[PATCH v7] xen/domain: rewrite emulation_flags_ok()

dmkhn@proton.me posted 1 patch 4 months, 3 weeks ago
Patches applied successfully (tree, apply log)
git fetch https://gitlab.com/xen-project/patchew/xen tags/patchew/20250610144500.3176661-1-dmukhin@ford.com
There is a newer version of this series
xen/arch/x86/domain.c | 78 +++++++++++++++++++++++++++++++++----------
1 file changed, 60 insertions(+), 18 deletions(-)
[PATCH v7] xen/domain: rewrite emulation_flags_ok()
Posted by dmkhn@proton.me 4 months, 3 weeks ago
From: Denis Mukhin <dmukhin@ford.com>

Rewrite emulation_flags_ok() to simplify future modifications.

No functional change intended.

Signed-off-by: Denis Mukhin <dmukhin@ford.com>
---
Changes since v6:
- simplified checks for PV further

Link to v6: https://lore.kernel.org/xen-devel/20250610004216.3012253-1-dmukhin@ford.com/
Link to CI: https://gitlab.com/xen-project/people/dmukhin/xen/-/pipelines/1862559089
---
 xen/arch/x86/domain.c | 78 +++++++++++++++++++++++++++++++++----------
 1 file changed, 60 insertions(+), 18 deletions(-)

diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index 7536b6c871..fdbd064ebf 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -743,32 +743,74 @@ int arch_sanitise_domain_config(struct xen_domctl_createdomain *config)
     return 0;
 }
 
+/*
+ * Verify that the domain's emulation flags resolve to a supported configuration.
+ *
+ * This ensures we only allow a known, safe subset of emulation combinations
+ * (for both functionality and security). Arbitrary mixes are likely to cause
+ * errors (e.g. null pointer dereferences).
+ *
+ * NB: use the internal X86_EMU_XXX symbols, not the public XEN_X86_EMU_XXX
+ * symbols.
+ */
 static bool emulation_flags_ok(const struct domain *d, uint32_t emflags)
 {
+    enum {
+        CAP_PV          = BIT(0, U),
+        CAP_HVM         = BIT(1, U),
+        CAP_HWDOM       = BIT(2, U),
+        CAP_DOMU        = BIT(3, U),
+    };
+    static const struct {
+        unsigned int caps;
+        uint32_t min;
+        uint32_t opt;
+    } configs[] = {
+#ifdef CONFIG_PV
+        /* PV dom0 and domU */
+        {
+            .caps   = CAP_PV | CAP_HWDOM | CAP_DOMU,
+            .opt    = X86_EMU_PIT,
+        },
+#endif /* #ifdef CONFIG_PV */
+
+#ifdef CONFIG_HVM
+        /* PVH dom0 */
+        {
+            .caps   = CAP_HVM | CAP_HWDOM,
+            .min    = X86_EMU_LAPIC | X86_EMU_IOAPIC | X86_EMU_VPCI,
+        },
+
+        /* PVH domU */
+        {
+            .caps   = CAP_HVM | CAP_DOMU,
+            .min    = X86_EMU_LAPIC,
+        },
+
+        /* HVM domU */
+        {
+            .caps   = CAP_HVM | CAP_DOMU,
+            .min    = X86_EMU_ALL & ~(X86_EMU_VPCI | X86_EMU_USE_PIRQ),
+            /* HVM PIRQ feature is user-selectable. */
+            .opt    = X86_EMU_USE_PIRQ,
+        },
+#endif /* #ifdef CONFIG_HVM */
+    };
+    unsigned int i;
+    unsigned int caps = (is_pv_domain(d) ? CAP_PV : CAP_HVM) |
+                        (is_hardware_domain(d) ? CAP_HWDOM : CAP_DOMU);
+
 #ifdef CONFIG_HVM
     /* This doesn't catch !CONFIG_HVM case but it is better than nothing */
     BUILD_BUG_ON(X86_EMU_ALL != XEN_X86_EMU_ALL);
 #endif
 
