Re: [Qemu-devel] [RFC 13/14] vga: Add endian control register

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Mon, 2014-06-30 at 13:38 +0200, Gerd Hoffmann wrote:
>   Hi,
> 
> > diff --git a/hw/display/vga_int.h b/hw/display/vga_int.h
> > index ae64321..894c6ab 100644
> > --- a/hw/display/vga_int.h
> > +++ b/hw/display/vga_int.h
> > @@ -47,6 +47,8 @@
> >  #define VBE_DISPI_INDEX_Y_OFFSET        0x9
> >  #define VBE_DISPI_INDEX_NB              0xa /* size of vbe_regs[] */
> >  #define VBE_DISPI_INDEX_VIDEO_MEMORY_64K 0xa /* read-only, not in vbe_regs */
> > +#define VBE_DISPI_INDEX_EXTENDED_CAPS   0xb /* read-only, not in vbe_regs */
> > +#define VBE_DISPI_INDEX_ENDIAN_CTRL     0xc /* not in vbe_regs */
> >  
> >  #define VBE_DISPI_ID0                   0xB0C0
> >  #define VBE_DISPI_ID1                   0xB0C1
> > @@ -55,13 +57,22 @@
> >  #define VBE_DISPI_ID4                   0xB0C4
> >  #define VBE_DISPI_ID5                   0xB0C5
> 
> I was more thinking to add ID6 to indicate the new interface revision
> with the additional VBE_DISPI_INDEX_ENDIAN_CTRL register.
> 
> I'm a bit worried that there is no response from the bochs guys yet, I
> don't want have two incompatible rev6 interfaces.  At least nobody seems
> to have defined one so far, google finds nothing for "bochs dispi
> 0xB0C6".

Ah ok, I haven't quite figured out how that DISPI_ID stuff works, I'll
dig a bit. Definitely an option though I like the idea of an
"capability" register in the new revision so we can easily add features
that don't have to be implemented by the host (for example there's
little point for Bochs to implement endian control).

The lack of response from Bochs worries me too. I've tried hanging out
on their IRC as well with no result so far.

Cheers,
Ben.