Clean up

added weighted parallel regions (by Mikhail K)

.gitignore

@@ -74,6 +74,7 @@ Sapfor/Sapc++/Sapc++/x64/
 Sapfor/Sapc++/x64/
 
 /build
+/cmake-build-debug
 
 Sapfor/out/
 Sapfor/_bin/*

src/CFGraph/CFGraph.h

@@ -24,7 +24,7 @@ namespace SAPFOR
 
         std::vector<BasicBlock*> next;
         std::vector<BasicBlock*> prev;
-        
+        BasicBlock* idom{};
         //reaching definition
         std::map<SAPFOR::Argument*, std::set<int>> RD_in, RD_out;
 
@@ -42,6 +42,7 @@ namespace SAPFOR
         void addInstruction(IR_Block* item);
         void addPrev(BasicBlock* prev_) { prev.push_back(prev_); }
         void addNext(BasicBlock* next_) { next.push_back(next_); }
+        void setIdom(BasicBlock* idom_) { idom = idom_; }
 
         int removePrev(BasicBlock* removed);
         int removeNext(BasicBlock* removed);
@@ -69,7 +70,8 @@ namespace SAPFOR
         const std::vector<IR_Block*>& getInstructions() const { return instructions; }
         const std::vector<BasicBlock*>& getNext() const { return next; }
         const std::vector<BasicBlock*>& getPrev() const { return prev; }
-        
+        BasicBlock* getIdom() const { return idom; }
+
         /*
         *  FOR LIVE ANALYSIS
         */                
@@ -146,4 +148,4 @@ static inline void deleteCFG(std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*
         byFunc.second.clear();
     }
     cfg.clear();
-}
+}

src/Distribution/Array.h

@@ -827,4 +827,4 @@ namespace Distribution
 #undef PAIR
 #undef MAP
 #undef SET
-#undef TO_STR
+#undef TO_STR

src/LoopAnalyzer/loop_analyzer.cpp

@@ -1635,9 +1635,9 @@ void loopAnalyzer(SgFile *file, vector<ParallelRegion*> &regions, map<tuple<int,
         string fName = file->functions(i)->symbol()->identifier();
 #if _WIN32
         if (file->functions(i)->variant() != MODULE_STMT)
-            sendMessage_2lvl(wstring(L"<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> '") + wstring(fName.begin(), fName.end()) + L"'");
+            sendMessage_2lvl(wstring(L"обработка функции '") + wstring(fName.begin(), fName.end()) + L"'");
         else
-            sendMessage_2lvl(wstring(L"<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD> '") + wstring(fName.begin(), fName.end()) + L"'");
+            sendMessage_2lvl(wstring(L"обработка модуля '") + wstring(fName.begin(), fName.end()) + L"'");
 #else
         if (file->functions(i)->variant() != MODULE_STMT)
             sendMessage_2lvl(wstring(L"processing function '") + wstring(fName.begin(), fName.end()) + L"'");
@@ -1710,7 +1710,10 @@ void loopAnalyzer(SgFile *file, vector<ParallelRegion*> &regions, map<tuple<int,
         map<string, pair<SgSymbol*, SgStatement*>> notMappedDistributedArrays;
         set<string> mappedDistrbutedArrays;
 
+        const ParallelRegionLines* prevParLines = NULL;
+        double prevLinesWeight = 1.0;
         double currentWeight = 1.0;
+
         while (st != lastNode)
         {
             createNeededException();
@@ -1733,13 +1736,31 @@ void loopAnalyzer(SgFile *file, vector<ParallelRegion*> &regions, map<tuple<int,
             }
 
             const int currentLine = st->lineNumber() < -1 ? st->localLineNumber() : st->lineNumber();
-            ParallelRegion *currReg = getRegionByLine(regions, st->fileName(), currentLine);
+            auto regAndLines = getRegionAndLinesByLine(regions, st->fileName(), currentLine);
+
+            auto *currReg = regAndLines.first;
+            auto *parLines = regAndLines.second;
             if (currReg == NULL)
             {
                 st = st->lexNext();
                 continue;
             }
 
+            if (parLines != prevParLines)
+            {
+                prevParLines = parLines;
+                auto newParLinesWeight = parLines ? parLines->weight : 1.0;
+
+                if (prevParLines)
+                    currentWeight /= prevLinesWeight;
+
+                if (parLines)
+                    currentWeight *= newParLinesWeight;
+
+                prevLinesWeight = newParLinesWeight;
+                prevParLines = parLines;
+            }
+
             if (isSgExecutableStatement(st) == NULL)
                 delcsStatViewed.insert(st);
             else if (!sharedMemoryParallelization && 
@@ -2168,7 +2189,7 @@ void loopAnalyzer(SgFile *file, vector<ParallelRegion*> &regions, map<tuple<int,
                 {
                     string fName = file->functions(i)->symbol()->identifier();
 #ifdef _WIN32
-                    sendMessage_2lvl(wstring(L"<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> <20><><EFBFBD><EFBFBD><EFBFBD> ") + std::to_wstring(idx) + L"/" + std::to_wstring(convertedLoopInfo.size()));
+                    sendMessage_2lvl(wstring(L"обработка цикла ") + std::to_wstring(idx) + L"/" + std::to_wstring(convertedLoopInfo.size()));
 #else
                     sendMessage_2lvl(wstring(L"processing loop ") + std::to_wstring(idx) + L"/" + std::to_wstring(convertedLoopInfo.size()));
 #endif

src/ParallelizationRegions/ParRegions.cpp

@@ -93,8 +93,10 @@ static inline SgStatement* getParentStat(SgStatement *st)
     return iterator;
 }
 
-static void updateRegionInfo(SgStatement *st, map<string, pair<Statement*, Statement*>> &startEnd, map<string, pair<int, int>> &lines_,
-                             set<string> &funcCallFromReg, const map<string, FuncInfo*> &mapFuncs)
+static void updateRegionInfo(SgStatement *st, map<string, pair<Statement*, Statement*>> &startEnd,
+                             map<string, pair<int, int>> &lines_,
+                             map<string, map<string, set<int>>> &funcCallFromReg,
+                             const map<string, FuncInfo*> &mapFuncs)
 {    
     string containsPrefix = "";
     SgStatement *st_ps = getParentStat(st);
@@ -103,17 +105,26 @@ static void updateRegionInfo(SgStatement *st, map<string, pair<Statement*, State
         containsPrefix = st_ps->symbol()->identifier() + string(".");
 
     extendRegionInfo(st, startEnd, lines_);
+
+    set<string> callsFromStatement;
+
     if (st->variant() == PROC_STAT)
     {
         string fullName = st->symbol()->identifier();
         //check contains
         if (mapFuncs.find(containsPrefix + fullName) != mapFuncs.end())
             fullName = containsPrefix + fullName;
-        funcCallFromReg.insert(fullName);
+        callsFromStatement.insert(fullName);
     }
 
     for (int z = 0; z < 3; ++z)
-        findFuncCalls(st->expr(z), funcCallFromReg, containsPrefix, mapFuncs);
+        findFuncCalls(st->expr(z), callsFromStatement, containsPrefix, mapFuncs);
+
+    string filename = st->fileName();
+    int line = st->lineNumber();
+
+    for (const auto &func_name : callsFromStatement)
+        funcCallFromReg[func_name][filename].insert(line);
 }
 
 static void fillArrayNamesInReg(set<string> &usedArrayInRegion, SgExpression *exp)
@@ -318,8 +329,9 @@ void fillRegionLines(SgFile *file, vector<ParallelRegion*> &regions, vector<Mess
     string regionName = "";
     map<string, pair<Statement*, Statement*>> startEnd;
     map<string, pair<int, int>> lines_;
-    set<string> funcCallFromReg;
+    map<string, map<string, set<int>>> funcCallFromReg;
     bool regionStarted = false;
+    double fragmentWeight = 1.0;
 
     vector<SgStatement*> toDel;
     for (int i = 0; i < funcNum; ++i)
@@ -368,6 +380,33 @@ void fillRegionLines(SgFile *file, vector<ParallelRegion*> &regions, vector<Mess
                         itFunc->second->callRegions.insert(0);
                     }
                 }
+
+                // parse SPF_APPLY_FRAGMENT clause
+                auto *apply_fragment = data->expr(1);
+                fragmentWeight = 1.0;
+
+                while (apply_fragment)
+                {
+                    auto *curr = apply_fragment->lhs();
+                    if (curr)
+                    {
+                        __spf_print(1, "%s %d\n", curr->unparse(), curr->variant());
+
+                        if (curr->variant() == SPF_WEIGHT_OP)
+                        {
+                            if (curr->lhs() &&
+                                isSgValueExp(curr->lhs()) &&
+                                isSgValueExp(curr->lhs())->doubleValue())
+                            {
+                                fragmentWeight = strtod(isSgValueExp(curr->lhs())->doubleValue(), NULL);
+                                __spf_print(1, "->> %lf\n", fragmentWeight);
+                            }
+                            else
+                                __spf_print(1, "WEIGHT clause without double argument\n");
+                        }
+                    }
+                    apply_fragment = apply_fragment->rhs();
+                }
             }
 
             if (next && next->variant() == SPF_END_PARALLEL_REG_DIR)
@@ -400,10 +439,12 @@ void fillRegionLines(SgFile *file, vector<ParallelRegion*> &regions, vector<Mess
 
                 extendRegionInfo(st, startEnd, lines_, true);
                 for (auto itRegInfo = startEnd.begin(); itRegInfo != startEnd.end(); ++itRegInfo)
-                    currReg->AddLines(lines_[itRegInfo->first], itRegInfo->first, &itRegInfo->second);
+                    currReg->AddLines(lines_[itRegInfo->first], itRegInfo->first, &itRegInfo->second, fragmentWeight);
 
-                for (auto &func : funcCallFromReg)
-                    currReg->AddFuncCalls(func);
+                for (auto &by_func : funcCallFromReg)
+                    for (auto &by_file : by_func.second)
+                        for(auto &by_line : by_file.second)
+                            currReg->AddFuncCalls(by_func.first, by_file.first, by_line);
 
                 filterUserDirectives(currReg, usedArrayInRegion, userDvmRedistrDirs, userDvmRealignDirs, userDvmShadowDirs);
                 currReg->AddUserDirectives(userDvmRealignDirs, DVM_REALIGN_DIR);
@@ -496,34 +537,48 @@ void fillRegionLinesStep2(vector<ParallelRegion*> &regions, const map<string, ve
     {
         if (regions[i]->GetName() != "DEFAULT")
             for (auto &func : regions[i]->GetFuncCalls())
-                setExplicitFlag(func, funcMap);        
+                setExplicitFlag(func.first, funcMap);
     }
 
     for (int i = 0; i < regions.size(); ++i)
     {
         if (regions[i]->GetName() != "DEFAULT")
         {
-            set<string> uniqFuncCalls;
-            for (auto &elem : regions[i]->GetFuncCalls())
-                uniqFuncCalls.insert(elem);
+            map<string, double> uniqFuncCalls;
+            map<string, map<string, set<int>>> callPlaces;
+
+            for (auto &elem : regions[i]->GetFuncCalls())
+            {
+                double max_weight = 0;
+                for (auto* fragment : elem.second)
+                    if (fragment->weight > max_weight)
+                        max_weight = fragment->weight;
+
+                uniqFuncCalls[elem.first] = max_weight;
+            }
+
+            bool wasChanged = true;
+            auto funcsBefore = uniqFuncCalls;
 
-            bool wasChanged = 1;
             while (wasChanged)
             {
-                wasChanged = 0;
+                wasChanged = false;
+                auto updated = uniqFuncCalls;
+
                 for (auto &uniqF : uniqFuncCalls)
                 {
-                    auto func = funcMap.find(uniqF);
+                    auto func = funcMap.find(uniqF.first);
                     if (func != funcMap.end())
                     {
-                        for (auto &calls : func->second->callsFrom)
+                        for (auto &call : func->second->callsFromDetailed)
                         {
-                            auto it = uniqFuncCalls.find(calls);
-                            if (it == uniqFuncCalls.end())
-                            {
-                                uniqFuncCalls.insert(it, calls);
-                                wasChanged = 1;
-                            }
+                            auto it = updated.find(call.detailCallsFrom.first);
+                            if (it == updated.end())
+                                updated.insert({call.detailCallsFrom.first, uniqF.second});
+                            else
+                                it->second = std::max(it->second, uniqF.second);
+
+                            callPlaces[call.detailCallsFrom.first][func->second->fileName].insert(call.detailCallsFrom.second);
                         }
                     }
                 }
@@ -532,21 +587,27 @@ void fillRegionLinesStep2(vector<ParallelRegion*> &regions, const map<string, ve
             string toPrint = "";
             for (auto &elem : uniqFuncCalls)
             {
-                auto it = funcMap.find(elem);
+                auto it = funcMap.find(elem.first);
                 if (it != funcMap.end())
                 {
-                    regions[i]->AddLines(it->second->linesNum, it->second->fileName);
-                    regions[i]->AddFuncCallsToAllCalls(it->second);
+                    regions[i]->AddLines(it->second->linesNum, it->second->fileName, NULL, elem.second);
 
                     if (it->second->inRegion == 0)
                         it->second->inRegion = 2;
 
                     it->second->callRegions.insert(i);
 
-                    toPrint += elem + " ";
+                    toPrint += elem.first + " ";
                 }
             }
 
+            for (auto &elem : callPlaces)
+            {
+                for (const auto &byFile : elem.second)
+                    for (auto byLine : byFile.second)
+                        regions[i]->AddFuncCalls(elem.first, byFile.first, byLine);
+            }
+
             if (toPrint != "")
                 __spf_print(1, "[%s]: funcs: %s\n", regions[i]->GetName().c_str(), toPrint.c_str());
         }

src/ParallelizationRegions/ParRegions.h

@@ -19,7 +19,7 @@
 
 struct ParallelRegionLines
 {
-    ParallelRegionLines() 
+    ParallelRegionLines(double weight = 1.0) : weight(weight)
     {
         lines = std::make_pair(-1, -1);
         stats = std::make_pair<Statement*, Statement*>(NULL, NULL);
@@ -27,14 +27,15 @@ struct ParallelRegionLines
         intervalAfter = std::make_pair<Statement*, Statement*>(NULL, NULL);
     }
 
-    ParallelRegionLines(const std::pair<int, int> &lines) : lines(lines) 
+    ParallelRegionLines(const std::pair<int, int> &lines, double weight = 1.0) : lines(lines), weight(weight)
     {
         stats = std::make_pair<Statement*, Statement*>(NULL, NULL);
         intervalBefore = std::make_pair<Statement*, Statement*>(NULL, NULL);
         intervalAfter = std::make_pair<Statement*, Statement*>(NULL, NULL);
     }
 
-    ParallelRegionLines(const std::pair<int, int> &lines, const std::pair<Statement*, Statement*> stats) : lines(lines), stats(stats)
+    ParallelRegionLines(const std::pair<int, int> &lines, const std::pair<Statement*, Statement*> stats, double weight = 1.0)
+        : lines(lines), stats(stats), weight(weight)
     {
         intervalBefore = std::make_pair<Statement*, Statement*>(NULL, NULL);
         intervalAfter = std::make_pair<Statement*, Statement*>(NULL, NULL);
@@ -62,6 +63,8 @@ struct ParallelRegionLines
     // <start, end> interval
     std::pair<Statement*, Statement*> intervalBefore;
     std::pair<Statement*, Statement*> intervalAfter;
+
+    double weight; // weight of the fragment among all fragments of this region
 };
 
 #if __SPF
@@ -116,7 +119,7 @@ public:
         currentVariant = copy.currentVariant;
     }
 
-    int AddLines(const std::pair<int, int> &linesToAdd, const std::string &file, const std::pair<Statement*, Statement*> *startEnd = NULL)
+    int AddLines(const std::pair<int, int> &linesToAdd, const std::string &file, const std::pair<Statement*, Statement*> *startEnd = NULL, double weight = 1.0)
     {
         if (linesToAdd.first > linesToAdd.second)
             return -1;
@@ -126,17 +129,29 @@ public:
             it = lines.insert(it, make_pair(file, std::vector<ParallelRegionLines>()));
 
         if (startEnd)
-            it->second.push_back(ParallelRegionLines(linesToAdd, *startEnd));
+            it->second.push_back(ParallelRegionLines(linesToAdd, *startEnd, weight));
         else
-            it->second.push_back(ParallelRegionLines(linesToAdd));
+            it->second.push_back(ParallelRegionLines(linesToAdd, weight));
 
         return 0;
     }
 
-    void AddFuncCalls(const std::string &func) { functionsCall.insert(func); }
+    void AddFuncCalls(const std::string &func, const std::string &file, const int line)
+    {
+        auto *found_lines = GetLinesByLine(file, line);
+
+        if (found_lines)
+            functionsCall[func].insert(found_lines);
+    }
 
 #if __SPF
-    void AddFuncCallsToAllCalls(FuncInfo *func) { allFunctionsCall.insert(func); }
+    void AddFuncCallsToAllCalls(FuncInfo *func, const std::string &file, const int line)
+    {
+        auto *found_lines = GetLinesByLine(file, line);
+
+        if (found_lines)
+            allFunctionsCall[func].insert(found_lines);
+    }
 #endif
 
     uint64_t GetId() const { return regionId; }
@@ -176,10 +191,10 @@ public:
     const DataDirective& GetDataDir() const { return dataDirectives; }
     DataDirective& GetDataDirToModify() { return dataDirectives; }
 
-    const std::set<std::string>& GetFuncCalls() const { return functionsCall; }
+    const std::map<std::string, std::set<const ParallelRegionLines*>>& GetFuncCalls() const { return functionsCall; }
 
 #if __SPF
-    const std::set<FuncInfo*>& GetAllFuncCalls() const { return allFunctionsCall; }
+    const std::map<FuncInfo*, std::set<const ParallelRegionLines*>>& GetAllFuncCalls() const { return allFunctionsCall; }
     const std::map<FuncInfo*, std::map<DIST::Array*, std::vector<ParallelRegionLines>>>& GetUsedLocalArrays() const { return usedLocalArrays; }
     const std::map<FuncInfo*, std::map<DIST::Array*, std::vector<ParallelRegionLines>>>& GetUsedCommonArrays() const { return usedCommonArrays; }
 
@@ -218,7 +233,7 @@ public:
     }
 #endif
 
-    bool HasThisLine(const int line, const std::string &file) const
+    bool HasThisLine(const int line, const std::string &file, const ParallelRegionLines** found = nullptr) const
     {
         bool retVal = false;
         auto it = lines.find(file);
@@ -229,6 +244,9 @@ public:
                 if (it->second[i].lines.first <= line && it->second[i].lines.second >= line)
                 {
                     retVal = true;
+                    if (found)
+                        *found = &(it->second[i]);
+
                     break;
                 }
             }
@@ -292,7 +310,7 @@ public:
         fprintf(fileOut, "  originalName '%s'\n", originalName.c_str());
         fprintf(fileOut, "  functions call from %d:\n", (int)functionsCall.size());
         for (auto &func : functionsCall)
-            fprintf(fileOut, "    '%s'\n", func.c_str());
+            fprintf(fileOut, "    '%s'\n", func.first.c_str());
         fprintf(fileOut, "  total lines %d:\n", (int)lines.size());
         for (auto &line : lines)
         {
@@ -362,11 +380,11 @@ private:
     std::string originalName;
     // file -> lines info
     std::map<std::string, std::vector<ParallelRegionLines>> lines;
-    std::set<std::string> functionsCall;
+    std::map<std::string, std::set<const ParallelRegionLines*>> functionsCall; // func name -> fragments with calls
 
 #if __SPF
     // for RESOLVE_PAR_REGIONS
-    std::set<FuncInfo*> allFunctionsCall;
+    std::map<FuncInfo*, std::set<const ParallelRegionLines*>> allFunctionsCall; // function -> fragments with calls
     std::map<FuncInfo*, std::map<DIST::Array*, std::vector<ParallelRegionLines>>> usedLocalArrays;  // func -> array -> lines
     std::map<FuncInfo*, std::map<DIST::Array*, std::vector<ParallelRegionLines>>> usedCommonArrays; // func -> array -> lines
     //
@@ -408,4 +426,5 @@ private:
 ParallelRegion* getRegionById(const std::vector<ParallelRegion*>& regions, const uint64_t regionId);
 ParallelRegion* getRegionByName(const std::vector<ParallelRegion*>& regions, const std::string& regionName);
 ParallelRegion* getRegionByLine(const std::vector<ParallelRegion*>& regions, const std::string& file, const int line);
+std::pair<ParallelRegion*, const ParallelRegionLines*> getRegionAndLinesByLine(const std::vector<ParallelRegion*>& regions, const std::string& file, const int line);
 std::set<ParallelRegion*> getAllRegionsByLine(const std::vector<ParallelRegion*>& regions, const std::string& file, const int line);

src/ProjectParameters/domTree.h

@@ -0,0 +1,111 @@
+#pragma once
+
+#include "vector"
+#include "map"
+
+#include "../CFGraph/CFGraph.h"
+#include <unordered_map>
+
+using namespace std;
+
+namespace SAPFOR {
+class DominatorFinder {
+private:
+    BasicBlock* entry;
+    std::vector<BasicBlock*> vertices;
+    std::unordered_map<BasicBlock*, int> dfs_num;
+    std::vector<int> parent, semi, vertex, ancestor, label;
+    std::vector<std::vector<int>> bucket;
+    int n;
+
+    void DFS(BasicBlock* v, int parent_num) {
+        dfs_num[v] = n;
+        vertex[n] = n;
+        semi[n] = n;
+        label[n] = n;
+        ancestor[n] = -1;
+        parent[n] = parent_num;
+        vertices[n++] = v;
+
+        for (const auto& w : v->getNext()) {
+            if (dfs_num[w] == -1) {
+                DFS(w, dfs_num[v]);
+            }
+        }
+    }
+
+    void Compress(int v) {
+        if (ancestor[ancestor[v]] != -1) {
+            Compress(ancestor[v]);
+            if (semi[label[ancestor[v]]] < semi[label[v]])
+                label[v] = label[ancestor[v]];
+            ancestor[v] = ancestor[ancestor[v]];
+        }
+    }
+
+    int Eval(int v) {
+        if (ancestor[v] == -1) return v;
+        Compress(v);
+        return label[v];
+    }
+
+    void Link(int v, int w) {
+        ancestor[w] = v;
+    }
+
+public:
+    DominatorFinder(std::vector<BasicBlock*>& blocks) {
+        if (blocks.empty()) return;
+        entry = blocks[0];
+        n = 0;
+
+        for (auto block : blocks) dfs_num[block] = -1;
+
+        int max_size = blocks.size();
+        vertices.resize(max_size);
+        parent.assign(max_size, -1);
+        semi.assign(max_size, -1);
+        vertex.assign(max_size, -1);
+        ancestor.assign(max_size, -1);
+        label.assign(max_size, -1);
+        bucket.resize(max_size);
+
+        DFS(entry, -1);
+
+        for (int i = n - 1; i > 0; --i) {
+            int w = vertex[i];
+
+            for (BasicBlock* v : vertices[w]->getPrev()) {
+                int u = Eval(dfs_num[v]);
+                if (semi[u] < semi[w])
+                    semi[w] = semi[u];
+            }
+            bucket[vertex[semi[w]]].push_back(w);
+            Link(parent[w], w);
+
+            for (int v : bucket[parent[w]])
+            {
+                int u = Eval(v);
+                if (semi[u] < semi[v])
+                    vertices[v]->setIdom(vertices[u]);
+                else
+                    vertices[v]->setIdom(vertices[parent[w]]);
+            }
+            bucket[parent[w]].clear();
+        }
+
+        for (int i = 1; i < n; ++i) {
+            int w = vertex[i];
+            if (vertices[w]->getIdom() != vertices[vertex[semi[w]]])
+                vertices[w]->setIdom(vertices[w]->getIdom()->getIdom());
+        }
+
+        entry->setIdom(nullptr);
+    }
+};
+
+void buildDominatorTreeLT(std::vector<BasicBlock*>& blocks) {
+    DominatorFinder finder(blocks);
+}
+
+}

src/ProjectParameters/projectParameters.cpp

@@ -1,33 +1,413 @@
-#include "../Utils/leak_detector.h"
-
 #include <cstdio>
 #include <cstring>
-#include <cstring>
 #include <fstream>
 #include <iostream>
 #include <cstdlib>
 #include <set>
+#include <string>
 #include <vector>
 #include <algorithm>
 #include <tuple>
 
