Series comparison

The following changes since commit 30aa19446d82358a30eac3b556b4d6641e00b7c1:

Merge remote-tracking branch 'remotes/cschoenebeck/tags/pull-9p-20200812' into staging (2020-08-24 16:39:53 +0100)

are available in the Git repository at:

https://github.com/XanClic/qemu.git tags/pull-block-2020-08-25

for you to fetch changes up to c576fd97d4ca77b5a1a27728df11a61083dbfa98:

iotests: Add tests for qcow2 images with extended L2 entries (2020-08-25 10:20:18 +0200)

----------------------------------------------------------------
Block patches:
- qcow2 subclusters (extended L2 entries)

----------------------------------------------------------------
Alberto Garcia (34):
  qcow2: Make Qcow2AioTask store the full host offset
  qcow2: Convert qcow2_get_cluster_offset() into qcow2_get_host_offset()
  qcow2: Add calculate_l2_meta()
  qcow2: Split cluster_needs_cow() out of count_cow_clusters()
  qcow2: Process QCOW2_CLUSTER_ZERO_ALLOC clusters in handle_copied()
  qcow2: Add get_l2_entry() and set_l2_entry()
  qcow2: Document the Extended L2 Entries feature
  qcow2: Add dummy has_subclusters() function
  qcow2: Add subcluster-related fields to BDRVQcow2State
  qcow2: Add offset_to_sc_index()
  qcow2: Add offset_into_subcluster() and size_to_subclusters()
  qcow2: Add l2_entry_size()
  qcow2: Update get/set_l2_entry() and add get/set_l2_bitmap()
  qcow2: Add QCow2SubclusterType and qcow2_get_subcluster_type()
  qcow2: Add qcow2_get_subcluster_range_type()
  qcow2: Add qcow2_cluster_is_allocated()
  qcow2: Add cluster type parameter to qcow2_get_host_offset()
  qcow2: Replace QCOW2_CLUSTER_* with QCOW2_SUBCLUSTER_*
  qcow2: Handle QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC
  qcow2: Add subcluster support to calculate_l2_meta()
  qcow2: Add subcluster support to qcow2_get_host_offset()
  qcow2: Add subcluster support to zero_in_l2_slice()
  qcow2: Add subcluster support to discard_in_l2_slice()
  qcow2: Add subcluster support to check_refcounts_l2()
  qcow2: Update L2 bitmap in qcow2_alloc_cluster_link_l2()
  qcow2: Clear the L2 bitmap when allocating a compressed cluster
  qcow2: Add subcluster support to handle_alloc_space()
  qcow2: Add subcluster support to qcow2_co_pwrite_zeroes()
  qcow2: Add subcluster support to qcow2_measure()
  qcow2: Add prealloc field to QCowL2Meta
  qcow2: Add the 'extended_l2' option and the QCOW2_INCOMPAT_EXTL2 bit
  qcow2: Allow preallocation and backing files if extended_l2 is set
  qcow2: Assert that expand_zero_clusters_in_l1() does not support
    subclusters
  iotests: Add tests for qcow2 images with extended L2 entries

-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

The file_cluster_offset field of Qcow2AioTask stores a cluster-aligned
host offset. In practice this is not very useful because all users(*)
of this structure need the final host offset into the cluster, which
they calculate using

host_offset = file_cluster_offset + offset_into_cluster(s, offset)

There is no reason why Qcow2AioTask cannot store host_offset directly
and that is what this patch does.

(*) compressed clusters are the exception: in this case what
    file_cluster_offset was storing was the full compressed cluster
    descriptor (offset + size). This does not change with this patch
    but it is documented now.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <07c4b15c644dcf06c9459f98846ac1c4ea96e26f.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.c      | 69 ++++++++++++++++++++++------------------------
 block/trace-events |  2 +-
 2 files changed, 34 insertions(+), 37 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ typedef struct {
 
 static int coroutine_fn
 qcow2_co_preadv_compressed(BlockDriverState *bs,
-                           uint64_t file_cluster_offset,
+                           uint64_t cluster_descriptor,
                            uint64_t offset,
                            uint64_t bytes,
                            QEMUIOVector *qiov,
@@ -XXX,XX +XXX,XX @@ out:
 
 static coroutine_fn int
 qcow2_co_preadv_encrypted(BlockDriverState *bs,
-                           uint64_t file_cluster_offset,
+                           uint64_t host_offset,
                            uint64_t offset,
                            uint64_t bytes,
                            QEMUIOVector *qiov,
@@ -XXX,XX +XXX,XX @@ qcow2_co_preadv_encrypted(BlockDriverState *bs,
     }
 
     BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
-    ret = bdrv_co_pread(s->data_file,
-                        file_cluster_offset + offset_into_cluster(s, offset),
-                        bytes, buf, 0);
+    ret = bdrv_co_pread(s->data_file, host_offset, bytes, buf, 0);
     if (ret < 0) {
         goto fail;
     }
 
-    if (qcow2_co_decrypt(bs,
-                         file_cluster_offset + offset_into_cluster(s, offset),
-                         offset, buf, bytes) < 0)
+    if (qcow2_co_decrypt(bs, host_offset, offset, buf, bytes) < 0)
     {
         ret = -EIO;
         goto fail;
@@ -XXX,XX +XXX,XX @@ typedef struct Qcow2AioTask {
 
     BlockDriverState *bs;
     QCow2ClusterType cluster_type; /* only for read */
-    uint64_t file_cluster_offset;
+    uint64_t host_offset; /* or full descriptor in compressed clusters */
     uint64_t offset;
     uint64_t bytes;
     QEMUIOVector *qiov;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
                                        AioTaskPool *pool,
                                        AioTaskFunc func,
                                        QCow2ClusterType cluster_type,
-                                       uint64_t file_cluster_offset,
+                                       uint64_t host_offset,
                                        uint64_t offset,
                                        uint64_t bytes,
                                        QEMUIOVector *qiov,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
         .bs = bs,
         .cluster_type = cluster_type,
         .qiov = qiov,
-        .file_cluster_offset = file_cluster_offset,
+        .host_offset = host_offset,
         .offset = offset,
         .bytes = bytes,
         .qiov_offset = qiov_offset,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
 
     trace_qcow2_add_task(qemu_coroutine_self(), bs, pool,
                          func == qcow2_co_preadv_task_entry ? "read" : "write",
-                         cluster_type, file_cluster_offset, offset, bytes,
+                         cluster_type, host_offset, offset, bytes,
                          qiov, qiov_offset);
 
     if (!pool) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
 
 static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs,
                                              QCow2ClusterType cluster_type,
-                                             uint64_t file_cluster_offset,
+                                             uint64_t host_offset,
                                              uint64_t offset, uint64_t bytes,
                                              QEMUIOVector *qiov,
                                              size_t qiov_offset)
 {
     BDRVQcow2State *s = bs->opaque;
-    int offset_in_cluster = offset_into_cluster(s, offset);
 
     switch (cluster_type) {
     case QCOW2_CLUSTER_ZERO_PLAIN:
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs,
                                    qiov, qiov_offset, 0);
 
     case QCOW2_CLUSTER_COMPRESSED:
-        return qcow2_co_preadv_compressed(bs, file_cluster_offset,
+        return qcow2_co_preadv_compressed(bs, host_offset,
                                           offset, bytes, qiov, qiov_offset);
 
     case QCOW2_CLUSTER_NORMAL:
-        assert(offset_into_cluster(s, file_cluster_offset) == 0);
         if (bs->encrypted) {
-            return qcow2_co_preadv_encrypted(bs, file_cluster_offset,
+            return qcow2_co_preadv_encrypted(bs, host_offset,
                                              offset, bytes, qiov, qiov_offset);
         }
 
         BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
-        return bdrv_co_preadv_part(s->data_file,
-                                   file_cluster_offset + offset_in_cluster,
+        return bdrv_co_preadv_part(s->data_file, host_offset,
                                    bytes, qiov, qiov_offset, 0);
 
     default:
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task_entry(AioTask *task)
 
     assert(!t->l2meta);
 
-    return qcow2_co_preadv_task(t->bs, t->cluster_type, t->file_cluster_offset,
+    return qcow2_co_preadv_task(t->bs, t->cluster_type, t->host_offset,
                                 t->offset, t->bytes, t->qiov, t->qiov_offset);
 }
 
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
         {
             qemu_iovec_memset(qiov, qiov_offset, 0, cur_bytes);
         } else {
+            /*
+             * For compressed clusters the variable cluster_offset
+             * does not actually store the offset but the full
+             * descriptor. We need to leave it unchanged because
+             * that's what qcow2_co_preadv_compressed() expects.
+             */
+            uint64_t host_offset = (ret == QCOW2_CLUSTER_COMPRESSED) ?
+                cluster_offset :
+                cluster_offset + offset_into_cluster(s, offset);
             if (!aio && cur_bytes != bytes) {
                 aio = aio_task_pool_new(QCOW2_MAX_WORKERS);
             }
             ret = qcow2_add_task(bs, aio, qcow2_co_preadv_task_entry, ret,
-                                 cluster_offset, offset, cur_bytes,
+                                 host_offset, offset, cur_bytes,
                                  qiov, qiov_offset, NULL);
             if (ret < 0) {
                 goto out;
@@ -XXX,XX +XXX,XX @@ static int handle_alloc_space(BlockDriverState *bs, QCowL2Meta *l2meta)
  *           not use it somehow after qcow2_co_pwritev_task() call
  */
 static coroutine_fn int qcow2_co_pwritev_task(BlockDriverState *bs,
-                                              uint64_t file_cluster_offset,
+                                              uint64_t host_offset,
                                               uint64_t offset, uint64_t bytes,
                                               QEMUIOVector *qiov,
                                               uint64_t qiov_offset,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_task(BlockDriverState *bs,
     int ret;
     BDRVQcow2State *s = bs->opaque;
     void *crypt_buf = NULL;
-    int offset_in_cluster = offset_into_cluster(s, offset);
     QEMUIOVector encrypted_qiov;
 
     if (bs->encrypted) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_task(BlockDriverState *bs,
         }
         qemu_iovec_to_buf(qiov, qiov_offset, crypt_buf, bytes);
 
-        if (qcow2_co_encrypt(bs, file_cluster_offset + offset_in_cluster,
-                             offset, crypt_buf, bytes) < 0)
-        {
+        if (qcow2_co_encrypt(bs, host_offset, offset, crypt_buf, bytes) < 0) {
             ret = -EIO;
             goto out_unlocked;
         }
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_task(BlockDriverState *bs,
      */
     if (!merge_cow(offset, bytes, qiov, qiov_offset, l2meta)) {
         BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO);
-        trace_qcow2_writev_data(qemu_coroutine_self(),
-                                file_cluster_offset + offset_in_cluster);
-        ret = bdrv_co_pwritev_part(s->data_file,
-                                   file_cluster_offset + offset_in_cluster,
+        trace_qcow2_writev_data(qemu_coroutine_self(), host_offset);
+        ret = bdrv_co_pwritev_part(s->data_file, host_offset,
                                    bytes, qiov, qiov_offset, 0);
         if (ret < 0) {
             goto out_unlocked;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_task_entry(AioTask *task)
 
     assert(!t->cluster_type);
 
-    return qcow2_co_pwritev_task(t->bs, t->file_cluster_offset,
+    return qcow2_co_pwritev_task(t->bs, t->host_offset,
                                  t->offset, t->bytes, t->qiov, t->qiov_offset,
                                  t->l2meta);
 }
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_part(
             aio = aio_task_pool_new(QCOW2_MAX_WORKERS);
         }
         ret = qcow2_add_task(bs, aio, qcow2_co_pwritev_task_entry, 0,
-                             cluster_offset, offset, cur_bytes,
-                             qiov, qiov_offset, l2meta);
+                             cluster_offset + offset_in_cluster, offset,
+                             cur_bytes, qiov, qiov_offset, l2meta);
         l2meta = NULL; /* l2meta is consumed by qcow2_co_pwritev_task() */
         if (ret < 0) {
             goto fail_nometa;
@@ -XXX,XX +XXX,XX @@ qcow2_co_pwritev_compressed_part(BlockDriverState *bs,
 
 static int coroutine_fn
 qcow2_co_preadv_compressed(BlockDriverState *bs,
-                           uint64_t file_cluster_offset,
+                           uint64_t cluster_descriptor,
                            uint64_t offset,
                            uint64_t bytes,
                            QEMUIOVector *qiov,
@@ -XXX,XX +XXX,XX @@ qcow2_co_preadv_compressed(BlockDriverState *bs,
     uint8_t *buf, *out_buf;
     int offset_in_cluster = offset_into_cluster(s, offset);
 
-    coffset = file_cluster_offset & s->cluster_offset_mask;
-    nb_csectors = ((file_cluster_offset >> s->csize_shift) & s->csize_mask) + 1;
+    coffset = cluster_descriptor & s->cluster_offset_mask;
+    nb_csectors = ((cluster_descriptor >> s->csize_shift) & s->csize_mask) + 1;
     csize = nb_csectors * QCOW2_COMPRESSED_SECTOR_SIZE -
         (coffset & ~QCOW2_COMPRESSED_SECTOR_MASK);
 
diff --git a/block/trace-events b/block/trace-events
index XXXXXXX..XXXXXXX 100644
--- a/block/trace-events
+++ b/block/trace-events
@@ -XXX,XX +XXX,XX @@ luring_io_uring_submit(void *s, int ret) "LuringState %p ret %d"
 luring_resubmit_short_read(void *s, void *luringcb, int nread) "LuringState %p luringcb %p nread %d"
 
 # qcow2.c
-qcow2_add_task(void *co, void *bs, void *pool, const char *action, int cluster_type, uint64_t file_cluster_offset, uint64_t offset, uint64_t bytes, void *qiov, size_t qiov_offset) "co %p bs %p pool %p: %s: cluster_type %d file_cluster_offset %" PRIu64 " offset %" PRIu64 " bytes %" PRIu64 " qiov %p qiov_offset %zu"
+qcow2_add_task(void *co, void *bs, void *pool, const char *action, int cluster_type, uint64_t host_offset, uint64_t offset, uint64_t bytes, void *qiov, size_t qiov_offset) "co %p bs %p pool %p: %s: cluster_type %d file_cluster_offset %" PRIu64 " offset %" PRIu64 " bytes %" PRIu64 " qiov %p qiov_offset %zu"
 qcow2_writev_start_req(void *co, int64_t offset, int bytes) "co %p offset 0x%" PRIx64 " bytes %d"
 qcow2_writev_done_req(void *co, int ret) "co %p ret %d"
 qcow2_writev_start_part(void *co) "co %p"
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

qcow2_get_cluster_offset() takes an (unaligned) guest offset and
returns the (aligned) offset of the corresponding cluster in the qcow2
image.

In practice none of the callers need to know where the cluster starts
so this patch makes the function calculate and return the final host
offset directly. The function is also renamed accordingly.

There is a pre-existing exception with compressed clusters: in this
case the function returns the complete cluster descriptor (containing
the offset and size of the compressed data). This does not change with
this patch but it is now documented.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <ffae6cdc5ca8950e8280ac0f696dcc376cb07095.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h         |  4 ++--
 block/qcow2-cluster.c | 41 +++++++++++++++++++++++------------------
 block/qcow2.c         | 24 +++++++-----------------
 3 files changed, 32 insertions(+), 37 deletions(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index);
 int qcow2_encrypt_sectors(BDRVQcow2State *s, int64_t sector_num,
                           uint8_t *buf, int nb_sectors, bool enc, Error **errp);
 
-int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
-                             unsigned int *bytes, uint64_t *cluster_offset);
+int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
+                          unsigned int *bytes, uint64_t *host_offset);
 int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
                                unsigned int *bytes, uint64_t *host_offset,
                                QCowL2Meta **m);
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn do_perform_cow_write(BlockDriverState *bs,
 
 
 /*
- * get_cluster_offset
+ * get_host_offset
  *
- * For a given offset of the virtual disk, find the cluster type and offset in
- * the qcow2 file. The offset is stored in *cluster_offset.
+ * For a given offset of the virtual disk find the equivalent host
+ * offset in the qcow2 file and store it in *host_offset. Neither
+ * offset needs to be aligned to a cluster boundary.
+ *
+ * If the cluster is unallocated then *host_offset will be 0.
+ * If the cluster is compressed then *host_offset will contain the
+ * complete compressed cluster descriptor.
  *
  * On entry, *bytes is the maximum number of contiguous bytes starting at
  * offset that we are interested in.
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn do_perform_cow_write(BlockDriverState *bs,
  * Returns the cluster type (QCOW2_CLUSTER_*) on success, -errno in error
  * cases.
  */
-int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
-                             unsigned int *bytes, uint64_t *cluster_offset)
+int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
+                          unsigned int *bytes, uint64_t *host_offset)
 {
     BDRVQcow2State *s = bs->opaque;
     unsigned int l2_index;
-    uint64_t l1_index, l2_offset, *l2_slice;
+    uint64_t l1_index, l2_offset, *l2_slice, l2_entry;
     int c;
     unsigned int offset_in_cluster;
     uint64_t bytes_available, bytes_needed, nb_clusters;
@@ -XXX,XX +XXX,XX @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
         bytes_needed = bytes_available;
     }
 
-    *cluster_offset = 0;
+    *host_offset = 0;
 
     /* seek to the l2 offset in the l1 table */
 
@@ -XXX,XX +XXX,XX @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
     /* find the cluster offset for the given disk offset */
 
     l2_index = offset_to_l2_slice_index(s, offset);
-    *cluster_offset = be64_to_cpu(l2_slice[l2_index]);
+    l2_entry = be64_to_cpu(l2_slice[l2_index]);
 
     nb_clusters = size_to_clusters(s, bytes_needed);
     /* bytes_needed <= *bytes + offset_in_cluster, both of which are unsigned
@@ -XXX,XX +XXX,XX @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
      * true */
     assert(nb_clusters <= INT_MAX);
 
-    type = qcow2_get_cluster_type(bs, *cluster_offset);
+    type = qcow2_get_cluster_type(bs, l2_entry);
     if (s->qcow_version < 3 && (type == QCOW2_CLUSTER_ZERO_PLAIN ||
                                 type == QCOW2_CLUSTER_ZERO_ALLOC)) {
         qcow2_signal_corruption(bs, true, -1, -1, "Zero cluster entry found"
@@ -XXX,XX +XXX,XX @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
         }
         /* Compressed clusters can only be processed one by one */
         c = 1;
-        *cluster_offset &= L2E_COMPRESSED_OFFSET_SIZE_MASK;
+        *host_offset = l2_entry & L2E_COMPRESSED_OFFSET_SIZE_MASK;
         break;
     case QCOW2_CLUSTER_ZERO_PLAIN:
     case QCOW2_CLUSTER_UNALLOCATED:
         /* how many empty clusters ? */
         c = count_contiguous_clusters_unallocated(bs, nb_clusters,
                                                   &l2_slice[l2_index], type);
-        *cluster_offset = 0;
         break;
     case QCOW2_CLUSTER_ZERO_ALLOC:
-    case QCOW2_CLUSTER_NORMAL:
+    case QCOW2_CLUSTER_NORMAL: {
+        uint64_t host_cluster_offset = l2_entry & L2E_OFFSET_MASK;
+        *host_offset = host_cluster_offset + offset_in_cluster;
         /* how many allocated clusters ? */
         c = count_contiguous_clusters(bs, nb_clusters, s->cluster_size,
                                       &l2_slice[l2_index], QCOW_OFLAG_ZERO);
-        *cluster_offset &= L2E_OFFSET_MASK;
-        if (offset_into_cluster(s, *cluster_offset)) {
+        if (offset_into_cluster(s, host_cluster_offset)) {
             qcow2_signal_corruption(bs, true, -1, -1,
                                     "Cluster allocation offset %#"
                                     PRIx64 " unaligned (L2 offset: %#" PRIx64
-                                    ", L2 index: %#x)", *cluster_offset,
+                                    ", L2 index: %#x)", host_cluster_offset,
                                     l2_offset, l2_index);
             ret = -EIO;
             goto fail;
         }
-        if (has_data_file(bs) && *cluster_offset != offset - offset_in_cluster)
-        {
+        if (has_data_file(bs) && *host_offset != offset) {
             qcow2_signal_corruption(bs, true, -1, -1,
                                     "External data file host cluster offset %#"
                                     PRIx64 " does not match guest cluster "
                                     "offset: %#" PRIx64
-                                    ", L2 index: %#x)", *cluster_offset,
+                                    ", L2 index: %#x)", host_cluster_offset,
                                     offset - offset_in_cluster, l2_index);
             ret = -EIO;
             goto fail;
         }
         break;
+    }
     default:
         abort();
     }
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
                                               BlockDriverState **file)
 {
     BDRVQcow2State *s = bs->opaque;
-    uint64_t cluster_offset;
+    uint64_t host_offset;
     unsigned int bytes;
     int ret, status = 0;
 
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
     }
 
     bytes = MIN(INT_MAX, count);
-    ret = qcow2_get_cluster_offset(bs, offset, &bytes, &cluster_offset);
+    ret = qcow2_get_host_offset(bs, offset, &bytes, &host_offset);
     qemu_co_mutex_unlock(&s->lock);
     if (ret < 0) {
         return ret;
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
 
     if ((ret == QCOW2_CLUSTER_NORMAL || ret == QCOW2_CLUSTER_ZERO_ALLOC) &&
         !s->crypto) {
-        *map = cluster_offset | offset_into_cluster(s, offset);
+        *map = host_offset;
         *file = s->data_file->bs;
         status |= BDRV_BLOCK_OFFSET_VALID;
     }
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
     BDRVQcow2State *s = bs->opaque;
     int ret = 0;
     unsigned int cur_bytes; /* number of bytes in current iteration */
-    uint64_t cluster_offset = 0;
+    uint64_t host_offset = 0;
     AioTaskPool *aio = NULL;
 
     while (bytes != 0 && aio_task_pool_status(aio) == 0) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
         }
 
         qemu_co_mutex_lock(&s->lock);
-        ret = qcow2_get_cluster_offset(bs, offset, &cur_bytes, &cluster_offset);
+        ret = qcow2_get_host_offset(bs, offset, &cur_bytes, &host_offset);
         qemu_co_mutex_unlock(&s->lock);
         if (ret < 0) {
             goto out;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
         {
             qemu_iovec_memset(qiov, qiov_offset, 0, cur_bytes);
         } else {
-            /*
-             * For compressed clusters the variable cluster_offset
-             * does not actually store the offset but the full
-             * descriptor. We need to leave it unchanged because
-             * that's what qcow2_co_preadv_compressed() expects.
-             */
-            uint64_t host_offset = (ret == QCOW2_CLUSTER_COMPRESSED) ?
-                cluster_offset :
-                cluster_offset + offset_into_cluster(s, offset);
             if (!aio && cur_bytes != bytes) {
                 aio = aio_task_pool_new(QCOW2_MAX_WORKERS);
             }
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
         offset = QEMU_ALIGN_DOWN(offset, s->cluster_size);
         bytes = s->cluster_size;
         nr = s->cluster_size;
-        ret = qcow2_get_cluster_offset(bs, offset, &nr, &off);
+        ret = qcow2_get_host_offset(bs, offset, &nr, &off);
         if (ret != QCOW2_CLUSTER_UNALLOCATED &&
             ret != QCOW2_CLUSTER_ZERO_PLAIN &&
             ret != QCOW2_CLUSTER_ZERO_ALLOC) {
@@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_from(BlockDriverState *bs,
         cur_bytes = MIN(bytes, INT_MAX);
         cur_write_flags = write_flags;
 
-        ret = qcow2_get_cluster_offset(bs, src_offset, &cur_bytes, &copy_offset);
+        ret = qcow2_get_host_offset(bs, src_offset, &cur_bytes, &copy_offset);
         if (ret < 0) {
             goto out;
         }
@@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_from(BlockDriverState *bs,
 
         case QCOW2_CLUSTER_NORMAL:
             child = s->data_file;
-            copy_offset += offset_into_cluster(s, src_offset);
             break;
 
         default:
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

handle_alloc() creates a QCowL2Meta structure in order to update the
image metadata and perform the necessary copy-on-write operations.

This patch moves that code to a separate function so it can be used
from other places.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <e5bc4a648dac31972bfa7a0e554be8064be78799.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 77 +++++++++++++++++++++++++++++--------------
 1 file changed, 53 insertions(+), 24 deletions(-)

From: Alberto Garcia <berto@igalia.com>

We are going to need it in other places.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <65e5d9627ca2ebe7e62deaeddf60949c33067d9d.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 34 +++++++++++++++++++---------------
 1 file changed, 19 insertions(+), 15 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static void calculate_l2_meta(BlockDriverState *bs,
     QLIST_INSERT_HEAD(&s->cluster_allocs, *m, next_in_flight);
 }
 
+/* Returns true if writing to a cluster requires COW */
+static bool cluster_needs_cow(BlockDriverState *bs, uint64_t l2_entry)
+{
+    switch (qcow2_get_cluster_type(bs, l2_entry)) {
+    case QCOW2_CLUSTER_NORMAL:
+        if (l2_entry & QCOW_OFLAG_COPIED) {
+            return false;
+        }
+    case QCOW2_CLUSTER_UNALLOCATED:
+    case QCOW2_CLUSTER_COMPRESSED:
+    case QCOW2_CLUSTER_ZERO_PLAIN:
+    case QCOW2_CLUSTER_ZERO_ALLOC:
+        return true;
+    default:
+        abort();
+    }
+}
+
 /*
  * Returns the number of contiguous clusters that can be used for an allocating
  * write, but require COW to be performed (this includes yet unallocated space,
@@ -XXX,XX +XXX,XX @@ static int count_cow_clusters(BlockDriverState *bs, int nb_clusters,
 
     for (i = 0; i < nb_clusters; i++) {
         uint64_t l2_entry = be64_to_cpu(l2_slice[l2_index + i]);
-        QCow2ClusterType cluster_type = qcow2_get_cluster_type(bs, l2_entry);
-
-        switch(cluster_type) {
-        case QCOW2_CLUSTER_NORMAL:
-            if (l2_entry & QCOW_OFLAG_COPIED) {
-                goto out;
-            }
+        if (!cluster_needs_cow(bs, l2_entry)) {
             break;
-        case QCOW2_CLUSTER_UNALLOCATED:
-        case QCOW2_CLUSTER_COMPRESSED:
-        case QCOW2_CLUSTER_ZERO_PLAIN:
-        case QCOW2_CLUSTER_ZERO_ALLOC:
-            break;
-        default:
-            abort();
         }
     }
 
-out:
     assert(i <= nb_clusters);
     return i;
 }
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

When writing to a qcow2 file there are two functions that take a
virtual offset and return a host offset, possibly allocating new
clusters if necessary:

- handle_copied() looks for normal data clusters that are already
     allocated and have a reference count of 1. In those clusters we
     can simply write the data and there is no need to perform any
     copy-on-write.

- handle_alloc() looks for clusters that do need copy-on-write,
     either because they haven't been allocated yet, because their
     reference count is != 1 or because they are ZERO_ALLOC clusters.

The ZERO_ALLOC case is a bit special because those are clusters that
are already allocated and they could perfectly be dealt with in
handle_copied() (as long as copy-on-write is performed when required).

In fact, there is extra code specifically for them in handle_alloc()
that tries to reuse the existing allocation if possible and frees them
otherwise.

This patch changes the handling of ZERO_ALLOC clusters so the
semantics of these two functions are now like this:

- handle_copied() looks for clusters that are already allocated and
     which we can overwrite (NORMAL and ZERO_ALLOC clusters with a
     reference count of 1).

- handle_alloc() looks for clusters for which we need a new
     allocation (all other cases).

One important difference after this change is that clusters found
in handle_copied() may now require copy-on-write, but this will be
necessary anyway once we add support for subclusters.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <eb17fc938f6be7be2e8d8ff42763d2c19241f866.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 252 +++++++++++++++++++++++-------------------
 1 file changed, 139 insertions(+), 113 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m)
 
 /*
  * For a given write request, create a new QCowL2Meta structure, add
- * it to @m and the BDRVQcow2State.cluster_allocs list.
+ * it to @m and the BDRVQcow2State.cluster_allocs list. If the write
+ * request does not need copy-on-write or changes to the L2 metadata
+ * then this function does nothing.
  *
  * @host_cluster_offset points to the beginning of the first cluster.
  *
  * @guest_offset and @bytes indicate the offset and length of the
  * request.
  *
+ * @l2_slice contains the L2 entries of all clusters involved in this
+ * write request.
+ *
  * If @keep_old is true it means that the clusters were already
  * allocated and will be overwritten. If false then the clusters are
  * new and we have to decrease the reference count of the old ones.
@@ -XXX,XX +XXX,XX @@ void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m)
 static void calculate_l2_meta(BlockDriverState *bs,
                               uint64_t host_cluster_offset,
                               uint64_t guest_offset, unsigned bytes,
-                              QCowL2Meta **m, bool keep_old)
+                              uint64_t *l2_slice, QCowL2Meta **m, bool keep_old)
 {
     BDRVQcow2State *s = bs->opaque;
-    unsigned cow_start_from = 0;
+    int l2_index = offset_to_l2_slice_index(s, guest_offset);
+    uint64_t l2_entry;
+    unsigned cow_start_from, cow_end_to;
     unsigned cow_start_to = offset_into_cluster(s, guest_offset);
     unsigned cow_end_from = cow_start_to + bytes;
-    unsigned cow_end_to = ROUND_UP(cow_end_from, s->cluster_size);
     unsigned nb_clusters = size_to_clusters(s, cow_end_from);
     QCowL2Meta *old_m = *m;
+    QCow2ClusterType type;
+
+    assert(nb_clusters <= s->l2_slice_size - l2_index);
+
+    /* Return if there's no COW (all clusters are normal and we keep them) */
+    if (keep_old) {
+        int i;
+        for (i = 0; i < nb_clusters; i++) {
+            l2_entry = be64_to_cpu(l2_slice[l2_index + i]);
+            if (qcow2_get_cluster_type(bs, l2_entry) != QCOW2_CLUSTER_NORMAL) {
+                break;
+            }
+        }
+        if (i == nb_clusters) {
+            return;
+        }
+    }
+
+    /* Get the L2 entry of the first cluster */
+    l2_entry = be64_to_cpu(l2_slice[l2_index]);
+    type = qcow2_get_cluster_type(bs, l2_entry);
+
+    if (type == QCOW2_CLUSTER_NORMAL && keep_old) {
+        cow_start_from = cow_start_to;
+    } else {
+        cow_start_from = 0;
+    }
+
+    /* Get the L2 entry of the last cluster */
+    l2_entry = be64_to_cpu(l2_slice[l2_index + nb_clusters - 1]);
+    type = qcow2_get_cluster_type(bs, l2_entry);
+
+    if (type == QCOW2_CLUSTER_NORMAL && keep_old) {
+        cow_end_to = cow_end_from;
+    } else {
+        cow_end_to = ROUND_UP(cow_end_from, s->cluster_size);
+    }
 
     *m = g_malloc0(sizeof(**m));
     **m = (QCowL2Meta) {
@@ -XXX,XX +XXX,XX @@ static void calculate_l2_meta(BlockDriverState *bs,
     QLIST_INSERT_HEAD(&s->cluster_allocs, *m, next_in_flight);
 }
 
-/* Returns true if writing to a cluster requires COW */
-static bool cluster_needs_cow(BlockDriverState *bs, uint64_t l2_entry)
+/*
+ * Returns true if writing to the cluster pointed to by @l2_entry
+ * requires a new allocation (that is, if the cluster is unallocated
+ * or has refcount > 1 and therefore cannot be written in-place).
+ */
+static bool cluster_needs_new_alloc(BlockDriverState *bs, uint64_t l2_entry)
 {
     switch (qcow2_get_cluster_type(bs, l2_entry)) {
     case QCOW2_CLUSTER_NORMAL:
+    case QCOW2_CLUSTER_ZERO_ALLOC:
         if (l2_entry & QCOW_OFLAG_COPIED) {
             return false;
         }
     case QCOW2_CLUSTER_UNALLOCATED:
     case QCOW2_CLUSTER_COMPRESSED:
     case QCOW2_CLUSTER_ZERO_PLAIN:
-    case QCOW2_CLUSTER_ZERO_ALLOC:
         return true;
     default:
         abort();
@@ -XXX,XX +XXX,XX @@ static bool cluster_needs_cow(BlockDriverState *bs, uint64_t l2_entry)
 }
 
 /*
- * Returns the number of contiguous clusters that can be used for an allocating
- * write, but require COW to be performed (this includes yet unallocated space,
- * which must copy from the backing file)
+ * Returns the number of contiguous clusters that can be written to
+ * using one single write request, starting from @l2_index.
+ * At most @nb_clusters are checked.
+ *
+ * If @new_alloc is true this counts clusters that are either
+ * unallocated, or allocated but with refcount > 1 (so they need to be
+ * newly allocated and COWed).
+ *
+ * If @new_alloc is false this counts clusters that are already
+ * allocated and can be overwritten in-place (this includes clusters
+ * of type QCOW2_CLUSTER_ZERO_ALLOC).
  */
-static int count_cow_clusters(BlockDriverState *bs, int nb_clusters,
-    uint64_t *l2_slice, int l2_index)
+static int count_single_write_clusters(BlockDriverState *bs, int nb_clusters,
+                                       uint64_t *l2_slice, int l2_index,
+                                       bool new_alloc)
 {
+    BDRVQcow2State *s = bs->opaque;
+    uint64_t l2_entry = be64_to_cpu(l2_slice[l2_index]);
+    uint64_t expected_offset = l2_entry & L2E_OFFSET_MASK;
     int i;
 
     for (i = 0; i < nb_clusters; i++) {
-        uint64_t l2_entry = be64_to_cpu(l2_slice[l2_index + i]);
-        if (!cluster_needs_cow(bs, l2_entry)) {
+        l2_entry = be64_to_cpu(l2_slice[l2_index + i]);
+        if (cluster_needs_new_alloc(bs, l2_entry) != new_alloc) {
             break;
         }
+        if (!new_alloc) {
+            if (expected_offset != (l2_entry & L2E_OFFSET_MASK)) {
+                break;
+            }
+            expected_offset += s->cluster_size;
+        }
     }
 
     assert(i <= nb_clusters);
@@ -XXX,XX +XXX,XX @@ static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
 }
 
 /*
- * Checks how many already allocated clusters that don't require a copy on
- * write there are at the given guest_offset (up to *bytes). If *host_offset is
- * not INV_OFFSET, only physically contiguous clusters beginning at this host
- * offset are counted.
+ * Checks how many already allocated clusters that don't require a new
+ * allocation there are at the given guest_offset (up to *bytes).
+ * If *host_offset is not INV_OFFSET, only physically contiguous clusters
+ * beginning at this host offset are counted.
  *
  * Note that guest_offset may not be cluster aligned. In this case, the
  * returned *host_offset points to exact byte referenced by guest_offset and
@@ -XXX,XX +XXX,XX @@ static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
  * Returns:
  *   0:     if no allocated clusters are available at the given offset.
  *          *bytes is normally unchanged. It is set to 0 if the cluster
- *          is allocated and doesn't need COW, but doesn't have the right
- *          physical offset.
+ *          is allocated and can be overwritten in-place but doesn't have
+ *          the right physical offset.
  *
- *   1:     if allocated clusters that don't require a COW are available at
- *          the requested offset. *bytes may have decreased and describes
- *          the length of the area that can be written to.
+ *   1:     if allocated clusters that can be overwritten in place are
+ *          available at the requested offset. *bytes may have decreased
+ *          and describes the length of the area that can be written to.
  *
  *  -errno: in error cases
  */
@@ -XXX,XX +XXX,XX @@ static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
 {
     BDRVQcow2State *s = bs->opaque;
     int l2_index;
-    uint64_t cluster_offset;
+    uint64_t l2_entry, cluster_offset;
     uint64_t *l2_slice;
     uint64_t nb_clusters;
     unsigned int keep_clusters;
@@ -XXX,XX +XXX,XX @@ static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
 
     l2_index = offset_to_l2_slice_index(s, guest_offset);
     nb_clusters = MIN(nb_clusters, s->l2_slice_size - l2_index);
-    assert(nb_clusters <= INT_MAX);
+    /* Limit total byte count to BDRV_REQUEST_MAX_BYTES */
+    nb_clusters = MIN(nb_clusters, BDRV_REQUEST_MAX_BYTES >> s->cluster_bits);
 
     /* Find L2 entry for the first involved cluster */
     ret = get_cluster_table(bs, guest_offset, &l2_slice, &l2_index);
@@ -XXX,XX +XXX,XX @@ static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
         return ret;
     }
 
-    cluster_offset = be64_to_cpu(l2_slice[l2_index]);
-
-    /* Check how many clusters are already allocated and don't need COW */
-    if (qcow2_get_cluster_type(bs, cluster_offset) == QCOW2_CLUSTER_NORMAL
-        && (cluster_offset & QCOW_OFLAG_COPIED))
-    {
-        /* If a specific host_offset is required, check it */
-        bool offset_matches =
-            (cluster_offset & L2E_OFFSET_MASK) == *host_offset;
-
-        if (offset_into_cluster(s, cluster_offset & L2E_OFFSET_MASK)) {
-            qcow2_signal_corruption(bs, true, -1, -1, "Data cluster offset "
-                                    "%#llx unaligned (guest offset: %#" PRIx64
-                                    ")", cluster_offset & L2E_OFFSET_MASK,
-                                    guest_offset);
+    l2_entry = be64_to_cpu(l2_slice[l2_index]);
+    cluster_offset = l2_entry & L2E_OFFSET_MASK;
+
+    if (!cluster_needs_new_alloc(bs, l2_entry)) {
+        if (offset_into_cluster(s, cluster_offset)) {
+            qcow2_signal_corruption(bs, true, -1, -1, "%s cluster offset "
+                                    "%#" PRIx64 " unaligned (guest offset: %#"
+                                    PRIx64 ")", l2_entry & QCOW_OFLAG_ZERO ?
+                                    "Preallocated zero" : "Data",
+                                    cluster_offset, guest_offset);
             ret = -EIO;
             goto out;
         }
 
-        if (*host_offset != INV_OFFSET && !offset_matches) {
+        /* If a specific host_offset is required, check it */
+        if (*host_offset != INV_OFFSET && cluster_offset != *host_offset) {
             *bytes = 0;
             ret = 0;
             goto out;
         }
 
         /* We keep all QCOW_OFLAG_COPIED clusters */
-        keep_clusters =
-            count_contiguous_clusters(bs, nb_clusters, s->cluster_size,
-                                      &l2_slice[l2_index],
-                                      QCOW_OFLAG_COPIED | QCOW_OFLAG_ZERO);
+        keep_clusters = count_single_write_clusters(bs, nb_clusters, l2_slice,
+                                                    l2_index, false);
         assert(keep_clusters <= nb_clusters);
 
         *bytes = MIN(*bytes,
                  keep_clusters * s->cluster_size
                  - offset_into_cluster(s, guest_offset));
+        assert(*bytes != 0);
+
+        calculate_l2_meta(bs, cluster_offset, guest_offset,
+                          *bytes, l2_slice, m, true);
 
         ret = 1;
     } else {
@@ -XXX,XX +XXX,XX @@ out:
     /* Only return a host offset if we actually made progress. Otherwise we
      * would make requirements for handle_alloc() that it can't fulfill */
     if (ret > 0) {
-        *host_offset = (cluster_offset & L2E_OFFSET_MASK)
-                     + offset_into_cluster(s, guest_offset);
+        *host_offset = cluster_offset + offset_into_cluster(s, guest_offset);
     }
 
     return ret;
@@ -XXX,XX +XXX,XX @@ static int do_alloc_cluster_offset(BlockDriverState *bs, uint64_t guest_offset,
 }
 
 /*
- * Allocates new clusters for an area that either is yet unallocated or needs a
- * copy on write. If *host_offset is not INV_OFFSET, clusters are only
- * allocated if the new allocation can match the specified host offset.
+ * Allocates new clusters for an area that is either still unallocated or
+ * cannot be overwritten in-place. If *host_offset is not INV_OFFSET,
+ * clusters are only allocated if the new allocation can match the specified
+ * host offset.
  *
  * Note that guest_offset may not be cluster aligned. In this case, the
  * returned *host_offset points to exact byte referenced by guest_offset and
@@ -XXX,XX +XXX,XX @@ static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
     BDRVQcow2State *s = bs->opaque;
     int l2_index;
     uint64_t *l2_slice;
-    uint64_t entry;
     uint64_t nb_clusters;
     int ret;
-    bool keep_old_clusters = false;
 
-    uint64_t alloc_cluster_offset = INV_OFFSET;
+    uint64_t alloc_cluster_offset;
 
     trace_qcow2_handle_alloc(qemu_coroutine_self(), guest_offset, *host_offset,
                              *bytes);
@@ -XXX,XX +XXX,XX @@ static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
 
     l2_index = offset_to_l2_slice_index(s, guest_offset);
     nb_clusters = MIN(nb_clusters, s->l2_slice_size - l2_index);
-    assert(nb_clusters <= INT_MAX);
-
-    /* Limit total allocation byte count to INT_MAX */
-    nb_clusters = MIN(nb_clusters, INT_MAX >> s->cluster_bits);
+    /* Limit total allocation byte count to BDRV_REQUEST_MAX_BYTES */
+    nb_clusters = MIN(nb_clusters, BDRV_REQUEST_MAX_BYTES >> s->cluster_bits);
 
     /* Find L2 entry for the first involved cluster */
     ret = get_cluster_table(bs, guest_offset, &l2_slice, &l2_index);
@@ -XXX,XX +XXX,XX @@ static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
         return ret;
     }
 
-    entry = be64_to_cpu(l2_slice[l2_index]);
-    nb_clusters = count_cow_clusters(bs, nb_clusters, l2_slice, l2_index);
+    nb_clusters = count_single_write_clusters(bs, nb_clusters,
+                                              l2_slice, l2_index, true);
 
     /* This function is only called when there were no non-COW clusters, so if
      * we can't find any unallocated or COW clusters either, something is
      * wrong with our code. */
     assert(nb_clusters > 0);
 
-    if (qcow2_get_cluster_type(bs, entry) == QCOW2_CLUSTER_ZERO_ALLOC &&
-        (entry & QCOW_OFLAG_COPIED) &&
-        (*host_offset == INV_OFFSET ||
-         start_of_cluster(s, *host_offset) == (entry & L2E_OFFSET_MASK)))
-    {
-        int preallocated_nb_clusters;
-
-        if (offset_into_cluster(s, entry & L2E_OFFSET_MASK)) {
-            qcow2_signal_corruption(bs, true, -1, -1, "Preallocated zero "
-                                    "cluster offset %#llx unaligned (guest "
-                                    "offset: %#" PRIx64 ")",
-                                    entry & L2E_OFFSET_MASK, guest_offset);
-            ret = -EIO;
-            goto fail;
-        }
-
-        /* Try to reuse preallocated zero clusters; contiguous normal clusters
-         * would be fine, too, but count_cow_clusters() above has limited
-         * nb_clusters already to a range of COW clusters */
-        preallocated_nb_clusters =
-            count_contiguous_clusters(bs, nb_clusters, s->cluster_size,
-                                      &l2_slice[l2_index], QCOW_OFLAG_COPIED);
-        assert(preallocated_nb_clusters > 0);
-
-        nb_clusters = preallocated_nb_clusters;
-        alloc_cluster_offset = entry & L2E_OFFSET_MASK;
-
-        /* We want to reuse these clusters, so qcow2_alloc_cluster_link_l2()
-         * should not free them. */
-        keep_old_clusters = true;
+    /* Allocate at a given offset in the image file */
+    alloc_cluster_offset = *host_offset == INV_OFFSET ? INV_OFFSET :
+        start_of_cluster(s, *host_offset);
+    ret = do_alloc_cluster_offset(bs, guest_offset, &alloc_cluster_offset,
+                                  &nb_clusters);
+    if (ret < 0) {
+        goto out;
     }
 
-    qcow2_cache_put(s->l2_table_cache, (void **) &l2_slice);
-
-    if (alloc_cluster_offset == INV_OFFSET) {
-        /* Allocate, if necessary at a given offset in the image file */
-        alloc_cluster_offset = *host_offset == INV_OFFSET ? INV_OFFSET :
-                               start_of_cluster(s, *host_offset);
-        ret = do_alloc_cluster_offset(bs, guest_offset, &alloc_cluster_offset,
-                                      &nb_clusters);
-        if (ret < 0) {
-            goto fail;
-        }
-
-        /* Can't extend contiguous allocation */
-        if (nb_clusters == 0) {
-            *bytes = 0;
-            return 0;
-        }
-
-        assert(alloc_cluster_offset != INV_OFFSET);
+    /* Can't extend contiguous allocation */
+    if (nb_clusters == 0) {
+        *bytes = 0;
+        ret = 0;
+        goto out;
     }
 
+    assert(alloc_cluster_offset != INV_OFFSET);
+
     /*
      * Save info needed for meta data update.
      *
@@ -XXX,XX +XXX,XX @@ static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
     *bytes = MIN(*bytes, nb_bytes - offset_into_cluster(s, guest_offset));
     assert(*bytes != 0);
 
-    calculate_l2_meta(bs, alloc_cluster_offset, guest_offset, *bytes,
-                      m, keep_old_clusters);
+    calculate_l2_meta(bs, alloc_cluster_offset, guest_offset, *bytes, l2_slice,
+                      m, false);
 
-    return 1;
+    ret = 1;
 
-fail:
-    if (*m && (*m)->nb_clusters > 0) {
+out:
+    qcow2_cache_put(s->l2_table_cache, (void **) &l2_slice);
+    if (ret < 0 && *m && (*m)->nb_clusters > 0) {
         QLIST_REMOVE(*m, next_in_flight);
     }
     return ret;
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

The size of an L2 entry is 64 bits, but if we want to have subclusters
we need extended L2 entries. This means that we have to access L2
tables and slices differently depending on whether an image has
extended L2 entries or not.

This patch replaces all l2_slice[] accesses with calls to
get_l2_entry() and set_l2_entry().

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <9586363531fec125ba1386e561762d3e4224e9fc.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h          | 12 ++++++++
 block/qcow2-cluster.c  | 63 ++++++++++++++++++++++--------------------
 block/qcow2-refcount.c | 17 ++++++------
 3 files changed, 54 insertions(+), 38 deletions(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ typedef enum QCow2MetadataOverlap {
 
 #define INV_OFFSET (-1ULL)
 
+static inline uint64_t get_l2_entry(BDRVQcow2State *s, uint64_t *l2_slice,
+                                    int idx)
+{
+    return be64_to_cpu(l2_slice[idx]);
+}
+
+static inline void set_l2_entry(BDRVQcow2State *s, uint64_t *l2_slice,
+                                int idx, uint64_t entry)
+{
+    l2_slice[idx] = cpu_to_be64(entry);
+}
+
 static inline bool has_data_file(BlockDriverState *bs)
 {
     BDRVQcow2State *s = bs->opaque;
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ fail:
  * cluster which may require a different handling)
  */
 static int count_contiguous_clusters(BlockDriverState *bs, int nb_clusters,
-        int cluster_size, uint64_t *l2_slice, uint64_t stop_flags)
+        int cluster_size, uint64_t *l2_slice, int l2_index, uint64_t stop_flags)
 {
+    BDRVQcow2State *s = bs->opaque;
     int i;
     QCow2ClusterType first_cluster_type;
     uint64_t mask = stop_flags | L2E_OFFSET_MASK | QCOW_OFLAG_COMPRESSED;
-    uint64_t first_entry = be64_to_cpu(l2_slice[0]);
+    uint64_t first_entry = get_l2_entry(s, l2_slice, l2_index);
     uint64_t offset = first_entry & mask;
 
     first_cluster_type = qcow2_get_cluster_type(bs, first_entry);
@@ -XXX,XX +XXX,XX @@ static int count_contiguous_clusters(BlockDriverState *bs, int nb_clusters,
            first_cluster_type == QCOW2_CLUSTER_ZERO_ALLOC);
 
     for (i = 0; i < nb_clusters; i++) {
-        uint64_t l2_entry = be64_to_cpu(l2_slice[i]) & mask;
+        uint64_t l2_entry = get_l2_entry(s, l2_slice, l2_index + i) & mask;
         if (offset + (uint64_t) i * cluster_size != l2_entry) {
             break;
         }
@@ -XXX,XX +XXX,XX @@ static int count_contiguous_clusters(BlockDriverState *bs, int nb_clusters,
 static int count_contiguous_clusters_unallocated(BlockDriverState *bs,
                                                  int nb_clusters,
                                                  uint64_t *l2_slice,
+                                                 int l2_index,
                                                  QCow2ClusterType wanted_type)
 {
+    BDRVQcow2State *s = bs->opaque;
     int i;
 
     assert(wanted_type == QCOW2_CLUSTER_ZERO_PLAIN ||
            wanted_type == QCOW2_CLUSTER_UNALLOCATED);
     for (i = 0; i < nb_clusters; i++) {
-        uint64_t entry = be64_to_cpu(l2_slice[i]);
+        uint64_t entry = get_l2_entry(s, l2_slice, l2_index + i);
         QCow2ClusterType type = qcow2_get_cluster_type(bs, entry);
 
         if (type != wanted_type) {
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
     /* find the cluster offset for the given disk offset */
 
     l2_index = offset_to_l2_slice_index(s, offset);
-    l2_entry = be64_to_cpu(l2_slice[l2_index]);
+    l2_entry = get_l2_entry(s, l2_slice, l2_index);
 
     nb_clusters = size_to_clusters(s, bytes_needed);
     /* bytes_needed <= *bytes + offset_in_cluster, both of which are unsigned
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
     case QCOW2_CLUSTER_UNALLOCATED:
         /* how many empty clusters ? */
         c = count_contiguous_clusters_unallocated(bs, nb_clusters,
-                                                  &l2_slice[l2_index], type);
+                                                  l2_slice, l2_index, type);
         break;
     case QCOW2_CLUSTER_ZERO_ALLOC:
     case QCOW2_CLUSTER_NORMAL: {
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
         *host_offset = host_cluster_offset + offset_in_cluster;
         /* how many allocated clusters ? */
         c = count_contiguous_clusters(bs, nb_clusters, s->cluster_size,
-                                      &l2_slice[l2_index], QCOW_OFLAG_ZERO);
+                                      l2_slice, l2_index, QCOW_OFLAG_ZERO);
         if (offset_into_cluster(s, host_cluster_offset)) {
             qcow2_signal_corruption(bs, true, -1, -1,
                                     "Cluster allocation offset %#"
@@ -XXX,XX +XXX,XX @@ int qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
 
     /* Compression can't overwrite anything. Fail if the cluster was already
      * allocated. */
-    cluster_offset = be64_to_cpu(l2_slice[l2_index]);
+    cluster_offset = get_l2_entry(s, l2_slice, l2_index);
     if (cluster_offset & L2E_OFFSET_MASK) {
         qcow2_cache_put(s->l2_table_cache, (void **) &l2_slice);
         return -EIO;
@@ -XXX,XX +XXX,XX @@ int qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
 
     BLKDBG_EVENT(bs->file, BLKDBG_L2_UPDATE_COMPRESSED);
     qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
-    l2_slice[l2_index] = cpu_to_be64(cluster_offset);
+    set_l2_entry(s, l2_slice, l2_index, cluster_offset);
     qcow2_cache_put(s->l2_table_cache, (void **) &l2_slice);
 
     *host_offset = cluster_offset & s->cluster_offset_mask;
@@ -XXX,XX +XXX,XX @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
          * cluster the second one has to do RMW (which is done above by
          * perform_cow()), update l2 table with its cluster pointer and free
          * old cluster. This is what this loop does */
-        if (l2_slice[l2_index + i] != 0) {
-            old_cluster[j++] = l2_slice[l2_index + i];
+        if (get_l2_entry(s, l2_slice, l2_index + i) != 0) {
+            old_cluster[j++] = get_l2_entry(s, l2_slice, l2_index + i);
         }
 
         /* The offset must fit in the offset field of the L2 table entry */
         assert((offset & L2E_OFFSET_MASK) == offset);
 
-        l2_slice[l2_index + i] = cpu_to_be64(offset | QCOW_OFLAG_COPIED);
+        set_l2_entry(s, l2_slice, l2_index + i, offset | QCOW_OFLAG_COPIED);
      }
 
 
@@ -XXX,XX +XXX,XX @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
      */
     if (!m->keep_old_clusters && j != 0) {
         for (i = 0; i < j; i++) {
-            qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1,
-                                    QCOW2_DISCARD_NEVER);
+            qcow2_free_any_clusters(bs, old_cluster[i], 1, QCOW2_DISCARD_NEVER);
         }
     }
 
@@ -XXX,XX +XXX,XX @@ static void calculate_l2_meta(BlockDriverState *bs,
     if (keep_old) {
         int i;
         for (i = 0; i < nb_clusters; i++) {
-            l2_entry = be64_to_cpu(l2_slice[l2_index + i]);
+            l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
             if (qcow2_get_cluster_type(bs, l2_entry) != QCOW2_CLUSTER_NORMAL) {
                 break;
             }
@@ -XXX,XX +XXX,XX @@ static void calculate_l2_meta(BlockDriverState *bs,
     }
 
     /* Get the L2 entry of the first cluster */
-    l2_entry = be64_to_cpu(l2_slice[l2_index]);
+    l2_entry = get_l2_entry(s, l2_slice, l2_index);
     type = qcow2_get_cluster_type(bs, l2_entry);
 
     if (type == QCOW2_CLUSTER_NORMAL && keep_old) {
@@ -XXX,XX +XXX,XX @@ static void calculate_l2_meta(BlockDriverState *bs,
     }
 
     /* Get the L2 entry of the last cluster */
-    l2_entry = be64_to_cpu(l2_slice[l2_index + nb_clusters - 1]);
+    l2_entry = get_l2_entry(s, l2_slice, l2_index + nb_clusters - 1);
     type = qcow2_get_cluster_type(bs, l2_entry);
 
     if (type == QCOW2_CLUSTER_NORMAL && keep_old) {
@@ -XXX,XX +XXX,XX @@ static int count_single_write_clusters(BlockDriverState *bs, int nb_clusters,
                                        bool new_alloc)
 {
     BDRVQcow2State *s = bs->opaque;
-    uint64_t l2_entry = be64_to_cpu(l2_slice[l2_index]);
+    uint64_t l2_entry = get_l2_entry(s, l2_slice, l2_index);
     uint64_t expected_offset = l2_entry & L2E_OFFSET_MASK;
     int i;
 
     for (i = 0; i < nb_clusters; i++) {
-        l2_entry = be64_to_cpu(l2_slice[l2_index + i]);
+        l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
         if (cluster_needs_new_alloc(bs, l2_entry) != new_alloc) {
             break;
         }
@@ -XXX,XX +XXX,XX @@ static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
         return ret;
     }
 
-    l2_entry = be64_to_cpu(l2_slice[l2_index]);
+    l2_entry = get_l2_entry(s, l2_slice, l2_index);
     cluster_offset = l2_entry & L2E_OFFSET_MASK;
 
     if (!cluster_needs_new_alloc(bs, l2_entry)) {
@@ -XXX,XX +XXX,XX @@ static int discard_in_l2_slice(BlockDriverState *bs, uint64_t offset,
     for (i = 0; i < nb_clusters; i++) {
         uint64_t old_l2_entry;
 
-        old_l2_entry = be64_to_cpu(l2_slice[l2_index + i]);
+        old_l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
 
         /*
          * If full_discard is false, make sure that a discarded area reads back
@@ -XXX,XX +XXX,XX @@ static int discard_in_l2_slice(BlockDriverState *bs, uint64_t offset,
         /* First remove L2 entries */
         qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
         if (!full_discard && s->qcow_version >= 3) {
-            l2_slice[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
+            set_l2_entry(s, l2_slice, l2_index + i, QCOW_OFLAG_ZERO);
         } else {
-            l2_slice[l2_index + i] = cpu_to_be64(0);
+            set_l2_entry(s, l2_slice, l2_index + i, 0);
         }
 
         /* Then decrease the refcount */
@@ -XXX,XX +XXX,XX @@ static int zero_in_l2_slice(BlockDriverState *bs, uint64_t offset,
         uint64_t old_offset;
         QCow2ClusterType cluster_type;
 
-        old_offset = be64_to_cpu(l2_slice[l2_index + i]);
+        old_offset = get_l2_entry(s, l2_slice, l2_index + i);
 
         /*
          * Minimize L2 changes if the cluster already reads back as
@@ -XXX,XX +XXX,XX @@ static int zero_in_l2_slice(BlockDriverState *bs, uint64_t offset,
 
         qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
         if (cluster_type == QCOW2_CLUSTER_COMPRESSED || unmap) {
-            l2_slice[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
+            set_l2_entry(s, l2_slice, l2_index + i, QCOW_OFLAG_ZERO);
             qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
         } else {
-            l2_slice[l2_index + i] |= cpu_to_be64(QCOW_OFLAG_ZERO);
+            uint64_t entry = get_l2_entry(s, l2_slice, l2_index + i);
+            set_l2_entry(s, l2_slice, l2_index + i, entry | QCOW_OFLAG_ZERO);
         }
     }
 
@@ -XXX,XX +XXX,XX @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
             }
 
             for (j = 0; j < s->l2_slice_size; j++) {
-                uint64_t l2_entry = be64_to_cpu(l2_slice[j]);
+                uint64_t l2_entry = get_l2_entry(s, l2_slice, j);
                 int64_t offset = l2_entry & L2E_OFFSET_MASK;
                 QCow2ClusterType cluster_type =
                     qcow2_get_cluster_type(bs, l2_entry);
@@ -XXX,XX +XXX,XX @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                     if (!bs->backing) {
                         /* not backed; therefore we can simply deallocate the
                          * cluster */
-                        l2_slice[j] = 0;
+                        set_l2_entry(s, l2_slice, j, 0);
                         l2_dirty = true;
                         continue;
                     }
@@ -XXX,XX +XXX,XX @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                 }
 
                 if (l2_refcount == 1) {
-                    l2_slice[j] = cpu_to_be64(offset | QCOW_OFLAG_COPIED);
+                    set_l2_entry(s, l2_slice, j, offset | QCOW_OFLAG_COPIED);
                 } else {
-                    l2_slice[j] = cpu_to_be64(offset);
+                    set_l2_entry(s, l2_slice, j, offset);
                 }
                 l2_dirty = true;
             }
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -XXX,XX +XXX,XX @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
                     uint64_t cluster_index;
                     uint64_t offset;
 
-                    entry = be64_to_cpu(l2_slice[j]);
+                    entry = get_l2_entry(s, l2_slice, j);
                     old_entry = entry;
                     entry &= ~QCOW_OFLAG_COPIED;
                     offset = entry & L2E_OFFSET_MASK;
@@ -XXX,XX +XXX,XX @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
                             qcow2_cache_set_dependency(bs, s->l2_table_cache,
                                                        s->refcount_block_cache);
                         }
-                        l2_slice[j] = cpu_to_be64(entry);
+                        set_l2_entry(s, l2_slice, j, entry);
                         qcow2_cache_entry_mark_dirty(s->l2_table_cache,
                                                      l2_slice);
                     }
@@ -XXX,XX +XXX,XX @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
 
     /* Do the actual checks */
     for(i = 0; i < s->l2_size; i++) {
-        l2_entry = be64_to_cpu(l2_table[i]);
+        l2_entry = get_l2_entry(s, l2_table, i);
 
         switch (qcow2_get_cluster_type(bs, l2_entry)) {
         case QCOW2_CLUSTER_COMPRESSED:
@@ -XXX,XX +XXX,XX @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
                                            QCOW2_OL_INACTIVE_L2;
 
                         l2_entry = QCOW_OFLAG_ZERO;
-                        l2_table[i] = cpu_to_be64(l2_entry);
+                        set_l2_entry(s, l2_table, i, l2_entry);
                         ret = qcow2_pre_write_overlap_check(bs, ign,
                                 l2e_offset, sizeof(uint64_t), false);
                         if (ret < 0) {
@@ -XXX,XX +XXX,XX @@ static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res,
         }
 
         for (j = 0; j < s->l2_size; j++) {
-            uint64_t l2_entry = be64_to_cpu(l2_table[j]);
+            uint64_t l2_entry = get_l2_entry(s, l2_table, j);
             uint64_t data_offset = l2_entry & L2E_OFFSET_MASK;
             QCow2ClusterType cluster_type = qcow2_get_cluster_type(bs, l2_entry);
 
@@ -XXX,XX +XXX,XX @@ static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res,
                             "l2_entry=%" PRIx64 " refcount=%" PRIu64 "\n",
                             repair ? "Repairing" : "ERROR", l2_entry, refcount);
                     if (repair) {
-                        l2_table[j] = cpu_to_be64(refcount == 1
-                                    ? l2_entry |  QCOW_OFLAG_COPIED
-                                    : l2_entry & ~QCOW_OFLAG_COPIED);
+                        set_l2_entry(s, l2_table, j,
+                                     refcount == 1 ?
+                                     l2_entry |  QCOW_OFLAG_COPIED :
+                                     l2_entry & ~QCOW_OFLAG_COPIED);
                         l2_dirty++;
                     }
                 }
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Subcluster allocation in qcow2 is implemented by extending the
existing L2 table entries and adding additional information to
indicate the allocation status of each subcluster.

