attempt to build new ast

.gitignore

@@ -78,3 +78,4 @@ Sapfor/Sapc++/x64/
 
 Sapfor/out/
 Sapfor/_bin/*
+_bin/*

CMakeLists.txt

@@ -204,6 +204,8 @@ set(TR_INLINER src/Transformations/FunctionInlining/inliner.cpp
         src/Transformations/FunctionInlining/inliner.h)
 set(TR_RENAME_SYMBOLS  src/Transformations/RenameSymbols/rename_symbols.cpp
                     src/Transformations/RenameSymbols/rename_symbols.h)
+SET(TR_SWAP_OPERATORS src/Transformations/SwapOperators/swap_operators.cpp
+        src/Transformations/SwapOperators/swap_operators.h)
 
 set(TRANSFORMS 
         ${TR_DEAD_CODE}
@@ -226,7 +228,8 @@ set(TRANSFORMS
         ${TR_REPLACE_ARRAYS_IN_IO}
         ${TR_EXPR_TRANSFORM}
         ${TR_INLINER}
-        ${TR_RENAME_SYMBOLS})
+        ${TR_RENAME_SYMBOLS}
+        ${TR_SWAP_OPERATORS})
         
 set(CFG src/CFGraph/IR.cpp
         src/CFGraph/IR.h
@@ -459,6 +462,7 @@ source_group (Transformations\\GlobalVariables FILES ${TR_GV})
 source_group (Transformations\\ConvertToC FILES ${TR_CONV})
 source_group (Transformations\\SetImplicitNone FILES ${TR_IMPLICIT_NONE})
 source_group (Transformations\\ReplaceArraysInIO FILES ${TR_REPLACE_ARRAYS_IN_IO})
+source_group (Transformations\\SwapOperators FILES ${TR_SWAP_OPERATORS})
 
 
 source_group (CreateIntervals FILES ${CREATE_INTER_T})

src/PrivateAnalyzer/private_arrays_search.cpp

@@ -124,32 +124,18 @@ static void SolveDataFlow(Region* DFG)
 map<LoopGraph*, ArrayAccessingIndexes>  FindPrivateArrays(map<string, vector<LoopGraph*>> &loopGraph, map<FuncInfo*, vector<SAPFOR::BasicBlock*>>& FullIR)
 {  
     map<LoopGraph*, ArrayAccessingIndexes> result;
-    for (const auto& [fileName, loops] : loopGraph)
+    for (const auto& [loopName, loops] : loopGraph)
     {
-        SgFile::switchToFile(fileName);
         for (const auto& loop : loops)
         {
-            SgStatement* search_func = loop->loop->GetOriginal();
-
-            while (search_func && (!isSgProgHedrStmt(search_func)))
-                search_func = search_func->controlParent();
-
             for (const auto& [funcInfo, blocks]: FullIR)
-            {   
-                if (funcInfo->fileName == fileName && funcInfo->funcPointer->GetOriginal() == search_func)
             {   
                 Region* loopRegion = new Region(loop, blocks);
-                    if (loopRegion->getBasickBlocks().size() <= 1)
-                    {
-                        delete(loopRegion);
-                        continue;
-                    }
                 SolveDataFlow(loopRegion);
                 result[loop] = loopRegion->array_priv;
                 delete(loopRegion);
             }
         }
     }
-    }
     return result;
 }

src/Sapfor.cpp

@@ -89,6 +89,7 @@
 #include "Transformations/DeadCodeRemoving/dead_code.h"
 #include "Transformations/RenameSymbols/rename_symbols.h"
 #include "Transformations/FunctionInlining/inliner.h"
+#include "Transformations/SwapOperators/swap_operators.h"
 
 #include "ProjectParameters/projectParameters.h"
 
@@ -939,6 +940,8 @@ static bool runAnalysis(SgProject &project, const int curr_regime, const bool ne
                     internalExit = err;
             }
         }
+        else if (curr_regime == SWAP_OPERATORS)        
+            runSwapOperators(file, loopGraph, fullIR, countOfTransform);        
         else if (curr_regime == PRIVATE_REMOVING_ANALYSIS)
         {
             auto itFound = loopGraph.find(file->filename());
@@ -1019,6 +1022,8 @@ static bool runAnalysis(SgProject &project, const int curr_regime, const bool ne
                 if(func->funcPointer->variant() != ENTRY_STAT)
                     countOfTransform += removeDeadCode(func->funcPointer, allFuncInfo, commonBlocks);
         }
+        else if (curr_regime == FIND_PRIVATE_ARRAYS)
+            FindPrivateArrays(loopGraph, fullIR);
         else if (curr_regime == TEST_PASS)
         {
         //test pass
@@ -1035,7 +1040,8 @@ static bool runAnalysis(SgProject &project, const int curr_regime, const bool ne
                           PRIVATE_REMOVING,
                           PRIVATE_ARRAYS_EXPANSION, 
                           PRIVATE_ARRAYS_SHRINKING,
-                          REMOVE_DEAD_CODE };
+                          REMOVE_DEAD_CODE,
+                          SWAP_OPERATORS };
 
     if ((countOfTransform == 0 || internalExit > 0) && applyFor.find(curr_regime) != applyFor.end())
     {
@@ -1914,8 +1920,6 @@ static bool runAnalysis(SgProject &project, const int curr_regime, const bool ne
     }
     else if (curr_regime == TRANSFORM_ASSUMED_SIZE_PARAMETERS)
         transformAssumedSizeParameters(allFuncInfo);
-    else if (curr_regime == FIND_PRIVATE_ARRAYS) 
-        auto result = FindPrivateArrays(loopGraph, fullIR);
 
     const float elapsed = duration_cast<milliseconds>(high_resolution_clock::now() - timeForPass).count() / 1000.;
     const float elapsedGlobal = duration_cast<milliseconds>(high_resolution_clock::now() - globalTime).count() / 1000.;
@@ -2334,6 +2338,7 @@ void runPass(const int curr_regime, const char *proj_name, const char *folderNam
     case INSERT_NO_DISTR_FLAGS_FROM_GUI:
     case PRIVATE_REMOVING:
     case RENAME_INLCUDES:
+    case SWAP_OPERATORS:
         runAnalysis(*project, curr_regime, true, "", folderName);
         break;
     case INLINE_PROCEDURES:

src/Sapfor.h

@@ -122,6 +122,8 @@ enum passes {
     CREATE_INTER_TREE,
     INSERT_INTER_TREE,
 
+    SWAP_OPERATORS,
+
     SHADOW_GROUPING,
     INLINE_PROCEDURES,
     FILL_PARALLEL_REG_IR,
@@ -319,6 +321,7 @@ static void setPassValues()
     passNames[CHECK_PAR_REG_DIR] = "CHECK_PAR_REG_DIR";
     passNames[CREATE_INTER_TREE] = "CREATE_INTER_TREE";
     passNames[INSERT_INTER_TREE] = "INSERT_INTER_TREE";
+    passNames[SWAP_OPERATORS] = "SWAP_OPERATORS";
     passNames[CREATE_PARALLEL_REGIONS] = "CREATE_PARALLEL_REGIONS";
     passNames[PRIVATE_REMOVING_ANALYSIS] = "PRIVATE_REMOVING_ANALYSIS";
     passNames[PRIVATE_REMOVING] = "PRIVATE_REMOVING";