+#include "../Utils/SgUtils.h"
+#include "../CFGraph/CFGraph.h"
+#include "CFGraph/IR.h"
+#include "Distribution/Array.h"
 #include "dvm.h"
 #include "../Utils/errors.h"
-#include "../Utils/SgUtils.h"
 #include "../GraphCall/graph_calls.h"
 #include "../GraphCall/graph_calls_func.h"
 
+#include "libSage++.h"
 #include "projectParameters.h"
+#include "domTree.h"
 
 using namespace std;
 
-map< pair<string, int>, set<string>> 
-	findParameters(const map<string, vector<DefUseList>> &defUseByFunctions, 
-                   const map<string, CommonBlock*> &commonBlocks, 
-				   const map<string, vector<FuncInfo*>> &allFuncInfo)
+tuple<FuncInfo*, SAPFOR::Instruction*, SAPFOR::BasicBlock*> stmtToIR(const map<FuncInfo*, vector<SAPFOR::BasicBlock*>>& CFGraph, SgStatement* stmt)
 {
-	map< pair<string, int>, set<string>> foundParameters;
-	
-	
-	return foundParameters;
-}
+    SgStatement* cur = stmt;
+    cur->switchToFile();
+    while (cur->variant() != PROC_HEDR && cur->variant() != PROG_HEDR && cur->variant() != FUNC_HEDR)
+        cur = cur->controlParent();
+
+    string funcName = ((SgProcHedrStmt*)cur)->nameWithContains();
+
+    int stmtID = stmt->id();
+    for (const auto& [func, bblocks] : CFGraph)
+    {
+        if (func->funcName != funcName)
+            continue;
+
+        for (auto basicBlock : bblocks)
+            for (auto ins : basicBlock->getInstructions())
+                if (stmtID == ins->getInstruction()->getOperator()->id())
+                    return make_tuple(func, ins->getInstruction(), basicBlock);
+    }
+
+    printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+    return { NULL, NULL, NULL };
+}
+
+tuple<FuncInfo*, SAPFOR::Instruction*, SAPFOR::BasicBlock*> IRByNumber(const map<FuncInfo*, vector<SAPFOR::BasicBlock*>>& CFGraph, int num)
+{
+    if (num < 0)
+        return { NULL, NULL, NULL };
+
+    for (const auto& [func, bblocks] : CFGraph)
+        for (auto byBB : bblocks)
+            if (byBB->getInstructions().front()->getNumber() <= num && byBB->getInstructions().back()->getNumber() >= num)
+                return make_tuple(func, getInstructionByNumber(byBB->getInstructions(), num), byBB);
+
+    printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+    return { NULL, NULL, NULL};
+}
+
+template<typename Iterator>
+void processArgument(set<SAPFOR::Argument*>& worklist,
+                     SAPFOR::Argument* arg,
+                     Iterator instr,
+                     Iterator first_instr)
+{
+    if (arg == NULL)
+        return;
+    if (arg->getType() == SAPFOR::CFG_ARG_TYPE::REG)
+        extract_vars_from_reg(worklist, arg, instr, first_instr);
+    else if (arg->getType() == SAPFOR::CFG_ARG_TYPE::VAR)
+    {
+        worklist.insert(arg);
+    }
+}
+
+template<typename Iterator>
+void extract_vars_from_reg(set<SAPFOR::Argument*>& worklist,
+                           SAPFOR::Argument* reg,
+                           Iterator instr,
+                           Iterator first_instr)
+{
+    for (; instr >= first_instr; instr--)
+    {
+        if ((*instr)->getInstruction()->getResult() == reg)
+        {
+            processArgument(worklist, (*instr)->getInstruction()->getArg1(), instr, first_instr);
+            processArgument(worklist, (*instr)->getInstruction()->getArg2(), instr, first_instr);
+            return;
+        }
+    }
+}
+
+void lookup_for_vars(std::set<std::tuple<SgStatement*, std::string, MODE>>& where_to_add,
+                     set<SAPFOR::Argument*>& worklist,
+                     SAPFOR::Instruction* instr,
+                     SAPFOR::BasicBlock* bblock,
+                     FuncInfo* cur_func,
+                     const std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& fullIR)
+{
+    while (bblock)
+    {
+        auto first_instr = bblock->getInstructions().begin();
+        auto cur_instr = std::find_if(first_instr, bblock->getInstructions().end(), [instr](SAPFOR::IR_Block* i) {
+            return i->getInstruction() == instr;
+        });
+
+        for (; cur_instr >= bblock->getInstructions().begin(); cur_instr--)
+        {
+            auto instr = (*cur_instr)->getInstruction();
+            auto result_arg = instr->getResult();
+            auto arg1 = instr->getArg1();
+            auto arg2 = instr->getArg2();
+
+            if (worklist.count(result_arg))
+            {
+                processArgument(worklist, arg1, cur_instr, first_instr);
+                processArgument(worklist, arg2, cur_instr, first_instr);
+                worklist.erase(result_arg);
+            }
+            if (instr->getOperation() == SAPFOR::CFG_OP::PARAM && worklist.count(arg1))
+            {
+                // skip to F_CALL
+                auto f_call_instr = cur_instr;
+                while ((*f_call_instr)->getInstruction()->getOperation() != SAPFOR::CFG_OP::F_CALL)
+                    f_call_instr++;
+
+                if ((*f_call_instr)->getInstruction()->getArg1()->getValue() == "_READ")
+                {
+                    auto stmt_before = (*f_call_instr)->getInstruction()->getOperator();
+                    auto filename = stmt_before->fileName();
+                    auto line = stmt_before->lineNumber();
+                    auto var_name = arg1->getValue().substr(arg1->getValue().find('%') + 1);
+                    __spf_print(1,"Please specify value of variable %s on line %d of file %s\n", arg1->getValue().c_str(), line, filename);
+                    auto toAdd = make_tuple(stmt_before, var_name, MODE::AFTER);
+                    where_to_add.insert(toAdd);
+                    worklist.erase(arg1);
+                }
+            }
+        }
+
+        const auto& RD = bblock->getRD_In();
+        map<SAPFOR::BasicBlock*, SAPFOR::Instruction*> group_by_block;
+        for (auto& arg : worklist)
+        {
+            if (RD.count(arg))
