2 * Copyright (C) 1995-2011 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
31 #include "irgraph_t.h"
34 #include "iredges_t.h"
37 #include "irprintf_t.h"
44 #include "raw_bitset.h"
55 #include "bessaconstr.h"
57 #include "betranshlp.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 be_abi_call_t *call; /**< The ABI call information. */
91 ir_node *init_sp; /**< The node representing the stack pointer
92 at the start of the function. */
94 ir_node *start; /**< The be_Start params node. */
95 pmap *regs; /**< A map of all callee-save and ignore regs to
96 their Projs to the RegParams node. */
97 int start_block_bias; /**< The stack bias at the end of the start block. */
99 pmap *keep_map; /**< mapping blocks to keep nodes. */
101 ir_node **calls; /**< flexible array containing all be_Call nodes */
104 static ir_heights_t *ir_heights;
106 /** Flag: if set, try to omit the frame pointer in all routines. */
107 static int be_omit_fp = 1;
109 static ir_node *be_abi_reg_map_get(pmap *map, const arch_register_t *reg)
111 return pmap_get(ir_node, map, reg);
114 static void be_abi_reg_map_set(pmap *map, const arch_register_t* reg,
117 pmap_insert(map, reg, node);
121 * Check if the given register is callee save, ie. will be saved by the callee.
123 static bool arch_register_is_callee_save(
124 const arch_env_t *arch_env,
125 const arch_register_t *reg)
127 if (arch_env->impl->register_saved_by)
128 return arch_env->impl->register_saved_by(reg, /*callee=*/1);
133 * Check if the given register is caller save, ie. must be saved by the caller.
135 static bool arch_register_is_caller_save(
136 const arch_env_t *arch_env,
137 const arch_register_t *reg)
139 if (arch_env->impl->register_saved_by)
140 return arch_env->impl->register_saved_by(reg, /*callee=*/0);
147 _ ____ ___ ____ _ _ _ _
148 / \ | __ )_ _| / ___|__ _| | | |__ __ _ ___| | _____
149 / _ \ | _ \| | | | / _` | | | '_ \ / _` |/ __| |/ / __|
150 / ___ \| |_) | | | |__| (_| | | | |_) | (_| | (__| <\__ \
151 /_/ \_\____/___| \____\__,_|_|_|_.__/ \__,_|\___|_|\_\___/
153 These callbacks are used by the backend to set the parameters
154 for a specific call type.
158 * Set compare function: compares two ABI call object arguments.
160 static int cmp_call_arg(const void *a, const void *b, size_t n)
162 const be_abi_call_arg_t *p = (const be_abi_call_arg_t*)a;
163 const be_abi_call_arg_t *q = (const be_abi_call_arg_t*)b;
165 return !(p->is_res == q->is_res && p->pos == q->pos && p->callee == q->callee);
169 * Get an ABI call object argument.
171 * @param call the abi call
172 * @param is_res true for call results, false for call arguments
173 * @param pos position of the argument
174 * @param callee context type - if we are callee or caller
176 static be_abi_call_arg_t *get_call_arg(be_abi_call_t *call, int is_res, int pos, int callee)
178 be_abi_call_arg_t arg;
181 memset(&arg, 0, sizeof(arg));
186 hash = is_res * 128 + pos;
188 return set_find(be_abi_call_arg_t, call->params, &arg, sizeof(arg), hash);
192 * Set an ABI call object argument.
194 static void remember_call_arg(be_abi_call_arg_t *arg, be_abi_call_t *call, be_abi_context_t context)
196 unsigned hash = arg->is_res * 128 + arg->pos;
197 if (context & ABI_CONTEXT_CALLEE) {
199 (void)set_insert(be_abi_call_arg_t, call->params, arg, sizeof(*arg), hash);
201 if (context & ABI_CONTEXT_CALLER) {
203 (void)set_insert(be_abi_call_arg_t, call->params, arg, sizeof(*arg), hash);
207 /* Set the flags for a call. */
208 void be_abi_call_set_flags(be_abi_call_t *call, be_abi_call_flags_t flags, const be_abi_callbacks_t *cb)
214 /* Sets the number of bytes the stackframe is shrinked by the callee on return */
215 void be_abi_call_set_pop(be_abi_call_t *call, int pop)
221 /* Set register class for call address */
222 void be_abi_call_set_call_address_reg_class(be_abi_call_t *call, const arch_register_class_t *cls)
224 call->cls_addr = cls;
228 void be_abi_call_param_stack(be_abi_call_t *call, int arg_pos,
229 ir_mode *load_mode, unsigned alignment,
230 unsigned space_before, unsigned space_after,
231 be_abi_context_t context)
233 be_abi_call_arg_t arg;
234 memset(&arg, 0, sizeof(arg));
235 assert(alignment > 0 && "Alignment must be greater than 0");
237 arg.load_mode = load_mode;
238 arg.alignment = alignment;
239 arg.space_before = space_before;
240 arg.space_after = space_after;
244 remember_call_arg(&arg, call, context);
247 void be_abi_call_param_reg(be_abi_call_t *call, int arg_pos, const arch_register_t *reg, be_abi_context_t context)
249 be_abi_call_arg_t arg;
250 memset(&arg, 0, sizeof(arg));
257 remember_call_arg(&arg, call, context);
260 void be_abi_call_res_reg(be_abi_call_t *call, int arg_pos, const arch_register_t *reg, be_abi_context_t context)
262 be_abi_call_arg_t arg;
263 memset(&arg, 0, sizeof(arg));
270 remember_call_arg(&arg, call, context);
273 /* Get the flags of a ABI call object. */
274 be_abi_call_flags_t be_abi_call_get_flags(const be_abi_call_t *call)
280 * Constructor for a new ABI call object.
282 * @param cls_addr register class of the call address
284 * @return the new ABI call object
286 static be_abi_call_t *be_abi_call_new(const arch_register_class_t *cls_addr)
288 be_abi_call_t *call = XMALLOCZ(be_abi_call_t);
291 call->params = new_set(cmp_call_arg, 16);
293 call->cls_addr = cls_addr;
295 call->flags.bits.try_omit_fp = be_omit_fp;
301 * Destructor for an ABI call object.
303 static void be_abi_call_free(be_abi_call_t *call)
305 del_set(call->params);
310 * Initializes the frame layout from parts
312 * @param frame the stack layout that will be initialized
313 * @param args the stack argument layout type
314 * @param between the between layout type
315 * @param locals the method frame type
317 * @return the initialized stack layout
319 static be_stack_layout_t *stack_frame_init(be_stack_layout_t *frame, ir_type *args,
320 ir_type *between, ir_type *locals)
322 frame->arg_type = args;
323 frame->between_type = between;
324 frame->frame_type = locals;
325 frame->initial_offset = 0;
326 frame->initial_bias = 0;
327 frame->order[1] = between;
329 /* typical decreasing stack: locals have the
330 * lowest addresses, arguments the highest */
331 frame->order[0] = locals;
332 frame->order[2] = args;
343 Adjustment of the calls inside a graph.
348 * Transform a call node into a be_Call node.
350 * @param env The ABI environment for the current irg.
351 * @param irn The call node.
352 * @param curr_sp The stack pointer node to use.
353 * @return The stack pointer after the call.
355 static ir_node *adjust_call(be_abi_irg_t *env, ir_node *irn, ir_node *curr_sp)
357 ir_graph *irg = get_irn_irg(irn);
358 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
359 ir_type *call_tp = get_Call_type(irn);
360 ir_node *call_ptr = get_Call_ptr(irn);
361 size_t n_params = get_method_n_params(call_tp);
362 ir_node *curr_mem = get_Call_mem(irn);
363 ir_node *bl = get_nodes_block(irn);
365 const arch_register_t *sp = arch_env->sp;
366 be_abi_call_t *call = be_abi_call_new(sp->reg_class);
367 ir_mode *mach_mode = sp->reg_class->mode;
368 int n_res = get_method_n_ress(call_tp);
370 ir_node *res_proj = NULL;
371 int n_reg_params = 0;
372 int n_stack_params = 0;
375 const arch_register_t **states = NEW_ARR_F(const arch_register_t*, 0);
376 const arch_register_t **destroyed_regs = NEW_ARR_F(const arch_register_t*, 0);
380 int n_reg_results = 0;
382 int *stack_param_idx;
384 int throws_exception;
389 /* Let the isa fill out the abi description for that call node. */
390 arch_env_get_call_abi(arch_env, call_tp, call);
392 /* Insert code to put the stack arguments on the stack. */
393 assert((size_t)get_Call_n_params(irn) == n_params);
394 stack_param_idx = ALLOCAN(int, n_params);
395 for (p = 0; p < n_params; ++p) {
396 be_abi_call_arg_t *arg = get_call_arg(call, 0, p, 0);
399 int arg_size = get_type_size_bytes(get_method_param_type(call_tp, p));
401 stack_size += round_up2(arg->space_before, arg->alignment);
402 stack_size += round_up2(arg_size, arg->alignment);
403 stack_size += round_up2(arg->space_after, arg->alignment);
405 stack_param_idx[n_stack_params++] = p;
409 /* Collect all arguments which are passed in registers. */
410 reg_param_idxs = ALLOCAN(int, n_params);
411 for (p = 0; p < n_params; ++p) {
412 be_abi_call_arg_t *arg = get_call_arg(call, 0, p, 0);
413 if (arg && arg->in_reg) {
414 reg_param_idxs[n_reg_params++] = p;
419 * If the stack is decreasing and we do not want to store sequentially,
420 * or someone else allocated the call frame
421 * we allocate as much space on the stack all parameters need, by
422 * moving the stack pointer along the stack's direction.
424 * Note: we also have to do this for stack_size == 0, because we may have
425 * to adjust stack alignment for the call.
427 curr_sp = be_new_IncSP(sp, bl, curr_sp, stack_size, 1);
429 dbgi = get_irn_dbg_info(irn);
430 /* If there are some parameters which shall be passed on the stack. */
431 if (n_stack_params > 0) {
433 ir_node **in = ALLOCAN(ir_node*, n_stack_params+1);
436 curr_mem = get_Call_mem(irn);
437 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.
453 curr_ofs += arg->space_before;
454 curr_ofs = round_up2(curr_ofs, arg->alignment);
456 /* Make the expression to compute the argument's offset. */
458 ir_mode *constmode = mach_mode;
459 if (mode_is_reference(mach_mode)) {
462 addr = new_r_Const_long(irg, constmode, curr_ofs);
463 addr = new_r_Add(bl, curr_sp, addr, mach_mode);
466 /* Insert a store for primitive arguments. */
467 if (is_atomic_type(param_type)) {
468 ir_node *nomem = get_irg_no_mem(irg);
469 ir_node *mem_input = nomem;
470 ir_node *store = new_rd_Store(dbgi, bl, mem_input, addr, param, cons_none);
471 mem = new_r_Proj(store, mode_M, pn_Store_M);
473 /* Make a mem copy for compound arguments. */
476 assert(mode_is_reference(get_irn_mode(param)));
477 copy = new_rd_CopyB(dbgi, bl, curr_mem, addr, param, param_type);
478 mem = new_r_Proj(copy, mode_M, pn_CopyB_M);
481 curr_ofs += param_size;
486 /* We need the sync only, if we didn't build the stores sequentially. */
487 if (n_stack_params >= 1) {
488 curr_mem = new_r_Sync(bl, n_in, in);
490 curr_mem = get_Call_mem(irn);
494 /* Put caller save into the destroyed set and state registers in the states
496 for (i = 0, n = arch_env->n_register_classes; i < n; ++i) {
498 const arch_register_class_t *cls = &arch_env->register_classes[i];
499 for (j = 0; j < cls->n_regs; ++j) {
500 const arch_register_t *reg = arch_register_for_index(cls, j);
502 /* even if destroyed all is specified, neither SP nor FP are
503 * destroyed (else bad things will happen) */
504 if (reg == arch_env->sp || reg == arch_env->bp)
507 if (reg->type & arch_register_type_state) {
508 ARR_APP1(const arch_register_t*, destroyed_regs, reg);
509 ARR_APP1(const arch_register_t*, states, reg);
510 /* we're already in the destroyed set so no need for further
514 if (arch_register_is_caller_save(arch_env, reg)) {
515 if (!(reg->type & arch_register_type_ignore)) {
516 ARR_APP1(const arch_register_t*, destroyed_regs, reg);
522 /* search the largest result proj number */
523 res_projs = ALLOCANZ(ir_node*, n_res);
525 foreach_out_edge(irn, edge) {
526 ir_node *irn = get_edge_src_irn(edge);
528 if (!is_Proj(irn) || get_Proj_proj(irn) != pn_Call_T_result)
531 foreach_out_edge(irn, res_edge) {
533 ir_node *res = get_edge_src_irn(res_edge);
535 assert(is_Proj(res));
537 proj = get_Proj_proj(res);
538 assert(proj < n_res);
539 assert(res_projs[proj] == NULL);
540 res_projs[proj] = res;
546 /** TODO: this is not correct for cases where return values are passed
547 * on the stack, but no known ABI does this currently...
549 n_reg_results = n_res;
552 in = ALLOCAN(ir_node*, n_reg_params + ARR_LEN(states));
554 /* make the back end call node and set its register requirements. */
555 for (i = 0; i < n_reg_params; ++i) {
556 in[n_ins++] = get_Call_param(irn, reg_param_idxs[i]);
559 /* add state registers ins */
560 for (s = 0; s < ARR_LEN(states); ++s) {
561 const arch_register_t *reg = states[s];
562 const arch_register_class_t *cls = reg->reg_class;
563 ir_node *regnode = new_r_Unknown(irg, cls->mode);
564 in[n_ins++] = regnode;
566 assert(n_ins == (int) (n_reg_params + ARR_LEN(states)));
568 /* ins collected, build the call */
569 throws_exception = ir_throws_exception(irn);
570 if (env->call->flags.bits.call_has_imm && is_SymConst(call_ptr)) {
572 low_call = be_new_Call(dbgi, irg, bl, curr_mem, curr_sp, curr_sp,
573 n_reg_results + pn_be_Call_first_res + ARR_LEN(destroyed_regs),
574 n_ins, in, get_Call_type(irn));
575 be_Call_set_entity(low_call, get_SymConst_entity(call_ptr));
578 low_call = be_new_Call(dbgi, irg, bl, curr_mem, curr_sp, call_ptr,
579 n_reg_results + pn_be_Call_first_res + ARR_LEN(destroyed_regs),
580 n_ins, in, get_Call_type(irn));
582 ir_set_throws_exception(low_call, throws_exception);
583 be_Call_set_pop(low_call, call->pop);
585 /* put the call into the list of all calls for later processing */
586 ARR_APP1(ir_node *, env->calls, low_call);
588 /* create new stack pointer */
589 curr_sp = new_r_Proj(low_call, get_irn_mode(curr_sp), pn_be_Call_sp);
590 be_set_constr_single_reg_out(low_call, pn_be_Call_sp, sp,
591 arch_register_req_type_ignore | arch_register_req_type_produces_sp);
592 arch_set_irn_register(curr_sp, sp);
594 /* now handle results */
595 for (i = 0; i < n_res; ++i) {
596 ir_node *proj = res_projs[i];
597 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 0);
598 long pn = i + pn_be_Call_first_res;
600 /* returns values on stack not supported yet */
604 shift the proj number to the right, since we will drop the
605 unspeakable Proj_T from the Call. Therefore, all real argument
606 Proj numbers must be increased by pn_be_Call_first_res
608 pn = i + pn_be_Call_first_res;
611 ir_type *res_type = get_method_res_type(call_tp, i);
612 ir_mode *mode = get_type_mode(res_type);
613 proj = new_r_Proj(low_call, mode, pn);
616 set_Proj_pred(proj, low_call);
617 set_Proj_proj(proj, pn);
621 /* remove register from destroyed regs */
623 size_t n = ARR_LEN(destroyed_regs);
624 for (j = 0; j < n; ++j) {
625 if (destroyed_regs[j] == arg->reg) {
626 destroyed_regs[j] = destroyed_regs[n-1];
627 ARR_SHRINKLEN(destroyed_regs,n-1);
635 Set the register class of the call address to
636 the backend provided class (default: stack pointer class)
638 be_node_set_reg_class_in(low_call, n_be_Call_ptr, call->cls_addr);
640 DBG((dbg, LEVEL_3, "\tcreated backend call %+F\n", low_call));
642 /* Set the register classes and constraints of the Call parameters. */
643 for (i = 0; i < n_reg_params; ++i) {
644 int index = reg_param_idxs[i];
645 be_abi_call_arg_t *arg = get_call_arg(call, 0, index, 0);
646 assert(arg->reg != NULL);
648 be_set_constr_single_reg_in(low_call, n_be_Call_first_arg + i,
649 arg->reg, arch_register_req_type_none);
652 /* Set the register constraints of the results. */
653 for (i = 0; i < n_res; ++i) {
654 ir_node *proj = res_projs[i];
655 const be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 0);
656 int pn = get_Proj_proj(proj);
659 be_set_constr_single_reg_out(low_call, pn, arg->reg,
660 arch_register_req_type_none);
661 arch_set_irn_register(proj, arg->reg);
663 exchange(irn, low_call);
665 /* kill the ProjT node */
666 if (res_proj != NULL) {
670 /* Make additional projs for the caller save registers
671 and the Keep node which keeps them alive. */
677 int curr_res_proj = pn_be_Call_first_res + n_reg_results;
680 n_ins = ARR_LEN(destroyed_regs) + n_reg_results + 1;
681 in = ALLOCAN(ir_node *, n_ins);
683 /* also keep the stack pointer */
684 set_irn_link(curr_sp, (void*) sp);
687 for (d = 0; d < ARR_LEN(destroyed_regs); ++d) {
688 const arch_register_t *reg = destroyed_regs[d];
689 ir_node *proj = new_r_Proj(low_call, reg->reg_class->mode, curr_res_proj);
691 /* memorize the register in the link field. we need afterwards to set the register class of the keep correctly. */
692 be_set_constr_single_reg_out(low_call, curr_res_proj, reg,
693 arch_register_req_type_none);
694 arch_set_irn_register(proj, reg);
696 set_irn_link(proj, (void*) reg);
701 for (i = 0; i < n_reg_results; ++i) {
702 ir_node *proj = res_projs[i];
703 const arch_register_t *reg = arch_get_irn_register(proj);
704 set_irn_link(proj, (void*) reg);
709 /* create the Keep for the caller save registers */
710 keep = be_new_Keep(bl, n, in);
711 for (i = 0; i < n; ++i) {
712 const arch_register_t *reg = (const arch_register_t*)get_irn_link(in[i]);
713 be_node_set_reg_class_in(keep, i, arch_register_get_class(reg));
717 /* Clean up the stack. */
718 assert(stack_size >= call->pop);
719 stack_size -= call->pop;
721 if (stack_size > 0) {
722 ir_node *mem_proj = NULL;
724 foreach_out_edge(low_call, edge) {
725 ir_node *irn = get_edge_src_irn(edge);
726 if (is_Proj(irn) && get_Proj_proj(irn) == pn_Call_M) {
733 mem_proj = new_r_Proj(low_call, mode_M, pn_be_Call_M);
734 keep_alive(mem_proj);
737 /* Clean up the stack frame or revert alignment fixes if we allocated it */
738 curr_sp = be_new_IncSP(sp, bl, curr_sp, -stack_size, 0);
740 be_abi_call_free(call);
743 DEL_ARR_F(destroyed_regs);
749 * Adjust the size of a node representing a stack alloc or free for the minimum stack alignment.
751 * @param alignment the minimum stack alignment
752 * @param size the node containing the non-aligned size
753 * @param block the block where new nodes are allocated on
754 * @param dbg debug info for new nodes
756 * @return a node representing the aligned size
758 static ir_node *adjust_alloc_size(unsigned stack_alignment, ir_node *size,
759 ir_node *block, dbg_info *dbg)
761 if (stack_alignment > 1) {
767 assert(is_po2(stack_alignment));
769 mode = get_irn_mode(size);
770 tv = new_tarval_from_long(stack_alignment-1, mode);
771 irg = get_Block_irg(block);
772 mask = new_r_Const(irg, tv);
773 size = new_rd_Add(dbg, block, size, mask, mode);
775 tv = new_tarval_from_long(-(long)stack_alignment, mode);
776 mask = new_r_Const(irg, tv);
777 size = new_rd_And(dbg, block, size, mask, mode);
783 * The alloca is transformed into a back end alloca node and connected to the stack nodes.
785 static ir_node *adjust_alloc(be_abi_irg_t *env, ir_node *alloc, ir_node *curr_sp)
787 ir_node *block = get_nodes_block(alloc);
788 ir_graph *irg = get_Block_irg(block);
789 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
790 ir_node *alloc_mem = NULL;
791 ir_node *alloc_res = NULL;
792 ir_type *type = get_Alloc_type(alloc);
799 unsigned stack_alignment;
801 /* all non-stack Alloc nodes should already be lowered before the backend */
802 assert(get_Alloc_where(alloc) == stack_alloc);
804 foreach_out_edge(alloc, edge) {
805 ir_node *irn = get_edge_src_irn(edge);
807 assert(is_Proj(irn));
808 switch (get_Proj_proj(irn)) {
820 /* Beware: currently Alloc nodes without a result might happen,
821 only escape analysis kills them and this phase runs only for object
822 oriented source. We kill the Alloc here. */
823 if (alloc_res == NULL && alloc_mem) {
824 exchange(alloc_mem, get_Alloc_mem(alloc));
828 dbg = get_irn_dbg_info(alloc);
829 count = get_Alloc_count(alloc);
831 /* we might need to multiply the count with the element size */
832 if (!is_unknown_type(type) && get_type_size_bytes(type) != 1) {
833 ir_mode *mode = get_irn_mode(count);
834 ir_tarval *tv = new_tarval_from_long(get_type_size_bytes(type),
836 ir_node *cnst = new_rd_Const(dbg, irg, tv);
837 size = new_rd_Mul(dbg, block, count, cnst, mode);
842 /* The stack pointer will be modified in an unknown manner.
843 We cannot omit it. */
844 env->call->flags.bits.try_omit_fp = 0;
846 stack_alignment = 1 << arch_env->stack_alignment;
847 size = adjust_alloc_size(stack_alignment, size, block, dbg);
848 new_alloc = be_new_AddSP(arch_env->sp, block, curr_sp, size);
849 set_irn_dbg_info(new_alloc, dbg);
851 if (alloc_mem != NULL) {
855 addsp_mem = new_r_Proj(new_alloc, mode_M, pn_be_AddSP_M);
857 /* We need to sync the output mem of the AddSP with the input mem
858 edge into the alloc node. */
859 ins[0] = get_Alloc_mem(alloc);
861 sync = new_r_Sync(block, 2, ins);
863 exchange(alloc_mem, sync);
866 exchange(alloc, new_alloc);
868 /* fix projnum of alloca res */
869 set_Proj_proj(alloc_res, pn_be_AddSP_res);
871 curr_sp = new_r_Proj(new_alloc, get_irn_mode(curr_sp), pn_be_AddSP_sp);
878 * The Free is transformed into a back end free node and connected to the stack nodes.
880 static ir_node *adjust_free(be_abi_irg_t *env, ir_node *free, ir_node *curr_sp)
882 ir_node *block = get_nodes_block(free);
883 ir_graph *irg = get_irn_irg(free);
884 ir_type *type = get_Free_type(free);
885 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
886 ir_mode *sp_mode = arch_env->sp->reg_class->mode;
887 dbg_info *dbg = get_irn_dbg_info(free);
888 ir_node *subsp, *mem, *res, *size, *sync;
890 unsigned stack_alignment;
892 /* all non-stack-alloc Free nodes should already be lowered before the
894 assert(get_Free_where(free) == stack_alloc);
896 /* we might need to multiply the size with the element size */
897 if (!is_unknown_type(type) && get_type_size_bytes(type) != 1) {
898 ir_tarval *tv = new_tarval_from_long(get_type_size_bytes(type), mode_Iu);
899 ir_node *cnst = new_rd_Const(dbg, irg, tv);
900 ir_node *mul = new_rd_Mul(dbg, block, get_Free_count(free),
904 size = get_Free_count(free);
907 stack_alignment = 1 << arch_env->stack_alignment;
908 size = adjust_alloc_size(stack_alignment, size, block, dbg);
910 /* The stack pointer will be modified in an unknown manner.
911 We cannot omit it. */
912 env->call->flags.bits.try_omit_fp = 0;
913 subsp = be_new_SubSP(arch_env->sp, block, curr_sp, size);
914 set_irn_dbg_info(subsp, dbg);
916 mem = new_r_Proj(subsp, mode_M, pn_be_SubSP_M);
917 res = new_r_Proj(subsp, sp_mode, pn_be_SubSP_sp);
919 /* we need to sync the memory */
920 in[0] = get_Free_mem(free);
922 sync = new_r_Sync(block, 2, in);
924 /* and make the AddSP dependent on the former memory */
925 add_irn_dep(subsp, get_Free_mem(free));
928 exchange(free, sync);
935 * Check if a node is somehow data dependent on another one.
936 * both nodes must be in the same basic block.
937 * @param n1 The first node.
938 * @param n2 The second node.
939 * @return 1, if n1 is data dependent (transitively) on n2, 0 if not.
941 static int dependent_on(ir_node *n1, ir_node *n2)
943 assert(get_nodes_block(n1) == get_nodes_block(n2));
945 return heights_reachable_in_block(ir_heights, n1, n2);
948 static int cmp_call_dependency(const void *c1, const void *c2)
950 ir_node *n1 = *(ir_node **) c1;
951 ir_node *n2 = *(ir_node **) c2;
955 Classical qsort() comparison function behavior:
956 0 if both elements are equal
957 1 if second is "smaller" that first
958 -1 if first is "smaller" that second
960 if (dependent_on(n1, n2))
963 if (dependent_on(n2, n1))
966 /* The nodes have no depth order, but we need a total order because qsort()
969 * Additionally, we need to respect transitive dependencies. Consider a
970 * Call a depending on Call b and an independent Call c.
971 * We MUST NOT order c > a and b > c. */
972 h1 = get_irn_height(ir_heights, n1);
973 h2 = get_irn_height(ir_heights, n2);
974 if (h1 < h2) return -1;
975 if (h1 > h2) return 1;
976 /* Same height, so use a random (but stable) order */
977 return get_irn_idx(n1) - get_irn_idx(n2);
981 * Walker: links all Call/Alloc/Free nodes to the Block they are contained.
983 static void link_ops_in_block_walker(ir_node *irn, void *data)
985 be_abi_irg_t *env = (be_abi_irg_t*)data;
986 unsigned code = get_irn_opcode(irn);
988 if (code == iro_Call ||
989 (code == iro_Alloc && get_Alloc_where(irn) == stack_alloc) ||
990 (code == iro_Free && get_Free_where(irn) == stack_alloc)) {
991 ir_node *bl = get_nodes_block(irn);
992 void *save = get_irn_link(bl);
994 set_irn_link(irn, save);
995 set_irn_link(bl, irn);
998 if (code == iro_Builtin && get_Builtin_kind(irn) == ir_bk_return_address) {
999 ir_node *param = get_Builtin_param(irn, 0);
1000 ir_tarval *tv = get_Const_tarval(param);
1001 unsigned long value = get_tarval_long(tv);
1002 /* use ebp, so the climbframe algo works... */
1004 env->call->flags.bits.try_omit_fp = 0;
1011 * Process all Call/Alloc/Free nodes inside a basic block.
1012 * Note that the link field of the block must contain a linked list of all
1013 * nodes inside the Block. We first order this list according to data dependency
1014 * and that connect the nodes together.
1016 static void process_ops_in_block(ir_node *bl, void *data)
1018 be_abi_irg_t *env = (be_abi_irg_t*)data;
1019 ir_node *curr_sp = env->init_sp;
1026 for (irn = (ir_node*)get_irn_link(bl); irn != NULL;
1027 irn = (ir_node*)get_irn_link(irn)) {
1031 nodes = ALLOCAN(ir_node*, n_nodes);
1032 for (irn = (ir_node*)get_irn_link(bl), n = 0; irn != NULL;
1033 irn = (ir_node*)get_irn_link(irn), ++n) {
1037 /* If there were call nodes in the block. */
1042 /* order the call nodes according to data dependency */
1043 qsort(nodes, n_nodes, sizeof(nodes[0]), cmp_call_dependency);
1045 for (i = n_nodes - 1; i >= 0; --i) {
1046 ir_node *irn = nodes[i];
1048 DBG((dbg, LEVEL_3, "\tprocessing call %+F\n", irn));
1049 switch (get_irn_opcode(irn)) {
1052 /* The stack pointer will be modified due to a call. */
1053 env->call->flags.bits.try_omit_fp = 0;
1055 curr_sp = adjust_call(env, irn, curr_sp);
1058 if (get_Alloc_where(irn) == stack_alloc)
1059 curr_sp = adjust_alloc(env, irn, curr_sp);
1062 if (get_Free_where(irn) == stack_alloc)
1063 curr_sp = adjust_free(env, irn, curr_sp);
1066 panic("invalid call");
1070 /* Keep the last stack state in the block by tying it to Keep node,
1071 * the proj from calls is already kept */
1072 if (curr_sp != env->init_sp &&
1073 !(is_Proj(curr_sp) && be_is_Call(get_Proj_pred(curr_sp)))) {
1075 keep = be_new_Keep(bl, 1, nodes);
1076 pmap_insert(env->keep_map, bl, keep);
1080 set_irn_link(bl, curr_sp);
1084 * Adjust all call nodes in the graph to the ABI conventions.
