🚀 go-pugleaf

In article <9011151743.AA20534@uunet.uu.net> mmodrall%greystone@UUNET.UU.NET writes:
>Hi...
>	I'm trying to get gcc put together on a concurrent machine 
>with the vanilla distribution from you guys and a pocketful of changes
>from concurrent.  The problem is that when i try to build it with the
>concurrent c compiler, using the .md file provided by them, the make
>procedure gets a syntax error reading masscomp.md and core dumps...

I got gcc and g++ up on a MASSCOMP 5600 machine some time ago. I
however do not have access to the machine or the company which owns
it any longer. So I can not answer detailed questions but I will
tell you the problem I found which sound similar to yours.

The RTU version I had was 4.x and 5.x(I think?) There was a bug in
the function 'setbuffer'. This function is a BSD function and
available in the 'bsd universe'. The function is suppose to allow
one to point a _iob buffer to a user defined region of memory. Well
the function I found after stepping through a whole lot of
dis-assembly code was a simple 'rts'. What effect this had was that
the writers of the tools which processed input from the *.md files
would use a trick to allocate a buffer, collect some stuff then put
that buffer into an 'iob' then use the standard fio functions to
process the buffer. Sort of a push stream then pop stream type of
idea. Well if setbuffer is broken then the trick fails with a memory
error.
-- 

John Clark
jclark@ucsd.edu