IBM Z: Fix usage of "f" constraint with long doubles
[official-gcc.git] / gcc / tree-vrp.c
blob62b90079a03f3ca1de8c5f51d7105901995b00da
1 /* Support routines for Value Range Propagation (VRP).
2 Copyright (C) 2005-2021 Free Software Foundation, Inc.
3 Contributed by Diego Novillo <dnovillo@redhat.com>.
5 This file is part of GCC.
7 GCC is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3, or (at your option)
10 any later version.
12 GCC is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with GCC; see the file COPYING3. If not see
19 <http://www.gnu.org/licenses/>. */
21 #include "config.h"
22 #include "system.h"
23 #include "coretypes.h"
24 #include "backend.h"
25 #include "insn-codes.h"
26 #include "rtl.h"
27 #include "tree.h"
28 #include "gimple.h"
29 #include "cfghooks.h"
30 #include "tree-pass.h"
31 #include "ssa.h"
32 #include "optabs-tree.h"
33 #include "gimple-pretty-print.h"
34 #include "flags.h"
35 #include "fold-const.h"
36 #include "stor-layout.h"
37 #include "calls.h"
38 #include "cfganal.h"
39 #include "gimple-fold.h"
40 #include "tree-eh.h"
41 #include "gimple-iterator.h"
42 #include "gimple-walk.h"
43 #include "tree-cfg.h"
44 #include "tree-ssa-loop-manip.h"
45 #include "tree-ssa-loop-niter.h"
46 #include "tree-ssa-loop.h"
47 #include "tree-into-ssa.h"
48 #include "tree-ssa.h"
49 #include "cfgloop.h"
50 #include "tree-scalar-evolution.h"
51 #include "tree-ssa-propagate.h"
52 #include "tree-chrec.h"
53 #include "tree-ssa-threadupdate.h"
54 #include "tree-ssa-scopedtables.h"
55 #include "tree-ssa-threadedge.h"
56 #include "omp-general.h"
57 #include "target.h"
58 #include "case-cfn-macros.h"
59 #include "alloc-pool.h"
60 #include "domwalk.h"
61 #include "tree-cfgcleanup.h"
62 #include "stringpool.h"
63 #include "attribs.h"
64 #include "vr-values.h"
65 #include "builtins.h"
66 #include "range-op.h"
67 #include "value-range-equiv.h"
68 #include "gimple-array-bounds.h"
70 /* Set of SSA names found live during the RPO traversal of the function
71 for still active basic-blocks. */
72 class live_names
74 public:
75 live_names ();
76 ~live_names ();
77 void set (tree, basic_block);
78 void clear (tree, basic_block);
79 void merge (basic_block dest, basic_block src);
80 bool live_on_block_p (tree, basic_block);
81 bool live_on_edge_p (tree, edge);
82 bool block_has_live_names_p (basic_block);
83 void clear_block (basic_block);
85 private:
86 sbitmap *live;
87 unsigned num_blocks;
88 void init_bitmap_if_needed (basic_block);
91 void
92 live_names::init_bitmap_if_needed (basic_block bb)
94 unsigned i = bb->index;
95 if (!live[i])
97 live[i] = sbitmap_alloc (num_ssa_names);
98 bitmap_clear (live[i]);
102 bool
103 live_names::block_has_live_names_p (basic_block bb)
105 unsigned i = bb->index;
106 return live[i] && bitmap_empty_p (live[i]);
109 void
110 live_names::clear_block (basic_block bb)
112 unsigned i = bb->index;
113 if (live[i])
115 sbitmap_free (live[i]);
116 live[i] = NULL;
120 void
121 live_names::merge (basic_block dest, basic_block src)
123 init_bitmap_if_needed (dest);
124 init_bitmap_if_needed (src);
125 bitmap_ior (live[dest->index], live[dest->index], live[src->index]);
128 void
129 live_names::set (tree name, basic_block bb)
131 init_bitmap_if_needed (bb);
132 bitmap_set_bit (live[bb->index], SSA_NAME_VERSION (name));
135 void
136 live_names::clear (tree name, basic_block bb)
138 unsigned i = bb->index;
139 if (live[i])
140 bitmap_clear_bit (live[i], SSA_NAME_VERSION (name));
143 live_names::live_names ()
145 num_blocks = last_basic_block_for_fn (cfun);
146 live = XCNEWVEC (sbitmap, num_blocks);
149 live_names::~live_names ()
151 for (unsigned i = 0; i < num_blocks; ++i)
152 if (live[i])
153 sbitmap_free (live[i]);
154 XDELETEVEC (live);
157 bool
158 live_names::live_on_block_p (tree name, basic_block bb)
160 return (live[bb->index]
161 && bitmap_bit_p (live[bb->index], SSA_NAME_VERSION (name)));
164 /* Return true if the SSA name NAME is live on the edge E. */
166 bool
167 live_names::live_on_edge_p (tree name, edge e)
169 return live_on_block_p (name, e->dest);
173 /* VR_TYPE describes a range with mininum value *MIN and maximum
174 value *MAX. Restrict the range to the set of values that have
175 no bits set outside NONZERO_BITS. Update *MIN and *MAX and
176 return the new range type.
178 SGN gives the sign of the values described by the range. */
180 enum value_range_kind
181 intersect_range_with_nonzero_bits (enum value_range_kind vr_type,
182 wide_int *min, wide_int *max,
183 const wide_int &nonzero_bits,
184 signop sgn)
186 if (vr_type == VR_ANTI_RANGE)
188 /* The VR_ANTI_RANGE is equivalent to the union of the ranges
189 A: [-INF, *MIN) and B: (*MAX, +INF]. First use NONZERO_BITS
190 to create an inclusive upper bound for A and an inclusive lower
191 bound for B. */
192 wide_int a_max = wi::round_down_for_mask (*min - 1, nonzero_bits);
193 wide_int b_min = wi::round_up_for_mask (*max + 1, nonzero_bits);
195 /* If the calculation of A_MAX wrapped, A is effectively empty
196 and A_MAX is the highest value that satisfies NONZERO_BITS.
197 Likewise if the calculation of B_MIN wrapped, B is effectively
198 empty and B_MIN is the lowest value that satisfies NONZERO_BITS. */
199 bool a_empty = wi::ge_p (a_max, *min, sgn);
200 bool b_empty = wi::le_p (b_min, *max, sgn);
202 /* If both A and B are empty, there are no valid values. */
203 if (a_empty && b_empty)
204 return VR_UNDEFINED;
206 /* If exactly one of A or B is empty, return a VR_RANGE for the
207 other one. */
208 if (a_empty || b_empty)
210 *min = b_min;
211 *max = a_max;
212 gcc_checking_assert (wi::le_p (*min, *max, sgn));
213 return VR_RANGE;
216 /* Update the VR_ANTI_RANGE bounds. */
217 *min = a_max + 1;
218 *max = b_min - 1;
219 gcc_checking_assert (wi::le_p (*min, *max, sgn));
221 /* Now check whether the excluded range includes any values that
222 satisfy NONZERO_BITS. If not, switch to a full VR_RANGE. */
223 if (wi::round_up_for_mask (*min, nonzero_bits) == b_min)
225 unsigned int precision = min->get_precision ();
226 *min = wi::min_value (precision, sgn);
227 *max = wi::max_value (precision, sgn);
228 vr_type = VR_RANGE;
231 if (vr_type == VR_RANGE)
233 *max = wi::round_down_for_mask (*max, nonzero_bits);
235 /* Check that the range contains at least one valid value. */
236 if (wi::gt_p (*min, *max, sgn))
237 return VR_UNDEFINED;
239 *min = wi::round_up_for_mask (*min, nonzero_bits);
240 gcc_checking_assert (wi::le_p (*min, *max, sgn));
242 return vr_type;
245 /* Return true if max and min of VR are INTEGER_CST. It's not necessary
246 a singleton. */
248 bool
249 range_int_cst_p (const value_range *vr)
251 return (vr->kind () == VR_RANGE && range_has_numeric_bounds_p (vr));
254 /* Return the single symbol (an SSA_NAME) contained in T if any, or NULL_TREE
255 otherwise. We only handle additive operations and set NEG to true if the
256 symbol is negated and INV to the invariant part, if any. */
258 tree
259 get_single_symbol (tree t, bool *neg, tree *inv)
261 bool neg_;
262 tree inv_;
264 *inv = NULL_TREE;
265 *neg = false;
267 if (TREE_CODE (t) == PLUS_EXPR
268 || TREE_CODE (t) == POINTER_PLUS_EXPR
269 || TREE_CODE (t) == MINUS_EXPR)
271 if (is_gimple_min_invariant (TREE_OPERAND (t, 0)))
273 neg_ = (TREE_CODE (t) == MINUS_EXPR);
274 inv_ = TREE_OPERAND (t, 0);
275 t = TREE_OPERAND (t, 1);
277 else if (is_gimple_min_invariant (TREE_OPERAND (t, 1)))
279 neg_ = false;
280 inv_ = TREE_OPERAND (t, 1);
281 t = TREE_OPERAND (t, 0);
283 else
284 return NULL_TREE;
286 else
288 neg_ = false;
289 inv_ = NULL_TREE;
292 if (TREE_CODE (t) == NEGATE_EXPR)
294 t = TREE_OPERAND (t, 0);
295 neg_ = !neg_;
298 if (TREE_CODE (t) != SSA_NAME)
299 return NULL_TREE;
301 if (inv_ && TREE_OVERFLOW_P (inv_))
302 inv_ = drop_tree_overflow (inv_);
304 *neg = neg_;
305 *inv = inv_;
306 return t;
309 /* The reverse operation: build a symbolic expression with TYPE
310 from symbol SYM, negated according to NEG, and invariant INV. */
312 static tree
313 build_symbolic_expr (tree type, tree sym, bool neg, tree inv)
315 const bool pointer_p = POINTER_TYPE_P (type);
316 tree t = sym;
318 if (neg)
319 t = build1 (NEGATE_EXPR, type, t);
321 if (integer_zerop (inv))
322 return t;
324 return build2 (pointer_p ? POINTER_PLUS_EXPR : PLUS_EXPR, type, t, inv);
327 /* Return
328 1 if VAL < VAL2
329 0 if !(VAL < VAL2)
330 -2 if those are incomparable. */
332 operand_less_p (tree val, tree val2)
334 /* LT is folded faster than GE and others. Inline the common case. */
335 if (TREE_CODE (val) == INTEGER_CST && TREE_CODE (val2) == INTEGER_CST)
336 return tree_int_cst_lt (val, val2);
337 else if (TREE_CODE (val) == SSA_NAME && TREE_CODE (val2) == SSA_NAME)
338 return val == val2 ? 0 : -2;
339 else
341 int cmp = compare_values (val, val2);
342 if (cmp == -1)
343 return 1;
344 else if (cmp == 0 || cmp == 1)
345 return 0;
346 else
347 return -2;
350 return 0;
353 /* Compare two values VAL1 and VAL2. Return
355 -2 if VAL1 and VAL2 cannot be compared at compile-time,
356 -1 if VAL1 < VAL2,
357 0 if VAL1 == VAL2,
358 +1 if VAL1 > VAL2, and
359 +2 if VAL1 != VAL2
361 This is similar to tree_int_cst_compare but supports pointer values
362 and values that cannot be compared at compile time.
364 If STRICT_OVERFLOW_P is not NULL, then set *STRICT_OVERFLOW_P to
365 true if the return value is only valid if we assume that signed
366 overflow is undefined. */
369 compare_values_warnv (tree val1, tree val2, bool *strict_overflow_p)
371 if (val1 == val2)
372 return 0;
374 /* Below we rely on the fact that VAL1 and VAL2 are both pointers or
375 both integers. */
376 gcc_assert (POINTER_TYPE_P (TREE_TYPE (val1))
377 == POINTER_TYPE_P (TREE_TYPE (val2)));
379 /* Convert the two values into the same type. This is needed because
380 sizetype causes sign extension even for unsigned types. */
381 if (!useless_type_conversion_p (TREE_TYPE (val1), TREE_TYPE (val2)))
382 val2 = fold_convert (TREE_TYPE (val1), val2);
384 const bool overflow_undefined
385 = INTEGRAL_TYPE_P (TREE_TYPE (val1))
386 && TYPE_OVERFLOW_UNDEFINED (TREE_TYPE (val1));
387 tree inv1, inv2;
388 bool neg1, neg2;
389 tree sym1 = get_single_symbol (val1, &neg1, &inv1);
390 tree sym2 = get_single_symbol (val2, &neg2, &inv2);
392 /* If VAL1 and VAL2 are of the form '[-]NAME [+ CST]', return -1 or +1
393 accordingly. If VAL1 and VAL2 don't use the same name, return -2. */
394 if (sym1 && sym2)
396 /* Both values must use the same name with the same sign. */
397 if (sym1 != sym2 || neg1 != neg2)
398 return -2;
400 /* [-]NAME + CST == [-]NAME + CST. */
401 if (inv1 == inv2)
402 return 0;
404 /* If overflow is defined we cannot simplify more. */
405 if (!overflow_undefined)
406 return -2;
408 if (strict_overflow_p != NULL
409 /* Symbolic range building sets TREE_NO_WARNING to declare
410 that overflow doesn't happen. */
411 && (!inv1 || !TREE_NO_WARNING (val1))
412 && (!inv2 || !TREE_NO_WARNING (val2)))
413 *strict_overflow_p = true;
415 if (!inv1)
416 inv1 = build_int_cst (TREE_TYPE (val1), 0);
417 if (!inv2)
418 inv2 = build_int_cst (TREE_TYPE (val2), 0);
420 return wi::cmp (wi::to_wide (inv1), wi::to_wide (inv2),
421 TYPE_SIGN (TREE_TYPE (val1)));
424 const bool cst1 = is_gimple_min_invariant (val1);
425 const bool cst2 = is_gimple_min_invariant (val2);
427 /* If one is of the form '[-]NAME + CST' and the other is constant, then
428 it might be possible to say something depending on the constants. */
429 if ((sym1 && inv1 && cst2) || (sym2 && inv2 && cst1))
431 if (!overflow_undefined)
432 return -2;
434 if (strict_overflow_p != NULL
435 /* Symbolic range building sets TREE_NO_WARNING to declare
436 that overflow doesn't happen. */
437 && (!sym1 || !TREE_NO_WARNING (val1))
438 && (!sym2 || !TREE_NO_WARNING (val2)))
439 *strict_overflow_p = true;
441 const signop sgn = TYPE_SIGN (TREE_TYPE (val1));
442 tree cst = cst1 ? val1 : val2;
443 tree inv = cst1 ? inv2 : inv1;
445 /* Compute the difference between the constants. If it overflows or
446 underflows, this means that we can trivially compare the NAME with
447 it and, consequently, the two values with each other. */
448 wide_int diff = wi::to_wide (cst) - wi::to_wide (inv);
449 if (wi::cmp (0, wi::to_wide (inv), sgn)
450 != wi::cmp (diff, wi::to_wide (cst), sgn))
452 const int res = wi::cmp (wi::to_wide (cst), wi::to_wide (inv), sgn);
453 return cst1 ? res : -res;
456 return -2;
459 /* We cannot say anything more for non-constants. */
460 if (!cst1 || !cst2)
461 return -2;
463 if (!POINTER_TYPE_P (TREE_TYPE (val1)))
465 /* We cannot compare overflowed values. */
466 if (TREE_OVERFLOW (val1) || TREE_OVERFLOW (val2))
467 return -2;
469 if (TREE_CODE (val1) == INTEGER_CST
470 && TREE_CODE (val2) == INTEGER_CST)
471 return tree_int_cst_compare (val1, val2);
473 if (poly_int_tree_p (val1) && poly_int_tree_p (val2))
475 if (known_eq (wi::to_poly_widest (val1),
476 wi::to_poly_widest (val2)))
477 return 0;
478 if (known_lt (wi::to_poly_widest (val1),
479 wi::to_poly_widest (val2)))
480 return -1;
481 if (known_gt (wi::to_poly_widest (val1),
482 wi::to_poly_widest (val2)))
483 return 1;
486 return -2;
488 else
490 if (TREE_CODE (val1) == INTEGER_CST && TREE_CODE (val2) == INTEGER_CST)
492 /* We cannot compare overflowed values. */
493 if (TREE_OVERFLOW (val1) || TREE_OVERFLOW (val2))
494 return -2;
496 return tree_int_cst_compare (val1, val2);
499 /* First see if VAL1 and VAL2 are not the same. */
500 if (operand_equal_p (val1, val2, 0))
501 return 0;
503 fold_defer_overflow_warnings ();
505 /* If VAL1 is a lower address than VAL2, return -1. */
506 tree t = fold_binary_to_constant (LT_EXPR, boolean_type_node, val1, val2);
507 if (t && integer_onep (t))
509 fold_undefer_and_ignore_overflow_warnings ();
510 return -1;
513 /* If VAL1 is a higher address than VAL2, return +1. */
514 t = fold_binary_to_constant (LT_EXPR, boolean_type_node, val2, val1);
515 if (t && integer_onep (t))
517 fold_undefer_and_ignore_overflow_warnings ();
518 return 1;
521 /* If VAL1 is different than VAL2, return +2. */
522 t = fold_binary_to_constant (NE_EXPR, boolean_type_node, val1, val2);
523 fold_undefer_and_ignore_overflow_warnings ();
524 if (t && integer_onep (t))
525 return 2;
527 return -2;
531 /* Compare values like compare_values_warnv. */
534 compare_values (tree val1, tree val2)
536 bool sop;
537 return compare_values_warnv (val1, val2, &sop);
540 /* If BOUND will include a symbolic bound, adjust it accordingly,
541 otherwise leave it as is.
543 CODE is the original operation that combined the bounds (PLUS_EXPR
544 or MINUS_EXPR).
546 TYPE is the type of the original operation.
548 SYM_OPn is the symbolic for OPn if it has a symbolic.
550 NEG_OPn is TRUE if the OPn was negated. */
552 static void
553 adjust_symbolic_bound (tree &bound, enum tree_code code, tree type,
554 tree sym_op0, tree sym_op1,
555 bool neg_op0, bool neg_op1)
557 bool minus_p = (code == MINUS_EXPR);
558 /* If the result bound is constant, we're done; otherwise, build the
559 symbolic lower bound. */
560 if (sym_op0 == sym_op1)
562 else if (sym_op0)
563 bound = build_symbolic_expr (type, sym_op0,
564 neg_op0, bound);
565 else if (sym_op1)
567 /* We may not negate if that might introduce
568 undefined overflow. */
569 if (!minus_p
570 || neg_op1
571 || TYPE_OVERFLOW_WRAPS (type))
572 bound = build_symbolic_expr (type, sym_op1,
573 neg_op1 ^ minus_p, bound);
574 else
575 bound = NULL_TREE;
579 /* Combine OP1 and OP1, which are two parts of a bound, into one wide
580 int bound according to CODE. CODE is the operation combining the
581 bound (either a PLUS_EXPR or a MINUS_EXPR).
583 TYPE is the type of the combine operation.
585 WI is the wide int to store the result.
587 OVF is -1 if an underflow occurred, +1 if an overflow occurred or 0
588 if over/underflow occurred. */
590 static void
591 combine_bound (enum tree_code code, wide_int &wi, wi::overflow_type &ovf,
592 tree type, tree op0, tree op1)
594 bool minus_p = (code == MINUS_EXPR);
595 const signop sgn = TYPE_SIGN (type);
596 const unsigned int prec = TYPE_PRECISION (type);
598 /* Combine the bounds, if any. */
599 if (op0 && op1)
601 if (minus_p)
602 wi = wi::sub (wi::to_wide (op0), wi::to_wide (op1), sgn, &ovf);
603 else
604 wi = wi::add (wi::to_wide (op0), wi::to_wide (op1), sgn, &ovf);
606 else if (op0)
607 wi = wi::to_wide (op0);
608 else if (op1)
610 if (minus_p)
611 wi = wi::neg (wi::to_wide (op1), &ovf);
612 else
613 wi = wi::to_wide (op1);
615 else
616 wi = wi::shwi (0, prec);
619 /* Given a range in [WMIN, WMAX], adjust it for possible overflow and
620 put the result in VR.
622 TYPE is the type of the range.
624 MIN_OVF and MAX_OVF indicate what type of overflow, if any,
625 occurred while originally calculating WMIN or WMAX. -1 indicates
626 underflow. +1 indicates overflow. 0 indicates neither. */
628 static void
629 set_value_range_with_overflow (value_range_kind &kind, tree &min, tree &max,
630 tree type,
631 const wide_int &wmin, const wide_int &wmax,
632 wi::overflow_type min_ovf,
633 wi::overflow_type max_ovf)
635 const signop sgn = TYPE_SIGN (type);
636 const unsigned int prec = TYPE_PRECISION (type);
638 /* For one bit precision if max < min, then the swapped
639 range covers all values. */
640 if (prec == 1 && wi::lt_p (wmax, wmin, sgn))
642 kind = VR_VARYING;
643 return;
646 if (TYPE_OVERFLOW_WRAPS (type))
648 /* If overflow wraps, truncate the values and adjust the
649 range kind and bounds appropriately. */
650 wide_int tmin = wide_int::from (wmin, prec, sgn);
651 wide_int tmax = wide_int::from (wmax, prec, sgn);
652 if ((min_ovf != wi::OVF_NONE) == (max_ovf != wi::OVF_NONE))
654 /* If the limits are swapped, we wrapped around and cover
655 the entire range. */
656 if (wi::gt_p (tmin, tmax, sgn))
657 kind = VR_VARYING;
658 else
660 kind = VR_RANGE;
661 /* No overflow or both overflow or underflow. The
662 range kind stays VR_RANGE. */
663 min = wide_int_to_tree (type, tmin);
664 max = wide_int_to_tree (type, tmax);
666 return;
668 else if ((min_ovf == wi::OVF_UNDERFLOW && max_ovf == wi::OVF_NONE)
669 || (max_ovf == wi::OVF_OVERFLOW && min_ovf == wi::OVF_NONE))
671 /* Min underflow or max overflow. The range kind
672 changes to VR_ANTI_RANGE. */
673 bool covers = false;
674 wide_int tem = tmin;
675 tmin = tmax + 1;
676 if (wi::cmp (tmin, tmax, sgn) < 0)
677 covers = true;
678 tmax = tem - 1;
679 if (wi::cmp (tmax, tem, sgn) > 0)
680 covers = true;
681 /* If the anti-range would cover nothing, drop to varying.
