REMOVE_DIST_ARRAYS_FROM_IO: regard generated intent statements, carefully detect assumed-shape arrays

src/Transformations/ReplaceArraysInIO/replace_dist_arrays_in_io.cpp

@@ -17,6 +17,159 @@ using std::pair;
 
 #define DEBUG_TRACE 0
 
+static SgStatement* getDeclStatement(int line_num)
+{
+    PTR_BFND node = current_file->firstStatement()->thebif;
+    for (; node; node = node->thread)
+    {
+        SgStatement *st = BfndMapping(node);
+        if (st->lineNumber() == line_num && isSgDeclarationStatement(st))
+            return st;
+    }
+
+    return NULL;
+}
+
+static SgExpression* findExprWithVariant(SgExpression* exp, int variant)
+{
+    if (exp)
+    {
+        if (exp->variant() == variant)
+            return exp;
+
+        auto *l = findExprWithVariant(exp->lhs(), variant);
+        if (l)
+            return l;
+
+        auto *r = findExprWithVariant(exp->rhs(), variant);
+        if (r)
+            return r;
+    }
+
+    return NULL;
+}
+
+bool switchToDeclarationFile(DIST::Array* array_p, 
+                             pair<string, int>& decl_place, 
+                             const string &current_file_name,
+                             FuncInfo *current_func)
+{
+    if (!array_p)
+        return false;
+
+    // try to find declaration from current function
+    for (const auto &p : array_p->GetDeclInfo())
+    {
+        if (p.first == current_file_name && 
+            current_func->linesNum.first <= p.second && 
+            p.second <= current_func->linesNum.second)
+        {
+            decl_place = p;
+            return true;
+        }
+    }
+
+    for (const auto &p : array_p->GetDeclInfo())
+    {
+        if (SgFile::switchToFile(p.first) != -1)
+        {
+            decl_place = p;
+            return true;
+        }
+    }
+
+    return false;
+}
+
+static bool checkAssumedShape(SgStatement* decl, const string& array_name)
+{
+    SgExpression* list = decl->expr(0);
+    SgExpression* dim_list = NULL;
+
+    while (list)
+    {
+        auto *arr_ref = list->lhs();
+
+        if (arr_ref && 
+            arr_ref->symbol() && 
+            arr_ref->symbol()->identifier() && 
+            arr_ref->symbol()->identifier() == array_name)
+        {
+            dim_list = arr_ref->lhs();
+            break;
+        }
+
+        list = list->rhs();
+    }
+
+    if (!dim_list)
+    {
+        auto *dim_expr = findExprWithVariant(decl->expr(2), DIMENSION_OP);
+        if (dim_expr)
+            dim_list = dim_expr->lhs();
+    }
+
+    while (dim_list)
+    {
+        auto *lhs = dim_list->lhs();
+
+        if (lhs && lhs->variant() == DDOT && (!lhs->lhs() || !lhs->rhs()))
+            return true;
+
+        dim_list = dim_list->rhs();
+    }
+
+    return false;
+}
+
+static bool checkAssumedSize(const string &array_name, DIST::Array* array_p, const string &current_file_name, FuncInfo *current_func)
+{
+    if (!array_p)
+        return true; // raise true to prevent array from copying
+
+    bool found = false;
+
+    pair<string, int> decl_place;
+
+    if (!switchToDeclarationFile(array_p, decl_place, current_file_name, current_func))
+        return true;
+
+    auto *st = getDeclStatement(decl_place.second);
+
+    if (!st)
+        printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+    
+    SgExpression* list = st->expr(0);    
+    while (list)
+    {
+        if (list->lhs() && 
+            list->lhs()->symbol() && 
+            list->lhs()->symbol()->identifier() && 
+            list->lhs()->symbol()->identifier() == array_name
+        )
+        {
+            if(findExprWithVariant(list->lhs(), STAR_RANGE))
+                found = true;
+
+            break;
+        }
+
+        list = list->rhs();
+    }
+
+    if (!found)
+    {
+        auto *dim_expr = findExprWithVariant(st->expr(2), DIMENSION_OP);
+        if (dim_expr && findExprWithVariant(dim_expr, STAR_RANGE))
+            found = true;
+    }
+
+    if (SgFile::switchToFile(current_file_name) == -1)
+        printInternalError(convertFileName(__FILE__).c_str(), __LINE__); // can't switch back to file with array usage
+
+    return found;
+}
+
 static void findArrays(SgExpression* exp, set<SgSymbol*>& arrays)
 {
     if (exp)
@@ -29,7 +182,10 @@ static void findArrays(SgExpression* exp, set<SgSymbol*>& arrays)
     }
 }
 
-static void populateDistributedIoArrays(map<SgSymbol*, set<SgStatement*>>& arrays, SgStatement* stat)
+static void populateDistributedIoArrays(map<SgSymbol*, set<SgStatement*>>& arrays, 
+                                        SgStatement* stat, 
+                                        const string& current_file_name,
+                                        FuncInfo *current_func)
 {
     auto var = stat->variant();
 
@@ -130,8 +286,16 @@ static void populateDistributedIoArrays(map<SgSymbol*, set<SgStatement*>>& array
     {
         string array_name = string(by_symb->identifier());
         DIST::Array* array_p = getArrayFromDeclarated(declaratedInStmt(by_symb), array_name);
-        if (array_p && array_p->GetDistributeFlagVal() == Distribution::distFlag::IO_PRIV && arrays[by_symb].insert(stat).second)
+        if (array_p && 
-            __spf_print(DEBUG_TRACE, "[%d]: add array %s\n", stat->lineNumber(), array_p->GetName().c_str());
+            array_p->GetDistributeFlagVal() == Distribution::distFlag::IO_PRIV && 
+            !checkAssumedSize(array_name, array_p, current_file_name, current_func)
+        )
+        {
+            auto inserted = arrays[by_symb].insert(stat).second;
+
+            if (inserted)
+                __spf_print(DEBUG_TRACE, "[%d]: add array %s %p\n", stat->lineNumber(), array_p->GetName().c_str(), by_symb);
+        }
     }
 
     __spf_print(DEBUG_TRACE, "[replace]\n");
@@ -204,7 +368,12 @@ static void replaceArrayRec(SgSymbol* arr, SgSymbol* replace_by, SgStatement* st
     }
 }
 
-static void copyArrayBetweenStatements(SgSymbol* replace_symb, SgSymbol* replace_by, SgStatement* start, SgStatement* last, bool start_is_scope)
+static void copyArrayBetweenStatements(SgSymbol* replace_symb, 
+                                       SgSymbol* replace_by,
+                                       SgStatement* start, 
+                                       SgStatement* last, 
+                                       bool start_is_scope,
+                                       FuncInfo *func_info)
 {
     while (start->lexNext() && !isSgExecutableStatement(start->lexNext()))
         start = start->lexNext();
@@ -227,6 +396,18 @@ static void copyArrayBetweenStatements(SgSymbol* replace_symb, SgSymbol* replace
         start->insertStmtAfter(*assign, *parent);
     }
 
+    if (has_write)
+    {
+        for (int i = 0; i < func_info->funcParams.identificators.size(); i++)
+        {
+            if (func_info->funcParams.identificators[i] == replace_symb->identifier())
+            {
+                has_write &= func_info->funcParams.isArgOut(i);
+                break;
+            }
+        }
+    }
+
     if (has_write)
     {
         // A = A_reg
@@ -237,9 +418,14 @@ static void copyArrayBetweenStatements(SgSymbol* replace_symb, SgSymbol* replace
     }
 }
 
-static void replaceArrayInFragment(SgSymbol* replace_symb, const set<SgStatement*> usages, SgSymbol* replace_by, SgStatement* start, SgStatement* last, const string& filename)
+static void replaceArrayInFragment(SgSymbol* replace_symb, 
+                                   const set<SgStatement*> usages, 
+                                   SgSymbol* replace_by, 
+                                   SgStatement* start, 
+                                   SgStatement* last, 
+                                   FuncInfo *func_info)
 {
-    while (start->lexNext() && !isSgExecutableStatement(start->lexNext()))
+    while (start->lexNext() && (!isSgExecutableStatement(start->lexNext()) || start->lexNext()->variant() == ALLOCATE_STMT))
         start = start->lexNext();
 
     set<SgStatement*> not_opened, not_closed, copied;
@@ -306,7 +492,7 @@ static void replaceArrayInFragment(SgSymbol* replace_symb, const set<SgStatement
         }
 
         __spf_print(DEBUG_TRACE, "[copy %s] [%d, %d]\n", replace_symb->identifier(), scope_start->lineNumber(), scope_end->lineNumber());
-        copyArrayBetweenStatements(replace_symb, replace_by, scope_start, scope_end, copy_scope == scope_start);
+        copyArrayBetweenStatements(replace_symb, replace_by, scope_start, scope_end, copy_scope == scope_start, func_info);
         copied.insert(copy_scope);
     }
 }
@@ -347,9 +533,9 @@ static bool ioReginBorder(SgStatement* stat, SgStatement* last_io_bound)
 }
 
 void replaceDistributedArraysInIO(vector<ParallelRegion*>& regions,
-    const map<string, vector<FuncInfo*>>& allFuncInfo,
+                                  const map<string, vector<FuncInfo*>>& allFuncInfo,
-    map<string, vector<Messages>>& SPF_messages,
+                                  map<string, vector<Messages>>& SPF_messages,
-    map<string, map<int, set<string>>>& newDeclsToInclude)
+                                  map<string, map<int, set<string>>>& newDeclsToInclude)
 {
     map<SgSymbol*, SgSymbol*> created_copies;
     map<string, map<int, set<string>>> copied_syms;
@@ -358,23 +544,32 @@ void replaceDistributedArraysInIO(vector<ParallelRegion*>& regions,
     {
         __spf_print(DEBUG_TRACE, "[%s]: enter region\n", region->GetName().c_str());
 
-        for (auto& linesByFile : region->GetAllLinesToModify())
+        for (auto& lines_by_file : region->GetAllLinesToModify())
         {
-            const auto& filename = linesByFile.first;
+            const auto& current_file_name = lines_by_file.first;
 
-            if (SgFile::switchToFile(filename) == -1)
+            if (SgFile::switchToFile(current_file_name) == -1)
                 printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
 
-            for (auto& lines : linesByFile.second)
+            auto func_info_it = allFuncInfo.find(current_file_name);
+            if (func_info_it == allFuncInfo.end())
             {
-                __spf_print(DEBUG_TRACE, "[fragment] %s: %d:%d %d\n", filename.c_str(), lines.lines.first,
+                printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+                return;
+            }
+
+            auto *lbound_symb = new SgSymbol(PROCEDURE_NAME, "lbound");
+            auto *ubound_symb = new SgSymbol(PROCEDURE_NAME, "ubound");
+
+            for (auto& lines : lines_by_file.second)
+            {
+                __spf_print(DEBUG_TRACE, "[fragment] %s: %d:%d %d\n", current_file_name.c_str(), lines.lines.first,
                     lines.lines.second, lines.isImplicit());
 
                 SgStatement* curr_stmt, * end;
 
                 if (lines.isImplicit())
                 {
-
                     curr_stmt = current_file->SgStatementAtLine(lines.lines.first);
                     end = current_file->SgStatementAtLine(lines.lines.second);
 
@@ -390,6 +585,7 @@ void replaceDistributedArraysInIO(vector<ParallelRegion*>& regions,
                 map<SgSymbol*, set<SgStatement*>> need_replace;
                 SgStatement* last_io_bound = NULL;
 
+                FuncInfo *current_func_info = NULL;
 
                 while (curr_stmt != end)
                 {
@@ -398,9 +594,24 @@ void replaceDistributedArraysInIO(vector<ParallelRegion*>& regions,
 
                     auto var = curr_stmt->variant();
 
-
                     if (var == PROC_HEDR || var == PROG_HEDR || var == FUNC_HEDR)
                     {
+                        current_func_info = NULL;
+                        for (auto *func_info : func_info_it->second)
+                        {
+                            if (func_info->funcName == curr_stmt->symbol()->identifier())
+                            {
+                                current_func_info = func_info;
+                                break;
+                            }
+                        }
+
+                        if (!current_func_info)
+                        {
+                            printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+                            return;
+                        }    
+
                         curr_stmt = curr_stmt->lexNext();
                         while (curr_stmt && !isSgExecutableStatement(curr_stmt))
                         {
@@ -424,77 +635,191 @@ void replaceDistributedArraysInIO(vector<ParallelRegion*>& regions,
                                 string array_name = string(array_to_copy->identifier());
                                 DIST::Array* array_p = getArrayFromDeclarated(declaratedInStmt(array_to_copy), array_name);
 
-                                bool fromModule = (array_p->GetLocation().first == DIST::l_MODULE);
+                                // at this point all considered arrays must have DIST::Array* references
-                                const string locationName = array_p->GetLocation().second;
+                                if (!array_p)
+                                    printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+
+                                // ... and be declared in switchable files (not in headers)
+                                pair<string, int> decl_place;
+                                if (!switchToDeclarationFile(array_p, decl_place, current_file_name, current_func_info))
+                                    printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
 
-                                auto place = *array_p->GetDeclInfo().begin();
-                                string fileName = place.first;
                                 string suffix = "_io_l";
 
-                                if (fromModule)
+                                switch (array_p->GetLocation().first)
-                                    suffix = "_io_m";
-
-                                pair<SgSymbol*, SgSymbol*> copied;
-                                copied.first = array_to_copy;
-
-                                if (SgFile::switchToFile(fileName) == -1)
                                 {
-                                    auto* func_stmt = curr_stmt->getScopeForDeclare();
+                                    case DIST::l_MODULE:
+                                        suffix = "_io_m";
+                                        break;
+                                    case DIST::l_COMMON:
+                                        suffix = "_io_c";
+                                        break;
+                                }
                                 
