Currently, AUD_register_card picks the audio backend from either: - the first audiodev that was created
- the audio_prio_list[] array, which can be customized at configure time This series instead extends -audio to define a default audio backend if no "model" is used. This preserves simple command line use where a single "-audio" option applies to all audio devices and captures, and also uses a single QAPI-based configuration syntax for both -audio and -audiodev. The current hack to use the first -audiodev as a default audio device is removed. For migration purposes, the first audiodev is suggested in case of an error: ./qemu-system-x86_64 -device sb16 -audiodev pa,id=default qemu-system-x86_64: -device sb16: no default audio driver available Perhaps you wanted to set audiodev=default? VNC is changed to reintroduce use of the default audio backend; still, compared to before the cleanup effort this will not be enabled if -nodefaults is use, which is an improvement as it removes magic. Paolo Paolo Bonzini (7): audio: error hints need a trailing \n audio: disable default backends if -audio/-audiodev is used audio: extract audio_define_default audio: extend -audio to allow creating a default backend audio: do not use first -audiodev as default audio device audio: reintroduce default audio backend for VNC audio, qtest: get rid of QEMU_AUDIO_DRV audio/audio.c | 85 +++++++++++++-------------------- audio/audio.h | 3 ++ docs/about/deprecated.rst | 6 --- docs/about/removed-features.rst | 14 ++++-- qemu-options.hx | 29 +++++++---- system/vl.c | 34 ++++++++----- tests/qtest/libqtest.c | 4 +- ui/vnc.c | 2 + 8 files changed, 93 insertions(+), 84 deletions(-) -- 2.41.0
