[Bug rtl-optimization/33828] Issues with the implementation of code hoisting in gcse.c

[steven at gcc dot gnu dot org]

------- Comment #6 from steven at gcc dot gnu dot org  2007-10-20 11:13 -------
Again from gcse.c:compute_transpout():

static void
compute_transpout (void)
{
  basic_block bb;
  unsigned int i;
  struct expr *expr;

  sbitmap_vector_ones (transpout, last_basic_block);

  FOR_EACH_BB (bb)
    {
      /* Note that flow inserted a nop a the end of basic blocks that
         end in call instructions for reasons other than abnormal
         control flow.  */
      if (! CALL_P (BB_END (bb)))
        continue;

      for (i = 0; i < expr_hash_table.size; i++)
        for (expr = expr_hash_table.table[i]; expr ; expr =
expr->next_same_hash)
          if (MEM_P (expr->expr))
            {
              if (GET_CODE (XEXP (expr->expr, 0)) == SYMBOL_REF
                  && CONSTANT_POOL_ADDRESS_P (XEXP (expr->expr, 0)))
                continue;

              /* ??? Optimally, we would use interprocedural alias
                 analysis to determine if this mem is actually killed
                 by this call.  */
              RESET_BIT (transpout[bb->index], expr->bitmap_index);
            }
    }
}

Notice the comment about interprocedural alias analysis.  That's all nice and
dandy, but it would be a start to handle const and pure calls here first. 
Calls to const and pure functions don't kill MEMs.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33828