src/Transformations/SwapOperators/swap_operators.cpp

@@ -0,0 +1,422 @@
+#include <map>
+#include <unordered_set>
+#include <vector>
+#include <queue>
+#include <iostream>
+
+#include "../../Utils/errors.h"
+#include "../../Utils/SgUtils.h"
+#include "../../GraphCall/graph_calls.h"
+#include "../../GraphCall/graph_calls_func.h"
+#include "../../CFGraph/CFGraph.h"
+#include "../../CFGraph/IR.h"
+#include "../../GraphLoop/graph_loops.h"
+#include "swap_operators.h"
+
+using namespace std;
+
+
+unordered_set<int> loop_tags = {FOR_NODE/*, FORALL_NODE, WHILE_NODE, DO_WHILE_NODE*/};
+unordered_set<int> importantDepsTags = {FOR_NODE, IF_NODE};
+unordered_set<int> importantUpdDepsTags = {ELSEIF_NODE};
+unordered_set<int> importantEndTags = {CONTROL_END};
+
+
+vector<SAPFOR::IR_Block*> findInstructionsFromOperator(SgStatement* st, vector<SAPFOR::BasicBlock*> Blocks)
+{
+    vector<SAPFOR::IR_Block*> result;
+    string filename = st -> fileName();
+    for (auto& block: Blocks)
+    {
+        vector<SAPFOR::IR_Block*> instructionsInBlock = block -> getInstructions();
+        for (auto& instruction: instructionsInBlock)
+        {
+            SgStatement* curOperator = instruction -> getInstruction() -> getOperator();
+            if (curOperator -> lineNumber() == st -> lineNumber())
+                result.push_back(instruction);
+        }
+    }
+    return result;
+}
+
+vector<SAPFOR::BasicBlock*> findFuncBlocksByFuncStatement(SgStatement *st, map<FuncInfo*, vector<SAPFOR::BasicBlock*>>& FullIR)
+{
+    vector<SAPFOR::BasicBlock*> result;
+    Statement* forSt = (Statement*)st;
+    for (auto& func: FullIR)
+    {
+        if (func.first -> funcPointer -> getCurrProcessFile() == forSt -> getCurrProcessFile()
+            && func.first -> funcPointer -> lineNumber() == forSt -> lineNumber())
+            result = func.second;
+    }
+    return result;
+}
+
+map<SgForStmt*, vector<SAPFOR::BasicBlock*>> findAndAnalyzeLoops(SgStatement *st, vector<SAPFOR::BasicBlock*> blocks)
+{
+    map<SgForStmt*, vector<SAPFOR::BasicBlock*>> result;
+    SgStatement *lastNode = st->lastNodeOfStmt();
+    while (st && st != lastNode)
+    {
+        if (loop_tags.find(st -> variant()) != loop_tags.end())
+        {
+            // part with find statements of loop
+            SgForStmt *forSt = (SgForStmt*)st;
+            SgStatement *loopBody = forSt -> body();
+            SgStatement *lastLoopNode = st->lastNodeOfStmt();
+            // part with find blocks and instructions of loops
+            unordered_set<int> blocks_nums;
+            while (loopBody && loopBody != lastLoopNode)
+            {
+                SAPFOR::IR_Block* IR = findInstructionsFromOperator(loopBody, blocks).front();
+                if (blocks_nums.find(IR -> getBasicBlock() -> getNumber()) == blocks_nums.end())
+                {
+                    result[forSt].push_back(IR -> getBasicBlock());
+                    blocks_nums.insert(IR -> getBasicBlock() -> getNumber());
+                }
+                loopBody = loopBody -> lexNext();
+            }
+            std::sort(result[forSt].begin(), result[forSt].end());
+        }
+        st = st -> lexNext();
+    }
+    return result;
+}
+
+map<SgStatement*, set<SgStatement*>> AnalyzeLoopAndFindDeps(SgForStmt* forStatement, vector<SAPFOR::BasicBlock*> loopBlocks, map<FuncInfo*, vector<SAPFOR::BasicBlock*>>& FullIR)
+{
+    map<SgStatement*, set<SgStatement*>> result;
+    for (SAPFOR::BasicBlock* bb: loopBlocks)
+    {
+        map<SAPFOR::Argument*, set<int>> blockReachingDefinitions = bb -> getRD_In();
+        vector<SAPFOR::IR_Block*> instructions = bb -> getInstructions();
+        for (SAPFOR::IR_Block* irBlock: instructions)
+        {
+            // TODO: Think about what to do with function calls and array references. Because there are also dependencies there that are not reflected in RD, but they must be taken into account
+            SAPFOR::Instruction* instr = irBlock -> getInstruction();
+            result[instr -> getOperator()];
+            // take Argument 1 and it's RD and push operators to final set
+            if (instr -> getArg1() != NULL)
+            {
+                SAPFOR::Argument* arg = instr -> getArg1();
