2 * Copyright (C) 1995-2008 University of Karlsruhe. All right reserved.
4 * This file is part of libFirm.
6 * This file may be distributed and/or modified under the terms of the
7 * GNU General Public License version 2 as published by the Free Software
8 * Foundation and appearing in the file LICENSE.GPL included in the
9 * packaging of this file.
11 * Licensees holding valid libFirm Professional Edition licenses may use
12 * this file in accordance with the libFirm Commercial License.
13 * Agreement provided with the Software.
15 * This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
16 * WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * @brief Backend ABI implementation.
23 * @author Sebastian Hack, Michael Beck
32 #include "irgraph_t.h"
35 #include "iredges_t.h"
38 #include "irprintf_t.h"
45 #include "raw_bitset.h"
56 #include "bessaconstr.h"
59 DEBUG_ONLY(static firm_dbg_module_t *dbg;)
61 typedef struct be_abi_call_arg_t {
62 unsigned is_res : 1; /**< 1: the call argument is a return value. 0: it's a call parameter. */
63 unsigned in_reg : 1; /**< 1: this argument is transmitted in registers. */
64 unsigned on_stack : 1; /**< 1: this argument is transmitted on the stack. */
65 unsigned callee : 1; /**< 1: someone called us. 0: We call another function */
68 const arch_register_t *reg;
71 unsigned alignment; /**< stack alignment */
72 unsigned space_before; /**< allocate space before */
73 unsigned space_after; /**< allocate space after */
76 struct be_abi_call_t {
77 be_abi_call_flags_t flags; /**< Flags describing the ABI behavior on calls */
78 int pop; /**< number of bytes the stack frame is shrinked by the callee on return. */
79 const be_abi_callbacks_t *cb;
80 ir_type *between_type;
82 const arch_register_class_t *cls_addr; /**< register class of the call address */
86 * The ABI information for the current graph.
89 survive_dce_t *dce_survivor;
91 be_abi_call_t *call; /**< The ABI call information. */
93 ir_node *init_sp; /**< The node representing the stack pointer
94 at the start of the function. */
96 ir_node *start; /**< The be_Start params node. */
97 pmap *regs; /**< A map of all callee-save and ignore regs to
98 their Projs to the RegParams node. */
100 int start_block_bias; /**< The stack bias at the end of the start block. */
102 void *cb; /**< ABI Callback self pointer. */
104 pmap *keep_map; /**< mapping blocks to keep nodes. */
105 pset *ignore_regs; /**< Additional registers which shall be ignored. */
107 ir_node **calls; /**< flexible array containing all be_Call nodes */
110 static ir_heights_t *ir_heights;
112 /** Flag: if set, try to omit the frame pointer in all routines. */
113 static int be_omit_fp = 1;
115 /** Flag: if set, try to omit the frame pointer in leaf routines only. */
116 static int be_omit_leaf_fp = 1;
119 _ ____ ___ ____ _ _ _ _
120 / \ | __ )_ _| / ___|__ _| | | |__ __ _ ___| | _____
121 / _ \ | _ \| | | | / _` | | | '_ \ / _` |/ __| |/ / __|
122 / ___ \| |_) | | | |__| (_| | | | |_) | (_| | (__| <\__ \
123 /_/ \_\____/___| \____\__,_|_|_|_.__/ \__,_|\___|_|\_\___/
125 These callbacks are used by the backend to set the parameters
126 for a specific call type.
130 * Set compare function: compares two ABI call object arguments.
132 static int cmp_call_arg(const void *a, const void *b, size_t n)
134 const be_abi_call_arg_t *p = a, *q = b;
136 return !(p->is_res == q->is_res && p->pos == q->pos && p->callee == q->callee);
140 * Get an ABI call object argument.
142 * @param call the abi call
143 * @param is_res true for call results, false for call arguments
144 * @param pos position of the argument
145 * @param callee context type - if we are callee or caller
147 static be_abi_call_arg_t *get_call_arg(be_abi_call_t *call, int is_res, int pos, int callee)
149 be_abi_call_arg_t arg;
152 memset(&arg, 0, sizeof(arg));
157 hash = is_res * 128 + pos;
159 return set_find(call->params, &arg, sizeof(arg), hash);
163 * Set an ABI call object argument.
165 static void remember_call_arg(be_abi_call_arg_t *arg, be_abi_call_t *call, be_abi_context_t context)
167 unsigned hash = arg->is_res * 128 + arg->pos;
168 if (context & ABI_CONTEXT_CALLEE) {
170 set_insert(call->params, arg, sizeof(*arg), hash);
172 if (context & ABI_CONTEXT_CALLER) {
174 set_insert(call->params, arg, sizeof(*arg), hash);
178 /* Set the flags for a call. */
179 void be_abi_call_set_flags(be_abi_call_t *call, be_abi_call_flags_t flags, const be_abi_callbacks_t *cb)
185 /* Sets the number of bytes the stackframe is shrinked by the callee on return */
186 void be_abi_call_set_pop(be_abi_call_t *call, int pop)
192 /* Set register class for call address */
193 void be_abi_call_set_call_address_reg_class(be_abi_call_t *call, const arch_register_class_t *cls)
195 call->cls_addr = cls;
199 void be_abi_call_param_stack(be_abi_call_t *call, int arg_pos,
200 ir_mode *load_mode, unsigned alignment,
201 unsigned space_before, unsigned space_after,
202 be_abi_context_t context)
204 be_abi_call_arg_t arg;
205 memset(&arg, 0, sizeof(arg));
206 assert(alignment > 0 && "Alignment must be greater than 0");
208 arg.load_mode = load_mode;
209 arg.alignment = alignment;
210 arg.space_before = space_before;
211 arg.space_after = space_after;
215 remember_call_arg(&arg, call, context);
218 void be_abi_call_param_reg(be_abi_call_t *call, int arg_pos, const arch_register_t *reg, be_abi_context_t context)
220 be_abi_call_arg_t arg;
221 memset(&arg, 0, sizeof(arg));
228 remember_call_arg(&arg, call, context);
231 void be_abi_call_res_reg(be_abi_call_t *call, int arg_pos, const arch_register_t *reg, be_abi_context_t context)
233 be_abi_call_arg_t arg;
234 memset(&arg, 0, sizeof(arg));
241 remember_call_arg(&arg, call, context);
244 /* Get the flags of a ABI call object. */
245 be_abi_call_flags_t be_abi_call_get_flags(const be_abi_call_t *call)
251 * Constructor for a new ABI call object.
253 * @param cls_addr register class of the call address
255 * @return the new ABI call object
257 static be_abi_call_t *be_abi_call_new(const arch_register_class_t *cls_addr)
259 be_abi_call_t *call = XMALLOCZ(be_abi_call_t);
262 call->params = new_set(cmp_call_arg, 16);
264 call->cls_addr = cls_addr;
266 call->flags.bits.try_omit_fp = be_omit_fp | be_omit_leaf_fp;
272 * Destructor for an ABI call object.
274 static void be_abi_call_free(be_abi_call_t *call)
276 del_set(call->params);
281 * Initializes the frame layout from parts
283 * @param frame the stack layout that will be initialized
284 * @param args the stack argument layout type
285 * @param between the between layout type
286 * @param locals the method frame type
287 * @param stack_dir the stack direction: < 0 decreasing, > 0 increasing addresses
288 * @param param_map an array mapping method argument positions to the stack argument type
290 * @return the initialized stack layout
292 static be_stack_layout_t *stack_frame_init(be_stack_layout_t *frame, ir_type *args,
293 ir_type *between, ir_type *locals, int stack_dir,
294 ir_entity *param_map[])
296 frame->arg_type = args;
297 frame->between_type = between;
298 frame->frame_type = locals;
299 frame->initial_offset = 0;
300 frame->initial_bias = 0;
301 frame->stack_dir = stack_dir;
302 frame->order[1] = between;
303 frame->param_map = param_map;
306 frame->order[0] = args;
307 frame->order[2] = locals;
309 /* typical decreasing stack: locals have the
310 * lowest addresses, arguments the highest */
311 frame->order[0] = locals;
312 frame->order[2] = args;
324 Adjustment of the calls inside a graph.
329 * Transform a call node into a be_Call node.
331 * @param env The ABI environment for the current irg.
332 * @param irn The call node.
333 * @param curr_sp The stack pointer node to use.
334 * @return The stack pointer after the call.
336 static ir_node *adjust_call(be_abi_irg_t *env, ir_node *irn, ir_node *curr_sp)
338 ir_graph *irg = get_irn_irg(irn);
339 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
340 ir_type *call_tp = get_Call_type(irn);
341 ir_node *call_ptr = get_Call_ptr(irn);
342 int n_params = get_method_n_params(call_tp);
343 ir_node *curr_mem = get_Call_mem(irn);
344 ir_node *bl = get_nodes_block(irn);
346 int stack_dir = arch_env->stack_dir;
347 const arch_register_t *sp = arch_env->sp;
348 be_abi_call_t *call = be_abi_call_new(sp->reg_class);
349 ir_mode *mach_mode = sp->reg_class->mode;
350 int no_alloc = call->flags.bits.frame_is_setup_on_call;
351 int n_res = get_method_n_ress(call_tp);
352 int do_seq = call->flags.bits.store_args_sequential && !no_alloc;
354 ir_node *res_proj = NULL;
355 int n_reg_params = 0;
356 int n_stack_params = 0;
359 const arch_register_t **states = NEW_ARR_F(const arch_register_t*, 0);
360 const arch_register_t **destroyed_regs = NEW_ARR_F(const arch_register_t*, 0);
364 int n_reg_results = 0;
365 const ir_edge_t *edge;
367 int *stack_param_idx;
368 int i, n, destroy_all_regs;
371 /* Let the isa fill out the abi description for that call node. */
372 arch_env_get_call_abi(arch_env, call_tp, call);
374 /* Insert code to put the stack arguments on the stack. */
375 assert(get_Call_n_params(irn) == n_params);
376 stack_param_idx = ALLOCAN(int, n_params);
377 for (i = 0; i < n_params; ++i) {
378 be_abi_call_arg_t *arg = get_call_arg(call, 0, i, 0);
381 int arg_size = get_type_size_bytes(get_method_param_type(call_tp, i));
383 stack_size += round_up2(arg->space_before, arg->alignment);
384 stack_size += round_up2(arg_size, arg->alignment);
385 stack_size += round_up2(arg->space_after, arg->alignment);
387 stack_param_idx[n_stack_params++] = i;
391 /* Collect all arguments which are passed in registers. */
392 reg_param_idxs = ALLOCAN(int, n_params);
393 for (i = 0; i < n_params; ++i) {
394 be_abi_call_arg_t *arg = get_call_arg(call, 0, i, 0);
395 if (arg && arg->in_reg) {
396 reg_param_idxs[n_reg_params++] = i;
401 * If the stack is decreasing and we do not want to store sequentially,
402 * or someone else allocated the call frame
403 * we allocate as much space on the stack all parameters need, by
404 * moving the stack pointer along the stack's direction.
406 * Note: we also have to do this for stack_size == 0, because we may have
407 * to adjust stack alignment for the call.
409 if (stack_dir < 0 && !do_seq && !no_alloc) {
410 curr_sp = be_new_IncSP(sp, bl, curr_sp, stack_size, 1);
413 dbgi = get_irn_dbg_info(irn);
414 /* If there are some parameters which shall be passed on the stack. */
415 if (n_stack_params > 0) {
417 ir_node **in = ALLOCAN(ir_node*, n_stack_params+1);
421 * Reverse list of stack parameters if call arguments are from left to right.
422 * We must them reverse again if they are pushed (not stored) and the stack
423 * direction is downwards.
425 if (call->flags.bits.left_to_right ^ (do_seq && stack_dir < 0)) {
426 for (i = 0; i < n_stack_params >> 1; ++i) {
427 int other = n_stack_params - i - 1;
428 int tmp = stack_param_idx[i];
429 stack_param_idx[i] = stack_param_idx[other];
430 stack_param_idx[other] = tmp;
434 curr_mem = get_Call_mem(irn);
436 in[n_in++] = curr_mem;
439 for (i = 0; i < n_stack_params; ++i) {
440 int p = stack_param_idx[i];
441 be_abi_call_arg_t *arg = get_call_arg(call, 0, p, 0);
442 ir_node *param = get_Call_param(irn, p);
443 ir_node *addr = curr_sp;
445 ir_type *param_type = get_method_param_type(call_tp, p);
446 int param_size = get_type_size_bytes(param_type) + arg->space_after;
449 * If we wanted to build the arguments sequentially,
450 * the stack pointer for the next must be incremented,
451 * and the memory value propagated.
455 addr = curr_sp = be_new_IncSP(sp, bl, curr_sp,
456 param_size + arg->space_before, 0);
457 add_irn_dep(curr_sp, curr_mem);
459 curr_ofs += arg->space_before;
460 curr_ofs = round_up2(curr_ofs, arg->alignment);
462 /* Make the expression to compute the argument's offset. */
464 ir_mode *constmode = mach_mode;
465 if (mode_is_reference(mach_mode)) {
468 addr = new_r_Const_long(irg, constmode, curr_ofs);
469 addr = new_r_Add(bl, curr_sp, addr, mach_mode);
473 /* Insert a store for primitive arguments. */
474 if (is_atomic_type(param_type)) {
476 ir_node *mem_input = do_seq ? curr_mem : new_NoMem();
477 store = new_rd_Store(dbgi, bl, mem_input, addr, param, 0);
478 mem = new_r_Proj(store, mode_M, pn_Store_M);
480 /* Make a mem copy for compound arguments. */
483 assert(mode_is_reference(get_irn_mode(param)));
484 copy = new_rd_CopyB(dbgi, bl, curr_mem, addr, param, param_type);
485 mem = new_r_Proj(copy, mode_M, pn_CopyB_M);
488 curr_ofs += param_size;
496 /* We need the sync only, if we didn't build the stores sequentially. */
498 if (n_stack_params >= 1) {
499 curr_mem = new_r_Sync(bl, n_in, in);
501 curr_mem = get_Call_mem(irn);
506 /* check for the return_twice property */
507 destroy_all_regs = 0;
508 if (is_SymConst_addr_ent(call_ptr)) {
509 ir_entity *ent = get_SymConst_entity(call_ptr);
511 if (get_entity_additional_properties(ent) & mtp_property_returns_twice)
512 destroy_all_regs = 1;
514 ir_type *call_tp = get_Call_type(irn);
516 if (get_method_additional_properties(call_tp) & mtp_property_returns_twice)
517 destroy_all_regs = 1;
520 /* Put caller save into the destroyed set and state registers in the states
522 for (i = 0, n = arch_env_get_n_reg_class(arch_env); i < n; ++i) {
524 const arch_register_class_t *cls = arch_env_get_reg_class(arch_env, i);
525 for (j = 0; j < cls->n_regs; ++j) {
526 const arch_register_t *reg = arch_register_for_index(cls, j);
528 /* even if destroyed all is specified, neither SP nor FP are
529 * destroyed (else bad things will happen) */
530 if (reg == arch_env->sp || reg == arch_env->bp)
533 if (arch_register_type_is(reg, state)) {
534 ARR_APP1(const arch_register_t*, destroyed_regs, reg);
535 ARR_APP1(const arch_register_t*, states, reg);
536 /* we're already in the destroyed set so no need for further
540 if (destroy_all_regs || arch_register_type_is(reg, caller_save)) {
541 if (! arch_register_type_is(reg, ignore)) {
542 ARR_APP1(const arch_register_t*, destroyed_regs, reg);
548 /* search the largest result proj number */
549 res_projs = ALLOCANZ(ir_node*, n_res);
551 foreach_out_edge(irn, edge) {
552 const ir_edge_t *res_edge;
553 ir_node *irn = get_edge_src_irn(edge);
555 if (!is_Proj(irn) || get_Proj_proj(irn) != pn_Call_T_result)
558 foreach_out_edge(irn, res_edge) {
560 ir_node *res = get_edge_src_irn(res_edge);
562 assert(is_Proj(res));
564 proj = get_Proj_proj(res);
565 assert(proj < n_res);
566 assert(res_projs[proj] == NULL);
567 res_projs[proj] = res;
573 /** TODO: this is not correct for cases where return values are passed
574 * on the stack, but no known ABI does this currently...
