EEPROM on A1200?

[Doobrey]

@Motorollin
 I normally remove workbench.library from ROM, and leave the 3.9 wb in libs:
 to make more space, also if that still doesn't leave enough room you could leave the 3.1 bootmenu in ROM as the 3.9 one is 13kb bigger.

 Just done a little test,
 Updated all residents from 3.9bb2 and no workbench leaves 37kb free
 Updated all except bootmenu and no workbench leaves about 50kb free.

Using EEPROMS should be fine, I've run the 'new' ROMS from flash on my A1200 and in E-UAE without problems.

A word of warning though, several people have reported big problems when using 3.9bb2's execv45, but have said that it works fine when they swapped it for Piru's Exec44...but Sod's law comes into play cos I can't reproduce the problem here  :roll:

[Doobrey]

_yak_ wrote:
Uhm, it's probably a dump question. I don't know where is the ROM mapped in the memory but is it possible to use a bigger EEPROM chip? :roll:

Yup :-)
 That's how I started out experimenting building my own ROMS, I just built an image containing all the relevant parts of the rom update into a 512kb rom, and added it to UAE as an extended ROM.
In the A1200/4000/CD32 (dunno about earlier machines) CBM left 512kb for the 2nd half of a 1mb ROM at $E00000-$E7FFFF, with the normal 512kb at $f80000-$ffffff as the 1st half...I guess any future upgrades wouldn't have fitted a 512k ROM, as someone else said the A1200 40.68 is already a tight fit
 
 The only trouble is that it's not a contiguous 1mb area, so you'll need to build it as 2 separate 512k images (1 at $e00000 the other at $f80000) and then join them together to make the single 1mb ROM :hammer:

[Doobrey]

motorollin wrote:
@Doobrey
Are there any other libraries I can safely leave out? I'll be using my Blizzard SCSI interface, so can I remove scsi.device? Also, I don't use the PCMCIA slot. Can I remove the card device and resource?

In the remus.guide there's a list of what libraries/devices etc depend on each other, there isn't really that much else that can safely be removed.
 Removing scsi.device,card.resource and carddisk.resource works fine on UAE, but I haven't tried it on my A1200 cos I need them for the HD and network card.
 I've also booted fine without the v40 icon.library, leaving the v45 one in libs: , but again others who tried it couldn't boot ! ..ahh, the joys of no 2 Amiga setups being the same :lol:

[Doobrey]

motorollin wrote:
I think I'll set up UAE and see what I can remove. How did you get your KickFlash device to work with 1MB?

If you're running UAE with host directories mounted as amiga drives, use $e00000 as the base address when building an extended ROM as UAE uses a filesystem thats hardcoded into the $f0000 space..plus it clashes with a 'magic window' for the Amiga side to communicate with UAE.
Also use the CD32 version of exec44/45 in the 'normal' rom ,as the other versions don't search for extended roms in the $e00000 space.
If the extended rom doesn't contain exec, you need to put the example romheader (in remus/others/romheader) as the first item in the build list to get the normal rom booting first.

I guess all this stuff should get added to the docs, at least I'm honest and admit they're crap :-D

As for the kickflash, I did it similar to the way RedSkull did it in his kf500, except I've mapped the $f8 part to the lower half and the $e0 part to the upper half  because I want to use the flash's 32kb(16kb in each chip) parameter block as non-volatile memory  :sealed:

[Doobrey]

_yak_ wrote:
 Does someone there actualy know how the ROM modules are linked together?

They're not linked, simply put one after another...with a couple of exceptions like input/gameport/keyboard ,exec/alerthook etc compiled as one as they share some common routines to save space.

Each resident module contains a resident structure that identifies it( see the Amiga NDK  includes/exec/resident.h)
Basically exec searches through the memory areas it knows could contain a ROM for the RTC_MATCHWORD value ($4afc), if it finds it and the next longword is the address of the $4afc then it knows it's found a resident and then adds it to the resident list ( only if a greater version hasn't already been found etc.)

[Doobrey]

@Motorollin,
  If it's lots of wgets for addresses between A80000 and B80000 it's OK, it's just looking for residents in an area of memory reserved for another 1MB ROM on the CD32. I'll do a simple patch to stop it.

[Doobrey]

motorollin wrote:
There was also an illegal instruction,

Hmm, that doesn't sound good.. was the CPU set to '020 in EUAE ?
If it still happens, could you mail me the .rpf file (if you saved your build list) so I can build it and check it out?

 and lost of lines saying "Byte put to custom register DFF110 PC=00FDD8AA" all with different addresses. Is this ok? Will any of these errors cause a problem if the ROM is used on a real Amiga?

Depends,how many is 'lots'?
It looks like that's trackdisk.device just checking for disks in DF0-3..or was it disk.resource..gah it's late and my brains fried  :crazy: ...anyway, some register accesses in the uae.log are normal..even the bug in exec's chipmem size routine that reads from $200000.
If you boot with your A1200's 40.68 ROM and look at the log you can get an idea of what normally happens.