All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] x86/xsaves: ebx may return wrong value using CPUID eax=0xdh, ecx =1
@ 2015-12-21  1:50 Shuai Ruan
  2015-12-21 12:06 ` Jan Beulich
  0 siblings, 1 reply; 2+ messages in thread
From: Shuai Ruan @ 2015-12-21  1:50 UTC (permalink / raw)
  To: xen-devel
  Cc: wei.liu2, Ian.Campbell, stefano.stabellini, andrew.cooper3,
	ian.jackson, jbeulich, keir

Refer to SDM 13.4.3. The value return by ecx[1] with cpuid
function 0xdh and sub-fucntion i (i>1) indicates the alignment
of the state component i when the compacted format of the extended
region of an xsave area is used.

So when hvm guest using CPUID eax=0xdh,ecx=1 to get the size of area
used for compacted format, we need to take alignment into consideration.

Signed-off-by: Shuai Ruan <shuai.ruan@linux.intel.com>
---
 tools/libxc/xc_cpuid_x86.c | 10 ++++++++--
 xen/arch/x86/hvm/hvm.c     | 12 ++++++++++++
 2 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/tools/libxc/xc_cpuid_x86.c b/tools/libxc/xc_cpuid_x86.c
index 8882c01..480bd83 100644
--- a/tools/libxc/xc_cpuid_x86.c
+++ b/tools/libxc/xc_cpuid_x86.c
@@ -285,6 +285,7 @@ static void intel_xc_cpuid_policy(xc_interface *xch,
 #define XSAVEC          (1 << 1)
 #define XGETBV1         (1 << 2)
 #define XSAVES          (1 << 3)
+#define XSTATE_ALIGN    (1 << 1)
 /* Configure extended state enumeration leaves (0x0000000D for xsave) */
 static void xc_cpuid_config_xsave(xc_interface *xch,
                                   const struct cpuid_domain_info *info,
@@ -333,8 +334,13 @@ static void xc_cpuid_config_xsave(xc_interface *xch,
             regs[0] = regs[1] = regs[2] = regs[3] = 0;
             break;
         }
-        /* Don't touch EAX, EBX. Also cleanup ECX and EDX */
-        regs[2] = regs[3] = 0;
+        /* Don't touch EAX, EBX.
+         * ECX[1] indicates whether the state component
+         * needs alignment when compacted format is used.
+         */
+        regs[2] &= XSTATE_ALIGN;
+        /* Clean up EDX */
+        regs[3] = 0;
         break;
     }
 }
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 08cef1f..c4ac677 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -4619,7 +4619,19 @@ void hvm_cpuid(unsigned int input, unsigned int *eax, unsigned int *ebx,
                     for ( sub_leaf = 2; sub_leaf < 63; sub_leaf++ )
                         if ( (v->arch.xcr0 | v->arch.hvm_vcpu.msr_xss) &
                              (1ULL << sub_leaf) )
+                        {
+                            domain_cpuid(d, input, sub_leaf, &_eax, &_ebx,
+                                         &_ecx, &_edx);
+                            /*
+                             * The value return by _ecx[1] indicates the
+                             * alignment of the state component i when the
+                             * compacted format of the extended region of
+                             *  an xsave area is used.
+                             */
+                            if (_ecx & 0x2)
+                                *ebx = ROUNDUP(*ebx, 64);
                             *ebx += xstate_sizes[sub_leaf];
+			}
             }
             else
                 *ebx = *ecx = *edx = 0;
-- 
1.9.1

^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] x86/xsaves: ebx may return wrong value using CPUID eax=0xdh, ecx =1
  2015-12-21  1:50 [PATCH] x86/xsaves: ebx may return wrong value using CPUID eax=0xdh, ecx =1 Shuai Ruan
@ 2015-12-21 12:06 ` Jan Beulich
  0 siblings, 0 replies; 2+ messages in thread
From: Jan Beulich @ 2015-12-21 12:06 UTC (permalink / raw)
  To: Shuai Ruan
  Cc: wei.liu2, Ian.Campbell, stefano.stabellini, andrew.cooper3,
	ian.jackson, xen-devel, keir

>>> On 21.12.15 at 02:50, <shuai.ruan@linux.intel.com> wrote:
> Refer to SDM 13.4.3. The value return by ecx[1] with cpuid

I think we've clarified before that SDM references should be by
section title, not section number.

> --- a/tools/libxc/xc_cpuid_x86.c
> +++ b/tools/libxc/xc_cpuid_x86.c
> @@ -285,6 +285,7 @@ static void intel_xc_cpuid_policy(xc_interface *xch,
>  #define XSAVEC          (1 << 1)
>  #define XGETBV1         (1 << 2)
>  #define XSAVES          (1 << 3)
> +#define XSTATE_ALIGN    (1 << 1)

XSTATE_ALIGN64? (There might well be another, higher aligment
flag in the future.)

> --- a/xen/arch/x86/hvm/hvm.c
> +++ b/xen/arch/x86/hvm/hvm.c
> @@ -4619,7 +4619,19 @@ void hvm_cpuid(unsigned int input, unsigned int *eax, unsigned int *ebx,
>                      for ( sub_leaf = 2; sub_leaf < 63; sub_leaf++ )
>                          if ( (v->arch.xcr0 | v->arch.hvm_vcpu.msr_xss) &
>                               (1ULL << sub_leaf) )
> +                        {
> +                            domain_cpuid(d, input, sub_leaf, &_eax, &_ebx,
> +                                         &_ecx, &_edx);
> +                            /*
> +                             * The value return by _ecx[1] indicates the
> +                             * alignment of the state component i when the
> +                             * compacted format of the extended region of
> +                             *  an xsave area is used.
> +                             */
> +                            if (_ecx & 0x2)

Needs a #define just like in libxc.

Jan

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2015-12-21 12:06 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-12-21  1:50 [PATCH] x86/xsaves: ebx may return wrong value using CPUID eax=0xdh, ecx =1 Shuai Ruan
2015-12-21 12:06 ` Jan Beulich

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.