Re: [Xen-devel] [[PATCH for-4.13]] xen/arm: mm: Allow generic xen page-tables helpers to be called early

["Xia, Hongyan" <hongyax@xxxxxxxxxx>]

Hi Julien,

Sure. You should be able to find it on directnonmap-v2.3 branch at
https://xenbits.xen.org/git-http/people/hx242/xen.git.

Commit: a4fef31b99388524d3f7748967c5d04a924cb7e3
    x86: add Persistent Map (PMAP) infrastructure

One thing to note is that the PMAP structure is really low-performance
because we do not want to burn too many fixmap entries. It should no
longer be in use the moment we bootstrapped other mapping
infrastructures. The only intention of it is to be able to map pages
very early on.

Hongyan

On Wed, 2019-10-16 at 13:46 +0100, Julien Grall wrote:
> Hi Hongyan,
> 
> On 11/10/2019 10:53, Xia, Hongyan wrote:
> > Not commenting on the patch, but I had exactly the same problem
> > when
> > removing the direct map in x86. map_domain_page has to be usable
> > without the direct map and even before alloc_boot_pages can be used
> > (so
> > that I can map the bootmem_regions_list, although I have made it
> > statically allocated now).
> > 
> > The direct map teardown series in the end introduces a persistent
> > mapping infrastructure to map page tables which occupies a few
> > fixmap
> > entries, and avoid set_fixmap but instead modify the entries
> > directly
> > to avoid invocation loops. The result is that map_xen_pagetable and
> > map_domain_page is usable from the very beginning of start_xen.
> 
> Would you mind to point to the patch adding the implementation on
> x86?
> 
> On arm32, we are not using a direct map. So we are using a per-cpu
> mapping (see 
> map_domain_page() implementation in arch/arm/mm.c).
> 
> On arm64, we are using the direct map so far. But we could use any of
> the two 
> solutions if needed.
> 
> Cheers,
> 
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel