Deallocating RAM from the Memory Pool

[Zetr0]

Hello there my forum chummies,

I have a question and I am sure that there is a solution as I used this atleast 14/15 years ago.

is there a way I can "software" deallocate memory that has been auto configured.

for example.

say I have an 8MB RTC memory card configured with 8MB of ram, can I call a small program to de-allocate all the ram from a given starting address ?

the idea is to be able to have a startup that checks the mouse button press to de-alloc 3.5MB from the Auto config memory thus allowing PCMCIA activity to go on without issues. ?

any ideas?

[Ed Cruse]

Originally Posted by Zetr0

Hello there my forum chummies,

I have a question and I am sure that there is a solution as I used this atleast 14/15 years ago.

is there a way I can "software" deallocate memory that has been auto configured.

for example.

say I have an 8MB RTC memory card configured with 8MB of ram, can I call a small program to de-allocate all the ram from a given starting address ?

the idea is to be able to have a startup that checks the mouse button press to de-alloc 3.5MB from the Auto config memory thus allowing PCMCIA activity to go on without issues. ?

any ideas?

If there isn't already a program that can do this, then I suppose a program could be written to minipulate the available-ram-list and make the ram of interest unavailable. Would that cause problems with the PCMCIA activity because to the system the ram wouldn't be available so it couldn't be allocated? If the PCMCIA just simply uses ram without allocating it then it seems like the above solution would work. My best guess anyway.

[TheCorfiot]

@Zeets
I've a feeling Toni Wilen may know the answer to this one my friend.

TC

[Toni Wilen]

I am quite sure above "fix" does not work.

Problem is that fast RAM (if >4M) and PCMCIA physical address space (Gayle) conflicts. It is not possible to have two different chips responding to same address, it most likely result in RAM errors or system crashing randomly and/or failing to boot.

It is exactly same situation if you have both fast ram (>4M) and PCMCIA SRAM card inserted (without disabling PCMCIA slot). SRAM and Fast RAM both would exist in same address space.

Perhaps it is possible to "disconnect" Gayle's PCMCIA common area address space (=hardware hack) because it isn't really needed when using CF card in PCMCIA adapter or PCMCIA network (SRAM cards need it but if you have SRAM card you won't need fast RAM..)

IO area is outside of fast RAM address space.

EDIT: if you want to enable PCMCIA even if you have >4M RAM, just clear bit 0 at 0xda8000 and see what happens.. (+insert card because slot is also disabled when it is empty)

[rkauer]

So a simple command to set bit 0 to 0 on 0xDA8000$ will stop Gayle from acting up?

Just a matter of booting the >4Mb Amiga, insert the card and then issue the command?

That's a precious information!

[TheCorfiot]

I told you, Toni knows everything Thanks Toni

TC

[Toni Wilen]

Originally Posted by rkauer

So a simple command to set bit 0 to 0 on 0xDA8000$ will stop Gayle from acting up?

Just a matter of booting the >4Mb Amiga, insert the card and then issue the command?

That's a precious information!

It re-enables PCMCIA port if KS has disabled it (if >4M RAM was detected)

I have no idea if it only causes instant crash

[ppill]

@Toni

How would you do it using POKE?

POKE	W=WORD/S,L=LONG/S,LIB=LIBRARY/K,DEV=DEVICE/K,RES=RESOURCE/K,AT/S
           C=CUSTOM/S,CIAA/S,CIAB/S,H=HEX/S,BIN=BINARY/S,STRING/S,ADDRESS/A

[ppill]

When I use:

peek 0xDA8000

I get 129 (decimal)

after

poke 0xDA8000 0

i get 200

is that correct?!

btw the CF0: fails to mount.

A1200+8megs (Elbox)

[Toni Wilen]

Originally Posted by ppill

When I use:

peek 0xDA8000

I get 129 (decimal)

after

poke 0xDA8000 0

i get 200

is that correct?!

btw the CF0: fails to mount.

A1200+8megs (Elbox)

Looks ok, other bits are status bits, for example bit 6 means card inserted. This means slot is now enabled.

I guess the slot isn't initialized properly or carddisk.device (or some other device driver) isn't loaded. It is also possible the device driver checks ram size and refuses to initialize without patching it.

[ppill]

I guess it was too good to be true...

[Toni Wilen]

At least system didn't become unstable, which means there is still hope if someone is interested in device driver patching

[rkauer]

Don't you need to read the other bits too before poking? Only the bit 0 have to be modified.

[Toni Wilen]

Originally Posted by rkauer

Don't you need to read the other bits too before poking? Only the bit 0 have to be modified.

Clearing other bits is safe, it does nothing.

[rkauer]

So just making it read an even number make it act correct, them? Interesting, really.

[Toni Wilen]

Find the gayle documention (gayle.pdf), it should be attached to some post here

[prowler]

Originally Posted by Toni Wilen

Find the gayle documention (gayle.pdf), it should be attached to some post here

http://eab.abime.net/showthread.php?t=35282

[cosmicfrog]

now this is a grovey idea if some1 can make it work
being able to have 4meg of fast and a working pcmia port without having to swap/configure the 8meg to 4, now that would be nice, gota love sofware solutions to hardware problems.

[Doobrey]

Originally Posted by Toni Wilen

It re-enables PCMCIA port if KS has disabled it (if >4M RAM was detected)

You'd also have to try to re-initialise card.resource too (It doesn't build it's library and add itself to exec's resource list if it finds memory at $00600100), since most of the PCMCIA drivers such as compactflash.device will fail if card.resource can't be opened.

[Toni Wilen]

Originally Posted by Doobrey

You'd also have to try to re-initialise card.resource too (It doesn't build it's library and add itself to exec's resource list if it finds memory at $00600100), since most of the PCMCIA drivers such as compactflash.device will fail if card.resource can't be opened.

Oops, I forgot it was card.resource, not the device driver (which is only for SRAM cards I think) that needs patching.

I can do the patch (should be really simple) but is there program that can load on-disk resources? (I am too lazy to write one..)