) > (h)), (l) > (h), 0)))
> > > | ^~~~~~~~~~~~~~
> > > ./include/linux/bits.h:38:3: note: in expansion of macro 'GENMASK_INPUT_CHECK'
> > > 38 | (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
> > > | ^~~~~~~~~~~~~~~~~~~
> > > ./include/linux/find.h:119:31: note: in expansion of macro 'GENMASK'
> > > 119 | unsigned long val = *addr & GENMASK(size - 1, 0);
> > > | ^~~~~~~
> > > ./include/linux/bits.h:25:34: warning: comparison of unsigned expression in '< 0' is always false [-Wtype-limits]
> > > 25 | __is_constexpr((l) > (h)), (l) > (h), 0)))
> > > | ^
> > > ./include/linux/build_bug.h:16:62: note: in definition of macro 'BUILD_BUG_ON_ZERO'
> > > 16 | #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
> > > | ^
> > > ./include/linux/bits.h:38:3: note: in expansion of macro 'GENMASK_INPUT_CHECK'
> > > 38 | (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
> > > | ^~~~~~~~~~~~~~~~~~~
> > > ./include/linux/find.h:119:31: note: in expansion of macro 'GENMASK'
> > > 119 | unsigned long val = *addr & GENMASK(size - 1, 0);
> > > | ^~~~~~~
> > >
> > > Signed-off-by: Qian Cai <quic_qiancai@xxxxxxxxxxx>
> > > ---
> > > include/linux/find.h | 28 ++++++++--------------------
> > > 1 file changed, 8 insertions(+), 20 deletions(-)
> > >
> > > diff --git a/include/linux/find.h b/include/linux/find.h
> > > index 5bb6db213bcb..5ce2b17aea42 100644
> > > --- a/include/linux/find.h
> > > +++ b/include/linux/find.h
> > > @@ -115,11 +115,8 @@ unsigned long find_next_zero_bit(const unsigned long *addr, unsigned long size,
> > > static inline
> > > unsigned long find_first_bit(const unsigned long *addr, unsigned long size)
> > > {
> > > - if (small_const_nbits(size)) {
> > > - unsigned long val = *addr & GENMASK(size - 1, 0);
> > > -
> > > - return val ? __ffs(val) : size;
> > > - }
> > > + if (small_const_nbits(size))
> > > + return size;
> > >
> > > return _find_first_bit(addr, size);
> > > }
> >
> > [...]
> >
> > Nice catch! I'm a bit concerned that small_const_nbits() will never
> > allow GENMASK() to be passed with size == 0, but the patch looks
> > good to me overall.
>
> Can you explain to me how it is supposed to work?
>
> For example,
>
> x = 0xaa55;
> size = 5;
>
> printf("%lu\n", find_first_bit(&x, size));
>
> In the resulting code we will always have 5 as the result,
> but is it correct one?

I think it would work really bad and fail to load the kernel
for many systems, especially those with NR_CPUS == 64 or less.

That's why I think Apr 1 branch is a good place for it.