Re: [Xen-devel] [RFC PATCH v2 06/17] libxl: create vkb device only for guests with graphics output

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxx>]

On Thu, 1 Nov 2018, Ian Jackson wrote:
> (CCing Stefano's new address.)

Thank you, I am still responsive at my kernel.org address too.


> Marek Marczykowski-Górecki writes ("[RFC PATCH v2 06/17] libxl: create vkb 
> device only for guests with graphics output"):
> > The forced vkb device is meant for better performance of qemu access
> > (at least according to ebbd2561b4cefb299f0f68a88b2788504223de18 "libxl:
> > Add a vkbd frontend/backend pair for HVM guests"), which isn't used if
> > there is no configured channel to actually access that keyboard.
> 
> I think the background here is that the "usb keyboard/mouse" referred
> to in ebbd2561 is supposedly used only for vnc ?

Yes. The idea is that if the guest doesn't use the usb keyboard/mouse,
QEMU will need far less frequent wakeups in Dom0.


> I think we still support SDL though, so maybe that needs to be checked
> too.  I did `git-grep -i sdl tools/libxl' and there's a few
> occurrences of something like this
>   tools/libxl/libxl_dm.c:        if (!sdl && !vnc)
> in particular, in one of these cases it passes -nographic to qemu.
> 
> Probably the situations where we shouldn't unconditionally provide a
> vkb are those where -nographic is passed to qemu ?

Yes, if the configured machine already is meant to be used without
emulated usb keyboard mouse, then vkb is not necessary.


> Maybe we need a helper inline function for sdl || vnc.  If you decide
> to introduce one, please do that in a patch before this one.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel