]>
Commit | Line | Data |
---|---|---|
0732f75f | 1 | /* SSA Jump Threading |
a945c346 | 2 | Copyright (C) 2005-2024 Free Software Foundation, Inc. |
0732f75f JL |
3 | |
4 | This file is part of GCC. | |
5 | ||
6 | GCC is free software; you can redistribute it and/or modify | |
7 | it under the terms of the GNU General Public License as published by | |
8 | the Free Software Foundation; either version 3, or (at your option) | |
9 | any later version. | |
10 | ||
11 | GCC is distributed in the hope that it will be useful, | |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | GNU General Public License for more details. | |
15 | ||
16 | You should have received a copy of the GNU General Public License | |
17 | along with GCC; see the file COPYING3. If not see | |
18 | <http://www.gnu.org/licenses/>. */ | |
19 | ||
20 | #include "config.h" | |
21 | #include "system.h" | |
22 | #include "coretypes.h" | |
23 | #include "backend.h" | |
24 | #include "predict.h" | |
25 | #include "tree.h" | |
26 | #include "gimple.h" | |
27 | #include "fold-const.h" | |
28 | #include "cfgloop.h" | |
29 | #include "gimple-iterator.h" | |
30 | #include "tree-cfg.h" | |
31 | #include "tree-ssa-threadupdate.h" | |
0732f75f JL |
32 | #include "tree-ssa-loop.h" |
33 | #include "cfganal.h" | |
34 | #include "tree-pass.h" | |
8b2ef235 JL |
35 | #include "gimple-ssa.h" |
36 | #include "tree-phinodes.h" | |
27bddc4a | 37 | #include "tree-inline.h" |
0a4e5cf3 | 38 | #include "tree-vectorizer.h" |
2e96b5f1 AH |
39 | #include "value-range.h" |
40 | #include "gimple-range.h" | |
41 | #include "tree-ssa-threadedge.h" | |
42 | #include "gimple-range-path.h" | |
43 | #include "ssa.h" | |
44 | #include "tree-cfgcleanup.h" | |
779275c0 | 45 | #include "tree-pretty-print.h" |
01b50387 | 46 | #include "cfghooks.h" |
bc5baac5 | 47 | #include "dbgcnt.h" |
0732f75f | 48 | |
69e55442 AH |
49 | // Path registry for the backwards threader. After all paths have been |
50 | // registered with register_path(), thread_through_all_blocks() is called | |
51 | // to modify the CFG. | |
52 | ||
7f6c2258 | 53 | class back_threader_registry : public back_jt_path_registry |
69e55442 AH |
54 | { |
55 | public: | |
69e55442 | 56 | bool register_path (const vec<basic_block> &, edge taken); |
69e55442 AH |
57 | }; |
58 | ||
59 | // Class to abstract the profitability code for the backwards threader. | |
60 | ||
61 | class back_threader_profitability | |
62 | { | |
63 | public: | |
bac07a1d | 64 | back_threader_profitability (bool speed_p, gimple *stmt); |
d45ddc2c | 65 | bool possibly_profitable_path_p (const vec<basic_block> &, bool *); |
bac07a1d RB |
66 | bool profitable_path_p (const vec<basic_block> &, |
67 | edge taken, bool *irreducible_loop); | |
69e55442 AH |
68 | private: |
69 | const bool m_speed_p; | |
bac07a1d RB |
70 | int m_exit_jump_benefit; |
71 | bool m_threaded_multiway_branch; | |
72 | // The following are computed by possibly_profitable_path_p | |
73 | bool m_threaded_through_latch; | |
74 | bool m_multiway_branch_in_path; | |
75 | bool m_contains_hot_bb; | |
76 | int m_n_insns; | |
69e55442 AH |
77 | }; |
78 | ||
bac07a1d RB |
79 | back_threader_profitability::back_threader_profitability (bool speed_p, |
80 | gimple *last) | |
81 | : m_speed_p (speed_p) | |
82 | { | |
83 | m_threaded_multiway_branch = (gimple_code (last) == GIMPLE_SWITCH | |
84 | || gimple_code (last) == GIMPLE_GOTO); | |
85 | // The forward threader has estimate_threading_killed_stmts, in | |
86 | // particular it estimates further DCE from eliminating the exit | |
87 | // control stmt. | |
88 | m_exit_jump_benefit = estimate_num_insns (last, &eni_size_weights); | |
89 | } | |
90 | ||
4e0f56d7 AH |
91 | // Back threader flags. |
92 | #define BT_NONE 0 | |
93 | // Generate fast code at the expense of code size. | |
94 | #define BT_SPEED 1 | |
95 | // Resolve unknown SSAs on entry to a threading path. If set, use the | |
96 | // ranger. If not, assume all ranges on entry to a path are VARYING. | |
97 | #define BT_RESOLVE 2 | |
98 | ||
2e96b5f1 AH |
99 | class back_threader |
100 | { | |
2e96b5f1 | 101 | public: |
bc5baac5 | 102 | back_threader (function *fun, unsigned flags, bool first); |
4e0f56d7 AH |
103 | ~back_threader (); |
104 | unsigned thread_blocks (); | |
2e96b5f1 | 105 | private: |
4e0f56d7 | 106 | void maybe_thread_block (basic_block bb); |
bc5baac5 | 107 | bool debug_counter (); |
bac07a1d | 108 | edge maybe_register_path (back_threader_profitability &); |
53080c5b | 109 | void maybe_register_path_dump (edge taken_edge); |
bac07a1d RB |
110 | void find_paths_to_names (basic_block bb, bitmap imports, unsigned, |
111 | back_threader_profitability &); | |
2e96b5f1 AH |
112 | edge find_taken_edge (const vec<basic_block> &path); |
113 | edge find_taken_edge_cond (const vec<basic_block> &path, gcond *); | |
114 | edge find_taken_edge_switch (const vec<basic_block> &path, gswitch *); | |
779275c0 AH |
115 | virtual void debug (); |
116 | virtual void dump (FILE *out); | |
2e96b5f1 | 117 | |
34cd97ff | 118 | back_threader_registry m_registry; |
2e96b5f1 AH |
119 | |
120 | // Current path being analyzed. | |
121 | auto_vec<basic_block> m_path; | |
122 | // Hash to mark visited BBs while analyzing a path. | |
123 | hash_set<basic_block> m_visited_bbs; | |
124 | // The set of SSA names, any of which could potentially change the | |
125 | // value of the final conditional in a path. | |
d71e1be7 | 126 | auto_bitmap m_imports; |
2e96b5f1 AH |
127 | // The last statement in the path. |
128 | gimple *m_last_stmt; | |
2e96b5f1 AH |
129 | // Marker to differentiate unreachable edges. |
130 | static const edge UNREACHABLE_EDGE; | |
401aaa59 AH |
131 | // Set to TRUE if unknown SSA names along a path should be resolved |
132 | // with the ranger. Otherwise, unknown SSA names are assumed to be | |
5d4d64fa | 133 | // VARYING. Setting to true is more precise but slower. |
4e0f56d7 | 134 | function *m_fun; |
011d0a03 AH |
135 | // Ranger for the path solver. |
136 | gimple_ranger *m_ranger; | |
4e0f56d7 | 137 | unsigned m_flags; |
bc5baac5 AH |
138 | // Set to TRUE for the first of each thread[12] pass or the first of |
139 | // each threadfull[12] pass. This is used to differentiate between | |
140 | // the different threading passes so we can set up debug counters. | |
141 | bool m_first; | |
2e96b5f1 AH |
142 | }; |
143 | ||
144 | // Used to differentiate unreachable edges, so we may stop the search | |
145 | // in a the given direction. | |
146 | const edge back_threader::UNREACHABLE_EDGE = (edge) -1; | |
147 | ||
bc5baac5 | 148 | back_threader::back_threader (function *fun, unsigned flags, bool first) |
bac07a1d | 149 | : m_first (first) |
2e96b5f1 | 150 | { |
4e0f56d7 AH |
151 | if (flags & BT_SPEED) |
152 | loop_optimizer_init (LOOPS_HAVE_PREHEADERS | LOOPS_HAVE_SIMPLE_LATCHES); | |
153 | else | |
154 | loop_optimizer_init (AVOID_CFG_MODIFICATIONS); | |
155 | ||
156 | m_fun = fun; | |
157 | m_flags = flags; | |
2e96b5f1 | 158 | m_last_stmt = NULL; |
eb5ee646 AH |
159 | |
160 | // The path solver needs EDGE_DFS_BACK in resolving mode. | |
161 | if (flags & BT_RESOLVE) | |
162 | mark_dfs_back_edges (); | |
011d0a03 AH |
163 | |
164 | m_ranger = new gimple_ranger; | |
4e0f56d7 AH |
165 | } |
166 | ||
167 | back_threader::~back_threader () | |
168 | { | |
011d0a03 | 169 | delete m_ranger; |
4e0f56d7 | 170 | loop_optimizer_finalize (); |
2e96b5f1 AH |
171 | } |
172 | ||
bc5baac5 AH |
173 | // A wrapper for the various debug counters for the threading passes. |
174 | // Returns TRUE if it's OK to register the current threading | |
175 | // candidate. | |
176 | ||
177 | bool | |
178 | back_threader::debug_counter () | |
179 | { | |
180 | // The ethread pass is mostly harmless ;-). | |
181 | if ((m_flags & BT_SPEED) == 0) | |
182 | return true; | |
183 | ||
184 | if (m_flags & BT_RESOLVE) | |
185 | { | |
186 | if (m_first && !dbg_cnt (back_threadfull1)) | |
187 | return false; | |
188 | ||
189 | if (!m_first && !dbg_cnt (back_threadfull2)) | |
190 | return false; | |
191 | } | |
192 | else | |
193 | { | |
194 | if (m_first && !dbg_cnt (back_thread1)) | |
195 | return false; | |
196 | ||
197 | if (!m_first && !dbg_cnt (back_thread2)) | |
198 | return false; | |
199 | } | |
200 | return true; | |
201 | } | |
202 | ||
a2ab1a5a AH |
203 | static void |
204 | dump_path (FILE *dump_file, const vec<basic_block> &path) | |
205 | { | |
206 | for (unsigned i = path.length (); i > 0; --i) | |
207 | { | |
208 | basic_block bb = path[i - 1]; | |
209 | fprintf (dump_file, "%d", bb->index); | |
210 | if (i > 1) | |
211 | fprintf (dump_file, "->"); | |
212 | } | |
213 | } | |
214 | ||
53080c5b AH |
215 | // Dump details of an attempt to register a path. |
216 | ||
217 | void | |
218 | back_threader::maybe_register_path_dump (edge taken) | |
219 | { | |
220 | if (m_path.is_empty ()) | |
221 | return; | |
222 | ||
223 | fprintf (dump_file, "path: "); | |
a2ab1a5a | 224 | dump_path (dump_file, m_path); |
53080c5b AH |
225 | fprintf (dump_file, "->"); |
226 | ||
227 | if (taken == UNREACHABLE_EDGE) | |
228 | fprintf (dump_file, "xx REJECTED (unreachable)\n"); | |
229 | else if (taken) | |
230 | fprintf (dump_file, "%d SUCCESS\n", taken->dest->index); | |
231 | else | |
232 | fprintf (dump_file, "xx REJECTED\n"); | |
233 | } | |
234 | ||
235 | // If an outgoing edge can be determined out of the current path, | |
236 | // register it for jump threading and return the taken edge. | |
cbeeadff | 237 | // |
2b59cf47 AH |
238 | // Return NULL if it is unprofitable to thread this path, or the |
239 | // outgoing edge is unknown. Return UNREACHABLE_EDGE if the path is | |
240 | // unreachable. | |
2e96b5f1 | 241 | |
cbeeadff | 242 | edge |
bac07a1d | 243 | back_threader::maybe_register_path (back_threader_profitability &profit) |
2e96b5f1 | 244 | { |
cbeeadff | 245 | edge taken_edge = find_taken_edge (m_path); |
2e96b5f1 | 246 | |
cbeeadff | 247 | if (taken_edge && taken_edge != UNREACHABLE_EDGE) |
2e96b5f1 | 248 | { |
837be6c7 RB |
249 | bool irreducible = false; |
250 | if (profit.profitable_path_p (m_path, taken_edge, &irreducible) | |
251 | && debug_counter () | |
252 | && m_registry.register_path (m_path, taken_edge)) | |
cbeeadff | 253 | { |
837be6c7 RB |
254 | if (irreducible) |
255 | vect_free_loop_info_assumptions (m_path[0]->loop_father); | |
2b59cf47 AH |
256 | } |
257 | else | |
837be6c7 | 258 | taken_edge = NULL; |
2e96b5f1 | 259 | } |
53080c5b AH |
260 | |
261 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
262 | maybe_register_path_dump (taken_edge); | |
263 | ||
cbeeadff | 264 | return taken_edge; |
2e96b5f1 AH |
265 | } |
266 | ||
267 | // Return the known taken edge out of a path. If the path can be | |
268 | // determined to be unreachable, return UNREACHABLE_EDGE. If no | |
269 | // outgoing edge can be calculated, return NULL. | |
270 | ||
271 | edge | |
272 | back_threader::find_taken_edge (const vec<basic_block> &path) | |
273 | { | |
274 | gcc_checking_assert (path.length () > 1); | |
275 | switch (gimple_code (m_last_stmt)) | |
276 | { | |
277 | case GIMPLE_COND: | |
278 | return find_taken_edge_cond (path, as_a<gcond *> (m_last_stmt)); | |
279 | ||
280 | case GIMPLE_SWITCH: | |
281 | return find_taken_edge_switch (path, as_a<gswitch *> (m_last_stmt)); | |
282 | ||
283 | default: | |
284 | return NULL; | |
285 | } | |
286 | } | |
287 | ||
288 | // Same as find_taken_edge, but for paths ending in a switch. | |
289 | ||
290 | edge | |
291 | back_threader::find_taken_edge_switch (const vec<basic_block> &path, | |
292 | gswitch *sw) | |
293 | { | |
294 | tree name = gimple_switch_index (sw); | |
295 | int_range_max r; | |
296 | ||
011d0a03 AH |
297 | path_range_query solver (*m_ranger, path, m_imports, m_flags & BT_RESOLVE); |
298 | solver.range_of_expr (r, name, sw); | |
2e96b5f1 AH |
299 | |
300 | if (r.undefined_p ()) | |
301 | return UNREACHABLE_EDGE; | |
302 | ||
303 | if (r.varying_p ()) | |
304 | return NULL; | |
305 | ||
113dab2b AH |
306 | tree label = find_case_label_range (sw, &r); |
307 | if (!label) | |
308 | return NULL; | |
2e96b5f1 | 309 | |
113dab2b | 310 | return find_edge (gimple_bb (sw), label_to_block (cfun, CASE_LABEL (label))); |
2e96b5f1 AH |
311 | } |
312 | ||
313 | // Same as find_taken_edge, but for paths ending in a GIMPLE_COND. | |
314 | ||
315 | edge | |
316 | back_threader::find_taken_edge_cond (const vec<basic_block> &path, | |
317 | gcond *cond) | |
318 | { | |
2e96b5f1 | 319 | int_range_max r; |
2e96b5f1 | 320 | |
011d0a03 AH |
321 | path_range_query solver (*m_ranger, path, m_imports, m_flags & BT_RESOLVE); |
322 | solver.range_of_stmt (r, cond); | |
2e96b5f1 | 323 | |
011d0a03 | 324 | if (solver.unreachable_path_p ()) |
90ef1535 AH |
325 | return UNREACHABLE_EDGE; |
326 | ||
cb779afe AH |
327 | int_range<2> true_range = range_true (); |
328 | int_range<2> false_range = range_false (); | |
2e96b5f1 AH |
329 | |
330 | if (r == true_range || r == false_range) | |
331 | { | |
332 | edge e_true, e_false; | |
333 | basic_block bb = gimple_bb (cond); | |
334 | extract_true_false_edges_from_block (bb, &e_true, &e_false); | |
335 | return r == true_range ? e_true : e_false; | |
336 | } | |
337 | return NULL; | |
338 | } | |
339 | ||
2e96b5f1 AH |
340 | // Find jump threading paths to any of the SSA names in the |
341 | // INTERESTING bitmap, and register any such paths. | |
342 | // | |
2e96b5f1 | 343 | // BB is the current path being processed. |
409978d5 RB |
344 | // |
345 | // OVERALL_PATHS is the search space up to this block | |
2e96b5f1 | 346 | |
98b212c1 | 347 | void |
409978d5 | 348 | back_threader::find_paths_to_names (basic_block bb, bitmap interesting, |
bac07a1d RB |
349 | unsigned overall_paths, |
350 | back_threader_profitability &profit) | |
2e96b5f1 AH |
351 | { |
352 | if (m_visited_bbs.add (bb)) | |
98b212c1 | 353 | return; |
2e96b5f1 AH |
354 | |
355 | m_path.safe_push (bb); | |
356 | ||
bac07a1d RB |
357 | // Try to resolve the path without looking back. Avoid resolving paths |
358 | // we know are large but are not (yet) recognized as Finite State Machine. | |
359 | // ??? Ideally we'd explore the cheapest path to the loop backedge here, | |
360 | // avoiding the exponential greedy search and only start that from there. | |
361 | // Precomputing a path-size-to-immediate-dominator-of-successor for each | |
362 | // edge might help here. Alternatively copying divergent control flow | |
363 | // on the way to the backedge could be worthwhile. | |
364 | bool large_non_fsm; | |
98b212c1 | 365 | if (m_path.length () > 1 |
d45ddc2c | 366 | && (!profit.possibly_profitable_path_p (m_path, &large_non_fsm) |
bac07a1d RB |
367 | || (!large_non_fsm |
368 | && maybe_register_path (profit)))) | |
d86d81a4 | 369 | ; |
401aaa59 | 370 | |
9594e04e RB |
371 | // The backwards thread copier cannot copy blocks that do not belong |
372 | // to the same loop, so when the new source of the path entry no | |
373 | // longer belongs to it we don't need to search further. | |
374 | else if (m_path[0]->loop_father != bb->loop_father) | |
375 | ; | |
376 | ||
409978d5 RB |
377 | // Continue looking for ways to extend the path but limit the |
378 | // search space along a branch | |
379 | else if ((overall_paths = overall_paths * EDGE_COUNT (bb->preds)) | |
380 | <= (unsigned)param_max_jump_thread_paths) | |
2e96b5f1 | 381 | { |
d86d81a4 RB |
382 | // For further greedy searching we want to remove interesting |
383 | // names defined in BB but add ones on the PHI edges for the | |
16b013c9 RB |
384 | // respective edges and adding imports from those stmts. |
385 | // We do this by starting with all names | |
d86d81a4 RB |
386 | // not defined in BB as interesting, collecting a list of |
387 | // interesting PHIs in BB on the fly. Then we iterate over | |
388 | // predecessor edges, adding interesting PHI edge defs to | |
389 | // the set of interesting names to consider when processing it. | |
390 | auto_bitmap new_interesting; | |
16b013c9 | 391 | auto_vec<int, 16> new_imports; |
d86d81a4 RB |
392 | auto_vec<gphi *, 4> interesting_phis; |
393 | bitmap_iterator bi; | |
394 | unsigned i; | |
16b013c9 | 395 | auto_vec<tree, 16> worklist; |
d86d81a4 | 396 | EXECUTE_IF_SET_IN_BITMAP (interesting, 0, i, bi) |
2e96b5f1 | 397 | { |
d86d81a4 RB |
398 | tree name = ssa_name (i); |
399 | gimple *def_stmt = SSA_NAME_DEF_STMT (name); | |
16b013c9 | 400 | /* Imports remain interesting. */ |
d86d81a4 | 401 | if (gimple_bb (def_stmt) != bb) |
d86d81a4 | 402 | { |
16b013c9 RB |
403 | bitmap_set_bit (new_interesting, i); |
404 | continue; | |
405 | } | |
406 | worklist.quick_push (name); | |
407 | while (!worklist.is_empty ()) | |
408 | { | |
409 | tree name = worklist.pop (); | |
410 | gimple *def_stmt = SSA_NAME_DEF_STMT (name); | |
411 | /* Newly discovered imports are interesting. */ | |
412 | if (gimple_bb (def_stmt) != bb) | |
413 | { | |
414 | bitmap_set_bit (new_interesting, SSA_NAME_VERSION (name)); | |
415 | continue; | |
416 | } | |
417 | /* Local PHIs participate in renaming below. */ | |
418 | if (gphi *phi = dyn_cast<gphi *> (def_stmt)) | |
419 | { | |
420 | tree res = gimple_phi_result (phi); | |
421 | if (!SSA_NAME_OCCURS_IN_ABNORMAL_PHI (res)) | |
422 | interesting_phis.safe_push (phi); | |
423 | } | |
424 | /* For other local defs process their uses, amending | |
425 | imports on the way. */ | |
3cba5cd6 | 426 | else |
16b013c9 RB |
427 | { |
428 | tree ssa[3]; | |
3cba5cd6 AM |
429 | unsigned lim = gimple_range_ssa_names (ssa, 3, def_stmt); |
430 | for (unsigned j = 0; j < lim; ++j) | |
16b013c9 RB |
431 | { |
432 | tree rhs = ssa[j]; | |
433 | if (rhs | |
16b013c9 RB |
434 | && bitmap_set_bit (m_imports, |
435 | SSA_NAME_VERSION (rhs))) | |
436 | { | |
437 | new_imports.safe_push (SSA_NAME_VERSION (rhs)); | |
438 | worklist.safe_push (rhs); | |
439 | } | |
440 | } | |
441 | } | |
d86d81a4 | 442 | } |
2e96b5f1 | 443 | } |
d86d81a4 RB |
444 | if (!bitmap_empty_p (new_interesting) |
445 | || !interesting_phis.is_empty ()) | |
98b212c1 | 446 | { |
16b013c9 RB |
447 | auto_vec<int, 4> unwind (interesting_phis.length ()); |
448 | auto_vec<int, 4> imports_unwind (interesting_phis.length ()); | |
98b212c1 AH |
449 | edge_iterator iter; |
450 | edge e; | |
451 | FOR_EACH_EDGE (e, iter, bb->preds) | |
d86d81a4 RB |
452 | { |
453 | if (e->flags & EDGE_ABNORMAL | |
454 | // This is like path_crosses_loops in profitable_path_p but | |
455 | // more restrictive to avoid peeling off loop iterations (see | |
456 | // tree-ssa/pr14341.c for an example). | |
457 | // ??? Note this restriction only applied when visiting an | |
458 | // interesting PHI with the former resolve_phi. | |
459 | || (!interesting_phis.is_empty () | |
460 | && m_path[0]->loop_father != e->src->loop_father)) | |
461 | continue; | |
462 | for (gphi *phi : interesting_phis) | |
463 | { | |
464 | tree def = PHI_ARG_DEF_FROM_EDGE (phi, e); | |
465 | if (TREE_CODE (def) == SSA_NAME) | |
16b013c9 RB |
466 | { |
467 | int ver = SSA_NAME_VERSION (def); | |
468 | if (bitmap_set_bit (new_interesting, ver)) | |
469 | { | |
470 | if (bitmap_set_bit (m_imports, ver)) | |
471 | imports_unwind.quick_push (ver); | |
472 | unwind.quick_push (ver); | |
473 | } | |
474 | } | |
d86d81a4 | 475 | } |
bac07a1d RB |
476 | find_paths_to_names (e->src, new_interesting, overall_paths, |
477 | profit); | |
16b013c9 RB |
478 | // Restore new_interesting. |
479 | for (int def : unwind) | |
480 | bitmap_clear_bit (new_interesting, def); | |
d86d81a4 | 481 | unwind.truncate (0); |
16b013c9 RB |
482 | // Restore and m_imports. |
483 | for (int def : imports_unwind) | |
484 | bitmap_clear_bit (m_imports, def); | |
485 | imports_unwind.truncate (0); | |
d86d81a4 | 486 | } |
98b212c1 | 487 | } |
16b013c9 RB |
488 | /* m_imports tracks all interesting names on the path, so when |
489 | backtracking we have to restore it. */ | |
490 | for (int j : new_imports) | |
491 | bitmap_clear_bit (m_imports, j); | |
2e96b5f1 | 492 | } |
409978d5 RB |
493 | else if (dump_file && (dump_flags & TDF_DETAILS)) |
494 | fprintf (dump_file, " FAIL: Search space limit %d reached.\n", | |
495 | param_max_jump_thread_paths); | |
2e96b5f1 | 496 | |
98b212c1 | 497 | // Reset things to their original state. |
2e96b5f1 AH |
498 | m_path.pop (); |
499 | m_visited_bbs.remove (bb); | |
2e96b5f1 AH |
500 | } |
501 | ||
34cd97ff AH |
502 | // Search backwards from BB looking for paths where the final |
503 | // conditional maybe threaded to a successor block. Record such paths | |
504 | // for jump threading. | |
0732f75f | 505 | |
34cd97ff AH |
506 | void |
507 | back_threader::maybe_thread_block (basic_block bb) | |
0732f75f | 508 | { |
bac07a1d RB |
509 | if (EDGE_COUNT (bb->succs) <= 1) |
510 | return; | |
511 | ||
60bf26a4 | 512 | gimple *stmt = *gsi_last_bb (bb); |
34cd97ff AH |
513 | if (!stmt) |
514 | return; | |
0732f75f | 515 | |
34cd97ff | 516 | enum gimple_code code = gimple_code (stmt); |
d45ddc2c RB |
517 | if (code != GIMPLE_SWITCH |
518 | && code != GIMPLE_COND) | |
bac07a1d | 519 | return; |
34cd97ff | 520 | |
bac07a1d RB |
521 | m_last_stmt = stmt; |
522 | m_visited_bbs.empty (); | |
523 | m_path.truncate (0); | |
bac07a1d RB |
524 | |
525 | // We compute imports of the path during discovery starting | |
526 | // just with names used in the conditional. | |
527 | bitmap_clear (m_imports); | |
528 | ssa_op_iter iter; | |
d45ddc2c | 529 | tree name; |
bac07a1d RB |
530 | FOR_EACH_SSA_TREE_OPERAND (name, stmt, iter, SSA_OP_USE) |
531 | { | |
532 | if (!gimple_range_ssa_p (name)) | |
533 | return; | |
534 | bitmap_set_bit (m_imports, SSA_NAME_VERSION (name)); | |
535 | } | |
536 | ||
537 | // Interesting is the set of imports we still not have see | |
538 | // the definition of. So while imports only grow, the | |
539 | // set of interesting defs dwindles and once empty we can | |
540 | // stop searching. | |
541 | auto_bitmap interesting; | |
542 | bitmap_copy (interesting, m_imports); | |
543 | back_threader_profitability profit (m_flags & BT_SPEED, stmt); | |
544 | find_paths_to_names (bb, interesting, 1, profit); | |
34cd97ff AH |
545 | } |
546 | ||
34cd97ff AH |
547 | DEBUG_FUNCTION void |
548 | debug (const vec <basic_block> &path) | |
549 | { | |
550 | dump_path (stderr, path); | |
a2ab1a5a | 551 | fputc ('\n', stderr); |
0732f75f JL |
552 | } |
553 | ||
779275c0 AH |
554 | void |
555 | back_threader::dump (FILE *out) | |
556 | { | |
779275c0 AH |
557 | fprintf (out, "\nCandidates for pre-computation:\n"); |
558 | fprintf (out, "===================================\n"); | |
559 | ||
560 | bitmap_iterator bi; | |
561 | unsigned i; | |
562 | ||
563 | EXECUTE_IF_SET_IN_BITMAP (m_imports, 0, i, bi) | |
564 | { | |
565 | tree name = ssa_name (i); | |
566 | print_generic_expr (out, name, TDF_NONE); | |
567 | fprintf (out, "\n"); | |
568 | } | |
569 | } | |
570 | ||
571 | void | |
572 | back_threader::debug () | |
573 | { | |
574 | dump (stderr); | |
575 | } | |
576 | ||
bac07a1d RB |
577 | /* Examine jump threading path PATH and return TRUE if it is possibly |
578 | profitable to thread it, otherwise return FALSE. If this function | |
579 | returns TRUE profitable_path_p might not be satisfied but when | |
580 | the path is extended it might be. In particular indicate in | |
581 | *LARGE_NON_FSM whether the thread is too large for a non-FSM thread | |
582 | but would be OK if we extend the path to cover the loop backedge. | |
bb5e62d6 | 583 | |
01b50387 AH |
584 | ?? It seems we should be able to loosen some of the restrictions in |
585 | this function after loop optimizations have run. */ | |
69e55442 AH |
586 | |
587 | bool | |
bac07a1d | 588 | back_threader_profitability::possibly_profitable_path_p |
d45ddc2c | 589 | (const vec<basic_block> &m_path, |
bac07a1d | 590 | bool *large_non_fsm) |
bb5e62d6 | 591 | { |
69e55442 | 592 | gcc_checking_assert (!m_path.is_empty ()); |
ad071b2b | 593 | |
69e55442 AH |
594 | /* We can an empty path here (excluding the DEF block) when the |
595 | statement that makes a conditional generate a compile-time | |
596 | constant result is in the same block as the conditional. | |
ad071b2b JL |
597 | |
598 | That's not really a jump threading opportunity, but instead is | |
599 | simple cprop & simplification. We could handle it here if we | |
600 | wanted by wiring up all the incoming edges. If we run this | |
601 | early in IPA, that might be worth doing. For now we just | |
602 | reject that case. */ | |
69e55442 AH |
603 | if (m_path.length () <= 1) |
604 | return false; | |
ad071b2b | 605 | |
bb5e62d6 | 606 | gimple_stmt_iterator gsi; |
cde30fe0 | 607 | loop_p loop = m_path[0]->loop_father; |
bac07a1d RB |
608 | |
609 | // We recompute the following, when we rewrite possibly_profitable_path_p | |
610 | // to work incrementally on added BBs we have to unwind them on backtracking | |
611 | m_n_insns = 0; | |
612 | m_threaded_through_latch = false; | |
613 | m_multiway_branch_in_path = false; | |
614 | m_contains_hot_bb = false; | |
0f0c2cc3 JH |
615 | |
616 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
617 | fprintf (dump_file, "Checking profitability of path (backwards): "); | |
bb5e62d6 JL |
618 | |
619 | /* Count the number of instructions on the path: as these instructions | |
620 | will have to be duplicated, we will not record the path if there | |
621 | are too many instructions on the path. Also check that all the | |
622 | blocks in the path belong to a single loop. */ | |
cde30fe0 | 623 | for (unsigned j = 0; j < m_path.length (); j++) |
bb5e62d6 | 624 | { |
cde30fe0 | 625 | basic_block bb = m_path[j]; |
bb5e62d6 | 626 | |
0f0c2cc3 JH |
627 | if (dump_file && (dump_flags & TDF_DETAILS)) |
628 | fprintf (dump_file, " bb:%i", bb->index); | |
a7753db4 AH |
629 | /* Remember, blocks in the path are stored in opposite order in |
630 | the PATH array. The last entry in the array represents the | |
631 | block with an outgoing edge that we will redirect to the jump | |
632 | threading path. Thus we don't care how many statements are | |
633 | in that block because it will not be copied or whether or not | |
634 | it ends in a multiway branch. */ | |
cde30fe0 | 635 | if (j < m_path.length () - 1) |
bb5e62d6 | 636 | { |
bac07a1d | 637 | int orig_n_insns = m_n_insns; |
bac07a1d RB |
638 | if (!m_contains_hot_bb && m_speed_p) |
639 | m_contains_hot_bb |= optimize_bb_for_speed_p (bb); | |
bb5e62d6 JL |
640 | for (gsi = gsi_after_labels (bb); |
641 | !gsi_end_p (gsi); | |
642 | gsi_next_nondebug (&gsi)) | |
643 | { | |
70a62009 IL |
644 | /* Do not allow OpenACC loop markers and __builtin_constant_p on |
645 | threading paths. The latter is disallowed, because an | |
646 | expression might be constant on two threading paths, and | |
647 | become non-constant (i.e.: phi) when they merge. */ | |
bb5e62d6 | 648 | gimple *stmt = gsi_stmt (gsi); |
70a62009 IL |
649 | if (gimple_call_internal_p (stmt, IFN_UNIQUE) |
650 | || gimple_call_builtin_p (stmt, BUILT_IN_CONSTANT_P)) | |
79db991e AH |
651 | { |
652 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
653 | fputc ('\n', dump_file); | |
654 | return false; | |
655 | } | |
bb5e62d6 JL |
656 | /* Do not count empty statements and labels. */ |
657 | if (gimple_code (stmt) != GIMPLE_NOP | |
bb5e62d6 | 658 | && !is_gimple_debug (stmt)) |
bac07a1d | 659 | m_n_insns += estimate_num_insns (stmt, &eni_size_weights); |
bb5e62d6 | 660 | } |
0f0c2cc3 | 661 | if (dump_file && (dump_flags & TDF_DETAILS)) |
bac07a1d | 662 | fprintf (dump_file, " (%i insns)", m_n_insns-orig_n_insns); |
bb5e62d6 JL |
663 | |
664 | /* We do not look at the block with the threaded branch | |
665 | in this loop. So if any block with a last statement that | |
666 | is a GIMPLE_SWITCH or GIMPLE_GOTO is seen, then we have a | |
667 | multiway branch on our path. | |
668 | ||
669 | The block in PATH[0] is special, it's the block were we're | |
670 | going to be able to eliminate its branch. */ | |
bac07a1d | 671 | if (j > 0) |
bb5e62d6 | 672 | { |
60bf26a4 | 673 | gimple *last = *gsi_last_bb (bb); |
bac07a1d RB |
674 | if (last |
675 | && (gimple_code (last) == GIMPLE_SWITCH | |
676 | || gimple_code (last) == GIMPLE_GOTO)) | |
677 | m_multiway_branch_in_path = true; | |
bb5e62d6 JL |
678 | } |
679 | } | |
680 | ||
681 | /* Note if we thread through the latch, we will want to include | |
682 | the last entry in the array when determining if we thread | |
683 | through the loop latch. */ | |
684 | if (loop->latch == bb) | |
01b50387 | 685 | { |
bac07a1d | 686 | m_threaded_through_latch = true; |
01b50387 AH |
687 | if (dump_file && (dump_flags & TDF_DETAILS)) |
688 | fprintf (dump_file, " (latch)"); | |
689 | } | |
bb5e62d6 JL |
690 | } |
691 | ||
692 | /* We are going to remove the control statement at the end of the | |
693 | last block in the threading path. So don't count it against our | |
694 | statement count. */ | |
bac07a1d | 695 | m_n_insns -= m_exit_jump_benefit; |
0f0c2cc3 JH |
696 | |
697 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
698 | fprintf (dump_file, "\n Control statement insns: %i\n" | |
699 | " Overall: %i insns\n", | |
bac07a1d | 700 | m_exit_jump_benefit, m_n_insns); |
bb5e62d6 | 701 | |
0f0c2cc3 JH |
702 | /* Threading is profitable if the path duplicated is hot but also |
703 | in a case we separate cold path from hot path and permit optimization | |
704 | of the hot path later. Be on the agressive side here. In some testcases, | |
705 | as in PR 78407 this leads to noticeable improvements. */ | |
bac07a1d | 706 | if (m_speed_p) |
27bddc4a | 707 | { |
bac07a1d | 708 | if (m_n_insns >= param_max_fsm_thread_path_insns) |
27bddc4a JH |
709 | { |
710 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
c7a669af | 711 | fprintf (dump_file, " FAIL: Jump-thread path not considered: " |
27bddc4a JH |
712 | "the number of instructions on the path " |
713 | "exceeds PARAM_MAX_FSM_THREAD_PATH_INSNS.\n"); | |
69e55442 | 714 | return false; |
27bddc4a | 715 | } |
bac07a1d RB |
716 | edge entry = find_edge (m_path[m_path.length () - 1], |
717 | m_path[m_path.length () - 2]); | |
718 | if (probably_never_executed_edge_p (cfun, entry)) | |
b9da6864 RB |
719 | { |
720 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
721 | fprintf (dump_file, " FAIL: Jump-thread path not considered: " | |
bac07a1d | 722 | "path entry is probably never executed.\n"); |
b9da6864 RB |
723 | return false; |
724 | } | |
bac07a1d RB |
725 | } |
726 | else if (m_n_insns > 1) | |
727 | { | |
728 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
729 | fprintf (dump_file, " FAIL: Jump-thread path not considered: " | |
730 | "duplication of %i insns is needed and optimizing for size.\n", | |
731 | m_n_insns); | |
732 | return false; | |
733 | } | |
734 | ||
735 | /* The generic copier used by the backthreader does not re-use an | |
736 | existing threading path to reduce code duplication. So for that | |
737 | case, drastically reduce the number of statements we are allowed | |
738 | to copy. We don't know yet whether we will thread through the latch | |
739 | so we have to be permissive and continue threading, but indicate | |
740 | to the caller the thread, if final, wouldn't be profitable. */ | |
741 | if ((!m_threaded_multiway_branch | |
742 | || !loop->latch | |
743 | || loop->latch->index == EXIT_BLOCK) | |
744 | && (m_n_insns * param_fsm_scale_path_stmts | |
745 | >= param_max_jump_thread_duplication_stmts)) | |
746 | { | |
747 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
748 | fprintf (dump_file, | |
749 | " FAIL: Did not thread around loop and would copy too " | |
750 | "many statements.\n"); | |
751 | return false; | |
752 | } | |
753 | *large_non_fsm = (!(m_threaded_through_latch && m_threaded_multiway_branch) | |
754 | && (m_n_insns * param_fsm_scale_path_stmts | |
755 | >= param_max_jump_thread_duplication_stmts)); | |
756 | ||
79db991e AH |
757 | if (dump_file && (dump_flags & TDF_DETAILS)) |
758 | fputc ('\n', dump_file); | |
bac07a1d RB |
759 | return true; |
760 | } | |
761 | ||
762 | /* Examine jump threading path PATH and return TRUE if it is profitable to | |
763 | thread it, otherwise return FALSE. | |
764 | ||
765 | The taken edge out of the path is TAKEN_EDGE. | |
766 | ||
767 | CREATES_IRREDUCIBLE_LOOP is set to TRUE if threading this path | |
768 | would create an irreducible loop. | |
769 | ||
770 | ?? It seems we should be able to loosen some of the restrictions in | |
771 | this function after loop optimizations have run. */ | |
772 | ||
773 | bool | |
774 | back_threader_profitability::profitable_path_p (const vec<basic_block> &m_path, | |
775 | edge taken_edge, | |
776 | bool *creates_irreducible_loop) | |
777 | { | |
778 | // We can assume that possibly_profitable_path_p holds here | |
779 | ||
780 | loop_p loop = m_path[0]->loop_father; | |
781 | ||
782 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
783 | fprintf (dump_file, "Checking profitability of path (backwards): "); | |
784 | ||
785 | /* If this path threaded through the loop latch back into the | |
786 | same loop and the destination does not dominate the loop | |
787 | latch, then this thread would create an irreducible loop. */ | |
788 | *creates_irreducible_loop = false; | |
789 | if (m_threaded_through_latch | |
790 | && loop == taken_edge->dest->loop_father | |
791 | && (determine_bb_domination_status (loop, taken_edge->dest) | |
792 | == DOMST_NONDOMINATING)) | |
793 | *creates_irreducible_loop = true; | |
794 | ||
795 | /* Threading is profitable if the path duplicated is hot but also | |
796 | in a case we separate cold path from hot path and permit optimization | |
797 | of the hot path later. Be on the agressive side here. In some testcases, | |
798 | as in PR 78407 this leads to noticeable improvements. */ | |
799 | if (m_speed_p | |
800 | && (optimize_edge_for_speed_p (taken_edge) || m_contains_hot_bb)) | |
801 | { | |
802 | if (probably_never_executed_edge_p (cfun, taken_edge)) | |
49ba4fde RB |
803 | { |
804 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
805 | fprintf (dump_file, " FAIL: Jump-thread path not considered: " | |
bac07a1d | 806 | "path leads to probably never executed edge.\n"); |
49ba4fde RB |
807 | return false; |
808 | } | |
27bddc4a | 809 | } |
bac07a1d | 810 | else if (m_n_insns > 1) |
bb5e62d6 JL |
811 | { |
812 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
c7a669af | 813 | fprintf (dump_file, " FAIL: Jump-thread path not considered: " |
27bddc4a | 814 | "duplication of %i insns is needed and optimizing for size.\n", |
bac07a1d | 815 | m_n_insns); |
69e55442 | 816 | return false; |
bb5e62d6 JL |
817 | } |
818 | ||
819 | /* We avoid creating irreducible inner loops unless we thread through | |
820 | a multiway branch, in which case we have deemed it worth losing | |
821 | other loop optimizations later. | |
822 | ||
7c9f20fc RB |
823 | We also consider it worth creating an irreducible inner loop after |
824 | loop optimizations if the number of copied statement is low. */ | |
bac07a1d | 825 | if (!m_threaded_multiway_branch |
69e55442 | 826 | && *creates_irreducible_loop |
7c9f20fc RB |
827 | && (!(cfun->curr_properties & PROP_loop_opts_done) |
828 | || (m_n_insns * param_fsm_scale_path_stmts | |
829 | >= param_max_jump_thread_duplication_stmts))) | |
bb5e62d6 JL |
830 | { |
831 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
832 | fprintf (dump_file, | |
7c9f20fc RB |
833 | " FAIL: Would create irreducible loop early without " |
834 | "threading multiway branch.\n"); | |
bac07a1d | 835 | /* We compute creates_irreducible_loop only late. */ |
69e55442 | 836 | return false; |
bb5e62d6 JL |
837 | } |
838 | ||
c7a669af AH |
839 | /* The generic copier used by the backthreader does not re-use an |
840 | existing threading path to reduce code duplication. So for that | |
841 | case, drastically reduce the number of statements we are allowed | |
842 | to copy. */ | |
bac07a1d RB |
843 | if (!(m_threaded_through_latch && m_threaded_multiway_branch) |
844 | && (m_n_insns * param_fsm_scale_path_stmts | |
028d4092 | 845 | >= param_max_jump_thread_duplication_stmts)) |
bb5e62d6 JL |
846 | { |
847 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
848 | fprintf (dump_file, | |
c7a669af | 849 | " FAIL: Did not thread around loop and would copy too " |
bb5e62d6 | 850 | "many statements.\n"); |
69e55442 | 851 | return false; |
bb5e62d6 JL |
852 | } |
853 | ||
854 | /* When there is a multi-way branch on the path, then threading can | |
855 | explode the CFG due to duplicating the edges for that multi-way | |
856 | branch. So like above, only allow a multi-way branch on the path | |
857 | if we actually thread a multi-way branch. */ | |
bac07a1d | 858 | if (!m_threaded_multiway_branch && m_multiway_branch_in_path) |
bb5e62d6 JL |
859 | { |
860 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
861 | fprintf (dump_file, | |
c7a669af | 862 | " FAIL: Thread through multiway branch without threading " |
bb5e62d6 | 863 | "a multiway branch.\n"); |
69e55442 | 864 | return false; |
bb5e62d6 | 865 | } |
01b50387 AH |
866 | |
867 | /* Threading through an empty latch would cause code to be added to | |
868 | the latch. This could alter the loop form sufficiently to cause | |
869 | loop optimizations to fail. Disable these threads until after | |
870 | loop optimizations have run. */ | |
bac07a1d | 871 | if ((m_threaded_through_latch || taken_edge->dest == loop->latch) |
01b50387 AH |
872 | && !(cfun->curr_properties & PROP_loop_opts_done) |
873 | && empty_block_p (loop->latch)) | |
874 | { | |
875 | if (dump_file && (dump_flags & TDF_DETAILS)) | |
876 | fprintf (dump_file, | |
bac07a1d RB |
877 | " FAIL: Thread through latch before loop opts would create " |
878 | "non-empty latch\n"); | |
01b50387 | 879 | return false; |
01b50387 | 880 | } |
79db991e AH |
881 | if (dump_file && (dump_flags & TDF_DETAILS)) |
882 | fputc ('\n', dump_file); | |
69e55442 AH |
883 | return true; |
884 | } | |
885 | ||
bac07a1d | 886 | |
cde30fe0 AH |
887 | /* The current path PATH is a vector of blocks forming a jump threading |
888 | path in reverse order. TAKEN_EDGE is the edge taken from path[0]. | |
081fdda6 | 889 | |
cde30fe0 | 890 | Convert the current path into the form used by register_jump_thread and |
69e55442 | 891 | register it. |
081fdda6 | 892 | |
69e55442 AH |
893 | Return TRUE if successful or FALSE otherwise. */ |
894 | ||
895 | bool | |
896 | back_threader_registry::register_path (const vec<basic_block> &m_path, | |
897 | edge taken_edge) | |
081fdda6 | 898 | { |
7f6c2258 | 899 | vec<jump_thread_edge *> *jump_thread_path = allocate_thread_path (); |
081fdda6 | 900 | |
c7a669af AH |
901 | // The generic copier ignores the edge type. We can build the |
902 | // thread edges with any type. | |
cde30fe0 | 903 | for (unsigned int j = 0; j + 1 < m_path.length (); j++) |
081fdda6 | 904 | { |
cde30fe0 AH |
905 | basic_block bb1 = m_path[m_path.length () - j - 1]; |
906 | basic_block bb2 = m_path[m_path.length () - j - 2]; | |
8af01c66 | 907 | |
081fdda6 JL |
908 | edge e = find_edge (bb1, bb2); |
909 | gcc_assert (e); | |
7f6c2258 | 910 | push_edge (jump_thread_path, e, EDGE_COPY_SRC_BLOCK); |
081fdda6 JL |
911 | } |
912 | ||
