X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=ir%2Fbe%2Fbemain.c;h=77af774ae3188b3c05d3bdc98f7f9c6dc4f7c674;hb=3c5deb9481abe11a876ea84bd55b0606596b9bd4;hp=41ca3502fe0efb0aea7afd2ca9c125b56ad061ce;hpb=e9d5686e6022c8ef5eccfc4b9828e08f395dc899;p=libfirm diff --git a/ir/be/bemain.c b/ir/be/bemain.c index 41ca3502f..77af774ae 100644 --- a/ir/be/bemain.c +++ b/ir/be/bemain.c @@ -1,18 +1,38 @@ +/* + * Copyright (C) 1995-2008 University of Karlsruhe. All right reserved. + * + * This file is part of libFirm. + * + * This file may be distributed and/or modified under the terms of the + * GNU General Public License version 2 as published by the Free Software + * Foundation and appearing in the file LICENSE.GPL included in the + * packaging of this file. + * + * Licensees holding valid libFirm Professional Edition licenses may use + * this file in accordance with the libFirm Commercial License. + * Agreement provided with the Software. + * + * This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE + * WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE. + */ + /** - * Backend driver. - * @author Sebastian Hack - * @date 25.11.2004 + * @file + * @brief Main Backend driver. + * @author Sebastian Hack + * @date 25.11.2004 + * @version $Id$ */ #ifdef HAVE_CONFIG_H #include "config.h" #endif #include +#include -#ifdef WITH_LIBCORE -#include -#include -#endif /* WITH_LIBCORE */ +#include "lc_opts.h" +#include "lc_opts_enum.h" #include "obst.h" #include "bitset.h" @@ -26,24 +46,23 @@ #include "iredges_t.h" #include "irloop_t.h" #include "irtools.h" - -#include "bearch.h" -#include "firm/bearch_firm.h" -#include "ia32/bearch_ia32.h" -#include "TEMPLATE/bearch_TEMPLATE.h" - +#include "irvrfy.h" +#include "irprintf.h" +#include "iroptimize.h" +#include "firmstat.h" +#include "execfreq.h" +#include "irprofile.h" + +#include "bearch_t.h" #include "be_t.h" -#include "benumb_t.h" +#include "bemodule.h" #include "beutil.h" #include "benode_t.h" #include "beirgmod.h" #include "besched_t.h" #include "belistsched.h" #include "belive_t.h" -#include "bespillilp.h" -#include "bespillbelady.h" #include "bera.h" -#include "beraextern.h" #include "bechordal_t.h" #include "beifg.h" #include "beifg_impl.h" @@ -51,275 +70,897 @@ #include "becopystat.h" #include "bessadestr.h" #include "beabi.h" +#include "belower.h" +#include "beschedmris.h" +#include "bestat.h" +#include "beverify.h" +#include "be_dbgout.h" +#include "beirg_t.h" +#define NEW_ID(s) new_id_from_chars(s, sizeof(s) - 1) -#define DUMP_INITIAL (1 << 0) -#define DUMP_ABI (1 << 1) -#define DUMP_SCHED (1 << 2) -#define DUMP_PREPARED (1 << 3) -#define DUMP_RA (1 << 4) -#define DUMP_FINAL (1 << 5) +#ifdef WITH_ILP +#include "beilpsched.h" +#endif /* WITH_ILP */ /* options visible for anyone */ -be_options_t be_options = { - /* ilp server */ - "i44pc52.info.uni-karlsruhe.de", - - /* ilp solver */ - "cplex" +static be_options_t be_options = { + DUMP_NONE, /* dump flags */ + BE_TIME_OFF, /* no timing */ + 0, /* no opt profile */ + 0, /* try to omit frame pointer */ + 0, /* try to omit leaf frame pointer */ + 0, /* create PIC code */ + 0, /* create gprof compatible profiling code */ + BE_VRFY_WARN, /* verification level: warn */ + BE_SCHED_LIST, /* scheduler: list scheduler */ + "linux", /* target OS name */ + "i44pc52.info.uni-karlsruhe.de", /* ilp server */ + "cplex", /* ilp solver */ + 0, /* enable statistic event dumping */ + "", /* print stat events */ }; -/* dump flags */ -static unsigned dump_flags = 2 * DUMP_FINAL - 1; - -/* register allocator to use. */ -static const be_ra_t *ra = &be_ra_chordal_allocator; +/* config file. */ +static char config_file[256] = { 0 }; /* back end instruction set architecture to use */ -static const arch_isa_if_t *isa_if = &ia32_isa_if; -#ifdef WITH_LIBCORE - -static lc_opt_entry_t *be_grp_root = NULL; +static const arch_isa_if_t *isa_if = NULL; /* possible dumping options */ static const lc_opt_enum_mask_items_t dump_items[] = { - { "none", 0 }, + { "none", DUMP_NONE }, { "initial", DUMP_INITIAL }, { "abi", DUMP_ABI }, { "sched", DUMP_SCHED }, { "prepared", DUMP_PREPARED }, { "regalloc", DUMP_RA }, { "final", DUMP_FINAL }, - { "all", 2 * DUMP_FINAL - 1 }, + { "be", DUMP_BE }, + { "all", 2 * DUMP_BE - 1 }, { NULL, 0 } }; -/* register allocators */ -static const lc_opt_enum_const_ptr_items_t ra_items[] = { - { "chordal", &be_ra_chordal_allocator }, - { "external", &be_ra_external_allocator }, - { NULL, NULL } +/* verify options. */ +static const lc_opt_enum_int_items_t vrfy_items[] = { + { "off", BE_VRFY_OFF }, + { "warn", BE_VRFY_WARN }, + { "assert", BE_VRFY_ASSERT }, + { NULL, 0 } }; -/* instruction set architectures. */ -static const lc_opt_enum_const_ptr_items_t isa_items[] = { - { "firm", &firm_isa }, - { "ia32", &ia32_isa_if }, - { "TEMPLATE",&TEMPLATE_isa_if }, - { NULL, NULL } +/* scheduling options. */ +static const lc_opt_enum_int_items_t sched_items[] = { + { "list", BE_SCHED_LIST }, +#ifdef WITH_ILP + { "ilp", BE_SCHED_ILP }, +#endif /* WITH_ILP */ + { NULL, 0 } }; static lc_opt_enum_mask_var_t dump_var = { - &dump_flags, dump_items + &be_options.dump_flags, dump_items }; -static lc_opt_enum_const_ptr_var_t ra_var = { - (const void **) &ra, ra_items +static lc_opt_enum_int_var_t vrfy_var = { + &be_options.vrfy_option, vrfy_items }; -static lc_opt_enum_const_ptr_var_t isa_var = { - (const void **) &isa_if, isa_items +static lc_opt_enum_int_var_t sched_var = { + &be_options.scheduler, sched_items }; static const lc_opt_table_entry_t be_main_options[] = { - LC_OPT_ENT_ENUM_MASK("dump", "dump irg on several occasions", &dump_var), - LC_OPT_ENT_ENUM_PTR("ra", "register allocator", &ra_var), - LC_OPT_ENT_ENUM_PTR("isa", "the instruction set architecture", &isa_var), + LC_OPT_ENT_STR ("config", "read another config file containing backend options", config_file, sizeof(config_file)), + LC_OPT_ENT_ENUM_MASK("dump", "dump irg on several occasions", &dump_var), + LC_OPT_ENT_BOOL ("omitfp", "omit frame pointer", &be_options.omit_fp), + LC_OPT_ENT_BOOL ("omitleaffp", "omit frame pointer in leaf routines", &be_options.omit_leaf_fp), + LC_OPT_ENT_BOOL ("pic", "create PIC code", &be_options.pic), + LC_OPT_ENT_BOOL ("gprof", "create gprof profiling code", &be_options.gprof), + LC_OPT_ENT_ENUM_PTR ("vrfy", "verify the backend irg", &vrfy_var), + LC_OPT_ENT_BOOL ("time", "get backend timing statistics", &be_options.timing), + LC_OPT_ENT_BOOL ("profile", "instrument the code for execution count profiling", &be_options.opt_profile), + LC_OPT_ENT_ENUM_PTR ("sched", "select a scheduler", &sched_var), + LC_OPT_ENT_STR ("os", "specify target operating system", &be_options.target_os, sizeof(be_options.target_os)), +#ifdef FIRM_STATISTICS + LC_OPT_ENT_BOOL ("statev", "dump statistic events", &be_options.statev), + LC_OPT_ENT_STR ("filtev", "filter for stat events (regex if support is active", &be_options.filtev, sizeof(be_options.filtev)), +#endif +#ifdef WITH_ILP LC_OPT_ENT_STR ("ilp.server", "the ilp server name", be_options.ilp_server, sizeof(be_options.ilp_server)), LC_OPT_ENT_STR ("ilp.solver", "the ilp solver name", be_options.ilp_solver, sizeof(be_options.ilp_solver)), - { NULL } +#endif /* WITH_ILP */ + LC_OPT_LAST }; -#endif /* WITH_LIBCORE */ +static be_module_list_entry_t *isa_ifs = NULL; -void be_opt_register(void) + +unsigned short asm_constraint_flags[256]; + +void be_init_default_asm_constraint_flags(void) { -#ifdef WITH_LIBCORE - int i; - lc_opt_entry_t *be_grp_ra; + asm_constraint_flags['?'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + asm_constraint_flags['!'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + asm_constraint_flags['&'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT + | ASM_CONSTRAINT_FLAG_MODIFIER_EARLYCLOBBER; + asm_constraint_flags['%'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT + | ASM_CONSTRAINT_FLAG_MODIFIER_COMMUTATIVE; + asm_constraint_flags['!'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + + asm_constraint_flags['='] = ASM_CONSTRAINT_FLAG_MODIFIER_WRITE + | ASM_CONSTRAINT_FLAG_MODIFIER_NO_READ; + asm_constraint_flags['+'] = ASM_CONSTRAINT_FLAG_MODIFIER_READ + | ASM_CONSTRAINT_FLAG_MODIFIER_WRITE; + + asm_constraint_flags['i'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['s'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['E'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['F'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['G'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['H'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['I'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['J'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['K'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['L'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['M'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['N'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['O'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + asm_constraint_flags['P'] = ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + + asm_constraint_flags['m'] = ASM_CONSTRAINT_FLAG_SUPPORTS_MEMOP; + asm_constraint_flags['o'] = ASM_CONSTRAINT_FLAG_SUPPORTS_MEMOP; + asm_constraint_flags['V'] = ASM_CONSTRAINT_FLAG_SUPPORTS_MEMOP; + asm_constraint_flags['<'] = ASM_CONSTRAINT_FLAG_SUPPORTS_MEMOP; + asm_constraint_flags['>'] = ASM_CONSTRAINT_FLAG_SUPPORTS_MEMOP; + + asm_constraint_flags['p'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['0'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['1'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['2'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['3'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['4'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['5'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['6'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['7'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['8'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + asm_constraint_flags['9'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER; + + asm_constraint_flags['X'] = ASM_CONSTRAINT_FLAG_SUPPORTS_REGISTER + | ASM_CONSTRAINT_FLAG_SUPPORTS_MEMOP + | ASM_CONSTRAINT_FLAG_SUPPORTS_IMMEDIATE; + + /* these should have been catched by the parsing code already */ + asm_constraint_flags['#'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + asm_constraint_flags['*'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + asm_constraint_flags[' '] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + asm_constraint_flags['\t'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + asm_constraint_flags['\n'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; + asm_constraint_flags['\r'] = ASM_CONSTRAINT_FLAG_NO_SUPPORT; +} + +asm_constraint_flags_t be_parse_asm_constraints(const char *constraint) +{ + asm_constraint_flags_t flags = 0; + const char *c; + asm_constraint_flags_t tflags; + + for (c = constraint; *c != '\0'; ++c) { + switch (*c) { + case '#': + /* 'comment' stuff */ + while(*c != 0 && *c != ',') + ++c; + break; + case '*': + /* 'comment' character */ + ++c; + break; + case ' ': + case '\t': + case '\n': + case '\r': + break; + default: + tflags = asm_constraint_flags[(int) *c]; + if (tflags != 0) { + flags |= tflags; + } else { + flags |= isa_if->parse_asm_constraint(isa_if, &c); + } + break; + } + } + + if ( ((flags & ASM_CONSTRAINT_FLAG_MODIFIER_WRITE) + && (flags & ASM_CONSTRAINT_FLAG_MODIFIER_NO_WRITE)) + || ((flags & ASM_CONSTRAINT_FLAG_MODIFIER_READ + && (flags & ASM_CONSTRAINT_FLAG_MODIFIER_NO_READ)))) { + flags |= ASM_CONSTRAINT_FLAG_INVALID; + } + if (! (flags & (ASM_CONSTRAINT_FLAG_MODIFIER_READ + | ASM_CONSTRAINT_FLAG_MODIFIER_WRITE + | ASM_CONSTRAINT_FLAG_MODIFIER_NO_WRITE + | ASM_CONSTRAINT_FLAG_MODIFIER_NO_READ))) { + flags |= ASM_CONSTRAINT_FLAG_MODIFIER_READ; + } - be_grp_root = lc_opt_get_grp(firm_opt_get_root(), "be"); - be_grp_ra = lc_opt_get_grp(be_grp_root, "ra"); + return flags; +} - lc_opt_add_table(be_grp_root, be_main_options); +bool be_is_valid_clobber(const char *clobber) +{ + /* memory is a valid clobber. (the frontend has to detect this case too, + * because it has to add memory edges to the asm) */ + if (strcmp(clobber, "memory") == 0) + return true; + /* cc (condition code) is always valid */ + if (strcmp(clobber, "cc") == 0) + return true; + + return isa_if->is_valid_clobber(isa_if, clobber); +} - /* register allocator options */ - for(i = 0; ra_items[i].name != NULL; ++i) { - const be_ra_t *ra = ra_items[i].value; - ra->register_options(be_grp_ra); +void be_register_isa_if(const char *name, const arch_isa_if_t *isa) +{ + if (isa_if == NULL) + isa_if = isa; + + be_add_module_to_list(&isa_ifs, name, (void*) isa); +} + +void be_opt_register(void) +{ + lc_opt_entry_t *be_grp; + static int run_once = 0; + + if (run_once) { + return; } + run_once = 1; + + be_init_modules(); - /* register isa options */ - for(i = 0; isa_items[i].name != NULL; ++i) { - const arch_isa_if_t *isa = isa_items[i].value; - isa->register_options(be_grp_root); + be_grp = lc_opt_get_grp(firm_opt_get_root(), "be"); + lc_opt_add_table(be_grp, be_main_options); + + be_add_module_list_opt(be_grp, "isa", "the instruction set architecture", + &isa_ifs, (void**) &isa_if); +} + +/* Parse one argument. */ +int be_parse_arg(const char *arg) { + lc_opt_entry_t *be_grp = lc_opt_get_grp(firm_opt_get_root(), "be"); + if (strcmp(arg, "help") == 0 || (arg[0] == '?' && arg[1] == '\0')) { + lc_opt_print_help(be_grp, stdout); + return -1; } -#endif /* WITH_LIBCORE */ + return lc_opt_from_single_arg(be_grp, NULL, arg, NULL); } +/** The be parameters returned by default, all off. */ +static const backend_params be_params = { + 0, /* need dword lowering */ + 0, /* don't support inline assembler yet */ + 0, /* no immediate floating point mode. */ + NULL, /* no additional opcodes */ + NULL, /* will be set later */ + NULL, /* but yet no creator function */ + NULL, /* context for create_intrinsic_fkt */ + NULL, /* no if conversion settings */ + NULL /* no immediate fp mode */ +}; + +/* Perform schedule verification if requested. */ +static void be_sched_vrfy(be_irg_t *birg, int vrfy_opt) { + if (vrfy_opt == BE_VRFY_WARN) { + be_verify_schedule(birg); + } else if (vrfy_opt == BE_VRFY_ASSERT) { + assert(be_verify_schedule(birg) && "Schedule verification failed."); + } +} -void be_init(void) +/* Initialize the Firm backend. Must be run first in init_firm()! */ +void firm_be_init(void) { be_opt_register(); + be_init_modules(); +} - be_sched_init(); - be_liveness_init(); - be_numbering_init(); - be_copy_opt_init(); - copystat_init(); - phi_class_init(); +/* Finalize the Firm backend. */ +void firm_be_finish(void) +{ + be_quit_modules(); } -static be_main_env_t *be_init_env(be_main_env_t *env) +/* Returns the backend parameter */ +const backend_params *be_get_backend_param(void) +{ + if (isa_if->get_params) + return isa_if->get_params(); + return &be_params; +} + +/** + * Initializes the main environment for the backend. + * + * @param env an empty environment + * @param file_handle the file handle where the output will be written to + */ +static be_main_env_t *be_init_env(be_main_env_t *env, FILE *file_handle) { memset(env, 0, sizeof(*env)); - obstack_init(&env->obst); - env->dbg = firm_dbg_register("be.main"); + env->options = &be_options; + env->ent_trampoline_map = pmap_create(); + env->pic_trampolines_type = new_type_class(NEW_ID("$PIC_TRAMPOLINE_TYPE")); + env->ent_pic_symbol_map = pmap_create(); + env->pic_symbols_type = new_type_struct(NEW_ID("$PIC_SYMBOLS_TYPE")); - env->arch_env = obstack_alloc(&env->obst, sizeof(env->arch_env[0])); - arch_env_init(env->arch_env, isa_if); + remove_irp_type(env->pic_trampolines_type); + remove_irp_type(env->pic_symbols_type); + set_class_final(env->pic_trampolines_type, 1); - /* Register the irn handler of the architecture */ - if (arch_isa_get_irn_handler(env->arch_env->isa)) - arch_env_add_irn_handler(env->arch_env, arch_isa_get_irn_handler(env->arch_env->isa)); + memset(asm_constraint_flags, 0, sizeof(asm_constraint_flags)); + env->arch_env = arch_env_init(isa_if, file_handle, env); - /* - * Register the node handler of the back end infrastructure. - * This irn handler takes care of the platform independent - * spill, reload and perm nodes. - */ - arch_env_add_irn_handler(env->arch_env, &be_node_irn_handler); + be_phi_handler_new(env); + be_dbg_open(); return env; } +/** + * Called when the be_main_env_t can be destroyed. + */ static void be_done_env(be_main_env_t *env) { - env->arch_env->isa->impl->done(env->arch_env->isa); - obstack_free(&env->obst, NULL); + arch_env_done(env->arch_env); + be_dbg_close(); + be_phi_handler_free(); + + pmap_destroy(env->ent_trampoline_map); + pmap_destroy(env->ent_pic_symbol_map); + free_type(env->pic_trampolines_type); + free_type(env->pic_symbols_type); } +/** + * A wrapper around a firm dumper. Dumps only, if + * flags are enabled. + * + * @param mask a bitmask containing the reason what will be dumped + * @param irg the IR graph to dump + * @param suffix the suffix for the dumper + * @param dumper the dumper to be called + */ static void dump(int mask, ir_graph *irg, const char *suffix, - void (*dumper)(ir_graph *, const char *)) + void (*dumper)(ir_graph *, const char *)) { - if(dump_flags & mask) - dumper(irg, suffix); + if(be_options.dump_flags & mask) + be_dump(irg, suffix, dumper); } -static void prepare_graph(be_irg_t *birg) +/** + * Prepare a backend graph for code generation and initialize its birg + */ +static void initialize_birg(be_irg_t *birg, ir_graph *irg, be_main_env_t *env) { - ir_graph *irg = birg->irg; + memset(birg, 0, sizeof(*birg)); + birg->irg = irg; + birg->main_env = env; + + edges_deactivate_kind(irg, EDGE_KIND_DEP); + edges_activate_kind(irg, EDGE_KIND_DEP); + + dump(DUMP_INITIAL, irg, "-begin", dump_ir_block_graph); + + be_stat_init_irg(env->arch_env, irg); + be_do_stat_nodes(irg, "01 Begin"); + + /* set the current graph (this is important for several firm functions) */ + current_ir_graph = irg; /* Normalize proj nodes. */ normalize_proj_nodes(irg); - /* Make just one return node. */ - // normalize_one_return(irg); + /* we do this before critical edge split. As this produces less returns, + because sometimes (= 164.gzip) multiple returns are slower */ + normalize_n_returns(irg); /* Remove critical edges */ - remove_critical_cf_edges(irg); - - /* Compute the dominance information. */ - free_dom(irg); - compute_doms(irg); + remove_critical_cf_edges_ex(irg, /*ignore_exception_edges=*/0); /* Ensure, that the ir_edges are computed. */ - edges_activate(irg); + edges_assure(irg); - /* check, if the dominance property is fulfilled. */ - be_check_dominance(irg); + set_irg_phase_state(irg, phase_backend); + dump(DUMP_INITIAL, irg, "-prepared", dump_ir_block_graph); } -static void be_main_loop(FILE *file_handle) +#define BE_TIMER_ONLY(code) do { if (be_timing) { code; } } while(0) + +int be_timing; +ir_timer_t *t_abi; +ir_timer_t *t_codegen; +ir_timer_t *t_sched; +ir_timer_t *t_constr; +ir_timer_t *t_finish; +ir_timer_t *t_emit; +ir_timer_t *t_other; +ir_timer_t *t_verify; +ir_timer_t *t_heights; +ir_timer_t *t_live; +ir_timer_t *t_execfreq; +ir_timer_t *t_ssa_constr; +ir_timer_t *t_ra_constr; +ir_timer_t *t_ra_prolog; +ir_timer_t *t_ra_epilog; +ir_timer_t *t_ra_spill; +ir_timer_t *t_ra_spill_apply; +ir_timer_t *t_ra_color; +ir_timer_t *t_ra_ifg; +ir_timer_t *t_ra_copymin; +ir_timer_t *t_ra_ssa; +ir_timer_t *t_ra_other; + +/** + * The Firm backend main loop. + * Do architecture specific lowering for all graphs + * and call the architecture specific code generator. + * + * @param file_handle the file handle the output will be written to + * @param cup_name name of the compilation unit + */ +static void be_main_loop(FILE *file_handle, const char *cup_name) { - int i, n; - arch_isa_t *isa; + int i; be_main_env_t env; + char prof_filename[256]; + static const char suffix[] = ".prof"; + be_irg_t *birgs; + int num_birgs; + ir_graph **irg_list, **backend_irg_list; + arch_env_t *arch_env; + + be_timing = (be_options.timing == BE_TIME_ON); + + if (be_timing) { + t_abi = ir_timer_register("time_beabi", "be abi introduction"); + t_codegen = ir_timer_register("time_codegen", "codegeneration"); + t_sched = ir_timer_register("time_sched", "scheduling"); + t_constr = ir_timer_register("time_constr", "assure constraints"); + t_finish = ir_timer_register("time_finish", "graph finish"); + t_emit = ir_timer_register("time_emiter", "code emiter"); + t_verify = ir_timer_register("time_verify", "graph verification"); + t_other = ir_timer_register("time_other", "other"); + t_heights = ir_timer_register("time_heights", "heights"); + t_live = ir_timer_register("time_liveness", "be liveness"); + t_execfreq = ir_timer_register("time_execfreq", "execfreq"); + t_ssa_constr = ir_timer_register("time_ssa_constr", "ssa reconstruction"); + t_ra_prolog = ir_timer_register("time_ra_prolog", "regalloc prolog"); + t_ra_epilog = ir_timer_register("time_ra_epilog", "regalloc epilog"); + t_ra_constr = ir_timer_register("time_ra_constr", "regalloc constraints"); + t_ra_spill = ir_timer_register("time_ra_spill", "spiller"); + t_ra_spill_apply + = ir_timer_register("time_ra_spill_apply", "apply spills"); + t_ra_color = ir_timer_register("time_ra_color", "graph coloring"); + t_ra_ifg = ir_timer_register("time_ra_ifg", "interference graph"); + t_ra_copymin = ir_timer_register("time_ra_copymin", "copy minimization"); + t_ra_ssa = ir_timer_register("time_ra_ssadestr", "ssa destruction"); + t_ra_other = ir_timer_register("time_ra_other", "regalloc other"); + } + + be_init_env(&env, file_handle); + env.cup_name = cup_name; + + be_dbg_so(cup_name); + be_dbg_types(); + + arch_env = env.arch_env; - be_init_env(&env); + /* backend may provide an ordered list of irgs where code should be generated for */ + irg_list = NEW_ARR_F(ir_graph *, 0); + backend_irg_list = arch_env_get_backend_irg_list(arch_env, &irg_list); - isa = arch_env_get_isa(env.arch_env); + /* we might need 1 birg more for instrumentation constructor */ + num_birgs = backend_irg_list ? ARR_LEN(backend_irg_list) : get_irp_n_irgs(); + birgs = alloca(sizeof(birgs[0]) * (num_birgs + 1)); + + /* First: initialize all birgs */ + for(i = 0; i < num_birgs; ++i) { + ir_graph *irg = backend_irg_list ? backend_irg_list[i] : get_irp_irg(i); + initialize_birg(&birgs[i], irg, &env); + } + DEL_ARR_F(irg_list); + + /* + Get the filename for the profiling data. + Beware: '\0' is already included in sizeof(suffix) + */ + memset(prof_filename, 0, sizeof(prof_filename)); + strncpy(prof_filename, cup_name, sizeof(prof_filename) - sizeof(suffix)); + strcat(prof_filename, suffix); + + /* + Next: Either instruments all irgs with profiling code + or try to read in profile data for current translation unit. + */ + if (be_options.opt_profile) { + ir_graph *prof_init_irg = ir_profile_instrument(prof_filename, profile_default); + initialize_birg(&birgs[num_birgs], prof_init_irg, &env); + num_birgs++; + } else { + ir_profile_read(prof_filename); + } /* For all graphs */ - for(i = 0, n = get_irp_n_irgs(); i < n; ++i) { - ir_graph *irg = get_irp_irg(i); + for (i = 0; i < num_birgs; ++i) { + be_irg_t *birg = &birgs[i]; + ir_graph *irg = birg->irg; + optimization_state_t state; const arch_code_generator_if_t *cg_if; - be_irg_t birg; - birg.irg = irg; - birg.main_env = &env; + /* set the current graph (this is important for several firm functions) */ + current_ir_graph = irg; - DBG((env.dbg, LEVEL_2, "====> IRG: %F\n", irg)); - dump(DUMP_INITIAL, irg, "-begin", dump_ir_block_graph); +#if 0 + { + unsigned percent = 100*i/num_birgs; + ir_printf("%u.%02u %+F\n", percent/100, percent%100, irg); + } +#endif + be_sched_init_phase(irg); - /* set the current graph (this is important for several firm functions) */ - current_ir_graph = birg.irg; + /* reset the phi handler. */ + be_phi_handler_reset(); + + stat_ev_ctx_push_fobj("bemain_irg", irg); + + /* stop and reset timers */ + BE_TIMER_PUSH(t_other); /* t_other */ + + /* Verify the initial graph */ + BE_TIMER_PUSH(t_verify); + if (be_options.vrfy_option == BE_VRFY_WARN) { + irg_verify(irg, VRFY_ENFORCE_SSA); + be_check_dominance(irg); + } else if (be_options.vrfy_option == BE_VRFY_ASSERT) { + assert(irg_verify(irg, VRFY_ENFORCE_SSA) && "irg verification failed"); + assert(be_check_dominance(irg) && "Dominance verification failed"); + } + BE_TIMER_POP(t_verify); /* Get the code generator interface. */ - cg_if = isa->impl->get_code_generator_if(isa); + cg_if = arch_env_get_code_generator_if(arch_env); /* get a code generator for this graph. */ - birg.cg = cg_if->init(file_handle, &birg); + birg->cg = cg_if->init(birg); - /* create the code generator and generate code. */ - prepare_graph(&birg); + /* some transformations need to be done before abi introduce */ + arch_code_generator_before_abi(birg->cg); /* implement the ABI conventions. */ - birg.abi = be_abi_introduce(&birg); + BE_TIMER_PUSH(t_abi); + birg->abi = be_abi_introduce(birg); + BE_TIMER_POP(t_abi); + dump(DUMP_ABI, irg, "-abi", dump_ir_block_graph); + be_do_stat_nodes(irg, "02 Abi"); + + if (be_options.vrfy_option == BE_VRFY_WARN) { + be_check_dominance(irg); + be_verify_out_edges(irg); + } else if (be_options.vrfy_option == BE_VRFY_ASSERT) { + assert(be_verify_out_edges(irg)); + assert(be_check_dominance(irg) && "Dominance verification failed"); + } /* generate code */ - arch_code_generator_prepare_graph(birg.cg); + stat_ev_ctx_push_str("bemain_phase", "prepare"); + BE_TIMER_PUSH(t_codegen); + arch_code_generator_prepare_graph(birg->cg); + BE_TIMER_POP(t_codegen); + stat_ev_ctx_pop("bemain_phase"); - /* - * Since the code generator made a lot of new nodes and skipped - * a lot of old ones, we should do dead node elim here. - * Note that this requires disabling the edges here. - */ - edges_deactivate(irg); - dead_node_elimination(irg); - edges_activate(irg); + /* reset the phi handler. */ + be_phi_handler_reset(); - /* Compute loop nesting