Re: Mac 840AV, was Re: Status of d-i kernel transition for 2.6.14
On Sun, Dec 11, 2005 at 02:22:44PM +1100, Finn Thain wrote:
> From my experience with the 660AV (which for all intents and purposes is
> the same as an 840AV, only slower) and from anecdotal reports from various
> testers (going back to about 2.6.7), You could expect intermittent trouble
> with the 840AV if you did acquire one. Neglecting ADB, the 650 isn't in
> bad shape by comparison. (BTW, I found an old post from Wouter Verhelst
> that comfirms that his mac is Centris 650.)
> As I recall, the relevant issues with Debian devs' macs are,
> - unreliable built-in (MACE) ethernet (840)
Yes, the driver is a little hackish, and is the only thing actually
using PSC DMA at the moment. It also has a tendency to DMA over the
kernel during boot if you aren't careful about using the right
version of Penguin and the right settings.
> - unreliable CUDA ADB (840?)
Apparently there are some issues with most of the ADB drivers. The
PMU and EGRET (IIsi) support has never been rock solid, and somewhere
along the line CUDA and MacII-style ADB both got broken, leaving only
the IOP ADB to be solid. It's easy to get that one working, since it's
a pretty trivial driver once you get the IOP chips up and running.
> - real time clock doesn't work (650 and 840)
This is a generic problem. I think the RTC is broken for all
models. In any case, the RTC is dependent on ADB for the 840, as
well as other CUDA, PMU, or EGRET based models. Even after we get
the generic problems fixed, anyone with a Q900 or Q950 will still
have to wait for someone to figure out the caboose chip.
> - Mac II ADB driver boot crash (650)
I guess I should get my Q650 running again to test this. I did have
it up and running earlier, but I really don't have room to have it
set up properly.