Excessive TLB flush ranges

[Uladzislau Rezki]

On Fri, May 19, 2023 at 04:56:53PM +0200, Thomas Gleixner wrote:
> On Fri, May 19 2023 at 12:01, Uladzislau Rezki wrote:
> > On Wed, May 17, 2023 at 06:32:25PM +0200, Thomas Gleixner wrote:
> >> That made me look into this coalescing code. I understand why you want
> >> to batch and coalesce and rather do a rare full tlb flush than sending
> >> gazillions of IPIs.
> >> 
> > Your issues has no connections with merging. But the place you looked
> > was correct :)
> 
> I'm not talking about merging. I'm talking about coalescing ranges.
> 
>        start = 0x95c8d000 end = 0x95c8e000
> 
> plus the VA from list which has
> 
>        start = 0xf08a1000 end = 0xf08a5000
> 
> which results in a flush range of:
> 
>        start = 0x95c8d000 end = 0xf08a5000
> No?
> 
Correct. 0x95c8d000 is a min, 0xf08a5000 is a max.

> > @@ -1739,15 +1739,14 @@ static bool __purge_vmap_area_lazy(unsigned long start, unsigned long end)
> >         if (unlikely(list_empty(&local_purge_list)))
> >                 goto out;
> >
> > -       start = min(start,
> > -               list_first_entry(&local_purge_list,
> > -                       struct vmap_area, list)->va_start);
> > +       /* OK. A per-cpu wants to flush an exact range. */
> > +       if (start != ULONG_MAX)
> > +               flush_tlb_kernel_range(start, end);
> >
> > -       end = max(end,
> > -               list_last_entry(&local_purge_list,
> > -                       struct vmap_area, list)->va_end);
> > +       /* Flush per-VA. */
> > +       list_for_each_entry(va, &local_purge_list, list)
> > +               flush_tlb_kernel_range(va->va_start, va->va_end);
> >
> > -       flush_tlb_kernel_range(start, end);
> >         resched_threshold = lazy_max_pages() << 1;
> 
> That's completely wrong, really.
> 
Absolutely. That is why we do not flush a range per-VA ;-) I provided the
data just to show what happens if we do it! A per-VA flushing works when
a system is not capable of doing a full flush, so it has to do it page
by page. In this scenario we should bypass ranges(not mapped) which are
between VAs in a purge-list.

--
Uladzislau Rezki