Re: Signal 11 bug in egcs-1.1.1

[Jeffrey A Law <law at cygnus dot com>]

  In message <199812221940.OAA10644@onsager.chem.mcgill.ca>you write:
  >          gcc version egcs-2.91.60 19981201 (egcs-1.1.1 release)
  > 
  > The build was configured with the default options on both machines.
  > The compilation fails when trying to build something called compobj.o (the 
  > preprocessed file is appended below).  The
  > command (with -save-temps -v added by me) used is:
[ ... ]


  >  /usr/local/lib/gcc-lib/i586-pc-linux-gnulibc1/egcs-2.91.60/cc1 compobj.i -  > quiet 
  > -dumpbase compobj.c -g -O2 -Wall -version -o compobj.s
  > GNU C version egcs-2.91.60 19981201 (egcs-1.1.1 release) 
  > (i586-pc-linux-gnulibc1) compiled by GNU C version egcs-2.91.60 19981201 
  > (egcs-1.1.1 release).
  > gcc: Internal compiler error: program cc1 got fatal signal 11
[ ... ]

  > I built an unmodified egcs, and although I no longer have the source files 
  > and cannot tell you what configure options were used,  I did a full
  > bootstrap install, including the comparison of stage2 and stage3 compilers.
  > Most likely, I upped the optimization options slightly, but nothing else.
I've been unable to reproduce this problem.  The file compiles fine regardless
of what optimization level I use.

The only plausible possibility I can think of is that when you raised the
optimization level for building the compiler that somehow the compiler mis-
compiled itself and triggers the bug you sent.

One thing you can do is try and debug this yourself a little.

Run gdb on "cc1" and start it with "-O2 -g -Wall compobj.i" and see what
happens.  cc1 will be found in
/usr/local/lib/gcc-lib/i586-pc-linux-gnulibc1/egcs-2.91.60/cc1

jeff