Re: [PATCH v2] xen: fix cpuid reporting on PVH Dom0

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 01/08/14 16:26, Roger Pau Monne wrote:
> dab11417d also caused some problems regarding HVM guest creation on
> PVH Dom0, mainly the CR4 mask returned by hvm_cr4_guest_reserved_bits
> changed from 0xfffffffffffff800 to 0xfffffffffffff893, which means HVM
> guests created from a PVH Dom0 are unable to set VME, PVI, PSE or PGE
> CR4 flags.
>
> This is because cpuid on PVH guests mask PSE, PGE, PSE36 and VME
> flags, so the white listing done in xc_cpuid_hvm_policy doesn't enable
> those features, and the guest ends up with a very restrictive cpuid
> policy.
>
> Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
> Cc: Jan Beulich <JBeulich@suse.com>
> Cc: Andrew Cooper <andrew.cooper3@citrix.com>
> ---
> Changes since RFC:
>  - Expose the cpuid flags to all PVH guests and not only
>    hardware/control domains.
> ---
>  xen/arch/x86/traps.c |   12 ++++++++----
>  1 files changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/xen/arch/x86/traps.c b/xen/arch/x86/traps.c
> index 677074b..465b196 100644
> --- a/xen/arch/x86/traps.c
> +++ b/xen/arch/x86/traps.c
> @@ -803,12 +803,16 @@ void pv_cpuid(struct cpu_user_regs *regs)
>      if ( (regs->eax & 0x7fffffff) == 0x00000001 )
>      {
>          /* Modify Feature Information. */
> -        __clear_bit(X86_FEATURE_VME, &d);
>          if ( !cpu_has_apic )
>              __clear_bit(X86_FEATURE_APIC, &d);
> -        __clear_bit(X86_FEATURE_PSE, &d);
> -        __clear_bit(X86_FEATURE_PGE, &d);
> -        __clear_bit(X86_FEATURE_PSE36, &d);

Newline here...

> +        if ( !is_pvh_vcpu(curr) )
> +        {
> +            __clear_bit(X86_FEATURE_PSE, &d);
> +            __clear_bit(X86_FEATURE_PGE, &d);
> +            __clear_bit(X86_FEATURE_PSE36, &d);
> +            __clear_bit(X86_FEATURE_VME, &d);

Why do PVH guests care about VME?  They can't use it as far as I am aware.

> +        }
> +

... and not here.

>      }
>  
>      switch ( (uint32_t)regs->eax )


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