1086 static void process_calls(ir_graph *irg)
1088 be_abi_irg_t *abi = be_get_irg_abi(irg);
1090 irg_walk_graph(irg, firm_clear_link, link_ops_in_block_walker, abi);
1092 ir_heights = heights_new(irg);
1093 irg_block_walk_graph(irg, NULL, process_ops_in_block, abi);
1094 heights_free(ir_heights);
1098 * Computes the stack argument layout type.
1099 * Changes a possibly allocated value param type by moving
1100 * entities to the stack layout type.
1102 * @param call the current call ABI
1103 * @param method_type the method type
1105 * @return the stack argument layout type
1107 static ir_type *compute_arg_type(ir_graph *irg, be_abi_call_t *call,
1108 ir_type *method_type)
1110 struct obstack *obst = be_get_be_obst(irg);
1111 ir_type *frame_type = get_irg_frame_type(irg);
1112 size_t n_params = get_method_n_params(method_type);
1113 size_t n_frame_members = get_compound_n_members(frame_type);
1114 ir_entity *va_start_entity = NULL;
1120 ir_entity **map = OALLOCNZ(obst, ir_entity*, n_params);
1121 res = new_type_struct(new_id_from_chars("arg_type", 8));
1123 /* collect existing entities for value_param_types */
1124 for (f = n_frame_members; f > 0; ) {
1125 ir_entity *entity = get_compound_member(frame_type, --f);
1128 set_entity_link(entity, NULL);
1129 if (!is_parameter_entity(entity))
1131 num = get_entity_parameter_number(entity);
1132 if (num == IR_VA_START_PARAMETER_NUMBER) {
1133 /* move entity to new arg_type */
1134 set_entity_owner(entity, res);
1135 va_start_entity = entity;
1138 assert(num < n_params);
1139 if (map[num] != NULL)
1140 panic("multiple entities for parameter %u in %+F found", f, irg);
1142 if (num != n_params && !get_call_arg(call, 0, num, 1)->on_stack) {
1143 /* don't move this entity */
1148 /* move entity to new arg_type */
1149 set_entity_owner(entity, res);
1152 for (i = 0; i < n_params; ++i) {
1153 be_abi_call_arg_t *arg = get_call_arg(call, 0, i, 1);
1154 ir_type *param_type = get_method_param_type(method_type, i);
1157 if (!arg->on_stack) {
1161 if (entity == NULL) {
1162 /* create a new entity */
1163 entity = new_parameter_entity(res, i, param_type);
1165 ofs += arg->space_before;
1166 ofs = round_up2(ofs, arg->alignment);
1167 set_entity_offset(entity, ofs);
1168 ofs += arg->space_after;
1169 ofs += get_type_size_bytes(param_type);
1170 arg->stack_ent = entity;
1172 if (va_start_entity != NULL) {
1173 set_entity_offset(va_start_entity, ofs);
1175 set_type_size_bytes(res, ofs);
1176 set_type_state(res, layout_fixed);
1182 const arch_register_t *reg;
1186 static int cmp_regs(const void *a, const void *b)
1188 const reg_node_map_t *p = (const reg_node_map_t*)a;
1189 const reg_node_map_t *q = (const reg_node_map_t*)b;
1191 if (p->reg->reg_class == q->reg->reg_class)
1192 return p->reg->index - q->reg->index;
1194 return p->reg->reg_class < q->reg->reg_class ? -1 : +1;
1197 static void reg_map_to_arr(reg_node_map_t *res, pmap *reg_map)
1200 size_t n = pmap_count(reg_map);
1203 foreach_pmap(reg_map, ent) {
1204 res[i].reg = (const arch_register_t*)ent->key;
1205 res[i].irn = (ir_node*)ent->value;
1209 qsort(res, n, sizeof(res[0]), cmp_regs);
1213 * Creates a be_Return for a Return node.
1215 * @param @env the abi environment
1216 * @param irn the Return node or NULL if there was none
1217 * @param bl the block where the be_Retun should be placed
1218 * @param mem the current memory
1219 * @param n_res number of return results
1221 static ir_node *create_be_return(be_abi_irg_t *env, ir_node *irn, ir_node *bl,
1222 ir_node *mem, int n_res)
1224 be_abi_call_t *call = env->call;
1225 ir_graph *irg = get_Block_irg(bl);
1226 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1228 pmap *reg_map = pmap_create();
1229 ir_node *keep = pmap_get(ir_node, env->keep_map, bl);
1236 const arch_register_t **regs;
1240 get the valid stack node in this block.
1241 If we had a call in that block there is a Keep constructed by process_calls()
1242 which points to the last stack modification in that block. we'll use
1243 it then. Else we use the stack from the start block and let
1244 the ssa construction fix the usage.
1246 stack = be_abi_reg_map_get(env->regs, arch_env->sp);
1248 stack = get_irn_n(keep, 0);
1250 remove_End_keepalive(get_irg_end(irg), keep);
1253 /* Insert results for Return into the register map. */
1254 for (i = 0; i < n_res; ++i) {
1255 ir_node *res = get_Return_res(irn, i);
1256 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 1);
1257 assert(arg->in_reg && "return value must be passed in register");
1258 pmap_insert(reg_map, (void *) arg->reg, res);
1261 /* Add uses of the callee save registers. */
1262 foreach_pmap(env->regs, ent) {
1263 const arch_register_t *reg = (const arch_register_t*)ent->key;
1264 if ((reg->type & arch_register_type_ignore) || arch_register_is_callee_save(arch_env, reg))
1265 pmap_insert(reg_map, ent->key, ent->value);
1268 be_abi_reg_map_set(reg_map, arch_env->sp, stack);
1271 Maximum size of the in array for Return nodes is
1272 return args + callee save/ignore registers + memory + stack pointer
1274 in_max = pmap_count(reg_map) + n_res + 2;
1276 in = ALLOCAN(ir_node*, in_max);
1277 regs = ALLOCAN(arch_register_t const*, in_max);
1280 in[1] = be_abi_reg_map_get(reg_map, arch_env->sp);
1282 regs[1] = arch_env->sp;
1285 /* clear SP entry, since it has already been grown. */
1286 pmap_insert(reg_map, (void *) arch_env->sp, NULL);
1287 for (i = 0; i < n_res; ++i) {
1288 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 1);
1290 in[n] = be_abi_reg_map_get(reg_map, arg->reg);
1291 regs[n++] = arg->reg;
1293 /* Clear the map entry to mark the register as processed. */
1294 be_abi_reg_map_set(reg_map, arg->reg, NULL);
1297 /* grow the rest of the stuff. */
1298 foreach_pmap(reg_map, ent) {
1300 in[n] = (ir_node*)ent->value;
1301 regs[n++] = (const arch_register_t*)ent->key;
1305 /* The in array for the new back end return is now ready. */
1307 dbgi = get_irn_dbg_info(irn);
1311 /* we have to pop the shadow parameter in in case of struct returns */
1313 ret = be_new_Return(dbgi, irg, bl, n_res, pop, n, in);
1315 /* Set the register classes of the return's parameter accordingly. */
1316 for (i = 0; i < n; ++i) {
1317 if (regs[i] == NULL)
1320 be_set_constr_single_reg_in(ret, i, regs[i], arch_register_req_type_none);
1323 /* Free the space of the Epilog's in array and the register <-> proj map. */
1324 pmap_destroy(reg_map);
1329 typedef struct lower_frame_sels_env_t {
1330 ir_node *frame; /**< the current frame */
1331 const arch_register_class_t *sp_class; /**< register class of the stack pointer */
1332 const arch_register_class_t *link_class; /**< register class of the link pointer */
1333 ir_type *frame_tp; /**< the frame type */
1334 int static_link_pos; /**< argument number of the hidden static link */
1335 } lower_frame_sels_env_t;
1338 * Walker: Replaces Sels of frame type and
1339 * value param type entities by FrameAddress.