+            {
+                if (RD.at(arg).size() == 1 && *RD.at(arg).begin() == SAPFOR::CFG_VAL::UNINIT)
+                    __spf_print(1, "variable %s has no definition\n", arg->getValue().c_str());
+                else if (RD.at(arg).size() > 1)
+                {
+                    auto stmt_after = (*first_instr)->getInstruction()->getOperator();
+                    auto filename = stmt_after->fileName();
+                    auto line = stmt_after->lineNumber();
+                    auto var_name = arg->getValue().substr(arg->getValue().find('%') + 1);
+                    __spf_print(1, "variable %s has multiple reaching definitions, further analysis is impossible\n", arg->getValue().c_str());
+                    __spf_print(1,"Please specify value of variable %s on line %d of file %s\n", arg->getValue().c_str(), line, filename);
+                    auto toAdd = make_tuple(stmt_after, var_name, MODE::BEFORE);
+                    where_to_add.insert(toAdd);
+                    worklist.erase(arg);
+                }
+                else
+                {
+                    auto instr_num = *RD.at(arg).begin();
+                    auto [func, instr, bblock] = IRByNumber(fullIR, instr_num);
+                    if (cur_func == func && (group_by_block[bblock] == NULL || group_by_block[bblock]->getNumber() < instr_num))
+                        group_by_block[bblock] = instr;
+                }
+            }
+        }
+
+        while (bblock && group_by_block.find(bblock) == group_by_block.end())
+            bblock = bblock->getIdom();
+        if (bblock)
+            instr = group_by_block[bblock];
+    }
+
+    // other variables are from global scope
+    const auto& RD = fullIR.at(cur_func).front()->getRD_In();
+    for (auto& arg : worklist)
+    {
+        if (arg->isMemGlobal())
+        {
+            set<int> found_rd;
+            if (RD.count(arg))
+                found_rd = RD.at(arg);
+            if (found_rd.size() == 0)
+            {
+                auto call_instr = call_sites[cur_func].size() ? call_sites[cur_func].front() : NULL;
+                while (call_instr && found_rd.size() == 0)
+                {
+                    auto [call_func, _, call_bblock] = IRByNumber(fullIR, call_instr->getNumber());
+                    if (call_bblock->getRD_Out().count(arg))
+                        found_rd = call_bblock->getRD_Out().at(arg);
+
+
+                    call_instr = call_sites[call_func].size() ? call_sites[call_func].front() : NULL;
+                }
+            }
+            if (found_rd.size() == 1 && *found_rd.begin() == SAPFOR::CFG_VAL::UNINIT)
+            {
+                __spf_print(1, "variable %s has no definition\n", arg->getValue().c_str());
+            } else if (found_rd.size() > 1)
+            {
+                auto first_instr = fullIR.at(cur_func).front()->getInstructions().begin();
+                auto stmt_after = (*first_instr)->getInstruction()->getOperator();
+                auto filename = stmt_after->fileName();
+                auto line = stmt_after->lineNumber();
+                auto var_name = arg->getValue().substr(arg->getValue().find('%') + 1);
+                __spf_print(1, "variable %s has multiple reaching definitions, further analysis is impossible\n", arg->getValue().c_str());
+                __spf_print(1,"Please specify value of variable %s on line %d of file %s\n", arg->getValue().c_str(), line, filename);
+                auto toAdd = make_tuple(stmt_after, var_name, MODE::BEFORE);
+                where_to_add.insert(toAdd);
+            }
+            else
+            {
+                auto instr_num = *found_rd.begin();
+                auto [func, instr, bblock] = IRByNumber(fullIR, instr_num);
+                set<SAPFOR::Argument*> new_worklist = {arg};
+
+                lookup_for_vars(where_to_add, new_worklist, instr, bblock, func, fullIR);
+            }
+        }
+    }
+
+
+    for (const auto& call_instr : call_sites[cur_func])
+    {
+        set<SAPFOR::Argument*> new_worklist;
+        auto params_num = cur_func->funcParams.countOfPars;
+
+        auto [call_func, _, call_bblock] = IRByNumber(fullIR, call_instr->getNumber());
+        auto first_instr = call_bblock->getInstructions().begin();
+        auto cur_instr = std::find_if(first_instr, call_bblock->getInstructions().end(), [call_instr](SAPFOR::IR_Block* i) {
+            return i->getInstruction() == call_instr;
+        });
+        for (auto& arg : worklist)
+        {
+            if (arg->getMemType() == SAPFOR::CFG_MEM_TYPE::FUNC_PARAM_)
+            {
+                auto param_num= stoi(arg->getValue().substr(arg->getValue().find('%', arg->getValue().find('%') + 1) + 1));
+                auto param_instr = (cur_instr - (params_num - param_num));
+                auto param_arg = (*param_instr)->getInstruction()->getArg1();
+                processArgument(new_worklist, param_arg, param_instr, first_instr);
+            }
+        }
+        lookup_for_vars(where_to_add, new_worklist, call_instr, call_bblock, call_func, fullIR);
+    }
+
+}
+
+void handle_single_allocate(std::set<tuple<SgStatement*, std::string, MODE>>& where_to_add,
+                            SgStatement* alloc_statement,
+                            const std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& fullIR)
+{
+    auto [func, instr, bblock] = stmtToIR(fullIR, alloc_statement);
+
+    auto first_instr = bblock->getInstructions().begin();
+    auto cur_instr = std::find_if(first_instr, bblock->getInstructions().end(), [instr](SAPFOR::IR_Block* i) {
+        return i->getInstruction() == instr;
+    });
+    auto alloc_instr = cur_instr;
+
+    // skip to F_CALL _ALLOC n
+    while ((*alloc_instr)->getInstruction()->getOperation() != SAPFOR::CFG_OP::F_CALL ||
