[OpenBIOS] HelenOS regression in QEMU implies OpenBIOS VGA issue

[Mark Cave-Ayland]

On 16/03/12 20:59, Benjamin Herrenschmidt wrote:

> I observed that as well, it hangs some time during the display of
> the boot messages.
>
>> A local git bisect shows that the culprit is actually this commit:
>>
>> commit b6de01a59394f1fafbfb2c65cb784c9d23067fa2
>> Author: laurent<laurent at f158a5a8-5612-0410-a976-696ce0be7e32>
>> Date:   Tue Aug 4 21:08:37 2009 +0000
>>
>>       Allows to configure screen size from Qemu command line options
>> using FW_CFG
>>       interface.
>>       Only enabled for PPC/QEMU and VGA/VBE.
>
> I'm surprised by that. This has worked well for a long time, I think
> something else in qemu causes that commit to fail.
>
>> git-svn-id: svn://openbios.org/openbios/trunk@538
>> f158a5a8-5612-0410-a976-696ce0be7e32
>>
>> Having stepped through vga_vbe_init() in drivers/vga_vbe.c, the only
>> difference is that reading the display depth from the QEMU FW CFG
>> interface returns 15 instead of the in-built OpenBIOS default of 8.
>
> AFAIK that's been working fine for a while...
>
>> So is a bit-depth of 15 supported by OpenBIOS? This seems quite unusual
>> given that it is not a power of 2; however if I use GDB to override the
>> value to 16 then HelenOS boots but with a blue screen BIOS on VGA so I'm
>> not sure that is the correct solution either. Does anyone know what the
>> default bit-depth is in Apple's firmware?

For those people not on the HelenOS/qemu-ppc lists, I can confirm that 
the bug is actually in HelenOS which currently doesn't work with 
non-8-bit displays. In the meantime, you can boot HelenOS with my QEMU 
PPC modified interrupt patch and a recent OpenBIOS by forcing the 
bit-depth to 8 using "-g 800x600x8" on the QEMU command line.


ATB,

Mark.