X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=ir%2Fopt%2Fscalar_replace.c;h=f585ea7b4d1dd98c9494210efdb43c8f62509fe6;hb=8d256d64559a5ee81a207046862ef8b284590f1c;hp=1011145d1a52c746b304925807ab5bc48e85f70c;hpb=2d0845f84ffddcc9e1b0fdf79a53bd78ebbc25a9;p=libfirm diff --git a/ir/opt/scalar_replace.c b/ir/opt/scalar_replace.c index 1011145d1..f585ea7b4 100644 --- a/ir/opt/scalar_replace.c +++ b/ir/opt/scalar_replace.c @@ -27,10 +27,9 @@ #include "config.h" #endif -#ifdef HAVE_STRING_H #include -#endif +#include "iroptimize.h" #include "scalar_replace.h" #include "irflag_t.h" #include "irouts.h" @@ -55,8 +54,8 @@ * accesses like a.b.c[8].d */ typedef union { - ir_entity *ent; - tarval *tv; + ir_entity *ent; + tarval *tv; } path_elem_t; /** @@ -64,17 +63,17 @@ typedef union { * to variables that will be scalar replaced. */ typedef struct _path_t { - unsigned vnum; /**< The value number. */ - unsigned path_len; /**< The length of the access path. */ - path_elem_t path[1]; /**< The path. */ + unsigned vnum; /**< The value number. */ + unsigned path_len; /**< The length of the access path. */ + path_elem_t path[1]; /**< The path. */ } path_t; /** The size of a path in bytes. */ #define PATH_SIZE(p) (sizeof(*(p)) + sizeof((p)->path[0]) * ((p)->path_len - 1)) typedef struct _scalars_t { - ir_entity *ent; /**< A entity for scalar replacement. */ - ir_type *ent_owner; /**< The owner of this entity. */ + ir_entity *ent; /**< A entity for scalar replacement. */ + ir_type *ent_owner; /**< The owner of this entity. */ } scalars_t; @@ -83,13 +82,13 @@ typedef struct _scalars_t { * * @return 0 if they are identically */ -static int path_cmp(const void *elt, const void *key, size_t size) -{ - const path_t *p1 = elt; - const path_t *p2 = key; +static int path_cmp(const void *elt, const void *key, size_t size) { + const path_t *p1 = elt; + const path_t *p2 = key; + (void) size; - /* we can use memcmp here, because identical tarvals should have identical addresses */ - return memcmp(p1->path, p2->path, p1->path_len * sizeof(p1->path[0])); + /* we can use memcmp here, because identical tarvals should have identical addresses */ + return memcmp(p1->path, p2->path, p1->path_len * sizeof(p1->path[0])); } /** @@ -97,26 +96,25 @@ static int path_cmp(const void *elt, const void *key, size_t size) * * @return 0 if they are identically */ -static int ent_cmp(const void *elt, const void *key, size_t size) -{ - const scalars_t *c1 = elt; - const scalars_t *c2 = key; +static int ent_cmp(const void *elt, const void *key, size_t size) { + const scalars_t *c1 = elt; + const scalars_t *c2 = key; + (void) size; - return c1->ent != c2->ent; + return c1->ent != c2->ent; } /** * Calculate a hash value for a path. */ -static unsigned path_hash(const path_t *path) -{ - unsigned hash = 0; - unsigned i; +static unsigned path_hash(const path_t *path) { + unsigned hash = 0; + unsigned i; - for (i = 0; i < path->path_len; ++i) - hash ^= (unsigned)PTR_TO_INT(path->path[i].ent); + for (i = 0; i < path->path_len; ++i) + hash ^= (unsigned)PTR_TO_INT(path->path[i].ent); - return hash >> 4; + return hash >> 4; } /** @@ -125,15 +123,15 @@ static unsigned path_hash(const path_t *path) * @param sel the Sel node that will be checked */ static int is_const_sel(ir_node *sel) { - int i, n = get_Sel_n_indexs(sel); + int i, n = get_Sel_n_indexs(sel); - for (i = 0; i < n; ++i) { - ir_node *idx = get_Sel_index(sel, i); + for (i = 0; i < n; ++i) { + ir_node *idx = get_Sel_index(sel, i); - if (get_irn_op(idx) != op_Const) - return 0; - } - return 1; + if (get_irn_op(idx) != op_Const) + return 0; + } + return 1; } /** @@ -154,15 +152,15 @@ static int is_const_sel(ir_node *sel) { * @param ent_mode the mode of the accessed entity */ static int check_load_store_mode(ir_mode *mode, ir_mode *ent_mode) { - if (ent_mode != mode) { - if (ent_mode == NULL || - get_mode_size_bits(ent_mode) != get_mode_size_bits(mode) || - get_mode_sort(ent_mode) != get_mode_sort(mode) || - get_mode_arithmetic(ent_mode) != irma_twos_complement || - get_mode_arithmetic(mode) != irma_twos_complement) - return 0; - } - return 1; + if (ent_mode != mode) { + if (ent_mode == NULL || + get_mode_size_bits(ent_mode) != get_mode_size_bits(mode) || + get_mode_sort(ent_mode) != get_mode_sort(mode) || + get_mode_arithmetic(ent_mode) != irma_twos_complement || + get_mode_arithmetic(mode) != irma_twos_complement) + return 0; + } + return 1; } /* @@ -171,62 +169,64 @@ static int check_load_store_mode(ir_mode *mode, ir_mode *ent_mode) { */ int is_address_taken(ir_node *sel) { - int i; - ir_mode *emode, *mode; - ir_node *value; - ir_entity *ent; - - if (! is_const_sel(sel)) - return 1; - - for (i = get_irn_n_outs(sel) - 1; i >= 0; --i) { - ir_node *succ = get_irn_out(sel, i); - - switch (get_irn_opcode(succ)) { - case iro_Load: - /* check if this load is not a hidden conversion */ - mode = get_Load_mode(succ); - ent = get_Sel_entity(sel); - emode = get_type_mode(get_entity_type(ent)); - if (! check_load_store_mode(mode, emode)) - return 1; - break; - - case iro_Store: - /* check that Sel is not the Store's value */ - value = get_Store_value(succ); - if (value == sel) - return 1; - /* check if this Store is not a hidden conversion */ - mode = get_irn_mode(value); - ent = get_Sel_entity(sel); - emode = get_type_mode(get_entity_type(ent)); - if (! check_load_store_mode(mode, emode)) - return 1; - break; - - case iro_Sel: { - /* Check the Sel successor of Sel */ - int res = is_address_taken(succ); - - if (res) - return 1; - break; - } - - case iro_Call: - /* The address of an entity is given as a parameter. - * As long as we do not have analyses that can tell what - * is done with parameters, think is taken. - */ - return 1; - - default: - /* another op, the address is taken */ - return 1; - } - } - return 0; + int i; + ir_mode *emode, *mode; + ir_node *value; + ir_entity *ent; + + if (! is_const_sel(sel)) + return 1; + + for (i = get_irn_n_outs(sel) - 1; i >= 0; --i) { + ir_node *succ = get_irn_out(sel, i); + + switch (get_irn_opcode(succ)) { + case iro_Load: + /* check if this load is not a hidden conversion */ + mode = get_Load_mode(succ); + ent = get_Sel_entity(sel); + emode = get_type_mode(get_entity_type(ent)); + if (! check_load_store_mode(mode, emode)) + return 1; + break; + + case iro_Store: + /* check that Sel is not the Store's value */ + value = get_Store_value(succ); + if (value == sel) + return 1; + /* check if this Store is not a hidden conversion */ + mode = get_irn_mode(value); + ent = get_Sel_entity(sel); + emode = get_type_mode(get_entity_type(ent)); + if (! check_load_store_mode(mode, emode)) + return 1; + break; + + case iro_Sel: { + /* Check the Sel successor of Sel */ + int res = is_address_taken(succ); + + if (res) + return 1; + break; + } + + case iro_Call: + /* The address of an entity is given as a parameter. + * As long as we do not have analyses that can tell what + * is done with parameters, think is taken. + * One special case: If the Call type tells that it's a + * value parameter, the address is NOT taken. + */ + return 1; + + default: + /* another op, the address is taken */ + return 1; + } + } + return 0; } /** @@ -235,37 +235,36 @@ int is_address_taken(ir_node *sel) * @param ent the entity that will be scalar replaced * @param sel a Sel node that selects some fields of this entity */ -static void link_all_leave_sels(ir_entity *ent, ir_node *sel) -{ - int i, n, flag = 1; - - n = get_irn_n_outs(sel); - for (i = 0; i < n; ++i) { - ir_node *succ = get_irn_out(sel, i); - - if (is_Sel(succ)) { - link_all_leave_sels(ent, succ); - flag = 0; - } - } - - if (flag) { - /* if Sel nodes with memory inputs are used, a entity can be - * visited more than once causing a ring here, so we use the - * node flag to mark linked nodes - */ - if (irn_visited(sel)) - return; - - /* we know we are at a leave, because this function is only - * called if the address is NOT taken, so succ must be a Load - * or a Store node - */ - set_irn_link(sel, get_entity_link(ent)); - set_entity_link(ent, sel); - - mark_irn_visited(sel); - } +static void link_all_leave_sels(ir_entity *ent, ir_node *sel) { + int i, n, flag = 1; + + n = get_irn_n_outs(sel); + for (i = 0; i < n; ++i) { + ir_node *succ = get_irn_out(sel, i); + + if (is_Sel(succ)) { + link_all_leave_sels(ent, succ); + flag = 0; + } + } + + if (flag) { + /* if Sel nodes with memory inputs are used, a entity can be + * visited more than once causing a ring here, so we use the + * node flag to mark linked nodes + */ + if (irn_visited(sel)) + return; + + /* we know we are at a leave, because this function is only + * called if the address is NOT taken, so succ must be a Load + * or a Store node + */ + set_irn_link(sel, get_entity_link(ent)); + set_entity_link(ent, sel); + + mark_irn_visited(sel); + } } /* we need a special address that serves as an address taken marker */ @@ -286,78 +285,76 @@ static void *ADDRESS_TAKEN = &_x; * @return non-zero if at least one entity could be replaced * potentially */ -static int find_possible_replacements(ir_graph *irg) -{ - ir_node *irg_frame = get_irg_frame(irg); - int i, n; - int res = 0; - - inc_irg_visited(irg); - - n = get_irn_n_outs(irg_frame); - - /* - * First, clear the link field of all interesting entities. - * Note that we did not rely on the fact that there is only - * one Sel node per entity, so we might access one entity - * more than once here. - * That's why we have need two loops. - */ - for (i = 0; i < n; ++i) { - ir_node *succ = get_irn_out(irg_frame, i); - - if (is_Sel(succ)) { - ir_entity *ent = get_Sel_entity(succ); - set_entity_link(ent, NULL); - } - } - - /* - * Check the ir_graph for Sel nodes. If the entity of Sel - * isn't a scalar replacement set the link of this entity - * equal ADDRESS_TAKEN. - */ - for (i = 0; i < n; ++i) { - ir_node *succ = get_irn_out(irg_frame, i); - - if (is_Sel(succ)) { - ir_entity *ent = get_Sel_entity(succ); - ir_type *ent_type; - - if (get_entity_link(ent) == ADDRESS_TAKEN) - continue; - - /* - * Beware: in rare cases even entities on the frame might be - * volatile. This might happen if the entity serves as a store - * to a value that must survive a exception. Do not optimize - * such entities away. - */ - if (get_entity_volatility(ent) == volatility_is_volatile) { - set_entity_link(ent, ADDRESS_TAKEN); - continue; - } - - ent_type = get_entity_type(ent); - - /* we can handle arrays, structs and atomic types yet */ - if (is_Array_type(ent_type) || is_Struct_type(ent_type) || is_atomic_type(ent_type)) { - if (is_address_taken(succ)) { - if (get_entity_link(ent)) /* killing one */ - --res; - set_entity_link(ent, ADDRESS_TAKEN); - } - else { - /* possible found one */ - if (get_entity_link(ent) == NULL) - ++res; - link_all_leave_sels(ent, succ); - } - } - } - } - - return res; +static int find_possible_replacements(ir_graph *irg) { + ir_node *irg_frame = get_irg_frame(irg); + int i, n; + int res = 0; + + inc_irg_visited(irg); + + n = get_irn_n_outs(irg_frame); + + /* + * First, clear the link field of all interesting entities. + * Note that we did not rely on the fact that there is only + * one Sel node per entity, so we might access one entity + * more than once here. + * That's why we have need two loops. + */ + for (i = 0; i < n; ++i) { + ir_node *succ = get_irn_out(irg_frame, i); + + if (is_Sel(succ)) { + ir_entity *ent = get_Sel_entity(succ); + set_entity_link(ent, NULL); + } + } + + /* + * Check the ir_graph for Sel nodes. If the entity of Sel + * isn't a scalar replacement set the link of this entity + * equal ADDRESS_TAKEN. + */ + for (i = 0; i < n; ++i) { + ir_node *succ = get_irn_out(irg_frame, i); + + if (is_Sel(succ)) { + ir_entity *ent = get_Sel_entity(succ); + ir_type *ent_type; + + if (get_entity_link(ent) == ADDRESS_TAKEN) + continue; + + /* + * Beware: in rare cases even entities on the frame might be + * volatile. This might happen if the entity serves as a store + * to a value that must survive a exception. Do not optimize + * such entities away. + */ + if (get_entity_volatility(ent) == volatility_is_volatile) { + set_entity_link(ent, ADDRESS_TAKEN); + continue; + } + + ent_type = get_entity_type(ent); + + /* we can handle arrays, structs and atomic types yet */ + if (is_Array_type(ent_type) || is_Struct_type(ent_type) || is_atomic_type(ent_type)) { + if (is_address_taken(succ)) { + if (get_entity_link(ent)) /* killing one */ + --res; + set_entity_link(ent, ADDRESS_TAKEN); + } else { + /* possible found one */ + if (get_entity_link(ent) == NULL) + ++res; + link_all_leave_sels(ent, succ); + } + } + } + } + + return res; } /** @@ -366,34 +363,32 @@ static int find_possible_replacements(ir_graph *irg) * @param sel the Sel node * @param len the length of the path so far */ -static path_t *find_path(ir_node *sel, unsigned len) -{ - int pos, i, n; - path_t *res; - ir_node *pred = get_Sel_ptr(sel); +static path_t *find_path(ir_node *sel, unsigned len) { + int pos, i, n; + path_t *res; + ir_node *pred = get_Sel_ptr(sel); - /* the current Sel node will add some path elements */ - n = get_Sel_n_indexs(sel); - len += n + 1; + /* the current Sel node will add some path elements */ + n = get_Sel_n_indexs(sel); + len += n + 1; - if (! is_Sel(pred)) { - /* we found the root */ + if (! is_Sel(pred)) { + /* we found the root */ - res = xmalloc(sizeof(*res) + (len - 1) * sizeof(res->path)); - res->path_len = len; - } - else - res = find_path(pred, len); + res = xmalloc(sizeof(*res) + (len - 1) * sizeof(res->path)); + res->path_len = len; + } else + res = find_path(pred, len); - pos = res->path_len - len; + pos = res->path_len - len; - res->path[pos++].ent = get_Sel_entity(sel); - for (i = 0; i < n; ++i) { - ir_node *index = get_Sel_index(sel, i); + res->path[pos++].ent = get_Sel_entity(sel); + for (i = 0; i < n; ++i) { + ir_node *index = get_Sel_index(sel, i); - res->path[pos++].tv = get_Const_tarval(index); - } - return res; + res->path[pos++].tv = get_Const_tarval(index); + } + return res; } @@ -411,302 +406,302 @@ static path_t *find_path(ir_node *sel, unsigned len) */ static unsigned allocate_value_numbers(pset *sels, ir_entity *ent, unsigned vnum, ir_mode ***modes) { - ir_node *sel, *next; - path_t *key, *path; - set *pathes = new_set(path_cmp, 8); + ir_node *sel, *next; + path_t *key, *path; + set *pathes = new_set(path_cmp, 8); - /* visit all Sel nodes in the chain of the entity */ - for (sel = get_entity_link(ent); sel; sel = next) { - next = get_irn_link(sel); + /* visit all Sel nodes in the chain of the entity */ + for (sel = get_entity_link(ent); sel; sel = next) { + next = get_irn_link(sel); - /* we must mark this sel for later */ - pset_insert_ptr(sels, sel); + /* we must mark this sel for later */ + pset_insert_ptr(sels, sel); - key = find_path(sel, 0); - path = set_find(pathes, key, PATH_SIZE(key), path_hash(key)); + key = find_path(sel, 0); + path = set_find(pathes, key, PATH_SIZE(key), path_hash(key)); - if (path) - SET_VNUM(sel, path->vnum); - else { - key->vnum = vnum++; + if (path) + SET_VNUM(sel, path->vnum); + else { + key->vnum = vnum++; - set_insert(pathes, key, PATH_SIZE(key), path_hash(key)); + set_insert(pathes, key, PATH_SIZE(key), path_hash(key)); - SET_VNUM(sel, key->vnum); - ARR_EXTO(ir_mode *, *modes, (key->vnum + 15) & ~15); + SET_VNUM(sel, key->vnum); + ARR_EXTO(ir_mode *, *modes, (int)((key->vnum + 15) & ~15)); - (*modes)[key->vnum] = get_type_mode(get_entity_type(get_Sel_entity(sel))); + (*modes)[key->vnum] = get_type_mode(get_entity_type(get_Sel_entity(sel))); - assert((*modes)[key->vnum] && "Value is not atomic"); + assert((*modes)[key->vnum] && "Value is not atomic"); #ifdef DEBUG_libfirm - /* Debug output */ - if (get_opt_scalar_replacement_verbose() && get_firm_verbosity() > 1) { - unsigned i; - printf(" %s", get_entity_name(key->path[0].ent)); - for (i = 1; i < key->path_len; ++i) { - if (is_entity(key->path[i].ent)) - printf(".%s", get_entity_name(key->path[i].ent)); - else - printf("[%ld]", get_tarval_long(key->path[i].tv)); - } - printf(" = %u (%s)\n", PTR_TO_INT(get_irn_link(sel)), get_mode_name((*modes)[key->vnum])); - } + /* Debug output */ + if (get_opt_scalar_replacement_verbose() && get_firm_verbosity() > 1) { + unsigned i; + printf(" %s", get_entity_name(key->path[0].ent)); + for (i = 1; i < key->path_len; ++i) { + if (is_entity(key->path[i].ent)) + printf(".%s", get_entity_name(key->path[i].ent)); + else + printf("[%ld]", get_tarval_long(key->path[i].tv)); + } + printf(" = %u (%s)\n", PTR_TO_INT(get_irn_link(sel)), get_mode_name((*modes)[key->vnum])); + } #endif /* DEBUG_libfirm */ - } - free(key); - } + } + free(key); + } - del_set(pathes); - set_entity_link(ent, NULL); - return vnum; + del_set(pathes); + set_entity_link(ent, NULL); + return vnum; } /** * A list entry for the fixing lists */ typedef struct _list_entry_t { - ir_node *node; /**< the node that must be fixed */ - unsigned vnum; /**< the value number of this node */ + ir_node *node; /**< the node that must be fixed */ + unsigned vnum; /**< the value number of this node */ } list_entry_t; /** * environment for memory walker */ typedef struct _env_t { - struct obstack obst; /**< a obstack for the value blocks */ - int nvals; /**< number of values */ - ir_mode **modes; /**< the modes of the values */ - list_entry_t *fix_phis; /**< list of all Phi nodes that must be fixed */ - list_entry_t *fix_loads; /**< list of all Load nodes that must be fixed */ - pset *sels; /**< A set of all Sel nodes that have a value number */ + struct obstack obst; /**< a obstack for the value blocks */ + int nvals; /**< number of values */ + ir_mode **modes; /**< the modes of the values */ + list_entry_t *fix_phis; /**< list of all Phi nodes that must be fixed */ + list_entry_t *fix_loads; /**< list of all Load nodes that must be fixed */ + pset *sels; /**< A set of all Sel nodes that have a value number */ } env_t; /** * topological walker. */ -static void topologic_walker(ir_node *node, void *ctx) -{ - env_t *env = ctx; - ir_op *op = get_irn_op(node); - ir_node *adr, *block, *mem, *unk, **value_arr, **in, *val; - ir_mode *mode; - unsigned vnum; - int i, j, n; - list_entry_t *l; - - if (op == op_Load) { - /* a load, check if we can resolve it */ - adr = get_Load_ptr(node); - - if (! is_Sel(adr)) - return; - - if (! pset_find_ptr(env->sels, adr)) - return; - - /* ok, we have a Load that will be replaced */ - vnum = GET_VNUM(adr); - - assert(vnum < (unsigned)env->nvals); - - block = get_nodes_block(node); - value_arr = get_irn_link(block); - - /* check, if we can replace this Load */ - if (value_arr[vnum]) { - mem = get_Load_mem(node); - - /* Beware: A Load can contain a hidden conversion in Firm. - This happens for instance in the following code: - - int i; - unsigned j = *(unsigned *)&i; - - Handle this here. */ - val = value_arr[vnum]; - mode = get_Load_mode(node); - if (mode != get_irn_mode(val)) - val = new_d_Conv(get_irn_dbg_info(node), val, mode); - - turn_into_tuple(node, pn_Load_max); - set_Tuple_pred(node, pn_Load_M, mem); - set_Tuple_pred(node, pn_Load_res, val); - set_Tuple_pred(node, pn_Load_X_except, new_Bad()); - } else { - l = obstack_alloc(&env->obst, sizeof(*l)); - l->node = node; - l->vnum = vnum; - - set_irn_link(node, env->fix_loads); - env->fix_loads = l; - } - } else if (op == op_Store) { - /* a Store always can be replaced */ - adr = get_Store_ptr(node); - - if (! is_Sel(adr)) - return; - - if (! pset_find_ptr(env->sels, adr)) - return; - - vnum = GET_VNUM(adr); - - assert(vnum < (unsigned)env->nvals); - - block = get_nodes_block(node); - value_arr = get_irn_link(block); - - /* Beware: A Store can contain a hidden conversion in Firm. */ - val = get_Store_value(node); - if (get_irn_mode(val) != env->modes[vnum]) - val = new_d_Conv(get_irn_dbg_info(node), val, env->modes[vnum]); - value_arr[vnum] = val; - - mem = get_Store_mem(node); - - turn_into_tuple(node, pn_Store_max); - set_Tuple_pred(node, pn_Store_M, mem); - set_Tuple_pred(node, pn_Store_X_except, new_Bad()); - } else if (op == op_Phi && get_irn_mode(node) == mode_M) { - /* - * found a memory Phi: Here, we must create new Phi nodes - */ - block = get_nodes_block(node); - value_arr = get_irn_link(block); - - n = get_Block_n_cfgpreds(block); - - in = alloca(sizeof(*in) * n); - - for (i = env->nvals - 1; i >= 0; --i) { - unk = new_Unknown(env->modes[i]); - for (j = n - 1; j >= 0; --j) - in[j] = unk; - - value_arr[i] = new_r_Phi(current_ir_graph, block, n, in, env->modes[i]); - - l = obstack_alloc(&env->obst, sizeof(*l)); - l->node = value_arr[i]; - l->vnum = i; +static void topologic_walker(ir_node *node, void *ctx) { + env_t *env = ctx; + ir_op *op = get_irn_op(node); + ir_node *adr, *block, *mem, *unk, **value_arr, **in, *val; + ir_mode *mode; + unsigned vnum; + int i, j, n; + list_entry_t *l; + + if (op == op_Load) { + /* a load, check if we can resolve it */ + adr = get_Load_ptr(node); + + if (! is_Sel(adr)) + return; + + if (! pset_find_ptr(env->sels, adr)) + return; + + /* ok, we have a Load that will be replaced */ + vnum = GET_VNUM(adr); + + assert(vnum < (unsigned)env->nvals); + + block = get_nodes_block(node); + value_arr = get_irn_link(block); + + /* check, if we can replace this Load */ + if (value_arr[vnum]) { + mem = get_Load_mem(node); + + /* Beware: A Load can contain a hidden conversion in Firm. + This happens for instance in the following code: + + int i; + unsigned j = *(unsigned *)&i; + + Handle this here. */ + val = value_arr[vnum]; + mode = get_Load_mode(node); + if (mode != get_irn_mode(val)) + val = new_d_Conv(get_irn_dbg_info(node), val, mode); + + turn_into_tuple(node, pn_Load_max); + set_Tuple_pred(node, pn_Load_M, mem); + set_Tuple_pred(node, pn_Load_res, val); + set_Tuple_pred(node, pn_Load_X_regular, new_r_Jmp(current_ir_graph, block)); + set_Tuple_pred(node, pn_Load_X_except, new_Bad()); + } else { + l = obstack_alloc(&env->obst, sizeof(*l)); + l->node = node; + l->vnum = vnum; + + set_irn_link(node, env->fix_loads); + env->fix_loads = l; + } + } else if (op == op_Store) { + /* a Store always can be replaced */ + adr = get_Store_ptr(node); + + if (! is_Sel(adr)) + return; + + if (! pset_find_ptr(env->sels, adr)) + return; + + vnum = GET_VNUM(adr); + + assert(vnum < (unsigned)env->nvals); + + block = get_nodes_block(node); + value_arr = get_irn_link(block); + + /* Beware: A Store can contain a hidden conversion in Firm. */ + val = get_Store_value(node); + if (get_irn_mode(val) != env->modes[vnum]) + val = new_d_Conv(get_irn_dbg_info(node), val, env->modes[vnum]); + value_arr[vnum] = val; + + mem = get_Store_mem(node); + block = get_nodes_block(node); + + turn_into_tuple(node, pn_Store_max); + set_Tuple_pred(node, pn_Store_M, mem); + set_Tuple_pred(node, pn_Store_X_regular, new_r_Jmp(current_ir_graph, block)); + set_Tuple_pred(node, pn_Store_X_except, new_Bad()); + } else if (op == op_Phi && get_irn_mode(node) == mode_M) { + /* + * found a memory Phi: Here, we must create new Phi nodes + */ + block = get_nodes_block(node); + value_arr = get_irn_link(block); + + n = get_Block_n_cfgpreds(block); + + in = alloca(sizeof(*in) * n); + + for (i = env->nvals - 1; i >= 0; --i) { + unk = new_Unknown(env->modes[i]); + for (j = n - 1; j >= 0; --j) + in[j] = unk; + + value_arr[i] = new_r_Phi(current_ir_graph, block, n, in, env->modes[i]); + + l = obstack_alloc(&env->obst, sizeof(*l)); + l->node = value_arr[i]; + l->vnum = i; - set_irn_link(value_arr[i], env->fix_phis); - env->fix_phis = l; - } - } + set_irn_link(value_arr[i], env->fix_phis); + env->fix_phis = l; + } + } } /** * Walker: allocate the value array for every block. */ -static void alloc_value_arr(ir_node *block, void *ctx) -{ - env_t *env = ctx; - ir_node **var_arr = obstack_alloc(&env->obst, sizeof(*var_arr) * env->nvals); +static void alloc_value_arr(ir_node *block, void *ctx) { + env_t *env = ctx; + ir_node **var_arr = obstack_alloc(&env->obst, sizeof(*var_arr) * env->nvals); - /* the value array is empty at start */ - memset(var_arr, 0, sizeof(*var_arr) * env->nvals); - set_irn_link(block, var_arr); + /* the value array is empty at start */ + memset(var_arr, 0, sizeof(*var_arr) * env->nvals); + set_irn_link(block, var_arr); } /** * searches through blocks beginning from block for value * vnum and return it. */ -static ir_node *find_vnum_value(ir_node *block, unsigned vnum) -{ - ir_node **value_arr; - int i; - ir_node *res; +static ir_node *find_vnum_value(ir_node *block, unsigned vnum) { + ir_node **value_arr; + int i; + ir_node *res; - if (Block_not_block_visited(block)) { - mark_Block_block_visited(block); + if (Block_not_block_visited(block)) { + mark_Block_block_visited(block); - value_arr = get_irn_link(block); + value_arr = get_irn_link(block); - if (value_arr[vnum]) - return value_arr[vnum]; + if (value_arr[vnum]) + return value_arr[vnum]; - for (i = get_Block_n_cfgpreds(block) - 1; i >= 0; --i) { - ir_node *pred = get_Block_cfgpred(block, i); + for (i = get_Block_n_cfgpreds(block) - 1; i >= 0; --i) { + ir_node *pred = get_Block_cfgpred(block, i); - res = find_vnum_value(get_nodes_block(pred), vnum); - if (res) - return res; - } - } - return NULL; + res = find_vnum_value(get_nodes_block(pred), vnum); + if (res) + return res; + } + } + return NULL; } /** * fix the Phi list */ -static void fix_phis(env_t *env) -{ - list_entry_t *l; - ir_node *phi, *block, *pred, *val; - int i; - - for (l = env->fix_phis; l; l = get_irn_link(phi)) { - phi = l->node; - - block = get_nodes_block(phi); - for (i = get_irn_arity(phi) - 1; i >= 0; --i) { - pred = get_Block_cfgpred(block, i); - pred = get_nodes_block(pred); - - inc_irg_block_visited(current_ir_graph); - val = find_vnum_value(pred, l->vnum); - - if (val) - set_irn_n(phi, i, val); - } - } +static void fix_phis(env_t *env) { + list_entry_t *l; + ir_node *phi, *block, *pred, *val; + int i; + + for (l = env->fix_phis; l; l = get_irn_link(phi)) { + phi = l->node; + + block = get_nodes_block(phi); + for (i = get_irn_arity(phi) - 1; i >= 0; --i) { + pred = get_Block_cfgpred(block, i); + pred = get_nodes_block(pred); + + inc_irg_block_visited(current_ir_graph); + val = find_vnum_value(pred, l->vnum); + + if (val) + set_irn_n(phi, i, val); + } + } } /** * fix the Load list */ -static void fix_loads(env_t *env) -{ - list_entry_t *l; - ir_node *load, *block, *pred, *val = NULL, *mem; - ir_mode *mode; - int i; - - for (l = env->fix_loads; l; l = get_irn_link(load)) { - load = l->node; - - block = get_nodes_block(load); - for (i = get_Block_n_cfgpreds(block) - 1; i >= 0; --i) { - pred = get_Block_cfgpred(block, i); - pred = get_nodes_block(pred); - - inc_irg_block_visited(current_ir_graph); - val = find_vnum_value(pred, l->vnum); - - if (val) - break; - } - - if (! val) { - /* access of an uninitialized value */ - val = new_Unknown(env->modes[l->vnum]); - } - - mem = get_Load_mem(load); - /* Beware: A Load can contain a hidden conversion in Firm. - Handle this here. */ - mode = get_Load_mode(load); - if (mode != get_irn_mode(val)) - val = new_d_Conv(get_irn_dbg_info(load), val, mode); - - turn_into_tuple(load, pn_Load_max); - set_Tuple_pred(load, pn_Load_M, mem); - set_Tuple_pred(load, pn_Load_res, val); - set_Tuple_pred(load, pn_Load_X_except, new_Bad()); - } +static void fix_loads(env_t *env) { + list_entry_t *l; + ir_node *load, *block, *pred, *val = NULL, *mem; + ir_mode *mode; + int i; + + for (l = env->fix_loads; l; l = get_irn_link(load)) { + load = l->node; + + block = get_nodes_block(load); + for (i = get_Block_n_cfgpreds(block) - 1; i >= 0; --i) { + pred = get_Block_cfgpred(block, i); + pred = get_nodes_block(pred); + + inc_irg_block_visited(current_ir_graph); + val = find_vnum_value(pred, l->vnum); + + if (val) + break; + } + + if (! val) { + /* access of an uninitialized value */ + val = new_Unknown(env->modes[l->vnum]); + } + + /* Beware: A Load can contain a hidden conversion in Firm. + Handle this here. */ + mode = get_Load_mode(load); + if (mode != get_irn_mode(val)) + val = new_d_Conv(get_irn_dbg_info(load), val, mode); + + mem = get_Load_mem(load); + + turn_into_tuple(load, pn_Load_max); + set_Tuple_pred(load, pn_Load_M, mem); + set_Tuple_pred(load, pn_Load_res, val); + set_Tuple_pred(load, pn_Load_X_regular, new_r_Jmp(current_ir_graph, block)); + set_Tuple_pred(load, pn_Load_X_except, new_Bad()); + } } /** @@ -717,31 +712,30 @@ static void fix_loads(env_t *env) * @param modes A flexible array, containing all the modes of * the value numbers. */ -static void do_scalar_replacements(pset *sels, int nvals, ir_mode **modes) -{ - env_t env; - - obstack_init(&env.obst); - env.nvals = nvals; - env.modes = modes; - env.fix_phis = NULL; - env.fix_loads = NULL; - env.sels = sels; - - /* first step: allocate the value arrays for every block */ - irg_block_walk_graph(current_ir_graph, NULL, alloc_value_arr, &env); - - /* - * second step: walk over the graph blockwise in topological order - * and fill the array as much as possible. - */ - irg_walk_blkwise_graph(current_ir_graph, NULL, topologic_walker, &env); - - /* third, fix the list of Phis, then the list of Loads */ - fix_phis(&env); - fix_loads(&env); - - obstack_free(&env.obst, NULL); +static void do_scalar_replacements(pset *sels, int nvals, ir_mode **modes) { + env_t env; + + obstack_init(&env.obst); + env.nvals = nvals; + env.modes = modes; + env.fix_phis = NULL; + env.fix_loads = NULL; + env.sels = sels; + + /* first step: allocate the value arrays for every block */ + irg_block_walk_graph(current_ir_graph, NULL, alloc_value_arr, &env); + + /* + * second step: walk over the graph blockwise in topological order + * and fill the array as much as possible. + */ + irg_walk_blkwise_graph(current_ir_graph, NULL, topologic_walker, &env); + + /* third, fix the list of Phis, then the list of Loads */ + fix_phis(&env); + fix_loads(&env); + + obstack_free(&env.obst, NULL); } /* @@ -749,100 +743,99 @@ static void do_scalar_replacements(pset *sels, int nvals, ir_mode **modes) * * @param irg The current ir graph. */ -void scalar_replacement_opt(ir_graph *irg) -{ - unsigned nvals; - int i; - scalars_t key, *value; - ir_node *irg_frame; - ir_mode **modes; - set *set_ent; - pset *sels; - ir_type *ent_type; - ir_graph *rem; - - if (! get_opt_scalar_replacement()) - return; - - rem = current_ir_graph; - - /* Call algorithm that computes