summaryrefslogtreecommitdiffhomepage
path: root/ir/be/ia32/ia32_intrinsics.c
blob: 22aa7050b37c758543280ba5002dd42fbd0fb97b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
/*
 * This file is part of libFirm.
 * Copyright (C) 2012 University of Karlsruhe.
 */

/**
 * @file
 * @brief       This file implements the mapping of 64Bit intrinsic
 *              functions to code or library calls.
 * @author      Michael Beck
 */
#include "array.h"
#include "begnuas.h"
#include "constbits.h"
#include "gen_ia32_regalloc_if.h"
#include "ia32_bearch_t.h"
#include "ia32_new_nodes.h"
#include "ircons_t.h"
#include "iredges_t.h"
#include "irgmod.h"
#include "irnode_t.h"
#include "irop_t.h"
#include "iroptimize.h"
#include "irprog_t.h"
#include "lower_dw.h"
#include "panic.h"
#include "tv_t.h"
#include "util.h"
#include "x86_x87.h"

typedef enum {
	no_carry = 0,
	can_carry,
	must_carry
} carry_result;

static ir_tarval *bitinfo_max(bitinfo *info)
{
	ir_tarval *z    = info->z;
	ir_tarval *o    = info->o;
	ir_mode   *mode = get_tarval_mode(z);
	ir_tarval *min  = get_mode_min(mode);

	assert(get_mode_arithmetic(mode) == irma_twos_complement);
	return tarval_and(z, tarval_ornot(o, min));
}

static ir_tarval *bitinfo_min(bitinfo *info)
{
	ir_tarval *z    = info->z;
	ir_tarval *o    = info->o;
	ir_mode   *mode = get_tarval_mode(z);
	ir_tarval *min  = get_mode_min(mode);

	assert(get_mode_arithmetic(mode) == irma_twos_complement);
	return tarval_or(o, tarval_and(z, min));
}

static carry_result lower_add_carry(ir_node *left, ir_node *right, ir_mode *mode)
{
	assert(!mode_is_signed(mode));

	bitinfo *bi_left = get_bitinfo(left);
	if (!bi_left) {
		return can_carry;
	}
	bitinfo *bi_right = get_bitinfo(right);
	// If we have bitinfo for one node, we should also have it for
	// the other
	assert(bi_right);

	ir_tarval    *lmin   = tarval_convert_to(bitinfo_min(bi_left),  mode);
	ir_tarval    *rmin   = tarval_convert_to(bitinfo_min(bi_right), mode);
	ir_tarval    *lmax   = tarval_convert_to(bitinfo_max(bi_left),  mode);
	ir_tarval    *rmax   = tarval_convert_to(bitinfo_max(bi_right), mode);
	carry_result  result = no_carry;

	int old_wrap_on_overflow = tarval_get_wrap_on_overflow();
	tarval_set_wrap_on_overflow(false);

	if (tarval_add(lmax, rmax) == tarval_bad) {
		result = can_carry;
		if (tarval_add(lmin, rmin) == tarval_bad) {
			result = must_carry;
		}
	}

	tarval_set_wrap_on_overflow(old_wrap_on_overflow);

	return result;
}

static carry_result lower_sub_borrow(ir_node *left, ir_node *right, ir_mode *mode)
{
	assert(!mode_is_signed(mode));

	bitinfo *bi_left = get_bitinfo(left);
	if (!bi_left) {
		return can_carry;
	}
	bitinfo *bi_right = get_bitinfo(right);
	// If we have bitinfo for one node, we should also have it for
	// the other
	assert(bi_right);

	ir_tarval    *lmin   = tarval_convert_to(bitinfo_min(bi_left),  mode);
	ir_tarval    *rmin   = tarval_convert_to(bitinfo_min(bi_right), mode);
	ir_tarval    *lmax   = tarval_convert_to(bitinfo_max(bi_left),  mode);
	ir_tarval    *rmax   = tarval_convert_to(bitinfo_max(bi_right), mode);
	carry_result  result = no_carry;

	int old_wrap_on_overflow = tarval_get_wrap_on_overflow();
	tarval_set_wrap_on_overflow(false);