-    if ( is_hvm_domain(d) )
-    {
-        if ( is_hardware_domain(d) &&
-             emflags != (X86_EMU_VPCI | X86_EMU_LAPIC | X86_EMU_IOAPIC) )
-            return false;
-        if ( !is_hardware_domain(d) &&
-             /* HVM PIRQ feature is user-selectable. */
-             (emflags & ~X86_EMU_USE_PIRQ) !=
-             (X86_EMU_ALL & ~(X86_EMU_VPCI | X86_EMU_USE_PIRQ)) &&
-             emflags != X86_EMU_LAPIC )
-            return false;
-    }
-    else if ( emflags != 0 && emflags != X86_EMU_PIT )
-    {
-        /* PV or classic PVH. */
-        return false;
-    }
+    for ( i = 0; i < ARRAY_SIZE(configs); i++ )
+        if ( (caps & configs[i].caps) == caps &&
+             (emflags & ~configs[i].opt) == configs[i].min )
+            return true;
 
-    return true;
+    return false;
 }
 
 void __init arch_init_idle_domain(struct domain *d)
-- 
2.34.1
Re: [PATCH v7] xen/domain: rewrite emulation_flags_ok()
Posted by Roger Pau Monné 4 months, 2 weeks ago
On Tue, Jun 10, 2025 at 02:45:12PM +0000, dmkhn@proton.me wrote:
> From: Denis Mukhin <dmukhin@ford.com>
> 
> Rewrite emulation_flags_ok() to simplify future modifications.
> 
> No functional change intended.
> 
> Signed-off-by: Denis Mukhin <dmukhin@ford.com>
> ---
> Changes since v6:
> - simplified checks for PV further
> 
> Link to v6: https://lore.kernel.org/xen-devel/20250610004216.3012253-1-dmukhin@ford.com/
> Link to CI: https://gitlab.com/xen-project/people/dmukhin/xen/-/pipelines/1862559089
> ---
>  xen/arch/x86/domain.c | 78 +++++++++++++++++++++++++++++++++----------
>  1 file changed, 60 insertions(+), 18 deletions(-)
> 
> diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
> index 7536b6c871..fdbd064ebf 100644
> --- a/xen/arch/x86/domain.c
> +++ b/xen/arch/x86/domain.c
> @@ -743,32 +743,74 @@ int arch_sanitise_domain_config(struct xen_domctl_createdomain *config)
>      return 0;
>  }
>  
> +/*
> + * Verify that the domain's emulation flags resolve to a supported configuration.
> + *
> + * This ensures we only allow a known, safe subset of emulation combinations
> + * (for both functionality and security). Arbitrary mixes are likely to cause
> + * errors (e.g. null pointer dereferences).
> + *
> + * NB: use the internal X86_EMU_XXX symbols, not the public XEN_X86_EMU_XXX
> + * symbols.

Not sure if it's worth expanding the sentence a bit to add the reason
why X86_EMU_XXX should be used (so that we take build-time config
options into account for short-circuited emulations).

> + */
>  static bool emulation_flags_ok(const struct domain *d, uint32_t emflags)
>  {
> +    enum {
> +        CAP_PV          = BIT(0, U),
> +        CAP_HVM         = BIT(1, U),
> +        CAP_HWDOM       = BIT(2, U),
> +        CAP_DOMU        = BIT(3, U),
> +    };

You probably want to name this type.

> +    static const struct {
> +        unsigned int caps;

So it can be used here

> +        uint32_t min;
> +        uint32_t opt;
> +    } configs[] = {
> +#ifdef CONFIG_PV
> +        /* PV dom0 and domU */
> +        {
> +            .caps   = CAP_PV | CAP_HWDOM | CAP_DOMU,
> +            .opt    = X86_EMU_PIT,
> +        },
> +#endif /* #ifdef CONFIG_PV */
> +
> +#ifdef CONFIG_HVM
> +        /* PVH dom0 */
> +        {
> +            .caps   = CAP_HVM | CAP_HWDOM,
> +            .min    = X86_EMU_LAPIC | X86_EMU_IOAPIC | X86_EMU_VPCI,
> +        },
> +
> +        /* PVH domU */
> +        {
> +            .caps   = CAP_HVM | CAP_DOMU,
> +            .min    = X86_EMU_LAPIC,
> +        },
> +
> +        /* HVM domU */
> +        {
> +            .caps   = CAP_HVM | CAP_DOMU,
> +            .min    = X86_EMU_ALL & ~(X86_EMU_VPCI | X86_EMU_USE_PIRQ),
> +            /* HVM PIRQ feature is user-selectable. */
> +            .opt    = X86_EMU_USE_PIRQ,
> +        },
> +#endif /* #ifdef CONFIG_HVM */
> +    };
> +    unsigned int i;
> +    unsigned int caps = (is_pv_domain(d) ? CAP_PV : CAP_HVM) |
> +                        (is_hardware_domain(d) ? CAP_HWDOM : CAP_DOMU);

And here instead of using unsigned int?

