Tons of testsuite failures on mips-sgi-irix6.2/ with egcs-1.1

[Jeffrey A Law]

  In message < 199809042048.WAA12202@achibm5.chemie.uni-karlsruhe.de >you write:
  > 
  > I have tried to run the testsuite on mips-sgi-irix6.2, and I get tons
  > of failures. I waded through the logs, and most (if not all) of the
  > unusual failures are of the form
  > 
  > Executing on host: /inge/georg/egcs-rel/gcc/testsuite/../xgcc -B/inge/georg
  > /egcs-rel/gcc/testsuite/../ /work/georg/egcs-rel/gcc/testsuite/g++.old-deja
  > /g++.pt/memclass6.C  -ansi -pedantic-errors  -I/work/georg/egcs-rel/libio -
  > I/inge/georg/egcs-rel/mips-sgi-irix6.2/libio -I/work/georg/egcs-rel/libstdc
  > ++ -I/work/georg/egcs-rel/libstdc++/stl -I/inge/georg/egcs-rel/gcc/testsuit
  > e/../include -lstdc++  -L/inge/georg/egcs-rel/mips-sgi-irix6.2//libstdc++ -
  > L/inge/georg/egcs-rel/mips-sgi-irix6.2//libiberty   -lm  -o /inge/georg/egc
  > s-rel/gcc/testsuite/g++-pt-memclass6-C   
  > output is:
  > ld32: WARNING 84: /inge/georg/egcs-rel/mips-sgi-irix6.2//libstdc++/libstdc+
  > +.a is not used for resolving any symbol.
  > ld32: WARNING 84: /usr/lib32/libm.so is not used for resolving any symbol.
  > FAIL: g++.pt/memclass6.C (test for excess errors)
  > ld32: WARNING 84: /inge/georg/egcs-rel/mips-sgi-irix6.2//libstdc++/libstdc+
  > +.a is not used for resolving any symbol.
  > ld32: WARNING 84: /usr/lib32/libm.so is not used for resolving any symbol.
  > PASS: g++.pt/memclass6.C  Execution test
  > 
  > So actually it is just a silly linker warning. Does anybody know how
  > to tweak the spec file to switch this off (I have tried a -woff 84
  > next to the existing -woff 131, but that didn't help)?
Hmmm.  There should be a procedure in dejagnu which prunes out such
warnings before passing the output to the rest of dejagnu.
prune_system_crud or something like that.  

jeff