lansd@eecg.toronto.edu (Robert Lansdale) (12/21/88)
I have read a few messages lately from MINIX gurus attesting to the long and tedious hours required to port MINIX to other machines. I am just putting the final touches on a 68000 board to be used as a host for my DSP multi-processor, and thought it would be a good lesson in O/S porting if I was to port the Atari code to this machine. The machine in question only has a SCSI port and two RS-232 lines, no floppies. For code compilation and development I could use my Atari-ST and Mark Williams C Version 3.0. What I would like to know is: how much time will I have to invest to get MINIX 1.3 up and running on the board? Of this time, what proportion will be spent updating the 1.1 sources to 1.3, looking for updates, etc. While I'm asking, has anyone ported MINIX to the Motel Cypher 68000 computer? -- ------------------------------------------------------------------------- ____ ____ ____ ____ |____| |____| | | | CSNET: lansd%eecg.toronto.edu |____ |____ |____ |____| ARPA: lansd%eecg.toronto.edu@relay.cs.net | Electrical Engineering Computer Group,
Dickson@pco-multics.hbi.honeywell.com (Paul Dickson) (01/05/89)
> I have read a few messages lately from MINIX gurus attesting to >the long and tedious hours required to port MINIX to other machines. I >am just putting the final touches on a 68000 board to be used as a host for >my DSP multi-processor, and thought it would be a good lesson in O/S >porting if I was to port the Atari code to this machine. > > The machine in question only has a SCSI port and two RS-232 >lines, no floppies. For code compilation and development I could use >my Atari-ST and Mark Williams C Version 3.0. > > What I would like to know is: how much time will I have to >invest to get MINIX 1.3 up and running on the board? Of this time, >what proportion will be spent updating the 1.1 sources to 1.3, looking >for updates, etc. I have been spending time in the past 7 months porting the PC Minix to the Slicer (an 80186 Single board computer), so I do have a little experience with porting to a slightly different hardware base. My friend (Kevin Fleming) and I had Minix mostly up and running within 2 months, only working most weekends. The major reason it has taken so much longer is because were waiting for a final version of Minix 1.3. Right now, we have the commands and libraries converted and compiled to Turbo C 2.0 and expect to have it running by the end of January (if we can find the time). After we get it running on a standard PC, we will add our changes that are necessary for the Slicer. Here are list of things you will find that you need and/or want: Fast compiler Separate development machine Listings of kernel files The listings are fairly important. They let you see all the code to figure out the dependencies and what constants you have to change. Taking a day to read and understand the kernel code will greatly shorten the porting time. Besides, once to see the code set on paper, it doesn't look as difficult as you would think. In your case your might want to consider hooking the SCSI port directly to the SCSI bus of your Atari ST (I believe the Atari does have SCSI port, doesn't it?). Then write a boot strap ROM code to read the Minix code directly from the SCSI port. After you get the code loaded into memory, take it one procedure at a time in the calling sequence, putting some kind of output flag to let you know where in the code you are so you can figure out were to look if it dies. This is pretty easy since the code already runs on the CPU/compiler you are using. Probably the most difficult problem you will face is how to get the updated code to the target machine. The faster the better. Personally, I wouldn't go through the serial port unless I absolutely had to. This is because the worst part is the debugging cycle, which is why it is a good thing to have a separate machine for editing and compiling the code. -Paul Dickson Arpanet: Dickson%pco @ BCO-Multics.HBI.Honeywell.COM