Re: Booting original iMac

[Michael <macallan%netbsd.org@localhost>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On Mar 11, 2010, at 3:39 PM, John D. Baker wrote:

> So, I was finally able to build a NetBSD-4.0_STABLE for macppc and  
> used
> it to netboot my old original bondi-blue iMac (233MHz MPC750).  See
> "dmesg.boot" at:
>
>  http://bobdbob.com/~jdbaker/imac-nb4-dmesg.txt
>
> The file contains boot messages of a GENERIC kernel, GENERIC with
> "machfb" disabled via userconf, and a custom kernel built from:
>
>  http://bobdbob.com/~jdbaker/BONDI-nb4.txt
>
> The machfb attachment leaves the display garbaged, but booting  
> completes
> and I can log in blind and at least reboot the machine.  Disabling  
> machfb
> via userconf lets ofb attach and everything works well from a text
> console.

Strange, I used to have access to a machine like that ( also with 6MB  
VRAM ) and machfb Just Worked. That was years ago though, around or  
before the 4.0 release.
The description ( garbage in the upper region of the screen ) sounds  
like something tried to switch to 24bit mode which makes no sense at  
all. Also, machfb isn't supposed to mess with video modes at all -  
guess I'll have a look at the code again.
On a hunch - if you have a MacOS CD handy - could you switch the iMac  
to 800x600 and see if that changes anything?
Also, is the picture stable when machfb messes up? If machfb does try  
to switch video modes it will probably use one of the standard modes,  
most of which don't work on the iMac's built in display. On the other  
hand, if it does try to switch modes it should use the firmware- 
provided EDID data.
Too bad I don't have that kind of machine around, machfb works fine on  
everything I have here :/

> I haven't tried X yet.  The custom kernel mostly just eliminates  
> stuff I don't/can't have.

The Rage Pro is well supported, the mach64 driver should work fine.  
You will have to add some config option to tell X that the built-in  
display has a fixed line frequency of 60kHz - any mode that doesn't  
match 60kHz horizontal will not display properly or not at all.

> Previously, in
>
>  http://mail-index.netbsd.org/port-macppc/2009/12/13/msg000981.html
>  http://mail-index.netbsd.org/port-macppc/2010/01/16/msg000998.html
>
> I described problems booting -current on the iMac (hangs after message
> about waiting for SCSI devices to settle).  I built and tried  
> 5.0_STABLE,
> but it has the same problem as -current.
>
> I then built a custom kernel for NetBSD-5.0_STABLE based on what I did
> for 4.0 just in case the problem was with other added drivers.
>
>  http://bobdbob.com/~jdbaker/BONDI-nb5.txt
>
> But this does not help.

I seriously doubt it's the SCSI chip, that's just (one of) the last  
line(s) displayed. At that time the kernel will also probe for USB  
devices and a bunch of other things, any of which could cause the hang.
You don't need any of the devices that attach to adb, only adb itself  
( or alternatively, cuda ). The iMac doesn't have any actual ADB  
hardware, you still need the adb driver to access the real time clock  
and to reboot the machine. Yes, the driver's name is misleading, it  
does more than just provide access to ADB.
In 5.0 - does the kernel probe USB and IDE devices before hanging?
Also, does anything change if you get rid of adb and use cuda instead?
( cuda does what adb does, only with clean separation of the actual  
ADB support code, a bunch of bug fixes and support for an i2c bus  
which is needed only on beige G3s so far )

have fun
Michael

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)

iQEVAwUBS5luL8pnzkX8Yg2nAQJk6gf9FWG8qTWEbKSBV8LT0OS+alhL4wYEscEz
EslDQOWzXjVOVZEOjVl1y6aLIk5Q1Mi+0ukVAReA5gtqIyG13yoYSoRY8mPQTx8D
cwA88v3nwfipDU2vi0ZzMxZa9A/qNHFQFIjb7ZK1+WFAZN1qL4WhBiRY60Vz2xNF
Wlh3Z9ahcTLhHAr3ItzMzT/3QWvFB2E0EOQ4U9SgbtYmuky2M5qe8qYyl1G+6rnA
ZJ9+xEWzKzfHkBXmovGAHQTK9oaASAIoQcnmQErJmgeiKRptR9Hi9STka9dhOWPN
WLl4gctrBuLHzfAw46cVj/3OjoAIK7iaZ46x3gedhtjGEUxDe4glfw==
=0vSS
-----END PGP SIGNATURE-----