Subject: new r4k code problems
To: Simon Burge <simonb@telstra.com.au>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
List: port-pmax
Date: 06/17/1997 03:53:56
>Howdy folks,

>I've built a kernel for a 5000/260 from last night's SUP, and both
>booting of disk and diskless it hangs after the "root file system type"
>message.  I've seen a quite a few commits during today, so I'll try and
>build a new kernel tomorrow.  The machine is at work, and I'm at home,
>so there's not to much I can do now it's wedged (all the reset-button
>pushers have gone home for the day).  In the meantime, any ideas?  Below
>are the boot messages booting of disk...

>Simon.

hi Simon,

This may be an alignment problem. Michael Hitch committed a change
that alignts the save-area for $at to an eight-byte boundary.  (could
this be the same chip bug that requires the bcopy() changes?)

Anyway, a kernel built from the latest sup will boot on a 5000/50, so
the 5000/260 should be okay. 

Most of the changes committed over the past day or so are aimed at
getting a single kernel to boot on either mips1 or mips3. We're very,
very close to that now.  Maybe in the day after tomorrow's sup.

>ps:  The kernel was too large to boot with tftp, but after an
>"addnode <hostname> -h <enet addr> -l <path-to-ecoff-kernel>", booting
>with MOP seemed fine.

This is very likely a PROM bug. (t's the same ecoff file, right?)
Conversely, 3100 can tftpboot a 3MByte Ultrix kernel just fine.