Commit 99997823bb ("ui/dbus: add p2p=on/off option")
introduced an asynchronous D-Bus client setup path, with the completion
handler reaching back into the global dbus_display state.
This makes the callback effectively operate on whatever request is
current when it runs, rather than the one that created it. A completion
from an older request can therefore clear a newer
add_client_cancellable or install its connection after a replacement
request has already been issued. It also relies on the DBusDisplay
instance remaining alive until completion.
Fix this by passing the DBusDisplay and GCancellable as callback data,
taking references while the async setup is in flight, and only acting
on completion if it still matches the current request. Also drop the
previous cancellable before creating a new request.
Fixes: 99997823bb ("ui/dbus: add p2p=on/off option")
Signed-off-by: GuoHan Zhao <zhaoguohan@kylinos.cn>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20260326065111.626236-1-zhaoguohan@kylinos.cn>
include/ui/shader.h and include/ui/surface.h are included by files that
do not depend on Epoxy so they shouldn't include epoxy/gl.h. Otherwise,
compilations of these files can fail because the path to the directory
containing epoxy/gl.h may not be passed to the compiler.
Signed-off-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20260303-gl-v1-3-d90f0a237a52@rsg.ci.i.u-tokyo.ac.jp>
console_gl_check_format() was supposed to check if the pixman format is
supported by surface_gl_create_texture(), but it missed
PIXMAN_BE_x8r8g8b8 and PIXMAN_BE_a8r8g8b8, which are properly mapped to
OpenGL formats by surface_gl_create_texture().
Fix the discrepancy of the two functions by sharing the code to map
pixman formats to OpenGL ones.
Signed-off-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20260303-gl-v1-1-d90f0a237a52@rsg.ci.i.u-tokyo.ac.jp>
The EGL surface and context are destroyed when a new GTK window is
created. We must ensure these are recreated and initialized before
any rendering happens in gd_egl_refresh.
Currently, the check for a pending draw is performed before the
surface initialization block. This can result in an attempt to
draw when the EGL surface (vc->gfx.esurface) is not yet available.
This patch moves the drawing check after the surface initialization
to ensure a valid surface exists before rendering in gd_egl_refresh.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20260303011151.1925827-1-dongwon.kim@intel.com>
The dbus filter holds a strong reference to the DBusDisplayListener
(via GDestroyNotify) to ensure the listener remains alive while the
filter may still be running in another thread. This creates a
reference cycle (ddl -> conn -> filter -> ddl) that prevents the
listener from being freed.
Break the cycle by connecting to the connection's "closed" signal
and removing the filter when the connection closes.
Fixes: commit fa88b85dea ("ui/dbus: filter out pending messages when scanout")
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
cursor_fb is a local egl_fb that gets an FBO allocated via
egl_fb_setup_for_tex but is never destroyed, leaking the
framebuffer object on every cursor update.
Add egl_fb_destroy() after the cursor data has been read.
Fixes: commit 142ca628a7 ("ui: add a D-Bus display backend")
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
In QEMU 10.1, commit 5d56bff11e ("ui/vdagent: add migration support")
added migration support for the vdagent chardev and commit 42000e0013
("ui/vdagent: remove migration blocker") removed the migration
blocker. No compat for older machine versions was added, so migration
with pre-10.1 machine version, from a 10.1 binary to a pre-10.1 binary
will result in a failure when loading the VM state in the target
instance:
> Unknown savevm section or instance 'vdagent' 0. Make sure that your
> current VM setup matches your saved VM setup, including any
> hotplugged devices
Add a compat flag to block migration when the machine version is less
than 10.1 to avoid this.
Cc: qemu-stable@nongnu.org
Fixes: 42000e0013 ("ui/vdagent: remove migration blocker")
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20260310142552.240877-1-f.ebner@proxmox.com>
Don't change bound GL context when creating new GL context for consistency
with behavior expected by virglrenderer that assumes context-creation doesn't
switch context. eglCreateContext() doesn't require GL context to be bound
when it's invoked. Update qemu_egl_create_context() to spawn GL sub-contexts
from a given shared GL context instead of a currently-bound context.
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Message-ID: <20260303151422.977399-4-dmitry.osipenko@collabora.com>
Message-ID: <20260304165043.1437519-6-alex.bennee@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
The vnc_display_print_local_addr() method is intended to print the VNC
listening address on the console at startup, so the user can see the
auto-chosen port address when using the 'to=' flag. This is only called
by vnc_display_open() which is in the QEMU startup callpath. The check
for not being in QMP is thus redundant and can be removed.
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Neither the VNC or SPICE code for password changes provides error
reporting at source, leading the callers to report a largely useless
generic error message.
Fixing this removes one of the two remaining needs for the undesirable
error_printf_unless_qmp() method.
While fixing this the error message hint is improved to recommend the
'password-secret' option which allows securely passing a password at
startup.
Reported-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Some time ago (commit facda5443f) I've added 'flat' mode (which
omits 'backing-image' key in reply) to 'query-named-block-nodes' to
minimize the size of the returned JSON for deeper backing chains.
While 'query-block' behaved slightly better it turns out that in libvirt
we do call 'query-block' to figure out some information about the
block device (e.g. throttling info) but we don't look at the backing
chain itself.
Wire up 'flat' for 'query-block' so that libvirt can ask for an
abbreviated output. The implementation is much simpler as the internals
are shared with 'query-named-block-nodes'.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <f4476e9f7e8fda74c02be3f806acaa9aa2df4d9a.1770210044.git.pkrempa@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The endianness field used an int to represent a boolean concept, with
0 meaning little-endian and 1 meaning big-endian. This required runtime
validation to reject invalid values and made the code less readable.
Replace with a bool big_endian field that is self-documenting and
type-safe. The compiler now enforces valid values, eliminating the
need for the validation check in audio_validate_settings().
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
Reviewed-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Introduce a sub-class for current "audio_driver" based implementations.
Future AudioBackend implementations can do without it.
Next cleanup will actually remove "audio_driver" struct altogether and
make the subclass proper QOM objects.
Public APIs still rely on backend being an AudioMixeng. They will
assert() if not. This will be addressed later to allow other backends.
Note that the initial naming proposed for this object was AudioDriver,
however the semantics for "driver" is already overloaded and leads to
confusion, in particular with the QAPI AudiodevDriver. The defining
characteristic is of using QEMU's software mixing engine, so
AudioMixengBackend.
Reviewed-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
Reviewed-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Start a simple test program that will exercise the QEMU audio APIs.
It is meant to run manually for now, as it accesses the sound system and
produces sound by default, and also runs for a few seconds. We may want
to make it silent or use the "none" (noaudio) backend by default though,
so it can run as part of the automated test suite.
Reviewed-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Set "using_dbus_display" during early_dbus_init(), so that we can try to
create the "dbus" audio backend by default from audio_prio_list.
This makes dbus audio work by default when using an audio device,
without having to setup and wire up the -audiodev manually.
The added FIXME is addressed in the following commits.
Reviewed-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
Reviewed-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
All three events are shared between precopy and postcopy, rather than
precopy specific.
For example, both precopy and postcopy will go through a SETUP process.
Meanwhile, both FAILED and DONE notifiers will be notified for either
precopy or postcopy on completions / failures.
Rename them to make them match what they do, and shorter.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/qemu-devel/20260126213614.3815900-6-peterx@redhat.com
[fixed-up entry in scsi-disk.c that got merged first]
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Migration notifiers will notify at any of three places: (1) SETUP
phase, (2) migration completes, (3) migration fails.
There's actually a special case for spice: one can refer to
b82fc321bf ("Postcopy+spice: Pass spice migration data earlier"). It
doesn't need another 4th event because in commit 9d9babf78d ("migration:
MigrationEvent for notifiers") we merged it together with the DONE event.
The merge makes some sense if we treat "switchover" of postcopy as "DONE",
however that also means for postcopy we'll notify DONE twice.. The other
one at the end of postcopy when migration_cleanup().
In reality, the current code base will also notify FAILED for postcopy
twice. It's because an (maybe accidental) change in commit
4af667f87c ("migration: notifier error checking").
First of all, we still need that notification when switchover as stated in
Dave's commit, however that's only needed for spice. To fix it, introduce
POSTCOPY_START event to differenciate it from DONE. Use that instead in
postcopy_start(). Then spice will need to capture this event too.
Then we remove the extra FAILED notification in postcopy_start().
If one wonder if other DONE users should also monitor POSTCOPY_START
event.. We have two more DONE users:
- kvm_arm_gicv3_notifier
- cpr_exec_notifier
Both of them do not need a notification for POSTCOPY_START, but only when
migration completed. Actually, both of them are used in CPR, which doesn't
support postcopy.
When at this, update the notifier transition graph in the comment, and move
it from migration_add_notifier() to be closer to where the enum is defined.
I didn't attach Fixes: because I am not aware of any real bug on such
double reporting. I'm wildly guessing the 2nd notify might be silently
ignored in many cases. However this is still worth fixing.
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dr. David Alan Gilbert <dave@treblig.org>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/qemu-devel/20260126213614.3815900-3-peterx@redhat.com
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Most handlers implementations has name like {unit_name}_{handler_name},
which is usual and well-recognized pattern. Convert the rest (especially
with useless qemu_ prefixes and misleading qmp_ prefixes) to the common
pattern.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20260115144606.233252-5-vsementsov@yandex-team.ru>
Create a new header corresponding to functions defined in
util/aiocb.c, and include it whenever AIOCBs are used but
AioContext is not.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
A recent change in glibc 2.42.9000 [1] changes the return type of
strstr() and other string functions to be 'const char *' when the
input is a 'const char *'.
This breaks the build in various files with errors such as :
error: initialization discards 'const' qualifier from pointer target type [-Werror=discarded-qualifiers]
208 | char *pidstr = strstr(filename, "%");
| ^~~~~~
Fix this by changing the type of the variables that store the result
of these functions to 'const char *'.
[1] https://sourceware.org/git/?p=glibc.git;a=commit;h=cd748a63ab1a7ae846175c532a3daab341c62690
Signed-off-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20251209174328.698774-1-clg@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
When there is no display device on qemu machine,
and user only access qemu by remote vnc.
At the same time user input `info vnc` by QMP,
the qemu will abort.
To avoid the abort above, I add display device check,
when query vnc info in qmp_query_vnc_servers().
Reviewed-by: Marc-AndréLureau <marcandre.lureau@redhat.com>
Signed-off-by: Alano Song <AlanoSong@163.com>
[ Marc-André - removed useless Error *err ]
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20251125131955.7024-1-AlanoSong@163.com>
Since commit f626116f ("ui/vdagent: factor out clipboard peer
registration"), the QEMU clipboard serial is reset whenever the vdagent
chardev receives the guest caps. This triggers a CHR_EVENT_CLOSED which
is handled by virtio_serial_close() to notify the guest.
The "reconnection logic" is there to reset the agent when a
client (dbus, spice etc) reconnects, or the agent is restarted.
It is required to sync the clipboard serials and to prevent races or
loops due to clipboard managers on both ends (but this is not
implemented by windows vdagent).
The Unix agent has been reconnecting without resending caps, thus
working with this approach.
However, the Windows agent does not seem to have a way to handle
VIRTIO_CONSOLE_PORT_OPEN=0 event and do not receive further data...
Let's not trigger this disconnection/reset logic if the agent does not
support VD_AGENT_CAP_CLIPBOARD_GRAB_SERIAL.
Fixes: f626116f ("ui/vdagent: factor out clipboard peer registration")
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reported-by: Lucas Kornicki <lucas.kornicki@nutanix.com>
Tested-by: Fiona Ebner <f.ebner@proxmox.com>
Reviewed-by: Fiona Ebner <f.ebner@proxmox.com>
Tested-by: Lucas Kornicki <lucas.kornicki@nutanix.com>
An upcoming patch needs to pass more than just sioc as the opaque
pointer to an AioContext; but since our AioContext code in general
(and its QIO Channel wrapper code) lacks a notify callback present
with GSource, we do not have the trivial option of just g_malloc'ing a
small struct to hold all that data coupled with a notify of g_free.
Instead, the data pointer must outlive the registered handler; in
fact, having the data pointer have the same lifetime as QIONetListener
is adequate.
But the cleanest way to stick such a helper struct in QIONetListener
will be to rearrange internal struct members. And that in turn means
that all existing code that currently directly accesses
listener->nsioc and listener->sioc[] should instead go through
accessor functions, to be immune to the upcoming struct layout
changes. So this patch adds accessor methods qio_net_listener_nsioc()
and qio_net_listener_sioc(), and puts them to use.
While at it, notice that the pattern of grabbing an sioc from the
listener only to turn around can call
qio_channel_socket_get_local_address is common enough to also warrant
the helper of qio_net_listener_get_local_address, and fix a copy-paste
error in the corresponding documentation.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20251113011625.878876-24-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
This prevents direct access of the class members by the VNC
display code.
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
For modularity/clarity reasons, move the capture API in a specific
header.
The current audio/ header license is MIT.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
