I have huge problems with my drives at the moment. At first I thought it must be because of a blow on the machine about 1/2 year ago.
BUT now the ramdisk is starting to fail. I cannot read simple executables from it.
SO I remembered, that I updgraded dos.library and ramlib.kmod some weeks ago.
Does anyone of the betatesters know, if there are issues with versions of those two entities?
I think it is a better bet than both my usb stick and harddrive going bust at the same time. The machine boots fine 19 out of 20 times, and the last time I assume it is an issue with my beta elf.library.
@elfpipe ramlib.kmod has nothing to do with the ram disk, that's for loading disk based libraries and devices for Exec.
ram-handler.kmod is the filesystem that handles the RAM: volume.
dos.library handles calls to all filesystems and handlers for application requests, if there was anything wrong with it, you'd soon be in deep trouble everywhere, it would be quite obvious, you probably wouldn't even be able to boot up.
elf.library is responsible for loading PPC elf format file (executables) into memory and that is likely an issue if you are using beta releases, it could be executable specific if one was put together differently.
Revert to the last "official" elf library and other components and do a cold boot to see if your issues fo away.
If you provide version numbers, we may be able to help you further, but I would most strongly suggest you don't play mix-n-match with early releases of kickstart components, as they may be release dependant on each other and some may not work at all.
@elfpipe RAM: is mounted by dos.library, a mountlist or DOSDriver isn't required for it.
DEVS:Mountlist was the old way to mount other dos devices, a single file with all dos device entries. It was replaced (but is still supported) in AmigaOS version 2.1 by using separate files in DEVS:DOSDrivers for each dos device instead.
ramdrive.device is a floppy/HD like RAM based device, usual DOSDriver name RAD: or RAD0:, unlike ramhandler/RAM: it has a fixed size. Additionally the contents survived a fast reboot (CTRL-Amiga-Amiga), but at least with current Radeon gfx drivers that doesn't work any more because you always have to do a cold reboot (CTRL-Alt-Alt) instead.
@elfpipe ram-handler.kmod 54.26 requires a reasonably recent V54 dos.library for the best results, it also requires V54.1+ of Exec if you want to have access to the ExtMem features.
As Joerg indicated, the dos.library mounts all the RTF_AFTERDOS modules, so if RAM: is not mounting, it can only be that the module is not being loaded from the kickstart directory (check your kicklayout), or, there is something failing during initialisation of the ram-handler itself.
Luckily, all failure points in ram-handler will print a message to the serial port using IExec->DebugPrintF() in the form of like; "[RAM] Cannot allocate blabla...." that sort of format, so check your serial debug by typing "dumpdebugbuffer" in a shell.
Also, if you set the exec debuglevel to 5 (or higher), you will always get a message in the serial debug; "[RAM] Handler has started successfully....".
Don't know if this may make any difference but maybe your Kickstart/ram-handler.kmod may not have the 'read' attributes set? Another idea: may it be possible that you have several kickstart and you somehow messed up with them? Ending up with mixing incompatible versions? On this last one please add asked by Colin provide version numbers of the listed kmod.
try to update to the very latest. i also had a problem with elf.library. that was gone before i could check and write a bugreport with the latest updates.
we're not discussing beta software here now? are we?
Those are the latest beta release versions, they work here fine. Also, you didn't mention the elf.library version. And, you don't need a serial hookup, just type "dumpdebugbuffer" in a shell. You can also flush the debug buffer with the "Clear" argument.