Re: [Qemu-devel] [RFC 23/23] 40p: Add an 8514/A graphics card

[Roy Tam <roytam@gmail.com>]

2011/6/16 Andreas Färber <andreas.faerber@web.de>:
> Am 15.06.2011 um 06:33 schrieb Roy Tam:
>
>> 2011/6/14 Andreas Färber <andreas.faerber@web.de>:
>>>
>>> The IBM E15 is equivalent to an S3 Vision864.
>>>
>>> Lacking S3 SDAC (86C716) support, the DAC indizes are translated
>>> to greyscale colors. This works sufficiently to observe firmware
>>> boot progress.
>>>
>>
>> IMO we can generalize it as a generic S3 Vision864 and use it in x86 guest
>> too.
>
> What in particular would we need to generalize? It's qdev'ified; I don't
> remember seeing anything PReP-specific in there.
> It still needs VMState of course.
>
>> http://wiki.qemu.org/Google_Summer_of_Code_2011#Add_a_S3_Trio_or_S3_Virge
>> CC: Natalia Portillo <claunia@claunia.com>
>>
>>> Cc: Hervé Poussineau <hpoussin@reactos.org>
>>>
>>> Fixed off-by-one drawing issue.
>>> Replaced hardcoded color for RECT.
>>> Separate I/O debug output for readability.
>>>
>>> Signed-off-by: Andreas Färber <andreas.faerber@web.de>
>>> ---
>>>  Makefile.objs                   |    1 +
>>>  default-configs/ppc-softmmu.mak |    1 +
>>>  hw/pci_ids.h                    |    3 +
>>>  hw/ppc_prep.c                   |    2 +
>>>  hw/vga-s3.c                     |  694
>>> +++++++++++++++++++++++++++++++++++++++
>>>  5 files changed, 701 insertions(+), 0 deletions(-)
>>>  create mode 100644 hw/vga-s3.c
>>>
>> [snip]
>>
>> IMO using vga-s3-864.c as the file name may be better as S3 produced
>> many display chips.
>
> Actually in the meantime I've already renamed the file to vga-ibm8514a.c
> (better: vga-ibm8514.c) and renamed the registration functions to reflect
> the s3_vision864.
>

Actually what I mean is that, you may make the code to be more S3
Vision864-ish because S3 Vision864 extends its display capability, not
only 1024*768 4/8bpp.

> I agree that rather than bundling different graphics cards by one vendor we
> should bundle different implementations of the same chipset. I was assuming
> that the IBM E15 uses a different vendor_id at least but could share the
> initialization with other cards, once mst's declarative PCI initialization
> reaches master (prepared that locally).
> Supposedly the Miro 20SD is compatible with the S3 Vision864, and the ATI
> Mach8 was based on IBM 8514/A, too. Not sure about the Weitek P9100-based
> IBM S15.
>
> Any suggestions what to do about the RAMDAC would be appreciated. The
> firmware only accesses the DAC_MASK register but does not write to the DAC
> itself, so for colors the card needs a reset handler that initializes the
> 256-color palette to the right RGB values.
>
> Andreas