[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 14/19] xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables



On Wed, 5 Jun 2019, Julien Grall wrote:
> On 05/06/2019 00:11, Stefano Stabellini wrote:
> > On Tue, 14 May 2019, Julien Grall wrote:
> > > The page-table walker is configured to use the same shareability and
> > > cacheability as the access performed when updating the page-tables. This
> > > means cleaning the cache for secondary CPUs runtime page-tables is
> > > unnecessary.
> > 
> > All right. Is there an explicit configuration for the shareability and
> > cacheability used by the page-table walker or is it specified as such in
> > the Arm Arm?
> 
> See the configuration of TCR_EL2, I can mention it.
 
That would be nice. I double-checked and it is as you wrote.


> > Also, isn't it possible that CPUs on a different cluster
> > (big.LITTLE) would have issues with this if the cache could be split
> > between the two clusters?
> 
> I don't understand this... Cache should be coherent when a CPU leaves EL3.
> But we already share some bits of the page tables between the processor (see
> create_xen_page_tables). So I don't see where there is a possible problem
> here.

If the cache is always coherent across the clusters and the
pagetable-walkers of different clusters, then this is fine.

 
> > > Signed-off-by: Julien Grall <julien.grall@xxxxxxx>
> > > Reviewed-by: Andrii Anisov <andrii_anisov@xxxxxxxx>
> > > 
> > > ---
> > >      Changes in v2:
> > >          - Add Andrii's reviewed-by
> > > ---
> > >   xen/arch/arm/mm.c | 3 ---
> > >   1 file changed, 3 deletions(-)
> > > 
> > > diff --git a/xen/arch/arm/mm.c b/xen/arch/arm/mm.c
> > > index cda2847d00..6db7dda0da 100644
> > > --- a/xen/arch/arm/mm.c
> > > +++ b/xen/arch/arm/mm.c
> > > @@ -769,9 +769,6 @@ int init_secondary_pagetables(int cpu)
> > >           
> > > write_pte(&first[first_table_offset(DOMHEAP_VIRT_START+i*FIRST_SIZE)],
> > > pte);
> > >       }
> > >   -    clean_dcache_va_range(first, PAGE_SIZE);
> > > -    clean_dcache_va_range(domheap, DOMHEAP_SECOND_PAGES*PAGE_SIZE);
> > > -
> > >       per_cpu(xen_pgtable, cpu) = first;
> > >       per_cpu(xen_dommap, cpu) = domheap;
> > >   
> > > -- 
> > > 2.11.0
> > > 
> 
> -- 
> Julien Grall
> 

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.