Merge remote-tracking branch 'mreitz/tags/pull-block-2017-09-26' into queue-block

Block patches

# gpg: Signature made Tue Sep 26 15:01:00 2017 CEST
# gpg:                using RSA key F407DB0061D5CF40
# gpg: Good signature from "Max Reitz <mreitz@redhat.com>"
# Primary key fingerprint: 91BE B60A 30DB 3E88 57D1  1829 F407 DB00 61D5 CF40

* mreitz/tags/pull-block-2017-09-26:
  block/qcow2-bitmap: fix use of uninitialized pointer
  qemu-iotests: add shrinking image test
  qcow2: add shrink image support
  qcow2: add qcow2_cache_discard
  qemu-img: add --shrink flag for resize

Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/qcow2-bitmap: fix use of uninitialized pointer

Without initialization to zero dirty_bitmap field may be not zero
for a bitmap which should not be stored and
qcow2_store_persistent_dirty_bitmaps will erroneously call
store_bitmap for it which leads to SIGSEGV on bdrv_dirty_bitmap_name.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-id: 20170922144353.4220-1-vsementsov@virtuozzo.com
Cc: qemu-stable@nongnu.org
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>

qemu-iotests: add shrinking image test

Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 20170918124230.8152-5-pbutsykin@virtuozzo.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

qcow2: add shrink image support

This patch add shrinking of the image file for qcow2. As a result, this allows
us to reduce the virtual image size and free up space on the disk without
copying the image. Image can be fragmented and shrink is done by punching holes
in the image file.

Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 20170918124230.8152-4-pbutsykin@virtuozzo.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

qcow2: add qcow2_cache_discard

Whenever l2/refcount table clusters are discarded from the file we can
automatically drop unnecessary content of the cache tables. This reduces
the chance of eviction useful cache data and eliminates inconsistent data
in the cache with the data in the file.

Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 20170918124230.8152-3-pbutsykin@virtuozzo.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

qemu-img: add --shrink flag for resize

The flag is additional precaution against data loss. Perhaps in the future the
operation shrink without this flag will be blocked for all formats, but for now
we need to maintain compatibility with raw.

Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 20170918124230.8152-2-pbutsykin@virtuozzo.com
[mreitz: Added a missing space to a warning]
Signed-off-by: Max Reitz <mreitz@redhat.com>

iotests: fix 181: enable postcopy-ram capability on target

Migration capabilities should be enabled on both source and
destination qemu processes.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: Test change-backing-file command

This involves a temporary read-write reopen if the backing file link in
the middle of a backing file chain should be changed and is therefore a
good test for the latest bdrv_reopen() vs. op blockers fixes.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Fix permissions after bdrv_reopen()

If we switch between read-only and read-write, the permissions that
image format drivers need on bs->file change, too. Make sure to update
the permissions during bdrv_reopen().

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

block: reopen: Queue children after their parents

We will calculate the required new permissions in the prepare stage of a
reopen. Required permissions of children can be influenced by the
changes made to their parents, but parents are independent from their
children. This means that permissions need to be calculated top-down. In
order to achieve this, queue parents before their children rather than
queuing the children first.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

block: Base permissions on rw state after reopen

When new permissions are calculated during bdrv_reopen(), they need to
be based on the state of the graph as it will be after the reopen has
completed, not on the current state of the involved nodes.

This patch makes bdrv_is_writable() optionally accept a BlockReopenQueue
from which the new flags are taken. This is then used for determining
the new bs->file permissions of format drivers as soon as we add the
code to actually pass a non-NULL reopen queue to the .bdrv_child_perm
callbacks.

While moving bdrv_is_writable(), make it static. It isn't used outside
block.c.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

block: Add reopen queue to bdrv_check_perm()

In the context of bdrv_reopen(), we'll have to look at the state of the
graph as it will be after the reopen. This interface addition is in
preparation for the change.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

block: Add reopen_queue to bdrv_child_perm()

In the context of bdrv_reopen(), we'll have to look at the state of the
graph as it will be after the reopen. This interface addition is in
preparation for the change.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

qemu-io: Drop write permissions before read-only reopen

qemu-io provides a 'reopen' command that allows switching from writable
to read-only access. We need to make sure that we don't try to keep
write permissions to a BlockBackend that becomes read-only, otherwise
things are going to fail.

