[PATCH 1/4] x86/domctl: Implement XEN_DOMCTL_{get, set}_vcpu_msrs

[Andrew Cooper <andrew.cooper3@citrix.com>]

Despite my 'Reviewed-by' tag on c/s 65e3554908 "x86/PV: support data
breakpoint extension registers", I have re-evaluated my position as far as the
hypercall interface is concerned.

Previously, for the sake of not modifying the migration code in libxc,
XEN_DOMCTL_get_ext_vcpucontext would jump though hoops to return -ENOBUFS if
and only if MSRs were in use and no buffer was present.

This is fragile, and awkward from a toolstack point-of-view when actually
sending MSR content in the migration stream.  It also complicates fixing a
further race condition, between querying the number of MSRs for a vcpu, and
the vcpu touching a new one.

As this code is still only in staging, take this opportunity to redesign the
interface.  This patch introduces the brand new XEN_DOMCTL_{get,set}_vcpu_msrs
subops.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
CC: Keir Fraser <keir@xen.org>
CC: Jan Beulich <JBeulich@suse.com>
---
 xen/arch/x86/domctl.c       |  126 +++++++++++++++++++++++++++++++++++++++++++
 xen/include/public/domctl.h |   31 +++++++++++
 2 files changed, 157 insertions(+)

diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
index 1285dd0..c0c2d1b 100644
--- a/xen/arch/x86/domctl.c
+++ b/xen/arch/x86/domctl.c
@@ -1341,6 +1341,132 @@ long arch_do_domctl(
     }
     break;
 
+    case XEN_DOMCTL_get_vcpu_msrs:
+    case XEN_DOMCTL_set_vcpu_msrs:
+    {
+        struct xen_domctl_vcpu_msrs *vmsrs = &domctl->u.vcpu_msrs;
+        struct xen_domctl_vcpu_msr msr;
+        struct vcpu *v;
+        uint32_t nr_msrs = 0;
+
+        ret = -ESRCH;
+        if ( (vmsrs->vcpu >= d->max_vcpus) ||
+             ((v = d->vcpu[vmsrs->vcpu]) == NULL) )
+            break;
+
+        ret = -EINVAL;
+        if ( (v == current) || /* no vcpu_pause() */
+             !is_pv_domain(d) )
+            break;
+
+        /* Count maximum number of optional msrs. */
+        if ( boot_cpu_has(X86_FEATURE_DBEXT) )
+            nr_msrs += 4;
+
+        if ( domctl->cmd == XEN_DOMCTL_get_vcpu_msrs )
+        {
+            /* NULL guest handle is a request for max size. */
+            if ( guest_handle_is_null(vmsrs->msrs) ||
+                 (vmsrs->msr_count < nr_msrs) )
+            {
+                vmsrs->msr_count = nr_msrs;
+                ret = guest_handle_is_null(vmsrs->msrs) ? 0 : -ENOBUFS;
+            }
+            else
+            {
+                ret = i = 0;
+
+                vcpu_pause(v);
+
+                if ( boot_cpu_has(X86_FEATURE_DBEXT) )
+                {
+                    unsigned int j;
+
+                    if ( v->arch.pv_vcpu.dr_mask[0] )
+                    {
+                        msr.index = MSR_AMD64_DR0_ADDRESS_MASK;
+                        msr.reserved = 0;
+                        msr.value = v->arch.pv_vcpu.dr_mask[0];
+                        if ( copy_to_guest_offset(vmsrs->msrs, i, &msr, 1) )
+                            ret = -EFAULT;
+                        else
+                            ++i;
+                    }
+
+                    for ( j = 0; j < 3; ++j )
+                    {
+                        if ( !v->arch.pv_vcpu.dr_mask[1 + j] )
+                            continue;
+                        if ( !ret )
+                        {
+                            msr.index = MSR_AMD64_DR1_ADDRESS_MASK + j;
+                            msr.reserved = 0;
+                            msr.value = v->arch.pv_vcpu.dr_mask[1 + j];
+                            if ( copy_to_guest_offset(vmsrs->msrs, i, &msr, 1) )
+                                ret = -EFAULT;
+                            else
+                                ++i;
+                        }
+                    }
+
+                }
+
+                vcpu_unpause(v);
+
+                /* Check we didn't lie to userspace then overflow the buffer */
+                BUG_ON(i > nr_msrs);
+                vmsrs->msr_count = i;
+            }
+
+            copyback = 1;
+        }
+        else
+        {
+            ret = -EINVAL;
+            if ( vmsrs->msr_count > nr_msrs )
+                break;
+
+            vcpu_pause(v);
+
+            for ( i = 0; i < vmsrs->msr_count; ++i )
+            {
+                ret = -EFAULT;
+                if ( copy_from_guest_offset(&msr, vmsrs->msrs, i, 1) )
+                    break;
+
+                ret = -EINVAL;
+                if ( msr.reserved )
+                    break;
+
+                switch ( msr.index )
+                {
+                case MSR_AMD64_DR0_ADDRESS_MASK:
+                    if ( !boot_cpu_has(X86_FEATURE_DBEXT) ||
+                         (msr.value >> 32) )
+                        break;
+                    v->arch.pv_vcpu.dr_mask[0] = msr.value;
+                    continue;
+
+                case MSR_AMD64_DR1_ADDRESS_MASK ...
+                    MSR_AMD64_DR3_ADDRESS_MASK:
+                    if ( !boot_cpu_has(X86_FEATURE_DBEXT) ||
+                         (msr.value >> 32) )
+                        break;
+                    msr.index -= MSR_AMD64_DR1_ADDRESS_MASK - 1;
+                    v->arch.pv_vcpu.dr_mask[msr.index] = msr.value;
+                    continue;
+                }
+                break;
+            }
+
+            vcpu_unpause(v);
+
+            if ( i == vmsrs->msr_count )
+                ret = 0;
+        }
+    }
+    break;
+
     default:
         ret = iommu_do_domctl(domctl, d, u_domctl);
         break;
diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index 385b053..7a13e25 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -895,6 +895,34 @@ struct xen_domctl_cacheflush {
 typedef struct xen_domctl_cacheflush xen_domctl_cacheflush_t;
 DEFINE_XEN_GUEST_HANDLE(xen_domctl_cacheflush_t);
 
