[PATCH RFC v3 10/11] KVM: Add dirty-gfn-count property

[Peter Xu <peterx@redhat.com>]

Add a parameter for dirty gfn count for dirty rings.  If zero, dirty ring is
disabled.  Otherwise dirty ring will be enabled with the per-vcpu gfn count as
specified.  If dirty ring cannot be enabled due to unsupported kernel or
illegal parameter, it'll fallback to dirty logging.

By default, dirty ring is not enabled (dirty-gfn-count default to 0).

Signed-off-by: Peter Xu <peterx@redhat.com>
---
 accel/kvm/kvm-all.c | 72 +++++++++++++++++++++++++++++++++++++++++++++
 qemu-options.hx     |  5 ++++
 2 files changed, 77 insertions(+)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index b9aaa7912c..dd017d0720 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -128,6 +128,9 @@ struct KVMState
         KVMMemoryListener *ml;
         AddressSpace *as;
     } *as;
+    bool kvm_dirty_ring_enabled;    /* Whether KVM dirty ring is enabled */
+    uint64_t kvm_dirty_ring_size;   /* Size of the per-vcpu dirty ring */
+    uint32_t kvm_dirty_gfn_count;   /* Number of dirty GFNs per ring */
 };
 
 KVMState *kvm_state;
@@ -2129,6 +2132,40 @@ static int kvm_init(MachineState *ms)
     s->memory_listener.listener.coalesced_io_add = kvm_coalesce_mmio_region;
     s->memory_listener.listener.coalesced_io_del = kvm_uncoalesce_mmio_region;
 
+    /*
+     * Enable KVM dirty ring if supported, otherwise fall back to
+     * dirty logging mode
+     */
+    if (s->kvm_dirty_gfn_count > 0) {
+        uint64_t ring_size;
+
+        ring_size = s->kvm_dirty_gfn_count * sizeof(struct kvm_dirty_gfn);
+
+        /* Read the max supported pages */
+        ret = kvm_vm_check_extension(kvm_state, KVM_CAP_DIRTY_LOG_RING);
+        if (ret > 0) {
+            if (ring_size > ret) {
+                error_report("KVM dirty GFN count %" PRIu32 " too big "
+                             "(maximum is %ld).  Please use a smaller value.",
+                             s->kvm_dirty_gfn_count,
+                             ret / sizeof(struct kvm_dirty_gfn));
+                ret = -EINVAL;
+                goto err;
+            }
+
+            ret = kvm_vm_enable_cap(s, KVM_CAP_DIRTY_LOG_RING, 0, ring_size);
+            if (ret) {
+                error_report("Enabling of KVM dirty ring failed: %d. "
+                             "Suggested mininum value is 1024. "
+                             "Please also make sure it's a power of two.", ret);
+                goto err;
+            }
+
+            s->kvm_dirty_ring_size = ring_size;
+            s->kvm_dirty_ring_enabled = true;
+        }
+    }
+
     kvm_memory_listener_register(s, &s->memory_listener,
                                  &address_space_memory, 0);
     memory_listener_register(&kvm_io_listener,
@@ -3089,6 +3126,33 @@ bool kvm_kernel_irqchip_split(void)
     return kvm_state->kernel_irqchip_split == ON_OFF_AUTO_ON;
 }
 
+static void kvm_get_dirty_gfn_count(Object *obj, Visitor *v,
+                                    const char *name, void *opaque,
+                                    Error **errp)
+{
+    KVMState *s = KVM_STATE(obj);
+    uint32_t value = s->kvm_dirty_gfn_count;
+
+    visit_type_uint32(v, name, &value, errp);
+}
+
+static void kvm_set_dirty_gfn_count(Object *obj, Visitor *v,
+                                    const char *name, void *opaque,
+                                    Error **errp)
+{
+    KVMState *s = KVM_STATE(obj);
+    Error *error = NULL;
+    uint32_t value;
+
+    visit_type_uint32(v, name, &value, &error);
+    if (error) {
+        error_propagate(errp, error);
+        return;
+    }
+
+    s->kvm_dirty_gfn_count = value;
+}
+
 static void kvm_accel_instance_init(Object *obj)
 {
     KVMState *s = KVM_STATE(obj);
@@ -3096,6 +3160,8 @@ static void kvm_accel_instance_init(Object *obj)
     s->kvm_shadow_mem = -1;
     s->kernel_irqchip_allowed = true;
     s->kernel_irqchip_split = ON_OFF_AUTO_AUTO;
+    /* KVM dirty ring is by default off */
+    s->kvm_dirty_gfn_count = 0;
 }
 
 static void kvm_accel_class_init(ObjectClass *oc, void *data)
@@ -3117,6 +3183,12 @@ static void kvm_accel_class_init(ObjectClass *oc, void *data)
         NULL, NULL);
     object_class_property_set_description(oc, "kvm-shadow-mem",
         "KVM shadow MMU size");
+
+    object_class_property_add(oc, "dirty-gfn-count", "uint32",
+        kvm_get_dirty_gfn_count, kvm_set_dirty_gfn_count,
+        NULL, NULL);
+    object_class_property_set_description(oc, "dirty-gfn-count",
+        "KVM dirty GFN count (=0 to disable dirty ring)");
 }
 
 static const TypeInfo kvm_accel_type = {
diff --git a/qemu-options.hx b/qemu-options.hx
index 93bde2bbc8..b59d47473e 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -124,6 +124,7 @@ DEF("accel", HAS_ARG, QEMU_OPTION_accel,
     "                kernel-irqchip=on|off|split controls accelerated irqchip support (default=on)\n"
     "                kvm-shadow-mem=size of KVM shadow MMU in bytes\n"
     "                tb-size=n (TCG translation block cache size)\n"
+    "                dirty-gfn-count=n (KVM dirty ring GFN count, default 0)\n"
     "                thread=single|multi (enable multi-threaded TCG)\n", QEMU_ARCH_ALL)
 SRST
 ``-accel name[,prop=value[,...]]``
@@ -158,6 +159,10 @@ SRST
         where both the back-end and front-ends support it and no
         incompatible TCG features have been enabled (e.g.
         icount/replay).
+
+    ``dirty-gfn-count=n``
+        Controls the per-vcpu KVM dirty ring GFN count (=0 to disable).
+
 ERST
 
 DEF("smp", HAS_ARG, QEMU_OPTION_smp,
-- 
2.26.2