Re: G4 Weirdness


Subject: Re: G4 Weirdness
From: Tom Rini (trini@kernel.crashing.org)
Date: Wed Nov 03 1999 - 14:39:18 MST


On Wed, 3 Nov 1999, Tom Tarka wrote:

> Tom Rini wrote:
>
> > On Wed, 3 Nov 1999, Tom Tarka wrote:
> >
> > > Something, I'm not sure what, happened to my G4 running YDL that, even with
> > > the BootX arguments
> > > "hda=noautotune", my machine gets a DMA timeout on hda. AFAICT, I haven't
> > > changed anything!
> >
> > noautotune means you wouldn't (read aren't) doing UDMA anyhow on those
> > drives. Unless you have an unflashed rev1 B&W, you won't be doing UDMA
> > regardless..
>
> if i do not use noautotune, my drive gets a DMA timeout on boot, the kernel
> panics,
> and it reboots before I even get into a second runlevel. This is a known issue
> for G4s...
> (in addition to some G3s)

For the moment, noautotune is _required_ on RevB B&Ws and "Yikes" G4s.
Newer kernels might not panic, but I've seen a good deal of corruption on
one of my machines. Short of it is we can't/won't/don't do UDMA on the
cmd646-7 and the cmd646-5 with an apple-sugjested OF update (G4 disabling
one). From what I can tell this might be some sort of a hard-coded "Fix"
for some of the flakeyness apple noted with the revAs (Yes, MacOS does
seem to do UDMA, which is why i'm not entrily sure of why it's not off,
however the OF update seemed to trigger the change.)

---
Tom Rini (TR1265)
http://gate.crashing.org/~trini/



This archive was generated by hypermail 2a24 : Fri Dec 03 1999 - 19:07:32 MST