OSDN Git Service

qmiga/qemu.git
6 years agotravis: test out-of-tree builds
Philippe Mathieu-Daudé [Thu, 21 Jun 2018 12:26:37 +0000 (09:26 -0300)]
travis: test out-of-tree builds

Force one config to build 'out-of-tree' (object files and executables
are created in a tree outside the project source code).

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
6 years agotravis: do not waste time cloning unused submodules
Philippe Mathieu-Daudé [Fri, 22 Jun 2018 02:15:46 +0000 (23:15 -0300)]
travis: do not waste time cloning unused submodules

Builds only require:
- dtc
- keycodemapdb
- capstone

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
[AJB: drop wget cache]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
6 years agoRevert "Makefile: Rename TARGET_DIRS to TARGET_LIST"
Alex Bennée [Wed, 4 Jul 2018 06:30:11 +0000 (07:30 +0100)]
Revert "Makefile: Rename TARGET_DIRS to TARGET_LIST"

This reverts commit 208ecb3e1acc8d55dab49fdf721a86d513691688. This was
causing problems by making DEF_TARGET_LIST pointless and having to
jump through hoops to build on mingw with a dully enabled config.
This includes a change to fix the per-guest TCG test probe which was
added after 208ecb3 and used TARGET_LIST.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/jnsnow/tags/bitmaps-pull-request' into staging
Peter Maydell [Thu, 5 Jul 2018 12:33:52 +0000 (13:33 +0100)]
Merge remote-tracking branch 'remotes/jnsnow/tags/bitmaps-pull-request' into staging

Pull request

# gpg: Signature made Wed 04 Jul 2018 07:33:42 BST
# gpg:                using RSA key 7DEF8106AAFC390E
# gpg: Good signature from "John Snow (John Huston) <jsnow@redhat.com>"
# Primary key fingerprint: FAEB 9711 A12C F475 812F  18F2 88A9 064D 1835 61EB
#      Subkey fingerprint: F9B7 ABDB BCAC DF95 BE76  CBD0 7DEF 8106 AAFC 390E

* remotes/jnsnow/tags/bitmaps-pull-request:
  dirty-bitmap: fix double lock on bitmap enabling
  block/dirty-bitmap: add bdrv_enable_dirty_bitmap_locked

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/armbru/tags/pull-monitor-2018-07-03-v2' into...
Peter Maydell [Thu, 5 Jul 2018 10:25:14 +0000 (11:25 +0100)]
Merge remote-tracking branch 'remotes/armbru/tags/pull-monitor-2018-07-03-v2' into staging

Monitor patches for 2018-07-03

# gpg: Signature made Tue 03 Jul 2018 22:20:13 BST
# gpg:                using RSA key 3870B400EB918653
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>"
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>"
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* remotes/armbru/tags/pull-monitor-2018-07-03-v2: (32 commits)
  qapi: Polish command flags documentation in qapi-code-gen.txt
  monitor: Improve some comments
  qmp: Clean up capability negotiation after commit 02130314d8c
  qobject: Let qobject_from_jsonf() fail instead of abort
  qmp: Switch timestamp_put() to qdict_from_jsonf_nofail()
  qmp: Add some comments around null responses
  qmp: Simplify monitor_qmp_respond()
  qmp: Replace get_qmp_greeting() by qmp_greeting()
  qmp: Replace monitor_json_emitter{,raw}() by qmp_{queue,send}_response()
  qmp: Use QDict * instead of QObject * for response objects
  qmp: De-duplicate error response building
  qobject: New qdict_from_jsonf_nofail()
  monitor: Peel off @mon_global wrapper
  monitor: Rename use_io_thr to use_io_thread
  qmp: Don't let JSON errors jump the queue
  qmp: Don't let malformed in-band commands jump the queue
  tests/qmp-test: Demonstrate QMP errors jumping the queue
  qmp: Simplify code around monitor_qmp_dispatch_one()
  qmp: Always free QMPRequest with qmp_request_free()
  qmp: Revert change to handle_qmp_command tracepoint
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2018-07-03' into staging
Peter Maydell [Thu, 5 Jul 2018 09:31:36 +0000 (10:31 +0100)]
Merge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2018-07-03' into staging

QAPI patches for 2018-07-03

# gpg: Signature made Tue 03 Jul 2018 21:52:55 BST
# gpg:                using RSA key 3870B400EB918653
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>"
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>"
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* remotes/armbru/tags/pull-qapi-2018-07-03:
  qapi: add conditions to SPICE type/commands/events on the schema
  qapi: add conditions to VNC type/commands/events on the schema
  qapi: add 'If:' section to generated documentation
  qapi-types: add #if conditions to types & visitors
  qapi/events: add #if conditions to events
  qapi/commands: add #if conditions to commands
  qapi-introspect: add preprocessor conditions to generated QLit
  qapi-introspect: modify to_qlit() to append ',' on level > 0
  qapi: add #if/#endif helpers
  qapi: mcgen() shouldn't indent # lines
  qapi: add 'ifcond' to visitor methods
  qapi: leave the ifcond attribute undefined until check()
  qapi: pass 'if' condition into QAPISchemaEntity objects
  qapi: add 'if' to top-level expressions

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/mdroth/tags/qga-pull-2018-07-03-tag' into staging
Peter Maydell [Thu, 5 Jul 2018 07:21:25 +0000 (08:21 +0100)]
Merge remote-tracking branch 'remotes/mdroth/tags/qga-pull-2018-07-03-tag' into staging

qemu-ga patch queue for soft-freeze

* add systemd suspend support
* add used/total space stats for guest-get-fsinfo
* fixes for guest-get-fsinfo over PCI bridges
* MSI installer and schema doc fixes
* guard against unbounded allocations in guest-file-read
* add some additional qga test cases

# gpg: Signature made Tue 03 Jul 2018 21:45:32 BST
# gpg:                using RSA key 3353C9CEF108B584
# gpg: Good signature from "Michael Roth <flukshun@gmail.com>"
# gpg:                 aka "Michael Roth <mdroth@utexas.edu>"
# gpg:                 aka "Michael Roth <mdroth@linux.vnet.ibm.com>"
# Primary key fingerprint: CEAC C9E1 5534 EBAB B82D  3FA0 3353 C9CE F108 B584

* remotes/mdroth/tags/qga-pull-2018-07-03-tag:
  qga: removing bios_supports_mode
  qga: systemd hibernate/suspend/hybrid-sleep support
  qga: removing switch statements, adding run_process_child
  qga: guest_suspend: decoupling pm-utils and sys logic
  qga: bios_supports_mode: decoupling pm-utils and sys logic
  qga: refactoring qmp_guest_suspend_* functions
  qemu-ga: make get-fsinfo work over pci bridges
  qga-win: Fixing msi upgrade disallow in WiX file
  qga/schema: fix documentation for GuestOSInfo
  test-qga: add trivial tests for some commands
  qga-win: add driver path usage to GuestFilesystemInfo
  qga: add mountpoint usage info to GuestFilesystemInfo
  qga: check bytes count read by guest-file-read
  qga: unset frozen state if no mount points are frozen

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/vivier2/tags/linux-user-for-3.0-pull-request...
Peter Maydell [Wed, 4 Jul 2018 21:38:10 +0000 (22:38 +0100)]
Merge remote-tracking branch 'remotes/vivier2/tags/linux-user-for-3.0-pull-request' into staging

Some strace cleanups
Minor fixes for setsockopt() and netlink
Manage 'F' flag with qemu-binfmt-conf.sh

# gpg: Signature made Tue 03 Jul 2018 17:34:20 BST
# gpg:                using RSA key F30C38BD3F2FBE3C
# gpg: Good signature from "Laurent Vivier <lvivier@redhat.com>"
# gpg:                 aka "Laurent Vivier <laurent@vivier.eu>"
# gpg:                 aka "Laurent Vivier (Red Hat) <lvivier@redhat.com>"
# Primary key fingerprint: CD2F 75DD C8E3 A4DC 2E4F  5173 F30C 38BD 3F2F BE3C

* remotes/vivier2/tags/linux-user-for-3.0-pull-request:
  linux-user/strace: Improve recvmsg() output
  linux-user/strace: Improve capget()/capset() output
  linux-user: update do_setsockopt()
  linux-user: add netlink CARRIER_UP_COUNT/CARRIER_DOWN_COUNT
  qemu-binfmt-conf.sh: allow to provide a suffix to the interpreter name
  qemu-binfmt-conf.sh: add persistent (F) flags
  qemu-binfmt-conf.sh: cleanup --credential

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/kraxel/tags/audio-20180703-pull-request' into...
Peter Maydell [Wed, 4 Jul 2018 06:28:11 +0000 (07:28 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/audio-20180703-pull-request' into staging

audio: hda fixes, timer tracing

# gpg: Signature made Tue 03 Jul 2018 13:10:03 BST
# 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-20180703-pull-request:
  audio: add audio timer trace points
  audio/hda: fix CID 1393631
  audio/hda: adjust larger gaps faster

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agodirty-bitmap: fix double lock on bitmap enabling
Vladimir Sementsov-Ogievskiy [Wed, 4 Jul 2018 06:12:49 +0000 (02:12 -0400)]
dirty-bitmap: fix double lock on bitmap enabling

Bitmap lock/unlock were added to bdrv_enable_dirty_bitmap in
8b1402ce80d, but some places were not updated correspondingly, which
leads to trying to take this lock twice, which is dead-lock. Fix this.

