Brian Baird writes:
> I'm pretty sure that Cisco 25xx routers have 20MHz 68030's in them,
> not Coldfire's.  In fact I can't offhand think of any Cisco routers
> that have coldfires:
> 
>      Cisco 1000/1400		Mot 68360 (Dragonball)
>      Cisco 2500			Mot 68030

2500:
CPU	Moto MC68EC30FE25C (25Mhz or underclocked to 20Mhz?)
Ether	AMD AM79C90JC (+ AM7992BJC?)
Async	Moto MC2681FN
Sync	Hitachi HD64570F

>      Cisco 1600/1700/2600	Mot MPC860 (ppc)
>      Cisco 3000			PPC of some kind

The 3000 (or "IGS") has a 68EC030 like the 2500.

>      Cisco 3600			R4700/RM5271 (mips)

3620:
mobo:
CPU	Orion SDT79R4700-80MS
Async	National? PC16552DV
PCI	Galileo GT-64010A-B-0
PCMCIA	Cirrus Logic, CL-PD6729-QC-C
ARB??	Altera, EPM7032LC44-7
REG??	Altera, EPF8452AQC160-4
IO??	Altera, EPM7032LC44-12

network module: (PCI of some format? PMC?)
Ether	AMD AM79C970AKC
Bridge?	Xilinx, XC4013E

With a PCCARD SCSI adapter this might be nice. :-)

>      Cisco 4000			R4400? (mips)

The 4000 is a 68EC030 (socketed?) AFAIK.  I could open one up if there
is real interest.

> > 	Anybody working on a Moto coldfire, m68kish, NetBSD port? Can't let
> > 	them oily penguins get all the Cisco 2501 glory! B^).

I started working on it before I fully realized they had a 68EC030 without
a proper MMU.  If I remember some details from the datasheet correctly the
'EC' part ignores the MMU instructions, but provides some limited ability
to map "segments" so that a program at 0xN00... thinks it is at 0x00000...
I don't know if this shadowing would be enough to allow a "normal" m68k
binary to run, just without any memory protection.
I guess the idea would be for the kernel to run at the top of memory and
at context switch the process to be run would be shadow mapped to 0x0.
It could still clobber anything else in memory though. :-)

I may be totally wrong about this, so take this with a shovel of salt. :)

-Andrew