	if (tarval_sub(lmin, rmax) == tarval_bad) {
		result = can_carry;
		if (tarval_sub(lmax, rmin) == tarval_bad) {
			result = must_carry;
		}
	}

	tarval_set_wrap_on_overflow(old_wrap_on_overflow);

	return result;
}

/**
 * lower 64bit addition: an 32bit add for the lower parts, an add with
 * carry for the higher parts. If the carry's value is known, fold it
 * into the upper add.
 */
static void ia32_lower_add64(ir_node *const node)
{
	dbg_info     *dbg        = get_irn_dbg_info(node);
	ir_node      *block      = get_nodes_block(node);
	ir_node      *left       = get_Add_left(node);
	ir_node      *right      = get_Add_right(node);
	ir_node      *left_low   = get_lowered_low(left);
	ir_node      *left_high  = get_lowered_high(left);
	ir_node      *right_low  = get_lowered_low(right);
	ir_node      *right_high = get_lowered_high(right);
	ir_mode      *low_mode   = get_irn_mode(left_low);
	ir_mode      *high_mode  = get_irn_mode(left_high);
	carry_result  cr         = lower_add_carry(left, right, low_mode);

	assert(get_irn_mode(left_low)  == get_irn_mode(right_low));
	assert(get_irn_mode(left_high) == get_irn_mode(right_high));

	if (cr == no_carry) {
		ir_node *add_low  = new_rd_Add(dbg, block, left_low,  right_low);
		ir_node *add_high = new_rd_Add(dbg, block, left_high, right_high);
		ir_set_dw_lowered(node, add_low, add_high);
	} else if (cr == must_carry && (is_Const(left_high) || is_Const(right_high))) {
		// We cannot assume that left_high and right_high form a normalized Add.
		ir_node *constant;
		ir_node *other;

		if (is_Const(left_high)) {
			constant = left_high;
			other    = right_high;
		} else {
			constant = right_high;
			other    = left_high;
		}

		ir_graph *irg            = get_irn_irg(right_high);
		ir_node  *one            = new_rd_Const(dbg, irg, get_mode_one(high_mode));
		ir_node  *const_plus_one = new_rd_Add(dbg, block, constant, one);
		ir_node  *add_high       = new_rd_Add(dbg, block, other, const_plus_one);
		ir_node  *add_low        = new_rd_Add(dbg, block, left_low, right_low);
		ir_set_dw_lowered(node, add_low, add_high);
	} else {
		/* l_res = a_l + b_l */
		ir_node  *add_low    = new_bd_ia32_l_Add(dbg, block, left_low, right_low);
		ir_mode  *mode_flags = ia32_reg_classes[CLASS_ia32_flags].mode;
		ir_node  *res_low    = new_r_Proj(add_low, ia32_mode_gp, pn_ia32_l_Add_res);
		ir_node  *flags      = new_r_Proj(add_low, mode_flags, pn_ia32_l_Add_flags);

		/* h_res = a_h + b_h + carry */
		ir_mode  *mode     = get_node_high_mode(node);
		ir_node  *add_high
			= new_bd_ia32_l_Adc(dbg, block, left_high, right_high, flags, mode);
		ir_set_dw_lowered(node, res_low, add_high);
	}
}

/**
 * lower 64bit subtraction: a 32bit sub for the lower parts, a sub
 * with borrow for the higher parts. If the borrow's value is known,
 * fold it into the upper sub.
 */
static void ia32_lower_sub64(ir_node *const node)
{
	dbg_info     *dbg        = get_irn_dbg_info(node);
	ir_node      *block      = get_nodes_block(node);
	ir_node      *left       = get_Sub_left(node);
	ir_node      *right      = get_Sub_right(node);
	ir_node      *left_low   = get_lowered_low(left);
	ir_node      *left_high  = get_lowered_high(left);
	ir_node      *right_low  = get_lowered_low(right);
	ir_node      *right_high = get_lowered_high(right);
	ir_mode      *low_mode   = get_irn_mode(left_low);
	carry_result  cr         = lower_sub_borrow(left, right, low_mode);