This requires a bdrv_drain() call because otherwise in-flight AIO
write requests could issue new internal requests while the permission
has already gone away, which would cause assertion failures. Draining
the queue doesn't break AIO requests in any new way, bdrv_reopen() would
drain it anyway only a few lines later.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>

block: Clean up some bad code in the vvfat driver

Remove the unnecessary home-grown redefinition of the assert() macro here,
and remove the unusable debug code at the end of the checkpoint() function.
The code there uses assert() with side-effects (assignment to the "mapping"
variable), which should be avoided. Looking more closely, it seems as it is
apparently also only usable for one certain directory layout (with a file
named USB.H in it) and thus is of no use for the rest of the world.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/throttle-groups.c: allocate RestartData on the heap

RestartData is the opaque data of the throttle_group_restart_queue_entry
coroutine. By being stack allocated, it isn't available anymore if
aio_co_enter schedules the coroutine with a bottom half and runs after
throttle_group_restart_queue returns.

Cc: qemu-stable@nongnu.org
Signed-off-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

throttle: Assert that bkt->max is valid in throttle_compute_wait()

If bkt->max == 0 and bkt->burst_length > 1 then we could have a
division by 0 in throttle_do_compute_wait(). That configuration is
however not permitted and is already detected by throttle_is_valid(),
but let's assert it in throttle_compute_wait() to make it explicit.

Found by Coverity (CID: 1381016).

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

iotests: Print full path of bad output if mismatch

So it is easier to copy paste the path.

Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

iotests: use virtio aliases for 067

The default cpu model on s390x does not provide zPCI, which is
not yet wired up on tcg. Moreover, virtio-ccw is the standard
on s390x.

Using virtio-scsi will implicitly pick the right device, so just
switch to that for simplicity.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: QingFeng Hao <haoqf@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

iotests: use -ccw on s390x for 051

The default cpu model on s390x does not provide zPCI, which is
not yet wired up on tcg. Moreover, virtio-ccw is the standard
on s390x, so use the -ccw instead of the -pci versions of virtio
devices on s390x.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: QingFeng Hao <haoqf@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

iotests: use -ccw on s390x for 040, 139, and 182

The default cpu model on s390x does not provide zPCI, which is
not yet wired up on tcg. Moreover, virtio-ccw is the standard
on s390x, so use the -ccw instead of the -pci versions of virtio
devices on s390x.

Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: QingFeng Hao <haoqf@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

docs: add qemu-block-drivers(7) man page

Block driver documentation is available in qemu-doc.html.  It would be
convenient to have documentation for formats, protocols, and filter
drivers in a man page.

Extract the relevant part of qemu-doc.html into a new file called
docs/qemu-block-drivers.texi.  This file can also be built as a
stand-alone document (man, html, etc).

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

file-posix: Clear out first sector in hdev_create

People get surprised when, after "qemu-img create -f raw /dev/sdX", they
still see qcow2 with "qemu-img info", if previously the bdev had a qcow2
header. While this is natural because raw doesn't need to write any
magic bytes during creation, hdev_create is free to clear out the first
sector to make sure the stale qcow2 header doesn't cause such confusion.

Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-img: Clarify about relative backing file options

It's not too surprising when a user specifies the backing file relative
to the current working directory instead of the top layer image. This
causes error when they differ. Though the error message has enough
information to infer the fact about the misunderstanding, it is better
if we document this explicitly, so that users don't have to learn from
mistakes.

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: Add missing -machine accel=qtest

A basic set of qemu options is initialised in ./common:

    export QEMU_OPTIONS="-nodefaults -machine accel=qtest"

However, two test cases (172 and 186) overwrite QEMU_OPTIONS and neglect
to manually set '-machine accel=qtest'. Add the missing option for 172.
186 probably only copied the code from 172, it doesn't actually need to
overwrite QEMU_OPTIONS, so remove that in 186.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Tested-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>

Merge remote-tracking branch 'remotes/thibault/tags/samuel-thibault' into staging

slirp updates

# gpg: Signature made Sun 24 Sep 2017 19:07:51 BST
# gpg:                using RSA key 0x9E511E01C737F075
# gpg: Good signature from "Samuel Thibault <samuel.thibault@aquilenet.fr>"
# gpg:                 aka "Samuel Thibault <sthibault@debian.org>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@gnu.org>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@inria.fr>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@labri.fr>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@ens-lyon.org>"
# gpg:                 aka "Samuel Thibault <samuel.thibault@u-bordeaux.fr>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 900C B024 B679 31D4 0F82  304B D017 8C76 7D06 9EE6
#      Subkey fingerprint: 9A37 3D36 64A8 DC62 DA0A  34FD 9E51 1E01 C737 F075

