Bounty: A500 KickStart v1.3 patch with support for scsi.device 40.63

[alexh]

Originally Posted by dJOS

The scsi.Device is 40.63 but i haven't tried 40.70.

1) Those are Kickstart version numbers not scsi.device version numbers?
2) I think you mean 40.68 and not 40.63?

Kickstart 40.68 = scsi.device 40.12 (A1200)
Kickstart 40.70 = scsi.device 40.20 (A4000)

[dJOS]

Originally Posted by alexh

1) Those are Kickstart version numbers not scsi.device version numbers?

2) I think you mean 40.68 and not 40.63?



Kickstart 40.68 = scsi.device 40.12 (A1200)

Kickstart 40.70 = scsi.device 40.20 (A4000)

I could well be mixing up my version numbers, doh!

Edit: as per pic above the patched Rom shows version 43.45.

[alexh]

I believe all patched versions report 43.45 regardless of what you patched?

Has this still not been resolved?

It is still impossible to boot a KS1.3 partition with a KS1.3 v34.5 patched to contain a scsi.device driver?

[kipper2k]

k..
time to up the ante.

i'll increase the bounty to $100 in goods from my store for this to be solved

[dJOS]

Originally Posted by alexh

I believe all patched versions report 43.45 regardless of what you patched?

Has this still not been resolved?

It is still impossible to boot a KS1.3 partition with a KS1.3 v34.5 patched to contain a scsi.device driver?

I've been able to get them working but only with a boot floppy and a mount-list.

Cheers kipper, I'll update the first post.

[Vot]

Originally Posted by dJOS

I've been able to get them working but only with a boot floppy and a mount-list.

Cheers kipper, I'll update the first post.

Im still going to have a look i just haven't had any time to fiddle with my a500.

[dJOS]

Cheers Vot

[kipper2k]

k...

I think we have a winner, i have just autobooted a KS 1.3 from CF card (i had a failat error, but who cares, it is booting!!

more info later

[kipper2k]

here we go...

http://www.kipper2k.com/_aaa/Untitled.html

[dJOS]

Awesome kipper, I'll have a look a bit later today (have some house Reno to do).

[dJOS]

Ok I finally got a chance to try your mod kipper (moved house 2 weeks ago and still unpacking) and it's no joy for me - my rev6a + 1mb chip-ram mod hates it. I get a purple screen then black with diagonal lines followed by a nice solid orange screen.

I tried 3 different 27c400's (known working) and even byte-swapped the last one. :P

[Beska]

I'd be interested in know how you did this Kipper, I've been playing around with this myself using a combination of LoadModule and custom romtags and have only managed to get a solid white screen on boot. I know I'm close but I'm missing something.

[A10001986]

Better late than never:

Just solved this (the original problem, that is), but please donate the bounty to some charity

Since I am running mkl's ram68k+ide68k on a machine similar to an A500 (to be exact: A1000 with Phoenix board from 1990), I used the A600 version of KS 3.1, which is running scsi.device 40.5.

I assume (but haven't tested this) that the same issue exists with the 1200's kickstarts.

1) Follow the instructions in kick1.3scsipatch

2) Before swapping bytes for programming:

a) open the new kickstart file a hex editor

b) look for the Romtag structure at the beginning of the file. It starts at around $50 and reads
4AFC 00F8 XXXX 00FB FFE8 0128 030A 00F8 XXXX 00F8 YYYY
Write down the values YYYY (should be something like 009x, but depend on the scsi.device version you are using)

c) search for the end of the scsi.device code. At around $2900 the data becomes all FFFFFFFF. Overwrite the first FFFFFFFFs with the following hex data:

48E78080 41F900DE 100010BC 00011010 10101010 10101010 10101010 10104CDF 01014EF9 00F8YYYY

where YYYY is what you wrote down in step b)

d) Note the location of the data you just entered. It should be around $2990 from the start. You need to note the EXACT offset from the start of the file to the first byte (48) from the data above.

d) go back to the romtag structure and replace YYYY with the offset you just noted.

Background:

What is actually missing from the combination of KS 1.3 and scsi.device from some later KS is some Gayle-initialization:

lea.l $de1000,a0
move.b #$0,(a0)
move.b (a0),d0
move.b (a0),d0
move.b (a0),d0
move.b (a0),d0
move.b (a0),d0
move.b (a0),d0
move.b (a0),d0
move.b (a0),d0

That's it. This code must be executed before scsi.device initializes. It somehow "enables" Gayle. Without it, scsi.device does not recognize Gayle and ... does nothing.

The procedure above adds this code at the end of the scsi.device hunk and changes the init-ptr of the romtag structure to run this code, which in turn jumps back to the original init routine.

Edit: Please be aware that this is a q&d solution. The correct procedure involves a check for mirrored dff000 custom chip registers at de1000. Please use this only in machines that actually have a Gayle or compatible device.

[dirkies]

Thanks for reviving this thread!

Your solution looks interesting indeed, but the last sentence ruins it...

Originally Posted by A10001986

Please use this only in machines that actually have a Gayle or compatible device.

The thread clearly says this should work on an A500, and since Gayle was A600 and later, the bounty is still out there...

[idrougge]

Depends on how compatible the Ide68k is with Gayle. Note that A10001986 wrote that he uses an Ide68k on a Phoenix board, not on an A600.

[A10001986]

I don't give a rat's behind about the bounty, I wanted a booting system. And since no one has had any luck with this issue until yesterday, I thought maybe others are interested.

The procedure is - in a more complicated form, as mentioned - what Gayle-enabled Kickstarts >=2.05 do. Look at KS 3.1(1200), it starts at $f8046a.

I guess if this is written for Gayle, well, it should work on Gayle. So we have A600 and A1200 covered (if anyone is really willing to run 1.3 on an A1200).

And since a lot of IDE controllers out there emulate Gayle (and those exist for A500, A2000, A1000, ...), it works on those too, otherwise those would really suck at emulating Gayle.

Long story short: If your Amiga (whatever model) can boot from your Gayle-based or Gayle-compatible IDE controller using KS >=2.05(37.350), the patch works.

Instead of judging how the solution looks, how about trying it out?

For obvious reasons, I can't share the kickstart file here. But you can PM me for assistance.

[dJOS]

Brilliant detective work A10001986, when I get a chance I'll try it out.

[A10001986]

Created a gpatch for scsi.device 40.5.

1) Use GetScsi11 for grabbing. scsi.device 40.5 is in KS 3.1 40.63 (which is the usual KS 3.1 for A500, A600, A2000). Please do not execute SetPatch before grabbing. Otherwise you might end up with a later scsi.device, for which the patch is not suitable.

2) Apply the attached patch using gpatch, just like when using the patches in kick13scsi

3) Join the result and your 1.3 image (NOT the other way round!).

4) Proceed as described in the docs of kick13scsi (ie swap bytes, burn)

[dirkies]

It's not that I think your hard work is irrelevant, I appreciate you're trying to find the holy Grail, but me too it's been years I hoped someone would tinker a ks1.3 that can be fitted to a standard A500 which would allow it to boot from a HDD/CF card without any floppy drives to load first. Hence there's a bounty on this, as for some reason it just doesn't seem to be possible (yet).

[idrougge]

Didn't he just describe how it was possible?