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
32 #include "irgraph_t.h"
35 #include "iredges_t.h"
38 #include "irprintf_t.h"
45 #include "raw_bitset.h"
56 #include "bessaconstr.h"
58 #include "betranshlp.h"
60 DEBUG_ONLY(static firm_dbg_module_t *dbg;)
62 typedef struct be_abi_call_arg_t {
63 unsigned is_res : 1; /**< 1: the call argument is a return value. 0: it's a call parameter. */
64 unsigned in_reg : 1; /**< 1: this argument is transmitted in registers. */
65 unsigned on_stack : 1; /**< 1: this argument is transmitted on the stack. */
66 unsigned callee : 1; /**< 1: someone called us. 0: We call another function */
69 const arch_register_t *reg;
72 unsigned alignment; /**< stack alignment */
73 unsigned space_before; /**< allocate space before */
74 unsigned space_after; /**< allocate space after */
77 struct be_abi_call_t {
78 be_abi_call_flags_t flags; /**< Flags describing the ABI behavior on calls */
79 int pop; /**< number of bytes the stack frame is shrinked by the callee on return. */
80 const be_abi_callbacks_t *cb;
81 ir_type *between_type;
83 const arch_register_class_t *cls_addr; /**< register class of the call address */
87 * The ABI information for the current graph.
90 be_abi_call_t *call; /**< The ABI call information. */
92 ir_node *init_sp; /**< The node representing the stack pointer
93 at the start of the function. */
95 ir_node *start; /**< The be_Start params node. */
96 pmap *regs; /**< A map of all callee-save and ignore regs to
97 their Projs to the RegParams node. */
98 int start_block_bias; /**< The stack bias at the end of the start block. */
100 pmap *keep_map; /**< mapping blocks to keep nodes. */
102 ir_node **calls; /**< flexible array containing all be_Call nodes */
105 static ir_heights_t *ir_heights;
107 /** Flag: if set, try to omit the frame pointer in all routines. */
108 static int be_omit_fp = 1;
110 static ir_node *be_abi_reg_map_get(pmap *map, const arch_register_t *reg)
112 return (ir_node*)pmap_get(map, reg);
115 static void be_abi_reg_map_set(pmap *map, const arch_register_t* reg,
118 pmap_insert(map, reg, node);
122 * Check if the given register is callee save, ie. will be saved by the callee.
124 static bool arch_register_is_callee_save(
125 const arch_env_t *arch_env,
126 const arch_register_t *reg)
128 if (arch_env->impl->register_saved_by)
129 return arch_env->impl->register_saved_by(reg, /*callee=*/1);
134 * Check if the given register is caller save, ie. must be saved by the caller.
136 static bool arch_register_is_caller_save(
137 const arch_env_t *arch_env,
138 const arch_register_t *reg)
140 if (arch_env->impl->register_saved_by)
141 return arch_env->impl->register_saved_by(reg, /*callee=*/0);
148 _ ____ ___ ____ _ _ _ _
149 / \ | __ )_ _| / ___|__ _| | | |__ __ _ ___| | _____
150 / _ \ | _ \| | | | / _` | | | '_ \ / _` |/ __| |/ / __|
151 / ___ \| |_) | | | |__| (_| | | | |_) | (_| | (__| <\__ \
152 /_/ \_\____/___| \____\__,_|_|_|_.__/ \__,_|\___|_|\_\___/
154 These callbacks are used by the backend to set the parameters
155 for a specific call type.
159 * Set compare function: compares two ABI call object arguments.
161 static int cmp_call_arg(const void *a, const void *b, size_t n)
163 const be_abi_call_arg_t *p = (const be_abi_call_arg_t*)a;
164 const be_abi_call_arg_t *q = (const be_abi_call_arg_t*)b;
166 return !(p->is_res == q->is_res && p->pos == q->pos && p->callee == q->callee);
170 * Get an ABI call object argument.
172 * @param call the abi call
173 * @param is_res true for call results, false for call arguments
174 * @param pos position of the argument
175 * @param callee context type - if we are callee or caller
177 static be_abi_call_arg_t *get_call_arg(be_abi_call_t *call, int is_res, int pos, int callee)
179 be_abi_call_arg_t arg;
182 memset(&arg, 0, sizeof(arg));
187 hash = is_res * 128 + pos;
189 return (be_abi_call_arg_t*)set_find(call->params, &arg, sizeof(arg), hash);
193 * Set an ABI call object argument.
195 static void remember_call_arg(be_abi_call_arg_t *arg, be_abi_call_t *call, be_abi_context_t context)
197 unsigned hash = arg->is_res * 128 + arg->pos;
198 if (context & ABI_CONTEXT_CALLEE) {
200 set_insert(call->params, arg, sizeof(*arg), hash);
202 if (context & ABI_CONTEXT_CALLER) {
204 set_insert(call->params, arg, sizeof(*arg), hash);
208 /* Set the flags for a call. */
209 void be_abi_call_set_flags(be_abi_call_t *call, be_abi_call_flags_t flags, const be_abi_callbacks_t *cb)
215 /* Sets the number of bytes the stackframe is shrinked by the callee on return */
216 void be_abi_call_set_pop(be_abi_call_t *call, int pop)
222 /* Set register class for call address */
223 void be_abi_call_set_call_address_reg_class(be_abi_call_t *call, const arch_register_class_t *cls)
225 call->cls_addr = cls;
229 void be_abi_call_param_stack(be_abi_call_t *call, int arg_pos,
230 ir_mode *load_mode, unsigned alignment,
231 unsigned space_before, unsigned space_after,
232 be_abi_context_t context)
234 be_abi_call_arg_t arg;
235 memset(&arg, 0, sizeof(arg));
236 assert(alignment > 0 && "Alignment must be greater than 0");
238 arg.load_mode = load_mode;
239 arg.alignment = alignment;
240 arg.space_before = space_before;
241 arg.space_after = space_after;
245 remember_call_arg(&arg, call, context);
248 void be_abi_call_param_reg(be_abi_call_t *call, int arg_pos, const arch_register_t *reg, be_abi_context_t context)
250 be_abi_call_arg_t arg;
251 memset(&arg, 0, sizeof(arg));
258 remember_call_arg(&arg, call, context);
261 void be_abi_call_res_reg(be_abi_call_t *call, int arg_pos, const arch_register_t *reg, be_abi_context_t context)
263 be_abi_call_arg_t arg;
264 memset(&arg, 0, sizeof(arg));
271 remember_call_arg(&arg, call, context);
274 /* Get the flags of a ABI call object. */
275 be_abi_call_flags_t be_abi_call_get_flags(const be_abi_call_t *call)
281 * Constructor for a new ABI call object.
283 * @param cls_addr register class of the call address
285 * @return the new ABI call object
287 static be_abi_call_t *be_abi_call_new(const arch_register_class_t *cls_addr)
289 be_abi_call_t *call = XMALLOCZ(be_abi_call_t);
292 call->params = new_set(cmp_call_arg, 16);
294 call->cls_addr = cls_addr;
296 call->flags.bits.try_omit_fp = be_omit_fp;
302 * Destructor for an ABI call object.
304 static void be_abi_call_free(be_abi_call_t *call)
306 del_set(call->params);
311 * Initializes the frame layout from parts
313 * @param frame the stack layout that will be initialized
314 * @param args the stack argument layout type
315 * @param between the between layout type
316 * @param locals the method frame type
317 * @param param_map an array mapping method argument positions to the stack argument type
319 * @return the initialized stack layout
321 static be_stack_layout_t *stack_frame_init(be_stack_layout_t *frame, ir_type *args,
322 ir_type *between, ir_type *locals,
323 ir_entity *param_map[])
325 frame->arg_type = args;
326 frame->between_type = between;
327 frame->frame_type = locals;
328 frame->initial_offset = 0;
329 frame->initial_bias = 0;
330 frame->order[1] = between;
331 frame->param_map = param_map;
333 /* typical decreasing stack: locals have the
334 * lowest addresses, arguments the highest */
335 frame->order[0] = locals;
336 frame->order[2] = args;
347 Adjustment of the calls inside a graph.
352 * Transform a call node into a be_Call node.
354 * @param env The ABI environment for the current irg.
355 * @param irn The call node.
356 * @param curr_sp The stack pointer node to use.
357 * @return The stack pointer after the call.
359 static ir_node *adjust_call(be_abi_irg_t *env, ir_node *irn, ir_node *curr_sp)
361 ir_graph *irg = get_irn_irg(irn);
362 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
363 ir_type *call_tp = get_Call_type(irn);
364 ir_node *call_ptr = get_Call_ptr(irn);
365 size_t n_params = get_method_n_params(call_tp);
366 ir_node *curr_mem = get_Call_mem(irn);
367 ir_node *bl = get_nodes_block(irn);
369 const arch_register_t *sp = arch_env->sp;
370 be_abi_call_t *call = be_abi_call_new(sp->reg_class);
371 ir_mode *mach_mode = sp->reg_class->mode;
372 int no_alloc = call->flags.bits.frame_is_setup_on_call;
373 int n_res = get_method_n_ress(call_tp);
374 int do_seq = call->flags.bits.store_args_sequential && !no_alloc;
376 ir_node *res_proj = NULL;
377 int n_reg_params = 0;
378 int n_stack_params = 0;
381 const arch_register_t **states = NEW_ARR_F(const arch_register_t*, 0);
382 const arch_register_t **destroyed_regs = NEW_ARR_F(const arch_register_t*, 0);
386 int n_reg_results = 0;
387 const ir_edge_t *edge;
389 int *stack_param_idx;
391 int throws_exception;
396 /* Let the isa fill out the abi description for that call node. */
397 arch_env_get_call_abi(arch_env, call_tp, call);
399 /* Insert code to put the stack arguments on the stack. */
400 assert(get_Call_n_params(irn) == n_params);
401 stack_param_idx = ALLOCAN(int, n_params);
402 for (p = 0; p < n_params; ++p) {
403 be_abi_call_arg_t *arg = get_call_arg(call, 0, p, 0);
406 int arg_size = get_type_size_bytes(get_method_param_type(call_tp, p));
408 stack_size += round_up2(arg->space_before, arg->alignment);
409 stack_size += round_up2(arg_size, arg->alignment);
410 stack_size += round_up2(arg->space_after, arg->alignment);
412 stack_param_idx[n_stack_params++] = p;
416 /* Collect all arguments which are passed in registers. */
417 reg_param_idxs = ALLOCAN(int, n_params);
418 for (p = 0; p < n_params; ++p) {
419 be_abi_call_arg_t *arg = get_call_arg(call, 0, p, 0);
420 if (arg && arg->in_reg) {
421 reg_param_idxs[n_reg_params++] = p;
426 * If the stack is decreasing and we do not want to store sequentially,
427 * or someone else allocated the call frame
428 * we allocate as much space on the stack all parameters need, by
429 * moving the stack pointer along the stack's direction.
431 * Note: we also have to do this for stack_size == 0, because we may have
432 * to adjust stack alignment for the call.
434 if (!do_seq && !no_alloc) {
435 curr_sp = be_new_IncSP(sp, bl, curr_sp, stack_size, 1);
438 dbgi = get_irn_dbg_info(irn);
439 /* If there are some parameters which shall be passed on the stack. */
440 if (n_stack_params > 0) {
442 ir_node **in = ALLOCAN(ir_node*, n_stack_params+1);
445 /* push params in reverse direction because stack grows downwards */
447 for (i = 0; i < n_stack_params >> 1; ++i) {
448 int other = n_stack_params - i - 1;
449 int tmp = stack_param_idx[i];
450 stack_param_idx[i] = stack_param_idx[other];
451 stack_param_idx[other] = tmp;
455 curr_mem = get_Call_mem(irn);
457 in[n_in++] = curr_mem;
460 for (i = 0; i < n_stack_params; ++i) {
461 int p = stack_param_idx[i];
462 be_abi_call_arg_t *arg = get_call_arg(call, 0, p, 0);
463 ir_node *param = get_Call_param(irn, p);
464 ir_node *addr = curr_sp;
466 ir_type *param_type = get_method_param_type(call_tp, p);
467 int param_size = get_type_size_bytes(param_type) + arg->space_after;
470 * If we wanted to build the arguments sequentially,
471 * the stack pointer for the next must be incremented,
472 * and the memory value propagated.
476 addr = curr_sp = be_new_IncSP(sp, bl, curr_sp,
477 param_size + arg->space_before, 0);
478 add_irn_dep(curr_sp, curr_mem);
480 curr_ofs += arg->space_before;
481 curr_ofs = round_up2(curr_ofs, arg->alignment);
483 /* Make the expression to compute the argument's offset. */
485 ir_mode *constmode = mach_mode;
486 if (mode_is_reference(mach_mode)) {
489 addr = new_r_Const_long(irg, constmode, curr_ofs);
490 addr = new_r_Add(bl, curr_sp, addr, mach_mode);
494 /* Insert a store for primitive arguments. */
495 if (is_atomic_type(param_type)) {
496 ir_node *nomem = get_irg_no_mem(irg);
497 ir_node *mem_input = do_seq ? curr_mem : nomem;
498 ir_node *store = new_rd_Store(dbgi, bl, mem_input, addr, param, cons_none);
499 mem = new_r_Proj(store, mode_M, pn_Store_M);
501 /* Make a mem copy for compound arguments. */
504 assert(mode_is_reference(get_irn_mode(param)));
505 copy = new_rd_CopyB(dbgi, bl, curr_mem, addr, param, param_type);
506 mem = new_r_Proj(copy, mode_M, pn_CopyB_M);
509 curr_ofs += param_size;
517 /* We need the sync only, if we didn't build the stores sequentially. */
519 if (n_stack_params >= 1) {
520 curr_mem = new_r_Sync(bl, n_in, in);
522 curr_mem = get_Call_mem(irn);
527 /* Put caller save into the destroyed set and state registers in the states
529 for (i = 0, n = arch_env->n_register_classes; i < n; ++i) {
531 const arch_register_class_t *cls = &arch_env->register_classes[i];
532 for (j = 0; j < cls->n_regs; ++j) {
533 const arch_register_t *reg = arch_register_for_index(cls, j);
535 /* even if destroyed all is specified, neither SP nor FP are
536 * destroyed (else bad things will happen) */
537 if (reg == arch_env->sp || reg == arch_env->bp)
540 if (reg->type & arch_register_type_state) {
541 ARR_APP1(const arch_register_t*, destroyed_regs, reg);
542 ARR_APP1(const arch_register_t*, states, reg);
543 /* we're already in the destroyed set so no need for further
547 if (arch_register_is_caller_save(arch_env, reg)) {
548 if (!(reg->type & arch_register_type_ignore)) {
549 ARR_APP1(const arch_register_t*, destroyed_regs, reg);
555 /* search the largest result proj number */
556 res_projs = ALLOCANZ(ir_node*, n_res);
558 foreach_out_edge(irn, edge) {
559 const ir_edge_t *res_edge;
560 ir_node *irn = get_edge_src_irn(edge);
562 if (!is_Proj(irn) || get_Proj_proj(irn) != pn_Call_T_result)
565 foreach_out_edge(irn, res_edge) {
567 ir_node *res = get_edge_src_irn(res_edge);
569 assert(is_Proj(res));
571 proj = get_Proj_proj(res);
572 assert(proj < n_res);
573 assert(res_projs[proj] == NULL);
574 res_projs[proj] = res;
580 /** TODO: this is not correct for cases where return values are passed
581 * on the stack, but no known ABI does this currently...
583 n_reg_results = n_res;
586 in = ALLOCAN(ir_node*, n_reg_params + ARR_LEN(states));
588 /* make the back end call node and set its register requirements. */
589 for (i = 0; i < n_reg_params; ++i) {
590 in[n_ins++] = get_Call_param(irn, reg_param_idxs[i]);
593 /* add state registers ins */
594 for (s = 0; s < ARR_LEN(states); ++s) {
595 const arch_register_t *reg = states[s];
596 const arch_register_class_t *cls = reg->reg_class;
597 ir_node *regnode = new_r_Unknown(irg, cls->mode);
598 in[n_ins++] = regnode;
600 assert(n_ins == (int) (n_reg_params + ARR_LEN(states)));
602 /* ins collected, build the call */
603 throws_exception = ir_throws_exception(irn);
604 if (env->call->flags.bits.call_has_imm && is_SymConst(call_ptr)) {
606 low_call = be_new_Call(dbgi, irg, bl, curr_mem, curr_sp, curr_sp,
607 n_reg_results + pn_be_Call_first_res + ARR_LEN(destroyed_regs),
608 n_ins, in, get_Call_type(irn));
609 be_Call_set_entity(low_call, get_SymConst_entity(call_ptr));
612 low_call = be_new_Call(dbgi, irg, bl, curr_mem, curr_sp, call_ptr,
613 n_reg_results + pn_be_Call_first_res + ARR_LEN(destroyed_regs),
614 n_ins, in, get_Call_type(irn));
616 ir_set_throws_exception(low_call, throws_exception);
617 be_Call_set_pop(low_call, call->pop);
619 /* put the call into the list of all calls for later processing */
620 ARR_APP1(ir_node *, env->calls, low_call);
622 /* create new stack pointer */
623 curr_sp = new_r_Proj(low_call, get_irn_mode(curr_sp), pn_be_Call_sp);
624 be_set_constr_single_reg_out(low_call, pn_be_Call_sp, sp,
625 arch_register_req_type_ignore | arch_register_req_type_produces_sp);
626 arch_set_irn_register(curr_sp, sp);
628 /* now handle results */
629 for (i = 0; i < n_res; ++i) {
630 ir_node *proj = res_projs[i];
631 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 0);
632 long pn = i + pn_be_Call_first_res;
634 /* returns values on stack not supported yet */
638 shift the proj number to the right, since we will drop the
639 unspeakable Proj_T from the Call. Therefore, all real argument
640 Proj numbers must be increased by pn_be_Call_first_res
642 pn = i + pn_be_Call_first_res;
645 ir_type *res_type = get_method_res_type(call_tp, i);
646 ir_mode *mode = get_type_mode(res_type);
647 proj = new_r_Proj(low_call, mode, pn);
650 set_Proj_pred(proj, low_call);
651 set_Proj_proj(proj, pn);
655 /* remove register from destroyed regs */
657 size_t n = ARR_LEN(destroyed_regs);
658 for (j = 0; j < n; ++j) {
659 if (destroyed_regs[j] == arg->reg) {
660 destroyed_regs[j] = destroyed_regs[n-1];
661 ARR_SHRINKLEN(destroyed_regs,n-1);
669 Set the register class of the call address to
670 the backend provided class (default: stack pointer class)
672 be_node_set_reg_class_in(low_call, n_be_Call_ptr, call->cls_addr);
674 DBG((dbg, LEVEL_3, "\tcreated backend call %+F\n", low_call));
676 /* Set the register classes and constraints of the Call parameters. */
677 for (i = 0; i < n_reg_params; ++i) {
678 int index = reg_param_idxs[i];
679 be_abi_call_arg_t *arg = get_call_arg(call, 0, index, 0);
680 assert(arg->reg != NULL);
682 be_set_constr_single_reg_in(low_call, n_be_Call_first_arg + i,
683 arg->reg, arch_register_req_type_none);
686 /* Set the register constraints of the results. */
687 for (i = 0; i < n_res; ++i) {
688 ir_node *proj = res_projs[i];
689 const be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 0);
690 int pn = get_Proj_proj(proj);
693 be_set_constr_single_reg_out(low_call, pn, arg->reg,
694 arch_register_req_type_none);
695 arch_set_irn_register(proj, arg->reg);
697 exchange(irn, low_call);
699 /* kill the ProjT node */
700 if (res_proj != NULL) {
704 /* Make additional projs for the caller save registers
705 and the Keep node which keeps them alive. */
711 int curr_res_proj = pn_be_Call_first_res + n_reg_results;
714 n_ins = ARR_LEN(destroyed_regs) + n_reg_results + 1;
715 in = ALLOCAN(ir_node *, n_ins);
717 /* also keep the stack pointer */
718 set_irn_link(curr_sp, (void*) sp);
721 for (d = 0; d < ARR_LEN(destroyed_regs); ++d) {
722 const arch_register_t *reg = destroyed_regs[d];
723 ir_node *proj = new_r_Proj(low_call, reg->reg_class->mode, curr_res_proj);
725 /* memorize the register in the link field. we need afterwards to set the register class of the keep correctly. */
726 be_set_constr_single_reg_out(low_call, curr_res_proj, reg,
727 arch_register_req_type_none);
728 arch_set_irn_register(proj, reg);
730 set_irn_link(proj, (void*) reg);
735 for (i = 0; i < n_reg_results; ++i) {
736 ir_node *proj = res_projs[i];
737 const arch_register_t *reg = arch_get_irn_register(proj);
738 set_irn_link(proj, (void*) reg);
743 /* create the Keep for the caller save registers */
744 keep = be_new_Keep(bl, n, in);
745 for (i = 0; i < n; ++i) {
746 const arch_register_t *reg = (const arch_register_t*)get_irn_link(in[i]);
747 be_node_set_reg_class_in(keep, i, arch_register_get_class(reg));
751 /* Clean up the stack. */
752 assert(stack_size >= call->pop);
753 stack_size -= call->pop;
755 if (stack_size > 0) {
756 ir_node *mem_proj = NULL;
758 foreach_out_edge(low_call, edge) {
759 ir_node *irn = get_edge_src_irn(edge);
760 if (is_Proj(irn) && get_Proj_proj(irn) == pn_Call_M) {
767 mem_proj = new_r_Proj(low_call, mode_M, pn_be_Call_M);
768 keep_alive(mem_proj);
771 /* Clean up the stack frame or revert alignment fixes if we allocated it */
773 curr_sp = be_new_IncSP(sp, bl, curr_sp, -stack_size, 0);
776 be_abi_call_free(call);
779 DEL_ARR_F(destroyed_regs);
785 * Adjust the size of a node representing a stack alloc or free for the minimum stack alignment.
787 * @param alignment the minimum stack alignment
788 * @param size the node containing the non-aligned size
789 * @param block the block where new nodes are allocated on
790 * @param dbg debug info for new nodes
792 * @return a node representing the aligned size
794 static ir_node *adjust_alloc_size(unsigned stack_alignment, ir_node *size,
795 ir_node *block, dbg_info *dbg)
797 if (stack_alignment > 1) {
803 assert(is_po2(stack_alignment));
805 mode = get_irn_mode(size);
806 tv = new_tarval_from_long(stack_alignment-1, mode);
807 irg = get_Block_irg(block);
808 mask = new_r_Const(irg, tv);
809 size = new_rd_Add(dbg, block, size, mask, mode);
811 tv = new_tarval_from_long(-(long)stack_alignment, mode);
812 mask = new_r_Const(irg, tv);
813 size = new_rd_And(dbg, block, size, mask, mode);
819 * The alloca is transformed into a back end alloca node and connected to the stack nodes.
821 static ir_node *adjust_alloc(be_abi_irg_t *env, ir_node *alloc, ir_node *curr_sp)
823 ir_node *block = get_nodes_block(alloc);
824 ir_graph *irg = get_Block_irg(block);
825 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
826 ir_node *alloc_mem = NULL;
827 ir_node *alloc_res = NULL;
828 ir_type *type = get_Alloc_type(alloc);
831 const ir_edge_t *edge;
836 unsigned stack_alignment;
838 /* all non-stack Alloc nodes should already be lowered before the backend */
839 assert(get_Alloc_where(alloc) == stack_alloc);
841 foreach_out_edge(alloc, edge) {
842 ir_node *irn = get_edge_src_irn(edge);
844 assert(is_Proj(irn));
845 switch (get_Proj_proj(irn)) {
857 /* Beware: currently Alloc nodes without a result might happen,
858 only escape analysis kills them and this phase runs only for object
859 oriented source. We kill the Alloc here. */
860 if (alloc_res == NULL && alloc_mem) {
861 exchange(alloc_mem, get_Alloc_mem(alloc));
865 dbg = get_irn_dbg_info(alloc);
866 count = get_Alloc_count(alloc);
868 /* we might need to multiply the count with the element size */
869 if (type != firm_unknown_type && get_type_size_bytes(type) != 1) {
870 ir_mode *mode = get_irn_mode(count);
871 ir_tarval *tv = new_tarval_from_long(get_type_size_bytes(type),
873 ir_node *cnst = new_rd_Const(dbg, irg, tv);
874 size = new_rd_Mul(dbg, block, count, cnst, mode);
879 /* The stack pointer will be modified in an unknown manner.
880 We cannot omit it. */
881 env->call->flags.bits.try_omit_fp = 0;
883 stack_alignment = 1 << arch_env->stack_alignment;
884 size = adjust_alloc_size(stack_alignment, size, block, dbg);
885 new_alloc = be_new_AddSP(arch_env->sp, block, curr_sp, size);
886 set_irn_dbg_info(new_alloc, dbg);
888 if (alloc_mem != NULL) {
892 addsp_mem = new_r_Proj(new_alloc, mode_M, pn_be_AddSP_M);
894 /* We need to sync the output mem of the AddSP with the input mem
895 edge into the alloc node. */
896 ins[0] = get_Alloc_mem(alloc);
898 sync = new_r_Sync(block, 2, ins);
900 exchange(alloc_mem, sync);
903 exchange(alloc, new_alloc);
905 /* fix projnum of alloca res */
906 set_Proj_proj(alloc_res, pn_be_AddSP_res);
908 curr_sp = new_r_Proj(new_alloc, get_irn_mode(curr_sp), pn_be_AddSP_sp);
915 * The Free is transformed into a back end free node and connected to the stack nodes.
917 static ir_node *adjust_free(be_abi_irg_t *env, ir_node *free, ir_node *curr_sp)
919 ir_node *block = get_nodes_block(free);
920 ir_graph *irg = get_irn_irg(free);
921 ir_type *type = get_Free_type(free);
922 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
923 ir_mode *sp_mode = arch_env->sp->reg_class->mode;
924 dbg_info *dbg = get_irn_dbg_info(free);
925 ir_node *subsp, *mem, *res, *size, *sync;
927 unsigned stack_alignment;
929 /* all non-stack-alloc Free nodes should already be lowered before the
931 assert(get_Free_where(free) == stack_alloc);
933 /* we might need to multiply the size with the element size */
934 if (type != firm_unknown_type && get_type_size_bytes(type) != 1) {
935 ir_tarval *tv = new_tarval_from_long(get_type_size_bytes(type), mode_Iu);
936 ir_node *cnst = new_rd_Const(dbg, irg, tv);
937 ir_node *mul = new_rd_Mul(dbg, block, get_Free_count(free),
941 size = get_Free_count(free);
944 stack_alignment = 1 << arch_env->stack_alignment;
945 size = adjust_alloc_size(stack_alignment, size, block, dbg);
947 /* The stack pointer will be modified in an unknown manner.
948 We cannot omit it. */
949 env->call->flags.bits.try_omit_fp = 0;
950 subsp = be_new_SubSP(arch_env->sp, block, curr_sp, size);
951 set_irn_dbg_info(subsp, dbg);
953 mem = new_r_Proj(subsp, mode_M, pn_be_SubSP_M);
954 res = new_r_Proj(subsp, sp_mode, pn_be_SubSP_sp);
956 /* we need to sync the memory */
957 in[0] = get_Free_mem(free);
959 sync = new_r_Sync(block, 2, in);
961 /* and make the AddSP dependent on the former memory */
962 add_irn_dep(subsp, get_Free_mem(free));
965 exchange(free, sync);
972 * Check if a node is somehow data dependent on another one.
973 * both nodes must be in the same basic block.
974 * @param n1 The first node.
975 * @param n2 The second node.
976 * @return 1, if n1 is data dependent (transitively) on n2, 0 if not.
978 static int dependent_on(ir_node *n1, ir_node *n2)
980 assert(get_nodes_block(n1) == get_nodes_block(n2));
982 return heights_reachable_in_block(ir_heights, n1, n2);
985 static int cmp_call_dependency(const void *c1, const void *c2)
987 ir_node *n1 = *(ir_node **) c1;
988 ir_node *n2 = *(ir_node **) c2;
992 Classical qsort() comparison function behavior:
993 0 if both elements are equal
994 1 if second is "smaller" that first
995 -1 if first is "smaller" that second
997 if (dependent_on(n1, n2))
1000 if (dependent_on(n2, n1))
1003 /* The nodes have no depth order, but we need a total order because qsort()
1006 * Additionally, we need to respect transitive dependencies. Consider a
1007 * Call a depending on Call b and an independent Call c.
1008 * We MUST NOT order c > a and b > c. */
1009 h1 = get_irn_height(ir_heights, n1);
1010 h2 = get_irn_height(ir_heights, n2);
1011 if (h1 < h2) return -1;
1012 if (h1 > h2) return 1;
1013 /* Same height, so use a random (but stable) order */
1014 return get_irn_idx(n1) - get_irn_idx(n2);
1018 * Walker: links all Call/Alloc/Free nodes to the Block they are contained.
1019 * Clears the irg_is_leaf flag if a Call is detected.
1021 static void link_ops_in_block_walker(ir_node *irn, void *data)
1023 be_abi_irg_t *env = (be_abi_irg_t*)data;
1024 unsigned code = get_irn_opcode(irn);
1026 if (code == iro_Call ||
1027 (code == iro_Alloc && get_Alloc_where(irn) == stack_alloc) ||
1028 (code == iro_Free && get_Free_where(irn) == stack_alloc)) {
1029 ir_node *bl = get_nodes_block(irn);
1030 void *save = get_irn_link(bl);
1032 if (code == iro_Call)
1033 env->call->flags.bits.irg_is_leaf = 0;
1035 set_irn_link(irn, save);
1036 set_irn_link(bl, irn);
1039 if (code == iro_Builtin && get_Builtin_kind(irn) == ir_bk_return_address) {
1040 ir_node *param = get_Builtin_param(irn, 0);
1041 ir_tarval *tv = get_Const_tarval(param);
1042 unsigned long value = get_tarval_long(tv);
1043 /* use ebp, so the climbframe algo works... */
1045 env->call->flags.bits.try_omit_fp = 0;
1052 * Process all Call/Alloc/Free nodes inside a basic block.
1053 * Note that the link field of the block must contain a linked list of all
1054 * nodes inside the Block. We first order this list according to data dependency
1055 * and that connect the nodes together.
