| 1 | /* |
| 2 | * Copyright (C) 2011-2017 Apple Inc. All rights reserved. |
| 3 | * |
| 4 | * Redistribution and use in source and binary forms, with or without |
| 5 | * modification, are permitted provided that the following conditions |
| 6 | * are met: |
| 7 | * 1. Redistributions of source code must retain the above copyright |
| 8 | * notice, this list of conditions and the following disclaimer. |
| 9 | * 2. Redistributions in binary form must reproduce the above copyright |
| 10 | * notice, this list of conditions and the following disclaimer in the |
| 11 | * documentation and/or other materials provided with the distribution. |
| 12 | * |
| 13 | * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY |
| 14 | * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 15 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
| 16 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL APPLE INC. OR |
| 17 | * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, |
| 18 | * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
| 19 | * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR |
| 20 | * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY |
| 21 | * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 22 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 23 | * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 24 | */ |
| 25 | |
| 26 | #pragma once |
| 27 | |
| 28 | #include <wtf/CommaPrinter.h> |
| 29 | #include <wtf/FastBitVector.h> |
| 30 | #include <wtf/GraphNodeWorklist.h> |
| 31 | |
| 32 | namespace WTF { |
| 33 | |
| 34 | // This is a utility for finding the dominators of a graph. Dominators are almost universally used |
| 35 | // for control flow graph analysis, so this code will refer to the graph's "nodes" as "blocks". In |
| 36 | // that regard this code is kind of specialized for the various JSC compilers, but you could use it |
| 37 | // for non-compiler things if you are OK with referring to your "nodes" as "blocks". |
| 38 | |
| 39 | template<typename Graph> |
| 40 | class Dominators { |
| 41 | public: |
| 42 | using List = typename Graph::List; |
| 43 | |
| 44 | Dominators(Graph& graph, bool selfCheck = false) |
| 45 | : m_graph(graph) |
| 46 | , m_data(graph.template newMap<BlockData>()) |
| 47 | { |
| 48 | LengauerTarjan lengauerTarjan(m_graph); |
| 49 | lengauerTarjan.compute(); |
| 50 | |
| 51 | // From here we want to build a spanning tree with both upward and downward links and we want |
| 52 | // to do a search over this tree to compute pre and post numbers that can be used for dominance |
| 53 | // tests. |
| 54 | |
| 55 | for (unsigned blockIndex = m_graph.numNodes(); blockIndex--;) { |
| 56 | typename Graph::Node block = m_graph.node(blockIndex); |
| 57 | if (!block) |
| 58 | continue; |
| 59 | |
| 60 | typename Graph::Node idomBlock = lengauerTarjan.immediateDominator(block); |
| 61 | m_data[block].idomParent = idomBlock; |
| 62 | if (idomBlock) |
| 63 | m_data[idomBlock].idomKids.append(block); |
| 64 | } |
| 65 | |
| 66 | unsigned nextPreNumber = 0; |
| 67 | unsigned nextPostNumber = 0; |
| 68 | |
| 69 | // Plain stack-based worklist because we are guaranteed to see each block exactly once anyway. |
| 70 | Vector<GraphNodeWithOrder<typename Graph::Node>> worklist; |
| 71 | worklist.append(GraphNodeWithOrder<typename Graph::Node>(m_graph.root(), GraphVisitOrder::Pre)); |
| 72 | while (!worklist.isEmpty()) { |
| 73 | GraphNodeWithOrder<typename Graph::Node> item = worklist.takeLast(); |
| 74 | switch (item.order) { |
| 75 | case GraphVisitOrder::Pre: |
| 76 | m_data[item.node].preNumber = nextPreNumber++; |
| 77 | worklist.append(GraphNodeWithOrder<typename Graph::Node>(item.node, GraphVisitOrder::Post)); |
