On 7/19/25 16:03, Demi Marie Obenour wrote:
> On 7/19/25 04:06, Alyssa Ross wrote:
>> Demi Marie Obenour <demiobenour@gmail.com> writes:
>>
>>> diff --git a/img/app/Makefile b/img/app/Makefile
>>> index d81337f99e2e83cf03ac73b9fc96fae6ce118537..691db75c287294104bb2fa332e002792bfa61f6e 100644
>>> --- a/img/app/Makefile
>>> +++ b/img/app/Makefile
>>> @@ -135,7 +152,7 @@ start-virtiofsd: scripts/start-virtiofsd.elb
>>>  .PHONY: start-virtiofsd
>>>  
>>>  run-qemu: $(imgdir)/appvm/blk/root.img start-vhost-user-net start-virtiofsd
>>> -	@../../scripts/run-qemu.sh -m 256 -cpu host -kernel $(KERNEL) -vga none \
>>> +	@../../scripts/run-qemu.sh -m 256 -kernel $(KERNEL) -vga none \
>>
>> Why this change?
> 
> Otherwise the script fails with TCG, but it seems to fail anyway due to
> rutabaga issues in what is probably an untested configuration.  I'll
> get another machine to do testing with KVM enabled.
> 
>>>  	    -drive file=$(imgdir)/appvm/blk/root.img,if=virtio,format=raw,readonly=on \
>>>  	    -append "root=PARTLABEL=root nokaslr" \
>>>  	    -gdb unix:build/gdb.sock,server,nowait \
>>> @@ -151,7 +168,8 @@ run-qemu: $(imgdir)/appvm/blk/root.img start-vhost-user-net start-virtiofsd
>>>  	    -parallel none \
>>>  	    -chardev vc,id=virtiocon0 \
>>>  	    -device virtio-serial-pci \
>>> -	    -device virtconsole,chardev=virtiocon0
>>> +	    -device virtconsole,chardev=virtiocon0 \
>>> +	    -audio driver=pipewire,model=virtio
>>
>> Let's group this with the fs and gpu arguments?  More similar to them
>> than the console.  Also, would it be better to use the pa driver, so it
>> can be tested on systems that don't have PipeWire yet?  There shouldn't
>> be any downside, right?
> 
> Will fix in v5.

I wrote too soon: the pa driver fails if PipeWire is in use, as it expects
"$XDG_RUNTIME_DIR/native/pid" to exist and pipewire-pulse doesn't create it.
Qubes OS doesn't alter pipewire-pulse so it isn't the cause.
-- 
Sincerely,
Demi Marie Obenour (she/her/hers)