[PATCH] [HVM] fix 64-bit hvm guest migration

[PATCH] [HVM] fix 64-bit hvm guest migration

[David Lively]

[-- Attachment #1: Type: text/plain, Size: 549 bytes --]

Hi -
  We (well, Robert) just (earlier today) noticed we couldn't migrate 
64-bit guests with the 3.1.1 and unstable hypervisors.  The culprit was 
a too-strict test on guest EFER bits, trivially fixed by the following 
patch, which applies to both unstable and 3.1.1.  Sorry we didn't catch 
this before today ...

Dave

The sanity check for ctxt.msr_efer used to fail when EFER_LMA was set.
This prevent migration of 64-bit guests.

Signed-off-by: David Lively <dlively@virtualiron.com>
Signed-off-by: Robert Phillips <rphillips@virtualiron.com>


[-- Attachment #2: xen-hvm-load-efer-test.patch --]
[-- Type: text/x-patch, Size: 615 bytes --]

diff -r e56402b6e155 xen/arch/x86/hvm/hvm.c
--- a/xen/arch/x86/hvm/hvm.c	Wed Oct 10 11:37:02 2007 -0400
+++ b/xen/arch/x86/hvm/hvm.c	Wed Oct 10 11:37:02 2007 -0400
@@ -389,8 +389,7 @@ static int hvm_load_cpu_ctxt(struct doma
                  ctxt.msr_efer);
         return -EINVAL;
     }
-
-    if ( (ctxt.msr_efer & ~(EFER_LME | EFER_NX | EFER_SCE)) ||
+    if ( (ctxt.msr_efer & ~(EFER_LME | EFER_LMA | EFER_NX | EFER_SCE)) ||
          ((sizeof(long) != 8) && (ctxt.msr_efer & EFER_LME)) ||
          (!cpu_has_nx && (ctxt.msr_efer & EFER_NX)) ||
          (!cpu_has_syscall && (ctxt.msr_efer & EFER_SCE)) ||

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

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

Re: [PATCH] [HVM] fix 64-bit hvm guest migration

[Keir Fraser]

Yep, that's a very obvious bug. I'll slip the 3.1.1 release tag for it and
get it pushed to the main public tree.

 Thanks,
 Keir

On 10/10/07 22:14, "David Lively" <dlively@virtualiron.com> wrote:

> Hi -
>   We (well, Robert) just (earlier today) noticed we couldn't migrate
> 64-bit guests with the 3.1.1 and unstable hypervisors.  The culprit was
> a too-strict test on guest EFER bits, trivially fixed by the following
> patch, which applies to both unstable and 3.1.1.  Sorry we didn't catch
> this before today ...
> 
> Dave
> 
> The sanity check for ctxt.msr_efer used to fail when EFER_LMA was set.
> This prevent migration of 64-bit guests.
> 
> Signed-off-by: David Lively <dlively@virtualiron.com>
> Signed-off-by: Robert Phillips <rphillips@virtualiron.com>
> 
> diff -r e56402b6e155 xen/arch/x86/hvm/hvm.c
> --- a/xen/arch/x86/hvm/hvm.c Wed Oct 10 11:37:02 2007 -0400
> +++ b/xen/arch/x86/hvm/hvm.c Wed Oct 10 11:37:02 2007 -0400
> @@ -389,8 +389,7 @@ static int hvm_load_cpu_ctxt(struct doma
>                   ctxt.msr_efer);
>          return -EINVAL;
>      }
> -
> -    if ( (ctxt.msr_efer & ~(EFER_LME | EFER_NX | EFER_SCE)) ||
> +    if ( (ctxt.msr_efer & ~(EFER_LME | EFER_LMA | EFER_NX | EFER_SCE)) ||
>           ((sizeof(long) != 8) && (ctxt.msr_efer & EFER_LME)) ||
>           (!cpu_has_nx && (ctxt.msr_efer & EFER_NX)) ||
>           (!cpu_has_syscall && (ctxt.msr_efer & EFER_SCE)) ||
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel