Re: PING edge-vector-branch merge

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Sep 30, 2004, at 3:00 AM, Eric Botcazou wrote:

> > http://gcc.gnu.org/ml/gcc-patches/2004-09/msg01903.html
> >
> > This patch is a first step to resolving some compile time
> > regressions.  The first is looking up PHI arguments in
> > constant time, the second is edge redirection out of basic
> > blocks ending in a SWITCH_EXPR.
>
> It breaks bootstrap with non GNU compilers, at least with Sun CC.  You 
> can
> reproduce the failure by bootstrapping with 'gcc 
> -fkeep-inline-functions':
>
> gcc -m64 -fkeep-inline-functions   -g -DENABLE_CHECKING
> -DENABLE_ASSERT_CHECKING -DIN_GCC   -W -Wall -Wwrite-strings
> -Wstrict-prototypes -Wmissing-prototypes  -fno-common   -DHAVE_CONFIG_H
> -DGENERATOR_FILE  -o build/genflags \
>         build/genflags.o build/rtl.o build/read-rtl.o build/ggc-none.o
> build/min-insn-modes.o build/gensupport.o build/insn-conditions.o
> build/print-rtl.o \
>         build/errors.o 
> ../build-sparc64-sun-solaris2.9/libiberty/libiberty.a
> Undefined                       first referenced
>  symbol                             in file
> vec_gc_p_reserve                    build/insn-conditions.o
> vec_gc_free                         build/insn-conditions.o
> bitmap_zero_bits                    build/insn-conditions.o
> ld: fatal: Symbol referencing errors. No output written to 
> build/genflags
> collect2: ld returned 1 exit status
> gmake[2]: *** [build/genflags] Error 1

bitmap_zero_bits was not caused the edge branch merge but Zdenek's
patch for bitmap iterator.  In fact I assume all of those problems
were caused by his patch rather than the edge-vector branch merge
as basic_block.h should not be included in insn-conditions.o at all.

Thanks,
Andrew Pinski