7f6c2258 | 913 | push_edge (jump_thread_path, taken_edge, EDGE_NO_COPY_SRC_BLOCK); |
6ca94826 | 914 | return register_jump_thread (jump_thread_path); |
081fdda6 JL |
915 | } |
916 | ||
4e0f56d7 | 917 | // Thread all suitable paths in the current function. |
5d4d64fa | 918 | // |
4e0f56d7 | 919 | // Return TODO_flags. |
8b2ef235 | 920 | |
4e0f56d7 AH |
921 | unsigned int |
922 | back_threader::thread_blocks () | |
8b2ef235 | 923 | { |
8b2ef235 | 924 | basic_block bb; |
4e0f56d7 AH |
925 | FOR_EACH_BB_FN (bb, m_fun) |
926 | if (EDGE_COUNT (bb->succs) > 1) | |
927 | maybe_thread_block (bb); | |
bb7ebad1 | 928 | |
4e0f56d7 | 929 | bool changed = m_registry.thread_through_all_blocks (true); |
5d4d64fa | 930 | |
4e0f56d7 AH |
931 | if (m_flags & BT_SPEED) |
932 | return changed ? TODO_cleanup_cfg : 0; | |
5d4d64fa | 933 | |
4e0f56d7 | 934 | return false; |
8b2ef235 | 935 | } |
b720e919 JH |
936 | |
937 | namespace { | |
938 | ||
939 | const pass_data pass_data_early_thread_jumps = | |
940 | { | |
941 | GIMPLE_PASS, | |
942 | "ethread", | |
943 | OPTGROUP_NONE, | |
944 | TV_TREE_SSA_THREAD_JUMPS, | |
945 | ( PROP_cfg | PROP_ssa ), | |
946 | 0, | |
947 | 0, | |
948 | 0, | |
949 | ( TODO_cleanup_cfg | TODO_update_ssa ), | |
950 | }; | |
951 | ||
5d4d64fa AH |
952 | const pass_data pass_data_thread_jumps = |
953 | { | |
954 | GIMPLE_PASS, | |
955 | "thread", | |
956 | OPTGROUP_NONE, | |
957 | TV_TREE_SSA_THREAD_JUMPS, | |
958 | ( PROP_cfg | PROP_ssa ), | |
959 | 0, | |
960 | 0, | |
961 | 0, | |
962 | TODO_update_ssa, | |
963 | }; | |
964 | ||
965 | const pass_data pass_data_thread_jumps_full = | |
966 | { | |
967 | GIMPLE_PASS, | |
4b3a325f | 968 | "threadfull", |
5d4d64fa AH |
969 | OPTGROUP_NONE, |
970 | TV_TREE_SSA_THREAD_JUMPS, | |
971 | ( PROP_cfg | PROP_ssa ), | |
972 | 0, | |
973 | 0, | |
974 | 0, | |
975 | TODO_update_ssa, | |
976 | }; | |
977 | ||
978 | // Early jump threading pass optimizing for size. | |
b720e919 JH |
979 | class pass_early_thread_jumps : public gimple_opt_pass |
980 | { | |
981 | public: | |
982 | pass_early_thread_jumps (gcc::context *ctxt) | |
983 | : gimple_opt_pass (pass_data_early_thread_jumps, ctxt) | |
984 | {} | |
985 | ||
5d4d64fa AH |
986 | opt_pass * clone () override |
987 | { | |
988 | return new pass_early_thread_jumps (m_ctxt); | |
989 | } | |
bc5baac5 AH |
990 | void set_pass_param (unsigned int, bool param) override |
991 | { | |
992 | m_first = param; | |
993 | } | |
5d4d64fa AH |
994 | bool gate (function *) override |
995 | { | |
996 | return flag_thread_jumps; | |
997 | } | |
998 | unsigned int execute (function *fun) override | |
999 | { | |
bc5baac5 | 1000 | back_threader threader (fun, BT_NONE, m_first); |
4e0f56d7 | 1001 | return threader.thread_blocks (); |
5d4d64fa | 1002 | } |
bc5baac5 AH |
1003 | private: |
1004 | bool m_first; | |
b720e919 JH |
1005 | }; |
1006 | ||
5d4d64fa AH |
1007 | // Jump threading pass without resolving of unknown SSAs. |
1008 | class pass_thread_jumps : public gimple_opt_pass | |
b720e919 | 1009 | { |
5d4d64fa AH |
1010 | public: |
1011 | pass_thread_jumps (gcc::context *ctxt) | |
1012 | : gimple_opt_pass (pass_data_thread_jumps, ctxt) | |
1013 | {} | |
1014 | opt_pass * clone (void) override | |
1015 | { | |
1016 | return new pass_thread_jumps (m_ctxt); | |
1017 | } | |
bc5baac5 AH |
1018 | void set_pass_param (unsigned int, bool param) override |
1019 | { | |
1020 | m_first = param; | |
1021 | } | |
5d4d64fa AH |
1022 | bool gate (function *) override |
1023 | { | |
1024 | return flag_thread_jumps && flag_expensive_optimizations; | |
1025 | } | |
1026 | unsigned int execute (function *fun) override | |
1027 | { | |
bc5baac5 | 1028 | back_threader threader (fun, BT_SPEED, m_first); |
4e0f56d7 | 1029 | return threader.thread_blocks (); |
5d4d64fa | 1030 | } |
bc5baac5 AH |
1031 | private: |
1032 | bool m_first; | |
5d4d64fa | 1033 | }; |
b720e919 | 1034 | |
5d4d64fa AH |
1035 | // Jump threading pass that fully resolves unknown SSAs. |
1036 | class pass_thread_jumps_full : public gimple_opt_pass | |
b720e919 | 1037 | { |
5d4d64fa AH |
1038 | public: |
1039 | pass_thread_jumps_full (gcc::context *ctxt) | |
1040 | : gimple_opt_pass (pass_data_thread_jumps_full, ctxt) | |
1041 | {} | |
1042 | opt_pass * clone (void) override | |
1043 | { | |
dece6ae7 | 1044 | return new pass_thread_jumps_full (m_ctxt); |
5d4d64fa | 1045 | } |
bc5baac5 AH |
1046 | void set_pass_param (unsigned int, bool param) override |
1047 | { | |
1048 | m_first = param; | |
1049 | } | |
5d4d64fa AH |
1050 | bool gate (function *) override |
1051 | { | |
1052 | return flag_thread_jumps && flag_expensive_optimizations; | |
1053 | } | |
1054 | unsigned int execute (function *fun) override | |
1055 | { | |
bc5baac5 | 1056 | back_threader threader (fun, BT_SPEED | BT_RESOLVE, m_first); |
4e0f56d7 | 1057 | return threader.thread_blocks (); |
5d4d64fa | 1058 | } |
bc5baac5 AH |
1059 | private: |
1060 | bool m_first; | |
5d4d64fa | 1061 | }; |
108fdd6d | 1062 | |
5d4d64fa | 1063 | } // namespace { |
108fdd6d | 1064 | |
5d4d64fa AH |
1065 | gimple_opt_pass * |
1066 | make_pass_thread_jumps (gcc::context *ctxt) | |
1067 | { | |
1068 | return new pass_thread_jumps (ctxt); | |
b720e919 JH |
1069 | } |
1070 | ||
5d4d64fa AH |
1071 | gimple_opt_pass * |
1072 | make_pass_thread_jumps_full (gcc::context *ctxt) | |
1073 | { | |
1074 | return new pass_thread_jumps_full (ctxt); | |
b720e919 JH |
1075 | } |
1076 | ||
1077 | gimple_opt_pass * | |
1078 | make_pass_early_thread_jumps (gcc::context *ctxt) | |
1079 | { | |
1080 | return new pass_early_thread_jumps (ctxt); | |
1081 | } |