This patch documents the changes to the qcow2 format and how they
affect the calculation of the L2 cache size.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <5199f2e1c717bcaa58b48142c9062b803145ff7f.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 docs/interop/qcow2.txt | 68 ++++++++++++++++++++++++++++++++++++++++--
 docs/qcow2-cache.txt   | 19 +++++++++++-
 2 files changed, 83 insertions(+), 4 deletions(-)

diff --git a/docs/interop/qcow2.txt b/docs/interop/qcow2.txt
index XXXXXXX..XXXXXXX 100644
--- a/docs/interop/qcow2.txt
+++ b/docs/interop/qcow2.txt
@@ -XXX,XX +XXX,XX @@ The first cluster of a qcow2 image contains the file header:
                     as the maximum cluster size and won't be able to open images
                     with larger cluster sizes.
 
+                    Note: if the image has Extended L2 Entries then cluster_bits
+                    must be at least 14 (i.e. 16384 byte clusters).
+
          24 - 31:   size
                     Virtual disk size in bytes.
 
@@ -XXX,XX +XXX,XX @@ the next fields through header_length.
                                 clusters. The compression_type field must be
                                 present and not zero.
 
-                    Bits 4-63:  Reserved (set to 0)
+                    Bit 4:      Extended L2 Entries.  If this bit is set then
+                                L2 table entries use an extended format that
+                                allows subcluster-based allocation. See the
+                                Extended L2 Entries section for more details.
+
+                    Bits 5-63:  Reserved (set to 0)
 
          80 -  87:  compatible_features
                     Bitmask of compatible features. An implementation can
@@ -XXX,XX +XXX,XX @@ cannot be relaxed without an incompatible layout change).
 Given an offset into the virtual disk, the offset into the image file can be
 obtained as follows:
 
-    l2_entries = (cluster_size / sizeof(uint64_t))
+    l2_entries = (cluster_size / sizeof(uint64_t))        [*]
 
     l2_index = (offset / cluster_size) % l2_entries
     l1_index = (offset / cluster_size) / l2_entries
@@ -XXX,XX +XXX,XX @@ obtained as follows:
 
     return cluster_offset + (offset % cluster_size)
 
+    [*] this changes if Extended L2 Entries are enabled, see next section
+
 L1 table entry:
 
     Bit  0 -  8:    Reserved (set to 0)
@@ -XXX,XX +XXX,XX @@ Standard Cluster Descriptor:
                     nor is data read from the backing file if the cluster is
                     unallocated.
 
-                    With version 2, this is always 0.
+                    With version 2 or with extended L2 entries (see the next
+                    section), this is always 0.
 
          1 -  8:    Reserved (set to 0)
 
@@ -XXX,XX +XXX,XX @@ file (except if bit 0 in the Standard Cluster Descriptor is set). If there is
 no backing file or the backing file is smaller than the image, they shall read
 zeros for all parts that are not covered by the backing file.
 
+== Extended L2 Entries ==
+
+An image uses Extended L2 Entries if bit 4 is set on the incompatible_features
+field of the header.
+
+In these images standard data clusters are divided into 32 subclusters of the
+same size. They are contiguous and start from the beginning of the cluster.
+Subclusters can be allocated independently and the L2 entry contains information
+indicating the status of each one of them. Compressed data clusters don't have
+subclusters so they are treated the same as in images without this feature.
+
+The size of an extended L2 entry is 128 bits so the number of entries per table
+is calculated using this formula:
+
+    l2_entries = (cluster_size / (2 * sizeof(uint64_t)))
+
+The first 64 bits have the same format as the standard L2 table entry described
+in the previous section, with the exception of bit 0 of the standard cluster
+descriptor.
+
+The last 64 bits contain a subcluster allocation bitmap with this format:
+
+Subcluster Allocation Bitmap (for standard clusters):
+
+    Bit  0 - 31:    Allocation status (one bit per subcluster)
+
+                    1: the subcluster is allocated. In this case the
+                       host cluster offset field must contain a valid
+                       offset.
+                    0: the subcluster is not allocated. In this case
+                       read requests shall go to the backing file or
+                       return zeros if there is no backing file data.
+
+                    Bits are assigned starting from the least significant
+                    one (i.e. bit x is used for subcluster x).
+
+        32 - 63     Subcluster reads as zeros (one bit per subcluster)
+
+                    1: the subcluster reads as zeros. In this case the
+                       allocation status bit must be unset. The host
+                       cluster offset field may or may not be set.
+                    0: no effect.
+
+                    Bits are assigned starting from the least significant
+                    one (i.e. bit x is used for subcluster x - 32).
+
+Subcluster Allocation Bitmap (for compressed clusters):
+
+    Bit  0 - 63:    Reserved (set to 0)
+                    Compressed clusters don't have subclusters,
+                    so this field is not used.
 
 == Snapshots ==
 
diff --git a/docs/qcow2-cache.txt b/docs/qcow2-cache.txt
index XXXXXXX..XXXXXXX 100644
--- a/docs/qcow2-cache.txt
+++ b/docs/qcow2-cache.txt
@@ -XXX,XX +XXX,XX @@
 qcow2 L2/refcount cache configuration
 =====================================
-Copyright (C) 2015, 2018 Igalia, S.L.
+Copyright (C) 2015, 2018-2020 Igalia, S.L.
 Author: Alberto Garcia <berto@igalia.com>
 
 This work is licensed under the terms of the GNU GPL, version 2 or
@@ -XXX,XX +XXX,XX @@ support this functionality, and is 0 (disabled) on other platforms.
 This functionality currently relies on the MADV_DONTNEED argument for
 madvise() to actually free the memory. This is a Linux-specific feature,
 so cache-clean-interval is not supported on other systems.
+
+
+Extended L2 Entries
+-------------------
+All numbers shown in this document are valid for qcow2 images with normal
+64-bit L2 entries.
+
+Images with extended L2 entries need twice as much L2 metadata, so the L2
+cache size must be twice as large for the same disk space.
+
+   disk_size = l2_cache_size * cluster_size / 16
+
+i.e.
+
+   l2_cache_size = disk_size * 16 / cluster_size
+
+Refcount blocks are not affected by this.
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This function will be used by the qcow2 code to check if an image has
subclusters or not.

At the moment this simply returns false. Once all patches needed for
subcluster support are ready then QEMU will be able to create and
read images with subclusters and this function will return the actual
value.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <905526221083581a1b7057bca1585487661c5c13.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ typedef enum QCow2MetadataOverlap {
 
 #define INV_OFFSET (-1ULL)
 
+static inline bool has_subclusters(BDRVQcow2State *s)
+{
+    /* FIXME: Return false until this feature is complete */
+    return false;
+}
+
 static inline uint64_t get_l2_entry(BDRVQcow2State *s, uint64_t *l2_slice,
                                     int idx)
 {
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This patch adds the following new fields to BDRVQcow2State:

- subclusters_per_cluster: Number of subclusters in a cluster
- subcluster_size: The size of each subcluster, in bytes
- subcluster_bits: No. of bits so 1 << subcluster_bits = subcluster_size

Images without subclusters are treated as if they had exactly one
subcluster per cluster (i.e. subcluster_size = cluster_size).

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <55bfeac86b092fa2c9d182a95cbeb479ff7eca4f.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h | 5 +++++
 block/qcow2.c | 5 +++++
 2 files changed, 10 insertions(+)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@
 /* The cluster reads as all zeros */
 #define QCOW_OFLAG_ZERO (1ULL << 0)
 
+#define QCOW_EXTL2_SUBCLUSTERS_PER_CLUSTER 32
+
 #define MIN_CLUSTER_BITS 9
 #define MAX_CLUSTER_BITS 21
 
@@ -XXX,XX +XXX,XX @@ typedef struct BDRVQcow2State {
     int cluster_bits;
     int cluster_size;
     int l2_slice_size;
+    int subcluster_bits;
+    int subcluster_size;
+    int subclusters_per_cluster;
     int l2_bits;
     int l2_size;
     int l1_size;
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_do_open(BlockDriverState *bs, QDict *options,
         }
     }
 
+    s->subclusters_per_cluster =
+        has_subclusters(s) ? QCOW_EXTL2_SUBCLUSTERS_PER_CLUSTER : 1;
+    s->subcluster_size = s->cluster_size / s->subclusters_per_cluster;
+    s->subcluster_bits = ctz32(s->subcluster_size);
+
     /* Check support for various header values */
     if (header.refcount_order > 6) {
         error_setg(errp, "Reference count entry width too large; may not "
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

For a given offset, return the subcluster number within its cluster
(i.e. with 32 subclusters per cluster it returns a number between 0
and 31).

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <56e3e4ac0d827c6a2f5f259106c5ddb7c4ca2653.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ static inline int offset_to_l2_slice_index(BDRVQcow2State *s, int64_t offset)
     return (offset >> s->cluster_bits) & (s->l2_slice_size - 1);
 }
 
+static inline int offset_to_sc_index(BDRVQcow2State *s, int64_t offset)
+{
+    return (offset >> s->subcluster_bits) & (s->subclusters_per_cluster - 1);
+}
+
 static inline int64_t qcow2_vm_state_offset(BDRVQcow2State *s)
 {
     return (int64_t)s->l1_vm_state_index << (s->cluster_bits + s->l2_bits);
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Like offset_into_cluster() and size_to_clusters(), but for
subclusters.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <3cc2390dcdef3d234d47c741b708bd8734490862.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ static inline int64_t offset_into_cluster(BDRVQcow2State *s, int64_t offset)
     return offset & (s->cluster_size - 1);
 }
 
+static inline int64_t offset_into_subcluster(BDRVQcow2State *s, int64_t offset)
+{
+    return offset & (s->subcluster_size - 1);
+}
+
 static inline uint64_t size_to_clusters(BDRVQcow2State *s, uint64_t size)
 {
     return (size + (s->cluster_size - 1)) >> s->cluster_bits;
 }
 
+static inline uint64_t size_to_subclusters(BDRVQcow2State *s, uint64_t size)
+{
+    return (size + (s->subcluster_size - 1)) >> s->subcluster_bits;
+}
+
 static inline int64_t size_to_l1(BDRVQcow2State *s, int64_t size)
 {
     int shift = s->cluster_bits + s->l2_bits;
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

qcow2 images with subclusters have 128-bit L2 entries. The first 64
bits contain the same information as traditional images and the last
64 bits form a bitmap with the status of each individual subcluster.

Because of that we cannot assume that L2 entries are sizeof(uint64_t)
anymore. This function returns the proper value for the image.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <d34d578bd0380e739e2dde3e8dd6187d3d249fa9.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h          |  9 +++++++++
 block/qcow2-cluster.c  | 12 ++++++------
 block/qcow2-refcount.c | 14 ++++++++------
 block/qcow2.c          |  8 ++++----
 4 files changed, 27 insertions(+), 16 deletions(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@
 
 #define QCOW_EXTL2_SUBCLUSTERS_PER_CLUSTER 32
 
+/* Size of normal and extended L2 entries */
+#define L2E_SIZE_NORMAL   (sizeof(uint64_t))
+#define L2E_SIZE_EXTENDED (sizeof(uint64_t) * 2)
+
 #define MIN_CLUSTER_BITS 9
 #define MAX_CLUSTER_BITS 21
 
@@ -XXX,XX +XXX,XX @@ static inline bool has_subclusters(BDRVQcow2State *s)
     return false;
 }
 
+static inline size_t l2_entry_size(BDRVQcow2State *s)
+{
+    return has_subclusters(s) ? L2E_SIZE_EXTENDED : L2E_SIZE_NORMAL;
+}
+
 static inline uint64_t get_l2_entry(BDRVQcow2State *s, uint64_t *l2_slice,
                                     int idx)
 {
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int l2_load(BlockDriverState *bs, uint64_t offset,
                    uint64_t l2_offset, uint64_t **l2_slice)
 {
     BDRVQcow2State *s = bs->opaque;
-    int start_of_slice = sizeof(uint64_t) *
+    int start_of_slice = l2_entry_size(s) *
         (offset_to_l2_index(s, offset) - offset_to_l2_slice_index(s, offset));
 
     return qcow2_cache_get(bs, s->l2_table_cache, l2_offset + start_of_slice,
@@ -XXX,XX +XXX,XX @@ static int l2_allocate(BlockDriverState *bs, int l1_index)
 
     /* allocate a new l2 entry */
 
-    l2_offset = qcow2_alloc_clusters(bs, s->l2_size * sizeof(uint64_t));
+    l2_offset = qcow2_alloc_clusters(bs, s->l2_size * l2_entry_size(s));
     if (l2_offset < 0) {
         ret = l2_offset;
         goto fail;
@@ -XXX,XX +XXX,XX @@ static int l2_allocate(BlockDriverState *bs, int l1_index)
 
     /* allocate a new entry in the l2 cache */
 
-    slice_size2 = s->l2_slice_size * sizeof(uint64_t);
+    slice_size2 = s->l2_slice_size * l2_entry_size(s);
     n_slices = s->cluster_size / slice_size2;
 
     trace_qcow2_l2_allocate_get_empty(bs, l1_index);
@@ -XXX,XX +XXX,XX @@ fail:
     }
     s->l1_table[l1_index] = old_l2_offset;
     if (l2_offset > 0) {
-        qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
+        qcow2_free_clusters(bs, l2_offset, s->l2_size * l2_entry_size(s),
                             QCOW2_DISCARD_ALWAYS);
     }
     return ret;
@@ -XXX,XX +XXX,XX @@ static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
 
         /* Then decrease the refcount of the old table */
         if (l2_offset) {
-            qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
+            qcow2_free_clusters(bs, l2_offset, s->l2_size * l2_entry_size(s),
                                 QCOW2_DISCARD_OTHER);
         }
 
@@ -XXX,XX +XXX,XX @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
     int ret;
     int i, j;
 
-    slice_size2 = s->l2_slice_size * sizeof(uint64_t);
+    slice_size2 = s->l2_slice_size * l2_entry_size(s);
     n_slices = s->cluster_size / slice_size2;
 
     if (!is_active_l1) {
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -XXX,XX +XXX,XX @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
     l2_slice = NULL;
     l1_table = NULL;
     l1_size2 = l1_size * sizeof(uint64_t);
-    slice_size2 = s->l2_slice_size * sizeof(uint64_t);
+    slice_size2 = s->l2_slice_size * l2_entry_size(s);
     n_slices = s->cluster_size / slice_size2;
 
     s->cache_discards = true;
@@ -XXX,XX +XXX,XX @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
     int i, l2_size, nb_csectors, ret;
 
     /* Read L2 table from disk */
-    l2_size = s->l2_size * sizeof(uint64_t);
+    l2_size = s->l2_size * l2_entry_size(s);
     l2_table = g_malloc(l2_size);
 
     ret = bdrv_pread(bs->file, l2_offset, l2_table, l2_size);
@@ -XXX,XX +XXX,XX @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
                             fix & BDRV_FIX_ERRORS ? "Repairing" : "ERROR",
                             offset);
                     if (fix & BDRV_FIX_ERRORS) {
+                        int idx = i * (l2_entry_size(s) / sizeof(uint64_t));
                         uint64_t l2e_offset =
-                            l2_offset + (uint64_t)i * sizeof(uint64_t);
+                            l2_offset + (uint64_t)i * l2_entry_size(s);
                         int ign = active ? QCOW2_OL_ACTIVE_L2 :
                                            QCOW2_OL_INACTIVE_L2;
 
                         l2_entry = QCOW_OFLAG_ZERO;
                         set_l2_entry(s, l2_table, i, l2_entry);
                         ret = qcow2_pre_write_overlap_check(bs, ign,
-                                l2e_offset, sizeof(uint64_t), false);
+                                l2e_offset, l2_entry_size(s), false);
                         if (ret < 0) {
                             fprintf(stderr, "ERROR: Overlap check failed\n");
                             res->check_errors++;
@@ -XXX,XX +XXX,XX @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
                         }
 
                         ret = bdrv_pwrite_sync(bs->file, l2e_offset,
-                                               &l2_table[i], sizeof(uint64_t));
+                                               &l2_table[idx],
+                                               l2_entry_size(s));
                         if (ret < 0) {
                             fprintf(stderr, "ERROR: Failed to overwrite L2 "
                                     "table entry: %s\n", strerror(-ret));
@@ -XXX,XX +XXX,XX @@ static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res,
         }
 
         ret = bdrv_pread(bs->file, l2_offset, l2_table,
-                         s->l2_size * sizeof(uint64_t));
+                         s->l2_size * l2_entry_size(s));
         if (ret < 0) {
             fprintf(stderr, "ERROR: Could not read L2 table: %s\n",
                     strerror(-ret));
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static void read_cache_sizes(BlockDriverState *bs, QemuOpts *opts,
     uint64_t max_l2_entries = DIV_ROUND_UP(virtual_disk_size, s->cluster_size);
     /* An L2 table is always one cluster in size so the max cache size
      * should be a multiple of the cluster size. */
-    uint64_t max_l2_cache = ROUND_UP(max_l2_entries * sizeof(uint64_t),
+    uint64_t max_l2_cache = ROUND_UP(max_l2_entries * l2_entry_size(s),
                                      s->cluster_size);
 
     combined_cache_size_set = qemu_opt_get(opts, QCOW2_OPT_CACHE_SIZE);
@@ -XXX,XX +XXX,XX @@ static int qcow2_update_options_prepare(BlockDriverState *bs,
         }
     }
 
-    r->l2_slice_size = l2_cache_entry_size / sizeof(uint64_t);
+    r->l2_slice_size = l2_cache_entry_size / l2_entry_size(s);
     r->l2_table_cache = qcow2_cache_create(bs, l2_cache_size,
                                            l2_cache_entry_size);
     r->refcount_block_cache = qcow2_cache_create(bs, refcount_cache_size,
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_do_open(BlockDriverState *bs, QDict *options,
         bs->encrypted = true;
     }
 
-    s->l2_bits = s->cluster_bits - 3; /* L2 is always one cluster */
+    s->l2_bits = s->cluster_bits - ctz32(l2_entry_size(s));
     s->l2_size = 1 << s->l2_bits;
     /* 2^(s->refcount_order - 3) is the refcount width in bytes */
     s->refcount_block_bits = s->cluster_bits - (s->refcount_order - 3);
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset,
          *  preallocation. All that matters is that we will not have to allocate
          *  new refcount structures for them.) */
         nb_new_l2_tables = DIV_ROUND_UP(nb_new_data_clusters,
-                                        s->cluster_size / sizeof(uint64_t));
+                                        s->cluster_size / l2_entry_size(s));
         /* The cluster range may not be aligned to L2 boundaries, so add one L2
          * table for a potential head/tail */
         nb_new_l2_tables++;
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Extended L2 entries are 128-bit wide: 64 bits for the entry itself and
64 bits for the subcluster allocation bitmap.

In order to support them correctly get/set_l2_entry() need to be
updated so they take the entry width into account in order to
calculate the correct offset.

This patch also adds the get/set_l2_bitmap() functions that are
used to access the bitmaps. For convenience we allow calling
get_l2_bitmap() on images without subclusters. In this case the
returned value is always 0 and has no meaning.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <6ee0f81ae3329c991de125618b3675e1e46acdbb.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ static inline size_t l2_entry_size(BDRVQcow2State *s)
 static inline uint64_t get_l2_entry(BDRVQcow2State *s, uint64_t *l2_slice,
                                     int idx)
 {
+    idx *= l2_entry_size(s) / sizeof(uint64_t);
     return be64_to_cpu(l2_slice[idx]);
 }
 
+static inline uint64_t get_l2_bitmap(BDRVQcow2State *s, uint64_t *l2_slice,
+                                     int idx)
+{
+    if (has_subclusters(s)) {
+        idx *= l2_entry_size(s) / sizeof(uint64_t);
+        return be64_to_cpu(l2_slice[idx + 1]);
+    } else {
+        return 0; /* For convenience only; this value has no meaning. */
+    }
+}
+
 static inline void set_l2_entry(BDRVQcow2State *s, uint64_t *l2_slice,
                                 int idx, uint64_t entry)
 {
+    idx *= l2_entry_size(s) / sizeof(uint64_t);
     l2_slice[idx] = cpu_to_be64(entry);
 }
 
+static inline void set_l2_bitmap(BDRVQcow2State *s, uint64_t *l2_slice,
+                                 int idx, uint64_t bitmap)
+{
+    assert(has_subclusters(s));
+    idx *= l2_entry_size(s) / sizeof(uint64_t);
+    l2_slice[idx + 1] = cpu_to_be64(bitmap);
+}
+
 static inline bool has_data_file(BlockDriverState *bs)
 {
     BDRVQcow2State *s = bs->opaque;
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This patch adds QCow2SubclusterType, which is the subcluster-level
version of QCow2ClusterType. All QCOW2_SUBCLUSTER_* values have the
the same meaning as their QCOW2_CLUSTER_* equivalents (when they
exist). See below for details and caveats.

In images without extended L2 entries clusters are treated as having
exactly one subcluster so it is possible to replace one data type with
the other while keeping the exact same semantics.

With extended L2 entries there are new possible values, and every
subcluster in the same cluster can obviously have a different
QCow2SubclusterType so functions need to be adapted to work on the
subcluster level.

There are several things that have to be taken into account:

a) QCOW2_SUBCLUSTER_COMPRESSED means that the whole cluster is
     compressed. We do not support compression at the subcluster
     level.

b) There are two different values for unallocated subclusters:
     QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN which means that the whole
     cluster is unallocated, and QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC
     which means that the cluster is allocated but the subcluster is
     not. The latter can only happen in images with extended L2
     entries.

c) QCOW2_SUBCLUSTER_INVALID is used to detect the cases where an L2
     entry has a value that violates the specification. The caller is
     responsible for handling these situations.

To prevent compatibility problems with images that have invalid
     values but are currently being read by QEMU without causing side
     effects, QCOW2_SUBCLUSTER_INVALID is only returned for images
     with extended L2 entries.

qcow2_cluster_to_subcluster_type() is added as a separate function
from qcow2_get_subcluster_type(), but this is only temporary and both
will be merged in a subsequent patch.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <26ef38e270f25851c98b51278852b4c4a7f97e69.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h | 126 +++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 125 insertions(+), 1 deletion(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@
 
 #define QCOW_EXTL2_SUBCLUSTERS_PER_CLUSTER 32
 
+/* The subcluster X [0..31] is allocated */
+#define QCOW_OFLAG_SUB_ALLOC(X)   (1ULL << (X))
+/* The subcluster X [0..31] reads as zeroes */
+#define QCOW_OFLAG_SUB_ZERO(X)    (QCOW_OFLAG_SUB_ALLOC(X) << 32)
+/* Subclusters [X, Y) (0 <= X <= Y <= 32) are allocated */
+#define QCOW_OFLAG_SUB_ALLOC_RANGE(X, Y) \
+    (QCOW_OFLAG_SUB_ALLOC(Y) - QCOW_OFLAG_SUB_ALLOC(X))
+/* Subclusters [X, Y) (0 <= X <= Y <= 32) read as zeroes */
+#define QCOW_OFLAG_SUB_ZERO_RANGE(X, Y) \
+    (QCOW_OFLAG_SUB_ALLOC_RANGE(X, Y) << 32)
+/* L2 entry bitmap with all allocation bits set */
+#define QCOW_L2_BITMAP_ALL_ALLOC  (QCOW_OFLAG_SUB_ALLOC_RANGE(0, 32))
+/* L2 entry bitmap with all "read as zeroes" bits set */
+#define QCOW_L2_BITMAP_ALL_ZEROES (QCOW_OFLAG_SUB_ZERO_RANGE(0, 32))
+
 /* Size of normal and extended L2 entries */
 #define L2E_SIZE_NORMAL   (sizeof(uint64_t))
 #define L2E_SIZE_EXTENDED (sizeof(uint64_t) * 2)
@@ -XXX,XX +XXX,XX @@ typedef struct QCowL2Meta
     QLIST_ENTRY(QCowL2Meta) next_in_flight;
 } QCowL2Meta;
 
+/*
+ * In images with standard L2 entries all clusters are treated as if
+ * they had one subcluster so QCow2ClusterType and QCow2SubclusterType
+ * can be mapped to each other and have the exact same meaning
+ * (QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC cannot happen in these images).
+ *
+ * In images with extended L2 entries QCow2ClusterType refers to the
+ * complete cluster and QCow2SubclusterType to each of the individual
+ * subclusters, so there are several possible combinations:
+ *
+ *     |--------------+---------------------------|
+ *     | Cluster type | Possible subcluster types |
+ *     |--------------+---------------------------|
+ *     | UNALLOCATED  |         UNALLOCATED_PLAIN |
+ *     |              |                ZERO_PLAIN |
+ *     |--------------+---------------------------|
+ *     | NORMAL       |         UNALLOCATED_ALLOC |
+ *     |              |                ZERO_ALLOC |
+ *     |              |                    NORMAL |
+ *     |--------------+---------------------------|
+ *     | COMPRESSED   |                COMPRESSED |
+ *     |--------------+---------------------------|
+ *
+ * QCOW2_SUBCLUSTER_INVALID means that the L2 entry is incorrect and
+ * the image should be marked corrupt.
+ */
+
 typedef enum QCow2ClusterType {
     QCOW2_CLUSTER_UNALLOCATED,
     QCOW2_CLUSTER_ZERO_PLAIN,
@@ -XXX,XX +XXX,XX @@ typedef enum QCow2ClusterType {
     QCOW2_CLUSTER_COMPRESSED,
 } QCow2ClusterType;
 
+typedef enum QCow2SubclusterType {
+    QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN,
+    QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC,
+    QCOW2_SUBCLUSTER_ZERO_PLAIN,
+    QCOW2_SUBCLUSTER_ZERO_ALLOC,
+    QCOW2_SUBCLUSTER_NORMAL,
+    QCOW2_SUBCLUSTER_COMPRESSED,
+    QCOW2_SUBCLUSTER_INVALID,
+} QCow2SubclusterType;
+
 typedef enum QCow2MetadataOverlap {
     QCOW2_OL_MAIN_HEADER_BITNR      = 0,
     QCOW2_OL_ACTIVE_L1_BITNR        = 1,
@@ -XXX,XX +XXX,XX @@ static inline int64_t qcow2_vm_state_offset(BDRVQcow2State *s)
 static inline QCow2ClusterType qcow2_get_cluster_type(BlockDriverState *bs,
                                                       uint64_t l2_entry)
 {
+    BDRVQcow2State *s = bs->opaque;
+
     if (l2_entry & QCOW_OFLAG_COMPRESSED) {
         return QCOW2_CLUSTER_COMPRESSED;
-    } else if (l2_entry & QCOW_OFLAG_ZERO) {
+    } else if ((l2_entry & QCOW_OFLAG_ZERO) && !has_subclusters(s)) {
         if (l2_entry & L2E_OFFSET_MASK) {
             return QCOW2_CLUSTER_ZERO_ALLOC;
         }
@@ -XXX,XX +XXX,XX @@ static inline QCow2ClusterType qcow2_get_cluster_type(BlockDriverState *bs,
     }
 }
 
+/*
+ * For an image without extended L2 entries, return the
+ * QCow2SubclusterType equivalent of a given QCow2ClusterType.
+ */
+static inline
+QCow2SubclusterType qcow2_cluster_to_subcluster_type(QCow2ClusterType type)
+{
+    switch (type) {
+    case QCOW2_CLUSTER_COMPRESSED:
+        return QCOW2_SUBCLUSTER_COMPRESSED;
+    case QCOW2_CLUSTER_ZERO_PLAIN:
+        return QCOW2_SUBCLUSTER_ZERO_PLAIN;
+    case QCOW2_CLUSTER_ZERO_ALLOC:
+        return QCOW2_SUBCLUSTER_ZERO_ALLOC;
+    case QCOW2_CLUSTER_NORMAL:
+        return QCOW2_SUBCLUSTER_NORMAL;
+    case QCOW2_CLUSTER_UNALLOCATED:
+        return QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN;
+    default:
+        g_assert_not_reached();
+    }
+}
+
+/*
+ * In an image without subsclusters @l2_bitmap is ignored and
+ * @sc_index must be 0.
+ * Return QCOW2_SUBCLUSTER_INVALID if an invalid l2 entry is detected
+ * (this checks the whole entry and bitmap, not only the bits related
+ * to subcluster @sc_index).
+ */
+static inline
+QCow2SubclusterType qcow2_get_subcluster_type(BlockDriverState *bs,
+                                              uint64_t l2_entry,
+                                              uint64_t l2_bitmap,
+                                              unsigned sc_index)
+{
+    BDRVQcow2State *s = bs->opaque;
+    QCow2ClusterType type = qcow2_get_cluster_type(bs, l2_entry);
+    assert(sc_index < s->subclusters_per_cluster);
+
+    if (has_subclusters(s)) {
+        switch (type) {
+        case QCOW2_CLUSTER_COMPRESSED:
+            return QCOW2_SUBCLUSTER_COMPRESSED;
+        case QCOW2_CLUSTER_NORMAL:
+            if ((l2_bitmap >> 32) & l2_bitmap) {
+                return QCOW2_SUBCLUSTER_INVALID;
+            } else if (l2_bitmap & QCOW_OFLAG_SUB_ZERO(sc_index)) {
+                return QCOW2_SUBCLUSTER_ZERO_ALLOC;
+            } else if (l2_bitmap & QCOW_OFLAG_SUB_ALLOC(sc_index)) {
+                return QCOW2_SUBCLUSTER_NORMAL;
+            } else {
+                return QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC;
+            }
+        case QCOW2_CLUSTER_UNALLOCATED:
+            if (l2_bitmap & QCOW_L2_BITMAP_ALL_ALLOC) {
+                return QCOW2_SUBCLUSTER_INVALID;
+            } else if (l2_bitmap & QCOW_OFLAG_SUB_ZERO(sc_index)) {
+                return QCOW2_SUBCLUSTER_ZERO_PLAIN;
+            } else {
+                return QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN;
+            }
+        default:
+            g_assert_not_reached();
+        }
+    } else {
+        return qcow2_cluster_to_subcluster_type(type);
+    }
+}
+
 /* Check whether refcounts are eager or lazy */
 static inline bool qcow2_need_accurate_refcounts(BDRVQcow2State *s)
 {
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

There are situations in which we want to know how many contiguous
subclusters of the same type there are in a given cluster. This can be
done by simply iterating over the subclusters and repeatedly calling
qcow2_get_subcluster_type() for each one of them.

However once we determined the type of a subcluster we can check the
rest efficiently by counting the number of adjacent ones (or zeroes)
in the bitmap. This is what this function does.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <db917263d568ec6ffb4a41cac3c9100f96bf6c18.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 51 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ fail:
     return ret;
 }
 
+/*
+ * For a given L2 entry, count the number of contiguous subclusters of
+ * the same type starting from @sc_from. Compressed clusters are
+ * treated as if they were divided into subclusters of size
+ * s->subcluster_size.
+ *
+ * Return the number of contiguous subclusters and set @type to the
+ * subcluster type.
+ *
+ * If the L2 entry is invalid return -errno and set @type to
+ * QCOW2_SUBCLUSTER_INVALID.
+ */
+G_GNUC_UNUSED
+static int qcow2_get_subcluster_range_type(BlockDriverState *bs,
+                                           uint64_t l2_entry,
+                                           uint64_t l2_bitmap,
+                                           unsigned sc_from,
+                                           QCow2SubclusterType *type)
+{
+    BDRVQcow2State *s = bs->opaque;
+    uint32_t val;
+
+    *type = qcow2_get_subcluster_type(bs, l2_entry, l2_bitmap, sc_from);
+
+    if (*type == QCOW2_SUBCLUSTER_INVALID) {
+        return -EINVAL;
+    } else if (!has_subclusters(s) || *type == QCOW2_SUBCLUSTER_COMPRESSED) {
+        return s->subclusters_per_cluster - sc_from;
+    }
+
+    switch (*type) {
+    case QCOW2_SUBCLUSTER_NORMAL:
+        val = l2_bitmap | QCOW_OFLAG_SUB_ALLOC_RANGE(0, sc_from);
+        return cto32(val) - sc_from;
+
+    case QCOW2_SUBCLUSTER_ZERO_PLAIN:
+    case QCOW2_SUBCLUSTER_ZERO_ALLOC:
+        val = (l2_bitmap | QCOW_OFLAG_SUB_ZERO_RANGE(0, sc_from)) >> 32;
+        return cto32(val) - sc_from;
+
+    case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
+    case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC:
+        val = ((l2_bitmap >> 32) | l2_bitmap)
+            & ~QCOW_OFLAG_SUB_ALLOC_RANGE(0, sc_from);
+        return ctz32(val) - sc_from;
+
+    default:
+        g_assert_not_reached();
+    }
+}
+
 /*
  * Checks how many clusters in a given L2 slice are contiguous in the image
  * file. As soon as one of the flags in the bitmask stop_flags changes compared
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This helper function tells us if a cluster is allocated (that is,
there is an associated host offset for it).

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <6d8771c5c79cbdc6c519875a5078e1cc85856d63.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ QCow2SubclusterType qcow2_get_subcluster_type(BlockDriverState *bs,
     }
 }
 
+static inline bool qcow2_cluster_is_allocated(QCow2ClusterType type)
+{
+    return (type == QCOW2_CLUSTER_COMPRESSED || type == QCOW2_CLUSTER_NORMAL ||
+            type == QCOW2_CLUSTER_ZERO_ALLOC);
+}
+
 /* Check whether refcounts are eager or lazy */
 static inline bool qcow2_need_accurate_refcounts(BDRVQcow2State *s)
 {
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This function returns an integer that can be either an error code or a
cluster type (a value from the QCow2ClusterType enum).

We are going to start using subcluster types instead of cluster types
in some functions so it's better to use the exact data types instead
of integers for clarity and in order to detect errors more easily.

This patch makes qcow2_get_host_offset() return 0 on success and
puts the returned cluster type in a separate parameter. There are no
semantic changes.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <396b6eab1859a271551dcd7dcba77f8934aa3c3f.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h         |  3 ++-
 block/qcow2-cluster.c | 11 +++++++----
 block/qcow2.c         | 37 ++++++++++++++++++++++---------------
 3 files changed, 31 insertions(+), 20 deletions(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ int qcow2_encrypt_sectors(BDRVQcow2State *s, int64_t sector_num,
                           uint8_t *buf, int nb_sectors, bool enc, Error **errp);
 
 int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
-                          unsigned int *bytes, uint64_t *host_offset);
+                          unsigned int *bytes, uint64_t *host_offset,
+                          QCow2ClusterType *cluster_type);
 int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
                                unsigned int *bytes, uint64_t *host_offset,
                                QCowL2Meta **m);
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn do_perform_cow_write(BlockDriverState *bs,
  *
  * On exit, *bytes is the number of bytes starting at offset that have the same
  * cluster type and (if applicable) are stored contiguously in the image file.
+ * The cluster type is stored in *cluster_type.
  * Compressed clusters are always returned one by one.
  *
- * Returns the cluster type (QCOW2_CLUSTER_*) on success, -errno in error
- * cases.
+ * Returns 0 on success, -errno in error cases.
  */
 int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
-                          unsigned int *bytes, uint64_t *host_offset)
+                          unsigned int *bytes, uint64_t *host_offset,
+                          QCow2ClusterType *cluster_type)
 {
     BDRVQcow2State *s = bs->opaque;
     unsigned int l2_index;
@@ -XXX,XX +XXX,XX @@ out:
     assert(bytes_available - offset_in_cluster <= UINT_MAX);
     *bytes = bytes_available - offset_in_cluster;
 
-    return type;
+    *cluster_type = type;
+
+    return 0;
 
 fail:
     qcow2_cache_put(s->l2_table_cache, (void **)&l2_slice);
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
     BDRVQcow2State *s = bs->opaque;
     uint64_t host_offset;
     unsigned int bytes;
+    QCow2ClusterType type;
     int ret, status = 0;
 
     qemu_co_mutex_lock(&s->lock);
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
     }
 
     bytes = MIN(INT_MAX, count);
-    ret = qcow2_get_host_offset(bs, offset, &bytes, &host_offset);
+    ret = qcow2_get_host_offset(bs, offset, &bytes, &host_offset, &type);
     qemu_co_mutex_unlock(&s->lock);
     if (ret < 0) {
         return ret;
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
 
     *pnum = bytes;
 
-    if ((ret == QCOW2_CLUSTER_NORMAL || ret == QCOW2_CLUSTER_ZERO_ALLOC) &&
+    if ((type == QCOW2_CLUSTER_NORMAL || type == QCOW2_CLUSTER_ZERO_ALLOC) &&
         !s->crypto) {
         *map = host_offset;
         *file = s->data_file->bs;
         status |= BDRV_BLOCK_OFFSET_VALID;
     }
-    if (ret == QCOW2_CLUSTER_ZERO_PLAIN || ret == QCOW2_CLUSTER_ZERO_ALLOC) {
+    if (type == QCOW2_CLUSTER_ZERO_PLAIN || type == QCOW2_CLUSTER_ZERO_ALLOC) {
         status |= BDRV_BLOCK_ZERO;
-    } else if (ret != QCOW2_CLUSTER_UNALLOCATED) {
+    } else if (type != QCOW2_CLUSTER_UNALLOCATED) {
         status |= BDRV_BLOCK_DATA;
     }
     if (s->metadata_preallocation && (status & BDRV_BLOCK_DATA) &&
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
     int ret = 0;
     unsigned int cur_bytes; /* number of bytes in current iteration */
     uint64_t host_offset = 0;
+    QCow2ClusterType type;
     AioTaskPool *aio = NULL;
 
     while (bytes != 0 && aio_task_pool_status(aio) == 0) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
         }
 
         qemu_co_mutex_lock(&s->lock);
-        ret = qcow2_get_host_offset(bs, offset, &cur_bytes, &host_offset);
+        ret = qcow2_get_host_offset(bs, offset, &cur_bytes,
+                                    &host_offset, &type);
         qemu_co_mutex_unlock(&s->lock);
         if (ret < 0) {
             goto out;
         }
 
-        if (ret == QCOW2_CLUSTER_ZERO_PLAIN ||
-            ret == QCOW2_CLUSTER_ZERO_ALLOC ||
-            (ret == QCOW2_CLUSTER_UNALLOCATED && !bs->backing))
+        if (type == QCOW2_CLUSTER_ZERO_PLAIN ||
+            type == QCOW2_CLUSTER_ZERO_ALLOC ||
+            (type == QCOW2_CLUSTER_UNALLOCATED && !bs->backing))
         {
             qemu_iovec_memset(qiov, qiov_offset, 0, cur_bytes);
         } else {
             if (!aio && cur_bytes != bytes) {
                 aio = aio_task_pool_new(QCOW2_MAX_WORKERS);
             }
-            ret = qcow2_add_task(bs, aio, qcow2_co_preadv_task_entry, ret,
+            ret = qcow2_add_task(bs, aio, qcow2_co_preadv_task_entry, type,
                                  host_offset, offset, cur_bytes,
                                  qiov, qiov_offset, NULL);
             if (ret < 0) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
     if (head || tail) {
         uint64_t off;
         unsigned int nr;
+        QCow2ClusterType type;
 
         assert(head + bytes <= s->cluster_size);
 
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
         offset = QEMU_ALIGN_DOWN(offset, s->cluster_size);
         bytes = s->cluster_size;
         nr = s->cluster_size;
-        ret = qcow2_get_host_offset(bs, offset, &nr, &off);
-        if (ret != QCOW2_CLUSTER_UNALLOCATED &&
-            ret != QCOW2_CLUSTER_ZERO_PLAIN &&
-            ret != QCOW2_CLUSTER_ZERO_ALLOC) {
+        ret = qcow2_get_host_offset(bs, offset, &nr, &off, &type);
+        if (ret < 0 ||
+            (type != QCOW2_CLUSTER_UNALLOCATED &&
+             type != QCOW2_CLUSTER_ZERO_PLAIN &&
+             type != QCOW2_CLUSTER_ZERO_ALLOC)) {
             qemu_co_mutex_unlock(&s->lock);
             return -ENOTSUP;
         }
@@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_from(BlockDriverState *bs,
 
     while (bytes != 0) {
         uint64_t copy_offset = 0;
+        QCow2ClusterType type;
         /* prepare next request */
         cur_bytes = MIN(bytes, INT_MAX);
         cur_write_flags = write_flags;
 
-        ret = qcow2_get_host_offset(bs, src_offset, &cur_bytes, &copy_offset);
+        ret = qcow2_get_host_offset(bs, src_offset, &cur_bytes,
+                                    &copy_offset, &type);
         if (ret < 0) {
             goto out;
         }
 
-        switch (ret) {
+        switch (type) {
         case QCOW2_CLUSTER_UNALLOCATED:
             if (bs->backing && bs->backing->bs) {
                 int64_t backing_length = bdrv_getlength(bs->backing->bs);
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

In order to support extended L2 entries some functions of the qcow2
driver need to start dealing with subclusters instead of clusters.

qcow2_get_host_offset() is modified to return the subcluster type
instead of the cluster type, and all callers are updated to replace
all values of QCow2ClusterType with their QCow2SubclusterType
equivalents.

This patch only changes the data types, there are no semantic changes.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <f6c29737c295f32cbee74c903c30b01820363b34.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h         |  2 +-
 block/qcow2-cluster.c | 10 +++----
 block/qcow2.c         | 70 ++++++++++++++++++++++---------------------
 3 files changed, 42 insertions(+), 40 deletions(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ int qcow2_encrypt_sectors(BDRVQcow2State *s, int64_t sector_num,
 
 int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
                           unsigned int *bytes, uint64_t *host_offset,
-                          QCow2ClusterType *cluster_type);
+                          QCow2SubclusterType *subcluster_type);
 int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
                                unsigned int *bytes, uint64_t *host_offset,
                                QCowL2Meta **m);
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn do_perform_cow_write(BlockDriverState *bs,
  * offset that we are interested in.
  *
  * On exit, *bytes is the number of bytes starting at offset that have the same
- * cluster type and (if applicable) are stored contiguously in the image file.
- * The cluster type is stored in *cluster_type.
- * Compressed clusters are always returned one by one.
+ * subcluster type and (if applicable) are stored contiguously in the image
+ * file. The subcluster type is stored in *subcluster_type.
+ * Compressed clusters are always processed one by one.
  *
  * Returns 0 on success, -errno in error cases.
  */
 int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
                           unsigned int *bytes, uint64_t *host_offset,
-                          QCow2ClusterType *cluster_type)
+                          QCow2SubclusterType *subcluster_type)
 {
     BDRVQcow2State *s = bs->opaque;
     unsigned int l2_index;
@@ -XXX,XX +XXX,XX @@ out:
     assert(bytes_available - offset_in_cluster <= UINT_MAX);
     *bytes = bytes_available - offset_in_cluster;
 
-    *cluster_type = type;
+    *subcluster_type = qcow2_cluster_to_subcluster_type(type);
 
     return 0;
 
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
     BDRVQcow2State *s = bs->opaque;
     uint64_t host_offset;
     unsigned int bytes;
-    QCow2ClusterType type;
+    QCow2SubclusterType type;
     int ret, status = 0;
 
     qemu_co_mutex_lock(&s->lock);
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
 
     *pnum = bytes;
 
-    if ((type == QCOW2_CLUSTER_NORMAL || type == QCOW2_CLUSTER_ZERO_ALLOC) &&
-        !s->crypto) {
+    if ((type == QCOW2_SUBCLUSTER_NORMAL ||
+         type == QCOW2_SUBCLUSTER_ZERO_ALLOC) && !s->crypto) {
         *map = host_offset;
         *file = s->data_file->bs;
         status |= BDRV_BLOCK_OFFSET_VALID;
     }
-    if (type == QCOW2_CLUSTER_ZERO_PLAIN || type == QCOW2_CLUSTER_ZERO_ALLOC) {
+    if (type == QCOW2_SUBCLUSTER_ZERO_PLAIN ||
+        type == QCOW2_SUBCLUSTER_ZERO_ALLOC) {
         status |= BDRV_BLOCK_ZERO;
-    } else if (type != QCOW2_CLUSTER_UNALLOCATED) {
+    } else if (type != QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN) {
         status |= BDRV_BLOCK_DATA;
     }
     if (s->metadata_preallocation && (status & BDRV_BLOCK_DATA) &&
@@ -XXX,XX +XXX,XX @@ typedef struct Qcow2AioTask {
     AioTask task;
 
     BlockDriverState *bs;
-    QCow2ClusterType cluster_type; /* only for read */
+    QCow2SubclusterType subcluster_type; /* only for read */
     uint64_t host_offset; /* or full descriptor in compressed clusters */
     uint64_t offset;
     uint64_t bytes;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task_entry(AioTask *task);
 static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
                                        AioTaskPool *pool,
                                        AioTaskFunc func,
-                                       QCow2ClusterType cluster_type,
+                                       QCow2SubclusterType subcluster_type,
                                        uint64_t host_offset,
                                        uint64_t offset,
                                        uint64_t bytes,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
     *task = (Qcow2AioTask) {
         .task.func = func,
         .bs = bs,
-        .cluster_type = cluster_type,
+        .subcluster_type = subcluster_type,
         .qiov = qiov,
         .host_offset = host_offset,
         .offset = offset,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
 
     trace_qcow2_add_task(qemu_coroutine_self(), bs, pool,
                          func == qcow2_co_preadv_task_entry ? "read" : "write",
-                         cluster_type, host_offset, offset, bytes,
+                         subcluster_type, host_offset, offset, bytes,
                          qiov, qiov_offset);
 
     if (!pool) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_add_task(BlockDriverState *bs,
 }
 
 static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs,
-                                             QCow2ClusterType cluster_type,
+                                             QCow2SubclusterType subc_type,
                                              uint64_t host_offset,
                                              uint64_t offset, uint64_t bytes,
                                              QEMUIOVector *qiov,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs,
 {
     BDRVQcow2State *s = bs->opaque;
 
-    switch (cluster_type) {
-    case QCOW2_CLUSTER_ZERO_PLAIN:
-    case QCOW2_CLUSTER_ZERO_ALLOC:
+    switch (subc_type) {
+    case QCOW2_SUBCLUSTER_ZERO_PLAIN:
+    case QCOW2_SUBCLUSTER_ZERO_ALLOC:
         /* Both zero types are handled in qcow2_co_preadv_part */
         g_assert_not_reached();
 
-    case QCOW2_CLUSTER_UNALLOCATED:
+    case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
         assert(bs->backing); /* otherwise handled in qcow2_co_preadv_part */
 
         BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO);
         return bdrv_co_preadv_part(bs->backing, offset, bytes,
                                    qiov, qiov_offset, 0);
 
-    case QCOW2_CLUSTER_COMPRESSED:
+    case QCOW2_SUBCLUSTER_COMPRESSED:
         return qcow2_co_preadv_compressed(bs, host_offset,
                                           offset, bytes, qiov, qiov_offset);
 
-    case QCOW2_CLUSTER_NORMAL:
+    case QCOW2_SUBCLUSTER_NORMAL:
         if (bs->encrypted) {
             return qcow2_co_preadv_encrypted(bs, host_offset,
                                              offset, bytes, qiov, qiov_offset);
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task_entry(AioTask *task)
 
     assert(!t->l2meta);
 
-    return qcow2_co_preadv_task(t->bs, t->cluster_type, t->host_offset,
-                                t->offset, t->bytes, t->qiov, t->qiov_offset);
+    return qcow2_co_preadv_task(t->bs, t->subcluster_type,
+                                t->host_offset, t->offset, t->bytes,
+                                t->qiov, t->qiov_offset);
 }
 
 static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
     int ret = 0;
     unsigned int cur_bytes; /* number of bytes in current iteration */
     uint64_t host_offset = 0;
-    QCow2ClusterType type;
+    QCow2SubclusterType type;
     AioTaskPool *aio = NULL;
 
     while (bytes != 0 && aio_task_pool_status(aio) == 0) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
             goto out;
         }
 
-        if (type == QCOW2_CLUSTER_ZERO_PLAIN ||
-            type == QCOW2_CLUSTER_ZERO_ALLOC ||
-            (type == QCOW2_CLUSTER_UNALLOCATED && !bs->backing))
+        if (type == QCOW2_SUBCLUSTER_ZERO_PLAIN ||
+            type == QCOW2_SUBCLUSTER_ZERO_ALLOC ||
+            (type == QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN && !bs->backing))
         {
             qemu_iovec_memset(qiov, qiov_offset, 0, cur_bytes);
         } else {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_task_entry(AioTask *task)
 {
     Qcow2AioTask *t = container_of(task, Qcow2AioTask, task);
 
-    assert(!t->cluster_type);
+    assert(!t->subcluster_type);
 
     return qcow2_co_pwritev_task(t->bs, t->host_offset,
                                  t->offset, t->bytes, t->qiov, t->qiov_offset,
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
     if (head || tail) {
         uint64_t off;
         unsigned int nr;
-        QCow2ClusterType type;
+        QCow2SubclusterType type;
 
         assert(head + bytes <= s->cluster_size);
 
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
         nr = s->cluster_size;
         ret = qcow2_get_host_offset(bs, offset, &nr, &off, &type);
         if (ret < 0 ||
-            (type != QCOW2_CLUSTER_UNALLOCATED &&
-             type != QCOW2_CLUSTER_ZERO_PLAIN &&
-             type != QCOW2_CLUSTER_ZERO_ALLOC)) {
+            (type != QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN &&
+             type != QCOW2_SUBCLUSTER_ZERO_PLAIN &&
+             type != QCOW2_SUBCLUSTER_ZERO_ALLOC)) {
             qemu_co_mutex_unlock(&s->lock);
             return -ENOTSUP;
         }
@@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_from(BlockDriverState *bs,
 
     while (bytes != 0) {
         uint64_t copy_offset = 0;
-        QCow2ClusterType type;
+        QCow2SubclusterType type;
         /* prepare next request */
         cur_bytes = MIN(bytes, INT_MAX);
         cur_write_flags = write_flags;
@@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_from(BlockDriverState *bs,
         }
 
         switch (type) {
-        case QCOW2_CLUSTER_UNALLOCATED:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
             if (bs->backing && bs->backing->bs) {
                 int64_t backing_length = bdrv_getlength(bs->backing->bs);
                 if (src_offset >= backing_length) {
@@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_from(BlockDriverState *bs,
             }
             break;
 
-        case QCOW2_CLUSTER_ZERO_PLAIN:
-        case QCOW2_CLUSTER_ZERO_ALLOC:
+        case QCOW2_SUBCLUSTER_ZERO_PLAIN:
+        case QCOW2_SUBCLUSTER_ZERO_ALLOC:
             cur_write_flags |= BDRV_REQ_ZERO_WRITE;
             break;
 
-        case QCOW2_CLUSTER_COMPRESSED:
+        case QCOW2_SUBCLUSTER_COMPRESSED:
             ret = -ENOTSUP;
             goto out;
 
-        case QCOW2_CLUSTER_NORMAL:
+        case QCOW2_SUBCLUSTER_NORMAL:
             child = s->data_file;
             break;
 
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwritev_compressed_task_entry(AioTask *task)
 {
     Qcow2AioTask *t = container_of(task, Qcow2AioTask, task);
 
-    assert(!t->cluster_type && !t->l2meta);
+    assert(!t->subcluster_type && !t->l2meta);
 
     return qcow2_co_pwritev_compressed_task(t->bs, t->offset, t->bytes, t->qiov,
                                             t->qiov_offset);
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

When dealing with subcluster types there is a new value called
QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC that has no equivalent in
QCow2ClusterType.

This patch handles that value in all places where subcluster types
are processed.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <bf09e2e2439a468a901bb96ace411eed9ee50295.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
     *pnum = bytes;
 
     if ((type == QCOW2_SUBCLUSTER_NORMAL ||
-         type == QCOW2_SUBCLUSTER_ZERO_ALLOC) && !s->crypto) {
+         type == QCOW2_SUBCLUSTER_ZERO_ALLOC ||
+         type == QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC) && !s->crypto) {
         *map = host_offset;
         *file = s->data_file->bs;
         status |= BDRV_BLOCK_OFFSET_VALID;
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_block_status(BlockDriverState *bs,
     if (type == QCOW2_SUBCLUSTER_ZERO_PLAIN ||
         type == QCOW2_SUBCLUSTER_ZERO_ALLOC) {
         status |= BDRV_BLOCK_ZERO;
-    } else if (type != QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN) {
+    } else if (type != QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN &&
+               type != QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC) {
         status |= BDRV_BLOCK_DATA;
     }
     if (s->metadata_preallocation && (status & BDRV_BLOCK_DATA) &&
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs,
         g_assert_not_reached();
 
     case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
+    case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC:
         assert(bs->backing); /* otherwise handled in qcow2_co_preadv_part */
 
         BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO);
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs,
 
         if (type == QCOW2_SUBCLUSTER_ZERO_PLAIN ||
             type == QCOW2_SUBCLUSTER_ZERO_ALLOC ||
-            (type == QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN && !bs->backing))
+            (type == QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN && !bs->backing) ||
+            (type == QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC && !bs->backing))
         {
             qemu_iovec_memset(qiov, qiov_offset, 0, cur_bytes);
         } else {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
         ret = qcow2_get_host_offset(bs, offset, &nr, &off, &type);
         if (ret < 0 ||
             (type != QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN &&
+             type != QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC &&
              type != QCOW2_SUBCLUSTER_ZERO_PLAIN &&
              type != QCOW2_SUBCLUSTER_ZERO_ALLOC)) {
             qemu_co_mutex_unlock(&s->lock);
@@ -XXX,XX +XXX,XX @@ qcow2_co_copy_range_from(BlockDriverState *bs,
 
         switch (type) {
         case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC:
             if (bs->backing && bs->backing->bs) {
                 int64_t backing_length = bdrv_getlength(bs->backing->bs);
                 if (src_offset >= backing_length) {
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

If an image has subclusters then there are more copy-on-write
scenarios that we need to consider. Let's say we have a write request
from the middle of subcluster #3 until the end of the cluster:

1) If we are writing to a newly allocated cluster then we need
   copy-on-write. The previous contents of subclusters #0 to #3 must
   be copied to the new cluster. We can optimize this process by
   skipping all leading unallocated or zero subclusters (the status of
   those skipped subclusters will be reflected in the new L2 bitmap).

2) If we are overwriting an existing cluster:

2.1) If subcluster #3 is unallocated or has the all-zeroes bit set
        then we need copy-on-write (on subcluster #3 only).

2.2) If subcluster #3 was already allocated then there is no need
        for any copy-on-write. However we still need to update the L2
        bitmap to reflect possible changes in the allocation status of
        subclusters #4 to #31. Because of this, this function checks
        if all the overwritten subclusters are already allocated and
        in this case it returns without creating a new QCowL2Meta
        structure.

After all these changes l2meta_cow_start() and l2meta_cow_end()
are not necessarily cluster-aligned anymore. We need to update the
calculation of old_start and old_end in handle_dependencies() to
guarantee that no two requests try to write on the same cluster.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <4292dd56e4446d386a2fe307311737a711c00708.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 167 +++++++++++++++++++++++++++++++++---------
 1 file changed, 133 insertions(+), 34 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ fail:
  * If the L2 entry is invalid return -errno and set @type to
  * QCOW2_SUBCLUSTER_INVALID.
  */
-G_GNUC_UNUSED
 static int qcow2_get_subcluster_range_type(BlockDriverState *bs,
                                            uint64_t l2_entry,
                                            uint64_t l2_bitmap,
@@ -XXX,XX +XXX,XX @@ void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m)
  * If @keep_old is true it means that the clusters were already
  * allocated and will be overwritten. If false then the clusters are
  * new and we have to decrease the reference count of the old ones.
+ *
+ * Returns 0 on success, -errno on failure.
  */
-static void calculate_l2_meta(BlockDriverState *bs,
-                              uint64_t host_cluster_offset,
-                              uint64_t guest_offset, unsigned bytes,
-                              uint64_t *l2_slice, QCowL2Meta **m, bool keep_old)
+static int calculate_l2_meta(BlockDriverState *bs, uint64_t host_cluster_offset,
+                             uint64_t guest_offset, unsigned bytes,
+                             uint64_t *l2_slice, QCowL2Meta **m, bool keep_old)
 {
     BDRVQcow2State *s = bs->opaque;
-    int l2_index = offset_to_l2_slice_index(s, guest_offset);
-    uint64_t l2_entry;
+    int sc_index, l2_index = offset_to_l2_slice_index(s, guest_offset);
+    uint64_t l2_entry, l2_bitmap;
     unsigned cow_start_from, cow_end_to;
     unsigned cow_start_to = offset_into_cluster(s, guest_offset);
     unsigned cow_end_from = cow_start_to + bytes;
     unsigned nb_clusters = size_to_clusters(s, cow_end_from);
     QCowL2Meta *old_m = *m;
-    QCow2ClusterType type;
+    QCow2SubclusterType type;
+    int i;
+    bool skip_cow = keep_old;
 
     assert(nb_clusters <= s->l2_slice_size - l2_index);
 
-    /* Return if there's no COW (all clusters are normal and we keep them) */
-    if (keep_old) {
-        int i;
-        for (i = 0; i < nb_clusters; i++) {
-            l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
-            if (qcow2_get_cluster_type(bs, l2_entry) != QCOW2_CLUSTER_NORMAL) {
-                break;
+    /* Check the type of all affected subclusters */
+    for (i = 0; i < nb_clusters; i++) {
+        l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
+        l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index + i);
+        if (skip_cow) {
+            unsigned write_from = MAX(cow_start_to, i << s->cluster_bits);
+            unsigned write_to = MIN(cow_end_from, (i + 1) << s->cluster_bits);
+            int first_sc = offset_to_sc_index(s, write_from);
+            int last_sc = offset_to_sc_index(s, write_to - 1);
+            int cnt = qcow2_get_subcluster_range_type(bs, l2_entry, l2_bitmap,
+                                                      first_sc, &type);
+            /* Is any of the subclusters of type != QCOW2_SUBCLUSTER_NORMAL ? */
+            if (type != QCOW2_SUBCLUSTER_NORMAL || first_sc + cnt <= last_sc) {
+                skip_cow = false;
             }
+        } else {
+            /* If we can't skip the cow we can still look for invalid entries */
+            type = qcow2_get_subcluster_type(bs, l2_entry, l2_bitmap, 0);
         }
-        if (i == nb_clusters) {
-            return;
+        if (type == QCOW2_SUBCLUSTER_INVALID) {
+            int l1_index = offset_to_l1_index(s, guest_offset);
+            uint64_t l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
+            qcow2_signal_corruption(bs, true, -1, -1, "Invalid cluster "
+                                    "entry found (L2 offset: %#" PRIx64
+                                    ", L2 index: %#x)",
+                                    l2_offset, l2_index + i);
+            return -EIO;
         }
     }
 
+    if (skip_cow) {
+        return 0;
+    }
+
     /* Get the L2 entry of the first cluster */
     l2_entry = get_l2_entry(s, l2_slice, l2_index);
-    type = qcow2_get_cluster_type(bs, l2_entry);
-
-    if (type == QCOW2_CLUSTER_NORMAL && keep_old) {
-        cow_start_from = cow_start_to;
+    l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index);
+    sc_index = offset_to_sc_index(s, guest_offset);
+    type = qcow2_get_subcluster_type(bs, l2_entry, l2_bitmap, sc_index);
+
+    if (!keep_old) {
+        switch (type) {
+        case QCOW2_SUBCLUSTER_COMPRESSED:
+            cow_start_from = 0;
+            break;
+        case QCOW2_SUBCLUSTER_NORMAL:
+        case QCOW2_SUBCLUSTER_ZERO_ALLOC:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC:
+            if (has_subclusters(s)) {
+                /* Skip all leading zero and unallocated subclusters */
+                uint32_t alloc_bitmap = l2_bitmap & QCOW_L2_BITMAP_ALL_ALLOC;
+                cow_start_from =
+                    MIN(sc_index, ctz32(alloc_bitmap)) << s->subcluster_bits;
+            } else {
+                cow_start_from = 0;
+            }
+            break;
+        case QCOW2_SUBCLUSTER_ZERO_PLAIN:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
+            cow_start_from = sc_index << s->subcluster_bits;
+            break;
+        default:
+            g_assert_not_reached();
+        }
     } else {
-        cow_start_from = 0;
+        switch (type) {
+        case QCOW2_SUBCLUSTER_NORMAL:
+            cow_start_from = cow_start_to;
+            break;
+        case QCOW2_SUBCLUSTER_ZERO_ALLOC:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC:
+            cow_start_from = sc_index << s->subcluster_bits;
+            break;
+        default:
+            g_assert_not_reached();
+        }
     }
 
     /* Get the L2 entry of the last cluster */
-    l2_entry = get_l2_entry(s, l2_slice, l2_index + nb_clusters - 1);
-    type = qcow2_get_cluster_type(bs, l2_entry);
-
-    if (type == QCOW2_CLUSTER_NORMAL && keep_old) {
-        cow_end_to = cow_end_from;
+    l2_index += nb_clusters - 1;
+    l2_entry = get_l2_entry(s, l2_slice, l2_index);
+    l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index);
+    sc_index = offset_to_sc_index(s, guest_offset + bytes - 1);
+    type = qcow2_get_subcluster_type(bs, l2_entry, l2_bitmap, sc_index);
+
+    if (!keep_old) {
+        switch (type) {
+        case QCOW2_SUBCLUSTER_COMPRESSED:
+            cow_end_to = ROUND_UP(cow_end_from, s->cluster_size);
+            break;
+        case QCOW2_SUBCLUSTER_NORMAL:
+        case QCOW2_SUBCLUSTER_ZERO_ALLOC:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC:
+            cow_end_to = ROUND_UP(cow_end_from, s->cluster_size);
+            if (has_subclusters(s)) {
+                /* Skip all trailing zero and unallocated subclusters */
+                uint32_t alloc_bitmap = l2_bitmap & QCOW_L2_BITMAP_ALL_ALLOC;
+                cow_end_to -=
+                    MIN(s->subclusters_per_cluster - sc_index - 1,
+                        clz32(alloc_bitmap)) << s->subcluster_bits;
+            }
+            break;
+        case QCOW2_SUBCLUSTER_ZERO_PLAIN:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
+            cow_end_to = ROUND_UP(cow_end_from, s->subcluster_size);
+            break;
+        default:
+            g_assert_not_reached();
+        }
     } else {
-        cow_end_to = ROUND_UP(cow_end_from, s->cluster_size);
+        switch (type) {
+        case QCOW2_SUBCLUSTER_NORMAL:
+            cow_end_to = cow_end_from;
+            break;
+        case QCOW2_SUBCLUSTER_ZERO_ALLOC:
+        case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC:
+            cow_end_to = ROUND_UP(cow_end_from, s->subcluster_size);
+            break;
+        default:
+            g_assert_not_reached();
+        }
     }
 
     *m = g_malloc0(sizeof(**m));
@@ -XXX,XX +XXX,XX @@ static void calculate_l2_meta(BlockDriverState *bs,
 
     qemu_co_queue_init(&(*m)->dependent_requests);
     QLIST_INSERT_HEAD(&s->cluster_allocs, *m, next_in_flight);
+
+    return 0;
 }
 
 /*
@@ -XXX,XX +XXX,XX @@ static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
 
         uint64_t start = guest_offset;
         uint64_t end = start + bytes;
-        uint64_t old_start = l2meta_cow_start(old_alloc);
-        uint64_t old_end = l2meta_cow_end(old_alloc);
+        uint64_t old_start = start_of_cluster(s, l2meta_cow_start(old_alloc));
+        uint64_t old_end = ROUND_UP(l2meta_cow_end(old_alloc), s->cluster_size);
 
         if (end <= old_start || start >= old_end) {
             /* No intersection */
@@ -XXX,XX +XXX,XX @@ static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
                  - offset_into_cluster(s, guest_offset));
         assert(*bytes != 0);
 
-        calculate_l2_meta(bs, cluster_offset, guest_offset,
-                          *bytes, l2_slice, m, true);
+        ret = calculate_l2_meta(bs, cluster_offset, guest_offset,
+                                *bytes, l2_slice, m, true);
+        if (ret < 0) {
+            goto out;
+        }
 
         ret = 1;
     } else {
@@ -XXX,XX +XXX,XX @@ static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
     *bytes = MIN(*bytes, nb_bytes - offset_into_cluster(s, guest_offset));
     assert(*bytes != 0);
 
-    calculate_l2_meta(bs, alloc_cluster_offset, guest_offset, *bytes, l2_slice,
-                      m, false);
+    ret = calculate_l2_meta(bs, alloc_cluster_offset, guest_offset, *bytes,
+                            l2_slice, m, false);
+    if (ret < 0) {
+        goto out;
+    }
 
     ret = 1;
 
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

The logic of this function remains pretty much the same, except that
it uses count_contiguous_subclusters(), which combines the logic of
count_contiguous_clusters() / count_contiguous_clusters_unallocated()
and checks individual subclusters.

qcow2_cluster_to_subcluster_type() is not necessary as a separate
function anymore so it's inlined into its caller.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <d2193fd48653a350d80f0eca1c67b1d9053fb2f3.1594396418.git.berto@igalia.com>
[mreitz: Initialize expected_type to anything]
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h         |  38 ++++-------
 block/qcow2-cluster.c | 150 ++++++++++++++++++++++--------------------
 2 files changed, 92 insertions(+), 96 deletions(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ static inline QCow2ClusterType qcow2_get_cluster_type(BlockDriverState *bs,
     }
 }
 
-/*
- * For an image without extended L2 entries, return the
- * QCow2SubclusterType equivalent of a given QCow2ClusterType.
- */
-static inline
-QCow2SubclusterType qcow2_cluster_to_subcluster_type(QCow2ClusterType type)
-{
-    switch (type) {
-    case QCOW2_CLUSTER_COMPRESSED:
-        return QCOW2_SUBCLUSTER_COMPRESSED;
-    case QCOW2_CLUSTER_ZERO_PLAIN:
-        return QCOW2_SUBCLUSTER_ZERO_PLAIN;
-    case QCOW2_CLUSTER_ZERO_ALLOC:
-        return QCOW2_SUBCLUSTER_ZERO_ALLOC;
-    case QCOW2_CLUSTER_NORMAL:
-        return QCOW2_SUBCLUSTER_NORMAL;
-    case QCOW2_CLUSTER_UNALLOCATED:
-        return QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN;
-    default:
-        g_assert_not_reached();
-    }
-}
-
 /*
  * In an image without subsclusters @l2_bitmap is ignored and
  * @sc_index must be 0.
@@ -XXX,XX +XXX,XX @@ QCow2SubclusterType qcow2_get_subcluster_type(BlockDriverState *bs,
             g_assert_not_reached();
         }
     } else {
-        return qcow2_cluster_to_subcluster_type(type);
+        switch (type) {
+        case QCOW2_CLUSTER_COMPRESSED:
+            return QCOW2_SUBCLUSTER_COMPRESSED;
+        case QCOW2_CLUSTER_ZERO_PLAIN:
+            return QCOW2_SUBCLUSTER_ZERO_PLAIN;
+        case QCOW2_CLUSTER_ZERO_ALLOC:
+            return QCOW2_SUBCLUSTER_ZERO_ALLOC;
+        case QCOW2_CLUSTER_NORMAL:
+            return QCOW2_SUBCLUSTER_NORMAL;
+        case QCOW2_CLUSTER_UNALLOCATED:
+            return QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN;
+        default:
+            g_assert_not_reached();
+        }
     }
 }
 
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int qcow2_get_subcluster_range_type(BlockDriverState *bs,
 }
 
 /*
- * Checks how many clusters in a given L2 slice are contiguous in the image
- * file. As soon as one of the flags in the bitmask stop_flags changes compared
- * to the first cluster, the search is stopped and the cluster is not counted
- * as contiguous. (This allows it, for example, to stop at the first compressed
- * cluster which may require a different handling)
+ * Return the number of contiguous subclusters of the exact same type
+ * in a given L2 slice, starting from cluster @l2_index, subcluster
+ * @sc_index. Allocated subclusters are required to be contiguous in
+ * the image file.
+ * At most @nb_clusters are checked (note that this means clusters,
+ * not subclusters).
+ * Compressed clusters are always processed one by one but for the
+ * purpose of this count they are treated as if they were divided into
+ * subclusters of size s->subcluster_size.
+ * On failure return -errno and update @l2_index to point to the
+ * invalid entry.
  */
-static int count_contiguous_clusters(BlockDriverState *bs, int nb_clusters,
-        int cluster_size, uint64_t *l2_slice, int l2_index, uint64_t stop_flags)
+static int count_contiguous_subclusters(BlockDriverState *bs, int nb_clusters,
+                                        unsigned sc_index, uint64_t *l2_slice,
+                                        unsigned *l2_index)
 {
     BDRVQcow2State *s = bs->opaque;
-    int i;
-    QCow2ClusterType first_cluster_type;
-    uint64_t mask = stop_flags | L2E_OFFSET_MASK | QCOW_OFLAG_COMPRESSED;
-    uint64_t first_entry = get_l2_entry(s, l2_slice, l2_index);
-    uint64_t offset = first_entry & mask;
+    int i, count = 0;
+    bool check_offset = false;
+    uint64_t expected_offset = 0;
+    QCow2SubclusterType expected_type = QCOW2_SUBCLUSTER_NORMAL, type;
 
-    first_cluster_type = qcow2_get_cluster_type(bs, first_entry);
-    if (first_cluster_type == QCOW2_CLUSTER_UNALLOCATED) {
-        return 0;
-    }
-
-    /* must be allocated */
-    assert(first_cluster_type == QCOW2_CLUSTER_NORMAL ||
-           first_cluster_type == QCOW2_CLUSTER_ZERO_ALLOC);
+    assert(*l2_index + nb_clusters <= s->l2_slice_size);
 
     for (i = 0; i < nb_clusters; i++) {
-        uint64_t l2_entry = get_l2_entry(s, l2_slice, l2_index + i) & mask;
-        if (offset + (uint64_t) i * cluster_size != l2_entry) {
+        unsigned first_sc = (i == 0) ? sc_index : 0;
+        uint64_t l2_entry = get_l2_entry(s, l2_slice, *l2_index + i);
+        uint64_t l2_bitmap = get_l2_bitmap(s, l2_slice, *l2_index + i);
+        int ret = qcow2_get_subcluster_range_type(bs, l2_entry, l2_bitmap,
+                                                  first_sc, &type);
+        if (ret < 0) {
+            *l2_index += i; /* Point to the invalid entry */
+            return -EIO;
+        }
+        if (i == 0) {
+            if (type == QCOW2_SUBCLUSTER_COMPRESSED) {
+                /* Compressed clusters are always processed one by one */
+                return ret;
+            }
+            expected_type = type;
+            expected_offset = l2_entry & L2E_OFFSET_MASK;
+            check_offset = (type == QCOW2_SUBCLUSTER_NORMAL ||
+                            type == QCOW2_SUBCLUSTER_ZERO_ALLOC ||
+                            type == QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC);
+        } else if (type != expected_type) {
             break;
+        } else if (check_offset) {
+            expected_offset += s->cluster_size;
+            if (expected_offset != (l2_entry & L2E_OFFSET_MASK)) {
+                break;
+            }
         }
-    }
-
-        return i;
-}
-
-/*
- * Checks how many consecutive unallocated clusters in a given L2
- * slice have the same cluster type.
- */
-static int count_contiguous_clusters_unallocated(BlockDriverState *bs,
-                                                 int nb_clusters,
-                                                 uint64_t *l2_slice,
-                                                 int l2_index,
-                                                 QCow2ClusterType wanted_type)
-{
-    BDRVQcow2State *s = bs->opaque;
-    int i;
-
-    assert(wanted_type == QCOW2_CLUSTER_ZERO_PLAIN ||
-           wanted_type == QCOW2_CLUSTER_UNALLOCATED);
-    for (i = 0; i < nb_clusters; i++) {
-        uint64_t entry = get_l2_entry(s, l2_slice, l2_index + i);
-        QCow2ClusterType type = qcow2_get_cluster_type(bs, entry);
-
-        if (type != wanted_type) {
+        count += ret;
+        /* Stop if there are type changes before the end of the cluster */
+        if (first_sc + ret < s->subclusters_per_cluster) {
             break;
         }
     }
 
-    return i;
+    return count;
 }
 
 static int coroutine_fn do_perform_cow_read(BlockDriverState *bs,
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
                           QCow2SubclusterType *subcluster_type)
 {
     BDRVQcow2State *s = bs->opaque;
-    unsigned int l2_index;
-    uint64_t l1_index, l2_offset, *l2_slice, l2_entry;
-    int c;
+    unsigned int l2_index, sc_index;
+    uint64_t l1_index, l2_offset, *l2_slice, l2_entry, l2_bitmap;
+    int sc;
     unsigned int offset_in_cluster;
     uint64_t bytes_available, bytes_needed, nb_clusters;
-    QCow2ClusterType type;
+    QCow2SubclusterType type;
     int ret;
 
     offset_in_cluster = offset_into_cluster(s, offset);
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
 
     l1_index = offset_to_l1_index(s, offset);
     if (l1_index >= s->l1_size) {
-        type = QCOW2_CLUSTER_UNALLOCATED;
+        type = QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN;
         goto out;
     }
 
     l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
     if (!l2_offset) {
-        type = QCOW2_CLUSTER_UNALLOCATED;
+        type = QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN;
         goto out;
     }
 
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
     /* find the cluster offset for the given disk offset */
 
     l2_index = offset_to_l2_slice_index(s, offset);
+    sc_index = offset_to_sc_index(s, offset);
     l2_entry = get_l2_entry(s, l2_slice, l2_index);
+    l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index);
 
     nb_clusters = size_to_clusters(s, bytes_needed);
     /* bytes_needed <= *bytes + offset_in_cluster, both of which are unsigned
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
      * true */
     assert(nb_clusters <= INT_MAX);
 
-    type = qcow2_get_cluster_type(bs, l2_entry);
-    if (s->qcow_version < 3 && (type == QCOW2_CLUSTER_ZERO_PLAIN ||
-                                type == QCOW2_CLUSTER_ZERO_ALLOC)) {
+    type = qcow2_get_subcluster_type(bs, l2_entry, l2_bitmap, sc_index);
+    if (s->qcow_version < 3 && (type == QCOW2_SUBCLUSTER_ZERO_PLAIN ||
+                                type == QCOW2_SUBCLUSTER_ZERO_ALLOC)) {
         qcow2_signal_corruption(bs, true, -1, -1, "Zero cluster entry found"
                                 " in pre-v3 image (L2 offset: %#" PRIx64
                                 ", L2 index: %#x)", l2_offset, l2_index);
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
         goto fail;
     }
     switch (type) {
-    case QCOW2_CLUSTER_COMPRESSED:
+    case QCOW2_SUBCLUSTER_INVALID:
+        break; /* This is handled by count_contiguous_subclusters() below */
+    case QCOW2_SUBCLUSTER_COMPRESSED:
         if (has_data_file(bs)) {
             qcow2_signal_corruption(bs, true, -1, -1, "Compressed cluster "
                                     "entry found in image with external data "
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
             ret = -EIO;
             goto fail;
         }
-        /* Compressed clusters can only be processed one by one */
-        c = 1;
         *host_offset = l2_entry & L2E_COMPRESSED_OFFSET_SIZE_MASK;
         break;
-    case QCOW2_CLUSTER_ZERO_PLAIN:
-    case QCOW2_CLUSTER_UNALLOCATED:
-        /* how many empty clusters ? */
-        c = count_contiguous_clusters_unallocated(bs, nb_clusters,
-                                                  l2_slice, l2_index, type);
+    case QCOW2_SUBCLUSTER_ZERO_PLAIN:
+    case QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN:
         break;
-    case QCOW2_CLUSTER_ZERO_ALLOC:
-    case QCOW2_CLUSTER_NORMAL: {
+    case QCOW2_SUBCLUSTER_ZERO_ALLOC:
+    case QCOW2_SUBCLUSTER_NORMAL:
+    case QCOW2_SUBCLUSTER_UNALLOCATED_ALLOC: {
         uint64_t host_cluster_offset = l2_entry & L2E_OFFSET_MASK;
         *host_offset = host_cluster_offset + offset_in_cluster;
-        /* how many allocated clusters ? */
-        c = count_contiguous_clusters(bs, nb_clusters, s->cluster_size,
-                                      l2_slice, l2_index, QCOW_OFLAG_ZERO);
         if (offset_into_cluster(s, host_cluster_offset)) {
             qcow2_signal_corruption(bs, true, -1, -1,
                                     "Cluster allocation offset %#"
@@ -XXX,XX +XXX,XX @@ int qcow2_get_host_offset(BlockDriverState *bs, uint64_t offset,
         abort();
     }
 
+    sc = count_contiguous_subclusters(bs, nb_clusters, sc_index,
+                                      l2_slice, &l2_index);
+    if (sc < 0) {
+        qcow2_signal_corruption(bs, true, -1, -1, "Invalid cluster entry found "
+                                " (L2 offset: %#" PRIx64 ", L2 index: %#x)",
+                                l2_offset, l2_index);
+        ret = -EIO;
+        goto fail;
+    }
     qcow2_cache_put(s->l2_table_cache, (void **) &l2_slice);
 
-    bytes_available = (int64_t)c * s->cluster_size;
+    bytes_available = ((int64_t)sc + sc_index) << s->subcluster_bits;
 
 out:
     if (bytes_available > bytes_needed) {
@@ -XXX,XX +XXX,XX @@ out:
     assert(bytes_available - offset_in_cluster <= UINT_MAX);
     *bytes = bytes_available - offset_in_cluster;
 
-    *subcluster_type = qcow2_cluster_to_subcluster_type(type);
+    *subcluster_type = type;
 
     return 0;
 
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

The QCOW_OFLAG_ZERO bit that indicates that a cluster reads as
zeroes is only used in standard L2 entries. Extended L2 entries use
individual 'all zeroes' bits for each subcluster.

This must be taken into account when updating the L2 entry and also
when deciding that an existing entry does not need to be updated.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <b61d61606d8c9b367bd641ab37351ddb9172799a.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 38 ++++++++++++++++++++------------------
 1 file changed, 20 insertions(+), 18 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int zero_in_l2_slice(BlockDriverState *bs, uint64_t offset,
     int l2_index;
     int ret;
     int i;
-    bool unmap = !!(flags & BDRV_REQ_MAY_UNMAP);
 
     ret = get_cluster_table(bs, offset, &l2_slice, &l2_index);
     if (ret < 0) {
@@ -XXX,XX +XXX,XX @@ static int zero_in_l2_slice(BlockDriverState *bs, uint64_t offset,
     assert(nb_clusters <= INT_MAX);
 
     for (i = 0; i < nb_clusters; i++) {
-        uint64_t old_offset;
-        QCow2ClusterType cluster_type;
-
-        old_offset = get_l2_entry(s, l2_slice, l2_index + i);
+        uint64_t old_l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
+        uint64_t old_l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index + i);
+        QCow2ClusterType type = qcow2_get_cluster_type(bs, old_l2_entry);
+        bool unmap = (type == QCOW2_CLUSTER_COMPRESSED) ||
+            ((flags & BDRV_REQ_MAY_UNMAP) && qcow2_cluster_is_allocated(type));
+        uint64_t new_l2_entry = unmap ? 0 : old_l2_entry;
+        uint64_t new_l2_bitmap = old_l2_bitmap;
+
+        if (has_subclusters(s)) {
+            new_l2_bitmap = QCOW_L2_BITMAP_ALL_ZEROES;
+        } else {
+            new_l2_entry |= QCOW_OFLAG_ZERO;
+        }
 
-        /*
-         * Minimize L2 changes if the cluster already reads back as
-         * zeroes with correct allocation.
-         */
-        cluster_type = qcow2_get_cluster_type(bs, old_offset);
-        if (cluster_type == QCOW2_CLUSTER_ZERO_PLAIN ||
-            (cluster_type == QCOW2_CLUSTER_ZERO_ALLOC && !unmap)) {
+        if (old_l2_entry == new_l2_entry && old_l2_bitmap == new_l2_bitmap) {
             continue;
         }
 
         qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
-        if (cluster_type == QCOW2_CLUSTER_COMPRESSED || unmap) {
-            set_l2_entry(s, l2_slice, l2_index + i, QCOW_OFLAG_ZERO);
-            qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
-        } else {
-            uint64_t entry = get_l2_entry(s, l2_slice, l2_index + i);
-            set_l2_entry(s, l2_slice, l2_index + i, entry | QCOW_OFLAG_ZERO);
+        if (unmap) {
+            qcow2_free_any_clusters(bs, old_l2_entry, 1, QCOW2_DISCARD_REQUEST);
+        }
+        set_l2_entry(s, l2_slice, l2_index + i, new_l2_entry);
+        if (has_subclusters(s)) {
+            set_l2_bitmap(s, l2_slice, l2_index + i, new_l2_bitmap);
         }
     }
 
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Two things need to be taken into account here:

1) With full_discard == true the L2 entry must be cleared completely.
   This also includes the L2 bitmap if the image has extended L2
   entries.

2) With full_discard == false we have to make the discarded cluster
   read back as zeroes. With normal L2 entries this is done with the
   QCOW_OFLAG_ZERO bit, whereas with extended L2 entries this is done
   with the individual 'all zeroes' bits for each subcluster.

Note however that QCOW_OFLAG_ZERO is not supported in v2 qcow2
   images so, if there is a backing file, discard cannot guarantee
   that the image will read back as zeroes. If this is important for
   the caller it should forbid it as qcow2_co_pdiscard() does (see
   80f5c01183 for more details).

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <5ef8274e628aa3ab559bfac467abf488534f2b76.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 52 +++++++++++++++++++------------------------
 1 file changed, 23 insertions(+), 29 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int discard_in_l2_slice(BlockDriverState *bs, uint64_t offset,
     assert(nb_clusters <= INT_MAX);
 
     for (i = 0; i < nb_clusters; i++) {
-        uint64_t old_l2_entry;
-
-        old_l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
+        uint64_t old_l2_entry = get_l2_entry(s, l2_slice, l2_index + i);
+        uint64_t old_l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index + i);
+        uint64_t new_l2_entry = old_l2_entry;
+        uint64_t new_l2_bitmap = old_l2_bitmap;
+        QCow2ClusterType cluster_type =
+            qcow2_get_cluster_type(bs, old_l2_entry);
 
         /*
+         * If full_discard is true, the cluster should not read back as zeroes,
+         * but rather fall through to the backing file.
+         *
          * If full_discard is false, make sure that a discarded area reads back
          * as zeroes for v3 images (we cannot do it for v2 without actually
          * writing a zero-filled buffer). We can skip the operation if the
@@ -XXX,XX +XXX,XX @@ static int discard_in_l2_slice(BlockDriverState *bs, uint64_t offset,
          *
          * TODO We might want to use bdrv_block_status(bs) here, but we're
          * holding s->lock, so that doesn't work today.
-         *
-         * If full_discard is true, the sector should not read back as zeroes,
-         * but rather fall through to the backing file.
          */
-        switch (qcow2_get_cluster_type(bs, old_l2_entry)) {
-        case QCOW2_CLUSTER_UNALLOCATED:
-            if (full_discard || !bs->backing) {
-                continue;
-            }
-            break;
-
-        case QCOW2_CLUSTER_ZERO_PLAIN:
-            if (!full_discard) {
-                continue;
+        if (full_discard) {
+            new_l2_entry = new_l2_bitmap = 0;
+        } else if (bs->backing || qcow2_cluster_is_allocated(cluster_type)) {
+            if (has_subclusters(s)) {
+                new_l2_entry = 0;
+                new_l2_bitmap = QCOW_L2_BITMAP_ALL_ZEROES;
+            } else {
+                new_l2_entry = s->qcow_version >= 3 ? QCOW_OFLAG_ZERO : 0;
             }
-            break;
-
-        case QCOW2_CLUSTER_ZERO_ALLOC:
-        case QCOW2_CLUSTER_NORMAL:
-        case QCOW2_CLUSTER_COMPRESSED:
-            break;
+        }
 
-        default:
-            abort();
+        if (old_l2_entry == new_l2_entry && old_l2_bitmap == new_l2_bitmap) {
+            continue;
         }
 
         /* First remove L2 entries */
         qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
-        if (!full_discard && s->qcow_version >= 3) {
-            set_l2_entry(s, l2_slice, l2_index + i, QCOW_OFLAG_ZERO);
-        } else {
-            set_l2_entry(s, l2_slice, l2_index + i, 0);
+        set_l2_entry(s, l2_slice, l2_index + i, new_l2_entry);
+        if (has_subclusters(s)) {
+            set_l2_bitmap(s, l2_slice, l2_index + i, new_l2_bitmap);
         }
-
         /* Then decrease the refcount */
         qcow2_free_any_clusters(bs, old_l2_entry, 1, type);
     }
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

The offset field of an uncompressed cluster's L2 entry must be aligned
to the cluster size, otherwise it is invalid. If the cluster has no
data then it means that the offset points to a preallocation, so we
can clear the offset field without affecting the guest-visible data.
This is what 'qemu-img check' does when run in repair mode.

On traditional qcow2 images this can only happen when QCOW_OFLAG_ZERO
is set, and repairing such entries turns the clusters from ZERO_ALLOC
into ZERO_PLAIN.

Extended L2 entries have no ZERO_ALLOC clusters and no QCOW_OFLAG_ZERO
but the idea is the same: if none of the subclusters are allocated
then we can clear the offset field and leave the bitmap untouched.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <9f4ed1d0a34b0a545b032c31ecd8c14734065342.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-refcount.c     | 16 +++++++++++-----
 tests/qemu-iotests/060.out |  2 +-
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -XXX,XX +XXX,XX @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
 
             /* Correct offsets are cluster aligned */
             if (offset_into_cluster(s, offset)) {
+                bool contains_data;
                 res->corruptions++;
 
-                if (qcow2_get_cluster_type(bs, l2_entry) ==
-                    QCOW2_CLUSTER_ZERO_ALLOC)
-                {
-                    fprintf(stderr, "%s offset=%" PRIx64 ": Preallocated zero "
+                if (has_subclusters(s)) {
+                    uint64_t l2_bitmap = get_l2_bitmap(s, l2_table, i);
+                    contains_data = (l2_bitmap & QCOW_L2_BITMAP_ALL_ALLOC);
+                } else {
+                    contains_data = !(l2_entry & QCOW_OFLAG_ZERO);
+                }
+
+                if (!contains_data) {
+                    fprintf(stderr, "%s offset=%" PRIx64 ": Preallocated "
                             "cluster is not properly aligned; L2 entry "
                             "corrupted.\n",
                             fix & BDRV_FIX_ERRORS ? "Repairing" : "ERROR",
@@ -XXX,XX +XXX,XX @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
                         int ign = active ? QCOW2_OL_ACTIVE_L2 :
                                            QCOW2_OL_INACTIVE_L2;
 
-                        l2_entry = QCOW_OFLAG_ZERO;
+                        l2_entry = has_subclusters(s) ? 0 : QCOW_OFLAG_ZERO;
                         set_l2_entry(s, l2_table, i, l2_entry);
                         ret = qcow2_pre_write_overlap_check(bs, ign,
                                 l2e_offset, l2_entry_size(s), false);
diff --git a/tests/qemu-iotests/060.out b/tests/qemu-iotests/060.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/060.out
+++ b/tests/qemu-iotests/060.out
@@ -XXX,XX +XXX,XX @@ discard 65536/65536 bytes at offset 0
 qcow2: Marking image as corrupt: Preallocated zero cluster offset 0x2a00 unaligned (guest offset: 0); further corruption events will be suppressed
 write failed: Input/output error
 --- Repairing ---
-Repairing offset=2a00: Preallocated zero cluster is not properly aligned; L2 entry corrupted.
+Repairing offset=2a00: Preallocated cluster is not properly aligned; L2 entry corrupted.
 The following inconsistencies were found and repaired:
 
     0 leaked clusters
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

The L2 bitmap needs to be updated after each write to indicate what
new subclusters are now allocated. This needs to happen even if the
cluster was already allocated and the L2 entry was otherwise valid.

In some cases however a write operation doesn't need change the L2
bitmap (because all affected subclusters were already allocated). This
is detected in calculate_l2_meta(), and qcow2_alloc_cluster_link_l2()
is never called in those cases.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <0875620d49f44320334b6a91c73b3f301f975f38.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
         assert((offset & L2E_OFFSET_MASK) == offset);
 
         set_l2_entry(s, l2_slice, l2_index + i, offset | QCOW_OFLAG_COPIED);
+
+        /* Update bitmap with the subclusters that were just written */
+        if (has_subclusters(s)) {
+            uint64_t l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index + i);
+            unsigned written_from = m->cow_start.offset;
+            unsigned written_to = m->cow_end.offset + m->cow_end.nb_bytes ?:
+                m->nb_clusters << s->cluster_bits;
+            int first_sc, last_sc;
+            /* Narrow written_from and written_to down to the current cluster */
+            written_from = MAX(written_from, i << s->cluster_bits);
+            written_to   = MIN(written_to, (i + 1) << s->cluster_bits);
+            assert(written_from < written_to);
+            first_sc = offset_to_sc_index(s, written_from);
+            last_sc  = offset_to_sc_index(s, written_to - 1);
+            l2_bitmap |= QCOW_OFLAG_SUB_ALLOC_RANGE(first_sc, last_sc + 1);
+            l2_bitmap &= ~QCOW_OFLAG_SUB_ZERO_RANGE(first_sc, last_sc + 1);
+            set_l2_bitmap(s, l2_slice, l2_index + i, l2_bitmap);
+        }
      }
 
 
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Compressed clusters always have the bitmap part of the extended L2
entry set to 0.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <04455b3de5dfeb9d1cfe1fc7b02d7060a6e09710.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ int qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
     BLKDBG_EVENT(bs->file, BLKDBG_L2_UPDATE_COMPRESSED);
     qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
     set_l2_entry(s, l2_slice, l2_index, cluster_offset);
+    if (has_subclusters(s)) {
+        set_l2_bitmap(s, l2_slice, l2_index, 0);
+    }
     qcow2_cache_put(s->l2_table_cache, (void **) &l2_slice);
 
     *host_offset = cluster_offset & s->cluster_offset_mask;
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

The bdrv_co_pwrite_zeroes() call here fills complete clusters with
zeroes, but it can happen that some subclusters are not part of the
write request or the copy-on-write. This patch makes sure that only
the affected subclusters are overwritten.

A potential improvement would be to also fill with zeroes the other
subclusters if we can guarantee that we are not overwriting existing
data. However this would waste more disk space, so we should first
evaluate if it's really worth doing.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <b3dc97e8e2240ddb5191a4f930e8fc9653f94621.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static int handle_alloc_space(BlockDriverState *bs, QCowL2Meta *l2meta)
 
     for (m = l2meta; m != NULL; m = m->next) {
         int ret;
+        uint64_t start_offset = m->alloc_offset + m->cow_start.offset;
+        unsigned nb_bytes = m->cow_end.offset + m->cow_end.nb_bytes -
+            m->cow_start.offset;
 
         if (!m->cow_start.nb_bytes && !m->cow_end.nb_bytes) {
             continue;
@@ -XXX,XX +XXX,XX @@ static int handle_alloc_space(BlockDriverState *bs, QCowL2Meta *l2meta)
          * efficiently zero out the whole clusters
          */
 
-        ret = qcow2_pre_write_overlap_check(bs, 0, m->alloc_offset,
-                                            m->nb_clusters * s->cluster_size,
+        ret = qcow2_pre_write_overlap_check(bs, 0, start_offset, nb_bytes,
                                             true);
         if (ret < 0) {
             return ret;
         }
 
         BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC_SPACE);
-        ret = bdrv_co_pwrite_zeroes(s->data_file, m->alloc_offset,
-                                    m->nb_clusters * s->cluster_size,
+        ret = bdrv_co_pwrite_zeroes(s->data_file, start_offset, nb_bytes,
                                     BDRV_REQ_NO_FALLBACK);
         if (ret < 0) {
             if (ret != -ENOTSUP && ret != -EAGAIN) {
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This works now at the subcluster level and pwrite_zeroes_alignment is
updated accordingly.

qcow2_cluster_zeroize() is turned into qcow2_subcluster_zeroize() with
the following changes:

- The request can now be subcluster-aligned.

- The cluster-aligned body of the request is still zeroized using
     zero_in_l2_slice() as before.

- The subcluster-aligned head and tail of the request are zeroized
     with the new zero_l2_subclusters() function.

There is just one thing to take into account for a possible future
improvement: compressed clusters cannot be partially zeroized so
zero_l2_subclusters() on the head or the tail can return -ENOTSUP.
This makes the caller repeat the *complete* request and write actual
zeroes to disk. This is sub-optimal because

1) if the head area was compressed we would still be able to use
      the fast path for the body and possibly the tail.

2) if the tail area was compressed we are writing zeroes to the
      head and the body areas, which are already zeroized.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <17e05e2ee7e12f10dcf012da81e83ebe27eb3bef.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h         |  4 +--
 block/qcow2-cluster.c | 81 +++++++++++++++++++++++++++++++++++++++----
 block/qcow2.c         | 33 +++++++++---------
 3 files changed, 94 insertions(+), 24 deletions(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ void qcow2_alloc_cluster_abort(BlockDriverState *bs, QCowL2Meta *m);
 int qcow2_cluster_discard(BlockDriverState *bs, uint64_t offset,
                           uint64_t bytes, enum qcow2_discard_type type,
                           bool full_discard);
-int qcow2_cluster_zeroize(BlockDriverState *bs, uint64_t offset,
-                          uint64_t bytes, int flags);
+int qcow2_subcluster_zeroize(BlockDriverState *bs, uint64_t offset,
+                             uint64_t bytes, int flags);
 
 int qcow2_expand_zero_clusters(BlockDriverState *bs,
                                BlockDriverAmendStatusCB *status_cb,
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int zero_in_l2_slice(BlockDriverState *bs, uint64_t offset,
     return nb_clusters;
 }
 
-int qcow2_cluster_zeroize(BlockDriverState *bs, uint64_t offset,
-                          uint64_t bytes, int flags)
+static int zero_l2_subclusters(BlockDriverState *bs, uint64_t offset,
+                               unsigned nb_subclusters)
+{
+    BDRVQcow2State *s = bs->opaque;
+    uint64_t *l2_slice;
+    uint64_t old_l2_bitmap, l2_bitmap;
+    int l2_index, ret, sc = offset_to_sc_index(s, offset);
+
+    /* For full clusters use zero_in_l2_slice() instead */
+    assert(nb_subclusters > 0 && nb_subclusters < s->subclusters_per_cluster);
+    assert(sc + nb_subclusters <= s->subclusters_per_cluster);
+    assert(offset_into_subcluster(s, offset) == 0);
+
+    ret = get_cluster_table(bs, offset, &l2_slice, &l2_index);
+    if (ret < 0) {
+        return ret;
+    }
+
+    switch (qcow2_get_cluster_type(bs, get_l2_entry(s, l2_slice, l2_index))) {
+    case QCOW2_CLUSTER_COMPRESSED:
+        ret = -ENOTSUP; /* We cannot partially zeroize compressed clusters */
+        goto out;
+    case QCOW2_CLUSTER_NORMAL:
+    case QCOW2_CLUSTER_UNALLOCATED:
+        break;
+    default:
+        g_assert_not_reached();
+    }
+
+    old_l2_bitmap = l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index);
+
+    l2_bitmap |=  QCOW_OFLAG_SUB_ZERO_RANGE(sc, sc + nb_subclusters);
+    l2_bitmap &= ~QCOW_OFLAG_SUB_ALLOC_RANGE(sc, sc + nb_subclusters);
+
+    if (old_l2_bitmap != l2_bitmap) {
+        set_l2_bitmap(s, l2_slice, l2_index, l2_bitmap);
+        qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_slice);
+    }
+
+    ret = 0;
+out:
+    qcow2_cache_put(s->l2_table_cache, (void **) &l2_slice);
+
+    return ret;
+}
+
+int qcow2_subcluster_zeroize(BlockDriverState *bs, uint64_t offset,
+                             uint64_t bytes, int flags)
 {
     BDRVQcow2State *s = bs->opaque;
     uint64_t end_offset = offset + bytes;
     uint64_t nb_clusters;
+    unsigned head, tail;
     int64_t cleared;
     int ret;
 
@@ -XXX,XX +XXX,XX @@ int qcow2_cluster_zeroize(BlockDriverState *bs, uint64_t offset,
     }
 
     /* Caller must pass aligned values, except at image end */
-    assert(QEMU_IS_ALIGNED(offset, s->cluster_size));
-    assert(QEMU_IS_ALIGNED(end_offset, s->cluster_size) ||
+    assert(offset_into_subcluster(s, offset) == 0);
+    assert(offset_into_subcluster(s, end_offset) == 0 ||
            end_offset >= bs->total_sectors << BDRV_SECTOR_BITS);
 
     /*
@@ -XXX,XX +XXX,XX @@ int qcow2_cluster_zeroize(BlockDriverState *bs, uint64_t offset,
         return -ENOTSUP;
     }
 
-    /* Each L2 slice is handled by its own loop iteration */
-    nb_clusters = size_to_clusters(s, bytes);
+    head = MIN(end_offset, ROUND_UP(offset, s->cluster_size)) - offset;
+    offset += head;
+
+    tail = (end_offset >= bs->total_sectors << BDRV_SECTOR_BITS) ? 0 :
+        end_offset - MAX(offset, start_of_cluster(s, end_offset));
+    end_offset -= tail;
 
     s->cache_discards = true;
 
+    if (head) {
+        ret = zero_l2_subclusters(bs, offset - head,
+                                  size_to_subclusters(s, head));
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
+    /* Each L2 slice is handled by its own loop iteration */
+    nb_clusters = size_to_clusters(s, end_offset - offset);
+
     while (nb_clusters > 0) {
         cleared = zero_in_l2_slice(bs, offset, nb_clusters, flags);
         if (cleared < 0) {
@@ -XXX,XX +XXX,XX @@ int qcow2_cluster_zeroize(BlockDriverState *bs, uint64_t offset,
         offset += (cleared * s->cluster_size);
     }
 
+    if (tail) {
+        ret = zero_l2_subclusters(bs, end_offset, size_to_subclusters(s, tail));
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
     ret = 0;
 fail:
     s->cache_discards = false;
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static void qcow2_refresh_limits(BlockDriverState *bs, Error **errp)
         /* Encryption works on a sector granularity */
         bs->bl.request_alignment = qcrypto_block_get_sector_size(s->crypto);
     }
-    bs->bl.pwrite_zeroes_alignment = s->cluster_size;
+    bs->bl.pwrite_zeroes_alignment = s->subcluster_size;
     bs->bl.pdiscard_alignment = s->cluster_size;
 }
 
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
     int ret;
     BDRVQcow2State *s = bs->opaque;
 
-    uint32_t head = offset % s->cluster_size;
-    uint32_t tail = (offset + bytes) % s->cluster_size;
+    uint32_t head = offset_into_subcluster(s, offset);
+    uint32_t tail = ROUND_UP(offset + bytes, s->subcluster_size) -
+        (offset + bytes);
 
     trace_qcow2_pwrite_zeroes_start_req(qemu_coroutine_self(), offset, bytes);
     if (offset + bytes == bs->total_sectors * BDRV_SECTOR_SIZE) {
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
         unsigned int nr;
         QCow2SubclusterType type;
 
-        assert(head + bytes <= s->cluster_size);
+        assert(head + bytes + tail <= s->subcluster_size);
 
         /* check whether remainder of cluster already reads as zero */
         if (!(is_zero(bs, offset - head, head) &&
-              is_zero(bs, offset + bytes,
-                      tail ? s->cluster_size - tail : 0))) {
+              is_zero(bs, offset + bytes, tail))) {
             return -ENOTSUP;
         }
 
         qemu_co_mutex_lock(&s->lock);
         /* We can have new write after previous check */
-        offset = QEMU_ALIGN_DOWN(offset, s->cluster_size);
-        bytes = s->cluster_size;
-        nr = s->cluster_size;
+        offset -= head;
+        bytes = s->subcluster_size;
+        nr = s->subcluster_size;
         ret = qcow2_get_host_offset(bs, offset, &nr, &off, &type);
         if (ret < 0 ||
             (type != QCOW2_SUBCLUSTER_UNALLOCATED_PLAIN &&
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
 
     trace_qcow2_pwrite_zeroes(qemu_coroutine_self(), offset, bytes);
 
-    /* Whatever is left can use real zero clusters */
-    ret = qcow2_cluster_zeroize(bs, offset, bytes, flags);
+    /* Whatever is left can use real zero subclusters */
+    ret = qcow2_subcluster_zeroize(bs, offset, bytes, flags);
     qemu_co_mutex_unlock(&s->lock);
 
     return ret;
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset,
     }
 
     if ((flags & BDRV_REQ_ZERO_WRITE) && offset > old_length) {
-        uint64_t zero_start = QEMU_ALIGN_UP(old_length, s->cluster_size);
+        uint64_t zero_start = QEMU_ALIGN_UP(old_length, s->subcluster_size);
 
         /*
-         * Use zero clusters as much as we can. qcow2_cluster_zeroize()
-         * requires a cluster-aligned start. The end may be unaligned if it is
-         * at the end of the image (which it is here).
+         * Use zero clusters as much as we can. qcow2_subcluster_zeroize()
+         * requires a subcluster-aligned start. The end may be unaligned if
+         * it is at the end of the image (which it is here).
          */
         if (offset > zero_start) {
-            ret = qcow2_cluster_zeroize(bs, zero_start, offset - zero_start, 0);
+            ret = qcow2_subcluster_zeroize(bs, zero_start, offset - zero_start,
+                                           0);
             if (ret < 0) {
                 error_setg_errno(errp, -ret, "Failed to zero out new clusters");
                 goto fail;
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Extended L2 entries are bigger than normal L2 entries so this has an
impact on the amount of metadata needed for a qcow2 file.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <7efae2efd5e36b42d2570743a12576d68ce53685.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.c | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ int64_t qcow2_refcount_metadata_size(int64_t clusters, size_t cluster_size,
  * @total_size: virtual disk size in bytes
  * @cluster_size: cluster size in bytes
  * @refcount_order: refcount bits power-of-2 exponent
+ * @extended_l2: true if the image has extended L2 entries
  *
  * Returns: Total number of bytes required for the fully allocated image
  * (including metadata).
  */
 static int64_t qcow2_calc_prealloc_size(int64_t total_size,
                                         size_t cluster_size,
-                                        int refcount_order)
+                                        int refcount_order,
+                                        bool extended_l2)
 {
     int64_t meta_size = 0;
     uint64_t nl1e, nl2e;
     int64_t aligned_total_size = ROUND_UP(total_size, cluster_size);
+    size_t l2e_size = extended_l2 ? L2E_SIZE_EXTENDED : L2E_SIZE_NORMAL;
 
     /* header: 1 cluster */
     meta_size += cluster_size;
 
     /* total size of L2 tables */
     nl2e = aligned_total_size / cluster_size;
-    nl2e = ROUND_UP(nl2e, cluster_size / sizeof(uint64_t));
-    meta_size += nl2e * sizeof(uint64_t);
+    nl2e = ROUND_UP(nl2e, cluster_size / l2e_size);
+    meta_size += nl2e * l2e_size;
 
     /* total size of L1 tables */
-    nl1e = nl2e * sizeof(uint64_t) / cluster_size;
+    nl1e = nl2e * l2e_size / cluster_size;
     nl1e = ROUND_UP(nl1e, cluster_size / sizeof(uint64_t));
     meta_size += nl1e * sizeof(uint64_t);
 
@@ -XXX,XX +XXX,XX @@ static BlockMeasureInfo *qcow2_measure(QemuOpts *opts, BlockDriverState *in_bs,
     PreallocMode prealloc;
     bool has_backing_file;
     bool has_luks;
+    bool extended_l2 = false; /* Set to false until the option is added */
+    size_t l2e_size;
 
     /* Parse image creation options */
     cluster_size = qcow2_opt_get_cluster_size_del(opts, &local_err);
@@ -XXX,XX +XXX,XX @@ static BlockMeasureInfo *qcow2_measure(QemuOpts *opts, BlockDriverState *in_bs,
     virtual_size = ROUND_UP(virtual_size, cluster_size);
 
     /* Check that virtual disk size is valid */
+    l2e_size = extended_l2 ? L2E_SIZE_EXTENDED : L2E_SIZE_NORMAL;
     l2_tables = DIV_ROUND_UP(virtual_size / cluster_size,
-                             cluster_size / sizeof(uint64_t));
+                             cluster_size / l2e_size);
     if (l2_tables * sizeof(uint64_t) > QCOW_MAX_L1_SIZE) {
         error_setg(&local_err, "The image size is too large "
                                "(try using a larger cluster size)");
@@ -XXX,XX +XXX,XX @@ static BlockMeasureInfo *qcow2_measure(QemuOpts *opts, BlockDriverState *in_bs,
     }
 
     info = g_new0(BlockMeasureInfo, 1);
-    info->fully_allocated =
+    info->fully_allocated = luks_payload_size +
         qcow2_calc_prealloc_size(virtual_size, cluster_size,
-                                 ctz32(refcount_bits)) + luks_payload_size;
+                                 ctz32(refcount_bits), extended_l2);
 
     /*
      * Remove data clusters that are not required.  This overestimates the
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This field allows us to indicate that the L2 metadata update does not
come from a write request with actual data but from a preallocation
request.

For traditional images this does not make any difference, but for
images with extended L2 entries this means that the clusters are
allocated normally in the L2 table but individual subclusters are
marked as unallocated.

This will allow preallocating images that have a backing file.

There is one special case: when we resize an existing image we can
also request that the new clusters are preallocated. If the image
already had a backing file then we have to hide any possible stale
data and zero out the new clusters (see commit 955c7d6687 for more
details).

In this case the subclusters cannot be left as unallocated so the L2
bitmap must be updated.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <960d4c444a4f5a870e2b47e5da322a73cd9a2f5a.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.h         | 8 ++++++++
 block/qcow2-cluster.c | 2 +-
 block/qcow2.c         | 6 ++++++
 3 files changed, 15 insertions(+), 1 deletion(-)

diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ typedef struct QCowL2Meta
      */
     bool skip_cow;
 
+    /**
+     * Indicates that this is not a normal write request but a preallocation.
+     * If the image has extended L2 entries this means that no new individual
+     * subclusters will be marked as allocated in the L2 bitmap (but any
+     * existing contents of that bitmap will be kept).
+     */
+    bool prealloc;
+
     /**
      * The I/O vector with the data from the actual guest write request.
      * If non-NULL, this is meant to be merged together with the data
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
         set_l2_entry(s, l2_slice, l2_index + i, offset | QCOW_OFLAG_COPIED);
 
         /* Update bitmap with the subclusters that were just written */
-        if (has_subclusters(s)) {
+        if (has_subclusters(s) && !m->prealloc) {
             uint64_t l2_bitmap = get_l2_bitmap(s, l2_slice, l2_index + i);
             unsigned written_from = m->cow_start.offset;
             unsigned written_to = m->cow_end.offset + m->cow_end.nb_bytes ?:
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int qcow2_handle_l2meta(BlockDriverState *bs,
         QCowL2Meta *next;
 
         if (link_l2) {
+            assert(!l2meta->prealloc);
             ret = qcow2_alloc_cluster_link_l2(bs, l2meta);
             if (ret) {
                 goto out;
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn preallocate_co(BlockDriverState *bs, uint64_t offset,
 
         while (meta) {
             QCowL2Meta *next = meta->next;
+            meta->prealloc = true;
 
             ret = qcow2_alloc_cluster_link_l2(bs, meta);
             if (ret < 0) {
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset,
         int64_t clusters_allocated;
         int64_t old_file_size, last_cluster, new_file_size;
         uint64_t nb_new_data_clusters, nb_new_l2_tables;
+        bool subclusters_need_allocation = false;
 
         /* With a data file, preallocation means just allocating the metadata
          * and forwarding the truncate request to the data file */
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset,
                                    BDRV_REQ_ZERO_WRITE, NULL);
             if (ret >= 0) {
                 flags &= ~BDRV_REQ_ZERO_WRITE;
+                /* Ensure that we read zeroes and not backing file data */
+                subclusters_need_allocation = true;
             }
         } else {
             ret = -1;
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_truncate(BlockDriverState *bs, int64_t offset,
                     .offset       = nb_clusters << s->cluster_bits,
                     .nb_bytes     = 0,
                 },
+                .prealloc     = !subclusters_need_allocation,
             };
             qemu_co_queue_init(&allocation.dependent_requests);
 
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Now that the implementation of subclusters is complete we can finally
add the necessary options to create and read images with this feature,
which we call "extended L2 entries".