+           (*alloc_instr)->getInstruction()->getArg1()->getValue() != "_ALLOC")
+        alloc_instr++;
+
+
+    auto arrays_num = stoi((*alloc_instr)->getInstruction()->getArg2()->getValue());
+
+    set<SAPFOR::Argument*> worklist;
+    for (int i = 0; i < arrays_num; i++)
+    {
+        auto param_instr = --alloc_instr;
+        auto param_reg = (*param_instr)->getInstruction()->getArg1();
+
+        while ((*param_instr)->getInstruction()->getOperation() != SAPFOR::CFG_OP::LOAD ||
+               (*param_instr)->getInstruction()->getResult() != param_reg)
+            param_instr--;
+
+
+        auto dimensions_num = stoi((*param_instr)->getInstruction()->getArg2()->getValue());
+
+        for (int j = 0; j < dimensions_num; j++)
+        {
+            auto ref_instr = --param_instr;
+            if ((*ref_instr)->getInstruction()->getOperation() == SAPFOR::CFG_OP::RANGE)
+            {
+                vector<SAPFOR::Argument*> range_args = {(*ref_instr)->getInstruction()->getArg1(),
+                                                        (*ref_instr)->getInstruction()->getArg2(),
+                                                        (*ref_instr)->getInstruction()->getResult()};
+                for (auto& arg : range_args)
+                    processArgument(worklist, arg, ref_instr, first_instr);
+
+            } else
+            {
+                auto arg = (*ref_instr)->getInstruction()->getArg1();
+                processArgument(worklist, arg, ref_instr, first_instr);
+            }
+        }
+    }
+    lookup_for_vars(where_to_add,worklist, instr, bblock, func, fullIR);
+}
+
+
+void handle_single_loop(std::set<tuple<SgStatement*, std::string, MODE>>& where_to_add,
+                        SgStatement* loop_stmt,
+                        const std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& fullIR)
+{
+    auto [func, instr, bblock] = stmtToIR(fullIR, loop_stmt);
+
+    auto cur_instr = bblock->getInstructions().end() - 1;
+
+    set<SAPFOR::Argument*> worklist;
+    extract_vars_from_reg(worklist, (*cur_instr)->getInstruction()->getResult(), cur_instr, bblock->getInstructions().begin());
+
+    lookup_for_vars(where_to_add, worklist, (*cur_instr)->getInstruction(), bblock, func, fullIR);
+}
+
+void
+findParameters(ResultSet& foundParameters,
+               std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& fullIR,
+               const std::map<std::tuple<int, std::string, std::string>, std::pair<DIST::Array*, DIST::ArrayAccessInfo*>>& declaredArrays)
+{
+    set<tuple<SgStatement*, string, MODE>> where_to_add;
+
+    map<string, FuncInfo*> name_to_func;
+    for (const auto& [func, _] : fullIR)
+        name_to_func[func->funcName] = func;
+
+    for (auto& [func, bblocks] : fullIR)
+    {
+        for (const auto& block : bblocks)
+            for (const auto& ir_block : block->getInstructions())
+            {
+                auto instr = ir_block->getInstruction();
+                if (instr->getOperation() == SAPFOR::CFG_OP::F_CALL)
+                {
+                    auto func_name = instr->getArg1()->getValue();
+                    auto func_info = name_to_func.find(func_name);
+
+                    if (func_info != name_to_func.end())
+                        call_sites[func_info->second].push_back(instr);
+                }
+            }
+        
+
+        SAPFOR::buildDominatorTreeLT(bblocks);
+    }
+
+    std::set<SgStatement*> alloc_statements;
+    for (const auto& [func, bblocks] : fullIR)
+        for (const auto& block : bblocks)
+            for (auto instr = block->getInstructions().begin(); instr != block->getInstructions().end(); ++instr)
+            {
+                auto op = (*instr)->getInstruction()->getOperator();
+                if (op && op->variant() == ALLOCATE_STMT) {
+                    alloc_statements.insert(op);
+                }
+            }
+
+    set<SgStatement*> for_statements;
+    // Find all FOR statements in the program
+    for (const auto& [func, bblocks] : fullIR)
+        for (const auto& block : bblocks)
+            for (auto instr = block->getInstructions().begin(); instr != block->getInstructions().end(); ++instr)
+            {
+                auto op = (*instr)->getInstruction()->getOperator();
+                if (op && op->variant() == FOR_NODE)
+                    for_statements.insert(op);
+            }
+
+    for (const auto& alloc_statement : alloc_statements)
+        handle_single_allocate(where_to_add, alloc_statement, fullIR);
+
+    for (const auto& stmt : for_statements)
+        handle_single_loop(where_to_add, stmt, fullIR);
+
+    for (const auto& [stmt_before, var_name, mode] : where_to_add)
+    {
+        stmt_before->switchToFile();
+
+        SgVariableSymb* var_symb = new SgVariableSymb(var_name.c_str());
+        SgVarRefExp* var = new SgVarRefExp(var_symb);
+        SgValueExp* zero = new SgValueExp(1337);
+        SgExprListExp* ex = new SgExprListExp();
+        auto assgn_op = new SgExpression(ASSGN_OP, var, zero);
+        ex->setLhs(assgn_op);
+
+        SgExpression* parameter_op = new SgExpression(SPF_PARAMETER_OP, ex);
+        auto dir_list = new SgExprListExp();
+        dir_list->setLhs(parameter_op);
+        SgStatement* toAdd = new SgStatement(SPF_ANALYSIS_DIR, NULL, NULL, dir_list, NULL, NULL);
+
+        toAdd->setlineNumber(stmt_before->lineNumber());
+        toAdd->setLocalLineNumber(stmt_before->lineNumber());
+        toAdd->setFileId(stmt_before->getFileId());
+        toAdd->setProject(stmt_before->getProject());
+
+        if (mode == MODE::AFTER)
+            stmt_before->insertStmtAfter(*toAdd, *stmt_before->controlParent());
+        else
+            stmt_before->insertStmtBefore(*toAdd, *stmt_before->controlParent());
+
+        foundParameters.insert(make_tuple(stmt_before->fileName(), stmt_before->lineNumber(), var_name));
+    }
+}