information (for weighting copies) */ - construct_cf_backedges(irg); + be_do_stat_nodes(irg, "03 Prepare"); dump(DUMP_PREPARED, irg, "-prepared", dump_ir_block_graph); - /* Schedule the graphs. */ - arch_code_generator_before_sched(birg.cg); - list_sched(isa, irg); + if (be_options.vrfy_option == BE_VRFY_WARN) { + be_check_dominance(irg); + be_verify_out_edges(irg); + } else if (be_options.vrfy_option == BE_VRFY_ASSERT) { + assert(be_verify_out_edges(irg)); + assert(be_check_dominance(irg) && "Dominance verification failed"); + } + + BE_TIMER_PUSH(t_execfreq); + /** + * Create execution frequencies from profile data or estimate some + */ + if (ir_profile_has_data()) + birg->exec_freq = ir_create_execfreqs_from_profile(irg); + else + birg->exec_freq = compute_execfreq(irg, 10); + BE_TIMER_POP(t_execfreq); + + + /* disabled for now, fails for EmptyFor.c and XXEndless.c */ + /* be_live_chk_compare(birg); */ + + /* let backend prepare scheduling */ + stat_ev_ctx_push_str("bemain_phase", "before_sched"); + BE_TIMER_PUSH(t_codegen); + arch_code_generator_before_sched(birg->cg); + BE_TIMER_POP(t_codegen); + stat_ev_ctx_pop("bemain_phase"); + + /* schedule the irg */ + BE_TIMER_PUSH(t_sched); + switch (be_options.scheduler) { + default: + fprintf(stderr, "Warning: invalid scheduler (%d) selected, falling back to list scheduler.\n", be_options.scheduler); + case BE_SCHED_LIST: + list_sched(birg, &be_options); + break; +#ifdef WITH_ILP + case BE_SCHED_ILP: + be_ilp_sched(birg, &be_options); + break; +#endif /* WITH_ILP */ + }; + BE_TIMER_POP(t_sched); dump(DUMP_SCHED, irg, "-sched", dump_ir_block_graph_sched); + /* check schedule */ + BE_TIMER_PUSH(t_verify); + be_sched_vrfy(birg, be_options.vrfy_option); + BE_TIMER_POP(t_verify); + + be_do_stat_nodes(irg, "04 Schedule"); + + /* introduce patterns to assure constraints */ + BE_TIMER_PUSH(t_constr); + /* we switch off optimizations here, because they might cause trouble */ + save_optimization_state(&state); + set_optimize(0); + set_opt_normalize(0); + set_opt_cse(0); + + assert(!get_opt_cse()); + + /* add Keeps for should_be_different constrained nodes */ + /* beware: needs schedule due to usage of be_ssa_constr */ + assure_constraints(birg); + BE_TIMER_POP(t_constr); + + dump(DUMP_SCHED, irg, "-assured", dump_ir_block_graph_sched); + be_do_stat_nodes(irg, "05 Constraints"); + + /* stuff needs to be done after scheduling but before register allocation */ + BE_TIMER_PUSH(t_codegen); + arch_code_generator_before_ra(birg->cg); + BE_TIMER_POP(t_codegen); + /* connect all stack modifying nodes together (see beabi.c) */ - be_abi_fix_stack_nodes(birg.abi); + BE_TIMER_PUSH(t_abi); + be_abi_fix_stack_nodes(birg->abi); + BE_TIMER_POP(t_abi); + + dump(DUMP_SCHED, irg, "-fix_stack", dump_ir_block_graph_sched); + + /* check schedule */ + BE_TIMER_PUSH(t_verify); + be_sched_vrfy(birg, be_options.vrfy_option); + BE_TIMER_POP(t_verify); - /* Verify the schedule */ - sched_verify_irg(irg); + /* do some statistics */ + //be_do_stat_reg_pressure(birg); + +#ifdef FIRM_STATISTICS + stat_ev_dbl("costs_before_ra", be_estimate_irg_costs(irg, arch_env, birg->exec_freq)); +#endif /* Do register allocation */ - arch_code_generator_before_ra(birg.cg); - ra->allocate(&birg); - dump(DUMP_RA, irg, "-ra", dump_ir_block_graph_sched); + be_allocate_registers(birg); + +#ifdef FIRM_STATISTICS + stat_ev_dbl("costs_before_ra", be_estimate_irg_costs(irg, arch_env, birg->exec_freq)); +#endif - /* This is not ready yet: */ - /* be_abi_fix_stack_bias(birg.abi); */ + dump(DUMP_RA, irg, "-ra", dump_ir_block_graph_sched); + be_do_stat_nodes(irg, "06 Register Allocation"); + + /* let the code generator prepare the graph for emitter */ + BE_TIMER_PUSH(t_finish); + arch_code_generator_after_ra(birg->cg); + BE_TIMER_POP(t_finish); + + /* fix stack offsets */ + BE_TIMER_PUSH(t_abi); + be_abi_fix_stack_nodes(birg->abi); + be_remove_dead_nodes_from_schedule(birg); + be_abi_fix_stack_bias(birg->abi); + BE_TIMER_POP(t_abi); + + dump(DUMP_SCHED, irg, "-fix_stack_after_ra", dump_ir_block_graph_sched); + + BE_TIMER_PUSH(t_finish); + arch_code_generator_finish(birg->cg); + BE_TIMER_POP(t_finish); + + dump(DUMP_FINAL, irg, "-finish", dump_ir_block_graph_sched); + + /* check schedule and register allocation */ + BE_TIMER_PUSH(t_verify); + if (be_options.vrfy_option == BE_VRFY_WARN) { + irg_verify(irg, VRFY_ENFORCE_SSA); + be_check_dominance(irg); + be_verify_out_edges(irg); + be_verify_schedule(birg); + be_verify_register_allocation(birg); + } else if (be_options.vrfy_option == BE_VRFY_ASSERT) { + assert(irg_verify(irg, VRFY_ENFORCE_SSA) && "irg verification failed"); + assert(be_verify_out_edges(irg) && "out edge verification failed"); + assert(be_check_dominance(irg) && "Dominance verification failed"); + assert(be_verify_schedule(birg) && "Schedule verification failed"); + assert(be_verify_register_allocation(birg) + && "register allocation verification failed"); + + } + BE_TIMER_POP(t_verify); + + /* emit assembler code */ + BE_TIMER_PUSH(t_emit); + arch_code_generator_done(birg->cg); + BE_TIMER_POP(t_emit); - arch_code_generator_done(birg.cg); dump(DUMP_FINAL, irg, "-end", dump_ir_block_graph_sched); - be_abi_free(birg.abi); - } + BE_TIMER_PUSH(t_abi); + be_abi_free(birg->abi); + BE_TIMER_POP(t_abi); + + be_do_stat_nodes(irg, "07 Final"); + restore_optimization_state(&state); + + BE_TIMER_POP(t_other); + +#define STOP_AND_RESET_TIMER(timer) do { ir_timer_stop(timer); ir_timer_reset(timer); } while(0) + +#define LC_EMIT(timer) \ + stat_ev_if { \ + stat_ev_dbl(ir_timer_get_name(timer), ir_timer_elapsed_msec(timer)); \ + } else { \ + printf("%-20s: %8.3lf msec\n", ir_timer_get_description(timer), (double)ir_timer_elapsed_usec(timer) / 1000.0); \ + } \ + STOP_AND_RESET_TIMER(timer); + + BE_TIMER_ONLY( + stat_ev_if { + } else { + printf("==>> IRG %s <<==\n", get_entity_name(get_irg_entity(irg))); + } + LC_EMIT(t_abi); + LC_EMIT(t_codegen); + LC_EMIT(t_sched); + LC_EMIT(t_live); + LC_EMIT(t_heights); + LC_EMIT(t_ssa_constr); + LC_EMIT(t_constr); + LC_EMIT(t_execfreq); + LC_EMIT(t_ra_prolog); + LC_EMIT(t_ra_spill); + LC_EMIT(t_ra_spill_apply); + LC_EMIT(t_ra_constr); + LC_EMIT(t_ra_color); + LC_EMIT(t_ra_ifg); + LC_EMIT(t_ra_copymin); + LC_EMIT(t_ra_ssa); + LC_EMIT(t_ra_epilog); + LC_EMIT(t_ra_other); + LC_EMIT(t_finish); + LC_EMIT(t_emit); + LC_EMIT(t_verify); + LC_EMIT(t_other); + ); +#undef LC_EMIT + + be_sched_free_phase(irg); + + be_free_birg(birg); + + /* switched off due to statistics (statistic module needs all irgs) */ +#if 0 /* STA needs irgs */ +#ifdef FIRM_STATISTICS + if (! stat_is_active()) +#endif /* FIRM_STATISTICS */ + free_ir_graph(irg); +#endif /* if 0 */ + stat_ev_ctx_pop("bemain_irg"); + } + ir_profile_free(); be_done_env(&env); } -void be_main(FILE *file_handle) +/* Main interface to the frontend. */ +void be_main(FILE *file_handle, const char *cup_name) { + ir_timer_t *t = NULL; + + /* The user specified another config file to read. do that now. */ + if(strlen(config_file) > 0) { + FILE *f; + + if((f = fopen(config_file, "rt")) != NULL) { + lc_opt_from_file(config_file, f, NULL); + fclose(f); + } + } + + if (be_options.timing == BE_TIME_ON) { + t = ir_timer_register("bemain", "measure complete bemain loop"); + + if (ir_timer_enter_high_priority()) { + fprintf(stderr, "Warning: Could not enter high priority mode.\n"); + } + + ir_timer_reset_and_start(t); + } + +#ifdef FIRM_STATISTICS + if (be_options.statev) { + const char *dot = strrchr(cup_name, '.'); + const char *pos = dot ? dot : cup_name + strlen(cup_name); + char *buf = alloca(pos - cup_name + 1); + strncpy(buf, cup_name, pos - cup_name); + buf[pos - cup_name] = '\0'; + + be_options.statev = 1; + stat_ev_begin(buf, be_options.filtev); + } +#endif + /* never build code for pseudo irgs */ set_visit_pseudo_irgs(0); - be_node_init(); - be_main_loop(file_handle); + be_node_init(); + + be_main_loop(file_handle, cup_name); + + if (be_options.timing == BE_TIME_ON) { + ir_timer_stop(t); + ir_timer_leave_high_priority(); + stat_ev_if { + stat_ev_dbl("backend_time", ir_timer_elapsed_msec(t)); + } else { + printf("%-20s: %lu msec\n", "BEMAINLOOP", ir_timer_elapsed_msec(t)); + } + } + +#ifdef FIRM_STATISTICS + if (be_options.statev) + stat_ev_end(); +#endif +} + +unsigned be_put_ignore_regs(const be_irg_t *birg, const arch_register_class_t *cls, bitset_t *bs) +{ + if (bs == NULL) + bs = bitset_alloca(cls->n_regs); + else + bitset_clear_all(bs); + + assert(bitset_size(bs) == (unsigned)cls->n_regs); + arch_put_non_ignore_regs(birg->main_env->arch_env, cls, bs); + bitset_flip_all(bs); + be_abi_put_ignore_regs(birg->abi, cls, bs); + + return bitset_popcnt(bs); }