Re: Fixing excessive shootdowns during FFS read (kern/53124) on x86

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Sun, Mar 25, 2018 at 11:48:02AM -0000, Michael van Elst wrote:
> [...]
> >3. change UVM code to not do this mapping via kernel map, so pmap_update()
> >and the IPIs are not triggered in first place
> 
> What I currently don't understand is that we only see one TLB shootdown
> per window and that happens when it is _unmapped_. Mapping the pages
> (with the exception of overwriting) is done by the regular fault handler
> and apparently this does not flush the TLB. This makes me think that it might
> not be required at all and is just an unwanted side-effect of pmap_update.

I think it's because if we get a fault, then this VA is not mapped
in the page table, and so it can't be cached in any TLB. So there is
nothing to flush in this case.

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--