qemu.git
7 years agohbitmap: fix missing restore count when finish deserialization
Liang Li [Wed, 7 Feb 2018 16:35:49 +0000 (11:35 -0500)]
hbitmap: fix missing restore count when finish deserialization

The .count of HBitmap is forgot to set in function
hbitmap_deserialize_finish, let's set it to the right value.

Cc: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Cc: Fam Zheng <famz@redhat.com>
Cc: Max Reitz <mreitz@redhat.com>
Cc: John Snow <jsnow@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Weiping Zhang <zhangweiping@didichuxing.com>
Signed-off-by: Liang Li <liliangleo@didichuxing.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 20180118131308.GA2181@liangdeMacBook-Pro.local
Signed-off-by: John Snow <jsnow@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20180206a' into...
Peter Maydell [Wed, 7 Feb 2018 12:07:23 +0000 (12:07 +0000)]
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20180206a' into staging

Migration pull 2018-02-06

This is based off Juan's last pull with a few extras, but
also removing:
   Add migration xbzrle test
   Add migration precopy test

As well as my normal test boxes, I also gave it a test
on a 32 bit ARM box and it seems happy (a Calxeda highbank)
and a big-endian power box.

Dave

# gpg: Signature made Tue 06 Feb 2018 15:33:31 GMT
# gpg:                using RSA key 0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <dgilbert@redhat.com>"
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A  9FA9 0516 331E BC5B FDE7

* remotes/dgilbert/tags/pull-migration-20180206a:
  migration: incoming postcopy advise sanity checks
  migration: Don't leak IO channels
  migration: Recover block devices if failure in device state
  tests: Adjust sleeps for migration test
  tests: Create migrate-start-postcopy command
  tests: Add deprecated commands migration test
  tests: Use consistent names for migration
  tests: Consolidate accelerators declaration
  tests: Remove deprecated migration tests commands
  migration: Drop current address parameter from save_zero_page()
  migration: use s->threshold_size inside migration_update_counters
  migration/savevm.c: set MAX_VM_CMD_PACKAGED_SIZE to 1ul << 32
  migration: Route errors down through migration_channel_connect
  migration: Allow migrate_fd_connect to take an Error *

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/ehabkost/tags/python-next-pull-request' into...
Peter Maydell [Tue, 6 Feb 2018 19:28:08 +0000 (19:28 +0000)]
Merge remote-tracking branch 'remotes/ehabkost/tags/python-next-pull-request' into staging

Python queue, 2018-02-05

# gpg: Signature made Mon 05 Feb 2018 23:07:57 GMT
# gpg:                using RSA key 2807936F984DC5A6
# 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: (21 commits)
  docker: change Fedora images to run with python3
  travis: improve python version test coverage
  ui: update keycodemapdb to get py3 fixes
  input: add missing JIS keys to virtio input
  qemu.py: don't launch again before shutdown()
  qemu.py: cleanup redundant calls in launch()
  qemu.py: use poll() instead of 'returncode'
  qemu.py: always cleanup on shutdown()
  qemu.py: refactor launch()
  qemu.py: better control of created files
  qemu.py: remove unused import
  configure: allow use of python 3
  scripts: ensure signrom treats data as bytes
  qapi: force a UTF-8 locale for running Python
  qapi: ensure stable sort ordering when checking QAPI entities
  qapi: remove '-q' arg to diff when comparing QAPI output
  qapi: Adapt to moved location of 'maketrans' function in py3
  qapi: adapt to moved location of StringIO module in py3
  qapi: Use OrderedDict from standard library if available
  qapi: use items()/values() intead of iteritems()/itervalues()
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agomigration: incoming postcopy advise sanity checks
Greg Kurz [Tue, 6 Feb 2018 11:23:30 +0000 (12:23 +0100)]
migration: incoming postcopy advise sanity checks

If postcopy-ram was set on the source but not on the destination,
migration doesn't occur, the destination prints an error and boots
the guest:

qemu-system-ppc64: Expected vmdescription section, but got 0

We end up with two running instances.

This behaviour was introduced in 2.11 by commit 58110f0acb1a "migration:
split common postcopy out of ram postcopy" to prepare ground for the
upcoming dirty bitmap postcopy support. It adds a new case where the
source may send an empty postcopy advise because dirty bitmap doesn't
need to check page sizes like RAM postcopy does.

If the source has enabled postcopy-ram, then it sends an advise with
the page size values. If the destination hasn't enabled postcopy-ram,
then loadvm_postcopy_handle_advise() leaves the page size values on
the stream and returns. This confuses qemu_loadvm_state() later on
and causes the destination to start execution.

As discussed several times, postcopy-ram should be enabled both sides
to be functional. This patch changes the destination to perform some
extra checks on the advise length to ensure this is the case. Otherwise
an error is returned and migration is aborted.

Reported-by: Balamuruhan S <bala24@linux.vnet.ibm.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <151791621042.19120.3103118434734245776.stgit@bahia>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Don't leak IO channels
Ross Lagerwall [Wed, 1 Nov 2017 14:25:23 +0000 (14:25 +0000)]
migration: Don't leak IO channels

Since qemu_fopen_channel_{in,out}put take references on the underlying
IO channels, make sure to release our references to them.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Message-Id: <20171101142526.1006-2-ross.lagerwall@citrix.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Recover block devices if failure in device state
Dr. David Alan Gilbert [Mon, 5 Feb 2018 09:13:37 +0000 (09:13 +0000)]
migration: Recover block devices if failure in device state

In e91d895 I added the new pause-before-switchover mechanism
to allow migration completion to be delayed; this changes the
last state prior to completion to MIGRATE_STATUS_DEVICE rather
than MIGRATE_STATUS_ACTIVE.

Fix the failure path in migration_completion to recover the block
devices if it fails in MIGRATE_STATUS_DEVICE, not just the
MIGRATE_STATUS_ACTIVE that it previously had.

This corresponds to rh bz:
  https://bugzilla.redhat.com/show_bug.cgi?id=1538494
whose symptom is an occasional source crash on a failed migration.