+                set<int> prevInstructionsNumbers = blockReachingDefinitions[arg];
+                for (int i: prevInstructionsNumbers)
+                {
+                    SAPFOR::Instruction* foundInstruction = getInstructionAndBlockByNumber(FullIR, i).first;
+                    if (foundInstruction != NULL)
+                    {
+                        SgStatement* prevOp = foundInstruction -> getOperator();
+                        if (prevOp != forStatement && instr -> getOperator() != forStatement && instr -> getOperator() -> lineNumber() > prevOp -> lineNumber()
+                            && prevOp -> lineNumber() > forStatement -> lineNumber())
+                            result[instr -> getOperator()].insert(prevOp);
+                    }
+                }
+            }
+            // take Argument 2 (if exists) and it's RD and push operators to final set
+            if (instr -> getArg2() != NULL)
+            {
+                SAPFOR::Argument* arg = instr -> getArg2();
+                set<int> prevInstructionsNumbers = blockReachingDefinitions[arg];
+                for (int i: prevInstructionsNumbers)
+                {
+                    SAPFOR::Instruction* foundInstruction = getInstructionAndBlockByNumber(FullIR, i).first;
+                    if (foundInstruction != NULL)
+                    {
+                        SgStatement* prevOp = foundInstruction -> getOperator();
+                        if (prevOp != forStatement && instr -> getOperator() != forStatement&& instr -> getOperator() -> lineNumber() > prevOp -> lineNumber()
+                            && prevOp -> lineNumber() > forStatement -> lineNumber())
+                            result[instr -> getOperator()].insert(prevOp);
+                    }
+                }
+            }
+            // update RD
+            if (instr -> getResult() != NULL)
+                blockReachingDefinitions[instr -> getResult()] = {instr -> getNumber()};
+        }
+    }
+    return result;
+}
+
+void buildAdditionalDeps(SgForStmt* forStatement, map<SgStatement*, set<SgStatement*>>& dependencies)
+{
+    SgStatement* lastNode = forStatement->lastNodeOfStmt();
+    vector<SgStatement*> importantDeps;
+    SgStatement* st = (SgStatement*) forStatement;
+    st = st -> lexNext();
+    SgStatement* logIfOp = NULL;
+    while (st && st != lastNode)
+    {
+        if(importantDeps.size() != 0)
+        {
+            if (st != importantDeps.back())
+            {
+                dependencies[st].insert(importantDeps.back());
+            }
+        }
+        if (logIfOp != NULL)
+        {
+            dependencies[st].insert(logIfOp);
+            logIfOp = NULL;
+        }
+        if (st -> variant() == LOGIF_NODE)
+        {
+            logIfOp = st;
+        }
+        if (importantDepsTags.find(st -> variant()) != importantDepsTags.end())
+        {
+            importantDeps.push_back(st);
+        }
+        if (importantUpdDepsTags.find(st -> variant()) != importantUpdDepsTags.end())
+        {
+            importantDeps.pop_back();
+            importantDeps.push_back(st);
+        }
+        if (importantEndTags.find(st -> variant()) != importantEndTags.end())
+        {
+            if(importantDeps.size() != 0)
+            {
+                importantDeps.pop_back();
+            }
+        }
+        st = st -> lexNext();
+    }
+}
+
+struct ReadyOp {
+    SgStatement* stmt;
+    int degree;
+    size_t arrival;
+    ReadyOp(SgStatement* s, int d, size_t a): stmt(s), degree(d), arrival(a) {}
+};
+
+struct ReadyOpCompare {
+    bool operator()(const ReadyOp& a, const ReadyOp& b) const {
+        if (a.degree != b.degree)
+            return a.degree > b.degree;
+        else
+            return a.arrival > b.arrival;
+    }
+};
+
+vector<SgStatement*> scheduleOperations(const map<SgStatement*, set<SgStatement*>>& dependencies)
+{
+    // get all statements
+    unordered_set<SgStatement*> allStmtsSet;
+    for (const auto& pair : dependencies)
+    {
+        allStmtsSet.insert(pair.first);
+        for (SgStatement* dep : pair.second)
+        {
+            allStmtsSet.insert(dep);
+        }
+    }
+    vector<SgStatement*> allStmts(allStmtsSet.begin(), allStmtsSet.end());
+    // count deps and build reversed graph
+    unordered_map<SgStatement*, vector<SgStatement*>> graph;
+    unordered_map<SgStatement*, int> inDegree;
+    unordered_map<SgStatement*, int> degree;
+    for (auto op : allStmts)
+        inDegree[op] = 0;
+    // find and remember initial dependencies
+    unordered_set<SgStatement*> dependentStmts;
+    for (const auto& pair : dependencies)
+    {
+        SgStatement* op = pair.first;
+        const auto& deps = pair.second;
+        degree[op] = deps.size();
+        inDegree[op] = deps.size();
+        if (!deps.empty())
+            dependentStmts.insert(op);
+        for (auto dep : deps)
+            graph[dep].push_back(op);
+    }
+    for (SgStatement* op : allStmts)
+    {
+        if (!degree.count(op))
+        {
+            degree[op] = 0;
+        }
+    }
+    // build queues
+    using PQ = priority_queue<ReadyOp, vector<ReadyOp>, ReadyOpCompare>;
+    PQ readyDependent;
+    queue<SgStatement*> readyIndependent;
+    size_t arrivalCounter = 0;
+    for (auto op : allStmts)
+    {
+        if (inDegree[op] == 0)
+        {
+            if (dependentStmts.count(op))
+            {
+                readyDependent.emplace(op, degree[op], arrivalCounter++);
+            }
+            else
+            {
+                readyIndependent.push(op);
+            }
+        }
+    }
+    // main sort algorythm
+    vector<SgStatement*> executionOrder;
+    while (!readyDependent.empty() || !readyIndependent.empty())
+    {
+        SgStatement* current = nullptr;
+        if (!readyDependent.empty())
+        {
+            current = readyDependent.top().stmt;
+            readyDependent.pop();
+        }
+        else
+        {
+            current = readyIndependent.front();
+            readyIndependent.pop();
+        }
+        executionOrder.push_back(current);
+        for (SgStatement* neighbor : graph[current])
+        {
+            inDegree[neighbor]--;
+            if (inDegree[neighbor] == 0) {
+                if (dependentStmts.count(neighbor))
+                {
+                    readyDependent.emplace(neighbor, degree[neighbor], arrivalCounter++);
+                }
+                else
+                {
+                    readyIndependent.push(neighbor);
+                }
+            }
+        }
+    }
+    return executionOrder;
+}
+
+static bool buildNewAST(SgStatement* loop, vector<SgStatement*>& newBody)
+{
+    if (!loop) {return false;}   
+    if (newBody.empty()) {return true;}
+    if (loop->variant() != FOR_NODE) {return false;}
+
+    SgStatement* loopStart = loop->lexNext();
+    SgStatement* loopEnd = loop->lastNodeOfStmt();
+    if (!loopStart || !loopEnd) {return false;}
+
+    for (SgStatement* stmt : newBody) {
+        if (stmt && stmt != loop && stmt != loopEnd) {
+            SgStatement* current = loopStart;
+            bool found = false;
+            while (current && current != loopEnd->lexNext()) {
+                if (current == stmt) {
+                    found = true;
+                    break;
+                }
+                current = current->lexNext();
+            }
+            if (!found) {return false;}
+        }
+    }
+
+    vector<SgStatement*> extractedStatements;
+    vector<char*> savedComments;
+    vector<int> savedLineNumbers;
+
+    for (SgStatement* stmt : newBody) {
+        if (stmt && stmt != loop && stmt != loopEnd) {
+            savedComments.push_back(stmt->comments() ? strdup(stmt->comments()) : nullptr);
+            savedLineNumbers.push_back(stmt->lineNumber());
+            SgStatement* extracted = stmt->extractStmt();
+            if (extracted) {extractedStatements.push_back(extracted);}
+        }
+    }
+
+    SgStatement* currentPos = loop;
+    int lineCounter = loop->lineNumber() + 1;
+    
+    for (size_t i = 0; i < extractedStatements.size(); i++) {
+        SgStatement* stmt = extractedStatements[i];
+        if (stmt) {
+            if (i < savedComments.size() && savedComments[i]) {
+                stmt->setComments(savedComments[i]);
+            }
+            stmt->setlineNumber(lineCounter++);
