[PATCH 0/3] x86/ept: force WB to foreign and grant mappings

Roger Pau Monne posted 3 patches 2 years, 11 months ago
Test gitlab-ci failed
Patches applied successfully (tree, apply log)
git fetch https://gitlab.com/xen-project/patchew/xen tags/patchew/20210528173935.29919-1-roger.pau@citrix.com
xen/arch/x86/hvm/mtrr.c           | 107 +---------------------
xen/arch/x86/hvm/vmx/vmx.c        |   6 +-
xen/arch/x86/mm/p2m-ept.c         | 145 ++++++++++++++++++++++++++++--
xen/include/asm-x86/hvm/vmx/vmx.h |   2 +
xen/include/asm-x86/mtrr.h        |   7 +-
5 files changed, 147 insertions(+), 120 deletions(-)
[PATCH 0/3] x86/ept: force WB to foreign and grant mappings
Posted by Roger Pau Monne 2 years, 11 months ago
Hello,

The aim of this series is to force the cache attribute of foreign and
grant mappings to WB for HVM/PVH guests. This is required because those
mappings will be likely be using unpopulated memory ranges in the p2m,
and those are usually UC in the MTRR state.

Having the guest set the correct MTRR attributes is also unlikely,
because the number of MTRR ranges is finite.

Roger Pau Monne (3):
  x86/mtrr: remove stale function prototype
  x86/mtrr: move epte_get_entry_emt to p2m-ept.c
  x86/ept: force WB cache attributes for grant and foreign maps

 xen/arch/x86/hvm/mtrr.c           | 107 +---------------------
 xen/arch/x86/hvm/vmx/vmx.c        |   6 +-
 xen/arch/x86/mm/p2m-ept.c         | 145 ++++++++++++++++++++++++++++--
 xen/include/asm-x86/hvm/vmx/vmx.h |   2 +
 xen/include/asm-x86/mtrr.h        |   7 +-
 5 files changed, 147 insertions(+), 120 deletions(-)

-- 
2.31.1


Re: [PATCH 0/3] x86/ept: force WB to foreign and grant mappings
Posted by Roger Pau Monné 2 years, 10 months ago
Ping?

This is missing an Ack or otherwise from the Intel maintainers.

Thanks, Roger.

On Fri, May 28, 2021 at 07:39:32PM +0200, Roger Pau Monne wrote:
> 
> Hello,
> 
> The aim of this series is to force the cache attribute of foreign and
> grant mappings to WB for HVM/PVH guests. This is required because those
> mappings will be likely be using unpopulated memory ranges in the p2m,
> and those are usually UC in the MTRR state.
> 
> Having the guest set the correct MTRR attributes is also unlikely,
> because the number of MTRR ranges is finite.
> 
> Roger Pau Monne (3):
>   x86/mtrr: remove stale function prototype
>   x86/mtrr: move epte_get_entry_emt to p2m-ept.c
>   x86/ept: force WB cache attributes for grant and foreign maps
> 
>  xen/arch/x86/hvm/mtrr.c           | 107 +---------------------
>  xen/arch/x86/hvm/vmx/vmx.c        |   6 +-
>  xen/arch/x86/mm/p2m-ept.c         | 145 ++++++++++++++++++++++++++++--
>  xen/include/asm-x86/hvm/vmx/vmx.h |   2 +
>  xen/include/asm-x86/mtrr.h        |   7 +-
>  5 files changed, 147 insertions(+), 120 deletions(-)
> 
> -- 
> 2.31.1
>