	assert(get_irn_mode(left_low)  == get_irn_mode(right_low));
	assert(get_irn_mode(left_high) == get_irn_mode(right_high));

	if (cr == no_carry) {
		ir_node *sub_low  = new_rd_Sub(dbg, block, left_low,  right_low);
		ir_node *sub_high = new_rd_Sub(dbg, block, left_high, right_high);
		ir_set_dw_lowered(node, sub_low, sub_high);
	} else if (cr == must_carry && (is_Const(left_high) || is_Const(right_high))) {
		ir_node  *sub_high;
		ir_graph *irg        = get_irn_irg(right_high);
		ir_mode  *high_mode  = get_irn_mode(left_high);
		ir_node  *one        = new_rd_Const(dbg, irg, get_mode_one(high_mode));

		if (is_Const(right_high)) {
			ir_node *new_const = new_rd_Add(dbg, block, right_high, one);
			sub_high = new_rd_Sub(dbg, block, left_high, new_const);
		} else if (is_Const(left_high)) {
			ir_node *new_const = new_rd_Sub(dbg, block, left_high, one);
			sub_high = new_rd_Sub(dbg, block, new_const, right_high);
		} else {
			panic("logic error");
		}

		ir_node  *sub_low  = new_rd_Sub(dbg, block, left_low, right_low);
		ir_set_dw_lowered(node, sub_low, sub_high);
	} else {
		/* l_res = a_l - b_l */
		ir_node  *sub_low    = new_bd_ia32_l_Sub(dbg, block, left_low, right_low);
		ir_mode  *mode_flags = ia32_reg_classes[CLASS_ia32_flags].mode;
		ir_node  *res_low    = new_r_Proj(sub_low, ia32_mode_gp, pn_ia32_l_Sub_res);
		ir_node  *flags      = new_r_Proj(sub_low, mode_flags, pn_ia32_l_Sub_flags);

		/* h_res = a_h - b_h - carry */
		ir_mode  *mode     = get_node_high_mode(node);
		ir_node  *sub_high
			= new_bd_ia32_l_Sbb(dbg, block, left_high, right_high, flags, mode);
		ir_set_dw_lowered(node, res_low, sub_high);
	}
}

/**
 * Checks whether node high is a sign extension of low.
 */
static bool is_sign_extend(ir_node *low, ir_node *high)
{
	if (is_Shrs(high)) {
		ir_node *high_r = get_Shrs_right(high);
		if (!is_Const(high_r)) return false;

		ir_tarval *shift_count = get_Const_tarval(high_r);
		if (!tarval_is_long(shift_count))       return false;
		if (get_tarval_long(shift_count) != 31) return false;

		ir_node *high_l = get_Shrs_left(high);

		if (is_Conv(low)    && get_Conv_op(low)    == high_l) return true;
		if (is_Conv(high_l) && get_Conv_op(high_l) == low)    return true;
	} else if (is_Const(low) && is_Const(high)) {
		ir_tarval *tl = get_Const_tarval(low);
		ir_tarval *th = get_Const_tarval(high);

		ir_mode *const tl_mode            = get_tarval_mode(tl);
		unsigned const tl_size            = get_mode_size_bits(tl_mode);
		bool     const tl_signed_negative = tarval_get_bit(tl, tl_size - 1);
		return tl_signed_negative ? tarval_is_all_one(th) : tarval_is_null(th);
	}

	return false;
}

/**
 * lower 64bit Mul operation.
 */
static void ia32_lower_mul64(ir_node *const node)
{
	dbg_info *const dbg        = get_irn_dbg_info(node);
	ir_node  *const block      = get_nodes_block(node);
	ir_node  *const left       = get_Mul_left(node);
	ir_node  *const right      = get_Mul_right(node);
	ir_node  *const left_low   = get_lowered_low(left);
	ir_node  *const left_high  = get_lowered_high(left);
	ir_node  *const right_low  = get_lowered_low(right);
	ir_node  *const right_high = get_lowered_high(right);

	/* h_res  = left_high * right_low
	 * h_res += left_low  * right_high
	 * hi:lo  = left_low  * right_low # 32x32 -> 64
	 * h_res += hi
	 * l_res  = lo */

	/* handle the often used case of 32x32=64 mul */
	ir_node *h_res;
	ir_node *l_res;
	ir_mode *mode = get_node_high_mode(node);
	if (is_sign_extend(left_low, left_high)
	    && is_sign_extend(right_low, right_high)) {
		ir_node *mul = new_bd_ia32_l_IMul(dbg, block, left_low, right_low);
		h_res = new_rd_Proj(dbg, mul, mode, pn_ia32_l_IMul_res_high);
		l_res = new_rd_Proj(dbg, mul, ia32_mode_gp, pn_ia32_l_IMul_res_low);
	} else {
		/* note that zero extension is handled here efficiently */
		ir_node *const right_lowc = new_rd_Conv(dbg, block, right_low, mode);
		ir_node *const lh_rl      = new_rd_Mul(dbg, block, left_high, right_lowc);
		h_res = lh_rl;

		ir_node *const left_lowc = new_rd_Conv(dbg, block, left_low, mode);
		ir_node *const ll_rh     = new_rd_Mul(dbg, block, left_lowc, right_high);
		h_res = new_rd_Add(dbg, block, h_res, ll_rh);

		if (is_irn_null(left_low) || is_irn_one(right_low)) {
			/* 0 * right_low, left_low * 1 -> left_low */
			l_res = left_low;
		} else if (is_irn_null(right_low) || is_irn_one(left_low)) {
			/* left_low * 0, 1 * right_low -> right_low */
			l_res = right_low;
		} else {
			ir_node *const ll_rl = new_bd_ia32_l_Mul(dbg, block, left_low, right_low);
			ir_node *const hi    = new_rd_Proj(dbg, ll_rl, mode,         pn_ia32_l_Mul_res_high);
			ir_node *const lo    = new_rd_Proj(dbg, ll_rl, ia32_mode_gp, pn_ia32_l_Mul_res_low);
			h_res = new_rd_Add(dbg, block, h_res, hi);
			l_res = lo;
		}
	}
	ir_set_dw_lowered(node, l_res, h_res);
}

/**
 * lower 64bit minus operation
 */
static void ia32_lower_minus64(ir_node *const node)
{
	dbg_info *dbg     = get_irn_dbg_info(node);
	ir_node  *block   = get_nodes_block(node);
	ir_node  *op      = get_Minus_op(node);
	ir_node  *op_low  = get_lowered_low(op);
	ir_node  *op_high = get_lowered_high(op);
	ir_node  *minus   = new_bd_ia32_l_Minus64(dbg, block, op_low, op_high);
	ir_node  *l_res   = new_r_Proj(minus, ia32_mode_gp, pn_ia32_Minus64_res_low);
	ir_mode  *mode    = get_node_high_mode(node);
	ir_node  *h_res   = new_r_Proj(minus, mode, pn_ia32_Minus64_res_high);
	ir_set_dw_lowered(node, l_res, h_res);
}

/**
 * lower 64bit conversions
 */
static void ia32_lower_conv64(ir_node *const node)
{
	dbg_info  *dbg       = get_irn_dbg_info(node);
	ir_node   *op        = get_Conv_op(node);
	ir_mode   *mode_from = get_irn_mode(op);
	ir_mode   *mode_to   = get_irn_mode(node);

	if (mode_is_float(mode_from) && get_mode_size_bits(mode_to) == 64
	    && get_mode_arithmetic(mode_to) == irma_twos_complement) {
		/* We have a Conv float -> long long here */
		ir_node *float_to_ll;
		ir_node *l_res;
		ir_node *h_res;
		ir_mode *mode = get_node_high_mode(node);
		if (mode_is_signed(mode)) {
			/* convert from float to signed 64bit */
			ir_node *block = get_nodes_block(node);
			float_to_ll = new_bd_ia32_l_FloattoLL(dbg, block, op);
			l_res = new_r_Proj(float_to_ll, ia32_mode_gp,
			                   pn_ia32_l_FloattoLL_res_low);
			h_res = new_r_Proj(float_to_ll, mode,
			                   pn_ia32_l_FloattoLL_res_high);
		} else {
			/* Convert from float to unsigned 64bit. */
			ir_graph  *irg = get_irn_irg(node);
			ir_tarval *flt_tv
				= new_tarval_from_str("9223372036854775808", 19, x86_mode_E);
			ir_node   *flt_corr  = new_r_Const(irg, flt_tv);

