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

[alexh]

1) Something has changed with the Gayle emulation in the CFIDE68k CPLD code since Thomas Köckerbauer did his patch?

2) It only works for A500+ (after a reset) because of address map conflicts? (The CFIDE68k connects directly to the 68k CPU but IIRC if something else responds to a read to address XXXX then there will be a conflict on the DATA bus?) WinUAE is unlikely to emulate data bus-conflicts?

3) Mixture of the two?

[Vot]

Originally Posted by alexh

1) Something has changed with the Gayle emulation in the CFIDE68k CPLD code since Thomas Köckerbauer did his patch?

2) It only works for A500+ (after a reset) because of address map conflicts? (The CFIDE68k connects directly to the 68k CPU but IIRC if something else responds to a read to address XXXX then there will be a conflict on the DATA bus?) WinUAE is unlikely to emulate data bus-conflicts?

3) Mixture of the two?

I will be able to try dJos's rom (the patched 1.3 rom created from Thomas's patch ) when my board from kipper turns up... at least we will know if it still works with my a500+.

Also i had a look what the patch does. I looked at what the patch applies to scsi.device 40.12 from an a1200.

it adds the asm shown in the distribution for adding a boot node near the end of the scsi.device (file offset 000029f4) where it looks like it calls expansion.library. (Note: the scsi.device only patch's only work on the dumps from the getscsi tool mentioned in the readme which generates a different dump from if you use romsplit - his read me says something about getscsi removes a call to adddevicenode)

The patch also modifies some bytes at the start of the device (at offset 00000017 and 00000019) from 76 to 87 but id have to look up what part of the device structure that is.

But as you have said, the patched 1.3 rom should work if the cpld is using the same codebase as when Thomas developed it. What doesn't make sense to me is djos says the 2.x roms with scsi.device do work....that would imply its the patch doing something wrong.

Anyway at-least i can test it in an a500+ which is discussed in Thomas's distribution.

[kipper2k]

Originally Posted by Vot

I will be able to try dJos's rom (the patched 1.3 rom created from Thomas's patch ) when my board from kipper turns up... at least we will know if it still works with my a500+.

Also i had a look what the patch does. I looked at what the patch applies to scsi.device 40.12 from an a1200.

it adds the asm shown in the distribution for adding a boot node near the end of the scsi.device (file offset 000029f4) where it looks like it calls expansion.library. (Note: the patch only works on the getscsi tool which is a different dump from if you use romsplit - his read me says something about getscsi removes a call to adddevicenode)

The patch also modifies some bytes at the start of the device (at offset 00000017 and 00000019) from 76 to 87 but id have to look up what part of the device structure that is.

But as you have said, the patched 1.3 rom should work if the cpld is using the same codebase as when Thomas developed it. What doesn't make sense to me is djos says the 2.x roms with scsi.device do work....that would imply its the patch doing something wrong.

Anyway at-least i can test it in an a500+ which is discussed in Thomas's distribution.

2.05 and higher will boot up, 2.04 doesn't

[Vot]

Originally Posted by kipper2k

2.05 and higher will boot up, 2.04 doesn't

Yea it has no scsi.device

[mark_k]

Originally Posted by Vot

Also i had a look what the patch does. I looked at what the patch applies to scsi.device 40.12 from an a1200.
..
But as you have said, the patched 1.3 rom should work if the cpld is using the same codebase as when Thomas developed it. What doesn't make sense to me is djos says the 2.x roms with scsi.device do work....that would imply its the patch doing something wrong.

I haven't looked at the patched code, but scsi.device from Kickstart 40.63 calls TaggedOpenLibrary() to open utility.library in its init routine. Presumably the Aminet patch must patch that and all calls to utility.library functions, in order to work at all under Kickstart 1.3 though.

[Vot]

Originally Posted by mark_k

I haven't looked at the patched code, but scsi.device from Kickstart 40.63 calls TaggedOpenLibrary() to open utility.library in its init routine. Presumably the Aminet patch must patch that and all calls to utility.library functions, in order to work at all under Kickstart 1.3 though.

Currently i think 43.45 is the highest one it has patches for:

http://aminet.net/driver/media/SCSI4345p.lha

Id have to boot my amiga back up to check

Not sure if it has calls to any other 2.x functions.

[Vot]

Originally Posted by dJOS

You can see the HD in HDToolBox and partition, but it but it never appears on the WorkBench Desktop and wont boot either. Also if you query the scsi.device version, it Guru's your a500.

I just had a thought.

Try making a mount-list for your cfcard and mounting the hdd from a floppy using the mount command.

To make a mount-list giggle-disk on a floppy should work.

See if it can communicate with the cfcard then..

[dJOS]

Make me an adf (hdd is dh0) and I'll burn the ROM and take my a500 apart again to test.

