[xpost,repost] VGA false positive in dmesg?

[Ottavio Caruso]

On Tue, 4 Aug 2020 at 18:59, Konstantin Belousov <kostikbel at gmail.com> wrote:
>
> On Tue, Aug 04, 2020 at 05:43:33PM +0100, Ottavio Caruso via freebsd-hackers wrote:
> > On 04/08/2020 16:46, Eugene Grosbein wrote:
> > > 04.08.2020 20:50, Ottavio Caruso via freebsd-hackers wrote:
> > >
> > > > On 04/08/2020 13:08, Eugene Grosbein wrote:
> > > > > 04.08.2020 18:37, Ottavio Caruso via freebsd-hackers wrote:
> > > > >
> > > > > > I'm a bit confused here. Do both ttyu0 and ttyv0 refer to the serial console? Do I need ttyv0? Is it safe to comment ttyv0 in /etc/ttys?
> > > > >
> > > > > Note that /etc/ttys serves multi-user mode only. For earlier stages, you may find useful
> > > > > to put these settings to /boot/loader.conf
> > > > >
> > > > > console="comconsole"
> > > > > comconsole_speed=115200
> > > > > comconsole_port="0x3f8"
> > > > > boot_multicons="NO"
> > > > >
> > > > > Check out conscontrol output after applying these.
> > > >
> > > >
> > > > That doesn't seem to work either:
> > >
> > > In fact, it works,
> >
> > Does it?
> In fact, why do you fight with the ghost VGA device ?  Does it break
> anything to you ?
>
> >
> >
> > > If ISA PNP VGA device is probed successfully, then qemu emulates it.
> >
> > But I don't understand how why it is probed successfully. the "-vga none"
> > option should remove any vga capability in the host.
> There is not much to probe for legacy ISA device.  As far as your hints
> file does not disable the attachment, system assumes that the device is
> present.
>
> Then there are framebuffer console drivers, vt and sc.  I do not want to
> confirm that by reading code, but I highly suspect that if vt(4) cannot
> attach, sc(4) would.  I remember that only vt(4) pays attention to the
> ACPI FADT flag NO_VGA, sc(4) does not.
>
> But the FADT reporting of missed ISA devices is usually buggy.
>
> > I have tried this
> > successfully on an OpenBSD guest and part-successfully on a NetBSD guest (I
> > had to comment all references to /usr/libexec/getty Pc). Why this is not the
> You tried what ?  How did you checked that OpenBSD and NetBSD did not claimed
> presence of VGA ?
>
> And again, what it changes ?

I get all of your points but they are irrelevant to my question: does
this vga device actually exist or is the kernel just spitting false
positives? What I want to achieve is secondary. If the qemu
documentation says that "-vga none" "disables vga", then we have a
problem somewhere.

Last minute edit: somebody pointed me out to this blob:
https://github.com/freebsd/freebsd/blob/1d6e4247415d264485ee94b59fdbc12e0c566fd0/sys/isa/vga_isa.c#L169

But I'm not really sure what to make of it.

-- 
Ottavio Caruso