Fixes: e91d8951d59d483f085f
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agotests: Adjust sleeps for migration test
Juan Quintela [Fri, 5 Jan 2018 12:56:48 +0000 (13:56 +0100)]
tests: Adjust sleeps for migration test

Also reorder code to not sleep when event already happened.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agotests: Create migrate-start-postcopy command
Juan Quintela [Fri, 5 Jan 2018 12:35:56 +0000 (13:35 +0100)]
tests: Create migrate-start-postcopy command

This way, it is like the rest of commands

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/rth/tags/pull-hppa-20180204' into staging
Peter Maydell [Tue, 6 Feb 2018 14:21:40 +0000 (14:21 +0000)]
Merge remote-tracking branch 'remotes/rth/tags/pull-hppa-20180204' into staging

hppa-softmmu update

# gpg: Signature made Sun 04 Feb 2018 22:20:40 GMT
# gpg:                using RSA key 64DF38E8AF7E215F
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>"
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* remotes/rth/tags/pull-hppa-20180204:
  roms/seabios-hppa: Update submodule and image
  tests: Enable boot-serial-test for hppa
  hw/hppa: Use qemu_log_mask instead of fprintf to stderr

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agotests: Add deprecated commands migration test
Juan Quintela [Fri, 5 Jan 2018 12:18:49 +0000 (13:18 +0100)]
tests: Add deprecated commands migration test

We add deprecated commands on a new test, so we don't have to add it
on normal tests.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agotests: Use consistent names for migration
Juan Quintela [Fri, 5 Jan 2018 11:51:49 +0000 (12:51 +0100)]
tests: Use consistent names for migration

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agotests: Consolidate accelerators declaration
Juan Quintela [Fri, 5 Jan 2018 11:49:59 +0000 (12:49 +0100)]
tests: Consolidate accelerators declaration

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agotests: Remove deprecated migration tests commands
Juan Quintela [Thu, 30 Nov 2017 20:12:03 +0000 (21:12 +0100)]
tests: Remove deprecated migration tests commands

We move to use migration_set_parameter() for everything.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Drop current address parameter from save_zero_page()
Juan Quintela [Mon, 8 Jan 2018 17:58:17 +0000 (18:58 +0100)]
migration: Drop current address parameter from save_zero_page()

It already has RAMBlock and offset, it can calculate it itself.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: use s->threshold_size inside migration_update_counters
Wei Wang [Mon, 22 Jan 2018 11:36:39 +0000 (19:36 +0800)]
migration: use s->threshold_size inside migration_update_counters

Fixes: b15df1ae50 ("migration: cleanup stats update into function")
The threshold size is changed to be recorded in s->threshold_size.

Signed-off-by: Wei Wang <wei.w.wang@intel.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration/savevm.c: set MAX_VM_CMD_PACKAGED_SIZE to 1ul << 32
Daniel Henrique Barboza [Fri, 26 Jan 2018 15:59:40 +0000 (13:59 -0200)]
migration/savevm.c: set MAX_VM_CMD_PACKAGED_SIZE to 1ul << 32

MAX_VM_CMD_PACKAGED_SIZE is a constant used in qemu_savevm_send_packaged
and loadvm_handle_cmd_packaged to determine whether a package is too
big to be sent or received. qemu_savevm_send_packaged is called inside
postcopy_start (migration/migration.c) to send the MigrationState
in a single blob to the destination, using the MIG_CMD_PACKAGED subcommand,
which will read it up using loadvm_handle_cmd_packaged. If the blob is
larger than MAX_VM_CMD_PACKAGED_SIZE, an error is thrown and the postcopy
migration is aborted. Both MAX_VM_CMD_PACKAGED_SIZE and MIG_CMD_PACKAGED
were introduced by commit 11cf1d984b ("MIG_CMD_PACKAGED: Send a packaged
chunk ..."). The constant has its original value of 1ul << 24 (16MB).

The current MAX_VM_CMD_PACKAGED_SIZE value is not enough to support postcopy
migration of bigger pseries guests. The blob size for a postcopy migration of
a pseries guest with the following setup:

qemu-system-ppc64 --nographic -vga none -machine pseries,accel=kvm -m 64G \
-smp 1,maxcpus=32 -device virtio-blk-pci,drive=rootdisk \
-drive file=f27.qcow2,if=none,cache=none,format=qcow2,id=rootdisk \
-netdev user,id=u1 -net nic,netdev=u1

Goes around 12MB. Bumping the RAM to 128G makes the blob sizes goes to 20MB.
With 256G the blob goes to 37MB - more than twice the current maximum size.
At this moment the pseries machine can handle guests with up to 1TB of RAM,
making this postcopy blob goes to 128MB of size approximately.

Following the discussions made in [1], there is a need to understand what
devices are aggressively consuming the blob in that manner and see if that
can be mitigated. Until then, we can set MAX_VM_CMD_PACKAGED_SIZE to the
maximum value allowed. Since the size is a 32 bit int variable, we can set
it as 1ul << 32, giving a maximum blob size of 4G that is enough to support
postcopy migration of 32TB RAM guests given the above constraints.

[1] https://lists.nongnu.org/archive/html/qemu-devel/2018-01/msg06313.html

Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
Reported-by: Balamuruhan S <bala24@linux.vnet.ibm.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Route errors down through migration_channel_connect
Dr. David Alan Gilbert [Fri, 15 Dec 2017 17:16:55 +0000 (17:16 +0000)]
migration: Route errors down through migration_channel_connect

Route async errors (especially from sockets) down through
migration_channel_connect and on to migrate_fd_connect where they
can be cleaned up.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agomigration: Allow migrate_fd_connect to take an Error *
Dr. David Alan Gilbert [Fri, 15 Dec 2017 17:16:54 +0000 (17:16 +0000)]
migration: Allow migrate_fd_connect to take an Error *

Allow whatever is performing the connection to pass migrate_fd_connect
an error to indicate there was a problem during connection, an allow
us to clean up.

The caller must free the error.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
7 years agodocker: change Fedora images to run with python3
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:17 +0000 (13:42 +0000)]
docker: change Fedora images to run with python3

Fedora has switched to Python 3 by default, so it makes sense to use that
for testing QEMU builds, so we get testing of Python 3 compatibility.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-15-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agotravis: improve python version test coverage
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:16 +0000 (13:42 +0000)]
travis: improve python version test coverage

Currently travis declares ancient python 2.4 is desired. Update that to
2.6 which is the oldest version any targetted distros still needs. If we
just list a python 3 version at the top level this will double the
number of travis jobs we run which is unreasonable.

So arbitrarily pick the clang test matrix entries to build with python
3.0 and 3.6, to extend coverage of python versions, without increasing
job count or build time.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-14-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoui: update keycodemapdb to get py3 fixes
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:15 +0000 (13:42 +0000)]
ui: update keycodemapdb to get py3 fixes

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-13-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoinput: add missing JIS keys to virtio input
Miika S [Tue, 16 Jan 2018 13:42:14 +0000 (13:42 +0000)]
input: add missing JIS keys to virtio input

keycodemapdb updated to add the QKeyCodes muhenkan and katakanahiragana

Signed-off-by: Miika S <miika9764@gmail.com>
Message-Id: <20180116134217.8725-12-berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu.py: don't launch again before shutdown()
Amador Pahim [Mon, 22 Jan 2018 20:50:33 +0000 (21:50 +0100)]
qemu.py: don't launch again before shutdown()

If a VM is launched, files are created and a cleanup is required before
a new launch. This cleanup is executed by shutdown(), so shutdown() must
be called even if the VM is manually terminated (i.e. using kill).

This patch creates a control to make sure launch() will not be executed
again if shutdown() is not called after the previous launch().

Signed-off-by: Amador Pahim <apahim@redhat.com>
Message-Id: <20180122205033.24893-7-apahim@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu.py: cleanup redundant calls in launch()
Amador Pahim [Mon, 22 Jan 2018 20:50:32 +0000 (21:50 +0100)]
qemu.py: cleanup redundant calls in launch()

Now that shutdown() is guaranteed to always execute self._load_io_log()
and self._post_shutdown(), their calls in 'except' became redundant and
we can safely replace it by a call to shutdown().

Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Amador Pahim <apahim@redhat.com>
Message-Id: <20180122205033.24893-6-apahim@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu.py: use poll() instead of 'returncode'
Amador Pahim [Mon, 22 Jan 2018 20:50:31 +0000 (21:50 +0100)]
qemu.py: use poll() instead of 'returncode'

The 'returncode' Popen attribute is not guaranteed to be updated. It
actually depends on a call to either poll(), wait() or communicate().

On the other hand, poll() will: "Check if child process has terminated.
Set and return returncode attribute."

Let's use the poll() to check whether the process is running and to get
the updated process exit code, when the process is finished.

Reviewed-by: Fam Zheng <famz@redhat.com>
eviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Amador Pahim <apahim@redhat.com>
Message-Id: <20180122205033.24893-5-apahim@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu.py: always cleanup on shutdown()
Amador Pahim [Mon, 22 Jan 2018 20:50:30 +0000 (21:50 +0100)]
qemu.py: always cleanup on shutdown()

Currently we only cleanup on shutdown() if the VM is running.

To make sure we will always cleanup, this patch makes the
self._load_io_log() and the self._post_shutdown() to
always be called on shutdown(), regardless the VM running state.

Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Amador Pahim <apahim@redhat.com>
Message-Id: <20180122205033.24893-4-apahim@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu.py: refactor launch()
Amador Pahim [Mon, 22 Jan 2018 20:50:29 +0000 (21:50 +0100)]
qemu.py: refactor launch()

This is just a refactor to separate the exception handler from the
actual launch procedure, improving the readability and making future
maintenances in this piece of code easier.

Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Amador Pahim <apahim@redhat.com>
Message-Id: <20180122205033.24893-3-apahim@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu.py: better control of created files
Amador Pahim [Mon, 22 Jan 2018 20:50:28 +0000 (21:50 +0100)]
qemu.py: better control of created files

To launch a VM, we need to create basically two files: the monitor
socket (if it's a UNIX socket) and the qemu log file.

For the qemu log file, we currently just open the path, which will
create the file if it does not exist or overwrite the file if it does
exist.

For the monitor socket, if it already exists, we are currently removing
it, even if it's not created by us.

This patch moves to _pre_launch() the responsibility to create a
temporary directory to host the files so we can remove the whole
directory on _post_shutdown().

Signed-off-by: Amador Pahim <apahim@redhat.com>
Message-Id: <20180122205033.24893-2-apahim@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqemu.py: remove unused import
Amador Pahim [Tue, 14 Nov 2017 10:22:39 +0000 (11:22 +0100)]
qemu.py: remove unused import

Removing 'import sys' as it's not used anywhere.

Signed-off-by: Amador Pahim <apahim@redhat.com>
Message-Id: <20171114102246.22221-2-apahim@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoconfigure: allow use of python 3
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:13 +0000 (13:42 +0000)]
configure: allow use of python 3

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-11-berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoscripts: ensure signrom treats data as bytes
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:12 +0000 (13:42 +0000)]
scripts: ensure signrom treats data as bytes

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-10-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: force a UTF-8 locale for running Python
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:11 +0000 (13:42 +0000)]
qapi: force a UTF-8 locale for running Python

Python2 did not validate locale correctness when reading input data, so
would happily read UTF-8 data in non-UTF-8 locales. Python3 is strict so
if you try to read UTF-8 data in the C locale, it will raise an error
for any UTF-8 bytes that aren't representable in 7-bit ascii encoding.
e.g.

UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 54: ordinal not in range(128)
Traceback (most recent call last):
  File "/tmp/qemu-test/src/scripts/qapi-commands.py", line 317, in <module>
    schema = QAPISchema(input_file)
  File "/tmp/qemu-test/src/scripts/qapi.py", line 1468, in __init__
    parser = QAPISchemaParser(open(fname, 'r'))
  File "/tmp/qemu-test/src/scripts/qapi.py", line 301, in __init__
    previously_included)
  File "/tmp/qemu-test/src/scripts/qapi.py", line 348, in _include
    exprs_include = QAPISchemaParser(fobj, previously_included, info)
  File "/tmp/qemu-test/src/scripts/qapi.py", line 271, in __init__
    self.src = fp.read()
  File "/usr/lib64/python3.5/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]

More background on this can be seen in

  https://www.python.org/dev/peps/pep-0538/

Many distros support a new C.UTF-8 locale that is like the C locale,
but with UTF-8 instead of 7-bit ASCII. That is not entirely portable
though. This patch thus sets the LANG to "C", but overrides LC_CTYPE
to be en_US.UTF-8 locale. This gets us pretty close to C.UTF-8, but
in a way that should be portable to everywhere QEMU builds.

This patch only forces UTF-8 for QAPI scripts, since that is the one
showing the immediate error under Python3 with C locale, but potentially
we ought to force this for all python scripts used in the build process.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-9-berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: ensure stable sort ordering when checking QAPI entities
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:10 +0000 (13:42 +0000)]
qapi: ensure stable sort ordering when checking QAPI entities

Some early python 3.x versions will have different default
ordering when calling the 'values()' method on a dict, compared
to python 2.x and later 3.x versions. Explicitly sort the items
to get a stable ordering.

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-8-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: remove '-q' arg to diff when comparing QAPI output
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:09 +0000 (13:42 +0000)]
qapi: remove '-q' arg to diff when comparing QAPI output

When the qapi schema tests fail they merely print that the expected
output didn't match the actual output. This is largely useless when
trying diagnose what went wrong. Removing the '-q' arg to diff
means that it is still silent on successful tests, but when it
fails we'll see details of the incorrect output.

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-7-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: Adapt to moved location of 'maketrans' function in py3
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:08 +0000 (13:42 +0000)]
qapi: Adapt to moved location of 'maketrans' function in py3

Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-6-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: adapt to moved location of StringIO module in py3
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:07 +0000 (13:42 +0000)]
qapi: adapt to moved location of StringIO module in py3

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-5-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: Use OrderedDict from standard library if available
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:06 +0000 (13:42 +0000)]
qapi: Use OrderedDict from standard library if available

The OrderedDict class appeared in the 'collections' module
from python 2.7 onwards, so use that in preference to our
local backport if available.

