x86/msr: fix handling of MSR_IA32_PERF_{STATUS/CTL}

Currently a PV hardware domain can also be given control over the CPU
frequency, and such guest is allowed to write to MSR_IA32_PERF_CTL.
However since commit 322ec7c89f6 the default behavior has been changed
to reject accesses to not explicitly handled MSRs, preventing PV
guests that manage CPU frequency from reading
MSR_IA32_PERF_{STATUS/CTL}.

Additionally some HVM guests (Windows at least) will attempt to read
MSR_IA32_PERF_CTL and will panic if given back a #GP fault:

  vmx.c:3035:d8v0 RDMSR 0x00000199 unimplemented
  d8v0 VIRIDIAN CRASH: 3b c0000096 fffff806871c1651 ffffda0253683720 0

Move the handling of MSR_IA32_PERF_{STATUS/CTL} to the common MSR
handling shared between HVM and PV guests, and add an explicit case
for reads to MSR_IA32_PERF_{STATUS/CTL}.

Restore previous behavior and allow PV guests with the required
permissions to read the contents of the mentioned MSRs. Non privileged
guests will get 0 when trying to read those registers, as writes to
MSR_IA32_PERF_CTL by such guest will already be silently dropped.

Fixes: 322ec7c89f6 ('x86/pv: disallow access to unknown MSRs')
Fixes: 84e848fd7a1 ('x86/hvm: disallow access to unknown MSRs')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

diff --git a/xen/arch/x86/msr.c b/xen/arch/x86/msr.c
index 9c69ef8792c00d5d12b7f7c4340a906fa62310a7..be8e36386250f7321549473e9b93d6511784977e 100644 (file)
--- a/xen/arch/x86/msr.c
+++ b/xen/arch/x86/msr.c
@@ -242,6 +242,25 @@ int guest_rdmsr(struct vcpu *v, uint32_t msr, uint64_t *val)
             goto gp_fault;
         break;
 
+        /*
+         * These MSRs are not enumerated in CPUID.  They have been around
+         * since the Pentium 4, and implemented by other vendors.
+         *
+         * Some versions of Windows try reading these before setting up a #GP
+         * handler, and Linux has several unguarded reads as well.  Provide
+         * RAZ semantics, in general, but permit a cpufreq controller dom0 to
+         * have full access.
+         */
+    case MSR_IA32_PERF_STATUS:
+    case MSR_IA32_PERF_CTL:
+        if ( !(cp->x86_vendor & (X86_VENDOR_INTEL | X86_VENDOR_CENTAUR)) )
+            goto gp_fault;
+
+        *val = 0;
+        if ( likely(!is_cpufreq_controller(d)) || rdmsr_safe(msr, *val) == 0 )
+            break;
+        goto gp_fault;
+
     case MSR_IA32_THERM_STATUS:
         if ( cp->x86_vendor != X86_VENDOR_INTEL )
             goto gp_fault;
@@ -448,6 +467,21 @@ int guest_wrmsr(struct vcpu *v, uint32_t msr, uint64_t val)
             goto gp_fault;
         break;
 
+        /*
+         * This MSR is not enumerated in CPUID.  It has been around since the
+         * Pentium 4, and implemented by other vendors.
+         *
+         * To match the RAZ semantics, implement as write-discard, except for
+         * a cpufreq controller dom0 which has full access.
+         */
+    case MSR_IA32_PERF_CTL:
+        if ( !(cp->x86_vendor & (X86_VENDOR_INTEL | X86_VENDOR_CENTAUR)) )
+            goto gp_fault;
+
+        if ( likely(!is_cpufreq_controller(d)) || wrmsr_safe(msr, val) == 0 )
+            break;
+        goto gp_fault;
+
     case MSR_X2APIC_FIRST ... MSR_X2APIC_LAST:
         if ( !is_hvm_domain(d) || v != curr )
             goto gp_fault;

diff --git a/xen/arch/x86/pv/emul-priv-op.c b/xen/arch/x86/pv/emul-priv-op.c
index 7cc16d6eda423b5e7c22b319f56ff380281001a7..dbceed8a05fdd3143dcd0e38fffa97ffb7109b9e 100644 (file)
--- a/xen/arch/x86/pv/emul-priv-op.c
+++ b/xen/arch/x86/pv/emul-priv-op.c
@@ -849,12 +849,6 @@ static inline uint64_t guest_misc_enable(uint64_t val)
     return val;
 }
 
-static inline bool is_cpufreq_controller(const struct domain *d)
-{
-    return ((cpufreq_controller == FREQCTL_dom0_kernel) &&
-            is_hardware_domain(d));
-}
-
 static uint64_t guest_efer(const struct domain *d)
 {
     uint64_t val;
@@ -1121,14 +1115,6 @@ static int write_msr(unsigned int reg, uint64_t val,
             return X86EMUL_OKAY;
         break;
 
-    case MSR_IA32_PERF_CTL:
-        if ( boot_cpu_data.x86_vendor != X86_VENDOR_INTEL )
-            break;
-        if ( likely(!is_cpufreq_controller(currd)) ||
-             wrmsr_safe(reg, val) == 0 )
-            return X86EMUL_OKAY;
-        break;
-
     case MSR_IA32_THERM_CONTROL:
     case MSR_IA32_ENERGY_PERF_BIAS:
         if ( boot_cpu_data.x86_vendor != X86_VENDOR_INTEL )

diff --git a/xen/include/xen/sched.h b/xen/include/xen/sched.h
index d8ed83f8697997a4db71a83963235985964a493e..ff8d0772b3667456f4b12e26c114478fa0caf0cd 100644 (file)
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -1069,6 +1069,22 @@ extern enum cpufreq_controller {
     FREQCTL_none, FREQCTL_dom0_kernel, FREQCTL_xen
 } cpufreq_controller;
 
+static always_inline bool is_cpufreq_controller(const struct domain *d)
+{
+    /*
+     * A PV dom0 can be nominated as the cpufreq controller, instead of using
+     * Xen's cpufreq driver, at which point dom0 gets direct access to certain
+     * MSRs.
+     *
+     * This interface only works when dom0 is identity pinned and has the same
+     * number of vCPUs as pCPUs on the system.
+     *
+     * It would be far better to paravirtualise the interface.
+     */
+    return (is_pv_domain(d) && is_hardware_domain(d) &&
+            cpufreq_controller == FREQCTL_dom0_kernel);
+}
+
 int cpupool_move_domain(struct domain *d, struct cpupool *c);
 int cpupool_do_sysctl(struct xen_sysctl_cpupool_op *op);
 int cpupool_get_id(const struct domain *d);