1057 static void process_ops_in_block(ir_node *bl, void *data)
1059 be_abi_irg_t *env = (be_abi_irg_t*)data;
1060 ir_node *curr_sp = env->init_sp;
1067 for (irn = (ir_node*)get_irn_link(bl); irn != NULL;
1068 irn = (ir_node*)get_irn_link(irn)) {
1072 nodes = ALLOCAN(ir_node*, n_nodes);
1073 for (irn = (ir_node*)get_irn_link(bl), n = 0; irn != NULL;
1074 irn = (ir_node*)get_irn_link(irn), ++n) {
1078 /* If there were call nodes in the block. */
1083 /* order the call nodes according to data dependency */
1084 qsort(nodes, n_nodes, sizeof(nodes[0]), cmp_call_dependency);
1086 for (i = n_nodes - 1; i >= 0; --i) {
1087 ir_node *irn = nodes[i];
1089 DBG((dbg, LEVEL_3, "\tprocessing call %+F\n", irn));
1090 switch (get_irn_opcode(irn)) {
1093 /* The stack pointer will be modified due to a call. */
1094 env->call->flags.bits.try_omit_fp = 0;
1096 curr_sp = adjust_call(env, irn, curr_sp);
1099 if (get_Alloc_where(irn) == stack_alloc)
1100 curr_sp = adjust_alloc(env, irn, curr_sp);
1103 if (get_Free_where(irn) == stack_alloc)
1104 curr_sp = adjust_free(env, irn, curr_sp);
1107 panic("invalid call");
1111 /* Keep the last stack state in the block by tying it to Keep node,
1112 * the proj from calls is already kept */
1113 if (curr_sp != env->init_sp &&
1114 !(is_Proj(curr_sp) && be_is_Call(get_Proj_pred(curr_sp)))) {
1116 keep = be_new_Keep(bl, 1, nodes);
1117 pmap_insert(env->keep_map, bl, keep);
1121 set_irn_link(bl, curr_sp);
1125 * Adjust all call nodes in the graph to the ABI conventions.
1127 static void process_calls(ir_graph *irg)
1129 be_abi_irg_t *abi = be_get_irg_abi(irg);
1131 abi->call->flags.bits.irg_is_leaf = 1;
1132 irg_walk_graph(irg, firm_clear_link, link_ops_in_block_walker, abi);
1134 ir_heights = heights_new(irg);
1135 irg_block_walk_graph(irg, NULL, process_ops_in_block, abi);
1136 heights_free(ir_heights);
1140 * Computes the stack argument layout type.
1141 * Changes a possibly allocated value param type by moving
1142 * entities to the stack layout type.
1144 * @param call the current call ABI
1145 * @param method_type the method type
1146 * @param param_map an array mapping method arguments to the stack layout
1149 * @return the stack argument layout type
1151 static ir_type *compute_arg_type(ir_graph *irg, be_abi_call_t *call,
1152 ir_type *method_type, ir_entity ***param_map)
1154 struct obstack *obst = be_get_be_obst(irg);
1155 ir_type *frame_type = get_irg_frame_type(irg);
1156 size_t n_params = get_method_n_params(method_type);
1157 size_t n_frame_members = get_compound_n_members(frame_type);
1158 ir_entity *va_start_entity = NULL;
1166 *param_map = map = OALLOCNZ(obst, ir_entity*, n_params);
1167 res = new_type_struct(new_id_from_chars("arg_type", 8));
1169 /* collect existing entities for value_param_types */
1170 for (f = n_frame_members; f > 0; ) {
1171 ir_entity *entity = get_compound_member(frame_type, --f);
1174 set_entity_link(entity, NULL);
1175 if (!is_parameter_entity(entity))
1177 num = get_entity_parameter_number(entity);
1178 if (num == IR_VA_START_PARAMETER_NUMBER) {
1179 /* move entity to new arg_type */
1180 set_entity_owner(entity, res);
1181 va_start_entity = entity;
1184 assert(num < n_params);
1185 if (map[num] != NULL)
1186 panic("multiple entities for parameter %u in %+F found", f, irg);
1188 if (num != n_params && !get_call_arg(call, 0, num, 1)->on_stack) {
1189 /* don't move this entity */
1194 /* move entity to new arg_type */
1195 set_entity_owner(entity, res);
1198 for (i = 0; i < n_params; ++i) {
1199 be_abi_call_arg_t *arg = get_call_arg(call, 0, i, 1);
1200 ir_type *param_type = get_method_param_type(method_type, i);
1203 if (!arg->on_stack) {
1207 if (entity == NULL) {
1208 /* create a new entity */
1209 entity = new_parameter_entity(res, i, param_type);
1211 ofs += arg->space_before;
1212 ofs = round_up2(ofs, arg->alignment);
1213 set_entity_offset(entity, ofs);
1214 ofs += arg->space_after;
1215 ofs += get_type_size_bytes(param_type);
1216 arg->stack_ent = entity;
1218 if (va_start_entity != NULL) {
1219 set_entity_offset(va_start_entity, ofs);
1221 set_type_size_bytes(res, ofs);
1222 set_type_state(res, layout_fixed);
1228 const arch_register_t *reg;
1232 static int cmp_regs(const void *a, const void *b)
1234 const reg_node_map_t *p = (const reg_node_map_t*)a;
1235 const reg_node_map_t *q = (const reg_node_map_t*)b;
1237 if (p->reg->reg_class == q->reg->reg_class)
1238 return p->reg->index - q->reg->index;
1240 return p->reg->reg_class < q->reg->reg_class ? -1 : +1;
1243 static void reg_map_to_arr(reg_node_map_t *res, pmap *reg_map)
1246 size_t n = pmap_count(reg_map);
1249 foreach_pmap(reg_map, ent) {
1250 res[i].reg = (const arch_register_t*)ent->key;
1251 res[i].irn = (ir_node*)ent->value;
1255 qsort(res, n, sizeof(res[0]), cmp_regs);
1259 * Creates a be_Return for a Return node.
1261 * @param @env the abi environment
1262 * @param irn the Return node or NULL if there was none
1263 * @param bl the block where the be_Retun should be placed
1264 * @param mem the current memory
1265 * @param n_res number of return results
1267 static ir_node *create_be_return(be_abi_irg_t *env, ir_node *irn, ir_node *bl,
1268 ir_node *mem, int n_res)
1270 be_abi_call_t *call = env->call;
1271 ir_graph *irg = get_Block_irg(bl);
1272 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1274 pmap *reg_map = pmap_create();
1275 ir_node *keep = (ir_node*)pmap_get(env->keep_map, bl);
1282 const arch_register_t **regs;
1286 get the valid stack node in this block.
1287 If we had a call in that block there is a Keep constructed by process_calls()
1288 which points to the last stack modification in that block. we'll use
1289 it then. Else we use the stack from the start block and let
1290 the ssa construction fix the usage.
