On Thu, May 4, 2023 at 6:54 PM Alexei Starovoitov
<alexei.starovoitov@xxxxxxxxx> wrote:
>
> On Thu, May 04, 2023 at 05:09:01PM -0700, Andrii Nakryiko wrote:
> > +struct backtrack_state {
> > + struct bpf_verifier_env *env;
> > + u32 frame;
> > + u32 bitcnt;
> > + u32 reg_masks[MAX_CALL_FRAMES];
> > + u64 stack_masks[MAX_CALL_FRAMES];
> > +};
> > +
>
> > +static inline u32 bt_empty(struct backtrack_state *bt)
> > +{
> > + u64 mask = 0;
> > + int i;
> > +
> > + for (i = 0; i < MAX_CALL_FRAMES; i++)
> > + mask |= bt->reg_masks[i] | bt->stack_masks[i];
> > +
> > + return mask == 0;
> > +}
> > +
> ...
> > +static inline void bt_set_frame_reg(struct backtrack_state *bt, u32 frame, u32 reg)
> > +{
> > + if (bt->reg_masks[frame] & (1 << reg))
> > + return;
> > +
> > + bt->reg_masks[frame] |= 1 << reg;
> > + bt->bitcnt++;
> > +}
>
> So you went with bitcnt and bt_empty ?
> I'm confused. I thought we discussed it's one or another.
> I have slight preference towards bt_empty() as above. fwiw.
sigh, forgot to drop bitcnt and ifs... will remove ifs and bitcnt in v3