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-4-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: use items()/values() intead of iteritems()/itervalues()
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:05 +0000 (13:42 +0000)]
qapi: use items()/values() intead of iteritems()/itervalues()

The iteritems()/itervalues() methods are gone in py3, but the
items()/values() methods are still around. The latter are less
efficient than the former in py2, but this has unmeasurably
small impact on QEMU build time, so taking portability over
efficiency is a net win.

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-3-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoqapi: convert to use python print function instead of statement
Daniel P. Berrange [Tue, 16 Jan 2018 13:42:04 +0000 (13:42 +0000)]
qapi: convert to use python print function instead of statement

Python 3 no longer supports the bare "print" statement, it must be
called as a normal function with round brackets. It is possible to
opt-in to this new syntax with Python 2.6 onwards by importing the
"print_function" from the "__future__" module, making it easy to
support Python 2 and 3 in parallel.

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <20180116134217.8725-2-berrange@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/stefanberger/tags/pull-tpm-2018-02-03-1' into...
Peter Maydell [Mon, 5 Feb 2018 09:31:37 +0000 (09:31 +0000)]
Merge remote-tracking branch 'remotes/stefanberger/tags/pull-tpm-2018-02-03-1' into staging

Merge tpm 2018/02/03 v1

# gpg: Signature made Sat 03 Feb 2018 14:02:35 GMT
# gpg:                using RSA key 75AD65802A0B4211
# gpg: Good signature from "Stefan Berger <stefanb@linux.vnet.ibm.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B818 B9CA DF90 89C2 D5CE  C66B 75AD 6580 2A0B 4211

* remotes/stefanberger/tags/pull-tpm-2018-02-03-1:
  tpm: tis: move one-line function into caller
  MAINTAINERS: add pointer to tpm-next repository
  tpm: wrap stX_be_p in tpm_cmd_set_XYZ functions
  tpm: Split off tpm_crb_reset function

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoroms/seabios-hppa: Update submodule and image
Richard Henderson [Sun, 4 Feb 2018 22:08:37 +0000 (14:08 -0800)]
roms/seabios-hppa: Update submodule and image

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotests: Enable boot-serial-test for hppa
Richard Henderson [Sun, 4 Feb 2018 06:46:15 +0000 (22:46 -0800)]
tests: Enable boot-serial-test for hppa

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agohw/hppa: Use qemu_log_mask instead of fprintf to stderr
Richard Henderson [Sun, 4 Feb 2018 06:41:41 +0000 (22:41 -0800)]
hw/hppa: Use qemu_log_mask instead of fprintf to stderr

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotpm: tis: move one-line function into caller
Stefan Berger [Fri, 2 Feb 2018 13:39:18 +0000 (08:39 -0500)]
tpm: tis: move one-line function into caller

Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agoMAINTAINERS: add pointer to tpm-next repository
Stefan Berger [Fri, 2 Feb 2018 11:51:38 +0000 (06:51 -0500)]
MAINTAINERS: add pointer to tpm-next repository

Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agotpm: wrap stX_be_p in tpm_cmd_set_XYZ functions
Stefan Berger [Thu, 1 Feb 2018 23:05:10 +0000 (18:05 -0500)]
tpm: wrap stX_be_p in tpm_cmd_set_XYZ functions

Wrap the calls to stl_be_p and stw_be_p in tpm_cmd_set_XYZ functions
that are similar to existing getters.

Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agotpm: Split off tpm_crb_reset function
Stefan Berger [Thu, 1 Feb 2018 22:55:50 +0000 (17:55 -0500)]
tpm: Split off tpm_crb_reset function

Split off the tpm_crb_reset function part from tpm_crb_realize
that we need to run every time the machine resets.

Also register our reset function with the system since TYPE_DEVICE
seems to not get a reset otherwise.

Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/kraxel/tags/ui-20180202-pull-request' into...
Peter Maydell [Fri, 2 Feb 2018 18:54:11 +0000 (18:54 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/ui-20180202-pull-request' into staging

ui: use QIONetListener in vnc, bugfixes for sdl1 and vnc.

# gpg: Signature made Fri 02 Feb 2018 07:17:36 GMT
# gpg:                using RSA key 4CB6D8EED3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"
# Primary key fingerprint: A032 8CFF B93A 17A7 9901  FE7D 4CB6 D8EE D3E8 7138

* remotes/kraxel/tags/ui-20180202-pull-request:
  ui: correctly advance output buffer when writing SASL data
  ui: convert VNC server to QIONetListener
  ui: fix mixup between qnum and qcode in SDL1 key handling

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/kraxel/tags/vga-20180202-pull-request' into...
Peter Maydell [Fri, 2 Feb 2018 17:24:00 +0000 (17:24 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/vga-20180202-pull-request' into staging

virtio-gpu: disallow vIOMMU

# gpg: Signature made Fri 02 Feb 2018 08:31:52 GMT
# gpg:                using RSA key 4CB6D8EED3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"
# Primary key fingerprint: A032 8CFF B93A 17A7 9901  FE7D 4CB6 D8EE D3E8 7138

* remotes/kraxel/tags/vga-20180202-pull-request:
  virtio-gpu: disallow vIOMMU

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/gkurz/tags/for-upstream' into staging
Peter Maydell [Fri, 2 Feb 2018 16:26:41 +0000 (16:26 +0000)]
Merge remote-tracking branch 'remotes/gkurz/tags/for-upstream' into staging

This series is mostly about 9p request cancellation. It fixes a
long standing bug (read "specification violation") where the server
would send an invalid response when the client has cancelled an
in-flight request. This was causing annoying spurious EINTR returns
in linux. The fix comes with some related testing in QTEST.

Other patches are code cleanup and improvements.

# gpg: Signature made Fri 02 Feb 2018 10:16:03 GMT
# gpg:                using RSA key 71D4D5E5822F73D6
# gpg: Good signature from "Greg Kurz <groug@kaod.org>"
# gpg:                 aka "Gregory Kurz <gregory.kurz@free.fr>"
# gpg:                 aka "[jpeg image of size 3330]"
# Primary key fingerprint: B482 8BAF 9431 40CE F2A3  4910 71D4 D5E5 822F 73D6

* remotes/gkurz/tags/for-upstream:
  tests/virtio-9p: explicitly handle potential integer overflows
  tests: virtio-9p: add FLUSH operation test
  libqos/virtio: return length written into used descriptor
  tests: virtio-9p: add WRITE operation test
  tests: virtio-9p: add LOPEN operation test
  tests: virtio-9p: use the synth backend
  tests: virtio-9p: wait for completion in the test code
  tests: virtio-9p: move request tag to the test functions
  9pfs: Correctly handle cancelled requests
  9pfs: drop v9fs_register_transport()

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/kraxel/tags/audio-20180202-pull-request' into...
Peter Maydell [Fri, 2 Feb 2018 15:33:57 +0000 (15:33 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/audio-20180202-pull-request' into staging

audio: two small fixes.

