Re: [PATCH v5 04/12] ltp/fsx.c: Add atomic writes support to fsx

[John Garry]

On 09/09/2025 07:57, Ojaswin Mujoo wrote:
> > > > > +
> > > > > +/*
> > > > > + * Round down n to nearest power of 2.
> > > > > + * If n is already a power of 2, return n;
> > > > > + */
> > > > > +static int rounddown_pow_of_2(int n) {
> > > > > +	int i = 0;
> > > > > +
> > > > > +	if (is_power_of_2(n))
> > > > > +		return n;
> > > > > +
> > > > > +	for (; (1 << i) < n; i++);
> > > > > +
> > > > > +	return 1 << (i - 1);
> > > > Is this the neatest way to do this?
> > > Well it is a straigforward o(logn) way. Do you have something else in
> > > mind?
> > check what the kernel does is always a good place to start...
> >
> > Thanks
> So kernel pretty much does same thing:
>
> unsigned long __rounddown_pow_of_two(unsigned long n)
> {
> 	return 1UL << (fls_long(n) - 1);
> }
>
> where fls*() variants jump into asm for efficiency. asm is obviously not
> needed here so we use the for loop to calculate the fls (find last bit
> set):
>
> 	for (; (1 << i) < n; i++);
>
> Ideally the compiler should do the right thing and optimize it.

I think that some versions of fls use builtin clz also, but I am not 
sure if that (clz) is always available for supported toolchains here.

Thanks,
John