[PATCH] x86: Correctly report max number of hypervisor leaves

[PATCH] x86: Correctly report max number of hypervisor leaves

[Boris Ostrovsky]

Commit def0bbd31 provided support for changing max number of
hypervisor cpuid leaves (in leaf 0x4000xx00). It also made the
hypervisor incorrectly report this number for guests that
use default value (i.e. don't specify leaf 0x4000xx00 in config
file)

Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reported-by: Roger Pau Monne <roger.pau@citrix.com>
---
 xen/arch/x86/traps.c |   16 +++++++++++-----
 1 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/xen/arch/x86/traps.c b/xen/arch/x86/traps.c
index 1722912..b82b1b3 100644
--- a/xen/arch/x86/traps.c
+++ b/xen/arch/x86/traps.c
@@ -692,13 +692,19 @@ int cpuid_hypervisor_leaves( uint32_t idx, uint32_t sub_idx,
     if ( idx > XEN_CPUID_MAX_NUM_LEAVES )
         return 0; /* Avoid unnecessary pass through domain_cpuid() */
 
-    /* Number of leaves may be user-specified */
     domain_cpuid(d, base, 0, &limit, &dummy, &dummy, &dummy);
-    limit &= 0xff;
-    if ( limit < 2 )
-        limit = 2;
-    else if ( limit > XEN_CPUID_MAX_NUM_LEAVES )
+    if ( limit == 0 )
+        /* Default number of leaves */
         limit = XEN_CPUID_MAX_NUM_LEAVES;
+    else
+    {
+        /* User-specified number of leaves */
+        limit &= 0xff;
+        if ( limit < 2 )
+            limit = 2;
+        else if ( limit > XEN_CPUID_MAX_NUM_LEAVES )
+            limit = XEN_CPUID_MAX_NUM_LEAVES;
+    }
 
     if ( idx > limit ) 
         return 0;
-- 
1.7.1

Re: [PATCH] x86: Correctly report max number of hypervisor leaves

[Roger Pau Monné]

On 30/05/14 22:11, Boris Ostrovsky wrote:
> Commit def0bbd31 provided support for changing max number of
> hypervisor cpuid leaves (in leaf 0x4000xx00). It also made the
> hypervisor incorrectly report this number for guests that
> use default value (i.e. don't specify leaf 0x4000xx00 in config
> file)
> 
> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
> Reported-by: Roger Pau Monne <roger.pau@citrix.com>

Tested-by: Roger Pau Monné <roger.pau@citrix.com>

However I think we should prevent PVH guests from seeing the newly
introduced XEN_HVM_CPUID_APIC_ACCESS_VIRT and
XEN_HVM_CPUID_X2APIC_VIRT flags. I have the following patch, but it
still adds one more is_pvh_vcpu into VMX code. I've tried checking for
vlapic_enabled, but it also returns true for PVH guests.

diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
index d45fb7f..ecc010e 100644
--- a/xen/arch/x86/hvm/vmx/vmx.c
+++ b/xen/arch/x86/hvm/vmx/vmx.c
@@ -1660,7 +1660,7 @@ void vmx_hypervisor_cpuid_leaf(uint32_t sub_idx,
                                uint32_t *eax, uint32_t *ebx,
                                uint32_t *ecx, uint32_t *edx)
 {
-    if ( sub_idx != 0 )
+    if ( sub_idx != 0 || is_pvh_vcpu(current) )
         return;
     if ( cpu_has_vmx_apic_reg_virt )
         *eax |= XEN_HVM_CPUID_APIC_ACCESS_VIRT;

Re: [PATCH] x86: Correctly report max number of hypervisor leaves

[Jan Beulich]

>>> On 02.06.14 at 11:29, <roger.pau@citrix.com> wrote:
> On 30/05/14 22:11, Boris Ostrovsky wrote:
>> Commit def0bbd31 provided support for changing max number of
>> hypervisor cpuid leaves (in leaf 0x4000xx00). It also made the
>> hypervisor incorrectly report this number for guests that
>> use default value (i.e. don't specify leaf 0x4000xx00 in config
>> file)
>> 
>> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
>> Reported-by: Roger Pau Monne <roger.pau@citrix.com>
> 
> Tested-by: Roger Pau Monné <roger.pau@citrix.com>
> 
> However I think we should prevent PVH guests from seeing the newly
> introduced XEN_HVM_CPUID_APIC_ACCESS_VIRT and
> XEN_HVM_CPUID_X2APIC_VIRT flags. I have the following patch, but it
> still adds one more is_pvh_vcpu into VMX code. I've tried checking for
> vlapic_enabled, but it also returns true for PVH guests.

And vlapic_enabled() wouldn't be the right qualifier anyway, as
feature availability doesn't depend on what other features are
enabled. I think is_pvh is fine for now here, until the folding of
PVH and HVM got far enough.

Jan

> --- a/xen/arch/x86/hvm/vmx/vmx.c
> +++ b/xen/arch/x86/hvm/vmx/vmx.c
> @@ -1660,7 +1660,7 @@ void vmx_hypervisor_cpuid_leaf(uint32_t sub_idx,
>                                 uint32_t *eax, uint32_t *ebx,
>                                 uint32_t *ecx, uint32_t *edx)
>  {
> -    if ( sub_idx != 0 )
> +    if ( sub_idx != 0 || is_pvh_vcpu(current) )
>          return;
>      if ( cpu_has_vmx_apic_reg_virt )
>          *eax |= XEN_HVM_CPUID_APIC_ACCESS_VIRT;



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