| 78 | for (typename Graph::Node kid : m_data[item.node].idomKids) |
| 79 | worklist.append(GraphNodeWithOrder<typename Graph::Node>(kid, GraphVisitOrder::Pre)); |
| 80 | break; |
| 81 | case GraphVisitOrder::Post: |
| 82 | m_data[item.node].postNumber = nextPostNumber++; |
| 83 | break; |
| 84 | } |
| 85 | } |
| 86 | |
| 87 | if (selfCheck) { |
| 88 | // Check our dominator calculation: |
| 89 | // 1) Check that our range-based ancestry test is the same as a naive ancestry test. |
| 90 | // 2) Check that our notion of who dominates whom is identical to a naive (not |
| 91 | // Lengauer-Tarjan) dominator calculation. |
| 92 | |
| 93 | ValidationContext context(m_graph, *this); |
| 94 | |
| 95 | for (unsigned fromBlockIndex = m_graph.numNodes(); fromBlockIndex--;) { |
| 96 | typename Graph::Node fromBlock = m_graph.node(fromBlockIndex); |
| 97 | if (!fromBlock || m_data[fromBlock].preNumber == UINT_MAX) |
| 98 | continue; |
| 99 | for (unsigned toBlockIndex = m_graph.numNodes(); toBlockIndex--;) { |
| 100 | typename Graph::Node toBlock = m_graph.node(toBlockIndex); |
| 101 | if (!toBlock || m_data[toBlock].preNumber == UINT_MAX) |
| 102 | continue; |
| 103 | |
| 104 | if (dominates(fromBlock, toBlock) != naiveDominates(fromBlock, toBlock)) |
| 105 | context.reportError(fromBlock, toBlock, "Range-based domination check is broken" ); |
| 106 | if (dominates(fromBlock, toBlock) != context.naiveDominators.dominates(fromBlock, toBlock)) |
| 107 | context.reportError(fromBlock, toBlock, "Lengauer-Tarjan domination is broken" ); |
| 108 | } |
| 109 | } |
| 110 | |
| 111 | context.handleErrors(); |
| 112 | } |
| 113 | } |
| 114 | |
| 115 | bool strictlyDominates(typename Graph::Node from, typename Graph::Node to) const |
| 116 | { |
| 117 | return m_data[to].preNumber > m_data[from].preNumber |
| 118 | && m_data[to].postNumber < m_data[from].postNumber; |
| 119 | } |
| 120 | |
| 121 | bool dominates(typename Graph::Node from, typename Graph::Node to) const |
| 122 | { |
| 123 | return from == to || strictlyDominates(from, to); |
| 124 | } |
| 125 | |
| 126 | // Returns the immediate dominator of this block. Returns null for the root block. |
| 127 | typename Graph::Node idom(typename Graph::Node block) const |
| 128 | { |
| 129 | return m_data[block].idomParent; |
| 130 | } |
| 131 | |
| 132 | template<typename Functor> |
| 133 | void forAllStrictDominatorsOf(typename Graph::Node to, const Functor& functor) const |
| 134 | { |
| 135 | for (typename Graph::Node block = m_data[to].idomParent; block; block = m_data[block].idomParent) |
| 136 | functor(block); |
| 137 | } |
| 138 | |
| 139 | // Note: This will visit the dominators starting with the 'to' node and moving up the idom tree |
| 140 | // until it gets to the root. Some clients of this function, like B3::moveConstants(), rely on this |
| 141 | // order. |
| 142 | template<typename Functor> |
| 143 | void forAllDominatorsOf(typename Graph::Node to, const Functor& functor) const |
| 144 | { |
| 145 | for (typename Graph::Node block = to; block; block = m_data[block].idomParent) |
| 146 | functor(block); |
| 147 | } |
| 148 | |
| 149 | template<typename Functor> |
| 150 | void forAllBlocksStrictlyDominatedBy(typename Graph::Node from, const Functor& functor) const |
| 151 | { |
| 152 | Vector<typename Graph::Node, 16> worklist; |
| 153 | worklist.appendVector(m_data[from].idomKids); |
| 154 | while (!worklist.isEmpty()) { |
| 155 | typename Graph::Node block = worklist.takeLast(); |
| 156 | functor(block); |
| 157 | worklist.appendVector(m_data[block].idomKids); |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | template<typename Functor> |
| 162 | void forAllBlocksDominatedBy(typename Graph::Node from, const Functor& functor) const |
| 163 | { |
| 164 | Vector<typename Graph::Node, 16> worklist; |
| 165 | worklist.append(from); |
| 166 | while (!worklist.isEmpty()) { |
| 167 | typename Graph::Node block = worklist.takeLast(); |
| 168 | functor(block); |
| 169 | worklist.appendVector(m_data[block].idomKids); |
| 170 | } |
| 171 | } |
| 172 | |
| 173 | typename Graph::Set strictDominatorsOf(typename Graph::Node to) const |
| 174 | { |
| 175 | typename Graph::Set result; |
| 176 | forAllStrictDominatorsOf( |
| 177 | to, |
| 178 | [&] (typename Graph::Node node) { |
| 179 | result.add(node); |
| 180 | }); |
| 181 | return result; |
| 182 | } |
| 183 | |
| 184 | typename Graph::Set dominatorsOf(typename Graph::Node to) const |
| 185 | { |
| 186 | typename Graph::Set result; |
| 187 | forAllDominatorsOf( |
| 188 | to, |
| 189 | [&] (typename Graph::Node node) { |
| 190 | result.add(node); |
| 191 | }); |
| 192 | return result; |
| 193 | } |
| 194 | |
| 195 | typename Graph::Set blocksStrictlyDominatedBy(typename Graph::Node from) const |
| 196 | { |
| 197 | typename Graph::Set result; |
| 198 | forAllBlocksStrictlyDominatedBy( |
| 199 | from, |
| 200 | [&] (typename Graph::Node node) { |
| 201 | result.add(node); |
| 202 | }); |
| 203 | return result; |
| 204 | } |
| 205 | |
| 206 | typename Graph::Set blocksDominatedBy(typename Graph::Node from) const |
| 207 | { |
| 208 | typename Graph::Set result; |
| 209 | forAllBlocksDominatedBy( |
| 210 | from, |
| 211 | [&] (typename Graph::Node node) { |
| 212 | result.add(node); |
| 213 | }); |
| 214 | return result; |
| 215 | } |
| 216 | |
| 217 | template<typename Functor> |
| 218 | void forAllBlocksInDominanceFrontierOf( |
| 219 | typename Graph::Node from, const Functor& functor) const |
| 220 | { |
| 221 | typename Graph::Set set; |
| 222 | forAllBlocksInDominanceFrontierOfImpl( |
| 223 | from, |
| 224 | [&] (typename Graph::Node block) { |
| 225 | if (set.add(block)) |
| 226 | functor(block); |
| 227 | }); |
| 228 | } |
| 229 | |
| 230 | typename Graph::Set dominanceFrontierOf(typename Graph::Node from) const |
| 231 | { |
| 232 | typename Graph::Set result; |
| 233 | forAllBlocksInDominanceFrontierOf( |
| 234 | from, |
| 235 | [&] (typename Graph::Node node) { |
| 236 | result.add(node); |
| 237 | }); |
| 238 | return result; |
| 239 | } |
| 240 | |
| 241 | template<typename Functor> |
| 242 | void forAllBlocksInIteratedDominanceFrontierOf(const List& from, const Functor& functor) |
| 243 | { |
| 244 | forAllBlocksInPrunedIteratedDominanceFrontierOf( |
| 245 | from, |
| 246 | [&] (typename Graph::Node block) -> bool { |
| 247 | functor(block); |
| 248 | return true; |
| 249 | }); |
| 250 | } |
| 251 | |
| 252 | // This is a close relative of forAllBlocksInIteratedDominanceFrontierOf(), which allows the |
| 253 | // given functor to return false to indicate that we don't wish to consider the given block. |
| 254 | // Useful for computing pruned SSA form. |
| 255 | template<typename Functor> |
| 256 | void forAllBlocksInPrunedIteratedDominanceFrontierOf( |
| 257 | const List& from, const Functor& functor) |
| 258 | { |
| 259 | typename Graph::Set set; |
| 260 | forAllBlocksInIteratedDominanceFrontierOfImpl( |
| 261 | from, |
| 262 | [&] (typename Graph::Node block) -> bool { |
| 263 | if (!set.add(block)) |
| 264 | return false; |
| 265 | return functor(block); |
| 266 | }); |
| 267 | } |
| 268 | |
| 269 | typename Graph::Set iteratedDominanceFrontierOf(const List& from) const |
| 270 | { |
| 271 | typename Graph::Set result; |
| 272 | forAllBlocksInIteratedDominanceFrontierOfImpl( |
| 273 | from, |
| 274 | [&] (typename Graph::Node node) -> bool { |
| 275 | return result.add(node); |
| 276 | }); |
| 277 | return result; |
| 278 | } |
| 279 | |
| 280 | void dump(PrintStream& out) const |
| 281 | { |