Ps I've never made a mount list and have no clue how.

Pss that won't stop a version query of scsi.Device crashing the Amiga.

[Vot]

Originally Posted by dJOS

Make me an adf (hdd is dh0) and I'll burn the ROM and take my a500 apart again to test.

Ps I've never made a mount list and have no clue how.

Pss that won't stop a version query of scsi.Device crashing the Amiga.

No but we can verify if transfers actually work

[kipper2k]

not sure if there is any relevant info here, but i thought i'd post it anyway...

http://eab.abime.net/showthread.php?t=54028

[Vot]

Originally Posted by kipper2k

not sure if there is any relevant info here, but i thought i'd post it anyway...



http://eab.abime.net/showthread.php?t=54028

Yep as suggested in the thread im going to get him to make a mount-list and try to mount the hdd from floppy.

Giggle disk just makes it easier as it will automatically make the mount-list. Trouble being giggle disk only runs on 2.x. So he will need to boot with 2.x create the mountlist then try the patched 1.3 kickstart with the mountlist

[dJOS]

Sorry chaps haven't had much time to play with my a500 this weekend (little kids are high maintenance especially when 1 of them is teething!).

Vot if you fancy a drive over to Vermont south I'll give you a patched ks1.3 ROM to play with.

[Vot]

Originally Posted by dJOS

Sorry chaps haven't had much time to play with my a500 this weekend (little kids are high maintenance especially when 1 of them is teething!).

Vot if you fancy a drive over to Vermont south I'll give you a patched ks1.3 ROM to play with.

I know your pain.

I have the same problem. My 2 year old has decided to not sleep all night.

EDIT: I know why. We both have some virus .. Sore throat

I did make you an adf but it got got invalidated (must have reset when it was writing) so i need to make it again. Also it seems giggledisk doesnt make a 1.3 compatible mountlist. Hopefully I will have a chance tomorrow.

As for a drive.. Hahahah I'm in Cairns thats about a 3000km drive.

[dJOS]

Lol, i thought you where in Melbourne for some reason.

Any way I burnt some new ROM's while the kids where sleeping and ran some more tests.

Querying the scsi device version no longer crashes workbench which is good (no idea why) but i still don't get a mounted drive even tho hdtoolbox can see the drive.

I'll post some pics shortly.

Edit, pics:

[dJOS]

So Vot very kindly made me a WB1.3 ADF with mountlist generator and it works great with patched KS1.3 using my existing DH0 and DH1 partitions.

So does that mean the v40.63 scsi.device is not automounting perfectly compatible partitions or is something else the root cause?

[lesta_smsc]

Originally Posted by dJOS

So Vot very kindly made me a WB1.3 ADF with mountlist generator and it works great with patched KS1.3 using my existing DH0 and DH1 partitions.

So does that mean the v40.63 scsi.device is not automounting perfectly compatible partitions or is something else the root cause?

Does that mean the bounty is no longer available lol.

[Vot]

Originally Posted by lesta_smsc

Does that mean the bounty is no longer available lol.

Well what it does tell us is that the scsi.device in the patched kickstart dJOS has works fine. Its just an autoboot issue. (Be that due to the ide emulation or the patch somehow.. Im guessing he would still prefer an auto-booting hdd rather than using a floppy to boot it.

I might have asked before dJOS but what version scsi.device is in the rom and how did you make it.

I.e with the gpatch that just patches the scsi.device & you used remus
or the gpatch that directly makes a patched kickstart from a stock one.

[dJOS]

Originally Posted by lesta_smsc

Does that mean the bounty is no longer available lol.

No, the bounty is to make a patch that allows ks1.3 to work fully with c.f. Ide hdd's. Booting off a floppy to mount the drives is just a step in the discovery process.

If it's a bug in the c.f. Ide code then I'd still pay the bounty to whoever fixes that and releases to code to the community.

[dJOS]

Originally Posted by Vot

Well what it does tell us is that the scsi.device in the patched kickstart dJOS has works fine. Its just an autoboot issue. (Be that due to the ide emulation or the patch somehow.. Im guessing he would still prefer an auto-booting hdd rather than using a floppy to boot it.

I might have asked before dJOS but what version scsi.device is in the rom and how did you make it.

I.e with the gpatch that just patches the scsi.device & you used remus
or the gpatch that directly makes a patched kickstart from a stock one.

Iirc it was gpatch, i followed the instructions in the linked aminet patch to the letter using my a1200.

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

[dJOS]

ROM Patch Link:
http://aminet.net/package/util/boot/kick13scsipatch

I might make some KS3.1 40.70 roms for my a1200 and try with that version - TBH I hadnt considered trying this till now. I think im running original KS3.1 40.68 roms atm so it'll be interesting to see if 40.70 has a newer scsi.device version.