qcow2: implement lazy refcounts

Lazy refcounts is a performance optimization for qcow2 that postpones
refcount metadata updates and instead marks the image dirty.  In the
case of crash or power failure the image will be left in a dirty state
and repaired next time it is opened.

Reducing metadata I/O is important for cache=writethrough and
cache=directsync because these modes guarantee that data is on disk
after each write (hence we cannot take advantage of caching updates in
RAM).  Refcount metadata is not needed for guest->file block address
translation and therefore does not need to be on-disk at the time of
write completion - this is the motivation behind the lazy refcount
optimization.

The lazy refcount optimization must be enabled at image creation time:

  qemu-img create -f qcow2 -o compat=1.1,lazy_refcounts=on a.qcow2 10G
  qemu-system-x86_64 -drive if=virtio,file=a.qcow2,cache=writethrough

Update qemu-iotests 031 and 036 since the extension header size changes
when we add feature bit table entries.

Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
diff --git a/block/qcow2.c b/block/qcow2.c
index 7fe1567..fd5e214 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -215,6 +215,41 @@
 }
 
 /*
+ * Sets the dirty bit and flushes afterwards if necessary.
+ *
+ * The incompatible_features bit is only set if the image file header was
+ * updated successfully.  Therefore it is not required to check the return
+ * value of this function.
+ */
+static int qcow2_mark_dirty(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t val;
+    int ret;
+
+    assert(s->qcow_version >= 3);
+
+    if (s->incompatible_features & QCOW2_INCOMPAT_DIRTY) {
+        return 0; /* already dirty */
+    }
+
+    val = cpu_to_be64(s->incompatible_features | QCOW2_INCOMPAT_DIRTY);
+    ret = bdrv_pwrite(bs->file, offsetof(QCowHeader, incompatible_features),
+                      &val, sizeof(val));
+    if (ret < 0) {
+        return ret;
+    }
+    ret = bdrv_flush(bs->file);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Only treat image as dirty if the header was updated successfully */
+    s->incompatible_features |= QCOW2_INCOMPAT_DIRTY;
+    return 0;
+}
+
+/*
  * Clears the dirty bit and flushes before if necessary.  Only call this
  * function when there are no pending requests, it does not guard against
  * concurrent requests dirtying the image.
@@ -752,6 +787,11 @@
             goto fail;
         }
 
+        if (l2meta.nb_clusters > 0 &&
+            (s->compatible_features & QCOW2_COMPAT_LAZY_REFCOUNTS)) {
+            qcow2_mark_dirty(bs);
+        }
+
         cluster_offset = l2meta.cluster_offset;
         assert((cluster_offset & 511) == 0);
 
@@ -994,6 +1034,11 @@
             .bit  = QCOW2_INCOMPAT_DIRTY_BITNR,
             .name = "dirty bit",
         },
+        {
+            .type = QCOW2_FEAT_TYPE_COMPATIBLE,
+            .bit  = QCOW2_COMPAT_LAZY_REFCOUNTS_BITNR,
+            .name = "lazy refcounts",
+        },
     };
 
     ret = header_ext_add(buf, QCOW2_EXT_MAGIC_FEATURE_TABLE,
@@ -1176,6 +1221,11 @@
         header.crypt_method = cpu_to_be32(QCOW_CRYPT_NONE);
     }
 
+    if (flags & BLOCK_FLAG_LAZY_REFCOUNTS) {
+        header.compatible_features |=
+            cpu_to_be64(QCOW2_COMPAT_LAZY_REFCOUNTS);
+    }
+
     ret = bdrv_pwrite(bs, 0, &header, sizeof(header));
     if (ret < 0) {
         goto out;
@@ -1289,6 +1339,8 @@
                     options->value.s);
                 return -EINVAL;
             }
+        } else if (!strcmp(options->name, BLOCK_OPT_LAZY_REFCOUNTS)) {
+            flags |= options->value.n ? BLOCK_FLAG_LAZY_REFCOUNTS : 0;
         }
         options++;
     }
@@ -1299,6 +1351,12 @@
         return -EINVAL;
     }
 
+    if (version < 3 && (flags & BLOCK_FLAG_LAZY_REFCOUNTS)) {
+        fprintf(stderr, "Lazy refcounts only supported with compatibility "
+                "level 1.1 and above (use compat=1.1 or greater)\n");
+        return -EINVAL;
+    }
+
     return qcow2_create2(filename, sectors, backing_file, backing_fmt, flags,
                          cluster_size, prealloc, options, version);
 }
@@ -1485,10 +1543,12 @@
         return ret;
     }
 
-    ret = qcow2_cache_flush(bs, s->refcount_block_cache);
-    if (ret < 0) {
-        qemu_co_mutex_unlock(&s->lock);
-        return ret;
+    if (qcow2_need_accurate_refcounts(s)) {
+        ret = qcow2_cache_flush(bs, s->refcount_block_cache);
+        if (ret < 0) {
+            qemu_co_mutex_unlock(&s->lock);
+            return ret;
+        }
     }
     qemu_co_mutex_unlock(&s->lock);
 
@@ -1603,6 +1663,11 @@
         .type = OPT_STRING,
         .help = "Preallocation mode (allowed values: off, metadata)"
     },
+    {
+        .name = BLOCK_OPT_LAZY_REFCOUNTS,
+        .type = OPT_FLAG,
+        .help = "Postpone refcount updates",
+    },
     { NULL }
 };