| 282 | for (unsigned blockIndex = 0; blockIndex < m_data.size(); ++blockIndex) { |
| 283 | if (m_data[blockIndex].preNumber == UINT_MAX) |
| 284 | continue; |
| 285 | |
| 286 | out.print(" Block #" , blockIndex, ": idom = " , m_graph.dump(m_data[blockIndex].idomParent), ", idomKids = [" ); |
| 287 | CommaPrinter comma; |
| 288 | for (unsigned i = 0; i < m_data[blockIndex].idomKids.size(); ++i) |
| 289 | out.print(comma, m_graph.dump(m_data[blockIndex].idomKids[i])); |
| 290 | out.print("], pre/post = " , m_data[blockIndex].preNumber, "/" , m_data[blockIndex].postNumber, "\n" ); |
| 291 | } |
| 292 | } |
| 293 | |
| 294 | private: |
| 295 | // This implements Lengauer and Tarjan's "A Fast Algorithm for Finding Dominators in a Flowgraph" |
| 296 | // (TOPLAS 1979). It uses the "simple" implementation of LINK and EVAL, which yields an O(n log n) |
| 297 | // solution. The full paper is linked below; this code attempts to closely follow the algorithm as |
| 298 | // it is presented in the paper; in particular sections 3 and 4 as well as appendix B. |
| 299 | // https://www.cs.princeton.edu/courses/archive/fall03/cs528/handouts/a%20fast%20algorithm%20for%20finding.pdf |
| 300 | // |
| 301 | // This code is very subtle. The Lengauer-Tarjan algorithm is incredibly deep to begin with. The |
| 302 | // goal of this code is to follow the code in the paper, however our implementation must deviate |
| 303 | // from the paper when it comes to recursion. The authors had used recursion to implement DFS, and |
| 304 | // also to implement the "simple" EVAL. We convert both of those into worklist-based solutions. |
| 305 | // Finally, once the algorithm gives us immediate dominators, we implement dominance tests by |
| 306 | // walking the dominator tree and computing pre and post numbers. We then use the range inclusion |
| 307 | // check trick that was first discovered by Paul F. Dietz in 1982 in "Maintaining order in a linked |
| 308 | // list" (see http://dl.acm.org/citation.cfm?id=802184). |
| 309 | |
| 310 | class LengauerTarjan { |
| 311 | public: |
| 312 | LengauerTarjan(Graph& graph) |
| 313 | : m_graph(graph) |
| 314 | , m_data(graph.template newMap<BlockData>()) |
| 315 | { |
| 316 | for (unsigned blockIndex = m_graph.numNodes(); blockIndex--;) { |
| 317 | typename Graph::Node block = m_graph.node(blockIndex); |
| 318 | if (!block) |
| 319 | continue; |
| 320 | m_data[block].label = block; |
| 321 | } |
| 322 | } |
| 323 | |
| 324 | void compute() |
| 325 | { |
| 326 | computeDepthFirstPreNumbering(); // Step 1. |
| 327 | computeSemiDominatorsAndImplicitImmediateDominators(); // Steps 2 and 3. |
| 328 | computeExplicitImmediateDominators(); // Step 4. |
| 329 | } |
| 330 | |
| 331 | typename Graph::Node immediateDominator(typename Graph::Node block) |
| 332 | { |
| 333 | return m_data[block].dom; |
| 334 | } |
| 335 | |
| 336 | private: |
| 337 | void computeDepthFirstPreNumbering() |
| 338 | { |
| 339 | // Use a block worklist that also tracks the index inside the successor list. This is |
| 340 | // necessary for ensuring that we don't attempt to visit a successor until the previous |
| 341 | // successors that we had visited are fully processed. This ends up being revealed in the |
| 342 | // output of this method because the first time we see an edge to a block, we set the |
| 343 | // block's parent. So, if we have: |
| 344 | // |
| 345 | // A -> B |
| 346 | // A -> C |
| 347 | // B -> C |
| 348 | // |
| 349 | // And we're processing A, then we want to ensure that if we see A->B first (and hence set |
| 350 | // B's prenumber before we set C's) then we also end up setting C's parent to B by virtue |
| 351 | // of not noticing A->C until we're done processing B. |
| 352 | |