1340 * Links all used entities.
1342 static void lower_frame_sels_walker(ir_node *irn, void *data)
1344 lower_frame_sels_env_t *ctx = (lower_frame_sels_env_t*)data;
1347 ir_node *ptr = get_Sel_ptr(irn);
1349 if (ptr == ctx->frame) {
1350 ir_entity *ent = get_Sel_entity(irn);
1351 ir_node *bl = get_nodes_block(irn);
1354 nw = be_new_FrameAddr(ctx->sp_class, bl, ctx->frame, ent);
1361 * The start block has no jump, instead it has an initial exec Proj.
1362 * The backend wants to handle all blocks the same way, so we replace
1363 * the out cfg edge with a real jump.
1365 static void fix_start_block(ir_graph *irg)
1367 ir_node *initial_X = get_irg_initial_exec(irg);
1368 ir_node *start_block = get_irg_start_block(irg);
1369 ir_node *jmp = new_r_Jmp(start_block);
1371 assert(is_Proj(initial_X));
1372 exchange(initial_X, jmp);
1373 set_irg_initial_exec(irg, new_r_Bad(irg, mode_X));
1375 /* merge start block with successor if possible */
1377 foreach_out_edge(jmp, edge) {
1378 ir_node *succ = get_edge_src_irn(edge);
1379 if (!is_Block(succ))
1382 if (get_irn_arity(succ) == 1) {
1383 exchange(succ, start_block);
1391 * Modify the irg itself and the frame type.
1393 static void modify_irg(ir_graph *irg)
1395 be_abi_irg_t *env = be_get_irg_abi(irg);
1396 be_abi_call_t *call = env->call;
1397 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1398 const arch_register_t *sp = arch_env->sp;
1399 ir_type *method_type = get_entity_type(get_irg_entity(irg));
1400 be_irg_t *birg = be_birg_from_irg(irg);
1401 struct obstack *obst = be_get_be_obst(irg);
1402 be_stack_layout_t *stack_layout = be_get_irg_stack_layout(irg);
1405 ir_node *new_mem_proj;
1411 unsigned frame_size;
1414 const arch_register_t *fp_reg;
1415 ir_node *frame_pointer;
1419 ir_type *arg_type, *bet_type;
1420 lower_frame_sels_env_t ctx;
1422 DBG((dbg, LEVEL_1, "introducing abi on %+F\n", irg));
1424 old_mem = get_irg_initial_mem(irg);
1426 irp_reserve_resources(irp, IRP_RESOURCE_ENTITY_LINK);
1428 arg_type = compute_arg_type(irg, call, method_type);
1430 /* Convert the Sel nodes in the irg to frame addr nodes: */
1431 ctx.frame = get_irg_frame(irg);
1432 ctx.sp_class = arch_env->sp->reg_class;
1433 ctx.link_class = arch_env->link_class;
1434 ctx.frame_tp = get_irg_frame_type(irg);
1436 /* layout the stackframe now */
1437 if (get_type_state(ctx.frame_tp) == layout_undefined) {
1438 default_layout_compound_type(ctx.frame_tp);
1441 /* align stackframe to 4 byte */
1442 frame_size = get_type_size_bytes(ctx.frame_tp);
1443 if (frame_size % 4 != 0) {
1444 set_type_size_bytes(ctx.frame_tp, frame_size + 4 - (frame_size % 4));
1447 env->regs = pmap_create();
1449 n_params = get_method_n_params(method_type);
1450 args = OALLOCNZ(obst, ir_node*, n_params);
1452 be_add_parameter_entity_stores(irg);
1454 irg_walk_graph(irg, lower_frame_sels_walker, NULL, &ctx);
1456 irp_free_resources(irp, IRP_RESOURCE_ENTITY_LINK);
1458 /* Fill the argument vector */
1459 arg_tuple = get_irg_args(irg);
1460 foreach_out_edge(arg_tuple, edge) {
1461 ir_node *irn = get_edge_src_irn(edge);
1462 if (! is_Anchor(irn)) {
1463 int nr = get_Proj_proj(irn);
1465 DBG((dbg, LEVEL_2, "\treading arg: %d -> %+F\n", nr, irn));
1469 stack_layout->sp_relative = call->flags.bits.try_omit_fp;
1470 bet_type = call->cb->get_between_type(irg);
1471 stack_frame_init(stack_layout, arg_type, bet_type,
1472 get_irg_frame_type(irg));
1474 /* Count the register params and add them to the number of Projs for the RegParams node */
1475 for (i = 0; i < n_params; ++i) {
1476 be_abi_call_arg_t *arg = get_call_arg(call, 0, i, 1);
1477 if (arg->in_reg && args[i]) {
1478 assert(arg->reg != sp && "cannot use stack pointer as parameter register");
1479 assert(i == get_Proj_proj(args[i]));
1481 /* For now, associate the register with the old Proj from Start representing that argument. */
1482 pmap_insert(env->regs, (void *) arg->reg, args[i]);
1483 DBG((dbg, LEVEL_2, "\targ #%d -> reg %s\n", i, arg->reg->name));
1487 /* Collect all callee-save registers */
1488 for (i = 0, n = arch_env->n_register_classes; i < n; ++i) {
1489 const arch_register_class_t *cls = &arch_env->register_classes[i];
1490 for (j = 0; j < cls->n_regs; ++j) {
1491 const arch_register_t *reg = &cls->regs[j];
1492 if ((reg->type & arch_register_type_state) || arch_register_is_callee_save(arch_env, reg)) {
1493 pmap_insert(env->regs, (void *) reg, NULL);
1498 fp_reg = call->flags.bits.try_omit_fp ? arch_env->sp : arch_env->bp;
1499 rbitset_clear(birg->allocatable_regs, fp_reg->global_index);
1501 /* handle start block here (place a jump in the block) */
1502 fix_start_block(irg);
1504 pmap_insert(env->regs, (void *) sp, NULL);
1505 pmap_insert(env->regs, (void *) arch_env->bp, NULL);
1506 start_bl = get_irg_start_block(irg);
1507 env->start = be_new_Start(NULL, start_bl, pmap_count(env->regs) + 1);
1508 set_irg_start(irg, env->start);
1511 * make proj nodes for the callee save registers.
1512 * memorize them, since Return nodes get those as inputs.
1514 * Note, that if a register corresponds to an argument, the regs map
1515 * contains the old Proj from start for that argument.
1517 rm = ALLOCAN(reg_node_map_t, pmap_count(env->regs));
1518 reg_map_to_arr(rm, env->regs);
1519 for (i = 0, n = pmap_count(env->regs); i < n; ++i) {
1520 const arch_register_t *reg = rm[i].reg;
1521 ir_mode *mode = reg->reg_class->mode;
1523 arch_register_req_type_t add_type = arch_register_req_type_none;
1527 add_type |= arch_register_req_type_produces_sp;
1528 if (!rbitset_is_set(birg->allocatable_regs, reg->global_index)) {
1529 add_type |= arch_register_req_type_ignore;
1533 proj = new_r_Proj(env->start, mode, nr + 1);
1534 pmap_insert(env->regs, (void *) reg, proj);
1535 be_set_constr_single_reg_out(env->start, nr + 1, reg, add_type);
1536 arch_set_irn_register(proj, reg);
1538 DBG((dbg, LEVEL_2, "\tregister save proj #%d -> reg %s\n", nr, reg->name));
1541 /* create a new initial memory proj */
1542 assert(is_Proj(old_mem));
1543 arch_set_irn_register_req_out(env->start, 0, arch_no_register_req);
1544 new_mem_proj = new_r_Proj(env->start, mode_M, 0);
1546 set_irg_initial_mem(irg, mem);
1548 env->init_sp = be_abi_reg_map_get(env->regs, sp);
1550 /* set new frame_pointer */
1551 frame_pointer = be_abi_reg_map_get(env->regs, fp_reg);
1552 set_irg_frame(irg, frame_pointer);
1554 /* rewire old mem users to new mem */
1555 exchange(old_mem, mem);
1557 /* keep the mem (for functions with an endless loop = no return) */
1560 set_irg_initial_mem(irg, mem);
1562 /* Now, introduce stack param nodes for all parameters passed on the stack */
1563 for (i = 0; i < n_params; ++i) {
1564 ir_node *arg_proj = args[i];
1565 ir_node *repl = NULL;
1567 if (arg_proj != NULL) {
1568 be_abi_call_arg_t *arg;
1569 ir_type *param_type;
1570 int nr = get_Proj_proj(arg_proj);
1573 nr = MIN(nr, n_params);
1574 arg = get_call_arg(call, 0, nr, 1);
1575 param_type = get_method_param_type(method_type, nr);
1578 repl = pmap_get(ir_node, env->regs, arg->reg);
1579 } else if (arg->on_stack) {
1580 ir_node *addr = be_new_FrameAddr(sp->reg_class, start_bl, frame_pointer, arg->stack_ent);
1582 /* For atomic parameters which are actually used, we create a Load node. */
1583 if (is_atomic_type(param_type) && get_irn_n_edges(args[i]) > 0) {
1584 ir_mode *mode = get_type_mode(param_type);
1585 ir_mode *load_mode = arg->load_mode;
1586 ir_node *nomem = get_irg_no_mem(irg);
1588 ir_node *load = new_r_Load(start_bl, nomem, addr, load_mode, cons_floats);
1589 repl = new_r_Proj(load, load_mode, pn_Load_res);
1591 if (mode != load_mode) {
1592 repl = new_r_Conv(start_bl, repl, mode);
1595 /* The stack parameter is not primitive (it is a struct or array),
1596 * we thus will create a node representing the parameter's address
1602 assert(repl != NULL);
1604 /* Beware: the mode of the register parameters is always the mode of the register class
1605 which may be wrong. Add Conv's then. */
1606 mode = get_irn_mode(args[i]);
1607 if (mode != get_irn_mode(repl)) {
1608 repl = new_r_Conv(get_nodes_block(repl), repl, mode);
1610 exchange(args[i], repl);
1614 /* the arg proj is not needed anymore now and should be only used by the anchor */
1615 assert(get_irn_n_edges(arg_tuple) == 1);
1616 kill_node(arg_tuple);
1617 set_irg_args(irg, new_r_Bad(irg, mode_T));
1619 /* All Return nodes hang on the End node, so look for them there. */
1620 end = get_irg_end_block(irg);
1621 for (i = 0, n = get_Block_n_cfgpreds(end); i < n; ++i) {
1622 ir_node *irn = get_Block_cfgpred(end, i);
1624 if (is_Return(irn)) {
1625 ir_node *blk = get_nodes_block(irn);
1626 ir_node *mem = get_Return_mem(irn);
1627 ir_node *ret = create_be_return(env, irn, blk, mem, get_Return_n_ress(irn));
1632 /* if we have endless loops here, n might be <= 0. Do NOT create a be_Return then,
1633 the code is dead and will never be executed. */
1636 /** Fix the state inputs of calls that still hang on unknowns */
1637 static void fix_call_state_inputs(ir_graph *irg)
1639 be_abi_irg_t *env = be_get_irg_abi(irg);
1640 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1642 arch_register_t **stateregs = NEW_ARR_F(arch_register_t*, 0);
1644 /* Collect caller save registers */
1645 n = arch_env->n_register_classes;
1646 for (i = 0; i < n; ++i) {
1648 const arch_register_class_t *cls = &arch_env->register_classes[i];
1649 for (j = 0; j < cls->n_regs; ++j) {
1650 const arch_register_t *reg = arch_register_for_index(cls, j);
1651 if (reg->type & arch_register_type_state) {
1652 ARR_APP1(arch_register_t*, stateregs, (arch_register_t *)reg);
1657 n = ARR_LEN(env->calls);
1658 n_states = ARR_LEN(stateregs);
1659 for (i = 0; i < n; ++i) {
1661 ir_node *call = env->calls[i];
1663 arity = get_irn_arity(call);
1665 /* the state reg inputs are the last n inputs of the calls */
1666 for (s = 0; s < n_states; ++s) {
1667 int inp = arity - n_states + s;
1668 const arch_register_t *reg = stateregs[s];
1669 ir_node *regnode = be_abi_reg_map_get(env->regs, reg);
1671 set_irn_n(call, inp, regnode);
1675 DEL_ARR_F(stateregs);
1679 * Create a trampoline entity for the given method.
1681 static ir_entity *create_trampoline(be_main_env_t *be, ir_entity *method)
1683 ir_type *type = get_entity_type(method);
1684 ident *old_id = get_entity_ld_ident(method);
1685 ident *id = id_mangle3("", old_id, "$stub");
1686 ir_type *parent = be->pic_trampolines_type;
1687 ir_entity *ent = new_entity(parent, old_id, type);
1688 set_entity_ld_ident(ent, id);
1689 set_entity_visibility(ent, ir_visibility_private);
1695 * Returns the trampoline entity for the given method.
1697 static ir_entity *get_trampoline(be_main_env_t *env, ir_entity *method)
1699 ir_entity *result = pmap_get(ir_entity, env->ent_trampoline_map, method);
1700 if (result == NULL) {
1701 result = create_trampoline(env, method);
1702 pmap_insert(env->ent_trampoline_map, method, result);
1708 static ir_entity *create_pic_symbol(be_main_env_t *be, ir_entity *entity)
1710 ident *old_id = get_entity_ld_ident(entity);
1711 ident *id = id_mangle3("", old_id, "$non_lazy_ptr");
1712 ir_type *e_type = get_entity_type(entity);
1713 ir_type *type = new_type_pointer(e_type);
1714 ir_type *parent = be->pic_symbols_type;
1715 ir_entity *ent = new_entity(parent, old_id, type);
1716 set_entity_ld_ident(ent, id);
1717 set_entity_visibility(ent, ir_visibility_private);
1722 static ir_entity *get_pic_symbol(be_main_env_t *env, ir_entity *entity)
1724 ir_entity *result = pmap_get(ir_entity, env->ent_pic_symbol_map, entity);
1725 if (result == NULL) {
1726 result = create_pic_symbol(env, entity);
1727 pmap_insert(env->ent_pic_symbol_map, entity, result);
1736 * Returns non-zero if a given entity can be accessed using a relative address.