+#if defined(__i386__) || defined(__x86_64__)
+struct xen_domctl_vcpu_msr {
+    uint32_t         index;
+    uint32_t         reserved;
+    uint64_aligned_t value;
+};
+typedef struct xen_domctl_vcpu_msr xen_domctl_vcpu_msr_t;
+DEFINE_XEN_GUEST_HANDLE(xen_domctl_vcpu_msr_t);
+
+/*
+ * XEN_DOMCTL_set_vcpu_msrs / XEN_DOMCTL_get_vcpu_msrs.
+ *
+ * For GET, a NULL 'msrs' guest handle is a request for the maximum
+ * 'msr_count' which is required to save the vcpu state.
+ *
+ * When actually getting the MSRs, 'msr_count' shall reflect the actual number
+ * of MSRs saved, which might be fewer than the maximum, if some are not
+ * currently used by the vcpu.
+ */
+struct xen_domctl_vcpu_msrs {
+    uint32_t vcpu;                                   /* IN     */
+    uint32_t msr_count;                              /* IN/OUT */
+    XEN_GUEST_HANDLE_64(xen_domctl_vcpu_msr_t) msrs; /* IN/OUT */
+};
+typedef struct xen_domctl_vcpu_msrs xen_domctl_vcpu_msrs_t;
+DEFINE_XEN_GUEST_HANDLE(xen_domctl_vcpu_msrs_t);
+#endif
+
 struct xen_domctl {
     uint32_t cmd;
 #define XEN_DOMCTL_createdomain                   1
@@ -965,6 +993,8 @@ struct xen_domctl {
 #define XEN_DOMCTL_getnodeaffinity               69
 #define XEN_DOMCTL_set_max_evtchn                70
 #define XEN_DOMCTL_cacheflush                    71
+#define XEN_DOMCTL_get_vcpu_msrs                 72
+#define XEN_DOMCTL_set_vcpu_msrs                 73
 #define XEN_DOMCTL_gdbsx_guestmemio            1000
 #define XEN_DOMCTL_gdbsx_pausevcpu             1001
 #define XEN_DOMCTL_gdbsx_unpausevcpu           1002
@@ -1014,6 +1044,7 @@ struct xen_domctl {
 #if defined(__i386__) || defined(__x86_64__)
         struct xen_domctl_cpuid             cpuid;
         struct xen_domctl_vcpuextstate      vcpuextstate;
+        struct xen_domctl_vcpu_msrs         vcpu_msrs;
 #endif
         struct xen_domctl_set_access_required access_required;
         struct xen_domctl_audit_p2m         audit_p2m;
-- 
1.7.10.4