| 353 | ExtendedGraphNodeWorklist<typename Graph::Node, unsigned, typename Graph::Set> worklist; |
| 354 | worklist.push(m_graph.root(), 0); |
| 355 | |
| 356 | while (GraphNodeWith<typename Graph::Node, unsigned> item = worklist.pop()) { |
| 357 | typename Graph::Node block = item.node; |
| 358 | unsigned successorIndex = item.data; |
| 359 | |
| 360 | // We initially push with successorIndex = 0 regardless of whether or not we have any |
| 361 | // successors. This is so that we can assign our prenumber. Subsequently we get pushed |
| 362 | // with higher successorIndex values, but only if they are in range. |
| 363 | ASSERT(!successorIndex || successorIndex < m_graph.successors(block).size()); |
| 364 | |
| 365 | if (!successorIndex) { |
| 366 | m_data[block].semiNumber = m_blockByPreNumber.size(); |
| 367 | m_blockByPreNumber.append(block); |
| 368 | } |
| 369 | |
| 370 | if (successorIndex < m_graph.successors(block).size()) { |
| 371 | unsigned nextSuccessorIndex = successorIndex + 1; |
| 372 | if (nextSuccessorIndex < m_graph.successors(block).size()) |
| 373 | worklist.forcePush(block, nextSuccessorIndex); |
| 374 | |
| 375 | typename Graph::Node successorBlock = m_graph.successors(block)[successorIndex]; |
| 376 | if (worklist.push(successorBlock, 0)) |
| 377 | m_data[successorBlock].parent = block; |
| 378 | } |
| 379 | } |
| 380 | } |
| 381 | |
| 382 | void computeSemiDominatorsAndImplicitImmediateDominators() |
| 383 | { |
| 384 | for (unsigned currentPreNumber = m_blockByPreNumber.size(); currentPreNumber-- > 1;) { |
| 385 | typename Graph::Node block = m_blockByPreNumber[currentPreNumber]; |
| 386 | BlockData& blockData = m_data[block]; |
| 387 | |
| 388 | // Step 2: |
| 389 | for (typename Graph::Node predecessorBlock : m_graph.predecessors(block)) { |
| 390 | typename Graph::Node intermediateBlock = eval(predecessorBlock); |
| 391 | blockData.semiNumber = std::min( |
| 392 | m_data[intermediateBlock].semiNumber, blockData.semiNumber); |
| 393 | } |
| 394 | unsigned bucketPreNumber = blockData.semiNumber; |
| 395 | ASSERT(bucketPreNumber <= currentPreNumber); |
| 396 | m_data[m_blockByPreNumber[bucketPreNumber]].bucket.append(block); |
| 397 | link(blockData.parent, block); |
| 398 | |
| 399 | // Step 3: |
| 400 | for (typename Graph::Node semiDominee : m_data[blockData.parent].bucket) { |
| 401 | typename Graph::Node possibleDominator = eval(semiDominee); |
| 402 | BlockData& semiDomineeData = m_data[semiDominee]; |
| 403 | ASSERT(m_blockByPreNumber[semiDomineeData.semiNumber] == blockData.parent); |
| 404 | BlockData& possibleDominatorData = m_data[possibleDominator]; |
| 405 | if (possibleDominatorData.semiNumber < semiDomineeData.semiNumber) |
| 406 | semiDomineeData.dom = possibleDominator; |
| 407 | else |
| 408 | semiDomineeData.dom = blockData.parent; |
| 409 | } |
| 410 | m_data[blockData.parent].bucket.clear(); |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | void computeExplicitImmediateDominators() |
| 415 | { |
| 416 | for (unsigned currentPreNumber = 1; currentPreNumber < m_blockByPreNumber.size(); ++currentPreNumber) { |
| 417 | typename Graph::Node block = m_blockByPreNumber[currentPreNumber]; |
| 418 | BlockData& blockData = m_data[block]; |
| 419 | |
| 420 | if (blockData.dom != m_blockByPreNumber[blockData.semiNumber]) |
| 421 | blockData.dom = m_data[blockData.dom].dom; |
| 422 | } |
| 423 | } |
| 424 | |
| 425 | void link(typename Graph::Node from, typename Graph::Node to) |
| 426 | { |
| 427 | m_data[to].ancestor = from; |
| 428 | } |
| 429 | |
| 430 | typename Graph::Node eval(typename Graph::Node block) |
| 431 | { |
| 432 | if (!m_data[block].ancestor) |
| 433 | return block; |
| 434 | |
| 435 | compress(block); |
| 436 | return m_data[block].label; |
| 437 | } |