682 Likewise if the anti-range bounds are outside of the
683 types values. */
684 if (covers || wi::cmp (tmin, tmax, sgn) > 0)
686 kind = VR_VARYING;
687 return;
689 kind = VR_ANTI_RANGE;
690 min = wide_int_to_tree (type, tmin);
691 max = wide_int_to_tree (type, tmax);
692 return;
694 else
696 /* Other underflow and/or overflow, drop to VR_VARYING. */
697 kind = VR_VARYING;
698 return;
701 else
703 /* If overflow does not wrap, saturate to the types min/max
704 value. */
705 wide_int type_min = wi::min_value (prec, sgn);
706 wide_int type_max = wi::max_value (prec, sgn);
707 kind = VR_RANGE;
708 if (min_ovf == wi::OVF_UNDERFLOW)
709 min = wide_int_to_tree (type, type_min);
710 else if (min_ovf == wi::OVF_OVERFLOW)
711 min = wide_int_to_tree (type, type_max);
712 else
713 min = wide_int_to_tree (type, wmin);
715 if (max_ovf == wi::OVF_UNDERFLOW)
716 max = wide_int_to_tree (type, type_min);
717 else if (max_ovf == wi::OVF_OVERFLOW)
718 max = wide_int_to_tree (type, type_max);
719 else
720 max = wide_int_to_tree (type, wmax);
724 /* Fold two value range's of a POINTER_PLUS_EXPR into VR. */
726 static void
727 extract_range_from_pointer_plus_expr (value_range *vr,
728 enum tree_code code,
729 tree expr_type,
730 const value_range *vr0,
731 const value_range *vr1)
733 gcc_checking_assert (POINTER_TYPE_P (expr_type)
734 && code == POINTER_PLUS_EXPR);
735 /* For pointer types, we are really only interested in asserting
736 whether the expression evaluates to non-NULL.
737 With -fno-delete-null-pointer-checks we need to be more
738 conservative. As some object might reside at address 0,
739 then some offset could be added to it and the same offset
740 subtracted again and the result would be NULL.
741 E.g.
742 static int a[12]; where &a[0] is NULL and
743 ptr = &a[6];
744 ptr -= 6;
745 ptr will be NULL here, even when there is POINTER_PLUS_EXPR
746 where the first range doesn't include zero and the second one
747 doesn't either. As the second operand is sizetype (unsigned),
748 consider all ranges where the MSB could be set as possible
749 subtractions where the result might be NULL. */
750 if ((!range_includes_zero_p (vr0)
751 || !range_includes_zero_p (vr1))
752 && !TYPE_OVERFLOW_WRAPS (expr_type)
753 && (flag_delete_null_pointer_checks
754 || (range_int_cst_p (vr1)
755 && !tree_int_cst_sign_bit (vr1->max ()))))
756 vr->set_nonzero (expr_type);
757 else if (vr0->zero_p () && vr1->zero_p ())
758 vr->set_zero (expr_type);
759 else
760 vr->set_varying (expr_type);
763 /* Extract range information from a PLUS/MINUS_EXPR and store the
764 result in *VR. */
766 static void
767 extract_range_from_plus_minus_expr (value_range *vr,
768 enum tree_code code,
769 tree expr_type,
770 const value_range *vr0_,
771 const value_range *vr1_)
773 gcc_checking_assert (code == PLUS_EXPR || code == MINUS_EXPR);
775 value_range vr0 = *vr0_, vr1 = *vr1_;
776 value_range vrtem0, vrtem1;
778 /* Now canonicalize anti-ranges to ranges when they are not symbolic
779 and express ~[] op X as ([]' op X) U ([]'' op X). */
780 if (vr0.kind () == VR_ANTI_RANGE
781 && ranges_from_anti_range (&vr0, &vrtem0, &vrtem1))
783 extract_range_from_plus_minus_expr (vr, code, expr_type, &vrtem0, vr1_);
784 if (!vrtem1.undefined_p ())
786 value_range vrres;
787 extract_range_from_plus_minus_expr (&vrres, code, expr_type,
788 &vrtem1, vr1_);
789 vr->union_ (&vrres);
791 return;
793 /* Likewise for X op ~[]. */
794 if (vr1.kind () == VR_ANTI_RANGE
795 && ranges_from_anti_range (&vr1, &vrtem0, &vrtem1))
797 extract_range_from_plus_minus_expr (vr, code, expr_type, vr0_, &vrtem0);
798 if (!vrtem1.undefined_p ())
800 value_range vrres;
801 extract_range_from_plus_minus_expr (&vrres, code, expr_type,
802 vr0_, &vrtem1);
803 vr->union_ (&vrres);
805 return;
808 value_range_kind kind;
809 value_range_kind vr0_kind = vr0.kind (), vr1_kind = vr1.kind ();
810 tree vr0_min = vr0.min (), vr0_max = vr0.max ();
811 tree vr1_min = vr1.min (), vr1_max = vr1.max ();
812 tree min = NULL_TREE, max = NULL_TREE;
814 /* This will normalize things such that calculating
815 [0,0] - VR_VARYING is not dropped to varying, but is
816 calculated as [MIN+1, MAX]. */
817 if (vr0.varying_p ())
819 vr0_kind = VR_RANGE;
820 vr0_min = vrp_val_min (expr_type);
821 vr0_max = vrp_val_max (expr_type);
823 if (vr1.varying_p ())
825 vr1_kind = VR_RANGE;
826 vr1_min = vrp_val_min (expr_type);
827 vr1_max = vrp_val_max (expr_type);
830 const bool minus_p = (code == MINUS_EXPR);
831 tree min_op0 = vr0_min;
832 tree min_op1 = minus_p ? vr1_max : vr1_min;
833 tree max_op0 = vr0_max;
834 tree max_op1 = minus_p ? vr1_min : vr1_max;
835 tree sym_min_op0 = NULL_TREE;
836 tree sym_min_op1 = NULL_TREE;
837 tree sym_max_op0 = NULL_TREE;
838 tree sym_max_op1 = NULL_TREE;
839 bool neg_min_op0, neg_min_op1, neg_max_op0, neg_max_op1;
841 neg_min_op0 = neg_min_op1 = neg_max_op0 = neg_max_op1 = false;
843 /* If we have a PLUS or MINUS with two VR_RANGEs, either constant or
844 single-symbolic ranges, try to compute the precise resulting range,
845 but only if we know that this resulting range will also be constant
846 or single-symbolic. */
847 if (vr0_kind == VR_RANGE && vr1_kind == VR_RANGE
848 && (TREE_CODE (min_op0) == INTEGER_CST
849 || (sym_min_op0
850 = get_single_symbol (min_op0, &neg_min_op0, &min_op0)))
851 && (TREE_CODE (min_op1) == INTEGER_CST
852 || (sym_min_op1
853 = get_single_symbol (min_op1, &neg_min_op1, &min_op1)))
854 && (!(sym_min_op0 && sym_min_op1)
855 || (sym_min_op0 == sym_min_op1
856 && neg_min_op0 == (minus_p ? neg_min_op1 : !neg_min_op1)))
857 && (TREE_CODE (max_op0) == INTEGER_CST
858 || (sym_max_op0
859 = get_single_symbol (max_op0, &neg_max_op0, &max_op0)))
860 && (TREE_CODE (max_op1) == INTEGER_CST
861 || (sym_max_op1
862 = get_single_symbol (max_op1, &neg_max_op1, &max_op1)))
863 && (!(sym_max_op0 && sym_max_op1)
864 || (sym_max_op0 == sym_max_op1
865 && neg_max_op0 == (minus_p ? neg_max_op1 : !neg_max_op1))))
867 wide_int wmin, wmax;
868 wi::overflow_type min_ovf = wi::OVF_NONE;
869 wi::overflow_type max_ovf = wi::OVF_NONE;
871 /* Build the bounds. */
872 combine_bound (code, wmin, min_ovf, expr_type, min_op0, min_op1);
873 combine_bound (code, wmax, max_ovf, expr_type, max_op0, max_op1);
875 /* If the resulting range will be symbolic, we need to eliminate any
876 explicit or implicit overflow introduced in the above computation
877 because compare_values could make an incorrect use of it. That's
878 why we require one of the ranges to be a singleton. */
879 if ((sym_min_op0 != sym_min_op1 || sym_max_op0 != sym_max_op1)
880 && ((bool)min_ovf || (bool)max_ovf
881 || (min_op0 != max_op0 && min_op1 != max_op1)))
883 vr->set_varying (expr_type);
884 return;
887 /* Adjust the range for possible overflow. */
888 set_value_range_with_overflow (kind, min, max, expr_type,
889 wmin, wmax, min_ovf, max_ovf);
890 if (kind == VR_VARYING)
892 vr->set_varying (expr_type);
893 return;
896 /* Build the symbolic bounds if needed. */
897 adjust_symbolic_bound (min, code, expr_type,
898 sym_min_op0, sym_min_op1,
899 neg_min_op0, neg_min_op1);
900 adjust_symbolic_bound (max, code, expr_type,
901 sym_max_op0, sym_max_op1,
902 neg_max_op0, neg_max_op1);
904 else
906 /* For other cases, for example if we have a PLUS_EXPR with two
907 VR_ANTI_RANGEs, drop to VR_VARYING. It would take more effort
908 to compute a precise range for such a case.
909 ??? General even mixed range kind operations can be expressed
910 by for example transforming ~[3, 5] + [1, 2] to range-only
911 operations and a union primitive:
912 [-INF, 2] + [1, 2] U [5, +INF] + [1, 2]
913 [-INF+1, 4] U [6, +INF(OVF)]
914 though usually the union is not exactly representable with
915 a single range or anti-range as the above is
916 [-INF+1, +INF(OVF)] intersected with ~[5, 5]
917 but one could use a scheme similar to equivalences for this. */
918 vr->set_varying (expr_type);
919 return;
922 /* If either MIN or MAX overflowed, then set the resulting range to
923 VARYING. */
924 if (min == NULL_TREE
925 || TREE_OVERFLOW_P (min)
926 || max == NULL_TREE
927 || TREE_OVERFLOW_P (max))
929 vr->set_varying (expr_type);
930 return;
933 int cmp = compare_values (min, max);
934 if (cmp == -2 || cmp == 1)
936 /* If the new range has its limits swapped around (MIN > MAX),
937 then the operation caused one of them to wrap around, mark
938 the new range VARYING. */
939 vr->set_varying (expr_type);
941 else
942 vr->set (min, max, kind);
945 /* Return the range-ops handler for CODE and EXPR_TYPE. If no
946 suitable operator is found, return NULL and set VR to VARYING. */
948 static const range_operator *
949 get_range_op_handler (value_range *vr,
950 enum tree_code code,
951 tree expr_type)
953 const range_operator *op = range_op_handler (code, expr_type);
954 if (!op)
955 vr->set_varying (expr_type);
956 return op;
959 /* If the types passed are supported, return TRUE, otherwise set VR to
960 VARYING and return FALSE. */
962 static bool
963 supported_types_p (value_range *vr,
964 tree type0,
965 tree type1 = NULL)
967 if (!value_range::supports_type_p (type0)
968 || (type1 && !value_range::supports_type_p (type1)))
970 vr->set_varying (type0);
971 return false;
973 return true;
976 /* If any of the ranges passed are defined, return TRUE, otherwise set
977 VR to UNDEFINED and return FALSE. */
979 static bool
980 defined_ranges_p (value_range *vr,
981 const value_range *vr0, const value_range *vr1 = NULL)
983 if (vr0->undefined_p () && (!vr1 || vr1->undefined_p ()))
985 vr->set_undefined ();
986 return false;
988 return true;
991 static value_range
992 drop_undefines_to_varying (const value_range *vr, tree expr_type)
994 if (vr->undefined_p ())
995 return value_range (expr_type);
996 else
997 return *vr;
1000 /* If any operand is symbolic, perform a binary operation on them and
1001 return TRUE, otherwise return FALSE. */
1003 static bool
1004 range_fold_binary_symbolics_p (value_range *vr,
1005 tree_code code,
1006 tree expr_type,
1007 const value_range *vr0_,
1008 const value_range *vr1_)
1010 if (vr0_->symbolic_p () || vr1_->symbolic_p ())
1012 value_range vr0 = drop_undefines_to_varying (vr0_, expr_type);
1013 value_range vr1 = drop_undefines_to_varying (vr1_, expr_type);
1014 if ((code == PLUS_EXPR || code == MINUS_EXPR))
1016 extract_range_from_plus_minus_expr (vr, code, expr_type,
1017 &vr0, &vr1);
1018 return true;
1020 if (POINTER_TYPE_P (expr_type) && code == POINTER_PLUS_EXPR)
1022 extract_range_from_pointer_plus_expr (vr, code, expr_type,
1023 &vr0, &vr1);
1024 return true;
1026 const range_operator *op = get_range_op_handler (vr, code, expr_type);
1027 vr0.normalize_symbolics ();
1028 vr1.normalize_symbolics ();
1029 return op->fold_range (*vr, expr_type, vr0, vr1);
1031 return false;
1034 /* If operand is symbolic, perform a unary operation on it and return
1035 TRUE, otherwise return FALSE. */
1037 static bool
1038 range_fold_unary_symbolics_p (value_range *vr,
1039 tree_code code,
1040 tree expr_type,
1041 const value_range *vr0)
1043 if (vr0->symbolic_p ())
1045 if (code == NEGATE_EXPR)
1047 /* -X is simply 0 - X. */
1048 value_range zero;
1049 zero.set_zero (vr0->type ());
1050 range_fold_binary_expr (vr, MINUS_EXPR, expr_type, &zero, vr0);
1051 return true;
1053 if (code == BIT_NOT_EXPR)
1055 /* ~X is simply -1 - X. */
1056 value_range minusone;
1057 minusone.set (build_int_cst (vr0->type (), -1));
1058 range_fold_binary_expr (vr, MINUS_EXPR, expr_type, &minusone, vr0);
1059 return true;
1061 const range_operator *op = get_range_op_handler (vr, code, expr_type);
1062 value_range vr0_cst (*vr0);
1063 vr0_cst.normalize_symbolics ();
1064 return op->fold_range (*vr, expr_type, vr0_cst, value_range (expr_type));
1066 return false;
1069 /* Perform a binary operation on a pair of ranges. */
1071 void
1072 range_fold_binary_expr (value_range *vr,
1073 enum tree_code code,
1074 tree expr_type,
1075 const value_range *vr0_,
1076 const value_range *vr1_)
1078 if (!supported_types_p (vr, expr_type)
1079 || !defined_ranges_p (vr, vr0_, vr1_))
1080 return;
1081 const range_operator *op = get_range_op_handler (vr, code, expr_type);
1082 if (!op)
1083 return;
1085 if (range_fold_binary_symbolics_p (vr, code, expr_type, vr0_, vr1_))
1086 return;
1088 value_range vr0 (*vr0_);
1089 value_range vr1 (*vr1_);
1090 if (vr0.undefined_p ())
1091 vr0.set_varying (expr_type);
1092 if (vr1.undefined_p ())
1093 vr1.set_varying (expr_type);
1094 vr0.normalize_addresses ();
1095 vr1.normalize_addresses ();
1096 op->fold_range (*vr, expr_type, vr0, vr1);
1099 /* Perform a unary operation on a range. */
1101 void
1102 range_fold_unary_expr (value_range *vr,
1103 enum tree_code code, tree expr_type,
1104 const value_range *vr0,
1105 tree vr0_type)
1107 if (!supported_types_p (vr, expr_type, vr0_type)
1108 || !defined_ranges_p (vr, vr0))
1109 return;
1110 const range_operator *op = get_range_op_handler (vr, code, expr_type);
1111 if (!op)
1112 return;
1114 if (range_fold_unary_symbolics_p (vr, code, expr_type, vr0))
1115 return;
1117 value_range vr0_cst (*vr0);
1118 vr0_cst.normalize_addresses ();
1119 op->fold_range (*vr, expr_type, vr0_cst, value_range (expr_type));
1122 /* If the range of values taken by OP can be inferred after STMT executes,
1123 return the comparison code (COMP_CODE_P) and value (VAL_P) that
1124 describes the inferred range. Return true if a range could be
1125 inferred. */
1127 bool
1128 infer_value_range (gimple *stmt, tree op, tree_code *comp_code_p, tree *val_p)
1130 *val_p = NULL_TREE;
1131 *comp_code_p = ERROR_MARK;
1133 /* Do not attempt to infer anything in names that flow through
1134 abnormal edges. */
1135 if (SSA_NAME_OCCURS_IN_ABNORMAL_PHI (op))
1136 return false;
1138 /* If STMT is the last statement of a basic block with no normal
1139 successors, there is no point inferring anything about any of its
1140 operands. We would not be able to find a proper insertion point
1141 for the assertion, anyway. */
1142 if (stmt_ends_bb_p (stmt))
1144 edge_iterator ei;
1145 edge e;
1147 FOR_EACH_EDGE (e, ei, gimple_bb (stmt)->succs)
1148 if (!(e->flags & (EDGE_ABNORMAL|EDGE_EH)))
1149 break;
1150 if (e == NULL)
1151 return false;
1154 if (infer_nonnull_range (stmt, op))
1156 *val_p = build_int_cst (TREE_TYPE (op), 0);
1157 *comp_code_p = NE_EXPR;
1158 return true;
1161 return false;
1164 /* Dump assert_info structure. */
1166 void
1167 dump_assert_info (FILE *file, const assert_info &assert)
1169 fprintf (file, "Assert for: ");
1170 print_generic_expr (file, assert.name);
1171 fprintf (file, "\n\tPREDICATE: expr=[");
1172 print_generic_expr (file, assert.expr);
1173 fprintf (file, "] %s ", get_tree_code_name (assert.comp_code));
1174 fprintf (file, "val=[");
1175 print_generic_expr (file, assert.val);
1176 fprintf (file, "]\n\n");
1179 DEBUG_FUNCTION void
1180 debug (const assert_info &assert)
1182 dump_assert_info (stderr, assert);
1185 /* Dump a vector of assert_info's. */
1187 void
1188 dump_asserts_info (FILE *file, const vec<assert_info> &asserts)
1190 for (unsigned i = 0; i < asserts.length (); ++i)
1192 dump_assert_info (file, asserts[i]);
1193 fprintf (file, "\n");
1197 DEBUG_FUNCTION void
1198 debug (const vec<assert_info> &asserts)
1200 dump_asserts_info (stderr, asserts);
1203 /* Push the assert info for NAME, EXPR, COMP_CODE and VAL to ASSERTS. */
1205 static void
1206 add_assert_info (vec<assert_info> &asserts,
1207 tree name, tree expr, enum tree_code comp_code, tree val)
1209 assert_info info;
1210 info.comp_code = comp_code;
1211 info.name = name;
1212 if (TREE_OVERFLOW_P (val))
1213 val = drop_tree_overflow (val);
1214 info.val = val;
1215 info.expr = expr;
1216 asserts.safe_push (info);
1217 if (dump_enabled_p ())
1218 dump_printf (MSG_NOTE | MSG_PRIORITY_INTERNALS,
1219 "Adding assert for %T from %T %s %T\n",
1220 name, expr, op_symbol_code (comp_code), val);
1223 /* (COND_OP0 COND_CODE COND_OP1) is a predicate which uses NAME.
1224 Extract a suitable test code and value and store them into *CODE_P and
1225 *VAL_P so the predicate is normalized to NAME *CODE_P *VAL_P.
1227 If no extraction was possible, return FALSE, otherwise return TRUE.
1229 If INVERT is true, then we invert the result stored into *CODE_P. */
1231 static bool
1232 extract_code_and_val_from_cond_with_ops (tree name, enum tree_code cond_code,
1233 tree cond_op0, tree cond_op1,
1234 bool invert, enum tree_code *code_p,
1235 tree *val_p)
1237 enum tree_code comp_code;
1238 tree val;
1240 /* Otherwise, we have a comparison of the form NAME COMP VAL
1241 or VAL COMP NAME. */
1242 if (name == cond_op1)
1244 /* If the predicate is of the form VAL COMP NAME, flip
1245 COMP around because we need to register NAME as the
1246 first operand in the predicate. */
1247 comp_code = swap_tree_comparison (cond_code);
1248 val = cond_op0;
1250 else if (name == cond_op0)
1252 /* The comparison is of the form NAME COMP VAL, so the
1253 comparison code remains unchanged. */
1254 comp_code = cond_code;
1255 val = cond_op1;
1257 else
1258 gcc_unreachable ();
1260 /* Invert the comparison code as necessary. */
1261 if (invert)
1262 comp_code = invert_tree_comparison (comp_code, 0);
1264 /* VRP only handles integral and pointer types. */
1265 if (! INTEGRAL_TYPE_P (TREE_TYPE (val))
1266 && ! POINTER_TYPE_P (TREE_TYPE (val)))
1267 return false;
1269 /* Do not register always-false predicates.
1270 FIXME: this works around a limitation in fold() when dealing with
1271 enumerations. Given 'enum { N1, N2 } x;', fold will not
1272 fold 'if (x > N2)' to 'if (0)'. */
1273 if ((comp_code == GT_EXPR || comp_code == LT_EXPR)
1274 && INTEGRAL_TYPE_P (TREE_TYPE (val)))
1276 tree min = TYPE_MIN_VALUE (TREE_TYPE (val));
1277 tree max = TYPE_MAX_VALUE (TREE_TYPE (val));
1279 if (comp_code == GT_EXPR
1280 && (!max
1281 || compare_values (val, max) == 0))
1282 return false;
1284 if (comp_code == LT_EXPR
1285 && (!min
1286 || compare_values (val, min) == 0))
1287 return false;
1289 *code_p = comp_code;
1290 *val_p = val;
1291 return true;
1294 /* Find out smallest RES where RES > VAL && (RES & MASK) == RES, if any
1295 (otherwise return VAL). VAL and MASK must be zero-extended for
1296 precision PREC. If SGNBIT is non-zero, first xor VAL with SGNBIT
1297 (to transform signed values into unsigned) and at the end xor
1298 SGNBIT back. */
1300 wide_int
1301 masked_increment (const wide_int &val_in, const wide_int &mask,
1302 const wide_int &sgnbit, unsigned int prec)
1304 wide_int bit = wi::one (prec), res;
1305 unsigned int i;
1307 wide_int val = val_in ^ sgnbit;
1308 for (i = 0; i < prec; i++, bit += bit)
1310 res = mask;
1311 if ((res & bit) == 0)
1312 continue;
1313 res = bit - 1;
1314 res = wi::bit_and_not (val + bit, res);
1315 res &= mask;
1316 if (wi::gtu_p (res, val))
1317 return res ^ sgnbit;
1319 return val ^ sgnbit;
1322 /* Helper for overflow_comparison_p
1324 OP0 CODE OP1 is a comparison. Examine the comparison and potentially
1325 OP1's defining statement to see if it ultimately has the form
1326 OP0 CODE (OP0 PLUS INTEGER_CST)
1328 If so, return TRUE indicating this is an overflow test and store into
1329 *NEW_CST an updated constant that can be used in a narrowed range test.
1331 REVERSED indicates if the comparison was originally:
1333 OP1 CODE' OP0.
1335 This affects how we build the updated constant. */
1337 static bool
1338 overflow_comparison_p_1 (enum tree_code code, tree op0, tree op1,
1339 bool follow_assert_exprs, bool reversed, tree *new_cst)
1341 /* See if this is a relational operation between two SSA_NAMES with
1342 unsigned, overflow wrapping values. If so, check it more deeply. */
1343 if ((code == LT_EXPR || code == LE_EXPR
1344 || code == GE_EXPR || code == GT_EXPR)
1345 && TREE_CODE (op0) == SSA_NAME
1346 && TREE_CODE (op1) == SSA_NAME
1347 && INTEGRAL_TYPE_P (TREE_TYPE (op0))
1348 && TYPE_UNSIGNED (TREE_TYPE (op0))
1349 && TYPE_OVERFLOW_WRAPS (TREE_TYPE (op0)))
1351 gimple *op1_def = SSA_NAME_DEF_STMT (op1);
1353 /* If requested, follow any ASSERT_EXPRs backwards for OP1. */
1354 if (follow_assert_exprs)
1356 while (gimple_assign_single_p (op1_def)
1357 && TREE_CODE (gimple_assign_rhs1 (op1_def)) == ASSERT_EXPR)
1359 op1 = TREE_OPERAND (gimple_assign_rhs1 (op1_def), 0);
1360 if (TREE_CODE (op1) != SSA_NAME)
1361 break;
1362 op1_def = SSA_NAME_DEF_STMT (op1);
1366 /* Now look at the defining statement of OP1 to see if it adds
1367 or subtracts a nonzero constant from another operand. */
1368 if (op1_def
1369 && is_gimple_assign (op1_def)
1370 && gimple_assign_rhs_code (op1_def) == PLUS_EXPR
1371 && TREE_CODE (gimple_assign_rhs2 (op1_def)) == INTEGER_CST
1372 && !integer_zerop (gimple_assign_rhs2 (op1_def)))
1374 tree target = gimple_assign_rhs1 (op1_def);
1376 /* If requested, follow ASSERT_EXPRs backwards for op0 looking
1377 for one where TARGET appears on the RHS. */
1378 if (follow_assert_exprs)
1380 /* Now see if that "other operand" is op0, following the chain
1381 of ASSERT_EXPRs if necessary. */
1382 gimple *op0_def = SSA_NAME_DEF_STMT (op0);
1383 while (op0 != target
1384 && gimple_assign_single_p (op0_def)
1385 && TREE_CODE (gimple_assign_rhs1 (op0_def)) == ASSERT_EXPR)
1387 op0 = TREE_OPERAND (gimple_assign_rhs1 (op0_def), 0);
1388 if (TREE_CODE (op0) != SSA_NAME)
1389 break;
1390 op0_def = SSA_NAME_DEF_STMT (op0);
1394 /* If we did not find our target SSA_NAME, then this is not
1395 an overflow test. */
1396 if (op0 != target)
1397 return false;
1399 tree type = TREE_TYPE (op0);
1400 wide_int max = wi::max_value (TYPE_PRECISION (type), UNSIGNED);
1401 tree inc = gimple_assign_rhs2 (op1_def);
1402 if (reversed)
1403 *new_cst = wide_int_to_tree (type, max + wi::to_wide (inc));
1404 else
1405 *new_cst = wide_int_to_tree (type, max - wi::to_wide (inc));
1406 return true;
1409 return false;
1412 /* OP0 CODE OP1 is a comparison. Examine the comparison and potentially
1413 OP1's defining statement to see if it ultimately has the form
1414 OP0 CODE (OP0 PLUS INTEGER_CST)
1416 If so, return TRUE indicating this is an overflow test and store into
1417 *NEW_CST an updated constant that can be used in a narrowed range test.
1419 These statements are left as-is in the IL to facilitate discovery of
1420 {ADD,SUB}_OVERFLOW sequences later in the optimizer pipeline. But
1421 the alternate range representation is often useful within VRP. */
1423 bool
1424 overflow_comparison_p (tree_code code, tree name, tree val,
1425 bool use_equiv_p, tree *new_cst)
1427 if (overflow_comparison_p_1 (code, name, val, use_equiv_p, false, new_cst))
1428 return true;
1429 return overflow_comparison_p_1 (swap_tree_comparison (code), val, name,
1430 use_equiv_p, true, new_cst);
1434 /* Try to register an edge assertion for SSA name NAME on edge E for
1435 the condition COND contributing to the conditional jump pointed to by BSI.
1436 Invert the condition COND if INVERT is true. */
1438 static void
1439 register_edge_assert_for_2 (tree name, edge e,
1440 enum tree_code cond_code,
1441 tree cond_op0, tree cond_op1, bool invert,
1442 vec<assert_info> &asserts)
1444 tree val;
1445 enum tree_code comp_code;
1447 if (!extract_code_and_val_from_cond_with_ops (name, cond_code,
1448 cond_op0,
1449 cond_op1,
1450 invert, &comp_code, &val))
1451 return;
1453 /* Queue the assert. */
1454 tree x;
1455 if (overflow_comparison_p (comp_code, name, val, false, &x))
1457 enum tree_code new_code = ((comp_code == GT_EXPR || comp_code == GE_EXPR)
1458 ? GT_EXPR : LE_EXPR);
1459 add_assert_info (asserts, name, name, new_code, x);
1461 add_assert_info (asserts, name, name, comp_code, val);
1463 /* In the case of NAME <= CST and NAME being defined as
1464 NAME = (unsigned) NAME2 + CST2 we can assert NAME2 >= -CST2
1465 and NAME2 <= CST - CST2. We can do the same for NAME > CST.
1466 This catches range and anti-range tests. */
1467 if ((comp_code == LE_EXPR
1468 || comp_code == GT_EXPR)
1469 && TREE_CODE (val) == INTEGER_CST
1470 && TYPE_UNSIGNED (TREE_TYPE (val)))
1472 gimple *def_stmt = SSA_NAME_DEF_STMT (name);
1473 tree cst2 = NULL_TREE, name2 = NULL_TREE, name3 = NULL_TREE;
1475 /* Extract CST2 from the (optional) addition. */
1476 if (is_gimple_assign (def_stmt)
1477 && gimple_assign_rhs_code (def_stmt) == PLUS_EXPR)
1479 name2 = gimple_assign_rhs1 (def_stmt);
1480 cst2 = gimple_assign_rhs2 (def_stmt);
1481 if (TREE_CODE (name2) == SSA_NAME
1482 && TREE_CODE (cst2) == INTEGER_CST)
1483 def_stmt = SSA_NAME_DEF_STMT (name2);
1486 /* Extract NAME2 from the (optional) sign-changing cast. */
1487 if (gimple_assign_cast_p (def_stmt))
1489 if (CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (def_stmt))
1490 && ! TYPE_UNSIGNED (TREE_TYPE (gimple_assign_rhs1 (def_stmt)))
1491 && (TYPE_PRECISION (gimple_expr_type (def_stmt))
1492 == TYPE_PRECISION (TREE_TYPE (gimple_assign_rhs1 (def_stmt)))))
1493 name3 = gimple_assign_rhs1 (def_stmt);
1496 /* If name3 is used later, create an ASSERT_EXPR for it. */
1497 if (name3 != NULL_TREE
1498 && TREE_CODE (name3) == SSA_NAME
1499 && (cst2 == NULL_TREE
1500 || TREE_CODE (cst2) == INTEGER_CST)
1501 && INTEGRAL_TYPE_P (TREE_TYPE (name3)))
1503 tree tmp;
1505 /* Build an expression for the range test. */
1506 tmp = build1 (NOP_EXPR, TREE_TYPE (name), name3);
1507 if (cst2 != NULL_TREE)
1508 tmp = build2 (PLUS_EXPR, TREE_TYPE (name), tmp, cst2);
1509 add_assert_info (asserts, name3, tmp, comp_code, val);
1512 /* If name2 is used later, create an ASSERT_EXPR for it. */
1513 if (name2 != NULL_TREE
1514 && TREE_CODE (name2) == SSA_NAME
1515 && TREE_CODE (cst2) == INTEGER_CST
1516 && INTEGRAL_TYPE_P (TREE_TYPE (name2)))
1518 tree tmp;
1520 /* Build an expression for the range test. */
1521 tmp = name2;
1522 if (TREE_TYPE (name) != TREE_TYPE (name2))
1523 tmp = build1 (NOP_EXPR, TREE_TYPE (name), tmp);
1524 if (cst2 != NULL_TREE)
1525 tmp = build2 (PLUS_EXPR, TREE_TYPE (name), tmp, cst2);
1526 add_assert_info (asserts, name2, tmp, comp_code, val);
1530 /* In the case of post-in/decrement tests like if (i++) ... and uses
1531 of the in/decremented value on the edge the extra name we want to
1532 assert for is not on the def chain of the name compared. Instead
1533 it is in the set of use stmts.
1534 Similar cases happen for conversions that were simplified through
1535 fold_{sign_changed,widened}_comparison. */
1536 if ((comp_code == NE_EXPR
1537 || comp_code == EQ_EXPR)
1538 && TREE_CODE (val) == INTEGER_CST)
1540 imm_use_iterator ui;
1541 gimple *use_stmt;
1542 FOR_EACH_IMM_USE_STMT (use_stmt, ui, name)
1544 if (!is_gimple_assign (use_stmt))
1545 continue;
1547 /* Cut off to use-stmts that are dominating the predecessor. */
1548 if (!dominated_by_p (CDI_DOMINATORS, e->src, gimple_bb (use_stmt)))
1549 continue;
1551 tree name2 = gimple_assign_lhs (use_stmt);
1552 if (TREE_CODE (name2) != SSA_NAME)
1553 continue;
1555 enum tree_code code = gimple_assign_rhs_code (use_stmt);
1556 tree cst;
1557 if (code == PLUS_EXPR
1558 || code == MINUS_EXPR)
1560 cst = gimple_assign_rhs2 (use_stmt);
1561 if (TREE_CODE (cst) != INTEGER_CST)
1562 continue;
1563 cst = int_const_binop (code, val, cst);
1565 else if (CONVERT_EXPR_CODE_P (code))
1567 /* For truncating conversions we cannot record
1568 an inequality. */
1569 if (comp_code == NE_EXPR
1570 && (TYPE_PRECISION (TREE_TYPE (name2))
1571 < TYPE_PRECISION (TREE_TYPE (name))))
1572 continue;
1573 cst = fold_convert (TREE_TYPE (name2), val);
1575 else
1576 continue;
1578 if (TREE_OVERFLOW_P (cst))
1579 cst = drop_tree_overflow (cst);
1580 add_assert_info (asserts, name2, name2, comp_code, cst);
1584 if (TREE_CODE_CLASS (comp_code) == tcc_comparison
1585 && TREE_CODE (val) == INTEGER_CST)
1587 gimple *def_stmt = SSA_NAME_DEF_STMT (name);
1588 tree name2 = NULL_TREE, names[2], cst2 = NULL_TREE;
1589 tree val2 = NULL_TREE;
1590 unsigned int prec = TYPE_PRECISION (TREE_TYPE (val));
1591 wide_int mask = wi::zero (prec);
1592 unsigned int nprec = prec;
1593 enum tree_code rhs_code = ERROR_MARK;
1595 if (is_gimple_assign (def_stmt))
1596 rhs_code = gimple_assign_rhs_code (def_stmt);
1598 /* In the case of NAME != CST1 where NAME = A +- CST2 we can
1599 assert that A != CST1 -+ CST2. */
1600 if ((comp_code == EQ_EXPR || comp_code == NE_EXPR)
1601 && (rhs_code == PLUS_EXPR || rhs_code == MINUS_EXPR))
1603 tree op0 = gimple_assign_rhs1 (def_stmt);
1604 tree op1 = gimple_assign_rhs2 (def_stmt);
1605 if (TREE_CODE (op0) == SSA_NAME
1606 && TREE_CODE (op1) == INTEGER_CST)
1608 enum tree_code reverse_op = (rhs_code == PLUS_EXPR
1609 ? MINUS_EXPR : PLUS_EXPR);
1610 op1 = int_const_binop (reverse_op, val, op1);
1611 if (TREE_OVERFLOW (op1))
1612 op1 = drop_tree_overflow (op1);
1613 add_assert_info (asserts, op0, op0, comp_code, op1);
1617 /* Add asserts for NAME cmp CST and NAME being defined
1618 as NAME = (int) NAME2. */
1619 if (!TYPE_UNSIGNED (TREE_TYPE (val))
1620 && (comp_code == LE_EXPR || comp_code == LT_EXPR
1621 || comp_code == GT_EXPR || comp_code == GE_EXPR)
1622 && gimple_assign_cast_p (def_stmt))
1624 name2 = gimple_assign_rhs1 (def_stmt);
1625 if (CONVERT_EXPR_CODE_P (rhs_code)
1626 && TREE_CODE (name2) == SSA_NAME
1627 && INTEGRAL_TYPE_P (TREE_TYPE (name2))
1628 && TYPE_UNSIGNED (TREE_TYPE (name2))
1629 && prec == TYPE_PRECISION (TREE_TYPE (name2))
1630 && (comp_code == LE_EXPR || comp_code == GT_EXPR
1631 || !tree_int_cst_equal (val,
1632 TYPE_MIN_VALUE (TREE_TYPE (val)))))
1634 tree tmp, cst;
1635 enum tree_code new_comp_code = comp_code;
1637 cst = fold_convert (TREE_TYPE (name2),
1638 TYPE_MIN_VALUE (TREE_TYPE (val)));
1639 /* Build an expression for the range test. */
1640 tmp = build2 (PLUS_EXPR, TREE_TYPE (name2), name2, cst);
1641 cst = fold_build2 (PLUS_EXPR, TREE_TYPE (name2), cst,
1642 fold_convert (TREE_TYPE (name2), val));
1643 if (comp_code == LT_EXPR || comp_code == GE_EXPR)
1645 new_comp_code = comp_code == LT_EXPR ? LE_EXPR : GT_EXPR;
1646 cst = fold_build2 (MINUS_EXPR, TREE_TYPE (name2), cst,
1647 build_int_cst (TREE_TYPE (name2), 1));
1649 add_assert_info (asserts, name2, tmp, new_comp_code, cst);
1653 /* Add asserts for NAME cmp CST and NAME being defined as
1654 NAME = NAME2 >> CST2.
1656 Extract CST2 from the right shift. */
1657 if (rhs_code == RSHIFT_EXPR)
1659 name2 = gimple_assign_rhs1 (def_stmt);
1660 cst2 = gimple_assign_rhs2 (def_stmt);
1661 if (TREE_CODE (name2) == SSA_NAME
1662 && tree_fits_uhwi_p (cst2)
1663 && INTEGRAL_TYPE_P (TREE_TYPE (name2))
1664 && IN_RANGE (tree_to_uhwi (cst2), 1, prec - 1)
1665 && type_has_mode_precision_p (TREE_TYPE (val)))
1667 mask = wi::mask (tree_to_uhwi (cst2), false, prec);
1668 val2 = fold_binary (LSHIFT_EXPR, TREE_TYPE (val), val, cst2);
1671 if (val2 != NULL_TREE
1672 && TREE_CODE (val2) == INTEGER_CST
1673 && simple_cst_equal (fold_build2 (RSHIFT_EXPR,
1674 TREE_TYPE (val),
1675 val2, cst2), val))
1677 enum tree_code new_comp_code = comp_code;
1678 tree tmp, new_val;
1680 tmp = name2;
1681 if (comp_code == EQ_EXPR || comp_code == NE_EXPR)
1683 if (!TYPE_UNSIGNED (TREE_TYPE (val)))
1685 tree type = build_nonstandard_integer_type (prec, 1);
1686 tmp = build1 (NOP_EXPR, type, name2);
1687 val2 = fold_convert (type, val2);
1689 tmp = fold_build2 (MINUS_EXPR, TREE_TYPE (tmp), tmp, val2);
1690 new_val = wide_int_to_tree (TREE_TYPE (tmp), mask);
1691 new_comp_code = comp_code == EQ_EXPR ? LE_EXPR : GT_EXPR;
1693 else if (comp_code == LT_EXPR || comp_code == GE_EXPR)
1695 wide_int minval
1696 = wi::min_value (prec, TYPE_SIGN (TREE_TYPE (val)));
1697 new_val = val2;
1698 if (minval == wi::to_wide (new_val))
1699 new_val = NULL_TREE;
1701 else
1703 wide_int maxval
1704 = wi::max_value (prec, TYPE_SIGN (TREE_TYPE (val)));
1705 mask |= wi::to_wide (val2);
1706 if (wi::eq_p (mask, maxval))
1707 new_val = NULL_TREE;
1708 else
1709 new_val = wide_int_to_tree (TREE_TYPE (val2), mask);
1712 if (new_val)
1713 add_assert_info (asserts, name2, tmp, new_comp_code, new_val);
1716 /* If we have a conversion that doesn't change the value of the source
1717 simply register the same assert for it. */
1718 if (CONVERT_EXPR_CODE_P (rhs_code))
1720 wide_int rmin, rmax;
1721 tree rhs1 = gimple_assign_rhs1 (def_stmt);
1722 if (INTEGRAL_TYPE_P (TREE_TYPE (rhs1))
1723 && TREE_CODE (rhs1) == SSA_NAME
1724 /* Make sure the relation preserves the upper/lower boundary of
1725 the range conservatively. */
1726 && (comp_code == NE_EXPR
1727 || comp_code == EQ_EXPR
1728 || (TYPE_SIGN (TREE_TYPE (name))
1729 == TYPE_SIGN (TREE_TYPE (rhs1)))
1730 || ((comp_code == LE_EXPR
1731 || comp_code == LT_EXPR)
1732 && !TYPE_UNSIGNED (TREE_TYPE (rhs1)))
1733 || ((comp_code == GE_EXPR
1734 || comp_code == GT_EXPR)
1735 && TYPE_UNSIGNED (TREE_TYPE (rhs1))))
1736 /* And the conversion does not alter the value we compare
1737 against and all values in rhs1 can be represented in
1738 the converted to type. */
1739 && int_fits_type_p (val, TREE_TYPE (rhs1))
1740 && ((TYPE_PRECISION (TREE_TYPE (name))
1741 > TYPE_PRECISION (TREE_TYPE (rhs1)))
1742 || (get_range_info (rhs1, &rmin, &rmax) == VR_RANGE
1743 && wi::fits_to_tree_p
1744 (widest_int::from (rmin,
1745 TYPE_SIGN (TREE_TYPE (rhs1))),
1746 TREE_TYPE (name))
1747 && wi::fits_to_tree_p
1748 (widest_int::from (rmax,
1749 TYPE_SIGN (TREE_TYPE (rhs1))),
1750 TREE_TYPE (name)))))
1751 add_assert_info (asserts, rhs1, rhs1,
1752 comp_code, fold_convert (TREE_TYPE (rhs1), val));
1755 /* Add asserts for NAME cmp CST and NAME being defined as
1756 NAME = NAME2 & CST2.
1758 Extract CST2 from the and.
1760 Also handle
1761 NAME = (unsigned) NAME2;
1762 casts where NAME's type is unsigned and has smaller precision
1763 than NAME2's type as if it was NAME = NAME2 & MASK. */
1764 names[0] = NULL_TREE;
1765 names[1] = NULL_TREE;
1766 cst2 = NULL_TREE;
1767 if (rhs_code == BIT_AND_EXPR
1768 || (CONVERT_EXPR_CODE_P (rhs_code)
1769 && INTEGRAL_TYPE_P (TREE_TYPE (val))
1770 && TYPE_UNSIGNED (TREE_TYPE (val))
1771 && TYPE_PRECISION (TREE_TYPE (gimple_assign_rhs1 (def_stmt)))
1772 > prec))
1774 name2 = gimple_assign_rhs1 (def_stmt);
1775 if (rhs_code == BIT_AND_EXPR)
1776 cst2 = gimple_assign_rhs2 (def_stmt);
1777 else
1779 cst2 = TYPE_MAX_VALUE (TREE_TYPE (val));
1780 nprec = TYPE_PRECISION (TREE_TYPE (name2));
1782 if (TREE_CODE (name2) == SSA_NAME
1783 && INTEGRAL_TYPE_P (TREE_TYPE (name2))
1784 && TREE_CODE (cst2) == INTEGER_CST
1785 && !integer_zerop (cst2)
1786 && (nprec > 1
1787 || TYPE_UNSIGNED (TREE_TYPE (val))))
1789 gimple *def_stmt2 = SSA_NAME_DEF_STMT (name2);
1790 if (gimple_assign_cast_p (def_stmt2))
1792 names[1] = gimple_assign_rhs1 (def_stmt2);
1793 if (!CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (def_stmt2))
1794 || TREE_CODE (names[1]) != SSA_NAME
1795 || !INTEGRAL_TYPE_P (TREE_TYPE (names[1]))
1796 || (TYPE_PRECISION (TREE_TYPE (name2))
1797 != TYPE_PRECISION (TREE_TYPE (names[1]))))
1798 names[1] = NULL_TREE;
1800 names[0] = name2;
1803 if (names[0] || names[1])
1805 wide_int minv, maxv, valv, cst2v;
1806 wide_int tem, sgnbit;
1807 bool valid_p = false, valn, cst2n;
1808 enum tree_code ccode = comp_code;
1810 valv = wide_int::from (wi::to_wide (val), nprec, UNSIGNED);
1811 cst2v = wide_int::from (wi::to_wide (cst2), nprec, UNSIGNED);
1812 valn = wi::neg_p (valv, TYPE_SIGN (TREE_TYPE (val)));
1813 cst2n = wi::neg_p (cst2v, TYPE_SIGN (TREE_TYPE (val)));
1814 /* If CST2 doesn't have most significant bit set,
1815 but VAL is negative, we have comparison like
1816 if ((x & 0x123) > -4) (always true). Just give up. */
1817 if (!cst2n && valn)
1818 ccode = ERROR_MARK;
1819 if (cst2n)
1820 sgnbit = wi::set_bit_in_zero (nprec - 1, nprec);
1821 else
1822 sgnbit = wi::zero (nprec);
1823 minv = valv & cst2v;
1824 switch (ccode)
1826 case EQ_EXPR:
1827 /* Minimum unsigned value for equality is VAL & CST2
1828 (should be equal to VAL, otherwise we probably should
1829 have folded the comparison into false) and
1830 maximum unsigned value is VAL | ~CST2. */
1831 maxv = valv | ~cst2v;
1832 valid_p = true;
1833 break;
1835 case NE_EXPR:
1836 tem = valv | ~cst2v;
1837 /* If VAL is 0, handle (X & CST2) != 0 as (X & CST2) > 0U. */
1838 if (valv == 0)
1840 cst2n = false;
1841 sgnbit = wi::zero (nprec);
1842 goto gt_expr;
1844 /* If (VAL | ~CST2) is all ones, handle it as
1845 (X & CST2) < VAL. */
1846 if (tem == -1)
1848 cst2n = false;
1849 valn = false;
1850 sgnbit = wi::zero (nprec);
1851 goto lt_expr;
1853 if (!cst2n && wi::neg_p (cst2v))
1854 sgnbit = wi::set_bit_in_zero (nprec - 1, nprec);
1855 if (sgnbit != 0)
1857 if (valv == sgnbit)
1859 cst2n = true;
1860 valn = true;
1861 goto gt_expr;
1863 if (tem == wi::mask (nprec - 1, false, nprec))
1865 cst2n = true;
1866 goto lt_expr;
1868 if (!cst2n)
1869 sgnbit = wi::zero (nprec);
1871 break;
1873 case GE_EXPR:
1874 /* Minimum unsigned value for >= if (VAL & CST2) == VAL
1875 is VAL and maximum unsigned value is ~0. For signed
1876 comparison, if CST2 doesn't have most significant bit
1877 set, handle it similarly. If CST2 has MSB set,
1878 the minimum is the same, and maximum is ~0U/2. */
1879 if (minv != valv)
1881 /* If (VAL & CST2) != VAL, X & CST2 can't be equal to
1882 VAL. */
1883 minv = masked_increment (valv, cst2v, sgnbit, nprec);
1884 if (minv == valv)
1885 break;
1887 maxv = wi::mask (nprec - (cst2n ? 1 : 0), false, nprec);
1888 valid_p = true;
1889 break;
1891 case GT_EXPR:
1892 gt_expr:
1893 /* Find out smallest MINV where MINV > VAL
1894 && (MINV & CST2) == MINV, if any. If VAL is signed and
1895 CST2 has MSB set, compute it biased by 1 << (nprec - 1). */
1896 minv = masked_increment (valv, cst2v, sgnbit, nprec);
1897 if (minv == valv)
1898 break;
1899 maxv = wi::mask (nprec - (cst2n ? 1 : 0), false, nprec);
1900 valid_p = true;
1901 break;
1903 case LE_EXPR:
1904 /* Minimum unsigned value for <= is 0 and maximum
1905 unsigned value is VAL | ~CST2 if (VAL & CST2) == VAL.
1906 Otherwise, find smallest VAL2 where VAL2 > VAL
1907 && (VAL2 & CST2) == VAL2 and use (VAL2 - 1) | ~CST2
1908 as maximum.
1909 For signed comparison, if CST2 doesn't have most
1910 significant bit set, handle it similarly. If CST2 has
1911 MSB set, the maximum is the same and minimum is INT_MIN. */
1912 if (minv == valv)
1913 maxv = valv;
1914 else
1916 maxv = masked_increment (valv, cst2v, sgnbit, nprec);
1917 if (maxv == valv)
1918 break;
1919 maxv -= 1;
1921 maxv |= ~cst2v;
1922 minv = sgnbit;
1923 valid_p = true;
1924 break;
1926 case LT_EXPR:
1927 lt_expr:
1928 /* Minimum unsigned value for < is 0 and maximum
1929 unsigned value is (VAL-1) | ~CST2 if (VAL & CST2) == VAL.
1930 Otherwise, find smallest VAL2 where VAL2 > VAL
1931 && (VAL2 & CST2) == VAL2 and use (VAL2 - 1) | ~CST2
1932 as maximum.
1933 For signed comparison, if CST2 doesn't have most
1934 significant bit set, handle it similarly. If CST2 has
1935 MSB set, the maximum is the same and minimum is INT_MIN. */
1936 if (minv == valv)
1938 if (valv == sgnbit)
1939 break;
1940 maxv = valv;
1942 else
1944 maxv = masked_increment (valv, cst2v, sgnbit, nprec);
1945 if (maxv == valv)
1946 break;
1948 maxv -= 1;
1949 maxv |= ~cst2v;
1950 minv = sgnbit;
1951 valid_p = true;
1952 break;
1954 default:
1955 break;
1957 if (valid_p
1958 && (maxv - minv) != -1)
1960 tree tmp, new_val, type;
1961 int i;
1963 for (i = 0; i < 2; i++)
1964 if (names[i])
1966 wide_int maxv2 = maxv;
1967 tmp = names[i];
1968 type = TREE_TYPE (names[i]);
1969 if (!TYPE_UNSIGNED (type))
1971 type = build_nonstandard_integer_type (nprec, 1);
1972 tmp = build1 (NOP_EXPR, type, names[i]);
1974 if (minv != 0)
1976 tmp = build2 (PLUS_EXPR, type, tmp,
1977 wide_int_to_tree (type, -minv));
1978 maxv2 = maxv - minv;
1980 new_val = wide_int_to_tree (type, maxv2);
1981 add_assert_info (asserts, names[i], tmp, LE_EXPR, new_val);
1988 /* OP is an operand of a truth value expression which is known to have
1989 a particular value. Register any asserts for OP and for any
1990 operands in OP's defining statement.
1992 If CODE is EQ_EXPR, then we want to register OP is zero (false),
1993 if CODE is NE_EXPR, then we want to register OP is nonzero (true). */
1995 static void
1996 register_edge_assert_for_1 (tree op, enum tree_code code,
1997 edge e, vec<assert_info> &asserts)
1999 gimple *op_def;
2000 tree val;
2001 enum tree_code rhs_code;
2003 /* We only care about SSA_NAMEs. */
2004 if (TREE_CODE (op) != SSA_NAME)
2005 return;
2007 /* We know that OP will have a zero or nonzero value. */
2008 val = build_int_cst (TREE_TYPE (op), 0);
2009 add_assert_info (asserts, op, op, code, val);
2011 /* Now look at how OP is set. If it's set from a comparison,
2012 a truth operation or some bit operations, then we may be able
2013 to register information about the operands of that assignment. */
2014 op_def = SSA_NAME_DEF_STMT (op);
2015 if (gimple_code (op_def) != GIMPLE_ASSIGN)
2016 return;
2018 rhs_code = gimple_assign_rhs_code (op_def);
2020 if (TREE_CODE_CLASS (rhs_code) == tcc_comparison)
2022 bool invert = (code == EQ_EXPR ? true : false);
2023 tree op0 = gimple_assign_rhs1 (op_def);
2024 tree op1 = gimple_assign_rhs2 (op_def);
2026 if (TREE_CODE (op0) == SSA_NAME)
2027 register_edge_assert_for_2 (op0, e, rhs_code, op0, op1, invert, asserts);
2028 if (TREE_CODE (op1) == SSA_NAME)
2029 register_edge_assert_for_2 (op1, e, rhs_code, op0, op1, invert, asserts);
2031 else if ((code == NE_EXPR
2032 && gimple_assign_rhs_code (op_def) == BIT_AND_EXPR)
2033 || (code == EQ_EXPR
2034 && gimple_assign_rhs_code (op_def) == BIT_IOR_EXPR))
2036 /* Recurse on each operand. */
2037 tree op0 = gimple_assign_rhs1 (op_def);
2038 tree op1 = gimple_assign_rhs2 (op_def);
2039 if (TREE_CODE (op0) == SSA_NAME
2040 && has_single_use (op0))
2041 register_edge_assert_for_1 (op0, code, e, asserts);
2042 if (TREE_CODE (op1) == SSA_NAME
2043 && has_single_use (op1))
2044 register_edge_assert_for_1 (op1, code, e, asserts);
2046 else if (gimple_assign_rhs_code (op_def) == BIT_NOT_EXPR
2047 && TYPE_PRECISION (TREE_TYPE (gimple_assign_lhs (op_def))) == 1)
2049 /* Recurse, flipping CODE. */
2050 code = invert_tree_comparison (code, false);
2051 register_edge_assert_for_1 (gimple_assign_rhs1 (op_def), code, e, asserts);
2053 else if (gimple_assign_rhs_code (op_def) == SSA_NAME)
2055 /* Recurse through the copy. */
2056 register_edge_assert_for_1 (gimple_assign_rhs1 (op_def), code, e, asserts);
2058 else if (CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (op_def)))
2060 /* Recurse through the type conversion, unless it is a narrowing
2061 conversion or conversion from non-integral type. */
2062 tree rhs = gimple_assign_rhs1 (op_def);
2063 if (INTEGRAL_TYPE_P (TREE_TYPE (rhs))
2064 && (TYPE_PRECISION (TREE_TYPE (rhs))
2065 <= TYPE_PRECISION (TREE_TYPE (op))))
2066 register_edge_assert_for_1 (rhs, code, e, asserts);
2070 /* Check if comparison
2071 NAME COND_OP INTEGER_CST
2072 has a form of
2073 (X & 11...100..0) COND_OP XX...X00...0
2074 Such comparison can yield assertions like
2075 X >= XX...X00...0
2076 X <= XX...X11...1
2077 in case of COND_OP being EQ_EXPR or
2078 X < XX...X00...0
2079 X > XX...X11...1
2080 in case of NE_EXPR. */
2082 static bool
2083 is_masked_range_test (tree name, tree valt, enum tree_code cond_code,
2084 tree *new_name, tree *low, enum tree_code *low_code,
2085 tree *high, enum tree_code *high_code)
2087 gimple *def_stmt = SSA_NAME_DEF_STMT (name);
2089 if (!is_gimple_assign (def_stmt)
2090 || gimple_assign_rhs_code (def_stmt) != BIT_AND_EXPR)
2091 return false;
2093 tree t = gimple_assign_rhs1 (def_stmt);
2094 tree maskt = gimple_assign_rhs2 (def_stmt);
2095 if (TREE_CODE (t) != SSA_NAME || TREE_CODE (maskt) != INTEGER_CST)
2096 return false;
2098 wi::tree_to_wide_ref mask = wi::to_wide (maskt);
2099 wide_int inv_mask = ~mask;
2100 /* Must have been removed by now so don't bother optimizing. */
2101 if (mask == 0 || inv_mask == 0)
2102 return false;
2104 /* Assume VALT is INTEGER_CST. */
2105 wi::tree_to_wide_ref val = wi::to_wide (valt);
2107 if ((inv_mask & (inv_mask + 1)) != 0
2108 || (val & mask) != val)
2109 return false;
2111 bool is_range = cond_code == EQ_EXPR;
2113 tree type = TREE_TYPE (t);
2114 wide_int min = wi::min_value (type),
2115 max = wi::max_value (type);
2117 if (is_range)
2119 *low_code = val == min ? ERROR_MARK : GE_EXPR;
2120 *high_code = val == max ? ERROR_MARK : LE_EXPR;
2122 else
2124 /* We can still generate assertion if one of alternatives
2125 is known to always be false. */
2126 if (val == min)
2128 *low_code = (enum tree_code) 0;
2129 *high_code = GT_EXPR;
2131 else if ((val | inv_mask) == max)
2133 *low_code = LT_EXPR;
2134 *high_code = (enum tree_code) 0;
2136 else
2137 return false;
2140 *new_name = t;
2141 *low = wide_int_to_tree (type, val);
2142 *high = wide_int_to_tree (type, val | inv_mask);
2144 return true;
2147 /* Try to register an edge assertion for SSA name NAME on edge E for
2148 the condition COND contributing to the conditional jump pointed to by
2149 SI. */
2151 void
2152 register_edge_assert_for (tree name, edge e,
2153 enum tree_code cond_code, tree cond_op0,
2154 tree cond_op1, vec<assert_info> &asserts)
2156 tree val;
2157 enum tree_code comp_code;
2158 bool is_else_edge = (e->flags & EDGE_FALSE_VALUE) != 0;
2160 /* Do not attempt to infer anything in names that flow through
2161 abnormal edges. */
2162 if (SSA_NAME_OCCURS_IN_ABNORMAL_PHI (name))
2163 return;
2165 if (!extract_code_and_val_from_cond_with_ops (name, cond_code,
2166 cond_op0, cond_op1,
2167 is_else_edge,
2168 &comp_code, &val))
2169 return;
2171 /* Register ASSERT_EXPRs for name. */
2172 register_edge_assert_for_2 (name, e, cond_code, cond_op0,
2173 cond_op1, is_else_edge, asserts);
2176 /* If COND is effectively an equality test of an SSA_NAME against
2177 the value zero or one, then we may be able to assert values
2178 for SSA_NAMEs which flow into COND. */
2180 /* In the case of NAME == 1 or NAME != 0, for BIT_AND_EXPR defining
2181 statement of NAME we can assert both operands of the BIT_AND_EXPR
2182 have nonzero value. */
2183 if ((comp_code == EQ_EXPR && integer_onep (val))
2184 || (comp_code == NE_EXPR && integer_zerop (val)))
2186 gimple *def_stmt = SSA_NAME_DEF_STMT (name);
2188 if (is_gimple_assign (def_stmt)
2189 && gimple_assign_rhs_code (def_stmt) == BIT_AND_EXPR)
2191 tree op0 = gimple_assign_rhs1 (def_stmt);
2192 tree op1 = gimple_assign_rhs2 (def_stmt);
2193 register_edge_assert_for_1 (op0, NE_EXPR, e, asserts);
2194 register_edge_assert_for_1 (op1, NE_EXPR, e, asserts);
2196 else if (is_gimple_assign (def_stmt)
2197 && (TREE_CODE_CLASS (gimple_assign_rhs_code (def_stmt))
2198 == tcc_comparison))
2199 register_edge_assert_for_1 (name, NE_EXPR, e, asserts);
2202 /* In the case of NAME == 0 or NAME != 1, for BIT_IOR_EXPR defining
2203 statement of NAME we can assert both operands of the BIT_IOR_EXPR
2204 have zero value. */
2205 if ((comp_code == EQ_EXPR && integer_zerop (val))
2206 || (comp_code == NE_EXPR
2207 && integer_onep (val)
2208 && TYPE_PRECISION (TREE_TYPE (name)) == 1))
2210 gimple *def_stmt = SSA_NAME_DEF_STMT (name);
2212 /* For BIT_IOR_EXPR only if NAME == 0 both operands have
2213 necessarily zero value, or if type-precision is one. */
2214 if (is_gimple_assign (def_stmt)
2215 && gimple_assign_rhs_code (def_stmt) == BIT_IOR_EXPR)
2217 tree op0 = gimple_assign_rhs1 (def_stmt);
2218 tree op1 = gimple_assign_rhs2 (def_stmt);
2219 register_edge_assert_for_1 (op0, EQ_EXPR, e, asserts);
2220 register_edge_assert_for_1 (op1, EQ_EXPR, e, asserts);
2222 else if (is_gimple_assign (def_stmt)
2223 && (TREE_CODE_CLASS (gimple_assign_rhs_code (def_stmt))
2224 == tcc_comparison))
2225 register_edge_assert_for_1 (name, EQ_EXPR, e, asserts);
2228 /* Sometimes we can infer ranges from (NAME & MASK) == VALUE. */
2229 if ((comp_code == EQ_EXPR || comp_code == NE_EXPR)
2230 && TREE_CODE (val) == INTEGER_CST)
2232 enum tree_code low_code, high_code;
2233 tree low, high;
2234 if (is_masked_range_test (name, val, comp_code, &name, &low,
2235 &low_code, &high, &high_code))
2237 if (low_code != ERROR_MARK)
2238 register_edge_assert_for_2 (name, e, low_code, name,
2239 low, /*invert*/false, asserts);
2240 if (high_code != ERROR_MARK)
2241 register_edge_assert_for_2 (name, e, high_code, name,
2242 high, /*invert*/false, asserts);
2247 /* Handle
2248 _4 = x_3 & 31;
2249 if (_4 != 0)
2250 goto <bb 6>;
2251 else
2252 goto <bb 7>;
2253 <bb 6>:
2254 __builtin_unreachable ();
2255 <bb 7>:
2256 x_5 = ASSERT_EXPR <x_3, ...>;
2257 If x_3 has no other immediate uses (checked by caller),
2258 var is the x_3 var from ASSERT_EXPR, we can clear low 5 bits
2259 from the non-zero bitmask. */
2261 void
2262 maybe_set_nonzero_bits (edge e, tree var)
2264 basic_block cond_bb = e->src;
2265 gimple *stmt = last_stmt (cond_bb);
2266 tree cst;
2268 if (stmt == NULL
2269 || gimple_code (stmt) != GIMPLE_COND
2270 || gimple_cond_code (stmt) != ((e->flags & EDGE_TRUE_VALUE)
2271 ? EQ_EXPR : NE_EXPR)
2272 || TREE_CODE (gimple_cond_lhs (stmt)) != SSA_NAME
2273 || !integer_zerop (gimple_cond_rhs (stmt)))
2274 return;
2276 stmt = SSA_NAME_DEF_STMT (gimple_cond_lhs (stmt));
2277 if (!is_gimple_assign (stmt)
2278 || gimple_assign_rhs_code (stmt) != BIT_AND_EXPR
2279 || TREE_CODE (gimple_assign_rhs2 (stmt)) != INTEGER_CST)
2280 return;
2281 if (gimple_assign_rhs1 (stmt) != var)
2283 gimple *stmt2;
2285 if (TREE_CODE (gimple_assign_rhs1 (stmt)) != SSA_NAME)
2286 return;
2287 stmt2 = SSA_NAME_DEF_STMT (gimple_assign_rhs1 (stmt));
2288 if (!gimple_assign_cast_p (stmt2)
2289 || gimple_assign_rhs1 (stmt2) != var
2290 || !CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (stmt2))
2291 || (TYPE_PRECISION (TREE_TYPE (gimple_assign_rhs1 (stmt)))
2292 != TYPE_PRECISION (TREE_TYPE (var))))
2293 return;
2295 cst = gimple_assign_rhs2 (stmt);
2296 set_nonzero_bits (var, wi::bit_and_not (get_nonzero_bits (var),
2297 wi::to_wide (cst)));
2300 /* Return true if STMT is interesting for VRP. */
2302 bool
2303 stmt_interesting_for_vrp (gimple *stmt)
2305 if (gimple_code (stmt) == GIMPLE_PHI)
2307 tree res = gimple_phi_result (stmt);
2308 return (!virtual_operand_p (res)
2309 && (INTEGRAL_TYPE_P (TREE_TYPE (res))
2310 || POINTER_TYPE_P (TREE_TYPE (res))));
2312 else if (is_gimple_assign (stmt) || is_gimple_call (stmt))
2314 tree lhs = gimple_get_lhs (stmt);
2316 /* In general, assignments with virtual operands are not useful
2317 for deriving ranges, with the obvious exception of calls to
2318 builtin functions. */
2319 if (lhs && TREE_CODE (lhs) == SSA_NAME
2320 && (INTEGRAL_TYPE_P (TREE_TYPE (lhs))
2321 || POINTER_TYPE_P (TREE_TYPE (lhs)))
2322 && (is_gimple_call (stmt)
2323 || !gimple_vuse (stmt)))
2324 return true;
2325 else if (is_gimple_call (stmt) && gimple_call_internal_p (stmt))
2326 switch (gimple_call_internal_fn (stmt))
2328 case IFN_ADD_OVERFLOW:
2329 case IFN_SUB_OVERFLOW:
2330 case IFN_MUL_OVERFLOW:
2331 case IFN_ATOMIC_COMPARE_EXCHANGE:
2332 /* These internal calls return _Complex integer type,
2333 but are interesting to VRP nevertheless. */
2334 if (lhs && TREE_CODE (lhs) == SSA_NAME)
2335 return true;
2336 break;
2337 default:
2338 break;
2341 else if (gimple_code (stmt) == GIMPLE_COND
2342 || gimple_code (stmt) == GIMPLE_SWITCH)
2343 return true;
2345 return false;
2349 /* Return the LHS of any ASSERT_EXPR where OP appears as the first
2350 argument to the ASSERT_EXPR and in which the ASSERT_EXPR dominates
2351 BB. If no such ASSERT_EXPR is found, return OP. */
2353 static tree
2354 lhs_of_dominating_assert (tree op, basic_block bb, gimple *stmt)
2356 imm_use_iterator imm_iter;
2357 gimple *use_stmt;
2358 use_operand_p use_p;
2360 if (TREE_CODE (op) == SSA_NAME)
2362 FOR_EACH_IMM_USE_FAST (use_p, imm_iter, op)
2364 use_stmt = USE_STMT (use_p);
2365 if (use_stmt != stmt
2366 && gimple_assign_single_p (use_stmt)
2367 && TREE_CODE (gimple_assign_rhs1 (use_stmt)) == ASSERT_EXPR
2368 && TREE_OPERAND (gimple_assign_rhs1 (use_stmt), 0) == op
2369 && dominated_by_p (CDI_DOMINATORS, bb, gimple_bb (use_stmt)))
2370 return gimple_assign_lhs (use_stmt);
2373 return op;
2376 /* A hack. */
2377 static class vr_values *x_vr_values;
2379 /* Searches the case label vector VEC for the index *IDX of the CASE_LABEL
2380 that includes the value VAL. The search is restricted to the range
2381 [START_IDX, n - 1] where n is the size of VEC.
2383 If there is a CASE_LABEL for VAL, its index is placed in IDX and true is
2384 returned.
2386 If there is no CASE_LABEL for VAL and there is one that is larger than VAL,
2387 it is placed in IDX and false is returned.
2389 If VAL is larger than any CASE_LABEL, n is placed on IDX and false is
2390 returned. */
2392 bool
2393 find_case_label_index (gswitch *stmt, size_t start_idx, tree val, size_t *idx)
2395 size_t n = gimple_switch_num_labels (stmt);
2396 size_t low, high;
2398 /* Find case label for minimum of the value range or the next one.
2399 At each iteration we are searching in [low, high - 1]. */
2401 for (low = start_idx, high = n; high != low; )
2403 tree t;
2404 int cmp;
2405 /* Note that i != high, so we never ask for n. */
2406 size_t i = (high + low) / 2;
2407 t = gimple_switch_label (stmt, i);
2409 /* Cache the result of comparing CASE_LOW and val. */
2410 cmp = tree_int_cst_compare (CASE_LOW (t), val);
2412 if (cmp == 0)
2414 /* Ranges cannot be empty. */
2415 *idx = i;
2416 return true;
2418 else if (cmp > 0)
2419 high = i;
2420 else
2422 low = i + 1;
2423 if (CASE_HIGH (t) != NULL
2424 && tree_int_cst_compare (CASE_HIGH (t), val) >= 0)
2426 *idx = i;
2427 return true;
2432 *idx = high;
2433 return false;
2436 /* Searches the case label vector VEC for the range of CASE_LABELs that is used
2437 for values between MIN and MAX. The first index is placed in MIN_IDX. The
2438 last index is placed in MAX_IDX. If the range of CASE_LABELs is empty
2439 then MAX_IDX < MIN_IDX.
2440 Returns true if the default label is not needed. */
2442 bool
2443 find_case_label_range (gswitch *stmt, tree min, tree max, size_t *min_idx,
2444 size_t *max_idx)
2446 size_t i, j;
2447 bool min_take_default = !find_case_label_index (stmt, 1, min, &i);
2448 bool max_take_default = !find_case_label_index (stmt, i, max, &j);
2450 if (i == j
2451 && min_take_default
2452 && max_take_default)
2454 /* Only the default case label reached.
2455 Return an empty range. */
2456 *min_idx = 1;
2457 *max_idx = 0;
2458 return false;
2460 else
2462 bool take_default = min_take_default || max_take_default;
2463 tree low, high;
2464 size_t k;
2466 if (max_take_default)
2467 j--;
2469 /* If the case label range is continuous, we do not need
2470 the default case label. Verify that. */
2471 high = CASE_LOW (gimple_switch_label (stmt, i));
2472 if (CASE_HIGH (gimple_switch_label (stmt, i)))
2473 high = CASE_HIGH (gimple_switch_label (stmt, i));
2474 for (k = i + 1; k <= j; ++k)
2476 low = CASE_LOW (gimple_switch_label (stmt, k));
2477 if (!integer_onep (int_const_binop (MINUS_EXPR, low, high)))
2479 take_default = true;
2480 break;
2482 high = low;
2483 if (CASE_HIGH (gimple_switch_label (stmt, k)))
2484 high = CASE_HIGH (gimple_switch_label (stmt, k));
2487 *min_idx = i;
2488 *max_idx = j;
2489 return !take_default;
2493 /* Given a SWITCH_STMT, return the case label that encompasses the
2494 known possible values for the switch operand. RANGE_OF_OP is a
2495 range for the known values of the switch operand. */
2497 tree
2498 find_case_label_range (gswitch *switch_stmt, const irange *range_of_op)
2500 if (range_of_op->undefined_p ()
2501 || range_of_op->varying_p ()
2502 || range_of_op->symbolic_p ())
2503 return NULL_TREE;
2505 size_t i, j;
2506 tree op = gimple_switch_index (switch_stmt);
2507 tree type = TREE_TYPE (op);
2508 tree tmin = wide_int_to_tree (type, range_of_op->lower_bound ());
2509 tree tmax = wide_int_to_tree (type, range_of_op->upper_bound ());
2510 find_case_label_range (switch_stmt, tmin, tmax, &i, &j);
2511 if (i == j)
2513 /* Look for exactly one label that encompasses the range of
2514 the operand. */
2515 tree label = gimple_switch_label (switch_stmt, i);
2516 tree case_high
2517 = CASE_HIGH (label) ? CASE_HIGH (label) : CASE_LOW (label);
2518 int_range_max label_range (CASE_LOW (label), case_high);
2519 if (!types_compatible_p (label_range.type (), range_of_op->type ()))
2520 range_cast (label_range, range_of_op->type ());
2521 label_range.intersect (range_of_op);
2522 if (label_range == *range_of_op)
2523 return label;
2525 else if (i > j)
2527 /* If there are no labels at all, take the default. */
2528 return gimple_switch_label (switch_stmt, 0);
2530 else
2532 /* Otherwise, there are various labels that can encompass
2533 the range of operand. In which case, see if the range of
2534 the operand is entirely *outside* the bounds of all the
2535 (non-default) case labels. If so, take the default. */
2536 unsigned n = gimple_switch_num_labels (switch_stmt);
2537 tree min_label = gimple_switch_label (switch_stmt, 1);
2538 tree max_label = gimple_switch_label (switch_stmt, n - 1);
2539 tree case_high = CASE_HIGH (max_label);
2540 if (!case_high)
2541 case_high = CASE_LOW (max_label);
2542 int_range_max label_range (CASE_LOW (min_label), case_high);
2543 if (!types_compatible_p (label_range.type (), range_of_op->type ()))
2544 range_cast (label_range, range_of_op->type ());
2545 label_range.intersect (range_of_op);
2546 if (label_range.undefined_p ())
2547 return gimple_switch_label (switch_stmt, 0);
2549 return NULL_TREE;
2552 struct case_info
2554 tree expr;
2555 basic_block bb;
2558 /* Location information for ASSERT_EXPRs. Each instance of this
2559 structure describes an ASSERT_EXPR for an SSA name. Since a single
2560 SSA name may have more than one assertion associated with it, these
2561 locations are kept in a linked list attached to the corresponding
2562 SSA name. */
2563 struct assert_locus
2565 /* Basic block where the assertion would be inserted. */
2566 basic_block bb;
2568 /* Some assertions need to be inserted on an edge (e.g., assertions
2569 generated by COND_EXPRs). In those cases, BB will be NULL. */
2570 edge e;
2572 /* Pointer to the statement that generated this assertion. */
2573 gimple_stmt_iterator si;
2575 /* Predicate code for the ASSERT_EXPR. Must be COMPARISON_CLASS_P. */
2576 enum tree_code comp_code;
2578 /* Value being compared against. */
2579 tree val;
2581 /* Expression to compare. */
2582 tree expr;
2584 /* Next node in the linked list. */
2585 assert_locus *next;
2588 /* Class to traverse the flowgraph looking for conditional jumps to
2589 insert ASSERT_EXPR range expressions. These range expressions are
2590 meant to provide information to optimizations that need to reason
2591 in terms of value ranges. They will not be expanded into RTL. */
2593 class vrp_asserts
2595 public:
2596 vrp_asserts (struct function *fn) : fun (fn) { }
2598 void insert_range_assertions ();
2600 /* Convert range assertion expressions into the implied copies and
2601 copy propagate away the copies. */
2602 void remove_range_assertions ();
2604 /* Dump all the registered assertions for all the names to FILE. */
2605 void dump (FILE *);
2607 /* Dump all the registered assertions for NAME to FILE. */
2608 void dump (FILE *file, tree name);
2610 /* Dump all the registered assertions for NAME to stderr. */
2611 void debug (tree name)
2613 dump (stderr, name);
2616 /* Dump all the registered assertions for all the names to stderr. */
2617 void debug ()
2619 dump (stderr);
2622 private:
2623 /* Set of SSA names found live during the RPO traversal of the function
2624 for still active basic-blocks. */
2625 live_names live;
2627 /* Function to work on. */
2628 struct function *fun;
2630 /* If bit I is present, it means that SSA name N_i has a list of
2631 assertions that should be inserted in the IL. */
2632 bitmap need_assert_for;
2634 /* Array of locations lists where to insert assertions. ASSERTS_FOR[I]
2635 holds a list of ASSERT_LOCUS_T nodes that describe where
2636 ASSERT_EXPRs for SSA name N_I should be inserted. */
2637 assert_locus **asserts_for;
2639 /* Finish found ASSERTS for E and register them at GSI. */
2640 void finish_register_edge_assert_for (edge e, gimple_stmt_iterator gsi,
2641 vec<assert_info> &asserts);
2643 /* Determine whether the outgoing edges of BB should receive an
2644 ASSERT_EXPR for each of the operands of BB's LAST statement. The
2645 last statement of BB must be a SWITCH_EXPR.
2647 If any of the sub-graphs rooted at BB have an interesting use of
2648 the predicate operands, an assert location node is added to the
2649 list of assertions for the corresponding operands. */
2650 void find_switch_asserts (basic_block bb, gswitch *last);
2652 /* Do an RPO walk over the function computing SSA name liveness
2653 on-the-fly and deciding on assert expressions to insert. */
2654 void find_assert_locations ();
2656 /* Traverse all the statements in block BB looking for statements that
2657 may generate useful assertions for the SSA names in their operand.
2658 See method implementation comentary for more information. */
2659 void find_assert_locations_in_bb (basic_block bb);
2661 /* Determine whether the outgoing edges of BB should receive an
2662 ASSERT_EXPR for each of the operands of BB's LAST statement.
2663 The last statement of BB must be a COND_EXPR.
2665 If any of the sub-graphs rooted at BB have an interesting use of
2666 the predicate operands, an assert location node is added to the
2667 list of assertions for the corresponding operands. */
2668 void find_conditional_asserts (basic_block bb, gcond *last);
2670 /* Process all the insertions registered for every name N_i registered
2671 in NEED_ASSERT_FOR. The list of assertions to be inserted are
2672 found in ASSERTS_FOR[i]. */
2673 void process_assert_insertions ();
2675 /* If NAME doesn't have an ASSERT_EXPR registered for asserting
2676 'EXPR COMP_CODE VAL' at a location that dominates block BB or
2677 E->DEST, then register this location as a possible insertion point
2678 for ASSERT_EXPR <NAME, EXPR COMP_CODE VAL>.
2680 BB, E and SI provide the exact insertion point for the new
2681 ASSERT_EXPR. If BB is NULL, then the ASSERT_EXPR is to be inserted
2682 on edge E. Otherwise, if E is NULL, the ASSERT_EXPR is inserted on
2683 BB. If SI points to a COND_EXPR or a SWITCH_EXPR statement, then E
2684 must not be NULL. */
2685 void register_new_assert_for (tree name, tree expr,
2686 enum tree_code comp_code,
2687 tree val, basic_block bb,
2688 edge e, gimple_stmt_iterator si);
2690 /* Given a COND_EXPR COND of the form 'V OP W', and an SSA name V,
2691 create a new SSA name N and return the assertion assignment
2692 'N = ASSERT_EXPR <V, V OP W>'. */
2693 gimple *build_assert_expr_for (tree cond, tree v);
2695 /* Create an ASSERT_EXPR for NAME and insert it in the location
2696 indicated by LOC. Return true if we made any edge insertions. */
2697 bool process_assert_insertions_for (tree name, assert_locus *loc);
2699 /* Qsort callback for sorting assert locations. */
2700 template <bool stable> static int compare_assert_loc (const void *,
2701 const void *);
2703 /* Return false if EXPR is a predicate expression involving floating
2704 point values. */
2705 bool fp_predicate (gimple *stmt)
2707 GIMPLE_CHECK (stmt, GIMPLE_COND);
2708 return FLOAT_TYPE_P (TREE_TYPE (gimple_cond_lhs (stmt)));
2711 bool all_imm_uses_in_stmt_or_feed_cond (tree var, gimple *stmt,
2712 basic_block cond_bb);
2714 static int compare_case_labels (const void *, const void *);
2717 /* Given a COND_EXPR COND of the form 'V OP W', and an SSA name V,
2718 create a new SSA name N and return the assertion assignment
2719 'N = ASSERT_EXPR <V, V OP W>'. */
2721 gimple *
2722 vrp_asserts::build_assert_expr_for (tree cond, tree v)
2724 tree a;
2725 gassign *assertion;
2727 gcc_assert (TREE_CODE (v) == SSA_NAME
2728 && COMPARISON_CLASS_P (cond));
2730 a = build2 (ASSERT_EXPR, TREE_TYPE (v), v, cond);
2731 assertion = gimple_build_assign (NULL_TREE, a);
2733 /* The new ASSERT_EXPR, creates a new SSA name that replaces the
2734 operand of the ASSERT_EXPR. Create it so the new name and the old one
2735 are registered in the replacement table so that we can fix the SSA web
2736 after adding all the ASSERT_EXPRs. */
2737 tree new_def = create_new_def_for (v, assertion, NULL);
2738 /* Make sure we preserve abnormalness throughout an ASSERT_EXPR chain
2739 given we have to be able to fully propagate those out to re-create
2740 valid SSA when removing the asserts. */
2741 if (SSA_NAME_OCCURS_IN_ABNORMAL_PHI (v))
2742 SSA_NAME_OCCURS_IN_ABNORMAL_PHI (new_def) = 1;
2744 return assertion;
2747 /* Dump all the registered assertions for NAME to FILE. */
2749 void
2750 vrp_asserts::dump (FILE *file, tree name)
2752 assert_locus *loc;
2754 fprintf (file, "Assertions to be inserted for ");
2755 print_generic_expr (file, name);
2756 fprintf (file, "\n");
2758 loc = asserts_for[SSA_NAME_VERSION (name)];
2759 while (loc)
2761 fprintf (file, "\t");
2762 print_gimple_stmt (file, gsi_stmt (loc->si), 0);
2763 fprintf (file, "\n\tBB #%d", loc->bb->index);
2764 if (loc->e)
2766 fprintf (file, "\n\tEDGE %d->%d", loc->e->src->index,
2767 loc->e->dest->index);
2768 dump_edge_info (file, loc->e, dump_flags, 0);
2770 fprintf (file, "\n\tPREDICATE: ");
2771 print_generic_expr (file, loc->expr);
2772 fprintf (file, " %s ", get_tree_code_name (loc->comp_code));
2773 print_generic_expr (file, loc->val);
2774 fprintf (file, "\n\n");
2775 loc = loc->next;
2778 fprintf (file, "\n");
2781 /* Dump all the registered assertions for all the names to FILE. */
2783 void
2784 vrp_asserts::dump (FILE *file)
2786 unsigned i;
2787 bitmap_iterator bi;
2789 fprintf (file, "\nASSERT_EXPRs to be inserted\n\n");
2790 EXECUTE_IF_SET_IN_BITMAP (need_assert_for, 0, i, bi)
2791 dump (file, ssa_name (i));
2792 fprintf (file, "\n");
2795 /* If NAME doesn't have an ASSERT_EXPR registered for asserting
2796 'EXPR COMP_CODE VAL' at a location that dominates block BB or
2797 E->DEST, then register this location as a possible insertion point
2798 for ASSERT_EXPR <NAME, EXPR COMP_CODE VAL>.
2800 BB, E and SI provide the exact insertion point for the new
2801 ASSERT_EXPR. If BB is NULL, then the ASSERT_EXPR is to be inserted
2802 on edge E. Otherwise, if E is NULL, the ASSERT_EXPR is inserted on
2803 BB. If SI points to a COND_EXPR or a SWITCH_EXPR statement, then E
2804 must not be NULL. */
2806 void
2807 vrp_asserts::register_new_assert_for (tree name, tree expr,
2808 enum tree_code comp_code,
2809 tree val,
2810 basic_block bb,
2811 edge e,
2812 gimple_stmt_iterator si)
2814 assert_locus *n, *loc, *last_loc;
2815 basic_block dest_bb;
2817 gcc_checking_assert (bb == NULL || e == NULL);
2819 if (e == NULL)
2820 gcc_checking_assert (gimple_code (gsi_stmt (si)) != GIMPLE_COND
2821 && gimple_code (gsi_stmt (si)) != GIMPLE_SWITCH);
2823 /* Never build an assert comparing against an integer constant with
2824 TREE_OVERFLOW set. This confuses our undefined overflow warning
2825 machinery. */
2826 if (TREE_OVERFLOW_P (val))
2827 val = drop_tree_overflow (val);
2829 /* The new assertion A will be inserted at BB or E. We need to
2830 determine if the new location is dominated by a previously
2831 registered location for A. If we are doing an edge insertion,
2832 assume that A will be inserted at E->DEST. Note that this is not
2833 necessarily true.
2835 If E is a critical edge, it will be split. But even if E is
2836 split, the new block will dominate the same set of blocks that
2837 E->DEST dominates.
2839 The reverse, however, is not true, blocks dominated by E->DEST
2840 will not be dominated by the new block created to split E. So,
2841 if the insertion location is on a critical edge, we will not use
2842 the new location to move another assertion previously registered
2843 at a block dominated by E->DEST. */
2844 dest_bb = (bb) ? bb : e->dest;
2846 /* If NAME already has an ASSERT_EXPR registered for COMP_CODE and
2847 VAL at a block dominating DEST_BB, then we don't need to insert a new
2848 one. Similarly, if the same assertion already exists at a block
2849 dominated by DEST_BB and the new location is not on a critical
2850 edge, then update the existing location for the assertion (i.e.,
2851 move the assertion up in the dominance tree).
2853 Note, this is implemented as a simple linked list because there
2854 should not be more than a handful of assertions registered per
2855 name. If this becomes a performance problem, a table hashed by
2856 COMP_CODE and VAL could be implemented. */
2857 loc = asserts_for[SSA_NAME_VERSION (name)];
2858 last_loc = loc;
2859 while (loc)
2861 if (loc->comp_code == comp_code
2862 && (loc->val == val
2863 || operand_equal_p (loc->val, val, 0))
2864 && (loc->expr == expr
2865 || operand_equal_p (loc->expr, expr, 0)))
2867 /* If E is not a critical edge and DEST_BB
2868 dominates the existing location for the assertion, move
2869 the assertion up in the dominance tree by updating its
2870 location information. */
2871 if ((e == NULL || !EDGE_CRITICAL_P (e))
2872 && dominated_by_p (CDI_DOMINATORS, loc->bb, dest_bb))
2874 loc->bb = dest_bb;
2875 loc->e = e;
2876 loc->si = si;
2877 return;
2881 /* Update the last node of the list and move to the next one. */
2882 last_loc = loc;
2883 loc = loc->next;
2886 /* If we didn't find an assertion already registered for
2887 NAME COMP_CODE VAL, add a new one at the end of the list of
2888 assertions associated with NAME. */
2889 n = XNEW (struct assert_locus);
2890 n->bb = dest_bb;
2891 n->e = e;
2892 n->si = si;
2893 n->comp_code = comp_code;
2894 n->val = val;
2895 n->expr = expr;
2896 n->next = NULL;
2898 if (last_loc)
2899 last_loc->next = n;
2900 else
2901 asserts_for[SSA_NAME_VERSION (name)] = n;
2903 bitmap_set_bit (need_assert_for, SSA_NAME_VERSION (name));
2906 /* Finish found ASSERTS for E and register them at GSI. */
2908 void
2909 vrp_asserts::finish_register_edge_assert_for (edge e,
2910 gimple_stmt_iterator gsi,
2911 vec<assert_info> &asserts)
2913 for (unsigned i = 0; i < asserts.length (); ++i)
2914 /* Only register an ASSERT_EXPR if NAME was found in the sub-graph
2915 reachable from E. */
2916 if (live.live_on_edge_p (asserts[i].name, e))
2917 register_new_assert_for (asserts[i].name, asserts[i].expr,
2918 asserts[i].comp_code, asserts[i].val,
2919 NULL, e, gsi);
2922 /* Determine whether the outgoing edges of BB should receive an
2923 ASSERT_EXPR for each of the operands of BB's LAST statement.
2924 The last statement of BB must be a COND_EXPR.
2926 If any of the sub-graphs rooted at BB have an interesting use of
2927 the predicate operands, an assert location node is added to the
2928 list of assertions for the corresponding operands. */
2930 void
2931 vrp_asserts::find_conditional_asserts (basic_block bb, gcond *last)
2933 gimple_stmt_iterator bsi;
2934 tree op;
2935 edge_iterator ei;
2936 edge e;
2937 ssa_op_iter iter;
2939 bsi = gsi_for_stmt (last);
2941 /* Look for uses of the operands in each of the sub-graphs
2942 rooted at BB. We need to check each of the outgoing edges
2943 separately, so that we know what kind of ASSERT_EXPR to
2944 insert. */
2945 FOR_EACH_EDGE (e, ei, bb->succs)
2947 if (e->dest == bb)
2948 continue;
2950 /* Register the necessary assertions for each operand in the
2951 conditional predicate. */
2952 auto_vec<assert_info, 8> asserts;
2953 FOR_EACH_SSA_TREE_OPERAND (op, last, iter, SSA_OP_USE)
2954 register_edge_assert_for (op, e,
2955 gimple_cond_code (last),
2956 gimple_cond_lhs (last),
2957 gimple_cond_rhs (last), asserts);
2958 finish_register_edge_assert_for (e, bsi, asserts);
2962 /* Compare two case labels sorting first by the destination bb index
2963 and then by the case value. */
2966 vrp_asserts::compare_case_labels (const void *p1, const void *p2)
2968 const struct case_info *ci1 = (const struct case_info *) p1;
2969 const struct case_info *ci2 = (const struct case_info *) p2;
2970 int idx1 = ci1->bb->index;
2971 int idx2 = ci2->bb->index;
2973 if (idx1 < idx2)
2974 return -1;
2975 else if (idx1 == idx2)
2977 /* Make sure the default label is first in a group. */
2978 if (!CASE_LOW (ci1->expr))
2979 return -1;
2980 else if (!CASE_LOW (ci2->expr))
2981 return 1;
2982 else
2983 return tree_int_cst_compare (CASE_LOW (ci1->expr),
2984 CASE_LOW (ci2->expr));
2986 else
2987 return 1;
2990 /* Determine whether the outgoing edges of BB should receive an
2991 ASSERT_EXPR for each of the operands of BB's LAST statement.
2992 The last statement of BB must be a SWITCH_EXPR.
2994 If any of the sub-graphs rooted at BB have an interesting use of
2995 the predicate operands, an assert location node is added to the
2996 list of assertions for the corresponding operands. */
2998 void
2999 vrp_asserts::find_switch_asserts (basic_block bb, gswitch *last)
3001 gimple_stmt_iterator bsi;
3002 tree op;
3003 edge e;
3004 struct case_info *ci;
3005 size_t n = gimple_switch_num_labels (last);
3006 #if GCC_VERSION >= 4000
3007 unsigned int idx;
3008 #else
3009 /* Work around GCC 3.4 bug (PR 37086). */
3010 volatile unsigned int idx;
3011 #endif
3013 bsi = gsi_for_stmt (last);
3014 op = gimple_switch_index (last);
3015 if (TREE_CODE (op) != SSA_NAME)
3016 return;
3018 /* Build a vector of case labels sorted by destination label. */
3019 ci = XNEWVEC (struct case_info, n);
3020 for (idx = 0; idx < n; ++idx)
3022 ci[idx].expr = gimple_switch_label (last, idx);
3023 ci[idx].bb = label_to_block (fun, CASE_LABEL (ci[idx].expr));
3025 edge default_edge = find_edge (bb, ci[0].bb);
3026 qsort (ci, n, sizeof (struct case_info), compare_case_labels);
3028 for (idx = 0; idx < n; ++idx)
3030 tree min, max;
3031 tree cl = ci[idx].expr;
3032 basic_block cbb = ci[idx].bb;
3034 min = CASE_LOW (cl);
3035 max = CASE_HIGH (cl);
3037 /* If there are multiple case labels with the same destination
3038 we need to combine them to a single value range for the edge. */
3039 if (idx + 1 < n && cbb == ci[idx + 1].bb)
3041 /* Skip labels until the last of the group. */
3042 do {
3043 ++idx;
3044 } while (idx < n && cbb == ci[idx].bb);
3045 --idx;
3047 /* Pick up the maximum of the case label range. */
3048 if (CASE_HIGH (ci[idx].expr))
3049 max = CASE_HIGH (ci[idx].expr);
3050 else
3051 max = CASE_LOW (ci[idx].expr);
3054 /* Can't extract a useful assertion out of a range that includes the
3055 default label. */
3056 if (min == NULL_TREE)
3057 continue;
3059 /* Find the edge to register the assert expr on. */
3060 e = find_edge (bb, cbb);
3062 /* Register the necessary assertions for the operand in the
3063 SWITCH_EXPR. */
3064 auto_vec<assert_info, 8> asserts;
3065 register_edge_assert_for (op, e,
3066 max ? GE_EXPR : EQ_EXPR,
3067 op, fold_convert (TREE_TYPE (op), min),
3068 asserts);
3069 if (max)
3070 register_edge_assert_for (op, e, LE_EXPR, op,
3071 fold_convert (TREE_TYPE (op), max),
3072 asserts);
3073 finish_register_edge_assert_for (e, bsi, asserts);
3076 XDELETEVEC (ci);
3078 if (!live.live_on_edge_p (op, default_edge))
3079 return;
3081 /* Now register along the default label assertions that correspond to the
3082 anti-range of each label. */
3083 int insertion_limit = param_max_vrp_switch_assertions;
3084 if (insertion_limit == 0)
3085 return;
3087 /* We can't do this if the default case shares a label with another case. */
3088 tree default_cl = gimple_switch_default_label (last);
3089 for (idx = 1; idx < n; idx++)
3091 tree min, max;
3092 tree cl = gimple_switch_label (last, idx);
3093 if (CASE_LABEL (cl) == CASE_LABEL (default_cl))
3094 continue;
3096 min = CASE_LOW (cl);
3097 max = CASE_HIGH (cl);
3099 /* Combine contiguous case ranges to reduce the number of assertions
3100 to insert. */
3101 for (idx = idx + 1; idx < n; idx++)
3103 tree next_min, next_max;
3104 tree next_cl = gimple_switch_label (last, idx);
3105 if (CASE_LABEL (next_cl) == CASE_LABEL (default_cl))
3106 break;
3108 next_min = CASE_LOW (next_cl);
3109 next_max = CASE_HIGH (next_cl);
3111 wide_int difference = (wi::to_wide (next_min)
3112 - wi::to_wide (max ? max : min));
3113 if (wi::eq_p (difference, 1))
3114 max = next_max ? next_max : next_min;
3115 else
3116 break;
3118 idx--;
3120 if (max == NULL_TREE)
3122 /* Register the assertion OP != MIN. */
3123 auto_vec<assert_info, 8> asserts;
3124 min = fold_convert (TREE_TYPE (op), min);
3125 register_edge_assert_for (op, default_edge, NE_EXPR, op, min,
3126 asserts);
3127 finish_register_edge_assert_for (default_edge, bsi, asserts);
3129 else
3131 /* Register the assertion (unsigned)OP - MIN > (MAX - MIN),
3132 which will give OP the anti-range ~[MIN,MAX]. */
3133 tree uop = fold_convert (unsigned_type_for (TREE_TYPE (op)), op);
3134 min = fold_convert (TREE_TYPE (uop), min);
3135 max = fold_convert (TREE_TYPE (uop), max);
3137 tree lhs = fold_build2 (MINUS_EXPR, TREE_TYPE (uop), uop, min);
3138 tree rhs = int_const_binop (MINUS_EXPR, max, min);
3139 register_new_assert_for (op, lhs, GT_EXPR, rhs,
3140 NULL, default_edge, bsi);
3143 if (--insertion_limit == 0)
3144 break;
3148 /* Traverse all the statements in block BB looking for statements that
3149 may generate useful assertions for the SSA names in their operand.
3150 If a statement produces a useful assertion A for name N_i, then the
3151 list of assertions already generated for N_i is scanned to
3152 determine if A is actually needed.
3154 If N_i already had the assertion A at a location dominating the
3155 current location, then nothing needs to be done. Otherwise, the
3156 new location for A is recorded instead.
3158 1- For every statement S in BB, all the variables used by S are
3159 added to bitmap FOUND_IN_SUBGRAPH.
3161 2- If statement S uses an operand N in a way that exposes a known
3162 value range for N, then if N was not already generated by an
3163 ASSERT_EXPR, create a new assert location for N. For instance,
3164 if N is a pointer and the statement dereferences it, we can
3165 assume that N is not NULL.
3167 3- COND_EXPRs are a special case of #2. We can derive range
3168 information from the predicate but need to insert different
3169 ASSERT_EXPRs for each of the sub-graphs rooted at the
3170 conditional block. If the last statement of BB is a conditional
3171 expression of the form 'X op Y', then
3173 a) Remove X and Y from the set FOUND_IN_SUBGRAPH.
3175 b) If the conditional is the only entry point to the sub-graph
3176 corresponding to the THEN_CLAUSE, recurse into it. On
3177 return, if X and/or Y are marked in FOUND_IN_SUBGRAPH, then
3178 an ASSERT_EXPR is added for the corresponding variable.
3180 c) Repeat step (b) on the ELSE_CLAUSE.
3182 d) Mark X and Y in FOUND_IN_SUBGRAPH.
3184 For instance,
3186 if (a == 9)
3187 b = a;
3188 else
3189 b = c + 1;
3191 In this case, an assertion on the THEN clause is useful to
3192 determine that 'a' is always 9 on that edge. However, an assertion
3193 on the ELSE clause would be unnecessary.
3195 4- If BB does not end in a conditional expression, then we recurse
3196 into BB's dominator children.
3198 At the end of the recursive traversal, every SSA name will have a
3199 list of locations where ASSERT_EXPRs should be added. When a new
3200 location for name N is found, it is registered by calling
3201 register_new_assert_for. That function keeps track of all the
3202 registered assertions to prevent adding unnecessary assertions.
3203 For instance, if a pointer P_4 is dereferenced more than once in a
3204 dominator tree, only the location dominating all the dereference of
3205 P_4 will receive an ASSERT_EXPR. */
3207 void
3208 vrp_asserts::find_assert_locations_in_bb (basic_block bb)
3210 gimple *last;
3212 last = last_stmt (bb);
3214 /* If BB's last statement is a conditional statement involving integer
3215 operands, determine if we need to add ASSERT_EXPRs. */
3216 if (last
3217 && gimple_code (last) == GIMPLE_COND
3218 && !fp_predicate (last)
3219 && !ZERO_SSA_OPERANDS (last, SSA_OP_USE))
3220 find_conditional_asserts (bb, as_a <gcond *> (last));
3222 /* If BB's last statement is a switch statement involving integer
3223 operands, determine if we need to add ASSERT_EXPRs. */
3224 if (last
3225 && gimple_code (last) == GIMPLE_SWITCH
3226 && !ZERO_SSA_OPERANDS (last, SSA_OP_USE))
3227 find_switch_asserts (bb, as_a <gswitch *> (last));
3229 /* Traverse all the statements in BB marking used names and looking
3230 for statements that may infer assertions for their used operands. */
3231 for (gimple_stmt_iterator si = gsi_last_bb (bb); !gsi_end_p (si);
3232 gsi_prev (&si))
3234 gimple *stmt;
3235 tree op;
3236 ssa_op_iter i;
3238 stmt = gsi_stmt (si);
3240 if (is_gimple_debug (stmt))
3241 continue;
3243 /* See if we can derive an assertion for any of STMT's operands. */
3244 FOR_EACH_SSA_TREE_OPERAND (op, stmt, i, SSA_OP_USE)
3246 tree value;
3247 enum tree_code comp_code;
3249 /* If op is not live beyond this stmt, do not bother to insert
3250 asserts for it. */
3251 if (!live.live_on_block_p (op, bb))
3252 continue;
3254 /* If OP is used in such a way that we can infer a value
3255 range for it, and we don't find a previous assertion for
3256 it, create a new assertion location node for OP. */
3257 if (infer_value_range (stmt, op, &comp_code, &value))
3259 /* If we are able to infer a nonzero value range for OP,
3260 then walk backwards through the use-def chain to see if OP
3261 was set via a typecast.
3263 If so, then we can also infer a nonzero value range
3264 for the operand of the NOP_EXPR. */
3265 if (comp_code == NE_EXPR && integer_zerop (value))
3267 tree t = op;
3268 gimple *def_stmt = SSA_NAME_DEF_STMT (t);
3270 while (is_gimple_assign (def_stmt)
3271 && CONVERT_EXPR_CODE_P
3272 (gimple_assign_rhs_code (def_stmt))
3273 && TREE_CODE
3274 (gimple_assign_rhs1 (def_stmt)) == SSA_NAME
3275 && POINTER_TYPE_P
3276 (TREE_TYPE (gimple_assign_rhs1 (def_stmt))))
3278 t = gimple_assign_rhs1 (def_stmt);
3279 def_stmt = SSA_NAME_DEF_STMT (t);
3281 /* Note we want to register the assert for the
3282 operand of the NOP_EXPR after SI, not after the
3283 conversion. */
3284 if (live.live_on_block_p (t, bb))
3285 register_new_assert_for (t, t, comp_code, value,
3286 bb, NULL, si);
3290 register_new_assert_for (op, op, comp_code, value, bb, NULL, si);
3294 /* Update live. */
3295 FOR_EACH_SSA_TREE_OPERAND (op, stmt, i, SSA_OP_USE)
3296 live.set (op, bb);
3297 FOR_EACH_SSA_TREE_OPERAND (op, stmt, i, SSA_OP_DEF)
3298 live.clear (op, bb);
3301 /* Traverse all PHI nodes in BB, updating live. */
3302 for (gphi_iterator si = gsi_start_phis (bb); !gsi_end_p (si);
3303 gsi_next (&si))
3305 use_operand_p arg_p;
3306 ssa_op_iter i;
3307 gphi *phi = si.phi ();
3308 tree res = gimple_phi_result (phi);
3310 if (virtual_operand_p (res))
3311 continue;
3313 FOR_EACH_PHI_ARG (arg_p, phi, i, SSA_OP_USE)
3315 tree arg = USE_FROM_PTR (arg_p);
3316 if (TREE_CODE (arg) == SSA_NAME)
3317 live.set (arg, bb);
3320 live.clear (res, bb);
3324 /* Do an RPO walk over the function computing SSA name liveness
3325 on-the-fly and deciding on assert expressions to insert. */
3327 void
3328 vrp_asserts::find_assert_locations (void)
3330 int *rpo = XNEWVEC (int, last_basic_block_for_fn (fun));
3331 int *bb_rpo = XNEWVEC (int, last_basic_block_for_fn (fun));
3332 int *last_rpo = XCNEWVEC (int, last_basic_block_for_fn (fun));
3333 int rpo_cnt, i;
3335 rpo_cnt = pre_and_rev_post_order_compute (NULL, rpo, false);
3336 for (i = 0; i < rpo_cnt; ++i)
3337 bb_rpo[rpo[i]] = i;
3339 /* Pre-seed loop latch liveness from loop header PHI nodes. Due to
3340 the order we compute liveness and insert asserts we otherwise
3341 fail to insert asserts into the loop latch. */
3342 loop_p loop;
3343 FOR_EACH_LOOP (loop, 0)
3345 i = loop->latch->index;
3346 unsigned int j = single_succ_edge (loop->latch)->dest_idx;
3347 for (gphi_iterator gsi = gsi_start_phis (loop->header);
3348 !gsi_end_p (gsi); gsi_next (&gsi))
3350 gphi *phi = gsi.phi ();
3351 if (virtual_operand_p (gimple_phi_result (phi)))
3352 continue;
3353 tree arg = gimple_phi_arg_def (phi, j);
3354 if (TREE_CODE (arg) == SSA_NAME)
3355 live.set (arg, loop->latch);
3359 for (i = rpo_cnt - 1; i >= 0; --i)
3361 basic_block bb = BASIC_BLOCK_FOR_FN (fun, rpo[i]);
3362 edge e;
3363 edge_iterator ei;
3365 /* Process BB and update the live information with uses in
3366 this block. */
3367 find_assert_locations_in_bb (bb);
3369 /* Merge liveness into the predecessor blocks and free it. */
3370 if (!live.block_has_live_names_p (bb))
3372 int pred_rpo = i;
3373 FOR_EACH_EDGE (e, ei, bb->preds)
3375 int pred = e->src->index;
3376 if ((e->flags & EDGE_DFS_BACK) || pred == ENTRY_BLOCK)
3377 continue;
3379 live.merge (e->src, bb);
3381 if (bb_rpo[pred] < pred_rpo)
3382 pred_rpo = bb_rpo[pred];
3385 /* Record the RPO number of the last visited block that needs
3386 live information from this block. */
3387 last_rpo[rpo[i]] = pred_rpo;
3389 else
3390 live.clear_block (bb);
3392 /* We can free all successors live bitmaps if all their
3393 predecessors have been visited already. */
3394 FOR_EACH_EDGE (e, ei, bb->succs)
3395 if (last_rpo[e->dest->index] == i)
3396 live.clear_block (e->dest);
3399 XDELETEVEC (rpo);
3400 XDELETEVEC (bb_rpo);
3401 XDELETEVEC (last_rpo);
3404 /* Create an ASSERT_EXPR for NAME and insert it in the location
3405 indicated by LOC. Return true if we made any edge insertions. */
3407 bool
3408 vrp_asserts::process_assert_insertions_for (tree name, assert_locus *loc)
3410 /* Build the comparison expression NAME_i COMP_CODE VAL. */
3411 gimple *stmt;
3412 tree cond;
3413 gimple *assert_stmt;
3414 edge_iterator ei;
3415 edge e;
3417 /* If we have X <=> X do not insert an assert expr for that. */
3418 if (loc->expr == loc->val)
3419 return false;
3421 cond = build2 (loc->comp_code, boolean_type_node, loc->expr, loc->val);
3422 assert_stmt = build_assert_expr_for (cond, name);
3423 if (loc->e)
3425 /* We have been asked to insert the assertion on an edge. This
3426 is used only by COND_EXPR and SWITCH_EXPR assertions. */
3427 gcc_checking_assert (gimple_code (gsi_stmt (loc->si)) == GIMPLE_COND
3428 || (gimple_code (gsi_stmt (loc->si))
3429 == GIMPLE_SWITCH));
3431 gsi_insert_on_edge (loc->e, assert_stmt);
3432 return true;
3435 /* If the stmt iterator points at the end then this is an insertion
3436 at the beginning of a block. */
3437 if (gsi_end_p (loc->si))
3439 gimple_stmt_iterator si = gsi_after_labels (loc->bb);
3440 gsi_insert_before (&si, assert_stmt, GSI_SAME_STMT);
3441 return false;
3444 /* Otherwise, we can insert right after LOC->SI iff the
3445 statement must not be the last statement in the block. */
3446 stmt = gsi_stmt (loc->si);
3447 if (!stmt_ends_bb_p (stmt))
3449 gsi_insert_after (&loc->si, assert_stmt, GSI_SAME_STMT);
3450 return false;
3453 /* If STMT must be the last statement in BB, we can only insert new
3454 assertions on the non-abnormal edge out of BB. Note that since
3455 STMT is not control flow, there may only be one non-abnormal/eh edge
3456 out of BB. */
3457 FOR_EACH_EDGE (e, ei, loc->bb->succs)
3458 if (!(e->flags & (EDGE_ABNORMAL|EDGE_EH)))
3460 gsi_insert_on_edge (e, assert_stmt);
3461 return true;
3464 gcc_unreachable ();
3467 /* Qsort helper for sorting assert locations. If stable is true, don't
3468 use iterative_hash_expr because it can be unstable for -fcompare-debug,
3469 on the other side some pointers might be NULL. */
3471 template <bool stable>
3473 vrp_asserts::compare_assert_loc (const void *pa, const void *pb)
3475 assert_locus * const a = *(assert_locus * const *)pa;
3476 assert_locus * const b = *(assert_locus * const *)pb;
3478 /* If stable, some asserts might be optimized away already, sort
3479 them last. */
3480 if (stable)
3482 if (a == NULL)
3483 return b != NULL;
3484 else if (b == NULL)
3485 return -1;
3488 if (a->e == NULL && b->e != NULL)
3489 return 1;
3490 else if (a->e != NULL && b->e == NULL)
3491 return -1;
3493 /* After the above checks, we know that (a->e == NULL) == (b->e == NULL),
3494 no need to test both a->e and b->e. */
3496 /* Sort after destination index. */
3497 if (a->e == NULL)
3499 else if (a->e->dest->index > b->e->dest->index)
3500 return 1;
3501 else if (a->e->dest->index < b->e->dest->index)
3502 return -1;
3504 /* Sort after comp_code. */
3505 if (a->comp_code > b->comp_code)
3506 return 1;
3507 else if (a->comp_code < b->comp_code)
3508 return -1;
3510 hashval_t ha, hb;
3512 /* E.g. if a->val is ADDR_EXPR of a VAR_DECL, iterative_hash_expr
3513 uses DECL_UID of the VAR_DECL, so sorting might differ between
3514 -g and -g0. When doing the removal of redundant assert exprs
3515 and commonization to successors, this does not matter, but for
3516 the final sort needs to be stable. */
3517 if (stable)
3519 ha = 0;
3520 hb = 0;
3522 else
3524 ha = iterative_hash_expr (a->expr, iterative_hash_expr (a->val, 0));
3525 hb = iterative_hash_expr (b->expr, iterative_hash_expr (b->val, 0));
3528 /* Break the tie using hashing and source/bb index. */
3529 if (ha == hb)
3530 return (a->e != NULL
3531 ? a->e->src->index - b->e->src->index
3532 : a->bb->index - b->bb->index);
3533 return ha > hb ? 1 : -1;
3536 /* Process all the insertions registered for every name N_i registered
3537 in NEED_ASSERT_FOR. The list of assertions to be inserted are
3538 found in ASSERTS_FOR[i]. */
3540 void
3541 vrp_asserts::process_assert_insertions ()
3543 unsigned i;
3544 bitmap_iterator bi;
3545 bool update_edges_p = false;
3546 int num_asserts = 0;
3548 if (dump_file && (dump_flags & TDF_DETAILS))
3549 dump (dump_file);
3551 EXECUTE_IF_SET_IN_BITMAP (need_assert_for, 0, i, bi)
3553 assert_locus *loc = asserts_for[i];
3554 gcc_assert (loc);
3556 auto_vec<assert_locus *, 16> asserts;
3557 for (; loc; loc = loc->next)
3558 asserts.safe_push (loc);
3559 asserts.qsort (compare_assert_loc<false>);
3561 /* Push down common asserts to successors and remove redundant ones. */
3562 unsigned ecnt = 0;
3563 assert_locus *common = NULL;
3564 unsigned commonj = 0;
3565 for (unsigned j = 0; j < asserts.length (); ++j)
3567 loc = asserts[j];
3568 if (! loc->e)
3569 common = NULL;
3570 else if (! common
3571 || loc->e->dest != common->e->dest
3572 || loc->comp_code != common->comp_code
3573 || ! operand_equal_p (loc->val, common->val, 0)
3574 || ! operand_equal_p (loc->expr, common->expr, 0))
3576 commonj = j;
3577 common = loc;
3578 ecnt = 1;
3580 else if (loc->e == asserts[j-1]->e)
3582 /* Remove duplicate asserts. */
3583 if (commonj == j - 1)
3585 commonj = j;
3586 common = loc;
3588 free (asserts[j-1]);
3589 asserts[j-1] = NULL;
3591 else
3593 ecnt++;
3594 if (EDGE_COUNT (common->e->dest->preds) == ecnt)
3596 /* We have the same assertion on all incoming edges of a BB.
3597 Insert it at the beginning of that block. */
3598 loc->bb = loc->e->dest;
3599 loc->e = NULL;
3600 loc->si = gsi_none ();
3601 common = NULL;
3602 /* Clear asserts commoned. */
3603 for (; commonj != j; ++commonj)
3604 if (asserts[commonj])
3606 free (asserts[commonj]);
3607 asserts[commonj] = NULL;
3613 /* The asserts vector sorting above might be unstable for
3614 -fcompare-debug, sort again to ensure a stable sort. */
3615 asserts.qsort (compare_assert_loc<true>);
3616 for (unsigned j = 0; j < asserts.length (); ++j)
3618 loc = asserts[j];
3619 if (! loc)
3620 break;
3621 update_edges_p |= process_assert_insertions_for (ssa_name (i), loc);
3622 num_asserts++;
3623 free (loc);
3627 if (update_edges_p)
3628 gsi_commit_edge_inserts ();
3630 statistics_counter_event (fun, "Number of ASSERT_EXPR expressions inserted",
3631 num_asserts);
3634 /* Traverse the flowgraph looking for conditional jumps to insert range
3635 expressions. These range expressions are meant to provide information
3636 to optimizations that need to reason in terms of value ranges. They
3637 will not be expanded into RTL. For instance, given:
3639 x = ...
3640 y = ...
3641 if (x < y)
3642 y = x - 2;
3643 else
3644 x = y + 3;
3646 this pass will transform the code into:
3648 x = ...
3649 y = ...
3650 if (x < y)
3652 x = ASSERT_EXPR <x, x < y>
3653 y = x - 2
3655 else
3657 y = ASSERT_EXPR <y, x >= y>
3658 x = y + 3
3661 The idea is that once copy and constant propagation have run, other
3662 optimizations will be able to determine what ranges of values can 'x'
3663 take in different paths of the code, simply by checking the reaching
3664 definition of 'x'. */
3666 void
3667 vrp_asserts::insert_range_assertions (void)
3669 need_assert_for = BITMAP_ALLOC (NULL);
3670 asserts_for = XCNEWVEC (assert_locus *, num_ssa_names);
3672 calculate_dominance_info (CDI_DOMINATORS);
3674 find_assert_locations ();
3675 if (!bitmap_empty_p (need_assert_for))
3677 process_assert_insertions ();
3678 update_ssa (TODO_update_ssa_no_phi);
3681 if (dump_file && (dump_flags & TDF_DETAILS))
3683 fprintf (dump_file, "\nSSA form after inserting ASSERT_EXPRs\n");
3684 dump_function_to_file (current_function_decl, dump_file, dump_flags);
3687 free (asserts_for);
3688 BITMAP_FREE (need_assert_for);
3691 /* Return true if all imm uses of VAR are either in STMT, or
3692 feed (optionally through a chain of single imm uses) GIMPLE_COND
3693 in basic block COND_BB. */
3695 bool
3696 vrp_asserts::all_imm_uses_in_stmt_or_feed_cond (tree var,
3697 gimple *stmt,
3698 basic_block cond_bb)
3700 use_operand_p use_p, use2_p;
3701 imm_use_iterator iter;
3703 FOR_EACH_IMM_USE_FAST (use_p, iter, var)
3704 if (USE_STMT (use_p) != stmt)
3706 gimple *use_stmt = USE_STMT (use_p), *use_stmt2;
3707 if (is_gimple_debug (use_stmt))
3708 continue;
3709 while (is_gimple_assign (use_stmt)
3710 && TREE_CODE (gimple_assign_lhs (use_stmt)) == SSA_NAME
3711 && single_imm_use (gimple_assign_lhs (use_stmt),
3712 &use2_p, &use_stmt2))
3713 use_stmt = use_stmt2;
3714 if (gimple_code (use_stmt) != GIMPLE_COND
3715 || gimple_bb (use_stmt) != cond_bb)
3716 return false;
3718 return true;
3721 /* Convert range assertion expressions into the implied copies and
3722 copy propagate away the copies. Doing the trivial copy propagation
3723 here avoids the need to run the full copy propagation pass after
3724 VRP.
3726 FIXME, this will eventually lead to copy propagation removing the
3727 names that had useful range information attached to them. For
3728 instance, if we had the assertion N_i = ASSERT_EXPR <N_j, N_j > 3>,
3729 then N_i will have the range [3, +INF].
3731 However, by converting the assertion into the implied copy
3732 operation N_i = N_j, we will then copy-propagate N_j into the uses
3733 of N_i and lose the range information. We may want to hold on to
3734 ASSERT_EXPRs a little while longer as the ranges could be used in
3735 things like jump threading.
3737 The problem with keeping ASSERT_EXPRs around is that passes after
3738 VRP need to handle them appropriately.
3740 Another approach would be to make the range information a first
3741 class property of the SSA_NAME so that it can be queried from
3742 any pass. This is made somewhat more complex by the need for
3743 multiple ranges to be associated with one SSA_NAME. */
3745 void
3746 vrp_asserts::remove_range_assertions ()
3748 basic_block bb;
3749 gimple_stmt_iterator si;
3750 /* 1 if looking at ASSERT_EXPRs immediately at the beginning of
3751 a basic block preceeded by GIMPLE_COND branching to it and
3752 __builtin_trap, -1 if not yet checked, 0 otherwise. */
3753 int is_unreachable;
3755 /* Note that the BSI iterator bump happens at the bottom of the
3756 loop and no bump is necessary if we're removing the statement
3757 referenced by the current BSI. */
3758 FOR_EACH_BB_FN (bb, fun)
3759 for (si = gsi_after_labels (bb), is_unreachable = -1; !gsi_end_p (si);)
3761 gimple *stmt = gsi_stmt (si);
3763 if (is_gimple_assign (stmt)
3764 && gimple_assign_rhs_code (stmt) == ASSERT_EXPR)
3766 tree lhs = gimple_assign_lhs (stmt);
3767 tree rhs = gimple_assign_rhs1 (stmt);
3768 tree var;
3770 var = ASSERT_EXPR_VAR (rhs);
3772 if (TREE_CODE (var) == SSA_NAME
3773 && !POINTER_TYPE_P (TREE_TYPE (lhs))
3774 && SSA_NAME_RANGE_INFO (lhs))
3776 if (is_unreachable == -1)
3778 is_unreachable = 0;
3779 if (single_pred_p (bb)
3780 && assert_unreachable_fallthru_edge_p
3781 (single_pred_edge (bb)))
3782 is_unreachable = 1;
3784 /* Handle
3785 if (x_7 >= 10 && x_7 < 20)
3786 __builtin_unreachable ();
3787 x_8 = ASSERT_EXPR <x_7, ...>;
3788 if the only uses of x_7 are in the ASSERT_EXPR and
3789 in the condition. In that case, we can copy the
3790 range info from x_8 computed in this pass also
3791 for x_7. */
3792 if (is_unreachable
3793 && all_imm_uses_in_stmt_or_feed_cond (var, stmt,
3794 single_pred (bb)))
3796 set_range_info (var, SSA_NAME_RANGE_TYPE (lhs),
3797 SSA_NAME_RANGE_INFO (lhs)->get_min (),
3798 SSA_NAME_RANGE_INFO (lhs)->get_max ());
3799 maybe_set_nonzero_bits (single_pred_edge (bb), var);
3803 /* Propagate the RHS into every use of the LHS. For SSA names
3804 also propagate abnormals as it merely restores the original
3805 IL in this case (an replace_uses_by would assert). */
3806 if (TREE_CODE (var) == SSA_NAME)
3808 imm_use_iterator iter;
3809 use_operand_p use_p;
3810 gimple *use_stmt;
3811 FOR_EACH_IMM_USE_STMT (use_stmt, iter, lhs)
3812 FOR_EACH_IMM_USE_ON_STMT (use_p, iter)
3813 SET_USE (use_p, var);
3815 else
3816 replace_uses_by (lhs, var);
3818 /* And finally, remove the copy, it is not needed. */
3819 gsi_remove (&si, true);
3820 release_defs (stmt);
3822 else
3824 if (!is_gimple_debug (gsi_stmt (si)))
3825 is_unreachable = 0;
3826 gsi_next (&si);
3831 class vrp_prop : public ssa_propagation_engine
3833 public:
3834 vrp_prop (vr_values *v)
3835 : ssa_propagation_engine (),
3836 m_vr_values (v) { }
3838 void initialize (struct function *);
3839 void finalize ();
3841 private:
3842 enum ssa_prop_result visit_stmt (gimple *, edge *, tree *) FINAL OVERRIDE;
3843 enum ssa_prop_result visit_phi (gphi *) FINAL OVERRIDE;
3845 struct function *fun;
3846 vr_values *m_vr_values;
3849 /* Initialization required by ssa_propagate engine. */
3851 void
3852 vrp_prop::initialize (struct function *fn)
3854 basic_block bb;
3855 fun = fn;
3857 FOR_EACH_BB_FN (bb, fun)
3859 for (gphi_iterator si = gsi_start_phis (bb); !gsi_end_p (si);
3860 gsi_next (&si))
3862 gphi *phi = si.phi ();
3863 if (!stmt_interesting_for_vrp (phi))
3865 tree lhs = PHI_RESULT (phi);
3866 m_vr_values->set_def_to_varying (lhs);
3867 prop_set_simulate_again (phi, false);
3869 else
3870 prop_set_simulate_again (phi, true);
3873 for (gimple_stmt_iterator si = gsi_start_bb (bb); !gsi_end_p (si);
3874 gsi_next (&si))
3876 gimple *stmt = gsi_stmt (si);
3878 /* If the statement is a control insn, then we do not
3879 want to avoid simulating the statement once. Failure
3880 to do so means that those edges will never get added. */
3881 if (stmt_ends_bb_p (stmt))
3882 prop_set_simulate_again (stmt, true);
3883 else if (!stmt_interesting_for_vrp (stmt))
3885 m_vr_values->set_defs_to_varying (stmt);
3886 prop_set_simulate_again (stmt, false);
3888 else
3889 prop_set_simulate_again (stmt, true);
3894 /* Evaluate statement STMT. If the statement produces a useful range,
3895 return SSA_PROP_INTERESTING and record the SSA name with the
3896 interesting range into *OUTPUT_P.
3898 If STMT is a conditional branch and we can determine its truth
3899 value, the taken edge is recorded in *TAKEN_EDGE_P.
3901 If STMT produces a varying value, return SSA_PROP_VARYING. */
3903 enum ssa_prop_result
3904 vrp_prop::visit_stmt (gimple *stmt, edge *taken_edge_p, tree *output_p)
3906 tree lhs = gimple_get_lhs (stmt);
3907 value_range_equiv vr;
3908 m_vr_values->extract_range_from_stmt (stmt, taken_edge_p, output_p, &vr);
3910 if (*output_p)
3912 if (m_vr_values->update_value_range (*output_p, &vr))
3914 if (dump_file && (dump_flags & TDF_DETAILS))
3916 fprintf (dump_file, "Found new range for ");
3917 print_generic_expr (dump_file, *output_p);
3918 fprintf (dump_file, ": ");
3919 dump_value_range (dump_file, &vr);
3920 fprintf (dump_file, "\n");
3923 if (vr.varying_p ())
3924 return SSA_PROP_VARYING;
3926 return SSA_PROP_INTERESTING;
3928 return SSA_PROP_NOT_INTERESTING;
3931 if (is_gimple_call (stmt) && gimple_call_internal_p (stmt))
3932 switch (gimple_call_internal_fn (stmt))
3934 case IFN_ADD_OVERFLOW:
3935 case IFN_SUB_OVERFLOW:
3936 case IFN_MUL_OVERFLOW:
3937 case IFN_ATOMIC_COMPARE_EXCHANGE:
3938 /* These internal calls return _Complex integer type,
3939 which VRP does not track, but the immediate uses
3940 thereof might be interesting. */
3941 if (lhs && TREE_CODE (lhs) == SSA_NAME)
3943 imm_use_iterator iter;
3944 use_operand_p use_p;
3945 enum ssa_prop_result res = SSA_PROP_VARYING;
3947 m_vr_values->set_def_to_varying (lhs);
3949 FOR_EACH_IMM_USE_FAST (use_p, iter, lhs)
3951 gimple *use_stmt = USE_STMT (use_p);
3952 if (!is_gimple_assign (use_stmt))
3953 continue;
3954 enum tree_code rhs_code = gimple_assign_rhs_code (use_stmt);
3955 if (rhs_code != REALPART_EXPR && rhs_code != IMAGPART_EXPR)
3956 continue;
3957 tree rhs1 = gimple_assign_rhs1 (use_stmt);
3958 tree use_lhs = gimple_assign_lhs (use_stmt);
3959 if (TREE_CODE (rhs1) != rhs_code
3960 || TREE_OPERAND (rhs1, 0) != lhs
3961 || TREE_CODE (use_lhs) != SSA_NAME
3962 || !stmt_interesting_for_vrp (use_stmt)
3963 || (!INTEGRAL_TYPE_P (TREE_TYPE (use_lhs))
3964 || !TYPE_MIN_VALUE (TREE_TYPE (use_lhs))
3965 || !TYPE_MAX_VALUE (TREE_TYPE (use_lhs))))
3966 continue;
3968 /* If there is a change in the value range for any of the
3969 REALPART_EXPR/IMAGPART_EXPR immediate uses, return
3970 SSA_PROP_INTERESTING. If there are any REALPART_EXPR
3971 or IMAGPART_EXPR immediate uses, but none of them have
3972 a change in their value ranges, return
3973 SSA_PROP_NOT_INTERESTING. If there are no
3974 {REAL,IMAG}PART_EXPR uses at all,
3975 return SSA_PROP_VARYING. */
3976 value_range_equiv new_vr;
3977 m_vr_values->extract_range_basic (&new_vr, use_stmt);
3978 const value_range_equiv *old_vr
3979 = m_vr_values->get_value_range (use_lhs);
3980 if (!old_vr->equal_p (new_vr, /*ignore_equivs=*/false))
3981 res = SSA_PROP_INTERESTING;
3982 else
3983 res = SSA_PROP_NOT_INTERESTING;
3984 new_vr.equiv_clear ();
3985 if (res == SSA_PROP_INTERESTING)
3987 *output_p = lhs;
3988 return res;
3992 return res;
3994 break;
3995 default:
3996 break;
3999 /* All other statements produce nothing of interest for VRP, so mark
4000 their outputs varying and prevent further simulation. */
4001 m_vr_values->set_defs_to_varying (stmt);
4003 return (*taken_edge_p) ? SSA_PROP_INTERESTING : SSA_PROP_VARYING;
4006 /* Visit all arguments for PHI node PHI that flow through executable
4007 edges. If a valid value range can be derived from all the incoming
4008 value ranges, set a new range for the LHS of PHI. */
4010 enum ssa_prop_result
4011 vrp_prop::visit_phi (gphi *phi)
4013 tree lhs = PHI_RESULT (phi);
4014 value_range_equiv vr_result;
4015 m_vr_values->extract_range_from_phi_node (phi, &vr_result);
4016 if (m_vr_values->update_value_range (lhs, &vr_result))
4018 if (dump_file && (dump_flags & TDF_DETAILS))
4020 fprintf (dump_file, "Found new range for ");
4021 print_generic_expr (dump_file, lhs);
4022 fprintf (dump_file, ": ");
4023 dump_value_range (dump_file, &vr_result);
4024 fprintf (dump_file, "\n");
4027 if (vr_result.varying_p ())
4028 return SSA_PROP_VARYING;
4030 return SSA_PROP_INTERESTING;
4033 /* Nothing changed, don't add outgoing edges. */
4034 return SSA_PROP_NOT_INTERESTING;
4037 /* Traverse all the blocks folding conditionals with known ranges. */
4039 void
4040 vrp_prop::finalize ()
4042 size_t i;
4044 /* We have completed propagating through the lattice. */
4045 m_vr_values->set_lattice_propagation_complete ();
4047 if (dump_file)
4049 fprintf (dump_file, "\nValue ranges after VRP:\n\n");
4050 m_vr_values->dump_all_value_ranges (dump_file);
4051 fprintf (dump_file, "\n");
4054 /* Set value range to non pointer SSA_NAMEs. */
4055 for (i = 0; i < num_ssa_names; i++)
4057 tree name = ssa_name (i);
4058 if (!name)
4059 continue;
4061 const value_range_equiv *vr = m_vr_values->get_value_range (name);
4062 if (!name || !vr->constant_p ())
4063 continue;
4065 if (POINTER_TYPE_P (TREE_TYPE (name))
4066 && range_includes_zero_p (vr) == 0)
4067 set_ptr_nonnull (name);
4068 else if (!POINTER_TYPE_P (TREE_TYPE (name)))
4069 set_range_info (name, *vr);
4073 class vrp_folder : public substitute_and_fold_engine
4075 public:
4076 vrp_folder (vr_values *v)
4077 : substitute_and_fold_engine (/* Fold all stmts. */ true),
4078 m_vr_values (v), simplifier (v)
4081 private:
4082 tree value_of_expr (tree name, gimple *stmt) OVERRIDE
4084 return m_vr_values->value_of_expr (name, stmt);
4086 bool fold_stmt (gimple_stmt_iterator *) FINAL OVERRIDE;
4087 bool fold_predicate_in (gimple_stmt_iterator *);
4089 vr_values *m_vr_values;
4090 simplify_using_ranges simplifier;
4093 /* If the statement pointed by SI has a predicate whose value can be
4094 computed using the value range information computed by VRP, compute
4095 its value and return true. Otherwise, return false. */
4097 bool
4098 vrp_folder::fold_predicate_in (gimple_stmt_iterator *si)
4100 bool assignment_p = false;
4101 tree val;
4102 gimple *stmt = gsi_stmt (*si);
4104 if (is_gimple_assign (stmt)
4105 && TREE_CODE_CLASS (gimple_assign_rhs_code (stmt)) == tcc_comparison)
4107 assignment_p = true;
4108 val = simplifier.vrp_evaluate_conditional (gimple_assign_rhs_code (stmt),
4109 gimple_assign_rhs1 (stmt),
4110 gimple_assign_rhs2 (stmt),
4111 stmt);
4113 else if (gcond *cond_stmt = dyn_cast <gcond *> (stmt))
4114 val = simplifier.vrp_evaluate_conditional (gimple_cond_code (cond_stmt),
4115 gimple_cond_lhs (cond_stmt),
4116 gimple_cond_rhs (cond_stmt),
4117 stmt);
4118 else
4119 return false;
4121 if (val)
4123 if (assignment_p)
4124 val = fold_convert (gimple_expr_type (stmt), val);
4126 if (dump_file)
4128 fprintf (dump_file, "Folding predicate ");
4129 print_gimple_expr (dump_file, stmt, 0);
4130 fprintf (dump_file, " to ");
4131 print_generic_expr (dump_file, val);
4132 fprintf (dump_file, "\n");
4135 if (is_gimple_assign (stmt))
4136 gimple_assign_set_rhs_from_tree (si, val);
4137 else
4139 gcc_assert (gimple_code (stmt) == GIMPLE_COND);
4140 gcond *cond_stmt = as_a <gcond *> (stmt);
4141 if (integer_zerop (val))
4142 gimple_cond_make_false (cond_stmt);
4143 else if (integer_onep (val))
4144 gimple_cond_make_true (cond_stmt);
4145 else
4146 gcc_unreachable ();
4149 return true;
4152 return false;
4155 /* Callback for substitute_and_fold folding the stmt at *SI. */
4157 bool
4158 vrp_folder::fold_stmt (gimple_stmt_iterator *si)
4160 if (fold_predicate_in (si))
4161 return true;
4163 return simplifier.simplify (si);
4166 /* Blocks which have more than one predecessor and more than
4167 one successor present jump threading opportunities, i.e.,
4168 when the block is reached from a specific predecessor, we
4169 may be able to determine which of the outgoing edges will
4170 be traversed. When this optimization applies, we are able
4171 to avoid conditionals at runtime and we may expose secondary
4172 optimization opportunities.
4174 This class is effectively a driver for the generic jump
4175 threading code. It basically just presents the generic code
4176 with edges that may be suitable for jump threading.
4178 Unlike DOM, we do not iterate VRP if jump threading was successful.
4179 While iterating may expose new opportunities for VRP, it is expected
4180 those opportunities would be very limited and the compile time cost
4181 to expose those opportunities would be significant.
4183 As jump threading opportunities are discovered, they are registered
4184 for later realization. */
4186 class vrp_jump_threader : public dom_walker
4188 public:
4189 vrp_jump_threader (struct function *, vr_values *);
4190 ~vrp_jump_threader ();
4192 void thread_jumps ()
4194 walk (m_fun->cfg->x_entry_block_ptr);
4197 private:
4198 static tree simplify_stmt (gimple *stmt, gimple *within_stmt,
4199 avail_exprs_stack *, basic_block);
4200 virtual edge before_dom_children (basic_block);
4201 virtual void after_dom_children (basic_block);
4203 function *m_fun;
4204 vr_values *m_vr_values;
4205 const_and_copies *m_const_and_copies;
4206 avail_exprs_stack *m_avail_exprs_stack;
4207 hash_table<expr_elt_hasher> *m_avail_exprs;
4208 gcond *m_dummy_cond;
4211 vrp_jump_threader::vrp_jump_threader (struct function *fun, vr_values *v)
4212 : dom_walker (CDI_DOMINATORS, REACHABLE_BLOCKS)
4214 /* Ugh. When substituting values earlier in this pass we can wipe
4215 the dominance information. So rebuild the dominator information
4216 as we need it within the jump threading code. */
4217 calculate_dominance_info (CDI_DOMINATORS);
4219 /* We do not allow VRP information to be used for jump threading
4220 across a back edge in the CFG. Otherwise it becomes too
4221 difficult to avoid eliminating loop exit tests. Of course
4222 EDGE_DFS_BACK is not accurate at this time so we have to
4223 recompute it. */
4224 mark_dfs_back_edges ();
4226 /* Allocate our unwinder stack to unwind any temporary equivalences
4227 that might be recorded. */
4228 m_const_and_copies = new const_and_copies ();
4230 m_dummy_cond = NULL;
4231 m_fun = fun;
4232 m_vr_values = v;
4233 m_avail_exprs = new hash_table<expr_elt_hasher> (1024);
4234 m_avail_exprs_stack = new avail_exprs_stack (m_avail_exprs);
4237 vrp_jump_threader::~vrp_jump_threader ()
4239 /* We do not actually update the CFG or SSA graphs at this point as
4240 ASSERT_EXPRs are still in the IL and cfg cleanup code does not
4241 yet handle ASSERT_EXPRs gracefully. */
4242 delete m_const_and_copies;
4243 delete m_avail_exprs;
4244 delete m_avail_exprs_stack;
4247 /* Called before processing dominator children of BB. We want to look
4248 at ASSERT_EXPRs and record information from them in the appropriate
4249 tables.
4251 We could look at other statements here. It's not seen as likely
4252 to significantly increase the jump threads we discover. */
4254 edge
4255 vrp_jump_threader::before_dom_children (basic_block bb)
4257 gimple_stmt_iterator gsi;
4259 m_avail_exprs_stack->push_marker ();
4260 m_const_and_copies->push_marker ();
4261 for (gsi = gsi_start_nondebug_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
4263 gimple *stmt = gsi_stmt (gsi);
4264 if (gimple_assign_single_p (stmt)
4265 && TREE_CODE (gimple_assign_rhs1 (stmt)) == ASSERT_EXPR)
4267 tree rhs1 = gimple_assign_rhs1 (stmt);
4268 tree cond = TREE_OPERAND (rhs1, 1);
4269 tree inverted = invert_truthvalue (cond);
4270 vec<cond_equivalence> p;
4271 p.create (3);
4272 record_conditions (&p, cond, inverted);
4273 for (unsigned int i = 0; i < p.length (); i++)
4274 m_avail_exprs_stack->record_cond (&p[i]);
4276 tree lhs = gimple_assign_lhs (stmt);
4277 m_const_and_copies->record_const_or_copy (lhs,
4278 TREE_OPERAND (rhs1, 0));
4279 p.release ();
4280 continue;
4282 break;
4284 return NULL;
4287 /* A trivial wrapper so that we can present the generic jump threading
4288 code with a simple API for simplifying statements. STMT is the
4289 statement we want to simplify, WITHIN_STMT provides the location
4290 for any overflow warnings.
4292 ?? This should be cleaned up. There's a virtually identical copy
4293 of this function in tree-ssa-dom.c. */
4295 tree
4296 vrp_jump_threader::simplify_stmt (gimple *stmt,
4297 gimple *within_stmt,
4298 avail_exprs_stack *avail_exprs_stack,
4299 basic_block bb)
4301 /* First see if the conditional is in the hash table. */
4302 tree cached_lhs = avail_exprs_stack->lookup_avail_expr (stmt, false, true);
4303 if (cached_lhs && is_gimple_min_invariant (cached_lhs))
4304 return cached_lhs;
4306 class vr_values *vr_values = x_vr_values;
4307 if (gcond *cond_stmt = dyn_cast <gcond *> (stmt))
4309 tree op0 = gimple_cond_lhs (cond_stmt);
4310 op0 = lhs_of_dominating_assert (op0, bb, stmt);
4312 tree op1 = gimple_cond_rhs (cond_stmt);
4313 op1 = lhs_of_dominating_assert (op1, bb, stmt);
4315 simplify_using_ranges simplifier (vr_values);
4316 return simplifier.vrp_evaluate_conditional (gimple_cond_code (cond_stmt),
4317 op0, op1, within_stmt);
4320 if (gswitch *switch_stmt = dyn_cast <gswitch *> (stmt))
4322 tree op = gimple_switch_index (switch_stmt);
4323 if (TREE_CODE (op) != SSA_NAME)
4324 return NULL_TREE;
4326 op = lhs_of_dominating_assert (op, bb, stmt);
4328 const value_range_equiv *vr = vr_values->get_value_range (op);
4329 return find_case_label_range (switch_stmt, vr);
4332 if (gassign *assign_stmt = dyn_cast <gassign *> (stmt))
4334 tree lhs = gimple_assign_lhs (assign_stmt);
4335 if (TREE_CODE (lhs) == SSA_NAME
4336 && (INTEGRAL_TYPE_P (TREE_TYPE (lhs))
4337 || POINTER_TYPE_P (TREE_TYPE (lhs)))
4338 && stmt_interesting_for_vrp (stmt))
4340 edge dummy_e;
4341 tree dummy_tree;
4342 value_range_equiv new_vr;
4343 vr_values->extract_range_from_stmt (stmt, &dummy_e,
4344 &dummy_tree, &new_vr);
4345 tree singleton;
4346 if (new_vr.singleton_p (&singleton))
4347 return singleton;
4351 return NULL_TREE;
4354 /* Called after processing dominator children of BB. This is where we
4355 actually call into the threader. */
4356 void
4357 vrp_jump_threader::after_dom_children (basic_block bb)
4359 if (!m_dummy_cond)
4360 m_dummy_cond = gimple_build_cond (NE_EXPR,
4361 integer_zero_node, integer_zero_node,
4362 NULL, NULL);
4364 x_vr_values = m_vr_values;
4365 thread_outgoing_edges (bb, m_dummy_cond, m_const_and_copies,
4366 m_avail_exprs_stack, NULL,
4367 simplify_stmt);
4368 x_vr_values = NULL;
4370 m_avail_exprs_stack->pop_to_marker ();
4371 m_const_and_copies->pop_to_marker ();
4374 /* STMT is a conditional at the end of a basic block.
4376 If the conditional is of the form SSA_NAME op constant and the SSA_NAME
4377 was set via a type conversion, try to replace the SSA_NAME with the RHS
4378 of the type conversion. Doing so makes the conversion dead which helps
4379 subsequent passes. */
4381 static void
4382 vrp_simplify_cond_using_ranges (vr_values *query, gcond *stmt)
4384 tree op0 = gimple_cond_lhs (stmt);
4385 tree op1 = gimple_cond_rhs (stmt);
4387 /* If we have a comparison of an SSA_NAME (OP0) against a constant,
4388 see if OP0 was set by a type conversion where the source of
4389 the conversion is another SSA_NAME with a range that fits
4390 into the range of OP0's type.
4392 If so, the conversion is redundant as the earlier SSA_NAME can be
4393 used for the comparison directly if we just massage the constant in the
4394 comparison. */
4395 if (TREE_CODE (op0) == SSA_NAME
4396 && TREE_CODE (op1) == INTEGER_CST)
4398 gimple *def_stmt = SSA_NAME_DEF_STMT (op0);
4399 tree innerop;
4401 if (!is_gimple_assign (def_stmt))
4402 return;
4404 switch (gimple_assign_rhs_code (def_stmt))
4406 CASE_CONVERT:
4407 innerop = gimple_assign_rhs1 (def_stmt);
4408 break;
4409 case VIEW_CONVERT_EXPR:
4410 innerop = TREE_OPERAND (gimple_assign_rhs1 (def_stmt), 0);
4411 if (!INTEGRAL_TYPE_P (TREE_TYPE (innerop)))
4412 return;
4413 break;
4414 default:
4415 return;
4418 if (TREE_CODE (innerop) == SSA_NAME
4419 && !POINTER_TYPE_P (TREE_TYPE (innerop))
4420 && !SSA_NAME_OCCURS_IN_ABNORMAL_PHI (innerop)
4421 && desired_pro_or_demotion_p (TREE_TYPE (innerop), TREE_TYPE (op0)))
4423 const value_range *vr = query->get_value_range (innerop);
4425 if (range_int_cst_p (vr)
4426 && range_fits_type_p (vr,
4427 TYPE_PRECISION (TREE_TYPE (op0)),
4428 TYPE_SIGN (TREE_TYPE (op0)))
4429 && int_fits_type_p (op1, TREE_TYPE (innerop)))
4431 tree newconst = fold_convert (TREE_TYPE (innerop), op1);
4432 gimple_cond_set_lhs (stmt, innerop);
4433 gimple_cond_set_rhs (stmt, newconst);
4434 update_stmt (stmt);
4435 if (dump_file && (dump_flags & TDF_DETAILS))
4437 fprintf (dump_file, "Folded into: ");
4438 print_gimple_stmt (dump_file, stmt, 0, TDF_SLIM);
4439 fprintf (dump_file, "\n");
4446 /* Main entry point to VRP (Value Range Propagation). This pass is
4447 loosely based on J. R. C. Patterson, ``Accurate Static Branch
4448 Prediction by Value Range Propagation,'' in SIGPLAN Conference on
4449 Programming Language Design and Implementation, pp. 67-78, 1995.
4450 Also available at http://citeseer.ist.psu.edu/patterson95accurate.html
4452 This is essentially an SSA-CCP pass modified to deal with ranges
4453 instead of constants.
4455 While propagating ranges, we may find that two or more SSA name
4456 have equivalent, though distinct ranges. For instance,
4458 1 x_9 = p_3->a;
4459 2 p_4 = ASSERT_EXPR <p_3, p_3 != 0>
4460 3 if (p_4 == q_2)
4461 4 p_5 = ASSERT_EXPR <p_4, p_4 == q_2>;
4462 5 endif
4463 6 if (q_2)
4465 In the code above, pointer p_5 has range [q_2, q_2], but from the
4466 code we can also determine that p_5 cannot be NULL and, if q_2 had
4467 a non-varying range, p_5's range should also be compatible with it.
4469 These equivalences are created by two expressions: ASSERT_EXPR and
4470 copy operations. Since p_5 is an assertion on p_4, and p_4 was the
4471 result of another assertion, then we can use the fact that p_5 and
4472 p_4 are equivalent when evaluating p_5's range.
4474 Together with value ranges, we also propagate these equivalences
4475 between names so that we can take advantage of information from
4476 multiple ranges when doing final replacement. Note that this
4477 equivalency relation is transitive but not symmetric.
4479 In the example above, p_5 is equivalent to p_4, q_2 and p_3, but we
4480 cannot assert that q_2 is equivalent to p_5 because q_2 may be used
4481 in contexts where that assertion does not hold (e.g., in line 6).
4483 TODO, the main difference between this pass and Patterson's is that
4484 we do not propagate edge probabilities. We only compute whether
4485 edges can be taken or not. That is, instead of having a spectrum
4486 of jump probabilities between 0 and 1, we only deal with 0, 1 and
4487 DON'T KNOW. In the future, it may be worthwhile to propagate
4488 probabilities to aid branch prediction. */
4490 static unsigned int
4491 execute_vrp (struct function *fun, bool warn_array_bounds_p)
4493 loop_optimizer_init (LOOPS_NORMAL | LOOPS_HAVE_RECORDED_EXITS);
4494 rewrite_into_loop_closed_ssa (NULL, TODO_update_ssa);
4495 scev_initialize ();
4497 /* ??? This ends up using stale EDGE_DFS_BACK for liveness computation.
4498 Inserting assertions may split edges which will invalidate
4499 EDGE_DFS_BACK. */
4500 vrp_asserts assert_engine (fun);
4501 assert_engine.insert_range_assertions ();
4503 threadedge_initialize_values ();
4505 /* For visiting PHI nodes we need EDGE_DFS_BACK computed. */
4506 mark_dfs_back_edges ();
4508 vr_values vrp_vr_values;
4510 class vrp_prop vrp_prop (&vrp_vr_values);
4511 vrp_prop.initialize (fun);
4512 vrp_prop.ssa_propagate ();
4514 /* Instantiate the folder here, so that edge cleanups happen at the
4515 end of this function. */
4516 vrp_folder folder (&vrp_vr_values);
4517 vrp_prop.finalize ();
4519 /* If we're checking array refs, we want to merge information on
4520 the executability of each edge between vrp_folder and the
4521 check_array_bounds_dom_walker: each can clear the
4522 EDGE_EXECUTABLE flag on edges, in different ways.
4524 Hence, if we're going to call check_all_array_refs, set
4525 the flag on every edge now, rather than in
4526 check_array_bounds_dom_walker's ctor; vrp_folder may clear
4527 it from some edges. */
4528 if (warn_array_bounds && warn_array_bounds_p)
4529 set_all_edges_as_executable (fun);
4531 folder.substitute_and_fold ();
4533 if (warn_array_bounds && warn_array_bounds_p)
4535 array_bounds_checker array_checker (fun, &vrp_vr_values);
4536 array_checker.check ();
4539 /* We must identify jump threading opportunities before we release
4540 the datastructures built by VRP. */
4541 vrp_jump_threader threader (fun, &vrp_vr_values);
4542 threader.thread_jumps ();
4544 /* A comparison of an SSA_NAME against a constant where the SSA_NAME
4545 was set by a type conversion can often be rewritten to use the
4546 RHS of the type conversion.
4548 However, doing so inhibits jump threading through the comparison.
4549 So that transformation is not performed until after jump threading
4550 is complete. */
4551 basic_block bb;
4552 FOR_EACH_BB_FN (bb, fun)
4554 gimple *last = last_stmt (bb);
4555 if (last && gimple_code (last) == GIMPLE_COND)
4556 vrp_simplify_cond_using_ranges (&vrp_vr_values,
4557 as_a <gcond *> (last));
4560 free_numbers_of_iterations_estimates (fun);
4562 /* ASSERT_EXPRs must be removed before finalizing jump threads
4563 as finalizing jump threads calls the CFG cleanup code which
4564 does not properly handle ASSERT_EXPRs. */
4565 assert_engine.remove_range_assertions ();
4567 /* If we exposed any new variables, go ahead and put them into
4568 SSA form now, before we handle jump threading. This simplifies
4569 interactions between rewriting of _DECL nodes into SSA form
4570 and rewriting SSA_NAME nodes into SSA form after block
4571 duplication and CFG manipulation. */
4572 update_ssa (TODO_update_ssa);
4574 /* We identified all the jump threading opportunities earlier, but could
4575 not transform the CFG at that time. This routine transforms the
4576 CFG and arranges for the dominator tree to be rebuilt if necessary.
4578 Note the SSA graph update will occur during the normal TODO
4579 processing by the pass manager. */
4580 thread_through_all_blocks (false);
4582 threadedge_finalize_values ();
4584 scev_finalize ();
4585 loop_optimizer_finalize ();
4586 return 0;
4589 namespace {
4591 const pass_data pass_data_vrp =
4593 GIMPLE_PASS, /* type */
4594 "vrp", /* name */
4595 OPTGROUP_NONE, /* optinfo_flags */
4596 TV_TREE_VRP, /* tv_id */
4597 PROP_ssa, /* properties_required */
4598 0, /* properties_provided */
4599 0, /* properties_destroyed */
4600 0, /* todo_flags_start */
4601 ( TODO_cleanup_cfg | TODO_update_ssa ), /* todo_flags_finish */
4604 class pass_vrp : public gimple_opt_pass
4606 public:
4607 pass_vrp (gcc::context *ctxt)
4608 : gimple_opt_pass (pass_data_vrp, ctxt), warn_array_bounds_p (false)
4611 /* opt_pass methods: */
4612 opt_pass * clone () { return new pass_vrp (m_ctxt); }
4613 void set_pass_param (unsigned int n, bool param)
4615 gcc_assert (n == 0);
4616 warn_array_bounds_p = param;
4618 virtual bool gate (function *) { return flag_tree_vrp != 0; }
4619 virtual unsigned int execute (function *fun)
4620 { return execute_vrp (fun, warn_array_bounds_p); }
4622 private:
4623 bool warn_array_bounds_p;
4624 }; // class pass_vrp
4626 } // anon namespace
4628 gimple_opt_pass *
4629 make_pass_vrp (gcc::context *ctxt)
4631 return new pass_vrp (ctxt);
4635 /* Worker for determine_value_range. */
4637 static void
4638 determine_value_range_1 (value_range *vr, tree expr)
4640 if (BINARY_CLASS_P (expr))
4642 value_range vr0, vr1;
4643 determine_value_range_1 (&vr0, TREE_OPERAND (expr, 0));
4644 determine_value_range_1 (&vr1, TREE_OPERAND (expr, 1));
4645 range_fold_binary_expr (vr, TREE_CODE (expr), TREE_TYPE (expr),
4646 &vr0, &vr1);
4648 else if (UNARY_CLASS_P (expr))
4650 value_range vr0;
4651 determine_value_range_1 (&vr0, TREE_OPERAND (expr, 0));
4652 range_fold_unary_expr (vr, TREE_CODE (expr), TREE_TYPE (expr),
4653 &vr0, TREE_TYPE (TREE_OPERAND (expr, 0)));
4655 else if (TREE_CODE (expr) == INTEGER_CST)
4656 vr->set (expr);
4657 else
4659 value_range_kind kind;
4660 wide_int min, max;
4661 /* For SSA names try to extract range info computed by VRP. Otherwise
4662 fall back to varying. */
4663 if (TREE_CODE (expr) == SSA_NAME
4664 && INTEGRAL_TYPE_P (TREE_TYPE (expr))
4665 && (kind = get_range_info (expr, &min, &max)) != VR_VARYING)
4666 vr->set (wide_int_to_tree (TREE_TYPE (expr), min),
4667 wide_int_to_tree (TREE_TYPE (expr), max),
4668 kind);
4669 else
4670 vr->set_varying (TREE_TYPE (expr));
4674 /* Compute a value-range for EXPR and set it in *MIN and *MAX. Return
4675 the determined range type. */
4677 value_range_kind
4678 determine_value_range (tree expr, wide_int *min, wide_int *max)
4680 value_range vr;
4681 determine_value_range_1 (&vr, expr);
4682 if (vr.constant_p ())
4684 *min = wi::to_wide (vr.min ());
4685 *max = wi::to_wide (vr.max ());
4686 return vr.kind ();
4689 return VR_VARYING;