Re: [Xen-devel] Page table sync. in Xen arm (Arndale) with SMP enabled (Bug fix included)

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Fri, 2013-04-19 at 14:08 +0100, Stefano Stabellini wrote:
> On Thu, 18 Apr 2013, Ian Campbell wrote:

> > This also explains why Stefano was able to implement
> > map_doman_page_global in terms of map_domain_page:
> > http://marc.info/?l=xen-devel&m=136551673224671
> 
> Regarding that patch, should I re-implement map_doman_page_global using
> the current map_domain_page code plus Sengul's patch?

I don't think we want a lock in map_domain_page for performance reasons.
I'm looking at making the domheap per-pcpu right now and although it
looks a bit fiddly I don't think it should take too long.

> Should I wait now and rebase later?

Ultimately we want to implement map_domain_page_global in terms of vmap.
That ought to be pretty easy. What I think needs to happen is that:
      * early_ioremap needs to flip around and start allocating from the
        top of the region (for vmap internal reasons, see below)
      * We need to initialise vmap and implement arch_vmap_virt_end() to
        return the address right below the used early_ioremap space
        (this is why it needs to allocate from the end)
      * Implement domain_map/unmap_page_global in terms of vmap/vunmap,
        which looks to be more of less trivial API impedance matching

I don't think that is blocked by the per-pcpu domheap stuff, is it?



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel