Hi On Tue, Mar 24, 2026 at 7:24 PM Daniel P. Berrangé <[email protected]> wrote: > > On Tue, Mar 17, 2026 at 12:51:14PM +0400, Marc-André Lureau wrote: > > Add a standalone VNC server binary that connects to a running QEMU > > instance via the D-Bus display interface (org.qemu.Display1, via the bus > > or directly p2p). This allows serving a VNC display without compiling > > VNC support directly into the QEMU system emulator, and enables running > > the VNC server as a separate process with independent lifecycle and > > privilege domain. > > IIUC, this new binary is linking to all the existnig ui/vnc.c > code, so there's no actual new impl of the RFB protocol, just > a new glue layer for launching the existing impl with comms > over DBus instead of from direct callback functions ? > > > Built only when both VNC and D-Bus display support are enabled. > > If we wanted to have qemu -vnc disabled, and qemu-vnc built, we would > > need to split CONFIG_VNC. This is left as a future exercise. > > > > I left out for now: > > - sasl & tls authz > > IMHO these are must have features, as the default VNC protocol > security is a very bad joke.
Ok that should be doable. One thing that bothers me a bit is mapping the -object options (authz) to a more regular CLI. Perhaps I'll just add -object args and wire it with the other options.. > > > - some runtime functionalities (better done by restarting) > > - a few legacy options > > - Windows support > > > > Signed-off-by: Marc-André Lureau <[email protected]> > > --- > > MAINTAINERS | 5 + > > docs/conf.py | 3 + > > docs/interop/dbus-display.rst | 2 + > > docs/interop/dbus-vnc.rst | 26 ++ > > docs/interop/index.rst | 1 + > > docs/meson.build | 1 + > > docs/tools/index.rst | 1 + > > docs/tools/qemu-vnc.rst | 199 +++++++++++ > > meson.build | 17 + > > contrib/qemu-vnc/qemu-vnc.h | 46 +++ > > contrib/qemu-vnc/trace.h | 4 + > > contrib/qemu-vnc/audio.c | 307 +++++++++++++++++ > > contrib/qemu-vnc/chardev.c | 127 +++++++ > > contrib/qemu-vnc/clipboard.c | 378 +++++++++++++++++++++ > > contrib/qemu-vnc/console.c | 168 ++++++++++ > > contrib/qemu-vnc/dbus.c | 439 ++++++++++++++++++++++++ > > contrib/qemu-vnc/display.c | 456 +++++++++++++++++++++++++ > > contrib/qemu-vnc/input.c | 239 ++++++++++++++ > > contrib/qemu-vnc/qemu-vnc.c | 450 +++++++++++++++++++++++++ > > contrib/qemu-vnc/stubs.c | 66 ++++ > > contrib/qemu-vnc/utils.c | 59 ++++ > > IMHO we should not really be adding more things to the contrib > directory, and rather wish most of the existing stuff was > not there. When I see "contrib" directories in projects, the > message is generally "unmaintained, abandoned junk". > > If this is a serious maintained feature we expect people to use, > it should be formally under the ui/ directory, possibly a > 'ui/dbus-vnc' subdir ? It is intended to be maintained. (although I hope we can deprecate it in favour of qemu-rdp in the long term). I'll follow the "Cleaning up contrib/ and tools/" thread and see where it could land. . > > > tests/qtest/dbus-vnc-test.c | 733 > > +++++++++++++++++++++++++++++++++++++++++ > > contrib/qemu-vnc/meson.build | 26 ++ > > contrib/qemu-vnc/qemu-vnc1.xml | 174 ++++++++++ > > contrib/qemu-vnc/trace-events | 20 ++ > > meson_options.txt | 2 + > > scripts/meson-buildoptions.sh | 3 + > > tests/dbus-daemon.sh | 14 +- > > tests/qtest/meson.build | 8 + > > 29 files changed, 3971 insertions(+), 3 deletions(-) > > > > > diff --git a/meson_options.txt b/meson_options.txt > > index 31d5916cfce..ef938e74793 100644 > > --- a/meson_options.txt > > +++ b/meson_options.txt > > @@ -119,6 +119,8 @@ option('vfio_user_server', type: 'feature', value: > > 'disabled', > > description: 'vfio-user server support') > > option('dbus_display', type: 'feature', value: 'auto', > > description: '-display dbus support') > > +option('qemu_vnc', type: 'feature', value: 'auto', > > + description: 'standalone VNC server over D-Bus') > > Every option is about QEMU, so 'qemu_' on a name feels redundant, > and QEMU alrady has a VNC server too. > > I suggest 'dbus_vnc' is possibly a better name > Well, I just named the option after the binary's name. I think that's easier. > > > option('tpm', type : 'feature', value : 'auto', > > description: 'TPM support') > > option('valgrind', type : 'feature', value: 'auto', > > With regards, > Daniel > -- > |: https://berrange.com ~~ https://hachyderm.io/@berrange :| > |: https://libvirt.org ~~ https://entangle-photo.org :| > |: https://pixelfed.art/berrange ~~ https://fstop138.berrange.com :| >