			ir_node *lower_blk = part_block_dw(node);
			ir_node *upper_blk = get_nodes_block(node);
			set_dw_control_flow_changed();

			ir_node *opc  = new_rd_Conv(dbg, upper_blk, op, x86_mode_E);
			ir_node *cmp  = new_rd_Cmp(dbg, upper_blk, opc, flt_corr,
			                           ir_relation_less);
			ir_node *cond = new_rd_Cond(dbg, upper_blk, cmp);
			ir_node *in[] = {
				new_r_Proj(cond, mode_X, pn_Cond_true),
				new_r_Proj(cond, mode_X, pn_Cond_false)
			};
			ir_node *blk   = new_r_Block(irg, 1, &in[1]);
			in[1] = new_r_Jmp(blk);

			set_irn_in(lower_blk, 2, in);

			/* create to Phis */
			ir_node *phi_in[] = {
				new_r_Const_null(irg, mode),
				new_r_Const_long(irg, mode, 0x80000000)
			};
			ir_node *int_phi
				= new_r_Phi(lower_blk, ARRAY_SIZE(phi_in), phi_in, mode);

			ir_node *fphi_in[] = {
				opc,
				new_rd_Sub(dbg, upper_blk, opc, flt_corr)
			};
			ir_node *flt_phi
				= new_r_Phi(lower_blk, ARRAY_SIZE(fphi_in), fphi_in,
				            x86_mode_E);

			/* fix Phi links for next part_block() */
			if (is_Phi(int_phi))
				add_Block_phi(lower_blk, int_phi);
			if (is_Phi(flt_phi))
				add_Block_phi(lower_blk, flt_phi);

			float_to_ll = new_bd_ia32_l_FloattoLL(dbg, lower_blk, flt_phi);
			l_res = new_r_Proj(float_to_ll, ia32_mode_gp,
			                   pn_ia32_l_FloattoLL_res_low);
			h_res = new_r_Proj(float_to_ll, mode,
			                   pn_ia32_l_FloattoLL_res_high);
			h_res = new_rd_Add(dbg, lower_blk, h_res, int_phi);

			/* move the call and its Proj's to the lower block */
			set_nodes_block(node, lower_blk);
			for (ir_node *proj = (ir_node*)get_irn_link(node); proj != NULL;
			     proj = (ir_node*)get_irn_link(proj)) {
				set_nodes_block(proj, lower_blk);
			}
		}
		ir_set_dw_lowered(node, l_res, h_res);
	} else if (get_mode_size_bits(mode_from) == 64
	           && get_mode_arithmetic(mode_from) == irma_twos_complement
	           && mode_is_float(mode_to)) {
		/* We have a Conv long long -> float here */
		ir_node *op_low  = get_lowered_low(op);
		ir_node *op_high = get_lowered_high(op);
		ir_node *block   = get_nodes_block(node);
		ir_node *ll_to_float
			= new_bd_ia32_l_LLtoFloat(dbg, block, op_high, op_low, mode_to);

		exchange(node, ll_to_float);
	} else {
		ir_default_lower_dw_Conv(node);
	}
}

static void ia32_lower_ASM(ir_node *const asmn)
{
	size_t             const pos_none      = (size_t)-1;
	size_t                   pos_out_cons  = pos_none;
	size_t             const n_constraints = get_ASM_n_constraints(asmn);
	ir_asm_constraint *const constraints   = get_ASM_constraints(asmn);
	for (size_t i = 0; i < n_constraints; ++i) {
		ir_asm_constraint const *const constraint = &constraints[i];
		if (needs_lowering(constraint->mode)) {
			char const *const constr = get_id_str(constraint->constraint);
			if (streq(constr, "=A")) {
				pos_out_cons = i;
			} else {
				panic("lowering ASM 64bit operands only supports '=A' currently");
			}
		}
	}

