3.0.1 regressions

[David Ronis]

Hi Mark,

Have the problems with fomit-frame-pointer, optimization and ix86's
been fixed?  It doesn't seem so from gnats:

PR   Category	State	Class	Responsible	Synopsis

3754 c++  open  wrong-code    unassigned  
     g++ 3.0: failed exception i386 -O -fomit-frame-pointer

3710 c    open  wrong-code    unassigned  
     segfaulting code generated with -O2 -fomit-frame-pointer (kernel!)

3480 optimization  open      wrong-cod  unassigned  
     gcc-3.0, bad code generated with -O3 -fomit-frame-pointer on intel x86

3374 bootstrap    open      wrong-cod  unassigned  
     Bootstrap failure with -O -fomit-frame-pointer

3323 optimization  open      wrong-cod  unassigned  
     gcc -O2 -fomit-frame-pointer -fPIC references unitialized %ebp on x86

3224 optimization  open      wrong-cod  unassigned  
     -fomit-frame-pointer generates wrong code (g77)

2447 c++    open      sw-bug    unassigned  
     g++ don't handle execption when using with -fomit-frame-pointer

1523 optimization  analyzed    wrong-cod  unassigned  
     GCC 2.97 miscompiles with -fomit-frame-pointer

141  c++    analyzed    wrong-cod  unassigned  C++: Is this a bug with throw and -fomit-frame-pointer?


I encountered this bug some time ago in trying to build groff-1.17.2
(the resulting code SIGSEGVS).  I just rebuilt it using gcc version
3.0.1 20010811 (prerelease), with the following CFLAGS/CXXFLAGS at the
configure step:

CFLAGS:  -O3 -march=i686 -ffast-math -funroll-loops -fomit-frame-pointer
 -fforce-mem -fforce-addr -malign-jumps=3 -malign-loops=3
 -malign-functions=3 -mpreferred-stack-boundary=3

CXXFLAGS: $(CFLAGS) -fpermissive

groff still SIGSEGV's.  If I use exactly the same flags with gcc
version 2.95.3 20010315 (release), on the same box, it works fine.

David