[Qemu-devel] extended vga modes?

[Qemu-devel] extended vga modes?

[Michael Tokarev]

After updating qemu-kvm Debian package to 0.12
we've a bugreport about missing video modes which
were present in previous versions.  Big thanks to
the original reporter, Bjørn Mork, who found what
the issue is.  See
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=572991
for the bugreport in question, together with the
resolution.

In short, when vgabios were dropped from qemu-kvm
(for whatever yet unknown reason), all local changes
to it were dropped too, including this patch:
http://git.kernel.org/?p=virt/kvm/qemu-kvm.git;a=commitdiff;h=ebfac597cf

Note that this patch is present in upstream 0.6c version
of vgabios, but not in 0.6b+ which is currently used in
qemu.

Should vgabios in qemu include that patch?

See also: https://bugzilla.redhat.com/show_bug.cgi?id=501545

Thanks!

/mjt

[Qemu-devel] Re: extended vga modes?

[Avi Kivity]

On 03/08/2010 12:20 PM, Michael Tokarev wrote:
> After updating qemu-kvm Debian package to 0.12
> we've a bugreport about missing video modes which
> were present in previous versions.  Big thanks to
> the original reporter, Bjørn Mork, who found what
> the issue is.  See
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=572991
> for the bugreport in question, together with the
> resolution.
>
> In short, when vgabios were dropped from qemu-kvm
> (for whatever yet unknown reason),

What do you mean?  qemu-kvm still carries a local vgabios (see 
kvm/vgabios in qemu-kvm.git).

> all local changes
> to it were dropped too, including this patch:
> http://git.kernel.org/?p=virt/kvm/qemu-kvm.git;a=commitdiff;h=ebfac597cf
>
> Note that this patch is present in upstream 0.6c version
> of vgabios, but not in 0.6b+ which is currently used in
> qemu.
>
> Should vgabios in qemu include that patch?
>
> See also: https://bugzilla.redhat.com/show_bug.cgi?id=501545
>
>    

Looks like Fedora was using the upstream vgabios for a while, not the 
version in qemu-kvm.git.

-- 
error compiling committee.c: too many arguments to function

[Qemu-devel] Re: extended vga modes?

[Michael Tokarev]

Avi Kivity wrote:
[]
>> In short, when vgabios were dropped from qemu-kvm
>> (for whatever yet unknown reason),
> 
> What do you mean?  qemu-kvm still carries a local vgabios (see
> kvm/vgabios in qemu-kvm.git).

Oh my.  So we all overlooked it.  I asked you several times
about the bios sources, in 0.12 seabios were supposed to be
in roms/seabios (which is still empty in the release), and
I thought vgabios should be in roms/vgabios (which is empty
too), and concluded it were dropped from qemu-kvm tarball.
But you're right, and I by mistake take vgabios sources from
upstream qemu when building Debian package, instead of using
the old'good sources from kvm/vgabios.  What a mess!... :(

And it looks like that it's time to remove at least parts of
this mess, don't you think?  How about pushing the vgabios
changes to qemu and moving it to the same place where it is
in qemu?  Does it make sense?

There were another patch mentioned recently when I asked for
bios sources origin a few days ago which probably should be
applied as well...

[]
>> Should vgabios in qemu include that patch?

>> See also: https://bugzilla.redhat.com/show_bug.cgi?id=501545
> 
> Looks like Fedora was using the upstream vgabios for a while, not the
> version in qemu-kvm.git.

Yes they are still using upstream but they also
applied the patches missing in there.

Thank you for the prompt response.  Now I think all
confusion is cleared.

/mjt

[Qemu-devel] Re: extended vga modes?

[Avi Kivity]

On 03/08/2010 01:07 PM, Michael Tokarev wrote:
> Avi Kivity wrote:
> []
>    
>>> In short, when vgabios were dropped from qemu-kvm
>>> (for whatever yet unknown reason),
>>>        
>> What do you mean?  qemu-kvm still carries a local vgabios (see
>> kvm/vgabios in qemu-kvm.git).
>>      
> Oh my.  So we all overlooked it.  I asked you several times
> about the bios sources, in 0.12 seabios were supposed to be
> in roms/seabios (which is still empty in the release), and
> I thought vgabios should be in roms/vgabios (which is empty
> too), and concluded it were dropped from qemu-kvm tarball.
> But you're right, and I by mistake take vgabios sources from
> upstream qemu when building Debian package, instead of using
> the old'good sources from kvm/vgabios.  What a mess!... :(
>
> And it looks like that it's time to remove at least parts of
> this mess, don't you think?  How about pushing the vgabios
> changes to qemu and moving it to the same place where it is
> in qemu?  Does it make sense?
>    

We can't push the changes to qemu since qemu.git doesn't have a vgabios 
fork.  We might push the changes upstream.  Best of all if the seabios 
thing repeats itself with vgabios so we have maintainable and maintained 
vga firmware.

-- 
error compiling committee.c: too many arguments to function

Re: [Qemu-devel] Re: extended vga modes?

[Roy Tam]

2010/3/8 Avi Kivity <avi@redhat.com>:
> On 03/08/2010 01:07 PM, Michael Tokarev wrote:
>>
>> Avi Kivity wrote:
>> []
>>
>>>>
>>>> In short, when vgabios were dropped from qemu-kvm
>>>> (for whatever yet unknown reason),
>>>>
>>>
>>> What do you mean?  qemu-kvm still carries a local vgabios (see
>>> kvm/vgabios in qemu-kvm.git).
>>>
>>
>> Oh my.  So we all overlooked it.  I asked you several times
>> about the bios sources, in 0.12 seabios were supposed to be
>> in roms/seabios (which is still empty in the release), and
>> I thought vgabios should be in roms/vgabios (which is empty
>> too), and concluded it were dropped from qemu-kvm tarball.
>> But you're right, and I by mistake take vgabios sources from
>> upstream qemu when building Debian package, instead of using
>> the old'good sources from kvm/vgabios.  What a mess!... :(
>>
>> And it looks like that it's time to remove at least parts of
>> this mess, don't you think?  How about pushing the vgabios
>> changes to qemu and moving it to the same place where it is
>> in qemu?  Does it make sense?
>>
>
> We can't push the changes to qemu since qemu.git doesn't have a vgabios
> fork.  We might push the changes upstream.  Best of all if the seabios thing
> repeats itself with vgabios so we have maintainable and maintained vga
> firmware.
>

actually they do. see vgasrc directory in seabios.git.
but it is very incomplete: no cirrus support, no vbe.

> --
> error compiling committee.c: too many arguments to function
>
>
>
>