	if (pos_out_cons == pos_none)
		return;

	size_t             const new_n_constraints = n_constraints + 1;
	ir_asm_constraint *const new_constraints   = ALLOCAN(ir_asm_constraint, new_n_constraints);
	int                const pos_out           = constraints[pos_out_cons].out_pos;
	for (size_t i = 0; i != n_constraints; ++i) {
		ir_asm_constraint *const c = &new_constraints[i];
		*c = constraints[i];
		if (c->out_pos > pos_out)
			c->out_pos += 1;
	}

	ir_mode *const lo_mode = ir_get_low_unsigned_mode();

	ir_asm_constraint *const lo = &new_constraints[pos_out_cons];
	ir_asm_constraint *const hi = &new_constraints[n_constraints];
	hi->in_pos     = -1;
	hi->out_pos    = lo->out_pos + 1;
	hi->constraint = new_id_from_str("=d");
	hi->mode       = get_high_mode(lo->mode);

	lo->constraint = new_id_from_str("=a");
	lo->mode       = lo_mode;

	ir_cons_flags flags = cons_none;
	if (get_irn_pinned(asmn) == op_pin_state_floats)
		flags |= cons_floats;
	if (ir_throws_exception(asmn))
		flags |= cons_throws_exception;

	dbg_info *const dbgi      = get_irn_dbg_info(asmn);
	ir_node  *const block     = get_nodes_block(asmn);
	ir_node  *const mem       = get_ASM_mem(asmn);
	int       const n_ins     = get_ASM_n_inputs(asmn);
	ir_node **const ins       = get_irn_in(asmn) + (n_ASM_max + 1);
	size_t    const n_clobber = get_ASM_n_clobbers(asmn);
	ident   **const clobbers  = get_ASM_clobbers(asmn);
	ident    *const asm_text  = get_ASM_text(asmn);
	ir_node  *const new_asm   = new_rd_ASM(dbgi, block, mem, n_ins, ins, asm_text, new_n_constraints, new_constraints, n_clobber, clobbers, flags);

	foreach_out_edge_safe(asmn, edge) {
		ir_node *const proj = get_edge_src_irn(edge);
		if (is_Proj(proj)) {
			ir_mode *const proj_mode = get_irn_mode(proj);
			unsigned       pn        = get_Proj_num(proj);
			if (pn == (unsigned)pos_out) {
				ir_mode *const hi_mode = get_high_mode(proj_mode);
				ir_node *const np_low  = new_r_Proj(new_asm, lo_mode, pn);
				ir_node *const np_high = new_r_Proj(new_asm, hi_mode, pn + 1);
				ir_set_dw_lowered(proj, np_low, np_high);
			} else {
				if (pn > (unsigned)pos_out)
					pn += 1;
				ir_node *const np = new_r_Proj(new_asm, proj_mode, pn);
				exchange(proj, np);
			}
		}
	}
}

void ia32_lower64(void)
{
	/* perform doubleword lowering */
	ir_mode *word_unsigned = ia32_reg_classes[CLASS_ia32_gp].mode;
	ir_mode *word_signed   = find_signed_mode(word_unsigned);
	lwrdw_param_t lower_dw_params = {
		word_unsigned,
		word_signed,
		64    /* doubleword size */
	};

	ir_prepare_dw_lowering(&lower_dw_params);
	ir_register_dw_lower_function(op_ASM,   ia32_lower_ASM);
	ir_register_dw_lower_function(op_Add,   ia32_lower_add64);
	ir_register_dw_lower_function(op_Sub,   ia32_lower_sub64);
	ir_register_dw_lower_function(op_Mul,   ia32_lower_mul64);
	ir_register_dw_lower_function(op_Minus, ia32_lower_minus64);
	ir_register_dw_lower_function(op_Conv,  ia32_lower_conv64);
	ir_lower_dw_ops();
}