[comp.sys.hp] Problems with HP9000/500 assembler

robbin@resrch.MoliEnergy.BC.CA (Robbin W. Johnson) (11/27/88)

While bring GNU CC up on our 500 I have encountered a problem with the
assembler.  First off, we are running HPUX 5.21 on a 2 CPU 550.

While compiling one of the files auto-magically generated during the
make'ing of cc1, the compiler fatally errors with the following:

  cc -g -DCROSSATARI=1 -DMINIX -c insn-extract.c
  as: /tmp/ctm3CAAa11176: fatal error in line 11312, proctable overflow

(NOTE: the two -D's in the command line do not apply to this source
file.  They come from the make command.)

From the format of the error message, I conclude that the problem is
with the assembler, and not (directly) due to the code being compiled. 
Obviously, there are too many proceedures being declared in the source
file (approx. 267.)  An array of pointer-to-function's is initialized
with 266 of these.

So the big question is - How can I increase the size of "proctable"?
There is nothing in the (meager) documentation that even hints that this
is possible.  I got around the problem by splitting the .c file into two
pieces and then manually compiling and "ld -r"'ing these two into an
insn-extract.o and that worked; but this is not a reasonable solution.

HP?  Anyone?

-- 
Robbin Johnson					...!ubc-cs!van-bc!resrch!robbin
robbin@MoliEnergy.BC.CA				...!uunet!/

dave@pmafire.UUCP (Dave Remien) (11/28/88)

In article <601@resrch.MoliEnergy.BC.CA> robbin@resrch.UUCP (Robbin W. Johnson) writes:
<>While bring GNU CC up on our 500 I have encountered a problem with the
<>assembler.  First off, we are running HPUX 5.21 on a 2 CPU 550.
<>
<>While compiling one of the files auto-magically generated during the
<>make'ing of cc1, the compiler fatally errors with the following:
<>
<>  cc -g -DCROSSATARI=1 -DMINIX -c insn-extract.c
<>  as: /tmp/ctm3CAAa11176: fatal error in line 11312, proctable overflow

Those of us with 500 series out in netland would love to hear more about
your porting of GNU CC to the 500 series. The way HP explained it to me
was that the assembler is unsupported; then they later came out with
"sort-of" documentation for it (when they put the disassembler on the SE
tapes). Given the architecture of the 550; I'd have thought that a GNU
cc port would be pretty difficult (read PAINFUL :-)). Don't get me
wrong; I *like* my 550s; they're the most reliable machines I've ever
used. One I've got has been up over a year non-stop now, and only the
ptydaemon gives me a little indigestion :-).  Please tell us more, if
you get it working.

>-- 
>Robbin Johnson					...!ubc-cs!van-bc!resrch!robbin
>robbin@MoliEnergy.BC.CA				...!uunet!/


-- 
Dave Remien - WINCO Computer Engineering Group (only somewhat confused, now)
208-526-3523 Paths: ...!bigtex!pmafire!dave | ...!ucdavis!egg-id!pma386!dave