> +
>  #ifdef CONFIG_HVM
>      /* This doesn't catch !CONFIG_HVM case but it is better than nothing */
>      BUILD_BUG_ON(X86_EMU_ALL != XEN_X86_EMU_ALL);
>  #endif
>  
> -    if ( is_hvm_domain(d) )
> -    {
> -        if ( is_hardware_domain(d) &&
> -             emflags != (X86_EMU_VPCI | X86_EMU_LAPIC | X86_EMU_IOAPIC) )
> -            return false;
> -        if ( !is_hardware_domain(d) &&
> -             /* HVM PIRQ feature is user-selectable. */
> -             (emflags & ~X86_EMU_USE_PIRQ) !=
> -             (X86_EMU_ALL & ~(X86_EMU_VPCI | X86_EMU_USE_PIRQ)) &&
> -             emflags != X86_EMU_LAPIC )
> -            return false;
> -    }
> -    else if ( emflags != 0 && emflags != X86_EMU_PIT )
> -    {
> -        /* PV or classic PVH. */
> -        return false;
> -    }
> +    for ( i = 0; i < ARRAY_SIZE(configs); i++ )
> +        if ( (caps & configs[i].caps) == caps &&
> +             (emflags & ~configs[i].opt) == configs[i].min )
> +            return true;
>  
> -    return true;
> +    return false;
>  }
>  
>  void __init arch_init_idle_domain(struct domain *d)
> -- 
> 2.34.1
> 
>
Re: [PATCH v7] xen/domain: rewrite emulation_flags_ok()
Posted by Roger Pau Monné 4 months, 2 weeks ago
On Mon, Jun 16, 2025 at 04:03:14PM +0200, Roger Pau Monné wrote:
> On Tue, Jun 10, 2025 at 02:45:12PM +0000, dmkhn@proton.me wrote:
> > From: Denis Mukhin <dmukhin@ford.com>
> > 
> > Rewrite emulation_flags_ok() to simplify future modifications.
> > 
> > No functional change intended.
> > 
> > Signed-off-by: Denis Mukhin <dmukhin@ford.com>
> > ---
> > Changes since v6:
> > - simplified checks for PV further
> > 
> > Link to v6: https://lore.kernel.org/xen-devel/20250610004216.3012253-1-dmukhin@ford.com/
> > Link to CI: https://gitlab.com/xen-project/people/dmukhin/xen/-/pipelines/1862559089
> > ---
> >  xen/arch/x86/domain.c | 78 +++++++++++++++++++++++++++++++++----------
> >  1 file changed, 60 insertions(+), 18 deletions(-)
> > 
> > diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
> > index 7536b6c871..fdbd064ebf 100644
> > --- a/xen/arch/x86/domain.c
> > +++ b/xen/arch/x86/domain.c
> > @@ -743,32 +743,74 @@ int arch_sanitise_domain_config(struct xen_domctl_createdomain *config)
> >      return 0;
> >  }
> >  
> > +/*
> > + * Verify that the domain's emulation flags resolve to a supported configuration.
> > + *
> > + * This ensures we only allow a known, safe subset of emulation combinations
> > + * (for both functionality and security). Arbitrary mixes are likely to cause
> > + * errors (e.g. null pointer dereferences).
> > + *
> > + * NB: use the internal X86_EMU_XXX symbols, not the public XEN_X86_EMU_XXX
> > + * symbols.
> 
> Not sure if it's worth expanding the sentence a bit to add the reason
> why X86_EMU_XXX should be used (so that we take build-time config
> options into account for short-circuited emulations).

With this comment adjusted as you have in v8:

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

Thanks, Roger.