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

Re: Fwd: Macintosh Quadra 950 Not Booting



On Sat, 8 Aug 2015, Greg Andrzejewski wrote:

> On 8/8/15, Finn Thain <fthain@telegraphics.com.au> wrote:
> >
> > I agree. And that reminds me: the Quadra 900 and 950 have an SCC IOP, 
> > like the Mac IIfx. That means you must tell MacOS to switch the serial 
> > ports to IOP-bypass ("compatible") mode. Otherwise the serial port 
> > won't work. (I don't know whether this applies to EMILE.)
> >
> > There is a control panel for that called Serial Switch, IIRC. That 
> > should get you a working serial console, and a backtrace. I'd be 
> > interested to know whether v3.x kernels also fail with the same BUG 
> > splat.
> 
> Incredible! I switch from Faster to Compatible and the kernel boots up 
> like nobody's business. Output on the framebuffer and all. Works great 
> with both Penguin and EMILE. Switch back and dead again.

Here's a couple more tests you could try, if you like.

With the SCC IOP in compatible mode and passing no kernel arguments --

Do you see the "ABCFGHIJK" on the quadra display immediately after my 
v4.1.4 kernel executes? (Perhaps video output only appears after macfb 
messages show up on the serial console?)

What about when you boot the v2.6.29 kernel? (I wouldn't expect any serial 
console with that one.)

> Earlyprink log is attached, in case anyone's interested. Not sure what 
> to make of the earlier scheduler warning. Unless it's something to do 
> with the IOP?

I don't know. Could be an early IOP interrupt? Certainly prev->pid was 
garbage in the __schedule_bug() error message you sent...

Note that the tty0 (macfb) framebuffer console and the ttyS0 (pmac_zilog) 
serial console both start fairly late, whereas the debug console should 
appear very early, well before the scheduler starts, and even before IOP 
is initialized.

See also comments in the source code regarding iop_preinit(). The kernel 
claims to be able to put the IOPs into "compatible" mode all by itself. 
Maybe it happens too late? We can't really do any iop initialization stuff 
earlier than "ABCFGHIJK" on the debug console. And the crash seems to 
happen before that, so we may have to depend on the boot loader or the 
control panel to do it.

> 
> Actually finding the Serial Switch control panel proved to be a bit of a 
> task, especially since Apple finally took down their venerable Older 
> Software Downloads page. I ended up using a program called TomeViewer to 
> extract it from the System 7.5.3 install files and dropped it into my 
> System folder. It's probably installed by default with 7.5.3, but I 
> 7.1's so much quicker and slimmer.

Running Penguin 19 on MacOS 7.5.3 should work with arbitrary 68k Mac 
hardware and avoids a known compatibility issue with Penguin on certain 
hardware with older MacOS that's described on www.mac.linux-m68k.org. (The 
difference in size and speed is marginal IMHO. I find Open Transport is 
helpful too.)

-- 

> 
> >
> > I can't explain why there is no output on the framebuffer. Lots of 
> > other Quadras have a DAFB chip and they generally work OK. Maybe try a 
> > different video mode and/or color depth? Since you have no expansion 
> > cards installed, I don't know what else to try.
> 
> Looks like something's going haywire with the serial ports set to 
> Faster. I wonder if you'd run into the same thing on the IIfx.
> 
> -Greg
> 


Reply to: