llvm.org GIT mirror llvm / 041b3f8 lib / Target / X86 / README.txt
041b3f8

Tree @041b3f8 (Download .tar.gz)

README.txt @041b3f8raw · history · blame

   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
 600
 601
 602
 603
 604
 605
 606
 607
 608
 609
 610
 611
 612
 613
 614
 615
 616
 617
 618
 619
 620
 621
 622
 623
 624
 625
 626
 627
 628
 629
 630
 631
 632
 633
 634
 635
 636
 637
 638
 639
 640
 641
 642
 643
 644
 645
 646
 647
 648
 649
 650
 651
 652
 653
 654
 655
 656
 657
 658
 659
 660
 661
 662
 663
 664
 665
 666
 667
 668
 669
 670
 671
 672
 673
 674
 675
 676
 677
 678
 679
 680
 681
 682
 683
 684
 685
 686
 687
 688
 689
 690
 691
 692
 693
 694
 695
 696
 697
 698
 699
 700
 701
 702
 703
 704
 705
 706
 707
 708
 709
 710
 711
 712
 713
 714
 715
 716
 717
 718
 719
 720
 721
 722
 723
 724
 725
 726
 727
 728
 729
 730
 731
 732
 733
 734
 735
 736
 737
 738
 739
 740
 741
 742
 743
 744
 745
 746
 747
 748
 749
 750
 751
 752
 753
 754
 755
 756
 757
 758
 759
 760
 761
 762
 763
 764
 765
 766
 767
 768
 769
 770
 771
 772
 773
 774
 775
 776
 777
 778
 779
 780
 781
 782
 783
 784
 785
 786
 787
 788
 789
 790
 791
 792
 793
 794
 795
 796
 797
 798
 799
 800
 801
 802
 803
 804
 805
 806
 807
 808
 809
 810
 811
 812
 813
 814
 815
 816
 817
 818
 819
 820
 821
 822
 823
 824
 825
 826
 827
 828
 829
 830
 831
 832
 833
 834
 835
 836
 837
 838
 839
 840
 841
 842
 843
 844
 845
 846
 847
 848
 849
 850
 851
 852
 853
 854
 855
 856
 857
 858
 859
 860
 861
 862
 863
 864
 865
 866
 867
 868
 869
 870
 871
 872
 873
 874
 875
 876
 877
 878
 879
 880
 881
 882
 883
 884
 885
 886
 887
 888
 889
 890
 891
 892
 893
 894
 895
 896
 897
 898
 899
 900
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
//===---------------------------------------------------------------------===//
// Random ideas for the X86 backend.
//===---------------------------------------------------------------------===//

Missing features:
  - Support for SSE4: http://www.intel.com/software/penryn
http://softwarecommunity.intel.com/isn/Downloads/Intel%20SSE4%20Programming%20Reference.pdf
  - support for 3DNow!
  - weird abis?

//===---------------------------------------------------------------------===//

CodeGen/X86/lea-3.ll:test3 should be a single LEA, not a shift/move.  The X86
backend knows how to three-addressify this shift, but it appears the register
allocator isn't even asking it to do so in this case.  We should investigate
why this isn't happening, it could have significant impact on other important
cases for X86 as well.

//===---------------------------------------------------------------------===//

This should be one DIV/IDIV instruction, not a libcall:

unsigned test(unsigned long long X, unsigned Y) {
        return X/Y;
}

This can be done trivially with a custom legalizer.  What about overflow 
though?  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14224

//===---------------------------------------------------------------------===//

Improvements to the multiply -> shift/add algorithm:
http://gcc.gnu.org/ml/gcc-patches/2004-08/msg01590.html

//===---------------------------------------------------------------------===//

Improve code like this (occurs fairly frequently, e.g. in LLVM):
long long foo(int x) { return 1LL << x; }

http://gcc.gnu.org/ml/gcc-patches/2004-09/msg01109.html
http://gcc.gnu.org/ml/gcc-patches/2004-09/msg01128.html
http://gcc.gnu.org/ml/gcc-patches/2004-09/msg01136.html

Another useful one would be  ~0ULL >> X and ~0ULL << X.

One better solution for 1LL << x is:
        xorl    %eax, %eax
        xorl    %edx, %edx
        testb   $32, %cl
        sete    %al
        setne   %dl
        sall    %cl, %eax
        sall    %cl, %edx

But that requires good 8-bit subreg support.

64-bit shifts (in general) expand to really bad code.  Instead of using
cmovs, we should expand to a conditional branch like GCC produces.

//===---------------------------------------------------------------------===//

Compile this:
_Bool f(_Bool a) { return a!=1; }

into:
        movzbl  %dil, %eax
        xorl    $1, %eax
        ret

//===---------------------------------------------------------------------===//

Some isel ideas:

1. Dynamic programming based approach when compile time if not an
   issue.
2. Code duplication (addressing mode) during isel.
3. Other ideas from "Register-Sensitive Selection, Duplication, and
   Sequencing of Instructions".
4. Scheduling for reduced register pressure.  E.g. "Minimum Register 
   Instruction Sequence Problem: Revisiting Optimal Code Generation for DAGs" 
   and other related papers.
   http://citeseer.ist.psu.edu/govindarajan01minimum.html

//===---------------------------------------------------------------------===//

Should we promote i16 to i32 to avoid partial register update stalls?

//===---------------------------------------------------------------------===//

Leave any_extend as pseudo instruction and hint to register
allocator. Delay codegen until post register allocation.
Note. any_extend is now turned into an INSERT_SUBREG. We still need to teach
the coalescer how to deal with it though.

//===---------------------------------------------------------------------===//

Count leading zeros and count trailing zeros:

int clz(int X) { return __builtin_clz(X); }
int ctz(int X) { return __builtin_ctz(X); }

$ gcc t.c -S -o - -O3  -fomit-frame-pointer -masm=intel
clz:
        bsr     %eax, DWORD PTR [%esp+4]
        xor     %eax, 31
        ret
ctz:
        bsf     %eax, DWORD PTR [%esp+4]
        ret

however, check that these are defined for 0 and 32.  Our intrinsics are, GCC's
aren't.

Another example (use predsimplify to eliminate a select):

int foo (unsigned long j) {
  if (j)
    return __builtin_ffs (j) - 1;
  else
    return 0;
}

//===---------------------------------------------------------------------===//

It appears icc use push for parameter passing. Need to investigate.

//===---------------------------------------------------------------------===//

Only use inc/neg/not instructions on processors where they are faster than
add/sub/xor.  They are slower on the P4 due to only updating some processor
flags.

//===---------------------------------------------------------------------===//

The instruction selector sometimes misses folding a load into a compare.  The
pattern is written as (cmp reg, (load p)).  Because the compare isn't 
commutative, it is not matched with the load on both sides.  The dag combiner
should be made smart enough to cannonicalize the load into the RHS of a compare
when it can invert the result of the compare for free.

//===---------------------------------------------------------------------===//

How about intrinsics? An example is:
  *res = _mm_mulhi_epu16(*A, _mm_mul_epu32(*B, *C));

compiles to
	pmuludq (%eax), %xmm0
	movl 8(%esp), %eax
	movdqa (%eax), %xmm1
	pmulhuw %xmm0, %xmm1

The transformation probably requires a X86 specific pass or a DAG combiner
target specific hook.

//===---------------------------------------------------------------------===//

In many cases, LLVM generates code like this:

_test:
        movl 8(%esp), %eax
        cmpl %eax, 4(%esp)
        setl %al
        movzbl %al, %eax
        ret

on some processors (which ones?), it is more efficient to do this:

_test:
        movl 8(%esp), %ebx
        xor  %eax, %eax
        cmpl %ebx, 4(%esp)
        setl %al
        ret

Doing this correctly is tricky though, as the xor clobbers the flags.