-                                    SgStatement* insertPlace = NULL;
+                                auto copied_symbol = copyArray(decl_place, 
+                                                               array_p, 
+                                                               lines_by_file.second, 
+                                                               suffix + to_string(region->GetId()), 
+                                                               decl_place.first, 
+                                                               newDeclsToInclude, 
+                                                               copied_syms);
+                            
+
+                                // original declaration statement
+                                auto* decl_stmt = getDeclStatement(decl_place.second);
+                                if (!decl_stmt || !isSgDeclarationStatement(decl_stmt))
+                                    printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+
+                                auto dynamic_array = checkAssumedShape(decl_stmt, array_name);  
+
+                                while (decl_stmt && !isSgExecutableStatement(decl_stmt))
+                                {
+                                    if (isSgDeclarationStatement(decl_stmt) &&
+                                        decl_stmt->expr(0) &&
+                                        decl_stmt->expr(0)->lhs() &&
+                                        decl_stmt->expr(0)->lhs()->symbol() == copied_symbol.second)
+                                    {
+                                        break;
+                                    }
+
+                                    decl_stmt = decl_stmt->lexNext();
+                                }
+                                
+                                // created declaration statement
+                                if (!decl_stmt || !isSgDeclarationStatement(decl_stmt))
+                                    printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
+
+                                // insert !$SPF ANALYSIS(PROCESS_PRIVATE(array)) directive before declaration statement
+                                
+                                string dir_str;
+                                if (decl_stmt->comments())
+                                {
+                                    auto str_comment = string(decl_stmt->comments());
+                                    if (str_comment.size() && str_comment.back() != '\n')
+                                        dir_str += "\n";
+                                }
+
+                                dir_str += "!$SPF ANALYSIS(PROCESS_PRIVATE(" + string(copied_symbol.second->identifier()) + "))\n";
+                                decl_stmt->addComment(dir_str.c_str());
+                            
+                                created_copies.insert({ array_to_copy, copied_symbol.second });
+
+                                // make array copy allocatable in case of assumed-shape array
+                                if(dynamic_array)
+                                {
+                                    // insert allocatable keyword in declaration
+                                    auto *kword_list = decl_stmt->expr(2);
+                                    if (!findExprWithVariant(kword_list, ALLOCATABLE_OP))
+                                    {
+                                        if (!kword_list)
+                                        {
+                                            kword_list = new SgExprListExp();
+                                            decl_stmt->setExpression(2, *kword_list);
+                                        }
+                                        
+                                        while (kword_list->rhs())
+                                            kword_list = kword_list->rhs();
+                                        
+                                        if (kword_list->lhs())
+                                        {
+                                            kword_list->setRhs(new SgExprListExp());
+                                            kword_list = kword_list->rhs();
+                                        }
+
+                                        kword_list->setLhs(new SgExpression(ALLOCATABLE_OP));
+                                    }
+
+                                    // can't switch back to file with array usage
+                                    if (SgFile::switchToFile(current_file_name) == -1)
+                                        printInternalError(convertFileName(__FILE__).c_str(), __LINE__); 
+                                        
+                                    // insert allocate(a_l(lbound(a, 1):ubound(a,1),...)) statement
+                                    SgStatement* allocate_stmt_place = NULL;
+
+                                    auto* func_stmt = curr_stmt->getScopeForDeclare();
                                     for (auto iterator = func_stmt->lexNext();
-                                        !isSgExecutableStatement(iterator) || isSPF_stat(iterator) &&
+                                        iterator && !isSgExecutableStatement(iterator);
-                                        !(iterator->variant() == SPF_PARALLEL_REG_DIR || iterator->variant() == SPF_END_PARALLEL_REG_DIR);
                                         iterator = iterator->lexNext())
                                     {
-                                        insertPlace = iterator;
+                                            allocate_stmt_place = iterator;
                                     }
 
                                     //NULL - no decl stats in function!
-                                    if (!insertPlace)
+                                    if (!allocate_stmt_place)
-                                        insertPlace = func_stmt;
+                                        allocate_stmt_place = func_stmt;
 
-                                    auto st = insertPlace->controlParent();
+                                    auto *allocate_stmt_parent = allocate_stmt_place->controlParent();
-                                    if (st->variant() == GLOBAL)
+                                    if (allocate_stmt_parent->variant() == GLOBAL)
-                                        st = insertPlace;
+                                        allocate_stmt_parent = allocate_stmt_place;
                                     
-                                    auto& copied_symb = array_to_copy->copy();
+                                    auto *created_array_ref = new SgArrayRefExp(*copied_symbol.second);
-                                    copied.second = &copied_symb;
 
-                                    auto new_name = string(array_to_copy->identifier()) + "_io_c";
+                                    auto* dim_list = new SgExprListExp();
-                                    copied_symb.changeName(new_name.c_str());
+                                    created_array_ref->setLhs(dim_list);
                                     
