Series comparison

-[PULL 0/8] Block patches
+[Qemu-devel] [PULL v2 0/6] Block patches
-The following changes since commit 013a18edbbc59cdad019100c7d03c0494642b74c:
+The following changes since commit b384cd95eb9c6f73ad84ed1bb0717a26e29cc78f:
-  Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-2020051=
+  Merge remote-tracking branch 'remotes/ehabkost/tags/machine-next-pull-request' into staging (2018-01-19 16:35:25 +0000)
 ' into staging (2020-05-14 16:17:55 +0100)
 are available in the Git repository at:
-  https://github.com/stefanha/qemu.git tags/block-pull-request
+  git://github.com/stefanha/qemu.git tags/block-pull-request
-for you to fetch changes up to ba607ca8bff4d2c2062902f8355657c865ac7c29:
+for you to fetch changes up to bcbb3866da19cce4360c828b6ec1c2a137757927:
-  aio-posix: disable fdmon-io_uring when GSource is used (2020-05-18 18:16:00=
+  block/parallels: add backing support to readv/writev (2018-01-22 14:02:33 +0000)
  +0100)
 ----------------------------------------------------------------
 Pull request
+v2:
+ * Drop merge failure from a previous pull request that broke virtio-blk on ARM
+   guests
+ * Add Parallels XML patch series
 ----------------------------------------------------------------
-Philippe Mathieu-Daud=C3=A9 (6):
+Edgar Kaziakhmedov (1):
-  tests/fuzz/Makefile: Do not link code using unavailable devices
+  block/parallels: add backing support to readv/writev
   Makefile: List fuzz targets in 'make help'
   tests/fuzz: Add missing space in test description
   tests/fuzz: Remove unuseful/unused typedefs
   tests/fuzz: Extract pciconfig_fuzz_qos() method
   tests/fuzz: Extract ioport_fuzz_qtest() method
-Stefan Hajnoczi (2):
+Klim Kireev (4):
-  aio-posix: don't duplicate fd handler deletion in
+  docs/interop/prl-xml: description of Parallels Disk format
-    fdmon_io_uring_destroy()
+  configure: add dependency
-  aio-posix: disable fdmon-io_uring when GSource is used
+  block/parallels: move some structures into header
   block/parallels: replace some magic numbers
- Makefile                          |  6 +++-
+Stefan Hajnoczi (1):
- tests/qtest/fuzz/Makefile.include |  6 ++--
+  block: add block_set_io_throttle virtio-blk-pci QMP example
  include/block/aio.h               |  3 ++
  tests/qtest/fuzz/i440fx_fuzz.c    | 47 ++++++++++++++++++++-----------
  util/aio-posix.c                  | 13 +++++++++
  util/aio-win32.c                  |  4 +++
  util/async.c                      |  1 +
  util/fdmon-io_uring.c             | 13 +++++++--
 files changed, 69 insertions(+), 24 deletions(-)
---=20
+ docs/interop/prl-xml.txt | 158 +++++++++++++++++++++++++++++++++++++++++++++++
-.25.3
+ qapi/block-core.json     |  18 ++++++
  configure                |  27 ++++++++
  block/Makefile.objs      |   2 +
  block/parallels.h        |  88 ++++++++++++++++++++++++++
  block/parallels.c        | 108 ++++++++++++++------------------
  scripts/checkpatch.pl    |   1 +
 files changed, 342 insertions(+), 60 deletions(-)
  create mode 100644 docs/interop/prl-xml.txt
  create mode 100644 block/parallels.h
+--
+.14.3

-[PULL 8/8] aio-posix: disable fdmon-io_uring when GSource is used
+[Qemu-devel] [PULL v2 1/6] block: add block_set_io_throttle virtio-blk-pci QMP example
-The glib event loop does not call fdmon_io_uring_wait() so fd handlers
+The block_set_io_throttle command can look up BlockBackends by the
-waiting to be submitted build up in the list. There is no benefit is
+attached qdev device ID.  virtio-blk-pci is a special case because the
-using io_uring when the glib GSource is being used, so disable it
+actual VirtIOBlock device is the "/virtio-backend" child of the PCI
-instead of implementing a more complex fix.
+adapter device.
-This fixes a memory leak where AioHandlers would build up and increasing
+Add a QMP schema example so clients will know how to use
-amounts of CPU time were spent iterating them in aio_pending(). The
+block_set_io_throttle on the virtio-blk-pci device.
 symptom is that guests become slow when QEMU is built with io_uring
 support.
-Buglink: https://bugs.launchpad.net/qemu/+bug/1877716
+The alternative is to implement some sort of aliasing for qmp_get_blk()
-Fixes: 73fd282e7b6dd4e4ea1c3bbb3d302c8db51e4ccf ("aio-posix: add io_uring fd monitoring implementation")
+but that is likely to cause confusion and could break future use cases.
 Let's not go there.
 Cc: Kevin Wolf <kwolf@redhat.com>
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
-Tested-by: Oleksandr Natalenko <oleksandr@redhat.com>
+Reviewed-by: Alberto Garcia <berto@igalia.com>
-Message-id: 20200511183630.279750-3-stefanha@redhat.com
+Message-id: 20180117090700.25811-1-stefanha@redhat.com
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- include/block/aio.h |  3 +++
+ qapi/block-core.json | 18 ++++++++++++++++++
- util/aio-posix.c    | 12 ++++++++++++
+file changed, 18 insertions(+)
  util/aio-win32.c    |  4 ++++
  util/async.c        |  1 +
 files changed, 20 insertions(+)
-diff --git a/include/block/aio.h b/include/block/aio.h
+diff --git a/qapi/block-core.json b/qapi/block-core.json
 index XXXXXXX..XXXXXXX 100644
---- a/include/block/aio.h
+--- a/qapi/block-core.json
-+++ b/include/block/aio.h
++++ b/qapi/block-core.json
-@@ -XXX,XX +XXX,XX @@ void aio_context_setup(AioContext *ctx);
+@@ -XXX,XX +XXX,XX @@
-  */
+ # Example:
- void aio_context_destroy(AioContext *ctx);
+ #
+ # -> { "execute": "block_set_io_throttle",
-+/* Used internally, do not call outside AioContext code */
++#      "arguments": { "id": "virtio-blk-pci0/virtio-backend",
-+void aio_context_use_g_source(AioContext *ctx);
++#                     "bps": 0,
-+
++#                     "bps_rd": 0,
- /**
++#                     "bps_wr": 0,
-  * aio_context_set_poll_params:
++#                     "iops": 512,
-  * @ctx: the aio context
++#                     "iops_rd": 0,
-diff --git a/util/aio-posix.c b/util/aio-posix.c
++#                     "iops_wr": 0,
-index XXXXXXX..XXXXXXX 100644
++#                     "bps_max": 0,
---- a/util/aio-posix.c
++#                     "bps_rd_max": 0,
-+++ b/util/aio-posix.c
++#                     "bps_wr_max": 0,
-@@ -XXX,XX +XXX,XX @@ void aio_context_destroy(AioContext *ctx)
++#                     "iops_max": 0,
-     aio_free_deleted_handlers(ctx);
++#                     "iops_rd_max": 0,
- }
++#                     "iops_wr_max": 0,
++#                     "bps_max_length": 0,
-+void aio_context_use_g_source(AioContext *ctx)
++#                     "iops_size": 0 } }
-+{
++# <- { "return": {} }
-+    /*
++#
-+     * Disable io_uring when the glib main loop is used because it doesn't
++# -> { "execute": "block_set_io_throttle",
-+     * support mixed glib/aio_poll() usage. It relies on aio_poll() being
+ #      "arguments": { "id": "ide0-1-0",
-+     * called regularly so that changes to the monitored file descriptors are
+ #                     "bps": 1000000,
-+     * submitted, otherwise a list of pending fd handlers builds up.
+ #                     "bps_rd": 0,
 +     */
 +    fdmon_io_uring_destroy(ctx);
 +    aio_free_deleted_handlers(ctx);
 +}
 +
  void aio_context_set_poll_params(AioContext *ctx, int64_t max_ns,
                                   int64_t grow, int64_t shrink, Error **errp)
  {
 diff --git a/util/aio-win32.c b/util/aio-win32.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/aio-win32.c
 +++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ void aio_context_destroy(AioContext *ctx)
  {
  }
 +void aio_context_use_g_source(AioContext *ctx)
 +{
 +}
 +
  void aio_context_set_poll_params(AioContext *ctx, int64_t max_ns,
                                   int64_t grow, int64_t shrink, Error **errp)
  {
 diff --git a/util/async.c b/util/async.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/async.c
 +++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ static GSourceFuncs aio_source_funcs = {
  GSource *aio_get_g_source(AioContext *ctx)
  {
 +    aio_context_use_g_source(ctx);
      g_source_ref(&ctx->source);
      return &ctx->source;
  }
 --
-.25.3
+.14.3