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <6476caaa73216bd05b7bb2d504a20415e1665176.1594396418.git.berto@igalia.com>
[mreitz: %s/5\.1/5.2/; fixed 302's and 303's reference output]
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 qapi/block-core.json             |   7 +++
 block/qcow2.h                    |   8 ++-
 include/block/block_int.h        |   1 +
 block/qcow2.c                    |  66 ++++++++++++++++++--
 tests/qemu-iotests/031.out       |   8 +--
 tests/qemu-iotests/036.out       |   4 +-
 tests/qemu-iotests/049.out       | 102 +++++++++++++++----------------
 tests/qemu-iotests/060.out       |   1 +
 tests/qemu-iotests/061.out       |  20 +++---
 tests/qemu-iotests/065           |  12 ++--
 tests/qemu-iotests/082.out       |  39 +++++++++---
 tests/qemu-iotests/085.out       |  38 ++++++------
 tests/qemu-iotests/144.out       |   4 +-
 tests/qemu-iotests/182.out       |   2 +-
 tests/qemu-iotests/185.out       |   8 +--
 tests/qemu-iotests/198           |   2 +
 tests/qemu-iotests/206.out       |   4 ++
 tests/qemu-iotests/242.out       |   5 ++
 tests/qemu-iotests/255.out       |   8 +--
 tests/qemu-iotests/274.out       |  49 ++++++++-------
 tests/qemu-iotests/280.out       |   2 +-
 tests/qemu-iotests/291.out       |   2 +
 tests/qemu-iotests/302.out       |   1 +
 tests/qemu-iotests/303.out       |   4 +-
 tests/qemu-iotests/common.filter |   1 +
 25 files changed, 256 insertions(+), 142 deletions(-)

diff --git a/qapi/block-core.json b/qapi/block-core.json
index XXXXXXX..XXXXXXX 100644
--- a/qapi/block-core.json
+++ b/qapi/block-core.json
@@ -XXX,XX +XXX,XX @@
 #                 standalone (read-only) raw image without looking at qcow2
 #                 metadata (since: 4.0)
 #
+# @extended-l2: true if the image has extended L2 entries; only valid for
+#               compat >= 1.1 (since 5.2)
+#
 # @lazy-refcounts: on or off; only valid for compat >= 1.1
 #
 # @corrupt: true if the image has been marked corrupt; only valid for
@@ -XXX,XX +XXX,XX @@
       'compat': 'str',
       '*data-file': 'str',
       '*data-file-raw': 'bool',
+      '*extended-l2': 'bool',
       '*lazy-refcounts': 'bool',
       '*corrupt': 'bool',
       'refcount-bits': 'int',
@@ -XXX,XX +XXX,XX @@
 # @data-file-raw: True if the external data file must stay valid as a
 #                 standalone (read-only) raw image without looking at qcow2
 #                 metadata (default: false; since: 4.0)
+# @extended-l2      True to make the image have extended L2 entries
+#                   (default: false; since 5.2)
 # @size: Size of the virtual disk in bytes
 # @version: Compatibility level (default: v3)
 # @backing-file: File name of the backing file if a backing file
@@ -XXX,XX +XXX,XX @@
   'data': { 'file':             'BlockdevRef',
             '*data-file':       'BlockdevRef',
             '*data-file-raw':   'bool',
+            '*extended-l2':     'bool',
             'size':             'size',
             '*version':         'BlockdevQcow2Version',
             '*backing-file':    'str',
diff --git a/block/qcow2.h b/block/qcow2.h
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -XXX,XX +XXX,XX @@ enum {
     QCOW2_INCOMPAT_CORRUPT_BITNR    = 1,
     QCOW2_INCOMPAT_DATA_FILE_BITNR  = 2,
     QCOW2_INCOMPAT_COMPRESSION_BITNR = 3,
+    QCOW2_INCOMPAT_EXTL2_BITNR      = 4,
     QCOW2_INCOMPAT_DIRTY            = 1 << QCOW2_INCOMPAT_DIRTY_BITNR,
     QCOW2_INCOMPAT_CORRUPT          = 1 << QCOW2_INCOMPAT_CORRUPT_BITNR,
     QCOW2_INCOMPAT_DATA_FILE        = 1 << QCOW2_INCOMPAT_DATA_FILE_BITNR,
     QCOW2_INCOMPAT_COMPRESSION      = 1 << QCOW2_INCOMPAT_COMPRESSION_BITNR,
+    QCOW2_INCOMPAT_EXTL2            = 1 << QCOW2_INCOMPAT_EXTL2_BITNR,
 
     QCOW2_INCOMPAT_MASK             = QCOW2_INCOMPAT_DIRTY
                                     | QCOW2_INCOMPAT_CORRUPT
                                     | QCOW2_INCOMPAT_DATA_FILE
-                                    | QCOW2_INCOMPAT_COMPRESSION,
+                                    | QCOW2_INCOMPAT_COMPRESSION
+                                    | QCOW2_INCOMPAT_EXTL2,
 };
 
 /* Compatible feature bits */
@@ -XXX,XX +XXX,XX @@ typedef enum QCow2MetadataOverlap {
 
 static inline bool has_subclusters(BDRVQcow2State *s)
 {
-    /* FIXME: Return false until this feature is complete */
-    return false;
+    return s->incompatible_features & QCOW2_INCOMPAT_EXTL2;
 }
 
 static inline size_t l2_entry_size(BDRVQcow2State *s)
diff --git a/include/block/block_int.h b/include/block/block_int.h
index XXXXXXX..XXXXXXX 100644
--- a/include/block/block_int.h
+++ b/include/block/block_int.h
@@ -XXX,XX +XXX,XX @@
 #define BLOCK_OPT_DATA_FILE         "data_file"
 #define BLOCK_OPT_DATA_FILE_RAW     "data_file_raw"
 #define BLOCK_OPT_COMPRESSION_TYPE  "compression_type"
+#define BLOCK_OPT_EXTL2             "extended_l2"
 
 #define BLOCK_PROBE_BUF_SIZE        512
 
diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_do_open(BlockDriverState *bs, QDict *options,
     s->subcluster_size = s->cluster_size / s->subclusters_per_cluster;
     s->subcluster_bits = ctz32(s->subcluster_size);
 
+    if (s->subcluster_size < (1 << MIN_CLUSTER_BITS)) {
+        error_setg(errp, "Unsupported subcluster size: %d", s->subcluster_size);
+        ret = -EINVAL;
+        goto fail;
+    }
+
     /* Check support for various header values */
     if (header.refcount_order > 6) {
         error_setg(errp, "Reference count entry width too large; may not "
@@ -XXX,XX +XXX,XX @@ int qcow2_update_header(BlockDriverState *bs)
                 .bit  = QCOW2_INCOMPAT_COMPRESSION_BITNR,
                 .name = "compression type",
             },
+            {
+                .type = QCOW2_FEAT_TYPE_INCOMPATIBLE,
+                .bit  = QCOW2_INCOMPAT_EXTL2_BITNR,
+                .name = "extended L2 entries",
+            },
             {
                 .type = QCOW2_FEAT_TYPE_COMPATIBLE,
                 .bit  = QCOW2_COMPAT_LAZY_REFCOUNTS_BITNR,
@@ -XXX,XX +XXX,XX @@ static int64_t qcow2_calc_prealloc_size(int64_t total_size,
     return meta_size + aligned_total_size;
 }
 
-static bool validate_cluster_size(size_t cluster_size, Error **errp)
+static bool validate_cluster_size(size_t cluster_size, bool extended_l2,
+                                  Error **errp)
 {
     int cluster_bits = ctz32(cluster_size);
     if (cluster_bits < MIN_CLUSTER_BITS || cluster_bits > MAX_CLUSTER_BITS ||
@@ -XXX,XX +XXX,XX @@ static bool validate_cluster_size(size_t cluster_size, Error **errp)
                    "%dk", 1 << MIN_CLUSTER_BITS, 1 << (MAX_CLUSTER_BITS - 10));
         return false;
     }
+
+    if (extended_l2) {
+        unsigned min_cluster_size =
+            (1 << MIN_CLUSTER_BITS) * QCOW_EXTL2_SUBCLUSTERS_PER_CLUSTER;
+        if (cluster_size < min_cluster_size) {
+            error_setg(errp, "Extended L2 entries are only supported with "
+                       "cluster sizes of at least %u bytes", min_cluster_size);
+            return false;
+        }
+    }
+
     return true;
 }
 
-static size_t qcow2_opt_get_cluster_size_del(QemuOpts *opts, Error **errp)
+static size_t qcow2_opt_get_cluster_size_del(QemuOpts *opts, bool extended_l2,
+                                             Error **errp)
 {
     size_t cluster_size;
 
     cluster_size = qemu_opt_get_size_del(opts, BLOCK_OPT_CLUSTER_SIZE,
                                          DEFAULT_CLUSTER_SIZE);
-    if (!validate_cluster_size(cluster_size, errp)) {
+    if (!validate_cluster_size(cluster_size, extended_l2, errp)) {
         return 0;
     }
     return cluster_size;
@@ -XXX,XX +XXX,XX @@ qcow2_co_create(BlockdevCreateOptions *create_options, Error **errp)
         cluster_size = DEFAULT_CLUSTER_SIZE;
     }
 
-    if (!validate_cluster_size(cluster_size, errp)) {
+    if (!qcow2_opts->has_extended_l2) {
+        qcow2_opts->extended_l2 = false;
+    }
+    if (qcow2_opts->extended_l2) {
+        if (version < 3) {
+            error_setg(errp, "Extended L2 entries are only supported with "
+                       "compatibility level 1.1 and above (use version=v3 or "
+                       "greater)");
+            ret = -EINVAL;
+            goto out;
+        }
+    }
+
+    if (!validate_cluster_size(cluster_size, qcow2_opts->extended_l2, errp)) {
         ret = -EINVAL;
         goto out;
     }
@@ -XXX,XX +XXX,XX @@ qcow2_co_create(BlockdevCreateOptions *create_options, Error **errp)
             cpu_to_be64(QCOW2_INCOMPAT_COMPRESSION);
     }
 
+    if (qcow2_opts->extended_l2) {
+        header->incompatible_features |=
+            cpu_to_be64(QCOW2_INCOMPAT_EXTL2);
+    }
+
     ret = blk_pwrite(blk, 0, header, cluster_size, 0);
     g_free(header);
     if (ret < 0) {
@@ -XXX,XX +XXX,XX @@ static int coroutine_fn qcow2_co_create_opts(BlockDriver *drv,
         { BLOCK_OPT_BACKING_FMT,        "backing-fmt" },
         { BLOCK_OPT_CLUSTER_SIZE,       "cluster-size" },
         { BLOCK_OPT_LAZY_REFCOUNTS,     "lazy-refcounts" },
+        { BLOCK_OPT_EXTL2,              "extended-l2" },
         { BLOCK_OPT_REFCOUNT_BITS,      "refcount-bits" },
         { BLOCK_OPT_ENCRYPT,            BLOCK_OPT_ENCRYPT_FORMAT },
         { BLOCK_OPT_COMPAT_LEVEL,       "version" },
@@ -XXX,XX +XXX,XX @@ static BlockMeasureInfo *qcow2_measure(QemuOpts *opts, BlockDriverState *in_bs,
     PreallocMode prealloc;
     bool has_backing_file;
     bool has_luks;
-    bool extended_l2 = false; /* Set to false until the option is added */
+    bool extended_l2;
     size_t l2e_size;
 
     /* Parse image creation options */
-    cluster_size = qcow2_opt_get_cluster_size_del(opts, &local_err);
+    extended_l2 = qemu_opt_get_bool_del(opts, BLOCK_OPT_EXTL2, false);
+
+    cluster_size = qcow2_opt_get_cluster_size_del(opts, extended_l2,
+                                                  &local_err);
     if (local_err) {
         goto err;
     }
@@ -XXX,XX +XXX,XX @@ static ImageInfoSpecific *qcow2_get_specific_info(BlockDriverState *bs,
             .corrupt            = s->incompatible_features &
                                   QCOW2_INCOMPAT_CORRUPT,
             .has_corrupt        = true,
+            .has_extended_l2    = true,
+            .extended_l2        = has_subclusters(s),
             .refcount_bits      = s->refcount_bits,
             .has_bitmaps        = !!bitmaps,
             .bitmaps            = bitmaps,
@@ -XXX,XX +XXX,XX @@ static QemuOptsList qcow2_create_opts = {
             .help = "qcow2 cluster size",                               \
             .def_value_str = stringify(DEFAULT_CLUSTER_SIZE)            \
         },                                                              \
+        {                                                               \
+            .name = BLOCK_OPT_EXTL2,                                    \
+            .type = QEMU_OPT_BOOL,                                      \
+            .help = "Extended L2 tables",                               \
+            .def_value_str = "off"                                      \
+        },                                                              \
         {                                                               \
             .name = BLOCK_OPT_PREALLOC,                                 \
             .type = QEMU_OPT_STRING,                                    \
diff --git a/tests/qemu-iotests/031.out b/tests/qemu-iotests/031.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/031.out
+++ b/tests/qemu-iotests/031.out
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 Header extension:
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 Header extension:
@@ -XXX,XX +XXX,XX @@ No errors were found on the image.
 
 magic                     0x514649fb
 version                   3
-backing_file_offset       0x210
+backing_file_offset       0x240
 backing_file_size         0x17
 cluster_bits              16
 size                      67108864
@@ -XXX,XX +XXX,XX @@ data                      'host_device'
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 Header extension:
diff --git a/tests/qemu-iotests/036.out b/tests/qemu-iotests/036.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/036.out
+++ b/tests/qemu-iotests/036.out
@@ -XXX,XX +XXX,XX @@ compatible_features       []
 autoclear_features        [63]
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 
@@ -XXX,XX +XXX,XX @@ compatible_features       []
 autoclear_features        []
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 *** done
diff --git a/tests/qemu-iotests/049.out b/tests/qemu-iotests/049.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/049.out
+++ b/tests/qemu-iotests/049.out
@@ -XXX,XX +XXX,XX @@ QA output created by 049
 == 1. Traditional size parameter ==
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1024
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1024b
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1k
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1K
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1048576 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1048576 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1G
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1073741824 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1073741824 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1T
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1099511627776 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1099511627776 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1024.0
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1024.0b
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1.5k
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1.5K
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1.5M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1572864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1572864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1.5G
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1610612736 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1610612736 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 TEST_DIR/t.qcow2 1.5T
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1649267441664 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1649267441664 lazy_refcounts=off refcount_bits=16
 
 == 2. Specifying size via -o ==
 
 qemu-img create -f qcow2 -o size=1024 TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1024b TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1k TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1K TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1M TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1048576 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1048576 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1G TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1073741824 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1073741824 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1T TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1099511627776 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1099511627776 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1024.0 TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1024.0b TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1024 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1.5k TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1.5K TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1536 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1.5M TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1572864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1572864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1.5G TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1610612736 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1610612736 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o size=1.5T TEST_DIR/t.qcow2
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1649267441664 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1649267441664 lazy_refcounts=off refcount_bits=16
 
 == 3. Invalid sizes ==
 
@@ -XXX,XX +XXX,XX @@ qemu-img: TEST_DIR/t.qcow2: The image size must be specified only once
 == Check correct interpretation of suffixes for cluster size ==
 
 qemu-img create -f qcow2 -o cluster_size=1024 TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=1024b TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=1k TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=1K TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=1M TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1048576 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1048576 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=1024.0 TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=1024.0b TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=1024 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=0.5k TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=512 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=512 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=0.5K TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=512 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=512 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o cluster_size=0.5M TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=524288 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=524288 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 == Check compat level option ==
 
 qemu-img create -f qcow2 -o compat=0.10 TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=0.10 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=0.10 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o compat=1.1 TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=1.1 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=1.1 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o compat=0.42 TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=0.42 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=0.42 lazy_refcounts=off refcount_bits=16
 qemu-img: TEST_DIR/t.qcow2: Invalid parameter '0.42'
 
 qemu-img create -f qcow2 -o compat=foobar TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=foobar lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=foobar lazy_refcounts=off refcount_bits=16
 qemu-img: TEST_DIR/t.qcow2: Invalid parameter 'foobar'
 
 == Check preallocation option ==
 
 qemu-img create -f qcow2 -o preallocation=off TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 preallocation=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off preallocation=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o preallocation=metadata TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 preallocation=metadata compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off preallocation=metadata compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o preallocation=1234 TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 preallocation=1234 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off preallocation=1234 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 qemu-img: TEST_DIR/t.qcow2: Invalid parameter '1234'
 
 == Check encryption option ==
 
 qemu-img create -f qcow2 -o encryption=off TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 encryption=off cluster_size=65536 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 encryption=off cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 --object secret,id=sec0,data=123456 -o encryption=on,encrypt.key-secret=sec0 TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 encryption=on encrypt.key-secret=sec0 cluster_size=65536 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 encryption=on encrypt.key-secret=sec0 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 == Check lazy_refcounts option (only with v3) ==
 
 qemu-img create -f qcow2 -o compat=1.1,lazy_refcounts=off TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=1.1 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=1.1 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o compat=1.1,lazy_refcounts=on TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=1.1 lazy_refcounts=on refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=1.1 lazy_refcounts=on refcount_bits=16
 
 qemu-img create -f qcow2 -o compat=0.10,lazy_refcounts=off TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=0.10 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=0.10 lazy_refcounts=off refcount_bits=16
 
 qemu-img create -f qcow2 -o compat=0.10,lazy_refcounts=on TEST_DIR/t.qcow2 64M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 compat=0.10 lazy_refcounts=on refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 compat=0.10 lazy_refcounts=on refcount_bits=16
 qemu-img: TEST_DIR/t.qcow2: Lazy refcounts only supported with compatibility level 1.1 and above (use version=v3 or greater)
 
 *** done
diff --git a/tests/qemu-iotests/060.out b/tests/qemu-iotests/060.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/060.out
+++ b/tests/qemu-iotests/060.out
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: true
+    extended l2: false
 qemu-io: can't open device TEST_DIR/t.IMGFMT: IMGFMT: Image is corrupt; cannot be opened read/write
 no file open, try 'help open'
 read 512/512 bytes at offset 0
diff --git a/tests/qemu-iotests/061.out b/tests/qemu-iotests/061.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/061.out
+++ b/tests/qemu-iotests/061.out
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 magic                     0x514649fb
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 magic                     0x514649fb
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 ERROR cluster 5 refcount=0 reference=1
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 magic                     0x514649fb
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 read 65536/65536 bytes at offset 44040192
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 ERROR cluster 5 refcount=0 reference=1
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 read 131072/131072 bytes at offset 0
@@ -XXX,XX +XXX,XX @@ Format specific information:
     data file: TEST_DIR/t.IMGFMT.data
     data file raw: false
     corrupt: false
+    extended l2: false
 No errors were found on the image.
 
 === Try changing the external data file ===
@@ -XXX,XX +XXX,XX @@ Format specific information:
     data file: foo
     data file raw: false
     corrupt: false
+    extended l2: false
 
 qemu-img: Could not open 'TEST_DIR/t.IMGFMT': 'data-file' is required for this image
 image: TEST_DIR/t.IMGFMT
@@ -XXX,XX +XXX,XX @@ Format specific information:
     refcount bits: 16
     data file raw: false
     corrupt: false
+    extended l2: false
 
 === Clearing and setting data-file-raw ===
 
@@ -XXX,XX +XXX,XX @@ Format specific information:
     data file: TEST_DIR/t.IMGFMT.data
     data file raw: true
     corrupt: false
+    extended l2: false
 No errors were found on the image.
 image: TEST_DIR/t.IMGFMT
 file format: IMGFMT
@@ -XXX,XX +XXX,XX @@ Format specific information:
     data file: TEST_DIR/t.IMGFMT.data
     data file raw: false
     corrupt: false
+    extended l2: false
 No errors were found on the image.
 qemu-img: data-file-raw cannot be set on existing images
 image: TEST_DIR/t.IMGFMT
@@ -XXX,XX +XXX,XX @@ Format specific information:
     data file: TEST_DIR/t.IMGFMT.data
     data file raw: false
     corrupt: false
+    extended l2: false
 No errors were found on the image.
 *** done
diff --git a/tests/qemu-iotests/065 b/tests/qemu-iotests/065
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/065
+++ b/tests/qemu-iotests/065
@@ -XXX,XX +XXX,XX @@ class TestQCow3NotLazy(TestQemuImgInfo):
     img_options = 'compat=1.1,lazy_refcounts=off'
     json_compare = { 'compat': '1.1', 'lazy-refcounts': False,
                      'refcount-bits': 16, 'corrupt': False,
-                     'compression-type': 'zlib' }
+                     'compression-type': 'zlib', 'extended-l2': False }
     human_compare = [ 'compat: 1.1', 'compression type: zlib',
                       'lazy refcounts: false', 'refcount bits: 16',
-                      'corrupt: false' ]
+                      'corrupt: false', 'extended l2: false' ]
 
 class TestQCow3Lazy(TestQemuImgInfo):
     '''Testing a qcow2 version 3 image with lazy refcounts enabled'''
     img_options = 'compat=1.1,lazy_refcounts=on'
     json_compare = { 'compat': '1.1', 'lazy-refcounts': True,
                      'refcount-bits': 16, 'corrupt': False,
-                     'compression-type': 'zlib' }
+                     'compression-type': 'zlib', 'extended-l2': False }
     human_compare = [ 'compat: 1.1', 'compression type: zlib',
                       'lazy refcounts: true', 'refcount bits: 16',
-                      'corrupt: false' ]
+                      'corrupt: false', 'extended l2: false' ]
 
 class TestQCow3NotLazyQMP(TestQMP):
     '''Testing a qcow2 version 3 image with lazy refcounts disabled, opening
@@ -XXX,XX +XXX,XX @@ class TestQCow3NotLazyQMP(TestQMP):
     qemu_options = 'lazy-refcounts=on'
     compare = { 'compat': '1.1', 'lazy-refcounts': False,
                 'refcount-bits': 16, 'corrupt': False,
-                'compression-type': 'zlib' }
+                'compression-type': 'zlib', 'extended-l2': False }
 
 
 class TestQCow3LazyQMP(TestQMP):
@@ -XXX,XX +XXX,XX @@ class TestQCow3LazyQMP(TestQMP):
     qemu_options = 'lazy-refcounts=off'
     compare = { 'compat': '1.1', 'lazy-refcounts': True,
                 'refcount-bits': 16, 'corrupt': False,
-                'compression-type': 'zlib' }
+                'compression-type': 'zlib', 'extended-l2': False }
 
 TestImageInfoSpecific = None
 TestQemuImgInfo = None
diff --git a/tests/qemu-iotests/082.out b/tests/qemu-iotests/082.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/082.out
+++ b/tests/qemu-iotests/082.out
@@ -XXX,XX +XXX,XX @@ QA output created by 082
 === create: Options specified more than once ===
 
 Testing: create -f foo -f qcow2 TEST_DIR/t.qcow2 128M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
 image: TEST_DIR/t.IMGFMT
 file format: IMGFMT
 virtual size: 128 MiB (134217728 bytes)
 cluster_size: 65536
 
 Testing: create -f qcow2 -o cluster_size=4k -o lazy_refcounts=on TEST_DIR/t.qcow2 128M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=4096 compression_type=zlib size=134217728 lazy_refcounts=on refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=4096 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=on refcount_bits=16
 image: TEST_DIR/t.IMGFMT
 file format: IMGFMT
 virtual size: 128 MiB (134217728 bytes)
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: true
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Testing: create -f qcow2 -o cluster_size=4k -o lazy_refcounts=on -o cluster_size=8k TEST_DIR/t.qcow2 128M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=8192 compression_type=zlib size=134217728 lazy_refcounts=on refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=8192 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=on refcount_bits=16
 image: TEST_DIR/t.IMGFMT
 file format: IMGFMT
 virtual size: 128 MiB (134217728 bytes)
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: true
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Testing: create -f qcow2 -o cluster_size=4k,cluster_size=8k TEST_DIR/t.qcow2 128M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=8192 compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=8192 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
 image: TEST_DIR/t.IMGFMT
 file format: IMGFMT
 virtual size: 128 MiB (134217728 bytes)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   size=<size>            - Virtual disk size
 
 Testing: create -f qcow2 -u -o backing_file=TEST_DIR/t.qcow2,,help -F qcow2 TEST_DIR/t.qcow2 128M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2,,help backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2,,help backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 Testing: create -f qcow2 -u -o backing_file=TEST_DIR/t.qcow2,,? -F qcow2 TEST_DIR/t.qcow2 128M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2,,? backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2,,? backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 Testing: create -f qcow2 -o backing_file=TEST_DIR/t.qcow2, -o help TEST_DIR/t.qcow2 128M
 qemu-img: Invalid option list: backing_file=TEST_DIR/t.qcow2,
@@ -XXX,XX +XXX,XX @@ Supported qcow2 options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
   refcount_bits=<num>    - Width of a reference count entry in bits
@@ -XXX,XX +XXX,XX @@ qemu-img: Format driver 'bochs' does not support image creation
 === convert: Options specified more than once ===
 
 Testing: create -f qcow2 TEST_DIR/t.qcow2 128M
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
 
 Testing: convert -f foo -f qcow2 TEST_DIR/t.qcow2 TEST_DIR/t.qcow2.base
 image: TEST_DIR/t.IMGFMT.base
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: true
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Testing: convert -O qcow2 -o cluster_size=4k -o lazy_refcounts=on -o cluster_size=8k TEST_DIR/t.qcow2 TEST_DIR/t.qcow2.base
 image: TEST_DIR/t.IMGFMT.base
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: true
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Testing: convert -O qcow2 -o cluster_size=4k,cluster_size=8k TEST_DIR/t.qcow2 TEST_DIR/t.qcow2.base
 image: TEST_DIR/t.IMGFMT.base
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
@@ -XXX,XX +XXX,XX @@ Supported qcow2 options:
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
+  extended_l2=<bool (on/off)> - Extended L2 tables
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
   refcount_bits=<num>    - Width of a reference count entry in bits
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: true
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Testing: amend -f qcow2 -o size=130M -o lazy_refcounts=off TEST_DIR/t.qcow2
 image: TEST_DIR/t.IMGFMT
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Testing: amend -f qcow2 -o size=8M -o lazy_refcounts=on -o size=132M TEST_DIR/t.qcow2
 image: TEST_DIR/t.IMGFMT
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: true
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Testing: amend -f qcow2 -o size=4M,size=148M TEST_DIR/t.qcow2
 image: TEST_DIR/t.IMGFMT
diff --git a/tests/qemu-iotests/085.out b/tests/qemu-iotests/085.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/085.out
+++ b/tests/qemu-iotests/085.out
@@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/t.IMGFMT.2', fmt=IMGFMT size=134217728
 === Create a single snapshot on virtio0 ===
 
 { 'execute': 'blockdev-snapshot-sync', 'arguments': { 'device': 'virtio0', 'snapshot-file':'TEST_DIR/1-snapshot-v0.IMGFMT', 'format': 'IMGFMT' } }
-Formatting 'TEST_DIR/1-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2.1 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/1-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2.1 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 
 === Invalid command - missing device and nodename ===
@@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/1-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compress
 === Create several transactional group snapshots ===
 
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/2-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/2-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/2-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/1-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/2-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2.2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/2-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/1-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/2-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/t.qcow2.2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/3-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/3-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/3-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/2-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/3-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/2-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/3-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/2-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/3-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/2-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/4-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/4-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/4-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/3-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/4-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/3-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/4-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/3-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/4-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/3-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/5-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/5-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/5-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/4-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/5-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/4-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/5-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/4-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/5-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/4-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/6-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/6-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/6-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/5-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/6-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/5-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/6-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/5-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/6-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/5-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/7-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/7-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/7-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/6-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/7-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/6-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/7-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/6-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/7-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/6-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/8-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/8-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/8-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/7-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/8-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/7-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/8-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/7-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/8-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/7-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/9-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/9-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/9-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/8-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/9-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/8-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/9-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/8-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/9-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/8-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'transaction', 'arguments': {'actions': [ { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio0', 'snapshot-file': 'TEST_DIR/10-snapshot-v0.IMGFMT' } }, { 'type': 'blockdev-snapshot-sync', 'data' : { 'device': 'virtio1', 'snapshot-file': 'TEST_DIR/10-snapshot-v1.IMGFMT' } } ] } }
-Formatting 'TEST_DIR/10-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/9-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
-Formatting 'TEST_DIR/10-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 backing_file=TEST_DIR/9-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/10-snapshot-v0.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/9-snapshot-v0.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/10-snapshot-v1.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 backing_file=TEST_DIR/9-snapshot-v1.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 
 === Create a couple of snapshots using blockdev-snapshot ===
diff --git a/tests/qemu-iotests/144.out b/tests/qemu-iotests/144.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/144.out
+++ b/tests/qemu-iotests/144.out
@@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=536870912
 { 'execute': 'qmp_capabilities' }
 {"return": {}}
 { 'execute': 'blockdev-snapshot-sync', 'arguments': { 'device': 'virtio0', 'snapshot-file':'TEST_DIR/tmp.IMGFMT', 'format': 'IMGFMT' } }
-Formatting 'TEST_DIR/tmp.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/tmp.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 
 === Performing block-commit on active layer ===
@@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/tmp.qcow2', fmt=qcow2 cluster_size=65536 compression_type=z
 === Performing Live Snapshot 2 ===
 
 { 'execute': 'blockdev-snapshot-sync', 'arguments': { 'device': 'virtio0', 'snapshot-file':'TEST_DIR/tmp2.IMGFMT', 'format': 'IMGFMT' } }
-Formatting 'TEST_DIR/tmp2.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/tmp2.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=536870912 backing_file=TEST_DIR/t.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 *** done
diff --git a/tests/qemu-iotests/182.out b/tests/qemu-iotests/182.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/182.out
+++ b/tests/qemu-iotests/182.out
@@ -XXX,XX +XXX,XX @@ Is another process using the image [TEST_DIR/t.qcow2]?
 {'execute': 'blockdev-add', 'arguments': { 'node-name': 'node0', 'driver': 'file', 'filename': 'TEST_DIR/t.IMGFMT', 'locking': 'on' } }
 {"return": {}}
 {'execute': 'blockdev-snapshot-sync', 'arguments': { 'node-name': 'node0', 'snapshot-file': 'TEST_DIR/t.IMGFMT.overlay', 'snapshot-node-name': 'node1' } }
-Formatting 'TEST_DIR/t.qcow2.overlay', fmt=qcow2 cluster_size=65536 compression_type=zlib size=197120 backing_file=TEST_DIR/t.qcow2 backing_fmt=file lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2.overlay', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=197120 backing_file=TEST_DIR/t.qcow2 backing_fmt=file lazy_refcounts=off refcount_bits=16
 {"return": {}}
 {'execute': 'blockdev-add', 'arguments': { 'node-name': 'node1', 'driver': 'file', 'filename': 'TEST_DIR/t.IMGFMT', 'locking': 'on' } }
 {"return": {}}
diff --git a/tests/qemu-iotests/185.out b/tests/qemu-iotests/185.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/185.out
+++ b/tests/qemu-iotests/185.out
@@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/t.IMGFMT.base', fmt=IMGFMT size=67108864
 === Creating backing chain ===
 
 { 'execute': 'blockdev-snapshot-sync', 'arguments': { 'device': 'disk', 'snapshot-file': 'TEST_DIR/t.IMGFMT.mid', 'format': 'IMGFMT', 'mode': 'absolute-paths' } }
-Formatting 'TEST_DIR/t.qcow2.mid', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 backing_file=TEST_DIR/t.qcow2.base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2.mid', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 backing_file=TEST_DIR/t.qcow2.base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 { 'execute': 'human-monitor-command', 'arguments': { 'command-line': 'qemu-io disk "write 0 4M"' } }
 wrote 4194304/4194304 bytes at offset 0
 4 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
 {"return": ""}
 { 'execute': 'blockdev-snapshot-sync', 'arguments': { 'device': 'disk', 'snapshot-file': 'TEST_DIR/t.IMGFMT', 'format': 'IMGFMT', 'mode': 'absolute-paths' } }
-Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 backing_file=TEST_DIR/t.qcow2.mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 backing_file=TEST_DIR/t.qcow2.mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 {"return": {}}
 
 === Start commit job and exit qemu ===
@@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zli
 { 'execute': 'qmp_capabilities' }
 {"return": {}}
 { 'execute': 'drive-mirror', 'arguments': { 'device': 'disk', 'target': 'TEST_DIR/t.IMGFMT.copy', 'format': 'IMGFMT', 'sync': 'full', 'speed': 65536 } }
-Formatting 'TEST_DIR/t.qcow2.copy', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2.copy', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "created", "id": "disk"}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "running", "id": "disk"}}
 {"return": {}}
@@ -XXX,XX +XXX,XX @@ Formatting 'TEST_DIR/t.qcow2.copy', fmt=qcow2 cluster_size=65536 compression_typ
 { 'execute': 'qmp_capabilities' }
 {"return": {}}
 { 'execute': 'drive-backup', 'arguments': { 'device': 'disk', 'target': 'TEST_DIR/t.IMGFMT.copy', 'format': 'IMGFMT', 'sync': 'full', 'speed': 65536 } }
-Formatting 'TEST_DIR/t.qcow2.copy', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/t.qcow2.copy', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "created", "id": "disk"}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "running", "id": "disk"}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "paused", "id": "disk"}}
diff --git a/tests/qemu-iotests/198 b/tests/qemu-iotests/198
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/198
+++ b/tests/qemu-iotests/198
@@ -XXX,XX +XXX,XX @@ echo "== checking image base =="
 $QEMU_IMG info --image-opts $IMGSPECBASE | _filter_img_info --format-specific \
     | sed -e "/^disk size:/ D" -e '/refcount bits:/ D' -e '/compat:/ D' \
           -e '/lazy refcounts:/ D' -e '/corrupt:/ D' -e '/^\s*data file/ D' \
+          -e '/extended l2:/ D' \
     | _filter_json_filename
 
 echo
@@ -XXX,XX +XXX,XX @@ echo "== checking image layer =="
 $QEMU_IMG info --image-opts $IMGSPECLAYER | _filter_img_info --format-specific \
     | sed -e "/^disk size:/ D" -e '/refcount bits:/ D' -e '/compat:/ D' \
           -e '/lazy refcounts:/ D' -e '/corrupt:/ D' -e '/^\s*data file/ D' \
+          -e '/extended l2:/ D' \
     | _filter_json_filename
 
 
diff --git a/tests/qemu-iotests/206.out b/tests/qemu-iotests/206.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/206.out
+++ b/tests/qemu-iotests/206.out
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 === Successful image creation (inline blockdev-add, explicit defaults) ===
 
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 === Successful image creation (v3 non-default options) ===
 
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: true
     refcount bits: 1
     corrupt: false
+    extended l2: false
 
 === Successful image creation (v2 non-default options) ===
 
@@ -XXX,XX +XXX,XX @@ Format specific information:
         payload offset: 528384
         master key iters: XXX
     corrupt: false
+    extended l2: false
 
 === Invalid BlockdevRef ===
 
diff --git a/tests/qemu-iotests/242.out b/tests/qemu-iotests/242.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/242.out
+++ b/tests/qemu-iotests/242.out
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 No bitmap in JSON format output
 
@@ -XXX,XX +XXX,XX @@ Format specific information:
             granularity: 32768
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 The same bitmaps in JSON format:
 [
@@ -XXX,XX +XXX,XX @@ Format specific information:
             granularity: 65536
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 The same bitmaps in JSON format:
 [
@@ -XXX,XX +XXX,XX @@ Format specific information:
             granularity: 65536
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 The same bitmaps in JSON format:
 [
@@ -XXX,XX +XXX,XX @@ Format specific information:
             granularity: 16384
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 Test complete
diff --git a/tests/qemu-iotests/255.out b/tests/qemu-iotests/255.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/255.out
+++ b/tests/qemu-iotests/255.out
@@ -XXX,XX +XXX,XX @@ Finishing a commit job with background reads
 
 === Create backing chain and start VM ===
 
-Formatting 'TEST_DIR/PID-t.qcow2.mid', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-t.qcow2.mid', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-t.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-t.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
 
 === Start background read requests ===
 
@@ -XXX,XX +XXX,XX @@ Closing the VM while a job is being cancelled
 
 === Create images and start VM ===
 
-Formatting 'TEST_DIR/PID-src.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-src.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-dst.qcow2', fmt=qcow2 cluster_size=65536 compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-dst.qcow2', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=134217728 lazy_refcounts=off refcount_bits=16
 
 wrote 1048576/1048576 bytes at offset 0
 1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
diff --git a/tests/qemu-iotests/274.out b/tests/qemu-iotests/274.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/274.out
+++ b/tests/qemu-iotests/274.out
@@ -XXX,XX +XXX,XX @@
 == Commit tests ==
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=2097152 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=2097152 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-mid', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1048576 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-mid', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1048576 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=2097152 backing_file=TEST_DIR/PID-mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=2097152 backing_file=TEST_DIR/PID-mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 2097152/2097152 bytes at offset 0
 2 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 read 1048576/1048576 bytes at offset 0
 1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 1048576/1048576 bytes at offset 1048576
 1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
 
 === Testing HMP commit (top -> mid) ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=2097152 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=2097152 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-mid', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1048576 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-mid', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1048576 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=2097152 backing_file=TEST_DIR/PID-mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=2097152 backing_file=TEST_DIR/PID-mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 2097152/2097152 bytes at offset 0
 2 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 read 1048576/1048576 bytes at offset 0
 1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 1048576/1048576 bytes at offset 1048576
 1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
 
 === Testing QMP active commit (top -> mid) ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=2097152 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=2097152 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-mid', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1048576 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-mid', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1048576 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=2097152 backing_file=TEST_DIR/PID-mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=2097152 backing_file=TEST_DIR/PID-mid backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 2097152/2097152 bytes at offset 0
 2 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 read 1048576/1048576 bytes at offset 0
 1 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 1048576/1048576 bytes at offset 1048576
 
 == Resize tests ==
 === preallocation=off ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=6442450944 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=6442450944 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=1073741824 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=1073741824 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 65536/65536 bytes at offset 5368709120
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 65536/65536 bytes at offset 5368709120
 { "start": 1073741824, "length": 7516192768, "depth": 0, "zero": true, "data": false}]
 
 === preallocation=metadata ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=34359738368 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=34359738368 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=32212254720 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=32212254720 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 65536/65536 bytes at offset 33285996544
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 65536/65536 bytes at offset 33285996544
 { "start": 34896609280, "length": 536870912, "depth": 0, "zero": true, "data": false, "offset": 2685075456}]
 
 === preallocation=falloc ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=10485760 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=10485760 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=5242880 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=5242880 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 65536/65536 bytes at offset 9437184
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 65536/65536 bytes at offset 9437184
 { "start": 5242880, "length": 10485760, "depth": 0, "zero": false, "data": true, "offset": 327680}]
 
 === preallocation=full ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=16777216 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=16777216 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=8388608 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=8388608 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 65536/65536 bytes at offset 11534336
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 65536/65536 bytes at offset 11534336
 { "start": 8388608, "length": 4194304, "depth": 0, "zero": false, "data": true, "offset": 327680}]
 
 === preallocation=off ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=393216 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=393216 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=259072 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=259072 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 65536/65536 bytes at offset 259072
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 65536/65536 bytes at offset 259072
 { "start": 262144, "length": 262144, "depth": 0, "zero": true, "data": false}]
 
 === preallocation=off ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=409600 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=409600 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=262144 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=262144 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 65536/65536 bytes at offset 344064
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
@@ -XXX,XX +XXX,XX @@ read 65536/65536 bytes at offset 344064
 { "start": 262144, "length": 262144, "depth": 0, "zero": true, "data": false}]
 
 === preallocation=off ===
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=524288 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=524288 lazy_refcounts=off refcount_bits=16
 
-Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 compression_type=zlib size=262144 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-top', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=262144 backing_file=TEST_DIR/PID-base backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
 
 wrote 65536/65536 bytes at offset 446464
 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
diff --git a/tests/qemu-iotests/280.out b/tests/qemu-iotests/280.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/280.out
+++ b/tests/qemu-iotests/280.out
@@ -XXX,XX +XXX,XX @@
-Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
+Formatting 'TEST_DIR/PID-base', fmt=qcow2 cluster_size=65536 extended_l2=off compression_type=zlib size=67108864 lazy_refcounts=off refcount_bits=16
 
 === Launch VM ===
 Enabling migration QMP events on VM...
diff --git a/tests/qemu-iotests/291.out b/tests/qemu-iotests/291.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/291.out
+++ b/tests/qemu-iotests/291.out
@@ -XXX,XX +XXX,XX @@ Format specific information:
             granularity: 65536
     refcount bits: 16
     corrupt: false
+    extended l2: false
 image: TEST_DIR/t.IMGFMT
 file format: IMGFMT
 virtual size: 10 MiB (10485760 bytes)
@@ -XXX,XX +XXX,XX @@ Format specific information:
             granularity: 65536
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 === Check bitmap contents ===
 
diff --git a/tests/qemu-iotests/302.out b/tests/qemu-iotests/302.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/302.out
+++ b/tests/qemu-iotests/302.out
@@ -XXX,XX +XXX,XX @@ Format specific information:
     lazy refcounts: false
     refcount bits: 16
     corrupt: false
+    extended l2: false
 
 === Converted image check ===
 No errors were found on the image.
diff --git a/tests/qemu-iotests/303.out b/tests/qemu-iotests/303.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/303.out
+++ b/tests/qemu-iotests/303.out
@@ -XXX,XX +XXX,XX @@ header_length             112
 
 Header extension:
 magic                     0x6803f857 (Feature table)
-length                    336
+length                    384
 data                      <binary>
 
 Header extension:
@@ -XXX,XX +XXX,XX @@ Bitmap table   type            size         offset
     {
         "name": "Feature table",
         "magic": 1745090647,
-        "length": 336,
+        "length": 384,
         "data_str": "<binary>"
     },
     {
diff --git a/tests/qemu-iotests/common.filter b/tests/qemu-iotests/common.filter
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/common.filter
+++ b/tests/qemu-iotests/common.filter
@@ -XXX,XX +XXX,XX @@ _filter_img_info()
         -e "/adapter_type: '[^']*'/d" \
         -e "/hwversion: '[^']*'/d" \
         -e "/lazy_refcounts: \$on\\|off\$/d" \
+        -e "/extended_l2=\$on\\|off\$/d" \
         -e "/block_size: [0-9]\\+/d" \
         -e "/block_state_zero: \$on\\|off\$/d" \
         -e "/log_size: [0-9]\\+/d" \
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Traditional qcow2 images don't allow preallocation if a backing file
is set. This is because once a cluster is allocated there is no way to
tell that its data should be read from the backing file.

Extended L2 entries have individual allocation bits for each
subcluster, and therefore it is perfectly possible to have an
allocated cluster with all its subclusters unallocated.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <6d5b0f38e7dc5f2f31d8cab1cb92044e9909aece.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2.c              | 7 ++++---
 tests/qemu-iotests/206.out | 2 +-
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -XXX,XX +XXX,XX @@ qcow2_co_create(BlockdevCreateOptions *create_options, Error **errp)
         qcow2_opts->preallocation = PREALLOC_MODE_OFF;
     }
     if (qcow2_opts->has_backing_file &&
-        qcow2_opts->preallocation != PREALLOC_MODE_OFF)
+        qcow2_opts->preallocation != PREALLOC_MODE_OFF &&
+        !qcow2_opts->extended_l2)
     {
-        error_setg(errp, "Backing file and preallocation cannot be used at "
-                   "the same time");
+        error_setg(errp, "Backing file and preallocation can only be used at "
+                   "the same time if extended_l2 is on");
         ret = -EINVAL;
         goto out;
     }
diff --git a/tests/qemu-iotests/206.out b/tests/qemu-iotests/206.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/206.out
+++ b/tests/qemu-iotests/206.out
@@ -XXX,XX +XXX,XX @@ Job failed: Different refcount widths than 16 bits require compatibility level 1
 === Invalid backing file options ===
 {"execute": "blockdev-create", "arguments": {"job-id": "job0", "options": {"backing-file": "/dev/null", "driver": "qcow2", "file": "node0", "preallocation": "full", "size": 67108864}}}
 {"return": {}}
-Job failed: Backing file and preallocation cannot be used at the same time
+Job failed: Backing file and preallocation can only be used at the same time if extended_l2 is on
 {"execute": "job-dismiss", "arguments": {"id": "job0"}}
 {"return": {}}
 
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

This function is only used by qcow2_expand_zero_clusters() to
downgrade a qcow2 image to a previous version. This would require
transforming all extended L2 entries into normal L2 entries but this
is not a simple task and there are no plans to implement this at the
moment.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <15e65112b4144381b4d8c0bdf8fb76b0d813e3d1.1594396418.git.berto@igalia.com>
[mreitz: Fixed comment style]
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c      | 14 ++++++++++++--
 tests/qemu-iotests/061     |  6 ++++++
 tests/qemu-iotests/061.out |  5 +++++
 3 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index XXXXXXX..XXXXXXX 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -XXX,XX +XXX,XX @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
     int ret;
     int i, j;
 
+    /* qcow2_downgrade() is not allowed in images with subclusters */
+    assert(!has_subclusters(s));
+
     slice_size2 = s->l2_slice_size * l2_entry_size(s);
     n_slices = s->cluster_size / slice_size2;
 
@@ -XXX,XX +XXX,XX @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
 
                 if (cluster_type == QCOW2_CLUSTER_ZERO_PLAIN) {
                     if (!bs->backing) {
-                        /* not backed; therefore we can simply deallocate the
-                         * cluster */
+                        /*
+                         * not backed; therefore we can simply deallocate the
+                         * cluster. No need to call set_l2_bitmap(), this
+                         * function doesn't support images with subclusters.
+                         */
                         set_l2_entry(s, l2_slice, j, 0);
                         l2_dirty = true;
                         continue;
@@ -XXX,XX +XXX,XX @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                 } else {
                     set_l2_entry(s, l2_slice, j, offset);
                 }
+                /*
+                 * No need to call set_l2_bitmap() after set_l2_entry() because
+                 * this function doesn't support images with subclusters.
+                 */
                 l2_dirty = true;
             }
 
diff --git a/tests/qemu-iotests/061 b/tests/qemu-iotests/061
index XXXXXXX..XXXXXXX 100755
--- a/tests/qemu-iotests/061
+++ b/tests/qemu-iotests/061
@@ -XXX,XX +XXX,XX @@ $QEMU_IMG amend -o "compat=0.10" "$TEST_IMG"
 _img_info --format-specific
 _check_test_img
 
+echo
+echo "=== Testing version downgrade with extended L2 entries ==="
+echo
+_make_test_img -o "compat=1.1,extended_l2=on" 64M
+$QEMU_IMG amend -o "compat=0.10" "$TEST_IMG"
+
 echo
 echo "=== Try changing the external data file ==="
 echo
diff --git a/tests/qemu-iotests/061.out b/tests/qemu-iotests/061.out
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/061.out
+++ b/tests/qemu-iotests/061.out
@@ -XXX,XX +XXX,XX @@ Format specific information:
     extended l2: false
 No errors were found on the image.
 
+=== Testing version downgrade with extended L2 entries ===
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+qemu-img: Cannot downgrade an image with incompatible features 0x10 set
+
 === Try changing the external data file ===
 
 Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
-- 
2.26.2

From: Alberto Garcia <berto@igalia.com>

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <e6dd0429cafe84ca603179c298a8703bddca2904.1594396418.git.berto@igalia.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 tests/qemu-iotests/271     | 901 +++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/271.out | 726 ++++++++++++++++++++++++++++++
 tests/qemu-iotests/group   |   1 +
 3 files changed, 1628 insertions(+)
 create mode 100755 tests/qemu-iotests/271
 create mode 100644 tests/qemu-iotests/271.out