| 438 | |
| 439 | void compress(typename Graph::Node initialBlock) |
| 440 | { |
| 441 | // This was meant to be a recursive function, but we don't like recursion because we don't |
| 442 | // want to blow the stack. The original function will call compress() recursively on the |
| 443 | // ancestor of anything that has an ancestor. So, we populate our worklist with the |
| 444 | // recursive ancestors of initialBlock. Then we process the list starting from the block |
| 445 | // that is furthest up the ancestor chain. |
| 446 | |
| 447 | typename Graph::Node ancestor = m_data[initialBlock].ancestor; |
| 448 | ASSERT(ancestor); |
| 449 | if (!m_data[ancestor].ancestor) |
| 450 | return; |
| 451 | |
| 452 | Vector<typename Graph::Node, 16> stack; |
| 453 | for (typename Graph::Node block = initialBlock; block; block = m_data[block].ancestor) |
| 454 | stack.append(block); |
| 455 | |
| 456 | // We only care about blocks that have an ancestor that has an ancestor. The last two |
| 457 | // elements in the stack won't satisfy this property. |
| 458 | ASSERT(stack.size() >= 2); |
| 459 | ASSERT(!m_data[stack[stack.size() - 1]].ancestor); |
| 460 | ASSERT(!m_data[m_data[stack[stack.size() - 2]].ancestor].ancestor); |
| 461 | |
| 462 | for (unsigned i = stack.size() - 2; i--;) { |
| 463 | typename Graph::Node block = stack[i]; |
| 464 | typename Graph::Node& labelOfBlock = m_data[block].label; |
| 465 | typename Graph::Node& ancestorOfBlock = m_data[block].ancestor; |
| 466 | ASSERT(ancestorOfBlock); |
| 467 | ASSERT(m_data[ancestorOfBlock].ancestor); |
| 468 | |
| 469 | typename Graph::Node labelOfAncestorOfBlock = m_data[ancestorOfBlock].label; |
| 470 | |
| 471 | if (m_data[labelOfAncestorOfBlock].semiNumber < m_data[labelOfBlock].semiNumber) |
| 472 | labelOfBlock = labelOfAncestorOfBlock; |
| 473 | ancestorOfBlock = m_data[ancestorOfBlock].ancestor; |
| 474 | } |
| 475 | } |
| 476 | |
| 477 | struct BlockData { |
| 478 | BlockData() |
| 479 | : parent(nullptr) |
| 480 | , preNumber(UINT_MAX) |
| 481 | , semiNumber(UINT_MAX) |
| 482 | , ancestor(nullptr) |
| 483 | , label(nullptr) |
| 484 | , dom(nullptr) |
| 485 | { |
| 486 | } |
| 487 | |
| 488 | typename Graph::Node parent; |
| 489 | unsigned preNumber; |
| 490 | unsigned semiNumber; |
| 491 | typename Graph::Node ancestor; |
| 492 | typename Graph::Node label; |
| 493 | Vector<typename Graph::Node> bucket; |
| 494 | typename Graph::Node dom; |
| 495 | }; |
| 496 | |
| 497 | Graph& m_graph; |
| 498 | typename Graph::template Map<BlockData> m_data; |
| 499 | Vector<typename Graph::Node> m_blockByPreNumber; |
| 500 | }; |
| 501 | |
| 502 | class NaiveDominators { |
| 503 | public: |
| 504 | NaiveDominators(Graph& graph) |
| 505 | : m_graph(graph) |
| 506 | { |
| 507 | // This implements a naive dominator solver. |
| 508 | |
| 509 | ASSERT(!graph.predecessors(graph.root()).size()); |
| 510 | |
| 511 | unsigned numBlocks = graph.numNodes(); |
| 512 | |
| 513 | // Allocate storage for the dense dominance matrix. |
| 514 | m_results.grow(numBlocks); |
| 515 | for (unsigned i = numBlocks; i--;) |
| 516 | m_results[i].resize(numBlocks); |
| 517 | m_scratch.resize(numBlocks); |
| 518 | |
| 519 | // We know that the entry block is only dominated by itself. |
| 520 | m_results[0].clearAll(); |
| 521 | m_results[0][0] = true; |
| 522 | |
| 523 | // Find all of the valid blocks. |
| 524 | m_scratch.clearAll(); |
| 525 | for (unsigned i = numBlocks; i--;) { |
| 526 | if (!graph.node(i)) |
| 527 | continue; |
| 528 | m_scratch[i] = true; |
| 529 | } |
| 530 | |
| 531 | // Mark all nodes as dominated by everything. |
| 532 | for (unsigned i = numBlocks; i-- > 1;) { |
| 533 | if (!graph.node(i) || !graph.predecessors(graph.node(i)).size()) |