Actually, iotest 199 (about dirty bitmap postcopy migration) is broken
now, and this fixes it.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-id: 20180625165745.25259-3-vsementsov@virtuozzo.com
Signed-off-by: John Snow <jsnow@redhat.com>
6 years agoblock/dirty-bitmap: add bdrv_enable_dirty_bitmap_locked
Vladimir Sementsov-Ogievskiy [Wed, 4 Jul 2018 06:12:49 +0000 (02:12 -0400)]
block/dirty-bitmap: add bdrv_enable_dirty_bitmap_locked

Add _locked version of bdrv_enable_dirty_bitmap, to fix dirty bitmap
migration in the following patch.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-id: 20180625165745.25259-2-vsementsov@virtuozzo.com
Signed-off-by: John Snow <jsnow@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/berrange/tags/qcrypto-next-pull-request' into...
Peter Maydell [Tue, 3 Jul 2018 22:06:18 +0000 (23:06 +0100)]
Merge remote-tracking branch 'remotes/berrange/tags/qcrypto-next-pull-request' into staging

Add support for PSK credentials with TLS

# gpg: Signature made Tue 03 Jul 2018 13:04:51 BST
# gpg:                using RSA key BE86EBB415104FDF
# gpg: Good signature from "Daniel P. Berrange <dan@berrange.com>"
# gpg:                 aka "Daniel P. Berrange <berrange@redhat.com>"
# Primary key fingerprint: DAF3 A6FD B26B 6291 2D0E  8E3F BE86 EBB4 1510 4FDF

* remotes/berrange/tags/qcrypto-next-pull-request:
  crypto: Implement TLS Pre-Shared Keys (PSK).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoqapi: Polish command flags documentation in qapi-code-gen.txt
Markus Armbruster [Tue, 3 Jul 2018 08:53:58 +0000 (10:53 +0200)]
qapi: Polish command flags documentation in qapi-code-gen.txt

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-33-armbru@redhat.com>

6 years agomonitor: Improve some comments
Markus Armbruster [Tue, 3 Jul 2018 08:53:57 +0000 (10:53 +0200)]
monitor: Improve some comments

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-32-armbru@redhat.com>

6 years agoqmp: Clean up capability negotiation after commit 02130314d8c
Markus Armbruster [Tue, 3 Jul 2018 08:53:56 +0000 (10:53 +0200)]
qmp: Clean up capability negotiation after commit 02130314d8c

qmp_greeting() offers capabilities to the client, and
qmp_qmp_capabilities() accepts or denies capabilities requested by the
client.  The two compute the set of available capabilities
independently.  Not nice.

Clean this up as follows.  Compute available capabilities just once in
monitor_qmp_caps_reset(), and store them in Monitor member
qmp.capab_offered[].  Have qmp_greeting() and qmp_qmp_capabilities()
use that.  Both are now oblivious of capability details.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-31-armbru@redhat.com>

6 years agoqobject: Let qobject_from_jsonf() fail instead of abort
Markus Armbruster [Tue, 3 Jul 2018 08:53:55 +0000 (10:53 +0200)]
qobject: Let qobject_from_jsonf() fail instead of abort

qobject_from_jsonf() aborts on error, unlike qobject_from_jsonv(),
which returns null.  Since all remaining users of qobject_from_jsonf()
cope fine with null, change it to return null.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-30-armbru@redhat.com>

6 years agoqmp: Switch timestamp_put() to qdict_from_jsonf_nofail()
Markus Armbruster [Tue, 3 Jul 2018 08:53:54 +0000 (10:53 +0200)]
qmp: Switch timestamp_put() to qdict_from_jsonf_nofail()

There's just one use of qobject_from_jsonf() to parse a JSON object
left: timestamp_put().  Switch it to qdict_from_jsonf_nofail().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-29-armbru@redhat.com>

6 years agoqmp: Add some comments around null responses
Markus Armbruster [Tue, 3 Jul 2018 08:53:53 +0000 (10:53 +0200)]
qmp: Add some comments around null responses

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-28-armbru@redhat.com>

6 years agoqmp: Simplify monitor_qmp_respond()
Markus Armbruster [Tue, 3 Jul 2018 08:53:52 +0000 (10:53 +0200)]
qmp: Simplify monitor_qmp_respond()

monitor_qmp_respond() takes both a response object and an error
object.  If an error object is non-null, the response object must be
null, and the response is built from the error object.

Of the two callers, one always passes a null response object, and one
a null error object.  Move building the response object from the error
object to the latter, and drop the error object parameter.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-27-armbru@redhat.com>

6 years agoqmp: Replace get_qmp_greeting() by qmp_greeting()
Markus Armbruster [Tue, 3 Jul 2018 08:53:51 +0000 (10:53 +0200)]
qmp: Replace get_qmp_greeting() by qmp_greeting()

get_qmp_greeting() returns a QDict * as QObject *.  It's caller
converts it right back.

Return QDict * instead.  While there, rename to qmp_greeting().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-26-armbru@redhat.com>

6 years agoqmp: Replace monitor_json_emitter{,raw}() by qmp_{queue,send}_response()
Markus Armbruster [Tue, 3 Jul 2018 08:53:50 +0000 (10:53 +0200)]
qmp: Replace monitor_json_emitter{,raw}() by qmp_{queue,send}_response()

monitor_json_emitter() and monitor_json_emitter_raw() are
unnecessarily general: they can send arbitrary JSON values, even
though we only ever use them for QMP, which may send only JSON
objects.

Specialize the argument from QObject * to QDict *, and rename to
qmp_queue_response(), qmp_send_response().

All callers but one lose an upcast.  The lone exception gains a
downcast; the next commit will get rid of it.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-25-armbru@redhat.com>

6 years agoqmp: Use QDict * instead of QObject * for response objects
Markus Armbruster [Tue, 3 Jul 2018 08:53:49 +0000 (10:53 +0200)]
qmp: Use QDict * instead of QObject * for response objects

By using the more specific type, we get fewer downcasts.  The
downcasts are safe, but not obviously so, at least not locally.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-24-armbru@redhat.com>

6 years agoqmp: De-duplicate error response building
Markus Armbruster [Tue, 3 Jul 2018 08:53:48 +0000 (10:53 +0200)]
qmp: De-duplicate error response building

All callers of qmp_build_error_object() duplicate the code to wrap it
in a response object.  Replace it by qmp_error_response() that
captures the duplicated code, including error_free().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-23-armbru@redhat.com>

6 years agoqobject: New qdict_from_jsonf_nofail()
Markus Armbruster [Tue, 3 Jul 2018 08:53:47 +0000 (10:53 +0200)]
qobject: New qdict_from_jsonf_nofail()

Many uses of qobject_from_jsonf() convert JSON objects.  Create new
convenience function qdict_from_jsonf_nofail() that includes the
conversion to QDict.  The next few commits will put it to use.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-22-armbru@redhat.com>

6 years agomonitor: Peel off @mon_global wrapper
Markus Armbruster [Tue, 3 Jul 2018 08:53:46 +0000 (10:53 +0200)]
monitor: Peel off @mon_global wrapper

Wrapping global variables in a struct without a use for the wrapper
struct buys us nothing but longer lines.  Unwrap them.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-21-armbru@redhat.com>

6 years agomonitor: Rename use_io_thr to use_io_thread
Markus Armbruster [Tue, 3 Jul 2018 08:53:45 +0000 (10:53 +0200)]
monitor: Rename use_io_thr to use_io_thread

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-20-armbru@redhat.com>

6 years agoqmp: Don't let JSON errors jump the queue
Markus Armbruster [Tue, 3 Jul 2018 08:53:44 +0000 (10:53 +0200)]
qmp: Don't let JSON errors jump the queue

handle_qmp_command() reports JSON syntax errors right away.  This is
wrong when OOB is enabled, because the errors can "jump the queue"
then.

