x86/cpu: Context switch cpuid masks and faulting state in context_switch()

A single ctxt_switch_levelling() function pointer is provided
(defaulting to an empty nop), which is overridden in the appropriate
$VENDOR_init_levelling().

set_cpuid_faulting() is made private and included within
intel_ctxt_switch_levelling().

One (attempted) functional change is that the faulting configuration should
not be special cased for dom0.  It turns out that the toolstack relies on the
special case (and indeed, on being a PV domain in the first place) to
correctly build HVM domains.

For now, the control domain is left as a special case, until futher work can
be completed to remove the restriction.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Jan Beulich <JBeulich@suse.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

xen/arch/x86/cpu/amd.c

@@ -331,6 +331,9 @@ static void __init noinline amd_init_levelling(void)
 		       (uint32_t)cpuidmask_defaults._7ab0,
 		       (uint32_t)cpuidmask_defaults._6c);
 	}
+
+	if (levelling_caps)
+		ctxt_switch_levelling = amd_ctxt_switch_levelling;
 }
 
 /*

xen/arch/x86/cpu/common.c

@@ -88,6 +88,13 @@ static const struct cpu_dev default_cpu = {
 };
 static const struct cpu_dev *this_cpu = &default_cpu;
 
+static void default_ctxt_switch_levelling(const struct domain *nextd)
+{
+	/* Nop */
+}
+void (* __read_mostly ctxt_switch_levelling)(const struct domain *nextd) =
+	default_ctxt_switch_levelling;
+
 bool_t opt_cpu_info;
 boolean_param("cpuinfo", opt_cpu_info);
 

xen/arch/x86/cpu/intel.c

@@ -32,13 +32,15 @@ static bool_t __init probe_intel_cpuid_faulting(void)
 	return 1;
 }
 
-static DEFINE_PER_CPU(bool_t, cpuid_faulting_enabled);
-void set_cpuid_faulting(bool_t enable)
+static void set_cpuid_faulting(bool_t enable)
 {
+	static DEFINE_PER_CPU(bool_t, cpuid_faulting_enabled);
+	bool_t *this_enabled = &this_cpu(cpuid_faulting_enabled);
 	uint32_t hi, lo;
 
-	if (!cpu_has_cpuid_faulting ||
-	    this_cpu(cpuid_faulting_enabled) == enable )
+	ASSERT(cpu_has_cpuid_faulting);
+
+	if (*this_enabled == enable)
 		return;
 
 	rdmsr(MSR_INTEL_MISC_FEATURES_ENABLES, lo, hi);
@@ -47,7 +49,7 @@ void set_cpuid_faulting(bool_t enable)
 		lo |= MSR_MISC_FEATURES_CPUID_FAULTING;
 	wrmsr(MSR_INTEL_MISC_FEATURES_ENABLES, lo, hi);
 
-	this_cpu(cpuid_faulting_enabled) = enable;
+	*this_enabled = enable;
 }
 
 /*
@@ -154,6 +156,28 @@ static void intel_ctxt_switch_levelling(const struct domain *nextd)
 	struct cpuidmasks *these_masks = &this_cpu(cpuidmasks);
 	const struct cpuidmasks *masks = &cpuidmask_defaults;
 
+	if (cpu_has_cpuid_faulting) {
+		/*
+		 * We *should* be enabling faulting for the control domain.
+		 *
+		 * Unfortunately, the domain builder (having only ever been a
+		 * PV guest) expects to be able to see host cpuid state in a
+		 * native CPUID instruction, to correctly build a CPUID policy
+		 * for HVM guests (notably the xstate leaves).
+		 *
+		 * This logic is fundimentally broken for HVM toolstack
+		 * domains, and faulting causes PV guests to behave like HVM
+		 * guests from their point of view.
+		 *
+		 * Future development plans will move responsibility for
+		 * generating the maximum full cpuid policy into Xen, at which
+		 * this problem will disappear.
+		 */
+		set_cpuid_faulting(nextd && is_pv_domain(nextd) &&
+				   !is_control_domain(nextd));
+		return;
+	}
+
 #define LAZY(msr, field)						\
 	({								\
 		if (unlikely(these_masks->field != masks->field) &&	\
@@ -227,6 +251,9 @@ static void __init noinline intel_init_levelling(void)
 			       (uint32_t)cpuidmask_defaults.e1cd,
 			       (uint32_t)cpuidmask_defaults.Da1);
 	}
+
+	if (levelling_caps)
+		ctxt_switch_levelling = intel_ctxt_switch_levelling;
 }
 
 static void early_init_intel(struct cpuinfo_x86 *c)

xen/arch/x86/crash.c

@@ -189,6 +189,9 @@ void machine_crash_shutdown(void)
 
     nmi_shootdown_cpus();
 
+    /* Reset CPUID masking and faulting to the host's default. */
+    ctxt_switch_levelling(NULL);
+
     info = kexec_crash_save_info();
     info->xen_phys_start = xen_phys_start;
     info->dom0_pfn_to_mfn_frame_list_list =

xen/arch/x86/domain.c

@@ -2087,9 +2087,7 @@ void context_switch(struct vcpu *prev, struct vcpu *next)
             load_segments(next);
         }
 
-        set_cpuid_faulting(is_pv_domain(nextd) &&
-                           !is_control_domain(nextd) &&
-                           !is_hardware_domain(nextd));
+        ctxt_switch_levelling(nextd);
     }
 
     context_saved(prev);

xen/include/asm-x86/processor.h

@@ -209,7 +209,7 @@ extern struct cpuinfo_x86 boot_cpu_data;
 extern struct cpuinfo_x86 cpu_data[];
 #define current_cpu_data cpu_data[smp_processor_id()]
 
-extern void set_cpuid_faulting(bool_t enable);
+extern void (*ctxt_switch_levelling)(const struct domain *nextd);
 
 extern u64 host_pat;
 extern bool_t opt_cpu_info;