the out edges */ - assure_irg_outs(irg); - - /* Find possible scalar replacements */ - if (find_possible_replacements(irg)) { - - if (get_opt_scalar_replacement_verbose()) { - printf("Scalar Replacement: %s\n", get_entity_name(get_irg_entity(irg))); - } - - /* Insert in set the scalar replacements. */ - irg_frame = get_irg_frame(irg); - nvals = 0; - modes = NEW_ARR_F(ir_mode *, 16); - set_ent = new_set(ent_cmp, 8); - sels = pset_new_ptr(8); - - for (i = 0 ; i < get_irn_n_outs(irg_frame); i++) { - ir_node *succ = get_irn_out(irg_frame, i); - - if (is_Sel(succ)) { - ir_entity *ent = get_Sel_entity(succ); - - if (get_entity_link(ent) == NULL || get_entity_link(ent) == ADDRESS_TAKEN) - continue; - - ent_type = get_entity_type(ent); - - key.ent = ent; - key.ent_owner = get_entity_owner(ent); - set_insert(set_ent, &key, sizeof(key), HASH_PTR(key.ent)); - - if (get_opt_scalar_replacement_verbose()) { - if (is_Array_type(ent_type)) { - printf(" found array %s\n", get_entity_name(ent)); - } - else if (is_Struct_type(ent_type)) { - printf(" found struct %s\n", get_entity_name(ent)); - } - else if (is_atomic_type(ent_type)) - printf(" found atomic value %s\n", get_entity_name(ent)); - else { - assert(0 && "Neither an array nor a struct or atomic value"); - } - } - - nvals = allocate_value_numbers(sels, ent, nvals, &modes); - } - } - - if (get_opt_scalar_replacement_verbose()) { - printf(" %u values will be needed\n", nvals); - } - - /* If scalars were found. */ - if (nvals) { - do_scalar_replacements(sels, nvals, modes); - - for (value = set_first(set_ent); value; value = set_next(set_ent)) { - remove_class_member(value->ent_owner, value->ent); - } - } - - del_pset(sels); - del_set(set_ent); - DEL_ARR_F(modes); - - if (nvals) { - /* - * We changed the graph, but did NOT introduce new blocks - * neither changed control flow, cf-backedges should be still - * consistent. - */ - set_irg_outs_inconsistent(irg); - set_irg_loopinfo_inconsistent(irg); - } - } - - current_ir_graph = rem; +void scalar_replacement_opt(ir_graph *irg) { + unsigned nvals; + int i; + scalars_t key, *value; + ir_node *irg_frame; + ir_mode **modes; + set *set_ent; + pset *sels; + ir_type *ent_type; + ir_graph *rem; + + if (! get_opt_scalar_replacement()) + return; + + rem = current_ir_graph; + + /* Call algorithm that computes the out edges */ + assure_irg_outs(irg); + + /* Find possible scalar replacements */ + if (find_possible_replacements(irg)) { + + if (get_opt_scalar_replacement_verbose()) { + printf("Scalar Replacement: %s\n", get_entity_name(get_irg_entity(irg))); + } + + /* Insert in set the scalar replacements. */ + irg_frame = get_irg_frame(irg); + nvals = 0; + modes = NEW_ARR_F(ir_mode *, 16); + set_ent = new_set(ent_cmp, 8); + sels = pset_new_ptr(8); + + for (i = 0 ; i < get_irn_n_outs(irg_frame); i++) { + ir_node *succ = get_irn_out(irg_frame, i); + + if (is_Sel(succ)) { + ir_entity *ent = get_Sel_entity(succ); + + if (get_entity_link(ent) == NULL || get_entity_link(ent) == ADDRESS_TAKEN) + continue; + + ent_type = get_entity_type(ent); + + key.ent = ent; + key.ent_owner = get_entity_owner(ent); + set_insert(set_ent, &key, sizeof(key), HASH_PTR(key.ent)); + + if (get_opt_scalar_replacement_verbose()) { + if (is_Array_type(ent_type)) { + printf(" found array %s\n", get_entity_name(ent)); + } + else if (is_Struct_type(ent_type)) { + printf(" found struct %s\n", get_entity_name(ent)); + } + else if (is_atomic_type(ent_type)) + printf(" found atomic value %s\n", get_entity_name(ent)); + else { + assert(0 && "Neither an array nor a struct or atomic value"); + } + } + + nvals = allocate_value_numbers(sels, ent, nvals, &modes); + } + } + + if (get_opt_scalar_replacement_verbose()) { + printf(" %u values will be needed\n", nvals); + } + + /* If scalars were found. */ + if (nvals) { + do_scalar_replacements(sels, nvals, modes); + + for (value = set_first(set_ent); value; value = set_next(set_ent)) { + remove_class_member(value->ent_owner, value->ent); + } + } + + del_pset(sels); + del_set(set_ent); + DEL_ARR_F(modes); + + if (nvals) { + /* + * We changed the graph, but did NOT introduce new blocks + * neither changed control flow, cf-backedges should be still + * consistent. + */ + set_irg_outs_inconsistent(irg); + set_irg_loopinfo_inconsistent(irg); + } + } + + current_ir_graph = rem; }