The previous commit fixed the same bug for semantic errors, by
delaying the checking until dispatch.  We can't delay the checking, so
delay the reporting.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-19-armbru@redhat.com>

6 years agoqmp: Don't let malformed in-band commands jump the queue
Markus Armbruster [Tue, 3 Jul 2018 08:53:43 +0000 (10:53 +0200)]
qmp: Don't let malformed in-band commands jump the queue

handle_qmp_command() reports certain errors right away.  This is wrong
when OOB is enabled, because the errors can "jump the queue" then, as
the previous commit demonstrates.

To fix, we need to delay errors until dispatch.  Do that for semantic
errors, mostly by reverting ill-advised parts of commit cf869d53172
"qmp: support out-of-band (oob) execution".  Bonus: doesn't run
qmp_dispatch_check_obj() twice, once in handle_qmp_command(), and
again in do_qmp_dispatch().  That's also due to commit cf869d53172.

The next commit will fix queue jumping for syntax errors.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-18-armbru@redhat.com>

6 years agotests/qmp-test: Demonstrate QMP errors jumping the queue
Markus Armbruster [Tue, 3 Jul 2018 08:53:42 +0000 (10:53 +0200)]
tests/qmp-test: Demonstrate QMP errors jumping the queue

When OOB is enabled, out-of-band commands are executed right away,
everything else is queued.  This lets out-of-band commands "jump the
queue".

However, certain errors are always reported right away, and therefore
can jump the queue even when the erroneous input does not request
out-of-band execution.  These errors are pretty unlikely to occur in
production, but it's wrong all the same.  Mark FIXME.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <20180703085358.13941-17-armbru@redhat.com>

6 years agoqmp: Simplify code around monitor_qmp_dispatch_one()
Markus Armbruster [Tue, 3 Jul 2018 08:53:41 +0000 (10:53 +0200)]
qmp: Simplify code around monitor_qmp_dispatch_one()

Change monitor_qmp_dispatch_one() to take its parameters unwrapped,
move monitor_resume() to the one caller that needs it, rename the
function to monitor_qmp_dispatch().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-16-armbru@redhat.com>

6 years agoqmp: Always free QMPRequest with qmp_request_free()
Markus Armbruster [Tue, 3 Jul 2018 08:53:40 +0000 (10:53 +0200)]
qmp: Always free QMPRequest with qmp_request_free()

monitor_qmp_dispatch_one() frees a QMPRequest manually, because it
needs to keep a reference to ->id.  Premature optimization.  Take an
additional reference so we can use qmp_request_free().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-15-armbru@redhat.com>

6 years agoqmp: Revert change to handle_qmp_command tracepoint
Markus Armbruster [Tue, 3 Jul 2018 08:53:39 +0000 (10:53 +0200)]
qmp: Revert change to handle_qmp_command tracepoint

Commit 71da4667db6 "monitor: separate QMP parser and dispatcher" moved
the handle_qmp_command tracepoint from handle_qmp_command() to
monitor_qmp_dispatch_one().  This delays tracing from enqueue time to
dequeue time.  Revert that.  Dequeue remains adequately visible via
tracepoint monitor_qmp_cmd_in_band.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-14-armbru@redhat.com>

6 years agoqmp: Redo how the client requests out-of-band execution
Markus Armbruster [Tue, 3 Jul 2018 08:53:38 +0000 (10:53 +0200)]
qmp: Redo how the client requests out-of-band execution

Commit cf869d53172 "qmp: support out-of-band (oob) execution" added a
general mechanism for command-independent arguments just for an
out-of-band flag:

    The "control" key is introduced to store this extra flag.  "control"
    field is used to store arguments that are shared by all the commands,
    rather than command specific arguments.  Let "run-oob" be the first.

However, it failed to reject unknown members of "control".  For
instance, in QMP command

    {"execute": "query-name", "id": 42, "control": {"crap": true}}

"crap" gets silently ignored.

Instead of fixing this, revert the general "control" mechanism
(because YAGNI), and do it the way I initially proposed, with key
"exec-oob".  Simpler code, simpler interface.

An out-of-band command

    {"execute": "migrate-pause", "id": 42, "control": {"run-oob": true}}

becomes

    {"exec-oob": "migrate-pause", "id": 42}

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-13-armbru@redhat.com>
[Commit message typo fixed]

6 years agoqmp qemu-ga: Fix qemu-ga not to accept "control"
Markus Armbruster [Tue, 3 Jul 2018 08:53:37 +0000 (10:53 +0200)]
qmp qemu-ga: Fix qemu-ga not to accept "control"

Commit cf869d53172 "qmp: support out-of-band (oob) execution"
accidentally made qemu-ga accept and ignore "control".  Fix that.

Out-of-band execution in a monitor that doesn't support it now fails
with

    {"error": {"class": "GenericError", "desc": "QMP input member 'control' is unexpected"}}

instead of

    {"error": {"class": "GenericError", "desc": "Please enable out-of-band first for the session during capabilities negotiation"}}

The old description is suboptimal when out-of-band cannot not be
enabled, or the command doesn't support out-of-band execution.

The new description is a bit unspecific, but it'll do.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-12-armbru@redhat.com>

6 years agotests/test-qga: Demonstrate the guest-agent ignores "control"
Markus Armbruster [Tue, 3 Jul 2018 08:53:36 +0000 (10:53 +0200)]
tests/test-qga: Demonstrate the guest-agent ignores "control"

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-11-armbru@redhat.com>

6 years agoqmp qemu-ga: Revert change that accidentally made qemu-ga accept "id"
Markus Armbruster [Tue, 3 Jul 2018 08:53:35 +0000 (10:53 +0200)]
qmp qemu-ga: Revert change that accidentally made qemu-ga accept "id"

Commit cf869d53172 "qmp: support out-of-band (oob) execution" changed
how we check "id":

    Note that in the patch I exported qmp_dispatch_check_obj() to be
    used to check the request earlier, and at the same time allowed
    "id" field to be there since actually we always allow that.

The part after "and" is ill-advised: it makes qemu-ga accept and
ignore "id".  Revert.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-10-armbru@redhat.com>

6 years agotests/test-qga: Demonstrate the guest-agent ignores "id"
Markus Armbruster [Tue, 3 Jul 2018 08:53:34 +0000 (10:53 +0200)]
tests/test-qga: Demonstrate the guest-agent ignores "id"

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-9-armbru@redhat.com>

6 years agoqmp: Make "id" optional again even in "oob" monitors
Markus Armbruster [Tue, 3 Jul 2018 08:53:33 +0000 (10:53 +0200)]
qmp: Make "id" optional again even in "oob" monitors

Commit cf869d53172 "qmp: support out-of-band (oob) execution" made
"id" mandatory for all commands when the client accepted capability
"oob".  This is rather onerous when you play with QMP by hand, and
unnecessarily so: only out-of-band commands need an ID for reliable
matching of response to command.

Revert that part of commit cf869d53172 for now, but have documentation
advise on the need to use "id" with out-of-band commands.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703085358.13941-8-armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
6 years agotests/qmp-test: Test in-band command doesn't overtake
Markus Armbruster [Tue, 3 Jul 2018 08:53:32 +0000 (10:53 +0200)]
tests/qmp-test: Test in-band command doesn't overtake

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-7-armbru@redhat.com>

6 years agoqmp: Get rid of x-oob-test command
Markus Armbruster [Tue, 3 Jul 2018 08:53:31 +0000 (10:53 +0200)]
qmp: Get rid of x-oob-test command

tests/qmp-test tests an out-of-band command overtaking a slow in-band
command.  To do that, it needs:

1. An in-band command that *reliably* takes long enough to be
   overtaken.

2. An out-of-band command to do the overtaking.

3. To avoid delays, a way to make the in-band command complete quickly
   after it was overtaken.

To satisfy these needs, commit 469638f9cb3 provides the rather
peculiar oob-capable QMP command x-oob-test:

* With "lock": true, it waits for a global semaphore.

* With "lock": false, it signals the global semaphore.

To satisfy 1., the test runs x-oob-test in-band with "lock": true.
To satisfy 2. and 3., it runs x-oob-test out-of-band with "lock": false.

