Welcome, Guest. Please login or register.

Author Topic: Highway USB causing some major Bootup Problems  (Read 3006 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline mboehmer_e3b

  • Sr. Member
  • ****
  • Join Date: Aug 2002
  • Posts: 312
    • Show all replies
    • http://www.e3b.de/usb/
Re: Highway USB causing some major Bootup Problems
« on: March 17, 2004, 11:12:32 PM »
Hmmm... why didn't you contact the manufacturer of this card? He might have some hints, but with the information you provide here I cannot deliver any guesses.
Did you send in your registration card, BTW ?
Your bug report here is like visiting the doctor and claiming "I don't feel good"...

Please contact mboehmere3b.de, providing for example:
- machine configuration
- processor card type / model
- Zorro board version / manufacturer
- other installed cards
- serial number of HIGHWAY
- Kickstart / OS version
- software failure number
- Early startup menu comments about Zorro cards

Don't be afraid of asking the guys who made this card - we provide support even to second hand users, which is not common in Amiga community these days...

Michael
 

Offline mboehmer_e3b

  • Sr. Member
  • ****
  • Join Date: Aug 2002
  • Posts: 312
    • Show all replies
    • http://www.e3b.de/usb/
Re: Highway USB causing some major Bootup Problems
« Reply #1 on: March 18, 2004, 08:34:41 AM »
Hi Steve,

known problem - but not in the HIGHWAY logic.

In Amiga Zorro bus systems, I/O expansion cards usually go to 0xe90000 ... 0xef0000, with minimum board space of 64k.
So some people concluded that the base address of the board will be 11101xxx (binary) anyhow, and "saved" some space in the buslogic by assuming this address assignment.
Unfortunately, they usually saved the space for "shutup" logic also, so the Kickstart can't deactivate cards if the I/O space is getting filled.

Two things may happen:
a) HIGHWAY / ALGOR is "last" in order and gets shutup by Kickstart => no longer accessible
b) HIGHWAY / ALGOR is "before" the bad guy and gets a valid address (like 0xea0000), while the bad guy receives an address in memory space (like 0x22000000) and is mapped by its malicious buslogic to the same address like the HIGHWAY / ALGOR
(00100010 binary => skip 5 bits => xxxxx010 => assumption 11101xxx => 11101010 = 0xea00000 which is I/O again :(

HIGHWAY / ALGOR can be mapped to any address within Zorro II, as well as supporting shutup logic. By changing the card order, you just leave the best places (at the beginning of autoconfig chain) to the bad guys, so they can't do any bullshit and get what they expect.

Hope this wasn't too technical, but exactly these problems were forseen by Dave Haynie when designing the Zorro autoconfig, so he stated clearly that all address bits have to be saved by an autoconfig board during address assignment.

Michael