SystemZInstrFP.td
26.2 KB
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
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
//==- SystemZInstrFP.td - Floating-point SystemZ instructions --*- tblgen-*-==//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// TODO: Most floating-point instructions (except for simple moves and the
// like) can raise exceptions -- should they have hasSideEffects=1 ?
//===----------------------------------------------------------------------===//
// Select instructions
//===----------------------------------------------------------------------===//
// C's ?: operator for floating-point operands.
let Predicates = [FeatureVector] in {
def SelectVR32 : SelectWrapper<f32, VR32>;
def SelectVR64 : SelectWrapper<f64, VR64>;
}
def SelectF32 : SelectWrapper<f32, FP32>;
def SelectF64 : SelectWrapper<f64, FP64>;
let Predicates = [FeatureNoVectorEnhancements1] in
def SelectF128 : SelectWrapper<f128, FP128>;
let Predicates = [FeatureVectorEnhancements1] in
def SelectVR128 : SelectWrapper<f128, VR128>;
defm CondStoreF32 : CondStores<FP32, simple_store,
simple_load, bdxaddr20only>;
defm CondStoreF64 : CondStores<FP64, simple_store,
simple_load, bdxaddr20only>;
//===----------------------------------------------------------------------===//
// Move instructions
//===----------------------------------------------------------------------===//
// Load zero.
let isAsCheapAsAMove = 1, isMoveImm = 1 in {
def LZER : InherentRRE<"lzer", 0xB374, FP32, fpimm0>;
def LZDR : InherentRRE<"lzdr", 0xB375, FP64, fpimm0>;
def LZXR : InherentRRE<"lzxr", 0xB376, FP128, fpimm0>;
}
// Moves between two floating-point registers.
def LER : UnaryRR <"ler", 0x38, null_frag, FP32, FP32>;
def LDR : UnaryRR <"ldr", 0x28, null_frag, FP64, FP64>;
def LXR : UnaryRRE<"lxr", 0xB365, null_frag, FP128, FP128>;
// For z13 we prefer LDR over LER to avoid partial register dependencies.
let isCodeGenOnly = 1 in
def LDR32 : UnaryRR<"ldr", 0x28, null_frag, FP32, FP32>;
// Moves between two floating-point registers that also set the condition
// codes.
let Uses = [FPC], mayRaiseFPException = 1,
Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
defm LTEBR : LoadAndTestRRE<"ltebr", 0xB302, FP32>;
defm LTDBR : LoadAndTestRRE<"ltdbr", 0xB312, FP64>;
defm LTXBR : LoadAndTestRRE<"ltxbr", 0xB342, FP128>;
}
// Note that LTxBRCompare is not available if we have vector support,
// since load-and-test instructions will partially clobber the target
// (vector) register.
let Predicates = [FeatureNoVector] in {
defm : CompareZeroFP<LTEBRCompare, FP32>;
defm : CompareZeroFP<LTDBRCompare, FP64>;
defm : CompareZeroFP<LTXBRCompare, FP128>;
}
// Use a normal load-and-test for compare against zero in case of
// vector support (via a pseudo to simplify instruction selection).
let Uses = [FPC], mayRaiseFPException = 1,
Defs = [CC], usesCustomInserter = 1, hasNoSchedulingInfo = 1 in {
def LTEBRCompare_VecPseudo : Pseudo<(outs), (ins FP32:$R1, FP32:$R2), []>;
def LTDBRCompare_VecPseudo : Pseudo<(outs), (ins FP64:$R1, FP64:$R2), []>;
def LTXBRCompare_VecPseudo : Pseudo<(outs), (ins FP128:$R1, FP128:$R2), []>;
}
let Predicates = [FeatureVector] in {
defm : CompareZeroFP<LTEBRCompare_VecPseudo, FP32>;
defm : CompareZeroFP<LTDBRCompare_VecPseudo, FP64>;
}
let Predicates = [FeatureVector, FeatureNoVectorEnhancements1] in
defm : CompareZeroFP<LTXBRCompare_VecPseudo, FP128>;
// Moves between 64-bit integer and floating-point registers.
def LGDR : UnaryRRE<"lgdr", 0xB3CD, bitconvert, GR64, FP64>;
def LDGR : UnaryRRE<"ldgr", 0xB3C1, bitconvert, FP64, GR64>;
// fcopysign with an FP32 result.
let isCodeGenOnly = 1 in {
def CPSDRss : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP32, FP32, FP32>;
def CPSDRsd : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP32, FP32, FP64>;
}
// The sign of an FP128 is in the high register.
let Predicates = [FeatureNoVectorEnhancements1] in
def : Pat<(fcopysign FP32:$src1, (f32 (fpround (f128 FP128:$src2)))),
(CPSDRsd FP32:$src1, (EXTRACT_SUBREG FP128:$src2, subreg_h64))>;
let Predicates = [FeatureVectorEnhancements1] in
def : Pat<(fcopysign FP32:$src1, (f32 (fpround (f128 VR128:$src2)))),
(CPSDRsd FP32:$src1, (EXTRACT_SUBREG VR128:$src2, subreg_h64))>;
// fcopysign with an FP64 result.
let isCodeGenOnly = 1 in
def CPSDRds : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP64, FP64, FP32>;
def CPSDRdd : BinaryRRFb<"cpsdr", 0xB372, fcopysign, FP64, FP64, FP64>;
// The sign of an FP128 is in the high register.
let Predicates = [FeatureNoVectorEnhancements1] in
def : Pat<(fcopysign FP64:$src1, (f64 (fpround (f128 FP128:$src2)))),
(CPSDRdd FP64:$src1, (EXTRACT_SUBREG FP128:$src2, subreg_h64))>;
let Predicates = [FeatureVectorEnhancements1] in
def : Pat<(fcopysign FP64:$src1, (f64 (fpround (f128 VR128:$src2)))),
(CPSDRdd FP64:$src1, (EXTRACT_SUBREG VR128:$src2, subreg_h64))>;
// fcopysign with an FP128 result. Use "upper" as the high half and leave
// the low half as-is.
class CopySign128<RegisterOperand cls, dag upper>
: Pat<(fcopysign FP128:$src1, cls:$src2),
(INSERT_SUBREG FP128:$src1, upper, subreg_h64)>;
let Predicates = [FeatureNoVectorEnhancements1] in {
def : CopySign128<FP32, (CPSDRds (EXTRACT_SUBREG FP128:$src1, subreg_h64),
FP32:$src2)>;
def : CopySign128<FP64, (CPSDRdd (EXTRACT_SUBREG FP128:$src1, subreg_h64),
FP64:$src2)>;
def : CopySign128<FP128, (CPSDRdd (EXTRACT_SUBREG FP128:$src1, subreg_h64),
(EXTRACT_SUBREG FP128:$src2, subreg_h64))>;
}
defm LoadStoreF32 : MVCLoadStore<load, f32, MVCSequence, 4>;
defm LoadStoreF64 : MVCLoadStore<load, f64, MVCSequence, 8>;
defm LoadStoreF128 : MVCLoadStore<load, f128, MVCSequence, 16>;
//===----------------------------------------------------------------------===//
// Load instructions
//===----------------------------------------------------------------------===//
let canFoldAsLoad = 1, SimpleBDXLoad = 1, mayLoad = 1 in {
defm LE : UnaryRXPair<"le", 0x78, 0xED64, load, FP32, 4>;
defm LD : UnaryRXPair<"ld", 0x68, 0xED65, load, FP64, 8>;
// For z13 we prefer LDE over LE to avoid partial register dependencies.
let isCodeGenOnly = 1 in
def LDE32 : UnaryRXE<"lde", 0xED24, null_frag, FP32, 4>;
// These instructions are split after register allocation, so we don't
// want a custom inserter.
let Has20BitOffset = 1, HasIndex = 1, Is128Bit = 1 in {
def LX : Pseudo<(outs FP128:$dst), (ins bdxaddr20only128:$src),
[(set FP128:$dst, (load bdxaddr20only128:$src))]>;
}
}
//===----------------------------------------------------------------------===//
// Store instructions
//===----------------------------------------------------------------------===//
let SimpleBDXStore = 1, mayStore = 1 in {
defm STE : StoreRXPair<"ste", 0x70, 0xED66, store, FP32, 4>;
defm STD : StoreRXPair<"std", 0x60, 0xED67, store, FP64, 8>;
// These instructions are split after register allocation, so we don't
// want a custom inserter.
let Has20BitOffset = 1, HasIndex = 1, Is128Bit = 1 in {
def STX : Pseudo<(outs), (ins FP128:$src, bdxaddr20only128:$dst),
[(store FP128:$src, bdxaddr20only128:$dst)]>;
}
}
//===----------------------------------------------------------------------===//
// Conversion instructions
//===----------------------------------------------------------------------===//
// Convert floating-point values to narrower representations, rounding
// according to the current mode. The destination of LEXBR and LDXBR
// is a 128-bit value, but only the first register of the pair is used.
let Uses = [FPC], mayRaiseFPException = 1 in {
def LEDBR : UnaryRRE<"ledbr", 0xB344, any_fpround, FP32, FP64>;
def LEXBR : UnaryRRE<"lexbr", 0xB346, null_frag, FP128, FP128>;
def LDXBR : UnaryRRE<"ldxbr", 0xB345, null_frag, FP128, FP128>;
def LEDBRA : TernaryRRFe<"ledbra", 0xB344, FP32, FP64>,
Requires<[FeatureFPExtension]>;
def LEXBRA : TernaryRRFe<"lexbra", 0xB346, FP128, FP128>,
Requires<[FeatureFPExtension]>;
def LDXBRA : TernaryRRFe<"ldxbra", 0xB345, FP128, FP128>,
Requires<[FeatureFPExtension]>;
}
let Predicates = [FeatureNoVectorEnhancements1] in {
def : Pat<(f32 (any_fpround FP128:$src)),
(EXTRACT_SUBREG (LEXBR FP128:$src), subreg_hh32)>;
def : Pat<(f64 (any_fpround FP128:$src)),
(EXTRACT_SUBREG (LDXBR FP128:$src), subreg_h64)>;
}
// Extend register floating-point values to wider representations.
let Uses = [FPC], mayRaiseFPException = 1 in {
def LDEBR : UnaryRRE<"ldebr", 0xB304, any_fpextend, FP64, FP32>;
def LXEBR : UnaryRRE<"lxebr", 0xB306, null_frag, FP128, FP32>;
def LXDBR : UnaryRRE<"lxdbr", 0xB305, null_frag, FP128, FP64>;
}
let Predicates = [FeatureNoVectorEnhancements1] in {
def : Pat<(f128 (any_fpextend (f32 FP32:$src))), (LXEBR FP32:$src)>;
def : Pat<(f128 (any_fpextend (f64 FP64:$src))), (LXDBR FP64:$src)>;
}
// Extend memory floating-point values to wider representations.
let Uses = [FPC], mayRaiseFPException = 1 in {
def LDEB : UnaryRXE<"ldeb", 0xED04, any_extloadf32, FP64, 4>;
def LXEB : UnaryRXE<"lxeb", 0xED06, null_frag, FP128, 4>;
def LXDB : UnaryRXE<"lxdb", 0xED05, null_frag, FP128, 8>;
}
let Predicates = [FeatureNoVectorEnhancements1] in {
def : Pat<(f128 (any_extloadf32 bdxaddr12only:$src)),
(LXEB bdxaddr12only:$src)>;
def : Pat<(f128 (any_extloadf64 bdxaddr12only:$src)),
(LXDB bdxaddr12only:$src)>;
}
// Convert a signed integer register value to a floating-point one.
let Uses = [FPC], mayRaiseFPException = 1 in {
def CEFBR : UnaryRRE<"cefbr", 0xB394, any_sint_to_fp, FP32, GR32>;
def CDFBR : UnaryRRE<"cdfbr", 0xB395, any_sint_to_fp, FP64, GR32>;
def CXFBR : UnaryRRE<"cxfbr", 0xB396, any_sint_to_fp, FP128, GR32>;
def CEGBR : UnaryRRE<"cegbr", 0xB3A4, any_sint_to_fp, FP32, GR64>;
def CDGBR : UnaryRRE<"cdgbr", 0xB3A5, any_sint_to_fp, FP64, GR64>;
def CXGBR : UnaryRRE<"cxgbr", 0xB3A6, any_sint_to_fp, FP128, GR64>;
}
// The FP extension feature provides versions of the above that allow
// specifying rounding mode and inexact-exception suppression flags.
let Uses = [FPC], mayRaiseFPException = 1, Predicates = [FeatureFPExtension] in {
def CEFBRA : TernaryRRFe<"cefbra", 0xB394, FP32, GR32>;
def CDFBRA : TernaryRRFe<"cdfbra", 0xB395, FP64, GR32>;
def CXFBRA : TernaryRRFe<"cxfbra", 0xB396, FP128, GR32>;
def CEGBRA : TernaryRRFe<"cegbra", 0xB3A4, FP32, GR64>;
def CDGBRA : TernaryRRFe<"cdgbra", 0xB3A5, FP64, GR64>;
def CXGBRA : TernaryRRFe<"cxgbra", 0xB3A6, FP128, GR64>;
}
// Convert am unsigned integer register value to a floating-point one.
let Predicates = [FeatureFPExtension] in {
let Uses = [FPC], mayRaiseFPException = 1 in {
def CELFBR : TernaryRRFe<"celfbr", 0xB390, FP32, GR32>;
def CDLFBR : TernaryRRFe<"cdlfbr", 0xB391, FP64, GR32>;
def CXLFBR : TernaryRRFe<"cxlfbr", 0xB392, FP128, GR32>;
def CELGBR : TernaryRRFe<"celgbr", 0xB3A0, FP32, GR64>;
def CDLGBR : TernaryRRFe<"cdlgbr", 0xB3A1, FP64, GR64>;
def CXLGBR : TernaryRRFe<"cxlgbr", 0xB3A2, FP128, GR64>;
}
def : Pat<(f32 (any_uint_to_fp GR32:$src)), (CELFBR 0, GR32:$src, 0)>;
def : Pat<(f64 (any_uint_to_fp GR32:$src)), (CDLFBR 0, GR32:$src, 0)>;
def : Pat<(f128 (any_uint_to_fp GR32:$src)), (CXLFBR 0, GR32:$src, 0)>;
def : Pat<(f32 (any_uint_to_fp GR64:$src)), (CELGBR 0, GR64:$src, 0)>;
def : Pat<(f64 (any_uint_to_fp GR64:$src)), (CDLGBR 0, GR64:$src, 0)>;
def : Pat<(f128 (any_uint_to_fp GR64:$src)), (CXLGBR 0, GR64:$src, 0)>;
}
// Convert a floating-point register value to a signed integer value,
// with the second operand (modifier M3) specifying the rounding mode.
let Uses = [FPC], mayRaiseFPException = 1, Defs = [CC] in {
def CFEBR : BinaryRRFe<"cfebr", 0xB398, GR32, FP32>;
def CFDBR : BinaryRRFe<"cfdbr", 0xB399, GR32, FP64>;
def CFXBR : BinaryRRFe<"cfxbr", 0xB39A, GR32, FP128>;
def CGEBR : BinaryRRFe<"cgebr", 0xB3A8, GR64, FP32>;
def CGDBR : BinaryRRFe<"cgdbr", 0xB3A9, GR64, FP64>;
def CGXBR : BinaryRRFe<"cgxbr", 0xB3AA, GR64, FP128>;
}
// fp_to_sint always rounds towards zero, which is modifier value 5.
def : Pat<(i32 (any_fp_to_sint FP32:$src)), (CFEBR 5, FP32:$src)>;
def : Pat<(i32 (any_fp_to_sint FP64:$src)), (CFDBR 5, FP64:$src)>;
def : Pat<(i32 (any_fp_to_sint FP128:$src)), (CFXBR 5, FP128:$src)>;
def : Pat<(i64 (any_fp_to_sint FP32:$src)), (CGEBR 5, FP32:$src)>;
def : Pat<(i64 (any_fp_to_sint FP64:$src)), (CGDBR 5, FP64:$src)>;
def : Pat<(i64 (any_fp_to_sint FP128:$src)), (CGXBR 5, FP128:$src)>;
// The FP extension feature provides versions of the above that allow
// also specifying the inexact-exception suppression flag.
let Uses = [FPC], mayRaiseFPException = 1,
Predicates = [FeatureFPExtension], Defs = [CC] in {
def CFEBRA : TernaryRRFe<"cfebra", 0xB398, GR32, FP32>;
def CFDBRA : TernaryRRFe<"cfdbra", 0xB399, GR32, FP64>;
def CFXBRA : TernaryRRFe<"cfxbra", 0xB39A, GR32, FP128>;
def CGEBRA : TernaryRRFe<"cgebra", 0xB3A8, GR64, FP32>;
def CGDBRA : TernaryRRFe<"cgdbra", 0xB3A9, GR64, FP64>;
def CGXBRA : TernaryRRFe<"cgxbra", 0xB3AA, GR64, FP128>;
}
// Convert a floating-point register value to an unsigned integer value.
let Predicates = [FeatureFPExtension] in {
let Uses = [FPC], mayRaiseFPException = 1, Defs = [CC] in {
def CLFEBR : TernaryRRFe<"clfebr", 0xB39C, GR32, FP32>;
def CLFDBR : TernaryRRFe<"clfdbr", 0xB39D, GR32, FP64>;
def CLFXBR : TernaryRRFe<"clfxbr", 0xB39E, GR32, FP128>;
def CLGEBR : TernaryRRFe<"clgebr", 0xB3AC, GR64, FP32>;
def CLGDBR : TernaryRRFe<"clgdbr", 0xB3AD, GR64, FP64>;
def CLGXBR : TernaryRRFe<"clgxbr", 0xB3AE, GR64, FP128>;
}
def : Pat<(i32 (any_fp_to_uint FP32:$src)), (CLFEBR 5, FP32:$src, 0)>;
def : Pat<(i32 (any_fp_to_uint FP64:$src)), (CLFDBR 5, FP64:$src, 0)>;
def : Pat<(i32 (any_fp_to_uint FP128:$src)), (CLFXBR 5, FP128:$src, 0)>;
def : Pat<(i64 (any_fp_to_uint FP32:$src)), (CLGEBR 5, FP32:$src, 0)>;
def : Pat<(i64 (any_fp_to_uint FP64:$src)), (CLGDBR 5, FP64:$src, 0)>;
def : Pat<(i64 (any_fp_to_uint FP128:$src)), (CLGXBR 5, FP128:$src, 0)>;
}
//===----------------------------------------------------------------------===//
// Unary arithmetic
//===----------------------------------------------------------------------===//
// We prefer generic instructions during isel, because they do not
// clobber CC and therefore give the scheduler more freedom. In cases
// the CC is actually useful, the SystemZElimCompare pass will try to
// convert generic instructions into opcodes that also set CC. Note
// that lcdf / lpdf / lndf only affect the sign bit, and can therefore
// be used with fp32 as well. This could be done for fp128, in which
// case the operands would have to be tied.
// Negation (Load Complement).
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
def LCEBR : UnaryRRE<"lcebr", 0xB303, null_frag, FP32, FP32>;
def LCDBR : UnaryRRE<"lcdbr", 0xB313, null_frag, FP64, FP64>;
def LCXBR : UnaryRRE<"lcxbr", 0xB343, fneg, FP128, FP128>;
}
// Generic form, which does not set CC.
def LCDFR : UnaryRRE<"lcdfr", 0xB373, fneg, FP64, FP64>;
let isCodeGenOnly = 1 in
def LCDFR_32 : UnaryRRE<"lcdfr", 0xB373, fneg, FP32, FP32>;
// Absolute value (Load Positive).
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
def LPEBR : UnaryRRE<"lpebr", 0xB300, null_frag, FP32, FP32>;
def LPDBR : UnaryRRE<"lpdbr", 0xB310, null_frag, FP64, FP64>;
def LPXBR : UnaryRRE<"lpxbr", 0xB340, fabs, FP128, FP128>;
}
// Generic form, which does not set CC.
def LPDFR : UnaryRRE<"lpdfr", 0xB370, fabs, FP64, FP64>;
let isCodeGenOnly = 1 in
def LPDFR_32 : UnaryRRE<"lpdfr", 0xB370, fabs, FP32, FP32>;
// Negative absolute value (Load Negative).
let Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
def LNEBR : UnaryRRE<"lnebr", 0xB301, null_frag, FP32, FP32>;
def LNDBR : UnaryRRE<"lndbr", 0xB311, null_frag, FP64, FP64>;
def LNXBR : UnaryRRE<"lnxbr", 0xB341, fnabs, FP128, FP128>;
}
// Generic form, which does not set CC.
def LNDFR : UnaryRRE<"lndfr", 0xB371, fnabs, FP64, FP64>;
let isCodeGenOnly = 1 in
def LNDFR_32 : UnaryRRE<"lndfr", 0xB371, fnabs, FP32, FP32>;
// Square root.
let Uses = [FPC], mayRaiseFPException = 1 in {
def SQEBR : UnaryRRE<"sqebr", 0xB314, any_fsqrt, FP32, FP32>;
def SQDBR : UnaryRRE<"sqdbr", 0xB315, any_fsqrt, FP64, FP64>;
def SQXBR : UnaryRRE<"sqxbr", 0xB316, any_fsqrt, FP128, FP128>;
def SQEB : UnaryRXE<"sqeb", 0xED14, loadu<any_fsqrt>, FP32, 4>;
def SQDB : UnaryRXE<"sqdb", 0xED15, loadu<any_fsqrt>, FP64, 8>;
}
// Round to an integer, with the second operand (modifier M3) specifying
// the rounding mode. These forms always check for inexact conditions.
let Uses = [FPC], mayRaiseFPException = 1 in {
def FIEBR : BinaryRRFe<"fiebr", 0xB357, FP32, FP32>;
def FIDBR : BinaryRRFe<"fidbr", 0xB35F, FP64, FP64>;
def FIXBR : BinaryRRFe<"fixbr", 0xB347, FP128, FP128>;
}
// frint rounds according to the current mode (modifier 0) and detects
// inexact conditions.
def : Pat<(any_frint FP32:$src), (FIEBR 0, FP32:$src)>;
def : Pat<(any_frint FP64:$src), (FIDBR 0, FP64:$src)>;
def : Pat<(any_frint FP128:$src), (FIXBR 0, FP128:$src)>;
let Predicates = [FeatureFPExtension] in {
// Extended forms of the FIxBR instructions. M4 can be set to 4
// to suppress detection of inexact conditions.
let Uses = [FPC], mayRaiseFPException = 1 in {
def FIEBRA : TernaryRRFe<"fiebra", 0xB357, FP32, FP32>;
def FIDBRA : TernaryRRFe<"fidbra", 0xB35F, FP64, FP64>;
def FIXBRA : TernaryRRFe<"fixbra", 0xB347, FP128, FP128>;
}
// fnearbyint is like frint but does not detect inexact conditions.
def : Pat<(any_fnearbyint FP32:$src), (FIEBRA 0, FP32:$src, 4)>;
def : Pat<(any_fnearbyint FP64:$src), (FIDBRA 0, FP64:$src, 4)>;
def : Pat<(any_fnearbyint FP128:$src), (FIXBRA 0, FP128:$src, 4)>;
// floor is no longer allowed to raise an inexact condition,
// so restrict it to the cases where the condition can be suppressed.
// Mode 7 is round towards -inf.
def : Pat<(any_ffloor FP32:$src), (FIEBRA 7, FP32:$src, 4)>;
def : Pat<(any_ffloor FP64:$src), (FIDBRA 7, FP64:$src, 4)>;
def : Pat<(any_ffloor FP128:$src), (FIXBRA 7, FP128:$src, 4)>;
// Same idea for ceil, where mode 6 is round towards +inf.
def : Pat<(any_fceil FP32:$src), (FIEBRA 6, FP32:$src, 4)>;
def : Pat<(any_fceil FP64:$src), (FIDBRA 6, FP64:$src, 4)>;
def : Pat<(any_fceil FP128:$src), (FIXBRA 6, FP128:$src, 4)>;
// Same idea for trunc, where mode 5 is round towards zero.
def : Pat<(any_ftrunc FP32:$src), (FIEBRA 5, FP32:$src, 4)>;
def : Pat<(any_ftrunc FP64:$src), (FIDBRA 5, FP64:$src, 4)>;
def : Pat<(any_ftrunc FP128:$src), (FIXBRA 5, FP128:$src, 4)>;
// Same idea for round, where mode 1 is round towards nearest with
// ties away from zero.
def : Pat<(any_fround FP32:$src), (FIEBRA 1, FP32:$src, 4)>;
def : Pat<(any_fround FP64:$src), (FIDBRA 1, FP64:$src, 4)>;
def : Pat<(any_fround FP128:$src), (FIXBRA 1, FP128:$src, 4)>;
}
//===----------------------------------------------------------------------===//
// Binary arithmetic
//===----------------------------------------------------------------------===//
// Addition.
let Uses = [FPC], mayRaiseFPException = 1,
Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
let isCommutable = 1 in {
def AEBR : BinaryRRE<"aebr", 0xB30A, any_fadd, FP32, FP32>;
def ADBR : BinaryRRE<"adbr", 0xB31A, any_fadd, FP64, FP64>;
def AXBR : BinaryRRE<"axbr", 0xB34A, any_fadd, FP128, FP128>;
}
defm AEB : BinaryRXEAndPseudo<"aeb", 0xED0A, any_fadd, FP32, load, 4>;
defm ADB : BinaryRXEAndPseudo<"adb", 0xED1A, any_fadd, FP64, load, 8>;
}
// Subtraction.
let Uses = [FPC], mayRaiseFPException = 1,
Defs = [CC], CCValues = 0xF, CompareZeroCCMask = 0xF in {
def SEBR : BinaryRRE<"sebr", 0xB30B, any_fsub, FP32, FP32>;
def SDBR : BinaryRRE<"sdbr", 0xB31B, any_fsub, FP64, FP64>;
def SXBR : BinaryRRE<"sxbr", 0xB34B, any_fsub, FP128, FP128>;
defm SEB : BinaryRXEAndPseudo<"seb", 0xED0B, any_fsub, FP32, load, 4>;
defm SDB : BinaryRXEAndPseudo<"sdb", 0xED1B, any_fsub, FP64, load, 8>;
}
// Multiplication.
let Uses = [FPC], mayRaiseFPException = 1 in {
let isCommutable = 1 in {
def MEEBR : BinaryRRE<"meebr", 0xB317, any_fmul, FP32, FP32>;
def MDBR : BinaryRRE<"mdbr", 0xB31C, any_fmul, FP64, FP64>;
def MXBR : BinaryRRE<"mxbr", 0xB34C, any_fmul, FP128, FP128>;
}
defm MEEB : BinaryRXEAndPseudo<"meeb", 0xED17, any_fmul, FP32, load, 4>;
defm MDB : BinaryRXEAndPseudo<"mdb", 0xED1C, any_fmul, FP64, load, 8>;
}
// f64 multiplication of two FP32 registers.
let Uses = [FPC], mayRaiseFPException = 1 in
def MDEBR : BinaryRRE<"mdebr", 0xB30C, null_frag, FP64, FP32>;
def : Pat<(any_fmul (f64 (any_fpextend FP32:$src1)),
(f64 (any_fpextend FP32:$src2))),
(MDEBR (INSERT_SUBREG (f64 (IMPLICIT_DEF)),
FP32:$src1, subreg_h32), FP32:$src2)>;
// f64 multiplication of an FP32 register and an f32 memory.
let Uses = [FPC], mayRaiseFPException = 1 in
def MDEB : BinaryRXE<"mdeb", 0xED0C, null_frag, FP64, load, 4>;
def : Pat<(any_fmul (f64 (any_fpextend FP32:$src1)),
(f64 (any_extloadf32 bdxaddr12only:$addr))),
(MDEB (INSERT_SUBREG (f64 (IMPLICIT_DEF)), FP32:$src1, subreg_h32),
bdxaddr12only:$addr)>;
// f128 multiplication of two FP64 registers.
let Uses = [FPC], mayRaiseFPException = 1 in
def MXDBR : BinaryRRE<"mxdbr", 0xB307, null_frag, FP128, FP64>;
let Predicates = [FeatureNoVectorEnhancements1] in
def : Pat<(any_fmul (f128 (any_fpextend FP64:$src1)),
(f128 (any_fpextend FP64:$src2))),
(MXDBR (INSERT_SUBREG (f128 (IMPLICIT_DEF)),
FP64:$src1, subreg_h64), FP64:$src2)>;
// f128 multiplication of an FP64 register and an f64 memory.
let Uses = [FPC], mayRaiseFPException = 1 in
def MXDB : BinaryRXE<"mxdb", 0xED07, null_frag, FP128, load, 8>;
let Predicates = [FeatureNoVectorEnhancements1] in
def : Pat<(any_fmul (f128 (any_fpextend FP64:$src1)),
(f128 (any_extloadf64 bdxaddr12only:$addr))),
(MXDB (INSERT_SUBREG (f128 (IMPLICIT_DEF)), FP64:$src1, subreg_h64),
bdxaddr12only:$addr)>;
// Fused multiply-add.
let Uses = [FPC], mayRaiseFPException = 1 in {
def MAEBR : TernaryRRD<"maebr", 0xB30E, z_any_fma, FP32, FP32>;
def MADBR : TernaryRRD<"madbr", 0xB31E, z_any_fma, FP64, FP64>;
defm MAEB : TernaryRXFAndPseudo<"maeb", 0xED0E, z_any_fma, FP32, FP32, load, 4>;
defm MADB : TernaryRXFAndPseudo<"madb", 0xED1E, z_any_fma, FP64, FP64, load, 8>;
}
// Fused multiply-subtract.
let Uses = [FPC], mayRaiseFPException = 1 in {
def MSEBR : TernaryRRD<"msebr", 0xB30F, z_any_fms, FP32, FP32>;
def MSDBR : TernaryRRD<"msdbr", 0xB31F, z_any_fms, FP64, FP64>;
defm MSEB : TernaryRXFAndPseudo<"mseb", 0xED0F, z_any_fms, FP32, FP32, load, 4>;
defm MSDB : TernaryRXFAndPseudo<"msdb", 0xED1F, z_any_fms, FP64, FP64, load, 8>;
}
// Division.
