Re: Another bootstrap failure on vax: label referred to in addr_vec deleted when basic block is moved to the fallthru position a

[Jan Hubicka <jh at suse dot cz>]

> 
> Enclosed is the patch that I worked out to fix the problem.  However, in

Thanks. It looks correct. Can you please post it to the mailing list?

> testing, I hit new problems:
> 
> 1) vax:
> 
> stage1/xgcc -Bstage1/ -B/usr/local/vax-dec-ultrix4.3/bin/ -c  -DIN_GCC    -g -O2 -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long  -DHAVE_CONFIG_H    -I. -I. -I../../gcc -I../../gcc/. -I../../gcc/config -I../../gcc/../include ../../gcc/cpplex.c -o cpplex.o
> ../../gcc/cpplex.c: In function `cpp_parse_escape':
> ../../gcc/cpplex.c:1980: Internal compiler error in fixup_var_refs_1, at function.c:1932

This happends when validate_replace_rtx fails to replace virtual register
by another register.  This is possible when the pattern is not in canonical
form.  Just breakpoint there, take a look what insn is offending and
fix the template to match.
> 
> 2) pa:
> 
> gcc -o c++filt  -DIN_GCC    -g  -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long  -DHAVE_CONFIG_H  \
>   cxxmain.o underscore.o obstack.o  -lintl     ../libiberty/libiberty.a
> cd ../../gcc && makeinfo   -I doc -I doc/include -o doc/cpp.info doc/cpp.texi
> cd ../../gcc && makeinfo   -I doc -I doc/include -o doc/gcc.info doc/gcc.texi
> doc/invoke.texi:3417: No matching `@end table'.
> makeinfo: Removing output file `doc/gcc.info' due to errors; use --force to preserve.

This is fixed already.
> 
> I don't think I will get any more time this weekend to work on these problems.

I will find time to build vax crosscompiler and track down the insn pattern
at tuesday, since then I am also busy by non-gcc stuff.

Honza
> 
> Dave
> -- 
> J. David Anglin                                  dave.anglin@nrc.ca
> National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)
> 
> 2001-09-28  John David Anglin  <dave@hiauly1.hia.nrc.ca>
> 
> 	* cfgcleanup.c (label_mentioned_p): New function.
> 	(try_optimize_cfg): Use it.
> 
> --- cfgcleanup.c.orig	Fri Sep 21 18:29:23 2001
> +++ cfgcleanup.c	Fri Sep 28 16:09:38 2001
> @@ -52,6 +52,7 @@
>  						 rtx *, rtx *));
>  
>  static bool delete_unreachable_blocks	PARAMS ((void));
> +static bool label_mentioned_p		PARAMS ((rtx, rtx));
>  static bool tail_recursion_label_p	PARAMS ((rtx));
>  static void merge_blocks_move_predecessor_nojumps PARAMS ((basic_block,
>  							  basic_block));
> @@ -248,6 +249,22 @@
>    return changed;
>  }
>  
> +/* Return true if LABEL is mentioned in any insn from INSN up to
> +   but not including the label.  */
> +
> +static bool
> +label_mentioned_p (label, insn)
> +     rtx label, insn;
> +{
> +  rtx x;
> +
> +  for (x = insn; x != label; x = NEXT_INSN (x))
> +    if (reg_mentioned_p (label, x))
> +      return true;
> +
> +  return false;
> +}
> +
>  /* Return true if LABEL is used for tail recursion.  */
>  
>  static bool
> @@ -1096,7 +1113,7 @@
>  	      /* If previous block ends with condjump jumping to next BB,
>  	         we can't delete the label.  */
>  	      && (b->pred->src == ENTRY_BLOCK_PTR
> -		  || !reg_mentioned_p (b->head, b->pred->src->end)))
> +		  || !label_mentioned_p (b->head, b->pred->src->end)))
>  	    {
>  	      rtx label = b->head;
>  	      b->head = NEXT_INSN (b->head);