576 n_reg_results = n_res;
579 in = ALLOCAN(ir_node*, n_reg_params + ARR_LEN(states));
581 /* make the back end call node and set its register requirements. */
582 for (i = 0; i < n_reg_params; ++i) {
583 in[n_ins++] = get_Call_param(irn, reg_param_idxs[i]);
586 /* add state registers ins */
587 for (i = 0; i < ARR_LEN(states); ++i) {
588 const arch_register_t *reg = states[i];
589 const arch_register_class_t *cls = arch_register_get_class(reg);
591 ir_node *regnode = be_abi_reg_map_get(env->regs, reg);
592 ir_fprintf(stderr, "Adding %+F\n", regnode);
594 ir_node *regnode = new_r_Unknown(irg, arch_register_class_mode(cls));
595 in[n_ins++] = regnode;
597 assert(n_ins == (int) (n_reg_params + ARR_LEN(states)));
599 /* ins collected, build the call */
600 if (env->call->flags.bits.call_has_imm && is_SymConst(call_ptr)) {
602 low_call = be_new_Call(dbgi, irg, bl, curr_mem, curr_sp, curr_sp,
603 n_reg_results + pn_be_Call_first_res + ARR_LEN(destroyed_regs),
604 n_ins, in, get_Call_type(irn));
605 be_Call_set_entity(low_call, get_SymConst_entity(call_ptr));
608 low_call = be_new_Call(dbgi, irg, bl, curr_mem, curr_sp, call_ptr,
609 n_reg_results + pn_be_Call_first_res + ARR_LEN(destroyed_regs),
610 n_ins, in, get_Call_type(irn));
612 be_Call_set_pop(low_call, call->pop);
614 /* put the call into the list of all calls for later processing */
615 ARR_APP1(ir_node *, env->calls, low_call);
617 /* create new stack pointer */
618 curr_sp = new_r_Proj(low_call, get_irn_mode(curr_sp), pn_be_Call_sp);
619 be_set_constr_single_reg_out(low_call, pn_be_Call_sp, sp,
620 arch_register_req_type_ignore | arch_register_req_type_produces_sp);
621 arch_set_irn_register(curr_sp, sp);
623 /* now handle results */
624 for (i = 0; i < n_res; ++i) {
626 ir_node *proj = res_projs[i];
627 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 0);
629 /* returns values on stack not supported yet */
633 shift the proj number to the right, since we will drop the
634 unspeakable Proj_T from the Call. Therefore, all real argument
635 Proj numbers must be increased by pn_be_Call_first_res
637 pn = i + pn_be_Call_first_res;
640 ir_type *res_type = get_method_res_type(call_tp, i);
641 ir_mode *mode = get_type_mode(res_type);
642 proj = new_r_Proj(low_call, mode, pn);
645 set_Proj_pred(proj, low_call);
646 set_Proj_proj(proj, pn);
650 /* remove register from destroyed regs */
652 int n = ARR_LEN(destroyed_regs);
653 for (j = 0; j < n; ++j) {
654 if (destroyed_regs[j] == arg->reg) {
655 destroyed_regs[j] = destroyed_regs[n-1];
656 ARR_SHRINKLEN(destroyed_regs,n-1);
664 Set the register class of the call address to
665 the backend provided class (default: stack pointer class)
667 be_node_set_reg_class_in(low_call, be_pos_Call_ptr, call->cls_addr);
669 DBG((dbg, LEVEL_3, "\tcreated backend call %+F\n", low_call));
671 /* Set the register classes and constraints of the Call parameters. */
672 for (i = 0; i < n_reg_params; ++i) {
673 int index = reg_param_idxs[i];
674 be_abi_call_arg_t *arg = get_call_arg(call, 0, index, 0);
675 assert(arg->reg != NULL);
677 be_set_constr_single_reg_in(low_call, be_pos_Call_first_arg + i,
681 /* Set the register constraints of the results. */
682 for (i = 0; i < n_res; ++i) {
683 ir_node *proj = res_projs[i];
684 const be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 0);
685 int pn = get_Proj_proj(proj);
688 be_set_constr_single_reg_out(low_call, pn, arg->reg, 0);
689 arch_set_irn_register(proj, arg->reg);
691 exchange(irn, low_call);
693 /* kill the ProjT node */
694 if (res_proj != NULL) {
698 /* Make additional projs for the caller save registers
699 and the Keep node which keeps them alive. */
704 int curr_res_proj = pn_be_Call_first_res + n_reg_results;
707 n_ins = ARR_LEN(destroyed_regs) + n_reg_results + 1;
708 in = ALLOCAN(ir_node *, n_ins);
710 /* also keep the stack pointer */
711 set_irn_link(curr_sp, (void*) sp);
714 for (i = 0; i < ARR_LEN(destroyed_regs); ++i) {
715 const arch_register_t *reg = destroyed_regs[i];
716 ir_node *proj = new_r_Proj(low_call, reg->reg_class->mode, curr_res_proj);
718 /* memorize the register in the link field. we need afterwards to set the register class of the keep correctly. */
719 be_set_constr_single_reg_out(low_call, curr_res_proj, reg, 0);
720 arch_set_irn_register(proj, reg);
722 set_irn_link(proj, (void*) reg);
727 for (i = 0; i < n_reg_results; ++i) {
728 ir_node *proj = res_projs[i];
729 const arch_register_t *reg = arch_get_irn_register(proj);
730 set_irn_link(proj, (void*) reg);
735 /* create the Keep for the caller save registers */
736 keep = be_new_Keep(bl, n, in);
737 for (i = 0; i < n; ++i) {
738 const arch_register_t *reg = get_irn_link(in[i]);
739 be_node_set_reg_class_in(keep, i, reg->reg_class);
743 /* Clean up the stack. */
744 assert(stack_size >= call->pop);
745 stack_size -= call->pop;
747 if (stack_size > 0) {
748 ir_node *mem_proj = NULL;
750 foreach_out_edge(low_call, edge) {
751 ir_node *irn = get_edge_src_irn(edge);
752 if (is_Proj(irn) && get_Proj_proj(irn) == pn_Call_M) {
759 mem_proj = new_r_Proj(low_call, mode_M, pn_be_Call_M_regular);
760 keep_alive(mem_proj);
763 /* Clean up the stack frame or revert alignment fixes if we allocated it */
765 curr_sp = be_new_IncSP(sp, bl, curr_sp, -stack_size, 0);
768 be_abi_call_free(call);
771 DEL_ARR_F(destroyed_regs);
777 * Adjust the size of a node representing a stack alloc or free for the minimum stack alignment.
779 * @param alignment the minimum stack alignment
780 * @param size the node containing the non-aligned size
781 * @param block the block where new nodes are allocated on
782 * @param dbg debug info for new nodes
784 * @return a node representing the aligned size
786 static ir_node *adjust_alloc_size(unsigned stack_alignment, ir_node *size,
787 ir_node *block, dbg_info *dbg)
789 if (stack_alignment > 1) {
795 assert(is_po2(stack_alignment));
797 mode = get_irn_mode(size);
798 tv = new_tarval_from_long(stack_alignment-1, mode);
799 irg = get_Block_irg(block);
800 mask = new_r_Const(irg, tv);
801 size = new_rd_Add(dbg, block, size, mask, mode);
803 tv = new_tarval_from_long(-(long)stack_alignment, mode);
804 mask = new_r_Const(irg, tv);
805 size = new_rd_And(dbg, block, size, mask, mode);
811 * The alloca is transformed into a back end alloca node and connected to the stack nodes.
813 static ir_node *adjust_alloc(be_abi_irg_t *env, ir_node *alloc, ir_node *curr_sp)
815 ir_node *block = get_nodes_block(alloc);
816 ir_graph *irg = get_Block_irg(block);
817 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
818 ir_node *alloc_mem = NULL;
819 ir_node *alloc_res = NULL;
820 ir_type *type = get_Alloc_type(alloc);
823 const ir_edge_t *edge;
828 unsigned stack_alignment;
830 /* all non-stack Alloc nodes should already be lowered before the backend */
831 assert(get_Alloc_where(alloc) == stack_alloc);
833 foreach_out_edge(alloc, edge) {
834 ir_node *irn = get_edge_src_irn(edge);
836 assert(is_Proj(irn));
837 switch (get_Proj_proj(irn)) {
849 /* Beware: currently Alloc nodes without a result might happen,
850 only escape analysis kills them and this phase runs only for object
851 oriented source. We kill the Alloc here. */
852 if (alloc_res == NULL && alloc_mem) {
853 exchange(alloc_mem, get_Alloc_mem(alloc));
857 dbg = get_irn_dbg_info(alloc);
858 count = get_Alloc_count(alloc);
860 /* we might need to multiply the count with the element size */
861 if (type != firm_unknown_type && get_type_size_bytes(type) != 1) {
862 ir_mode *mode = get_irn_mode(count);
863 tarval *tv = new_tarval_from_long(get_type_size_bytes(type),
865 ir_node *cnst = new_rd_Const(dbg, irg, tv);
866 size = new_rd_Mul(dbg, block, count, cnst, mode);
871 /* The stack pointer will be modified in an unknown manner.
872 We cannot omit it. */
873 env->call->flags.bits.try_omit_fp = 0;
875 stack_alignment = 1 << arch_env->stack_alignment;
876 size = adjust_alloc_size(stack_alignment, size, block, dbg);
877 new_alloc = be_new_AddSP(arch_env->sp, block, curr_sp, size);
878 set_irn_dbg_info(new_alloc, dbg);
880 if (alloc_mem != NULL) {
884 addsp_mem = new_r_Proj(new_alloc, mode_M, pn_be_AddSP_M);
886 /* We need to sync the output mem of the AddSP with the input mem
887 edge into the alloc node. */
888 ins[0] = get_Alloc_mem(alloc);
890 sync = new_r_Sync(block, 2, ins);
892 exchange(alloc_mem, sync);
895 exchange(alloc, new_alloc);
897 /* fix projnum of alloca res */
898 set_Proj_proj(alloc_res, pn_be_AddSP_res);
900 curr_sp = new_r_Proj(new_alloc, get_irn_mode(curr_sp), pn_be_AddSP_sp);
907 * The Free is transformed into a back end free node and connected to the stack nodes.
909 static ir_node *adjust_free(be_abi_irg_t *env, ir_node *free, ir_node *curr_sp)
911 ir_node *block = get_nodes_block(free);
912 ir_graph *irg = get_irn_irg(free);
913 ir_type *type = get_Free_type(free);
914 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
915 ir_mode *sp_mode = arch_env->sp->reg_class->mode;
916 dbg_info *dbg = get_irn_dbg_info(free);
917 ir_node *subsp, *mem, *res, *size, *sync;
919 unsigned stack_alignment;
921 /* all non-stack-alloc Free nodes should already be lowered before the
923 assert(get_Free_where(free) == stack_alloc);
925 /* we might need to multiply the size with the element size */
926 if (type != firm_unknown_type && get_type_size_bytes(type) != 1) {
927 tarval *tv = new_tarval_from_long(get_type_size_bytes(type), mode_Iu);
928 ir_node *cnst = new_rd_Const(dbg, irg, tv);
929 ir_node *mul = new_rd_Mul(dbg, block, get_Free_size(free),
933 size = get_Free_size(free);
936 stack_alignment = 1 << arch_env->stack_alignment;
937 size = adjust_alloc_size(stack_alignment, size, block, dbg);
939 /* The stack pointer will be modified in an unknown manner.