diff --git a/tests/qemu-iotests/271 b/tests/qemu-iotests/271
new file mode 100755
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/tests/qemu-iotests/271
@@ -XXX,XX +XXX,XX @@
+#!/bin/bash
+#
+# Test qcow2 images with extended L2 entries
+#
+# Copyright (C) 2019-2020 Igalia, S.L.
+# Author: Alberto Garcia <berto@igalia.com>
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+
+# creator
+owner=berto@igalia.com
+
+seq="$(basename $0)"
+echo "QA output created by $seq"
+
+here="$PWD"
+status=1	# failure is the default!
+
+_cleanup()
+{
+        _cleanup_test_img
+        rm -f "$TEST_IMG.raw"
+}
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common.rc
+. ./common.filter
+
+_supported_fmt qcow2
+_supported_proto file nfs
+_supported_os Linux
+_unsupported_imgopts extended_l2 compat=0.10 cluster_size data_file refcount_bits=1[^0-9]
+
+l2_offset=$((0x40000))
+
+_verify_img()
+{
+    $QEMU_IMG compare "$TEST_IMG" "$TEST_IMG.raw" | grep -v 'Images are identical'
+    $QEMU_IMG check "$TEST_IMG" | _filter_qemu_img_check | \
+        grep -v 'No errors were found on the image'
+}
+
+# Compare the bitmap of an extended L2 entry against an expected value
+_verify_l2_bitmap()
+{
+    entry_no="$1"            # L2 entry number, starting from 0
+    expected_alloc="$alloc"  # Space-separated list of allocated subcluster indexes
+    expected_zero="$zero"    # Space-separated list of zero subcluster indexes
+
+    offset=$(($l2_offset + $entry_no * 16))
+    entry=$(peek_file_be "$TEST_IMG" $offset 8)
+    offset=$(($offset + 8))
+    bitmap=$(peek_file_be "$TEST_IMG" $offset 8)
+
+    expected_bitmap=0
+    for bit in $expected_alloc; do
+        expected_bitmap=$(($expected_bitmap | (1 << $bit)))
+    done
+    for bit in $expected_zero; do
+        expected_bitmap=$(($expected_bitmap | (1 << (32 + $bit))))
+    done
+    printf -v expected_bitmap "%u" $expected_bitmap # Convert to unsigned
+
+    printf "L2 entry #%d: 0x%016x %016x\n" "$entry_no" "$entry" "$bitmap"
+    if [ "$bitmap" != "$expected_bitmap" ]; then
+        printf "ERROR: expecting bitmap       0x%016x\n" "$expected_bitmap"
+    fi
+}
+
+# This should be called as _run_test c=XXX sc=XXX off=XXX len=XXX cmd=XXX
+# c:   cluster number (0 if unset)
+# sc:  subcluster number inside cluster @c (0 if unset)
+# off: offset inside subcluster @sc, in kilobytes (0 if unset)
+# len: request length, passed directly to qemu-io (e.g: 256, 4k, 1M, ...)
+# cmd: the command to pass to qemu-io, must be one of
+#      write    -> write
+#      zero     -> write -z
+#      unmap    -> write -z -u
+#      compress -> write -c
+#      discard  -> discard
+_run_test()
+{
+    unset c sc off len cmd
+    for var in "$@"; do eval "$var"; done
+    case "${cmd:-write}" in
+        zero)
+            cmd="write -q -z";;
+        unmap)
+            cmd="write -q -z -u";;
+        compress)
+            pat=$((${pat:-0} + 1))
+            cmd="write -q -c -P ${pat}";;
+        write)
+            pat=$((${pat:-0} + 1))
+            cmd="write -q -P ${pat}";;
+        discard)
+            cmd="discard -q";;
+        *)
+            echo "Unknown option $cmd"
+            exit 1;;
+    esac
+    c="${c:-0}"
+    sc="${sc:-0}"
+    off="${off:-0}"
+    offset="$(($c * 64 + $sc * 2 + $off))"
+    [ "$offset" != 0 ] && offset="${offset}k"
+    cmd="$cmd ${offset} ${len}"
+    raw_cmd=$(echo $cmd | sed s/-c//) # Raw images don't support -c
+    echo $cmd | sed 's/-P [0-9][0-9]\?/-P PATTERN/'
+    $QEMU_IO -c "$cmd" "$TEST_IMG" | _filter_qemu_io
+    $QEMU_IO -c "$raw_cmd" -f raw "$TEST_IMG.raw" | _filter_qemu_io
+    _verify_img
+    _verify_l2_bitmap "$c"
+}
+
+_reset_img()
+{
+    size="$1"
+    $QEMU_IMG create -f raw "$TEST_IMG.raw" "$size" | _filter_img_create
+    if [ "$use_backing_file" = "yes" ]; then
+        $QEMU_IMG create -f raw "$TEST_IMG.base" "$size" | _filter_img_create
+        $QEMU_IO -c "write -q -P 0xFF 0 $size" -f raw "$TEST_IMG.base" | _filter_qemu_io
+        $QEMU_IO -c "write -q -P 0xFF 0 $size" -f raw "$TEST_IMG.raw" | _filter_qemu_io
+        _make_test_img -o extended_l2=on -F raw -b "$TEST_IMG.base" "$size"
+    else
+        _make_test_img -o extended_l2=on "$size"
+    fi
+}
+
+############################################################
+############################################################
+############################################################
+
+# Test that writing to an image with subclusters produces the expected
+# results, in images with and without backing files
+for use_backing_file in yes no; do
+    echo
+    echo "### Standard write tests (backing file: $use_backing_file) ###"
+    echo
+    _reset_img 1M
+    ### Write subcluster #0 (beginning of subcluster) ###
+    alloc="0"; zero=""
+    _run_test sc=0 len=1k
+
+    ### Write subcluster #1 (middle of subcluster) ###
+    alloc="0 1"; zero=""
+    _run_test sc=1 off=1 len=512
+
+    ### Write subcluster #2 (end of subcluster) ###
+    alloc="0 1 2"; zero=""
+    _run_test sc=2 off=1 len=1k
+
+    ### Write subcluster #3 (full subcluster) ###
+    alloc="0 1 2 3"; zero=""
+    _run_test sc=3 len=2k
+
+    ### Write subclusters #4-6 (full subclusters) ###
+    alloc="$(seq 0 6)"; zero=""
+    _run_test sc=4 len=6k
+
+    ### Write subclusters #7-9 (partial subclusters) ###
+    alloc="$(seq 0 9)"; zero=""
+    _run_test sc=7 off=1 len=4k
+
+    ### Write subcluster #16 (partial subcluster) ###
+    alloc="$(seq 0 9) 16"; zero=""
+    _run_test sc=16 len=1k
+
+    ### Write subcluster #31-#33 (cluster overlap) ###
+    alloc="$(seq 0 9) 16 31"; zero=""
+    _run_test sc=31 off=1 len=4k
+    alloc="0 1" ; zero=""
+    _verify_l2_bitmap 1
+
+    ### Zero subcluster #1
+    alloc="0 $(seq 2 9) 16 31"; zero="1"
+    _run_test sc=1 len=2k cmd=zero
+
+    ### Zero cluster #0
+    alloc=""; zero="$(seq 0 31)"
+    _run_test sc=0 len=64k cmd=zero
+
+    ### Fill cluster #0 with data
+    alloc="$(seq 0 31)"; zero=""
+    _run_test sc=0 len=64k
+
+    ### Zero and unmap half of cluster #0 (this won't unmap it)
+    alloc="$(seq 16 31)"; zero="$(seq 0 15)"
+    _run_test sc=0 len=32k cmd=unmap
+
+    ### Zero and unmap cluster #0
+    alloc=""; zero="$(seq 0 31)"
+    _run_test sc=0 len=64k cmd=unmap
+
+    ### Write subcluster #1 (middle of subcluster)
+    alloc="1"; zero="0 $(seq 2 31)"
+    _run_test sc=1 off=1 len=512
+
+    ### Fill cluster #0 with data
+    alloc="$(seq 0 31)"; zero=""
+    _run_test sc=0 len=64k
+
+    ### Discard cluster #0
+    alloc=""; zero="$(seq 0 31)"
+    _run_test sc=0 len=64k cmd=discard
+
+    ### Write compressed data to cluster #0
+    alloc=""; zero=""
+    _run_test sc=0 len=64k cmd=compress
+
+    ### Write subcluster #1 (middle of subcluster)
+    alloc="$(seq 0 31)"; zero=""
+    _run_test sc=1 off=1 len=512
+done
+
+############################################################
+############################################################
+############################################################
+
+# calculate_l2_meta() checks if none of the clusters affected by a
+# write operation need COW or changes to their L2 metadata and simply
+# returns when they don't. This is a test for that optimization.
+# Here clusters #0-#3 are overwritten but only #1 and #2 need changes.
+echo
+echo '### Overwriting several clusters without COW ###'
+echo
+use_backing_file="no" _reset_img 1M
+# Write cluster #0, subclusters #12-#31
+alloc="$(seq 12 31)"; zero=""
+_run_test sc=12 len=40k
+
+# Write cluster #1, subcluster #13
+alloc="13"; zero=""
+_run_test c=1 sc=13 len=2k
+
+# Zeroize cluster #2, subcluster #14
+alloc="14"; zero=""
+_run_test c=2 sc=14 len=2k
+alloc=""; zero="14"
+_run_test c=2 sc=14 len=2k cmd=zero
+
+# Write cluster #3, subclusters #0-#16
+alloc="$(seq 0 16)"; zero=""
+_run_test c=3 sc=0 len=34k
+
+# Write from cluster #0, subcluster #12 to cluster #3, subcluster #11
+alloc="$(seq 12 31)"; zero=""
+_run_test sc=12 len=192k
+alloc="$(seq 0 31)"; zero=""
+_verify_l2_bitmap 1
+_verify_l2_bitmap 2
+
+alloc="$(seq 0 16)"; zero=""
+_verify_l2_bitmap 3
+
+############################################################
+############################################################
+############################################################
+
+# Test different patterns of writing zeroes
+for use_backing_file in yes no; do
+    echo
+    echo "### Writing zeroes 1: unallocated clusters (backing file: $use_backing_file) ###"
+    echo
+    # Note that the image size is not a multiple of the cluster size
+    _reset_img 2083k
+
+    # Cluster-aligned request from clusters #0 to #2
+    alloc=""; zero="$(seq 0 31)"
+    _run_test c=0 sc=0 len=192k cmd=zero
+    _verify_l2_bitmap 1
+    _verify_l2_bitmap 2
+
+    # Subcluster-aligned request from clusters #3 to #5
+    alloc=""; zero="$(seq 16 31)"
+    _run_test c=3 sc=16 len=128k cmd=zero
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 4
+    alloc=""; zero="$(seq 0 15)"
+    _verify_l2_bitmap 5
+
+    # Unaligned request from clusters #6 to #8
+    if [ "$use_backing_file" = "yes" ]; then
+        alloc="15"; zero="$(seq 16 31)" # copy-on-write happening here
+    else
+        alloc=""; zero="$(seq 15 31)"
+    fi
+    _run_test c=6 sc=15 off=1 len=128k cmd=zero
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 7
+    if [ "$use_backing_file" = "yes" ]; then
+        alloc="15"; zero="$(seq 0 14)" # copy-on-write happening here
+    else
+        alloc=""; zero="$(seq 0 15)"
+    fi
+    _verify_l2_bitmap 8
+
+    echo
+    echo "### Writing zeroes 2: allocated clusters (backing file: $use_backing_file) ###"
+    echo
+    alloc="$(seq 0 31)"; zero=""
+    _run_test c=9 sc=0 len=576k
+    _verify_l2_bitmap 10
+    _verify_l2_bitmap 11
+    _verify_l2_bitmap 12
+    _verify_l2_bitmap 13
+    _verify_l2_bitmap 14
+    _verify_l2_bitmap 15
+    _verify_l2_bitmap 16
+    _verify_l2_bitmap 17
+
+    # Cluster-aligned request from clusters #9 to #11
+    alloc=""; zero="$(seq 0 31)"
+    _run_test c=9 sc=0 len=192k cmd=zero
+    _verify_l2_bitmap 10
+    _verify_l2_bitmap 11
+
+    # Subcluster-aligned request from clusters #12 to #14
+    alloc="$(seq 0 15)"; zero="$(seq 16 31)"
+    _run_test c=12 sc=16 len=128k cmd=zero
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 13
+    alloc="$(seq 16 31)"; zero="$(seq 0 15)"
+    _verify_l2_bitmap 14
+
+    # Unaligned request from clusters #15 to #17
+    alloc="$(seq 0 15)"; zero="$(seq 16 31)"
+    _run_test c=15 sc=15 off=1 len=128k cmd=zero
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 16
+    alloc="$(seq 15 31)"; zero="$(seq 0 14)"
+    _verify_l2_bitmap 17
+
+    echo
+    echo "### Writing zeroes 3: compressed clusters (backing file: $use_backing_file) ###"
+    echo
+    alloc=""; zero=""
+    for c in $(seq 18 28); do
+        _run_test c=$c sc=0 len=64k cmd=compress
+    done
+
+    # Cluster-aligned request from clusters #18 to #20
+    alloc=""; zero="$(seq 0 31)"
+    _run_test c=18 sc=0 len=192k cmd=zero
+    _verify_l2_bitmap 19
+    _verify_l2_bitmap 20
+
+    # Subcluster-aligned request from clusters #21 to #23.
+    # We cannot partially zero a compressed cluster so the code
+    # returns -ENOTSUP, which means copy-on-write of the compressed
+    # data and fill the rest with actual zeroes on disk.
+    # TODO: cluster #22 should use the 'all zeroes' bits.
+    alloc="$(seq 0 31)"; zero=""
+    _run_test c=21 sc=16 len=128k cmd=zero
+    _verify_l2_bitmap 22
+    _verify_l2_bitmap 23
+
+    # Unaligned request from clusters #24 to #26
+    # In this case QEMU internally sends a 1k request followed by a
+    # subcluster-aligned 128k request. The first request decompresses
+    # cluster #24, but that's not enough to perform the second request
+    # efficiently because it partially writes to cluster #26 (which is
+    # compressed) so we hit the same problem as before.
+    alloc="$(seq 0 31)"; zero=""
+    _run_test c=24 sc=15 off=1 len=129k cmd=zero
+    _verify_l2_bitmap 25
+    _verify_l2_bitmap 26
+
+    # Unaligned request from clusters #27 to #29
+    # Similar to the previous case, but this time the tail of the
+    # request does not correspond to a compressed cluster, so it can
+    # be zeroed efficiently.
+    # Note that the very last subcluster is partially written, so if
+    # there's a backing file we need to perform cow.
+    alloc="$(seq 0 15)"; zero="$(seq 16 31)"
+    _run_test c=27 sc=15 off=1 len=128k cmd=zero
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 28
+    if [ "$use_backing_file" = "yes" ]; then
+        alloc="15"; zero="$(seq 0 14)" # copy-on-write happening here
+    else
+        alloc=""; zero="$(seq 0 15)"
+    fi
+    _verify_l2_bitmap 29
+
+    echo
+    echo "### Writing zeroes 4: other tests (backing file: $use_backing_file) ###"
+    echo
+    # Unaligned request in the middle of cluster #30.
+    # If there's a backing file we need to allocate and do
+    # copy-on-write on the partially zeroed subclusters.
+    # If not we can set the 'all zeroes' bit on them.
+    if [ "$use_backing_file" = "yes" ]; then
+        alloc="15 19"; zero="$(seq 16 18)" # copy-on-write happening here
+    else
+        alloc=""; zero="$(seq 15 19)"
+    fi
+    _run_test c=30 sc=15 off=1 len=8k cmd=zero
+
+    # Fill the last cluster with zeroes, up to the end of the image
+    # (the image size is not a multiple of the cluster or subcluster size).
+    alloc=""; zero="$(seq 0 17)"
+    _run_test c=32 sc=0 len=35k cmd=zero
+done
+
+############################################################
+############################################################
+############################################################
+
+# Zero + unmap
+for use_backing_file in yes no; do
+    echo
+    echo "### Zero + unmap 1: allocated clusters (backing file: $use_backing_file) ###"
+    echo
+    # Note that the image size is not a multiple of the cluster size
+    _reset_img 2083k
+    alloc="$(seq 0 31)"; zero=""
+    _run_test c=9 sc=0 len=576k
+    _verify_l2_bitmap 10
+    _verify_l2_bitmap 11
+    _verify_l2_bitmap 12
+    _verify_l2_bitmap 13
+    _verify_l2_bitmap 14
+    _verify_l2_bitmap 15
+    _verify_l2_bitmap 16
+    _verify_l2_bitmap 17
+
+    # Cluster-aligned request from clusters #9 to #11
+    alloc=""; zero="$(seq 0 31)"
+    _run_test c=9 sc=0 len=192k cmd=unmap
+    _verify_l2_bitmap 10
+    _verify_l2_bitmap 11
+
+    # Subcluster-aligned request from clusters #12 to #14
+    alloc="$(seq 0 15)"; zero="$(seq 16 31)"
+    _run_test c=12 sc=16 len=128k cmd=unmap
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 13
+    alloc="$(seq 16 31)"; zero="$(seq 0 15)"
+    _verify_l2_bitmap 14
+
+    # Unaligned request from clusters #15 to #17
+    alloc="$(seq 0 15)"; zero="$(seq 16 31)"
+    _run_test c=15 sc=15 off=1 len=128k cmd=unmap
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 16
+    alloc="$(seq 15 31)"; zero="$(seq 0 14)"
+    _verify_l2_bitmap 17
+
+    echo
+    echo "### Zero + unmap 2: compressed clusters (backing file: $use_backing_file) ###"
+    echo
+    alloc=""; zero=""
+    for c in $(seq 18 28); do
+        _run_test c=$c sc=0 len=64k cmd=compress
+    done
+
+    # Cluster-aligned request from clusters #18 to #20
+    alloc=""; zero="$(seq 0 31)"
+    _run_test c=18 sc=0 len=192k cmd=unmap
+    _verify_l2_bitmap 19
+    _verify_l2_bitmap 20
+
+    # Subcluster-aligned request from clusters #21 to #23.
+    # We cannot partially zero a compressed cluster so the code
+    # returns -ENOTSUP, which means copy-on-write of the compressed
+    # data and fill the rest with actual zeroes on disk.
+    # TODO: cluster #22 should use the 'all zeroes' bits.
+    alloc="$(seq 0 31)"; zero=""
+    _run_test c=21 sc=16 len=128k cmd=unmap
+    _verify_l2_bitmap 22
+    _verify_l2_bitmap 23
+
+    # Unaligned request from clusters #24 to #26
+    # In this case QEMU internally sends a 1k request followed by a
+    # subcluster-aligned 128k request. The first request decompresses
+    # cluster #24, but that's not enough to perform the second request
+    # efficiently because it partially writes to cluster #26 (which is
+    # compressed) so we hit the same problem as before.
+    alloc="$(seq 0 31)"; zero=""
+    _run_test c=24 sc=15 off=1 len=129k cmd=unmap
+    _verify_l2_bitmap 25
+    _verify_l2_bitmap 26
+
+    # Unaligned request from clusters #27 to #29
+    # Similar to the previous case, but this time the tail of the
+    # request does not correspond to a compressed cluster, so it can
+    # be zeroed efficiently.
+    # Note that the very last subcluster is partially written, so if
+    # there's a backing file we need to perform cow.
+    alloc="$(seq 0 15)"; zero="$(seq 16 31)"
+    _run_test c=27 sc=15 off=1 len=128k cmd=unmap
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 28
+    if [ "$use_backing_file" = "yes" ]; then
+        alloc="15"; zero="$(seq 0 14)" # copy-on-write happening here
+    else
+        alloc=""; zero="$(seq 0 15)"
+    fi
+    _verify_l2_bitmap 29
+done
+
+############################################################
+############################################################
+############################################################
+
+# Test qcow2_cluster_discard() with full and normal discards
+for use_backing_file in yes no; do
+    echo
+    echo "### Discarding clusters with non-zero bitmaps (backing file: $use_backing_file) ###"
+    echo
+    if [ "$use_backing_file" = "yes" ]; then
+        _make_test_img -o extended_l2=on -F raw -b "$TEST_IMG.base" 1M
+    else
+        _make_test_img -o extended_l2=on 1M
+    fi
+    # Write clusters #0-#2 and then discard them
+    $QEMU_IO -c 'write -q 0 128k' "$TEST_IMG"
+    $QEMU_IO -c 'discard -q 0 128k' "$TEST_IMG"
+    # 'qemu-io discard' doesn't do a full discard, it zeroizes the
+    # cluster, so both clusters have all zero bits set now
+    alloc=""; zero="$(seq 0 31)"
+    _verify_l2_bitmap 0
+    _verify_l2_bitmap 1
+    # Now mark the 2nd half of the subclusters from cluster #0 as unallocated
+    poke_file "$TEST_IMG" $(($l2_offset+8)) "\x00\x00"
+    # Discard cluster #0 again to see how the zero bits have changed
+    $QEMU_IO -c 'discard -q 0 64k' "$TEST_IMG"
+    # And do a full discard of cluster #1 by shrinking and growing the image
+    $QEMU_IMG resize --shrink "$TEST_IMG" 64k
+    $QEMU_IMG resize "$TEST_IMG" 1M
+    # A normal discard sets all 'zero' bits only if the image has a
+    # backing file, otherwise it won't touch them.
+    if [ "$use_backing_file" = "yes" ]; then
+        alloc=""; zero="$(seq 0 31)"
+    else
+        alloc=""; zero="$(seq 0 15)"
+    fi
+    _verify_l2_bitmap 0
+    # A full discard should clear the L2 entry completely. However
+    # when growing an image with a backing file the new clusters are
+    # zeroized to hide the stale data from the backing file
+    if [ "$use_backing_file" = "yes" ]; then
+        alloc=""; zero="$(seq 0 31)"
+    else
+        alloc=""; zero=""
+    fi
+    _verify_l2_bitmap 1
+done
+
+############################################################
+############################################################
+############################################################
+
+# Test that corrupted L2 entries are detected in both read and write
+# operations
+for corruption_test_cmd in read write; do
+    echo
+    echo "### Corrupted L2 entries - $corruption_test_cmd test (allocated) ###"
+    echo
+    echo "# 'cluster is zero' bit set on the standard cluster descriptor"
+    echo
+    # We actually don't consider this a corrupted image.
+    # The 'cluster is zero' bit is unused in extended L2 entries so
+    # QEMU ignores it.
+    # TODO: maybe treat the image as corrupted and make qemu-img check fix it?
+    _make_test_img -o extended_l2=on 1M
+    $QEMU_IO -c 'write -q -P 0x11 0 2k' "$TEST_IMG"
+    poke_file "$TEST_IMG" $(($l2_offset+7)) "\x01"
+    alloc="0"; zero=""
+    _verify_l2_bitmap 0
+    $QEMU_IO -c "$corruption_test_cmd -q -P 0x11 0 1k" "$TEST_IMG"
+    if [ "$corruption_test_cmd" = "write" ]; then
+        alloc="0"; zero=""
+    fi
+    _verify_l2_bitmap 0
+
+    echo
+    echo "# Both 'subcluster is zero' and 'subcluster is allocated' bits set"
+    echo
+    _make_test_img -o extended_l2=on 1M
+    # Write from the middle of cluster #0 to the middle of cluster #2
+    $QEMU_IO -c 'write -q 32k 128k' "$TEST_IMG"
+    # Corrupt the L2 entry from cluster #1
+    poke_file_be "$TEST_IMG" $(($l2_offset+24)) 4 1
+    alloc="$(seq 0 31)"; zero="0"
+    _verify_l2_bitmap 1
+    $QEMU_IO -c "$corruption_test_cmd 0 192k" "$TEST_IMG"
+
+    echo
+    echo "### Corrupted L2 entries - $corruption_test_cmd test (unallocated) ###"
+    echo
+    echo "# 'cluster is zero' bit set on the standard cluster descriptor"
+    echo
+    # We actually don't consider this a corrupted image.
+    # The 'cluster is zero' bit is unused in extended L2 entries so
+    # QEMU ignores it.
+    # TODO: maybe treat the image as corrupted and make qemu-img check fix it?
+    _make_test_img -o extended_l2=on 1M
+    # We want to modify the (empty) L2 entry from cluster #0,
+    # but we write to #4 in order to initialize the L2 table first
+    $QEMU_IO -c 'write -q 256k 1k' "$TEST_IMG"
+    poke_file "$TEST_IMG" $(($l2_offset+7)) "\x01"
+    alloc=""; zero=""
+    _verify_l2_bitmap 0
+    $QEMU_IO -c "$corruption_test_cmd -q 0 1k" "$TEST_IMG"
+    if [ "$corruption_test_cmd" = "write" ]; then
+        alloc="0"; zero=""
+    fi
+    _verify_l2_bitmap 0
+
+    echo
+    echo "# 'subcluster is allocated' bit set"
+    echo
+    _make_test_img -o extended_l2=on 1M
+    # We want to corrupt the (empty) L2 entry from cluster #0,
+    # but we write to #4 in order to initialize the L2 table first
+    $QEMU_IO -c 'write -q 256k 1k' "$TEST_IMG"
+    poke_file "$TEST_IMG" $(($l2_offset+15)) "\x01"
+    alloc="0"; zero=""
+    _verify_l2_bitmap 0
+    $QEMU_IO -c "$corruption_test_cmd 0 1k" "$TEST_IMG"
+
+    echo
+    echo "# Both 'subcluster is zero' and 'subcluster is allocated' bits set"
+    echo
+    _make_test_img -o extended_l2=on 1M
+    # We want to corrupt the (empty) L2 entry from cluster #1,
+    # but we write to #4 in order to initialize the L2 table first
+    $QEMU_IO -c 'write -q 256k 1k' "$TEST_IMG"
+    # Corrupt the L2 entry from cluster #1
+    poke_file_be "$TEST_IMG" $(($l2_offset+24)) 8 $(((1 << 32) | 1))
+    alloc="0"; zero="0"
+    _verify_l2_bitmap 1
+    $QEMU_IO -c "$corruption_test_cmd 0 192k" "$TEST_IMG"
+
+    echo
+    echo "### Compressed cluster with subcluster bitmap != 0 - $corruption_test_cmd test ###"
+    echo
+    # We actually don't consider this a corrupted image.
+    # The bitmap in compressed clusters is unused so QEMU should just ignore it.
+    _make_test_img -o extended_l2=on 1M
+    $QEMU_IO -c 'write -q -P 11 -c 0 64k' "$TEST_IMG"
+    # Change the L2 bitmap to allocate subcluster #31 and zeroize subcluster #0
+    poke_file "$TEST_IMG" $(($l2_offset+11)) "\x01\x80"
+    alloc="31"; zero="0"
+    _verify_l2_bitmap 0
+    $QEMU_IO -c "$corruption_test_cmd -P 11 0 64k" "$TEST_IMG" | _filter_qemu_io
+    # Writing allocates a new uncompressed cluster so we get a new bitmap
+    if [ "$corruption_test_cmd" = "write" ]; then
+        alloc="$(seq 0 31)"; zero=""
+    fi
+    _verify_l2_bitmap 0
+done
+
+############################################################
+############################################################
+############################################################
+
+echo
+echo "### Detect and repair unaligned clusters ###"
+echo
+# Create a backing file and fill it with data
+$QEMU_IMG create -f raw "$TEST_IMG.base" 128k | _filter_img_create
+$QEMU_IO -c "write -q -P 0xff 0 128k" -f raw "$TEST_IMG.base" | _filter_qemu_io
+
+echo "# Corrupted L2 entry, allocated subcluster #"
+# Create a new image, allocate a cluster and write some data to it
+_make_test_img -o extended_l2=on -F raw -b "$TEST_IMG.base"
+$QEMU_IO -c 'write -q -P 1 4k 2k' "$TEST_IMG"
+# Corrupt the L2 entry by making the offset unaligned
+poke_file "$TEST_IMG" "$(($l2_offset+6))" "\x02"
+# This cannot be repaired, qemu-img check will fail to fix it
+_check_test_img -r all
+# Attempting to read the image will still show that it's corrupted
+$QEMU_IO -c 'read -q 0 2k' "$TEST_IMG"
+
+echo "# Corrupted L2 entry, no allocated subclusters #"
+# Create a new image, allocate a cluster and zeroize subcluster #2
+_make_test_img -o extended_l2=on -F raw -b "$TEST_IMG.base"
+$QEMU_IO -c 'write -q -P 1 4k 2k' "$TEST_IMG"
+$QEMU_IO -c 'write -q -z   4k 2k' "$TEST_IMG"
+# Corrupt the L2 entry by making the offset unaligned
+poke_file "$TEST_IMG" "$(($l2_offset+6))" "\x02"
+# This time none of the subclusters are allocated so we can repair the image
+_check_test_img -r all
+# And the data can be read normally
+$QEMU_IO -c 'read -q -P 0xff  0   4k' "$TEST_IMG"
+$QEMU_IO -c 'read -q -P 0x00 4k   2k' "$TEST_IMG"
+$QEMU_IO -c 'read -q -P 0xff 6k 122k' "$TEST_IMG"
+
+############################################################
+############################################################
+############################################################
+
+echo
+echo "### Image creation options ###"
+echo
+echo "# cluster_size < 16k"
+_make_test_img -o extended_l2=on,cluster_size=8k 1M
+
+echo "# backing file and preallocation=metadata"
+# For preallocation with backing files, create a backing file first
+$QEMU_IMG create -f raw "$TEST_IMG.base" 1M | _filter_img_create
+$QEMU_IO -c "write -q -P 0xff 0 1M" -f raw "$TEST_IMG.base" | _filter_qemu_io
+
+_make_test_img -o extended_l2=on,preallocation=metadata -F raw -b "$TEST_IMG.base" 512k
+$QEMU_IMG resize "$TEST_IMG" 1M
+$QEMU_IO -c 'read -P 0xff    0 512k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 512k 512k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IMG map "$TEST_IMG" | _filter_testdir
+
+echo "# backing file and preallocation=falloc"
+_make_test_img -o extended_l2=on,preallocation=falloc -F raw -b "$TEST_IMG.base" 512k
+$QEMU_IMG resize "$TEST_IMG" 1M
+$QEMU_IO -c 'read -P 0xff    0 512k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 512k 512k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IMG map "$TEST_IMG" | _filter_testdir
+
+echo "# backing file and preallocation=full"
+_make_test_img -o extended_l2=on,preallocation=full -F raw -b "$TEST_IMG.base" 512k
+$QEMU_IMG resize "$TEST_IMG" 1M
+$QEMU_IO -c 'read -P 0xff    0 512k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 512k 512k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IMG map "$TEST_IMG" | _filter_testdir
+
+echo
+echo "### Image resizing with preallocation and backing files ###"
+echo
+# In this case the new subclusters must have the 'all zeroes' bit set
+echo "# resize --preallocation=metadata"
+_make_test_img -o extended_l2=on -F raw -b "$TEST_IMG.base" 503k
+$QEMU_IMG resize --preallocation=metadata "$TEST_IMG" 1013k
+$QEMU_IO -c 'read -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 503k 510k' "$TEST_IMG" | _filter_qemu_io
+
+# In this case and the next one the new subclusters must be allocated
+echo "# resize --preallocation=falloc"
+_make_test_img -o extended_l2=on -F raw -b "$TEST_IMG.base" 503k
+$QEMU_IMG resize --preallocation=falloc "$TEST_IMG" 1013k
+$QEMU_IO -c 'read -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 503k 510k' "$TEST_IMG" | _filter_qemu_io
+
+echo "# resize --preallocation=full"
+_make_test_img -o extended_l2=on -F raw -b "$TEST_IMG.base" 503k
+$QEMU_IMG resize --preallocation=full "$TEST_IMG" 1013k
+$QEMU_IO -c 'read -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 503k 510k' "$TEST_IMG" | _filter_qemu_io
+
+echo
+echo "### Image resizing with preallocation without backing files ###"
+echo
+# In this case the new subclusters must have the 'all zeroes' bit set
+echo "# resize --preallocation=metadata"
+_make_test_img -o extended_l2=on 503k
+$QEMU_IO -c 'write -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IMG resize --preallocation=metadata "$TEST_IMG" 1013k
+$QEMU_IO -c 'read -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 503k 510k' "$TEST_IMG" | _filter_qemu_io
+
+# In this case and the next one the new subclusters must be allocated
+echo "# resize --preallocation=falloc"
+_make_test_img -o extended_l2=on 503k
+$QEMU_IO -c 'write -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IMG resize --preallocation=falloc "$TEST_IMG" 1013k
+$QEMU_IO -c 'read -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 503k 510k' "$TEST_IMG" | _filter_qemu_io
+
+echo "# resize --preallocation=full"
+_make_test_img -o extended_l2=on 503k
+$QEMU_IO -c 'write -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IMG resize --preallocation=full "$TEST_IMG" 1013k
+$QEMU_IO -c 'read -P 0xff    0 503k' "$TEST_IMG" | _filter_qemu_io
+$QEMU_IO -c 'read -P 0x00 503k 510k' "$TEST_IMG" | _filter_qemu_io
+
+echo
+echo "### qemu-img measure ###"
+echo
+echo "# 512MB, extended_l2=off" # This needs one L2 table
+$QEMU_IMG measure --size 512M -O qcow2 -o extended_l2=off
+echo "# 512MB, extended_l2=on"  # This needs two L2 tables
+$QEMU_IMG measure --size 512M -O qcow2 -o extended_l2=on
+
+echo "# 16K clusters, 64GB, extended_l2=off" # This needs one full L1 table cluster
+$QEMU_IMG measure --size 64G -O qcow2 -o cluster_size=16k,extended_l2=off
+echo "# 16K clusters, 64GB, extended_l2=on"  # This needs two full L2 table clusters
+$QEMU_IMG measure --size 64G -O qcow2 -o cluster_size=16k,extended_l2=on
+
+echo "# 8k clusters" # This should fail
+$QEMU_IMG measure --size 1M -O qcow2 -o cluster_size=8k,extended_l2=on
+
+echo "# 1024 TB" # Maximum allowed size with extended_l2=on and 64K clusters
+$QEMU_IMG measure --size 1024T -O qcow2 -o extended_l2=on
+echo "# 1025 TB" # This should fail
+$QEMU_IMG measure --size 1025T -O qcow2 -o extended_l2=on
+
+echo
+echo "### qemu-img amend ###"
+echo
+_make_test_img -o extended_l2=on 1M
+$QEMU_IMG amend -o extended_l2=off "$TEST_IMG" && echo "Unexpected pass"
+
+_make_test_img -o extended_l2=off 1M
+$QEMU_IMG amend -o extended_l2=on "$TEST_IMG" && echo "Unexpected pass"
+
+echo
+echo "### Test copy-on-write on an image with snapshots ###"
+echo
+_make_test_img -o extended_l2=on 1M
+
+# For each cluster from #0 to #9 this loop zeroes subcluster #7
+# and allocates subclusters #13 and #18.
+alloc="13 18"; zero="7"
+for c in $(seq 0 9); do
+    $QEMU_IO -c "write -q -z $((64*$c+14))k 2k" \
+             -c "write -q -P $((0xd0+$c)) $((64*$c+26))k 2k" \
+             -c "write -q -P $((0xe0+$c)) $((64*$c+36))k 2k" "$TEST_IMG"
+    _verify_l2_bitmap "$c"
+done
+
+# Create a snapshot and set l2_offset to the new L2 table
+$QEMU_IMG snapshot -c snap1 "$TEST_IMG"
+l2_offset=$((0x110000))
+
+# Write different patterns to each one of the clusters
+# in order to see how copy-on-write behaves in each case.
+$QEMU_IO -c "write -q -P 0xf0 $((64*0+30))k 1k" \
+         -c "write -q -P 0xf1 $((64*1+20))k 1k" \
+         -c "write -q -P 0xf2 $((64*2+40))k 1k" \
+         -c "write -q -P 0xf3 $((64*3+26))k 1k" \
+         -c "write -q -P 0xf4 $((64*4+14))k 1k" \
+         -c "write -q -P 0xf5 $((64*5+1))k  1k" \
+         -c "write -q -z      $((64*6+30))k 3k" \
+         -c "write -q -z      $((64*7+26))k 2k" \
+         -c "write -q -z      $((64*8+26))k 1k" \
+         -c "write -q -z      $((64*9+12))k 1k" \
+         "$TEST_IMG"
+alloc="$(seq 13 18)"; zero="7" _verify_l2_bitmap 0
+alloc="$(seq 10 18)"; zero="7" _verify_l2_bitmap 1
+alloc="$(seq 13 20)"; zero="7" _verify_l2_bitmap 2
+alloc="$(seq 13 18)"; zero="7" _verify_l2_bitmap 3
+alloc="$(seq 7 18)";  zero=""  _verify_l2_bitmap 4
+alloc="$(seq 0 18)";  zero=""  _verify_l2_bitmap 5
+alloc="13 18";  zero="7 15 16" _verify_l2_bitmap 6
+alloc="18";        zero="7 13" _verify_l2_bitmap 7
+alloc="$(seq 13 18)"; zero="7" _verify_l2_bitmap 8
+alloc="13 18";      zero="6 7" _verify_l2_bitmap 9
+
+echo
+echo "### Test concurrent requests ###"
+echo
+
+_concurrent_io()
+{
+# Allocate three subclusters in the same cluster.
+# This works because handle_dependencies() checks whether the requests
+# allocate the same cluster, even if the COW regions don't overlap (in
+# this case they don't).
+cat <<EOF
+open -o driver=$IMGFMT blkdebug::$TEST_IMG
+break write_aio A
+aio_write -P 10 30k 2k
+wait_break A
+aio_write -P 11 20k 2k
+aio_write -P 12 40k 2k
+resume A
+aio_flush
+EOF
+}
+
+_concurrent_verify()
+{
+cat <<EOF
+open -o driver=$IMGFMT $TEST_IMG
+read -q -P 10 30k 2k
+read -q -P 11 20k 2k
+read -q -P 12 40k 2k
+EOF
+}
+
+_make_test_img -o extended_l2=on 1M
+_concurrent_io     | $QEMU_IO | _filter_qemu_io
+_concurrent_verify | $QEMU_IO | _filter_qemu_io
+
+# success, all done
+echo "*** done"
+rm -f $seq.full
+status=0
diff --git a/tests/qemu-iotests/271.out b/tests/qemu-iotests/271.out
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/tests/qemu-iotests/271.out
@@ -XXX,XX +XXX,XX @@
+QA output created by 271
+
+### Standard write tests (backing file: yes) ###
+
+Formatting 'TEST_DIR/t.IMGFMT.raw', fmt=raw size=1048576
+Formatting 'TEST_DIR/t.IMGFMT.base', fmt=raw size=1048576
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+write -q -P PATTERN 0 1k
+L2 entry #0: 0x8000000000050000 0000000000000001
+write -q -P PATTERN 3k 512
+L2 entry #0: 0x8000000000050000 0000000000000003
+write -q -P PATTERN 5k 1k
+L2 entry #0: 0x8000000000050000 0000000000000007
+write -q -P PATTERN 6k 2k
+L2 entry #0: 0x8000000000050000 000000000000000f
+write -q -P PATTERN 8k 6k
+L2 entry #0: 0x8000000000050000 000000000000007f
+write -q -P PATTERN 15k 4k
+L2 entry #0: 0x8000000000050000 00000000000003ff
+write -q -P PATTERN 32k 1k
+L2 entry #0: 0x8000000000050000 00000000000103ff
+write -q -P PATTERN 63k 4k
+L2 entry #0: 0x8000000000050000 00000000800103ff
+L2 entry #1: 0x8000000000060000 0000000000000003
+write -q -z 2k 2k
+L2 entry #0: 0x8000000000050000 00000002800103fd
+write -q -z 0 64k
+L2 entry #0: 0x8000000000050000 ffffffff00000000
+write -q -P PATTERN 0 64k
+L2 entry #0: 0x8000000000050000 00000000ffffffff
+write -q -z -u 0 32k
+L2 entry #0: 0x8000000000050000 0000ffffffff0000
+write -q -z -u 0 64k
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+write -q -P PATTERN 3k 512
+L2 entry #0: 0x8000000000050000 fffffffd00000002
+write -q -P PATTERN 0 64k
+L2 entry #0: 0x8000000000050000 00000000ffffffff
+discard -q 0 64k
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+write -q -c -P PATTERN 0 64k
+L2 entry #0: 0x4000000000050000 0000000000000000
+write -q -P PATTERN 3k 512
+L2 entry #0: 0x8000000000070000 00000000ffffffff
+
+### Standard write tests (backing file: no) ###
+
+Formatting 'TEST_DIR/t.IMGFMT.raw', fmt=raw size=1048576
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+write -q -P PATTERN 0 1k
+L2 entry #0: 0x8000000000050000 0000000000000001
+write -q -P PATTERN 3k 512
+L2 entry #0: 0x8000000000050000 0000000000000003
+write -q -P PATTERN 5k 1k
+L2 entry #0: 0x8000000000050000 0000000000000007
+write -q -P PATTERN 6k 2k
+L2 entry #0: 0x8000000000050000 000000000000000f
+write -q -P PATTERN 8k 6k
+L2 entry #0: 0x8000000000050000 000000000000007f
+write -q -P PATTERN 15k 4k
+L2 entry #0: 0x8000000000050000 00000000000003ff
+write -q -P PATTERN 32k 1k
+L2 entry #0: 0x8000000000050000 00000000000103ff
+write -q -P PATTERN 63k 4k
+L2 entry #0: 0x8000000000050000 00000000800103ff
+L2 entry #1: 0x8000000000060000 0000000000000003
+write -q -z 2k 2k
+L2 entry #0: 0x8000000000050000 00000002800103fd
+write -q -z 0 64k
+L2 entry #0: 0x8000000000050000 ffffffff00000000
+write -q -P PATTERN 0 64k
+L2 entry #0: 0x8000000000050000 00000000ffffffff
+write -q -z -u 0 32k
+L2 entry #0: 0x8000000000050000 0000ffffffff0000
+write -q -z -u 0 64k
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+write -q -P PATTERN 3k 512
+L2 entry #0: 0x8000000000050000 fffffffd00000002
+write -q -P PATTERN 0 64k
+L2 entry #0: 0x8000000000050000 00000000ffffffff
+discard -q 0 64k
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+write -q -c -P PATTERN 0 64k
+L2 entry #0: 0x4000000000050000 0000000000000000
+write -q -P PATTERN 3k 512
+L2 entry #0: 0x8000000000070000 00000000ffffffff
+
+### Overwriting several clusters without COW ###
+
+Formatting 'TEST_DIR/t.IMGFMT.raw', fmt=raw size=1048576
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+write -q -P PATTERN 24k 40k
+L2 entry #0: 0x8000000000050000 00000000fffff000
+write -q -P PATTERN 90k 2k
+L2 entry #1: 0x8000000000060000 0000000000002000
+write -q -P PATTERN 156k 2k
+L2 entry #2: 0x8000000000070000 0000000000004000
+write -q -z 156k 2k
+L2 entry #2: 0x8000000000070000 0000400000000000
+write -q -P PATTERN 192k 34k
+L2 entry #3: 0x8000000000080000 000000000001ffff
+write -q -P PATTERN 24k 192k
+L2 entry #0: 0x8000000000050000 00000000fffff000
+L2 entry #1: 0x8000000000060000 00000000ffffffff
+L2 entry #2: 0x8000000000070000 00000000ffffffff
+L2 entry #3: 0x8000000000080000 000000000001ffff
+
+### Writing zeroes 1: unallocated clusters (backing file: yes) ###
+
+Formatting 'TEST_DIR/t.IMGFMT.raw', fmt=raw size=2132992
+Formatting 'TEST_DIR/t.IMGFMT.base', fmt=raw size=2132992
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=2132992 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+write -q -z 0 192k
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+L2 entry #1: 0x0000000000000000 ffffffff00000000
+L2 entry #2: 0x0000000000000000 ffffffff00000000
+write -q -z 224k 128k
+L2 entry #3: 0x0000000000000000 ffff000000000000
+L2 entry #4: 0x0000000000000000 ffffffff00000000
+L2 entry #5: 0x0000000000000000 0000ffff00000000
+write -q -z 415k 128k
+L2 entry #6: 0x8000000000050000 ffff000000008000
+L2 entry #7: 0x0000000000000000 ffffffff00000000
+L2 entry #8: 0x8000000000060000 00007fff00008000
+
+### Writing zeroes 2: allocated clusters (backing file: yes) ###
+
+write -q -P PATTERN 576k 576k
+L2 entry #9: 0x8000000000070000 00000000ffffffff
+L2 entry #10: 0x8000000000080000 00000000ffffffff
+L2 entry #11: 0x8000000000090000 00000000ffffffff
+L2 entry #12: 0x80000000000a0000 00000000ffffffff
+L2 entry #13: 0x80000000000b0000 00000000ffffffff
+L2 entry #14: 0x80000000000c0000 00000000ffffffff
+L2 entry #15: 0x80000000000d0000 00000000ffffffff
+L2 entry #16: 0x80000000000e0000 00000000ffffffff
+L2 entry #17: 0x80000000000f0000 00000000ffffffff
+write -q -z 576k 192k
+L2 entry #9: 0x8000000000070000 ffffffff00000000
+L2 entry #10: 0x8000000000080000 ffffffff00000000
+L2 entry #11: 0x8000000000090000 ffffffff00000000
+write -q -z 800k 128k
+L2 entry #12: 0x80000000000a0000 ffff00000000ffff
+L2 entry #13: 0x80000000000b0000 ffffffff00000000
+L2 entry #14: 0x80000000000c0000 0000ffffffff0000
+write -q -z 991k 128k
+L2 entry #15: 0x80000000000d0000 ffff00000000ffff
+L2 entry #16: 0x80000000000e0000 ffffffff00000000
+L2 entry #17: 0x80000000000f0000 00007fffffff8000
+
+### Writing zeroes 3: compressed clusters (backing file: yes) ###
+
+write -q -c -P PATTERN 1152k 64k
+L2 entry #18: 0x4000000000100000 0000000000000000
+write -q -c -P PATTERN 1216k 64k
+L2 entry #19: 0x4000000000110000 0000000000000000
+write -q -c -P PATTERN 1280k 64k
+L2 entry #20: 0x4000000000120000 0000000000000000
+write -q -c -P PATTERN 1344k 64k
+L2 entry #21: 0x4000000000130000 0000000000000000
+write -q -c -P PATTERN 1408k 64k
+L2 entry #22: 0x4000000000140000 0000000000000000
+write -q -c -P PATTERN 1472k 64k
+L2 entry #23: 0x4000000000150000 0000000000000000
+write -q -c -P PATTERN 1536k 64k
+L2 entry #24: 0x4000000000160000 0000000000000000
+write -q -c -P PATTERN 1600k 64k
+L2 entry #25: 0x4000000000170000 0000000000000000
+write -q -c -P PATTERN 1664k 64k
+L2 entry #26: 0x4000000000180000 0000000000000000
+write -q -c -P PATTERN 1728k 64k
+L2 entry #27: 0x4000000000190000 0000000000000000
+write -q -c -P PATTERN 1792k 64k
+L2 entry #28: 0x40000000001a0000 0000000000000000
+write -q -z 1152k 192k
+L2 entry #18: 0x0000000000000000 ffffffff00000000
+L2 entry #19: 0x0000000000000000 ffffffff00000000
+L2 entry #20: 0x0000000000000000 ffffffff00000000
+write -q -z 1376k 128k
+L2 entry #21: 0x8000000000100000 00000000ffffffff
+L2 entry #22: 0x8000000000110000 00000000ffffffff
+L2 entry #23: 0x8000000000120000 00000000ffffffff
+write -q -z 1567k 129k
+L2 entry #24: 0x8000000000130000 00000000ffffffff
+L2 entry #25: 0x8000000000140000 00000000ffffffff
+L2 entry #26: 0x8000000000150000 00000000ffffffff
+write -q -z 1759k 128k
+L2 entry #27: 0x8000000000160000 ffff00000000ffff
+L2 entry #28: 0x0000000000000000 ffffffff00000000
+L2 entry #29: 0x8000000000170000 00007fff00008000
+
+### Writing zeroes 4: other tests (backing file: yes) ###
+
+write -q -z 1951k 8k
+L2 entry #30: 0x8000000000180000 0007000000088000
+write -q -z 2048k 35k
+L2 entry #32: 0x0000000000000000 0003ffff00000000
+
+### Writing zeroes 1: unallocated clusters (backing file: no) ###
+
+Formatting 'TEST_DIR/t.IMGFMT.raw', fmt=raw size=2132992
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=2132992
+write -q -z 0 192k
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+L2 entry #1: 0x0000000000000000 ffffffff00000000
+L2 entry #2: 0x0000000000000000 ffffffff00000000
+write -q -z 224k 128k
+L2 entry #3: 0x0000000000000000 ffff000000000000
+L2 entry #4: 0x0000000000000000 ffffffff00000000
+L2 entry #5: 0x0000000000000000 0000ffff00000000
+write -q -z 415k 128k
+L2 entry #6: 0x0000000000000000 ffff800000000000
+L2 entry #7: 0x0000000000000000 ffffffff00000000
+L2 entry #8: 0x0000000000000000 0000ffff00000000
+
+### Writing zeroes 2: allocated clusters (backing file: no) ###
+
+write -q -P PATTERN 576k 576k
+L2 entry #9: 0x8000000000050000 00000000ffffffff
+L2 entry #10: 0x8000000000060000 00000000ffffffff
+L2 entry #11: 0x8000000000070000 00000000ffffffff
+L2 entry #12: 0x8000000000080000 00000000ffffffff
+L2 entry #13: 0x8000000000090000 00000000ffffffff
+L2 entry #14: 0x80000000000a0000 00000000ffffffff
+L2 entry #15: 0x80000000000b0000 00000000ffffffff
+L2 entry #16: 0x80000000000c0000 00000000ffffffff
+L2 entry #17: 0x80000000000d0000 00000000ffffffff
+write -q -z 576k 192k
+L2 entry #9: 0x8000000000050000 ffffffff00000000
+L2 entry #10: 0x8000000000060000 ffffffff00000000
+L2 entry #11: 0x8000000000070000 ffffffff00000000
+write -q -z 800k 128k
+L2 entry #12: 0x8000000000080000 ffff00000000ffff
+L2 entry #13: 0x8000000000090000 ffffffff00000000
+L2 entry #14: 0x80000000000a0000 0000ffffffff0000
+write -q -z 991k 128k
+L2 entry #15: 0x80000000000b0000 ffff00000000ffff
+L2 entry #16: 0x80000000000c0000 ffffffff00000000
+L2 entry #17: 0x80000000000d0000 00007fffffff8000
+
+### Writing zeroes 3: compressed clusters (backing file: no) ###
+
+write -q -c -P PATTERN 1152k 64k
+L2 entry #18: 0x40000000000e0000 0000000000000000
+write -q -c -P PATTERN 1216k 64k
+L2 entry #19: 0x40000000000f0000 0000000000000000
+write -q -c -P PATTERN 1280k 64k
+L2 entry #20: 0x4000000000100000 0000000000000000
+write -q -c -P PATTERN 1344k 64k
+L2 entry #21: 0x4000000000110000 0000000000000000
+write -q -c -P PATTERN 1408k 64k
+L2 entry #22: 0x4000000000120000 0000000000000000
+write -q -c -P PATTERN 1472k 64k
+L2 entry #23: 0x4000000000130000 0000000000000000
+write -q -c -P PATTERN 1536k 64k
+L2 entry #24: 0x4000000000140000 0000000000000000
+write -q -c -P PATTERN 1600k 64k
+L2 entry #25: 0x4000000000150000 0000000000000000
+write -q -c -P PATTERN 1664k 64k
+L2 entry #26: 0x4000000000160000 0000000000000000
+write -q -c -P PATTERN 1728k 64k
+L2 entry #27: 0x4000000000170000 0000000000000000
+write -q -c -P PATTERN 1792k 64k
+L2 entry #28: 0x4000000000180000 0000000000000000
+write -q -z 1152k 192k
+L2 entry #18: 0x0000000000000000 ffffffff00000000
+L2 entry #19: 0x0000000000000000 ffffffff00000000
+L2 entry #20: 0x0000000000000000 ffffffff00000000
+write -q -z 1376k 128k
+L2 entry #21: 0x80000000000e0000 00000000ffffffff
+L2 entry #22: 0x80000000000f0000 00000000ffffffff
+L2 entry #23: 0x8000000000100000 00000000ffffffff
+write -q -z 1567k 129k
+L2 entry #24: 0x8000000000110000 00000000ffffffff
+L2 entry #25: 0x8000000000120000 00000000ffffffff
+L2 entry #26: 0x8000000000130000 00000000ffffffff
+write -q -z 1759k 128k
+L2 entry #27: 0x8000000000140000 ffff00000000ffff
+L2 entry #28: 0x0000000000000000 ffffffff00000000
+L2 entry #29: 0x0000000000000000 0000ffff00000000
+
+### Writing zeroes 4: other tests (backing file: no) ###
+
+write -q -z 1951k 8k
+L2 entry #30: 0x0000000000000000 000f800000000000
+write -q -z 2048k 35k
+L2 entry #32: 0x0000000000000000 0003ffff00000000
+
+### Zero + unmap 1: allocated clusters (backing file: yes) ###
+
+Formatting 'TEST_DIR/t.IMGFMT.raw', fmt=raw size=2132992
+Formatting 'TEST_DIR/t.IMGFMT.base', fmt=raw size=2132992
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=2132992 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+write -q -P PATTERN 576k 576k
+L2 entry #9: 0x8000000000050000 00000000ffffffff
+L2 entry #10: 0x8000000000060000 00000000ffffffff
+L2 entry #11: 0x8000000000070000 00000000ffffffff
+L2 entry #12: 0x8000000000080000 00000000ffffffff
+L2 entry #13: 0x8000000000090000 00000000ffffffff
+L2 entry #14: 0x80000000000a0000 00000000ffffffff
+L2 entry #15: 0x80000000000b0000 00000000ffffffff
+L2 entry #16: 0x80000000000c0000 00000000ffffffff
+L2 entry #17: 0x80000000000d0000 00000000ffffffff
+write -q -z -u 576k 192k
+L2 entry #9: 0x0000000000000000 ffffffff00000000
+L2 entry #10: 0x0000000000000000 ffffffff00000000
+L2 entry #11: 0x0000000000000000 ffffffff00000000
+write -q -z -u 800k 128k
+L2 entry #12: 0x8000000000080000 ffff00000000ffff
+L2 entry #13: 0x0000000000000000 ffffffff00000000
+L2 entry #14: 0x80000000000a0000 0000ffffffff0000
+write -q -z -u 991k 128k
+L2 entry #15: 0x80000000000b0000 ffff00000000ffff
+L2 entry #16: 0x0000000000000000 ffffffff00000000
+L2 entry #17: 0x80000000000d0000 00007fffffff8000
+
+### Zero + unmap 2: compressed clusters (backing file: yes) ###
+
+write -q -c -P PATTERN 1152k 64k
+L2 entry #18: 0x4000000000050000 0000000000000000
+write -q -c -P PATTERN 1216k 64k
+L2 entry #19: 0x4000000000060000 0000000000000000
+write -q -c -P PATTERN 1280k 64k
+L2 entry #20: 0x4000000000070000 0000000000000000
+write -q -c -P PATTERN 1344k 64k
+L2 entry #21: 0x4000000000090000 0000000000000000
+write -q -c -P PATTERN 1408k 64k
+L2 entry #22: 0x40000000000c0000 0000000000000000
+write -q -c -P PATTERN 1472k 64k
+L2 entry #23: 0x40000000000e0000 0000000000000000
+write -q -c -P PATTERN 1536k 64k
+L2 entry #24: 0x40000000000f0000 0000000000000000
+write -q -c -P PATTERN 1600k 64k
+L2 entry #25: 0x4000000000100000 0000000000000000
+write -q -c -P PATTERN 1664k 64k
+L2 entry #26: 0x4000000000110000 0000000000000000
+write -q -c -P PATTERN 1728k 64k
+L2 entry #27: 0x4000000000120000 0000000000000000
+write -q -c -P PATTERN 1792k 64k
+L2 entry #28: 0x4000000000130000 0000000000000000
+write -q -z -u 1152k 192k
+L2 entry #18: 0x0000000000000000 ffffffff00000000
+L2 entry #19: 0x0000000000000000 ffffffff00000000
+L2 entry #20: 0x0000000000000000 ffffffff00000000
+write -q -z -u 1376k 128k
+L2 entry #21: 0x8000000000050000 00000000ffffffff
+L2 entry #22: 0x8000000000060000 00000000ffffffff
+L2 entry #23: 0x8000000000070000 00000000ffffffff
+write -q -z -u 1567k 129k
+L2 entry #24: 0x8000000000090000 00000000ffffffff
+L2 entry #25: 0x80000000000e0000 00000000ffffffff
+L2 entry #26: 0x80000000000f0000 00000000ffffffff
+write -q -z -u 1759k 128k
+L2 entry #27: 0x80000000000c0000 ffff00000000ffff
+L2 entry #28: 0x0000000000000000 ffffffff00000000
+L2 entry #29: 0x8000000000100000 00007fff00008000
+
+### Zero + unmap 1: allocated clusters (backing file: no) ###
+
+Formatting 'TEST_DIR/t.IMGFMT.raw', fmt=raw size=2132992
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=2132992
+write -q -P PATTERN 576k 576k
+L2 entry #9: 0x8000000000050000 00000000ffffffff
+L2 entry #10: 0x8000000000060000 00000000ffffffff
+L2 entry #11: 0x8000000000070000 00000000ffffffff
+L2 entry #12: 0x8000000000080000 00000000ffffffff
+L2 entry #13: 0x8000000000090000 00000000ffffffff
+L2 entry #14: 0x80000000000a0000 00000000ffffffff
+L2 entry #15: 0x80000000000b0000 00000000ffffffff
+L2 entry #16: 0x80000000000c0000 00000000ffffffff
+L2 entry #17: 0x80000000000d0000 00000000ffffffff
+write -q -z -u 576k 192k
+L2 entry #9: 0x0000000000000000 ffffffff00000000
+L2 entry #10: 0x0000000000000000 ffffffff00000000
+L2 entry #11: 0x0000000000000000 ffffffff00000000
+write -q -z -u 800k 128k
+L2 entry #12: 0x8000000000080000 ffff00000000ffff
+L2 entry #13: 0x0000000000000000 ffffffff00000000
+L2 entry #14: 0x80000000000a0000 0000ffffffff0000
+write -q -z -u 991k 128k
+L2 entry #15: 0x80000000000b0000 ffff00000000ffff
+L2 entry #16: 0x0000000000000000 ffffffff00000000
+L2 entry #17: 0x80000000000d0000 00007fffffff8000
+
+### Zero + unmap 2: compressed clusters (backing file: no) ###
+
+write -q -c -P PATTERN 1152k 64k
+L2 entry #18: 0x4000000000050000 0000000000000000
+write -q -c -P PATTERN 1216k 64k
+L2 entry #19: 0x4000000000060000 0000000000000000
+write -q -c -P PATTERN 1280k 64k
+L2 entry #20: 0x4000000000070000 0000000000000000
+write -q -c -P PATTERN 1344k 64k
+L2 entry #21: 0x4000000000090000 0000000000000000
+write -q -c -P PATTERN 1408k 64k
+L2 entry #22: 0x40000000000c0000 0000000000000000
+write -q -c -P PATTERN 1472k 64k
+L2 entry #23: 0x40000000000e0000 0000000000000000
+write -q -c -P PATTERN 1536k 64k
+L2 entry #24: 0x40000000000f0000 0000000000000000
+write -q -c -P PATTERN 1600k 64k
+L2 entry #25: 0x4000000000100000 0000000000000000
+write -q -c -P PATTERN 1664k 64k
+L2 entry #26: 0x4000000000110000 0000000000000000
+write -q -c -P PATTERN 1728k 64k
+L2 entry #27: 0x4000000000120000 0000000000000000
+write -q -c -P PATTERN 1792k 64k
+L2 entry #28: 0x4000000000130000 0000000000000000
+write -q -z -u 1152k 192k
+L2 entry #18: 0x0000000000000000 ffffffff00000000
+L2 entry #19: 0x0000000000000000 ffffffff00000000
+L2 entry #20: 0x0000000000000000 ffffffff00000000
+write -q -z -u 1376k 128k
+L2 entry #21: 0x8000000000050000 00000000ffffffff
+L2 entry #22: 0x8000000000060000 00000000ffffffff
+L2 entry #23: 0x8000000000070000 00000000ffffffff
+write -q -z -u 1567k 129k
+L2 entry #24: 0x8000000000090000 00000000ffffffff
+L2 entry #25: 0x80000000000e0000 00000000ffffffff
+L2 entry #26: 0x80000000000f0000 00000000ffffffff
+write -q -z -u 1759k 128k
+L2 entry #27: 0x80000000000c0000 ffff00000000ffff
+L2 entry #28: 0x0000000000000000 ffffffff00000000
+L2 entry #29: 0x0000000000000000 0000ffff00000000
+
+### Discarding clusters with non-zero bitmaps (backing file: yes) ###
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+L2 entry #1: 0x0000000000000000 ffffffff00000000
+Image resized.
+Image resized.
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+L2 entry #1: 0x0000000000000000 ffffffff00000000
+
+### Discarding clusters with non-zero bitmaps (backing file: no) ###
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x0000000000000000 ffffffff00000000
+L2 entry #1: 0x0000000000000000 ffffffff00000000
+Image resized.
+Image resized.
+L2 entry #0: 0x0000000000000000 0000ffff00000000
+L2 entry #1: 0x0000000000000000 0000000000000000
+
+### Corrupted L2 entries - read test (allocated) ###
+
+# 'cluster is zero' bit set on the standard cluster descriptor
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x8000000000050001 0000000000000001
+L2 entry #0: 0x8000000000050001 0000000000000001
+
+# Both 'subcluster is zero' and 'subcluster is allocated' bits set
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #1: 0x8000000000060000 00000001ffffffff
+qcow2: Marking image as corrupt: Invalid cluster entry found  (L2 offset: 0x40000, L2 index: 0x1); further corruption events will be suppressed
+read failed: Input/output error
+
+### Corrupted L2 entries - read test (unallocated) ###
+
+# 'cluster is zero' bit set on the standard cluster descriptor
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x0000000000000001 0000000000000000
+L2 entry #0: 0x0000000000000001 0000000000000000
+
+# 'subcluster is allocated' bit set
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x0000000000000000 0000000000000001
+qcow2: Marking image as corrupt: Invalid cluster entry found  (L2 offset: 0x40000, L2 index: 0); further corruption events will be suppressed
+read failed: Input/output error
+
+# Both 'subcluster is zero' and 'subcluster is allocated' bits set
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #1: 0x0000000000000000 0000000100000001
+qcow2: Marking image as corrupt: Invalid cluster entry found  (L2 offset: 0x40000, L2 index: 0x1); further corruption events will be suppressed
+read failed: Input/output error
+
+### Compressed cluster with subcluster bitmap != 0 - read test ###
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x4000000000050000 0000000180000000
+read 65536/65536 bytes at offset 0
+64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+L2 entry #0: 0x4000000000050000 0000000180000000
+
+### Corrupted L2 entries - write test (allocated) ###
+
+# 'cluster is zero' bit set on the standard cluster descriptor
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x8000000000050001 0000000000000001
+L2 entry #0: 0x8000000000050001 0000000000000001
+
+# Both 'subcluster is zero' and 'subcluster is allocated' bits set
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #1: 0x8000000000060000 00000001ffffffff
+qcow2: Marking image as corrupt: Invalid cluster entry found  (L2 offset: 0x40000, L2 index: 0x1); further corruption events will be suppressed
+write failed: Input/output error
+
+### Corrupted L2 entries - write test (unallocated) ###
+
+# 'cluster is zero' bit set on the standard cluster descriptor
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x0000000000000001 0000000000000000
+L2 entry #0: 0x8000000000060000 0000000000000001
+
+# 'subcluster is allocated' bit set
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x0000000000000000 0000000000000001
+qcow2: Marking image as corrupt: Invalid cluster entry found  (L2 offset: 0x40000, L2 index: 0); further corruption events will be suppressed
+write failed: Input/output error
+
+# Both 'subcluster is zero' and 'subcluster is allocated' bits set
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #1: 0x0000000000000000 0000000100000001
+qcow2: Marking image as corrupt: Invalid cluster entry found  (L2 offset: 0x40000, L2 index: 0x1); further corruption events will be suppressed
+write failed: Input/output error
+
+### Compressed cluster with subcluster bitmap != 0 - write test ###
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x4000000000050000 0000000180000000
+wrote 65536/65536 bytes at offset 0
+64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+L2 entry #0: 0x8000000000060000 00000000ffffffff
+
+### Detect and repair unaligned clusters ###
+
+Formatting 'TEST_DIR/t.IMGFMT.base', fmt=raw size=131072
+# Corrupted L2 entry, allocated subcluster #
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=131072 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+ERROR offset=50200: Data cluster is not properly aligned; L2 entry corrupted.
+ERROR cluster 6 refcount=0 reference=1
+Rebuilding refcount structure
+ERROR offset=50200: Data cluster is not properly aligned; L2 entry corrupted.
+Repairing cluster 1 refcount=1 reference=0
+Repairing cluster 2 refcount=1 reference=0
+ERROR offset=50200: Data cluster is not properly aligned; L2 entry corrupted.
+The following inconsistencies were found and repaired:
+
+    0 leaked clusters
+    1 corruptions
+
+Double checking the fixed image now...
+
+1 errors were found on the image.
+Data may be corrupted, or further writes to the image may corrupt it.
+qcow2: Marking image as corrupt: Cluster allocation offset 0x50200 unaligned (L2 offset: 0x40000, L2 index: 0); further corruption events will be suppressed
+read failed: Input/output error
+# Corrupted L2 entry, no allocated subclusters #
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=131072 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+Repairing offset=50200: Preallocated cluster is not properly aligned; L2 entry corrupted.
+Leaked cluster 5 refcount=1 reference=0
+Repairing cluster 5 refcount=1 reference=0
+The following inconsistencies were found and repaired:
+
+    1 leaked clusters
+    1 corruptions
+
+Double checking the fixed image now...
+No errors were found on the image.
+
+### Image creation options ###
+
+# cluster_size < 16k
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+qemu-img: TEST_DIR/t.IMGFMT: Extended L2 entries are only supported with cluster sizes of at least 16384 bytes
+# backing file and preallocation=metadata
+Formatting 'TEST_DIR/t.IMGFMT.base', fmt=raw size=1048576
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=524288 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw preallocation=metadata
+Image resized.
+read 524288/524288 bytes at offset 0
+512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 524288/524288 bytes at offset 524288
+512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+Offset          Length          Mapped to       File
+0               0x80000        0               TEST_DIR/t.qcow2.base
+# backing file and preallocation=falloc
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=524288 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw preallocation=falloc
+Image resized.
+read 524288/524288 bytes at offset 0
+512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 524288/524288 bytes at offset 524288
+512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+Offset          Length          Mapped to       File
+0               0x80000        0               TEST_DIR/t.qcow2.base
+# backing file and preallocation=full
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=524288 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw preallocation=full
+Image resized.
+read 524288/524288 bytes at offset 0
+512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 524288/524288 bytes at offset 524288
+512 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+Offset          Length          Mapped to       File
+0               0x80000        0               TEST_DIR/t.qcow2.base
+
+### Image resizing with preallocation and backing files ###
+
+# resize --preallocation=metadata
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=515072 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+Image resized.
+read 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 522240/522240 bytes at offset 515072
+510 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+# resize --preallocation=falloc
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=515072 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+Image resized.
+read 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 522240/522240 bytes at offset 515072
+510 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+# resize --preallocation=full
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=515072 backing_file=TEST_DIR/t.IMGFMT.base backing_fmt=raw
+Image resized.
+read 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 522240/522240 bytes at offset 515072
+510 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+
+### Image resizing with preallocation without backing files ###
+
+# resize --preallocation=metadata
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=515072
+wrote 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+Image resized.
+read 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 522240/522240 bytes at offset 515072
+510 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+# resize --preallocation=falloc
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=515072
+wrote 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+Image resized.
+read 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 522240/522240 bytes at offset 515072
+510 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+# resize --preallocation=full
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=515072
+wrote 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+Image resized.
+read 515072/515072 bytes at offset 0
+503 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+read 522240/522240 bytes at offset 515072
+510 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+
+### qemu-img measure ###
+
+# 512MB, extended_l2=off
+required size: 327680
+fully allocated size: 537198592
+# 512MB, extended_l2=on
+required size: 393216
+fully allocated size: 537264128
+# 16K clusters, 64GB, extended_l2=off
+required size: 42008576
+fully allocated size: 68761485312
+# 16K clusters, 64GB, extended_l2=on
+required size: 75579392
+fully allocated size: 68795056128
+# 8k clusters
+qemu-img: Extended L2 entries are only supported with cluster sizes of at least 16384 bytes
+# 1024 TB
+required size: 309285027840
+fully allocated size: 1126209191870464
+# 1025 TB
+qemu-img: The image size is too large (try using a larger cluster size)
+
+### qemu-img amend ###
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+qemu-img: Invalid parameter 'extended_l2'
+This option is only supported for image creation
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+qemu-img: Invalid parameter 'extended_l2'
+This option is only supported for image creation
+
+### Test copy-on-write on an image with snapshots ###
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+L2 entry #0: 0x8000000000050000 0000008000042000
+L2 entry #1: 0x8000000000060000 0000008000042000
+L2 entry #2: 0x8000000000070000 0000008000042000
+L2 entry #3: 0x8000000000080000 0000008000042000
+L2 entry #4: 0x8000000000090000 0000008000042000
+L2 entry #5: 0x80000000000a0000 0000008000042000
+L2 entry #6: 0x80000000000b0000 0000008000042000
+L2 entry #7: 0x80000000000c0000 0000008000042000
+L2 entry #8: 0x80000000000d0000 0000008000042000
+L2 entry #9: 0x80000000000e0000 0000008000042000
+L2 entry #0: 0x8000000000120000 000000800007e000
+L2 entry #1: 0x8000000000130000 000000800007fc00
+L2 entry #2: 0x8000000000140000 00000080001fe000
+L2 entry #3: 0x8000000000150000 000000800007e000
+L2 entry #4: 0x8000000000160000 000000000007ff80
+L2 entry #5: 0x8000000000170000 000000000007ffff
+L2 entry #6: 0x00000000000b0000 0001808000042000
+L2 entry #7: 0x00000000000c0000 0000208000040000
+L2 entry #8: 0x8000000000180000 000000800007e000
+L2 entry #9: 0x00000000000e0000 000000c000042000
+
+### Test concurrent requests ###
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=1048576
+blkdebug: Suspended request 'A'
+blkdebug: Resuming request 'A'
+wrote 2048/2048 bytes at offset 30720
+2 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+wrote 2048/2048 bytes at offset 20480
+2 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+wrote 2048/2048 bytes at offset 40960
+2 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
+*** done
diff --git a/tests/qemu-iotests/group b/tests/qemu-iotests/group
index XXXXXXX..XXXXXXX 100644
--- a/tests/qemu-iotests/group
+++ b/tests/qemu-iotests/group
@@ -XXX,XX +XXX,XX @@
 267 rw auto quick snapshot
 268 rw auto quick
 270 rw backing quick
+271 rw auto
 272 rw
 273 backing quick
 274 rw backing
-- 
2.26.2

The following changes since commit 1214d55d1c41fbab3a9973a05085b8760647e411:

Merge remote-tracking branch 'remotes/nvme/tags/nvme-next-pull-request' into staging (2021-02-09 13:24:37 +0000)

are available in the Git repository at:

https://gitlab.com/stefanha/qemu.git tags/block-pull-request

for you to fetch changes up to eb847c42296497978942f738cd41dc29a35a49b2:

docs: fix Parallels Image "dirty bitmap" section (2021-02-10 09:23:28 +0000)

----------------------------------------------------------------
Pull request

v4:
 * Add PCI_EXPRESS Kconfig dependency to fix s390x in "multi-process: setup PCI
   host bridge for remote device" [Philippe and Thomas]

----------------------------------------------------------------

Denis V. Lunev (1):
  docs: fix Parallels Image "dirty bitmap" section

Elena Ufimtseva (8):
  multi-process: add configure and usage information
  io: add qio_channel_writev_full_all helper
  io: add qio_channel_readv_full_all_eof & qio_channel_readv_full_all
    helpers
  multi-process: define MPQemuMsg format and transmission functions
  multi-process: introduce proxy object
  multi-process: add proxy communication functions
  multi-process: Forward PCI config space acceses to the remote process
  multi-process: perform device reset in the remote process

Jagannathan Raman (11):
  memory: alloc RAM from file at offset
  multi-process: Add config option for multi-process QEMU
  multi-process: setup PCI host bridge for remote device
  multi-process: setup a machine object for remote device process
  multi-process: Initialize message handler in remote device
  multi-process: Associate fd of a PCIDevice with its object
  multi-process: setup memory manager for remote device
  multi-process: PCI BAR read/write handling for proxy & remote
    endpoints
  multi-process: Synchronize remote memory
  multi-process: create IOHUB object to handle irq
  multi-process: Retrieve PCI info from remote process

John G Johnson (1):
  multi-process: add the concept description to
    docs/devel/qemu-multiprocess

Stefan Hajnoczi (6):
  .github: point Repo Lockdown bot to GitLab repo
  gitmodules: use GitLab repos instead of qemu.org
  gitlab-ci: remove redundant GitLab repo URL command
  docs: update README to use GitLab repo URLs
  pc-bios: update mirror URLs to GitLab
  get_maintainer: update repo URL to GitLab

MAINTAINERS                               |  24 +
 README.rst                                |   4 +-
 docs/devel/index.rst                      |   1 +
 docs/devel/multi-process.rst              | 966 ++++++++++++++++++++++
 docs/system/index.rst                     |   1 +
 docs/system/multi-process.rst             |  64 ++
 docs/interop/parallels.txt                |   2 +-
 configure                                 |  10 +
 meson.build                               |   5 +-
 hw/remote/trace.h                         |   1 +
 include/exec/memory.h                     |   2 +
 include/exec/ram_addr.h                   |   4 +-
 include/hw/pci-host/remote.h              |  30 +
 include/hw/pci/pci_ids.h                  |   3 +
 include/hw/remote/iohub.h                 |  42 +
 include/hw/remote/machine.h               |  38 +
 include/hw/remote/memory.h                |  19 +
 include/hw/remote/mpqemu-link.h           |  99 +++
 include/hw/remote/proxy-memory-listener.h |  28 +
 include/hw/remote/proxy.h                 |  48 ++
 include/io/channel.h                      |  78 ++
 include/qemu/mmap-alloc.h                 |   4 +-
 include/sysemu/iothread.h                 |   6 +
 backends/hostmem-memfd.c                  |   2 +-
 hw/misc/ivshmem.c                         |   3 +-
 hw/pci-host/remote.c                      |  75 ++
 hw/remote/iohub.c                         | 119 +++
 hw/remote/machine.c                       |  80 ++
 hw/remote/memory.c                        |  65 ++
 hw/remote/message.c                       | 230 ++++++
 hw/remote/mpqemu-link.c                   | 267 ++++++
 hw/remote/proxy-memory-listener.c         | 227 +++++
 hw/remote/proxy.c                         | 379 +++++++++
 hw/remote/remote-obj.c                    | 203 +++++
 io/channel.c                              | 116 ++-
 iothread.c                                |   6 +
 softmmu/memory.c                          |   3 +-
 softmmu/physmem.c                         |  12 +-
 util/mmap-alloc.c                         |   8 +-
 util/oslib-posix.c                        |   2 +-
 .github/lockdown.yml                      |   8 +-
 .gitlab-ci.yml                            |   1 -
 .gitmodules                               |  44 +-
 Kconfig.host                              |   4 +
 hw/Kconfig                                |   1 +
 hw/meson.build                            |   1 +
 hw/pci-host/Kconfig                       |   3 +
 hw/pci-host/meson.build                   |   1 +
 hw/remote/Kconfig                         |   4 +
 hw/remote/meson.build                     |  13 +
 hw/remote/trace-events                    |   4 +
 pc-bios/README                            |   4 +-
 scripts/get_maintainer.pl                 |   2 +-
 53 files changed, 3296 insertions(+), 70 deletions(-)
 create mode 100644 docs/devel/multi-process.rst
 create mode 100644 docs/system/multi-process.rst
 create mode 100644 hw/remote/trace.h
 create mode 100644 include/hw/pci-host/remote.h
 create mode 100644 include/hw/remote/iohub.h
 create mode 100644 include/hw/remote/machine.h
 create mode 100644 include/hw/remote/memory.h
 create mode 100644 include/hw/remote/mpqemu-link.h
 create mode 100644 include/hw/remote/proxy-memory-listener.h
 create mode 100644 include/hw/remote/proxy.h
 create mode 100644 hw/pci-host/remote.c
 create mode 100644 hw/remote/iohub.c
 create mode 100644 hw/remote/machine.c
 create mode 100644 hw/remote/memory.c
 create mode 100644 hw/remote/message.c
 create mode 100644 hw/remote/mpqemu-link.c
 create mode 100644 hw/remote/proxy-memory-listener.c
 create mode 100644 hw/remote/proxy.c
 create mode 100644 hw/remote/remote-obj.c
 create mode 100644 hw/remote/Kconfig
 create mode 100644 hw/remote/meson.build
 create mode 100644 hw/remote/trace-events

-- 
2.29.2

Use the GitLab repo URL as the main repo location in order to reduce
load on qemu.org.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-id: 20210111115017.156802-2-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 .github/lockdown.yml | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/.github/lockdown.yml b/.github/lockdown.yml
index XXXXXXX..XXXXXXX 100644
--- a/.github/lockdown.yml
+++ b/.github/lockdown.yml
@@ -XXX,XX +XXX,XX @@ issues:
   comment: |
     Thank you for your interest in the QEMU project.
 