-[PULL 7/8] aio-posix: don't duplicate fd handler deletion in fdmon_io_uring_destroy()
+[Qemu-devel] [PULL v2 2/6] docs/interop/prl-xml: description of Parallels Disk format
-The io_uring file descriptor monitoring implementation has an internal
+From: Klim Kireev <klim.kireev@virtuozzo.com>
 list of fd handlers that are pending submission to io_uring.
 fdmon_io_uring_destroy() deletes all fd handlers on the list.
-Don't delete fd handlers directly in fdmon_io_uring_destroy() for two
+This patch adds main information about Parallels Disk
-reasons:
+format, which consists of DiskDescriptor.xml and other files.
 . This duplicates the aio-posix.c AioHandler deletion code and could
    become outdated if the struct changes.
 . Only handlers with the FDMON_IO_URING_REMOVE flag set are safe to
    remove. If the flag is not set then something still has a pointer to
    the fd handler. Let aio-posix.c and its user worry about that. In
    practice this isn't an issue because fdmon_io_uring_destroy() is only
    called when shutting down so all users have removed their fd
    handlers, but the next patch will need this!
-Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
+Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
-Tested-by: Oleksandr Natalenko <oleksandr@redhat.com>
+Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
-Message-id: 20200511183630.279750-2-stefanha@redhat.com
+Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
 Signed-off-by: Denis V. Lunev <den@openvz.org>
 Message-id: 20180112090122.1702-2-klim.kireev@virtuozzo.com
 CC: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- util/aio-posix.c      |  1 +
+ docs/interop/prl-xml.txt | 158 +++++++++++++++++++++++++++++++++++++++++++++++
- util/fdmon-io_uring.c | 13 ++++++++++---
+file changed, 158 insertions(+)
-files changed, 11 insertions(+), 3 deletions(-)
+ create mode 100644 docs/interop/prl-xml.txt
-diff --git a/util/aio-posix.c b/util/aio-posix.c
+diff --git a/docs/interop/prl-xml.txt b/docs/interop/prl-xml.txt
-index XXXXXXX..XXXXXXX 100644
+new file mode 100644
---- a/util/aio-posix.c
+index XXXXXXX..XXXXXXX
-+++ b/util/aio-posix.c
+--- /dev/null
-@@ -XXX,XX +XXX,XX @@ void aio_context_destroy(AioContext *ctx)
++++ b/docs/interop/prl-xml.txt
- {
+@@ -XXX,XX +XXX,XX @@
-     fdmon_io_uring_destroy(ctx);
++= License =
      fdmon_epoll_disable(ctx);
 +    aio_free_deleted_handlers(ctx);
  }
  void aio_context_set_poll_params(AioContext *ctx, int64_t max_ns,
 diff --git a/util/fdmon-io_uring.c b/util/fdmon-io_uring.c
 index XXXXXXX..XXXXXXX 100644
 --- a/util/fdmon-io_uring.c
 +++ b/util/fdmon-io_uring.c
@@ -XXX,XX +XXX,XX @@ void fdmon_io_uring_destroy(AioContext *ctx)
          io_uring_queue_exit(&ctx->fdmon_io_uring);
 -        /* No need to submit these anymore, just free them. */
 +        /* Move handlers due to be removed onto the deleted list */
          while ((node = QSLIST_FIRST_RCU(&ctx->submit_list))) {
 +            unsigned flags = atomic_fetch_and(&node->flags,
 +                    ~(FDMON_IO_URING_PENDING |
 +                      FDMON_IO_URING_ADD |
 +                      FDMON_IO_URING_REMOVE));
 +
-+            if (flags & FDMON_IO_URING_REMOVE) {
++Copyright (c) 2015-2017, Virtuozzo, Inc.
-+                QLIST_INSERT_HEAD_RCU(&ctx->deleted_aio_handlers, node, node_deleted);
++Authors:
-+            }
++        2015 Denis Lunev <den@openvz.org>
 +        2015 Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
 +        2016-2017 Klim Kireev <klim.kireev@virtuozzo.com>
 +        2016-2017 Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
 +
-             QSLIST_REMOVE_HEAD_RCU(&ctx->submit_list, node_submitted);
++This work is licensed under the terms of the GNU GPL, version 2 or later.
--            QLIST_REMOVE(node, node);
++See the COPYING file in the top-level directory.
--            g_free(node);
++
-         }
++This specification contains minimal information about Parallels Disk Format,
++which is enough to proper work with QEMU. Nevertheless, Parallels Cloud Server
-         ctx->fdmon_ops = &fdmon_poll_ops;
++and Parallels Desktop are able to add some unspecified nodes to xml and use
 +them, but they are for internal work and don't affect functionality. Also it
 +uses auxiliary xml "Snapshot.xml", which allows to store optional snapshot
 +information, but it doesn't influence open/read/write functionality. QEMU and
 +other software should not use fields not covered in this document and
 +Snapshot.xml file and must leave them as is.
 +
 += Parallels Disk Format =
 +
 +Parallels disk consists of two parts: the set of snapshots and the disk
 +descriptor file, which stores information about all files and snapshots.
 +
 +== Definitions ==
 +    Snapshot       a record of the contents captured at a particular time,
 +                   capable of storing current state. A snapshot has UUID and
 +                   parent UUID.
 +
 + Snapshot image    an overlay representing the difference between this
 +                   snapshot and some earlier snapshot.
 +
 +    Overlay        an image storing the different sectors between two captured
 +                   states.
 +
 +   Root image      snapshot image with no parent, the root of snapshot tree.
 +
 +    Storage        the backing storage for a subset of the virtual disk. When
 +                   there is more than one storage in a Parallels disk then that
 +                   is referred to as a split image. In this case every storage
 +                   covers specific address space area of the disk and has its
 +                   particular root image. Split images are not considered here
 +                   and are not supported. Each storage consists of disk
 +                   parameters and a list of images. The list of images always
 +                   contains a root image and may also contain overlays. The
 +                   root image can be an expandable Parallels image file or
 +                   plain. Overlays must be expandable.
 +
 +  Description      DiskDescriptor.xml stores information about disk parameters,
 +     file          snapshots, storages.
 +
 +     Top           The overlay between actual state and some previous snapshot.
 +   Snapshot        It is not a snapshot in the classical sense because it
 +                   serves as the active image that the guest writes to.
 +
 +    Sector         a 512-byte data chunk.
 +
 +== Description file ==
 +All information is placed in a single XML element Parallels_disk_image.
 +The element has only one attribute "Version", that must be 1.0.
 +Schema of DiskDescriptor.xml:
 +
 +<Parallels_disk_image Version="1.0">
 +    <Disk_Parameters>
 +        ...
 +    </Disk_Parameters>
 +    <StorageData>
 +        ...
 +    </StorageData>
 +    <Snapshots>
 +        ...
 +    </Snapshots>
 +</Parallels_disk_image>
 +
 +== Disk_Parameters element ==
 +The Disk_Parameters element describes the physical layout of the virtual disk
 +and some general settings.
 +
 +The Disk_Parameters element MUST contain the following child elements:
 +    * Disk_size - number of sectors in the disk,
 +                  desired size of the disk.
 +    * Cylinders - number of the disk cylinders.
 +    * Heads     - number of the disk heads.
 +    * Sectors   - number of the disk sectors per cylinder
 +                  (sector size is 512 bytes)
 +                  Limitation: Product of the Heads, Sectors and Cylinders
 +                  values MUST be equal to the value of the Disk_size parameter.
 +    * Padding   - must be 0. Parallels Cloud Server and Parallels Desktop may
 +                  use padding set to 1, however this case is not covered
 +                  by this spec, QEMU and other software should not open
 +                  such disks and should not create them.
 +
 +== StorageData element ==
 +This element of the file describes the root image and all snapshot images.
 +
 +The StorageData element consists of the Storage child element, as shown below:
 +<StorageData>
 +    <Storage>
 +        ...
 +    </Storage>
 +</StorageData>
 +
 +A Storage element has following child elements:
 +    * Start     - start sector of the storage, in case of non split storage
 +                  equals to 0.
 +    * End       - number of sector following the last sector, in case of non
 +                  split storage equals to Disk_size.
 +    * Blocksize - storage cluster size, number of sectors per one cluster.
 +                  Cluster size for each "Compressed" (see below) image in
 +                  parallels disk must be equal to this field. Note: cluster
 +                  size for Parallels Expandable Image is in 'tracks' field of
 +                  its header (see docs/interop/parallels.txt).
 +    * Several Image child elements.
 +
 +Each Image element has following child elements:
 +    * GUID - image identifier, UUID in curly brackets.
 +             For instance, {12345678-9abc-def1-2345-6789abcdef12}.
 +             The GUID is used by the Snapshots element to reference images
 +             (see below)
 +    * Type - image type of the element. It can be:
 +             "Plain" for raw files.
 +             "Compressed" for expanding disks.
 +    * File - path to image file. Path can be relative to DiskDecriptor.xml or
 +             absolute.
 +
 +== Snapshots element ==
 +The Snapshots element describes the snapshot relations with the snapshot tree.
 +
 +The element contains the set of Shot child elements, as shown below:
 +<Snapshots>
 +    <TopGUID> ... </TopGUID> /* Optional child element */
 +    <Shot>
 +        ...
 +    </Shot>
 +    <Shot>
 +        ...
 +    </Shot>
 +    ...
 +</Snapshots>
 +
 +Each Shot element contains the following child elements:
 +    * GUID       - an image GUID.
 +    * ParentGUID - GUID of the image of the parent snapshot.
 +
 +The software may traverse snapshots from child to parent using <ParentGUID>
 +field as reference. ParentGUID of root snapshot is
 +{00000000-0000-0000-0000-000000000000}. There should be only one root
 +snapshot. Top snapshot could be described via two ways: via TopGUID child
 +element of the Snapshots element or via predefined GUID
 +{5fbaabe3-6958-40ff-92a7-860e329aab41}. If TopGUID is defined, predefined GUID is
 +interpreted as usual GUID. All snapshot images (except Top Snapshot) should be
 +opened read-only. There is another predefined GUID,
 +BackupID = {704718e1-2314-44c8-9087-d78ed36b0f4e}, which is used by original and
 +some third-party software for backup, QEMU and other software may operate with
 +images with GUID = BackupID as usual, however, it is not recommended to use this
 +GUID for new disks. Top snapshot cannot have this GUID.
 --
-.25.3
+.14.3

-[PULL 6/8] tests/fuzz: Extract ioport_fuzz_qtest() method
+[Qemu-devel] [PULL v2 3/6] configure: add dependency
-From: Philippe Mathieu-Daudé <philmd@redhat.com>
+From: Klim Kireev <klim.kireev@virtuozzo.com>
-Extract generic ioport_fuzz_qtest() method from
+This dependency is required for adequate Parallels images support.
-i440fx_fuzz_qtest(). This will help to write tests
+Typically the disk consists of several images which are glued by
-not specific to the i440FX controller.
+XML disk descriptor. Also XML hides inside several important parameters
 which are not available in the image header.
-Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
+The patch also adds clause to checkpatch.pl to understand libxml2 types.
-Message-id: 20200514143433.18569-7-philmd@redhat.com
 Signed-off-by: Denis V. Lunev <den@openvz.org>
 Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
 Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
 Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
 Message-id: 20180112090122.1702-3-klim.kireev@virtuozzo.com
 CC: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- tests/qtest/fuzz/i440fx_fuzz.c | 11 +++++++++--
+ configure             | 27 +++++++++++++++++++++++++++
-file changed, 9 insertions(+), 2 deletions(-)
+ block/Makefile.objs   |  2 ++
  scripts/checkpatch.pl |  1 +
 files changed, 30 insertions(+)
-diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
+diff --git a/configure b/configure
 index XXXXXXX..XXXXXXX 100755
 --- a/configure
 +++ b/configure
@@ -XXX,XX +XXX,XX @@ tcmalloc="no"
  jemalloc="no"
  replication="yes"
  vxhs=""
 +libxml2=""
  supported_cpu="no"
  supported_os="no"
@@ -XXX,XX +XXX,XX @@ for opt do
    ;;
    --enable-numa) numa="yes"
    ;;
 +  --disable-libxml2) libxml2="no"
 +  ;;
 +  --enable-libxml2) libxml2="yes"
 +  ;;
    --disable-tcmalloc) tcmalloc="no"
    ;;
    --enable-tcmalloc) tcmalloc="yes"
@@ -XXX,XX +XXX,XX @@ disabled with --disable-FEATURE, default is enabled if available:
    tpm             TPM support
    libssh2         ssh block device support
    numa            libnuma support
 +  libxml2         for Parallels image format
    tcmalloc        tcmalloc support
    jemalloc        jemalloc support
    replication     replication support
@@ -XXX,XX +XXX,XX @@ EOF
    fi
  fi
 +##########################################
 +# libxml2 probe
 +if test "$libxml2" != "no" ; then
 +    if $pkg_config --exists libxml-2.0; then
 +        libxml2="yes"
 +        libxml2_cflags=$($pkg_config --cflags libxml-2.0)
 +        libxml2_libs=$($pkg_config --libs libxml-2.0)
 +    else
 +        if test "$libxml2" = "yes"; then
 +            feature_not_found "libxml2" "Install libxml2 devel"
 +        fi
 +        libxml2="no"
 +    fi
 +fi
  ##########################################
  # glusterfs probe
@@ -XXX,XX +XXX,XX @@ echo "lzo support       $lzo"
  echo "snappy support    $snappy"
  echo "bzip2 support     $bzip2"
  echo "NUMA host support $numa"
 +echo "libxml2           $libxml2"
  echo "tcmalloc support  $tcmalloc"
  echo "jemalloc support  $jemalloc"
  echo "avx2 optimization $avx2_opt"
@@ -XXX,XX +XXX,XX @@ if test "$have_rtnetlink" = "yes" ; then
    echo "CONFIG_RTNETLINK=y" >> $config_host_mak
  fi
 +if test "$libxml2" = "yes" ; then
 +  echo "CONFIG_LIBXML2=y" >> $config_host_mak
 +  echo "LIBXML2_CFLAGS=$libxml2_cflags" >> $config_host_mak
 +  echo "LIBXML2_LIBS=$libxml2_libs" >> $config_host_mak
 +fi
 +
  if test "$replication" = "yes" ; then
    echo "CONFIG_REPLICATION=y" >> $config_host_mak
  fi
 diff --git a/block/Makefile.objs b/block/Makefile.objs
 index XXXXXXX..XXXXXXX 100644
---- a/tests/qtest/fuzz/i440fx_fuzz.c
+--- a/block/Makefile.objs
-+++ b/tests/qtest/fuzz/i440fx_fuzz.c
++++ b/block/Makefile.objs
-@@ -XXX,XX +XXX,XX @@ enum action_id {
+@@ -XXX,XX +XXX,XX @@ block-obj-$(if $(CONFIG_BZIP2),m,n) += dmg-bz2.o
-     ACTION_MAX
+ dmg-bz2.o-libs     := $(BZIP2_LIBS)
- };
+ qcow.o-libs        := -lz
+ linux-aio.o-libs   := -laio
--static void i440fx_fuzz_qtest(QTestState *s,
++parallels.o-cflags := $(LIBXML2_CFLAGS)
-+static void ioport_fuzz_qtest(QTestState *s,
++parallels.o-libs   := $(LIBXML2_LIBS)
-         const unsigned char *Data, size_t Size) {
+diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
-     /*
+index XXXXXXX..XXXXXXX 100755
-      * loop over the Data, breaking it up into actions. each action has an
+--- a/scripts/checkpatch.pl
-@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qtest(QTestState *s,
++++ b/scripts/checkpatch.pl
-     flush_events(s);
+@@ -XXX,XX +XXX,XX @@ our @typeList = (
- }
+     qr{${Ident}_handler_fn},
+     qr{target_(?:u)?long},
-+static void i440fx_fuzz_qtest(QTestState *s,
+     qr{hwaddr},
-+                              const unsigned char *Data,
++    qr{xml${Ident}},
-+                              size_t Size)
+ );
-+{
-+    ioport_fuzz_qtest(s, Data, Size);
+ # This can be modified by sub possible.  Since it can be empty, be careful
 +}
 +
  static void pciconfig_fuzz_qos(QTestState *s, QPCIBus *bus,
          const unsigned char *Data, size_t Size) {
      /*
 -     * Same as i440fx_fuzz_qtest, but using QOS. devfn is incorporated into the
 +     * Same as ioport_fuzz_qtest, but using QOS. devfn is incorporated into the
       * value written over Port IO
       */
      struct {
 --
-.25.3
+.14.3