940 We cannot omit it. */
941 env->call->flags.bits.try_omit_fp = 0;
942 subsp = be_new_SubSP(arch_env->sp, block, curr_sp, size);
943 set_irn_dbg_info(subsp, dbg);
945 mem = new_r_Proj(subsp, mode_M, pn_be_SubSP_M);
946 res = new_r_Proj(subsp, sp_mode, pn_be_SubSP_sp);
948 /* we need to sync the memory */
949 in[0] = get_Free_mem(free);
951 sync = new_r_Sync(block, 2, in);
953 /* and make the AddSP dependent on the former memory */
954 add_irn_dep(subsp, get_Free_mem(free));
957 exchange(free, sync);
964 * Check if a node is somehow data dependent on another one.
965 * both nodes must be in the same basic block.
966 * @param n1 The first node.
967 * @param n2 The second node.
968 * @return 1, if n1 is data dependent (transitively) on n2, 0 if not.
970 static int dependent_on(ir_node *n1, ir_node *n2)
972 assert(get_nodes_block(n1) == get_nodes_block(n2));
974 return heights_reachable_in_block(ir_heights, n1, n2);
977 static int cmp_call_dependency(const void *c1, const void *c2)
979 ir_node *n1 = *(ir_node **) c1;
980 ir_node *n2 = *(ir_node **) c2;
983 Classical qsort() comparison function behavior:
984 0 if both elements are equal
985 1 if second is "smaller" that first
986 -1 if first is "smaller" that second
988 if (dependent_on(n1, n2))
991 if (dependent_on(n2, n1))
994 /* The nodes have no depth order, but we need a total order because qsort()
996 return get_irn_idx(n1) - get_irn_idx(n2);
1000 * Walker: links all Call/Alloc/Free nodes to the Block they are contained.
1001 * Clears the irg_is_leaf flag if a Call is detected.
1003 static void link_ops_in_block_walker(ir_node *irn, void *data)
1005 be_abi_irg_t *env = data;
1006 ir_opcode code = get_irn_opcode(irn);
1008 if (code == iro_Call ||
1009 (code == iro_Alloc && get_Alloc_where(irn) == stack_alloc) ||
1010 (code == iro_Free && get_Free_where(irn) == stack_alloc)) {
1011 ir_node *bl = get_nodes_block(irn);
1012 void *save = get_irn_link(bl);
1014 if (code == iro_Call)
1015 env->call->flags.bits.irg_is_leaf = 0;
1017 set_irn_link(irn, save);
1018 set_irn_link(bl, irn);
1021 if (code == iro_Builtin && get_Builtin_kind(irn) == ir_bk_return_address) {
1022 ir_node *param = get_Builtin_param(irn, 0);
1023 tarval *tv = get_Const_tarval(param);
1024 unsigned long value = get_tarval_long(tv);
1025 /* use ebp, so the climbframe algo works... */
1027 env->call->flags.bits.try_omit_fp = 0;
1034 * Process all Call/Alloc/Free nodes inside a basic block.
1035 * Note that the link field of the block must contain a linked list of all
1036 * Call nodes inside the Block. We first order this list according to data dependency
1037 * and that connect the calls together.
1039 static void process_ops_in_block(ir_node *bl, void *data)
1041 be_abi_irg_t *env = data;
1042 ir_node *curr_sp = env->init_sp;
1049 for (irn = get_irn_link(bl); irn != NULL; irn = get_irn_link(irn)) {
1053 nodes = ALLOCAN(ir_node*, n_nodes);
1054 for (irn = get_irn_link(bl), n = 0; irn; irn = get_irn_link(irn), ++n) {
1058 /* If there were call nodes in the block. */
1063 /* order the call nodes according to data dependency */
1064 qsort(nodes, n_nodes, sizeof(nodes[0]), cmp_call_dependency);
1066 for (i = n_nodes - 1; i >= 0; --i) {
1067 ir_node *irn = nodes[i];
1069 DBG((dbg, LEVEL_3, "\tprocessing call %+F\n", irn));
1070 switch (get_irn_opcode(irn)) {
1073 /* The stack pointer will be modified due to a call. */
1074 env->call->flags.bits.try_omit_fp = 0;
1076 curr_sp = adjust_call(env, irn, curr_sp);
1079 if (get_Alloc_where(irn) == stack_alloc)
1080 curr_sp = adjust_alloc(env, irn, curr_sp);
1083 if (get_Free_where(irn) == stack_alloc)
1084 curr_sp = adjust_free(env, irn, curr_sp);
1087 panic("invalid call");
1091 /* Keep the last stack state in the block by tying it to Keep node,
1092 * the proj from calls is already kept */
1093 if (curr_sp != env->init_sp &&
1094 !(is_Proj(curr_sp) && be_is_Call(get_Proj_pred(curr_sp)))) {
1096 keep = be_new_Keep(bl, 1, nodes);
1097 pmap_insert(env->keep_map, bl, keep);
1101 set_irn_link(bl, curr_sp);
1105 * Adjust all call nodes in the graph to the ABI conventions.
1107 static void process_calls(ir_graph *irg)
1109 be_abi_irg_t *abi = be_get_irg_abi(irg);
1111 abi->call->flags.bits.irg_is_leaf = 1;
1112 irg_walk_graph(irg, firm_clear_link, link_ops_in_block_walker, abi);
1114 ir_heights = heights_new(irg);
1115 irg_block_walk_graph(irg, NULL, process_ops_in_block, abi);
1116 heights_free(ir_heights);
1120 * Computes the stack argument layout type.
1121 * Changes a possibly allocated value param type by moving
1122 * entities to the stack layout type.
1124 * @param env the ABI environment
1125 * @param call the current call ABI
1126 * @param method_type the method type
1127 * @param val_param_tp the value parameter type, will be destroyed
1128 * @param param_map an array mapping method arguments to the stack layout type
1130 * @return the stack argument layout type
1132 static ir_type *compute_arg_type(be_abi_irg_t *env, ir_graph *irg,
1133 be_abi_call_t *call,
1134 ir_type *method_type, ir_type *val_param_tp,
1135 ir_entity ***param_map)
1137 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1138 int dir = env->call->flags.bits.left_to_right ? 1 : -1;
1139 int inc = arch_env->stack_dir * dir;
1140 int n = get_method_n_params(method_type);
1141 int curr = inc > 0 ? 0 : n - 1;
1142 struct obstack *obst = be_get_be_obst(irg);
1148 ident *id = get_entity_ident(get_irg_entity(irg));
1151 *param_map = map = OALLOCN(obst, ir_entity*, n);
1152 res = new_type_struct(id_mangle_u(id, new_id_from_chars("arg_type", 8)));
1153 for (i = 0; i < n; ++i, curr += inc) {
1154 ir_type *param_type = get_method_param_type(method_type, curr);
1155 be_abi_call_arg_t *arg = get_call_arg(call, 0, curr, 1);
1158 if (arg->on_stack) {
1159 if (val_param_tp != NULL) {
1160 /* the entity was already created, create a copy in the param type */
1161 ir_entity *val_ent = get_method_value_param_ent(method_type, i);
1162 arg->stack_ent = copy_entity_own(val_ent, res);
1163 set_entity_link(val_ent, arg->stack_ent);
1164 set_entity_link(arg->stack_ent, NULL);
1166 /* create a new entity */
1167 snprintf(buf, sizeof(buf), "param_%d", i);
1168 arg->stack_ent = new_entity(res, new_id_from_str(buf), param_type);
1170 ofs += arg->space_before;
1171 ofs = round_up2(ofs, arg->alignment);
1172 set_entity_offset(arg->stack_ent, ofs);
1173 ofs += arg->space_after;
1174 ofs += get_type_size_bytes(param_type);
1175 map[i] = arg->stack_ent;
1178 set_type_size_bytes(res, ofs);
1179 set_type_state(res, layout_fixed);
1184 const arch_register_t *reg;
1188 static int cmp_regs(const void *a, const void *b)
1190 const reg_node_map_t *p = a;
1191 const reg_node_map_t *q = b;
1193 if (p->reg->reg_class == q->reg->reg_class)
1194 return p->reg->index - q->reg->index;
1196 return p->reg->reg_class - q->reg->reg_class;
1199 static void reg_map_to_arr(reg_node_map_t *res, pmap *reg_map)
1202 int n = pmap_count(reg_map);
1205 foreach_pmap(reg_map, ent) {
1206 res[i].reg = ent->key;
1207 res[i].irn = ent->value;
1211 qsort(res, n, sizeof(res[0]), cmp_regs);
1215 * Creates a barrier.
