[Xen-ia64-devel] Re: [patch 2/8] Kexec: Allow page fault handler to handle EFI regions

[Simon Horman <horms@xxxxxxxxxxxx>]

On Tue, Oct 23, 2007 at 02:13:18PM +0200, tgingold@xxxxxxx wrote:
> Quoting Simon Horman <horms@xxxxxxxxxxxx>:
> 
> > If the EFI region is mapped into a region within the address space usually
> > used by guests, then the page_fault handler needs to know about it. A
> > description of why this mapping is made is included in the patch that makes
> > the mapping.
> >
> > The patch to actually make use of this code is separate.
> >
> > This does not take into account other ways that relevant functions are
> > called.
> >
> > In particular:
> > - vcpu_itc_d() and vcpu_itc_i() call translate_domain_pte()
> > - ia64_shadow_fault() and vcpu_tpa() call vcpu_translate()
> >
> > This patch needs work, and I am positing it for discussion.
> 
> Honestly I don't really like this part of the patch: it adds complexity to an
> already complex area.
> 
> EFI_[UN]CACHED_OFFSET areas are id-mapped.  So why can't the tlb miss be
> handled directly in the ivt (like we do for id-mapped Xen area) ?
> 
> Maybe I don't really understand the issues so do not hesitate to explain!

I took a stab at implemeting this purely in assembly in ivt.S,
but for some reason that I couldn't put a finger on, it didn't work.
It was also rather messy.

I'll try again now that we have the EFI RID thing working,
perhaps ivt.S will work and not be a complete mess this time around.

-- 
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/


_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel