Stuart Krivis wrote:

This still seems an over simplification.. There are two seperate firmware issues, the "console" and the "palcode". The "console" is the program loader and high level machine management for loading things like unix, NT, etc. In the console code is embedded a copy of "palcode". Palcode is both a blessing and a curse. It's the low level pseudo-microcode that provides a personality for the execution environment. It is meant to handle differences in different hardware and provide a common interface to the given OS. As such it controls things like interrupt management and virtual memory management of page tables or whatever. It is the palcode that effectively determines the address space layout and things like whether is supposed to use page tables (32 bit or 64 bit!), etc etc.

The problem is that the SRM console contains a copy of the OSF and VMS palcode. ARC and AlphaBIOS contain the NT palcode (which is limited to 32 bit address spaces. has a very different interrupt management system etc). The NT palcode basically tries to make the Alpha look as much like a PC as it can get away with.

And there's the kicker.. palcode is very machine specific and has to be adapted for each new hardware type. The Alpha "clones" coming from people like Samsung (I'm told) simply don't have a SRM console with OSF palcode available.

MILO is an interesting animal. As I understand it, it is a pseudo-console in that it can be "run" from ARC or AlphaBIOS and it then takes over the machine and then switches to it's own hacked OSF palcode derivative to restore the 64 bit environment or something like that. It can then run Unix-like OS's that were intended to run under a SRM environment and real OSF palcode with a lot less pain than trying to run in a sanitized-to-avoid-confusing-NT environment.

Or something along those lines anyway. :-)

Cheers, -Peter

To Unsubscribe: send mail to majo...@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message