Kaffe vs. Alpha-Linux

[Felix von Leitner]

Thus spake Godmar Back (gback at cs.utah.edu):
> these problems have been reported before by Petter Reinholdtsen on July 26.
> The alpha JIT port is currently listed as one of the platforms that
> have "fallen behind recent changes"
> Rick Geer provided a sysdepCallMethod macro with which the interpreter
> runs (if I interpret his mail from July 15 right).  If you bring the 
> jit up-to-date, I encourage you to send the fixes.

Actually, my time is quite limited getting the kernel, libc, egcs, X and
the Qt+KDE up and running (which only partially succeeded), so I'd
rather not like to add one more project, but if you tell me what I
should do, maybe even send me a small test program that uses the macros
I should write and works fine on Intel with the Intel version, so that I
can understand what I'm supposed to do, I might have a look and see what
I can do about it.

> Basically, you'll have to determine SP_OFFSET for the alpha;

SP is the stack pointer?
What does that offset signify?
How much bytes a standard subroutine call puts on the stack?

> you'll have to find out whether its frame pointer needs to be
> manipulated too, and if so, define FP_OFFSET.  You can either look in
> <setjmp.h>, or I can send you a test program that finds out what it
> is.

Please do so!

> The macro FRAMEOKAY() is only used by the version that uses jit AND
> unix-internal.  For unix-jthreads, I felt that putting this macro in jit.h
> would violate encapsulation, so I replaced it with jthread_on_current_stack.
> It is only used from within the threading system, namely in TnextFrame.

Is there some FAQ or HOWTO about this?
What are jthreads and why would anyone want to use them in lieu of
pthreads?

> The macro NEXTFRAME is obsolete, no longer used, should be removed.

OK.

> The other macros in jit.h, as far as I can see, are still needed and you must 
> implement them properly for the alpha.

Mhh.
If you give me a small program that uses them, I will try my best.

Felix