# gpg: Signature made Fri 02 Feb 2018 07:49:20 GMT
# gpg:                using RSA key 4CB6D8EED3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"
# Primary key fingerprint: A032 8CFF B93A 17A7 9901  FE7D 4CB6 D8EE D3E8 7138

* remotes/kraxel/tags/audio-20180202-pull-request:
  hw/audio/sb16.c: change dolog() to qemu_log_mask()
  hw/audio/wm8750: move WM8750 declarations from i2c/i2c.h to audio/wm8750.h

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/cminyard/tags/for-release-20180201' into staging
Peter Maydell [Fri, 2 Feb 2018 14:10:13 +0000 (14:10 +0000)]
Merge remote-tracking branch 'remotes/cminyard/tags/for-release-20180201' into staging

Lots of litte miscellaneous fixes for the IPMI code, plus
add me as the IPMI maintainer.

# gpg: Signature made Thu 01 Feb 2018 18:44:55 GMT
# gpg:                using RSA key 61F38C90919BFF81
# gpg: Good signature from "Corey Minyard <cminyard@mvista.com>"
# gpg:                 aka "Corey Minyard <minyard@acm.org>"
# gpg:                 aka "Corey Minyard <corey@minyard.net>"
# gpg:                 aka "Corey Minyard <minyard@mvista.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: FD0D 5CE6 7CE0 F59A 6688  2686 61F3 8C90 919B FF81

* remotes/cminyard/tags/for-release-20180201:
  ipmi: Allow BMC device properties to be set
  ipmi: disable IRQ and ATN on an external disconnect
  ipmi: Fix macro issues
  ipmi: Add the platform event message command
  ipmi: Don't set the timestamp on add events that don't have it
  ipmi: Fix SEL get/set time commands
  Add maintainer for the IPMI code

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agoMerge remote-tracking branch 'remotes/elmarco/tags/dump-pull-request' into staging
Peter Maydell [Fri, 2 Feb 2018 10:57:50 +0000 (10:57 +0000)]
Merge remote-tracking branch 'remotes/elmarco/tags/dump-pull-request' into staging

# gpg: Signature made Thu 01 Feb 2018 11:15:42 GMT
# gpg:                using RSA key DAE8E10975969CE5
# gpg: Good signature from "Marc-André Lureau <marcandre.lureau@redhat.com>"
# gpg:                 aka "Marc-André Lureau <marcandre.lureau@gmail.com>"
# Primary key fingerprint: 87A9 BD93 3F87 C606 D276  F62D DAE8 E109 7596 9CE5

* remotes/elmarco/tags/dump-pull-request:
  dump-guest-memory.py: skip vmcoreinfo section if not available

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agotests/virtio-9p: explicitly handle potential integer overflows
Greg Kurz [Thu, 1 Feb 2018 20:21:28 +0000 (21:21 +0100)]
tests/virtio-9p: explicitly handle potential integer overflows

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
7 years agotests: virtio-9p: add FLUSH operation test
Greg Kurz [Thu, 1 Feb 2018 20:21:28 +0000 (21:21 +0100)]
tests: virtio-9p: add FLUSH operation test

The idea is to send a victim request that will possibly block in the
server and to send a flush request to cancel the victim request.

This patch adds two test to verifiy that:
- the server does not reply to a victim request that was actually
  cancelled
- the server replies to the flush request after replying to the
  victim request if it could not cancel it

9p request cancellation reference:

http://man.cat-v.org/plan_9/5/flush

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
(groug, change the test to only write a single byte to avoid
        any alignment or endianess consideration)

7 years agolibqos/virtio: return length written into used descriptor
Greg Kurz [Thu, 1 Feb 2018 20:21:28 +0000 (21:21 +0100)]
libqos/virtio: return length written into used descriptor

When a 9p request is flushed (ie, cancelled) by the guest, the device
is expected to simply mark the request as used, without sending a 9p
reply (ie, without writing anything into the used buffer).

To be able to test this, we need access to the length written by the
device into the used descriptor. This patch adds a uint32_t * argument
to qvirtqueue_get_buf() and qvirtio_wait_used_elem() for this purpose.

All existing users are updated accordingly.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/cody/tags/block-pull-request' into staging
Peter Maydell [Fri, 2 Feb 2018 09:53:15 +0000 (09:53 +0000)]
Merge remote-tracking branch 'remotes/cody/tags/block-pull-request' into staging

# gpg: Signature made Thu 01 Feb 2018 04:05:22 GMT
# gpg:                using RSA key BDBE7B27C0DE3057
# gpg: Good signature from "Jeffrey Cody <jcody@redhat.com>"
# gpg:                 aka "Jeffrey Cody <jeff@codyprime.org>"
# gpg:                 aka "Jeffrey Cody <codyprime@gmail.com>"
# Primary key fingerprint: 9957 4B4D 3474 90E7 9D98  D624 BDBE 7B27 C0DE 3057

* remotes/cody/tags/block-pull-request:
  iotests: Make 200 run on tmpfs
  block/ssh: fix possible segmentation fault when .desc is not null-terminated

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agovirtio-gpu: disallow vIOMMU
Peter Xu [Wed, 31 Jan 2018 04:04:01 +0000 (12:04 +0800)]
virtio-gpu: disallow vIOMMU

virtio-gpu has special code path that bypassed vIOMMU protection.  So
for now let's disable iommu_platform for the device until we fully
support that (if needed).

After the patch, both virtio-vga and virtio-gpu won't allow to boot with
iommu_platform parameter set.

CC: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-id: 20180131040401.3550-1-peterx@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
7 years agohw/audio/sb16.c: change dolog() to qemu_log_mask()
John Arbuckle [Thu, 1 Feb 2018 17:27:44 +0000 (12:27 -0500)]
hw/audio/sb16.c: change dolog() to qemu_log_mask()

Changes all the occurrances of dolog() to qemu_log_mask().

Signed-off-by: John Arbuckle <programmingkidx@gmail.com>
Message-id: 20180201172744.7504-1-programmingkidx@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
7 years agohw/audio/wm8750: move WM8750 declarations from i2c/i2c.h to audio/wm8750.h
Philippe Mathieu-Daudé [Tue, 19 Sep 2017 12:30:53 +0000 (09:30 -0300)]
hw/audio/wm8750: move WM8750 declarations from i2c/i2c.h to audio/wm8750.h

while here use TYPE_WM8750 and declare a data_req_cb() typedef.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20170919123053.32675-1-f4bug@amsat.org
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
7 years agoui: correctly advance output buffer when writing SASL data
Daniel P. Berrangé [Thu, 1 Feb 2018 15:58:41 +0000 (15:58 +0000)]
ui: correctly advance output buffer when writing SASL data

In this previous commit:

  commit 8f61f1c5a6bc06438a1172efa80bc7606594fa07
  Author: Daniel P. Berrange <berrange@redhat.com>
  Date:   Mon Dec 18 19:12:20 2017 +0000

    ui: track how much decoded data we consumed when doing SASL encoding

I attempted to fix a flaw with tracking how much data had actually been
processed when encoding with SASL. With that flaw, the VNC server could
mistakenly discard queued data that had not been sent.