-[PULL 1/8] tests/fuzz/Makefile: Do not link code using unavailable devices
+[Qemu-devel] [PULL v2 4/6] block/parallels: move some structures into header
-From: Philippe Mathieu-Daudé <philmd@redhat.com>
+From: Klim Kireev <klim.kireev@virtuozzo.com>
-Some devices availability depends on CONFIG options.
+To implement xml format, some defines and structures
-Use these options to only link tests when requested device
+from parallels.c are required.
 is available.
-Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
+Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
-Message-id: 20200514143433.18569-2-philmd@redhat.com
+Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
 Signed-off-by: Denis V. Lunev <den@openvz.org>
 Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
 Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
 Message-id: 20180112090122.1702-4-klim.kireev@virtuozzo.com
 CC: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- tests/qtest/fuzz/Makefile.include | 6 +++---
+ block/parallels.h | 86 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
-file changed, 3 insertions(+), 3 deletions(-)
+ block/parallels.c | 53 +---------------------------------
 files changed, 87 insertions(+), 52 deletions(-)
  create mode 100644 block/parallels.h
-diff --git a/tests/qtest/fuzz/Makefile.include b/tests/qtest/fuzz/Makefile.include
+diff --git a/block/parallels.h b/block/parallels.h
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/block/parallels.h
@@ -XXX,XX +XXX,XX @@
 +/*
 +* Block driver for Parallels disk image format
 +*
 +* Copyright (c) 2015-2017 Virtuozzo, Inc.
 +* Authors:
 +*         2016-2017 Klim S. Kireev <klim.kireev@virtuozzo.com>
 +*         2015 Denis V. Lunev <den@openvz.org>
 +*
 +* This code was originally based on comparing different disk images created
 +* by Parallels. Currently it is based on opened OpenVZ sources
 +* available at
 +*     https://github.com/OpenVZ/ploop
 +*
 +* Permission is hereby granted, free of charge, to any person obtaining a copy
 +* of this software and associated documentation files (the "Software"), to deal
 +* in the Software without restriction, including without limitation the rights
 +* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 +* copies of the Software, and to permit persons to whom the Software is
 +* furnished to do so, subject to the following conditions:
 +*
 +* The above copyright notice and this permission notice shall be included in
 +* all copies or substantial portions of the Software.
 +*
 +* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 +* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 +* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 +* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 +* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 +* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 +* THE SOFTWARE.
 +*/
 +#ifndef BLOCK_PARALLELS_H
 +#define BLOCK_PARALLELS_H
 +#include "qemu/coroutine.h"
 +#include "qemu/typedefs.h"
 +
 +#define DEFAULT_CLUSTER_SIZE 1048576        /* 1 MiB */
 +
 +/* always little-endian */
 +typedef struct ParallelsHeader {
 +    char magic[16]; /* "WithoutFreeSpace" */
 +    uint32_t version;
 +    uint32_t heads;
 +    uint32_t cylinders;
 +    uint32_t tracks;
 +    uint32_t bat_entries;
 +    uint64_t nb_sectors;
 +    uint32_t inuse;
 +    uint32_t data_off;
 +    char padding[12];
 +} QEMU_PACKED ParallelsHeader;
 +
 +typedef enum ParallelsPreallocMode {
 +    PRL_PREALLOC_MODE_FALLOCATE = 0,
 +    PRL_PREALLOC_MODE_TRUNCATE = 1,
 +    PRL_PREALLOC_MODE__MAX = 2,
 +} ParallelsPreallocMode;
 +
 +typedef struct BDRVParallelsState {
 +    /** Locking is conservative, the lock protects
 +     *   - image file extending (truncate, fallocate)
 +     *   - any access to block allocation table
 +     */
 +    CoMutex lock;
 +
 +    ParallelsHeader *header;
 +    uint32_t header_size;
 +    bool header_unclean;
 +
 +    unsigned long *bat_dirty_bmap;
 +    unsigned int  bat_dirty_block;
 +
 +    uint32_t *bat_bitmap;
 +    unsigned int bat_size;
 +
 +    int64_t  data_end;
 +    uint64_t prealloc_size;
 +    ParallelsPreallocMode prealloc_mode;
 +
 +    unsigned int tracks;
 +
 +    unsigned int off_multiplier;
 +    Error *migration_blocker;
 +} BDRVParallelsState;
 +
 +#endif
 diff --git a/block/parallels.c b/block/parallels.c
 index XXXXXXX..XXXXXXX 100644
---- a/tests/qtest/fuzz/Makefile.include
+--- a/block/parallels.c
-+++ b/tests/qtest/fuzz/Makefile.include
++++ b/block/parallels.c
-@@ -XXX,XX +XXX,XX @@ fuzz-obj-y += tests/qtest/fuzz/fork_fuzz.o
+@@ -XXX,XX +XXX,XX @@
- fuzz-obj-y += tests/qtest/fuzz/qos_fuzz.o
+ #include "qemu/bswap.h"
+ #include "qemu/bitmap.h"
- # Targets
+ #include "migration/blocker.h"
--fuzz-obj-y += tests/qtest/fuzz/i440fx_fuzz.o
++#include "parallels.h"
--fuzz-obj-y += tests/qtest/fuzz/virtio_net_fuzz.o
--fuzz-obj-y += tests/qtest/fuzz/virtio_scsi_fuzz.o
+ /**************************************************************/
-+fuzz-obj-$(CONFIG_PCI_I440FX) += tests/qtest/fuzz/i440fx_fuzz.o
-+fuzz-obj-$(CONFIG_VIRTIO_NET) += tests/qtest/fuzz/virtio_net_fuzz.o
+@@ -XXX,XX +XXX,XX @@
-+fuzz-obj-$(CONFIG_SCSI) += tests/qtest/fuzz/virtio_scsi_fuzz.o
+ #define HEADER_INUSE_MAGIC  (0x746F6E59)
+ #define MAX_PARALLELS_IMAGE_FACTOR (1ull << 32)
- FUZZ_CFLAGS += -I$(SRC_PATH)/tests -I$(SRC_PATH)/tests/qtest
 -#define DEFAULT_CLUSTER_SIZE 1048576        /* 1 MiB */
 -
 -
 -// always little-endian
 -typedef struct ParallelsHeader {
 -    char magic[16]; // "WithoutFreeSpace"
 -    uint32_t version;
 -    uint32_t heads;
 -    uint32_t cylinders;
 -    uint32_t tracks;
 -    uint32_t bat_entries;
 -    uint64_t nb_sectors;
 -    uint32_t inuse;
 -    uint32_t data_off;
 -    char padding[12];
 -} QEMU_PACKED ParallelsHeader;
 -
 -
 -typedef enum ParallelsPreallocMode {
 -    PRL_PREALLOC_MODE_FALLOCATE = 0,
 -    PRL_PREALLOC_MODE_TRUNCATE = 1,
 -    PRL_PREALLOC_MODE__MAX = 2,
 -} ParallelsPreallocMode;
 -
  static QEnumLookup prealloc_mode_lookup = {
      .array = (const char *const[]) {
          "falloc",
@@ -XXX,XX +XXX,XX @@ static QEnumLookup prealloc_mode_lookup = {
      .size = PRL_PREALLOC_MODE__MAX
  };
 -typedef struct BDRVParallelsState {
 -    /** Locking is conservative, the lock protects
 -     *   - image file extending (truncate, fallocate)
 -     *   - any access to block allocation table
 -     */
 -    CoMutex lock;
 -
 -    ParallelsHeader *header;
 -    uint32_t header_size;
 -    bool header_unclean;
 -
 -    unsigned long *bat_dirty_bmap;
 -    unsigned int  bat_dirty_block;
 -
 -    uint32_t *bat_bitmap;
 -    unsigned int bat_size;
 -
 -    int64_t  data_end;
 -    uint64_t prealloc_size;
 -    ParallelsPreallocMode prealloc_mode;
 -
 -    unsigned int tracks;
 -
 -    unsigned int off_multiplier;
 -    Error *migration_blocker;
 -} BDRVParallelsState;
 -
 -
  #define PARALLELS_OPT_PREALLOC_MODE     "prealloc-mode"
  #define PARALLELS_OPT_PREALLOC_SIZE     "prealloc-size"
 --
-.25.3
+.14.3