* remotes/thibault/tags/samuel-thibault:
  slirp: Add a special case for the NULL socket
  slirp: Fix intermittent send queue hangs on a socket
  slirp: Add explanation for hostfwd parsing failure

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

slirp: Add a special case for the NULL socket

NULL sockets are used for NDP, BOOTP, and other critical operations.
If the topmost mbuf in a NULL session is blocked pending resolution,
it may cause problems if it blocks other packets with a NULL socket.
So do not add mbufs with a NULL socket field to the same session.

Signed-off-by: Kevin Cernekee <cernekee@chromium.org>
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>

slirp: Fix intermittent send queue hangs on a socket

if_output() originally sent one mbuf per call and used the slirp->next_m
variable to keep track of where it left off.  But nowadays it tries to
send all of the mbufs from the fastq, and one mbuf from each session on
the batchq.  The next_m variable is both redundant and harmful: there is
a case[0] involving delayed packets in which next_m ends up pointing
to &slirp->if_batchq when an active session still exists, and this
blocks all traffic for that session until qemu is restarted.

The test case was created to reproduce a problem that was seen on
long-running Chromium OS VM tests[1] which rapidly create and
destroy ssh connections through hostfwd.

[0] https://pastebin.com/NNy6LreF
[1] https://bugs.chromium.org/p/chromium/issues/detail?id=766323

Signed-off-by: Kevin Cernekee <cernekee@chromium.org>
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>

slirp: Add explanation for hostfwd parsing failure

e.g.
./x86_64-softmmu/qemu-system-x86_64 -nographic -netdev 'user,id=vnet,hostfwd=:555.0.0.0:0-:22'
qemu-system-x86_64: -netdev user,id=vnet,hostfwd=:555.0.0.0:0-:22: Invalid host forwarding rule ':555.0.0.0:0-:22' (Bad host address)

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>

Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

* Speed up AddressSpaceDispatch creation (Alexey)
* Fix kvm.c assert (David)
* Memory fixes and further speedup (me)
* Persistent reservation manager infrastructure (me)
* virtio-serial: add enable_backend callback (Pavel)
* chardev GMainContext fixes (Peter)

# gpg: Signature made Fri 22 Sep 2017 20:07:33 BST
# gpg:                using RSA key 0xBFFBD25F78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>"
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini/tags/for-upstream: (32 commits)
  chardev: remove context in chr_update_read_handler
  chardev: use per-dev context for io_add_watch_poll
  chardev: add Chardev.gcontext field
  chardev: new qemu_chr_be_update_read_handlers()
  scsi: add persistent reservation manager using qemu-pr-helper
  scsi: add multipath support to qemu-pr-helper
  scsi: build qemu-pr-helper
  scsi, file-posix: add support for persistent reservation management
  memory: Share special empty FlatView
  memory: seek FlatView sharing candidates among children subregions
  memory: trace FlatView creation and destruction
  memory: Create FlatView directly
  memory: Get rid of address_space_init_shareable
  memory: Rework "info mtree" to print flat views and dispatch trees
  memory: Do not allocate FlatView in address_space_init
  memory: Share FlatView's and dispatch trees between address spaces
  memory: Move address_space_update_ioeventfds
  memory: Alloc dispatch tree where topology is generared
  memory: Store physical root MR in FlatView
  memory: Rename mem_begin/mem_commit/mem_add helpers
  ...

# Conflicts:
# configure

chardev: remove context in chr_update_read_handler

We had a per-chardev cache for context, then we don't need this
parameter to be passed in every time when chr_update_read_handler()
called.  As long as we are calling chr_update_read_handler() using
qemu_chr_be_update_read_handlers() we'll be fine.

Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1505975754-21555-5-git-send-email-peterx@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

chardev: use per-dev context for io_add_watch_poll

It was only passed in by chr_update_read_handlers().  However when
reconnect, we'll lose that context information.  So if a chardev was
running on another context (rather than the default context, the NULL
pointer), it'll switch back to the default context if reconnection
happens.  But, it should really stick to the old context.

Convert all the callers of io_add_watch_poll() to use the internally
cached gcontext.  Then the context should be able to survive even after
reconnections.

Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1505975754-21555-4-git-send-email-peterx@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

chardev: add Chardev.gcontext field

It caches the gcontext that is used to poll the chardev IO.  Before this
patch, we only passed it in via chr_update_read_handlers().  However
that may not be enough if the char backend is disconnected and
reconnected afterward.  There are chardev codes that still assumed the
context be NULL (which is the main context).  Will fix that up in
following up patches.

Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1505975754-21555-3-git-send-email-peterx@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

chardev: new qemu_chr_be_update_read_handlers()

Add a wrapper for the chr_update_read_handler().

Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1505975754-21555-2-git-send-email-peterx@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

scsi: add persistent reservation manager using qemu-pr-helper

This adds a concrete subclass of pr-manager that talks to qemu-pr-helper.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

scsi: add multipath support to qemu-pr-helper

Proper support of persistent reservation for multipath devices requires
communication with the multipath daemon, so that the reservation is
registered and applied when a path comes up.  The device mapper
utilities provide a library to do so; this patch makes qemu-pr-helper.c
detect multipath devices and, when one is found, delegate the operation
to libmpathpersist.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

scsi: build qemu-pr-helper

Introduce a privileged helper to run persistent reservation commands.
This lets virtual machines send persistent reservations without using
CAP_SYS_RAWIO or out-of-tree patches.  The helper uses Unix permissions
and SCM_RIGHTS to restrict access to processes that can access its socket
and prove that they have an open file descriptor for a raw SCSI device.

The next patch will also correct the usage of persistent reservations
with multipath devices.

It would also be possible to support for Linux's IOC_PR_* ioctls in
the future, to support NVMe devices.  For now, however, only SCSI is
supported.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge remote-tracking branch 'remotes/ehabkost/tags/python-next-pull-request' into staging

Python queue, 2017-09-22

* MAINTAINERS update
* Fix logging issue on test scripts using qemu.py

# gpg: Signature made Fri 22 Sep 2017 15:41:43 BST
# gpg:                using RSA key 0x2807936F984DC5A6
# gpg: Good signature from "Eduardo Habkost <ehabkost@redhat.com>"
# Primary key fingerprint: 5A32 2FD5 ABC4 D3DB ACCF  D1AA 2807 936F 984D C5A6

* remotes/ehabkost/tags/python-next-pull-request:
  MAINTAINERS: Add Python scripts
  qemu.py: Call logging.basicConfig() automatically

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

MAINTAINERS: Add Python scripts

Cleber and I are volunteering to review and queue patches for the
Python scripts and modules in scripts/.

I'm setting "M: Odd fixes" because not all scripts are actively
maintained.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170915230744.22942-1-ehabkost@redhat.com>
[ehabkost: add tests/*.py too]
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>

qemu.py: Call logging.basicConfig() automatically

Not all scripts using qemu.py configure the Python logging
module, and end up generating a "No handlers could be found for
logger" message instead of actual log messages.

To avoid requiring every script using qemu.py to configure
logging manually, call basicConfig() when creating a QEMUMachine
object.  This won't affect scripts that already set up logging,
but will ensure that scripts that don't configure logging keep
working.

Reported-by: Kevin Wolf <kwolf@redhat.com>
Fixes: 4738b0a85a0c2031fddc71b51cccebce0c4ba6b1
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170921162234.847-1-ehabkost@redhat.com>
Reviewed-by: Cleber Rosa <crosa@redhat.com>
Acked-by: Lukáš Doktor <ldoktor@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>

Merge remote-tracking branch 'remotes/juanquintela/tags/migration/20170922-1' into staging

migration/next for 20170922

# gpg: Signature made Fri 22 Sep 2017 13:15:06 BST
# gpg:                using RSA key 0xF487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>"
# gpg:                 aka "Juan Quintela <quintela@trasno.org>"
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* remotes/juanquintela/tags/migration/20170922-1:
  migration: split ufd_version_check onto receive/request features part
  migration: fix hardcoded function name in error report
  migration: pass MigrationIncomingState* into migration check functions
  migration: split common postcopy out of ram postcopy
  migration: fix ram_save_pending
  migration: add has_postcopy savevm handler
  bitmap: provide to_le/from_le helpers
  bitmap: introduce bitmap_count_one()
  bitmap: remove BITOP_WORD()
  migration: Split migration_fd_process_incoming
  migration: Create multifd migration threads
  migration: Create x-multifd-page-count parameter
  migration: Create x-multifd-channels parameter
  migration: Add multifd capability
  migration: Create migration_has_all_channels
  migration: Add comments to channel functions
  migration: Teach it about G_SOURCE_REMOVE
  migration: Create migration_ioc_process_incoming()

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

ide: fix enum comparison for gcc 4.7

Apparently GCC gets bent over comparing enum values against zero.
Replace the conditional with something less readable.

