From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jeffrey A Law <law@cygnus.com>
To: "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu>
Cc: egcs@cygnus.com
Subject: Re: egcs-971127 hppa1.1-hp-hpux9.01 results 
Date: Sun, 30 Nov 1997 20:14:00 -0000
Message-id: <9272.880943247@hurl.cygnus.com>
References: <199711291709.MAA28369@caip.rutgers.edu>
X-SW-Source: 1997-11/msg00990.html

  In message < 199711291709.MAA28369@caip.rutgers.edu >you write:
  > PS:     I think at least some (many?) of the "make check" errors for
  > libio, libstdc++ and g++ are due to the testsuite passing -g and
  > getting a warning "cc1plus: warning: -g is only supported when using
  > GAS on this processor".
And here's the appropriate magic to add to target.exp to filter out
those warnings too.

    # When using the HP assembler, -g isn't supported.
    regsub -all "(^|\n)(cc1: warning: -g is only supported when using GAS on this processor\[^\n\]*\ncc1: warning:\[^\n\]*\n?)+" $text "\\1" text
    regsub -all "(^|\n)(cc1plus: warning: -g is only supported when using GAS on this processor\[^\n\]*\ncc1plus: warning:\[^\n\]*\n?)+" $text "\\1" text


However, to get accurate g++ results (or even a g++ compiler that
works) you're going to need to add -fexceptions to the command
line because of the dumb linker on hpux9 core dumping on having a
local label in a .word pseudo-op.

At this point I'm not sure what else can be done to make the hpux9
toolchain happy when using the HP assembler.


jeff