src/ProjectParameters/projectParameters.h

@@ -1,3 +1,42 @@
 #pragma once
 
-std::map< std::pair<std::string, int>, std::set<std::string>> findParameters(const std::map<std::string, std::vector<DefUseList>> &defUseByFunctions, const std::map<std::string, CommonBlock*> &commonBlocks, const std::map<std::string, std::vector<FuncInfo*>> &allFuncInfo);
+#include "libSage++.h"
+#include <map>
+#include <string>
+#include <set>
+#include <vector>
+
+using ResultSet = std::set<std::tuple<std::string, int, std::string>>;
+
+enum class MODE 
+{
+    BEFORE,
+    AFTER
+};
+
+static std::map<FuncInfo*, std::vector<SAPFOR::Instruction*>> call_sites;
+
+template<typename Iterator>
+void extract_vars_from_reg(std::set<SAPFOR::Argument*>& worklist,
+                           SAPFOR::Argument* reg,
+                           Iterator instr,
+                           Iterator first_instr);
+
+
+template<typename Iterator>
+void processArgument(std::set<SAPFOR::Argument*>& worklist,
+                     SAPFOR::Argument* arg,
+                     Iterator instr,
+                     Iterator first_instr);
+
+void lookup_for_vars(std::set<std::tuple<SgStatement*, std::string, MODE>>& where_to_add,
+                     std::set<SAPFOR::Argument*>& worklist,
+                     SAPFOR::Instruction* instr,
+                     SAPFOR::BasicBlock* bblock,
+                     FuncInfo* cur_func,
+                     const std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& fullIR);
+
+void
+findParameters(ResultSet& foundParameters,
+               std::map<FuncInfo*, std::vector<SAPFOR::BasicBlock*>>& fullIR,
+			   const std::map<std::tuple<int, std::string, std::string>, std::pair<DIST::Array*, DIST::ArrayAccessInfo*>>& declaredArrays);

src/Sapfor.cpp

@@ -1894,7 +1894,9 @@ static bool runAnalysis(SgProject &project, const int curr_regime, const bool ne
     else if (curr_regime == RENAME_SYMBOLS)
         runRenameSymbols(&project, commonBlocks);
     else if (curr_regime == FIND_PARAMETERS)
-        parametersOfProject = findParameters(defUseByFunctions, commonBlocks, allFuncInfo);
+    {
+        findParameters(parametersOfProject, fullIR, declaredArrays);
+    }
     else if (curr_regime == BUILD_IR)
     {
         auto CFG_forFile = buildCFG(commonBlocks, allFuncInfo_IR, SAPFOR::CFG_Settings(0));
@@ -2372,6 +2374,7 @@ void runPass(const int curr_regime, const char *proj_name, const char *folderNam
     case LOOPS_COMBINER:
     case FIX_COMMON_BLOCKS:
     case TEST_PASS:
+    case FIND_PARAMETERS:
     case SET_IMPLICIT_NONE:
         runAnalysis(*project, curr_regime, false);
     case SUBST_EXPR_RD_AND_UNPARSE:

src/SapforData.h

@@ -168,7 +168,7 @@ std::map<int, UserFiles> filesInfo; // information about open,close,write and re
 //
 
 //for FIND_PARAMETERS
-std::map< std::pair<std::string, int>, std::set<std::string>> parametersOfProject; // [file, line] -> set[vars]
+std::set<std::tuple<std::string, int, std::string>> parametersOfProject; // [file, line, varname]
 //
 
 //for GET_MIN_MAX_BLOCK_DIST

src/Utils/utils.cpp

@@ -1415,6 +1415,35 @@ ParallelRegion* getRegionByLine(const vector<ParallelRegion*>& regions, const st
     return NULL;
 }
 
+std::pair<ParallelRegion*, const ParallelRegionLines*> getRegionAndLinesByLine(const vector<ParallelRegion*>& regions, const string& file, const int line)
+{
+    if (regions.size() == 1 && regions[0]->GetName() == "DEFAULT") // only default
+        return { regions[0], NULL };
+    else if (regions.size() > 0)
+    {
+        map<ParallelRegion*, const ParallelRegionLines*> regFound;
+        const ParallelRegionLines* foundLines = NULL;
+
+        for (int i = 0; i < regions.size(); ++i)
+            if (regions[i]->HasThisLine(line, file, &foundLines))
+                regFound[regions[i]] = foundLines;
+
+        if (regFound.size() == 0)
+            return { NULL, NULL };
+        else if (regFound.size() == 1)
+            return *regFound.begin();
+        else
+        {
+            __spf_print(1, "WARN: this lines included in more than one region!!\n");
+            return { NULL, NULL };
+        }
+    }
+    else
+        return { NULL, NULL };
+
+    return { NULL, NULL };
+}
+
 set<ParallelRegion*> getAllRegionsByLine(const vector<ParallelRegion*>& regions, const string& file, const int line)
 {
     set<ParallelRegion*> regFound;

tests/sapfor/parameter/dynamic_array_maximum.f90

@@ -0,0 +1,52 @@
+program dynamic_array_maximum_3d
+    implicit none
+    integer :: n1, n2, n3, n4 , k, i, j, l, a
+    integer :: sum3
+    real :: max_element
+    real, allocatable :: array(:,:,:), array2(:,:,:), array3(:,:,:)
+
+    write(*, *) "Enter 3 integers"
+    read(*, *) n, m, k
+    m = 100
+
+    if (1 .eq. 1) then
+        a = 3
+    else if (2 .eq. 1) then
+        a = 4
+    endif
+
+    m = m + 1
+    k = m * 1000 + n * 10
+
+    allocate(array(n, m + n, k + m + n), &
+            &array2(k, m + n, k), &
+            &array3(k, m, k + n))
+
+    call random_seed()
+    do i = 1, n1
+        do j = 1, m * n1
+            do l = 1, k * m * n1
+                call random_number(array(i,j,l))
+                array(i,j,l) = int(array(i,j,l) * 100)
+            end do
+        end do
+    end do
+
+    max_element = array(1,1,1)
+    do i = 1, n1
+        do j = 1, m
+            do l = 1, k
+                max_element = MAX(array(i,j,l), max_element)
+            end do
+        end do
+    end do
+    deallocate(array, array2, array3)
+    write(*, *) max_element
+end program dynamic_array_maximum_3d
+
+! function sum3(x, y, z)
+!  implicit none
+!  integer :: x, y, z
+!  integer :: sum3
+!  sum3 = x + y + z
+! end function sum3