Note that waiting for a semaphore violates the rules for oob-capable
commands.  Running x-oob-test with "lock": true hangs the monitor
until you run x-oob-test with "lock": false on another monitor (which
you might not have set up).

Having an externally visible QMP command that may hang the monitor is
not nice.  Let's apply a little more ingenuity to the problem.  Idea:
have an existing command block on reading a FIFO special file, unblock
it by opening the FIFO for writing.

For 1., use

    {"execute": "blockdev-add",  "id": ID1,
     "arguments": {
        "driver": "blkdebug", "node-name": ID1, "config": FIFO,
        "image": { "driver": "null-co"}}}

where ID1 is an arbitrary string, and FIFO is the name of the FIFO.

For 2., use

    {"execute": "migrate-pause", "id": ID2, "control": {"run-oob": true}}

where ID2 is a different arbitrary string.  Since there's no migration
to pause, the command will fail, but that's fine; instant failure is
still a test of out-of-band responses overtaking in-band commands.

For 3., open FIFO for writing.

Drop QMP command x-oob-test.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-6-armbru@redhat.com>
[Error checking tweaked]

6 years agoqmp: Document COMMAND_DROPPED design flaw
Markus Armbruster [Tue, 3 Jul 2018 08:53:30 +0000 (10:53 +0200)]
qmp: Document COMMAND_DROPPED design flaw

Events are broadcast to all monitors.  If another monitor's client has
a command with the same ID in flight, the event will incorrectly claim
that command was dropped.  This must be fixed before out-of-band
execution can graduate from "experimental".

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-5-armbru@redhat.com>

6 years agodocs/interop/qmp: Improve OOB documentation
Markus Armbruster [Tue, 3 Jul 2018 08:53:29 +0000 (10:53 +0200)]
docs/interop/qmp: Improve OOB documentation

OOB documentation is spread over qmp-spec.txt sections 2.2.1
Capabilities and 2.3 Issuing Commands.  The amount of detail is a bit
distracting there.  Move the meat of the matter to new section 2.3.1
Out of band execution.

Throw in a few other improvements while there:

* 2.2 Server Greeting: Drop advice to search entire capabilities
  array; should be obvious.

* 3. QMP Examples

  - 3.1 Server Greeting: Update greeting to the one we expect for the
    release.  Now shows capability "oob".  Update qmp-intro.txt
    likewise.

  - 3.2 Capabilities negotiation: Show client accepting capability
    "oob".

  - 3.7 Out-of-band execution: New.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-4-armbru@redhat.com>
[Whitespace tidied up]

6 years agoqga: removing bios_supports_mode
Daniel Henrique Barboza [Thu, 21 Jun 2018 10:21:53 +0000 (07:21 -0300)]
qga: removing bios_supports_mode

bios_support_mode verifies if the guest has support for a certain
suspend mode but it doesn't inform back which suspend tool
provides it. The caller, guest_suspend, executes all suspend
strategies in order again.

After adding systemd suspend support, bios_support_mode now will
verify for support for systemd, then pmutils, then Linux sys state
file. In a worst case scenario where both systemd and pmutils isn't
supported but Linux sys state is:

- bios_supports_mode will check for systemd, then pmutils, then
Linux sys state. It will tell guest_suspend that there is support,
but it will not tell who provides it;

- guest_suspend will try to execute (and fail) systemd suspend,
then pmutils suspend, to only then use the Linux sys suspend.
The time spent executing systemd and pmutils suspend was wasted
and could be avoided, but only bios_support_mode knew it but
didn't inform it back.

A quicker approach is to nuke bios_supports_mode and control
whether we found support at all with a bool flag inside
guest_suspend. guest_suspend will search for suspend support
and execute it as soon as possible. If the a given suspend
mechanism fails, continue to the next. If no suspend
support is found, the "not supported" message is still being
sent back to the user.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga: systemd hibernate/suspend/hybrid-sleep support
Daniel Henrique Barboza [Thu, 21 Jun 2018 10:21:52 +0000 (07:21 -0300)]
qga: systemd hibernate/suspend/hybrid-sleep support

pmutils isn't being supported by newer OSes like Fedora 27
or Mint. This means that the only suspend option QGA offers
for these guests are writing directly into the Linux sys state
file. This also means that QGA also loses the ability to do
hybrid suspend in those guests - this suspend mode is only
available when using pmutils.

Newer guests can use systemd facilities to do all the suspend
types QGA supports. The mapping in comparison with pmutils is:

- pm-hibernate -> systemctl hibernate
- pm-suspend -> systemctl suspend
- pm-suspend-hybrid -> systemctl hybrid-sleep

To discover whether systemd supports these functions, we inspect
the status of the services that implements them.

With this patch, we can offer hybrid suspend again for newer
guests that do not have pmutils support anymore.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga: removing switch statements, adding run_process_child
Daniel Henrique Barboza [Thu, 21 Jun 2018 10:21:51 +0000 (07:21 -0300)]
qga: removing switch statements, adding run_process_child

This is a cleanup of the resulting code after detaching
pmutils and Linux sys state file logic:

- remove the SUSPEND_MODE_* macros and use an enumeration
instead. At the same time, drop the switch statements
at the start of each function and use the enumeration
index to get the right binary/argument;

- create a new function called run_process_child(). This
function uses g_spawn_sync() to execute a shell command,
returning the exit code. This is a common operation in the
pmutils functions and will be used in the systemd implementation
as well, so this function will avoid code repetition.

There are more places inside commands-posix.c where this new
run_process_child function can also be used, but one step
at a time.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
*check/propagate local_err before setting errp directly
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga: guest_suspend: decoupling pm-utils and sys logic
Daniel Henrique Barboza [Thu, 21 Jun 2018 10:21:50 +0000 (07:21 -0300)]
qga: guest_suspend: decoupling pm-utils and sys logic

Following the same logic of the previous patch, let's also
decouple the suspend logic from guest_suspend into specialized
functions, one for each strategy we support at this moment.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga: bios_supports_mode: decoupling pm-utils and sys logic
Daniel Henrique Barboza [Thu, 21 Jun 2018 10:21:49 +0000 (07:21 -0300)]
qga: bios_supports_mode: decoupling pm-utils and sys logic

In bios_supports_mode there is a verification to assert if
the chosen suspend mode is supported by the pmutils tools and,
if not, we see if the Linux sys state files supports it.

This verification is done in the same function, one after
the other, and it works for now. But, when adding a new
suspend mechanism that will not necessarily follow the same
return 0 or 1 logic of pmutils, this code will be hard
to deal with.

This patch decouple the two existing logics into their own
functions, pmutils_supports_mode and linux_sys_state_supports_mode,
which in turn are used inside bios_support_mode. The existing
logic is kept but now it's easier to extend it.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga: refactoring qmp_guest_suspend_* functions
Daniel Henrique Barboza [Thu, 21 Jun 2018 10:21:48 +0000 (07:21 -0300)]
qga: refactoring qmp_guest_suspend_* functions

To be able to add new suspend mechanisms we need to detach
the existing QMP functions from the current implementation
specifics.

At this moment we have functions such as qmp_guest_suspend_ram
calling bios_suspend_mode and guest_suspend passing the
pmutils command and arguments as parameters. This patch
removes this logic from the QMP functions, moving them to
the respective functions that will have to deal with which
binary to use.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqemu-ga: make get-fsinfo work over pci bridges
Marc-André Lureau [Mon, 14 May 2018 12:38:55 +0000 (14:38 +0200)]
qemu-ga: make get-fsinfo work over pci bridges

Iterate over the PCI bridges to lookup the PCI device associated with
the block device.

This allows to lookup the driver under the following syspath:
/sys/devices/pci0000:00/0000:00:02.2/0000:03:00.0/virtio2/block/vda/vda3

It also works with an "old-style" Q35 libvirt hierarchy: root complex
-> DMI-PCI bridge -> PCI-PCI bridge -> virtio controller, ex:
/sys/devices/pci0000:00/0000:00:03.0/0000:01:01.0/0000:02:01.0/virtio1/block/vda/vda3

The setup can be reproduced with the following qemu command line
(Thanks Marcel for help):

qemu-system-x86_64 -M q35 \
  -device i82801b11-bridge,id=dmi2pci_bridge,bus=pcie.0
  -device pci-bridge,id=pci_bridge,bus=dmi2pci_bridge,addr=0x1,chassis_nr=1
  -device virtio-blk-pci,scsi=off,drive=drive-virtio-disk0,id=virtio-disk0,bootindex=1,bus=pci_bridge,addr=0x1

For consistency with other syspath-related debug messages, replace a
\"%s\" in the message with '%s'.

