Re: MIPS problem with structures/varargs

[Jeffrey A Law <law at cygnus dot com>]

  In message <19990903165218I.mitchell@codesourcery.com>you write:
  > I'm still unclear on one thing: is the fix split between the
  > middle-end and the back-end, or all in the back-end?  I understand
  > what you wrote; I'm unclear on whether or not doing something similar
  > in the MIPS backend is possible, without waiting for some
  > machine-independent part of the PA stuff to hit the tree.
There's generic code in the front-end (already in the tree) that is activiated
by function_arg returning that ugly expr_list/parallel thingie.  For the 
machine independent code look for emit_group_load or something like that.


The current PA code looks something like this:

     /* If the argument is more than a word long, then we need to align
         the base registers.  Same caveats as above.  */
      if (FUNCTION_ARG_SIZE (mode, type) > 1)
        {
          if (mode != BLKmode)
            {
              /* This seems backwards, but it is what HP specifies.  We need
                 gpr_reg_base to point to the smaller numbered register of
                 the integer register pair.  So if we have an even register
                  number, then decrement the gpr base.  */
              gpr_reg_base -= ((gpr_reg_base % 2) == 0);

              /* FP values behave sanely, except that each FP reg is only
                 half of word.  */
              fpr_reg_base += ((fpr_reg_base % 4) == 2) * 2;
            }
          else
            {
              rtx loc[8];
              int i, offset = 0, ub;
              ub = FUNCTION_ARG_SIZE (mode, type);
              ub = MIN(ub,
                       MAX(0, max_arg_words - cum->words - (cum->words & 1)));
              gpr_reg_base -= (cum->words & 1);
              for (i = 0; i < ub; i++)
                {
                  loc[i] = gen_rtx_EXPR_LIST (VOIDmode,
                                              gen_rtx_REG (DImode,
                                                           gpr_reg_base),
                                              GEN_INT(offset));
                  gpr_reg_base -= 1;
                  offset += 8;
                }
              if (ub == 0)
                return NULL_RTX;
              else if (ub == 1)
                return XEXP (loc[0], 0);
              else
                return gen_rtx_PARALLEL(mode, gen_rtvec_v(ub, loc));
            }
 

The key is that weird PARALLEL/EXPR_LIST stuff.  Hopefully I have a little time
to wrap my brain around this stuff soon since it can be used to fix a long
standing bug in the PA32 support.

jeff