-    This repository is a read-only mirror of the project's master
-    repostories hosted on https://git.qemu.org/git/qemu.git.
+    This repository is a read-only mirror of the project's repostories hosted
+    at https://gitlab.com/qemu-project/qemu.git.
     The project does not process issues filed on GitHub.
 
     The project issues are tracked on Launchpad:
@@ -XXX,XX +XXX,XX @@ pulls:
   comment: |
     Thank you for your interest in the QEMU project.
 
-    This repository is a read-only mirror of the project's master
-    repostories hosted on https://git.qemu.org/git/qemu.git.
+    This repository is a read-only mirror of the project's repostories hosted
+    on https://gitlab.com/qemu-project/qemu.git.
     The project does not process merge requests filed on GitHub.
 
     QEMU welcomes contributions of code (either fixing bugs or adding new
-- 
2.29.2

qemu.org is running out of bandwidth and the QEMU project is moving
towards a gating CI on GitLab. Use the GitLab repos instead of qemu.org
(they will become mirrors).

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210111115017.156802-3-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 .gitmodules | 44 ++++++++++++++++++++++----------------------
 1 file changed, 22 insertions(+), 22 deletions(-)

diff --git a/.gitmodules b/.gitmodules
index XXXXXXX..XXXXXXX 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -XXX,XX +XXX,XX @@
 [submodule "roms/seabios"]
 	path = roms/seabios
-	url = https://git.qemu.org/git/seabios.git/
+	url = https://gitlab.com/qemu-project/seabios.git/
 [submodule "roms/SLOF"]
 	path = roms/SLOF
-	url = https://git.qemu.org/git/SLOF.git
+	url = https://gitlab.com/qemu-project/SLOF.git
 [submodule "roms/ipxe"]
 	path = roms/ipxe
-	url = https://git.qemu.org/git/ipxe.git
+	url = https://gitlab.com/qemu-project/ipxe.git
 [submodule "roms/openbios"]
 	path = roms/openbios
-	url = https://git.qemu.org/git/openbios.git
+	url = https://gitlab.com/qemu-project/openbios.git
 [submodule "roms/qemu-palcode"]
 	path = roms/qemu-palcode
-	url = https://git.qemu.org/git/qemu-palcode.git
+	url = https://gitlab.com/qemu-project/qemu-palcode.git
 [submodule "roms/sgabios"]
 	path = roms/sgabios
-	url = https://git.qemu.org/git/sgabios.git
+	url = https://gitlab.com/qemu-project/sgabios.git
 [submodule "dtc"]
 	path = dtc
-	url = https://git.qemu.org/git/dtc.git
+	url = https://gitlab.com/qemu-project/dtc.git
 [submodule "roms/u-boot"]
 	path = roms/u-boot
-	url = https://git.qemu.org/git/u-boot.git
+	url = https://gitlab.com/qemu-project/u-boot.git
 [submodule "roms/skiboot"]
 	path = roms/skiboot
-	url = https://git.qemu.org/git/skiboot.git
+	url = https://gitlab.com/qemu-project/skiboot.git
 [submodule "roms/QemuMacDrivers"]
 	path = roms/QemuMacDrivers
-	url = https://git.qemu.org/git/QemuMacDrivers.git
+	url = https://gitlab.com/qemu-project/QemuMacDrivers.git
 [submodule "ui/keycodemapdb"]
 	path = ui/keycodemapdb
-	url = https://git.qemu.org/git/keycodemapdb.git
+	url = https://gitlab.com/qemu-project/keycodemapdb.git
 [submodule "capstone"]
 	path = capstone
-	url = https://git.qemu.org/git/capstone.git
+	url = https://gitlab.com/qemu-project/capstone.git
 [submodule "roms/seabios-hppa"]
 	path = roms/seabios-hppa
-	url = https://git.qemu.org/git/seabios-hppa.git
+	url = https://gitlab.com/qemu-project/seabios-hppa.git
 [submodule "roms/u-boot-sam460ex"]
 	path = roms/u-boot-sam460ex
-	url = https://git.qemu.org/git/u-boot-sam460ex.git
+	url = https://gitlab.com/qemu-project/u-boot-sam460ex.git
 [submodule "tests/fp/berkeley-testfloat-3"]
 	path = tests/fp/berkeley-testfloat-3
-	url = https://git.qemu.org/git/berkeley-testfloat-3.git
+	url = https://gitlab.com/qemu-project/berkeley-testfloat-3.git
 [submodule "tests/fp/berkeley-softfloat-3"]
 	path = tests/fp/berkeley-softfloat-3
-	url = https://git.qemu.org/git/berkeley-softfloat-3.git
+	url = https://gitlab.com/qemu-project/berkeley-softfloat-3.git
 [submodule "roms/edk2"]
 	path = roms/edk2
-	url = https://git.qemu.org/git/edk2.git
+	url = https://gitlab.com/qemu-project/edk2.git
 [submodule "slirp"]
 	path = slirp
-	url = https://git.qemu.org/git/libslirp.git
+	url = https://gitlab.com/qemu-project/libslirp.git
 [submodule "roms/opensbi"]
 	path = roms/opensbi
-	url = 	https://git.qemu.org/git/opensbi.git
+	url = 	https://gitlab.com/qemu-project/opensbi.git
 [submodule "roms/qboot"]
 	path = roms/qboot
-	url = https://git.qemu.org/git/qboot.git
+	url = https://gitlab.com/qemu-project/qboot.git
 [submodule "meson"]
 	path = meson
-	url = https://git.qemu.org/git/meson.git
+	url = https://gitlab.com/qemu-project/meson.git
 [submodule "roms/vbootrom"]
 	path = roms/vbootrom
-	url = https://git.qemu.org/git/vbootrom.git
+	url = https://gitlab.com/qemu-project/vbootrom.git
-- 
2.29.2

qemu.org is running out of bandwidth and the QEMU project is moving
towards a gating CI on GitLab. Use the GitLab repos instead of qemu.org
(they will become mirrors).

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210111115017.156802-5-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 README.rst | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/README.rst b/README.rst
index XXXXXXX..XXXXXXX 100644
--- a/README.rst
+++ b/README.rst
@@ -XXX,XX +XXX,XX @@ The QEMU source code is maintained under the GIT version control system.
 
 .. code-block:: shell
 
-   git clone https://git.qemu.org/git/qemu.git
+   git clone https://gitlab.com/qemu-project/qemu.git
 
 When submitting patches, one common approach is to use 'git
 format-patch' and/or 'git send-email' to format & send the mail to the
@@ -XXX,XX +XXX,XX @@ The QEMU website is also maintained under source control.
 
 .. code-block:: shell
 
-  git clone https://git.qemu.org/git/qemu-web.git
+  git clone https://gitlab.com/qemu-project/qemu-web.git
 
 * `<https://www.qemu.org/2017/02/04/the-new-qemu-website-is-up/>`_
 
-- 
2.29.2

qemu.org is running out of bandwidth and the QEMU project is moving
towards a gating CI on GitLab. Use the GitLab repos instead of qemu.org
(they will become mirrors).

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210111115017.156802-6-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 pc-bios/README | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/pc-bios/README b/pc-bios/README
index XXXXXXX..XXXXXXX 100644
--- a/pc-bios/README
+++ b/pc-bios/README
@@ -XXX,XX +XXX,XX @@
   legacy x86 software to communicate with an attached serial console as
   if a video card were attached.  The master sources reside in a subversion
   repository at http://sgabios.googlecode.com/svn/trunk.  A git mirror is
-  available at https://git.qemu.org/git/sgabios.git.
+  available at https://gitlab.com/qemu-project/sgabios.git.
 
 - The PXE roms come from the iPXE project. Built with BANNER_TIME 0.
   Sources available at http://ipxe.org.  Vendor:Device ID -> ROM mapping:
@@ -XXX,XX +XXX,XX @@
 
 - The u-boot binary for e500 comes from the upstream denx u-boot project where
   it was compiled using the qemu-ppce500 target.
-  A git mirror is available at: https://git.qemu.org/git/u-boot.git
+  A git mirror is available at: https://gitlab.com/qemu-project/u-boot.git
   The hash used to compile the current version is: 2072e72
 
 - Skiboot (https://github.com/open-power/skiboot/) is an OPAL
-- 
2.29.2

qemu.org is running out of bandwidth and the QEMU project is moving
towards a gating CI on GitLab. Use the GitLab repos instead of qemu.org
(they will become mirrors).

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20210111115017.156802-7-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 scripts/get_maintainer.pl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/get_maintainer.pl b/scripts/get_maintainer.pl
index XXXXXXX..XXXXXXX 100755
--- a/scripts/get_maintainer.pl
+++ b/scripts/get_maintainer.pl
@@ -XXX,XX +XXX,XX @@ sub vcs_exists {
 	warn("$P: No supported VCS found.  Add --nogit to options?\n");
 	warn("Using a git repository produces better results.\n");
 	warn("Try latest git repository using:\n");
-	warn("git clone https://git.qemu.org/git/qemu.git\n");
+	warn("git clone https://gitlab.com/qemu-project/qemu.git\n");
 	$printed_novcs = 1;
     }
     return 0;
-- 
2.29.2

From: John G Johnson <john.g.johnson@oracle.com>

Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 02a68adef99f5df6a380bf8fd7b90948777e411c.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                  |   7 +
 docs/devel/index.rst         |   1 +
 docs/devel/multi-process.rst | 966 +++++++++++++++++++++++++++++++++++
 3 files changed, 974 insertions(+)
 create mode 100644 docs/devel/multi-process.rst

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ S: Maintained
 F: hw/semihosting/
 F: include/hw/semihosting/
 
+Multi-process QEMU
+M: Elena Ufimtseva <elena.ufimtseva@oracle.com>
+M: Jagannathan Raman <jag.raman@oracle.com>
+M: John G Johnson <john.g.johnson@oracle.com>
+S: Maintained
+F: docs/devel/multi-process.rst
+
 Build and test automation
 -------------------------
 Build and test automation
diff --git a/docs/devel/index.rst b/docs/devel/index.rst
index XXXXXXX..XXXXXXX 100644
--- a/docs/devel/index.rst
+++ b/docs/devel/index.rst
@@ -XXX,XX +XXX,XX @@ Contents:
    clocks
    qom
    block-coroutine-wrapper
+   multi-process
diff --git a/docs/devel/multi-process.rst b/docs/devel/multi-process.rst
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/docs/devel/multi-process.rst
@@ -XXX,XX +XXX,XX @@
+This is the design document for multi-process QEMU. It does not
+necessarily reflect the status of the current implementation, which
+may lack features or be considerably different from what is described
+in this document. This document is still useful as a description of
+the goals and general direction of this feature.
+
+Please refer to the following wiki for latest details:
+https://wiki.qemu.org/Features/MultiProcessQEMU
+
+Multi-process QEMU
+===================
+
+QEMU is often used as the hypervisor for virtual machines running in the
+Oracle cloud. Since one of the advantages of cloud computing is the
+ability to run many VMs from different tenants in the same cloud
+infrastructure, a guest that compromised its hypervisor could
+potentially use the hypervisor's access privileges to access data it is
+not authorized for.
+
+QEMU can be susceptible to security attacks because it is a large,
+monolithic program that provides many features to the VMs it services.
+Many of these features can be configured out of QEMU, but even a reduced
+configuration QEMU has a large amount of code a guest can potentially
+attack. Separating QEMU reduces the attack surface by aiding to
+limit each component in the system to only access the resources that
+it needs to perform its job.
+
+QEMU services
+-------------
+
+QEMU can be broadly described as providing three main services. One is a
+VM control point, where VMs can be created, migrated, re-configured, and
+destroyed. A second is to emulate the CPU instructions within the VM,
+often accelerated by HW virtualization features such as Intel's VT
+extensions. Finally, it provides IO services to the VM by emulating HW
+IO devices, such as disk and network devices.
+
+A multi-process QEMU
+~~~~~~~~~~~~~~~~~~~~
+
+A multi-process QEMU involves separating QEMU services into separate
+host processes. Each of these processes can be given only the privileges
+it needs to provide its service, e.g., a disk service could be given
+access only to the disk images it provides, and not be allowed to
+access other files, or any network devices. An attacker who compromised
+this service would not be able to use this exploit to access files or
+devices beyond what the disk service was given access to.
+
+A QEMU control process would remain, but in multi-process mode, will
+have no direct interfaces to the VM. During VM execution, it would still
+provide the user interface to hot-plug devices or live migrate the VM.
+
+A first step in creating a multi-process QEMU is to separate IO services
+from the main QEMU program, which would continue to provide CPU
+emulation. i.e., the control process would also be the CPU emulation
+process. In a later phase, CPU emulation could be separated from the
+control process.
+
+Separating IO services
+----------------------
+
+Separating IO services into individual host processes is a good place to
+begin for a couple of reasons. One is the sheer number of IO devices QEMU
+can emulate provides a large surface of interfaces which could potentially
+be exploited, and, indeed, have been a source of exploits in the past.
+Another is the modular nature of QEMU device emulation code provides
+interface points where the QEMU functions that perform device emulation
+can be separated from the QEMU functions that manage the emulation of
+guest CPU instructions. The devices emulated in the separate process are
+referred to as remote devices.
+
+QEMU device emulation
+~~~~~~~~~~~~~~~~~~~~~
+
+QEMU uses an object oriented SW architecture for device emulation code.
+Configured objects are all compiled into the QEMU binary, then objects
+are instantiated by name when used by the guest VM. For example, the
+code to emulate a device named "foo" is always present in QEMU, but its
+instantiation code is only run when the device is included in the target
+VM. (e.g., via the QEMU command line as *-device foo*)
+
+The object model is hierarchical, so device emulation code names its
+parent object (such as "pci-device" for a PCI device) and QEMU will
+instantiate a parent object before calling the device's instantiation
+code.
+
+Current separation models
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+In order to separate the device emulation code from the CPU emulation
+code, the device object code must run in a different process. There are
+a couple of existing QEMU features that can run emulation code
+separately from the main QEMU process. These are examined below.
+
+vhost user model
+^^^^^^^^^^^^^^^^
+
+Virtio guest device drivers can be connected to vhost user applications
+in order to perform their IO operations. This model uses special virtio
+device drivers in the guest and vhost user device objects in QEMU, but
+once the QEMU vhost user code has configured the vhost user application,
+mission-mode IO is performed by the application. The vhost user
+application is a daemon process that can be contacted via a known UNIX
+domain socket.
+
+vhost socket
+''''''''''''
+
+As mentioned above, one of the tasks of the vhost device object within
+QEMU is to contact the vhost application and send it configuration
+information about this device instance. As part of the configuration
+process, the application can also be sent other file descriptors over
+the socket, which then can be used by the vhost user application in
+various ways, some of which are described below.
+
+vhost MMIO store acceleration
+'''''''''''''''''''''''''''''
+
+VMs are often run using HW virtualization features via the KVM kernel
+driver. This driver allows QEMU to accelerate the emulation of guest CPU
+instructions by running the guest in a virtual HW mode. When the guest
+executes instructions that cannot be executed by virtual HW mode,
+execution returns to the KVM driver so it can inform QEMU to emulate the
+instructions in SW.
+
+One of the events that can cause a return to QEMU is when a guest device
+driver accesses an IO location. QEMU then dispatches the memory
+operation to the corresponding QEMU device object. In the case of a
+vhost user device, the memory operation would need to be sent over a
+socket to the vhost application. This path is accelerated by the QEMU
+virtio code by setting up an eventfd file descriptor that the vhost
+application can directly receive MMIO store notifications from the KVM
+driver, instead of needing them to be sent to the QEMU process first.
+
+vhost interrupt acceleration
+''''''''''''''''''''''''''''
+
+Another optimization used by the vhost application is the ability to
+directly inject interrupts into the VM via the KVM driver, again,
+bypassing the need to send the interrupt back to the QEMU process first.
+The QEMU virtio setup code configures the KVM driver with an eventfd
+that triggers the device interrupt in the guest when the eventfd is
+written. This irqfd file descriptor is then passed to the vhost user
+application program.
+
+vhost access to guest memory
+''''''''''''''''''''''''''''
+
+The vhost application is also allowed to directly access guest memory,
+instead of needing to send the data as messages to QEMU. This is also
+done with file descriptors sent to the vhost user application by QEMU.
+These descriptors can be passed to ``mmap()`` by the vhost application
+to map the guest address space into the vhost application.
+
+IOMMUs introduce another level of complexity, since the address given to
+the guest virtio device to DMA to or from is not a guest physical
+address. This case is handled by having vhost code within QEMU register
+as a listener for IOMMU mapping changes. The vhost application maintains
+a cache of IOMMMU translations: sending translation requests back to
+QEMU on cache misses, and in turn receiving flush requests from QEMU
+when mappings are purged.
+
+applicability to device separation
+''''''''''''''''''''''''''''''''''
+
+Much of the vhost model can be re-used by separated device emulation. In
+particular, the ideas of using a socket between QEMU and the device
+emulation application, using a file descriptor to inject interrupts into
+the VM via KVM, and allowing the application to ``mmap()`` the guest
+should be re used.
+
+There are, however, some notable differences between how a vhost
+application works and the needs of separated device emulation. The most
+basic is that vhost uses custom virtio device drivers which always
+trigger IO with MMIO stores. A separated device emulation model must
+work with existing IO device models and guest device drivers. MMIO loads
+break vhost store acceleration since they are synchronous - guest
+progress cannot continue until the load has been emulated. By contrast,
+stores are asynchronous, the guest can continue after the store event
+has been sent to the vhost application.
+
+Another difference is that in the vhost user model, a single daemon can
+support multiple QEMU instances. This is contrary to the security regime
+desired, in which the emulation application should only be allowed to
+access the files or devices the VM it's running on behalf of can access.
+#### qemu-io model
+
+Qemu-io is a test harness used to test changes to the QEMU block backend
+object code. (e.g., the code that implements disk images for disk driver
+emulation) Qemu-io is not a device emulation application per se, but it
+does compile the QEMU block objects into a separate binary from the main
+QEMU one. This could be useful for disk device emulation, since its
+emulation applications will need to include the QEMU block objects.
+
+New separation model based on proxy objects
+-------------------------------------------
+
+A different model based on proxy objects in the QEMU program
+communicating with remote emulation programs could provide separation
+while minimizing the changes needed to the device emulation code. The
+rest of this section is a discussion of how a proxy object model would
+work.
+
+Remote emulation processes
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The remote emulation process will run the QEMU object hierarchy without
+modification. The device emulation objects will be also be based on the
+QEMU code, because for anything but the simplest device, it would not be
+a tractable to re-implement both the object model and the many device
+backends that QEMU has.
+
+The processes will communicate with the QEMU process over UNIX domain
+sockets. The processes can be executed either as standalone processes,
+or be executed by QEMU. In both cases, the host backends the emulation
+processes will provide are specified on its command line, as they would
+be for QEMU. For example:
+
+::
+
+    disk-proc -blockdev driver=file,node-name=file0,filename=disk-file0  \
+    -blockdev driver=qcow2,node-name=drive0,file=file0
+
+would indicate process *disk-proc* uses a qcow2 emulated disk named
+*file0* as its backend.
+
+Emulation processes may emulate more than one guest controller. A common
+configuration might be to put all controllers of the same device class
+(e.g., disk, network, etc.) in a single process, so that all backends of
+the same type can be managed by a single QMP monitor.
+
+communication with QEMU
+^^^^^^^^^^^^^^^^^^^^^^^
+
+The first argument to the remote emulation process will be a Unix domain
+socket that connects with the Proxy object. This is a required argument.
+
+::
+
+    disk-proc <socket number> <backend list>
+
+remote process QMP monitor
+^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Remote emulation processes can be monitored via QMP, similar to QEMU
+itself. The QMP monitor socket is specified the same as for a QEMU
+process:
+
+::
+
+    disk-proc -qmp unix:/tmp/disk-mon,server
+
+can be monitored over the UNIX socket path */tmp/disk-mon*.
+
+QEMU command line
+~~~~~~~~~~~~~~~~~
+
+Each remote device emulated in a remote process on the host is
+represented as a *-device* of type *pci-proxy-dev*. A socket
+sub-option to this option specifies the Unix socket that connects
+to the remote process. An *id* sub-option is required, and it should
+be the same id as used in the remote process.
+
+::
+
+    qemu-system-x86_64 ... -device pci-proxy-dev,id=lsi0,socket=3
+
+can be used to add a device emulated in a remote process
+
+
+QEMU management of remote processes
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+QEMU is not aware of the type of type of the remote PCI device. It is
+a pass through device as far as QEMU is concerned.
+
+communication with emulation process
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+primary channel
+'''''''''''''''
+
+The primary channel (referred to as com in the code) is used to bootstrap
+the remote process. It is also used to pass on device-agnostic commands
+like reset.
+
+per-device channels
+'''''''''''''''''''
+
+Each remote device communicates with QEMU using a dedicated communication
+channel. The proxy object sets up this channel using the primary
+channel during its initialization.
+
+QEMU device proxy objects
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+QEMU has an object model based on sub-classes inherited from the
+"object" super-class. The sub-classes that are of interest here are the
+"device" and "bus" sub-classes whose child sub-classes make up the
+device tree of a QEMU emulated system.
+
+The proxy object model will use device proxy objects to replace the
+device emulation code within the QEMU process. These objects will live
+in the same place in the object and bus hierarchies as the objects they
+replace. i.e., the proxy object for an LSI SCSI controller will be a
+sub-class of the "pci-device" class, and will have the same PCI bus
+parent and the same SCSI bus child objects as the LSI controller object
+it replaces.
+
+It is worth noting that the same proxy object is used to mediate with
+all types of remote PCI devices.
+
+object initialization
+^^^^^^^^^^^^^^^^^^^^^
+
+The Proxy device objects are initialized in the exact same manner in
+which any other QEMU device would be initialized.
+
+In addition, the Proxy objects perform the following two tasks:
+- Parses the "socket" sub option and connects to the remote process
+using this channel
+- Uses the "id" sub-option to connect to the emulated device on the
+separate process
+
+class\_init
+'''''''''''
+
+The ``class_init()`` method of a proxy object will, in general behave
+similarly to the object it replaces, including setting any static
+properties and methods needed by the proxy.
+
+instance\_init / realize
+''''''''''''''''''''''''
+
+The ``instance_init()`` and ``realize()`` functions would only need to
+perform tasks related to being a proxy, such are registering its own
+MMIO handlers, or creating a child bus that other proxy devices can be
+attached to later.
+
+Other tasks will be device-specific. For example, PCI device objects
+will initialize the PCI config space in order to make a valid PCI device
+tree within the QEMU process.
+
+address space registration
+^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+Most devices are driven by guest device driver accesses to IO addresses
+or ports. The QEMU device emulation code uses QEMU's memory region
+function calls (such as ``memory_region_init_io()``) to add callback
+functions that QEMU will invoke when the guest accesses the device's
+areas of the IO address space. When a guest driver does access the
+device, the VM will exit HW virtualization mode and return to QEMU,
+which will then lookup and execute the corresponding callback function.
+
+A proxy object would need to mirror the memory region calls the actual
+device emulator would perform in its initialization code, but with its
+own callbacks. When invoked by QEMU as a result of a guest IO operation,
+they will forward the operation to the device emulation process.
+
+PCI config space
+^^^^^^^^^^^^^^^^
+
+PCI devices also have a configuration space that can be accessed by the
+guest driver. Guest accesses to this space is not handled by the device
+emulation object, but by its PCI parent object. Much of this space is
+read-only, but certain registers (especially BAR and MSI-related ones)
+need to be propagated to the emulation process.
+
+PCI parent proxy
+''''''''''''''''
+
+One way to propagate guest PCI config accesses is to create a
+"pci-device-proxy" class that can serve as the parent of a PCI device
+proxy object. This class's parent would be "pci-device" and it would
+override the PCI parent's ``config_read()`` and ``config_write()``
+methods with ones that forward these operations to the emulation
+program.
+
+interrupt receipt
+^^^^^^^^^^^^^^^^^
+
+A proxy for a device that generates interrupts will need to create a
+socket to receive interrupt indications from the emulation process. An
+incoming interrupt indication would then be sent up to its bus parent to
+be injected into the guest. For example, a PCI device object may use
+``pci_set_irq()``.
+
+live migration
+^^^^^^^^^^^^^^
+
+The proxy will register to save and restore any *vmstate* it needs over
+a live migration event. The device proxy does not need to manage the
+remote device's *vmstate*; that will be handled by the remote process
+proxy (see below).
+
+QEMU remote device operation
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Generic device operations, such as DMA, will be performed by the remote
+process proxy by sending messages to the remote process.
+
+DMA operations
+^^^^^^^^^^^^^^
+
+DMA operations would be handled much like vhost applications do. One of
+the initial messages sent to the emulation process is a guest memory
+table. Each entry in this table consists of a file descriptor and size
+that the emulation process can ``mmap()`` to directly access guest
+memory, similar to ``vhost_user_set_mem_table()``. Note guest memory
+must be backed by file descriptors, such as when QEMU is given the
+*-mem-path* command line option.
+
+IOMMU operations
+^^^^^^^^^^^^^^^^
+
+When the emulated system includes an IOMMU, the remote process proxy in
+QEMU will need to create a socket for IOMMU requests from the emulation
+process. It will handle those requests with an
+``address_space_get_iotlb_entry()`` call. In order to handle IOMMU
+unmaps, the remote process proxy will also register as a listener on the
+device's DMA address space. When an IOMMU memory region is created
+within the DMA address space, an IOMMU notifier for unmaps will be added
+to the memory region that will forward unmaps to the emulation process
+over the IOMMU socket.
+
+device hot-plug via QMP
+^^^^^^^^^^^^^^^^^^^^^^^
+
+An QMP "device\_add" command can add a device emulated by a remote
+process. It will also have "rid" option to the command, just as the
+*-device* command line option does. The remote process may either be one
+started at QEMU startup, or be one added by the "add-process" QMP
+command described above. In either case, the remote process proxy will
+forward the new device's JSON description to the corresponding emulation
+process.
+
+live migration
+^^^^^^^^^^^^^^
+
+The remote process proxy will also register for live migration
+notifications with ``vmstate_register()``. When called to save state,
+the proxy will send the remote process a secondary socket file
+descriptor to save the remote process's device *vmstate* over. The
+incoming byte stream length and data will be saved as the proxy's
+*vmstate*. When the proxy is resumed on its new host, this *vmstate*
+will be extracted, and a secondary socket file descriptor will be sent
+to the new remote process through which it receives the *vmstate* in
+order to restore the devices there.
+
+device emulation in remote process
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The parts of QEMU that the emulation program will need include the
+object model; the memory emulation objects; the device emulation objects
+of the targeted device, and any dependent devices; and, the device's
+backends. It will also need code to setup the machine environment,
+handle requests from the QEMU process, and route machine-level requests
+(such as interrupts or IOMMU mappings) back to the QEMU process.
+
+initialization
+^^^^^^^^^^^^^^
+
+The process initialization sequence will follow the same sequence
+followed by QEMU. It will first initialize the backend objects, then
+device emulation objects. The JSON descriptions sent by the QEMU process
+will drive which objects need to be created.
+
+-  address spaces
+
+Before the device objects are created, the initial address spaces and
+memory regions must be configured with ``memory_map_init()``. This
+creates a RAM memory region object (*system\_memory*) and an IO memory
+region object (*system\_io*).
+
+-  RAM
+
+RAM memory region creation will follow how ``pc_memory_init()`` creates
+them, but must use ``memory_region_init_ram_from_fd()`` instead of
+``memory_region_allocate_system_memory()``. The file descriptors needed
+will be supplied by the guest memory table from above. Those RAM regions
+would then be added to the *system\_memory* memory region with
+``memory_region_add_subregion()``.
+
+-  PCI
+
+IO initialization will be driven by the JSON descriptions sent from the
+QEMU process. For a PCI device, a PCI bus will need to be created with
+``pci_root_bus_new()``, and a PCI memory region will need to be created
+and added to the *system\_memory* memory region with
+``memory_region_add_subregion_overlap()``. The overlap version is
+required for architectures where PCI memory overlaps with RAM memory.
+
+MMIO handling
+^^^^^^^^^^^^^
+
+The device emulation objects will use ``memory_region_init_io()`` to
+install their MMIO handlers, and ``pci_register_bar()`` to associate
+those handlers with a PCI BAR, as they do within QEMU currently.
+
+In order to use ``address_space_rw()`` in the emulation process to
+handle MMIO requests from QEMU, the PCI physical addresses must be the
+same in the QEMU process and the device emulation process. In order to
+accomplish that, guest BAR programming must also be forwarded from QEMU
+to the emulation process.
+
+interrupt injection
+^^^^^^^^^^^^^^^^^^^
+
+When device emulation wants to inject an interrupt into the VM, the
+request climbs the device's bus object hierarchy until the point where a
+bus object knows how to signal the interrupt to the guest. The details
+depend on the type of interrupt being raised.
+
+-  PCI pin interrupts
+
+On x86 systems, there is an emulated IOAPIC object attached to the root
+PCI bus object, and the root PCI object forwards interrupt requests to
+it. The IOAPIC object, in turn, calls the KVM driver to inject the
+corresponding interrupt into the VM. The simplest way to handle this in
+an emulation process would be to setup the root PCI bus driver (via
+``pci_bus_irqs()``) to send a interrupt request back to the QEMU
+process, and have the device proxy object reflect it up the PCI tree
+there.
+
+-  PCI MSI/X interrupts
+
+PCI MSI/X interrupts are implemented in HW as DMA writes to a
+CPU-specific PCI address. In QEMU on x86, a KVM APIC object receives
+these DMA writes, then calls into the KVM driver to inject the interrupt
+into the VM. A simple emulation process implementation would be to send
+the MSI DMA address from QEMU as a message at initialization, then
+install an address space handler at that address which forwards the MSI
+message back to QEMU.
+
+DMA operations
+^^^^^^^^^^^^^^
+
+When a emulation object wants to DMA into or out of guest memory, it
+first must use dma\_memory\_map() to convert the DMA address to a local
+virtual address. The emulation process memory region objects setup above
+will be used to translate the DMA address to a local virtual address the
+device emulation code can access.
+
+IOMMU
+^^^^^
+
+When an IOMMU is in use in QEMU, DMA translation uses IOMMU memory
+regions to translate the DMA address to a guest physical address before
+that physical address can be translated to a local virtual address. The
+emulation process will need similar functionality.
+
+-  IOTLB cache
+
+The emulation process will maintain a cache of recent IOMMU translations
+(the IOTLB). When the translate() callback of an IOMMU memory region is
+invoked, the IOTLB cache will be searched for an entry that will map the
+DMA address to a guest PA. On a cache miss, a message will be sent back
+to QEMU requesting the corresponding translation entry, which be both be
+used to return a guest address and be added to the cache.
+
+-  IOTLB purge
+
+The IOMMU emulation will also need to act on unmap requests from QEMU.
+These happen when the guest IOMMU driver purges an entry from the
+guest's translation table.
+
+live migration
+^^^^^^^^^^^^^^
+
+When a remote process receives a live migration indication from QEMU, it
+will set up a channel using the received file descriptor with
+``qio_channel_socket_new_fd()``. This channel will be used to create a
+*QEMUfile* that can be passed to ``qemu_save_device_state()`` to send
+the process's device state back to QEMU. This method will be reversed on
+restore - the channel will be passed to ``qemu_loadvm_state()`` to
+restore the device state.
+
+Accelerating device emulation
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The messages that are required to be sent between QEMU and the emulation
+process can add considerable latency to IO operations. The optimizations
+described below attempt to ameliorate this effect by allowing the
+emulation process to communicate directly with the kernel KVM driver.
+The KVM file descriptors created would be passed to the emulation process
+via initialization messages, much like the guest memory table is done.
+#### MMIO acceleration
+
+Vhost user applications can receive guest virtio driver stores directly
+from KVM. The issue with the eventfd mechanism used by vhost user is
+that it does not pass any data with the event indication, so it cannot
+handle guest loads or guest stores that carry store data. This concept
+could, however, be expanded to cover more cases.
+
+The expanded idea would require a new type of KVM device:
+*KVM\_DEV\_TYPE\_USER*. This device has two file descriptors: a master
+descriptor that QEMU can use for configuration, and a slave descriptor
+that the emulation process can use to receive MMIO notifications. QEMU
+would create both descriptors using the KVM driver, and pass the slave
+descriptor to the emulation process via an initialization message.
+
+data structures
+^^^^^^^^^^^^^^^
+
+-  guest physical range
+
+The guest physical range structure describes the address range that a
+device will respond to. It includes the base and length of the range, as
+well as which bus the range resides on (e.g., on an x86machine, it can
+specify whether the range refers to memory or IO addresses).
+
+A device can have multiple physical address ranges it responds to (e.g.,
+a PCI device can have multiple BARs), so the structure will also include
+an enumerated identifier to specify which of the device's ranges is
+being referred to.
+
++--------+----------------------------+
+| Name   | Description                |
++========+============================+
+| addr   | range base address         |
++--------+----------------------------+
+| len    | range length               |
++--------+----------------------------+
+| bus    | addr type (memory or IO)   |
++--------+----------------------------+
+| id     | range ID (e.g., PCI BAR)   |
++--------+----------------------------+
+
+-  MMIO request structure
+
+This structure describes an MMIO operation. It includes which guest
+physical range the MMIO was within, the offset within that range, the
+MMIO type (e.g., load or store), and its length and data. It also
+includes a sequence number that can be used to reply to the MMIO, and
+the CPU that issued the MMIO.
+
++----------+------------------------+
+| Name     | Description            |
++==========+========================+
+| rid      | range MMIO is within   |
++----------+------------------------+
+| offset   | offset withing *rid*   |
++----------+------------------------+
+| type     | e.g., load or store    |
++----------+------------------------+
+| len      | MMIO length            |
++----------+------------------------+
+| data     | store data             |
++----------+------------------------+
+| seq      | sequence ID            |
++----------+------------------------+
+
+-  MMIO request queues
+
+MMIO request queues are FIFO arrays of MMIO request structures. There
+are two queues: pending queue is for MMIOs that haven't been read by the
+emulation program, and the sent queue is for MMIOs that haven't been
+acknowledged. The main use of the second queue is to validate MMIO
+replies from the emulation program.
+
+-  scoreboard
+
+Each CPU in the VM is emulated in QEMU by a separate thread, so multiple
+MMIOs may be waiting to be consumed by an emulation program and multiple
+threads may be waiting for MMIO replies. The scoreboard would contain a
+wait queue and sequence number for the per-CPU threads, allowing them to
+be individually woken when the MMIO reply is received from the emulation
+program. It also tracks the number of posted MMIO stores to the device
+that haven't been replied to, in order to satisfy the PCI constraint
+that a load to a device will not complete until all previous stores to
+that device have been completed.
+
+-  device shadow memory
+
+Some MMIO loads do not have device side-effects. These MMIOs can be
+completed without sending a MMIO request to the emulation program if the
+emulation program shares a shadow image of the device's memory image
+with the KVM driver.
+
+The emulation program will ask the KVM driver to allocate memory for the
+shadow image, and will then use ``mmap()`` to directly access it. The
+emulation program can control KVM access to the shadow image by sending
+KVM an access map telling it which areas of the image have no
+side-effects (and can be completed immediately), and which require a
+MMIO request to the emulation program. The access map can also inform
+the KVM drive which size accesses are allowed to the image.
+
+master descriptor
+^^^^^^^^^^^^^^^^^
+
+The master descriptor is used by QEMU to configure the new KVM device.
+The descriptor would be returned by the KVM driver when QEMU issues a
+*KVM\_CREATE\_DEVICE* ``ioctl()`` with a *KVM\_DEV\_TYPE\_USER* type.
+
+KVM\_DEV\_TYPE\_USER device ops
+
+
+The *KVM\_DEV\_TYPE\_USER* operations vector will be registered by a
+``kvm_register_device_ops()`` call when the KVM system in initialized by
+``kvm_init()``. These device ops are called by the KVM driver when QEMU
+executes certain ``ioctl()`` operations on its KVM file descriptor. They
+include:
+
+-  create
+
+This routine is called when QEMU issues a *KVM\_CREATE\_DEVICE*
+``ioctl()`` on its per-VM file descriptor. It will allocate and
+initialize a KVM user device specific data structure, and assign the
+*kvm\_device* private field to it.
+
+-  ioctl
+
+This routine is invoked when QEMU issues an ``ioctl()`` on the master
+descriptor. The ``ioctl()`` commands supported are defined by the KVM
+device type. *KVM\_DEV\_TYPE\_USER* ones will need several commands:
+
+*KVM\_DEV\_USER\_SLAVE\_FD* creates the slave file descriptor that will
+be passed to the device emulation program. Only one slave can be created
+by each master descriptor. The file operations performed by this
+descriptor are described below.
+
+The *KVM\_DEV\_USER\_PA\_RANGE* command configures a guest physical
+address range that the slave descriptor will receive MMIO notifications
+for. The range is specified by a guest physical range structure
+argument. For buses that assign addresses to devices dynamically, this
+command can be executed while the guest is running, such as the case
+when a guest changes a device's PCI BAR registers.
+
+*KVM\_DEV\_USER\_PA\_RANGE* will use ``kvm_io_bus_register_dev()`` to
+register *kvm\_io\_device\_ops* callbacks to be invoked when the guest
+performs a MMIO operation within the range. When a range is changed,
+``kvm_io_bus_unregister_dev()`` is used to remove the previous
+instantiation.
+
+*KVM\_DEV\_USER\_TIMEOUT* will configure a timeout value that specifies
+how long KVM will wait for the emulation process to respond to a MMIO
+indication.
+
+-  destroy
+
+This routine is called when the VM instance is destroyed. It will need
+to destroy the slave descriptor; and free any memory allocated by the
+driver, as well as the *kvm\_device* structure itself.
+
+slave descriptor
+^^^^^^^^^^^^^^^^
+
+The slave descriptor will have its own file operations vector, which
+responds to system calls on the descriptor performed by the device
+emulation program.
+
+-  read
+
+A read returns any pending MMIO requests from the KVM driver as MMIO
+request structures. Multiple structures can be returned if there are
+multiple MMIO operations pending. The MMIO requests are moved from the
+pending queue to the sent queue, and if there are threads waiting for
+space in the pending to add new MMIO operations, they will be woken
+here.
+
+-  write
+
+A write also consists of a set of MMIO requests. They are compared to
+the MMIO requests in the sent queue. Matches are removed from the sent
+queue, and any threads waiting for the reply are woken. If a store is
+removed, then the number of posted stores in the per-CPU scoreboard is
+decremented. When the number is zero, and a non side-effect load was
+waiting for posted stores to complete, the load is continued.
+
+-  ioctl
+
+There are several ioctl()s that can be performed on the slave
+descriptor.
+
+A *KVM\_DEV\_USER\_SHADOW\_SIZE* ``ioctl()`` causes the KVM driver to
+allocate memory for the shadow image. This memory can later be
+``mmap()``\ ed by the emulation process to share the emulation's view of
+device memory with the KVM driver.
+
+A *KVM\_DEV\_USER\_SHADOW\_CTRL* ``ioctl()`` controls access to the
+shadow image. It will send the KVM driver a shadow control map, which
+specifies which areas of the image can complete guest loads without
+sending the load request to the emulation program. It will also specify
+the size of load operations that are allowed.
+
+-  poll
+
+An emulation program will use the ``poll()`` call with a *POLLIN* flag
+to determine if there are MMIO requests waiting to be read. It will
+return if the pending MMIO request queue is not empty.
+
+-  mmap
+
+This call allows the emulation program to directly access the shadow
+image allocated by the KVM driver. As device emulation updates device
+memory, changes with no side-effects will be reflected in the shadow,
+and the KVM driver can satisfy guest loads from the shadow image without
+needing to wait for the emulation program.
+
+kvm\_io\_device ops
+^^^^^^^^^^^^^^^^^^^
+
+Each KVM per-CPU thread can handle MMIO operation on behalf of the guest
+VM. KVM will use the MMIO's guest physical address to search for a
+matching *kvm\_io\_device* to see if the MMIO can be handled by the KVM
+driver instead of exiting back to QEMU. If a match is found, the
+corresponding callback will be invoked.
+
+-  read
+
+This callback is invoked when the guest performs a load to the device.
+Loads with side-effects must be handled synchronously, with the KVM
+driver putting the QEMU thread to sleep waiting for the emulation
+process reply before re-starting the guest. Loads that do not have
+side-effects may be optimized by satisfying them from the shadow image,
+if there are no outstanding stores to the device by this CPU. PCI memory
+ordering demands that a load cannot complete before all older stores to
+the same device have been completed.
+
+-  write
+
+Stores can be handled asynchronously unless the pending MMIO request
+queue is full. In this case, the QEMU thread must sleep waiting for
+space in the queue. Stores will increment the number of posted stores in
+the per-CPU scoreboard, in order to implement the PCI ordering
+constraint above.
+
+interrupt acceleration
+^^^^^^^^^^^^^^^^^^^^^^
+
+This performance optimization would work much like a vhost user
+application does, where the QEMU process sets up *eventfds* that cause
+the device's corresponding interrupt to be triggered by the KVM driver.
+These irq file descriptors are sent to the emulation process at
+initialization, and are used when the emulation code raises a device
+interrupt.
+
+intx acceleration
+'''''''''''''''''
+
+Traditional PCI pin interrupts are level based, so, in addition to an
+irq file descriptor, a re-sampling file descriptor needs to be sent to
+the emulation program. This second file descriptor allows multiple
+devices sharing an irq to be notified when the interrupt has been
+acknowledged by the guest, so they can re-trigger the interrupt if their
+device has not de-asserted its interrupt.
+
+intx irq descriptor
+
+
+The irq descriptors are created by the proxy object
+``using event_notifier_init()`` to create the irq and re-sampling
+*eventds*, and ``kvm_vm_ioctl(KVM_IRQFD)`` to bind them to an interrupt.
+The interrupt route can be found with
+``pci_device_route_intx_to_irq()``.
+
+intx routing changes
+
+
+Intx routing can be changed when the guest programs the APIC the device
+pin is connected to. The proxy object in QEMU will use
+``pci_device_set_intx_routing_notifier()`` to be informed of any guest
+changes to the route. This handler will broadly follow the VFIO
+interrupt logic to change the route: de-assigning the existing irq
+descriptor from its route, then assigning it the new route. (see
+``vfio_intx_update()``)
+
+MSI/X acceleration
+''''''''''''''''''
+
+MSI/X interrupts are sent as DMA transactions to the host. The interrupt
+data contains a vector that is programmed by the guest, A device may have
+multiple MSI interrupts associated with it, so multiple irq descriptors
+may need to be sent to the emulation program.
+
+MSI/X irq descriptor
+
+
+This case will also follow the VFIO example. For each MSI/X interrupt,
+an *eventfd* is created, a virtual interrupt is allocated by
+``kvm_irqchip_add_msi_route()``, and the virtual interrupt is bound to
+the eventfd with ``kvm_irqchip_add_irqfd_notifier()``.
+
+MSI/X config space changes
+
+
+The guest may dynamically update several MSI-related tables in the
+device's PCI config space. These include per-MSI interrupt enables and
+vector data. Additionally, MSIX tables exist in device memory space, not
+config space. Much like the BAR case above, the proxy object must look
+at guest config space programming to keep the MSI interrupt state
+consistent between QEMU and the emulation program.
+
+--------------
+
+Disaggregated CPU emulation
+---------------------------
+
+After IO services have been disaggregated, a second phase would be to
+separate a process to handle CPU instruction emulation from the main
+QEMU control function. There are no object separation points for this
+code, so the first task would be to create one.
+
+Host access controls
+--------------------
+
+Separating QEMU relies on the host OS's access restriction mechanisms to
+enforce that the differing processes can only access the objects they
+are entitled to. There are a couple types of mechanisms usually provided
+by general purpose OSs.
+
+Discretionary access control
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Discretionary access control allows each user to control who can access
+their files. In Linux, this type of control is usually too coarse for
+QEMU separation, since it only provides three separate access controls:
+one for the same user ID, the second for users IDs with the same group
+ID, and the third for all other user IDs. Each device instance would
+need a separate user ID to provide access control, which is likely to be
+unwieldy for dynamically created VMs.
+
+Mandatory access control
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+Mandatory access control allows the OS to add an additional set of
+controls on top of discretionary access for the OS to control. It also
+adds other attributes to processes and files such as types, roles, and
+categories, and can establish rules for how processes and files can
+interact.
+
+Type enforcement
+^^^^^^^^^^^^^^^^
+
+Type enforcement assigns a *type* attribute to processes and files, and
+allows rules to be written on what operations a process with a given
+type can perform on a file with a given type. QEMU separation could take
+advantage of type enforcement by running the emulation processes with
+different types, both from the main QEMU process, and from the emulation
+processes of different classes of devices.
+
+For example, guest disk images and disk emulation processes could have
+types separate from the main QEMU process and non-disk emulation
+processes, and the type rules could prevent processes other than disk
+emulation ones from accessing guest disk images. Similarly, network
+emulation processes can have a type separate from the main QEMU process
+and non-network emulation process, and only that type can access the
+host tun/tap device used to provide guest networking.
+
+Category enforcement
+^^^^^^^^^^^^^^^^^^^^
+
+Category enforcement assigns a set of numbers within a given range to
+the process or file. The process is granted access to the file if the
+process's set is a superset of the file's set. This enforcement can be
+used to separate multiple instances of devices in the same class.
+
+For example, if there are multiple disk devices provides to a guest,
+each device emulation process could be provisioned with a separate
+category. The different device emulation processes would not be able to
+access each other's backing disk images.
+
+Alternatively, categories could be used in lieu of the type enforcement
+scheme described above. In this scenario, different categories would be
+used to prevent device emulation processes in different classes from
+accessing resources assigned to other classes.
-- 
2.29.2

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Adds documentation explaining the command-line arguments needed
to use multi-process.

[Move orphan docs/multi-process.rst document into docs/system/ and add
it to index.rst to prevent Sphinx "document isn't included in any
toctree" error.
--Stefan]

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                   |  1 +
 docs/system/index.rst         |  1 +
 docs/system/multi-process.rst | 64 +++++++++++++++++++++++++++++++++++
 3 files changed, 66 insertions(+)
 create mode 100644 docs/system/multi-process.rst

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ M: Jagannathan Raman <jag.raman@oracle.com>
 M: John G Johnson <john.g.johnson@oracle.com>
 S: Maintained
 F: docs/devel/multi-process.rst
+F: docs/system/multi-process.rst
 
 Build and test automation
 -------------------------
diff --git a/docs/system/index.rst b/docs/system/index.rst
index XXXXXXX..XXXXXXX 100644
--- a/docs/system/index.rst
+++ b/docs/system/index.rst
@@ -XXX,XX +XXX,XX @@ Contents:
    pr-manager
    targets
    security
+   multi-process
    deprecated
    removed-features
    build-platforms
diff --git a/docs/system/multi-process.rst b/docs/system/multi-process.rst
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/docs/system/multi-process.rst
@@ -XXX,XX +XXX,XX @@
+Multi-process QEMU
+==================
+
+This document describes how to configure and use multi-process qemu.
+For the design document refer to docs/devel/qemu-multiprocess.
+
+1) Configuration
+----------------
+
+multi-process is enabled by default for targets that enable KVM
+
+
+2) Usage
+--------
+
+Multi-process QEMU requires an orchestrator to launch.
+
+Following is a description of command-line used to launch mpqemu.
+
+* Orchestrator:
+
+  - The Orchestrator creates a unix socketpair
+
+  - It launches the remote process and passes one of the
+    sockets to it via command-line.
+
+  - It then launches QEMU and specifies the other socket as an option
+    to the Proxy device object
+
+* Remote Process:
+
+  - QEMU can enter remote process mode by using the "remote" machine
+    option.
+
+  - The orchestrator creates a "remote-object" with details about
+    the device and the file descriptor for the device
+
+  - The remaining options are no different from how one launches QEMU with
+    devices.
+
+  - Example command-line for the remote process is as follows:
+
+      /usr/bin/qemu-system-x86_64                                        \
+      -machine x-remote                                                  \
+      -device lsi53c895a,id=lsi0                                         \
+      -drive id=drive_image2,file=/build/ol7-nvme-test-1.qcow2           \
+      -device scsi-hd,id=drive2,drive=drive_image2,bus=lsi0.0,scsi-id=0  \
+      -object x-remote-object,id=robj1,devid=lsi1,fd=4,
+
+* QEMU:
+
+  - Since parts of the RAM are shared between QEMU & remote process, a
+    memory-backend-memfd is required to facilitate this, as follows:
+
+    -object memory-backend-memfd,id=mem,size=2G
+
+  - A "x-pci-proxy-dev" device is created for each of the PCI devices emulated
+    in the remote process. A "socket" sub-option specifies the other end of
+    unix channel created by orchestrator. The "id" sub-option must be specified
+    and should be the same as the "id" specified for the remote PCI device
+
+  - Example commandline for QEMU is as follows:
+
+      -device x-pci-proxy-dev,id=lsi0,socket=3
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

Allow RAM MemoryRegion to be created from an offset in a file, instead
of allocating at offset of 0 by default. This is needed to synchronize
RAM between QEMU & remote process.

