[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: where's tfm?

Hi all,

On  14 Feb, this message from Benjamin Herrenschmidt echoed through
> On Sun, 2005-02-13 at 13:46 -0800, Brad Boyer wrote:
>> On Sun, Feb 13, 2005 at 02:08:19PM -0200, Felipe Fonseca wrote:
>> > yes. that looks like a problem...
>> > 
>> > On Sun, 13 Feb 2005 18:02:13 +0200, Eddy Petrisor
>> > <eddy.petrisor@gmail.com> wrote:
>> > > >"00:0b.0 Host bridge: Apple Computer Inc. Bandit PowerPC host
>> > > >bridge (rev 03) 00:10.0 Class ff00: Apple Computer Inc. Grand
>> > > >Central I/O (rev 02)"
>> Actually, the 1st gen PCI powermacs had almost no PCI devices.

Not really true.... they don't have a high number of PCI devices, but
everything _is_ connected to PCI.

>> The
>> builtin video is not actually a PCI device.

Well, though it s not a PCI card, and it is electrically probably not
really compatible to PCI (but who knows...), it does have a PCI
personality on a bus that has PCI semantics. You know, it may have been
designed as a real PCI device, but legend has it that Apple's hardware
engineers were more after design than function....

>> You need the controlfb
>> driver in the kernel and the fbdev driver in X. It won't be very
>> fast since it's not accelerated, but it should work.

This, on the other hand, is the real truth, brother :-)

> Actually ... control is behind the "chaos" chip which is a PCI device

Yes and no. It is a PCI host bridge, but unlike 'real' host bridges, it
has no PCI personality on the child bus. So, if you lspci the chaos PCI
bus, there is no chas device on there.

> (it contains DBDMA engines so one could theorically write some blit
> acceleration there ...)

No, chaos has no DBDMA engine AFAIK. PlanB has (you are surely confusing
here :-). PlanB could in theory do the bitblit, but noone has succeeded
in making it move data _between_ memory locations. From video to memory,
yes, but not from memory to memory.

> I'm surprised that chaos isn't showing up there, could be a problem
> with the PCI code on these old machines.

No, this is a case of 'special' hardware. Nobody will ever know if it is
a hardware bug that was not deemed important enough to fix, or if it is
so be concious design decision. Apple hardware :-) Talk to the Mac68k
guys, they can tell you some :-))



Michel Lanners                 |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes            |    Ask Questions.  Make Mistakes.
L-1710 Luxembourg              |
email   mlan@cpu.lu            |
http://www.cpu.lu/~mlan        |                     Learn Always. "

Reply to: