PR target/65871
[official-gcc.git] / gcc / tree-ssa-loop-niter.c
blob042f8dff2f17c87e190b6ea02d64a2b4c78d3008
1 /* Functions to determine/estimate number of iterations of a loop.
2 Copyright (C) 2004-2015 Free Software Foundation, Inc.
4 This file is part of GCC.
6 GCC is free software; you can redistribute it and/or modify it
7 under the terms of the GNU General Public License as published by the
8 Free Software Foundation; either version 3, or (at your option) any
9 later version.
11 GCC is distributed in the hope that it will be useful, but WITHOUT
12 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
14 for more details.
16 You should have received a copy of the GNU General Public License
17 along with GCC; see the file COPYING3. If not see
18 <http://www.gnu.org/licenses/>. */
20 #include "config.h"
21 #include "system.h"
22 #include "coretypes.h"
23 #include "tm.h"
24 #include "hash-set.h"
25 #include "machmode.h"
26 #include "vec.h"
27 #include "double-int.h"
28 #include "input.h"
29 #include "alias.h"
30 #include "symtab.h"
31 #include "wide-int.h"
32 #include "inchash.h"
33 #include "tree.h"
34 #include "fold-const.h"
35 #include "calls.h"
36 #include "hashtab.h"
37 #include "hard-reg-set.h"
38 #include "function.h"
39 #include "rtl.h"
40 #include "flags.h"
41 #include "statistics.h"
42 #include "real.h"
43 #include "fixed-value.h"
44 #include "insn-config.h"
45 #include "expmed.h"
46 #include "dojump.h"
47 #include "explow.h"
48 #include "emit-rtl.h"
49 #include "varasm.h"
50 #include "stmt.h"
51 #include "expr.h"
52 #include "tm_p.h"
53 #include "predict.h"
54 #include "dominance.h"
55 #include "cfg.h"
56 #include "basic-block.h"
57 #include "gimple-pretty-print.h"
58 #include "intl.h"
59 #include "tree-ssa-alias.h"
60 #include "internal-fn.h"
61 #include "gimple-expr.h"
62 #include "is-a.h"
63 #include "gimple.h"
64 #include "gimplify.h"
65 #include "gimple-iterator.h"
66 #include "gimple-ssa.h"
67 #include "tree-cfg.h"
68 #include "tree-phinodes.h"
69 #include "ssa-iterators.h"
70 #include "tree-ssa-loop-ivopts.h"
71 #include "tree-ssa-loop-niter.h"
72 #include "tree-ssa-loop.h"
73 #include "dumpfile.h"
74 #include "cfgloop.h"
75 #include "tree-chrec.h"
76 #include "tree-scalar-evolution.h"
77 #include "tree-data-ref.h"
78 #include "params.h"
79 #include "diagnostic-core.h"
80 #include "tree-inline.h"
81 #include "tree-pass.h"
82 #include "stringpool.h"
83 #include "tree-ssanames.h"
84 #include "wide-int-print.h"
87 #define SWAP(X, Y) do { affine_iv *tmp = (X); (X) = (Y); (Y) = tmp; } while (0)
89 /* The maximum number of dominator BBs we search for conditions
90 of loop header copies we use for simplifying a conditional
91 expression. */
92 #define MAX_DOMINATORS_TO_WALK 8
96 Analysis of number of iterations of an affine exit test.
100 /* Bounds on some value, BELOW <= X <= UP. */
102 typedef struct
104 mpz_t below, up;
105 } bounds;
108 /* Splits expression EXPR to a variable part VAR and constant OFFSET. */
110 static void
111 split_to_var_and_offset (tree expr, tree *var, mpz_t offset)
113 tree type = TREE_TYPE (expr);
114 tree op0, op1;
115 bool negate = false;
117 *var = expr;
118 mpz_set_ui (offset, 0);
120 switch (TREE_CODE (expr))
122 case MINUS_EXPR:
123 negate = true;
124 /* Fallthru. */
126 case PLUS_EXPR:
127 case POINTER_PLUS_EXPR:
128 op0 = TREE_OPERAND (expr, 0);
129 op1 = TREE_OPERAND (expr, 1);
131 if (TREE_CODE (op1) != INTEGER_CST)
132 break;
134 *var = op0;
135 /* Always sign extend the offset. */
136 wi::to_mpz (op1, offset, SIGNED);
137 if (negate)
138 mpz_neg (offset, offset);
139 break;
141 case INTEGER_CST:
142 *var = build_int_cst_type (type, 0);
143 wi::to_mpz (expr, offset, TYPE_SIGN (type));
144 break;
146 default:
147 break;
151 /* Stores estimate on the minimum/maximum value of the expression VAR + OFF
152 in TYPE to MIN and MAX. */
154 static void
155 determine_value_range (struct loop *loop, tree type, tree var, mpz_t off,
156 mpz_t min, mpz_t max)
158 wide_int minv, maxv;
159 enum value_range_type rtype = VR_VARYING;
161 /* If the expression is a constant, we know its value exactly. */
162 if (integer_zerop (var))
164 mpz_set (min, off);
165 mpz_set (max, off);
166 return;
169 get_type_static_bounds (type, min, max);
171 /* See if we have some range info from VRP. */
172 if (TREE_CODE (var) == SSA_NAME && INTEGRAL_TYPE_P (type))
174 edge e = loop_preheader_edge (loop);
175 signop sgn = TYPE_SIGN (type);
176 gphi_iterator gsi;
178 /* Either for VAR itself... */
179 rtype = get_range_info (var, &minv, &maxv);
180 /* Or for PHI results in loop->header where VAR is used as
181 PHI argument from the loop preheader edge. */
182 for (gsi = gsi_start_phis (loop->header); !gsi_end_p (gsi); gsi_next (&gsi))
184 gphi *phi = gsi.phi ();
185 wide_int minc, maxc;
186 if (PHI_ARG_DEF_FROM_EDGE (phi, e) == var
187 && (get_range_info (gimple_phi_result (phi), &minc, &maxc)
188 == VR_RANGE))
190 if (rtype != VR_RANGE)
192 rtype = VR_RANGE;
193 minv = minc;
194 maxv = maxc;
196 else
198 minv = wi::max (minv, minc, sgn);
199 maxv = wi::min (maxv, maxc, sgn);
200 /* If the PHI result range are inconsistent with
201 the VAR range, give up on looking at the PHI
202 results. This can happen if VR_UNDEFINED is
203 involved. */
204 if (wi::gt_p (minv, maxv, sgn))
206 rtype = get_range_info (var, &minv, &maxv);
207 break;
212 if (rtype == VR_RANGE)
214 mpz_t minm, maxm;
215 gcc_assert (wi::le_p (minv, maxv, sgn));
216 mpz_init (minm);
217 mpz_init (maxm);
218 wi::to_mpz (minv, minm, sgn);
219 wi::to_mpz (maxv, maxm, sgn);
220 mpz_add (minm, minm, off);
221 mpz_add (maxm, maxm, off);
222 /* If the computation may not wrap or off is zero, then this
223 is always fine. If off is negative and minv + off isn't
224 smaller than type's minimum, or off is positive and
225 maxv + off isn't bigger than type's maximum, use the more
226 precise range too. */
227 if (nowrap_type_p (type)
228 || mpz_sgn (off) == 0
229 || (mpz_sgn (off) < 0 && mpz_cmp (minm, min) >= 0)
230 || (mpz_sgn (off) > 0 && mpz_cmp (maxm, max) <= 0))
232 mpz_set (min, minm);
233 mpz_set (max, maxm);
234 mpz_clear (minm);
235 mpz_clear (maxm);
236 return;
238 mpz_clear (minm);
239 mpz_clear (maxm);
243 /* If the computation may wrap, we know nothing about the value, except for
244 the range of the type. */
245 if (!nowrap_type_p (type))
246 return;
248 /* Since the addition of OFF does not wrap, if OFF is positive, then we may
249 add it to MIN, otherwise to MAX. */
250 if (mpz_sgn (off) < 0)
251 mpz_add (max, max, off);
252 else
253 mpz_add (min, min, off);
256 /* Stores the bounds on the difference of the values of the expressions
257 (var + X) and (var + Y), computed in TYPE, to BNDS. */
259 static void
260 bound_difference_of_offsetted_base (tree type, mpz_t x, mpz_t y,
261 bounds *bnds)
263 int rel = mpz_cmp (x, y);
264 bool may_wrap = !nowrap_type_p (type);
265 mpz_t m;
267 /* If X == Y, then the expressions are always equal.
268 If X > Y, there are the following possibilities:
269 a) neither of var + X and var + Y overflow or underflow, or both of
270 them do. Then their difference is X - Y.
271 b) var + X overflows, and var + Y does not. Then the values of the
272 expressions are var + X - M and var + Y, where M is the range of
273 the type, and their difference is X - Y - M.
274 c) var + Y underflows and var + X does not. Their difference again
275 is M - X + Y.
276 Therefore, if the arithmetics in type does not overflow, then the
277 bounds are (X - Y, X - Y), otherwise they are (X - Y - M, X - Y)
278 Similarly, if X < Y, the bounds are either (X - Y, X - Y) or
279 (X - Y, X - Y + M). */
281 if (rel == 0)
283 mpz_set_ui (bnds->below, 0);
284 mpz_set_ui (bnds->up, 0);
285 return;
288 mpz_init (m);
289 wi::to_mpz (wi::minus_one (TYPE_PRECISION (type)), m, UNSIGNED);
290 mpz_add_ui (m, m, 1);
291 mpz_sub (bnds->up, x, y);
292 mpz_set (bnds->below, bnds->up);
294 if (may_wrap)
296 if (rel > 0)
297 mpz_sub (bnds->below, bnds->below, m);
298 else
299 mpz_add (bnds->up, bnds->up, m);
302 mpz_clear (m);
305 /* From condition C0 CMP C1 derives information regarding the
306 difference of values of VARX + OFFX and VARY + OFFY, computed in TYPE,
307 and stores it to BNDS. */
309 static void
310 refine_bounds_using_guard (tree type, tree varx, mpz_t offx,
311 tree vary, mpz_t offy,
312 tree c0, enum tree_code cmp, tree c1,
313 bounds *bnds)
315 tree varc0, varc1, tmp, ctype;
316 mpz_t offc0, offc1, loffx, loffy, bnd;
317 bool lbound = false;
318 bool no_wrap = nowrap_type_p (type);
319 bool x_ok, y_ok;
321 switch (cmp)
323 case LT_EXPR:
324 case LE_EXPR:
325 case GT_EXPR:
326 case GE_EXPR:
327 STRIP_SIGN_NOPS (c0);
328 STRIP_SIGN_NOPS (c1);
329 ctype = TREE_TYPE (c0);
330 if (!useless_type_conversion_p (ctype, type))
331 return;
333 break;
335 case EQ_EXPR:
336 /* We could derive quite precise information from EQ_EXPR, however, such
337 a guard is unlikely to appear, so we do not bother with handling
338 it. */
339 return;
341 case NE_EXPR:
342 /* NE_EXPR comparisons do not contain much of useful information, except for
343 special case of comparing with the bounds of the type. */
344 if (TREE_CODE (c1) != INTEGER_CST
345 || !INTEGRAL_TYPE_P (type))
346 return;
348 /* Ensure that the condition speaks about an expression in the same type
349 as X and Y. */
350 ctype = TREE_TYPE (c0);
351 if (TYPE_PRECISION (ctype) != TYPE_PRECISION (type))
352 return;
353 c0 = fold_convert (type, c0);
354 c1 = fold_convert (type, c1);
356 if (TYPE_MIN_VALUE (type)
357 && operand_equal_p (c1, TYPE_MIN_VALUE (type), 0))
359 cmp = GT_EXPR;
360 break;
362 if (TYPE_MAX_VALUE (type)
363 && operand_equal_p (c1, TYPE_MAX_VALUE (type), 0))
365 cmp = LT_EXPR;
366 break;
369 return;
370 default:
371 return;
374 mpz_init (offc0);
375 mpz_init (offc1);
376 split_to_var_and_offset (expand_simple_operations (c0), &varc0, offc0);
377 split_to_var_and_offset (expand_simple_operations (c1), &varc1, offc1);
379 /* We are only interested in comparisons of expressions based on VARX and
380 VARY. TODO -- we might also be able to derive some bounds from
381 expressions containing just one of the variables. */
383 if (operand_equal_p (varx, varc1, 0))
385 tmp = varc0; varc0 = varc1; varc1 = tmp;
386 mpz_swap (offc0, offc1);
387 cmp = swap_tree_comparison (cmp);
390 if (!operand_equal_p (varx, varc0, 0)
391 || !operand_equal_p (vary, varc1, 0))
392 goto end;
394 mpz_init_set (loffx, offx);
395 mpz_init_set (loffy, offy);
397 if (cmp == GT_EXPR || cmp == GE_EXPR)
399 tmp = varx; varx = vary; vary = tmp;
400 mpz_swap (offc0, offc1);
401 mpz_swap (loffx, loffy);
402 cmp = swap_tree_comparison (cmp);
403 lbound = true;
406 /* If there is no overflow, the condition implies that
408 (VARX + OFFX) cmp (VARY + OFFY) + (OFFX - OFFY + OFFC1 - OFFC0).
410 The overflows and underflows may complicate things a bit; each
411 overflow decreases the appropriate offset by M, and underflow
412 increases it by M. The above inequality would not necessarily be
413 true if
415 -- VARX + OFFX underflows and VARX + OFFC0 does not, or
416 VARX + OFFC0 overflows, but VARX + OFFX does not.
417 This may only happen if OFFX < OFFC0.
418 -- VARY + OFFY overflows and VARY + OFFC1 does not, or
419 VARY + OFFC1 underflows and VARY + OFFY does not.
420 This may only happen if OFFY > OFFC1. */
422 if (no_wrap)
424 x_ok = true;
425 y_ok = true;
427 else
429 x_ok = (integer_zerop (varx)
430 || mpz_cmp (loffx, offc0) >= 0);
431 y_ok = (integer_zerop (vary)
432 || mpz_cmp (loffy, offc1) <= 0);
435 if (x_ok && y_ok)
437 mpz_init (bnd);
438 mpz_sub (bnd, loffx, loffy);
439 mpz_add (bnd, bnd, offc1);
440 mpz_sub (bnd, bnd, offc0);
442 if (cmp == LT_EXPR)
443 mpz_sub_ui (bnd, bnd, 1);
445 if (lbound)
447 mpz_neg (bnd, bnd);
448 if (mpz_cmp (bnds->below, bnd) < 0)
449 mpz_set (bnds->below, bnd);
451 else
453 if (mpz_cmp (bnd, bnds->up) < 0)
454 mpz_set (bnds->up, bnd);
456 mpz_clear (bnd);
459 mpz_clear (loffx);
460 mpz_clear (loffy);
461 end:
462 mpz_clear (offc0);
463 mpz_clear (offc1);
466 /* Stores the bounds on the value of the expression X - Y in LOOP to BNDS.
467 The subtraction is considered to be performed in arbitrary precision,
468 without overflows.
470 We do not attempt to be too clever regarding the value ranges of X and
471 Y; most of the time, they are just integers or ssa names offsetted by
472 integer. However, we try to use the information contained in the
473 comparisons before the loop (usually created by loop header copying). */
475 static void
476 bound_difference (struct loop *loop, tree x, tree y, bounds *bnds)
478 tree type = TREE_TYPE (x);
479 tree varx, vary;
480 mpz_t offx, offy;
481 mpz_t minx, maxx, miny, maxy;
482 int cnt = 0;
483 edge e;
484 basic_block bb;
485 tree c0, c1;
486 gimple cond;
487 enum tree_code cmp;
489 /* Get rid of unnecessary casts, but preserve the value of
490 the expressions. */
491 STRIP_SIGN_NOPS (x);
492 STRIP_SIGN_NOPS (y);
494 mpz_init (bnds->below);
495 mpz_init (bnds->up);
496 mpz_init (offx);
497 mpz_init (offy);
498 split_to_var_and_offset (x, &varx, offx);
499 split_to_var_and_offset (y, &vary, offy);
501 if (!integer_zerop (varx)
502 && operand_equal_p (varx, vary, 0))
504 /* Special case VARX == VARY -- we just need to compare the
505 offsets. The matters are a bit more complicated in the
506 case addition of offsets may wrap. */
507 bound_difference_of_offsetted_base (type, offx, offy, bnds);
509 else
511 /* Otherwise, use the value ranges to determine the initial
512 estimates on below and up. */
513 mpz_init (minx);
514 mpz_init (maxx);
515 mpz_init (miny);
516 mpz_init (maxy);
517 determine_value_range (loop, type, varx, offx, minx, maxx);
518 determine_value_range (loop, type, vary, offy, miny, maxy);
520 mpz_sub (bnds->below, minx, maxy);
521 mpz_sub (bnds->up, maxx, miny);
522 mpz_clear (minx);
523 mpz_clear (maxx);
524 mpz_clear (miny);
525 mpz_clear (maxy);
528 /* If both X and Y are constants, we cannot get any more precise. */
529 if (integer_zerop (varx) && integer_zerop (vary))
530 goto end;
532 /* Now walk the dominators of the loop header and use the entry
533 guards to refine the estimates. */
534 for (bb = loop->header;
535 bb != ENTRY_BLOCK_PTR_FOR_FN (cfun) && cnt < MAX_DOMINATORS_TO_WALK;
536 bb = get_immediate_dominator (CDI_DOMINATORS, bb))
538 if (!single_pred_p (bb))
539 continue;
540 e = single_pred_edge (bb);
542 if (!(e->flags & (EDGE_TRUE_VALUE | EDGE_FALSE_VALUE)))
543 continue;
545 cond = last_stmt (e->src);
546 c0 = gimple_cond_lhs (cond);
547 cmp = gimple_cond_code (cond);
548 c1 = gimple_cond_rhs (cond);
550 if (e->flags & EDGE_FALSE_VALUE)
551 cmp = invert_tree_comparison (cmp, false);
553 refine_bounds_using_guard (type, varx, offx, vary, offy,
554 c0, cmp, c1, bnds);
555 ++cnt;
558 end:
559 mpz_clear (offx);
560 mpz_clear (offy);
563 /* Update the bounds in BNDS that restrict the value of X to the bounds
564 that restrict the value of X + DELTA. X can be obtained as a
565 difference of two values in TYPE. */
567 static void
568 bounds_add (bounds *bnds, const widest_int &delta, tree type)
570 mpz_t mdelta, max;
572 mpz_init (mdelta);
573 wi::to_mpz (delta, mdelta, SIGNED);
575 mpz_init (max);
576 wi::to_mpz (wi::minus_one (TYPE_PRECISION (type)), max, UNSIGNED);
578 mpz_add (bnds->up, bnds->up, mdelta);
579 mpz_add (bnds->below, bnds->below, mdelta);
581 if (mpz_cmp (bnds->up, max) > 0)
582 mpz_set (bnds->up, max);
584 mpz_neg (max, max);
585 if (mpz_cmp (bnds->below, max) < 0)
586 mpz_set (bnds->below, max);
588 mpz_clear (mdelta);
589 mpz_clear (max);
592 /* Update the bounds in BNDS that restrict the value of X to the bounds
593 that restrict the value of -X. */
595 static void
596 bounds_negate (bounds *bnds)
598 mpz_t tmp;
600 mpz_init_set (tmp, bnds->up);
601 mpz_neg (bnds->up, bnds->below);
602 mpz_neg (bnds->below, tmp);
603 mpz_clear (tmp);
606 /* Returns inverse of X modulo 2^s, where MASK = 2^s-1. */
608 static tree
609 inverse (tree x, tree mask)
611 tree type = TREE_TYPE (x);
612 tree rslt;
613 unsigned ctr = tree_floor_log2 (mask);
615 if (TYPE_PRECISION (type) <= HOST_BITS_PER_WIDE_INT)
617 unsigned HOST_WIDE_INT ix;
618 unsigned HOST_WIDE_INT imask;
619 unsigned HOST_WIDE_INT irslt = 1;
621 gcc_assert (cst_and_fits_in_hwi (x));
622 gcc_assert (cst_and_fits_in_hwi (mask));
624 ix = int_cst_value (x);
625 imask = int_cst_value (mask);
627 for (; ctr; ctr--)
629 irslt *= ix;
630 ix *= ix;
632 irslt &= imask;
634 rslt = build_int_cst_type (type, irslt);
636 else
638 rslt = build_int_cst (type, 1);
639 for (; ctr; ctr--)
641 rslt = int_const_binop (MULT_EXPR, rslt, x);
642 x = int_const_binop (MULT_EXPR, x, x);
644 rslt = int_const_binop (BIT_AND_EXPR, rslt, mask);
647 return rslt;
650 /* Derives the upper bound BND on the number of executions of loop with exit
651 condition S * i <> C. If NO_OVERFLOW is true, then the control variable of
652 the loop does not overflow. EXIT_MUST_BE_TAKEN is true if we are guaranteed
653 that the loop ends through this exit, i.e., the induction variable ever
654 reaches the value of C.
656 The value C is equal to final - base, where final and base are the final and
657 initial value of the actual induction variable in the analysed loop. BNDS
658 bounds the value of this difference when computed in signed type with
659 unbounded range, while the computation of C is performed in an unsigned
660 type with the range matching the range of the type of the induction variable.
661 In particular, BNDS.up contains an upper bound on C in the following cases:
662 -- if the iv must reach its final value without overflow, i.e., if
663 NO_OVERFLOW && EXIT_MUST_BE_TAKEN is true, or
664 -- if final >= base, which we know to hold when BNDS.below >= 0. */
666 static void
667 number_of_iterations_ne_max (mpz_t bnd, bool no_overflow, tree c, tree s,
668 bounds *bnds, bool exit_must_be_taken)
670 widest_int max;
671 mpz_t d;
672 tree type = TREE_TYPE (c);
673 bool bnds_u_valid = ((no_overflow && exit_must_be_taken)
674 || mpz_sgn (bnds->below) >= 0);
676 if (integer_onep (s)
677 || (TREE_CODE (c) == INTEGER_CST
678 && TREE_CODE (s) == INTEGER_CST
679 && wi::mod_trunc (c, s, TYPE_SIGN (type)) == 0)
680 || (TYPE_OVERFLOW_UNDEFINED (type)
681 && multiple_of_p (type, c, s)))
683 /* If C is an exact multiple of S, then its value will be reached before
684 the induction variable overflows (unless the loop is exited in some
685 other way before). Note that the actual induction variable in the
686 loop (which ranges from base to final instead of from 0 to C) may
687 overflow, in which case BNDS.up will not be giving a correct upper
688 bound on C; thus, BNDS_U_VALID had to be computed in advance. */
689 no_overflow = true;
690 exit_must_be_taken = true;
693 /* If the induction variable can overflow, the number of iterations is at
694 most the period of the control variable (or infinite, but in that case
695 the whole # of iterations analysis will fail). */
696 if (!no_overflow)
698 max = wi::mask <widest_int> (TYPE_PRECISION (type) - wi::ctz (s), false);
699 wi::to_mpz (max, bnd, UNSIGNED);
700 return;
703 /* Now we know that the induction variable does not overflow, so the loop
704 iterates at most (range of type / S) times. */
705 wi::to_mpz (wi::minus_one (TYPE_PRECISION (type)), bnd, UNSIGNED);
707 /* If the induction variable is guaranteed to reach the value of C before
708 overflow, ... */
709 if (exit_must_be_taken)
711 /* ... then we can strengthen this to C / S, and possibly we can use
712 the upper bound on C given by BNDS. */
713 if (TREE_CODE (c) == INTEGER_CST)
714 wi::to_mpz (c, bnd, UNSIGNED);
715 else if (bnds_u_valid)
716 mpz_set (bnd, bnds->up);
719 mpz_init (d);
720 wi::to_mpz (s, d, UNSIGNED);
721 mpz_fdiv_q (bnd, bnd, d);
722 mpz_clear (d);
725 /* Determines number of iterations of loop whose ending condition
726 is IV <> FINAL. TYPE is the type of the iv. The number of
727 iterations is stored to NITER. EXIT_MUST_BE_TAKEN is true if
728 we know that the exit must be taken eventually, i.e., that the IV
729 ever reaches the value FINAL (we derived this earlier, and possibly set
730 NITER->assumptions to make sure this is the case). BNDS contains the
731 bounds on the difference FINAL - IV->base. */
733 static bool
734 number_of_iterations_ne (tree type, affine_iv *iv, tree final,
735 struct tree_niter_desc *niter, bool exit_must_be_taken,
736 bounds *bnds)
738 tree niter_type = unsigned_type_for (type);
739 tree s, c, d, bits, assumption, tmp, bound;
740 mpz_t max;
742 niter->control = *iv;
743 niter->bound = final;
744 niter->cmp = NE_EXPR;
746 /* Rearrange the terms so that we get inequality S * i <> C, with S
747 positive. Also cast everything to the unsigned type. If IV does
748 not overflow, BNDS bounds the value of C. Also, this is the
749 case if the computation |FINAL - IV->base| does not overflow, i.e.,
750 if BNDS->below in the result is nonnegative. */
751 if (tree_int_cst_sign_bit (iv->step))
753 s = fold_convert (niter_type,
754 fold_build1 (NEGATE_EXPR, type, iv->step));
755 c = fold_build2 (MINUS_EXPR, niter_type,
756 fold_convert (niter_type, iv->base),
757 fold_convert (niter_type, final));
758 bounds_negate (bnds);
760 else
762 s = fold_convert (niter_type, iv->step);
763 c = fold_build2 (MINUS_EXPR, niter_type,
764 fold_convert (niter_type, final),
765 fold_convert (niter_type, iv->base));
768 mpz_init (max);
769 number_of_iterations_ne_max (max, iv->no_overflow, c, s, bnds,
770 exit_must_be_taken);
771 niter->max = widest_int::from (wi::from_mpz (niter_type, max, false),
772 TYPE_SIGN (niter_type));
773 mpz_clear (max);
775 /* First the trivial cases -- when the step is 1. */
776 if (integer_onep (s))
778 niter->niter = c;
779 return true;
782 /* Let nsd (step, size of mode) = d. If d does not divide c, the loop
783 is infinite. Otherwise, the number of iterations is
784 (inverse(s/d) * (c/d)) mod (size of mode/d). */
785 bits = num_ending_zeros (s);
786 bound = build_low_bits_mask (niter_type,
787 (TYPE_PRECISION (niter_type)
788 - tree_to_uhwi (bits)));
790 d = fold_binary_to_constant (LSHIFT_EXPR, niter_type,
791 build_int_cst (niter_type, 1), bits);
792 s = fold_binary_to_constant (RSHIFT_EXPR, niter_type, s, bits);
794 if (!exit_must_be_taken)
796 /* If we cannot assume that the exit is taken eventually, record the
797 assumptions for divisibility of c. */
798 assumption = fold_build2 (FLOOR_MOD_EXPR, niter_type, c, d);
799 assumption = fold_build2 (EQ_EXPR, boolean_type_node,
800 assumption, build_int_cst (niter_type, 0));
801 if (!integer_nonzerop (assumption))
802 niter->assumptions = fold_build2 (TRUTH_AND_EXPR, boolean_type_node,
803 niter->assumptions, assumption);
806 c = fold_build2 (EXACT_DIV_EXPR, niter_type, c, d);
807 tmp = fold_build2 (MULT_EXPR, niter_type, c, inverse (s, bound));
808 niter->niter = fold_build2 (BIT_AND_EXPR, niter_type, tmp, bound);
809 return true;
812 /* Checks whether we can determine the final value of the control variable
813 of the loop with ending condition IV0 < IV1 (computed in TYPE).
814 DELTA is the difference IV1->base - IV0->base, STEP is the absolute value
815 of the step. The assumptions necessary to ensure that the computation
816 of the final value does not overflow are recorded in NITER. If we
817 find the final value, we adjust DELTA and return TRUE. Otherwise
818 we return false. BNDS bounds the value of IV1->base - IV0->base,
819 and will be updated by the same amount as DELTA. EXIT_MUST_BE_TAKEN is
820 true if we know that the exit must be taken eventually. */
822 static bool
823 number_of_iterations_lt_to_ne (tree type, affine_iv *iv0, affine_iv *iv1,
824 struct tree_niter_desc *niter,
825 tree *delta, tree step,
826 bool exit_must_be_taken, bounds *bnds)
828 tree niter_type = TREE_TYPE (step);
829 tree mod = fold_build2 (FLOOR_MOD_EXPR, niter_type, *delta, step);
830 tree tmod;
831 mpz_t mmod;
832 tree assumption = boolean_true_node, bound, noloop;
833 bool ret = false, fv_comp_no_overflow;
834 tree type1 = type;
835 if (POINTER_TYPE_P (type))
836 type1 = sizetype;
838 if (TREE_CODE (mod) != INTEGER_CST)
839 return false;
840 if (integer_nonzerop (mod))
841 mod = fold_build2 (MINUS_EXPR, niter_type, step, mod);
842 tmod = fold_convert (type1, mod);
844 mpz_init (mmod);
845 wi::to_mpz (mod, mmod, UNSIGNED);
846 mpz_neg (mmod, mmod);
848 /* If the induction variable does not overflow and the exit is taken,
849 then the computation of the final value does not overflow. This is
850 also obviously the case if the new final value is equal to the
851 current one. Finally, we postulate this for pointer type variables,
852 as the code cannot rely on the object to that the pointer points being
853 placed at the end of the address space (and more pragmatically,
854 TYPE_{MIN,MAX}_VALUE is not defined for pointers). */
855 if (integer_zerop (mod) || POINTER_TYPE_P (type))
856 fv_comp_no_overflow = true;
857 else if (!exit_must_be_taken)
858 fv_comp_no_overflow = false;
859 else
860 fv_comp_no_overflow =
861 (iv0->no_overflow && integer_nonzerop (iv0->step))
862 || (iv1->no_overflow && integer_nonzerop (iv1->step));
864 if (integer_nonzerop (iv0->step))
866 /* The final value of the iv is iv1->base + MOD, assuming that this
867 computation does not overflow, and that
868 iv0->base <= iv1->base + MOD. */
869 if (!fv_comp_no_overflow)
871 bound = fold_build2 (MINUS_EXPR, type1,
872 TYPE_MAX_VALUE (type1), tmod);
873 assumption = fold_build2 (LE_EXPR, boolean_type_node,
874 iv1->base, bound);
875 if (integer_zerop (assumption))
876 goto end;
878 if (mpz_cmp (mmod, bnds->below) < 0)
879 noloop = boolean_false_node;
880 else if (POINTER_TYPE_P (type))
881 noloop = fold_build2 (GT_EXPR, boolean_type_node,
882 iv0->base,
883 fold_build_pointer_plus (iv1->base, tmod));
884 else
885 noloop = fold_build2 (GT_EXPR, boolean_type_node,
886 iv0->base,
887 fold_build2 (PLUS_EXPR, type1,
888 iv1->base, tmod));
890 else
892 /* The final value of the iv is iv0->base - MOD, assuming that this
893 computation does not overflow, and that
894 iv0->base - MOD <= iv1->base. */
895 if (!fv_comp_no_overflow)
897 bound = fold_build2 (PLUS_EXPR, type1,
898 TYPE_MIN_VALUE (type1), tmod);
899 assumption = fold_build2 (GE_EXPR, boolean_type_node,
900 iv0->base, bound);
901 if (integer_zerop (assumption))
902 goto end;
904 if (mpz_cmp (mmod, bnds->below) < 0)
905 noloop = boolean_false_node;
906 else if (POINTER_TYPE_P (type))
907 noloop = fold_build2 (GT_EXPR, boolean_type_node,
908 fold_build_pointer_plus (iv0->base,
909 fold_build1 (NEGATE_EXPR,
910 type1, tmod)),
911 iv1->base);
912 else
913 noloop = fold_build2 (GT_EXPR, boolean_type_node,
914 fold_build2 (MINUS_EXPR, type1,
915 iv0->base, tmod),
916 iv1->base);
919 if (!integer_nonzerop (assumption))
920 niter->assumptions = fold_build2 (TRUTH_AND_EXPR, boolean_type_node,
921 niter->assumptions,
922 assumption);
923 if (!integer_zerop (noloop))
924 niter->may_be_zero = fold_build2 (TRUTH_OR_EXPR, boolean_type_node,
925 niter->may_be_zero,
926 noloop);
927 bounds_add (bnds, wi::to_widest (mod), type);
928 *delta = fold_build2 (PLUS_EXPR, niter_type, *delta, mod);
930 ret = true;
931 end:
932 mpz_clear (mmod);
933 return ret;
936 /* Add assertions to NITER that ensure that the control variable of the loop
937 with ending condition IV0 < IV1 does not overflow. Types of IV0 and IV1
938 are TYPE. Returns false if we can prove that there is an overflow, true
939 otherwise. STEP is the absolute value of the step. */
941 static bool
942 assert_no_overflow_lt (tree type, affine_iv *iv0, affine_iv *iv1,
943 struct tree_niter_desc *niter, tree step)
945 tree bound, d, assumption, diff;
946 tree niter_type = TREE_TYPE (step);
948 if (integer_nonzerop (iv0->step))
950 /* for (i = iv0->base; i < iv1->base; i += iv0->step) */
951 if (iv0->no_overflow)
952 return true;
954 /* If iv0->base is a constant, we can determine the last value before
955 overflow precisely; otherwise we conservatively assume
956 MAX - STEP + 1. */
958 if (TREE_CODE (iv0->base) == INTEGER_CST)
960 d = fold_build2 (MINUS_EXPR, niter_type,
961 fold_convert (niter_type, TYPE_MAX_VALUE (type)),
962 fold_convert (niter_type, iv0->base));
963 diff = fold_build2 (FLOOR_MOD_EXPR, niter_type, d, step);
965 else
966 diff = fold_build2 (MINUS_EXPR, niter_type, step,
967 build_int_cst (niter_type, 1));
968 bound = fold_build2 (MINUS_EXPR, type,
969 TYPE_MAX_VALUE (type), fold_convert (type, diff));
970 assumption = fold_build2 (LE_EXPR, boolean_type_node,
971 iv1->base, bound);
973 else
975 /* for (i = iv1->base; i > iv0->base; i += iv1->step) */
976 if (iv1->no_overflow)
977 return true;
979 if (TREE_CODE (iv1->base) == INTEGER_CST)
981 d = fold_build2 (MINUS_EXPR, niter_type,
982 fold_convert (niter_type, iv1->base),
983 fold_convert (niter_type, TYPE_MIN_VALUE (type)));
984 diff = fold_build2 (FLOOR_MOD_EXPR, niter_type, d, step);
986 else
987 diff = fold_build2 (MINUS_EXPR, niter_type, step,
988 build_int_cst (niter_type, 1));
989 bound = fold_build2 (PLUS_EXPR, type,
990 TYPE_MIN_VALUE (type), fold_convert (type, diff));
991 assumption = fold_build2 (GE_EXPR, boolean_type_node,
992 iv0->base, bound);
995 if (integer_zerop (assumption))
996 return false;
997 if (!integer_nonzerop (assumption))
998 niter->assumptions = fold_build2 (TRUTH_AND_EXPR, boolean_type_node,
999 niter->assumptions, assumption);
1001 iv0->no_overflow = true;
1002 iv1->no_overflow = true;
1003 return true;
1006 /* Add an assumption to NITER that a loop whose ending condition
1007 is IV0 < IV1 rolls. TYPE is the type of the control iv. BNDS
1008 bounds the value of IV1->base - IV0->base. */
1010 static void
1011 assert_loop_rolls_lt (tree type, affine_iv *iv0, affine_iv *iv1,
1012 struct tree_niter_desc *niter, bounds *bnds)
1014 tree assumption = boolean_true_node, bound, diff;
1015 tree mbz, mbzl, mbzr, type1;
1016 bool rolls_p, no_overflow_p;
1017 widest_int dstep;
1018 mpz_t mstep, max;
1020 /* We are going to compute the number of iterations as
1021 (iv1->base - iv0->base + step - 1) / step, computed in the unsigned
1022 variant of TYPE. This formula only works if
1024 -step + 1 <= (iv1->base - iv0->base) <= MAX - step + 1
1026 (where MAX is the maximum value of the unsigned variant of TYPE, and
1027 the computations in this formula are performed in full precision,
1028 i.e., without overflows).
1030 Usually, for loops with exit condition iv0->base + step * i < iv1->base,
1031 we have a condition of the form iv0->base - step < iv1->base before the loop,
1032 and for loops iv0->base < iv1->base - step * i the condition
1033 iv0->base < iv1->base + step, due to loop header copying, which enable us
1034 to prove the lower bound.
1036 The upper bound is more complicated. Unless the expressions for initial
1037 and final value themselves contain enough information, we usually cannot
1038 derive it from the context. */
1040 /* First check whether the answer does not follow from the bounds we gathered
1041 before. */
1042 if (integer_nonzerop (iv0->step))
1043 dstep = wi::to_widest (iv0->step);
1044 else
1046 dstep = wi::sext (wi::to_widest (iv1->step), TYPE_PRECISION (type));
1047 dstep = -dstep;
1050 mpz_init (mstep);
1051 wi::to_mpz (dstep, mstep, UNSIGNED);
1052 mpz_neg (mstep, mstep);
1053 mpz_add_ui (mstep, mstep, 1);
1055 rolls_p = mpz_cmp (mstep, bnds->below) <= 0;
1057 mpz_init (max);
1058 wi::to_mpz (wi::minus_one (TYPE_PRECISION (type)), max, UNSIGNED);
1059 mpz_add (max, max, mstep);
1060 no_overflow_p = (mpz_cmp (bnds->up, max) <= 0
1061 /* For pointers, only values lying inside a single object
1062 can be compared or manipulated by pointer arithmetics.
1063 Gcc in general does not allow or handle objects larger
1064 than half of the address space, hence the upper bound
1065 is satisfied for pointers. */
1066 || POINTER_TYPE_P (type));
1067 mpz_clear (mstep);
1068 mpz_clear (max);
1070 if (rolls_p && no_overflow_p)
1071 return;
1073 type1 = type;
1074 if (POINTER_TYPE_P (type))
1075 type1 = sizetype;
1077 /* Now the hard part; we must formulate the assumption(s) as expressions, and
1078 we must be careful not to introduce overflow. */
1080 if (integer_nonzerop (iv0->step))
1082 diff = fold_build2 (MINUS_EXPR, type1,
1083 iv0->step, build_int_cst (type1, 1));
1085 /* We need to know that iv0->base >= MIN + iv0->step - 1. Since
1086 0 address never belongs to any object, we can assume this for
1087 pointers. */
1088 if (!POINTER_TYPE_P (type))
1090 bound = fold_build2 (PLUS_EXPR, type1,
1091 TYPE_MIN_VALUE (type), diff);
1092 assumption = fold_build2 (GE_EXPR, boolean_type_node,
1093 iv0->base, bound);
1096 /* And then we can compute iv0->base - diff, and compare it with
1097 iv1->base. */
1098 mbzl = fold_build2 (MINUS_EXPR, type1,
1099 fold_convert (type1, iv0->base), diff);
1100 mbzr = fold_convert (type1, iv1->base);
1102 else
1104 diff = fold_build2 (PLUS_EXPR, type1,
1105 iv1->step, build_int_cst (type1, 1));
1107 if (!POINTER_TYPE_P (type))
1109 bound = fold_build2 (PLUS_EXPR, type1,
1110 TYPE_MAX_VALUE (type), diff);
1111 assumption = fold_build2 (LE_EXPR, boolean_type_node,
1112 iv1->base, bound);
1115 mbzl = fold_convert (type1, iv0->base);
1116 mbzr = fold_build2 (MINUS_EXPR, type1,
1117 fold_convert (type1, iv1->base), diff);
1120 if (!integer_nonzerop (assumption))
1121 niter->assumptions = fold_build2 (TRUTH_AND_EXPR, boolean_type_node,
1122 niter->assumptions, assumption);
1123 if (!rolls_p)
1125 mbz = fold_build2 (GT_EXPR, boolean_type_node, mbzl, mbzr);
1126 niter->may_be_zero = fold_build2 (TRUTH_OR_EXPR, boolean_type_node,
1127 niter->may_be_zero, mbz);
1131 /* Determines number of iterations of loop whose ending condition
1132 is IV0 < IV1. TYPE is the type of the iv. The number of
1133 iterations is stored to NITER. BNDS bounds the difference
1134 IV1->base - IV0->base. EXIT_MUST_BE_TAKEN is true if we know
1135 that the exit must be taken eventually. */
1137 static bool
1138 number_of_iterations_lt (tree type, affine_iv *iv0, affine_iv *iv1,
1139 struct tree_niter_desc *niter,
1140 bool exit_must_be_taken, bounds *bnds)
1142 tree niter_type = unsigned_type_for (type);
1143 tree delta, step, s;
1144 mpz_t mstep, tmp;
1146 if (integer_nonzerop (iv0->step))
1148 niter->control = *iv0;
1149 niter->cmp = LT_EXPR;
1150 niter->bound = iv1->base;
1152 else
1154 niter->control = *iv1;
1155 niter->cmp = GT_EXPR;
1156 niter->bound = iv0->base;
1159 delta = fold_build2 (MINUS_EXPR, niter_type,
1160 fold_convert (niter_type, iv1->base),
1161 fold_convert (niter_type, iv0->base));
1163 /* First handle the special case that the step is +-1. */
1164 if ((integer_onep (iv0->step) && integer_zerop (iv1->step))
1165 || (integer_all_onesp (iv1->step) && integer_zerop (iv0->step)))
1167 /* for (i = iv0->base; i < iv1->base; i++)
1171 for (i = iv1->base; i > iv0->base; i--).
1173 In both cases # of iterations is iv1->base - iv0->base, assuming that
1174 iv1->base >= iv0->base.
1176 First try to derive a lower bound on the value of
1177 iv1->base - iv0->base, computed in full precision. If the difference
1178 is nonnegative, we are done, otherwise we must record the
1179 condition. */
1181 if (mpz_sgn (bnds->below) < 0)
1182 niter->may_be_zero = fold_build2 (LT_EXPR, boolean_type_node,
1183 iv1->base, iv0->base);
1184 niter->niter = delta;
1185 niter->max = widest_int::from (wi::from_mpz (niter_type, bnds->up, false),
1186 TYPE_SIGN (niter_type));
1187 return true;
1190 if (integer_nonzerop (iv0->step))
1191 step = fold_convert (niter_type, iv0->step);
1192 else
1193 step = fold_convert (niter_type,
1194 fold_build1 (NEGATE_EXPR, type, iv1->step));
1196 /* If we can determine the final value of the control iv exactly, we can
1197 transform the condition to != comparison. In particular, this will be
1198 the case if DELTA is constant. */
1199 if (number_of_iterations_lt_to_ne (type, iv0, iv1, niter, &delta, step,
1200 exit_must_be_taken, bnds))
1202 affine_iv zps;
1204 zps.base = build_int_cst (niter_type, 0);
1205 zps.step = step;
1206 /* number_of_iterations_lt_to_ne will add assumptions that ensure that
1207 zps does not overflow. */
1208 zps.no_overflow = true;
1210 return number_of_iterations_ne (type, &zps, delta, niter, true, bnds);
1213 /* Make sure that the control iv does not overflow. */
1214 if (!assert_no_overflow_lt (type, iv0, iv1, niter, step))
1215 return false;
1217 /* We determine the number of iterations as (delta + step - 1) / step. For
1218 this to work, we must know that iv1->base >= iv0->base - step + 1,
1219 otherwise the loop does not roll. */
1220 assert_loop_rolls_lt (type, iv0, iv1, niter, bnds);
1222 s = fold_build2 (MINUS_EXPR, niter_type,
1223 step, build_int_cst (niter_type, 1));
1224 delta = fold_build2 (PLUS_EXPR, niter_type, delta, s);
1225 niter->niter = fold_build2 (FLOOR_DIV_EXPR, niter_type, delta, step);
1227 mpz_init (mstep);
1228 mpz_init (tmp);
1229 wi::to_mpz (step, mstep, UNSIGNED);
1230 mpz_add (tmp, bnds->up, mstep);
1231 mpz_sub_ui (tmp, tmp, 1);
1232 mpz_fdiv_q (tmp, tmp, mstep);
1233 niter->max = widest_int::from (wi::from_mpz (niter_type, tmp, false),
1234 TYPE_SIGN (niter_type));
1235 mpz_clear (mstep);
1236 mpz_clear (tmp);
1238 return true;
1241 /* Determines number of iterations of loop whose ending condition
1242 is IV0 <= IV1. TYPE is the type of the iv. The number of
1243 iterations is stored to NITER. EXIT_MUST_BE_TAKEN is true if
1244 we know that this condition must eventually become false (we derived this
1245 earlier, and possibly set NITER->assumptions to make sure this
1246 is the case). BNDS bounds the difference IV1->base - IV0->base. */
1248 static bool
1249 number_of_iterations_le (tree type, affine_iv *iv0, affine_iv *iv1,
1250 struct tree_niter_desc *niter, bool exit_must_be_taken,
1251 bounds *bnds)
1253 tree assumption;
1254 tree type1 = type;
1255 if (POINTER_TYPE_P (type))
1256 type1 = sizetype;
1258 /* Say that IV0 is the control variable. Then IV0 <= IV1 iff
1259 IV0 < IV1 + 1, assuming that IV1 is not equal to the greatest
1260 value of the type. This we must know anyway, since if it is
1261 equal to this value, the loop rolls forever. We do not check
1262 this condition for pointer type ivs, as the code cannot rely on
1263 the object to that the pointer points being placed at the end of
1264 the address space (and more pragmatically, TYPE_{MIN,MAX}_VALUE is
1265 not defined for pointers). */
1267 if (!exit_must_be_taken && !POINTER_TYPE_P (type))
1269 if (integer_nonzerop (iv0->step))
1270 assumption = fold_build2 (NE_EXPR, boolean_type_node,
1271 iv1->base, TYPE_MAX_VALUE (type));
1272 else
1273 assumption = fold_build2 (NE_EXPR, boolean_type_node,
1274 iv0->base, TYPE_MIN_VALUE (type));
1276 if (integer_zerop (assumption))
1277 return false;
1278 if (!integer_nonzerop (assumption))
1279 niter->assumptions = fold_build2 (TRUTH_AND_EXPR, boolean_type_node,
1280 niter->assumptions, assumption);
1283 if (integer_nonzerop (iv0->step))
1285 if (POINTER_TYPE_P (type))
1286 iv1->base = fold_build_pointer_plus_hwi (iv1->base, 1);
1287 else
1288 iv1->base = fold_build2 (PLUS_EXPR, type1, iv1->base,
1289 build_int_cst (type1, 1));
1291 else if (POINTER_TYPE_P (type))
1292 iv0->base = fold_build_pointer_plus_hwi (iv0->base, -1);
1293 else
1294 iv0->base = fold_build2 (MINUS_EXPR, type1,
1295 iv0->base, build_int_cst (type1, 1));
1297 bounds_add (bnds, 1, type1);
1299 return number_of_iterations_lt (type, iv0, iv1, niter, exit_must_be_taken,
1300 bnds);
1303 /* Dumps description of affine induction variable IV to FILE. */
1305 static void
1306 dump_affine_iv (FILE *file, affine_iv *iv)
1308 if (!integer_zerop (iv->step))
1309 fprintf (file, "[");
1311 print_generic_expr (dump_file, iv->base, TDF_SLIM);
1313 if (!integer_zerop (iv->step))
1315 fprintf (file, ", + , ");
1316 print_generic_expr (dump_file, iv->step, TDF_SLIM);
1317 fprintf (file, "]%s", iv->no_overflow ? "(no_overflow)" : "");
1321 /* Determine the number of iterations according to condition (for staying
1322 inside loop) which compares two induction variables using comparison
1323 operator CODE. The induction variable on left side of the comparison
1324 is IV0, the right-hand side is IV1. Both induction variables must have
1325 type TYPE, which must be an integer or pointer type. The steps of the
1326 ivs must be constants (or NULL_TREE, which is interpreted as constant zero).
1328 LOOP is the loop whose number of iterations we are determining.
1330 ONLY_EXIT is true if we are sure this is the only way the loop could be
1331 exited (including possibly non-returning function calls, exceptions, etc.)
1332 -- in this case we can use the information whether the control induction
1333 variables can overflow or not in a more efficient way.
1335 if EVERY_ITERATION is true, we know the test is executed on every iteration.
1337 The results (number of iterations and assumptions as described in
1338 comments at struct tree_niter_desc in tree-ssa-loop.h) are stored to NITER.
1339 Returns false if it fails to determine number of iterations, true if it
1340 was determined (possibly with some assumptions). */
1342 static bool
1343 number_of_iterations_cond (struct loop *loop,
1344 tree type, affine_iv *iv0, enum tree_code code,
1345 affine_iv *iv1, struct tree_niter_desc *niter,
1346 bool only_exit, bool every_iteration)
1348 bool exit_must_be_taken = false, ret;
1349 bounds bnds;
1351 /* If the test is not executed every iteration, wrapping may make the test
1352 to pass again.