| 534 | m_results[i].clearAll(); |
| 535 | else |
| 536 | m_results[i] = m_scratch; |
| 537 | } |
| 538 | |
| 539 | // Iteratively eliminate nodes that are not dominator. |
| 540 | bool changed; |
| 541 | do { |
| 542 | changed = false; |
| 543 | // Prune dominators in all non entry blocks: forward scan. |
| 544 | for (unsigned i = 1; i < numBlocks; ++i) |
| 545 | changed |= pruneDominators(i); |
| 546 | |
| 547 | if (!changed) |
| 548 | break; |
| 549 | |
| 550 | // Prune dominators in all non entry blocks: backward scan. |
| 551 | changed = false; |
| 552 | for (unsigned i = numBlocks; i-- > 1;) |
| 553 | changed |= pruneDominators(i); |
| 554 | } while (changed); |
| 555 | } |
| 556 | |
| 557 | bool dominates(unsigned from, unsigned to) const |
| 558 | { |
| 559 | return m_results[to][from]; |
| 560 | } |
| 561 | |
| 562 | bool dominates(typename Graph::Node from, typename Graph::Node to) const |
| 563 | { |
| 564 | return dominates(m_graph.index(from), m_graph.index(to)); |
| 565 | } |
| 566 | |
| 567 | void dump(PrintStream& out) const |
| 568 | { |
| 569 | for (unsigned blockIndex = 0; blockIndex < m_graph.numNodes(); ++blockIndex) { |
| 570 | typename Graph::Node block = m_graph.node(blockIndex); |
| 571 | if (!block) |
| 572 | continue; |
| 573 | out.print(" Block " , m_graph.dump(block), ":" ); |
| 574 | for (unsigned otherIndex = 0; otherIndex < m_graph.numNodes(); ++otherIndex) { |
| 575 | if (!dominates(m_graph.index(block), otherIndex)) |
| 576 | continue; |
| 577 | out.print(" " , m_graph.dump(m_graph.node(otherIndex))); |
| 578 | } |
| 579 | out.print("\n" ); |
| 580 | } |
| 581 | } |
| 582 | |
| 583 | private: |
| 584 | bool pruneDominators(unsigned idx) |
| 585 | { |
| 586 | typename Graph::Node block = m_graph.node(idx); |
| 587 | |
| 588 | if (!block || !m_graph.predecessors(block).size()) |
| 589 | return false; |
| 590 | |
| 591 | // Find the intersection of dom(preds). |
| 592 | m_scratch = m_results[m_graph.index(m_graph.predecessors(block)[0])]; |
| 593 | for (unsigned j = m_graph.predecessors(block).size(); j-- > 1;) |
| 594 | m_scratch &= m_results[m_graph.index(m_graph.predecessors(block)[j])]; |
| 595 | |
| 596 | // The block is also dominated by itself. |
| 597 | m_scratch[idx] = true; |
| 598 | |
| 599 | return m_results[idx].setAndCheck(m_scratch); |
| 600 | } |
| 601 | |
| 602 | Graph& m_graph; |
| 603 | Vector<FastBitVector> m_results; // For each block, the bitvector of blocks that dominate it. |
| 604 | FastBitVector m_scratch; // A temporary bitvector with bit for each block. We recycle this to save new/deletes. |
| 605 | }; |
| 606 | |
| 607 | struct ValidationContext { |
| 608 | ValidationContext(Graph& graph, Dominators& dominators) |
| 609 | : graph(graph) |
| 610 | , dominators(dominators) |
| 611 | , naiveDominators(graph) |
| 612 | { |
| 613 | } |
| 614 | |
| 615 | void reportError(typename Graph::Node from, typename Graph::Node to, const char* message) |
| 616 | { |
| 617 | Error error; |
| 618 | error.from = from; |
| 619 | error.to = to; |
| 620 | error.message = message; |
| 621 | errors.append(error); |
| 622 | } |
| 623 | |
| 624 | void handleErrors() |
| 625 | { |
| 626 | if (errors.isEmpty()) |
| 627 | return; |
| 628 | |
| 629 | dataLog("DFG DOMINATOR VALIDATION FAILED:\n" ); |
| 630 | dataLog("\n" ); |
| 631 | dataLog("For block domination relationships:\n" ); |
| 632 | for (unsigned i = 0; i < errors.size(); ++i) { |
| 633 | dataLog( |
| 634 | " " , graph.dump(errors[i].from), " -> " , graph.dump(errors[i].to), |
| 635 | " (" , errors[i].message, ")\n" ); |
| 636 | } |
| 637 | dataLog("\n" ); |
| 638 | dataLog("Control flow graph:\n" ); |
| 639 | for (unsigned blockIndex = 0; blockIndex < graph.numNodes(); ++blockIndex) { |
