Episteme PROMENEUR <
episteme....@gmail.com> 於 2021年4月3日 週六 上午2:33寫道:
>
> Some notes about
qemu.android.
>
> 1.
>
> "-usb -device usb-tablet,bus=usb-bus.0 "
>
> is deprecated and will be removed
You probably misread it. It's the new syntax.
> the recent syntax is
>
> -device usb-ehci \
> -device usb-mouse \
> -device usb-kbd \
> -device usb-tablet \
>
> You must use "-device usb-tablet"" to get a seamless capture of the mouse by the host and the guest.
>
> 2.
>
> "-soundhw ac97"
>
> is deprecated and will be removed
Hmm? I didn't see any word that this is deprecated, at least
in man page of my qemu 4.2.1.
Checking the latest qemu doc, it's said
-soundhw is deprecated since 5.1. OK...
> the recent syntax is
>
> -device ich9-intel-hda \
> -device hda-duplex,audiodev=snd0 \ <<<<< this is my choice there are several choices
> -audiodev pa,id=snd0 \ <<<<< the virtual sound card "snd0" uses the pulseaudio server of the host
>
> "-device hda-duplex,audiodev=snd0" this defines a sound card "snd0" with intel hda codecs and which uses the host line in where a jack micro is plugged and the host line out where host speakers are plugged. Micro and speakers are dynamically shared between host and guest.
Are these options generic enough?
I meant, they shouldn't depend on qemu version
and sound cards configurations of the host.
If yes, I can consider to use these in the future.
Note the script could be run in very different distributions or
environments so it should be generic as possible.
> I did not find any generic syntax to define a sound card using intel hda codecs with recent syntax.
>
> 3.
>
> "-net user,"
>
> with this the vm gets an ip address like 10.0.0.5 because using the dhcp of the vm net.
> you can't establish any connection between host and guest.
>
> the better way is to use a host bridge. this way the vm gets an ip address of your lan and is a member of your lan.
>
> i advise the following generic syntax
>
> "-nic bridge" <<<<< this means the vm uses the first host bridge it finds.
Using bridge mode is much more complex than using -net user.
Configuring bridge mode is far beyond the mission of our rpm.
The script should just work -- people don't need to do step 1,2,3,... to use it.
So there is no plan to use the bridge mode by default.
People who need bridge mode could configure that themselves.
Actually there is no problem to make connection from the guest to host
in the net user mode. You just can't make connection from host to the guest
directly. If you need that, probably using hostfwd= is simpler.
> Le vendredi 2 avril 2021 à 19:33:11 UTC+2, Episteme PROMENEUR a écrit :
>> I have just tried 8.1-r5. Esc does not work.
>> Le vendredi 2 avril 2021 à 19:11:21 UTC+2, Chih-Wei Huang a écrit :
>>> Episteme PROMENEUR <
episteme....@gmail.com> 於 2021年4月3日 週六 上午12:46寫道:
>>> >
>>> > android-x86-8.1-r5, android-x86-9.0-r2
>>> >
>>> > type "esc" instead of clicking the "back arrow" of a window then no action.
>>>
>>> Oh, I thought you talked about 7.1-r5...
>>>
>>> ESC still works as BACK in 8.1, but not work in 9.0.
>>> I think that's changed by AOSP.
>>>
>>> > Le vendredi 2 avril 2021 à 18:14:53 UTC+2, Chih-Wei Huang a écrit :
>>> >> Episteme PROMENEUR <
episteme....@gmail.com> 於 2021年4月1日 週四 上午12:00寫道:
>>> >> >
>>> >> > 0. General explanation
>>> >> >
>>> >> > After many trials with Virtualbox. I abandonnned it.
>>> >> >
>>> >> > Because there is no guest additions for Android OS.
>>> >> >
>>> >> > You get problems with mouse, keyboard, GUI. Even with the best mature android-x86-14.1-r5.
>>> >> >
>>> >> > Thus, I went to explore qemu/kvm another virtualization technology.
>>> >> >
>>> >> > I choose android-x86-7.1-r5 after trials with android-x86-14.1-r5 and android-x86-9.0-r2.
>>> >> >
>>> >> > They are not mature to use them as a PC OS.
>>> >> > There are :
>>> >> > - key problem ("Esc" for example),
>>> >>
>>> >> What's the problem exactly?