The fix was not quite right though, because it merely decremented the
vs->output.offset value. This is effectively discarding data from the
end of the pending output buffer. We actually need to discard data from
the start of the pending output buffer. We also want to free memory that
is no longer required. The correct way to handle this is to use the
buffer_advance() helper method instead of directly manipulating the
offset value.

Reported-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Message-id: 20180201155841.27509-1-berrange@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
7 years agoui: convert VNC server to QIONetListener
Daniel P. Berrange [Thu, 1 Feb 2018 16:45:14 +0000 (16:45 +0000)]
ui: convert VNC server to QIONetListener

The VNC server already has the ability to listen on multiple sockets.
Converting it to use the QIONetListener APIs though, will reduce the
amount of code in the VNC server and improve the clarity of what is
left.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20180201164514.10330-1-berrange@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
7 years agoui: fix mixup between qnum and qcode in SDL1 key handling
Daniel P. Berrangé [Thu, 1 Feb 2018 18:00:33 +0000 (18:00 +0000)]
ui: fix mixup between qnum and qcode in SDL1 key handling

The previous commit:

  commit 2ec78706d188df7d3dab43d07b19b05ef7800a44
  Author: Daniel P. Berrange <berrange@redhat.com>
  Date:   Wed Jan 17 16:47:15 2018 +0000

    ui: convert GTK and SDL1 frontends to keycodemapdb

changed the x_keymap.c keymap so that its target was qcodes instead of
qnums. It updated the GTK frontend to take account of this change, but
forgot to update the SDL1 frontend. Thus the SDL frontend was getting
qcodes but dispatching them as if they were qnums. IOW, keyboard input
was completely hosed with SDL1. Since the keyboard layout tables are
still all based on qnums, it is easier to just keep SDL1 using qnums as
it will be deleted in a few releases time.

Reported-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Tested-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-id: 20180201180033.14255-1-berrange@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
7 years agotests: virtio-9p: add WRITE operation test
Greg Kurz [Thu, 1 Feb 2018 20:21:28 +0000 (21:21 +0100)]
tests: virtio-9p: add WRITE operation test

Trivial test of a successful write.

Signed-off-by: Greg Kurz <groug@kaod.org>
(groug, handle potential overflow when computing request size,
        add missing g_free(buf),
        backend handles one written byte at a time to validate
        the server doesn't do short-reads)
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
7 years agotests: virtio-9p: add LOPEN operation test
Greg Kurz [Thu, 1 Feb 2018 20:21:28 +0000 (21:21 +0100)]
tests: virtio-9p: add LOPEN operation test

Trivial test of a successful open.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
7 years agotests: virtio-9p: use the synth backend
Greg Kurz [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
tests: virtio-9p: use the synth backend

The purpose of virtio-9p-test is to test the virtio-9p device, especially
the 9p server state machine. We don't really care what fsdev backend we're
using. Moreover, if we want to be able to test the flush request or a
device reset with in-flights I/O, it is close to impossible to achieve
with a physical backend because we cannot ask it reliably to put an I/O
on hold at a specific point in time.

Fortunately, we can do that with the synthetic backend, which allows to
register callbacks on read/write accesses to a specific file. This will
be used by a later patch to test the 9P flush request.

The walk request test is converted to using the synth backend.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
7 years agotests: virtio-9p: wait for completion in the test code
Greg Kurz [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
tests: virtio-9p: wait for completion in the test code

In order to test request cancellation, we will need to send multiple
requests and wait for the associated replies. Since we poll the ISR
to know if a request completed, we may have several replies to parse
when we detect ISR was set to 1.

This patch moves the waiting out of the reply parsing path, up into
the functional tests.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
7 years agotests: virtio-9p: move request tag to the test functions
Greg Kurz [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
tests: virtio-9p: move request tag to the test functions

It doesn't really makes sense to hide the request tag from the test
functions. It prevents to test the 9p server behavior when passed
a wrong tag (ie, still in use or different from P9_NOTAG for a
version request). Also the spec says that a tag is reusable as soon
as the corresponding request was replied or flushed: no need to
always increment tags like we do now. And finaly, an upcoming test
of the flush command will need to manipulate tags explicitely.

This simply changes all request functions to have a tag argument.
Except for the version request which needs P9_NOTAG, all other
tests can pass 0 since they wait for the reply before sending
another request.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
7 years ago9pfs: Correctly handle cancelled requests
Keno Fischer [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
9pfs: Correctly handle cancelled requests

# Background

I was investigating spurious non-deterministic EINTR returns from
various 9p file system operations in a Linux guest served from the
qemu 9p server.

 ## EINTR, ERESTARTSYS and the linux kernel

When a signal arrives that the Linux kernel needs to deliver to user-space
while a given thread is blocked (in the 9p case waiting for a reply to its
request in 9p_client_rpc -> wait_event_interruptible), it asks whatever
driver is currently running to abort its current operation (in the 9p case
causing the submission of a TFLUSH message) and return to user space.
In these situations, the error message reported is generally ERESTARTSYS.
If the userspace processes specified SA_RESTART, this means that the
system call will get restarted upon completion of the signal handler
delivery (assuming the signal handler doesn't modify the process state
in complicated ways not relevant here). If SA_RESTART is not specified,
ERESTARTSYS gets translated to EINTR and user space is expected to handle
the restart itself.

 ## The 9p TFLUSH command

The 9p TFLUSH commands requests that the server abort an ongoing operation.
The man page [1] specifies:

```
If it recognizes oldtag as the tag of a pending transaction, it should
abort any pending response and discard that tag.
[...]
When the client sends a Tflush, it must wait to receive the corresponding
Rflush before reusing oldtag for subsequent messages. If a response to the
flushed request is received before the Rflush, the client must honor the
response as if it had not been flushed, since the completed request may
signify a state change in the server
```

In particular, this means that the server must not send a reply with the
orignal tag in response to the cancellation request, because the client is
obligated to interpret such a reply as a coincidental reply to the original
request.