let Uses = [FPC], mayRaiseFPException = 1 in {
def DEBR : BinaryRRE<"debr", 0xB30D, any_fdiv, FP32, FP32>;
def DDBR : BinaryRRE<"ddbr", 0xB31D, any_fdiv, FP64, FP64>;
def DXBR : BinaryRRE<"dxbr", 0xB34D, any_fdiv, FP128, FP128>;
defm DEB : BinaryRXEAndPseudo<"deb", 0xED0D, any_fdiv, FP32, load, 4>;
defm DDB : BinaryRXEAndPseudo<"ddb", 0xED1D, any_fdiv, FP64, load, 8>;
}
// Divide to integer.
let Uses = [FPC], mayRaiseFPException = 1, Defs = [CC] in {
def DIEBR : TernaryRRFb<"diebr", 0xB353, FP32, FP32, FP32>;
def DIDBR : TernaryRRFb<"didbr", 0xB35B, FP64, FP64, FP64>;
}
//===----------------------------------------------------------------------===//
// Comparisons
//===----------------------------------------------------------------------===//
let Uses = [FPC], mayRaiseFPException = 1, Defs = [CC], CCValues = 0xF in {
def CEBR : CompareRRE<"cebr", 0xB309, z_any_fcmp, FP32, FP32>;
def CDBR : CompareRRE<"cdbr", 0xB319, z_any_fcmp, FP64, FP64>;
def CXBR : CompareRRE<"cxbr", 0xB349, z_any_fcmp, FP128, FP128>;
def CEB : CompareRXE<"ceb", 0xED09, z_any_fcmp, FP32, load, 4>;
def CDB : CompareRXE<"cdb", 0xED19, z_any_fcmp, FP64, load, 8>;
def KEBR : CompareRRE<"kebr", 0xB308, z_strict_fcmps, FP32, FP32>;
def KDBR : CompareRRE<"kdbr", 0xB318, z_strict_fcmps, FP64, FP64>;
def KXBR : CompareRRE<"kxbr", 0xB348, z_strict_fcmps, FP128, FP128>;
def KEB : CompareRXE<"keb", 0xED08, z_strict_fcmps, FP32, load, 4>;
def KDB : CompareRXE<"kdb", 0xED18, z_strict_fcmps, FP64, load, 8>;
}
// Test Data Class.
let Defs = [CC], CCValues = 0xC in {
def TCEB : TestRXE<"tceb", 0xED10, z_tdc, FP32>;
def TCDB : TestRXE<"tcdb", 0xED11, z_tdc, FP64>;
def TCXB : TestRXE<"tcxb", 0xED12, z_tdc, FP128>;
}
//===----------------------------------------------------------------------===//
// Floating-point control register instructions
//===----------------------------------------------------------------------===//
let hasSideEffects = 1 in {
let mayLoad = 1, mayStore = 1 in {
// TODO: EFPC and SFPC do not touch memory at all
let Uses = [FPC] in {
def EFPC : InherentRRE<"efpc", 0xB38C, GR32, int_s390_efpc>;
def STFPC : StoreInherentS<"stfpc", 0xB29C, storei<int_s390_efpc>, 4>;
}
let Defs = [FPC] in {
def SFPC : SideEffectUnaryRRE<"sfpc", 0xB384, GR32, int_s390_sfpc>;
def LFPC : SideEffectUnaryS<"lfpc", 0xB29D, loadu<int_s390_sfpc>, 4>;
}
}
let Defs = [FPC], mayRaiseFPException = 1 in {
def SFASR : SideEffectUnaryRRE<"sfasr", 0xB385, GR32, null_frag>;
def LFAS : SideEffectUnaryS<"lfas", 0xB2BD, null_frag, 4>;
}
let Uses = [FPC], Defs = [FPC] in {
def SRNMB : SideEffectAddressS<"srnmb", 0xB2B8, null_frag, shift12only>,
Requires<[FeatureFPExtension]>;
def SRNM : SideEffectAddressS<"srnm", 0xB299, null_frag, shift12only>;
def SRNMT : SideEffectAddressS<"srnmt", 0xB2B9, null_frag, shift12only>;
}
}
//===----------------------------------------------------------------------===//
// Peepholes
//===----------------------------------------------------------------------===//
def : Pat<(f32 fpimmneg0), (LCDFR_32 (LZER))>;
def : Pat<(f64 fpimmneg0), (LCDFR (LZDR))>;
def : Pat<(f128 fpimmneg0), (LCXBR (LZXR))>;