//===---------------------------------------------------------------------===//

We should generate bts/btr/etc instructions on targets where they are cheap or
when codesize is important.  e.g., for:

void setbit(int *target, int bit) {
    *target |= (1 << bit);
}
void clearbit(int *target, int bit) {
    *target &= ~(1 << bit);
}

//===---------------------------------------------------------------------===//

Instead of the following for memset char*, 1, 10:

	movl $16843009, 4(%edx)
	movl $16843009, (%edx)
	movw $257, 8(%edx)

It might be better to generate

	movl $16843009, %eax
	movl %eax, 4(%edx)
	movl %eax, (%edx)
	movw al, 8(%edx)
	
when we can spare a register. It reduces code size.

//===---------------------------------------------------------------------===//

Evaluate what the best way to codegen sdiv X, (2^C) is.  For X/8, we currently
get this:

int %test1(int %X) {
        %Y = div int %X, 8
        ret int %Y
}

_test1:
        movl 4(%esp), %eax
        movl %eax, %ecx
        sarl $31, %ecx
        shrl $29, %ecx
        addl %ecx, %eax
        sarl $3, %eax
        ret

GCC knows several different ways to codegen it, one of which is this:

_test1:
        movl    4(%esp), %eax
        cmpl    $-1, %eax
        leal    7(%eax), %ecx
        cmovle  %ecx, %eax
        sarl    $3, %eax
        ret

which is probably slower, but it's interesting at least :)

//===---------------------------------------------------------------------===//

The first BB of this code:

declare bool %foo()
int %bar() {
        %V = call bool %foo()
        br bool %V, label %T, label %F
T:
        ret int 1
F:
        call bool %foo()
        ret int 12
}

compiles to:

_bar:
        subl $12, %esp
        call L_foo$stub
        xorb $1, %al
        testb %al, %al
        jne LBB_bar_2   # F

It would be better to emit "cmp %al, 1" than a xor and test.

//===---------------------------------------------------------------------===//

We are currently lowering large (1MB+) memmove/memcpy to rep/stosl and rep/movsl
We should leave these as libcalls for everything over a much lower threshold,
since libc is hand tuned for medium and large mem ops (avoiding RFO for large
stores, TLB preheating, etc)

//===---------------------------------------------------------------------===//

Optimize this into something reasonable:
 x * copysign(1.0, y) * copysign(1.0, z)

//===---------------------------------------------------------------------===//

Optimize copysign(x, *y) to use an integer load from y.

//===---------------------------------------------------------------------===//

%X = weak global int 0

void %foo(int %N) {
	%N = cast int %N to uint
	%tmp.24 = setgt int %N, 0
	br bool %tmp.24, label %no_exit, label %return

no_exit:
	%indvar = phi uint [ 0, %entry ], [ %indvar.next, %no_exit ]
	%i.0.0 = cast uint %indvar to int
	volatile store int %i.0.0, int* %X
	%indvar.next = add uint %indvar, 1
	%exitcond = seteq uint %indvar.next, %N
	br bool %exitcond, label %return, label %no_exit

return:
	ret void
}

compiles into:

	.text
	.align	4
	.globl	_foo
_foo:
	movl 4(%esp), %eax
	cmpl $1, %eax
	jl LBB_foo_4	# return
LBB_foo_1:	# no_exit.preheader
	xorl %ecx, %ecx
LBB_foo_2:	# no_exit
	movl L_X$non_lazy_ptr, %edx
	movl %ecx, (%edx)
	incl %ecx
	cmpl %eax, %ecx
	jne LBB_foo_2	# no_exit
LBB_foo_3:	# return.loopexit
LBB_foo_4:	# return
	ret

We should hoist "movl L_X$non_lazy_ptr, %edx" out of the loop after
remateralization is implemented. This can be accomplished with 1) a target
dependent LICM pass or 2) makeing SelectDAG represent the whole function. 

//===---------------------------------------------------------------------===//

The following tests perform worse with LSR:

lambda, siod, optimizer-eval, ackermann, hash2, nestedloop, strcat, and Treesor.

//===---------------------------------------------------------------------===//

We are generating far worse code than gcc:

volatile short X, Y;

void foo(int N) {
  int i;
  for (i = 0; i < N; i++) { X = i; Y = i*4; }
}

LBB1_1:	# entry.bb_crit_edge
	xorl	%ecx, %ecx
	xorw	%dx, %dx
LBB1_2:	# bb
	movl	L_X$non_lazy_ptr, %esi
	movw	%cx, (%esi)
	movl	L_Y$non_lazy_ptr, %esi
	movw	%dx, (%esi)
	addw	$4, %dx
	incl	%ecx
	cmpl	%eax, %ecx
	jne	LBB1_2	# bb

vs.

	xorl	%edx, %edx
	movl	L_X$non_lazy_ptr-"L00000000001$pb"(%ebx), %esi
	movl	L_Y$non_lazy_ptr-"L00000000001$pb"(%ebx), %ecx
L4:
	movw	%dx, (%esi)
	leal	0(,%edx,4), %eax
	movw	%ax, (%ecx)
	addl	$1, %edx
	cmpl	%edx, %edi
	jne	L4

This is due to the lack of post regalloc LICM.

//===---------------------------------------------------------------------===//

Teach the coalescer to coalesce vregs of different register classes. e.g. FR32 /
FR64 to VR128.

//===---------------------------------------------------------------------===//

mov $reg, 48(%esp)
...
leal 48(%esp), %eax
mov %eax, (%esp)
call _foo

Obviously it would have been better for the first mov (or any op) to store
directly %esp[0] if there are no other uses.

//===---------------------------------------------------------------------===//

Adding to the list of cmp / test poor codegen issues:

int test(__m128 *A, __m128 *B) {
  if (_mm_comige_ss(*A, *B))
    return 3;
  else
    return 4;
}

_test:
	movl 8(%esp), %eax
	movaps (%eax), %xmm0
	movl 4(%esp), %eax
	movaps (%eax), %xmm1
	comiss %xmm0, %xmm1
	setae %al
	movzbl %al, %ecx
	movl $3, %eax
	movl $4, %edx
	cmpl $0, %ecx
	cmove %edx, %eax
	ret

Note the setae, movzbl, cmpl, cmove can be replaced with a single cmovae. There
are a number of issues. 1) We are introducing a setcc between the result of the
intrisic call and select. 2) The intrinsic is expected to produce a i32 value
so a any extend (which becomes a zero extend) is added.

We probably need some kind of target DAG combine hook to fix this.

//===---------------------------------------------------------------------===//

We generate significantly worse code for this than GCC:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21150
http://gcc.gnu.org/bugzilla/attachment.cgi?id=8701

There is also one case we do worse on PPC.

//===---------------------------------------------------------------------===//

If shorter, we should use things like:
movzwl %ax, %eax
instead of:
andl $65535, %EAX

The former can also be used when the two-addressy nature of the 'and' would
require a copy to be inserted (in X86InstrInfo::convertToThreeAddress).

//===---------------------------------------------------------------------===//

Consider this:

typedef struct pair { float A, B; } pair;
void pairtest(pair P, float *FP) {
        *FP = P.A+P.B;
}

We currently generate this code with llvmgcc4:

_pairtest:
        movl 8(%esp), %eax
        movl 4(%esp), %ecx
        movd %eax, %xmm0
        movd %ecx, %xmm1
        addss %xmm0, %xmm1
        movl 12(%esp), %eax
        movss %xmm1, (%eax)
        ret

we should be able to generate:
_pairtest:
        movss 4(%esp), %xmm0
        movl 12(%esp), %eax
        addss 8(%esp), %xmm0
        movss %xmm0, (%eax)
        ret

The issue is that llvmgcc4 is forcing the struct to memory, then passing it as
integer chunks.  It does this so that structs like {short,short} are passed in
a single 32-bit integer stack slot.  We should handle the safe cases above much
nicer, while still handling the hard cases.

