On Mon, Feb 04, 2013 at 11:23:59PM -0500, Programmingkid wrote:
On Feb 3, 2013, at 9:47 PM, Programmingkid wrote:
On Feb 3, 2013, at 7:17 PM, Amadeusz Sławiński wrote:
On Mon, Feb 04, 2013 at 12:24:26AM +0100, Andreas Färber wrote:
Am 02.02.2013 22:27, schrieb Programmingkid:
I am investigating the device tree of the mac99 model. It looks like it is wrong. The part that I am concerned with is the via-cuda node. Both my iMac and iBook don't even have a via-cuda. They both have a via-pmu@16000 node. I'm hoping someone has an original blue and white PowerMac G3. I would like the full device tree printout from it. The question I am trying to answer is does any newworld Mac have a via-cuda? If they don't, we know that it should be changed to a via-pmu@16000 node.
I have one around, and it has a via-cuda@16000 node. There's a power-mgt@0 node at the same level (under mac-io@5) but no pmu node.
With regards to mac99 target I would rather be concerned with lack of /uni-n, as darwin seems to check for this (it panics when checking machine type)
Do you think this is a QEMU or OpenBIOS issue?
Fast grep through qemu newworld machine shows that it is at least partly implemented:
% grep uni hw/ppc/mac_newworld.c static void unin_write(void *opaque, hwaddr addr, uint64_t value, static uint64_t unin_read(void *opaque, hwaddr addr, unsigned size) static const MemoryRegionOps unin_ops = { .read = unin_read, .write = unin_write, MemoryRegion *unin_memory = g_new(MemoryRegion, 1); memory_region_init_io(unin_memory, &unin_ops, NULL, "unin", 0x1000); memory_region_add_subregion(get_system_memory(), 0xf8000000, unin_memory);
In OpenBios it is only mentioned in PCI database and I think it is just used to setup /pci properties
As far as I can see the /uni-n node has following properties name = "uni-n" model = "AAPL,UniNorth" device_type = "memory-controller" compatible = "uni-north" reg #address-cells #size-cells device-rev = 0x7
Amadeusz