ppcg_scop_from_pet_scop: abort on data dependent conditions

[ppcg.git] / cpu.c

/*
 * Copyright 2012 INRIA Paris-Rocquencourt
 *
 * Use of this software is governed by the MIT license
 *
 * Written by Tobias Grosser, INRIA Paris-Rocquencourt,
 * Domaine de Voluceau, Rocquenqourt, B.P. 105,
 * 78153 Le Chesnay Cedex France
 */

#include <limits.h>
#include <stdio.h>
#include <string.h>

#include <isl/aff.h>
#include <isl/ctx.h>
#include <isl/map.h>
#include <isl/ast_build.h>
#include <pet.h>

#include "ppcg.h"
#include "ppcg_options.h"
#include "cpu.h"
#include "pet_printer.h"
#include "print.h"
#include "rewrite.h"

/* Representation of a statement inside a generated AST.
 *
 * "stmt" refers to the original statement.
 * "n_access" is the number of accesses in the statement.
 * "access" is the list of accesses transformed to refer to the iterators
 * in the generated AST.
 */
struct ppcg_stmt {
        struct pet_stmt *stmt;

        int n_access;
        isl_ast_expr_list **access;
};

static void ppcg_stmt_free(void *user)
{
        struct ppcg_stmt *stmt = user;
        int i;

        if (!stmt)
                return;

        for (i = 0; i < stmt->n_access; ++i)
                isl_ast_expr_list_free(stmt->access[i]);

        free(stmt->access);
        free(stmt);
}

/* Derive the output file name from the input file name.
 * 'input' is the entire path of the input file. The output
 * is the file name plus the additional extension.
 *
 * We will basically replace everything after the last point
 * with '.ppcg.c'. This means file.c becomes file.ppcg.c
 */
static FILE *get_output_file(const char *input, const char *output)
{
        char name[PATH_MAX];
        const char *base;
        const char *ext;
        const char ppcg_marker[] = ".ppcg";
        int len;

        base = strrchr(input, '/');
        if (base)
                base++;
        else
                base = input;
        ext = strrchr(base, '.');
        len = ext ? ext - base : strlen(base);

        memcpy(name, base, len);
        strcpy(name + len, ppcg_marker);
        strcpy(name + len + sizeof(ppcg_marker) - 1, ext ? ext : ".c");

        if (!output)
                output = name;

        return fopen(output, "w");
}

/* Data used to annotate for nodes in the ast.
 */
struct ast_node_userinfo {
        /* The for node is an openmp parallel for node. */
        int is_openmp;
};

/* Information used while building the ast.
 */
struct ast_build_userinfo {
        /* The current ppcg scop. */
        struct ppcg_scop *scop;

        /* Are we currently in a parallel for loop? */
        int in_parallel_for;
};

/* Check if the current scheduling dimension is parallel.
 *
 * We check for parallelism by verifying that the loop does not carry any
 * dependences.
 *
 * Parallelism test: if the distance is zero in all outer dimensions, then it
 * has to be zero in the current dimension as well.
 * Implementation: first, translate dependences into time space, then force
 * outer dimensions to be equal.  If the distance is zero in the current
 * dimension, then the loop is parallel.
 * The distance is zero in the current dimension if it is a subset of a map
 * with equal values for the current dimension.
 */
static int ast_schedule_dim_is_parallel(__isl_keep isl_ast_build *build,
        struct ppcg_scop *scop)
{
        isl_union_map *schedule_node, *schedule, *deps;
        isl_map *schedule_deps, *test;
        isl_space *schedule_space;
        unsigned i, dimension, is_parallel;

        schedule = isl_ast_build_get_schedule(build);
        schedule_space = isl_ast_build_get_schedule_space(build);

        dimension = isl_space_dim(schedule_space, isl_dim_out) - 1;

        deps = isl_union_map_copy(scop->dep_flow);
        deps = isl_union_map_union(deps, isl_union_map_copy(scop->dep_false));
        deps = isl_union_map_apply_range(deps, isl_union_map_copy(schedule));
        deps = isl_union_map_apply_domain(deps, schedule);

        if (isl_union_map_is_empty(deps)) {
                isl_union_map_free(deps);
                isl_space_free(schedule_space);
                return 1;
        }

        schedule_deps = isl_map_from_union_map(deps);

        for (i = 0; i < dimension; i++)
                schedule_deps = isl_map_equate(schedule_deps, isl_dim_out, i,
                                               isl_dim_in, i);

        test = isl_map_universe(isl_map_get_space(schedule_deps));
        test = isl_map_equate(test, isl_dim_out, dimension, isl_dim_in,
                              dimension);
        is_parallel = isl_map_is_subset(schedule_deps, test);

        isl_space_free(schedule_space);
        isl_map_free(test);
        isl_map_free(schedule_deps);

        return is_parallel;
}

/* Mark a for node openmp parallel, if it is the outermost parallel for node.
 */
static void mark_openmp_parallel(__isl_keep isl_ast_build *build,
        struct ast_build_userinfo *build_info,
        struct ast_node_userinfo *node_info)
{
        if (build_info->in_parallel_for)
                return;

        if (ast_schedule_dim_is_parallel(build, build_info->scop)) {
                build_info->in_parallel_for = 1;
                node_info->is_openmp = 1;
        }
}

/* Allocate an ast_node_info structure and initialize it with default values.
 */
static struct ast_node_userinfo *allocate_ast_node_userinfo()
{
        struct ast_node_userinfo *node_info;
        node_info = (struct ast_node_userinfo *)
                malloc(sizeof(struct ast_node_userinfo));
        node_info->is_openmp = 0;
        return node_info;
}

/* Free an ast_node_info structure.
 */
static void free_ast_node_userinfo(void *ptr)
{
        struct ast_node_userinfo *info;
        info = (struct ast_node_userinfo *) ptr;
        free(info);
}

/* This method is executed before the construction of a for node. It creates
 * an isl_id that is used to annotate the subsequently generated ast for nodes.
 *
 * In this function we also run the following analyses:
 *
 *      - Detection of openmp parallel loops
 */
static __isl_give isl_id *ast_build_before_for(
        __isl_keep isl_ast_build *build, void *user)
{
        isl_id *id;
        struct ast_build_userinfo *build_info;
        struct ast_node_userinfo *node_info;

        build_info = (struct ast_build_userinfo *) user;
        node_info = allocate_ast_node_userinfo();
        id = isl_id_alloc(isl_ast_build_get_ctx(build), "", node_info);
        id = isl_id_set_free_user(id, free_ast_node_userinfo);

        mark_openmp_parallel(build, build_info, node_info);

        return id;
}

/* This method is executed after the construction of a for node.
 *
 * It performs the following actions:
 *
 *      - Reset the 'in_parallel_for' flag, as soon as we leave a for node,
 *        that is marked as openmp parallel.
 *
 */
static __isl_give isl_ast_node *ast_build_after_for(__isl_take isl_ast_node *node,
        __isl_keep isl_ast_build *build, void *user) {
        isl_id *id;
        struct ast_build_userinfo *build_info;
        struct ast_node_userinfo *info;

        id = isl_ast_node_get_annotation(node);
        info = isl_id_get_user(id);

        if (info && info->is_openmp) {
                build_info = (struct ast_build_userinfo *) user;
                build_info->in_parallel_for = 0;
        }

        isl_id_free(id);

        return node;
}

/* Print a memory access 'access' to the printer 'p'.
 *
 * "expr" refers to the original access.
 * "access" is the list of index expressions transformed to refer
 * to the iterators of the generated AST.
 *
 * In case the original access is unnamed (and presumably single-dimensional),
 * we assume this is not a memory access, but just an expression.
 */
static __isl_give isl_printer *print_access(__isl_take isl_printer *p,
        struct pet_expr *expr, __isl_keep isl_ast_expr_list *access)
{
        int i;
        const char *name;
        unsigned n_index;

        n_index = isl_ast_expr_list_n_ast_expr(access);
        name = isl_map_get_tuple_name(expr->acc.access, isl_dim_out);

        if (name == NULL) {
                isl_ast_expr *index;
                index = isl_ast_expr_list_get_ast_expr(access, 0);
                p = isl_printer_print_str(p, "(");
                p = isl_printer_print_ast_expr(p, index);
                p = isl_printer_print_str(p, ")");
                isl_ast_expr_free(index);
                return p;
        }

        p = isl_printer_print_str(p, name);

        for (i = 0; i < n_index; ++i) {
                isl_ast_expr *index;

                index = isl_ast_expr_list_get_ast_expr(access, i);

                p = isl_printer_print_str(p, "[");
                p = isl_printer_print_ast_expr(p, index);
                p = isl_printer_print_str(p, "]");
                isl_ast_expr_free(index);
        }

        return p;
}

/* Find the element in scop->stmts that has the given "id".
 */
static struct pet_stmt *find_stmt(struct ppcg_scop *scop, __isl_keep isl_id *id)
{
        int i;

        for (i = 0; i < scop->n_stmt; ++i) {
                struct pet_stmt *stmt = scop->stmts[i];
                isl_id *id_i;

                id_i = isl_set_get_tuple_id(stmt->domain);
                isl_id_free(id_i);

                if (id_i == id)
                        return stmt;
        }

        isl_die(isl_id_get_ctx(id), isl_error_internal,
                "statement not found", return NULL);
}

/* To print the transformed accesses we walk the list of transformed accesses
 * simultaneously with the pet printer. This means that whenever
 * the pet printer prints a pet access expression we have
 * the corresponding transformed access available for printing.
 */
static __isl_give isl_printer *print_access_expr(__isl_take isl_printer *p,
        struct pet_expr *expr, void *user)
{
        isl_ast_expr_list ***access = user;

        p = print_access(p, expr, **access);
        (*access)++;

        return p;
}

/* Print a user statement in the generated AST.
 * The ppcg_stmt has been attached to the node in at_each_domain.
 */
static __isl_give isl_printer *print_user(__isl_take isl_printer *p,
        __isl_take isl_ast_print_options *print_options,
        __isl_keep isl_ast_node *node, void *user)
{
        struct ppcg_stmt *stmt;
        isl_ast_expr_list **access;
        isl_id *id;

        id = isl_ast_node_get_annotation(node);
        stmt = isl_id_get_user(id);
        isl_id_free(id);

        access = stmt->access;

        p = isl_printer_start_line(p);
        p = print_pet_expr(p, stmt->stmt->body, &print_access_expr, &access);
        p = isl_printer_print_str(p, ";");
        p = isl_printer_end_line(p);

        isl_ast_print_options_free(print_options);

        return p;
}


/* Print a for loop node as an openmp parallel loop.
 *
 * To print an openmp parallel loop we print a normal for loop, but add
 * "#pragma openmp parallel for" in front.
 *
 * Variables that are declared within the body of this for loop are
 * automatically openmp 'private'. Iterators declared outside of the
 * for loop are automatically openmp 'shared'. As ppcg declares all iterators
 * at the position where they are assigned, there is no need to explicitly mark
 * variables. Their automatically assigned type is already correct.
 *
 * This function only generates valid OpenMP code, if the ast was generated
 * with the 'atomic-bounds' option enabled.
 *
 */
static __isl_give isl_printer *print_for_with_openmp(
        __isl_keep isl_ast_node *node, __isl_take isl_printer *p,
        __isl_take isl_ast_print_options *print_options)
{
        p = isl_printer_start_line(p);
        p = isl_printer_print_str(p, "#pragma omp parallel for");
        p = isl_printer_end_line(p);

        p = isl_ast_node_for_print(node, p, print_options);

        return p;
}

/* Print a for node.
 *
 * Depending on how the node is annotated, we either print a normal
 * for node or an openmp parallel for node.
 */
static __isl_give isl_printer *print_for(__isl_take isl_printer *p,
        __isl_take isl_ast_print_options *print_options,
        __isl_keep isl_ast_node *node, void *user)
{
        struct ppcg_print_info *print_info;
        isl_id *id;
        int openmp;