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20170921013821.1673-1-jsnow@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

migration: split ufd_version_check onto receive/request features part

This modification is necessary for userfault fd features which are
required to be requested from userspace.
UFFD_FEATURE_THREAD_ID is a one of such "on demand" feature, which will
be introduced in the next patch.

QEMU have to use separate userfault file descriptor, due to
userfault context has internal state, and after first call of
ioctl UFFD_API it changes its state to UFFD_STATE_RUNNING (in case of
success), but kernel while handling ioctl UFFD_API expects UFFD_STATE_WAIT_API.
So only one ioctl with UFFD_API is possible per ufd.

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Alexey Perevalov <a.perevalov@samsung.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: fix hardcoded function name in error report

Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Alexey Perevalov <a.perevalov@samsung.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: pass MigrationIncomingState* into migration check functions

That tiny refactoring is necessary to be able to set
UFFD_FEATURE_THREAD_ID while requesting features, and then
to create downtime context in case when kernel supports it.

Signed-off-by: Alexey Perevalov <a.perevalov@samsung.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: split common postcopy out of ram postcopy

Split common postcopy staff from ram postcopy staff.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: fix ram_save_pending

Fill postcopy-able pending only if ram postcopy is enabled.
It is necessary because of there will be other postcopy-able states and
when ram postcopy is disabled, it should not spoil common postcopy
related pending.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: add has_postcopy savevm handler

Now postcopy-able states are recognized by not NULL
save_live_complete_postcopy handler. But when we have several different
postcopy-able states, it is not convenient. Ram postcopy may be
disabled, while some other postcopy enabled, in this case Ram state
should behave as it is not postcopy-able.

This patch add separate has_postcopy handler to specify behaviour of
savevm state.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

bitmap: provide to_le/from_le helpers

Provide helpers to convert bitmaps to little endian format. It can be
used when we want to send one bitmap via network to some other hosts.

One thing to mention is that, these helpers only solve the problem of
endianess, but it does not solve the problem of different word size on
machines (the bitmaps managing same count of bits may contains different
size when malloced). So we need to take care of the size alignment issue
on the callers for now.

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

bitmap: introduce bitmap_count_one()

Count how many bits set in the bitmap.

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

bitmap: remove BITOP_WORD()

We have BIT_WORD(). It's the same.

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

migration: Split migration_fd_process_incoming

We need that on later patches.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>

migration: Create multifd migration threads

Creation of the threads, nothing inside yet.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
--

Use pointers instead of long array names
Move to use semaphores instead of conditions as paolo suggestion

Put all the state inside one struct.
Use a counter for the number of threads created.  Needed during cancellation.

Add error return to thread creation

Add id field

Rename functions to multifd_save/load_setup/cleanup
Change recv parameters to a pointer to struct
Change back to a struct
Use Error * for _cleanup

migration: Create x-multifd-page-count parameter

Indicates how many pages we are going to send in each batch to a multifd
thread.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
--

Be consistent with defaults and documentation
Use new DEFINE_PROP_*
Rename x-multifd-group to x-multifd-page-count

migration: Create x-multifd-channels parameter

Indicates the number of channels that we will create.  By default we
create 2 channels.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
--

Catch inconsistent defaults (eric).
Improve comment stating that number of threads is the same than number
of sockets
Use new DEFIN_PROP_*
Rename x-multifd-threads to x-multifd-threads

migration: Add multifd capability

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
--

Use new DEFINE_PROP

migration: Create migration_has_all_channels

This function allows us to decide when to close the listener socket.
For now, we only need one connection.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>

migration: Add comments to channel functions

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>

migration: Teach it about G_SOURCE_REMOVE

As this is defined on glib 2.32, add compatibility macros for older glibs.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>

migration: Create migration_ioc_process_incoming()

We pass the ioc instead of the fd.  This will allow us to have more
than one channel open.  We also make sure that we set the
from_src_file sooner, so we don't need to pass it as a parameter.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
--

Do not assing mis->from_src_file (peterxu)