Fixes:
https://bugzilla.redhat.com/show_bug.cgi?id=1567041

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga-win: Fixing msi upgrade disallow in WiX file
Bishara AbuHattoum [Thu, 28 Jun 2018 14:58:16 +0000 (17:58 +0300)]
qga-win: Fixing msi upgrade disallow in WiX file

Issue: When upgrading qemu-ga using the msi from an old version
       to a newer one, the upgrade is not allowed by the msi
       showing this error message "Another version of this product
       is already installed."

BZ# 1536331: https://bugzilla.redhat.com/show_bug.cgi?id=1536331

Fix: For the upgrade to be allowed by the msi the WiX file must
     provide three things:
      1. Changing product's Id. (assigning it to "*")
      2. Constant product's UpgradeId. (exists)
      3. Changing version. (exists)

Reference: http://wixtoolset.org/documentation/manual/v3/howtos/updates/major_upgrade.html

Signed-off-by: Bishara AbuHattoum <bishara@daynix.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga/schema: fix documentation for GuestOSInfo
Tomáš Golembiovský [Fri, 20 Apr 2018 18:12:45 +0000 (20:12 +0200)]
qga/schema: fix documentation for GuestOSInfo

The documentation for kernel-version and kernel-release on Windows was
swapped.

Signed-off-by: Tomáš Golembiovský <tgolembi@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agotest-qga: add trivial tests for some commands
Tomáš Golembiovský [Fri, 20 Apr 2018 18:04:34 +0000 (20:04 +0200)]
test-qga: add trivial tests for some commands

These commands did not get their tests in the original commits:
- guest-get-host-name
- guest-get-timezone
- guest-get-users

Trivial tests that mostly only call the commands were added.

Signed-off-by: Tomáš Golembiovský <tgolembi@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
* replace QDECREF() with qobject_unref()
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoMerge remote-tracking branch 'remotes/kraxel/tags/vga-20180703-pull-request' into...
Peter Maydell [Tue, 3 Jul 2018 20:09:27 +0000 (21:09 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/vga-20180703-pull-request' into staging

vga: disable global_vmstate, virtio-gpu scanout tracking fixes.

# gpg: Signature made Tue 03 Jul 2018 10:44:56 BST
# 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-20180703-pull-request:
  vga: disable global_vmstate for 3.0+ machine types
  virtio-gpu: disable scanout when backing resource is destroyed
  virtio-gpu: update old resource too.
  virtio-gpu: tweak scanout disable.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
# Conflicts:
# hw/display/qxl.c
# hw/display/vga-isa-mm.c

6 years agoMerge remote-tracking branch 'remotes/kraxel/tags/ui-20180703-pull-request' into...
Peter Maydell [Tue, 3 Jul 2018 17:43:32 +0000 (18:43 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/ui-20180703-pull-request' into staging

ui: drop libdrm dependency.

# gpg: Signature made Tue 03 Jul 2018 10:18:36 BST
# 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-20180703-pull-request:
  ui: do not build-depend or link with libdrm, it is not needed

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoqapi: add conditions to SPICE type/commands/events on the schema
Marc-André Lureau [Tue, 3 Jul 2018 15:56:48 +0000 (17:56 +0200)]
qapi: add conditions to SPICE type/commands/events on the schema

Add #if defined(CONFIG_SPICE) in generated code, and adjust the
qmp/hmp code accordingly.

query-qmp-schema no longer reports the command/events etc as
available when disabled at compile time.

Commands made conditional:

* query-spice

  Before the patch, the command for !CONFIG_SPICE is unregistered. It
  will fail with the same error.

Events made conditional:

* SPICE_CONNECTED, SPICE_INITIALIZED, SPICE_DISCONNECTED,
  SPICE_MIGRATE_COMPLETED

Add TODO for conditional SPICE chardevs, delayed until the supports
for conditional members lands.

No HMP change, the code was already conditional.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-15-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi: add conditions to VNC type/commands/events on the schema
Marc-André Lureau [Tue, 3 Jul 2018 15:56:47 +0000 (17:56 +0200)]
qapi: add conditions to VNC type/commands/events on the schema

Add #if defined(CONFIG_VNC) in generated code, and adjust the
qmp/hmp code accordingly.

query-qmp-schema no longer reports the command/events etc as
available when disabled at compile.

Commands made conditional:

* query-vnc, query-vnc-servers, change-vnc-password

  Before the patch, the commands for !CONFIG_VNC are stubs that fail
  like this:

    {"error": {"class": "GenericError",
               "desc": "The feature 'vnc' is not enabled"}}

  Afterwards, they fail like this:

    {"error": {"class": "CommandNotFound",
               "desc": "The command FOO has not been found"}}

  I call that an improvement, because it lets clients distinguish
  between command unavailable (class CommandNotFound) and command failed
  (class GenericError).

Events made conditional:

* VNC_CONNECTED, VNC_INITIALIZED, VNC_DISCONNECTED

HMP change:

* info vnc

  Will return "unknown command: 'info vnc'" when VNC is compiled
  out (same as error for spice when --disable-spice)

Occurrences of VNC (case insensitive) in the schema that aren't
covered by this change:

* add_client

  Command has other uses, including "socket bases character devices".
  These are unconditional as far as I can tell.

* set_password, expire_password

  In theory, these commands could be used for managing any service's
  password.  In practice, they're used for VNC and SPICE services.
  They're documented for "remote display session" / "remote display
  server".

  The service is selected by argument @protocol.  The code special-cases
  protocol-specific argument checking, then calls a protocol-specific
  function to do the work.  If it fails, the command fails with "Could
  not set password".  It does when the service isn't compiled in (it's a
  stub then).

  We could make these commands conditional on the conjunction of all
  services [currently: defined(CONFIG_VNC) || defined(CONFIG_SPICE)],
  but I doubt it's worthwhile.

* change

  Command has other uses, namely changing media.
  This patch inlines a stub; no functional change.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-14-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi: add 'If:' section to generated documentation
Marc-André Lureau [Tue, 3 Jul 2018 15:56:46 +0000 (17:56 +0200)]
qapi: add 'If:' section to generated documentation

The documentation is generated only once, and doesn't know C
pre-conditions. Add 'If:' sections for top-level entities.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-13-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi-types: add #if conditions to types & visitors
Markus Armbruster [Tue, 3 Jul 2018 15:56:45 +0000 (17:56 +0200)]
qapi-types: add #if conditions to types & visitors

Types & visitors are coupled and must be handled together to avoid
temporary build regression.

Wrap generated types/visitor code with #if/#endif using the context
helpers. Derived from a patch by Marc-André.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-12-marcandre.lureau@redhat.com>

6 years agoqapi/events: add #if conditions to events
Marc-André Lureau [Tue, 3 Jul 2018 15:56:44 +0000 (17:56 +0200)]
qapi/events: add #if conditions to events

Wrap generated code with #if/#endif using an 'ifcontext' on
QAPIGenCSnippet objects.

This makes a conditional event's qapi_event_send_FOO() compile-time
conditional, but its enum QAPIEvent member remains unconditional for
now. A follow up patch "qapi-event: add 'if' condition to implicit
event enum" will improve this.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-11-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi/commands: add #if conditions to commands
Marc-André Lureau [Tue, 3 Jul 2018 15:56:43 +0000 (17:56 +0200)]
qapi/commands: add #if conditions to commands

Wrap generated code with #if/#endif using an 'ifcontext' on
QAPIGenCSnippet objects.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20180703155648.11933-10-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Line breaks tweaked]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi-introspect: add preprocessor conditions to generated QLit
Marc-André Lureau [Tue, 3 Jul 2018 15:56:42 +0000 (17:56 +0200)]
qapi-introspect: add preprocessor conditions to generated QLit

This commit adds 'ifcond' conditions to top-level QLit objects.
Future work will add them to object and enum type members, i.e. within
QLit objects.

Extend the QLit generator to_qlit() to accept (@obj, @cond) tuples in
addition to just @obj.  The tuple causes the QLit generated for
objects for @obj with #if/#endif conditions for @cond.