1217 static ir_node *create_barrier(ir_node *bl, ir_node **mem, pmap *regs,
1220 int n_regs = pmap_count(regs);
1226 in = ALLOCAN(ir_node*, n_regs+1);
1227 rm = ALLOCAN(reg_node_map_t, n_regs);
1228 reg_map_to_arr(rm, regs);
1229 for (n = 0; n < n_regs; ++n) {
1237 irn = be_new_Barrier(bl, n, in);
1239 for (n = 0; n < n_regs; ++n) {
1240 ir_node *pred = rm[n].irn;
1241 const arch_register_t *reg = rm[n].reg;
1242 arch_register_type_t add_type = 0;
1244 const backend_info_t *info;
1246 /* stupid workaround for now... as not all nodes report register
1248 info = be_get_info(skip_Proj(pred));
1249 if (info != NULL && info->out_infos != NULL) {
1250 const arch_register_req_t *ireq = arch_get_register_req_out(pred);
1251 if (ireq->type & arch_register_req_type_ignore)
1252 add_type |= arch_register_req_type_ignore;
1253 if (ireq->type & arch_register_req_type_produces_sp)
1254 add_type |= arch_register_req_type_produces_sp;
1257 proj = new_r_Proj(irn, get_irn_mode(pred), n);
1258 be_node_set_reg_class_in(irn, n, reg->reg_class);
1260 be_set_constr_single_reg_in(irn, n, reg, 0);
1261 be_set_constr_single_reg_out(irn, n, reg, add_type);
1262 arch_set_irn_register(proj, reg);
1264 pmap_insert(regs, (void *) reg, proj);
1268 *mem = new_r_Proj(irn, mode_M, n);
1275 * Creates a be_Return for a Return node.
1277 * @param @env the abi environment
1278 * @param irn the Return node or NULL if there was none
1279 * @param bl the block where the be_Retun should be placed
1280 * @param mem the current memory
1281 * @param n_res number of return results
1283 static ir_node *create_be_return(be_abi_irg_t *env, ir_node *irn, ir_node *bl,
1284 ir_node *mem, int n_res)
1286 be_abi_call_t *call = env->call;
1287 ir_graph *irg = get_Block_irg(bl);
1288 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1290 pmap *reg_map = pmap_create();
1291 ir_node *keep = pmap_get(env->keep_map, bl);
1298 const arch_register_t **regs;
1302 get the valid stack node in this block.
1303 If we had a call in that block there is a Keep constructed by process_calls()
1304 which points to the last stack modification in that block. we'll use
1305 it then. Else we use the stack from the start block and let
1306 the ssa construction fix the usage.
1308 stack = be_abi_reg_map_get(env->regs, arch_env->sp);
1310 stack = get_irn_n(keep, 0);
1312 remove_End_keepalive(get_irg_end(irg), keep);
1315 /* Insert results for Return into the register map. */
1316 for (i = 0; i < n_res; ++i) {
1317 ir_node *res = get_Return_res(irn, i);
1318 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 1);
1319 assert(arg->in_reg && "return value must be passed in register");
1320 pmap_insert(reg_map, (void *) arg->reg, res);
1323 /* Add uses of the callee save registers. */
1324 foreach_pmap(env->regs, ent) {
1325 const arch_register_t *reg = ent->key;
1326 if (arch_register_type_is(reg, callee_save) || arch_register_type_is(reg, ignore))
1327 pmap_insert(reg_map, ent->key, ent->value);
1330 be_abi_reg_map_set(reg_map, arch_env->sp, stack);
1332 /* Make the Epilogue node and call the arch's epilogue maker. */
1333 create_barrier(bl, &mem, reg_map, 1);
1334 call->cb->epilogue(env->cb, bl, &mem, reg_map);
1337 Maximum size of the in array for Return nodes is
1338 return args + callee save/ignore registers + memory + stack pointer
1340 in_max = pmap_count(reg_map) + n_res + 2;
1342 in = ALLOCAN(ir_node*, in_max);
1343 regs = ALLOCAN(arch_register_t const*, in_max);
1346 in[1] = be_abi_reg_map_get(reg_map, arch_env->sp);
1348 regs[1] = arch_env->sp;
1351 /* clear SP entry, since it has already been grown. */
1352 pmap_insert(reg_map, (void *) arch_env->sp, NULL);
1353 for (i = 0; i < n_res; ++i) {
1354 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 1);
1356 in[n] = be_abi_reg_map_get(reg_map, arg->reg);
1357 regs[n++] = arg->reg;
1359 /* Clear the map entry to mark the register as processed. */
1360 be_abi_reg_map_set(reg_map, arg->reg, NULL);
1363 /* grow the rest of the stuff. */
1364 foreach_pmap(reg_map, ent) {
1367 regs[n++] = ent->key;
1371 /* The in array for the new back end return is now ready. */
1373 dbgi = get_irn_dbg_info(irn);
1377 /* we have to pop the shadow parameter in in case of struct returns */
1379 ret = be_new_Return(dbgi, irg, bl, n_res, pop, n, in);
1381 /* Set the register classes of the return's parameter accordingly. */
1382 for (i = 0; i < n; ++i) {
1383 if (regs[i] == NULL)
1386 be_node_set_reg_class_in(ret, i, regs[i]->reg_class);
1389 /* Free the space of the Epilog's in array and the register <-> proj map. */
1390 pmap_destroy(reg_map);
1395 typedef struct ent_pos_pair ent_pos_pair;
1396 struct ent_pos_pair {
1397 ir_entity *ent; /**< a value param entity */
1398 int pos; /**< its parameter number */
1399 ent_pos_pair *next; /**< for linking */
1402 typedef struct lower_frame_sels_env_t {
1403 ent_pos_pair *value_param_list; /**< the list of all value param entities */
1404 ir_node *frame; /**< the current frame */
1405 const arch_register_class_t *sp_class; /**< register class of the stack pointer */
1406 const arch_register_class_t *link_class; /**< register class of the link pointer */
1407 ir_type *value_tp; /**< the value type if any */
1408 ir_type *frame_tp; /**< the frame type */
1409 int static_link_pos; /**< argument number of the hidden static link */
1410 } lower_frame_sels_env_t;
1413 * Return an entity from the backend for an value param entity.
1415 * @param ent an value param type entity
1416 * @param ctx context
1418 static ir_entity *get_argument_entity(ir_entity *ent, lower_frame_sels_env_t *ctx)
1420 ir_entity *argument_ent = get_entity_link(ent);
1422 if (argument_ent == NULL) {
1423 /* we have NO argument entity yet: This is bad, as we will
1424 * need one for backing store.
1427 ir_type *frame_tp = ctx->frame_tp;
1428 unsigned offset = get_type_size_bytes(frame_tp);
1429 ir_type *tp = get_entity_type(ent);
1430 unsigned align = get_type_alignment_bytes(tp);
1432 offset += align - 1;
1433 offset &= ~(align - 1);
1435 argument_ent = copy_entity_own(ent, frame_tp);
1437 /* must be automatic to set a fixed layout */
1438 set_entity_offset(argument_ent, offset);
1439 offset += get_type_size_bytes(tp);
1441 set_type_size_bytes(frame_tp, offset);
1442 set_entity_link(ent, argument_ent);
1444 return argument_ent;
1447 * Walker: Replaces Sels of frame type and
1448 * value param type entities by FrameAddress.
1449 * Links all used entities.
1451 static void lower_frame_sels_walker(ir_node *irn, void *data)
1453 lower_frame_sels_env_t *ctx = data;
1456 ir_node *ptr = get_Sel_ptr(irn);
1458 if (ptr == ctx->frame) {
1459 ir_entity *ent = get_Sel_entity(irn);
1460 ir_node *bl = get_nodes_block(irn);
1463 int is_value_param = 0;
1465 if (get_entity_owner(ent) == ctx->value_tp) {
1468 /* replace by its copy from the argument type */
1469 pos = get_struct_member_index(ctx->value_tp, ent);
1470 ent = get_argument_entity(ent, ctx);
1473 nw = be_new_FrameAddr(ctx->sp_class, bl, ctx->frame, ent);
1476 /* check, if it's a param Sel and if have not seen this entity before */
1477 if (is_value_param && get_entity_link(ent) == NULL) {
1483 ARR_APP1(ent_pos_pair, ctx->value_param_list, pair);
1485 set_entity_link(ent, ctx->value_param_list);
1492 * Check if a value parameter is transmitted as a register.
1493 * This might happen if the address of an parameter is taken which is
1494 * transmitted in registers.
1496 * Note that on some architectures this case must be handled specially
1497 * because the place of the backing store is determined by their ABI.
1499 * In the default case we move the entity to the frame type and create
1500 * a backing store into the first block.