Merge remote-tracking branch 'remotes/famz/tags/build-and-test-automation-pull-request' into staging

# gpg: Signature made Fri 22 Sep 2017 08:28:38 BST
# gpg:                using RSA key 0xCA35624C6A9171C6
# gpg: Good signature from "Fam Zheng <famz@redhat.com>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 5003 7CB7 9706 0F76 F021  AD56 CA35 624C 6A91 71C6

* remotes/famz/tags/build-and-test-automation-pull-request: (36 commits)
  docker: Drop 'set -e' from run script
  docker: Use archive-source.py
  tests: Add README for vm tests
  MAINTAINERS: Add tests/vm entry
  Makefile: Add rules to run vm tests
  tests: Add OpenBSD image
  tests: Add NetBSD image
  tests: Add FreeBSD image
  tests: Add ubuntu.i386 image
  tests: Add vm test lib
  tests: Add a test key pair
  scripts: Add archive-source.sh
  qemu.py: Add "wait()" method
  gitignore: Ignore vm test images
  MAINTAINERS: Fix subsystem name for "Build and test automation"
  buildsys: Move rdma libs to per object
  buildsys: Move brlapi libs to per object
  buildsys: Move usb redir cflags/libs to per object
  buildsys: Move libusb cflags/libs to per object
  buildsys: Move libcacard cflags/libs to per object
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20170922' into staging

Fix an s390x migration breakage up for 2.10 stable.
This will be fixed properly for 2.11.

# gpg: Signature made Fri 22 Sep 2017 08:28:22 BST
# gpg:                using RSA key 0xDECF6B93C6F02FAF
# gpg: Good signature from "Cornelia Huck <conny@cornelia-huck.de>"
# gpg:                 aka "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>"
# gpg:                 aka "Cornelia Huck <cohuck@kernel.org>"
# gpg:                 aka "Cornelia Huck <cohuck@redhat.com>"
# Primary key fingerprint: C3D0 D66D C362 4FF6 A8C0  18CE DECF 6B93 C6F0 2FAF

* remotes/cohuck/tags/s390x-20170922:
  s390x/ais: for 2.10 stable: disable ais facility

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

seccomp: Don't include libseccomp from QEMU header

The only prototype doesn't need anything from the lib header, and not
including it here allows files that include this header, for example
vl.c, to compile without the libseccomp cflags.

The breakage is since c3883e1f93 for environments where `pkg-config
--cflags libseccomp" is non-empty.

Reported-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Acked-by: Eduardo Otubo <otubo@redhat.com>
Message-id: 20170920083647.14599-1-famz@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

s390x/ais: for 2.10 stable: disable ais facility

The migration interface for ais was introduced with kernel 4.13
but the capability itself had been active since 4.12. As migration
support is considered necessary lets disable ais in the 2.10
stable version. A proper fix and re-enablement will be done
for qemu 2.11.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Message-Id: <20170921140834.14233-2-borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>

docker: Drop 'set -e' from run script

Signed-off-by: Fam Zheng <famz@redhat.com>

docker: Use archive-source.py

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

tests: Add README for vm tests

Signed-off-by: Fam Zheng <famz@redhat.com>

MAINTAINERS: Add tests/vm entry

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>

Makefile: Add rules to run vm tests

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Tested-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

tests: Add OpenBSD image

The image is prepared following instructions as in:

https://wiki.qemu.org/Hosts/BSD

Signed-off-by: Fam Zheng <famz@redhat.com>

tests: Add NetBSD image

The image is prepared following instructions as in:

https://wiki.qemu.org/Hosts/BSD

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Kamil Rytarowski <n54@gmx.com>

tests: Add FreeBSD image

The image is prepared following instructions as in:

https://wiki.qemu.org/Hosts/BSD

Signed-off-by: Fam Zheng <famz@redhat.com>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

tests: Add ubuntu.i386 image

This adds a 32bit guest.

The official LTS cloud image is downloaded and initialized with
cloud-init.

Signed-off-by: Fam Zheng <famz@redhat.com>

tests: Add vm test lib

This is the common code to implement a "VM test" to

  1) Download and initialize a pre-defined VM that has necessary
  dependencies to build QEMU and SSH access.

  2) Archive $SRC_PATH to a .tar file.

  3) Boot the VM, and pass the source tar file to the guest.

  4) SSH into the VM, untar the source tarball, build from the source.

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

tests: Add a test key pair

This will be used by setup test user ssh.