See generated tests/test-qmp-introspect.c. Example diff after this
patch:

    --- before 2018-01-08 11:55:24.757083654 +0100
    +++ tests/test-qmp-introspect.c 2018-01-08 13:08:44.477641629 +0100
    @@ -51,6 +51,8 @@
             { "name", QLIT_QSTR("EVENT_F"), },
             {}
         })),
    +#if defined(TEST_IF_CMD)
    +#if defined(TEST_IF_STRUCT)
         QLIT_QDICT(((QLitDictEntry[]) {
             { "arg-type", QLIT_QSTR("5"), },
             { "meta-type", QLIT_QSTR("command"), },
    @@ -58,12 +60,16 @@
             { "ret-type", QLIT_QSTR("0"), },
             {}
         })),
    +#endif /* defined(TEST_IF_STRUCT) */
    +#endif /* defined(TEST_IF_CMD) */

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-9-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi-introspect: modify to_qlit() to append ',' on level > 0
Marc-André Lureau [Tue, 3 Jul 2018 15:56:41 +0000 (17:56 +0200)]
qapi-introspect: modify to_qlit() to append ',' on level > 0

The following patch is going to break list entries with #if/#endif, so
they should have the trailing ',' as suffix.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-8-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi: add #if/#endif helpers
Marc-André Lureau [Tue, 3 Jul 2018 15:56:40 +0000 (17:56 +0200)]
qapi: add #if/#endif helpers

Add helpers to wrap generated code with #if/#endif lines.

A later patch wants to use QAPIGen for generating C snippets rather
than full C files with copyright headers etc.  Splice in class
QAPIGenCCode between QAPIGen and QAPIGenC.

Add a 'with' statement context manager that will be used to wrap
generator visitor methods.  The manager will check if code was
generated before adding #if/#endif lines on QAPIGenCSnippet
objects. Used in the following patches.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20180703155648.11933-7-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi: mcgen() shouldn't indent # lines
Marc-André Lureau [Tue, 3 Jul 2018 15:56:39 +0000 (17:56 +0200)]
qapi: mcgen() shouldn't indent # lines

Skip preprocessor lines when adding indentation, since that would
likely result in invalid code.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-6-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi: add 'ifcond' to visitor methods
Marc-André Lureau [Tue, 3 Jul 2018 15:56:38 +0000 (17:56 +0200)]
qapi: add 'ifcond' to visitor methods

Modify the test visitor to check correct passing of values.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-5-marcandre.lureau@redhat.com>
[Accidental change to roms/seabios dropped]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi: leave the ifcond attribute undefined until check()
Marc-André Lureau [Tue, 3 Jul 2018 15:56:37 +0000 (17:56 +0200)]
qapi: leave the ifcond attribute undefined until check()

We commonly initialize attributes to None in .init(), then set their
real value in .check().  Accessing the attribute before .check()
yields None.  If we're lucky, the code that accesses the attribute
prematurely chokes on None.

It won't for .ifcond, because None is a legitimate value.

Leave the ifcond attribute undefined until check().

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-4-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqapi: pass 'if' condition into QAPISchemaEntity objects
Marc-André Lureau [Tue, 3 Jul 2018 15:56:36 +0000 (17:56 +0200)]
qapi: pass 'if' condition into QAPISchemaEntity objects

Built-in objects remain unconditional.  Explicitly defined objects use
the condition specified in the schema.  Implicitly defined objects
inherit their condition from their users.  For most of them, there is
exactly one user, so the condition to use is obvious.  The exception
is wrapped types generated for simple union variants, which can be
shared by any number of simple unions.  The tight condition would be
the disjunction of the conditions of these simple unions.  For now,
use the wrapped type's condition instead.  Much simpler and good
enough for now.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-3-marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoqga-win: add driver path usage to GuestFilesystemInfo
Chen Hanxiao [Thu, 14 Jun 2018 08:06:07 +0000 (16:06 +0800)]
qga-win: add driver path usage to GuestFilesystemInfo

This patch adds support for getting the usage of
windows driver path.
The usage of fs stored as used_bytes and total_bytes.

Cc: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Chen Hanxiao <chenhanxiao@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga: add mountpoint usage info to GuestFilesystemInfo
Chen Hanxiao [Thu, 14 Jun 2018 08:06:06 +0000 (16:06 +0800)]
qga: add mountpoint usage info to GuestFilesystemInfo

This patch adds support for getting the usage of mounted
filesystem.
The usage of fs stored as used_bytes and total_bytes.
It's very useful when we try to monitor guest's filesystem.

Cc: Michael Roth <mdroth@linux.vnet.ibm.com>
Cc: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Chen Hanxiao <chenhanxiao@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqapi: add 'if' to top-level expressions
Marc-André Lureau [Tue, 3 Jul 2018 15:56:35 +0000 (17:56 +0200)]
qapi: add 'if' to top-level expressions

Accept 'if' key in top-level elements, accepted as string or list of
string type. The following patches will modify the test visitor to
check the value is correctly saved, and generate #if/#endif code (as a
single #if/endif line or a series for a list).

Example of 'if' key:
{ 'struct': 'TestIfStruct', 'data': { 'foo': 'int' },
  'if': 'defined(TEST_IF_STRUCT)' }

The generated code is for now *unconditional*. Later patches generate
the conditionals.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180703155648.11933-2-marcandre.lureau@redhat.com>
[Commit message and Documentation improved]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/kraxel/tags/usb-20180703-pull-request' into...
Peter Maydell [Tue, 3 Jul 2018 16:10:39 +0000 (17:10 +0100)]
Merge remote-tracking branch 'remotes/kraxel/tags/usb-20180703-pull-request' into staging

usb: bugfixes for ehci and xhci.

# gpg: Signature made Tue 03 Jul 2018 09:26:20 BST
# 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/usb-20180703-pull-request:
  xhci: fix guest-triggerable assert
  ehci: Don't fetch a NULL current qtd but advance the queue instead.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoqga: check bytes count read by guest-file-read
Prasad J Pandit [Wed, 13 Jun 2018 06:16:57 +0000 (11:46 +0530)]
qga: check bytes count read by guest-file-read

While reading file content via 'guest-file-read' command,
'qmp_guest_file_read' routine allocates buffer of count+1
bytes. It could overflow for large values of 'count'.
Add check to avoid it.

Reported-by: Fakhri Zulkifli <mohdfakhrizulkifli@gmail.com>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Cc: qemu-stable@nongnu.org
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoqga: unset frozen state if no mount points are frozen
Chen Hanxiao [Thu, 14 Jun 2018 08:10:13 +0000 (16:10 +0800)]
qga: unset frozen state if no mount points are frozen

If we set mountpoints to qmp_guest_fsfreeze_freeze_list,
we may got nothing to freeze as all mountpoints are
not valid.
So call ga_unset_frozen in this senario.

Also, if we return 0 frozen fs, there is no need to call
guest-fsfreeze-thaw.

Cc: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Chen Hanxiao <chenhanxiao@gmail.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
6 years agoMerge remote-tracking branch 'remotes/shorne/tags/pull-or-20180703' into staging
Peter Maydell [Tue, 3 Jul 2018 15:04:41 +0000 (16:04 +0100)]
Merge remote-tracking branch 'remotes/shorne/tags/pull-or-20180703' into staging

OpenRISC cleanups and Fixes for QEMU 3.0

Mostly patches from Richard Henderson fixing multiple things:
 * Fix singlestepping in GDB.
 * Use more TB linking.
 * Fixes to exit TB after updating SPRs to enable registering of state
   changes.
 * Significant optimizations and refactors to the TLB
 * Split out disassembly from translation.
 * Add qemu-or1k to qemu-binfmt-conf.sh.
 * Implement signal handling for linux-user.

