Updating a kickstart rom

scrappysphinx

Confused retro man
VIP
Joined
Jan 30, 2008
Posts
7,964
Country
UK
Region
Nottingham
Hi all, I have a gq4x and have burned kickstart roms successfully onto 27c400 and dual kickstarts onto 27c800 eproms.

I am now trying to create a custom kickstart 3.1 with updated scsi.device for larger HDD partitions.

I have used romsplit to extract the ROM and rebuilt it using Remus omitting card.resource and carddisk.resource in order to fit the newer scsi.device

After compiling the ROM and testing in WinUAE it all worked well so I went ahead and burned it to the EPROM

Placed it into the 500+ and i am presented with the 3.1 kickstart screen. Booting from floppy into workbench and using shell to check scsi.device version shows the newer version has loaded fine. However when a HDD is attached I am still shown the kickstart screen so it seems it's not able to autoboot.

Anyone have any tips on where I've gone wrong?

Do I need to add the modules to the ROM in a specific order or something?

Thanks for any tips
 
Hi all, I have a gq4x and have burned kickstart roms successfully onto 27c400 and dual kickstarts onto 27c800 eproms.

I am now trying to create a custom kickstart 3.1 with updated scsi.device for larger HDD partitions.

I have used romsplit to extract the ROM and rebuilt it using Remus omitting card.resource and carddisk.resource in order to fit the newer scsi.device

After compiling the ROM and testing in WinUAE it all worked well so I went ahead and burned it to the EPROM

Placed it into the 500+ and i am presented with the 3.1 kickstart screen. Booting from floppy into workbench and using shell to check scsi.device version shows the newer version has loaded fine. However when a HDD is attached I am still shown the kickstart screen so it seems it's not able to autoboot.

Anyone have any tips on where I've gone wrong?

Do I need to add the modules to the ROM in a specific order or something?

Thanks for any tips

Just to give to a sanity check here, I too have attempted this and was left with the same result. I followed a guide that explicitly specified the order and excluded the same two files. Is it possible they’re required for correct gayle activation?

I ended up using Cloanto’s 3.X ROM and their workbench.library to prevent the reboots on my A1000.
 
i found on my 500+ v8.1 that the 1.1 rom doesn't run but runs fine on rev 6a a500, i think 3.x ran fine but obvious old stuff that loaded wb dont work anymore since not workbench.library not in rom anymore
 
I'm not an expert on the updates to scsi.device over the years (one day I will dig in, but not today), but be sure you have the variant that is intended for the IDE-only host(s), and not one that was intended for actual SCSI (33C93A on A3000), or the A4KT's IDE+NCR SCSI systems. It might need the effort to take apart an A600 3.1 or 3.x image, apply a patch the scsi.device driver, and insert it back in. The components of an A600 3.1/3.x ROM to me would seem like the bare minimum requirements, and maybe try removing only one of those two card/carddisk components mentioned, if space/dependency allows. I've tripped over discussions with some module docs while searching on REMUS, but not currently able to do much where I am now (daytime).

Test the driver you have built by pulling up HDToolBox and accessing a spare disk/media to see if it sees the media - don't rely on a boot-up as your verification of basic function. I've read of people that have had problems sometimes transferring prepped media from emulation environments - I don't have any emulation environment, but I suspect the layers sometimes mask out some information that needs to be on the actual media if a setting isn't just right.

Also, is the updated FFS to support >2GB partitions in the ROM? Or in the boot block? Are you counting on it suporting >2GB for boot? An older FFS probably wouldn't like >2GB partitioning.

Workbench: If not accounted for, there is a stub that gets added to the modules list to make the system pull workbench.library from libs:. Look up the module tinkering done on the A4000T images which is very tight on space with the larger scsi.device needed (IDE+NCR53C710).
 
Last edited:
Hi, thanks for the reply.

Let me try and clarify

Currently I have a 10gb HDD partitioned as a 500mb dh0 and the remaining 9gb as dh1. Both partitions are formatted with PFS3 filesystem.

I use classicwb and have loadmodule in the startup sequence to soft load a newer version of SCSI.device which means the Amiga starts to boot, loads the new scsi.device, soft reboots and finally boots into workbench. It all works fine however I'd like to try and stop the additional reboot.

This can be achieved by having the newer scsi.device in the kickstart rom instead of soft loading it hence my desire to burn a new rom.

I have burned the rom and using a workbench floppy I can use shell to check the scsi.device version and indeed the newer version is detected and loaded. I can also access the early boot menu and dh0 is listed as a bootable device however it does not boot. Every time I am dumped back at the kickstart screen :)
 
As clear as mud... ;-) At least as it applies to why it's not working - It logically seems to be right.

Is there something about the Kickstart image you have built (vs the soft kick image used) that might have a problem with loading the PFS3 FileSystem (or loading it from the source location?) at boot time? Drop a floppy with SysInfo on it and write down all the devices, libraries, and such, and compare. Maybe one is off or unaccounted for.

Any chance to validate off a small piece of IDE media and with basic FFS?
 
Update,

the newly burned rom when put into an A600 with 2.5" ide hdd boots into workbench fine!?

Why won't it autoboot on my A500+ or CDTV (using mkl's IDE+8MB board).....

Its very odd! lol

I'll try formatting a small hdd with FFS filesystem instead.
 
Last edited:
What SCSI controller are you using on your A500+?

The scsi.device found in the kickstart 3.1 ROM for the A2000, A500(+) and A600 is only for the A600 as the A2000 and A500 do not have inbuilt SCSI or IDE controllers. You can safely remove scsi.device from a custom kickstart in an A2000/A500 and there will be no difference.

SCSI drivers for the A2000/A500 are found in the ROM incorporated in whatever SCSI controller you have.
 
I am using one of mkl's 8mb+ide board which is based on the a600 ide. Standard 3.1 ROM boots fine, home made 3.1 ROM with updated scsi.device (again designed for a600) doesn't boot
 
I'm also using the mkl IDE68K on my A1000 (w/ROM adapter+switcher) and it's working fine with both 2.05 and 3.1 or 3.x. I pick all of my signals off the 86-pin edge - just solder in L header pins into the VIAs and run short Dupont jumper lines to the card. Easy disconnect if it's ever needed, too.
 
I did take the signals from the edge connector. As i mentioned above, it boots fine with any stock rom that supports hdd's 2.05 (37.300) and 2.05 (37.350), 3.1 (40.60) tested and boot fine. It will not boot with any custom rom i make however.

I have no idea why that is, its quite frustrating but i will continue use loadmodule to softload the new scsi.device for the time being.
 
Did you give the module version comparison I mentioned above a try? Maybe one is missing.

You can also try pushing one (or the least amount) of module(s) in at a time.

Also, there may be an order dependency of the modules in the ROM that is possibly getting messed up.
 
Back
Top Bottom