Re: bootstrap failure while compiling gcc/tree.c (ICE)

[Brian Sidebotham <micros at valvers dot com>]

Brian Sidebotham wrote:
> Brian Sidebotham wrote:
> > ../../gcc/gcc/tree.c: In function "build_string":
> > ../../gcc/gcc/tree.c:1197: internal compiler error: in 
> > iterative_hash_expr, at tree.c:4189
> > Please submit a full bug report,
> > with preprocessed source if appropriate.
> >
> > I have placed the pre-processed file here:
> > http://www.valvers.com/gcc/arm-elf/tree.i (1.5Mb)
> >
> > The following configure line was used:
> >
> > ../gcc/configure --target=arm-elf --prefix=${installdir} --with-newlib 
> > --with-headers=${newlibdir}/src/newlib/libc/include 
> > --enable-languages=c,c++ --enable-interwork --enable-multilib
> >
> > codebase is svn revision 127732
> >
> > The failing command line is:
> >
> > gcc -c   -g -O2 -DIN_GCC -DCROSS_DIRECTORY_STRUCTURE  -W -Wall 
> > -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes 
> > -Wold-style-definition -Wmissing-format-attribute -fno-common   
> > -DHAVE_CONFIG_H -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. 
> > -I../../gcc/gcc/../include -I../../gcc/gcc/../libcpp/include 
> > -I../../gcc/gcc/../libdecnumber -I../../gcc/gcc/../libdecnumber/dpd 
> > -I../libdecnumber    ../../gcc/gcc/tree.c -o tree.o
> >
> > and lastly, the specs of the gcc compiler being used are:
> >
> > Using built-in specs.
> > Target: i486-linux-gnu
> > Configured with: ../src/configure -v 
> > --enable-languages=c,c++,java,f95,objc,ada,treelang --prefix=/usr 
> > --enable-shared --with-system-zlib --libexecdir=/usr/lib 
> > --without-included-gettext --enable-threads=posix --enable-nls 
> > --program-suffix=-4.0 --enable-__cxa_atexit --enable-clocale=gnu 
> > --enable-libstdcxx-debug --enable-java-awt=gtk-default 
> > --enable-gtk-cairo 
> > --with-java-home=/usr/lib/jvm/java-1.4.2-gcj-4.0-1.4.2.0/jre 
> > --enable-mpfr --disable-werror --with-tune=pentium4 
> > --enable-checking=release i486-linux-gnu
> > Thread model: posix
> > gcc version 4.0.3 (Ubuntu 4.0.3-1ubuntu5)
>
> These are the two lines that fail:
>
> tree.c:1197: memcpy (CONST_CAST (TREE_STRING_POINTER (s)), str, len);
>   memcpy (((__extension__(union {__typeof(((const char 
> *)(STRING_CST_CHECK (s)->string.str)))_q; void *_v;})(((const char 
> *)(STRING_CST_CHECK (s)->string.str))))._v), str, len);
>
> tree.c:1198: ((char *) CONST_CAST (TREE_STRING_POINTER (s)))[len] = '\0';
>   ((char *) ((__extension__(union {__typeof(((const char 
> *)(STRING_CST_CHECK (s)->string.str)))_q; void *_v;})(((const char 
> *)(STRING_CST_CHECK (s)->string.str))))._v))[len] = '\0';
>
> Removing the CONST_CAST delcarlations allows compilation of gcc to 
> complete with a warning (as expected)
>
> ../../gcc/gcc/tree.c: In function 'build_string':
> ../../gcc/gcc/tree.c:1197: warning: passing argument 1 of âmemcpyâ 
> discards qualifiers from pointer target type
This ICE is caused by the following patch:

2007-08-10  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>

	* system.h (CONST_CAST): New.
	* c-decl.c (c_make_fname_decl): Use it.
	* c-lex.c (cb_ident, lex_string): Likewise.
	* c-typeck.c (free_all_tagged_tu_seen_up_to): Likewise.
	* gcc.c (set_spec, read_specs, for_each_path, execute, do_spec_1,
	give_switch, set_multilib_dir): Likewise.
	* gengtype-parse.c (string_seq, typedef_name): Likewise.
	* passes.c (execute_one_pass): Likewise.
	* prefix.c (update_path): Likewise.
	* pretty-print.c (pp_base_destroy_prefix): Likewise.
	* tree.c (build_string): Likewise.

Reverting this patch lets gcc build again.

Unfortunately, CONST_CAST is a mystery to me (I don't understand it!) so there 
is no hope of me changing it to compile okay AND be right!

Best Regards,

Brian Sidebotham.