+            currentPos->insertStmtAfter(*stmt, *loop);
+            currentPos = stmt;
+        }
+    }
+
+    for (char* comment : savedComments) {
+        if (comment) {
+            free(comment);
+        }
+    }
+
+    if (currentPos && currentPos->lexNext() != loopEnd) {
+        currentPos->setLexNext(*loopEnd);
+    }
+    
+    return true;
+}
+
+static bool validateNewOrder(SgStatement* loop, const vector<SgStatement*>& newOrder)
+{
+    if (!loop || newOrder.empty()) {
+        return true;
+    }
+    unordered_set<SgStatement*> seen;
+    for (SgStatement* stmt : newOrder) {
+        if (stmt && stmt != loop && stmt != loop->lastNodeOfStmt()) {
+            if (seen.count(stmt)) {
+                return false;
+            }
+            seen.insert(stmt);
+        }
+    }
+    return true;
+}
+
+void runSwapOperators(SgFile *file, std::map<std::string, std::vector<LoopGraph*>>& loopGraph, std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& FullIR, int& countOfTransform)
+{
+    std::cout << "SWAP_OPERATORS Pass" << std::endl;    // to remove
+    countOfTransform += 1;                              // to remove
+
+    const int funcNum = file -> numberOfFunctions();
+    for (int i = 0; i < funcNum; ++i)
+    {
+        SgStatement *st = file -> functions(i);
+        vector<SAPFOR::BasicBlock*> blocks = findFuncBlocksByFuncStatement(st, FullIR);
+        map<SgForStmt*, vector<SAPFOR::BasicBlock*>> loopsMapping = findAndAnalyzeLoops(st, blocks);
+        for (pair<SgForStmt*, vector<SAPFOR::BasicBlock*>> loopForAnalyze: loopsMapping)
+        {
+            map<SgStatement*, set<SgStatement*>> dependencyGraph = AnalyzeLoopAndFindDeps(loopForAnalyze.first, loopForAnalyze.second, FullIR);
+            // TODO: Write a function that will go through the operators and update all dependencies so that there are no mix-ups and splits inside the semantic blocks (for if, do and may be some other cases)
+            buildAdditionalDeps(loopForAnalyze.first, dependencyGraph);
+            cout << endl;
+            int firstLine = loopForAnalyze.first -> lineNumber();
+            int lastLine = loopForAnalyze.first -> lastNodeOfStmt() -> lineNumber();
+            cout << "LOOP ANALYZE FROM " << firstLine << " TO " << lastLine << " RES" << endl;
+            // for (auto &v: dependencyGraph) {
+            //     cout << "OPERATOR: " << v.first -> lineNumber()  << "    " << v.first -> variant() << "\nDEPENDS ON:" << endl;
+            //     if (v.second.size() != 0)
+            //         for (auto vv: v.second)
+            //             cout << vv -> lineNumber() << "  ";
+            //     cout << endl;
+            // }
+            vector<SgStatement*> new_order = scheduleOperations(dependencyGraph);
+            cout << "RESULT ORDER:" << endl;
+            for (auto v: new_order) 
+                if (v -> lineNumber() > firstLine) 
+                    cout << v -> lineNumber() << endl;
+
+            if (validateNewOrder(loopForAnalyze.first, new_order)) {
+                buildNewAST(loopForAnalyze.first, new_order);
+            }
+            st = loopForAnalyze.first -> lexNext();
+            while (st != loopForAnalyze.first -> lastNodeOfStmt())
+            {
+                cout << st -> lineNumber() << "    " << st -> sunparse() << endl;
+                st = st -> lexNext();
+            }
+        }
+    }
+
+    return;
+};

src/Transformations/SwapOperators/swap_operators.h

@@ -0,0 +1,6 @@
+#pragma once
+
+#include "../../GraphLoop/graph_loops.h"
+#include "../../CFGraph/CFGraph.h"
+
+void runSwapOperators(SgFile *file, std::map<std::string, std::vector<LoopGraph*>>& loopGraph, std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& FullIR, int& countOfTransform);

src/Utils/PassManager.h

@@ -212,6 +212,8 @@ void InitPassesDependencies(map<passes, vector<passes>> &passDepsIn, set<passes>
 
     Pass(BUILD_IR) <= Pass(SUBST_EXPR_RD) <= Pass(SUBST_EXPR_RD_AND_UNPARSE);
 
+    list({BUILD_IR, CALL_GRAPH2}) <= Pass(SWAP_OPERATORS);
+
     list({ LOOP_ANALYZER_DATA_DIST_S1, SUBST_EXPR_RD } ) <= Pass(PRIVATE_REMOVING_ANALYSIS);
     list({ PRIVATE_REMOVING_ANALYSIS, REVERT_SUBST_EXPR_RD }) <= Pass(PRIVATE_REMOVING);