1292 stack = be_abi_reg_map_get(env->regs, arch_env->sp);
1294 stack = get_irn_n(keep, 0);
1296 remove_End_keepalive(get_irg_end(irg), keep);
1299 /* Insert results for Return into the register map. */
1300 for (i = 0; i < n_res; ++i) {
1301 ir_node *res = get_Return_res(irn, i);
1302 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 1);
1303 assert(arg->in_reg && "return value must be passed in register");
1304 pmap_insert(reg_map, (void *) arg->reg, res);
1307 /* Add uses of the callee save registers. */
1308 foreach_pmap(env->regs, ent) {
1309 const arch_register_t *reg = (const arch_register_t*)ent->key;
1310 if ((reg->type & arch_register_type_ignore) || arch_register_is_callee_save(arch_env, reg))
1311 pmap_insert(reg_map, ent->key, ent->value);
1314 be_abi_reg_map_set(reg_map, arch_env->sp, stack);
1317 Maximum size of the in array for Return nodes is
1318 return args + callee save/ignore registers + memory + stack pointer
1320 in_max = pmap_count(reg_map) + n_res + 2;
1322 in = ALLOCAN(ir_node*, in_max);
1323 regs = ALLOCAN(arch_register_t const*, in_max);
1326 in[1] = be_abi_reg_map_get(reg_map, arch_env->sp);
1328 regs[1] = arch_env->sp;
1331 /* clear SP entry, since it has already been grown. */
1332 pmap_insert(reg_map, (void *) arch_env->sp, NULL);
1333 for (i = 0; i < n_res; ++i) {
1334 be_abi_call_arg_t *arg = get_call_arg(call, 1, i, 1);
1336 in[n] = be_abi_reg_map_get(reg_map, arg->reg);
1337 regs[n++] = arg->reg;
1339 /* Clear the map entry to mark the register as processed. */
1340 be_abi_reg_map_set(reg_map, arg->reg, NULL);
1343 /* grow the rest of the stuff. */
1344 foreach_pmap(reg_map, ent) {
1346 in[n] = (ir_node*)ent->value;
1347 regs[n++] = (const arch_register_t*)ent->key;
1351 /* The in array for the new back end return is now ready. */
1353 dbgi = get_irn_dbg_info(irn);
1357 /* we have to pop the shadow parameter in in case of struct returns */
1359 ret = be_new_Return(dbgi, irg, bl, n_res, pop, n, in);
1361 /* Set the register classes of the return's parameter accordingly. */
1362 for (i = 0; i < n; ++i) {
1363 if (regs[i] == NULL)
1366 be_set_constr_single_reg_in(ret, i, regs[i], arch_register_req_type_none);
1369 /* Free the space of the Epilog's in array and the register <-> proj map. */
1370 pmap_destroy(reg_map);
1375 typedef struct lower_frame_sels_env_t {
1376 ir_node *frame; /**< the current frame */
1377 const arch_register_class_t *sp_class; /**< register class of the stack pointer */
1378 const arch_register_class_t *link_class; /**< register class of the link pointer */
1379 ir_type *frame_tp; /**< the frame type */
1380 int static_link_pos; /**< argument number of the hidden static link */
1381 } lower_frame_sels_env_t;
1384 * Walker: Replaces Sels of frame type and
1385 * value param type entities by FrameAddress.
1386 * Links all used entities.
1388 static void lower_frame_sels_walker(ir_node *irn, void *data)
1390 lower_frame_sels_env_t *ctx = (lower_frame_sels_env_t*)data;
1393 ir_node *ptr = get_Sel_ptr(irn);
1395 if (ptr == ctx->frame) {
1396 ir_entity *ent = get_Sel_entity(irn);
1397 ir_node *bl = get_nodes_block(irn);
1400 nw = be_new_FrameAddr(ctx->sp_class, bl, ctx->frame, ent);
1407 * The start block has no jump, instead it has an initial exec Proj.
1408 * The backend wants to handle all blocks the same way, so we replace
1409 * the out cfg edge with a real jump.
1411 static void fix_start_block(ir_graph *irg)
1413 ir_node *initial_X = get_irg_initial_exec(irg);
1414 ir_node *start_block = get_irg_start_block(irg);
1415 ir_node *jmp = new_r_Jmp(start_block);
1417 assert(is_Proj(initial_X));
1418 exchange(initial_X, jmp);
1419 set_irg_initial_exec(irg, new_r_Bad(irg, mode_X));
1421 /* merge start block with successor if possible */
1423 const ir_edge_t *edge;
1424 foreach_out_edge(jmp, edge) {
1425 ir_node *succ = get_edge_src_irn(edge);
1426 if (!is_Block(succ))
1429 if (get_irn_arity(succ) == 1) {
1430 exchange(succ, start_block);
1438 * Modify the irg itself and the frame type.
1440 static void modify_irg(ir_graph *irg)
1442 be_abi_irg_t *env = be_get_irg_abi(irg);
1443 be_abi_call_t *call = env->call;
1444 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1445 const arch_register_t *sp = arch_env->sp;
1446 ir_type *method_type = get_entity_type(get_irg_entity(irg));
1447 be_irg_t *birg = be_birg_from_irg(irg);
1448 struct obstack *obst = be_get_be_obst(irg);
1449 be_stack_layout_t *stack_layout = be_get_irg_stack_layout(irg);
1452 ir_node *new_mem_proj;
1458 unsigned frame_size;
1461 const arch_register_t *fp_reg;
1462 ir_node *frame_pointer;
1466 const ir_edge_t *edge;
1467 ir_type *arg_type, *bet_type;
1468 lower_frame_sels_env_t ctx;
1469 ir_entity **param_map;
1471 DBG((dbg, LEVEL_1, "introducing abi on %+F\n", irg));
1473 old_mem = get_irg_initial_mem(irg);
1475 irp_reserve_resources(irp, IRP_RESOURCE_ENTITY_LINK);
1477 arg_type = compute_arg_type(irg, call, method_type, ¶m_map);
1479 /* Convert the Sel nodes in the irg to frame addr nodes: */
1480 ctx.frame = get_irg_frame(irg);
1481 ctx.sp_class = arch_env->sp->reg_class;
1482 ctx.link_class = arch_env->link_class;
1483 ctx.frame_tp = get_irg_frame_type(irg);
1485 /* layout the stackframe now */
1486 if (get_type_state(ctx.frame_tp) == layout_undefined) {
1487 default_layout_compound_type(ctx.frame_tp);
1490 /* align stackframe to 4 byte */
1491 frame_size = get_type_size_bytes(ctx.frame_tp);
1492 if (frame_size % 4 != 0) {
1493 set_type_size_bytes(ctx.frame_tp, frame_size + 4 - (frame_size % 4));
1496 env->regs = pmap_create();
1498 n_params = get_method_n_params(method_type);
1499 args = OALLOCNZ(obst, ir_node*, n_params);
1501 be_add_parameter_entity_stores(irg);
1503 irg_walk_graph(irg, lower_frame_sels_walker, NULL, &ctx);
1505 irp_free_resources(irp, IRP_RESOURCE_ENTITY_LINK);
1507 /* Fill the argument vector */
1508 arg_tuple = get_irg_args(irg);
1509 foreach_out_edge(arg_tuple, edge) {
1510 ir_node *irn = get_edge_src_irn(edge);
1511 if (! is_Anchor(irn)) {
1512 int nr = get_Proj_proj(irn);
1514 DBG((dbg, LEVEL_2, "\treading arg: %d -> %+F\n", nr, irn));
1518 stack_layout->sp_relative = call->flags.bits.try_omit_fp;
1519 bet_type = call->cb->get_between_type(irg);
1520 stack_frame_init(stack_layout, arg_type, bet_type,
1521 get_irg_frame_type(irg), param_map);
1523 /* Count the register params and add them to the number of Projs for the RegParams node */
1524 for (i = 0; i < n_params; ++i) {
1525 be_abi_call_arg_t *arg = get_call_arg(call, 0, i, 1);
1526 if (arg->in_reg && args[i]) {
1527 assert(arg->reg != sp && "cannot use stack pointer as parameter register");
1528 assert(i == get_Proj_proj(args[i]));
1530 /* For now, associate the register with the old Proj from Start representing that argument. */
1531 pmap_insert(env->regs, (void *) arg->reg, args[i]);
1532 DBG((dbg, LEVEL_2, "\targ #%d -> reg %s\n", i, arg->reg->name));
1536 /* Collect all callee-save registers */
1537 for (i = 0, n = arch_env->n_register_classes; i < n; ++i) {
1538 const arch_register_class_t *cls = &arch_env->register_classes[i];
1539 for (j = 0; j < cls->n_regs; ++j) {
1540 const arch_register_t *reg = &cls->regs[j];
1541 if ((reg->type & arch_register_type_state) || arch_register_is_callee_save(arch_env, reg)) {
1542 pmap_insert(env->regs, (void *) reg, NULL);
1547 fp_reg = call->flags.bits.try_omit_fp ? arch_env->sp : arch_env->bp;
1548 rbitset_clear(birg->allocatable_regs, fp_reg->global_index);
1550 /* handle start block here (place a jump in the block) */
1551 fix_start_block(irg);
1553 pmap_insert(env->regs, (void *) sp, NULL);
1554 pmap_insert(env->regs, (void *) arch_env->bp, NULL);
1555 start_bl = get_irg_start_block(irg);
1556 env->start = be_new_Start(NULL, start_bl, pmap_count(env->regs) + 1);
1557 set_irg_start(irg, env->start);
1560 * make proj nodes for the callee save registers.
