SPARC, register renaming, __builtin_return_address

[David Edelsohn]

>>>>> Mark Mitchell writes:

Mark> But, I'm not sure what "force a stack frame" means.

	I meant something similar to rs6000.c:rs6000_stack_info() which
does:

  /* Determine if we need to allocate any stack frame:

     For AIX we need to push the stack if a frame pointer is needed (because
     the stack might be dynamically adjusted), if we are debugging, if we
     make calls, or if the sum of fp_save, gp_save, and local variables
     are more than the space needed to save all non-volatile registers:
     32-bit: 18*8 + 19*4 = 220 or 64-bit: 18*8 + 18*8 = 288 (GPR13 reserved).

     For V.4 we don't have the stack cushion that AIX uses, but assume that
     the debugger can handle stackless frames.  */

  if (info_ptr->calls_p)
    info_ptr->push_p = 1;

  else if (abi == ABI_V4 || abi == ABI_SOLARIS)
    info_ptr->push_p = (total_raw_size > info_ptr->fixed_size
                        || info_ptr->calls_p);

  else
    info_ptr->push_p = (frame_pointer_needed
                        || write_symbols != NO_DEBUG
                        || ((total_raw_size - info_ptr->fixed_size)
                            > (TARGET_32BIT ? 220 : 288)));

Note that for the AIX case, write_symbols != NO_DEBUG is one case.  (I am
not sure if the debugger still has problems with stackess frames.)

	I was suggesting that the equivalent part of the SPARC machine
description find some conservative test to require a stack frame in the
sort of cases that you are encountering.

David