Then there are a few fixups from me:
 * Fix delay slot detections to match hardware, this was masking a bug
   in the linus kernel.
 * Fix stores to the PIC mask register

# gpg: Signature made Tue 03 Jul 2018 14:44:10 BST
# gpg:                using RSA key C3B31C2D5E6627E4
# gpg: Good signature from "Stafford Horne <shorne@gmail.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: D9C4 7354 AEF8 6C10 3A25  EFF1 C3B3 1C2D 5E66 27E4

* remotes/shorne/tags/pull-or-20180703: (25 commits)
  target/openrisc: Fix writes to interrupt mask register
  target/openrisc: Fix delay slot exception flag to match spec
  linux-user: Fix struct sigaltstack for openrisc
  linux-user: Implement signals for openrisc
  target/openrisc: Add support in scripts/qemu-binfmt-conf.sh
  target/openrisc: Reorg tlb lookup
  target/openrisc: Increase the TLB size
  target/openrisc: Stub out handle_mmu_fault for softmmu
  target/openrisc: Use identical sizes for ITLB and DTLB
  target/openrisc: Fix cpu_mmu_index
  target/openrisc: Fix tlb flushing in mtspr
  target/openrisc: Reduce tlb to a single dimension
  target/openrisc: Merge mmu_helper.c into mmu.c
  target/openrisc: Remove indirect function calls for mmu
  target/openrisc: Merge tlb allocation into CPUOpenRISCState
  target/openrisc: Form the spr index from tcg
  target/openrisc: Exit the TB after l.mtspr
  target/openrisc: Split out is_user
  target/openrisc: Link more translation blocks
  target/openrisc: Fix singlestep_enabled
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agolinux-user/strace: Improve recvmsg() output
Philippe Mathieu-Daudé [Thu, 28 Jun 2018 03:46:46 +0000 (00:46 -0300)]
linux-user/strace: Improve recvmsg() output

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-By: Guido Günther <agx@sigxcpu.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20180628034652.24152-7-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
6 years agolinux-user/strace: Improve capget()/capset() output
Philippe Mathieu-Daudé [Thu, 28 Jun 2018 03:46:42 +0000 (00:46 -0300)]
linux-user/strace: Improve capget()/capset() output

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-By: Guido Günther <agx@sigxcpu.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20180628034652.24152-3-f4bug@amsat.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
6 years agoMerge remote-tracking branch 'remotes/dgibson/tags/ppc-for-3.0-20180703' into staging
Peter Maydell [Tue, 3 Jul 2018 13:59:27 +0000 (14:59 +0100)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-3.0-20180703' into staging

ppc patch queue 2018-07-03

Here's a last minue pull request before today's soft freeze.  Ideally
I would have sent this earlier, but I was waiting for a couple of
extra fixes I knew were close.  And the freeze crept up on me, like
always.

Most of the changes here are bugfixes in any case.  There are some
cleanups as well, which have been in my staging tree for a little
while.  There are a couple of truly new features (some extensions to
the sam460ex platform), but these are low risk, since they only affect
a new and not really stabilized machine type anyway.

Higlights are:
  * Mac platform improvements from Mark Cave-Ayland
  * Sam460ex improvements from BALATON Zoltan et al.
  * XICS interrupt handler cleanups from Cédric Le Goater
  * TCG improvements for atomic loads and stores from Richard
    Henderson
  * Assorted other bugfixes

# gpg: Signature made Tue 03 Jul 2018 06:55:22 BST
# gpg:                using RSA key 6C38CACA20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# gpg:                 aka "David Gibson (kernel.org) <dwg@kernel.org>"
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* remotes/dgibson/tags/ppc-for-3.0-20180703: (35 commits)
  ppc: Include vga cirrus card into the compiling process
  target/ppc: Relax reserved bitmask of indexed store instructions
  target/ppc: set is_jmp on ppc_tr_breakpoint_check
  spapr: compute default value of "hpt-max-page-size" later
  target/ppc/kvm: don't pass cpu to kvm_get_smmu_info()
  target/ppc/kvm: get rid of kvm_get_fallback_smmu_info()
  ppc440_uc: Basic emulation of PPC440 DMA controller
  sam460ex: Add RTC device
  hw/timer: Add basic M41T80 emulation
  ppc4xx_i2c: Rewrite to model hardware more closely
  hw/ppc: Give sam46ex its own config option
  fpu_helper.c: fix setting FPSCR[FI] bit
  target/ppc: Implement the rest of gen_st_atomic
  target/ppc: Implement the rest of gen_ld_atomic
  target/ppc: Use atomic min/max helpers
  target/ppc: Use MO_ALIGN for EXIWX and ECOWX
  target/ppc: Split out gen_st_atomic
  target/ppc: Split out gen_ld_atomic
  target/ppc: Split out gen_load_locked
  target/ppc: Tidy gen_conditional_store
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
# Conflicts:
# hw/ppc/spapr.c

6 years agotarget/openrisc: Fix writes to interrupt mask register
Stafford Horne [Sun, 1 Jul 2018 08:02:54 +0000 (17:02 +0900)]
target/openrisc: Fix writes to interrupt mask register

The interrupt controller mask register (PICMR) allows writing any value
to any of the 32 interrupt mask bits.  Writing a 0 masks the interrupt
writing a 1 unmasks (enables) the the interrupt.

For some reason the old code was or'ing the write values to the PICMR
meaning it was not possible to ever mask a interrupt once it was
enabled.

I have tested this by running linux 4.18 and my regular checks, I don't
see any issues.

Reported-by: Davidson Francis <davidsondfgl@gmail.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
6 years agotarget/openrisc: Fix delay slot exception flag to match spec
Stafford Horne [Sun, 1 Jul 2018 05:02:11 +0000 (14:02 +0900)]
target/openrisc: Fix delay slot exception flag to match spec

The delay slot exception flag is only set on the SR register during
exception.  Previously it was being set on both the ESR and SR this
caused QEMU to differ from the spec.  The was apparent as the linux
kernel had a bug where it could boot on QEMU but not on real hardware.

The fixed logic now matches hardware.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
6 years agolinux-user: Fix struct sigaltstack for openrisc
Richard Henderson [Mon, 28 May 2018 02:51:34 +0000 (21:51 -0500)]
linux-user: Fix struct sigaltstack for openrisc

Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
6 years agolinux-user: Implement signals for openrisc
Richard Henderson [Sun, 27 May 2018 19:02:17 +0000 (14:02 -0500)]
linux-user: Implement signals for openrisc

All of the existing code was boilerplate from elsewhere,
and would crash the guest upon the first signal.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
---
v2:
  Add a comment to the new definition of target_pt_regs.
  Install the signal mask into the ucontext.
v3:
  Incorporate feedback from Laurent.

6 years agotarget/openrisc: Add support in scripts/qemu-binfmt-conf.sh
Richard Henderson [Sun, 27 May 2018 13:41:53 +0000 (08:41 -0500)]
target/openrisc: Add support in scripts/qemu-binfmt-conf.sh

Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
6 years agotarget/openrisc: Reorg tlb lookup
Richard Henderson [Wed, 23 May 2018 06:14:02 +0000 (23:14 -0700)]
target/openrisc: Reorg tlb lookup

While openrisc has a split i/d tlb, qemu does not.  Perform a
lookup on both i & d tlbs in parallel and put the composite
rights into qemu's tlb.  This avoids ping-ponging the qemu tlb
between EXEC and READ.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
6 years agocrypto: Implement TLS Pre-Shared Keys (PSK).
Richard W.M. Jones [Tue, 3 Jul 2018 08:03:03 +0000 (09:03 +0100)]
crypto: Implement TLS Pre-Shared Keys (PSK).

Pre-Shared Keys (PSK) is a simpler mechanism for enabling TLS
connections than using certificates.  It requires only a simple secret
key:

  $ mkdir -m 0700 /tmp/keys
  $ psktool -u rjones -p /tmp/keys/keys.psk
  $ cat /tmp/keys/keys.psk
  rjones:d543770c15ad93d76443fb56f501a31969235f47e999720ae8d2336f6a13fcbc

The key can be secretly shared between clients and servers.  Clients
must specify the directory containing the "keys.psk" file and a
username (defaults to "qemu").  Servers must specify only the
directory.

Example NBD client:

  $ qemu-img info \
    --object tls-creds-psk,id=tls0,dir=/tmp/keys,username=rjones,endpoint=client \
    --image-opts \
    file.driver=nbd,file.host=localhost,file.port=10809,file.tls-creds=tls0,file.export=/

Example NBD server using qemu-nbd:

  $ qemu-nbd -t -x / \
    --object tls-creds-psk,id=tls0,endpoint=server,dir=/tmp/keys \
    --tls-creds tls0 \
    image.qcow2

Example NBD server using nbdkit:

  $ nbdkit -n -e / -fv \
    --tls=on --tls-psk=/tmp/keys/keys.psk \
    file file=disk.img

Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
6 years agoMerge remote-tracking branch 'remotes/cody/tags/block-pull-request' into staging
Peter Maydell [Tue, 3 Jul 2018 10:49:51 +0000 (11:49 +0100)]
Merge remote-tracking branch 'remotes/cody/tags/block-pull-request' into staging

# gpg: Signature made Tue 03 Jul 2018 04:42:11 BST
# 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:
  backup: Use copy offloading
  block: Honour BDRV_REQ_NO_SERIALISING in copy range
  block: Fix parameter checking in bdrv_co_copy_range_internal

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoMerge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2018-07-02' into staging
Peter Maydell [Tue, 3 Jul 2018 09:47:02 +0000 (10:47 +0100)]
Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2018-07-02' into staging

nbd patches for 2018-07-02

Bug fixes and iotest exposure of fleecing via NBD (serving a
read-only point-in-time view via blockdev-backup sync:none,
as well as serving dirty bitmaps over NBD), including a new
x-dirty-bitmap parameter when opening NBD clients as the
counterpart to x-nbd-server-add-bitmap. Also a random fix
for iscsi block_status spotted by Coverity that missed other
miscellaneous trees.

