Re: [PATCH bpf-next 2/4] bpf: track find_equal_scalars history on per-instruction level

[Eduard Zingerman <eddyz87@gmail.com>]

On Wed, 2024-02-28 at 13:36 -0800, Andrii Nakryiko wrote:
[...]

> I'd say it's not about a number of lines, it's about ease of
> understanding, reasoning, and using these helpers.
> 
> I do prefer the code you wrote below, but I'm not going to die on this
> hill if you insist. I'll go think about the rest of the logic.

Ok, code is meant to be read, so I'll switch to below in v2.

[...]

> > static u64 linked_reg_set_pack(struct linked_reg_set *s)
> > {
> >         u64 val = 0;
> >         int i;
> > 
> >         for (i = 0; i < s->cnt; ++i) {
> >                 struct reg_or_spill *r = &s->reg_set[i];
> >                 u64 tmp = 0;
> > 
> >                 tmp |= r->frameno & ES_FRAMENO_MASK;
> >                 tmp |= (r->spi & ES_SPI_MASK) << ES_SPI_OFF;
> 
> nit: we shouldn't mask anything here, it just makes an impression that
> r->frameno can be bigger than we have bits for it in a bitmask

Ok, I'll add bitmasks to field definitions and remove masks here.

[...]