Hi,

Made a nice patch series out of it, first cleaning up graphics_* variables,
then adding support for -g to virtio (just works), qxl (needs guest driver
fixes) and stdvga (needs virtual hardware update, and therefore also an
guest driver update for obvious reasons).

linux kernel patches to test qxl and stdvga are here:
  https://www.kraxel.org/cgit/linux/log/?h=qxl-display-size

please review & test,
  Gerd

Gerd Hoffmann (9):
  sysemu: add defines for sparc & ppc graphics_*
  sparc: use sparc_graphic_* defines
  ppc: use ppc_graphic_* defines
  drop arch-specific graphic_* initialization.
  enable -g command line switch for all archs
  virtio-gpu: use graphic_* variables.
  qxl: use graphic_* variables.
  update -g command line switch documentation
  [RfC] stdvga: use graphic_* variables.

 arch_init.c             | 11 -----------
 hw/display/qxl.c        | 32 ++++++++++++++++++++----------
 hw/display/vga-pci.c    | 52 ++++++++++++++++++++++++++++++++++---------------
 hw/display/vga_int.h    |  4 ++--
 hw/display/virtio-gpu.c |  9 +++++++--
 hw/display/virtio-vga.c |  3 ++-
 hw/ppc/mac_newworld.c   |  6 +++---
 hw/ppc/mac_oldworld.c   |  6 +++---
 hw/ppc/prep.c           | 10 +++++-----
 hw/ppc/spapr.c          |  6 +++---
 hw/sparc/sun4m.c        | 36 +++++++++++++++++-----------------
 include/sysemu/sysemu.h |  7 +++++++
 qemu-options.hx         | 15 ++++++++++++--
 vl.c                    |  3 +++
 14 files changed, 124 insertions(+), 76 deletions(-)

-- 
1.8.3.1

On 21/02/17 22:14, Gerd Hoffmann wrote:

>   Hi,
> 
> Made a nice patch series out of it, first cleaning up graphics_* variables,
> then adding support for -g to virtio (just works), qxl (needs guest driver
> fixes) and stdvga (needs virtual hardware update, and therefore also an
> guest driver update for obvious reasons).
> 
> linux kernel patches to test qxl and stdvga are here:
>   https://www.kraxel.org/cgit/linux/log/?h=qxl-display-size
> 
> please review & test,
>   Gerd

Interestingly enough, there have been recent discussions from the guys
over at emaculation as to whether it would be possible to pass in valid
resolutions from the host graphics driver for full-screen MacOS 9
emulation in QEMU.

With this patchset I believe Ben's OS 9 VGA driver could read the
suggested resolution from the command line via the QEMU extended VGA
registers, so I'm wondering if it would make sense to have a similar
mechanism for passing in a range of resolutions from the local video
driver that could be offered by the guest driver? Or is the normal guest
driver behaviour to just offer the standard set of *GA resolutions plus
whatever custom resolution is supplied via -g?


ATB,

Mark.

  Hi,

> With this patchset I believe Ben's OS 9 VGA driver could read the
> suggested resolution from the command line via the QEMU extended VGA
> registers,

Yes.

> so I'm wondering if it would make sense to have a similar
> mechanism for passing in a range of resolutions from the local video
> driver that could be offered by the guest driver? Or is the normal guest
> driver behaviour to just offer the standard set of *GA resolutions plus
> whatever custom resolution is supplied via -g?

The linux kms drivers simply use the kernel's mode database, to provide
a list of standard resolutions.  Additionally you also can define your
own resolutions, using xrandr --newmode & --addmode).

If we want go beyond passing a simple "please use <width>y<height>" we
probably should generate a edid block and pass that to the guest, but I
don't see a strong reason to do so.

cheers,
  Gerd