[PATCH] Fix regno_in_use_p for RTL checking bootstrap (PR rtl-optimization/72821)

[Jakub Jelinek]

On Wed, Aug 03, 2016 at 02:59:30PM -0400, Vladimir N Makarov wrote:
> --- lra-spills.c      (revision 239000)
> +++ lra-spills.c      (working copy)
> @@ -686,16 +686,40 @@ return_regno_p (unsigned int regno)
>    return false;
>  }
>  
> -/* Return true if REGNO is one of subsequent USE after INSN.  */
> +/* Return true if REGNO is in one of subsequent USE after INSN in the
> +   same BB.  */
>  static bool
>  regno_in_use_p (rtx_insn *insn, unsigned int regno)
>  {
> +  static lra_insn_recog_data_t id;
> +  static struct lra_static_insn_data *static_id;
> +  struct lra_insn_reg *reg;
> +  int i, arg_regno;
> +  basic_block bb = BLOCK_FOR_INSN (insn);
> +
>    while ((insn = next_nondebug_insn (insn)) != NULL_RTX
> -      && INSN_P (insn) && GET_CODE (PATTERN (insn)) == USE)
> +      && bb == BLOCK_FOR_INSN (insn))

This broke rtl checking bootstrap, because BLOCK_FOR_INSN can't be used on
BARRIERs.

Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok for
trunk?

2016-08-06  Jakub Jelinek  <ja...@redhat.com>

        PR rtl-optimization/72821
        * lra-spills.c (regno_in_use_p): Don't use BLOCK_FOR_INSN on barriers,
        just return false for them.

--- gcc/lra-spills.c.jj 2016-08-06 12:11:49.000000000 +0200
+++ gcc/lra-spills.c    2016-08-06 14:05:22.751688854 +0200
@@ -697,9 +697,10 @@ regno_in_use_p (rtx_insn *insn, unsigned
   int i, arg_regno;
   basic_block bb = BLOCK_FOR_INSN (insn);
 
-  while ((insn = next_nondebug_insn (insn)) != NULL_RTX
-        && bb == BLOCK_FOR_INSN (insn))
+  while ((insn = next_nondebug_insn (insn)) != NULL_RTX)
     {
+      if (BARRIER_P (insn) || bb != BLOCK_FOR_INSN (insn))
+       return false;
       if (! INSN_P (insn))
        continue;
       if (GET_CODE (PATTERN (insn)) == USE

        Jakub