1502 static void fix_address_of_parameter_access(be_abi_irg_t *env, ir_graph *irg,
1503 ent_pos_pair *value_param_list)
1505 be_abi_call_t *call = env->call;
1506 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1507 ent_pos_pair *entry, *new_list;
1509 int i, n = ARR_LEN(value_param_list);
1512 for (i = 0; i < n; ++i) {
1513 int pos = value_param_list[i].pos;
1514 be_abi_call_arg_t *arg = get_call_arg(call, 0, pos, 1);
1517 DBG((dbg, LEVEL_2, "\targ #%d need backing store\n", pos));
1518 value_param_list[i].next = new_list;
1519 new_list = &value_param_list[i];
1522 if (new_list != NULL) {
1523 /* ok, change the graph */
1524 ir_node *start_bl = get_irg_start_block(irg);
1525 ir_node *first_bl = get_first_block_succ(start_bl);
1526 ir_node *frame, *imem, *nmem, *store, *mem, *args;
1527 optimization_state_t state;
1530 assert(first_bl && first_bl != start_bl);
1531 /* we had already removed critical edges, so the following
1532 assertion should be always true. */
1533 assert(get_Block_n_cfgpreds(first_bl) == 1);
1535 /* now create backing stores */
1536 frame = get_irg_frame(irg);
1537 imem = get_irg_initial_mem(irg);
1539 save_optimization_state(&state);
1541 nmem = new_r_Proj(get_irg_start(irg), mode_M, pn_Start_M);
1542 restore_optimization_state(&state);
1544 /* reroute all edges to the new memory source */
1545 edges_reroute(imem, nmem, irg);
1549 args = get_irg_args(irg);
1550 for (entry = new_list; entry != NULL; entry = entry->next) {
1552 ir_type *tp = get_entity_type(entry->ent);
1553 ir_mode *mode = get_type_mode(tp);
1556 /* address for the backing store */
1557 addr = be_new_FrameAddr(arch_env->sp->reg_class, first_bl, frame, entry->ent);
1560 mem = new_r_Proj(store, mode_M, pn_Store_M);
1562 /* the backing store itself */
1563 store = new_r_Store(first_bl, mem, addr,
1564 new_r_Proj(args, mode, i), 0);
1566 /* the new memory Proj gets the last Proj from store */
1567 set_Proj_pred(nmem, store);
1568 set_Proj_proj(nmem, pn_Store_M);
1570 /* move all entities to the frame type */
1571 frame_tp = get_irg_frame_type(irg);
1572 offset = get_type_size_bytes(frame_tp);
1574 /* we will add new entities: set the layout to undefined */
1575 assert(get_type_state(frame_tp) == layout_fixed);
1576 set_type_state(frame_tp, layout_undefined);
1577 for (entry = new_list; entry != NULL; entry = entry->next) {
1578 ir_entity *ent = entry->ent;
1580 /* If the entity is still on the argument type, move it to the
1582 * This happens if the value_param type was build due to compound
1584 if (get_entity_owner(ent) != frame_tp) {
1585 ir_type *tp = get_entity_type(ent);
1586 unsigned align = get_type_alignment_bytes(tp);
1588 offset += align - 1;
1589 offset &= ~(align - 1);
1590 set_entity_owner(ent, frame_tp);
1591 /* must be automatic to set a fixed layout */
1592 set_entity_offset(ent, offset);
1593 offset += get_type_size_bytes(tp);
1596 set_type_size_bytes(frame_tp, offset);
1597 /* fix the layout again */
1598 set_type_state(frame_tp, layout_fixed);
1603 * The start block has no jump, instead it has an initial exec Proj.
1604 * The backend wants to handle all blocks the same way, so we replace
1605 * the out cfg edge with a real jump.
1607 static void fix_start_block(ir_graph *irg)
1609 ir_node *initial_X = get_irg_initial_exec(irg);
1610 ir_node *start_block = get_irg_start_block(irg);
1611 const ir_edge_t *edge;
1613 assert(is_Proj(initial_X));
1615 foreach_out_edge(initial_X, edge) {
1616 ir_node *block = get_edge_src_irn(edge);
1618 if (is_Anchor(block))
1620 if (block != start_block) {
1621 ir_node *jmp = new_r_Jmp(start_block);
1622 set_Block_cfgpred(block, get_edge_src_pos(edge), jmp);
1623 set_irg_initial_exec(irg, jmp);
1627 panic("Initial exec has no follow block in %+F", irg);
1631 * Update the entity of Sels to the outer value parameters.
1633 static void update_outer_frame_sels(ir_node *irn, void *env)
1635 lower_frame_sels_env_t *ctx = env;
1642 ptr = get_Sel_ptr(irn);
1643 if (! is_arg_Proj(ptr))
1645 if (get_Proj_proj(ptr) != ctx->static_link_pos)
1647 ent = get_Sel_entity(irn);
1649 if (get_entity_owner(ent) == ctx->value_tp) {
1650 /* replace by its copy from the argument type */
1651 pos = get_struct_member_index(ctx->value_tp, ent);
1652 ent = get_argument_entity(ent, ctx);
1653 set_Sel_entity(irn, ent);
1655 /* check, if we have not seen this entity before */
1656 if (get_entity_link(ent) == NULL) {
1662 ARR_APP1(ent_pos_pair, ctx->value_param_list, pair);
1664 set_entity_link(ent, ctx->value_param_list);
1670 * Fix access to outer local variables.
1672 static void fix_outer_variable_access(be_abi_irg_t *env,
1673 lower_frame_sels_env_t *ctx)
1679 for (i = get_class_n_members(ctx->frame_tp) - 1; i >= 0; --i) {
1680 ir_entity *ent = get_class_member(ctx->frame_tp, i);
1682 if (! is_method_entity(ent))
1685 irg = get_entity_irg(ent);
1690 * FIXME: find the number of the static link parameter
1691 * for now we assume 0 here
1693 ctx->static_link_pos = 0;
1695 irg_walk_graph(irg, NULL, update_outer_frame_sels, ctx);
1700 * Modify the irg itself and the frame type.
1702 static void modify_irg(ir_graph *irg)
1704 be_abi_irg_t *env = be_get_irg_abi(irg);
1705 be_abi_call_t *call = env->call;
1706 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1707 const arch_register_t *sp = arch_env->sp;
1708 ir_type *method_type = get_entity_type(get_irg_entity(irg));
1709 struct obstack *obst = be_get_be_obst(irg);
1710 be_stack_layout_t *stack_layout = be_get_irg_stack_layout(irg);
1713 ir_node *new_mem_proj;
1719 unsigned frame_size;
1722 const arch_register_t *fp_reg;
1723 ir_node *frame_pointer;
1727 const ir_edge_t *edge;
1728 ir_type *arg_type, *bet_type, *tp;
1729 lower_frame_sels_env_t ctx;
1730 ir_entity **param_map;
1732 DBG((dbg, LEVEL_1, "introducing abi on %+F\n", irg));
1734 /* Must fetch memory here, otherwise the start Barrier gets the wrong
1735 * memory, which leads to loops in the DAG. */
1736 old_mem = get_irg_initial_mem(irg);
1738 irp_reserve_resources(irp, IR_RESOURCE_ENTITY_LINK);
1740 /* set the links of all frame entities to NULL, we use it
1741 to detect if an entity is already linked in the value_param_list */
1742 tp = get_method_value_param_type(method_type);
1745 /* clear the links of the clone type, let the
1746 original entities point to its clones */
1747 for (i = get_struct_n_members(tp) - 1; i >= 0; --i) {
1748 ir_entity *mem = get_struct_member(tp, i);
1749 set_entity_link(mem, NULL);
1753 arg_type = compute_arg_type(env, irg, call, method_type, tp, ¶m_map);
1755 /* Convert the Sel nodes in the irg to frame addr nodes: */
1756 ctx.value_param_list = NEW_ARR_F(ent_pos_pair, 0);
1757 ctx.frame = get_irg_frame(irg);
1758 ctx.sp_class = arch_env->sp->reg_class;
1759 ctx.link_class = arch_env->link_class;
1760 ctx.frame_tp = get_irg_frame_type(irg);
1762 /* layout the stackframe now */
1763 if (get_type_state(ctx.frame_tp) == layout_undefined) {
1764 default_layout_compound_type(ctx.frame_tp);
1767 /* we will possible add new entities to the frame: set the layout to undefined */
1768 assert(get_type_state(ctx.frame_tp) == layout_fixed);
1769 set_type_state(ctx.frame_tp, layout_undefined);
1771 irg_walk_graph(irg, lower_frame_sels_walker, NULL, &ctx);
1773 /* fix the frame type layout again */
1774 set_type_state(ctx.frame_tp, layout_fixed);
1775 /* align stackframe to 4 byte */
1776 frame_size = get_type_size_bytes(ctx.frame_tp);
1777 if (frame_size % 4 != 0) {
1778 set_type_size_bytes(ctx.frame_tp, frame_size + 4 - (frame_size % 4));
1781 env->regs = pmap_create();
1783 n_params = get_method_n_params(method_type);
1784 args = OALLOCNZ(obst, ir_node*, n_params);
1787 * for inner function we must now fix access to outer frame entities.
1789 fix_outer_variable_access(env, &ctx);
1791 /* Check if a value parameter is transmitted as a register.
1792 * This might happen if the address of an parameter is taken which is
1793 * transmitted in registers.
1795 * Note that on some architectures this case must be handled specially
1796 * because the place of the backing store is determined by their ABI.
1798 * In the default case we move the entity to the frame type and create
1799 * a backing store into the first block.
1801 fix_address_of_parameter_access(env, irg, ctx.value_param_list);
1803 DEL_ARR_F(ctx.value_param_list);
1804 irp_free_resources(irp, IR_RESOURCE_ENTITY_LINK);
1806 /* Fill the argument vector */
1807 arg_tuple = get_irg_args(irg);
1808 foreach_out_edge(arg_tuple, edge) {
1809 ir_node *irn = get_edge_src_irn(edge);
1810 if (! is_Anchor(irn)) {
1811 int nr = get_Proj_proj(irn);
1813 DBG((dbg, LEVEL_2, "\treading arg: %d -> %+F\n", nr, irn));
1817 bet_type = call->cb->get_between_type(env->cb);
1818 stack_frame_init(stack_layout, arg_type, bet_type,
1819 get_irg_frame_type(irg), arch_env->stack_dir, param_map);
1820 stack_layout->sp_relative = call->flags.bits.try_omit_fp;
1822 /* Count the register params and add them to the number of Projs for the RegParams node */
1823 for (i = 0; i < n_params; ++i) {
1824 be_abi_call_arg_t *arg = get_call_arg(call, 0, i, 1);
1825 if (arg->in_reg && args[i]) {
1826 assert(arg->reg != sp && "cannot use stack pointer as parameter register");
1827 assert(i == get_Proj_proj(args[i]));
1829 /* For now, associate the register with the old Proj from Start representing that argument. */
1830 pmap_insert(env->regs, (void *) arg->reg, args[i]);
1831 DBG((dbg, LEVEL_2, "\targ #%d -> reg %s\n", i, arg->reg->name));
1835 /* Collect all callee-save registers */
1836 for (i = 0, n = arch_env_get_n_reg_class(arch_env); i < n; ++i) {
1837 const arch_register_class_t *cls = arch_env_get_reg_class(arch_env, i);
1838 for (j = 0; j < cls->n_regs; ++j) {
1839 const arch_register_t *reg = &cls->regs[j];
1840 if (arch_register_type_is(reg, callee_save) ||
1841 arch_register_type_is(reg, state)) {
1842 pmap_insert(env->regs, (void *) reg, NULL);
1847 /* handle start block here (place a jump in the block) */
1848 fix_start_block(irg);
1850 pmap_insert(env->regs, (void *) sp, NULL);
1851 pmap_insert(env->regs, (void *) arch_env->bp, NULL);
1852 start_bl = get_irg_start_block(irg);
1853 env->start = be_new_Start(NULL, start_bl, pmap_count(env->regs) + 1);
1856 * make proj nodes for the callee save registers.
1857 * memorize them, since Return nodes get those as inputs.
1859 * Note, that if a register corresponds to an argument, the regs map contains
1860 * the old Proj from start for that argument.
1863 rm = ALLOCAN(reg_node_map_t, pmap_count(env->regs));
1864 reg_map_to_arr(rm, env->regs);
1865 for (i = 0, n = pmap_count(env->regs); i < n; ++i) {
1866 arch_register_t *reg = (void *) rm[i].reg;
1867 ir_mode *mode = reg->reg_class->mode;
1869 arch_register_req_type_t add_type = 0;
1873 add_type |= arch_register_req_type_produces_sp | arch_register_req_type_ignore;
1876 proj = new_r_Proj(env->start, mode, nr + 1);
1877 pmap_insert(env->regs, (void *) reg, proj);
1878 be_set_constr_single_reg_out(env->start, nr + 1, reg, add_type);
1879 arch_set_irn_register(proj, reg);
1881 DBG((dbg, LEVEL_2, "\tregister save proj #%d -> reg %s\n", nr, reg->name));
1884 /* create a new initial memory proj */
1885 assert(is_Proj(old_mem));
1886 arch_set_out_register_req(env->start, 0, arch_no_register_req);
1887 new_mem_proj = new_r_Proj(env->start, mode_M, 0);
1889 set_irg_initial_mem(irg, mem);
1891 /* Generate the Prologue */
1892 fp_reg = call->cb->prologue(env->cb, &mem, env->regs, &stack_layout->initial_bias);
1894 /* do the stack allocation BEFORE the barrier, or spill code
1895 might be added before it */
1896 env->init_sp = be_abi_reg_map_get(env->regs, sp);
1897 env->init_sp = be_new_IncSP(sp, start_bl, env->init_sp, BE_STACK_FRAME_SIZE_EXPAND, 0);
1898 be_abi_reg_map_set(env->regs, sp, env->init_sp);
1900 create_barrier(start_bl, &mem, env->regs, 0);
1902 env->init_sp = be_abi_reg_map_get(env->regs, sp);
1903 arch_set_irn_register(env->init_sp, sp);
1905 frame_pointer = be_abi_reg_map_get(env->regs, fp_reg);
1906 set_irg_frame(irg, frame_pointer);
1907 pset_insert_ptr(env->ignore_regs, fp_reg);
1909 /* rewire old mem users to new mem */
1910 exchange(old_mem, mem);
1912 /* keep the mem (for functions with an endless loop = no return) */
1915 set_irg_initial_mem(irg, mem);
1917 /* Now, introduce stack param nodes for all parameters passed on the stack */
1918 for (i = 0; i < n_params; ++i) {
1919 ir_node *arg_proj = args[i];
1920 ir_node *repl = NULL;
1922 if (arg_proj != NULL) {
1923 be_abi_call_arg_t *arg;
1924 ir_type *param_type;
1925 int nr = get_Proj_proj(arg_proj);
1928 nr = MIN(nr, n_params);
1929 arg = get_call_arg(call, 0, nr, 1);
1930 param_type = get_method_param_type(method_type, nr);
1933 repl = pmap_get(env->regs, (void *) arg->reg);
1934 } else if (arg->on_stack) {
1935 ir_node *addr = be_new_FrameAddr(sp->reg_class, start_bl, frame_pointer, arg->stack_ent);
1937 /* For atomic parameters which are actually used, we create a Load node. */
1938 if (is_atomic_type(param_type) && get_irn_n_edges(args[i]) > 0) {
1939 ir_mode *mode = get_type_mode(param_type);
1940 ir_mode *load_mode = arg->load_mode;
1942 ir_node *load = new_r_Load(start_bl, new_NoMem(), addr, load_mode, cons_floats);
1943 repl = new_r_Proj(load, load_mode, pn_Load_res);
1945 if (mode != load_mode) {
1946 repl = new_r_Conv(start_bl, repl, mode);
1949 /* The stack parameter is not primitive (it is a struct or array),
1950 * we thus will create a node representing the parameter's address
1956 assert(repl != NULL);
1958 /* Beware: the mode of the register parameters is always the mode of the register class
1959 which may be wrong. Add Conv's then. */
1960 mode = get_irn_mode(args[i]);
1961 if (mode != get_irn_mode(repl)) {
1962 repl = new_r_Conv(get_nodes_block(repl), repl, mode);
1964 exchange(args[i], repl);
1968 /* the arg proj is not needed anymore now and should be only used by the anchor */
1969 assert(get_irn_n_edges(arg_tuple) == 1);
1970 kill_node(arg_tuple);
1971 set_irg_args(irg, new_r_Bad(irg));
1973 /* All Return nodes hang on the End node, so look for them there. */
1974 end = get_irg_end_block(irg);
1975 for (i = 0, n = get_Block_n_cfgpreds(end); i < n; ++i) {
1976 ir_node *irn = get_Block_cfgpred(end, i);
1978 if (is_Return(irn)) {
1979 ir_node *blk = get_nodes_block(irn);
1980 ir_node *mem = get_Return_mem(irn);
1981 ir_node *ret = create_be_return(env, irn, blk, mem, get_Return_n_ress(irn));
1986 /* if we have endless loops here, n might be <= 0. Do NOT create a be_Return then,
1987 the code is dead and will never be executed. */
1990 /** Fix the state inputs of calls that still hang on unknowns */
1991 static void fix_call_state_inputs(ir_graph *irg)
1993 be_abi_irg_t *env = be_get_irg_abi(irg);
1994 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1996 arch_register_t **stateregs = NEW_ARR_F(arch_register_t*, 0);
1998 /* Collect caller save registers */
1999 n = arch_env_get_n_reg_class(arch_env);
2000 for (i = 0; i < n; ++i) {
2002 const arch_register_class_t *cls = arch_env_get_reg_class(arch_env, i);
2003 for (j = 0; j < cls->n_regs; ++j) {
2004 const arch_register_t *reg = arch_register_for_index(cls, j);
2005 if (arch_register_type_is(reg, state)) {
2006 ARR_APP1(arch_register_t*, stateregs, (arch_register_t *)reg);
2011 n = ARR_LEN(env->calls);
2012 n_states = ARR_LEN(stateregs);
2013 for (i = 0; i < n; ++i) {
2015 ir_node *call = env->calls[i];
2017 arity = get_irn_arity(call);
2019 /* the state reg inputs are the last n inputs of the calls */
2020 for (s = 0; s < n_states; ++s) {
2021 int inp = arity - n_states + s;
2022 const arch_register_t *reg = stateregs[s];
2023 ir_node *regnode = be_abi_reg_map_get(env->regs, reg);
2025 set_irn_n(call, inp, regnode);
2029 DEL_ARR_F(stateregs);
2033 * Create a trampoline entity for the given method.