        openmp = 0;
        id = isl_ast_node_get_annotation(node);

        if (id) {
                struct ast_node_userinfo *info;

                info = (struct ast_node_userinfo *) isl_id_get_user(id);
                if (info && info->is_openmp)
                        openmp = 1;
        }

        if (openmp)
                p = print_for_with_openmp(node, p, print_options);
        else
                p = isl_ast_node_for_print(node, p, print_options);

        isl_id_free(id);

        return p;
}

/* Call "fn" on each access expression in "expr".
 */
static int foreach_access_expr(struct pet_expr *expr,
        int (*fn)(struct pet_expr *expr, void *user), void *user)
{
        int i;

        if (!expr)
                return -1;

        if (expr->type == pet_expr_access)
                return fn(expr, user);

        for (i = 0; i < expr->n_arg; ++i)
                if (foreach_access_expr(expr->args[i], fn, user) < 0)
                        return -1;

        return 0;
}

static int inc_n_access(struct pet_expr *expr, void *user)
{
        struct ppcg_stmt *stmt = user;
        stmt->n_access++;
        return 0;
}

/* Internal data for add_access.
 *
 * "stmt" is the statement to which an access needs to be added.
 * "build" is the current AST build.
 * "map" maps the AST loop iterators to the iteration domain of the statement.
 */
struct ppcg_add_access_data {
        struct ppcg_stmt *stmt;
        isl_ast_build *build;
        isl_map *map;
};

/* Given an access expression, add it to data->stmt after
 * transforming it to refer to the AST loop iterators.
 */
static int add_access(struct pet_expr *expr, void *user)
{
        int i, n;
        isl_ctx *ctx;
        isl_map *access;
        isl_pw_multi_aff *pma;
        struct ppcg_add_access_data *data = user;
        isl_ast_expr_list *index;

        ctx = isl_map_get_ctx(expr->acc.access);
        n = isl_map_dim(expr->acc.access, isl_dim_out);
        access = isl_map_copy(expr->acc.access);
        access = isl_map_apply_range(isl_map_copy(data->map), access);
        pma = isl_pw_multi_aff_from_map(access);
        pma = isl_pw_multi_aff_coalesce(pma);

        index = isl_ast_expr_list_alloc(ctx, n);
        for (i = 0; i < n; ++i) {
                isl_pw_aff *pa;
                isl_ast_expr *expr;

                pa = isl_pw_multi_aff_get_pw_aff(pma, i);
                expr = isl_ast_build_expr_from_pw_aff(data->build, pa);
                index = isl_ast_expr_list_add(index, expr);
        }
        isl_pw_multi_aff_free(pma);

        data->stmt->access[data->stmt->n_access] = index;
        data->stmt->n_access++;
        return 0;
}

/* Transform the accesses in the statement associated to the domain
 * called by "node" to refer to the AST loop iterators,
 * collect them in a ppcg_stmt and annotate the node with the ppcg_stmt.
 */
static __isl_give isl_ast_node *at_each_domain(__isl_take isl_ast_node *node,
        __isl_keep isl_ast_build *build, void *user)
{
        struct ppcg_scop *scop = user;
        isl_ast_expr *expr, *arg;
        isl_ctx *ctx;
        isl_id *id;
        isl_map *map;
        struct ppcg_stmt *stmt;
        struct ppcg_add_access_data data;

        ctx = isl_ast_node_get_ctx(node);
        stmt = isl_calloc_type(ctx, struct ppcg_stmt);
        if (!stmt)
                goto error;

        expr = isl_ast_node_user_get_expr(node);
        arg = isl_ast_expr_get_op_arg(expr, 0);
        isl_ast_expr_free(expr);
        id = isl_ast_expr_get_id(arg);
        isl_ast_expr_free(arg);
        stmt->stmt = find_stmt(scop, id);
        isl_id_free(id);
        if (!stmt->stmt)
                goto error;

        stmt->n_access = 0;
        if (foreach_access_expr(stmt->stmt->body, &inc_n_access, stmt) < 0)
                goto error;

        stmt->access = isl_calloc_array(ctx, isl_ast_expr_list *,
                                        stmt->n_access);
        if (!stmt->access)
                goto error;

        map = isl_map_from_union_map(isl_ast_build_get_schedule(build));
        map = isl_map_reverse(map);

        stmt->n_access = 0;
        data.stmt = stmt;
        data.build = build;
        data.map = map;
        if (foreach_access_expr(stmt->stmt->body, &add_access, &data) < 0)
                node = isl_ast_node_free(node);

        isl_map_free(map);

        id = isl_id_alloc(isl_ast_node_get_ctx(node), NULL, stmt);
        id = isl_id_set_free_user(id, &ppcg_stmt_free);
        return isl_ast_node_set_annotation(node, id);
error:
        ppcg_stmt_free(stmt);
        return isl_ast_node_free(node);
}

/* Code generate the scop 'scop' and print the corresponding C code to 'p'.
 */
static __isl_give isl_printer *print_scop(isl_ctx *ctx, struct ppcg_scop *scop,
        __isl_take isl_printer *p, struct ppcg_options *options)
{
        isl_set *context;
        isl_union_set *domain_set;
        isl_union_map *schedule_map;
        isl_ast_build *build;
        isl_ast_print_options *print_options;
        isl_ast_node *tree;
        struct ast_build_userinfo build_info;

        context = isl_set_copy(scop->context);
        domain_set = isl_union_set_copy(scop->domain);
        schedule_map = isl_union_map_copy(scop->schedule);
        schedule_map = isl_union_map_intersect_domain(schedule_map, domain_set);

        build = isl_ast_build_from_context(context);
        build = isl_ast_build_set_at_each_domain(build, &at_each_domain, scop);

        if (options->openmp) {
                build_info.scop = scop;
                build_info.in_parallel_for = 0;

                build = isl_ast_build_set_before_each_for(build,
                                                        &ast_build_before_for,
                                                        &build_info);
                build = isl_ast_build_set_after_each_for(build,
                                                        &ast_build_after_for,
                                                        &build_info);
        }

        tree = isl_ast_build_ast_from_schedule(build, schedule_map);
        isl_ast_build_free(build);

        print_options = isl_ast_print_options_alloc(ctx);
        print_options = isl_ast_print_options_set_print_user(print_options,
                                                        &print_user, NULL);

        print_options = isl_ast_print_options_set_print_for(print_options,
                                                        &print_for, NULL);

        p = isl_ast_node_print_macros(tree, p);
        p = isl_ast_node_print(tree, p, print_options);

        isl_ast_node_free(tree);

        return p;
}

/* Does "scop" refer to any arrays that are declared, but not
 * exposed to the code after the scop?
 */
static int any_hidden_declarations(struct ppcg_scop *scop)
{
        int i;

        if (!scop)
                return 0;

        for (i = 0; i < scop->n_array; ++i)
                if (scop->arrays[i]->declared && !scop->arrays[i]->exposed)
                        return 1;

        return 0;
}

int generate_cpu(isl_ctx *ctx, struct ppcg_scop *ps,
        struct ppcg_options *options, const char *input, const char *output)
{
        FILE *input_file;
        FILE *output_file;
        isl_printer *p;
        int hidden;

        if (!ps)
                return -1;

        input_file = fopen(input, "r");
        output_file = get_output_file(input, output);

        copy(input_file, output_file, 0, ps->start);
        fprintf(output_file, "/* ppcg generated CPU code */\n\n");
        p = isl_printer_to_file(ctx, output_file);
        p = isl_printer_set_output_format(p, ISL_FORMAT_C);
        p = ppcg_print_exposed_declarations(p, ps);
        hidden = any_hidden_declarations(ps);
        if (hidden) {
                p = ppcg_start_block(p);
                p = ppcg_print_hidden_declarations(p, ps);
        }
        p = print_scop(ctx, ps, p, options);
        if (hidden)
                p = ppcg_end_block(p);
        isl_printer_free(p);
        copy(input_file, output_file, ps->end, -1);

        fclose(output_file);
        fclose(input_file);

        return 0;
}