1561 * memorize them, since Return nodes get those as inputs.
1563 * Note, that if a register corresponds to an argument, the regs map
1564 * contains the old Proj from start for that argument.
1566 rm = ALLOCAN(reg_node_map_t, pmap_count(env->regs));
1567 reg_map_to_arr(rm, env->regs);
1568 for (i = 0, n = pmap_count(env->regs); i < n; ++i) {
1569 const arch_register_t *reg = rm[i].reg;
1570 ir_mode *mode = reg->reg_class->mode;
1572 arch_register_req_type_t add_type = arch_register_req_type_none;
1576 add_type |= arch_register_req_type_produces_sp;
1577 if (!rbitset_is_set(birg->allocatable_regs, reg->global_index)) {
1578 add_type |= arch_register_req_type_ignore;
1582 proj = new_r_Proj(env->start, mode, nr + 1);
1583 pmap_insert(env->regs, (void *) reg, proj);
1584 be_set_constr_single_reg_out(env->start, nr + 1, reg, add_type);
1585 arch_set_irn_register(proj, reg);
1587 DBG((dbg, LEVEL_2, "\tregister save proj #%d -> reg %s\n", nr, reg->name));
1590 /* create a new initial memory proj */
1591 assert(is_Proj(old_mem));
1592 arch_set_irn_register_req_out(env->start, 0, arch_no_register_req);
1593 new_mem_proj = new_r_Proj(env->start, mode_M, 0);
1595 set_irg_initial_mem(irg, mem);
1597 env->init_sp = be_abi_reg_map_get(env->regs, sp);
1599 /* set new frame_pointer */
1600 frame_pointer = be_abi_reg_map_get(env->regs, fp_reg);
1601 set_irg_frame(irg, frame_pointer);
1603 /* rewire old mem users to new mem */
1604 exchange(old_mem, mem);
1606 /* keep the mem (for functions with an endless loop = no return) */
1609 set_irg_initial_mem(irg, mem);
1611 /* Now, introduce stack param nodes for all parameters passed on the stack */
1612 for (i = 0; i < n_params; ++i) {
1613 ir_node *arg_proj = args[i];
1614 ir_node *repl = NULL;
1616 if (arg_proj != NULL) {
1617 be_abi_call_arg_t *arg;
1618 ir_type *param_type;
1619 int nr = get_Proj_proj(arg_proj);
1622 nr = MIN(nr, n_params);
1623 arg = get_call_arg(call, 0, nr, 1);
1624 param_type = get_method_param_type(method_type, nr);
1627 repl = (ir_node*)pmap_get(env->regs, arg->reg);
1628 } else if (arg->on_stack) {
1629 ir_node *addr = be_new_FrameAddr(sp->reg_class, start_bl, frame_pointer, arg->stack_ent);
1631 /* For atomic parameters which are actually used, we create a Load node. */
1632 if (is_atomic_type(param_type) && get_irn_n_edges(args[i]) > 0) {
1633 ir_mode *mode = get_type_mode(param_type);
1634 ir_mode *load_mode = arg->load_mode;
1635 ir_node *nomem = get_irg_no_mem(irg);
1637 ir_node *load = new_r_Load(start_bl, nomem, addr, load_mode, cons_floats);
1638 repl = new_r_Proj(load, load_mode, pn_Load_res);
1640 if (mode != load_mode) {
1641 repl = new_r_Conv(start_bl, repl, mode);
1644 /* The stack parameter is not primitive (it is a struct or array),
1645 * we thus will create a node representing the parameter's address
1651 assert(repl != NULL);
1653 /* Beware: the mode of the register parameters is always the mode of the register class
1654 which may be wrong. Add Conv's then. */
1655 mode = get_irn_mode(args[i]);
1656 if (mode != get_irn_mode(repl)) {
1657 repl = new_r_Conv(get_nodes_block(repl), repl, mode);
1659 exchange(args[i], repl);
1663 /* the arg proj is not needed anymore now and should be only used by the anchor */
1664 assert(get_irn_n_edges(arg_tuple) == 1);
1665 kill_node(arg_tuple);
1666 set_irg_args(irg, new_r_Bad(irg, mode_T));
1668 /* All Return nodes hang on the End node, so look for them there. */
1669 end = get_irg_end_block(irg);
1670 for (i = 0, n = get_Block_n_cfgpreds(end); i < n; ++i) {
1671 ir_node *irn = get_Block_cfgpred(end, i);
1673 if (is_Return(irn)) {
1674 ir_node *blk = get_nodes_block(irn);
1675 ir_node *mem = get_Return_mem(irn);
1676 ir_node *ret = create_be_return(env, irn, blk, mem, get_Return_n_ress(irn));
1681 /* if we have endless loops here, n might be <= 0. Do NOT create a be_Return then,
1682 the code is dead and will never be executed. */
1685 /** Fix the state inputs of calls that still hang on unknowns */
1686 static void fix_call_state_inputs(ir_graph *irg)
1688 be_abi_irg_t *env = be_get_irg_abi(irg);
1689 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1691 arch_register_t **stateregs = NEW_ARR_F(arch_register_t*, 0);
1693 /* Collect caller save registers */
1694 n = arch_env->n_register_classes;
1695 for (i = 0; i < n; ++i) {
1697 const arch_register_class_t *cls = &arch_env->register_classes[i];
1698 for (j = 0; j < cls->n_regs; ++j) {
1699 const arch_register_t *reg = arch_register_for_index(cls, j);
1700 if (reg->type & arch_register_type_state) {
1701 ARR_APP1(arch_register_t*, stateregs, (arch_register_t *)reg);
1706 n = ARR_LEN(env->calls);
1707 n_states = ARR_LEN(stateregs);
1708 for (i = 0; i < n; ++i) {
1710 ir_node *call = env->calls[i];
1712 arity = get_irn_arity(call);
1714 /* the state reg inputs are the last n inputs of the calls */
1715 for (s = 0; s < n_states; ++s) {
1716 int inp = arity - n_states + s;
1717 const arch_register_t *reg = stateregs[s];
1718 ir_node *regnode = be_abi_reg_map_get(env->regs, reg);
1720 set_irn_n(call, inp, regnode);
1724 DEL_ARR_F(stateregs);
1728 * Create a trampoline entity for the given method.
1730 static ir_entity *create_trampoline(be_main_env_t *be, ir_entity *method)
1732 ir_type *type = get_entity_type(method);
1733 ident *old_id = get_entity_ld_ident(method);
1734 ident *id = id_mangle3("", old_id, "$stub");
1735 ir_type *parent = be->pic_trampolines_type;
1736 ir_entity *ent = new_entity(parent, old_id, type);
1737 set_entity_ld_ident(ent, id);
1738 set_entity_visibility(ent, ir_visibility_private);
1744 * Returns the trampoline entity for the given method.