-[PULL 2/8] Makefile: List fuzz targets in 'make help'
+Deleted patch
-From: Philippe Mathieu-Daudé <philmd@redhat.com>
-List softmmu fuzz targets in 'make help' output:
-  $ make help
-  ...
-  Architecture specific targets:
-  aarch64-softmmu/all            - Build for aarch64-softmmu
-  aarch64-softmmu/fuzz           - Build fuzzer for aarch64-softmmu
-  alpha-softmmu/all              - Build for alpha-softmmu
-  alpha-softmmu/fuzz             - Build fuzzer for alpha-softmmu
-  arm-softmmu/all                - Build for arm-softmmu
-  arm-softmmu/fuzz               - Build fuzzer for arm-softmmu
-  ...
-Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
-Message-id: 20200514143433.18569-3-philmd@redhat.com
-Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
----
- Makefile | 6 +++++-
-file changed, 5 insertions(+), 1 deletion(-)
-diff --git a/Makefile b/Makefile
-index XXXXXXX..XXXXXXX 100644
---- a/Makefile
-+++ b/Makefile
-@@ -XXX,XX +XXX,XX @@ endif
-     @$(if $(TARGET_DIRS), \
-         echo 'Architecture specific targets:'; \
-         $(foreach t, $(TARGET_DIRS), \
--        $(call print-help-run,$(t)/all,Build for $(t));) \
-+        $(call print-help-run,$(t)/all,Build for $(t)); \
-+        $(if $(CONFIG_FUZZ), \
-+            $(if $(findstring softmmu,$(t)), \
-+                $(call print-help-run,$(t)/fuzz,Build fuzzer for $(t)); \
-+        ))) \
-         echo '')
-     @$(if $(TOOLS), \
-         echo 'Tools targets:'; \
---
-.25.3

-[PULL 3/8] tests/fuzz: Add missing space in test description
+Deleted patch
-From: Philippe Mathieu-Daudé <philmd@redhat.com>
-Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
-Message-id: 20200514143433.18569-4-philmd@redhat.com
-Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
----
- tests/qtest/fuzz/i440fx_fuzz.c | 6 +++---
-file changed, 3 insertions(+), 3 deletions(-)
-diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
-index XXXXXXX..XXXXXXX 100644
---- a/tests/qtest/fuzz/i440fx_fuzz.c
-+++ b/tests/qtest/fuzz/i440fx_fuzz.c
-@@ -XXX,XX +XXX,XX @@ static void register_pci_fuzz_targets(void)
-     /* Uses simple qtest commands and reboots to reset state */
-     fuzz_add_target(&(FuzzTarget){
-                 .name = "i440fx-qtest-reboot-fuzz",
--                .description = "Fuzz the i440fx using raw qtest commands and"
-+                .description = "Fuzz the i440fx using raw qtest commands and "
-                                "rebooting after each run",
-                 .get_init_cmdline = i440fx_argv,
-                 .fuzz = i440fx_fuzz_qtest});
-@@ -XXX,XX +XXX,XX @@ static void register_pci_fuzz_targets(void)
-     /* Uses libqos and forks to prevent state leakage */
-     fuzz_add_qos_target(&(FuzzTarget){
-                 .name = "i440fx-qos-fork-fuzz",
--                .description = "Fuzz the i440fx using raw qtest commands and"
-+                .description = "Fuzz the i440fx using raw qtest commands and "
-                                "rebooting after each run",
-                 .pre_vm_init = &fork_init,
-                 .fuzz = i440fx_fuzz_qos_fork,},
-@@ -XXX,XX +XXX,XX @@ static void register_pci_fuzz_targets(void)
-      */
-     fuzz_add_qos_target(&(FuzzTarget){
-                 .name = "i440fx-qos-noreset-fuzz",
--                .description = "Fuzz the i440fx using raw qtest commands and"
-+                .description = "Fuzz the i440fx using raw qtest commands and "
-                                "rebooting after each run",
-                 .fuzz = i440fx_fuzz_qos,},
-                 "i440FX-pcihost",
---
-.25.3

-[PULL 4/8] tests/fuzz: Remove unuseful/unused typedefs
+[Qemu-devel] [PULL v2 5/6] block/parallels: replace some magic numbers
-From: Philippe Mathieu-Daudé <philmd@redhat.com>
+From: Klim Kireev <klim.kireev@virtuozzo.com>
-These typedefs are not used. Use a simple structure,
+Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
-remote the typedefs.
+Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
+Signed-off-by: Denis V. Lunev <den@openvz.org>
-Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
+Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
-Message-id: 20200514143433.18569-5-philmd@redhat.com
+Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
 Message-id: 20180112090122.1702-5-klim.kireev@virtuozzo.com
 CC: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- tests/qtest/fuzz/i440fx_fuzz.c | 10 ++++------
+ block/parallels.h | 2 ++
-file changed, 4 insertions(+), 6 deletions(-)
+ block/parallels.c | 5 +++--
 files changed, 5 insertions(+), 2 deletions(-)
-diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
+diff --git a/block/parallels.h b/block/parallels.h
 index XXXXXXX..XXXXXXX 100644
---- a/tests/qtest/fuzz/i440fx_fuzz.c
+--- a/block/parallels.h
-+++ b/tests/qtest/fuzz/i440fx_fuzz.c
++++ b/block/parallels.h
-@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qtest(QTestState *s,
+@@ -XXX,XX +XXX,XX @@
-      * loop over the Data, breaking it up into actions. each action has an
+ #include "qemu/coroutine.h"
-      * opcode, address offset and value
+ #include "qemu/typedefs.h"
-      */
--    typedef struct QTestFuzzAction {
++#define HEADS_NUMBER 16
-+    struct {
++#define SEC_IN_CYL 32
-         uint8_t opcode;
+ #define DEFAULT_CLUSTER_SIZE 1048576        /* 1 MiB */
-         uint8_t addr;
-         uint32_t value;
+ /* always little-endian */
--    } QTestFuzzAction;
+diff --git a/block/parallels.c b/block/parallels.c
--    QTestFuzzAction a;
+index XXXXXXX..XXXXXXX 100644
-+    } a;
+--- a/block/parallels.c
++++ b/block/parallels.c
-     while (Size >= sizeof(a)) {
+@@ -XXX,XX +XXX,XX @@ static int parallels_create(const char *filename, QemuOpts *opts, Error **errp)
-         /* make a copy of the action so we can normalize the values in-place */
+     memcpy(header.magic, HEADER_MAGIC2, sizeof(header.magic));
-@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qos(QTestState *s,
+     header.version = cpu_to_le32(HEADER_VERSION);
-      * Same as i440fx_fuzz_qtest, but using QOS. devfn is incorporated into the
+     /* don't care much about geometry, it is not used on image level */
-      * value written over Port IO
+-    header.heads = cpu_to_le32(16);
-      */
+-    header.cylinders = cpu_to_le32(total_size / BDRV_SECTOR_SIZE / 16 / 32);
--    typedef struct QOSFuzzAction {
++    header.heads = cpu_to_le32(HEADS_NUMBER);
-+    struct {
++    header.cylinders = cpu_to_le32(total_size / BDRV_SECTOR_SIZE
-         uint8_t opcode;
++                                   / HEADS_NUMBER / SEC_IN_CYL);
-         uint8_t offset;
+     header.tracks = cpu_to_le32(cl_size >> BDRV_SECTOR_BITS);
-         int devfn;
+     header.bat_entries = cpu_to_le32(bat_entries);
-         uint32_t value;
+     header.nb_sectors = cpu_to_le64(DIV_ROUND_UP(total_size, BDRV_SECTOR_SIZE));
 -    } QOSFuzzAction;
 +    } a;
      static QPCIBus *bus;
      if (!bus) {
          bus = qpci_new_pc(s, fuzz_qos_alloc);
      }
 -    QOSFuzzAction a;
      while (Size >= sizeof(a)) {
          memcpy(&a, Data, sizeof(a));
          switch (a.opcode % ACTION_MAX) {
 --
-.25.3
+.14.3

-[PULL 5/8] tests/fuzz: Extract pciconfig_fuzz_qos() method
+[Qemu-devel] [PULL v2 6/6] block/parallels: add backing support to readv/writev
-From: Philippe Mathieu-Daudé <philmd@redhat.com>
+From: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
-Extract the generic pciconfig_fuzz_qos() method from
+Since parallels format supports backing files, refine
-i440fx_fuzz_qos(). This will help to write tests not
+readv/writev (allocate_clusters) to redirect read/write requests
-specific to the i440FX controller.
+to a backing file (if cluster is not available in the current bs).
-Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
+Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
-Message-id: 20200514143433.18569-6-philmd@redhat.com
+Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
 Signed-off-by: Denis V. Lunev <den@openvz.org>
 Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
 Message-id: 20180112090122.1702-6-klim.kireev@virtuozzo.com
 CC: Stefan Hajnoczi <stefanha@redhat.com>
 Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
 ---
- tests/qtest/fuzz/i440fx_fuzz.c | 20 ++++++++++++++------
+ block/parallels.c | 50 ++++++++++++++++++++++++++++++++++++++++++++------
-file changed, 14 insertions(+), 6 deletions(-)
+file changed, 44 insertions(+), 6 deletions(-)
-diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
+diff --git a/block/parallels.c b/block/parallels.c
 index XXXXXXX..XXXXXXX 100644
---- a/tests/qtest/fuzz/i440fx_fuzz.c
+--- a/block/parallels.c
-+++ b/tests/qtest/fuzz/i440fx_fuzz.c
++++ b/block/parallels.c
-@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qtest(QTestState *s,
+@@ -XXX,XX +XXX,XX @@ static int64_t block_status(BDRVParallelsState *s, int64_t sector_num,
-     flush_events(s);
+ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
- }
+                                  int nb_sectors, int *pnum)
+ {
--static void i440fx_fuzz_qos(QTestState *s,
++    int ret;
-+static void pciconfig_fuzz_qos(QTestState *s, QPCIBus *bus,
+     BDRVParallelsState *s = bs->opaque;
-         const unsigned char *Data, size_t Size) {
+     int64_t pos, space, idx, to_allocate, i, len;
-     /*
-      * Same as i440fx_fuzz_qtest, but using QOS. devfn is incorporated into the
+@@ -XXX,XX +XXX,XX @@ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
-@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qos(QTestState *s,
+         return len;
-         uint32_t value;
+     }
-     } a;
+     if (s->data_end + space > (len >> BDRV_SECTOR_BITS)) {
+-        int ret;
--    static QPCIBus *bus;
+         space += s->prealloc_size;
--    if (!bus) {
+         if (s->prealloc_mode == PRL_PREALLOC_MODE_FALLOCATE) {
--        bus = qpci_new_pc(s, fuzz_qos_alloc);
+             ret = bdrv_pwrite_zeroes(bs->file,
--    }
+@@ -XXX,XX +XXX,XX @@ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
--
+         }
-     while (Size >= sizeof(a)) {
+     }
-         memcpy(&a, Data, sizeof(a));
-         switch (a.opcode % ACTION_MAX) {
++    /* Try to read from backing to fill empty clusters
-@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qos(QTestState *s,
++     * FIXME: 1. previous write_zeroes may be redundant
-     flush_events(s);
++     *        2. most of data we read from backing will be rewritten by
- }
++     *           parallels_co_writev. On aligned-to-cluster write we do not need
++     *           this read at all.
-+static void i440fx_fuzz_qos(QTestState *s,
++     *        3. it would be good to combine write of data from backing and new
-+                            const unsigned char *Data,
++     *           data into one write call */
-+                            size_t Size)
++    if (bs->backing) {
-+{
++        int64_t nb_cow_sectors = to_allocate * s->tracks;
-+    static QPCIBus *bus;
++        int64_t nb_cow_bytes = nb_cow_sectors << BDRV_SECTOR_BITS;
 +        QEMUIOVector qiov;
 +        struct iovec iov = {
 +            .iov_len = nb_cow_bytes,
 +            .iov_base = qemu_blockalign(bs, nb_cow_bytes)
 +        };
 +        qemu_iovec_init_external(&qiov, &iov, 1);
 +
-+    if (!bus) {
++        ret = bdrv_co_readv(bs->backing, idx * s->tracks, nb_cow_sectors,
-+        bus = qpci_new_pc(s, fuzz_qos_alloc);
++                            &qiov);
 +        if (ret < 0) {
 +            qemu_vfree(iov.iov_base);
 +            return ret;
 +        }
 +
 +        ret = bdrv_co_writev(bs->file, s->data_end, nb_cow_sectors, &qiov);
 +        qemu_vfree(iov.iov_base);
 +        if (ret < 0) {
 +            return ret;
 +        }
 +    }
 +
-+    pciconfig_fuzz_qos(s, bus, Data, Size);
+     for (i = 0; i < to_allocate; i++) {
-+}
+         s->bat_bitmap[idx + i] = cpu_to_le32(s->data_end / s->off_multiplier);
          s->data_end += s->tracks;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int parallels_co_readv(BlockDriverState *bs,
          nbytes = n << BDRV_SECTOR_BITS;
 +        qemu_iovec_reset(&hd_qiov);
 +        qemu_iovec_concat(&hd_qiov, qiov, bytes_done, nbytes);
 +
- static void i440fx_fuzz_qos_fork(QTestState *s,
+         if (position < 0) {
-         const unsigned char *Data, size_t Size) {
+-            qemu_iovec_memset(qiov, bytes_done, 0, nbytes);
-     if (fork() == 0) {
++            if (bs->backing) {
 +                ret = bdrv_co_readv(bs->backing, sector_num, n, &hd_qiov);
 +                if (ret < 0) {
 +                    break;
 +                }
 +            } else {
 +                qemu_iovec_memset(&hd_qiov, 0, 0, nbytes);
 +            }
          } else {
 -            qemu_iovec_reset(&hd_qiov);
 -            qemu_iovec_concat(&hd_qiov, qiov, bytes_done, nbytes);
 -
              ret = bdrv_co_readv(bs->file, position, n, &hd_qiov);
              if (ret < 0) {
                  break;
@@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_parallels = {
      .bdrv_co_flush_to_os      = parallels_co_flush_to_os,
      .bdrv_co_readv  = parallels_co_readv,
      .bdrv_co_writev = parallels_co_writev,
 -
 +    .supports_backing = true,
      .bdrv_create    = parallels_create,
      .bdrv_check     = parallels_check,
      .create_opts    = &parallels_create_opts,
 --
-.25.3
+.14.3