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

scripts: Add archive-source.sh

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

qemu.py: Add "wait()" method

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

gitignore: Ignore vm test images

Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

MAINTAINERS: Fix subsystem name for "Build and test automation"

The subsystem name for the "Build test automation" section is
"-------------------------", because an actual subsystem name
line is missing:

  $ ./scripts/get_maintainer.pl -f tests/docker/docker.py
  "Alex Bennée" <alex.bennee@linaro.org> (maintainer:-----------------...)
  Fam Zheng <famz@redhat.com> (maintainer:-----------------...)
  "Philippe Mathieu-Daudé" <f4bug@amsat.org> (reviewer:-----------------...)
  qemu-devel@nongnu.org (open list:-----------------...)

Fix the issue by inserting a subsystem name line where
get_maintainer.pl expects it.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20170921170209.9101-1-ehabkost@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move rdma libs to per object

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907084230.26493-1-famz@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move brlapi libs to per object

baum.o already receives the sdl cflags in its per object variable, do
the same for brlapi libs to avoid cluttering libs_softmmu.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907084700.952-1-famz@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move usb redir cflags/libs to per object

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907082918.7299-10-famz@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move libusb cflags/libs to per object

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907082918.7299-9-famz@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move libcacard cflags/libs to per object

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907082918.7299-8-famz@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move audio libs to per object

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907082918.7299-5-famz@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move sdl cflags/libs to per object

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907082918.7299-3-famz@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

buildsys: Move vde libs to per object

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907083552.17725-3-famz@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>

vl: Don't include vde header

Nothing in vl.c uses anything from the vde package, do remove the
unnecessary include.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907083552.17725-2-famz@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>

docker: Add test-block

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170905025614.579-6-famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Based-on: 20170905021201.25684-1-famz@redhat.com

docker: Add nettle-devel to fedora image

The LUKS cases in qemu-iotests requires this.

Reviewed-by: Kashyap Chamarthy <kchamart@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170905025614.579-5-famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Based-on: 20170905021201.25684-1-famz@redhat.com

docker: Use unconfined security profile

Some by default blocked syscalls are required to run tests for example
userfaultfd.

Reviewed-by: Kashyap Chamarthy <kchamart@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170905025614.579-4-famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Based-on: 20170905021201.25684-1-famz@redhat.com

docker: Add test_fail and prep_fail

They both print a message and exit, but with different status code so
distinguish real test errors from env preparation failures.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170905025614.579-3-famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Based-on: 20170905021201.25684-1-famz@redhat.com

docker: Fix return code of build_qemu()

Without "set -e", the "&&" makes sure that the return code reflects the
result status, and that make only runs if configure succeeds.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170905025614.579-2-famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Based-on: 20170905021201.25684-1-famz@redhat.com

tests/docker: Clean up paths

The 'run' script already creats src, build and install directories under
$TEST_DIR, use it in common.rc.

Also the tests always run from $QEMU_SRC/tests/docker, so use a relative
$CMD string.

Message-Id: <20170817035721.11064-1-famz@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>

docker: Enable features explicitly in test-full

Also avoid "set -e".

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907141245.31946-3-famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

docker: Update ubuntu image

Base on the newer ubuntu-lts (16.06) and include more packages for
better build coverage.

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170907141245.31946-2-famz@redhat.com>

docker: reduce noise when building travis.docker

Set the DEBIAN_FRONTEND and locale env vars to stop apt complaining so
much as we build the image.

Suggested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20170725133425.436-7-alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

docker: don't install device-tree-compiler build-deps in travis.docker

Installing the device-tree-compiler build-deps is a little extreme. We
only actually need the binary so include it with the other packages.

Suggested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20170725133425.436-6-alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

docker: docker.py make --no-cache skip checksum test

If you invoke with NOCACHE=1 we pass --no-cache in the argv to
docker.py but may still not force a rebuild if the dockerfile checksum
hasn't changed. By testing for its presence we can force builds
without having to manually remove the docker image.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20170725133425.436-5-alex.bennee@linaro.org>
Signed-off-by: Fam Zheng <famz@redhat.com>

docker: ensure NOUSER for travis images

While adding the current user is a useful default behaviour for
creating new images it is not appropriate for Travis which already has
a default user.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20170725133425.436-2-alex.bennee@linaro.org>
Signed-off-by: Fam Zheng <famz@redhat.com>