-                                    auto stat = array_to_copy->makeVarDeclStmt();
+                                    int dim_len = array_p->GetSizes().size();
-                                    auto res = CalculateInteger(stat->expr(0)->copyPtr());
+                                    for (int i = 1; i <= dim_len; i++)
-                                    res->lhs()->setSymbol(copied_symb);
-                                    stat->setExpression(0, res);
-
-                                    insertPlace->insertStmtAfter(*stat, *st);
-                                }
-                                else
-                                    copied = copyArray(place, array_p, linesByFile.second, suffix + to_string(region->GetId()), fileName, newDeclsToInclude, copied_syms);
-
-                                SgStatement* decl = SgStatement::getStatementByFileAndLine(place.first, place.second);
-
-                                if (decl)
-                                    decl = decl->lexNext();
-
-                                if (decl)
-                                {
-                                    string dir_str;
-                                    if (decl->comments())
                                     {
-                                        string str_comment = string(decl->comments());
+                                        auto *lcall = new SgFunctionCallExp(*lbound_symb);
-                                        if (str_comment.size() && str_comment.back() != '\n')
+                                        auto *rcall = new SgFunctionCallExp(*ubound_symb);
-                                            dir_str += "\n";
+                                        
+                                        for (auto *call : {lcall, rcall})
+                                        {
+                                            call->setLhs(new SgExprListExp());
+
+                                            call->lhs()->setLhs(new SgArrayRefExp(*array_to_copy));
+                                            call->lhs()->setRhs(new SgValueExp(i));
+                                        }
+                                        
+                                        auto *dot_expr = new SgExpression(DDOT);
+                                        dot_expr->setLhs(lcall);
+                                        dot_expr->setRhs(rcall);
+
+                                        dim_list->setLhs(dot_expr);
+
+                                        if (i < dim_len)
+                                        {
+                                            auto *next = new SgExprListExp();
+                                            dim_list->setRhs(next);
+                                            dim_list = next;
+                                        }
                                     }
 
-                                    dir_str += "!$SPF ANALYSIS(PROCESS_PRIVATE(" + string(copied.second->identifier()) + "))\n";
+                                    auto *allocate_stmt = new SgStatement(ALLOCATE_STMT);
-                                    decl->addComment(dir_str.c_str());
+                                    allocate_stmt->setExpression(0, created_array_ref);
-                                }
-                                created_copies.insert({ array_to_copy, copied.second });
 
-                                if (curr_stmt->switchToFile() == -1)
+                                    allocate_stmt_place->insertStmtAfter(*allocate_stmt, *allocate_stmt_parent);
+
+                                    // insert deallocate statemens before all returns
+                                    auto *find_return_stmt = func_stmt;
+                                    while (find_return_stmt != func_stmt->lastNodeOfStmt())
+                                    {
+                                        auto *next = find_return_stmt->lexNext();
+                                        if (next && (isSgReturnStmt(next) || next == func_stmt->lastNodeOfStmt()))
+                                        {
+                                            if (next->hasLabel())
+                                            {
+                                                moveLabelBefore(next);
+                                                find_return_stmt = next->lexPrev();
+                                            }
+                                            
+                                            auto *dealloc_stmt = new SgStatement(DEALLOCATE_STMT);
+
+                                            dealloc_stmt->setExpression(0, new SgExprListExp());
+                                            dealloc_stmt->expr(0)->setLhs(new SgArrayRefExp(*copied_symbol.second));
+
+                                            find_return_stmt->insertStmtAfter(*dealloc_stmt, *next->controlParent());
+
+                                            if (next == curr_stmt)
+                                                curr_stmt = dealloc_stmt;
+                                        }
+
+                                        find_return_stmt = next;
+                                    }
+                                }
+
+                                // can't switch back to file with array usage
+                                if (SgFile::switchToFile(current_file_name) == -1)
                                     printInternalError(convertFileName(__FILE__).c_str(), __LINE__); 
                             }
                         }
@@ -507,7 +832,7 @@ void replaceDistributedArraysInIO(vector<ParallelRegion*>& regions,
                             {
                                 auto it = created_copies.find(p.first);
                                 if (it != created_copies.end())
-                                    replaceArrayInFragment(p.first, p.second, it->second, last_io_bound, curr_stmt, filename);
+                                    replaceArrayInFragment(p.first, p.second, it->second, last_io_bound, curr_stmt, current_func_info);
                                 else
                                     printInternalError(convertFileName(__FILE__).c_str(), __LINE__);
                             }
@@ -527,7 +852,7 @@ void replaceDistributedArraysInIO(vector<ParallelRegion*>& regions,
                         }
                     }
 
-                    populateDistributedIoArrays(need_replace, curr_stmt);
+                    populateDistributedIoArrays(need_replace, curr_stmt, current_file_name, current_func_info);
                     curr_stmt = curr_stmt->lexNext();
                 }
             }