The following changes since commit 013a18edbbc59cdad019100c7d03c0494642b74c:

Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-2020051=
4' into staging (2020-05-14 16:17:55 +0100)

are available in the Git repository at:

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

for you to fetch changes up to ba607ca8bff4d2c2062902f8355657c865ac7c29:

aio-posix: disable fdmon-io_uring when GSource is used (2020-05-18 18:16:00=
 +0100)

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

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

Philippe Mathieu-Daud=C3=A9 (6):
  tests/fuzz/Makefile: Do not link code using unavailable devices
  Makefile: List fuzz targets in 'make help'
  tests/fuzz: Add missing space in test description
  tests/fuzz: Remove unuseful/unused typedefs
  tests/fuzz: Extract pciconfig_fuzz_qos() method
  tests/fuzz: Extract ioport_fuzz_qtest() method

Stefan Hajnoczi (2):
  aio-posix: don't duplicate fd handler deletion in
    fdmon_io_uring_destroy()
  aio-posix: disable fdmon-io_uring when GSource is used

--=20
2.25.3

From: Philippe Mathieu-Daudé <philmd@redhat.com>

Some devices availability depends on CONFIG options.
Use these options to only link tests when requested device
is available.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20200514143433.18569-2-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/qtest/fuzz/Makefile.include | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tests/qtest/fuzz/Makefile.include b/tests/qtest/fuzz/Makefile.include
index XXXXXXX..XXXXXXX 100644
--- a/tests/qtest/fuzz/Makefile.include
+++ b/tests/qtest/fuzz/Makefile.include
@@ -XXX,XX +XXX,XX @@ fuzz-obj-y += tests/qtest/fuzz/fork_fuzz.o
 fuzz-obj-y += tests/qtest/fuzz/qos_fuzz.o
 
 # Targets
-fuzz-obj-y += tests/qtest/fuzz/i440fx_fuzz.o
-fuzz-obj-y += tests/qtest/fuzz/virtio_net_fuzz.o
-fuzz-obj-y += tests/qtest/fuzz/virtio_scsi_fuzz.o
+fuzz-obj-$(CONFIG_PCI_I440FX) += tests/qtest/fuzz/i440fx_fuzz.o
+fuzz-obj-$(CONFIG_VIRTIO_NET) += tests/qtest/fuzz/virtio_net_fuzz.o
+fuzz-obj-$(CONFIG_SCSI) += tests/qtest/fuzz/virtio_scsi_fuzz.o
 
 FUZZ_CFLAGS += -I$(SRC_PATH)/tests -I$(SRC_PATH)/tests/qtest
 
-- 
2.25.3

From: Philippe Mathieu-Daudé <philmd@redhat.com>

List softmmu fuzz targets in 'make help' output:

$ make help
  ...
  Architecture specific targets:
  aarch64-softmmu/all            - Build for aarch64-softmmu
  aarch64-softmmu/fuzz           - Build fuzzer for aarch64-softmmu
  alpha-softmmu/all              - Build for alpha-softmmu
  alpha-softmmu/fuzz             - Build fuzzer for alpha-softmmu
  arm-softmmu/all                - Build for arm-softmmu
  arm-softmmu/fuzz               - Build fuzzer for arm-softmmu
  ...

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20200514143433.18569-3-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 Makefile | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/Makefile b/Makefile
index XXXXXXX..XXXXXXX 100644
--- a/Makefile
+++ b/Makefile
@@ -XXX,XX +XXX,XX @@ endif
 	@$(if $(TARGET_DIRS), \
 		echo 'Architecture specific targets:'; \
 		$(foreach t, $(TARGET_DIRS), \
-		$(call print-help-run,$(t)/all,Build for $(t));) \
+		$(call print-help-run,$(t)/all,Build for $(t)); \
+		$(if $(CONFIG_FUZZ), \
+			$(if $(findstring softmmu,$(t)), \
+				$(call print-help-run,$(t)/fuzz,Build fuzzer for $(t)); \
+		))) \
 		echo '')
 	@$(if $(TOOLS), \
 		echo 'Tools targets:'; \
-- 
2.25.3

From: Philippe Mathieu-Daudé <philmd@redhat.com>

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20200514143433.18569-4-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/qtest/fuzz/i440fx_fuzz.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
index XXXXXXX..XXXXXXX 100644
--- a/tests/qtest/fuzz/i440fx_fuzz.c
+++ b/tests/qtest/fuzz/i440fx_fuzz.c
@@ -XXX,XX +XXX,XX @@ static void register_pci_fuzz_targets(void)
     /* Uses simple qtest commands and reboots to reset state */
     fuzz_add_target(&(FuzzTarget){
                 .name = "i440fx-qtest-reboot-fuzz",
-                .description = "Fuzz the i440fx using raw qtest commands and"
+                .description = "Fuzz the i440fx using raw qtest commands and "
                                "rebooting after each run",
                 .get_init_cmdline = i440fx_argv,
                 .fuzz = i440fx_fuzz_qtest});
@@ -XXX,XX +XXX,XX @@ static void register_pci_fuzz_targets(void)
     /* Uses libqos and forks to prevent state leakage */
     fuzz_add_qos_target(&(FuzzTarget){
                 .name = "i440fx-qos-fork-fuzz",
-                .description = "Fuzz the i440fx using raw qtest commands and"
+                .description = "Fuzz the i440fx using raw qtest commands and "
                                "rebooting after each run",
                 .pre_vm_init = &fork_init,
                 .fuzz = i440fx_fuzz_qos_fork,},
@@ -XXX,XX +XXX,XX @@ static void register_pci_fuzz_targets(void)
      */
     fuzz_add_qos_target(&(FuzzTarget){
                 .name = "i440fx-qos-noreset-fuzz",
-                .description = "Fuzz the i440fx using raw qtest commands and"
+                .description = "Fuzz the i440fx using raw qtest commands and "
                                "rebooting after each run",
                 .fuzz = i440fx_fuzz_qos,},
                 "i440FX-pcihost",
-- 
2.25.3

From: Philippe Mathieu-Daudé <philmd@redhat.com>

These typedefs are not used. Use a simple structure,
remote the typedefs.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20200514143433.18569-5-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/qtest/fuzz/i440fx_fuzz.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
index XXXXXXX..XXXXXXX 100644
--- a/tests/qtest/fuzz/i440fx_fuzz.c
+++ b/tests/qtest/fuzz/i440fx_fuzz.c
@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qtest(QTestState *s,
      * loop over the Data, breaking it up into actions. each action has an
      * opcode, address offset and value
      */
-    typedef struct QTestFuzzAction {
+    struct {
         uint8_t opcode;
         uint8_t addr;
         uint32_t value;
-    } QTestFuzzAction;
-    QTestFuzzAction a;
+    } a;
 
     while (Size >= sizeof(a)) {
         /* make a copy of the action so we can normalize the values in-place */
@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qos(QTestState *s,
      * Same as i440fx_fuzz_qtest, but using QOS. devfn is incorporated into the
      * value written over Port IO
      */
-    typedef struct QOSFuzzAction {
+    struct {
         uint8_t opcode;
         uint8_t offset;
         int devfn;
         uint32_t value;
-    } QOSFuzzAction;
+    } a;
 
     static QPCIBus *bus;
     if (!bus) {
         bus = qpci_new_pc(s, fuzz_qos_alloc);
     }
 
-    QOSFuzzAction a;
     while (Size >= sizeof(a)) {
         memcpy(&a, Data, sizeof(a));
         switch (a.opcode % ACTION_MAX) {
-- 
2.25.3

From: Philippe Mathieu-Daudé <philmd@redhat.com>

Extract the generic pciconfig_fuzz_qos() method from
i440fx_fuzz_qos(). This will help to write tests not
specific to the i440FX controller.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20200514143433.18569-6-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/qtest/fuzz/i440fx_fuzz.c | 20 ++++++++++++++------
 1 file changed, 14 insertions(+), 6 deletions(-)

diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
index XXXXXXX..XXXXXXX 100644
--- a/tests/qtest/fuzz/i440fx_fuzz.c
+++ b/tests/qtest/fuzz/i440fx_fuzz.c
@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qtest(QTestState *s,
     flush_events(s);
 }
 
-static void i440fx_fuzz_qos(QTestState *s,
+static void pciconfig_fuzz_qos(QTestState *s, QPCIBus *bus,
         const unsigned char *Data, size_t Size) {
     /*
      * Same as i440fx_fuzz_qtest, but using QOS. devfn is incorporated into the
@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qos(QTestState *s,
         uint32_t value;
     } a;
 
-    static QPCIBus *bus;
-    if (!bus) {
-        bus = qpci_new_pc(s, fuzz_qos_alloc);
-    }
-
     while (Size >= sizeof(a)) {
         memcpy(&a, Data, sizeof(a));
         switch (a.opcode % ACTION_MAX) {
@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qos(QTestState *s,
     flush_events(s);
 }
 
