Xen devel - Aug 2005 - [PATCH 1/2] Writable pagetable support for x86_64 xenlinux (xen part)

Attached is a patch to xen. I would like thank Ian/Keir for the design
suggestions.

 arch/x86/audit.c              |   62 ++++++----
 arch/x86/domain.c             |    6 -
 arch/x86/mm.c                 |  238
++++++++++++++++++++++++++++++------------
 arch/x86/shadow32.c           |    2 
 arch/x86/shadow_public.c      |    6 -
 arch/x86/traps.c              |   11 +
 common/grant_table.c          |    2 
 include/asm-x86/mm.h          |   28 +++-
 include/asm-x86/page.h        |   27 ++--
 include/asm-x86/shadow.h      |   13 +-
 include/asm-x86/x86_64/page.h |    3 
 11 files changed, 271 insertions(+), 127 deletions(-)

Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>

Jun
---
Intel Open Source Technology Center 


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

Now _SMP_ x86_64 kernel works fine with writable pagetable!

This _additional_ patch ptwr-smp.patch (just a single line change,
effectively + minor bug fix) is required to get SMP kernel working. The
first patch (ptwr-xen.patch) is also attached for convenience. Once
these two are (ptwr-xen.patch & ptwr-smp.patch) applied to Xen, the
patch  for x86_64 xenlinux (already sent) is required to activate
writable pagetable.

One ugly code that I don''t like is __not_mapped(l2_pgentry_t *pl2e).
Any
idea for doing it better? 

I used a patch from Chris/Xin for the x86_64 SMP kernel. Probably he has
more updates there.

Nakajima, Jun wrote:
> Attached is a patch to xen. I would like thank Ian/Keir for the
> design suggestions. 
> 
>  arch/x86/audit.c              |   62 ++++++----
>  arch/x86/domain.c             |    6 -
>  arch/x86/mm.c                 |  238
>  ++++++++++++++++++++++++++++++------------ arch/x86/shadow32.c      
>  |    2 arch/x86/shadow_public.c      |    6 -
>  arch/x86/traps.c              |   11 +
>  common/grant_table.c          |    2
>  include/asm-x86/mm.h          |   28 +++-
>  include/asm-x86/page.h        |   27 ++--
>  include/asm-x86/shadow.h      |   13 +-
>  include/asm-x86/x86_64/page.h |    3
>  11 files changed, 271 insertions(+), 127 deletions(-)
> 
> Signed-off-by: Jun Nakajima <jun.nakajima@intel.com>
> 
> Jun
> ---
> Intel Open Source Technology Center

Jun
---
Intel Open Source Technology Center



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

On 9 Aug 2005, at 08:59, Nakajima, Jun wrote:
> One ugly code that I don''t like is __not_mapped(l2_pgentry_t
*pl2e).
> Any
> idea for doing it better?
>
> I used a patch from Chris/Xin for the x86_64 SMP kernel. Probably he 
> has
> more updates there.
i checked in a cleaned up version of your SMP patch, as part of some 
wider-ranging cleanups of your previous wrpt patch. A couple of notes 
on that:

  1. I got rid of __not_mapped(). It should suffice to attempt the 
access through the linear mapping and handle the possible fault.
  2. You had gross CONFIG_X86_64-specific tests in get_page_type(). I 
cannot see how any good could possible come of them so I removed them. 
Either you need a good justification why they''re needed or, if they 
actually fixed observed problems, I think there must be some underlying 
bug that those conditional sections of code were masking. Either way, 
this needs looking at.

  -- Keir


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