| 640 | typename Graph::Node block = graph.node(blockIndex); |
| 641 | if (!block) |
| 642 | continue; |
| 643 | dataLog(" Block " , graph.dump(graph.node(blockIndex)), ": successors = [" ); |
| 644 | CommaPrinter comma; |
| 645 | for (auto successor : graph.successors(block)) |
| 646 | dataLog(comma, graph.dump(successor)); |
| 647 | dataLog("], predecessors = [" ); |
| 648 | comma = CommaPrinter(); |
| 649 | for (auto predecessor : graph.predecessors(block)) |
| 650 | dataLog(comma, graph.dump(predecessor)); |
| 651 | dataLog("]\n" ); |
| 652 | } |
| 653 | dataLog("\n" ); |
| 654 | dataLog("Lengauer-Tarjan Dominators:\n" ); |
| 655 | dataLog(dominators); |
| 656 | dataLog("\n" ); |
| 657 | dataLog("Naive Dominators:\n" ); |
| 658 | naiveDominators.dump(WTF::dataFile()); |
| 659 | dataLog("\n" ); |
| 660 | dataLog("Graph at time of failure:\n" ); |
| 661 | dataLog(graph); |
| 662 | dataLog("\n" ); |
| 663 | dataLog("DFG DOMINATOR VALIDATION FAILIED!\n" ); |
| 664 | CRASH(); |
| 665 | } |
| 666 | |
| 667 | Graph& graph; |
| 668 | Dominators& dominators; |
| 669 | NaiveDominators naiveDominators; |
| 670 | |
| 671 | struct Error { |
| 672 | typename Graph::Node from; |
| 673 | typename Graph::Node to; |
| 674 | const char* message; |
| 675 | }; |
| 676 | |
| 677 | Vector<Error> errors; |
| 678 | }; |
| 679 | |
| 680 | bool naiveDominates(typename Graph::Node from, typename Graph::Node to) const |
| 681 | { |
| 682 | for (typename Graph::Node block = to; block; block = m_data[block].idomParent) { |
| 683 | if (block == from) |
| 684 | return true; |
| 685 | } |
| 686 | return false; |
| 687 | } |
| 688 | |
| 689 | template<typename Functor> |
| 690 | void forAllBlocksInDominanceFrontierOfImpl( |
| 691 | typename Graph::Node from, const Functor& functor) const |
| 692 | { |
| 693 | // Paraphrasing from http://en.wikipedia.org/wiki/Dominator_(graph_theory): |
| 694 | // "The dominance frontier of a block 'from' is the set of all blocks 'to' such that |
| 695 | // 'from' dominates an immediate predecessor of 'to', but 'from' does not strictly |
| 696 | // dominate 'to'." |
| 697 | // |
| 698 | // A useful corner case to remember: a block may be in its own dominance frontier if it has |
| 699 | // a loop edge to itself, since it dominates itself and so it dominates its own immediate |
| 700 | // predecessor, and a block never strictly dominates itself. |
| 701 | |
| 702 | forAllBlocksDominatedBy( |
| 703 | from, |
| 704 | [&] (typename Graph::Node block) { |
| 705 | for (typename Graph::Node to : m_graph.successors(block)) { |
| 706 | if (!strictlyDominates(from, to)) |
| 707 | functor(to); |
| 708 | } |
| 709 | }); |
| 710 | } |
| 711 | |
| 712 | template<typename Functor> |
| 713 | void forAllBlocksInIteratedDominanceFrontierOfImpl( |
| 714 | const List& from, const Functor& functor) const |
| 715 | { |
| 716 | List worklist = from; |
| 717 | while (!worklist.isEmpty()) { |
| 718 | typename Graph::Node block = worklist.takeLast(); |
| 719 | forAllBlocksInDominanceFrontierOfImpl( |
| 720 | block, |
| 721 | [&] (typename Graph::Node otherBlock) { |
| 722 | if (functor(otherBlock)) |
| 723 | worklist.append(otherBlock); |
| 724 | }); |
| 725 | } |
| 726 | } |
| 727 | |
| 728 | struct BlockData { |
| 729 | BlockData() |
| 730 | : idomParent(nullptr) |
| 731 | , preNumber(UINT_MAX) |
| 732 | , postNumber(UINT_MAX) |
| 733 | { |
| 734 | } |
| 735 | |
| 736 | Vector<typename Graph::Node> idomKids; |
| 737 | typename Graph::Node idomParent; |
| 738 | |
| 739 | unsigned preNumber; |
| 740 | unsigned postNumber; |
| 741 | }; |
| 742 | |
| 743 | Graph& m_graph; |
| 744 | typename Graph::template Map<BlockData> m_data; |
| 745 | }; |
| 746 | |
| 747 | } // namespace WTF |
| 748 | |
| 749 | using WTF::Dominators; |
| 750 | |