Re: bug in a29k-coff

Re: bug in a29k-coff

[Richard Kenner]

        store 0,0,AI0,gr116

    I don't know whether that is valid assembly language or not for the
    a29k so I can't blame it on gcc or binutils.

This is not valid.  AI0 is a "fake" register used to denote incoming
arguments in the variable-size register frame (I don't remember the details:
I wrote this stuff nearly 15 years ago).

bug in a29k-coff

[Joel Sherrill]

CC:      gcc 2.95.2
BINUTILS: binutils 2.10
TARGET:   a29k-rtems (basically a29k-coff)

I don't know if anyone cares about this or not.  We have an old
a29k port of RTEMS and I was (perhaps foolishly) trying to build
a29k tools to see how badly it was bit-rotted.  When building C++,
it blew up on this gcc invocation:

/usr1/rtems/work-tools/tools/b-a29k-gcc/gcc/xgcc \
  -B/usr1/rtems/work-tools/tools/b-a29k-gcc/gcc/ \
  -B/opt/a29k-rtems/a29k-rtems/bin/ \
  -I/opt/a29k-rtems/a29k-rtems/include -O2 \
  -I../../gcc-2.95.2/gcc/../newlib/libc/sys/rtems/include
-DCROSS_COMPILE \
  -DIN_GCC    -O2 -I./include   -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 \
  -D__GCC_FLOAT_NOT_NEEDED -Dinhibit_libc -O2 -I. \
  -I../../gcc-2.95.2/gcc -I../../gcc-2.95.2/gcc/config \
  -I../../gcc-2.95.2/gcc/../include -c \
  ../../gcc-2.95.2/gcc/cp/new1.cc -DL_op_new

Tracking it down, I cut new1.s down to this and it still failed.

==================                                                      
        .text
        store 0,0,AI0,gr116
==================
sh$ a29k-rtems-as new1.s
new1.s: Assembler messages:
new1.s:2: Error: Invalid operands
==================

I don't know whether that is valid assembly language or not for the
a29k so I can't blame it on gcc or binutils.

Thanks.
                                                

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel@OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985