 # The bug

When qemu receives a TFlush request, it sets the `cancelled` flag on the
relevant pdu. This flag is periodically checked, e.g. in
`v9fs_co_name_to_path`, and if set, the operation is aborted and the error
is set to EINTR. However, the server then violates the spec, by returning
to the client an Rerror response, rather than discarding the message
entirely. As a result, the client is required to assume that said Rerror
response is a result of the original request, not a result of the
cancellation and thus passes the EINTR error back to user space.
This is not the worst thing it could do, however as discussed above, the
correct error code would have been ERESTARTSYS, such that user space
programs with SA_RESTART set get correctly restarted upon completion of
the signal handler.
Instead, such programs get spurious EINTR results that they were not
expecting to handle.

It should be noted that there are plenty of user space programs that do not
set SA_RESTART and do not correctly handle EINTR either. However, that is
then a userspace bug. It should also be noted that this bug has been
mitigated by a recent commit to the Linux kernel [2], which essentially
prevents the kernel from sending Tflush requests unless the process is about
to die (in which case the process likely doesn't care about the response).
Nevertheless, for older kernels and to comply with the spec, I believe this
change is beneficial.

 # Implementation

The fix is fairly simple, just skipping notification of a reply if
the pdu was previously cancelled. We do however, also notify the transport
layer that we're doing this, so it can clean up any resources it may be
holding. I also added a new trace event to distinguish
operations that caused an error reply from those that were cancelled.

One complication is that we only omit sending the message on EINTR errors in
order to avoid confusing the rest of the code (which may assume that a
client knows about a fid if it sucessfully passed it off to pud_complete
without checking for cancellation status). This does mean that if the server
acts upon the cancellation flag, it always needs to set err to EINTR. I
believe this is true of the current code.

[1] https://9fans.github.io/plan9port/man/man9/flush.html
[2] https://github.com/torvalds/linux/commit/9523feac272ccad2ad8186ba4fcc891

Signed-off-by: Keno Fischer <keno@juliacomputing.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
[groug, send a zero-sized reply instead of detaching the buffer]
Signed-off-by: Greg Kurz <groug@kaod.org>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
7 years ago9pfs: drop v9fs_register_transport()
Greg Kurz [Thu, 1 Feb 2018 20:21:27 +0000 (21:21 +0100)]
9pfs: drop v9fs_register_transport()

No good reasons to do this outside of v9fs_device_realize_common().

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
7 years agodump-guest-memory.py: skip vmcoreinfo section if not available
Marc-André Lureau [Wed, 31 Jan 2018 10:37:24 +0000 (11:37 +0100)]
dump-guest-memory.py: skip vmcoreinfo section if not available

On some architectures, qemu doesn't support vmcoreinfo device,
and dump-guest-memory fails:

(gdb) dump-guest-memory /tmp/vmcore  ppc64-le
guest RAM blocks:
target_start     target_end       host_addr        message count
---------------- ---------------- ---------------- ------- -----
0000000000000000 0000000200000000 00003ffd86980000 added       1
0000200080000000 0000200080800000 00003ffd86170000 added       2
Python Exception <class 'gdb.error'> No symbol "vmcoreinfo_realize" in current context.:
Error occurred in Python command: No symbol "vmcoreinfo_realize" in current context.

Check that vmcoreinfo_realize symbol exists before evaluating an
expression with it.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
7 years agoiotests: Make 200 run on tmpfs
Max Reitz [Wed, 17 Jan 2018 13:50:15 +0000 (14:50 +0100)]
iotests: Make 200 run on tmpfs

200 currently fails on tmpfs because it sets cache=none.  However,
without that (and aio=native), the test still works now and it fails
before Jeff's series (on fc7dbc119e0852a70dc9fa68bb41a318e49e4cd6).  So
we can probably remove the aio=native safely, and replace cache=none by
cache=$CACHEMODE.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Message-id: 20180117135015.15051-1-mreitz@redhat.com
Signed-off-by: Jeff Cody <jcody@redhat.com>
7 years agoblock/ssh: fix possible segmentation fault when .desc is not null-terminated
Murilo Opsfelder Araujo [Fri, 5 Jan 2018 14:44:40 +0000 (12:44 -0200)]
block/ssh: fix possible segmentation fault when .desc is not null-terminated

This patch prevents a possible segmentation fault when .desc members are checked
against NULL.

The ssh_runtime_opts was added by commit
8a6a80896d6af03b8ee0c17cdf37219eca2588a7 ("block/ssh: Use QemuOpts for runtime
options").

This fix was inspired by
http://lists.nongnu.org/archive/html/qemu-devel/2018-01/msg00883.html.

Fixes: 8a6a80896d6af03b8ee0c17cdf37219eca2588a7 ("block/ssh: Use QemuOpts for runtime options")
Cc: Max Reitz <mreitz@redhat.com>
Cc: Eric Blake <eblake@redhat.com>
Signed-off-by: Murilo Opsfelder Araujo <muriloo@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
7 years agoMerge remote-tracking branch 'remotes/rth/tags/pull-hppa-20180131' into staging
Peter Maydell [Wed, 31 Jan 2018 15:50:29 +0000 (15:50 +0000)]
Merge remote-tracking branch 'remotes/rth/tags/pull-hppa-20180131' into staging

Implement hppa-softmmu

# gpg: Signature made Wed 31 Jan 2018 14:19:06 GMT
# gpg:                using RSA key 0x64DF38E8AF7E215F
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>"
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* remotes/rth/tags/pull-hppa-20180131: (43 commits)
  target/hppa: Implement PROBE for system mode
  target/hppa: Fix 32-bit operand masks for 0E FCVT
  hw/hppa: Add MAINTAINERS entry
  pc-bios: Add hppa-firmware.img and git submodule
  hw/hppa: Implement DINO system board
  target/hppa: Enable MTTCG
  target/hppa: Implement STWA
  target/hppa: Implement a pause instruction
  target/hppa: Implement LDSID for system mode
  target/hppa: Fix comment
  target/hppa: Increase number of temp regs
  target/hppa: Only use EXCP_DTLB_MISS
  target/hppa: Implement B,GATE insn
  target/hppa: Add migration for the cpu
  target/hppa: Add system registers to gdbstub
  target/hppa: Optimize for flat addressing space
  target/hppa: Implement halt and reset instructions
  target/hppa: Implement SYNCDMA insn
  target/hppa: Implement LCI
  target/hppa: Implement LPA
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7 years agotarget/hppa: Implement PROBE for system mode
Richard Henderson [Wed, 24 Jan 2018 23:03:25 +0000 (15:03 -0800)]
target/hppa: Implement PROBE for system mode

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Fix 32-bit operand masks for 0E FCVT
Richard Henderson [Wed, 24 Jan 2018 22:24:53 +0000 (14:24 -0800)]
target/hppa: Fix 32-bit operand masks for 0E FCVT

We masked the wrong bits, which prevented some of the
32-bit R registers.  E.g. "fcnvxf,sgl,sgl fr22R,fr6R".