1746 static ir_entity *get_trampoline(be_main_env_t *env, ir_entity *method)
1748 ir_entity *result = (ir_entity*)pmap_get(env->ent_trampoline_map, method);
1749 if (result == NULL) {
1750 result = create_trampoline(env, method);
1751 pmap_insert(env->ent_trampoline_map, method, result);
1757 static ir_entity *create_pic_symbol(be_main_env_t *be, ir_entity *entity)
1759 ident *old_id = get_entity_ld_ident(entity);
1760 ident *id = id_mangle3("", old_id, "$non_lazy_ptr");
1761 ir_type *e_type = get_entity_type(entity);
1762 ir_type *type = new_type_pointer(e_type);
1763 ir_type *parent = be->pic_symbols_type;
1764 ir_entity *ent = new_entity(parent, old_id, type);
1765 set_entity_ld_ident(ent, id);
1766 set_entity_visibility(ent, ir_visibility_private);
1771 static ir_entity *get_pic_symbol(be_main_env_t *env, ir_entity *entity)
1773 ir_entity *result = (ir_entity*)pmap_get(env->ent_pic_symbol_map, entity);
1774 if (result == NULL) {
1775 result = create_pic_symbol(env, entity);
1776 pmap_insert(env->ent_pic_symbol_map, entity, result);
1785 * Returns non-zero if a given entity can be accessed using a relative address.
1787 static int can_address_relative(ir_entity *entity)
1789 return get_entity_visibility(entity) != ir_visibility_external
1790 && !(get_entity_linkage(entity) & IR_LINKAGE_MERGE);
1793 static ir_node *get_pic_base(ir_graph *irg)
1795 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1796 if (arch_env->impl->get_pic_base == NULL)
1798 return arch_env->impl->get_pic_base(irg);
1801 /** patches SymConsts to work in position independent code */
1802 static void fix_pic_symconsts(ir_node *node, void *data)
1804 ir_graph *irg = get_irn_irg(node);
1805 be_main_env_t *be = be_get_irg_main_env(irg);
1815 arity = get_irn_arity(node);
1816 for (i = 0; i < arity; ++i) {
1818 ir_node *pred = get_irn_n(node, i);
1820 ir_entity *pic_symbol;
1821 ir_node *pic_symconst;
1823 if (!is_SymConst(pred))
1826 entity = get_SymConst_entity(pred);
1827 block = get_nodes_block(pred);
1829 /* calls can jump to relative addresses, so we can directly jump to
1830 the (relatively) known call address or the trampoline */
1831 if (i == 1 && is_Call(node)) {
1832 ir_entity *trampoline;
1833 ir_node *trampoline_const;
1835 if (can_address_relative(entity))
1838 dbgi = get_irn_dbg_info(pred);
1839 trampoline = get_trampoline(be, entity);
1840 trampoline_const = new_rd_SymConst_addr_ent(dbgi, irg, mode_P_code,
1842 set_irn_n(node, i, trampoline_const);
1846 /* everything else is accessed relative to EIP */
1847 mode = get_irn_mode(pred);
1848 pic_base = get_pic_base(irg);
1850 /* all ok now for locally constructed stuff */
1851 if (can_address_relative(entity)) {
1852 ir_node *add = new_r_Add(block, pic_base, pred, mode);
1854 /* make sure the walker doesn't visit this add again */
1855 mark_irn_visited(add);
1856 set_irn_n(node, i, add);
1860 /* get entry from pic symbol segment */
1861 dbgi = get_irn_dbg_info(pred);
1862 pic_symbol = get_pic_symbol(be, entity);
1863 pic_symconst = new_rd_SymConst_addr_ent(dbgi, irg, mode_P_code,
1865 add = new_r_Add(block, pic_base, pic_symconst, mode);
1866 mark_irn_visited(add);
1868 /* we need an extra indirection for global data outside our current
1869 module. The loads are always safe and can therefore float
1870 and need no memory input */
1871 load = new_r_Load(block, get_irg_no_mem(irg), add, mode, cons_floats);
1872 load_res = new_r_Proj(load, mode, pn_Load_res);
1874 set_irn_n(node, i, load_res);
1878 void be_abi_introduce(ir_graph *irg)
1880 be_abi_irg_t *env = XMALLOCZ(be_abi_irg_t);
1881 ir_node *old_frame = get_irg_frame(irg);
1882 be_options_t *options = be_get_irg_options(irg);
1883 const arch_env_t *arch_env = be_get_irg_arch_env(irg);
1884 ir_entity *entity = get_irg_entity(irg);
1885 ir_type *method_type = get_entity_type(entity);
1886 be_irg_t *birg = be_birg_from_irg(irg);
1887 struct obstack *obst = &birg->obst;
1888 ir_node *dummy = new_r_Dummy(irg,
1889 arch_env->sp->reg_class->mode);
1892 /* determine allocatable registers */
1893 assert(birg->allocatable_regs == NULL);
1894 birg->allocatable_regs = rbitset_obstack_alloc(obst, arch_env->n_registers);
1895 for (r = 0; r < arch_env->n_registers; ++r) {
1896 const arch_register_t *reg = &arch_env->registers[r];
1897 if ( !(reg->type & arch_register_type_ignore)) {
1898 rbitset_set(birg->allocatable_regs, r);
1902 /* break here if backend provides a custom API.
1903 * Note: we shouldn't have to setup any be_abi_irg_t* stuff at all,
1904 * but need more cleanup to make this work
1906 be_set_irg_abi(irg, env);
1908 be_omit_fp = options->omit_fp;
1910 env->keep_map = pmap_create();
1911 env->call = be_abi_call_new(arch_env->sp->reg_class);
1912 arch_env_get_call_abi(arch_env, method_type, env->call);
1914 env->init_sp = dummy;
1915 env->calls = NEW_ARR_F(ir_node*, 0);
1920 irg_walk_graph(irg, fix_pic_symconsts, NULL, env);
1923 /* Lower all call nodes in the IRG. */
1926 /* Process the IRG */
1929 /* fix call inputs for state registers */
1930 fix_call_state_inputs(irg);
1932 /* We don't need the keep map anymore. */
1933 pmap_destroy(env->keep_map);
1934 env->keep_map = NULL;
1936 /* calls array is not needed anymore */
1937 DEL_ARR_F(env->calls);
1940 /* reroute the stack origin of the calls to the true stack origin. */
1941 exchange(dummy, env->init_sp);
1942 exchange(old_frame, get_irg_frame(irg));
1944 pmap_destroy(env->regs);
1948 void be_abi_free(ir_graph *irg)
1950 be_abi_irg_t *env = be_get_irg_abi(irg);
1952 if (env->call != NULL)
1953 be_abi_call_free(env->call);
1954 assert(env->regs == NULL);
1957 be_set_irg_abi(irg, NULL);
1960 void be_put_allocatable_regs(const ir_graph *irg,
1961 const arch_register_class_t *cls, bitset_t *bs)
1963 be_irg_t *birg = be_birg_from_irg(irg);
1964 unsigned *allocatable_regs = birg->allocatable_regs;
1967 assert(bitset_size(bs) == cls->n_regs);
1968 bitset_clear_all(bs);
1969 for (i = 0; i < cls->n_regs; ++i) {
1970 const arch_register_t *reg = &cls->regs[i];
1971 if (rbitset_is_set(allocatable_regs, reg->global_index))
1976 unsigned be_get_n_allocatable_regs(const ir_graph *irg,
1977 const arch_register_class_t *cls)
1979 bitset_t *bs = bitset_alloca(cls->n_regs);
1980 be_put_allocatable_regs(irg, cls, bs);
1981 return bitset_popcount(bs);
1984 void be_set_allocatable_regs(const ir_graph *irg,
1985 const arch_register_class_t *cls,
1986 unsigned *raw_bitset)
1988 be_irg_t *birg = be_birg_from_irg(irg);
1989 unsigned *allocatable_regs = birg->allocatable_regs;
1992 rbitset_clear_all(raw_bitset, cls->n_regs);
1993 for (i = 0; i < cls->n_regs; ++i) {
1994 const arch_register_t *reg = &cls->regs[i];
1995 if (rbitset_is_set(allocatable_regs, reg->global_index))
1996 rbitset_set(raw_bitset, i);
2000 BE_REGISTER_MODULE_CONSTRUCTOR(be_init_abi)
2001 void be_init_abi(void)
2003 FIRM_DBG_REGISTER(dbg, "firm.be.abi");