This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: calls.c code permutation patch
> On Wed, May 03, 2000 at 02:55:36PM +0200, Jan Hubicka wrote:
> > * calls.c (expand_call): Reorganize somewhat in order
> > to avoid unnecesary recalculation inside loop and make
> > tail call failure code cleaner.
>
> This broke Alpha. I'm not sure why you were insisting on moving
> down the increment of currently_expanding_call, but it was still wrong.
I wanted to keep all the expanding stuff together..
>
> If it's not correct before optimize_tail_calls, it's possible
> to create circular rtl, which makes for all sorts of fun.
OK. I see.
sorry for another breakage.
Honza
>
>
> r~
>
>
> * calls.c (expand_call): Increment currently_expanding_call
> before calling optimize_tail_recursion.
>
> Index: calls.c
> ===================================================================
> RCS file: /cvs/gcc/egcs/gcc/calls.c,v
> retrieving revision 1.131
> diff -c -p -d -r1.131 calls.c
> *** calls.c 2000/05/09 14:33:41 1.131
> --- calls.c 2000/05/10 02:49:48
> *************** expand_call (exp, target, ignore)
> *** 2305,2313 ****
> This is most often true of sjlj-exceptions, which we couldn't
> tail-call to anyway. */
>
> ! if (!flag_optimize_sibling_calls
> || !rtx_equal_function_value_matters
> - || currently_expanding_call
> || !stmt_loop_nest_empty ()
> || any_pending_cleanups (1)
> || args_size.var)
> --- 2305,2313 ----
> This is most often true of sjlj-exceptions, which we couldn't
> tail-call to anyway. */
>
> ! if (currently_expanding_call++ != 0
> ! || !flag_optimize_sibling_calls
> || !rtx_equal_function_value_matters
> || !stmt_loop_nest_empty ()
> || any_pending_cleanups (1)
> || args_size.var)
> *************** expand_call (exp, target, ignore)
> *** 2447,2454 ****
> pending_stack_adjust = save_pending_stack_adjust;
> stack_pointer_delta = save_stack_pointer_delta;
> }
> -
> - currently_expanding_call++;
>
> if (profile_arc_flag && (flags & ECF_FORK_OR_EXEC))
> {
> --- 2447,2452 ----