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agohw/hppa: Add MAINTAINERS entry
Richard Henderson [Sun, 21 Jan 2018 22:05:03 +0000 (14:05 -0800)]
hw/hppa: Add MAINTAINERS entry

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agopc-bios: Add hppa-firmware.img and git submodule
Richard Henderson [Sun, 21 Jan 2018 21:04:45 +0000 (13:04 -0800)]
pc-bios: Add hppa-firmware.img and git submodule

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agohw/hppa: Implement DINO system board
Helge Deller [Sun, 8 Oct 2017 20:47:27 +0000 (16:47 -0400)]
hw/hppa: Implement DINO system board

Now that we have the prerequisites in target/hppa/,
implement the hardware for a PA7100LC.

This also enables build for hppa-softmmu.

Signed-off-by: Helge Deller <deller@gmx.de>
[rth: Since it is all new code, squashed all branch development
withing hw/hppa/ to a single patch.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Enable MTTCG
Richard Henderson [Sun, 7 Jan 2018 00:02:27 +0000 (16:02 -0800)]
target/hppa: Enable MTTCG

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Implement STWA
Richard Henderson [Sat, 6 Jan 2018 23:30:45 +0000 (15:30 -0800)]
target/hppa: Implement STWA

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Implement a pause instruction
Richard Henderson [Fri, 5 Jan 2018 00:49:29 +0000 (16:49 -0800)]
target/hppa: Implement a pause instruction

This is an extension to the base ISA, but we can use this in
the kernel idle loop to reduce the host cpu time consumed.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Implement LDSID for system mode
Helge Deller [Tue, 2 Jan 2018 20:31:45 +0000 (21:31 +0100)]
target/hppa: Implement LDSID for system mode

Signed-off-by: Helge Deller <deller@gmx.de>
Message-Id: <20180102203145.GA17059@ls3530.fritz.box>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Fix comment
Helge Deller [Tue, 12 Dec 2017 21:23:19 +0000 (22:23 +0100)]
target/hppa: Fix comment

Signed-off-by: Helge Deller <deller@gmx.de>
Message-Id: <20171212212319.GA31494@ls3530.fritz.box>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Increase number of temp regs
Richard Henderson [Wed, 1 Nov 2017 12:03:09 +0000 (05:03 -0700)]
target/hppa: Increase number of temp regs

HP-UX 10.20 CD contains "add r0, r0, r27" in a delay slot,
which uses at least 5 temps.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Only use EXCP_DTLB_MISS
Richard Henderson [Fri, 29 Dec 2017 03:26:52 +0000 (19:26 -0800)]
target/hppa: Only use EXCP_DTLB_MISS

Unknown why this works, but if we return EXCP_ITLB_MISS we
will triple-fault the first userland instruction fetch.
Is it something to do with having a combined I/DTLB?

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Implement B,GATE insn
Richard Henderson [Fri, 15 Dec 2017 20:37:26 +0000 (14:37 -0600)]
target/hppa: Implement B,GATE insn

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Add migration for the cpu
Richard Henderson [Mon, 20 Nov 2017 10:06:04 +0000 (11:06 +0100)]
target/hppa: Add migration for the cpu

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Add system registers to gdbstub
Richard Henderson [Tue, 7 Nov 2017 04:42:14 +0000 (05:42 +0100)]
target/hppa: Add system registers to gdbstub

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Optimize for flat addressing space
Richard Henderson [Mon, 6 Nov 2017 20:10:33 +0000 (21:10 +0100)]
target/hppa: Optimize for flat addressing space

Linux sets sr4-sr7 all to the same value, which means that we
need not do any runtime computation to find out what space to
use in forming the GVA.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Implement halt and reset instructions
Helge Deller [Fri, 29 Dec 2017 06:04:57 +0000 (22:04 -0800)]
target/hppa: Implement halt and reset instructions

Real hardware would use an external device to control the power.
But for the moment let's invent instructions in reserved space,
to be used by our custom firmware.

Signed-off-by: Helge Deller <deller@gmx.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agoipmi: Allow BMC device properties to be set
Corey Minyard [Mon, 28 Aug 2017 17:48:44 +0000 (12:48 -0500)]
ipmi: Allow BMC device properties to be set

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agoipmi: disable IRQ and ATN on an external disconnect
Corey Minyard [Tue, 19 Sep 2017 20:19:26 +0000 (15:19 -0500)]
ipmi: disable IRQ and ATN on an external disconnect

Otherwise there's no way to clear them without an external command,
and it could lock the OS in the VM if they were stuck.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
7 years agoipmi: Fix macro issues
Corey Minyard [Thu, 22 Dec 2016 14:22:11 +0000 (08:22 -0600)]
ipmi: Fix macro issues

Macro parameters should almost always have () around them when used.
llvm reported an error on this.

Remove redundant parenthesis and put parenthesis around the entire
macros with assignments in case they are used in an expression.

The macros were doing ((v) & 1) for a binary input, but that only works
if v == 0 or if v & 1.  Changed to !!(v) so they work for all values.

Remove some unused macros.

Reported in https://bugs.launchpad.net/bugs/1651167

An audit of these changes found no semantic changes; this is just
cleanups for proper style and to avoid a compiler warning.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
7 years agoipmi: Add the platform event message command
Corey Minyard [Sat, 19 Aug 2017 01:17:48 +0000 (20:17 -0500)]
ipmi: Add the platform event message command

This lets an event be added to the SEL as if a sensor had generated
it.  The OpenIPMI driver uses it for storing panic event information.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
7 years agoipmi: Don't set the timestamp on add events that don't have it
Corey Minyard [Sat, 19 Aug 2017 01:15:02 +0000 (20:15 -0500)]
ipmi: Don't set the timestamp on add events that don't have it

According to the spec, from section "32.3 OEM SEL Record - Type
E0h-FFh", event types from 0x0e to 0xff do not have a timestamp.
So don't set it when adding those types.  This required putting
the timestamp in a temporary buffer, since it's still required
to set the last addition time.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
7 years agoipmi: Fix SEL get/set time commands
Corey Minyard [Sat, 19 Aug 2017 01:13:10 +0000 (20:13 -0500)]
ipmi: Fix SEL get/set time commands

The minimum message size was on the wrong commands, for getting
the time it's zero and for setting the time it's 6.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agoAdd maintainer for the IPMI code
Corey Minyard [Thu, 7 Dec 2017 20:55:31 +0000 (14:55 -0600)]
Add maintainer for the IPMI code

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
7 years agotarget/hppa: Implement SYNCDMA insn
Richard Henderson [Sun, 29 Oct 2017 14:57:51 +0000 (15:57 +0100)]
target/hppa: Implement SYNCDMA insn

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7 years agotarget/hppa: Implement LCI
Richard Henderson [Sun, 5 Nov 2017 10:02:40 +0000 (11:02 +0100)]
target/hppa: Implement LCI

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>