- Eric Blake: nbd/server: Fix dirty bitmap logic regression
- Eric Blake: iscsi: Avoid potential for get_status overflow
- John Snow/Vladimir Sementsov-Ogievskiy: 0/2 block: formalize and test fleecing
- Eric Blake: 0/2 test NBD bitmap export

# gpg: Signature made Tue 03 Jul 2018 02:33:03 BST
# gpg:                using RSA key A7A16B4A2527436A
# gpg: Good signature from "Eric Blake <eblake@redhat.com>"
# gpg:                 aka "Eric Blake (Free Software Programmer) <ebb9@byu.net>"
# gpg:                 aka "[jpeg image of size 6874]"
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2  F3AA A7A1 6B4A 2527 436A

* remotes/ericb/tags/pull-nbd-2018-07-02:
  iotests: New test 223 for exporting dirty bitmap over NBD
  nbd/client: Add x-dirty-bitmap to query bitmap from server
  iotests: add 222 to test basic fleecing
  blockdev: enable non-root nodes for backup source
  iscsi: Avoid potential for get_status overflow
  nbd/server: Fix dirty bitmap logic regression

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agomonitor: Spell "I/O thread" consistently in comments
Markus Armbruster [Tue, 3 Jul 2018 08:53:28 +0000 (10:53 +0200)]
monitor: Spell "I/O thread" consistently in comments

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-3-armbru@redhat.com>

6 years agoqmp: Say "out-of-band" instead of "Out-Of-Band"
Markus Armbruster [Tue, 3 Jul 2018 08:53:27 +0000 (10:53 +0200)]
qmp: Say "out-of-band" instead of "Out-Of-Band"

Affects documentation and a few error messages.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180703085358.13941-2-armbru@redhat.com>

6 years agoaudio: add audio timer trace points
Gerd Hoffmann [Mon, 2 Jul 2018 16:15:24 +0000 (18:15 +0200)]
audio: add audio timer trace points

Track audio timer starts and stops.  Also trace delayed audio timer calls.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180702161524.17268-1-kraxel@redhat.com

6 years agoaudio/hda: fix CID 1393631
Gerd Hoffmann [Mon, 2 Jul 2018 14:55:12 +0000 (16:55 +0200)]
audio/hda: fix CID 1393631

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180702145513.11481-2-kraxel@redhat.com

6 years agoaudio/hda: adjust larger gaps faster
Gerd Hoffmann [Mon, 2 Jul 2018 14:55:11 +0000 (16:55 +0200)]
audio/hda: adjust larger gaps faster

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20180702145513.11481-1-kraxel@redhat.com

6 years agovga: disable global_vmstate for 3.0+ machine types
Gerd Hoffmann [Mon, 2 Jul 2018 16:33:44 +0000 (18:33 +0200)]
vga: disable global_vmstate for 3.0+ machine types

Move global_vmstate from vga_common_init() parameter to VGACommonState
field.  Set global_vmstate to true for isa vga devices, so nothing
changes here.  virtio-vga and secondary-vga already set global_vmstate
to false so no change here either.  All other pci vga devices get a new
global-vmstate property, defaulting to false.  A compat property flips
it to true for older machine types.

With this in place you don't get a vmstate section naming conflict any
more when adding multiple pci vga devices to your vm.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-Id: <20180702163345.17892-1-kraxel@redhat.com>

6 years agovirtio-gpu: disable scanout when backing resource is destroyed
Gerd Hoffmann [Mon, 2 Jul 2018 16:24:43 +0000 (18:24 +0200)]
virtio-gpu: disable scanout when backing resource is destroyed

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20180702162443.16796-4-kraxel@redhat.com

6 years agovirtio-gpu: update old resource too.
Gerd Hoffmann [Mon, 2 Jul 2018 16:24:42 +0000 (18:24 +0200)]
virtio-gpu: update old resource too.

When switching scanout from one resource to another we must update the
scanout_bitmask field for both new (set bit) and old (clear bit)
resource.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20180702162443.16796-3-kraxel@redhat.com

6 years agovirtio-gpu: tweak scanout disable.
Gerd Hoffmann [Mon, 2 Jul 2018 16:24:41 +0000 (18:24 +0200)]
virtio-gpu: tweak scanout disable.

- Factor out the code to virtio_gpu_disable_scanout().
- Allow disable scanout 0, show a message then.
- Clear scanout->resource_id.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20180702162443.16796-2-kraxel@redhat.com

6 years agoMerge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
Peter Maydell [Tue, 3 Jul 2018 08:49:20 +0000 (09:49 +0100)]
Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

pc, virtio: fixes

A couple of fixes to amd iommu, and a fix to virtio iommu.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Thu 28 Jun 2018 02:46:45 BST
# gpg:                using RSA key 281F0DB8D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>"
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* remotes/mst/tags/for_upstream:
  virtio-rng: process pending requests on DRIVER_OK
  hw/i386: Fix AMDVI GATS and HATS encodings
  hw/i386: Fix IVHD entry length for AMD IOMMU

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6 years agoui: do not build-depend or link with libdrm, it is not needed
Michael Tokarev [Sat, 30 Jun 2018 16:54:48 +0000 (19:54 +0300)]
ui: do not build-depend or link with libdrm, it is not needed

Opengl support brings up libdrm. But actually nothing uses this library
or includes any of its headers. Just remove checking for it from configure.

Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180630165448.30795-1-mjt@msgid.tls.msk.ru
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
6 years agoxhci: fix guest-triggerable assert
Gerd Hoffmann [Mon, 2 Jul 2018 16:27:52 +0000 (18:27 +0200)]
xhci: fix guest-triggerable assert

Set xhci into error state instead of throwing a core dump.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20180702162752.29233-1-kraxel@redhat.com

6 years agoehci: Don't fetch a NULL current qtd but advance the queue instead.
Sebastian Bauer [Mon, 25 Jun 2018 22:27:18 +0000 (00:27 +0200)]
ehci: Don't fetch a NULL current qtd but advance the queue instead.

Fetching qtd with the NULL address most likely makes no sense so from now
on, we handle it this case similarly as if the terminate (T) bit is not
set, which is already an exception as according to section 3.6 of the EHCI
spec there is no T bit defined for the current_qtd field.

The spec is a bit vague on how an EHCI driver should initialize these
fields: "The general operational model is that the host controller can
detect whether the overlay area contains a description of an active
transfer" (p. 49). QEMU primarily uses the QTD_TOKEN_ACTIVE bit of the
queue header to infer the activity state but there are other ways
conceivable.

This change allows QEMU to boot further into AmigaOS. The public available
version of the EHCI driver recycles queue heads in some rare conditions but
only clears the current_qtd field but not the status field. This works with
many available EHCI PCI cards but e.g., not with the Freescale USB
controller's found on the P5040. On the emulated EHCI controller of QEMU
the consequence is that some garbage was read in, which resulted in a
reset of the controller. This change fixes the problem.

Signed-off-by: Sebastian Bauer <mail@sebastianbauer.info>
Tested-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-id: 20180625222718.4488-1-mail@sebastianbauer.info
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
6 years agobackup: Use copy offloading
Fam Zheng [Tue, 3 Jul 2018 02:37:58 +0000 (10:37 +0800)]
backup: Use copy offloading

The implementation is similar to the 'qemu-img convert'. In the
beginning of the job, offloaded copy is attempted. If it fails, further
I/O will go through the existing bounce buffer code path.

Then, as Kevin pointed out, both this and qemu-img convert can benefit
from a local check if one request fails because of, for example, the
offset is beyond EOF, but another may well be accepted by the protocol
layer. This will be implemented separately.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-id: 20180703023758.14422-4-famz@redhat.com
Signed-off-by: Jeff Cody <jcody@redhat.com>