+static void i440fx_fuzz_qos(QTestState *s,
+                            const unsigned char *Data,
+                            size_t Size)
+{
+    static QPCIBus *bus;
+
+    if (!bus) {
+        bus = qpci_new_pc(s, fuzz_qos_alloc);
+    }
+
+    pciconfig_fuzz_qos(s, bus, Data, Size);
+}
+
 static void i440fx_fuzz_qos_fork(QTestState *s,
         const unsigned char *Data, size_t Size) {
     if (fork() == 0) {
-- 
2.25.3

From: Philippe Mathieu-Daudé <philmd@redhat.com>

Extract generic ioport_fuzz_qtest() method from
i440fx_fuzz_qtest(). This will help to write tests
not specific to the i440FX controller.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20200514143433.18569-7-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/qtest/fuzz/i440fx_fuzz.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/tests/qtest/fuzz/i440fx_fuzz.c b/tests/qtest/fuzz/i440fx_fuzz.c
index XXXXXXX..XXXXXXX 100644
--- a/tests/qtest/fuzz/i440fx_fuzz.c
+++ b/tests/qtest/fuzz/i440fx_fuzz.c
@@ -XXX,XX +XXX,XX @@ enum action_id {
     ACTION_MAX
 };
 
-static void i440fx_fuzz_qtest(QTestState *s,
+static void ioport_fuzz_qtest(QTestState *s,
         const unsigned char *Data, size_t Size) {
     /*
      * loop over the Data, breaking it up into actions. each action has an
@@ -XXX,XX +XXX,XX @@ static void i440fx_fuzz_qtest(QTestState *s,
     flush_events(s);
 }
 
+static void i440fx_fuzz_qtest(QTestState *s,
+                              const unsigned char *Data,
+                              size_t Size)
+{
+    ioport_fuzz_qtest(s, Data, Size);
+}
+
 static void pciconfig_fuzz_qos(QTestState *s, QPCIBus *bus,
         const unsigned char *Data, size_t Size) {
     /*
-     * Same as i440fx_fuzz_qtest, but using QOS. devfn is incorporated into the
+     * Same as ioport_fuzz_qtest, but using QOS. devfn is incorporated into the
      * value written over Port IO
      */
     struct {
-- 
2.25.3

The io_uring file descriptor monitoring implementation has an internal
list of fd handlers that are pending submission to io_uring.
fdmon_io_uring_destroy() deletes all fd handlers on the list.

Don't delete fd handlers directly in fdmon_io_uring_destroy() for two
reasons:
1. This duplicates the aio-posix.c AioHandler deletion code and could
   become outdated if the struct changes.
2. Only handlers with the FDMON_IO_URING_REMOVE flag set are safe to
   remove. If the flag is not set then something still has a pointer to
   the fd handler. Let aio-posix.c and its user worry about that. In
   practice this isn't an issue because fdmon_io_uring_destroy() is only
   called when shutting down so all users have removed their fd
   handlers, but the next patch will need this!

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-by: Oleksandr Natalenko <oleksandr@redhat.com>
Message-id: 20200511183630.279750-2-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 util/aio-posix.c      |  1 +
 util/fdmon-io_uring.c | 13 ++++++++++---
 2 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/util/aio-posix.c b/util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ void aio_context_destroy(AioContext *ctx)
 {
     fdmon_io_uring_destroy(ctx);
     fdmon_epoll_disable(ctx);
+    aio_free_deleted_handlers(ctx);
 }
 
 void aio_context_set_poll_params(AioContext *ctx, int64_t max_ns,
diff --git a/util/fdmon-io_uring.c b/util/fdmon-io_uring.c
index XXXXXXX..XXXXXXX 100644
--- a/util/fdmon-io_uring.c
+++ b/util/fdmon-io_uring.c
@@ -XXX,XX +XXX,XX @@ void fdmon_io_uring_destroy(AioContext *ctx)
 
         io_uring_queue_exit(&ctx->fdmon_io_uring);
 
-        /* No need to submit these anymore, just free them. */
+        /* Move handlers due to be removed onto the deleted list */
         while ((node = QSLIST_FIRST_RCU(&ctx->submit_list))) {
+            unsigned flags = atomic_fetch_and(&node->flags,
+                    ~(FDMON_IO_URING_PENDING |
+                      FDMON_IO_URING_ADD |
+                      FDMON_IO_URING_REMOVE));
+
+            if (flags & FDMON_IO_URING_REMOVE) {
+                QLIST_INSERT_HEAD_RCU(&ctx->deleted_aio_handlers, node, node_deleted);
+            }
+
             QSLIST_REMOVE_HEAD_RCU(&ctx->submit_list, node_submitted);
-            QLIST_REMOVE(node, node);
-            g_free(node);
         }
 
         ctx->fdmon_ops = &fdmon_poll_ops;
-- 
2.25.3

The glib event loop does not call fdmon_io_uring_wait() so fd handlers
waiting to be submitted build up in the list. There is no benefit is
using io_uring when the glib GSource is being used, so disable it
instead of implementing a more complex fix.

This fixes a memory leak where AioHandlers would build up and increasing
amounts of CPU time were spent iterating them in aio_pending(). The
symptom is that guests become slow when QEMU is built with io_uring
support.

Buglink: https://bugs.launchpad.net/qemu/+bug/1877716
Fixes: 73fd282e7b6dd4e4ea1c3bbb3d302c8db51e4ccf ("aio-posix: add io_uring fd monitoring implementation")
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-by: Oleksandr Natalenko <oleksandr@redhat.com>
Message-id: 20200511183630.279750-3-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/block/aio.h |  3 +++
 util/aio-posix.c    | 12 ++++++++++++
 util/aio-win32.c    |  4 ++++
 util/async.c        |  1 +
 4 files changed, 20 insertions(+)

diff --git a/include/block/aio.h b/include/block/aio.h
index XXXXXXX..XXXXXXX 100644
--- a/include/block/aio.h
+++ b/include/block/aio.h
@@ -XXX,XX +XXX,XX @@ void aio_context_setup(AioContext *ctx);
  */
 void aio_context_destroy(AioContext *ctx);
 
+/* Used internally, do not call outside AioContext code */
+void aio_context_use_g_source(AioContext *ctx);
+
 /**
  * aio_context_set_poll_params:
  * @ctx: the aio context
diff --git a/util/aio-posix.c b/util/aio-posix.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -XXX,XX +XXX,XX @@ void aio_context_destroy(AioContext *ctx)
     aio_free_deleted_handlers(ctx);
 }
 
+void aio_context_use_g_source(AioContext *ctx)
+{
+    /*
+     * Disable io_uring when the glib main loop is used because it doesn't
+     * support mixed glib/aio_poll() usage. It relies on aio_poll() being
+     * called regularly so that changes to the monitored file descriptors are
+     * submitted, otherwise a list of pending fd handlers builds up.
+     */
+    fdmon_io_uring_destroy(ctx);
+    aio_free_deleted_handlers(ctx);
+}
+
 void aio_context_set_poll_params(AioContext *ctx, int64_t max_ns,
                                  int64_t grow, int64_t shrink, Error **errp)
 {
diff --git a/util/aio-win32.c b/util/aio-win32.c
index XXXXXXX..XXXXXXX 100644
--- a/util/aio-win32.c
+++ b/util/aio-win32.c
@@ -XXX,XX +XXX,XX @@ void aio_context_destroy(AioContext *ctx)
 {
 }
 
+void aio_context_use_g_source(AioContext *ctx)
+{
+}
+
 void aio_context_set_poll_params(AioContext *ctx, int64_t max_ns,
                                  int64_t grow, int64_t shrink, Error **errp)
 {
diff --git a/util/async.c b/util/async.c
index XXXXXXX..XXXXXXX 100644
--- a/util/async.c
+++ b/util/async.c
@@ -XXX,XX +XXX,XX @@ static GSourceFuncs aio_source_funcs = {
 
 GSource *aio_get_g_source(AioContext *ctx)
 {
+    aio_context_use_g_source(ctx);
     g_source_ref(&ctx->source);
     return &ctx->source;
 }
-- 
2.25.3

The following changes since commit b384cd95eb9c6f73ad84ed1bb0717a26e29cc78f:

Merge remote-tracking branch 'remotes/ehabkost/tags/machine-next-pull-request' into staging (2018-01-19 16:35:25 +0000)

are available in the Git repository at:

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

for you to fetch changes up to bcbb3866da19cce4360c828b6ec1c2a137757927:

block/parallels: add backing support to readv/writev (2018-01-22 14:02:33 +0000)

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

v2:
 * Drop merge failure from a previous pull request that broke virtio-blk on ARM
   guests
 * Add Parallels XML patch series

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

Edgar Kaziakhmedov (1):
  block/parallels: add backing support to readv/writev

Klim Kireev (4):
  docs/interop/prl-xml: description of Parallels Disk format
  configure: add dependency
  block/parallels: move some structures into header
  block/parallels: replace some magic numbers

Stefan Hajnoczi (1):
  block: add block_set_io_throttle virtio-blk-pci QMP example

docs/interop/prl-xml.txt | 158 +++++++++++++++++++++++++++++++++++++++++++++++
 qapi/block-core.json     |  18 ++++++
 configure                |  27 ++++++++
 block/Makefile.objs      |   2 +
 block/parallels.h        |  88 ++++++++++++++++++++++++++
 block/parallels.c        | 108 ++++++++++++++------------------
 scripts/checkpatch.pl    |   1 +
 7 files changed, 342 insertions(+), 60 deletions(-)
 create mode 100644 docs/interop/prl-xml.txt
 create mode 100644 block/parallels.h

-- 
2.14.3

The block_set_io_throttle command can look up BlockBackends by the
attached qdev device ID.  virtio-blk-pci is a special case because the
actual VirtIOBlock device is the "/virtio-backend" child of the PCI
adapter device.

Add a QMP schema example so clients will know how to use
block_set_io_throttle on the virtio-blk-pci device.

The alternative is to implement some sort of aliasing for qmp_get_blk()
but that is likely to cause confusion and could break future use cases.
Let's not go there.

Cc: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Message-id: 20180117090700.25811-1-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 qapi/block-core.json | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

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 @@
 # Example:
 #
 # -> { "execute": "block_set_io_throttle",
+#      "arguments": { "id": "virtio-blk-pci0/virtio-backend",
+#                     "bps": 0,
+#                     "bps_rd": 0,
+#                     "bps_wr": 0,
+#                     "iops": 512,
+#                     "iops_rd": 0,
+#                     "iops_wr": 0,
+#                     "bps_max": 0,
+#                     "bps_rd_max": 0,
+#                     "bps_wr_max": 0,
+#                     "iops_max": 0,
+#                     "iops_rd_max": 0,
+#                     "iops_wr_max": 0,
+#                     "bps_max_length": 0,
+#                     "iops_size": 0 } }
+# <- { "return": {} }
+#
+# -> { "execute": "block_set_io_throttle",
 #      "arguments": { "id": "ide0-1-0",
 #                     "bps": 1000000,
 #                     "bps_rd": 0,
-- 
2.14.3

From: Klim Kireev <klim.kireev@virtuozzo.com>

This patch adds main information about Parallels Disk
format, which consists of DiskDescriptor.xml and other files.

Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Message-id: 20180112090122.1702-2-klim.kireev@virtuozzo.com
CC: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 docs/interop/prl-xml.txt | 158 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 158 insertions(+)
 create mode 100644 docs/interop/prl-xml.txt

diff --git a/docs/interop/prl-xml.txt b/docs/interop/prl-xml.txt
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/docs/interop/prl-xml.txt
@@ -XXX,XX +XXX,XX @@
+= License =
+
+Copyright (c) 2015-2017, Virtuozzo, Inc.
+Authors:
+        2015 Denis Lunev <den@openvz.org>
+        2015 Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
+        2016-2017 Klim Kireev <klim.kireev@virtuozzo.com>
+        2016-2017 Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
+
+This work is licensed under the terms of the GNU GPL, version 2 or later.
+See the COPYING file in the top-level directory.
+
+This specification contains minimal information about Parallels Disk Format,
+which is enough to proper work with QEMU. Nevertheless, Parallels Cloud Server
+and Parallels Desktop are able to add some unspecified nodes to xml and use
+them, but they are for internal work and don't affect functionality. Also it
+uses auxiliary xml "Snapshot.xml", which allows to store optional snapshot
+information, but it doesn't influence open/read/write functionality. QEMU and
+other software should not use fields not covered in this document and
+Snapshot.xml file and must leave them as is.
+
+= Parallels Disk Format =
+
+Parallels disk consists of two parts: the set of snapshots and the disk
+descriptor file, which stores information about all files and snapshots.
+
+== Definitions ==
+    Snapshot       a record of the contents captured at a particular time,
+                   capable of storing current state. A snapshot has UUID and
+                   parent UUID.
+
+ Snapshot image    an overlay representing the difference between this
+                   snapshot and some earlier snapshot.
+
+    Overlay        an image storing the different sectors between two captured
+                   states.
+
+   Root image      snapshot image with no parent, the root of snapshot tree.
+
+    Storage        the backing storage for a subset of the virtual disk. When
+                   there is more than one storage in a Parallels disk then that
+                   is referred to as a split image. In this case every storage
+                   covers specific address space area of the disk and has its
+                   particular root image. Split images are not considered here
+                   and are not supported. Each storage consists of disk
+                   parameters and a list of images. The list of images always
+                   contains a root image and may also contain overlays. The
+                   root image can be an expandable Parallels image file or
+                   plain. Overlays must be expandable.
+
+  Description      DiskDescriptor.xml stores information about disk parameters,
+     file          snapshots, storages.
+
+     Top           The overlay between actual state and some previous snapshot.
+   Snapshot        It is not a snapshot in the classical sense because it
+                   serves as the active image that the guest writes to.
+
+    Sector         a 512-byte data chunk.
+
+== Description file ==
+All information is placed in a single XML element Parallels_disk_image.
+The element has only one attribute "Version", that must be 1.0.
+Schema of DiskDescriptor.xml:
+
+<Parallels_disk_image Version="1.0">
+    <Disk_Parameters>
+        ...
+    </Disk_Parameters>
+    <StorageData>
+        ...
+    </StorageData>
+    <Snapshots>
+        ...
+    </Snapshots>
+</Parallels_disk_image>
+
+== Disk_Parameters element ==
+The Disk_Parameters element describes the physical layout of the virtual disk
+and some general settings.
+
+The Disk_Parameters element MUST contain the following child elements:
+    * Disk_size - number of sectors in the disk,
+                  desired size of the disk.
+    * Cylinders - number of the disk cylinders.
+    * Heads     - number of the disk heads.
+    * Sectors   - number of the disk sectors per cylinder
+                  (sector size is 512 bytes)
+                  Limitation: Product of the Heads, Sectors and Cylinders
+                  values MUST be equal to the value of the Disk_size parameter.
+    * Padding   - must be 0. Parallels Cloud Server and Parallels Desktop may
+                  use padding set to 1, however this case is not covered
+                  by this spec, QEMU and other software should not open
+                  such disks and should not create them.
+
+== StorageData element ==
+This element of the file describes the root image and all snapshot images.
+
+The StorageData element consists of the Storage child element, as shown below:
+<StorageData>
+    <Storage>
+        ...
+    </Storage>
+</StorageData>
+
+A Storage element has following child elements:
+    * Start     - start sector of the storage, in case of non split storage
+                  equals to 0.
+    * End       - number of sector following the last sector, in case of non
+                  split storage equals to Disk_size.
+    * Blocksize - storage cluster size, number of sectors per one cluster.
+                  Cluster size for each "Compressed" (see below) image in
+                  parallels disk must be equal to this field. Note: cluster
+                  size for Parallels Expandable Image is in 'tracks' field of
+                  its header (see docs/interop/parallels.txt).
+    * Several Image child elements.
+
+Each Image element has following child elements:
+    * GUID - image identifier, UUID in curly brackets.
+             For instance, {12345678-9abc-def1-2345-6789abcdef12}.
+             The GUID is used by the Snapshots element to reference images
+             (see below)
+    * Type - image type of the element. It can be:
+             "Plain" for raw files.
+             "Compressed" for expanding disks.
+    * File - path to image file. Path can be relative to DiskDecriptor.xml or
+             absolute.
+
+== Snapshots element ==
+The Snapshots element describes the snapshot relations with the snapshot tree.
+
+The element contains the set of Shot child elements, as shown below:
+<Snapshots>
+    <TopGUID> ... </TopGUID> /* Optional child element */
+    <Shot>
+        ...
+    </Shot>
+    <Shot>
+        ...
+    </Shot>
+    ...
+</Snapshots>
+
+Each Shot element contains the following child elements:
+    * GUID       - an image GUID.
+    * ParentGUID - GUID of the image of the parent snapshot.
+
+The software may traverse snapshots from child to parent using <ParentGUID>
+field as reference. ParentGUID of root snapshot is
+{00000000-0000-0000-0000-000000000000}. There should be only one root
+snapshot. Top snapshot could be described via two ways: via TopGUID child
+element of the Snapshots element or via predefined GUID
+{5fbaabe3-6958-40ff-92a7-860e329aab41}. If TopGUID is defined, predefined GUID is
+interpreted as usual GUID. All snapshot images (except Top Snapshot) should be
+opened read-only. There is another predefined GUID,
+BackupID = {704718e1-2314-44c8-9087-d78ed36b0f4e}, which is used by original and
+some third-party software for backup, QEMU and other software may operate with
+images with GUID = BackupID as usual, however, it is not recommended to use this
+GUID for new disks. Top snapshot cannot have this GUID.
-- 
2.14.3

From: Klim Kireev <klim.kireev@virtuozzo.com>

This dependency is required for adequate Parallels images support.
Typically the disk consists of several images which are glued by
XML disk descriptor. Also XML hides inside several important parameters
which are not available in the image header.

The patch also adds clause to checkpatch.pl to understand libxml2 types.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-id: 20180112090122.1702-3-klim.kireev@virtuozzo.com
CC: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 configure             | 27 +++++++++++++++++++++++++++
 block/Makefile.objs   |  2 ++
 scripts/checkpatch.pl |  1 +
 3 files changed, 30 insertions(+)

diff --git a/configure b/configure
index XXXXXXX..XXXXXXX 100755
--- a/configure
+++ b/configure
@@ -XXX,XX +XXX,XX @@ tcmalloc="no"
 jemalloc="no"
 replication="yes"
 vxhs=""
+libxml2=""
 
 supported_cpu="no"
 supported_os="no"
@@ -XXX,XX +XXX,XX @@ for opt do
   ;;
   --enable-numa) numa="yes"
   ;;
+  --disable-libxml2) libxml2="no"
+  ;;
+  --enable-libxml2) libxml2="yes"
+  ;;
   --disable-tcmalloc) tcmalloc="no"
   ;;
   --enable-tcmalloc) tcmalloc="yes"
@@ -XXX,XX +XXX,XX @@ disabled with --disable-FEATURE, default is enabled if available:
   tpm             TPM support
   libssh2         ssh block device support
   numa            libnuma support
+  libxml2         for Parallels image format
   tcmalloc        tcmalloc support
   jemalloc        jemalloc support
   replication     replication support
@@ -XXX,XX +XXX,XX @@ EOF
   fi
 fi
 
+##########################################
+# libxml2 probe
+if test "$libxml2" != "no" ; then
+    if $pkg_config --exists libxml-2.0; then
+        libxml2="yes"
+        libxml2_cflags=$($pkg_config --cflags libxml-2.0)
+        libxml2_libs=$($pkg_config --libs libxml-2.0)
+    else
+        if test "$libxml2" = "yes"; then
+            feature_not_found "libxml2" "Install libxml2 devel"
+        fi
+        libxml2="no"
+    fi
+fi
 
 ##########################################
 # glusterfs probe
@@ -XXX,XX +XXX,XX @@ echo "lzo support       $lzo"
 echo "snappy support    $snappy"
 echo "bzip2 support     $bzip2"
 echo "NUMA host support $numa"
+echo "libxml2           $libxml2"
 echo "tcmalloc support  $tcmalloc"
 echo "jemalloc support  $jemalloc"
 echo "avx2 optimization $avx2_opt"
@@ -XXX,XX +XXX,XX @@ if test "$have_rtnetlink" = "yes" ; then
   echo "CONFIG_RTNETLINK=y" >> $config_host_mak
 fi
 
+if test "$libxml2" = "yes" ; then
+  echo "CONFIG_LIBXML2=y" >> $config_host_mak
+  echo "LIBXML2_CFLAGS=$libxml2_cflags" >> $config_host_mak
+  echo "LIBXML2_LIBS=$libxml2_libs" >> $config_host_mak
+fi
+
 if test "$replication" = "yes" ; then
   echo "CONFIG_REPLICATION=y" >> $config_host_mak
 fi
diff --git a/block/Makefile.objs b/block/Makefile.objs
index XXXXXXX..XXXXXXX 100644
--- a/block/Makefile.objs
+++ b/block/Makefile.objs
@@ -XXX,XX +XXX,XX @@ block-obj-$(if $(CONFIG_BZIP2),m,n) += dmg-bz2.o
 dmg-bz2.o-libs     := $(BZIP2_LIBS)
 qcow.o-libs        := -lz
 linux-aio.o-libs   := -laio
+parallels.o-cflags := $(LIBXML2_CFLAGS)
+parallels.o-libs   := $(LIBXML2_LIBS)
diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
index XXXXXXX..XXXXXXX 100755
--- a/scripts/checkpatch.pl
+++ b/scripts/checkpatch.pl
@@ -XXX,XX +XXX,XX @@ our @typeList = (
 	qr{${Ident}_handler_fn},
 	qr{target_(?:u)?long},
 	qr{hwaddr},
+	qr{xml${Ident}},
 );
 
 # This can be modified by sub possible.  Since it can be empty, be careful
-- 
2.14.3

From: Klim Kireev <klim.kireev@virtuozzo.com>

To implement xml format, some defines and structures
from parallels.c are required.

Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
Message-id: 20180112090122.1702-4-klim.kireev@virtuozzo.com
CC: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/parallels.h | 86 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 block/parallels.c | 53 +---------------------------------
 2 files changed, 87 insertions(+), 52 deletions(-)
 create mode 100644 block/parallels.h

diff --git a/block/parallels.h b/block/parallels.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/block/parallels.h
@@ -XXX,XX +XXX,XX @@
+/*
+* Block driver for Parallels disk image format
+*
+* Copyright (c) 2015-2017 Virtuozzo, Inc.
+* Authors:
+*         2016-2017 Klim S. Kireev <klim.kireev@virtuozzo.com>
+*         2015 Denis V. Lunev <den@openvz.org>
+*
+* This code was originally based on comparing different disk images created
+* by Parallels. Currently it is based on opened OpenVZ sources
+* available at
+*     https://github.com/OpenVZ/ploop
+*
+* Permission is hereby granted, free of charge, to any person obtaining a copy
+* of this software and associated documentation files (the "Software"), to deal
+* in the Software without restriction, including without limitation the rights
+* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+* copies of the Software, and to permit persons to whom the Software is
+* furnished to do so, subject to the following conditions:
+*
+* The above copyright notice and this permission notice shall be included in
+* all copies or substantial portions of the Software.
+*
+* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+* THE SOFTWARE.
+*/
+#ifndef BLOCK_PARALLELS_H
+#define BLOCK_PARALLELS_H
+#include "qemu/coroutine.h"
+#include "qemu/typedefs.h"
+
+#define DEFAULT_CLUSTER_SIZE 1048576        /* 1 MiB */
+
+/* always little-endian */
+typedef struct ParallelsHeader {
+    char magic[16]; /* "WithoutFreeSpace" */
+    uint32_t version;
+    uint32_t heads;
+    uint32_t cylinders;
+    uint32_t tracks;
+    uint32_t bat_entries;
+    uint64_t nb_sectors;
+    uint32_t inuse;
+    uint32_t data_off;
+    char padding[12];
+} QEMU_PACKED ParallelsHeader;
+
+typedef enum ParallelsPreallocMode {
+    PRL_PREALLOC_MODE_FALLOCATE = 0,
+    PRL_PREALLOC_MODE_TRUNCATE = 1,
+    PRL_PREALLOC_MODE__MAX = 2,
+} ParallelsPreallocMode;
+
+typedef struct BDRVParallelsState {
+    /** Locking is conservative, the lock protects
+     *   - image file extending (truncate, fallocate)
+     *   - any access to block allocation table
+     */
+    CoMutex lock;
+
+    ParallelsHeader *header;
+    uint32_t header_size;
+    bool header_unclean;
+
+    unsigned long *bat_dirty_bmap;
+    unsigned int  bat_dirty_block;
+
+    uint32_t *bat_bitmap;
+    unsigned int bat_size;
+
+    int64_t  data_end;
+    uint64_t prealloc_size;
+    ParallelsPreallocMode prealloc_mode;
+
+    unsigned int tracks;
+
+    unsigned int off_multiplier;
+    Error *migration_blocker;
+} BDRVParallelsState;
+
+#endif
diff --git a/block/parallels.c b/block/parallels.c
index XXXXXXX..XXXXXXX 100644
--- a/block/parallels.c
+++ b/block/parallels.c
@@ -XXX,XX +XXX,XX @@
 #include "qemu/bswap.h"
 #include "qemu/bitmap.h"
 #include "migration/blocker.h"
+#include "parallels.h"
 
 /**************************************************************/
 
@@ -XXX,XX +XXX,XX @@
 #define HEADER_INUSE_MAGIC  (0x746F6E59)
 #define MAX_PARALLELS_IMAGE_FACTOR (1ull << 32)
 
-#define DEFAULT_CLUSTER_SIZE 1048576        /* 1 MiB */
-
-
-// always little-endian
-typedef struct ParallelsHeader {
-    char magic[16]; // "WithoutFreeSpace"
-    uint32_t version;
-    uint32_t heads;
-    uint32_t cylinders;
-    uint32_t tracks;
-    uint32_t bat_entries;
-    uint64_t nb_sectors;
-    uint32_t inuse;
-    uint32_t data_off;
-    char padding[12];
-} QEMU_PACKED ParallelsHeader;
-
-
-typedef enum ParallelsPreallocMode {
-    PRL_PREALLOC_MODE_FALLOCATE = 0,
-    PRL_PREALLOC_MODE_TRUNCATE = 1,
-    PRL_PREALLOC_MODE__MAX = 2,
-} ParallelsPreallocMode;
-
 static QEnumLookup prealloc_mode_lookup = {
     .array = (const char *const[]) {
         "falloc",
@@ -XXX,XX +XXX,XX @@ static QEnumLookup prealloc_mode_lookup = {
     .size = PRL_PREALLOC_MODE__MAX
 };
 
-typedef struct BDRVParallelsState {
-    /** Locking is conservative, the lock protects
-     *   - image file extending (truncate, fallocate)
-     *   - any access to block allocation table
-     */
-    CoMutex lock;
-
-    ParallelsHeader *header;
-    uint32_t header_size;
-    bool header_unclean;
-
-    unsigned long *bat_dirty_bmap;
-    unsigned int  bat_dirty_block;
-
-    uint32_t *bat_bitmap;
-    unsigned int bat_size;
-
-    int64_t  data_end;
-    uint64_t prealloc_size;
-    ParallelsPreallocMode prealloc_mode;
-
-    unsigned int tracks;
-
-    unsigned int off_multiplier;
-    Error *migration_blocker;
-} BDRVParallelsState;
-
-
 #define PARALLELS_OPT_PREALLOC_MODE     "prealloc-mode"
 #define PARALLELS_OPT_PREALLOC_SIZE     "prealloc-size"
 
-- 
2.14.3

From: Klim Kireev <klim.kireev@virtuozzo.com>

Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
Message-id: 20180112090122.1702-5-klim.kireev@virtuozzo.com
CC: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/parallels.h | 2 ++
 block/parallels.c | 5 +++--
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/block/parallels.h b/block/parallels.h
index XXXXXXX..XXXXXXX 100644
--- a/block/parallels.h
+++ b/block/parallels.h
@@ -XXX,XX +XXX,XX @@
 #include "qemu/coroutine.h"
 #include "qemu/typedefs.h"
 
+#define HEADS_NUMBER 16
+#define SEC_IN_CYL 32
 #define DEFAULT_CLUSTER_SIZE 1048576        /* 1 MiB */
 
 /* always little-endian */
diff --git a/block/parallels.c b/block/parallels.c
index XXXXXXX..XXXXXXX 100644
--- a/block/parallels.c
+++ b/block/parallels.c
@@ -XXX,XX +XXX,XX @@ static int parallels_create(const char *filename, QemuOpts *opts, Error **errp)
     memcpy(header.magic, HEADER_MAGIC2, sizeof(header.magic));
     header.version = cpu_to_le32(HEADER_VERSION);
     /* don't care much about geometry, it is not used on image level */
-    header.heads = cpu_to_le32(16);
-    header.cylinders = cpu_to_le32(total_size / BDRV_SECTOR_SIZE / 16 / 32);
+    header.heads = cpu_to_le32(HEADS_NUMBER);
+    header.cylinders = cpu_to_le32(total_size / BDRV_SECTOR_SIZE
+                                   / HEADS_NUMBER / SEC_IN_CYL);
     header.tracks = cpu_to_le32(cl_size >> BDRV_SECTOR_BITS);
     header.bat_entries = cpu_to_le32(bat_entries);
     header.nb_sectors = cpu_to_le64(DIV_ROUND_UP(total_size, BDRV_SECTOR_SIZE));
-- 
2.14.3

From: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>

Since parallels format supports backing files, refine
readv/writev (allocate_clusters) to redirect read/write requests
to a backing file (if cluster is not available in the current bs).

Signed-off-by: Edgar Kaziakhmedov <edgar.kaziakhmedov@virtuozzo.com>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Klim Kireev <klim.kireev@virtuozzo.com>
Message-id: 20180112090122.1702-6-klim.kireev@virtuozzo.com
CC: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 block/parallels.c | 50 ++++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 44 insertions(+), 6 deletions(-)

diff --git a/block/parallels.c b/block/parallels.c
index XXXXXXX..XXXXXXX 100644
--- a/block/parallels.c
+++ b/block/parallels.c
@@ -XXX,XX +XXX,XX @@ static int64_t block_status(BDRVParallelsState *s, int64_t sector_num,
 static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
                                  int nb_sectors, int *pnum)
 {
+    int ret;
     BDRVParallelsState *s = bs->opaque;
     int64_t pos, space, idx, to_allocate, i, len;
 
@@ -XXX,XX +XXX,XX @@ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
         return len;
     }
     if (s->data_end + space > (len >> BDRV_SECTOR_BITS)) {
-        int ret;
         space += s->prealloc_size;
         if (s->prealloc_mode == PRL_PREALLOC_MODE_FALLOCATE) {
             ret = bdrv_pwrite_zeroes(bs->file,
@@ -XXX,XX +XXX,XX @@ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
         }
     }
 
+    /* Try to read from backing to fill empty clusters
+     * FIXME: 1. previous write_zeroes may be redundant
+     *        2. most of data we read from backing will be rewritten by
+     *           parallels_co_writev. On aligned-to-cluster write we do not need
+     *           this read at all.
+     *        3. it would be good to combine write of data from backing and new
+     *           data into one write call */
+    if (bs->backing) {
+        int64_t nb_cow_sectors = to_allocate * s->tracks;
+        int64_t nb_cow_bytes = nb_cow_sectors << BDRV_SECTOR_BITS;
+        QEMUIOVector qiov;
+        struct iovec iov = {
+            .iov_len = nb_cow_bytes,
+            .iov_base = qemu_blockalign(bs, nb_cow_bytes)
+        };
+        qemu_iovec_init_external(&qiov, &iov, 1);
+
+        ret = bdrv_co_readv(bs->backing, idx * s->tracks, nb_cow_sectors,
+                            &qiov);
+        if (ret < 0) {
+            qemu_vfree(iov.iov_base);
+            return ret;
+        }
+
+        ret = bdrv_co_writev(bs->file, s->data_end, nb_cow_sectors, &qiov);
+        qemu_vfree(iov.iov_base);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
     for (i = 0; i < to_allocate; i++) {
         s->bat_bitmap[idx + i] = cpu_to_le32(s->data_end / s->off_multiplier);
         s->data_end += s->tracks;
@@ -XXX,XX +XXX,XX @@ static coroutine_fn int parallels_co_readv(BlockDriverState *bs,
 
         nbytes = n << BDRV_SECTOR_BITS;
 
+        qemu_iovec_reset(&hd_qiov);
+        qemu_iovec_concat(&hd_qiov, qiov, bytes_done, nbytes);
+
         if (position < 0) {
-            qemu_iovec_memset(qiov, bytes_done, 0, nbytes);
+            if (bs->backing) {
+                ret = bdrv_co_readv(bs->backing, sector_num, n, &hd_qiov);
+                if (ret < 0) {
+                    break;
+                }
+            } else {
+                qemu_iovec_memset(&hd_qiov, 0, 0, nbytes);
+            }
         } else {
-            qemu_iovec_reset(&hd_qiov);
-            qemu_iovec_concat(&hd_qiov, qiov, bytes_done, nbytes);
-
             ret = bdrv_co_readv(bs->file, position, n, &hd_qiov);
             if (ret < 0) {
                 break;
@@ -XXX,XX +XXX,XX @@ static BlockDriver bdrv_parallels = {
     .bdrv_co_flush_to_os      = parallels_co_flush_to_os,
     .bdrv_co_readv  = parallels_co_readv,
     .bdrv_co_writev = parallels_co_writev,
-
+    .supports_backing = true,
     .bdrv_create    = parallels_create,
     .bdrv_check     = parallels_check,
     .create_opts    = &parallels_create_opts,
-- 
2.14.3