2035 static ir_entity *create_trampoline(be_main_env_t *be, ir_entity *method)
2037 ir_type *type = get_entity_type(method);
2038 ident *old_id = get_entity_ld_ident(method);
2039 ident *id = id_mangle3("", old_id, "$stub");
2040 ir_type *parent = be->pic_trampolines_type;
2041 ir_entity *ent = new_entity(parent, old_id, type);
2042 set_entity_ld_ident(ent, id);
2043 set_entity_visibility(ent, ir_visibility_private);
2049 * Returns the trampoline entity for the given method.
2051 static ir_entity *get_trampoline(be_main_env_t *env, ir_entity *method)
2053 ir_entity *result = pmap_get(env->ent_trampoline_map, method);
2054 if (result == NULL) {
2055 result = create_trampoline(env, method);
2056 pmap_insert(env->ent_trampoline_map, method, result);
2062 static ir_entity *create_pic_symbol(be_main_env_t *be, ir_entity *entity)
2064 ident *old_id = get_entity_ld_ident(entity);
2065 ident *id = id_mangle3("", old_id, "$non_lazy_ptr");
2066 ir_type *e_type = get_entity_type(entity);
2067 ir_type *type = new_type_pointer(e_type);
2068 ir_type *parent = be->pic_symbols_type;
2069 ir_entity *ent = new_entity(parent, old_id, type);
2070 set_entity_ld_ident(ent, id);
2071 set_entity_visibility(ent, ir_visibility_private);
2076 static ir_entity *get_pic_symbol(be_main_env_t *env, ir_entity *entity)
2078 ir_entity *result = pmap_get(env->ent_pic_symbol_map, entity);
2079 if (result == NULL) {
2080 result = create_pic_symbol(env, entity);
2081 pmap_insert(env->ent_pic_symbol_map, entity, result);
2090 * Returns non-zero if a given entity can be accessed using a relative address.
2092 static int can_address_relative(ir_entity *entity)
2094 return get_entity_visibility(entity) != ir_visibility_external
2095 && !(get_entity_linkage(entity) & IR_LINKAGE_MERGE);
2098 /** patches SymConsts to work in position independent code */
2099 static void fix_pic_symconsts(ir_node *node, void *data)
2107 ir_graph *irg = get_irn_irg(node);
2109 be_main_env_t *be = be_get_irg_main_env(irg);
2112 arity = get_irn_arity(node);
2113 for (i = 0; i < arity; ++i) {
2115 ir_node *pred = get_irn_n(node, i);
2117 ir_entity *pic_symbol;
2118 ir_node *pic_symconst;
2120 if (!is_SymConst(pred))
2123 entity = get_SymConst_entity(pred);
2124 block = get_nodes_block(pred);
2126 /* calls can jump to relative addresses, so we can directly jump to
2127 the (relatively) known call address or the trampoline */
2128 if (i == 1 && is_Call(node)) {
2129 ir_entity *trampoline;
2130 ir_node *trampoline_const;
2132 if (can_address_relative(entity))
2135 dbgi = get_irn_dbg_info(pred);
2136 trampoline = get_trampoline(be, entity);
2137 trampoline_const = new_rd_SymConst_addr_ent(dbgi, irg, mode_P_code,
2139 set_irn_n(node, i, trampoline_const);
2143 /* everything else is accessed relative to EIP */
2144 mode = get_irn_mode(pred);
2145 pic_base = arch_code_generator_get_pic_base(be_get_irg_cg(irg));
2147 /* all ok now for locally constructed stuff */
2148 if (can_address_relative(entity)) {
2149 ir_node *add = new_r_Add(block, pic_base, pred, mode);
2151 /* make sure the walker doesn't visit this add again */
2152 mark_irn_visited(add);
2153 set_irn_n(node, i, add);
2157 /* get entry from pic symbol segment */
2158 dbgi = get_irn_dbg_info(pred);
2159 pic_symbol = get_pic_symbol(be, entity);
2160 pic_symconst = new_rd_SymConst_addr_ent(dbgi, irg, mode_P_code,
2162 add = new_r_Add(block, pic_base, pic_symconst, mode);
2163 mark_irn_visited(add);
2165 /* we need an extra indirection for global data outside our current
2166 module. The loads are always safe and can therefore float
2167 and need no memory input */
2168 load = new_r_Load(block, new_NoMem(), add, mode, cons_floats);
2169 load_res = new_r_Proj(load, mode, pn_Load_res);
2171 set_irn_n(node, i, load_res);
2175 be_abi_irg_t *be_abi_introduce(ir_graph *irg)
2177 be_abi_irg_t *env = XMALLOCZ(be_abi_irg_t);
2178 ir_node *old_frame = get_irg_frame(irg);
2179 be_options_t *options = be_get_irg_options(irg);
2180 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
2181 ir_entity *entity = get_irg_entity(irg);
2182 ir_type *method_type = get_entity_type(entity);
2187 env->ignore_regs = pset_new_ptr_default();
2189 /* break here if backend provides a custom API.
2190 * Note: we shouldn't have to setup any be_abi_irg_t* stuff at all,
2191 * but need more cleanup to make this work
2193 be_set_irg_abi(irg, env);
2194 if (arch_env->custom_abi)
2197 be_omit_fp = options->omit_fp;
2198 be_omit_leaf_fp = options->omit_leaf_fp;
2200 env->dce_survivor = new_survive_dce();
2201 env->keep_map = pmap_create();
2202 env->call = be_abi_call_new(arch_env->sp->reg_class);
2203 arch_env_get_call_abi(arch_env, method_type, env->call);
2205 env->init_sp = dummy = new_r_Dummy(irg, arch_env->sp->reg_class->mode);
2206 env->calls = NEW_ARR_F(ir_node*, 0);
2209 irg_walk_graph(irg, fix_pic_symconsts, NULL, env);
2212 /* Lower all call nodes in the IRG. */
2216 Beware: init backend abi call object after processing calls,
2217 otherwise some information might be not yet available.
2219 env->cb = env->call->cb->init(env->call, irg);
2221 /* Process the IRG */
2224 /* fix call inputs for state registers */
2225 fix_call_state_inputs(irg);
2227 /* We don't need the keep map anymore. */
2228 pmap_destroy(env->keep_map);
2229 env->keep_map = NULL;
2231 /* calls array is not needed anymore */
2232 DEL_ARR_F(env->calls);
2235 /* reroute the stack origin of the calls to the true stack origin. */
2236 exchange(dummy, env->init_sp);
2237 exchange(old_frame, get_irg_frame(irg));
2239 /* Make some important node pointers survive the dead node elimination. */
2240 survive_dce_register_irn(env->dce_survivor, &env->init_sp);
2241 foreach_pmap(env->regs, ent) {
2242 survive_dce_register_irn(env->dce_survivor, (ir_node **) &ent->value);
2245 env->call->cb->done(env->cb);
2250 void be_abi_free(ir_graph *irg)
2252 be_abi_irg_t *env = be_get_irg_abi(irg);
2254 if (env->call != NULL)
2255 be_abi_call_free(env->call);
2256 if (env->dce_survivor != NULL)
2257 free_survive_dce(env->dce_survivor);
2258 if (env->ignore_regs != NULL)
2259 del_pset(env->ignore_regs);
2260 if (env->regs != NULL)
2261 pmap_destroy(env->regs);
2264 be_set_irg_abi(irg, NULL);
2267 void be_abi_put_ignore_regs(be_abi_irg_t *abi, const arch_register_class_t *cls, bitset_t *bs)
2269 arch_register_t *reg;
2271 for (reg = pset_first(abi->ignore_regs); reg; reg = pset_next(abi->ignore_regs))
2272 if (reg->reg_class == cls)
2273 bitset_set(bs, reg->index);
2276 void be_abi_set_non_ignore_regs(be_abi_irg_t *abi, const arch_register_class_t *cls, unsigned *raw_bitset)
2279 arch_register_t *reg;
2281 for (i = 0; i < cls->n_regs; ++i) {
2282 if (arch_register_type_is(&cls->regs[i], ignore))
2285 rbitset_set(raw_bitset, i);
2288 for (reg = pset_first(abi->ignore_regs); reg != NULL;
2289 reg = pset_next(abi->ignore_regs)) {
2290 if (reg->reg_class != cls)
2293 rbitset_clear(raw_bitset, reg->index);
2297 ir_node *be_abi_get_callee_save_irn(be_abi_irg_t *abi, const arch_register_t *reg)
2299 assert(arch_register_type_is(reg, callee_save));
2300 assert(pmap_contains(abi->regs, (void *) reg));
2301 return pmap_get(abi->regs, (void *) reg);
2304 ir_node *be_abi_get_ignore_irn(be_abi_irg_t *abi, const arch_register_t *reg)
2306 assert(arch_register_type_is(reg, ignore));
2307 assert(pmap_contains(abi->regs, (void *) reg));
2308 return pmap_get(abi->regs, (void *) reg);
2311 BE_REGISTER_MODULE_CONSTRUCTOR(be_init_abi);
2312 void be_init_abi(void)
2314 FIRM_DBG_REGISTER(dbg, "firm.be.abi");