SUMMARY: gcc mips-tfile compile error

• Previous message: North, Walter: "v5.1B PK4 dupatch behavior change"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Date: Sat, 17 Jul 2004 23:41:33 -0500
To: "tru64-unix-managers@ornl. gov ((E-mail))" <tru64-unix-managers@ornl.gov>

Original question below.

The problem has to do with an upgrade done to the assembler, as,
during a 5.1A upgrade or patch, where the newer as had a change
that confused the older 2.9x gcc compiler.

A workaround is to invoke the assembler with an "-oldas" option.
This can be done more globally with a script put in, say, /sbin that
contains:
    #!/bin/sh
    /usr/bin/as -oldas $*
Then before doing a build that requires this, make sure your PATH
is set to find this /sbin/as script before it can find /usr/bin/as.

There is a suggestion that an upgrade to gcc 3.2 is the better fix, but
I couldn't find a setld package for it, and the workaround satisfied
our current needs. So I didn't bother building gcc 3.2 from source.
Maybe later.

Thanks to Joe Mario for the basic answer and the inspiration to
google out the rest.

-Neil

Original Question:
> Machine: Alpha ES45 4x1GHz
> OS: True-64 5.1A, PK 6
> gcc version: 2.9-gnupro-99r1
> f90 version: Compaq Fortran V5.5-1877
> Compaq Fortran Compiler V5.5-1877-48BBF
>
> One of our users is trying to build the WRF atmopheric meso-scale
> model. The compilers specified are gcc, cpp, and f90.
>
> During running of the perl build script, the following error occurs:
> mips-tfile:Symbol table is not last (symbol table ends at 4776, .o
> ends at 4808
>
> Any clues as to what is the problem here?
> Is this an issue with the gcc compiler and/or its version?

--
Neil R. Smith				neils@ariel.met.tamu.edu
Comp.Sys.Mngr.			(979)845-6272
Dept. Atmospheric Sciences/Texas A&M University

• Previous message: North, Walter: "v5.1B PK4 dupatch behavior change"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]