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 609996697ad8617e3b01df38accc5c208c24d74e.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/exec/memory.h     |  2 ++
 include/exec/ram_addr.h   |  4 ++--
 include/qemu/mmap-alloc.h |  4 +++-
 backends/hostmem-memfd.c  |  2 +-
 hw/misc/ivshmem.c         |  3 ++-
 softmmu/memory.c          |  3 ++-
 softmmu/physmem.c         | 12 +++++++-----
 util/mmap-alloc.c         |  8 +++++---
 util/oslib-posix.c        |  2 +-
 9 files changed, 25 insertions(+), 15 deletions(-)

diff --git a/include/exec/memory.h b/include/exec/memory.h
index XXXXXXX..XXXXXXX 100644
--- a/include/exec/memory.h
+++ b/include/exec/memory.h
@@ -XXX,XX +XXX,XX @@ void memory_region_init_ram_from_file(MemoryRegion *mr,
  * @size: size of the region.
  * @share: %true if memory must be mmaped with the MAP_SHARED flag
  * @fd: the fd to mmap.
+ * @offset: offset within the file referenced by fd
  * @errp: pointer to Error*, to store an error if it happens.
  *
  * Note that this function does not do anything to cause the data in the
@@ -XXX,XX +XXX,XX @@ void memory_region_init_ram_from_fd(MemoryRegion *mr,
                                     uint64_t size,
                                     bool share,
                                     int fd,
+                                    ram_addr_t offset,
                                     Error **errp);
 #endif
 
diff --git a/include/exec/ram_addr.h b/include/exec/ram_addr.h
index XXXXXXX..XXXXXXX 100644
--- a/include/exec/ram_addr.h
+++ b/include/exec/ram_addr.h
@@ -XXX,XX +XXX,XX @@ RAMBlock *qemu_ram_alloc_from_file(ram_addr_t size, MemoryRegion *mr,
                                    uint32_t ram_flags, const char *mem_path,
                                    bool readonly, Error **errp);
 RAMBlock *qemu_ram_alloc_from_fd(ram_addr_t size, MemoryRegion *mr,
-                                 uint32_t ram_flags, int fd, bool readonly,
-                                 Error **errp);
+                                 uint32_t ram_flags, int fd, off_t offset,
+                                 bool readonly, Error **errp);
 
 RAMBlock *qemu_ram_alloc_from_ptr(ram_addr_t size, void *host,
                                   MemoryRegion *mr, Error **errp);
diff --git a/include/qemu/mmap-alloc.h b/include/qemu/mmap-alloc.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu/mmap-alloc.h
+++ b/include/qemu/mmap-alloc.h
@@ -XXX,XX +XXX,XX @@ size_t qemu_mempath_getpagesize(const char *mem_path);
  *  @readonly: true for a read-only mapping, false for read/write.
  *  @shared: map has RAM_SHARED flag.
  *  @is_pmem: map has RAM_PMEM flag.
+ *  @map_offset: map starts at offset of map_offset from the start of fd
  *
  * Return:
  *  On success, return a pointer to the mapped area.
@@ -XXX,XX +XXX,XX @@ void *qemu_ram_mmap(int fd,
                     size_t align,
                     bool readonly,
                     bool shared,
-                    bool is_pmem);
+                    bool is_pmem,
+                    off_t map_offset);
 
 void qemu_ram_munmap(int fd, void *ptr, size_t size);
 
diff --git a/backends/hostmem-memfd.c b/backends/hostmem-memfd.c
index XXXXXXX..XXXXXXX 100644
--- a/backends/hostmem-memfd.c
+++ b/backends/hostmem-memfd.c
@@ -XXX,XX +XXX,XX @@ memfd_backend_memory_alloc(HostMemoryBackend *backend, Error **errp)
     name = host_memory_backend_get_name(backend);
     memory_region_init_ram_from_fd(&backend->mr, OBJECT(backend),
                                    name, backend->size,
-                                   backend->share, fd, errp);
+                                   backend->share, fd, 0, errp);
     g_free(name);
 }
 
diff --git a/hw/misc/ivshmem.c b/hw/misc/ivshmem.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/misc/ivshmem.c
+++ b/hw/misc/ivshmem.c
@@ -XXX,XX +XXX,XX @@ static void process_msg_shmem(IVShmemState *s, int fd, Error **errp)
 
     /* mmap the region and map into the BAR2 */
     memory_region_init_ram_from_fd(&s->server_bar2, OBJECT(s),
-                                   "ivshmem.bar2", size, true, fd, &local_err);
+                                   "ivshmem.bar2", size, true, fd, 0,
+                                   &local_err);
     if (local_err) {
         error_propagate(errp, local_err);
         return;
diff --git a/softmmu/memory.c b/softmmu/memory.c
index XXXXXXX..XXXXXXX 100644
--- a/softmmu/memory.c
+++ b/softmmu/memory.c
@@ -XXX,XX +XXX,XX @@ void memory_region_init_ram_from_fd(MemoryRegion *mr,
                                     uint64_t size,
                                     bool share,
                                     int fd,
+                                    ram_addr_t offset,
                                     Error **errp)
 {
     Error *err = NULL;
@@ -XXX,XX +XXX,XX @@ void memory_region_init_ram_from_fd(MemoryRegion *mr,
     mr->destructor = memory_region_destructor_ram;
     mr->ram_block = qemu_ram_alloc_from_fd(size, mr,
                                            share ? RAM_SHARED : 0,
-                                           fd, false, &err);
+                                           fd, offset, false, &err);
     if (err) {
         mr->size = int128_zero();
         object_unparent(OBJECT(mr));
diff --git a/softmmu/physmem.c b/softmmu/physmem.c
index XXXXXXX..XXXXXXX 100644
--- a/softmmu/physmem.c
+++ b/softmmu/physmem.c
@@ -XXX,XX +XXX,XX @@ static void *file_ram_alloc(RAMBlock *block,
                             int fd,
                             bool readonly,
                             bool truncate,
+                            off_t offset,
                             Error **errp)
 {
     void *area;
@@ -XXX,XX +XXX,XX @@ static void *file_ram_alloc(RAMBlock *block,
     }
 
     area = qemu_ram_mmap(fd, memory, block->mr->align, readonly,
-                         block->flags & RAM_SHARED, block->flags & RAM_PMEM);
+                         block->flags & RAM_SHARED, block->flags & RAM_PMEM,
+                         offset);
     if (area == MAP_FAILED) {
         error_setg_errno(errp, errno,
                          "unable to map backing store for guest RAM");
@@ -XXX,XX +XXX,XX @@ static void ram_block_add(RAMBlock *new_block, Error **errp, bool shared)
 
 #ifdef CONFIG_POSIX
 RAMBlock *qemu_ram_alloc_from_fd(ram_addr_t size, MemoryRegion *mr,
-                                 uint32_t ram_flags, int fd, bool readonly,
-                                 Error **errp)
+                                 uint32_t ram_flags, int fd, off_t offset,
+                                 bool readonly, Error **errp)
 {
     RAMBlock *new_block;
     Error *local_err = NULL;
@@ -XXX,XX +XXX,XX @@ RAMBlock *qemu_ram_alloc_from_fd(ram_addr_t size, MemoryRegion *mr,
     new_block->max_length = size;
     new_block->flags = ram_flags;
     new_block->host = file_ram_alloc(new_block, size, fd, readonly,
-                                     !file_size, errp);
+                                     !file_size, offset, errp);
     if (!new_block->host) {
         g_free(new_block);
         return NULL;
@@ -XXX,XX +XXX,XX @@ RAMBlock *qemu_ram_alloc_from_file(ram_addr_t size, MemoryRegion *mr,
         return NULL;
     }
 
-    block = qemu_ram_alloc_from_fd(size, mr, ram_flags, fd, readonly, errp);
+    block = qemu_ram_alloc_from_fd(size, mr, ram_flags, fd, 0, readonly, errp);
     if (!block) {
         if (created) {
             unlink(mem_path);
diff --git a/util/mmap-alloc.c b/util/mmap-alloc.c
index XXXXXXX..XXXXXXX 100644
--- a/util/mmap-alloc.c
+++ b/util/mmap-alloc.c
@@ -XXX,XX +XXX,XX @@ void *qemu_ram_mmap(int fd,
                     size_t align,
                     bool readonly,
                     bool shared,
-                    bool is_pmem)
+                    bool is_pmem,
+                    off_t map_offset)
 {
     int prot;
     int flags;
@@ -XXX,XX +XXX,XX @@ void *qemu_ram_mmap(int fd,
 
     prot = PROT_READ | (readonly ? 0 : PROT_WRITE);
 
-    ptr = mmap(guardptr + offset, size, prot, flags | map_sync_flags, fd, 0);
+    ptr = mmap(guardptr + offset, size, prot,
+               flags | map_sync_flags, fd, map_offset);
 
     if (ptr == MAP_FAILED && map_sync_flags) {
         if (errno == ENOTSUP) {
@@ -XXX,XX +XXX,XX @@ void *qemu_ram_mmap(int fd,
          * if map failed with MAP_SHARED_VALIDATE | MAP_SYNC,
          * we will remove these flags to handle compatibility.
          */
-        ptr = mmap(guardptr + offset, size, prot, flags, fd, 0);
+        ptr = mmap(guardptr + offset, size, prot, flags, fd, map_offset);
     }
 
     if (ptr == MAP_FAILED) {
diff --git a/util/oslib-posix.c b/util/oslib-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/oslib-posix.c
+++ b/util/oslib-posix.c
@@ -XXX,XX +XXX,XX @@ void *qemu_memalign(size_t alignment, size_t size)
 void *qemu_anon_ram_alloc(size_t size, uint64_t *alignment, bool shared)
 {
     size_t align = QEMU_VMALLOC_ALIGN;
-    void *ptr = qemu_ram_mmap(-1, size, align, false, shared, false);
+    void *ptr = qemu_ram_mmap(-1, size, align, false, shared, false, 0);
 
     if (ptr == MAP_FAILED) {
         return NULL;
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

Add configuration options to enable or disable multiprocess QEMU code

diff --git a/configure b/configure
index XXXXXXX..XXXXXXX 100755
--- a/configure
+++ b/configure
@@ -XXX,XX +XXX,XX @@ skip_meson=no
 gettext="auto"
 fuse="auto"
 fuse_lseek="auto"
+multiprocess="no"
 
 malloc_trim="auto"
 
@@ -XXX,XX +XXX,XX @@ Linux)
   linux="yes"
   linux_user="yes"
   vhost_user=${default_feature:-yes}
+  multiprocess=${default_feature:-yes}
 ;;
 esac
 
@@ -XXX,XX +XXX,XX @@ for opt do
   ;;
   --disable-fuse-lseek) fuse_lseek="disabled"
   ;;
+  --enable-multiprocess) multiprocess="yes"
+  ;;
+  --disable-multiprocess) multiprocess="no"
+  ;;
   *)
       echo "ERROR: unknown option $opt"
       echo "Try '$0 --help' for more information"
@@ -XXX,XX +XXX,XX @@ disabled with --disable-FEATURE, default is enabled if available
   libdaxctl       libdaxctl support
   fuse            FUSE block device export
   fuse-lseek      SEEK_HOLE/SEEK_DATA support for FUSE exports
+  multiprocess    Multiprocess QEMU support
 
 NOTE: The object files are built at the place where configure is launched
 EOF
@@ -XXX,XX +XXX,XX @@ fi
 if test "$have_mlockall" = "yes" ; then
   echo "HAVE_MLOCKALL=y" >> $config_host_mak
 fi
+if test "$multiprocess" = "yes" ; then
+  echo "CONFIG_MULTIPROCESS_ALLOWED=y" >> $config_host_mak
+fi
 if test "$fuzzing" = "yes" ; then
   # If LIB_FUZZING_ENGINE is set, assume we are running on OSS-Fuzz, and the
   # needed CFLAGS have already been provided
diff --git a/meson.build b/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/meson.build
+++ b/meson.build
@@ -XXX,XX +XXX,XX @@ host_kconfig = \
   ('CONFIG_VHOST_KERNEL' in config_host ? ['CONFIG_VHOST_KERNEL=y'] : []) + \
   (have_virtfs ? ['CONFIG_VIRTFS=y'] : []) + \
   ('CONFIG_LINUX' in config_host ? ['CONFIG_LINUX=y'] : []) + \
-  ('CONFIG_PVRDMA' in config_host ? ['CONFIG_PVRDMA=y'] : [])
+  ('CONFIG_PVRDMA' in config_host ? ['CONFIG_PVRDMA=y'] : []) + \
+  ('CONFIG_MULTIPROCESS_ALLOWED' in config_host ? ['CONFIG_MULTIPROCESS_ALLOWED=y'] : [])
 
 ignored = [ 'TARGET_XML_FILES', 'TARGET_ABI_DIR', 'TARGET_ARCH' ]
 
@@ -XXX,XX +XXX,XX @@ summary_info += {'libpmem support':   config_host.has_key('CONFIG_LIBPMEM')}
 summary_info += {'libdaxctl support': config_host.has_key('CONFIG_LIBDAXCTL')}
 summary_info += {'libudev':           libudev.found()}
 summary_info += {'FUSE lseek':        fuse_lseek.found()}
+summary_info += {'Multiprocess QEMU': config_host.has_key('CONFIG_MULTIPROCESS_ALLOWED')}
 summary(summary_info, bool_yn: true, section: 'Dependencies')
 
 if not supported_cpus.contains(cpu)
diff --git a/Kconfig.host b/Kconfig.host
index XXXXXXX..XXXXXXX 100644
--- a/Kconfig.host
+++ b/Kconfig.host
@@ -XXX,XX +XXX,XX @@ config VIRTFS
 
 config PVRDMA
     bool
+
+config MULTIPROCESS_ALLOWED
+    bool
+    imply MULTIPROCESS
diff --git a/hw/Kconfig b/hw/Kconfig
index XXXXXXX..XXXXXXX 100644
--- a/hw/Kconfig
+++ b/hw/Kconfig
@@ -XXX,XX +XXX,XX @@ source pci-host/Kconfig
 source pcmcia/Kconfig
 source pci/Kconfig
 source rdma/Kconfig
+source remote/Kconfig
 source rtc/Kconfig
 source scsi/Kconfig
 source sd/Kconfig
diff --git a/hw/remote/Kconfig b/hw/remote/Kconfig
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/Kconfig
@@ -XXX,XX +XXX,XX @@
+config MULTIPROCESS
+    bool
+    depends on PCI && KVM
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

PCI host bridge is setup for the remote device process. It is
implemented using remote-pcihost object. It is an extension of the PCI
host bridge setup by QEMU.
Remote-pcihost configures a PCI bus which could be used by the remote
PCI device to latch on to.

[Added PCI_EXPRESS condition in hw/remote/Kconfig since remote-pcihost
needs PCIe. This solves "make check" failure on s390x. Fix suggested by
Philippe Mathieu-Daudé <philmd@redhat.com> and Thomas Huth
<thuth@redhat.com>.
--Stefan]

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                  |  2 +
 include/hw/pci-host/remote.h | 29 ++++++++++++++
 hw/pci-host/remote.c         | 75 ++++++++++++++++++++++++++++++++++++
 hw/pci-host/Kconfig          |  3 ++
 hw/pci-host/meson.build      |  1 +
 hw/remote/Kconfig            |  3 +-
 6 files changed, 112 insertions(+), 1 deletion(-)
 create mode 100644 include/hw/pci-host/remote.h
 create mode 100644 hw/pci-host/remote.c

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ M: John G Johnson <john.g.johnson@oracle.com>
 S: Maintained
 F: docs/devel/multi-process.rst
 F: docs/system/multi-process.rst
+F: hw/pci-host/remote.c
+F: include/hw/pci-host/remote.h
 
 Build and test automation
 -------------------------
diff --git a/include/hw/pci-host/remote.h b/include/hw/pci-host/remote.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/include/hw/pci-host/remote.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * PCI Host for remote device
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef REMOTE_PCIHOST_H
+#define REMOTE_PCIHOST_H
+
+#include "exec/memory.h"
+#include "hw/pci/pcie_host.h"
+
+#define TYPE_REMOTE_PCIHOST "remote-pcihost"
+OBJECT_DECLARE_SIMPLE_TYPE(RemotePCIHost, REMOTE_PCIHOST)
+
+struct RemotePCIHost {
+    /*< private >*/
+    PCIExpressHost parent_obj;
+    /*< public >*/
+
+    MemoryRegion *mr_pci_mem;
+    MemoryRegion *mr_sys_io;
+};
+
+#endif
diff --git a/hw/pci-host/remote.c b/hw/pci-host/remote.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/pci-host/remote.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Remote PCI host device
+ *
+ * Unlike PCI host devices that model physical hardware, the purpose
+ * of this PCI host is to host multi-process QEMU devices.
+ *
+ * Multi-process QEMU extends the PCI host of a QEMU machine into a
+ * remote process. Any PCI device attached to the remote process is
+ * visible in the QEMU guest. This allows existing QEMU device models
+ * to be reused in the remote process.
+ *
+ * This PCI host is purely a container for PCI devices. It's fake in the
+ * sense that the guest never sees this PCI host and has no way of
+ * accessing it. Its job is just to provide the environment that QEMU
+ * PCI device models need when running in a remote process.
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "hw/pci/pci.h"
+#include "hw/pci/pci_host.h"
+#include "hw/pci/pcie_host.h"
+#include "hw/qdev-properties.h"
+#include "hw/pci-host/remote.h"
+#include "exec/memory.h"
+
+static const char *remote_pcihost_root_bus_path(PCIHostState *host_bridge,
+                                                PCIBus *rootbus)
+{
+    return "0000:00";
+}
+
+static void remote_pcihost_realize(DeviceState *dev, Error **errp)
+{
+    PCIHostState *pci = PCI_HOST_BRIDGE(dev);
+    RemotePCIHost *s = REMOTE_PCIHOST(dev);
+
+    pci->bus = pci_root_bus_new(DEVICE(s), "remote-pci",
+                                s->mr_pci_mem, s->mr_sys_io,
+                                0, TYPE_PCIE_BUS);
+}
+
+static void remote_pcihost_class_init(ObjectClass *klass, void *data)
+{
+    DeviceClass *dc = DEVICE_CLASS(klass);
+    PCIHostBridgeClass *hc = PCI_HOST_BRIDGE_CLASS(klass);
+
+    hc->root_bus_path = remote_pcihost_root_bus_path;
+    dc->realize = remote_pcihost_realize;
+
+    dc->user_creatable = false;
+    set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
+    dc->fw_name = "pci";
+}
+
+static const TypeInfo remote_pcihost_info = {
+    .name = TYPE_REMOTE_PCIHOST,
+    .parent = TYPE_PCIE_HOST_BRIDGE,
+    .instance_size = sizeof(RemotePCIHost),
+    .class_init = remote_pcihost_class_init,
+};
+
+static void remote_pcihost_register(void)
+{
+    type_register_static(&remote_pcihost_info);
+}
+
+type_init(remote_pcihost_register)
diff --git a/hw/pci-host/Kconfig b/hw/pci-host/Kconfig
index XXXXXXX..XXXXXXX 100644
--- a/hw/pci-host/Kconfig
+++ b/hw/pci-host/Kconfig
@@ -XXX,XX +XXX,XX @@ config PCI_POWERNV
     select PCI_EXPRESS
     select MSI_NONBROKEN
     select PCIE_PORT
+
+config REMOTE_PCIHOST
+    bool
diff --git a/hw/pci-host/meson.build b/hw/pci-host/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/pci-host/meson.build
+++ b/hw/pci-host/meson.build
@@ -XXX,XX +XXX,XX @@ pci_ss.add(when: 'CONFIG_PCI_EXPRESS_XILINX', if_true: files('xilinx-pcie.c'))
 pci_ss.add(when: 'CONFIG_PCI_I440FX', if_true: files('i440fx.c'))
 pci_ss.add(when: 'CONFIG_PCI_SABRE', if_true: files('sabre.c'))
 pci_ss.add(when: 'CONFIG_XEN_IGD_PASSTHROUGH', if_true: files('xen_igd_pt.c'))
+pci_ss.add(when: 'CONFIG_REMOTE_PCIHOST', if_true: files('remote.c'))
 
 # PPC devices
 pci_ss.add(when: 'CONFIG_PREP_PCI', if_true: files('prep.c'))
diff --git a/hw/remote/Kconfig b/hw/remote/Kconfig
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/Kconfig
+++ b/hw/remote/Kconfig
@@ -XXX,XX +XXX,XX @@
 config MULTIPROCESS
     bool
-    depends on PCI && KVM
+    depends on PCI && PCI_EXPRESS && KVM
+    select REMOTE_PCIHOST
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

x-remote-machine object sets up various subsystems of the remote
device process. Instantiate PCI host bridge object and initialize RAM, IO &
PCI memory regions.

Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: c537f38d17f90453ca610c6b70cf3480274e0ba1.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                  |  2 ++
 include/hw/pci-host/remote.h |  1 +
 include/hw/remote/machine.h  | 27 ++++++++++++++
 hw/remote/machine.c          | 70 ++++++++++++++++++++++++++++++++++++
 hw/meson.build               |  1 +
 hw/remote/meson.build        |  5 +++
 6 files changed, 106 insertions(+)
 create mode 100644 include/hw/remote/machine.h
 create mode 100644 hw/remote/machine.c
 create mode 100644 hw/remote/meson.build

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: docs/devel/multi-process.rst
 F: docs/system/multi-process.rst
 F: hw/pci-host/remote.c
 F: include/hw/pci-host/remote.h
+F: hw/remote/machine.c
+F: include/hw/remote/machine.h
 
 Build and test automation
 -------------------------
diff --git a/include/hw/pci-host/remote.h b/include/hw/pci-host/remote.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/pci-host/remote.h
+++ b/include/hw/pci-host/remote.h
@@ -XXX,XX +XXX,XX @@ struct RemotePCIHost {
 
     MemoryRegion *mr_pci_mem;
     MemoryRegion *mr_sys_io;
+    MemoryRegion *mr_sys_mem;
 };
 
 #endif
diff --git a/include/hw/remote/machine.h b/include/hw/remote/machine.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/include/hw/remote/machine.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * Remote machine configuration
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef REMOTE_MACHINE_H
+#define REMOTE_MACHINE_H
+
+#include "qom/object.h"
+#include "hw/boards.h"
+#include "hw/pci-host/remote.h"
+
+struct RemoteMachineState {
+    MachineState parent_obj;
+
+    RemotePCIHost *host;
+};
+
+#define TYPE_REMOTE_MACHINE "x-remote-machine"
+OBJECT_DECLARE_SIMPLE_TYPE(RemoteMachineState, REMOTE_MACHINE)
+
+#endif
diff --git a/hw/remote/machine.c b/hw/remote/machine.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/machine.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Machine for remote device
+ *
+ *  This machine type is used by the remote device process in multi-process
+ *  QEMU. QEMU device models depend on parent busses, interrupt controllers,
+ *  memory regions, etc. The remote machine type offers this environment so
+ *  that QEMU device models can be used as remote devices.
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "hw/remote/machine.h"
+#include "exec/address-spaces.h"
+#include "exec/memory.h"
+#include "qapi/error.h"
+
+static void remote_machine_init(MachineState *machine)
+{
+    MemoryRegion *system_memory, *system_io, *pci_memory;
+    RemoteMachineState *s = REMOTE_MACHINE(machine);
+    RemotePCIHost *rem_host;
+
+    system_memory = get_system_memory();
+    system_io = get_system_io();
+
+    pci_memory = g_new(MemoryRegion, 1);
+    memory_region_init(pci_memory, NULL, "pci", UINT64_MAX);
+
+    rem_host = REMOTE_PCIHOST(qdev_new(TYPE_REMOTE_PCIHOST));
+
+    rem_host->mr_pci_mem = pci_memory;
+    rem_host->mr_sys_mem = system_memory;
+    rem_host->mr_sys_io = system_io;
+
+    s->host = rem_host;
+
+    object_property_add_child(OBJECT(s), "remote-pcihost", OBJECT(rem_host));
+    memory_region_add_subregion_overlap(system_memory, 0x0, pci_memory, -1);
+
+    qdev_realize(DEVICE(rem_host), sysbus_get_default(), &error_fatal);
+}
+
+static void remote_machine_class_init(ObjectClass *oc, void *data)
+{
+    MachineClass *mc = MACHINE_CLASS(oc);
+
+    mc->init = remote_machine_init;
+    mc->desc = "Experimental remote machine";
+}
+
+static const TypeInfo remote_machine = {
+    .name = TYPE_REMOTE_MACHINE,
+    .parent = TYPE_MACHINE,
+    .instance_size = sizeof(RemoteMachineState),
+    .class_init = remote_machine_class_init,
+};
+
+static void remote_machine_register_types(void)
+{
+    type_register_static(&remote_machine);
+}
+
+type_init(remote_machine_register_types);
diff --git a/hw/meson.build b/hw/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/meson.build
+++ b/hw/meson.build
@@ -XXX,XX +XXX,XX @@ subdir('moxie')
 subdir('nios2')
 subdir('openrisc')
 subdir('ppc')
+subdir('remote')
 subdir('riscv')
 subdir('rx')
 subdir('s390x')
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@
+remote_ss = ss.source_set()
+
+remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('machine.c'))
+
+softmmu_ss.add_all(when: 'CONFIG_MULTIPROCESS', if_true: remote_ss)
-- 
2.29.2

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Adds qio_channel_writev_full_all() to transmit both data and FDs.
Refactors existing code to use this helper.

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 480fbf1fe4152495d60596c9b665124549b426a5.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/io/channel.h | 25 +++++++++++++++++++++++++
 io/channel.c         | 15 ++++++++++++++-
 2 files changed, 39 insertions(+), 1 deletion(-)

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Adds qio_channel_readv_full_all_eof() and qio_channel_readv_full_all()
to read both data and FDs. Refactors existing code to use these helpers.

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: b059c4cc0fb741e794d644c144cc21372cad877d.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/io/channel.h |  53 +++++++++++++++++++++++
 io/channel.c         | 101 ++++++++++++++++++++++++++++++++++---------
 2 files changed, 134 insertions(+), 20 deletions(-)

diff --git a/include/io/channel.h b/include/io/channel.h
index XXXXXXX..XXXXXXX 100644
--- a/include/io/channel.h
+++ b/include/io/channel.h
@@ -XXX,XX +XXX,XX @@ void qio_channel_set_aio_fd_handler(QIOChannel *ioc,
                                     IOHandler *io_write,
                                     void *opaque);
 
+/**
+ * qio_channel_readv_full_all_eof:
+ * @ioc: the channel object
+ * @iov: the array of memory regions to read data to
+ * @niov: the length of the @iov array
+ * @fds: an array of file handles to read
+ * @nfds: number of file handles in @fds
+ * @errp: pointer to a NULL-initialized error object
+ *
+ *
+ * Performs same function as qio_channel_readv_all_eof.
+ * Additionally, attempts to read file descriptors shared
+ * over the channel. The function will wait for all
+ * requested data to be read, yielding from the current
+ * coroutine if required. data refers to both file
+ * descriptors and the iovs.
+ *
+ * Returns: 1 if all bytes were read, 0 if end-of-file
+ *          occurs without data, or -1 on error
+ */
+
+int qio_channel_readv_full_all_eof(QIOChannel *ioc,
+                                   const struct iovec *iov,
+                                   size_t niov,
+                                   int **fds, size_t *nfds,
+                                   Error **errp);
+
+/**
+ * qio_channel_readv_full_all:
+ * @ioc: the channel object
+ * @iov: the array of memory regions to read data to
+ * @niov: the length of the @iov array
+ * @fds: an array of file handles to read
+ * @nfds: number of file handles in @fds
+ * @errp: pointer to a NULL-initialized error object
+ *
+ *
+ * Performs same function as qio_channel_readv_all_eof.
+ * Additionally, attempts to read file descriptors shared
+ * over the channel. The function will wait for all
+ * requested data to be read, yielding from the current
+ * coroutine if required. data refers to both file
+ * descriptors and the iovs.
+ *
+ * Returns: 0 if all bytes were read, or -1 on error
+ */
+
+int qio_channel_readv_full_all(QIOChannel *ioc,
+                               const struct iovec *iov,
+                               size_t niov,
+                               int **fds, size_t *nfds,
+                               Error **errp);
+
 /**
  * qio_channel_writev_full_all:
  * @ioc: the channel object
diff --git a/io/channel.c b/io/channel.c
index XXXXXXX..XXXXXXX 100644
--- a/io/channel.c
+++ b/io/channel.c
@@ -XXX,XX +XXX,XX @@ int qio_channel_readv_all_eof(QIOChannel *ioc,
                               const struct iovec *iov,
                               size_t niov,
                               Error **errp)
+{
+    return qio_channel_readv_full_all_eof(ioc, iov, niov, NULL, NULL, errp);
+}
+
+int qio_channel_readv_all(QIOChannel *ioc,
+                          const struct iovec *iov,
+                          size_t niov,
+                          Error **errp)
+{
+    return qio_channel_readv_full_all(ioc, iov, niov, NULL, NULL, errp);
+}
+
+int qio_channel_readv_full_all_eof(QIOChannel *ioc,
+                                   const struct iovec *iov,
+                                   size_t niov,
+                                   int **fds, size_t *nfds,
+                                   Error **errp)
 {
     int ret = -1;
     struct iovec *local_iov = g_new(struct iovec, niov);
     struct iovec *local_iov_head = local_iov;
     unsigned int nlocal_iov = niov;
+    int **local_fds = fds;
+    size_t *local_nfds = nfds;
     bool partial = false;
 
+    if (nfds) {
+        *nfds = 0;
+    }
+
+    if (fds) {
+        *fds = NULL;
+    }
+
     nlocal_iov = iov_copy(local_iov, nlocal_iov,
                           iov, niov,
                           0, iov_size(iov, niov));
 
-    while (nlocal_iov > 0) {
+    while ((nlocal_iov > 0) || local_fds) {
         ssize_t len;
-        len = qio_channel_readv(ioc, local_iov, nlocal_iov, errp);
+        len = qio_channel_readv_full(ioc, local_iov, nlocal_iov, local_fds,
+                                     local_nfds, errp);
         if (len == QIO_CHANNEL_ERR_BLOCK) {
             if (qemu_in_coroutine()) {
                 qio_channel_yield(ioc, G_IO_IN);
@@ -XXX,XX +XXX,XX @@ int qio_channel_readv_all_eof(QIOChannel *ioc,
                 qio_channel_wait(ioc, G_IO_IN);
             }
             continue;
-        } else if (len < 0) {
-            goto cleanup;
-        } else if (len == 0) {
-            if (partial) {
-                error_setg(errp,
-                           "Unexpected end-of-file before all bytes were read");
-            } else {
+        }
+
+        if (len == 0) {
+            if (local_nfds && *local_nfds) {
+                /*
+                 * Got some FDs, but no data yet. This isn't an EOF
+                 * scenario (yet), so carry on to try to read data
+                 * on next loop iteration
+                 */
+                goto next_iter;
+            } else if (!partial) {
+                /* No fds and no data - EOF before any data read */
                 ret = 0;
+                goto cleanup;
+            } else {
+                len = -1;
+                error_setg(errp,
+                           "Unexpected end-of-file before all data were read");
+                /* Fallthrough into len < 0 handling */
+            }
+        }
+
+        if (len < 0) {
+            /* Close any FDs we previously received */
+            if (nfds && fds) {
+                size_t i;
+                for (i = 0; i < (*nfds); i++) {
+                    close((*fds)[i]);
+                }
+                g_free(*fds);
+                *fds = NULL;
+                *nfds = 0;
             }
             goto cleanup;
         }
 
+        if (nlocal_iov) {
+            iov_discard_front(&local_iov, &nlocal_iov, len);
+        }
+
+next_iter:
         partial = true;
-        iov_discard_front(&local_iov, &nlocal_iov, len);
+        local_fds = NULL;
+        local_nfds = NULL;
     }
 
     ret = 1;
@@ -XXX,XX +XXX,XX @@ int qio_channel_readv_all_eof(QIOChannel *ioc,
     return ret;
 }
 
-int qio_channel_readv_all(QIOChannel *ioc,
-                          const struct iovec *iov,
-                          size_t niov,
-                          Error **errp)
+int qio_channel_readv_full_all(QIOChannel *ioc,
+                               const struct iovec *iov,
+                               size_t niov,
+                               int **fds, size_t *nfds,
+                               Error **errp)
 {
-    int ret = qio_channel_readv_all_eof(ioc, iov, niov, errp);
+    int ret = qio_channel_readv_full_all_eof(ioc, iov, niov, fds, nfds, errp);
 
     if (ret == 0) {
-        ret = -1;
-        error_setg(errp,
-                   "Unexpected end-of-file before all bytes were read");
-    } else if (ret == 1) {
-        ret = 0;
+        error_prepend(errp,
+                      "Unexpected end-of-file before all data were read.");
+        return -1;
     }
+    if (ret == 1) {
+        return 0;
+    }
+
     return ret;
 }
 
-- 
2.29.2

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Defines MPQemuMsg, which is the message that is sent to the remote
process. This message is sent over QIOChannel and is used to
command the remote process to perform various tasks.
Define transmission functions used by proxy and by remote.

[Replace struct iovec send[2] = {0} with {} to make clang happy as
suggested by Peter Maydell <peter.maydell@linaro.org>.
--Stefan]

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                     |   2 +
 meson.build                     |   1 +
 hw/remote/trace.h               |   1 +
 include/hw/remote/mpqemu-link.h |  63 ++++++++++
 include/sysemu/iothread.h       |   6 +
 hw/remote/mpqemu-link.c         | 205 ++++++++++++++++++++++++++++++++
 iothread.c                      |   6 +
 hw/remote/meson.build           |   1 +
 hw/remote/trace-events          |   4 +
 9 files changed, 289 insertions(+)
 create mode 100644 hw/remote/trace.h
 create mode 100644 include/hw/remote/mpqemu-link.h
 create mode 100644 hw/remote/mpqemu-link.c
 create mode 100644 hw/remote/trace-events

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: hw/pci-host/remote.c
 F: include/hw/pci-host/remote.h
 F: hw/remote/machine.c
 F: include/hw/remote/machine.h
+F: hw/remote/mpqemu-link.c
+F: include/hw/remote/mpqemu-link.h
 
 Build and test automation
 -------------------------
diff --git a/meson.build b/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/meson.build
+++ b/meson.build
@@ -XXX,XX +XXX,XX @@ if have_system
     'net',
     'softmmu',
     'ui',
+    'hw/remote',
   ]
 endif
 if have_system or have_user
diff --git a/hw/remote/trace.h b/hw/remote/trace.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/trace.h
@@ -0,0 +1 @@
+#include "trace/trace-hw_remote.h"
diff --git a/include/hw/remote/mpqemu-link.h b/include/hw/remote/mpqemu-link.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/include/hw/remote/mpqemu-link.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * Communication channel between QEMU and remote device process
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef MPQEMU_LINK_H
+#define MPQEMU_LINK_H
+
+#include "qom/object.h"
+#include "qemu/thread.h"
+#include "io/channel.h"
+
+#define REMOTE_MAX_FDS 8
+
+#define MPQEMU_MSG_HDR_SIZE offsetof(MPQemuMsg, data.u64)
+
+/**
+ * MPQemuCmd:
+ *
+ * MPQemuCmd enum type to specify the command to be executed on the remote
+ * device.
+ *
+ * This uses a private protocol between QEMU and the remote process. vfio-user
+ * protocol would supersede this in the future.
+ *
+ */
+typedef enum {
+    MPQEMU_CMD_MAX,
+} MPQemuCmd;
+
+/**
+ * MPQemuMsg:
+ * @cmd: The remote command
+ * @size: Size of the data to be shared
+ * @data: Structured data
+ * @fds: File descriptors to be shared with remote device
+ *
+ * MPQemuMsg Format of the message sent to the remote device from QEMU.
+ *
+ */
+typedef struct {
+    int cmd;
+    size_t size;
+
+    union {
+        uint64_t u64;
+    } data;
+
+    int fds[REMOTE_MAX_FDS];
+    int num_fds;
+} MPQemuMsg;
+
+bool mpqemu_msg_send(MPQemuMsg *msg, QIOChannel *ioc, Error **errp);
+bool mpqemu_msg_recv(MPQemuMsg *msg, QIOChannel *ioc, Error **errp);
+
+bool mpqemu_msg_valid(MPQemuMsg *msg);
+
+#endif
diff --git a/include/sysemu/iothread.h b/include/sysemu/iothread.h
index XXXXXXX..XXXXXXX 100644
--- a/include/sysemu/iothread.h
+++ b/include/sysemu/iothread.h
@@ -XXX,XX +XXX,XX @@ IOThread *iothread_create(const char *id, Error **errp);
 void iothread_stop(IOThread *iothread);
 void iothread_destroy(IOThread *iothread);
 
+/*
+ * Returns true if executing withing IOThread context,
+ * false otherwise.
+ */
+bool qemu_in_iothread(void);
+
 #endif /* IOTHREAD_H */
diff --git a/hw/remote/mpqemu-link.c b/hw/remote/mpqemu-link.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/mpqemu-link.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Communication channel between QEMU and remote device process
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "qemu/module.h"
+#include "hw/remote/mpqemu-link.h"
+#include "qapi/error.h"
+#include "qemu/iov.h"
+#include "qemu/error-report.h"
+#include "qemu/main-loop.h"
+#include "io/channel.h"
+#include "sysemu/iothread.h"
+#include "trace.h"
+
+/*
+ * Send message over the ioc QIOChannel.
+ * This function is safe to call from:
+ * - main loop in co-routine context. Will block the main loop if not in
+ *   co-routine context;
+ * - vCPU thread with no co-routine context and if the channel is not part
+ *   of the main loop handling;
+ * - IOThread within co-routine context, outside of co-routine context
+ *   will block IOThread;
+ * Returns true if no errors were encountered, false otherwise.
+ */
+bool mpqemu_msg_send(MPQemuMsg *msg, QIOChannel *ioc, Error **errp)
+{
+    ERRP_GUARD();
+    bool iolock = qemu_mutex_iothread_locked();
+    bool iothread = qemu_in_iothread();
+    struct iovec send[2] = {};
+    int *fds = NULL;
+    size_t nfds = 0;
+    bool ret = false;
+
+    send[0].iov_base = msg;
+    send[0].iov_len = MPQEMU_MSG_HDR_SIZE;
+
+    send[1].iov_base = (void *)&msg->data;
+    send[1].iov_len = msg->size;
+
+    if (msg->num_fds) {
+        nfds = msg->num_fds;
+        fds = msg->fds;
+    }
+
+    /*
+     * Dont use in IOThread out of co-routine context as
+     * it will block IOThread.
+     */
+    assert(qemu_in_coroutine() || !iothread);
+
+    /*
+     * Skip unlocking/locking iothread lock when the IOThread is running
+     * in co-routine context. Co-routine context is asserted above
+     * for IOThread case.
+     * Also skip lock handling while in a co-routine in the main context.
+     */
+    if (iolock && !iothread && !qemu_in_coroutine()) {
+        qemu_mutex_unlock_iothread();
+    }
+
+    if (!qio_channel_writev_full_all(ioc, send, G_N_ELEMENTS(send),
+                                    fds, nfds, errp)) {
+        ret = true;
+    } else {
+        trace_mpqemu_send_io_error(msg->cmd, msg->size, nfds);
+    }
+
+    if (iolock && !iothread && !qemu_in_coroutine()) {
+        /* See above comment why skip locking here. */
+        qemu_mutex_lock_iothread();
+    }
+
+    return ret;
+}
+
+/*
+ * Read message from the ioc QIOChannel.
+ * This function is safe to call from:
+ * - From main loop in co-routine context. Will block the main loop if not in
+ *   co-routine context;
+ * - From vCPU thread with no co-routine context and if the channel is not part
+ *   of the main loop handling;
+ * - From IOThread within co-routine context, outside of co-routine context
+ *   will block IOThread;
+ */
+static ssize_t mpqemu_read(QIOChannel *ioc, void *buf, size_t len, int **fds,
+                           size_t *nfds, Error **errp)
+{
+    ERRP_GUARD();
+    struct iovec iov = { .iov_base = buf, .iov_len = len };
+    bool iolock = qemu_mutex_iothread_locked();
+    bool iothread = qemu_in_iothread();
+    int ret = -1;
+
+    /*
+     * Dont use in IOThread out of co-routine context as
+     * it will block IOThread.
+     */
+    assert(qemu_in_coroutine() || !iothread);
+
+    if (iolock && !iothread && !qemu_in_coroutine()) {
+        qemu_mutex_unlock_iothread();
+    }
+
+    ret = qio_channel_readv_full_all_eof(ioc, &iov, 1, fds, nfds, errp);
+
+    if (iolock && !iothread && !qemu_in_coroutine()) {
+        qemu_mutex_lock_iothread();
+    }
+
+    return (ret <= 0) ? ret : iov.iov_len;
+}
+
+bool mpqemu_msg_recv(MPQemuMsg *msg, QIOChannel *ioc, Error **errp)
+{
+    ERRP_GUARD();
+    g_autofree int *fds = NULL;
+    size_t nfds = 0;
+    ssize_t len;
+    bool ret = false;
+
+    len = mpqemu_read(ioc, msg, MPQEMU_MSG_HDR_SIZE, &fds, &nfds, errp);
+    if (len <= 0) {
+        goto fail;
+    } else if (len != MPQEMU_MSG_HDR_SIZE) {
+        error_setg(errp, "Message header corrupted");
+        goto fail;
+    }
+
+    if (msg->size > sizeof(msg->data)) {
+        error_setg(errp, "Invalid size for message");
+        goto fail;
+    }
+
+    if (!msg->size) {
+        goto copy_fds;
+    }
+
+    len = mpqemu_read(ioc, &msg->data, msg->size, NULL, NULL, errp);
+    if (len <= 0) {
+        goto fail;
+    }
+    if (len != msg->size) {
+        error_setg(errp, "Unable to read full message");
+        goto fail;
+    }
+
+copy_fds:
+    msg->num_fds = nfds;
+    if (nfds > G_N_ELEMENTS(msg->fds)) {
+        error_setg(errp,
+                   "Overflow error: received %zu fds, more than max of %d fds",
+                   nfds, REMOTE_MAX_FDS);
+        goto fail;
+    }
+    if (nfds) {
+        memcpy(msg->fds, fds, nfds * sizeof(int));
+    }
+
+    ret = true;
+
+fail:
+    if (*errp) {
+        trace_mpqemu_recv_io_error(msg->cmd, msg->size, nfds);
+    }
+    while (*errp && nfds) {
+        close(fds[nfds - 1]);
+        nfds--;
+    }
+
+    return ret;
+}
+
+bool mpqemu_msg_valid(MPQemuMsg *msg)
+{
+    if (msg->cmd >= MPQEMU_CMD_MAX && msg->cmd < 0) {
+        return false;
+    }
+
+    /* Verify FDs. */
+    if (msg->num_fds >= REMOTE_MAX_FDS) {
+        return false;
+    }
+
+    if (msg->num_fds > 0) {
+        for (int i = 0; i < msg->num_fds; i++) {
+            if (fcntl(msg->fds[i], F_GETFL) == -1) {
+                return false;
+            }
+        }
+    }
+
+    return true;
+}
diff --git a/iothread.c b/iothread.c
index XXXXXXX..XXXXXXX 100644
--- a/iothread.c
+++ b/iothread.c
@@ -XXX,XX +XXX,XX @@ IOThread *iothread_by_id(const char *id)
 {
     return IOTHREAD(object_resolve_path_type(id, TYPE_IOTHREAD, NULL));
 }
+
+bool qemu_in_iothread(void)
+{
+    return qemu_get_current_aio_context() == qemu_get_aio_context() ?
+                    false : true;
+}
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/meson.build
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@
 remote_ss = ss.source_set()
 
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('machine.c'))
+remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('mpqemu-link.c'))
 
 softmmu_ss.add_all(when: 'CONFIG_MULTIPROCESS', if_true: remote_ss)
diff --git a/hw/remote/trace-events b/hw/remote/trace-events
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/trace-events
@@ -XXX,XX +XXX,XX @@
+# multi-process trace events
+
+mpqemu_send_io_error(int cmd, int size, int nfds) "send command %d size %d, %d file descriptors to remote process"
+mpqemu_recv_io_error(int cmd, int size, int nfds) "failed to receive %d size %d, %d file descriptors to remote process"
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

Initializes the message handler function in the remote process. It is
called whenever there's an event pending on QIOChannel that registers
this function.

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 99d38d8b93753a6409ac2340e858858cda59ab1b.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                 |  1 +
 include/hw/remote/machine.h |  9 ++++++
 hw/remote/message.c         | 57 +++++++++++++++++++++++++++++++++++++
 hw/remote/meson.build       |  1 +
 4 files changed, 68 insertions(+)
 create mode 100644 hw/remote/message.c

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: hw/remote/machine.c
 F: include/hw/remote/machine.h
 F: hw/remote/mpqemu-link.c
 F: include/hw/remote/mpqemu-link.h
+F: hw/remote/message.c
 
 Build and test automation
 -------------------------
diff --git a/include/hw/remote/machine.h b/include/hw/remote/machine.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/machine.h
+++ b/include/hw/remote/machine.h
@@ -XXX,XX +XXX,XX @@
 #include "qom/object.h"
 #include "hw/boards.h"
 #include "hw/pci-host/remote.h"
+#include "io/channel.h"
 
 struct RemoteMachineState {
     MachineState parent_obj;
@@ -XXX,XX +XXX,XX @@ struct RemoteMachineState {
     RemotePCIHost *host;
 };
 
+/* Used to pass to co-routine device and ioc. */
+typedef struct RemoteCommDev {
+    PCIDevice *dev;
+    QIOChannel *ioc;
+} RemoteCommDev;
+
 #define TYPE_REMOTE_MACHINE "x-remote-machine"
 OBJECT_DECLARE_SIMPLE_TYPE(RemoteMachineState, REMOTE_MACHINE)
 
+void coroutine_fn mpqemu_remote_msg_loop_co(void *data);
+
 #endif
diff --git a/hw/remote/message.c b/hw/remote/message.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/message.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Copyright © 2020, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL-v2, version 2 or later.
+ *
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "hw/remote/machine.h"
+#include "io/channel.h"
+#include "hw/remote/mpqemu-link.h"
+#include "qapi/error.h"
+#include "sysemu/runstate.h"
+
+void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
+{
+    g_autofree RemoteCommDev *com = (RemoteCommDev *)data;
+    PCIDevice *pci_dev = NULL;
+    Error *local_err = NULL;
+
+    assert(com->ioc);
+
+    pci_dev = com->dev;
+    for (; !local_err;) {
+        MPQemuMsg msg = {0};
+
+        if (!mpqemu_msg_recv(&msg, com->ioc, &local_err)) {
+            break;
+        }
+
+        if (!mpqemu_msg_valid(&msg)) {
+            error_setg(&local_err, "Received invalid message from proxy"
+                                   "in remote process pid="FMT_pid"",
+                                   getpid());
+            break;
+        }
+
+        switch (msg.cmd) {
+        default:
+            error_setg(&local_err,
+                       "Unknown command (%d) received for device %s"
+                       " (pid="FMT_pid")",
+                       msg.cmd, DEVICE(pci_dev)->id, getpid());
+        }
+    }
+
+    if (local_err) {
+        error_report_err(local_err);
+        qemu_system_shutdown_request(SHUTDOWN_CAUSE_HOST_ERROR);
+    } else {
+        qemu_system_shutdown_request(SHUTDOWN_CAUSE_GUEST_SHUTDOWN);
+    }
+}
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/meson.build
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@ remote_ss = ss.source_set()
 
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('machine.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('mpqemu-link.c'))
+remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('message.c'))
 
 softmmu_ss.add_all(when: 'CONFIG_MULTIPROCESS', if_true: remote_ss)
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

Associate the file descriptor for a PCIDevice in remote process with
DeviceState object.

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: f405a2ed5d7518b87bea7c59cfdf334d67e5ee51.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS            |   1 +
 hw/remote/remote-obj.c | 203 +++++++++++++++++++++++++++++++++++++++++
 hw/remote/meson.build  |   1 +
 3 files changed, 205 insertions(+)
 create mode 100644 hw/remote/remote-obj.c

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: include/hw/remote/machine.h
 F: hw/remote/mpqemu-link.c
 F: include/hw/remote/mpqemu-link.h
 F: hw/remote/message.c
+F: hw/remote/remote-obj.c
 
 Build and test automation
 -------------------------
diff --git a/hw/remote/remote-obj.c b/hw/remote/remote-obj.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/remote-obj.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Copyright © 2020, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL-v2, version 2 or later.
+ *
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "qemu/error-report.h"
+#include "qemu/notify.h"
+#include "qom/object_interfaces.h"
+#include "hw/qdev-core.h"
+#include "io/channel.h"
+#include "hw/qdev-core.h"
+#include "hw/remote/machine.h"
+#include "io/channel-util.h"
+#include "qapi/error.h"
+#include "sysemu/sysemu.h"
+#include "hw/pci/pci.h"
+#include "qemu/sockets.h"
+#include "monitor/monitor.h"
+
+#define TYPE_REMOTE_OBJECT "x-remote-object"
+OBJECT_DECLARE_TYPE(RemoteObject, RemoteObjectClass, REMOTE_OBJECT)
+
+struct RemoteObjectClass {
+    ObjectClass parent_class;
+
+    unsigned int nr_devs;
+    unsigned int max_devs;
+};
+
+struct RemoteObject {
+    /* private */
+    Object parent;
+
+    Notifier machine_done;
+
+    int32_t fd;
+    char *devid;
+
+    QIOChannel *ioc;
+
+    DeviceState *dev;
+    DeviceListener listener;
+};
+
+static void remote_object_set_fd(Object *obj, const char *str, Error **errp)
+{
+    RemoteObject *o = REMOTE_OBJECT(obj);
+    int fd = -1;
+
+    fd = monitor_fd_param(monitor_cur(), str, errp);
+    if (fd == -1) {
+        error_prepend(errp, "Could not parse remote object fd %s:", str);
+        return;
+    }
+
+    if (!fd_is_socket(fd)) {
+        error_setg(errp, "File descriptor '%s' is not a socket", str);
+        close(fd);
+        return;
+    }
+
+    o->fd = fd;
+}
+
+static void remote_object_set_devid(Object *obj, const char *str, Error **errp)
+{
+    RemoteObject *o = REMOTE_OBJECT(obj);
+
+    g_free(o->devid);
+
+    o->devid = g_strdup(str);
+}
+
+static void remote_object_unrealize_listener(DeviceListener *listener,
+                                             DeviceState *dev)
+{
+    RemoteObject *o = container_of(listener, RemoteObject, listener);
+
+    if (o->dev == dev) {
+        object_unref(OBJECT(o));
+    }
+}
+
+static void remote_object_machine_done(Notifier *notifier, void *data)
+{
+    RemoteObject *o = container_of(notifier, RemoteObject, machine_done);
+    DeviceState *dev = NULL;
+    QIOChannel *ioc = NULL;
+    Coroutine *co = NULL;
+    RemoteCommDev *comdev = NULL;
+    Error *err = NULL;
+
+    dev = qdev_find_recursive(sysbus_get_default(), o->devid);
+    if (!dev || !object_dynamic_cast(OBJECT(dev), TYPE_PCI_DEVICE)) {
+        error_report("%s is not a PCI device", o->devid);
+        return;
+    }
+
+    ioc = qio_channel_new_fd(o->fd, &err);
+    if (!ioc) {
+        error_report_err(err);
+        return;
+    }
+    qio_channel_set_blocking(ioc, false, NULL);
+
+    o->dev = dev;
+
+    o->listener.unrealize = remote_object_unrealize_listener;
+    device_listener_register(&o->listener);
+
+    /* co-routine should free this. */
+    comdev = g_new0(RemoteCommDev, 1);
+    *comdev = (RemoteCommDev) {
+        .ioc = ioc,
+        .dev = PCI_DEVICE(dev),
+    };
+
+    co = qemu_coroutine_create(mpqemu_remote_msg_loop_co, comdev);
+    qemu_coroutine_enter(co);
+}
+
+static void remote_object_init(Object *obj)
+{
+    RemoteObjectClass *k = REMOTE_OBJECT_GET_CLASS(obj);
+    RemoteObject *o = REMOTE_OBJECT(obj);
+
+    if (k->nr_devs >= k->max_devs) {
+        error_report("Reached maximum number of devices: %u", k->max_devs);
+        return;
+    }
+
+    o->ioc = NULL;
+    o->fd = -1;
+    o->devid = NULL;
+
+    k->nr_devs++;
+
+    o->machine_done.notify = remote_object_machine_done;
+    qemu_add_machine_init_done_notifier(&o->machine_done);
+}
+
+static void remote_object_finalize(Object *obj)
+{
+    RemoteObjectClass *k = REMOTE_OBJECT_GET_CLASS(obj);
+    RemoteObject *o = REMOTE_OBJECT(obj);
+
+    device_listener_unregister(&o->listener);
+
+    if (o->ioc) {
+        qio_channel_shutdown(o->ioc, QIO_CHANNEL_SHUTDOWN_BOTH, NULL);
+        qio_channel_close(o->ioc, NULL);
+    }
+
+    object_unref(OBJECT(o->ioc));
+
+    k->nr_devs--;
+    g_free(o->devid);
+}
+
+static void remote_object_class_init(ObjectClass *klass, void *data)
+{
+    RemoteObjectClass *k = REMOTE_OBJECT_CLASS(klass);
+
+    /*
+     * Limit number of supported devices to 1. This is done to avoid devices
+     * from one VM accessing the RAM of another VM. This is done until we
+     * start using separate address spaces for individual devices.
+     */
+    k->max_devs = 1;
+    k->nr_devs = 0;
+
+    object_class_property_add_str(klass, "fd", NULL, remote_object_set_fd);
+    object_class_property_add_str(klass, "devid", NULL,
+                                  remote_object_set_devid);
+}
+
+static const TypeInfo remote_object_info = {
+    .name = TYPE_REMOTE_OBJECT,
+    .parent = TYPE_OBJECT,
+    .instance_size = sizeof(RemoteObject),
+    .instance_init = remote_object_init,
+    .instance_finalize = remote_object_finalize,
+    .class_size = sizeof(RemoteObjectClass),
+    .class_init = remote_object_class_init,
+    .interfaces = (InterfaceInfo[]) {
+        { TYPE_USER_CREATABLE },
+        { }
+    }
+};
+
+static void register_types(void)
+{
+    type_register_static(&remote_object_info);
+}
+
+type_init(register_types);
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/meson.build
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@ remote_ss = ss.source_set()
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('machine.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('mpqemu-link.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('message.c'))
+remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('remote-obj.c'))
 
 softmmu_ss.add_all(when: 'CONFIG_MULTIPROCESS', if_true: remote_ss)
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

SyncSysMemMsg message format is defined. It is used to send
file descriptors of the RAM regions to remote device.
RAM on the remote device is configured with a set of file descriptors.
Old RAM regions are deleted and new regions, each with an fd, is
added to the RAM.

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 7d2d1831d812e85f681e7a8ab99e032cf4704689.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                     |  2 +
 include/hw/remote/memory.h      | 19 ++++++++++
 include/hw/remote/mpqemu-link.h | 10 +++++
 hw/remote/memory.c              | 65 +++++++++++++++++++++++++++++++++
 hw/remote/mpqemu-link.c         | 11 ++++++
 hw/remote/meson.build           |  2 +
 6 files changed, 109 insertions(+)
 create mode 100644 include/hw/remote/memory.h
 create mode 100644 hw/remote/memory.c

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: hw/remote/mpqemu-link.c
 F: include/hw/remote/mpqemu-link.h
 F: hw/remote/message.c
 F: hw/remote/remote-obj.c
+F: include/hw/remote/memory.h
+F: hw/remote/memory.c
 
 Build and test automation
 -------------------------
diff --git a/include/hw/remote/memory.h b/include/hw/remote/memory.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/include/hw/remote/memory.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * Memory manager for remote device
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef REMOTE_MEMORY_H
+#define REMOTE_MEMORY_H
+
+#include "exec/hwaddr.h"
+#include "hw/remote/mpqemu-link.h"
+
+void remote_sysmem_reconfig(MPQemuMsg *msg, Error **errp);
+
+#endif
diff --git a/include/hw/remote/mpqemu-link.h b/include/hw/remote/mpqemu-link.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/mpqemu-link.h
+++ b/include/hw/remote/mpqemu-link.h
@@ -XXX,XX +XXX,XX @@
 #include "qom/object.h"
 #include "qemu/thread.h"
 #include "io/channel.h"
+#include "exec/hwaddr.h"
 
 #define REMOTE_MAX_FDS 8
 
@@ -XXX,XX +XXX,XX @@
  *
  */
 typedef enum {
+    MPQEMU_CMD_SYNC_SYSMEM,
     MPQEMU_CMD_MAX,
 } MPQemuCmd;
 
+typedef struct {
+    hwaddr gpas[REMOTE_MAX_FDS];
+    uint64_t sizes[REMOTE_MAX_FDS];
+    off_t offsets[REMOTE_MAX_FDS];
+} SyncSysmemMsg;
+
 /**
  * MPQemuMsg:
  * @cmd: The remote command
@@ -XXX,XX +XXX,XX @@ typedef enum {
  * MPQemuMsg Format of the message sent to the remote device from QEMU.
  *
  */
+
 typedef struct {
     int cmd;
     size_t size;
 
     union {
         uint64_t u64;
+        SyncSysmemMsg sync_sysmem;
     } data;
 
     int fds[REMOTE_MAX_FDS];
diff --git a/hw/remote/memory.c b/hw/remote/memory.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/memory.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Memory manager for remote device
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "hw/remote/memory.h"
+#include "exec/address-spaces.h"
+#include "exec/ram_addr.h"
+#include "qapi/error.h"
+
+static void remote_sysmem_reset(void)
+{
+    MemoryRegion *sysmem, *subregion, *next;
+
+    sysmem = get_system_memory();
+
+    QTAILQ_FOREACH_SAFE(subregion, &sysmem->subregions, subregions_link, next) {
+        if (subregion->ram) {
+            memory_region_del_subregion(sysmem, subregion);
+            object_unparent(OBJECT(subregion));
+        }
+    }
+}
+
+void remote_sysmem_reconfig(MPQemuMsg *msg, Error **errp)
+{
+    ERRP_GUARD();
+    SyncSysmemMsg *sysmem_info = &msg->data.sync_sysmem;
+    MemoryRegion *sysmem, *subregion;
+    static unsigned int suffix;
+    int region;
+
+    sysmem = get_system_memory();
+
+    remote_sysmem_reset();
+
+    for (region = 0; region < msg->num_fds; region++) {
+        g_autofree char *name;
+        subregion = g_new(MemoryRegion, 1);
+        name = g_strdup_printf("remote-mem-%u", suffix++);
+        memory_region_init_ram_from_fd(subregion, NULL,
+                                       name, sysmem_info->sizes[region],
+                                       true, msg->fds[region],
+                                       sysmem_info->offsets[region],
+                                       errp);
+
+        if (*errp) {
+            g_free(subregion);
+            remote_sysmem_reset();
+            return;
+        }
+
+        memory_region_add_subregion(sysmem, sysmem_info->gpas[region],
+                                    subregion);
+
+    }
+}
diff --git a/hw/remote/mpqemu-link.c b/hw/remote/mpqemu-link.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/mpqemu-link.c
+++ b/hw/remote/mpqemu-link.c
@@ -XXX,XX +XXX,XX @@ bool mpqemu_msg_valid(MPQemuMsg *msg)
         }
     }
 
+     /* Verify message specific fields. */
+    switch (msg->cmd) {
+    case MPQEMU_CMD_SYNC_SYSMEM:
+        if (msg->num_fds == 0 || msg->size != sizeof(SyncSysmemMsg)) {
+            return false;
+        }
+        break;
+    default:
+        break;
+    }
+
     return true;
 }
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/meson.build
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@ remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('mpqemu-link.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('message.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('remote-obj.c'))
 
+specific_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('memory.c'))
+
 softmmu_ss.add_all(when: 'CONFIG_MULTIPROCESS', if_true: remote_ss)
-- 
2.29.2

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Defines a PCI Device proxy object as a child of TYPE_PCI_DEVICE.