1353 TODO: the overflow case can be still used as unreliable estimate of upper
1354 bound. But we have no API to pass it down to number of iterations code
1355 and, at present, it will not use it anyway. */
1356 if (!every_iteration
1357 && (!iv0->no_overflow || !iv1->no_overflow
1358 || code == NE_EXPR || code == EQ_EXPR))
1359 return false;
1361 /* The meaning of these assumptions is this:
1362 if !assumptions
1363 then the rest of information does not have to be valid
1364 if may_be_zero then the loop does not roll, even if
1365 niter != 0. */
1366 niter->assumptions = boolean_true_node;
1367 niter->may_be_zero = boolean_false_node;
1368 niter->niter = NULL_TREE;
1369 niter->max = 0;
1370 niter->bound = NULL_TREE;
1371 niter->cmp = ERROR_MARK;
1373 /* Make < comparison from > ones, and for NE_EXPR comparisons, ensure that
1374 the control variable is on lhs. */
1375 if (code == GE_EXPR || code == GT_EXPR
1376 || (code == NE_EXPR && integer_zerop (iv0->step)))
1378 SWAP (iv0, iv1);
1379 code = swap_tree_comparison (code);
1382 if (POINTER_TYPE_P (type))
1384 /* Comparison of pointers is undefined unless both iv0 and iv1 point
1385 to the same object. If they do, the control variable cannot wrap
1386 (as wrap around the bounds of memory will never return a pointer
1387 that would be guaranteed to point to the same object, even if we
1388 avoid undefined behavior by casting to size_t and back). */
1389 iv0->no_overflow = true;
1390 iv1->no_overflow = true;
1393 /* If the control induction variable does not overflow and the only exit
1394 from the loop is the one that we analyze, we know it must be taken
1395 eventually. */
1396 if (only_exit)
1398 if (!integer_zerop (iv0->step) && iv0->no_overflow)
1399 exit_must_be_taken = true;
1400 else if (!integer_zerop (iv1->step) && iv1->no_overflow)
1401 exit_must_be_taken = true;
1404 /* We can handle the case when neither of the sides of the comparison is
1405 invariant, provided that the test is NE_EXPR. This rarely occurs in
1406 practice, but it is simple enough to manage. */
1407 if (!integer_zerop (iv0->step) && !integer_zerop (iv1->step))
1409 tree step_type = POINTER_TYPE_P (type) ? sizetype : type;
1410 if (code != NE_EXPR)
1411 return false;
1413 iv0->step = fold_binary_to_constant (MINUS_EXPR, step_type,
1414 iv0->step, iv1->step);
1415 iv0->no_overflow = false;
1416 iv1->step = build_int_cst (step_type, 0);
1417 iv1->no_overflow = true;
1420 /* If the result of the comparison is a constant, the loop is weird. More
1421 precise handling would be possible, but the situation is not common enough
1422 to waste time on it. */
1423 if (integer_zerop (iv0->step) && integer_zerop (iv1->step))
1424 return false;
1426 /* Ignore loops of while (i-- < 10) type. */
1427 if (code != NE_EXPR)
1429 if (iv0->step && tree_int_cst_sign_bit (iv0->step))
1430 return false;
1432 if (!integer_zerop (iv1->step) && !tree_int_cst_sign_bit (iv1->step))
1433 return false;
1436 /* If the loop exits immediately, there is nothing to do. */
1437 tree tem = fold_binary (code, boolean_type_node, iv0->base, iv1->base);
1438 if (tem && integer_zerop (tem))
1440 niter->niter = build_int_cst (unsigned_type_for (type), 0);
1441 niter->max = 0;
1442 return true;
1445 /* OK, now we know we have a senseful loop. Handle several cases, depending
1446 on what comparison operator is used. */
1447 bound_difference (loop, iv1->base, iv0->base, &bnds);
1449 if (dump_file && (dump_flags & TDF_DETAILS))
1451 fprintf (dump_file,
1452 "Analyzing # of iterations of loop %d\n", loop->num);
1454 fprintf (dump_file, " exit condition ");
1455 dump_affine_iv (dump_file, iv0);
1456 fprintf (dump_file, " %s ",
1457 code == NE_EXPR ? "!="
1458 : code == LT_EXPR ? "<"
1459 : "<=");
1460 dump_affine_iv (dump_file, iv1);
1461 fprintf (dump_file, "\n");
1463 fprintf (dump_file, " bounds on difference of bases: ");
1464 mpz_out_str (dump_file, 10, bnds.below);
1465 fprintf (dump_file, " ... ");
1466 mpz_out_str (dump_file, 10, bnds.up);
1467 fprintf (dump_file, "\n");
1470 switch (code)
1472 case NE_EXPR:
1473 gcc_assert (integer_zerop (iv1->step));
1474 ret = number_of_iterations_ne (type, iv0, iv1->base, niter,
1475 exit_must_be_taken, &bnds);
1476 break;
1478 case LT_EXPR:
1479 ret = number_of_iterations_lt (type, iv0, iv1, niter, exit_must_be_taken,
1480 &bnds);
1481 break;
1483 case LE_EXPR:
1484 ret = number_of_iterations_le (type, iv0, iv1, niter, exit_must_be_taken,
1485 &bnds);
1486 break;
1488 default:
1489 gcc_unreachable ();
1492 mpz_clear (bnds.up);
1493 mpz_clear (bnds.below);
1495 if (dump_file && (dump_flags & TDF_DETAILS))
1497 if (ret)
1499 fprintf (dump_file, " result:\n");
1500 if (!integer_nonzerop (niter->assumptions))
1502 fprintf (dump_file, " under assumptions ");
1503 print_generic_expr (dump_file, niter->assumptions, TDF_SLIM);
1504 fprintf (dump_file, "\n");
1507 if (!integer_zerop (niter->may_be_zero))
1509 fprintf (dump_file, " zero if ");
1510 print_generic_expr (dump_file, niter->may_be_zero, TDF_SLIM);
1511 fprintf (dump_file, "\n");
1514 fprintf (dump_file, " # of iterations ");
1515 print_generic_expr (dump_file, niter->niter, TDF_SLIM);
1516 fprintf (dump_file, ", bounded by ");
1517 print_decu (niter->max, dump_file);
1518 fprintf (dump_file, "\n");
1520 else
1521 fprintf (dump_file, " failed\n\n");
1523 return ret;
1526 /* Substitute NEW for OLD in EXPR and fold the result. */
1528 static tree
1529 simplify_replace_tree (tree expr, tree old, tree new_tree)
1531 unsigned i, n;
1532 tree ret = NULL_TREE, e, se;
1534 if (!expr)
1535 return NULL_TREE;
1537 /* Do not bother to replace constants. */
1538 if (CONSTANT_CLASS_P (old))
1539 return expr;
1541 if (expr == old
1542 || operand_equal_p (expr, old, 0))
1543 return unshare_expr (new_tree);
1545 if (!EXPR_P (expr))
1546 return expr;
1548 n = TREE_OPERAND_LENGTH (expr);
1549 for (i = 0; i < n; i++)
1551 e = TREE_OPERAND (expr, i);
1552 se = simplify_replace_tree (e, old, new_tree);
1553 if (e == se)
1554 continue;
1556 if (!ret)
1557 ret = copy_node (expr);
1559 TREE_OPERAND (ret, i) = se;
1562 return (ret ? fold (ret) : expr);
1565 /* Expand definitions of ssa names in EXPR as long as they are simple
1566 enough, and return the new expression. If STOP is specified, stop
1567 expanding if EXPR equals to it. */
1569 tree
1570 expand_simple_operations (tree expr, tree stop)
1572 unsigned i, n;
1573 tree ret = NULL_TREE, e, ee, e1;
1574 enum tree_code code;
1575 gimple stmt;
1577 if (expr == NULL_TREE)
1578 return expr;
1580 if (is_gimple_min_invariant (expr))
1581 return expr;
1583 code = TREE_CODE (expr);
1584 if (IS_EXPR_CODE_CLASS (TREE_CODE_CLASS (code)))
1586 n = TREE_OPERAND_LENGTH (expr);
1587 for (i = 0; i < n; i++)
1589 e = TREE_OPERAND (expr, i);
1590 ee = expand_simple_operations (e, stop);
1591 if (e == ee)
1592 continue;
1594 if (!ret)
1595 ret = copy_node (expr);
1597 TREE_OPERAND (ret, i) = ee;
1600 if (!ret)
1601 return expr;
1603 fold_defer_overflow_warnings ();
1604 ret = fold (ret);
1605 fold_undefer_and_ignore_overflow_warnings ();
1606 return ret;
1609 /* Stop if it's not ssa name or the one we don't want to expand. */
1610 if (TREE_CODE (expr) != SSA_NAME || expr == stop)
1611 return expr;
1613 stmt = SSA_NAME_DEF_STMT (expr);
1614 if (gimple_code (stmt) == GIMPLE_PHI)
1616 basic_block src, dest;
1618 if (gimple_phi_num_args (stmt) != 1)
1619 return expr;
1620 e = PHI_ARG_DEF (stmt, 0);
1622 /* Avoid propagating through loop exit phi nodes, which
1623 could break loop-closed SSA form restrictions. */
1624 dest = gimple_bb (stmt);
1625 src = single_pred (dest);
1626 if (TREE_CODE (e) == SSA_NAME
1627 && src->loop_father != dest->loop_father)
1628 return expr;
1630 return expand_simple_operations (e, stop);
1632 if (gimple_code (stmt) != GIMPLE_ASSIGN)
1633 return expr;
1635 /* Avoid expanding to expressions that contain SSA names that need
1636 to take part in abnormal coalescing. */
1637 ssa_op_iter iter;
1638 FOR_EACH_SSA_TREE_OPERAND (e, stmt, iter, SSA_OP_USE)
1639 if (SSA_NAME_OCCURS_IN_ABNORMAL_PHI (e))
1640 return expr;
1642 e = gimple_assign_rhs1 (stmt);
1643 code = gimple_assign_rhs_code (stmt);
1644 if (get_gimple_rhs_class (code) == GIMPLE_SINGLE_RHS)
1646 if (is_gimple_min_invariant (e))
1647 return e;
1649 if (code == SSA_NAME)
1650 return expand_simple_operations (e, stop);
1652 return expr;
1655 switch (code)
1657 CASE_CONVERT:
1658 /* Casts are simple. */
1659 ee = expand_simple_operations (e, stop);
1660 return fold_build1 (code, TREE_TYPE (expr), ee);
1662 case PLUS_EXPR:
1663 case MINUS_EXPR:
1664 if (ANY_INTEGRAL_TYPE_P (TREE_TYPE (expr))
1665 && TYPE_OVERFLOW_TRAPS (TREE_TYPE (expr)))
1666 return expr;
1667 /* Fallthru. */
1668 case POINTER_PLUS_EXPR:
1669 /* And increments and decrements by a constant are simple. */
1670 e1 = gimple_assign_rhs2 (stmt);
1671 if (!is_gimple_min_invariant (e1))
1672 return expr;
1674 ee = expand_simple_operations (e, stop);
1675 return fold_build2 (code, TREE_TYPE (expr), ee, e1);
1677 default:
1678 return expr;
1682 /* Tries to simplify EXPR using the condition COND. Returns the simplified
1683 expression (or EXPR unchanged, if no simplification was possible). */
1685 static tree
1686 tree_simplify_using_condition_1 (tree cond, tree expr)
1688 bool changed;
1689 tree e, te, e0, e1, e2, notcond;
1690 enum tree_code code = TREE_CODE (expr);
1692 if (code == INTEGER_CST)
1693 return expr;
1695 if (code == TRUTH_OR_EXPR
1696 || code == TRUTH_AND_EXPR
1697 || code == COND_EXPR)
1699 changed = false;
1701 e0 = tree_simplify_using_condition_1 (cond, TREE_OPERAND (expr, 0));
1702 if (TREE_OPERAND (expr, 0) != e0)
1703 changed = true;
1705 e1 = tree_simplify_using_condition_1 (cond, TREE_OPERAND (expr, 1));
1706 if (TREE_OPERAND (expr, 1) != e1)
1707 changed = true;
1709 if (code == COND_EXPR)
1711 e2 = tree_simplify_using_condition_1 (cond, TREE_OPERAND (expr, 2));
1712 if (TREE_OPERAND (expr, 2) != e2)
1713 changed = true;
1715 else
1716 e2 = NULL_TREE;
1718 if (changed)
1720 if (code == COND_EXPR)
1721 expr = fold_build3 (code, boolean_type_node, e0, e1, e2);
1722 else
1723 expr = fold_build2 (code, boolean_type_node, e0, e1);
1726 return expr;
1729 /* In case COND is equality, we may be able to simplify EXPR by copy/constant
1730 propagation, and vice versa. Fold does not handle this, since it is
1731 considered too expensive. */
1732 if (TREE_CODE (cond) == EQ_EXPR)
1734 e0 = TREE_OPERAND (cond, 0);
1735 e1 = TREE_OPERAND (cond, 1);
1737 /* We know that e0 == e1. Check whether we cannot simplify expr
1738 using this fact. */
1739 e = simplify_replace_tree (expr, e0, e1);
1740 if (integer_zerop (e) || integer_nonzerop (e))
1741 return e;
1743 e = simplify_replace_tree (expr, e1, e0);
1744 if (integer_zerop (e) || integer_nonzerop (e))
1745 return e;
1747 if (TREE_CODE (expr) == EQ_EXPR)
1749 e0 = TREE_OPERAND (expr, 0);
1750 e1 = TREE_OPERAND (expr, 1);
1752 /* If e0 == e1 (EXPR) implies !COND, then EXPR cannot be true. */
1753 e = simplify_replace_tree (cond, e0, e1);
1754 if (integer_zerop (e))
1755 return e;
1756 e = simplify_replace_tree (cond, e1, e0);
1757 if (integer_zerop (e))
1758 return e;
1760 if (TREE_CODE (expr) == NE_EXPR)
1762 e0 = TREE_OPERAND (expr, 0);
1763 e1 = TREE_OPERAND (expr, 1);
1765 /* If e0 == e1 (!EXPR) implies !COND, then EXPR must be true. */
1766 e = simplify_replace_tree (cond, e0, e1);
1767 if (integer_zerop (e))
1768 return boolean_true_node;
1769 e = simplify_replace_tree (cond, e1, e0);
1770 if (integer_zerop (e))
1771 return boolean_true_node;
1774 te = expand_simple_operations (expr);
1776 /* Check whether COND ==> EXPR. */
1777 notcond = invert_truthvalue (cond);
1778 e = fold_binary (TRUTH_OR_EXPR, boolean_type_node, notcond, te);
1779 if (e && integer_nonzerop (e))
1780 return e;
1782 /* Check whether COND ==> not EXPR. */
1783 e = fold_binary (TRUTH_AND_EXPR, boolean_type_node, cond, te);
1784 if (e && integer_zerop (e))
1785 return e;
1787 return expr;
1790 /* Tries to simplify EXPR using the condition COND. Returns the simplified
1791 expression (or EXPR unchanged, if no simplification was possible).
1792 Wrapper around tree_simplify_using_condition_1 that ensures that chains
1793 of simple operations in definitions of ssa names in COND are expanded,
1794 so that things like casts or incrementing the value of the bound before
1795 the loop do not cause us to fail. */
1797 static tree
1798 tree_simplify_using_condition (tree cond, tree expr)
1800 cond = expand_simple_operations (cond);
1802 return tree_simplify_using_condition_1 (cond, expr);
1805 /* Tries to simplify EXPR using the conditions on entry to LOOP.
1806 Returns the simplified expression (or EXPR unchanged, if no
1807 simplification was possible).*/
1809 static tree
1810 simplify_using_initial_conditions (struct loop *loop, tree expr)
1812 edge e;
1813 basic_block bb;
1814 gimple stmt;
1815 tree cond;
1816 int cnt = 0;
1818 if (TREE_CODE (expr) == INTEGER_CST)
1819 return expr;
1821 /* Limit walking the dominators to avoid quadraticness in
1822 the number of BBs times the number of loops in degenerate
1823 cases. */
1824 for (bb = loop->header;
1825 bb != ENTRY_BLOCK_PTR_FOR_FN (cfun) && cnt < MAX_DOMINATORS_TO_WALK;
1826 bb = get_immediate_dominator (CDI_DOMINATORS, bb))
1828 if (!single_pred_p (bb))
1829 continue;
1830 e = single_pred_edge (bb);
1832 if (!(e->flags & (EDGE_TRUE_VALUE | EDGE_FALSE_VALUE)))
1833 continue;
1835 stmt = last_stmt (e->src);
1836 cond = fold_build2 (gimple_cond_code (stmt),
1837 boolean_type_node,
1838 gimple_cond_lhs (stmt),
1839 gimple_cond_rhs (stmt));
1840 if (e->flags & EDGE_FALSE_VALUE)
1841 cond = invert_truthvalue (cond);
1842 expr = tree_simplify_using_condition (cond, expr);
1843 ++cnt;
1846 return expr;
1849 /* Tries to simplify EXPR using the evolutions of the loop invariants
1850 in the superloops of LOOP. Returns the simplified expression
1851 (or EXPR unchanged, if no simplification was possible). */
1853 static tree
1854 simplify_using_outer_evolutions (struct loop *loop, tree expr)
1856 enum tree_code code = TREE_CODE (expr);
1857 bool changed;
1858 tree e, e0, e1, e2;
1860 if (is_gimple_min_invariant (expr))
1861 return expr;
1863 if (code == TRUTH_OR_EXPR
1864 || code == TRUTH_AND_EXPR
1865 || code == COND_EXPR)
1867 changed = false;
1869 e0 = simplify_using_outer_evolutions (loop, TREE_OPERAND (expr, 0));
1870 if (TREE_OPERAND (expr, 0) != e0)
1871 changed = true;
1873 e1 = simplify_using_outer_evolutions (loop, TREE_OPERAND (expr, 1));
1874 if (TREE_OPERAND (expr, 1) != e1)
1875 changed = true;
1877 if (code == COND_EXPR)
1879 e2 = simplify_using_outer_evolutions (loop, TREE_OPERAND (expr, 2));
1880 if (TREE_OPERAND (expr, 2) != e2)
1881 changed = true;
1883 else
1884 e2 = NULL_TREE;
1886 if (changed)
1888 if (code == COND_EXPR)
1889 expr = fold_build3 (code, boolean_type_node, e0, e1, e2);
1890 else
1891 expr = fold_build2 (code, boolean_type_node, e0, e1);
1894 return expr;
1897 e = instantiate_parameters (loop, expr);
1898 if (is_gimple_min_invariant (e))
1899 return e;
1901 return expr;
1904 /* Returns true if EXIT is the only possible exit from LOOP. */
1906 bool
1907 loop_only_exit_p (const struct loop *loop, const_edge exit)
1909 basic_block *body;
1910 gimple_stmt_iterator bsi;
1911 unsigned i;
1912 gimple call;
1914 if (exit != single_exit (loop))
1915 return false;
1917 body = get_loop_body (loop);
1918 for (i = 0; i < loop->num_nodes; i++)
1920 for (bsi = gsi_start_bb (body[i]); !gsi_end_p (bsi); gsi_next (&bsi))
1922 call = gsi_stmt (bsi);
1923 if (gimple_code (call) != GIMPLE_CALL)
1924 continue;
1926 if (gimple_has_side_effects (call))
1928 free (body);
1929 return false;
1934 free (body);
1935 return true;
1938 /* Stores description of number of iterations of LOOP derived from
1939 EXIT (an exit edge of the LOOP) in NITER. Returns true if some
1940 useful information could be derived (and fields of NITER has
1941 meaning described in comments at struct tree_niter_desc
1942 declaration), false otherwise. If WARN is true and
1943 -Wunsafe-loop-optimizations was given, warn if the optimizer is going to use
1944 potentially unsafe assumptions.
1945 When EVERY_ITERATION is true, only tests that are known to be executed
1946 every iteration are considered (i.e. only test that alone bounds the loop).
1949 bool
1950 number_of_iterations_exit (struct loop *loop, edge exit,
1951 struct tree_niter_desc *niter,
1952 bool warn, bool every_iteration)
1954 gimple last;
1955 gcond *stmt;
1956 tree type;
1957 tree op0, op1;
1958 enum tree_code code;
1959 affine_iv iv0, iv1;
1960 bool safe;
1962 safe = dominated_by_p (CDI_DOMINATORS, loop->latch, exit->src);
1964 if (every_iteration && !safe)
1965 return false;
1967 niter->assumptions = boolean_false_node;
1968 last = last_stmt (exit->src);
1969 if (!last)
1970 return false;
1971 stmt = dyn_cast <gcond *> (last);
1972 if (!stmt)
1973 return false;
1975 /* We want the condition for staying inside loop. */
1976 code = gimple_cond_code (stmt);
1977 if (exit->flags & EDGE_TRUE_VALUE)
1978 code = invert_tree_comparison (code, false);
1980 switch (code)
1982 case GT_EXPR:
1983 case GE_EXPR:
1984 case LT_EXPR:
1985 case LE_EXPR:
1986 case NE_EXPR:
1987 break;
1989 default:
1990 return false;
1993 op0 = gimple_cond_lhs (stmt);
1994 op1 = gimple_cond_rhs (stmt);
1995 type = TREE_TYPE (op0);
1997 if (TREE_CODE (type) != INTEGER_TYPE
1998 && !POINTER_TYPE_P (type))
1999 return false;
2001 if (!simple_iv (loop, loop_containing_stmt (stmt), op0, &iv0, false))
2002 return false;
2003 if (!simple_iv (loop, loop_containing_stmt (stmt), op1, &iv1, false))
2004 return false;
2006 /* We don't want to see undefined signed overflow warnings while
2007 computing the number of iterations. */
2008 fold_defer_overflow_warnings ();
2010 iv0.base = expand_simple_operations (iv0.base);
2011 iv1.base = expand_simple_operations (iv1.base);
2012 if (!number_of_iterations_cond (loop, type, &iv0, code, &iv1, niter,
2013 loop_only_exit_p (loop, exit), safe))
2015 fold_undefer_and_ignore_overflow_warnings ();
2016 return false;
2019 if (optimize >= 3)
2021 niter->assumptions = simplify_using_outer_evolutions (loop,
2022 niter->assumptions);
2023 niter->may_be_zero = simplify_using_outer_evolutions (loop,
2024 niter->may_be_zero);
2025 niter->niter = simplify_using_outer_evolutions (loop, niter->niter);
2028 niter->assumptions
2029 = simplify_using_initial_conditions (loop,
2030 niter->assumptions);
2031 niter->may_be_zero
2032 = simplify_using_initial_conditions (loop,
2033 niter->may_be_zero);
2035 fold_undefer_and_ignore_overflow_warnings ();
2037 /* If NITER has simplified into a constant, update MAX. */
2038 if (TREE_CODE (niter->niter) == INTEGER_CST)
2039 niter->max = wi::to_widest (niter->niter);
2041 if (integer_onep (niter->assumptions))
2042 return true;
2044 /* With -funsafe-loop-optimizations we assume that nothing bad can happen.
2045 But if we can prove that there is overflow or some other source of weird
2046 behavior, ignore the loop even with -funsafe-loop-optimizations. */
2047 if (integer_zerop (niter->assumptions) || !single_exit (loop))
2048 return false;
2050 if (flag_unsafe_loop_optimizations)
2051 niter->assumptions = boolean_true_node;
2053 if (warn)
2055 const char *wording;
2056 location_t loc = gimple_location (stmt);
2058 /* We can provide a more specific warning if one of the operator is
2059 constant and the other advances by +1 or -1. */
2060 if (!integer_zerop (iv1.step)
2061 ? (integer_zerop (iv0.step)
2062 && (integer_onep (iv1.step) || integer_all_onesp (iv1.step)))
2063 : (integer_onep (iv0.step) || integer_all_onesp (iv0.step)))
2064 wording =
2065 flag_unsafe_loop_optimizations
2066 ? N_("assuming that the loop is not infinite")
2067 : N_("cannot optimize possibly infinite loops");
2068 else
2069 wording =
2070 flag_unsafe_loop_optimizations
2071 ? N_("assuming that the loop counter does not overflow")
2072 : N_("cannot optimize loop, the loop counter may overflow");
2074 warning_at ((LOCATION_LINE (loc) > 0) ? loc : input_location,
2075 OPT_Wunsafe_loop_optimizations, "%s", gettext (wording));
2078 return flag_unsafe_loop_optimizations;
2081 /* Try to determine the number of iterations of LOOP. If we succeed,
2082 expression giving number of iterations is returned and *EXIT is
2083 set to the edge from that the information is obtained. Otherwise
2084 chrec_dont_know is returned. */
2086 tree
2087 find_loop_niter (struct loop *loop, edge *exit)
2089 unsigned i;
2090 vec<edge> exits = get_loop_exit_edges (loop);
2091 edge ex;
2092 tree niter = NULL_TREE, aniter;
2093 struct tree_niter_desc desc;
2095 *exit = NULL;
2096 FOR_EACH_VEC_ELT (exits, i, ex)
2098 if (!number_of_iterations_exit (loop, ex, &desc, false))
2099 continue;
2101 if (integer_nonzerop (desc.may_be_zero))
2103 /* We exit in the first iteration through this exit.
2104 We won't find anything better. */
2105 niter = build_int_cst (unsigned_type_node, 0);
2106 *exit = ex;
2107 break;
2110 if (!integer_zerop (desc.may_be_zero))
2111 continue;
2113 aniter = desc.niter;
2115 if (!niter)
2117 /* Nothing recorded yet. */
2118 niter = aniter;
2119 *exit = ex;
2120 continue;
2123 /* Prefer constants, the lower the better. */
2124 if (TREE_CODE (aniter) != INTEGER_CST)
2125 continue;
2127 if (TREE_CODE (niter) != INTEGER_CST)
2129 niter = aniter;
2130 *exit = ex;
2131 continue;
2134 if (tree_int_cst_lt (aniter, niter))
2136 niter = aniter;
2137 *exit = ex;
2138 continue;
2141 exits.release ();
2143 return niter ? niter : chrec_dont_know;
2146 /* Return true if loop is known to have bounded number of iterations. */
2148 bool
2149 finite_loop_p (struct loop *loop)
2151 widest_int nit;
2152 int flags;
2154 if (flag_unsafe_loop_optimizations)
2155 return true;
2156 flags = flags_from_decl_or_type (current_function_decl);
2157 if ((flags & (ECF_CONST|ECF_PURE)) && !(flags & ECF_LOOPING_CONST_OR_PURE))
2159 if (dump_file && (dump_flags & TDF_DETAILS))
2160 fprintf (dump_file, "Found loop %i to be finite: it is within pure or const function.\n",
2161 loop->num);
2162 return true;
2165 if (loop->any_upper_bound
2166 || max_loop_iterations (loop, &nit))
2168 if (dump_file && (dump_flags & TDF_DETAILS))
2169 fprintf (dump_file, "Found loop %i to be finite: upper bound found.\n",
2170 loop->num);
2171 return true;
2173 return false;
2178 Analysis of a number of iterations of a loop by a brute-force evaluation.
2182 /* Bound on the number of iterations we try to evaluate. */
2184 #define MAX_ITERATIONS_TO_TRACK \
2185 ((unsigned) PARAM_VALUE (PARAM_MAX_ITERATIONS_TO_TRACK))
2187 /* Returns the loop phi node of LOOP such that ssa name X is derived from its
2188 result by a chain of operations such that all but exactly one of their
2189 operands are constants. */
2191 static gphi *
2192 chain_of_csts_start (struct loop *loop, tree x)
2194 gimple stmt = SSA_NAME_DEF_STMT (x);
2195 tree use;
2196 basic_block bb = gimple_bb (stmt);
2197 enum tree_code code;
2199 if (!bb
2200 || !flow_bb_inside_loop_p (loop, bb))
2201 return NULL;
2203 if (gimple_code (stmt) == GIMPLE_PHI)
2205 if (bb == loop->header)
2206 return as_a <gphi *> (stmt);
2208 return NULL;
2211 if (gimple_code (stmt) != GIMPLE_ASSIGN
2212 || gimple_assign_rhs_class (stmt) == GIMPLE_TERNARY_RHS)
2213 return NULL;
2215 code = gimple_assign_rhs_code (stmt);
2216 if (gimple_references_memory_p (stmt)
2217 || TREE_CODE_CLASS (code) == tcc_reference
2218 || (code == ADDR_EXPR
2219 && !is_gimple_min_invariant (gimple_assign_rhs1 (stmt))))
2220 return NULL;
2222 use = SINGLE_SSA_TREE_OPERAND (stmt, SSA_OP_USE);
2223 if (use == NULL_TREE)
2224 return NULL;
2226 return chain_of_csts_start (loop, use);
2229 /* Determines whether the expression X is derived from a result of a phi node
2230 in header of LOOP such that
2232 * the derivation of X consists only from operations with constants
2233 * the initial value of the phi node is constant
2234 * the value of the phi node in the next iteration can be derived from the
2235 value in the current iteration by a chain of operations with constants.
2237 If such phi node exists, it is returned, otherwise NULL is returned. */
2239 static gphi *
2240 get_base_for (struct loop *loop, tree x)
2242 gphi *phi;
2243 tree init, next;
2245 if (is_gimple_min_invariant (x))
2246 return NULL;
2248 phi = chain_of_csts_start (loop, x);
2249 if (!phi)
2250 return NULL;
2252 init = PHI_ARG_DEF_FROM_EDGE (phi, loop_preheader_edge (loop));
2253 next = PHI_ARG_DEF_FROM_EDGE (phi, loop_latch_edge (loop));
2255 if (TREE_CODE (next) != SSA_NAME)
2256 return NULL;
2258 if (!is_gimple_min_invariant (init))
2259 return NULL;
2261 if (chain_of_csts_start (loop, next) != phi)
2262 return NULL;
2264 return phi;
2267 /* Given an expression X, then
2269 * if X is NULL_TREE, we return the constant BASE.
2270 * otherwise X is a SSA name, whose value in the considered loop is derived
2271 by a chain of operations with constant from a result of a phi node in
2272 the header of the loop. Then we return value of X when the value of the
2273 result of this phi node is given by the constant BASE. */
2275 static tree
2276 get_val_for (tree x, tree base)
2278 gimple stmt;
2280 gcc_checking_assert (is_gimple_min_invariant (base));
2282 if (!x)
2283 return base;
2285 stmt = SSA_NAME_DEF_STMT (x);
2286 if (gimple_code (stmt) == GIMPLE_PHI)
2287 return base;
2289 gcc_checking_assert (is_gimple_assign (stmt));
2291 /* STMT must be either an assignment of a single SSA name or an
2292 expression involving an SSA name and a constant. Try to fold that
2293 expression using the value for the SSA name. */
2294 if (gimple_assign_ssa_name_copy_p (stmt))
2295 return get_val_for (gimple_assign_rhs1 (stmt), base);
2296 else if (gimple_assign_rhs_class (stmt) == GIMPLE_UNARY_RHS
2297 && TREE_CODE (gimple_assign_rhs1 (stmt)) == SSA_NAME)
2299 return fold_build1 (gimple_assign_rhs_code (stmt),
2300 gimple_expr_type (stmt),
2301 get_val_for (gimple_assign_rhs1 (stmt), base));
2303 else if (gimple_assign_rhs_class (stmt) == GIMPLE_BINARY_RHS)
2305 tree rhs1 = gimple_assign_rhs1 (stmt);
2306 tree rhs2 = gimple_assign_rhs2 (stmt);
2307 if (TREE_CODE (rhs1) == SSA_NAME)
2308 rhs1 = get_val_for (rhs1, base);
2309 else if (TREE_CODE (rhs2) == SSA_NAME)
2310 rhs2 = get_val_for (rhs2, base);
2311 else
2312 gcc_unreachable ();
2313 return fold_build2 (gimple_assign_rhs_code (stmt),
2314 gimple_expr_type (stmt), rhs1, rhs2);
2316 else
2317 gcc_unreachable ();
2321 /* Tries to count the number of iterations of LOOP till it exits by EXIT
2322 by brute force -- i.e. by determining the value of the operands of the
2323 condition at EXIT in first few iterations of the loop (assuming that
2324 these values are constant) and determining the first one in that the
2325 condition is not satisfied. Returns the constant giving the number
2326 of the iterations of LOOP if successful, chrec_dont_know otherwise. */
2328 tree
2329 loop_niter_by_eval (struct loop *loop, edge exit)
2331 tree acnd;
2332 tree op[2], val[2], next[2], aval[2];
2333 gphi *phi;
2334 gimple cond;
2335 unsigned i, j;
2336 enum tree_code cmp;
2338 cond = last_stmt (exit->src);
2339 if (!cond || gimple_code (cond) != GIMPLE_COND)
2340 return chrec_dont_know;
2342 cmp = gimple_cond_code (cond);
2343 if (exit->flags & EDGE_TRUE_VALUE)
2344 cmp = invert_tree_comparison (cmp, false);
2346 switch (cmp)
2348 case EQ_EXPR:
2349 case NE_EXPR:
2350 case GT_EXPR:
2351 case GE_EXPR:
2352 case LT_EXPR:
2353 case LE_EXPR:
2354 op[0] = gimple_cond_lhs (cond);
2355 op[1] = gimple_cond_rhs (cond);
2356 break;
2358 default:
2359 return chrec_dont_know;
2362 for (j = 0; j < 2; j++)
2364 if (is_gimple_min_invariant (op[j]))
2366 val[j] = op[j];
2367 next[j] = NULL_TREE;
2368 op[j] = NULL_TREE;
2370 else
2372 phi = get_base_for (loop, op[j]);
2373 if (!phi)
2374 return chrec_dont_know;
2375 val[j] = PHI_ARG_DEF_FROM_EDGE (phi, loop_preheader_edge (loop));
2376 next[j] = PHI_ARG_DEF_FROM_EDGE (phi, loop_latch_edge (loop));
2380 /* Don't issue signed overflow warnings. */
2381 fold_defer_overflow_warnings ();
2383 for (i = 0; i < MAX_ITERATIONS_TO_TRACK; i++)
2385 for (j = 0; j < 2; j++)
2386 aval[j] = get_val_for (op[j], val[j]);
2388 acnd = fold_binary (cmp, boolean_type_node, aval[0], aval[1]);
2389 if (acnd && integer_zerop (acnd))
2391 fold_undefer_and_ignore_overflow_warnings ();
2392 if (dump_file && (dump_flags & TDF_DETAILS))
2393 fprintf (dump_file,
2394 "Proved that loop %d iterates %d times using brute force.\n",
2395 loop->num, i);
2396 return build_int_cst (unsigned_type_node, i);
2399 for (j = 0; j < 2; j++)
2401 val[j] = get_val_for (next[j], val[j]);
2402 if (!is_gimple_min_invariant (val[j]))
2404 fold_undefer_and_ignore_overflow_warnings ();
2405 return chrec_dont_know;
2410 fold_undefer_and_ignore_overflow_warnings ();
2412 return chrec_dont_know;
2415 /* Finds the exit of the LOOP by that the loop exits after a constant
2416 number of iterations and stores the exit edge to *EXIT. The constant
2417 giving the number of iterations of LOOP is returned. The number of
2418 iterations is determined using loop_niter_by_eval (i.e. by brute force
2419 evaluation). If we are unable to find the exit for that loop_niter_by_eval
2420 determines the number of iterations, chrec_dont_know is returned. */
2422 tree
2423 find_loop_niter_by_eval (struct loop *loop, edge *exit)
2425 unsigned i;
2426 vec<edge> exits = get_loop_exit_edges (loop);
2427 edge ex;
2428 tree niter = NULL_TREE, aniter;
2430 *exit = NULL;
2432 /* Loops with multiple exits are expensive to handle and less important. */
2433 if (!flag_expensive_optimizations
2434 && exits.length () > 1)
2436 exits.release ();
2437 return chrec_dont_know;
2440 FOR_EACH_VEC_ELT (exits, i, ex)
2442 if (!just_once_each_iteration_p (loop, ex->src))
2443 continue;
2445 aniter = loop_niter_by_eval (loop, ex);
2446 if (chrec_contains_undetermined (aniter))
2447 continue;
2449 if (niter
2450 && !tree_int_cst_lt (aniter, niter))
2451 continue;
2453 niter = aniter;
2454 *exit = ex;
2456 exits.release ();
2458 return niter ? niter : chrec_dont_know;
2463 Analysis of upper bounds on number of iterations of a loop.
2467 static widest_int derive_constant_upper_bound_ops (tree, tree,
2468 enum tree_code, tree);
2470 /* Returns a constant upper bound on the value of the right-hand side of
2471 an assignment statement STMT. */
2473 static widest_int
2474 derive_constant_upper_bound_assign (gimple stmt)
2476 enum tree_code code = gimple_assign_rhs_code (stmt);
2477 tree op0 = gimple_assign_rhs1 (stmt);
2478 tree op1 = gimple_assign_rhs2 (stmt);
2480 return derive_constant_upper_bound_ops (TREE_TYPE (gimple_assign_lhs (stmt)),
2481 op0, code, op1);
2484 /* Returns a constant upper bound on the value of expression VAL. VAL
2485 is considered to be unsigned. If its type is signed, its value must
2486 be nonnegative. */
2488 static widest_int
2489 derive_constant_upper_bound (tree val)
2491 enum tree_code code;
2492 tree op0, op1;
2494 extract_ops_from_tree (val, &code, &op0, &op1);
2495 return derive_constant_upper_bound_ops (TREE_TYPE (val), op0, code, op1);
2498 /* Returns a constant upper bound on the value of expression OP0 CODE OP1,
2499 whose type is TYPE. The expression is considered to be unsigned. If
2500 its type is signed, its value must be nonnegative. */
2502 static widest_int
2503 derive_constant_upper_bound_ops (tree type, tree op0,
2504 enum tree_code code, tree op1)
2506 tree subtype, maxt;
2507 widest_int bnd, max, mmax, cst;
2508 gimple stmt;
2510 if (INTEGRAL_TYPE_P (type))
2511 maxt = TYPE_MAX_VALUE (type);
2512 else
2513 maxt = upper_bound_in_type (type, type);
2515 max = wi::to_widest (maxt);
2517 switch (code)
2519 case INTEGER_CST:
2520 return wi::to_widest (op0);
2522 CASE_CONVERT:
2523 subtype = TREE_TYPE (op0);
2524 if (!TYPE_UNSIGNED (subtype)
2525 /* If TYPE is also signed, the fact that VAL is nonnegative implies
2526 that OP0 is nonnegative. */
2527 && TYPE_UNSIGNED (type)
2528 && !tree_expr_nonnegative_p (op0))
2530 /* If we cannot prove that the casted expression is nonnegative,
2531 we cannot establish more useful upper bound than the precision
2532 of the type gives us. */
2533 return max;
2536 /* We now know that op0 is an nonnegative value. Try deriving an upper
2537 bound for it. */
2538 bnd = derive_constant_upper_bound (op0);
2540 /* If the bound does not fit in TYPE, max. value of TYPE could be
2541 attained. */
2542 if (wi::ltu_p (max, bnd))
2543 return max;
2545 return bnd;
2547 case PLUS_EXPR:
2548 case POINTER_PLUS_EXPR:
2549 case MINUS_EXPR:
2550 if (TREE_CODE (op1) != INTEGER_CST
2551 || !tree_expr_nonnegative_p (op0))
2552 return max;
2554 /* Canonicalize to OP0 - CST. Consider CST to be signed, in order to
2555 choose the most logical way how to treat this constant regardless
2556 of the signedness of the type. */
2557 cst = wi::sext (wi::to_widest (op1), TYPE_PRECISION (type));
2558 if (code != MINUS_EXPR)
2559 cst = -cst;
2561 bnd = derive_constant_upper_bound (op0);
2563 if (wi::neg_p (cst))
2565 cst = -cst;
2566 /* Avoid CST == 0x80000... */
2567 if (wi::neg_p (cst))
2568 return max;
2570 /* OP0 + CST. We need to check that
2571 BND <= MAX (type) - CST. */
2573 mmax -= cst;
2574 if (wi::ltu_p (bnd, max))
2575 return max;
2577 return bnd + cst;
2579 else
2581 /* OP0 - CST, where CST >= 0.
2583 If TYPE is signed, we have already verified that OP0 >= 0, and we
2584 know that the result is nonnegative. This implies that
2585 VAL <= BND - CST.
2587 If TYPE is unsigned, we must additionally know that OP0 >= CST,
2588 otherwise the operation underflows.
2591 /* This should only happen if the type is unsigned; however, for
2592 buggy programs that use overflowing signed arithmetics even with
2593 -fno-wrapv, this condition may also be true for signed values. */
2594 if (wi::ltu_p (bnd, cst))
2595 return max;
2597 if (TYPE_UNSIGNED (type))
2599 tree tem = fold_binary (GE_EXPR, boolean_type_node, op0,
2600 wide_int_to_tree (type, cst));
2601 if (!tem || integer_nonzerop (tem))
2602 return max;
2605 bnd -= cst;
2608 return bnd;
2610 case FLOOR_DIV_EXPR:
2611 case EXACT_DIV_EXPR:
2612 if (TREE_CODE (op1) != INTEGER_CST
2613 || tree_int_cst_sign_bit (op1))
2614 return max;
2616 bnd = derive_constant_upper_bound (op0);
2617 return wi::udiv_floor (bnd, wi::to_widest (op1));
2619 case BIT_AND_EXPR:
2620 if (TREE_CODE (op1) != INTEGER_CST
2621 || tree_int_cst_sign_bit (op1))
2622 return max;
2623 return wi::to_widest (op1);
2625 case SSA_NAME:
2626 stmt = SSA_NAME_DEF_STMT (op0);
2627 if (gimple_code (stmt) != GIMPLE_ASSIGN
2628 || gimple_assign_lhs (stmt) != op0)
2629 return max;
2630 return derive_constant_upper_bound_assign (stmt);
2632 default:
2633 return max;
2637 /* Emit a -Waggressive-loop-optimizations warning if needed. */
2639 static void
2640 do_warn_aggressive_loop_optimizations (struct loop *loop,
2641 widest_int i_bound, gimple stmt)
2643 /* Don't warn if the loop doesn't have known constant bound. */
2644 if (!loop->nb_iterations
2645 || TREE_CODE (loop->nb_iterations) != INTEGER_CST
2646 || !warn_aggressive_loop_optimizations
2647 /* To avoid warning multiple times for the same loop,
2648 only start warning when we preserve loops. */
2649 || (cfun->curr_properties & PROP_loops) == 0
2650 /* Only warn once per loop. */
2651 || loop->warned_aggressive_loop_optimizations
2652 /* Only warn if undefined behavior gives us lower estimate than the
2653 known constant bound. */
2654 || wi::cmpu (i_bound, wi::to_widest (loop->nb_iterations)) >= 0
2655 /* And undefined behavior happens unconditionally. */
2656 || !dominated_by_p (CDI_DOMINATORS, loop->latch, gimple_bb (stmt)))
2657 return;
2659 edge e = single_exit (loop);
2660 if (e == NULL)
2661 return;
2663 gimple estmt = last_stmt (e->src);
2664 if (warning_at (gimple_location (stmt), OPT_Waggressive_loop_optimizations,
2665 "iteration %E invokes undefined behavior",
2666 wide_int_to_tree (TREE_TYPE (loop->nb_iterations),
2667 i_bound)))
2668 inform (gimple_location (estmt), "containing loop");
2669 loop->warned_aggressive_loop_optimizations = true;
2672 /* Records that AT_STMT is executed at most BOUND + 1 times in LOOP. IS_EXIT
2673 is true if the loop is exited immediately after STMT, and this exit
2674 is taken at last when the STMT is executed BOUND + 1 times.
2675 REALISTIC is true if BOUND is expected to be close to the real number
2676 of iterations. UPPER is true if we are sure the loop iterates at most
2677 BOUND times. I_BOUND is a widest_int upper estimate on BOUND. */
2679 static void
2680 record_estimate (struct loop *loop, tree bound, const widest_int &i_bound,
2681 gimple at_stmt, bool is_exit, bool realistic, bool upper)
2683 widest_int delta;
2685 if (dump_file && (dump_flags & TDF_DETAILS))
2687 fprintf (dump_file, "Statement %s", is_exit ? "(exit)" : "");
2688 print_gimple_stmt (dump_file, at_stmt, 0, TDF_SLIM);
2689 fprintf (dump_file, " is %sexecuted at most ",
2690 upper ? "" : "probably ");
2691 print_generic_expr (dump_file, bound, TDF_SLIM);
2692 fprintf (dump_file, " (bounded by ");
2693 print_decu (i_bound, dump_file);
2694 fprintf (dump_file, ") + 1 times in loop %d.\n", loop->num);
2697 /* If the I_BOUND is just an estimate of BOUND, it rarely is close to the
2698 real number of iterations. */
2699 if (TREE_CODE (bound) != INTEGER_CST)
2700 realistic = false;
2701 else
2702 gcc_checking_assert (i_bound == wi::to_widest (bound));
2703 if (!upper && !realistic)
2704 return;
2706 /* If we have a guaranteed upper bound, record it in the appropriate
2707 list, unless this is an !is_exit bound (i.e. undefined behavior in
2708 at_stmt) in a loop with known constant number of iterations. */
2709 if (upper
2710 && (is_exit
2711 || loop->nb_iterations == NULL_TREE
2712 || TREE_CODE (loop->nb_iterations) != INTEGER_CST))
2714 struct nb_iter_bound *elt = ggc_alloc<nb_iter_bound> ();
2716 elt->bound = i_bound;
2717 elt->stmt = at_stmt;
2718 elt->is_exit = is_exit;
2719 elt->next = loop->bounds;
2720 loop->bounds = elt;
2723 /* If statement is executed on every path to the loop latch, we can directly
2724 infer the upper bound on the # of iterations of the loop. */
2725 if (!dominated_by_p (CDI_DOMINATORS, loop->latch, gimple_bb (at_stmt)))
2726 return;
2728 /* Update the number of iteration estimates according to the bound.
2729 If at_stmt is an exit then the loop latch is executed at most BOUND times,
2730 otherwise it can be executed BOUND + 1 times. We will lower the estimate
2731 later if such statement must be executed on last iteration */
2732 if (is_exit)
2733 delta = 0;
2734 else
2735 delta = 1;
2736 widest_int new_i_bound = i_bound + delta;
2738 /* If an overflow occurred, ignore the result. */
2739 if (wi::ltu_p (new_i_bound, delta))
2740 return;
2742 if (upper && !is_exit)
2743 do_warn_aggressive_loop_optimizations (loop, new_i_bound, at_stmt);
2744 record_niter_bound (loop, new_i_bound, realistic, upper);
2747 /* Record the estimate on number of iterations of LOOP based on the fact that
2748 the induction variable BASE + STEP * i evaluated in STMT does not wrap and
2749 its values belong to the range <LOW, HIGH>. REALISTIC is true if the
2750 estimated number of iterations is expected to be close to the real one.
2751 UPPER is true if we are sure the induction variable does not wrap. */
2753 static void
2754 record_nonwrapping_iv (struct loop *loop, tree base, tree step, gimple stmt,
2755 tree low, tree high, bool realistic, bool upper)
2757 tree niter_bound, extreme, delta;
2758 tree type = TREE_TYPE (base), unsigned_type;
2759 tree orig_base = base;
2761 if (TREE_CODE (step) != INTEGER_CST || integer_zerop (step))
2762 return;
2764 if (dump_file && (dump_flags & TDF_DETAILS))
2766 fprintf (dump_file, "Induction variable (");
2767 print_generic_expr (dump_file, TREE_TYPE (base), TDF_SLIM);
2768 fprintf (dump_file, ") ");
2769 print_generic_expr (dump_file, base, TDF_SLIM);
2770 fprintf (dump_file, " + ");
2771 print_generic_expr (dump_file, step, TDF_SLIM);
2772 fprintf (dump_file, " * iteration does not wrap in statement ");
2773 print_gimple_stmt (dump_file, stmt, 0, TDF_SLIM);
2774 fprintf (dump_file, " in loop %d.\n", loop->num);
2777 unsigned_type = unsigned_type_for (type);
2778 base = fold_convert (unsigned_type, base);
2779 step = fold_convert (unsigned_type, step);
2781 if (tree_int_cst_sign_bit (step))
2783 wide_int min, max;
2784 extreme = fold_convert (unsigned_type, low);
2785 if (TREE_CODE (orig_base) == SSA_NAME
2786 && TREE_CODE (high) == INTEGER_CST
2787 && INTEGRAL_TYPE_P (TREE_TYPE (orig_base))
2788 && get_range_info (orig_base, &min, &max) == VR_RANGE
2789 && wi::gts_p (high, max))
2790 base = wide_int_to_tree (unsigned_type, max);
2791 else if (TREE_CODE (base) != INTEGER_CST)
2792 base = fold_convert (unsigned_type, high);
2793 delta = fold_build2 (MINUS_EXPR, unsigned_type, base, extreme);
2794 step = fold_build1 (NEGATE_EXPR, unsigned_type, step);
2796 else
2798 wide_int min, max;
2799 extreme = fold_convert (unsigned_type, high);
2800 if (TREE_CODE (orig_base) == SSA_NAME
2801 && TREE_CODE (low) == INTEGER_CST
2802 && INTEGRAL_TYPE_P (TREE_TYPE (orig_base))
2803 && get_range_info (orig_base, &min, &max) == VR_RANGE
2804 && wi::gts_p (min, low))
2805 base = wide_int_to_tree (unsigned_type, min);
2806 else if (TREE_CODE (base) != INTEGER_CST)
2807 base = fold_convert (unsigned_type, low);
2808 delta = fold_build2 (MINUS_EXPR, unsigned_type, extreme, base);
2811 /* STMT is executed at most NITER_BOUND + 1 times, since otherwise the value
2812 would get out of the range. */
2813 niter_bound = fold_build2 (FLOOR_DIV_EXPR, unsigned_type, delta, step);
2814 widest_int max = derive_constant_upper_bound (niter_bound);
2815 record_estimate (loop, niter_bound, max, stmt, false, realistic, upper);
2818 /* Determine information about number of iterations a LOOP from the index
2819 IDX of a data reference accessed in STMT. RELIABLE is true if STMT is
2820 guaranteed to be executed in every iteration of LOOP. Callback for
2821 for_each_index. */
2823 struct ilb_data
2825 struct loop *loop;
2826 gimple stmt;
2829 static bool
2830 idx_infer_loop_bounds (tree base, tree *idx, void *dta)
2832 struct ilb_data *data = (struct ilb_data *) dta;
2833 tree ev, init, step;
2834 tree low, high, type, next;
2835 bool sign, upper = true, at_end = false;
2836 struct loop *loop = data->loop;
2837 bool reliable = true;
2839 if (TREE_CODE (base) != ARRAY_REF)
2840 return true;
2842 /* For arrays at the end of the structure, we are not guaranteed that they
2843 do not really extend over their declared size. However, for arrays of
2844 size greater than one, this is unlikely to be intended. */
2845 if (array_at_struct_end_p (base))
2847 at_end = true;
2848 upper = false;
2851 struct loop *dloop = loop_containing_stmt (data->stmt);
2852 if (!dloop)
2853 return true;
2855 ev = analyze_scalar_evolution (dloop, *idx);
2856 ev = instantiate_parameters (loop, ev);
2857 init = initial_condition (ev);
2858 step = evolution_part_in_loop_num (ev, loop->num);
2860 if (!init
2861 || !step
2862 || TREE_CODE (step) != INTEGER_CST
2863 || integer_zerop (step)
2864 || tree_contains_chrecs (init, NULL)
2865 || chrec_contains_symbols_defined_in_loop (init, loop->num))
2866 return true;
2868 low = array_ref_low_bound (base);
2869 high = array_ref_up_bound (base);
2871 /* The case of nonconstant bounds could be handled, but it would be
2872 complicated. */
2873 if (TREE_CODE (low) != INTEGER_CST
2874 || !high
2875 || TREE_CODE (high) != INTEGER_CST)
2876 return true;
2877 sign = tree_int_cst_sign_bit (step);
2878 type = TREE_TYPE (step);
2880 /* The array of length 1 at the end of a structure most likely extends
2881 beyond its bounds. */
2882 if (at_end
2883 && operand_equal_p (low, high, 0))
2884 return true;
2886 /* In case the relevant bound of the array does not fit in type, or
2887 it does, but bound + step (in type) still belongs into the range of the
2888 array, the index may wrap and still stay within the range of the array
2889 (consider e.g. if the array is indexed by the full range of
2890 unsigned char).
2892 To make things simpler, we require both bounds to fit into type, although
2893 there are cases where this would not be strictly necessary. */
2894 if (!int_fits_type_p (high, type)
2895 || !int_fits_type_p (low, type))
2896 return true;
2897 low = fold_convert (type, low);
2898 high = fold_convert (type, high);
2900 if (sign)
2901 next = fold_binary (PLUS_EXPR, type, low, step);
2902 else
2903 next = fold_binary (PLUS_EXPR, type, high, step);
2905 if (tree_int_cst_compare (low, next) <= 0
2906 && tree_int_cst_compare (next, high) <= 0)
2907 return true;
2909 /* If access is not executed on every iteration, we must ensure that overlow may
2910 not make the access valid later. */
2911 if (!dominated_by_p (CDI_DOMINATORS, loop->latch, gimple_bb (data->stmt))
2912 && scev_probably_wraps_p (initial_condition_in_loop_num (ev, loop->num),
2913 step, data->stmt, loop, true))
2914 reliable = false;
2916 record_nonwrapping_iv (loop, init, step, data->stmt, low, high, reliable, upper);
2917 return true;
2920 /* Determine information about number of iterations a LOOP from the bounds
2921 of arrays in the data reference REF accessed in STMT. RELIABLE is true if
2922 STMT is guaranteed to be executed in every iteration of LOOP.*/
2924 static void
2925 infer_loop_bounds_from_ref (struct loop *loop, gimple stmt, tree ref)
2927 struct ilb_data data;
2929 data.loop = loop;
2930 data.stmt = stmt;
2931 for_each_index (&ref, idx_infer_loop_bounds, &data);
2934 /* Determine information about number of iterations of a LOOP from the way
2935 arrays are used in STMT. RELIABLE is true if STMT is guaranteed to be
2936 executed in every iteration of LOOP. */
2938 static void
2939 infer_loop_bounds_from_array (struct loop *loop, gimple stmt)
2941 if (is_gimple_assign (stmt))
2943 tree op0 = gimple_assign_lhs (stmt);
2944 tree op1 = gimple_assign_rhs1 (stmt);
2946 /* For each memory access, analyze its access function
2947 and record a bound on the loop iteration domain. */
2948 if (REFERENCE_CLASS_P (op0))
2949 infer_loop_bounds_from_ref (loop, stmt, op0);
2951 if (REFERENCE_CLASS_P (op1))
2952 infer_loop_bounds_from_ref (loop, stmt, op1);
2954 else if (is_gimple_call (stmt))
2956 tree arg, lhs;
2957 unsigned i, n = gimple_call_num_args (stmt);
2959 lhs = gimple_call_lhs (stmt);
2960 if (lhs && REFERENCE_CLASS_P (lhs))
2961 infer_loop_bounds_from_ref (loop, stmt, lhs);
2963 for (i = 0; i < n; i++)
2965 arg = gimple_call_arg (stmt, i);
2966 if (REFERENCE_CLASS_P (arg))
2967 infer_loop_bounds_from_ref (loop, stmt, arg);
2972 /* Determine information about number of iterations of a LOOP from the fact
2973 that pointer arithmetics in STMT does not overflow. */
2975 static void
2976 infer_loop_bounds_from_pointer_arith (struct loop *loop, gimple stmt)
2978 tree def, base, step, scev, type, low, high;
2979 tree var, ptr;
2981 if (!is_gimple_assign (stmt)
2982 || gimple_assign_rhs_code (stmt) != POINTER_PLUS_EXPR)
2983 return;
2985 def = gimple_assign_lhs (stmt);
2986 if (TREE_CODE (def) != SSA_NAME)
2987 return;
2989 type = TREE_TYPE (def);
2990 if (!nowrap_type_p (type))
2991 return;
2993 ptr = gimple_assign_rhs1 (stmt);
2994 if (!expr_invariant_in_loop_p (loop, ptr))
2995 return;
2997 var = gimple_assign_rhs2 (stmt);
2998 if (TYPE_PRECISION (type) != TYPE_PRECISION (TREE_TYPE (var)))
2999 return;
3001 scev = instantiate_parameters (loop, analyze_scalar_evolution (loop, def));
3002 if (chrec_contains_undetermined (scev))
3003 return;
3005 base = initial_condition_in_loop_num (scev, loop->num);
3006 step = evolution_part_in_loop_num (scev, loop->num);
3008 if (!base || !step
3009 || TREE_CODE (step) != INTEGER_CST
3010 || tree_contains_chrecs (base, NULL)
3011 || chrec_contains_symbols_defined_in_loop (base, loop->num))
3012 return;
3014 low = lower_bound_in_type (type, type);
3015 high = upper_bound_in_type (type, type);
3017 /* In C, pointer arithmetic p + 1 cannot use a NULL pointer, and p - 1 cannot
3018 produce a NULL pointer. The contrary would mean NULL points to an object,
3019 while NULL is supposed to compare unequal with the address of all objects.
3020 Furthermore, p + 1 cannot produce a NULL pointer and p - 1 cannot use a
3021 NULL pointer since that would mean wrapping, which we assume here not to
3022 happen. So, we can exclude NULL from the valid range of pointer
3023 arithmetic. */
3024 if (flag_delete_null_pointer_checks && int_cst_value (low) == 0)
3025 low = build_int_cstu (TREE_TYPE (low), TYPE_ALIGN_UNIT (TREE_TYPE (type)));
3027 record_nonwrapping_iv (loop, base, step, stmt, low, high, false, true);
3030 /* Determine information about number of iterations of a LOOP from the fact
3031 that signed arithmetics in STMT does not overflow. */
3033 static void
3034 infer_loop_bounds_from_signedness (struct loop *loop, gimple stmt)
3036 tree def, base, step, scev, type, low, high;
3038 if (gimple_code (stmt) != GIMPLE_ASSIGN)
3039 return;
3041 def = gimple_assign_lhs (stmt);
3043 if (TREE_CODE (def) != SSA_NAME)
3044 return;
3046 type = TREE_TYPE (def);
3047 if (!INTEGRAL_TYPE_P (type)
3048 || !TYPE_OVERFLOW_UNDEFINED (type))
3049 return;
3051 scev = instantiate_parameters (loop, analyze_scalar_evolution (loop, def));
3052 if (chrec_contains_undetermined (scev))
3053 return;
3055 base = initial_condition_in_loop_num (scev, loop->num);
3056 step = evolution_part_in_loop_num (scev, loop->num);
3058 if (!base || !step
3059 || TREE_CODE (step) != INTEGER_CST
3060 || tree_contains_chrecs (base, NULL)
3061 || chrec_contains_symbols_defined_in_loop (base, loop->num))
3062 return;
3064 low = lower_bound_in_type (type, type);
3065 high = upper_bound_in_type (type, type);
3067 record_nonwrapping_iv (loop, base, step, stmt, low, high, false, true);
3070 /* The following analyzers are extracting informations on the bounds
3071 of LOOP from the following undefined behaviors:
3073 - data references should not access elements over the statically
3074 allocated size,
3076 - signed variables should not overflow when flag_wrapv is not set.
3079 static void
3080 infer_loop_bounds_from_undefined (struct loop *loop)
3082 unsigned i;
3083 basic_block *bbs;
3084 gimple_stmt_iterator bsi;
3085 basic_block bb;
3086 bool reliable;
3088 bbs = get_loop_body (loop);
3090 for (i = 0; i < loop->num_nodes; i++)
3092 bb = bbs[i];
3094 /* If BB is not executed in each iteration of the loop, we cannot
3095 use the operations in it to infer reliable upper bound on the
3096 # of iterations of the loop. However, we can use it as a guess.
3097 Reliable guesses come only from array bounds. */
3098 reliable = dominated_by_p (CDI_DOMINATORS, loop->latch, bb);
3100 for (bsi = gsi_start_bb (bb); !gsi_end_p (bsi); gsi_next (&bsi))
3102 gimple stmt = gsi_stmt (bsi);
3104 infer_loop_bounds_from_array (loop, stmt);
3106 if (reliable)
3108 infer_loop_bounds_from_signedness (loop, stmt);
3109 infer_loop_bounds_from_pointer_arith (loop, stmt);
3115 free (bbs);
3118 /* Compare wide ints, callback for qsort. */
3120 static int
3121 wide_int_cmp (const void *p1, const void *p2)
3123 const widest_int *d1 = (const widest_int *) p1;
3124 const widest_int *d2 = (const widest_int *) p2;
3125 return wi::cmpu (*d1, *d2);
3128 /* Return index of BOUND in BOUNDS array sorted in increasing order.
3129 Lookup by binary search. */
3131 static int
3132 bound_index (vec<widest_int> bounds, const widest_int &bound)
3134 unsigned int end = bounds.length ();
3135 unsigned int begin = 0;
3137 /* Find a matching index by means of a binary search. */
3138 while (begin != end)
3140 unsigned int middle = (begin + end) / 2;
3141 widest_int index = bounds[middle];
3143 if (index == bound)
3144 return middle;
3145 else if (wi::ltu_p (index, bound))
3146 begin = middle + 1;
3147 else
3148 end = middle;
3150 gcc_unreachable ();
3153 /* We recorded loop bounds only for statements dominating loop latch (and thus
3154 executed each loop iteration). If there are any bounds on statements not
3155 dominating the loop latch we can improve the estimate by walking the loop
3156 body and seeing if every path from loop header to loop latch contains
3157 some bounded statement. */
3159 static void
3160 discover_iteration_bound_by_body_walk (struct loop *loop)
3162 struct nb_iter_bound *elt;
3163 vec<widest_int> bounds = vNULL;
3164 vec<vec<basic_block> > queues = vNULL;
3165 vec<basic_block> queue = vNULL;
3166 ptrdiff_t queue_index;
3167 ptrdiff_t latch_index = 0;
3169 /* Discover what bounds may interest us. */
3170 for (elt = loop->bounds; elt; elt = elt->next)
3172 widest_int bound = elt->bound;
3174 /* Exit terminates loop at given iteration, while non-exits produce undefined
3175 effect on the next iteration. */
3176 if (!elt->is_exit)
3178 bound += 1;
3179 /* If an overflow occurred, ignore the result. */
3180 if (bound == 0)
3181 continue;
3184 if (!loop->any_upper_bound
3185 || wi::ltu_p (bound, loop->nb_iterations_upper_bound))
3186 bounds.safe_push (bound);
3189 /* Exit early if there is nothing to do. */
3190 if (!bounds.exists ())
3191 return;
3193 if (dump_file && (dump_flags & TDF_DETAILS))
3194 fprintf (dump_file, " Trying to walk loop body to reduce the bound.\n");
3196 /* Sort the bounds in decreasing order. */
3197 bounds.qsort (wide_int_cmp);
3199 /* For every basic block record the lowest bound that is guaranteed to
3200 terminate the loop. */
3202 hash_map<basic_block, ptrdiff_t> bb_bounds;
3203 for (elt = loop->bounds; elt; elt = elt->next)
3205 widest_int bound = elt->bound;
3206 if (!elt->is_exit)
3208 bound += 1;
3209 /* If an overflow occurred, ignore the result. */
3210 if (bound == 0)
3211 continue;
3214 if (!loop->any_upper_bound
3215 || wi::ltu_p (bound, loop->nb_iterations_upper_bound))
3217 ptrdiff_t index = bound_index (bounds, bound);
3218 ptrdiff_t *entry = bb_bounds.get (gimple_bb (elt->stmt));
3219 if (!entry)
3220 bb_bounds.put (gimple_bb (elt->stmt), index);
3221 else if ((ptrdiff_t)*entry > index)
3222 *entry = index;
3226 hash_map<basic_block, ptrdiff_t> block_priority;
3228 /* Perform shortest path discovery loop->header ... loop->latch.
3230 The "distance" is given by the smallest loop bound of basic block
3231 present in the path and we look for path with largest smallest bound
3232 on it.
3234 To avoid the need for fibonacci heap on double ints we simply compress
3235 double ints into indexes to BOUNDS array and then represent the queue
3236 as arrays of queues for every index.
3237 Index of BOUNDS.length() means that the execution of given BB has
3238 no bounds determined.
3240 VISITED is a pointer map translating basic block into smallest index
3241 it was inserted into the priority queue with. */
3242 latch_index = -1;
3244 /* Start walk in loop header with index set to infinite bound. */
3245 queue_index = bounds.length ();
3246 queues.safe_grow_cleared (queue_index + 1);
3247 queue.safe_push (loop->header);
3248 queues[queue_index] = queue;
3249 block_priority.put (loop->header, queue_index);
3251 for (; queue_index >= 0; queue_index--)
3253 if (latch_index < queue_index)
3255 while (queues[queue_index].length ())
3257 basic_block bb;
3258 ptrdiff_t bound_index = queue_index;
3259 edge e;
3260 edge_iterator ei;
3262 queue = queues[queue_index];
3263 bb = queue.pop ();
3265 /* OK, we later inserted the BB with lower priority, skip it. */
3266 if (*block_priority.get (bb) > queue_index)
3267 continue;
3269 /* See if we can improve the bound. */
3270 ptrdiff_t *entry = bb_bounds.get (bb);
3271 if (entry && *entry < bound_index)
3272 bound_index = *entry;
3274 /* Insert succesors into the queue, watch for latch edge
3275 and record greatest index we saw. */
3276 FOR_EACH_EDGE (e, ei, bb->succs)
3278 bool insert = false;
3280 if (loop_exit_edge_p (loop, e))
3281 continue;
3283 if (e == loop_latch_edge (loop)
3284 && latch_index < bound_index)
3285 latch_index = bound_index;
3286 else if (!(entry = block_priority.get (e->dest)))
3288 insert = true;
3289 block_priority.put (e->dest, bound_index);
3291 else if (*entry < bound_index)
3293 insert = true;
3294 *entry = bound_index;
3297 if (insert)
3298 queues[bound_index].safe_push (e->dest);
3302 queues[queue_index].release ();
3305 gcc_assert (latch_index >= 0);
3306 if ((unsigned)latch_index < bounds.length ())
3308 if (dump_file && (dump_flags & TDF_DETAILS))
3310 fprintf (dump_file, "Found better loop bound ");
3311 print_decu (bounds[latch_index], dump_file);
3312 fprintf (dump_file, "\n");
3314 record_niter_bound (loop, bounds[latch_index], false, true);
3317 queues.release ();
3318 bounds.release ();
3321 /* See if every path cross the loop goes through a statement that is known
3322 to not execute at the last iteration. In that case we can decrese iteration
3323 count by 1. */
3325 static void
3326 maybe_lower_iteration_bound (struct loop *loop)
3328 hash_set<gimple> *not_executed_last_iteration = NULL;
3329 struct nb_iter_bound *elt;
3330 bool found_exit = false;
3331 vec<basic_block> queue = vNULL;
3332 bitmap visited;
3334 /* Collect all statements with interesting (i.e. lower than
3335 nb_iterations_upper_bound) bound on them.
3337 TODO: Due to the way record_estimate choose estimates to store, the bounds
3338 will be always nb_iterations_upper_bound-1. We can change this to record
3339 also statements not dominating the loop latch and update the walk bellow
3340 to the shortest path algorthm. */
3341 for (elt = loop->bounds; elt; elt = elt->next)
3343 if (!elt->is_exit
3344 && wi::ltu_p (elt->bound, loop->nb_iterations_upper_bound))
3346 if (!not_executed_last_iteration)
3347 not_executed_last_iteration = new hash_set<gimple>;
3348 not_executed_last_iteration->add (elt->stmt);
3351 if (!not_executed_last_iteration)
3352 return;
3354 /* Start DFS walk in the loop header and see if we can reach the
3355 loop latch or any of the exits (including statements with side
3356 effects that may terminate the loop otherwise) without visiting
3357 any of the statements known to have undefined effect on the last
3358 iteration. */
3359 queue.safe_push (loop->header);
3360 visited = BITMAP_ALLOC (NULL);
3361 bitmap_set_bit (visited, loop->header->index);
3362 found_exit = false;
3366 basic_block bb = queue.pop ();
3367 gimple_stmt_iterator gsi;
3368 bool stmt_found = false;
3370 /* Loop for possible exits and statements bounding the execution. */
3371 for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
3373 gimple stmt = gsi_stmt (gsi);
3374 if (not_executed_last_iteration->contains (stmt))
3376 stmt_found = true;
3377 break;
3379 if (gimple_has_side_effects (stmt))
3381 found_exit = true;
3382 break;
3385 if (found_exit)
3386 break;
3388 /* If no bounding statement is found, continue the walk. */
3389 if (!stmt_found)
3391 edge e;
3392 edge_iterator ei;
3394 FOR_EACH_EDGE (e, ei, bb->succs)
3396 if (loop_exit_edge_p (loop, e)
3397 || e == loop_latch_edge (loop))
3399 found_exit = true;
3400 break;
3402 if (bitmap_set_bit (visited, e->dest->index))
3403 queue.safe_push (e->dest);
3407 while (queue.length () && !found_exit);
3409 /* If every path through the loop reach bounding statement before exit,
3410 then we know the last iteration of the loop will have undefined effect
3411 and we can decrease number of iterations. */
3413 if (!found_exit)
3415 if (dump_file && (dump_flags & TDF_DETAILS))
3416 fprintf (dump_file, "Reducing loop iteration estimate by 1; "
3417 "undefined statement must be executed at the last iteration.\n");
3418 record_niter_bound (loop, loop->nb_iterations_upper_bound - 1,
3419 false, true);
3422 BITMAP_FREE (visited);
3423 queue.release ();
3424 delete not_executed_last_iteration;
3427 /* Records estimates on numbers of iterations of LOOP. If USE_UNDEFINED_P
3428 is true also use estimates derived from undefined behavior. */
3430 static void
3431 estimate_numbers_of_iterations_loop (struct loop *loop)
3433 vec<edge> exits;
3434 tree niter, type;
3435 unsigned i;
3436 struct tree_niter_desc niter_desc;
3437 edge ex;
3438 widest_int bound;
3439 edge likely_exit;
3441 /* Give up if we already have tried to compute an estimation. */
3442 if (loop->estimate_state != EST_NOT_COMPUTED)
3443 return;
3445 loop->estimate_state = EST_AVAILABLE;
3446 /* Force estimate compuation but leave any existing upper bound in place. */
3447 loop->any_estimate = false;
3449 /* Ensure that loop->nb_iterations is computed if possible. If it turns out
3450 to be constant, we avoid undefined behavior implied bounds and instead
3451 diagnose those loops with -Waggressive-loop-optimizations. */
3452 number_of_latch_executions (loop);
3454 exits = get_loop_exit_edges (loop);
3455 likely_exit = single_likely_exit (loop);
3456 FOR_EACH_VEC_ELT (exits, i, ex)
3458 if (!number_of_iterations_exit (loop, ex, &niter_desc, false, false))
3459 continue;
3461 niter = niter_desc.niter;
3462 type = TREE_TYPE (niter);
3463 if (TREE_CODE (niter_desc.may_be_zero) != INTEGER_CST)
3464 niter = build3 (COND_EXPR, type, niter_desc.may_be_zero,
3465 build_int_cst (type, 0),
3466 niter);
3467 record_estimate (loop, niter, niter_desc.max,
3468 last_stmt (ex->src),
3469 true, ex == likely_exit, true);
3471 exits.release ();
3473 if (flag_aggressive_loop_optimizations)
3474 infer_loop_bounds_from_undefined (loop);
3476 discover_iteration_bound_by_body_walk (loop);
3478 maybe_lower_iteration_bound (loop);
3480 /* If we have a measured profile, use it to estimate the number of
3481 iterations. */
3482 if (loop->header->count != 0)
3484 gcov_type nit = expected_loop_iterations_unbounded (loop) + 1;
3485 bound = gcov_type_to_wide_int (nit);
3486 record_niter_bound (loop, bound, true, false);
3489 /* If we know the exact number of iterations of this loop, try to
3490 not break code with undefined behavior by not recording smaller
3491 maximum number of iterations. */
3492 if (loop->nb_iterations
3493 && TREE_CODE (loop->nb_iterations) == INTEGER_CST)
3495 loop->any_upper_bound = true;
3496 loop->nb_iterations_upper_bound = wi::to_widest (loop->nb_iterations);
3500 /* Sets NIT to the estimated number of executions of the latch of the
3501 LOOP. If CONSERVATIVE is true, we must be sure that NIT is at least as
3502 large as the number of iterations. If we have no reliable estimate,
3503 the function returns false, otherwise returns true. */
3505 bool
3506 estimated_loop_iterations (struct loop *loop, widest_int *nit)
3508 /* When SCEV information is available, try to update loop iterations
3509 estimate. Otherwise just return whatever we recorded earlier. */
3510 if (scev_initialized_p ())
3511 estimate_numbers_of_iterations_loop (loop);
3513 return (get_estimated_loop_iterations (loop, nit));
3516 /* Similar to estimated_loop_iterations, but returns the estimate only
3517 if it fits to HOST_WIDE_INT. If this is not the case, or the estimate
3518 on the number of iterations of LOOP could not be derived, returns -1. */
3520 HOST_WIDE_INT
3521 estimated_loop_iterations_int (struct loop *loop)
3523 widest_int nit;
3524 HOST_WIDE_INT hwi_nit;
3526 if (!estimated_loop_iterations (loop, &nit))
3527 return -1;
3529 if (!wi::fits_shwi_p (nit))
3530 return -1;
3531 hwi_nit = nit.to_shwi ();
3533 return hwi_nit < 0 ? -1 : hwi_nit;
3537 /* Sets NIT to an upper bound for the maximum number of executions of the
3538 latch of the LOOP. If we have no reliable estimate, the function returns
3539 false, otherwise returns true. */
3541 bool
3542 max_loop_iterations (struct loop *loop, widest_int *nit)
3544 /* When SCEV information is available, try to update loop iterations
3545 estimate. Otherwise just return whatever we recorded earlier. */
3546 if (scev_initialized_p ())
3547 estimate_numbers_of_iterations_loop (loop);
3549 return get_max_loop_iterations (loop, nit);
3552 /* Similar to max_loop_iterations, but returns the estimate only
3553 if it fits to HOST_WIDE_INT. If this is not the case, or the estimate
3554 on the number of iterations of LOOP could not be derived, returns -1. */
3556 HOST_WIDE_INT
3557 max_loop_iterations_int (struct loop *loop)
3559 widest_int nit;
3560 HOST_WIDE_INT hwi_nit;
3562 if (!max_loop_iterations (loop, &nit))
3563 return -1;
3565 if (!wi::fits_shwi_p (nit))
3566 return -1;
3567 hwi_nit = nit.to_shwi ();
3569 return hwi_nit < 0 ? -1 : hwi_nit;
3572 /* Returns an estimate for the number of executions of statements
3573 in the LOOP. For statements before the loop exit, this exceeds
3574 the number of execution of the latch by one. */
3576 HOST_WIDE_INT
3577 estimated_stmt_executions_int (struct loop *loop)
3579 HOST_WIDE_INT nit = estimated_loop_iterations_int (loop);
3580 HOST_WIDE_INT snit;
3582 if (nit == -1)
3583 return -1;
3585 snit = (HOST_WIDE_INT) ((unsigned HOST_WIDE_INT) nit + 1);
3587 /* If the computation overflows, return -1. */
3588 return snit < 0 ? -1 : snit;
3591 /* Sets NIT to the estimated maximum number of executions of the latch of the
3592 LOOP, plus one. If we have no reliable estimate, the function returns
3593 false, otherwise returns true. */
3595 bool
3596 max_stmt_executions (struct loop *loop, widest_int *nit)
3598 widest_int nit_minus_one;
3600 if (!max_loop_iterations (loop, nit))
3601 return false;
3603 nit_minus_one = *nit;
3605 *nit += 1;
3607 return wi::gtu_p (*nit, nit_minus_one);
3610 /* Sets NIT to the estimated number of executions of the latch of the
3611 LOOP, plus one. If we have no reliable estimate, the function returns
3612 false, otherwise returns true. */
3614 bool
3615 estimated_stmt_executions (struct loop *loop, widest_int *nit)
3617 widest_int nit_minus_one;
3619 if (!estimated_loop_iterations (loop, nit))
3620 return false;
3622 nit_minus_one = *nit;
3624 *nit += 1;
3626 return wi::gtu_p (*nit, nit_minus_one);
3629 /* Records estimates on numbers of iterations of loops. */
3631 void
3632 estimate_numbers_of_iterations (void)
3634 struct loop *loop;
3636 /* We don't want to issue signed overflow warnings while getting
3637 loop iteration estimates. */
3638 fold_defer_overflow_warnings ();
3640 FOR_EACH_LOOP (loop, 0)
3642 estimate_numbers_of_iterations_loop (loop);
3645 fold_undefer_and_ignore_overflow_warnings ();
3648 /* Returns true if statement S1 dominates statement S2. */
3650 bool
3651 stmt_dominates_stmt_p (gimple s1, gimple s2)
3653 basic_block bb1 = gimple_bb (s1), bb2 = gimple_bb (s2);
3655 if (!bb1
3656 || s1 == s2)
3657 return true;
3659 if (bb1 == bb2)
3661 gimple_stmt_iterator bsi;
3663 if (gimple_code (s2) == GIMPLE_PHI)
3664 return false;
3666 if (gimple_code (s1) == GIMPLE_PHI)
3667 return true;
3669 for (bsi = gsi_start_bb (bb1); gsi_stmt (bsi) != s2; gsi_next (&bsi))
3670 if (gsi_stmt (bsi) == s1)
3671 return true;
3673 return false;
3676 return dominated_by_p (CDI_DOMINATORS, bb2, bb1);
3679 /* Returns true when we can prove that the number of executions of
3680 STMT in the loop is at most NITER, according to the bound on
3681 the number of executions of the statement NITER_BOUND->stmt recorded in
3682 NITER_BOUND and fact that NITER_BOUND->stmt dominate STMT.
3684 ??? This code can become quite a CPU hog - we can have many bounds,
3685 and large basic block forcing stmt_dominates_stmt_p to be queried
3686 many times on a large basic blocks, so the whole thing is O(n^2)
3687 for scev_probably_wraps_p invocation (that can be done n times).
3689 It would make more sense (and give better answers) to remember BB
3690 bounds computed by discover_iteration_bound_by_body_walk. */
3692 static bool
3693 n_of_executions_at_most (gimple stmt,
3694 struct nb_iter_bound *niter_bound,
3695 tree niter)
3697 widest_int bound = niter_bound->bound;
3698 tree nit_type = TREE_TYPE (niter), e;
3699 enum tree_code cmp;
3701 gcc_assert (TYPE_UNSIGNED (nit_type));
3703 /* If the bound does not even fit into NIT_TYPE, it cannot tell us that
3704 the number of iterations is small. */
3705 if (!wi::fits_to_tree_p (bound, nit_type))
3706 return false;
3708 /* We know that NITER_BOUND->stmt is executed at most NITER_BOUND->bound + 1
3709 times. This means that:
3711 -- if NITER_BOUND->is_exit is true, then everything after
3712 it at most NITER_BOUND->bound times.
3714 -- If NITER_BOUND->is_exit is false, then if we can prove that when STMT
3715 is executed, then NITER_BOUND->stmt is executed as well in the same
3716 iteration then STMT is executed at most NITER_BOUND->bound + 1 times.
3718 If we can determine that NITER_BOUND->stmt is always executed
3719 after STMT, then STMT is executed at most NITER_BOUND->bound + 2 times.
3720 We conclude that if both statements belong to the same
3721 basic block and STMT is before NITER_BOUND->stmt and there are no
3722 statements with side effects in between. */
3724 if (niter_bound->is_exit)
3726 if (stmt == niter_bound->stmt
3727 || !stmt_dominates_stmt_p (niter_bound->stmt, stmt))
3728 return false;
3729 cmp = GE_EXPR;
3731 else
3733 if (!stmt_dominates_stmt_p (niter_bound->stmt, stmt))
3735 gimple_stmt_iterator bsi;
3736 if (gimple_bb (stmt) != gimple_bb (niter_bound->stmt)
3737 || gimple_code (stmt) == GIMPLE_PHI
3738 || gimple_code (niter_bound->stmt) == GIMPLE_PHI)
3739 return false;
3741 /* By stmt_dominates_stmt_p we already know that STMT appears
3742 before NITER_BOUND->STMT. Still need to test that the loop
3743 can not be terinated by a side effect in between. */
3744 for (bsi = gsi_for_stmt (stmt); gsi_stmt (bsi) != niter_bound->stmt;
3745 gsi_next (&bsi))
3746 if (gimple_has_side_effects (gsi_stmt (bsi)))
3747 return false;
3748 bound += 1;
3749 if (bound == 0
3750 || !wi::fits_to_tree_p (bound, nit_type))
3751 return false;
3753 cmp = GT_EXPR;
3756 e = fold_binary (cmp, boolean_type_node,
3757 niter, wide_int_to_tree (nit_type, bound));
3758 return e && integer_nonzerop (e);
3761 /* Returns true if the arithmetics in TYPE can be assumed not to wrap. */
3763 bool
3764 nowrap_type_p (tree type)
3766 if (INTEGRAL_TYPE_P (type)
3767 && TYPE_OVERFLOW_UNDEFINED (type))
3768 return true;
3770 if (POINTER_TYPE_P (type))
3771 return true;
3773 return false;
3776 /* Return false only when the induction variable BASE + STEP * I is
3777 known to not overflow: i.e. when the number of iterations is small
3778 enough with respect to the step and initial condition in order to
3779 keep the evolution confined in TYPEs bounds. Return true when the
3780 iv is known to overflow or when the property is not computable.
3782 USE_OVERFLOW_SEMANTICS is true if this function should assume that
3783 the rules for overflow of the given language apply (e.g., that signed
3784 arithmetics in C does not overflow). */
3786 bool
3787 scev_probably_wraps_p (tree base, tree step,
3788 gimple at_stmt, struct loop *loop,
3789 bool use_overflow_semantics)
3791 tree delta, step_abs;
3792 tree unsigned_type, valid_niter;
3793 tree type = TREE_TYPE (step);
3794 tree e;
3795 widest_int niter;
3796 struct nb_iter_bound *bound;
3798 /* FIXME: We really need something like
3799 http://gcc.gnu.org/ml/gcc-patches/2005-06/msg02025.html.
3801 We used to test for the following situation that frequently appears
3802 during address arithmetics:
3804 D.1621_13 = (long unsigned intD.4) D.1620_12;
3805 D.1622_14 = D.1621_13 * 8;
3806 D.1623_15 = (doubleD.29 *) D.1622_14;
3808 And derived that the sequence corresponding to D_14
3809 can be proved to not wrap because it is used for computing a
3810 memory access; however, this is not really the case -- for example,
3811 if D_12 = (unsigned char) [254,+,1], then D_14 has values
3812 2032, 2040, 0, 8, ..., but the code is still legal. */
3814 if (chrec_contains_undetermined (base)
3815 || chrec_contains_undetermined (step))
3816 return true;
3818 if (integer_zerop (step))
3819 return false;
3821 /* If we can use the fact that signed and pointer arithmetics does not
3822 wrap, we are done. */
3823 if (use_overflow_semantics && nowrap_type_p (TREE_TYPE (base)))
3824 return false;
3826 /* To be able to use estimates on number of iterations of the loop,
3827 we must have an upper bound on the absolute value of the step. */
3828 if (TREE_CODE (step) != INTEGER_CST)
3829 return true;
3831 /* Don't issue signed overflow warnings. */
3832 fold_defer_overflow_warnings ();
3834 /* Otherwise, compute the number of iterations before we reach the
3835 bound of the type, and verify that the loop is exited before this
3836 occurs. */
3837 unsigned_type = unsigned_type_for (type);
3838 base = fold_convert (unsigned_type, base);
3840 if (tree_int_cst_sign_bit (step))
3842 tree extreme = fold_convert (unsigned_type,
3843 lower_bound_in_type (type, type));
3844 delta = fold_build2 (MINUS_EXPR, unsigned_type, base, extreme);
3845 step_abs = fold_build1 (NEGATE_EXPR, unsigned_type,
3846 fold_convert (unsigned_type, step));
3848 else
3850 tree extreme = fold_convert (unsigned_type,
3851 upper_bound_in_type (type, type));
3852 delta = fold_build2 (MINUS_EXPR, unsigned_type, extreme, base);
3853 step_abs = fold_convert (unsigned_type, step);
3856 valid_niter = fold_build2 (FLOOR_DIV_EXPR, unsigned_type, delta, step_abs);
3858 estimate_numbers_of_iterations_loop (loop);
3860 if (max_loop_iterations (loop, &niter)
3861 && wi::fits_to_tree_p (niter, TREE_TYPE (valid_niter))
3862 && (e = fold_binary (GT_EXPR, boolean_type_node, valid_niter,
3863 wide_int_to_tree (TREE_TYPE (valid_niter),
3864 niter))) != NULL
3865 && integer_nonzerop (e))
3867 fold_undefer_and_ignore_overflow_warnings ();
3868 return false;
3870 if (at_stmt)
3871 for (bound = loop->bounds; bound; bound = bound->next)
3873 if (n_of_executions_at_most (at_stmt, bound, valid_niter))
3875 fold_undefer_and_ignore_overflow_warnings ();
3876 return false;
3880 fold_undefer_and_ignore_overflow_warnings ();
3882 /* At this point we still don't have a proof that the iv does not
3883 overflow: give up. */
3884 return true;
3887 /* Frees the information on upper bounds on numbers of iterations of LOOP. */
3889 void
3890 free_numbers_of_iterations_estimates_loop (struct loop *loop)
3892 struct nb_iter_bound *bound, *next;
3894 loop->nb_iterations = NULL;
3895 loop->estimate_state = EST_NOT_COMPUTED;
3896 for (bound = loop->bounds; bound; bound = next)
3898 next = bound->next;
3899 ggc_free (bound);
3902 loop->bounds = NULL;
3905 /* Frees the information on upper bounds on numbers of iterations of loops. */
3907 void
3908 free_numbers_of_iterations_estimates (void)
3910 struct loop *loop;
3912 FOR_EACH_LOOP (loop, 0)
3914 free_numbers_of_iterations_estimates_loop (loop);
3918 /* Substitute value VAL for ssa name NAME inside expressions held
3919 at LOOP. */
3921 void
3922 substitute_in_loop_info (struct loop *loop, tree name, tree val)
3924 loop->nb_iterations = simplify_replace_tree (loop->nb_iterations, name, val);