1738 static int can_address_relative(ir_entity *entity)
1740 return entity_has_definition(entity) && !(get_entity_linkage(entity) & IR_LINKAGE_MERGE);
1743 static ir_node *get_pic_base(ir_graph *irg)
1745 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1746 if (arch_env->impl->get_pic_base == NULL)
1748 return arch_env->impl->get_pic_base(irg);
1751 /** patches SymConsts to work in position independent code */
1752 static void fix_pic_symconsts(ir_node *node, void *data)
1754 ir_graph *irg = get_irn_irg(node);
1755 be_main_env_t *be = be_get_irg_main_env(irg);
1765 arity = get_irn_arity(node);
1766 for (i = 0; i < arity; ++i) {
1768 ir_node *pred = get_irn_n(node, i);
1770 ir_entity *pic_symbol;
1771 ir_node *pic_symconst;
1773 if (!is_SymConst(pred))
1776 entity = get_SymConst_entity(pred);
1777 block = get_nodes_block(pred);
1779 /* calls can jump to relative addresses, so we can directly jump to
1780 the (relatively) known call address or the trampoline */
1781 if (i == 1 && is_Call(node)) {
1782 ir_entity *trampoline;
1783 ir_node *trampoline_const;
1785 if (can_address_relative(entity))
1788 dbgi = get_irn_dbg_info(pred);
1789 trampoline = get_trampoline(be, entity);
1790 trampoline_const = new_rd_SymConst_addr_ent(dbgi, irg, mode_P_code,
1792 set_irn_n(node, i, trampoline_const);
1796 /* everything else is accessed relative to EIP */
1797 mode = get_irn_mode(pred);
1798 pic_base = get_pic_base(irg);
1800 /* all ok now for locally constructed stuff */
1801 if (can_address_relative(entity)) {
1802 ir_node *add = new_r_Add(block, pic_base, pred, mode);
1804 /* make sure the walker doesn't visit this add again */
1805 mark_irn_visited(add);
1806 set_irn_n(node, i, add);
1810 /* get entry from pic symbol segment */
1811 dbgi = get_irn_dbg_info(pred);
1812 pic_symbol = get_pic_symbol(be, entity);
1813 pic_symconst = new_rd_SymConst_addr_ent(dbgi, irg, mode_P_code,
1815 add = new_r_Add(block, pic_base, pic_symconst, mode);
1816 mark_irn_visited(add);
1818 /* we need an extra indirection for global data outside our current
1819 module. The loads are always safe and can therefore float
1820 and need no memory input */
1821 load = new_r_Load(block, get_irg_no_mem(irg), add, mode, cons_floats);
1822 load_res = new_r_Proj(load, mode, pn_Load_res);
1824 set_irn_n(node, i, load_res);
1828 void be_abi_introduce(ir_graph *irg)
1830 be_abi_irg_t *env = XMALLOCZ(be_abi_irg_t);
1831 ir_node *old_frame = get_irg_frame(irg);
1832 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1833 ir_entity *entity = get_irg_entity(irg);
1834 ir_type *method_type = get_entity_type(entity);
1835 be_irg_t *birg = be_birg_from_irg(irg);
1836 struct obstack *obst = &birg->obst;
1837 ir_node *dummy = new_r_Dummy(irg,
1838 arch_env->sp->reg_class->mode);
1841 /* determine allocatable registers */
1842 assert(birg->allocatable_regs == NULL);
1843 birg->allocatable_regs = rbitset_obstack_alloc(obst, arch_env->n_registers);
1844 for (r = 0; r < arch_env->n_registers; ++r) {
1845 const arch_register_t *reg = &arch_env->registers[r];
1846 if ( !(reg->type & arch_register_type_ignore)) {
1847 rbitset_set(birg->allocatable_regs, r);
1851 /* break here if backend provides a custom API.
1852 * Note: we shouldn't have to setup any be_abi_irg_t* stuff at all,
1853 * but need more cleanup to make this work
1855 be_set_irg_abi(irg, env);
1857 be_omit_fp = be_options.omit_fp;
1859 env->keep_map = pmap_create();
1860 env->call = be_abi_call_new(arch_env->sp->reg_class);
1861 arch_env_get_call_abi(arch_env, method_type, env->call);
1863 env->init_sp = dummy;
1864 env->calls = NEW_ARR_F(ir_node*, 0);
1868 if (be_options.pic) {
1869 irg_walk_graph(irg, fix_pic_symconsts, NULL, env);
1872 /* Lower all call nodes in the IRG. */
1875 /* Process the IRG */
1878 /* fix call inputs for state registers */
1879 fix_call_state_inputs(irg);
1881 /* We don't need the keep map anymore. */
1882 pmap_destroy(env->keep_map);
1883 env->keep_map = NULL;
1885 /* calls array is not needed anymore */
1886 DEL_ARR_F(env->calls);
1889 /* reroute the stack origin of the calls to the true stack origin. */
1890 exchange(dummy, env->init_sp);
1891 exchange(old_frame, get_irg_frame(irg));
1893 pmap_destroy(env->regs);
1897 void be_abi_free(ir_graph *irg)
1899 be_abi_irg_t *env = be_get_irg_abi(irg);
1901 if (env->call != NULL)
1902 be_abi_call_free(env->call);
1903 assert(env->regs == NULL);
1906 be_set_irg_abi(irg, NULL);
1909 void be_put_allocatable_regs(const ir_graph *irg,
1910 const arch_register_class_t *cls, bitset_t *bs)
1912 be_irg_t *birg = be_birg_from_irg(irg);
1913 unsigned *allocatable_regs = birg->allocatable_regs;
1916 assert(bitset_size(bs) == cls->n_regs);
1917 bitset_clear_all(bs);
1918 for (i = 0; i < cls->n_regs; ++i) {
1919 const arch_register_t *reg = &cls->regs[i];
1920 if (rbitset_is_set(allocatable_regs, reg->global_index))
1925 unsigned be_get_n_allocatable_regs(const ir_graph *irg,
1926 const arch_register_class_t *cls)
1928 bitset_t *bs = bitset_alloca(cls->n_regs);
1929 be_put_allocatable_regs(irg, cls, bs);
1930 return bitset_popcount(bs);
1933 void be_set_allocatable_regs(const ir_graph *irg,
1934 const arch_register_class_t *cls,
1935 unsigned *raw_bitset)
1937 be_irg_t *birg = be_birg_from_irg(irg);
1938 unsigned *allocatable_regs = birg->allocatable_regs;
1941 rbitset_clear_all(raw_bitset, cls->n_regs);
1942 for (i = 0; i < cls->n_regs; ++i) {
1943 const arch_register_t *reg = &cls->regs[i];
1944 if (rbitset_is_set(allocatable_regs, reg->global_index))
1945 rbitset_set(raw_bitset, i);
1949 BE_REGISTER_MODULE_CONSTRUCTOR(be_init_abi)
1950 void be_init_abi(void)
1952 FIRM_DBG_REGISTER(dbg, "firm.be.abi");