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: b5186ebfedf8e557044d09a768846c59230ad3a7.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS               |  2 +
 include/hw/remote/proxy.h | 33 +++++++++++++
 hw/remote/proxy.c         | 99 +++++++++++++++++++++++++++++++++++++++
 hw/remote/meson.build     |  1 +
 4 files changed, 135 insertions(+)
 create mode 100644 include/hw/remote/proxy.h
 create mode 100644 hw/remote/proxy.c

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: hw/remote/message.c
 F: hw/remote/remote-obj.c
 F: include/hw/remote/memory.h
 F: hw/remote/memory.c
+F: hw/remote/proxy.c
+F: include/hw/remote/proxy.h
 
 Build and test automation
 -------------------------
diff --git a/include/hw/remote/proxy.h b/include/hw/remote/proxy.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/include/hw/remote/proxy.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef PROXY_H
+#define PROXY_H
+
+#include "hw/pci/pci.h"
+#include "io/channel.h"
+
+#define TYPE_PCI_PROXY_DEV "x-pci-proxy-dev"
+OBJECT_DECLARE_SIMPLE_TYPE(PCIProxyDev, PCI_PROXY_DEV)
+
+struct PCIProxyDev {
+    PCIDevice parent_dev;
+    char *fd;
+
+    /*
+     * Mutex used to protect the QIOChannel fd from
+     * the concurrent access by the VCPUs since proxy
+     * blocks while awaiting for the replies from the
+     * process remote.
+     */
+    QemuMutex io_mutex;
+    QIOChannel *ioc;
+    Error *migration_blocker;
+};
+
+#endif /* PROXY_H */
diff --git a/hw/remote/proxy.c b/hw/remote/proxy.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/proxy.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "hw/remote/proxy.h"
+#include "hw/pci/pci.h"
+#include "qapi/error.h"
+#include "io/channel-util.h"
+#include "hw/qdev-properties.h"
+#include "monitor/monitor.h"
+#include "migration/blocker.h"
+#include "qemu/sockets.h"
+
+static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
+{
+    ERRP_GUARD();
+    PCIProxyDev *dev = PCI_PROXY_DEV(device);
+    int fd;
+
+    if (!dev->fd) {
+        error_setg(errp, "fd parameter not specified for %s",
+                   DEVICE(device)->id);
+        return;
+    }
+
+    fd = monitor_fd_param(monitor_cur(), dev->fd, errp);
+    if (fd == -1) {
+        error_prepend(errp, "proxy: unable to parse fd %s: ", dev->fd);
+        return;
+    }
+
+    if (!fd_is_socket(fd)) {
+        error_setg(errp, "proxy: fd %d is not a socket", fd);
+        close(fd);
+        return;
+    }
+
+    dev->ioc = qio_channel_new_fd(fd, errp);
+
+    error_setg(&dev->migration_blocker, "%s does not support migration",
+               TYPE_PCI_PROXY_DEV);
+    migrate_add_blocker(dev->migration_blocker, errp);
+
+    qemu_mutex_init(&dev->io_mutex);
+    qio_channel_set_blocking(dev->ioc, true, NULL);
+}
+
+static void pci_proxy_dev_exit(PCIDevice *pdev)
+{
+    PCIProxyDev *dev = PCI_PROXY_DEV(pdev);
+
+    if (dev->ioc) {
+        qio_channel_close(dev->ioc, NULL);
+    }
+
+    migrate_del_blocker(dev->migration_blocker);
+
+    error_free(dev->migration_blocker);
+}
+
+static Property proxy_properties[] = {
+    DEFINE_PROP_STRING("fd", PCIProxyDev, fd),
+    DEFINE_PROP_END_OF_LIST(),
+};
+
+static void pci_proxy_dev_class_init(ObjectClass *klass, void *data)
+{
+    DeviceClass *dc = DEVICE_CLASS(klass);
+    PCIDeviceClass *k = PCI_DEVICE_CLASS(klass);
+
+    k->realize = pci_proxy_dev_realize;
+    k->exit = pci_proxy_dev_exit;
+    device_class_set_props(dc, proxy_properties);
+}
+
+static const TypeInfo pci_proxy_dev_type_info = {
+    .name          = TYPE_PCI_PROXY_DEV,
+    .parent        = TYPE_PCI_DEVICE,
+    .instance_size = sizeof(PCIProxyDev),
+    .class_init    = pci_proxy_dev_class_init,
+    .interfaces = (InterfaceInfo[]) {
+        { INTERFACE_CONVENTIONAL_PCI_DEVICE },
+        { },
+    },
+};
+
+static void pci_proxy_dev_register_types(void)
+{
+    type_register_static(&pci_proxy_dev_type_info);
+}
+
+type_init(pci_proxy_dev_register_types)
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/meson.build
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@ remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('machine.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('mpqemu-link.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('message.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('remote-obj.c'))
+remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('proxy.c'))
 
 specific_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('memory.c'))
 
-- 
2.29.2

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: d54edb4176361eed86b903e8f27058363b6c83b3.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/hw/remote/mpqemu-link.h |  4 ++++
 hw/remote/mpqemu-link.c         | 34 +++++++++++++++++++++++++++++++++
 2 files changed, 38 insertions(+)

diff --git a/include/hw/remote/mpqemu-link.h b/include/hw/remote/mpqemu-link.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/mpqemu-link.h
+++ b/include/hw/remote/mpqemu-link.h
@@ -XXX,XX +XXX,XX @@
 #include "qemu/thread.h"
 #include "io/channel.h"
 #include "exec/hwaddr.h"
+#include "io/channel-socket.h"
+#include "hw/remote/proxy.h"
 
 #define REMOTE_MAX_FDS 8
 
@@ -XXX,XX +XXX,XX @@ typedef struct {
 bool mpqemu_msg_send(MPQemuMsg *msg, QIOChannel *ioc, Error **errp);
 bool mpqemu_msg_recv(MPQemuMsg *msg, QIOChannel *ioc, Error **errp);
 
+uint64_t mpqemu_msg_send_and_await_reply(MPQemuMsg *msg, PCIProxyDev *pdev,
+                                         Error **errp);
 bool mpqemu_msg_valid(MPQemuMsg *msg);
 
 #endif
diff --git a/hw/remote/mpqemu-link.c b/hw/remote/mpqemu-link.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/mpqemu-link.c
+++ b/hw/remote/mpqemu-link.c
@@ -XXX,XX +XXX,XX @@ fail:
     return ret;
 }
 
+/*
+ * Send msg and wait for a reply with command code RET_MSG.
+ * Returns the message received of size u64 or UINT64_MAX
+ * on error.
+ * Called from VCPU thread in non-coroutine context.
+ * Used by the Proxy object to communicate to remote processes.
+ */
+uint64_t mpqemu_msg_send_and_await_reply(MPQemuMsg *msg, PCIProxyDev *pdev,
+                                         Error **errp)
+{
+    ERRP_GUARD();
+    MPQemuMsg msg_reply = {0};
+    uint64_t ret = UINT64_MAX;
+
+    assert(!qemu_in_coroutine());
+
+    QEMU_LOCK_GUARD(&pdev->io_mutex);
+    if (!mpqemu_msg_send(msg, pdev->ioc, errp)) {
+        return ret;
+    }
+
+    if (!mpqemu_msg_recv(&msg_reply, pdev->ioc, errp)) {
+        return ret;
+    }
+
+    if (!mpqemu_msg_valid(&msg_reply)) {
+        error_setg(errp, "ERROR: Invalid reply received for command %d",
+                         msg->cmd);
+        return ret;
+    }
+
+    return msg_reply.data.u64;
+}
+
 bool mpqemu_msg_valid(MPQemuMsg *msg)
 {
     if (msg->cmd >= MPQEMU_CMD_MAX && msg->cmd < 0) {
-- 
2.29.2

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

The Proxy Object sends the PCI config space accesses as messages
to the remote process over the communication channel

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: d3c94f4618813234655356c60e6f0d0362ff42d6.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/hw/remote/mpqemu-link.h | 10 ++++++
 hw/remote/message.c             | 60 +++++++++++++++++++++++++++++++++
 hw/remote/mpqemu-link.c         |  8 ++++-
 hw/remote/proxy.c               | 55 ++++++++++++++++++++++++++++++
 4 files changed, 132 insertions(+), 1 deletion(-)

diff --git a/include/hw/remote/mpqemu-link.h b/include/hw/remote/mpqemu-link.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/mpqemu-link.h
+++ b/include/hw/remote/mpqemu-link.h
@@ -XXX,XX +XXX,XX @@
  */
 typedef enum {
     MPQEMU_CMD_SYNC_SYSMEM,
+    MPQEMU_CMD_RET,
+    MPQEMU_CMD_PCI_CFGWRITE,
+    MPQEMU_CMD_PCI_CFGREAD,
     MPQEMU_CMD_MAX,
 } MPQemuCmd;
 
@@ -XXX,XX +XXX,XX @@ typedef struct {
     off_t offsets[REMOTE_MAX_FDS];
 } SyncSysmemMsg;
 
+typedef struct {
+    uint32_t addr;
+    uint32_t val;
+    int len;
+} PciConfDataMsg;
+
 /**
  * MPQemuMsg:
  * @cmd: The remote command
@@ -XXX,XX +XXX,XX @@ typedef struct {
 
     union {
         uint64_t u64;
+        PciConfDataMsg pci_conf_data;
         SyncSysmemMsg sync_sysmem;
     } data;
 
diff --git a/hw/remote/message.c b/hw/remote/message.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/message.c
+++ b/hw/remote/message.c
@@ -XXX,XX +XXX,XX @@
 #include "hw/remote/mpqemu-link.h"
 #include "qapi/error.h"
 #include "sysemu/runstate.h"
+#include "hw/pci/pci.h"
+
+static void process_config_write(QIOChannel *ioc, PCIDevice *dev,
+                                 MPQemuMsg *msg, Error **errp);
+static void process_config_read(QIOChannel *ioc, PCIDevice *dev,
+                                MPQemuMsg *msg, Error **errp);
 
 void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
 {
@@ -XXX,XX +XXX,XX @@ void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
         }
 
         switch (msg.cmd) {
+        case MPQEMU_CMD_PCI_CFGWRITE:
+            process_config_write(com->ioc, pci_dev, &msg, &local_err);
+            break;
+        case MPQEMU_CMD_PCI_CFGREAD:
+            process_config_read(com->ioc, pci_dev, &msg, &local_err);
+            break;
         default:
             error_setg(&local_err,
                        "Unknown command (%d) received for device %s"
@@ -XXX,XX +XXX,XX @@ void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
         qemu_system_shutdown_request(SHUTDOWN_CAUSE_GUEST_SHUTDOWN);
     }
 }
+
+static void process_config_write(QIOChannel *ioc, PCIDevice *dev,
+                                 MPQemuMsg *msg, Error **errp)
+{
+    ERRP_GUARD();
+    PciConfDataMsg *conf = (PciConfDataMsg *)&msg->data.pci_conf_data;
+    MPQemuMsg ret = { 0 };
+
+    if ((conf->addr + sizeof(conf->val)) > pci_config_size(dev)) {
+        error_setg(errp, "Bad address for PCI config write, pid "FMT_pid".",
+                   getpid());
+        ret.data.u64 = UINT64_MAX;
+    } else {
+        pci_default_write_config(dev, conf->addr, conf->val, conf->len);
+    }
+
+    ret.cmd = MPQEMU_CMD_RET;
+    ret.size = sizeof(ret.data.u64);
+
+    if (!mpqemu_msg_send(&ret, ioc, NULL)) {
+        error_prepend(errp, "Error returning code to proxy, pid "FMT_pid": ",
+                      getpid());
+    }
+}
+
+static void process_config_read(QIOChannel *ioc, PCIDevice *dev,
+                                MPQemuMsg *msg, Error **errp)
+{
+    ERRP_GUARD();
+    PciConfDataMsg *conf = (PciConfDataMsg *)&msg->data.pci_conf_data;
+    MPQemuMsg ret = { 0 };
+
+    if ((conf->addr + sizeof(conf->val)) > pci_config_size(dev)) {
+        error_setg(errp, "Bad address for PCI config read, pid "FMT_pid".",
+                   getpid());
+        ret.data.u64 = UINT64_MAX;
+    } else {
+        ret.data.u64 = pci_default_read_config(dev, conf->addr, conf->len);
+    }
+
+    ret.cmd = MPQEMU_CMD_RET;
+    ret.size = sizeof(ret.data.u64);
+
+    if (!mpqemu_msg_send(&ret, ioc, NULL)) {
+        error_prepend(errp, "Error returning code to proxy, pid "FMT_pid": ",
+                      getpid());
+    }
+}
diff --git a/hw/remote/mpqemu-link.c b/hw/remote/mpqemu-link.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/mpqemu-link.c
+++ b/hw/remote/mpqemu-link.c
@@ -XXX,XX +XXX,XX @@ uint64_t mpqemu_msg_send_and_await_reply(MPQemuMsg *msg, PCIProxyDev *pdev,
         return ret;
     }
 
-    if (!mpqemu_msg_valid(&msg_reply)) {
+    if (!mpqemu_msg_valid(&msg_reply) || msg_reply.cmd != MPQEMU_CMD_RET) {
         error_setg(errp, "ERROR: Invalid reply received for command %d",
                          msg->cmd);
         return ret;
@@ -XXX,XX +XXX,XX @@ bool mpqemu_msg_valid(MPQemuMsg *msg)
             return false;
         }
         break;
+    case MPQEMU_CMD_PCI_CFGWRITE:
+    case MPQEMU_CMD_PCI_CFGREAD:
+        if (msg->size != sizeof(PciConfDataMsg)) {
+            return false;
+        }
+        break;
     default:
         break;
     }
diff --git a/hw/remote/proxy.c b/hw/remote/proxy.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/proxy.c
+++ b/hw/remote/proxy.c
@@ -XXX,XX +XXX,XX @@
 #include "monitor/monitor.h"
 #include "migration/blocker.h"
 #include "qemu/sockets.h"
+#include "hw/remote/mpqemu-link.h"
+#include "qemu/error-report.h"
 
 static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
 {
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_exit(PCIDevice *pdev)
     error_free(dev->migration_blocker);
 }
 
+static void config_op_send(PCIProxyDev *pdev, uint32_t addr, uint32_t *val,
+                           int len, unsigned int op)
+{
+    MPQemuMsg msg = { 0 };
+    uint64_t ret = -EINVAL;
+    Error *local_err = NULL;
+
+    msg.cmd = op;
+    msg.data.pci_conf_data.addr = addr;
+    msg.data.pci_conf_data.val = (op == MPQEMU_CMD_PCI_CFGWRITE) ? *val : 0;
+    msg.data.pci_conf_data.len = len;
+    msg.size = sizeof(PciConfDataMsg);
+
+    ret = mpqemu_msg_send_and_await_reply(&msg, pdev, &local_err);
+    if (local_err) {
+        error_report_err(local_err);
+    }
+
+    if (ret == UINT64_MAX) {
+        error_report("Failed to perform PCI config %s operation",
+                     (op == MPQEMU_CMD_PCI_CFGREAD) ? "READ" : "WRITE");
+    }
+
+    if (op == MPQEMU_CMD_PCI_CFGREAD) {
+        *val = (uint32_t)ret;
+    }
+}
+
+static uint32_t pci_proxy_read_config(PCIDevice *d, uint32_t addr, int len)
+{
+    uint32_t val;
+
+    config_op_send(PCI_PROXY_DEV(d), addr, &val, len, MPQEMU_CMD_PCI_CFGREAD);
+
+    return val;
+}
+
+static void pci_proxy_write_config(PCIDevice *d, uint32_t addr, uint32_t val,
+                                   int len)
+{
+    /*
+     * Some of the functions access the copy of remote device's PCI config
+     * space which is cached in the proxy device. Therefore, maintain
+     * it updated.
+     */
+    pci_default_write_config(d, addr, val, len);
+
+    config_op_send(PCI_PROXY_DEV(d), addr, &val, len, MPQEMU_CMD_PCI_CFGWRITE);
+}
+
 static Property proxy_properties[] = {
     DEFINE_PROP_STRING("fd", PCIProxyDev, fd),
     DEFINE_PROP_END_OF_LIST(),
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_class_init(ObjectClass *klass, void *data)
 
     k->realize = pci_proxy_dev_realize;
     k->exit = pci_proxy_dev_exit;
+    k->config_read = pci_proxy_read_config;
+    k->config_write = pci_proxy_write_config;
+
     device_class_set_props(dc, proxy_properties);
 }
 
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

Proxy device object implements handler for PCI BAR writes and reads.
The handler uses BAR_WRITE/BAR_READ message to communicate to the
remote process with the BAR address and value to be written/read.
The remote process implements handler for BAR_WRITE/BAR_READ
message.

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: a8b76714a9688be5552c4c92d089bc9e8a4707ff.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/hw/remote/mpqemu-link.h | 10 ++++
 include/hw/remote/proxy.h       |  9 ++++
 hw/remote/message.c             | 83 +++++++++++++++++++++++++++++++++
 hw/remote/mpqemu-link.c         |  6 +++
 hw/remote/proxy.c               | 60 ++++++++++++++++++++++++
 5 files changed, 168 insertions(+)

diff --git a/include/hw/remote/mpqemu-link.h b/include/hw/remote/mpqemu-link.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/mpqemu-link.h
+++ b/include/hw/remote/mpqemu-link.h
@@ -XXX,XX +XXX,XX @@ typedef enum {
     MPQEMU_CMD_RET,
     MPQEMU_CMD_PCI_CFGWRITE,
     MPQEMU_CMD_PCI_CFGREAD,
+    MPQEMU_CMD_BAR_WRITE,
+    MPQEMU_CMD_BAR_READ,
     MPQEMU_CMD_MAX,
 } MPQemuCmd;
 
@@ -XXX,XX +XXX,XX @@ typedef struct {
     int len;
 } PciConfDataMsg;
 
+typedef struct {
+    hwaddr addr;
+    uint64_t val;
+    unsigned size;
+    bool memory;
+} BarAccessMsg;
+
 /**
  * MPQemuMsg:
  * @cmd: The remote command
@@ -XXX,XX +XXX,XX @@ typedef struct {
         uint64_t u64;
         PciConfDataMsg pci_conf_data;
         SyncSysmemMsg sync_sysmem;
+        BarAccessMsg bar_access;
     } data;
 
     int fds[REMOTE_MAX_FDS];
diff --git a/include/hw/remote/proxy.h b/include/hw/remote/proxy.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/proxy.h
+++ b/include/hw/remote/proxy.h
@@ -XXX,XX +XXX,XX @@
 #define TYPE_PCI_PROXY_DEV "x-pci-proxy-dev"
 OBJECT_DECLARE_SIMPLE_TYPE(PCIProxyDev, PCI_PROXY_DEV)
 
+typedef struct ProxyMemoryRegion {
+    PCIProxyDev *dev;
+    MemoryRegion mr;
+    bool memory;
+    bool present;
+    uint8_t type;
+} ProxyMemoryRegion;
+
 struct PCIProxyDev {
     PCIDevice parent_dev;
     char *fd;
@@ -XXX,XX +XXX,XX @@ struct PCIProxyDev {
     QemuMutex io_mutex;
     QIOChannel *ioc;
     Error *migration_blocker;
+    ProxyMemoryRegion region[PCI_NUM_REGIONS];
 };
 
 #endif /* PROXY_H */
diff --git a/hw/remote/message.c b/hw/remote/message.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/message.c
+++ b/hw/remote/message.c
@@ -XXX,XX +XXX,XX @@
 #include "qapi/error.h"
 #include "sysemu/runstate.h"
 #include "hw/pci/pci.h"
+#include "exec/memattrs.h"
 
 static void process_config_write(QIOChannel *ioc, PCIDevice *dev,
                                  MPQemuMsg *msg, Error **errp);
 static void process_config_read(QIOChannel *ioc, PCIDevice *dev,
                                 MPQemuMsg *msg, Error **errp);
+static void process_bar_write(QIOChannel *ioc, MPQemuMsg *msg, Error **errp);
+static void process_bar_read(QIOChannel *ioc, MPQemuMsg *msg, Error **errp);
 
 void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
 {
@@ -XXX,XX +XXX,XX @@ void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
         case MPQEMU_CMD_PCI_CFGREAD:
             process_config_read(com->ioc, pci_dev, &msg, &local_err);
             break;
+        case MPQEMU_CMD_BAR_WRITE:
+            process_bar_write(com->ioc, &msg, &local_err);
+            break;
+        case MPQEMU_CMD_BAR_READ:
+            process_bar_read(com->ioc, &msg, &local_err);
+            break;
         default:
             error_setg(&local_err,
                        "Unknown command (%d) received for device %s"
@@ -XXX,XX +XXX,XX @@ static void process_config_read(QIOChannel *ioc, PCIDevice *dev,
                       getpid());
     }
 }
+
+static void process_bar_write(QIOChannel *ioc, MPQemuMsg *msg, Error **errp)
+{
+    ERRP_GUARD();
+    BarAccessMsg *bar_access = &msg->data.bar_access;
+    AddressSpace *as =
+        bar_access->memory ? &address_space_memory : &address_space_io;
+    MPQemuMsg ret = { 0 };
+    MemTxResult res;
+    uint64_t val;
+
+    if (!is_power_of_2(bar_access->size) ||
+       (bar_access->size > sizeof(uint64_t))) {
+        ret.data.u64 = UINT64_MAX;
+        goto fail;
+    }
+
+    val = cpu_to_le64(bar_access->val);
+
+    res = address_space_rw(as, bar_access->addr, MEMTXATTRS_UNSPECIFIED,
+                           (void *)&val, bar_access->size, true);
+
+    if (res != MEMTX_OK) {
+        error_setg(errp, "Bad address %"PRIx64" for mem write, pid "FMT_pid".",
+                   bar_access->addr, getpid());
+        ret.data.u64 = -1;
+    }
+
+fail:
+    ret.cmd = MPQEMU_CMD_RET;
+    ret.size = sizeof(ret.data.u64);
+
+    if (!mpqemu_msg_send(&ret, ioc, NULL)) {
+        error_prepend(errp, "Error returning code to proxy, pid "FMT_pid": ",
+                      getpid());
+    }
+}
+
+static void process_bar_read(QIOChannel *ioc, MPQemuMsg *msg, Error **errp)
+{
+    ERRP_GUARD();
+    BarAccessMsg *bar_access = &msg->data.bar_access;
+    MPQemuMsg ret = { 0 };
+    AddressSpace *as;
+    MemTxResult res;
+    uint64_t val = 0;
+
+    as = bar_access->memory ? &address_space_memory : &address_space_io;
+
+    if (!is_power_of_2(bar_access->size) ||
+       (bar_access->size > sizeof(uint64_t))) {
+        val = UINT64_MAX;
+        goto fail;
+    }
+
+    res = address_space_rw(as, bar_access->addr, MEMTXATTRS_UNSPECIFIED,
+                           (void *)&val, bar_access->size, false);
+
+    if (res != MEMTX_OK) {
+        error_setg(errp, "Bad address %"PRIx64" for mem read, pid "FMT_pid".",
+                   bar_access->addr, getpid());
+        val = UINT64_MAX;
+    }
+
+fail:
+    ret.cmd = MPQEMU_CMD_RET;
+    ret.data.u64 = le64_to_cpu(val);
+    ret.size = sizeof(ret.data.u64);
+
+    if (!mpqemu_msg_send(&ret, ioc, NULL)) {
+        error_prepend(errp, "Error returning code to proxy, pid "FMT_pid": ",
+                      getpid());
+    }
+}
diff --git a/hw/remote/mpqemu-link.c b/hw/remote/mpqemu-link.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/mpqemu-link.c
+++ b/hw/remote/mpqemu-link.c
@@ -XXX,XX +XXX,XX @@ bool mpqemu_msg_valid(MPQemuMsg *msg)
             return false;
         }
         break;
+    case MPQEMU_CMD_BAR_WRITE:
+    case MPQEMU_CMD_BAR_READ:
+        if ((msg->size != sizeof(BarAccessMsg)) || (msg->num_fds != 0)) {
+            return false;
+        }
+        break;
     default:
         break;
     }
diff --git a/hw/remote/proxy.c b/hw/remote/proxy.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/proxy.c
+++ b/hw/remote/proxy.c
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_register_types(void)
 }
 
 type_init(pci_proxy_dev_register_types)
+
+static void send_bar_access_msg(PCIProxyDev *pdev, MemoryRegion *mr,
+                                bool write, hwaddr addr, uint64_t *val,
+                                unsigned size, bool memory)
+{
+    MPQemuMsg msg = { 0 };
+    long ret = -EINVAL;
+    Error *local_err = NULL;
+
+    msg.size = sizeof(BarAccessMsg);
+    msg.data.bar_access.addr = mr->addr + addr;
+    msg.data.bar_access.size = size;
+    msg.data.bar_access.memory = memory;
+
+    if (write) {
+        msg.cmd = MPQEMU_CMD_BAR_WRITE;
+        msg.data.bar_access.val = *val;
+    } else {
+        msg.cmd = MPQEMU_CMD_BAR_READ;
+    }
+
+    ret = mpqemu_msg_send_and_await_reply(&msg, pdev, &local_err);
+    if (local_err) {
+        error_report_err(local_err);
+    }
+
+    if (!write) {
+        *val = ret;
+    }
+}
+
+static void proxy_bar_write(void *opaque, hwaddr addr, uint64_t val,
+                            unsigned size)
+{
+    ProxyMemoryRegion *pmr = opaque;
+
+    send_bar_access_msg(pmr->dev, &pmr->mr, true, addr, &val, size,
+                        pmr->memory);
+}
+
+static uint64_t proxy_bar_read(void *opaque, hwaddr addr, unsigned size)
+{
+    ProxyMemoryRegion *pmr = opaque;
+    uint64_t val;
+
+    send_bar_access_msg(pmr->dev, &pmr->mr, false, addr, &val, size,
+                        pmr->memory);
+
+    return val;
+}
+
+const MemoryRegionOps proxy_mr_ops = {
+    .read = proxy_bar_read,
+    .write = proxy_bar_write,
+    .endianness = DEVICE_NATIVE_ENDIAN,
+    .impl = {
+        .min_access_size = 1,
+        .max_access_size = 8,
+    },
+};
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

Add ProxyMemoryListener object which is used to keep the view of the RAM
in sync between QEMU and remote process.
A MemoryListener is registered for system-memory AddressSpace. The
listener sends SYNC_SYSMEM message to the remote process when memory
listener commits the changes to memory, the remote process receives
the message and processes it in the handler for SYNC_SYSMEM message.

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 04fe4e6a9ca90d4f11ab6f59be7652f5b086a071.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                               |   2 +
 include/hw/remote/proxy-memory-listener.h |  28 +++
 include/hw/remote/proxy.h                 |   2 +
 hw/remote/message.c                       |   4 +
 hw/remote/proxy-memory-listener.c         | 227 ++++++++++++++++++++++
 hw/remote/proxy.c                         |   6 +
 hw/remote/meson.build                     |   1 +
 7 files changed, 270 insertions(+)
 create mode 100644 include/hw/remote/proxy-memory-listener.h
 create mode 100644 hw/remote/proxy-memory-listener.c

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: include/hw/remote/memory.h
 F: hw/remote/memory.c
 F: hw/remote/proxy.c
 F: include/hw/remote/proxy.h
+F: hw/remote/proxy-memory-listener.c
+F: include/hw/remote/proxy-memory-listener.h
 
 Build and test automation
 -------------------------
diff --git a/include/hw/remote/proxy-memory-listener.h b/include/hw/remote/proxy-memory-listener.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/include/hw/remote/proxy-memory-listener.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef PROXY_MEMORY_LISTENER_H
+#define PROXY_MEMORY_LISTENER_H
+
+#include "exec/memory.h"
+#include "io/channel.h"
+
+typedef struct ProxyMemoryListener {
+    MemoryListener listener;
+
+    int n_mr_sections;
+    MemoryRegionSection *mr_sections;
+
+    QIOChannel *ioc;
+} ProxyMemoryListener;
+
+void proxy_memory_listener_configure(ProxyMemoryListener *proxy_listener,
+                                     QIOChannel *ioc);
+void proxy_memory_listener_deconfigure(ProxyMemoryListener *proxy_listener);
+
+#endif
diff --git a/include/hw/remote/proxy.h b/include/hw/remote/proxy.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/proxy.h
+++ b/include/hw/remote/proxy.h
@@ -XXX,XX +XXX,XX @@
 
 #include "hw/pci/pci.h"
 #include "io/channel.h"
+#include "hw/remote/proxy-memory-listener.h"
 
 #define TYPE_PCI_PROXY_DEV "x-pci-proxy-dev"
 OBJECT_DECLARE_SIMPLE_TYPE(PCIProxyDev, PCI_PROXY_DEV)
@@ -XXX,XX +XXX,XX @@ struct PCIProxyDev {
     QemuMutex io_mutex;
     QIOChannel *ioc;
     Error *migration_blocker;
+    ProxyMemoryListener proxy_listener;
     ProxyMemoryRegion region[PCI_NUM_REGIONS];
 };
 
diff --git a/hw/remote/message.c b/hw/remote/message.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/message.c
+++ b/hw/remote/message.c
@@ -XXX,XX +XXX,XX @@
 #include "sysemu/runstate.h"
 #include "hw/pci/pci.h"
 #include "exec/memattrs.h"
+#include "hw/remote/memory.h"
 
 static void process_config_write(QIOChannel *ioc, PCIDevice *dev,
                                  MPQemuMsg *msg, Error **errp);
@@ -XXX,XX +XXX,XX @@ void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
         case MPQEMU_CMD_BAR_READ:
             process_bar_read(com->ioc, &msg, &local_err);
             break;
+        case MPQEMU_CMD_SYNC_SYSMEM:
+            remote_sysmem_reconfig(&msg, &local_err);
+            break;
         default:
             error_setg(&local_err,
                        "Unknown command (%d) received for device %s"
diff --git a/hw/remote/proxy-memory-listener.c b/hw/remote/proxy-memory-listener.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/proxy-memory-listener.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "qemu/compiler.h"
+#include "qemu/int128.h"
+#include "qemu/range.h"
+#include "exec/memory.h"
+#include "exec/cpu-common.h"
+#include "cpu.h"
+#include "exec/ram_addr.h"
+#include "exec/address-spaces.h"
+#include "qapi/error.h"
+#include "hw/remote/mpqemu-link.h"
+#include "hw/remote/proxy-memory-listener.h"
+
+/*
+ * TODO: get_fd_from_hostaddr(), proxy_mrs_can_merge() and
+ * proxy_memory_listener_commit() defined below perform tasks similar to the
+ * functions defined in vhost-user.c. These functions are good candidates
+ * for refactoring.
+ *
+ */
+
+static void proxy_memory_listener_reset(MemoryListener *listener)
+{
+    ProxyMemoryListener *proxy_listener = container_of(listener,
+                                                       ProxyMemoryListener,
+                                                       listener);
+    int mrs;
+
+    for (mrs = 0; mrs < proxy_listener->n_mr_sections; mrs++) {
+        memory_region_unref(proxy_listener->mr_sections[mrs].mr);
+    }
+
+    g_free(proxy_listener->mr_sections);
+    proxy_listener->mr_sections = NULL;
+    proxy_listener->n_mr_sections = 0;
+}
+
+static int get_fd_from_hostaddr(uint64_t host, ram_addr_t *offset)
+{
+    MemoryRegion *mr;
+    ram_addr_t off;
+
+    /**
+     * Assumes that the host address is a valid address as it's
+     * coming from the MemoryListener system. In the case host
+     * address is not valid, the following call would return
+     * the default subregion of "system_memory" region, and
+     * not NULL. So it's not possible to check for NULL here.
+     */
+    mr = memory_region_from_host((void *)(uintptr_t)host, &off);
+
+    if (offset) {
+        *offset = off;
+    }
+
+    return memory_region_get_fd(mr);
+}
+
+static bool proxy_mrs_can_merge(uint64_t host, uint64_t prev_host, size_t size)
+{
+    if (((prev_host + size) != host)) {
+        return false;
+    }
+
+    if (get_fd_from_hostaddr(host, NULL) !=
+            get_fd_from_hostaddr(prev_host, NULL)) {
+        return false;
+    }
+
+    return true;
+}
+
+static bool try_merge(ProxyMemoryListener *proxy_listener,
+                      MemoryRegionSection *section)
+{
+    uint64_t mrs_size, mrs_gpa, mrs_page;
+    MemoryRegionSection *prev_sec;
+    bool merged = false;
+    uintptr_t mrs_host;
+    RAMBlock *mrs_rb;
+
+    if (!proxy_listener->n_mr_sections) {
+        return false;
+    }
+
+    mrs_rb = section->mr->ram_block;
+    mrs_page = (uint64_t)qemu_ram_pagesize(mrs_rb);
+    mrs_size = int128_get64(section->size);
+    mrs_gpa = section->offset_within_address_space;
+    mrs_host = (uintptr_t)memory_region_get_ram_ptr(section->mr) +
+               section->offset_within_region;
+
+    if (get_fd_from_hostaddr(mrs_host, NULL) < 0) {
+        return true;
+    }
+
+    mrs_host = mrs_host & ~(mrs_page - 1);
+    mrs_gpa = mrs_gpa & ~(mrs_page - 1);
+    mrs_size = ROUND_UP(mrs_size, mrs_page);
+
+    prev_sec = proxy_listener->mr_sections +
+               (proxy_listener->n_mr_sections - 1);
+    uint64_t prev_gpa_start = prev_sec->offset_within_address_space;
+    uint64_t prev_size = int128_get64(prev_sec->size);
+    uint64_t prev_gpa_end   = range_get_last(prev_gpa_start, prev_size);
+    uint64_t prev_host_start =
+        (uintptr_t)memory_region_get_ram_ptr(prev_sec->mr) +
+        prev_sec->offset_within_region;
+    uint64_t prev_host_end = range_get_last(prev_host_start, prev_size);
+
+    if (mrs_gpa <= (prev_gpa_end + 1)) {
+        g_assert(mrs_gpa > prev_gpa_start);
+
+        if ((section->mr == prev_sec->mr) &&
+            proxy_mrs_can_merge(mrs_host, prev_host_start,
+                                (mrs_gpa - prev_gpa_start))) {
+            uint64_t max_end = MAX(prev_host_end, mrs_host + mrs_size);
+            merged = true;
+            prev_sec->offset_within_address_space =
+                MIN(prev_gpa_start, mrs_gpa);
+            prev_sec->offset_within_region =
+                MIN(prev_host_start, mrs_host) -
+                (uintptr_t)memory_region_get_ram_ptr(prev_sec->mr);
+            prev_sec->size = int128_make64(max_end - MIN(prev_host_start,
+                                                         mrs_host));
+        }
+    }
+
+    return merged;
+}
+
+static void proxy_memory_listener_region_addnop(MemoryListener *listener,
+                                                MemoryRegionSection *section)
+{
+    ProxyMemoryListener *proxy_listener = container_of(listener,
+                                                       ProxyMemoryListener,
+                                                       listener);
+
+    if (!memory_region_is_ram(section->mr) ||
+            memory_region_is_rom(section->mr)) {
+        return;
+    }
+
+    if (try_merge(proxy_listener, section)) {
+        return;
+    }
+
+    ++proxy_listener->n_mr_sections;
+    proxy_listener->mr_sections = g_renew(MemoryRegionSection,
+                                          proxy_listener->mr_sections,
+                                          proxy_listener->n_mr_sections);
+    proxy_listener->mr_sections[proxy_listener->n_mr_sections - 1] = *section;
+    proxy_listener->mr_sections[proxy_listener->n_mr_sections - 1].fv = NULL;
+    memory_region_ref(section->mr);
+}
+
+static void proxy_memory_listener_commit(MemoryListener *listener)
+{
+    ProxyMemoryListener *proxy_listener = container_of(listener,
+                                                       ProxyMemoryListener,
+                                                       listener);
+    MPQemuMsg msg;
+    MemoryRegionSection *section;
+    ram_addr_t offset;
+    uintptr_t host_addr;
+    int region;
+    Error *local_err = NULL;
+
+    memset(&msg, 0, sizeof(MPQemuMsg));
+
+    msg.cmd = MPQEMU_CMD_SYNC_SYSMEM;
+    msg.num_fds = proxy_listener->n_mr_sections;
+    msg.size = sizeof(SyncSysmemMsg);
+    if (msg.num_fds > REMOTE_MAX_FDS) {
+        error_report("Number of fds is more than %d", REMOTE_MAX_FDS);
+        return;
+    }
+
+    for (region = 0; region < proxy_listener->n_mr_sections; region++) {
+        section = &proxy_listener->mr_sections[region];
+        msg.data.sync_sysmem.gpas[region] =
+            section->offset_within_address_space;
+        msg.data.sync_sysmem.sizes[region] = int128_get64(section->size);
+        host_addr = (uintptr_t)memory_region_get_ram_ptr(section->mr) +
+                    section->offset_within_region;
+        msg.fds[region] = get_fd_from_hostaddr(host_addr, &offset);
+        msg.data.sync_sysmem.offsets[region] = offset;
+    }
+    if (!mpqemu_msg_send(&msg, proxy_listener->ioc, &local_err)) {
+        error_report_err(local_err);
+    }
+}
+
+void proxy_memory_listener_deconfigure(ProxyMemoryListener *proxy_listener)
+{
+    memory_listener_unregister(&proxy_listener->listener);
+
+    proxy_memory_listener_reset(&proxy_listener->listener);
+}
+
+void proxy_memory_listener_configure(ProxyMemoryListener *proxy_listener,
+                                     QIOChannel *ioc)
+{
+    proxy_listener->n_mr_sections = 0;
+    proxy_listener->mr_sections = NULL;
+
+    proxy_listener->ioc = ioc;
+
+    proxy_listener->listener.begin = proxy_memory_listener_reset;
+    proxy_listener->listener.commit = proxy_memory_listener_commit;
+    proxy_listener->listener.region_add = proxy_memory_listener_region_addnop;
+    proxy_listener->listener.region_nop = proxy_memory_listener_region_addnop;
+    proxy_listener->listener.priority = 10;
+
+    memory_listener_register(&proxy_listener->listener,
+                             &address_space_memory);
+}
diff --git a/hw/remote/proxy.c b/hw/remote/proxy.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/proxy.c
+++ b/hw/remote/proxy.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/sockets.h"
 #include "hw/remote/mpqemu-link.h"
 #include "qemu/error-report.h"
+#include "hw/remote/proxy-memory-listener.h"
+#include "qom/object.h"
 
 static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
 {
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
 
     qemu_mutex_init(&dev->io_mutex);
     qio_channel_set_blocking(dev->ioc, true, NULL);
+
+    proxy_memory_listener_configure(&dev->proxy_listener, dev->ioc);
 }
 
 static void pci_proxy_dev_exit(PCIDevice *pdev)
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_exit(PCIDevice *pdev)
     migrate_del_blocker(dev->migration_blocker);
 
     error_free(dev->migration_blocker);
+
+    proxy_memory_listener_deconfigure(&dev->proxy_listener);
 }
 
 static void config_op_send(PCIProxyDev *pdev, uint32_t addr, uint32_t *val,
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/meson.build
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@ remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('remote-obj.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('proxy.c'))
 
 specific_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('memory.c'))
+specific_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('proxy-memory-listener.c'))
 
 softmmu_ss.add_all(when: 'CONFIG_MULTIPROCESS', if_true: remote_ss)
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

IOHUB object is added to manage PCI IRQs. It uses KVM_IRQFD
ioctl to create irqfd to injecting PCI interrupts to the guest.
IOHUB object forwards the irqfd to the remote process. Remote process
uses this fd to directly send interrupts to the guest, bypassing QEMU.

Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 51d5c3d54e28a68b002e3875c59599c9f5a424a1.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 MAINTAINERS                     |   2 +
 include/hw/pci/pci_ids.h        |   3 +
 include/hw/remote/iohub.h       |  42 +++++++++++
 include/hw/remote/machine.h     |   2 +
 include/hw/remote/mpqemu-link.h |   1 +
 include/hw/remote/proxy.h       |   4 ++
 hw/remote/iohub.c               | 119 ++++++++++++++++++++++++++++++++
 hw/remote/machine.c             |  10 +++
 hw/remote/message.c             |   4 ++
 hw/remote/mpqemu-link.c         |   5 ++
 hw/remote/proxy.c               |  56 +++++++++++++++
 hw/remote/meson.build           |   1 +
 12 files changed, 249 insertions(+)
 create mode 100644 include/hw/remote/iohub.h
 create mode 100644 hw/remote/iohub.c

diff --git a/MAINTAINERS b/MAINTAINERS
index XXXXXXX..XXXXXXX 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -XXX,XX +XXX,XX @@ F: hw/remote/proxy.c
 F: include/hw/remote/proxy.h
 F: hw/remote/proxy-memory-listener.c
 F: include/hw/remote/proxy-memory-listener.h
+F: hw/remote/iohub.c
+F: include/hw/remote/iohub.h
 
 Build and test automation
 -------------------------
diff --git a/include/hw/pci/pci_ids.h b/include/hw/pci/pci_ids.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/pci/pci_ids.h
+++ b/include/hw/pci/pci_ids.h
@@ -XXX,XX +XXX,XX @@
 #define PCI_DEVICE_ID_SUN_SIMBA          0x5000
 #define PCI_DEVICE_ID_SUN_SABRE          0xa000
 
+#define PCI_VENDOR_ID_ORACLE             0x108e
+#define PCI_DEVICE_ID_REMOTE_IOHUB       0xb000
+
 #define PCI_VENDOR_ID_CMD                0x1095
 #define PCI_DEVICE_ID_CMD_646            0x0646
 
diff --git a/include/hw/remote/iohub.h b/include/hw/remote/iohub.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/include/hw/remote/iohub.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * IO Hub for remote device
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef REMOTE_IOHUB_H
+#define REMOTE_IOHUB_H
+
+#include "hw/pci/pci.h"
+#include "qemu/event_notifier.h"
+#include "qemu/thread-posix.h"
+#include "hw/remote/mpqemu-link.h"
+
+#define REMOTE_IOHUB_NB_PIRQS    PCI_DEVFN_MAX
+
+typedef struct ResampleToken {
+    void *iohub;
+    int pirq;
+} ResampleToken;
+
+typedef struct RemoteIOHubState {
+    PCIDevice d;
+    EventNotifier irqfds[REMOTE_IOHUB_NB_PIRQS];
+    EventNotifier resamplefds[REMOTE_IOHUB_NB_PIRQS];
+    unsigned int irq_level[REMOTE_IOHUB_NB_PIRQS];
+    ResampleToken token[REMOTE_IOHUB_NB_PIRQS];
+    QemuMutex irq_level_lock[REMOTE_IOHUB_NB_PIRQS];
+} RemoteIOHubState;
+
+int remote_iohub_map_irq(PCIDevice *pci_dev, int intx);
+void remote_iohub_set_irq(void *opaque, int pirq, int level);
+void process_set_irqfd_msg(PCIDevice *pci_dev, MPQemuMsg *msg);
+
+void remote_iohub_init(RemoteIOHubState *iohub);
+void remote_iohub_finalize(RemoteIOHubState *iohub);
+
+#endif
diff --git a/include/hw/remote/machine.h b/include/hw/remote/machine.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/machine.h
+++ b/include/hw/remote/machine.h
@@ -XXX,XX +XXX,XX @@
 #include "hw/boards.h"
 #include "hw/pci-host/remote.h"
 #include "io/channel.h"
+#include "hw/remote/iohub.h"
 
 struct RemoteMachineState {
     MachineState parent_obj;
 
     RemotePCIHost *host;
+    RemoteIOHubState iohub;
 };
 
 /* Used to pass to co-routine device and ioc. */
diff --git a/include/hw/remote/mpqemu-link.h b/include/hw/remote/mpqemu-link.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/mpqemu-link.h
+++ b/include/hw/remote/mpqemu-link.h
@@ -XXX,XX +XXX,XX @@ typedef enum {
     MPQEMU_CMD_PCI_CFGREAD,
     MPQEMU_CMD_BAR_WRITE,
     MPQEMU_CMD_BAR_READ,
+    MPQEMU_CMD_SET_IRQFD,
     MPQEMU_CMD_MAX,
 } MPQemuCmd;
 
diff --git a/include/hw/remote/proxy.h b/include/hw/remote/proxy.h
index XXXXXXX..XXXXXXX 100644
--- a/include/hw/remote/proxy.h
+++ b/include/hw/remote/proxy.h
@@ -XXX,XX +XXX,XX @@
 #include "hw/pci/pci.h"
 #include "io/channel.h"
 #include "hw/remote/proxy-memory-listener.h"
+#include "qemu/event_notifier.h"
 
 #define TYPE_PCI_PROXY_DEV "x-pci-proxy-dev"
 OBJECT_DECLARE_SIMPLE_TYPE(PCIProxyDev, PCI_PROXY_DEV)
@@ -XXX,XX +XXX,XX @@ struct PCIProxyDev {
     QIOChannel *ioc;
     Error *migration_blocker;
     ProxyMemoryListener proxy_listener;
+    int virq;
+    EventNotifier intr;
+    EventNotifier resample;
     ProxyMemoryRegion region[PCI_NUM_REGIONS];
 };
 
diff --git a/hw/remote/iohub.c b/hw/remote/iohub.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/hw/remote/iohub.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Remote IO Hub
+ *
+ * Copyright © 2018, 2021 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+
+#include "hw/pci/pci.h"
+#include "hw/pci/pci_ids.h"
+#include "hw/pci/pci_bus.h"
+#include "qemu/thread.h"
+#include "hw/boards.h"
+#include "hw/remote/machine.h"
+#include "hw/remote/iohub.h"
+#include "qemu/main-loop.h"
+
+void remote_iohub_init(RemoteIOHubState *iohub)
+{
+    int pirq;
+
+    memset(&iohub->irqfds, 0, sizeof(iohub->irqfds));
+    memset(&iohub->resamplefds, 0, sizeof(iohub->resamplefds));
+
+    for (pirq = 0; pirq < REMOTE_IOHUB_NB_PIRQS; pirq++) {
+        qemu_mutex_init(&iohub->irq_level_lock[pirq]);
+        iohub->irq_level[pirq] = 0;
+        event_notifier_init_fd(&iohub->irqfds[pirq], -1);
+        event_notifier_init_fd(&iohub->resamplefds[pirq], -1);
+    }
+}
+
+void remote_iohub_finalize(RemoteIOHubState *iohub)
+{
+    int pirq;
+
+    for (pirq = 0; pirq < REMOTE_IOHUB_NB_PIRQS; pirq++) {
+        qemu_set_fd_handler(event_notifier_get_fd(&iohub->resamplefds[pirq]),
+                            NULL, NULL, NULL);
+        event_notifier_cleanup(&iohub->irqfds[pirq]);
+        event_notifier_cleanup(&iohub->resamplefds[pirq]);
+        qemu_mutex_destroy(&iohub->irq_level_lock[pirq]);
+    }
+}
+
+int remote_iohub_map_irq(PCIDevice *pci_dev, int intx)
+{
+    return pci_dev->devfn;
+}
+
+void remote_iohub_set_irq(void *opaque, int pirq, int level)
+{
+    RemoteIOHubState *iohub = opaque;
+
+    assert(pirq >= 0);
+    assert(pirq < PCI_DEVFN_MAX);
+
+    QEMU_LOCK_GUARD(&iohub->irq_level_lock[pirq]);
+
+    if (level) {
+        if (++iohub->irq_level[pirq] == 1) {
+            event_notifier_set(&iohub->irqfds[pirq]);
+        }
+    } else if (iohub->irq_level[pirq] > 0) {
+        iohub->irq_level[pirq]--;
+    }
+}
+
+static void intr_resample_handler(void *opaque)
+{
+    ResampleToken *token = opaque;
+    RemoteIOHubState *iohub = token->iohub;
+    int pirq, s;
+
+    pirq = token->pirq;
+
+    s = event_notifier_test_and_clear(&iohub->resamplefds[pirq]);
+
+    assert(s >= 0);
+
+    QEMU_LOCK_GUARD(&iohub->irq_level_lock[pirq]);
+
+    if (iohub->irq_level[pirq]) {
+        event_notifier_set(&iohub->irqfds[pirq]);
+    }
+}
+
+void process_set_irqfd_msg(PCIDevice *pci_dev, MPQemuMsg *msg)
+{
+    RemoteMachineState *machine = REMOTE_MACHINE(current_machine);
+    RemoteIOHubState *iohub = &machine->iohub;
+    int pirq, intx;
+
+    intx = pci_get_byte(pci_dev->config + PCI_INTERRUPT_PIN) - 1;
+
+    pirq = remote_iohub_map_irq(pci_dev, intx);
+
+    if (event_notifier_get_fd(&iohub->irqfds[pirq]) != -1) {
+        qemu_set_fd_handler(event_notifier_get_fd(&iohub->resamplefds[pirq]),
+                            NULL, NULL, NULL);
+        event_notifier_cleanup(&iohub->irqfds[pirq]);
+        event_notifier_cleanup(&iohub->resamplefds[pirq]);
+        memset(&iohub->token[pirq], 0, sizeof(ResampleToken));
+    }
+
+    event_notifier_init_fd(&iohub->irqfds[pirq], msg->fds[0]);
+    event_notifier_init_fd(&iohub->resamplefds[pirq], msg->fds[1]);
+
+    iohub->token[pirq].iohub = iohub;
+    iohub->token[pirq].pirq = pirq;
+
+    qemu_set_fd_handler(msg->fds[1], intr_resample_handler, NULL,
+                        &iohub->token[pirq]);
+}
diff --git a/hw/remote/machine.c b/hw/remote/machine.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/machine.c
+++ b/hw/remote/machine.c
@@ -XXX,XX +XXX,XX @@
 #include "exec/address-spaces.h"
 #include "exec/memory.h"
 #include "qapi/error.h"
+#include "hw/pci/pci_host.h"
+#include "hw/remote/iohub.h"
 
 static void remote_machine_init(MachineState *machine)
 {
     MemoryRegion *system_memory, *system_io, *pci_memory;
     RemoteMachineState *s = REMOTE_MACHINE(machine);
     RemotePCIHost *rem_host;
+    PCIHostState *pci_host;
 
     system_memory = get_system_memory();
     system_io = get_system_io();
@@ -XXX,XX +XXX,XX @@ static void remote_machine_init(MachineState *machine)
     memory_region_add_subregion_overlap(system_memory, 0x0, pci_memory, -1);
 
     qdev_realize(DEVICE(rem_host), sysbus_get_default(), &error_fatal);
+
+    pci_host = PCI_HOST_BRIDGE(rem_host);
+
+    remote_iohub_init(&s->iohub);
+
+    pci_bus_irqs(pci_host->bus, remote_iohub_set_irq, remote_iohub_map_irq,
+                 &s->iohub, REMOTE_IOHUB_NB_PIRQS);
 }
 
 static void remote_machine_class_init(ObjectClass *oc, void *data)
diff --git a/hw/remote/message.c b/hw/remote/message.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/message.c
+++ b/hw/remote/message.c
@@ -XXX,XX +XXX,XX @@
 #include "hw/pci/pci.h"
 #include "exec/memattrs.h"
 #include "hw/remote/memory.h"
+#include "hw/remote/iohub.h"
 
 static void process_config_write(QIOChannel *ioc, PCIDevice *dev,
                                  MPQemuMsg *msg, Error **errp);
@@ -XXX,XX +XXX,XX @@ void coroutine_fn mpqemu_remote_msg_loop_co(void *data)
         case MPQEMU_CMD_SYNC_SYSMEM:
             remote_sysmem_reconfig(&msg, &local_err);
             break;
+        case MPQEMU_CMD_SET_IRQFD:
+            process_set_irqfd_msg(pci_dev, &msg);
+            break;
         default:
             error_setg(&local_err,
                        "Unknown command (%d) received for device %s"
diff --git a/hw/remote/mpqemu-link.c b/hw/remote/mpqemu-link.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/mpqemu-link.c
+++ b/hw/remote/mpqemu-link.c
@@ -XXX,XX +XXX,XX @@ bool mpqemu_msg_valid(MPQemuMsg *msg)
             return false;
         }
         break;
+    case MPQEMU_CMD_SET_IRQFD:
+        if (msg->size || (msg->num_fds != 2)) {
+            return false;
+        }
+        break;
     default:
         break;
     }
diff --git a/hw/remote/proxy.c b/hw/remote/proxy.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/proxy.c
+++ b/hw/remote/proxy.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/error-report.h"
 #include "hw/remote/proxy-memory-listener.h"
 #include "qom/object.h"
+#include "qemu/event_notifier.h"
+#include "sysemu/kvm.h"
+#include "util/event_notifier-posix.c"
+
+static void proxy_intx_update(PCIDevice *pci_dev)
+{
+    PCIProxyDev *dev = PCI_PROXY_DEV(pci_dev);
+    PCIINTxRoute route;
+    int pin = pci_get_byte(pci_dev->config + PCI_INTERRUPT_PIN) - 1;
+
+    if (dev->virq != -1) {
+        kvm_irqchip_remove_irqfd_notifier_gsi(kvm_state, &dev->intr, dev->virq);
+        dev->virq = -1;
+    }
+
+    route = pci_device_route_intx_to_irq(pci_dev, pin);
+
+    dev->virq = route.irq;
+
+    if (dev->virq != -1) {
+        kvm_irqchip_add_irqfd_notifier_gsi(kvm_state, &dev->intr,
+                                           &dev->resample, dev->virq);
+    }
+}
+
+static void setup_irqfd(PCIProxyDev *dev)
+{
+    PCIDevice *pci_dev = PCI_DEVICE(dev);
+    MPQemuMsg msg;
+    Error *local_err = NULL;
+
+    event_notifier_init(&dev->intr, 0);
+    event_notifier_init(&dev->resample, 0);
+
+    memset(&msg, 0, sizeof(MPQemuMsg));
+    msg.cmd = MPQEMU_CMD_SET_IRQFD;
+    msg.num_fds = 2;
+    msg.fds[0] = event_notifier_get_fd(&dev->intr);
+    msg.fds[1] = event_notifier_get_fd(&dev->resample);
+    msg.size = 0;
+
+    if (!mpqemu_msg_send(&msg, dev->ioc, &local_err)) {
+        error_report_err(local_err);
+    }
+
+    dev->virq = -1;
+
+    proxy_intx_update(pci_dev);
+
+    pci_device_set_intx_routing_notifier(pci_dev, proxy_intx_update);
+}
 
 static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
 {
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
     qio_channel_set_blocking(dev->ioc, true, NULL);
 
     proxy_memory_listener_configure(&dev->proxy_listener, dev->ioc);
+
+    setup_irqfd(dev);
 }
 
 static void pci_proxy_dev_exit(PCIDevice *pdev)
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_exit(PCIDevice *pdev)
     error_free(dev->migration_blocker);
 
     proxy_memory_listener_deconfigure(&dev->proxy_listener);
+
+    event_notifier_cleanup(&dev->intr);
+    event_notifier_cleanup(&dev->resample);
 }
 
 static void config_op_send(PCIProxyDev *pdev, uint32_t addr, uint32_t *val,
diff --git a/hw/remote/meson.build b/hw/remote/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/meson.build
+++ b/hw/remote/meson.build
@@ -XXX,XX +XXX,XX @@ remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('mpqemu-link.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('message.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('remote-obj.c'))
 remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('proxy.c'))
+remote_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('iohub.c'))
 
 specific_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('memory.c'))
 specific_ss.add(when: 'CONFIG_MULTIPROCESS', if_true: files('proxy-memory-listener.c'))
-- 
2.29.2

From: Jagannathan Raman <jag.raman@oracle.com>

Retrieve PCI configuration info about the remote device and
configure the Proxy PCI object based on the returned information

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 85ee367bbb993aa23699b44cfedd83b4ea6d5221.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 hw/remote/proxy.c | 84 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 84 insertions(+)

diff --git a/hw/remote/proxy.c b/hw/remote/proxy.c
index XXXXXXX..XXXXXXX 100644
--- a/hw/remote/proxy.c
+++ b/hw/remote/proxy.c
@@ -XXX,XX +XXX,XX @@
 #include "sysemu/kvm.h"
 #include "util/event_notifier-posix.c"
 
+static void probe_pci_info(PCIDevice *dev, Error **errp);
+
 static void proxy_intx_update(PCIDevice *pci_dev)
 {
     PCIProxyDev *dev = PCI_PROXY_DEV(pci_dev);
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
 {
     ERRP_GUARD();
     PCIProxyDev *dev = PCI_PROXY_DEV(device);
+    uint8_t *pci_conf = device->config;
     int fd;
 
     if (!dev->fd) {
@@ -XXX,XX +XXX,XX @@ static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
     qemu_mutex_init(&dev->io_mutex);
     qio_channel_set_blocking(dev->ioc, true, NULL);
 
+    pci_conf[PCI_LATENCY_TIMER] = 0xff;
+    pci_conf[PCI_INTERRUPT_PIN] = 0x01;
+
     proxy_memory_listener_configure(&dev->proxy_listener, dev->ioc);
 
     setup_irqfd(dev);
+
+    probe_pci_info(PCI_DEVICE(dev), errp);
 }
 
 static void pci_proxy_dev_exit(PCIDevice *pdev)
@@ -XXX,XX +XXX,XX @@ const MemoryRegionOps proxy_mr_ops = {
         .max_access_size = 8,
     },
 };
+
+static void probe_pci_info(PCIDevice *dev, Error **errp)
+{
+    PCIDeviceClass *pc = PCI_DEVICE_GET_CLASS(dev);
+    uint32_t orig_val, new_val, base_class, val;
+    PCIProxyDev *pdev = PCI_PROXY_DEV(dev);
+    DeviceClass *dc = DEVICE_CLASS(pc);
+    uint8_t type;
+    int i, size;
+
+    config_op_send(pdev, PCI_VENDOR_ID, &val, 2, MPQEMU_CMD_PCI_CFGREAD);
+    pc->vendor_id = (uint16_t)val;
+
+    config_op_send(pdev, PCI_DEVICE_ID, &val, 2, MPQEMU_CMD_PCI_CFGREAD);
+    pc->device_id = (uint16_t)val;
+
+    config_op_send(pdev, PCI_CLASS_DEVICE, &val, 2, MPQEMU_CMD_PCI_CFGREAD);
+    pc->class_id = (uint16_t)val;
+
+    config_op_send(pdev, PCI_SUBSYSTEM_ID, &val, 2, MPQEMU_CMD_PCI_CFGREAD);
+    pc->subsystem_id = (uint16_t)val;
+
+    base_class = pc->class_id >> 4;
+    switch (base_class) {
+    case PCI_BASE_CLASS_BRIDGE:
+        set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
+        break;
+    case PCI_BASE_CLASS_STORAGE:
+        set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
+        break;
+    case PCI_BASE_CLASS_NETWORK:
+        set_bit(DEVICE_CATEGORY_NETWORK, dc->categories);
+        break;
+    case PCI_BASE_CLASS_INPUT:
+        set_bit(DEVICE_CATEGORY_INPUT, dc->categories);
+        break;
+    case PCI_BASE_CLASS_DISPLAY:
+        set_bit(DEVICE_CATEGORY_DISPLAY, dc->categories);
+        break;
+    case PCI_BASE_CLASS_PROCESSOR:
+        set_bit(DEVICE_CATEGORY_CPU, dc->categories);
+        break;
+    default:
+        set_bit(DEVICE_CATEGORY_MISC, dc->categories);
+        break;
+    }
+
+    for (i = 0; i < PCI_NUM_REGIONS; i++) {
+        config_op_send(pdev, PCI_BASE_ADDRESS_0 + (4 * i), &orig_val, 4,
+                       MPQEMU_CMD_PCI_CFGREAD);
+        new_val = 0xffffffff;
+        config_op_send(pdev, PCI_BASE_ADDRESS_0 + (4 * i), &new_val, 4,
+                       MPQEMU_CMD_PCI_CFGWRITE);
+        config_op_send(pdev, PCI_BASE_ADDRESS_0 + (4 * i), &new_val, 4,
+                       MPQEMU_CMD_PCI_CFGREAD);
+        size = (~(new_val & 0xFFFFFFF0)) + 1;
+        config_op_send(pdev, PCI_BASE_ADDRESS_0 + (4 * i), &orig_val, 4,
+                       MPQEMU_CMD_PCI_CFGWRITE);
+        type = (new_val & 0x1) ?
+                   PCI_BASE_ADDRESS_SPACE_IO : PCI_BASE_ADDRESS_SPACE_MEMORY;
+
+        if (size) {
+            g_autofree char *name;
+            pdev->region[i].dev = pdev;
+            pdev->region[i].present = true;
+            if (type == PCI_BASE_ADDRESS_SPACE_MEMORY) {
+                pdev->region[i].memory = true;
+            }
+            name = g_strdup_printf("bar-region-%d", i);
+            memory_region_init_io(&pdev->region[i].mr, OBJECT(pdev),
+                                  &proxy_mr_ops, &pdev->region[i],
+                                  name, size);
+            pci_register_bar(dev, i, type, &pdev->region[i].mr);
+        }
+    }
+}
-- 
2.29.2

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Perform device reset in the remote process when QEMU performs
device reset. This is required to reset the internal state
(like registers, etc...) of emulated devices

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 7cb220a51f565dc0817bd76e2f540e89c2d2b850.1611938319.git.jag.raman@oracle.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/hw/remote/mpqemu-link.h |  1 +
 hw/remote/message.c             | 22 ++++++++++++++++++++++
 hw/remote/proxy.c               | 19 +++++++++++++++++++
 3 files changed, 42 insertions(+)

From: "Denis V. Lunev" <den@openvz.org>

Original specification says that l1 table size if 64 * l1_size, which
is obviously wrong. The size of the l1 entry is 64 _bits_, not bytes.
Thus 64 is to be replaces with 8 as specification says about bytes.

There is also minor tweak, field name is renamed from l1 to l1_table,
which matches with the later text.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-id: 20210128171313.2210947-1-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

[Replace the original commit message "docs: fix mistake in dirty bitmap
feature description" as suggested by Eric Blake.
--Stefan]

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 docs/interop/parallels.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/docs/interop/parallels.txt b/docs/interop/parallels.txt
index XXXXXXX..XXXXXXX 100644
--- a/docs/interop/parallels.txt
+++ b/docs/interop/parallels.txt
@@ -XXX,XX +XXX,XX @@ of its data area are:
   28 - 31:    l1_size
               The number of entries in the L1 table of the bitmap.
 
-  variable:   l1 (64 * l1_size bytes)
+  variable:   l1_table (8 * l1_size bytes)
               L1 offset table (in bytes)
 
 A dirty bitmap is stored using a one-level structure for the mapping to host
-- 
2.29.2