While true in general, in this specific case we could do better by promoting
load int + bitcast to float -> load fload.  This basically needs alignment info,
the code is already implemented (but disabled) in dag combine).

//===---------------------------------------------------------------------===//

Another instruction selector deficiency:

void %bar() {
	%tmp = load int (int)** %foo
	%tmp = tail call int %tmp( int 3 )
	ret void
}

_bar:
	subl $12, %esp
	movl L_foo$non_lazy_ptr, %eax
	movl (%eax), %eax
	call *%eax
	addl $12, %esp
	ret

The current isel scheme will not allow the load to be folded in the call since
the load's chain result is read by the callseq_start.

//===---------------------------------------------------------------------===//

For this:

int test(int a)
{
  return a * 3;
}

We currently emits
	imull $3, 4(%esp), %eax

Perhaps this is what we really should generate is? Is imull three or four
cycles? Note: ICC generates this:
	movl	4(%esp), %eax
	leal	(%eax,%eax,2), %eax

The current instruction priority is based on pattern complexity. The former is
more "complex" because it folds a load so the latter will not be emitted.

Perhaps we should use AddedComplexity to give LEA32r a higher priority? We
should always try to match LEA first since the LEA matching code does some
estimate to determine whether the match is profitable.

However, if we care more about code size, then imull is better. It's two bytes
shorter than movl + leal.

//===---------------------------------------------------------------------===//

Implement CTTZ, CTLZ with bsf and bsr. GCC produces:

int ctz_(unsigned X) { return __builtin_ctz(X); }
int clz_(unsigned X) { return __builtin_clz(X); }
int ffs_(unsigned X) { return __builtin_ffs(X); }

_ctz_:
        bsfl    4(%esp), %eax
        ret
_clz_:
        bsrl    4(%esp), %eax
        xorl    $31, %eax
        ret
_ffs_:
        movl    $-1, %edx
        bsfl    4(%esp), %eax
        cmove   %edx, %eax
        addl    $1, %eax
        ret

//===---------------------------------------------------------------------===//

It appears gcc place string data with linkonce linkage in
.section __TEXT,__const_coal,coalesced instead of
.section __DATA,__const_coal,coalesced.
Take a look at darwin.h, there are other Darwin assembler directives that we
do not make use of.

//===---------------------------------------------------------------------===//

int %foo(int* %a, int %t) {
entry:
        br label %cond_true

cond_true:              ; preds = %cond_true, %entry
        %x.0.0 = phi int [ 0, %entry ], [ %tmp9, %cond_true ]  
        %t_addr.0.0 = phi int [ %t, %entry ], [ %tmp7, %cond_true ]
        %tmp2 = getelementptr int* %a, int %x.0.0              
        %tmp3 = load int* %tmp2         ; <int> [#uses=1]
        %tmp5 = add int %t_addr.0.0, %x.0.0             ; <int> [#uses=1]
        %tmp7 = add int %tmp5, %tmp3            ; <int> [#uses=2]
        %tmp9 = add int %x.0.0, 1               ; <int> [#uses=2]
        %tmp = setgt int %tmp9, 39              ; <bool> [#uses=1]
        br bool %tmp, label %bb12, label %cond_true

bb12:           ; preds = %cond_true
        ret int %tmp7
}

is pessimized by -loop-reduce and -indvars

//===---------------------------------------------------------------------===//

u32 to float conversion improvement:

float uint32_2_float( unsigned u ) {
  float fl = (int) (u & 0xffff);
  float fh = (int) (u >> 16);
  fh *= 0x1.0p16f;
  return fh + fl;
}

00000000        subl    $0x04,%esp
00000003        movl    0x08(%esp,1),%eax
00000007        movl    %eax,%ecx
00000009        shrl    $0x10,%ecx
0000000c        cvtsi2ss        %ecx,%xmm0
00000010        andl    $0x0000ffff,%eax
00000015        cvtsi2ss        %eax,%xmm1
00000019        mulss   0x00000078,%xmm0
00000021        addss   %xmm1,%xmm0
00000025        movss   %xmm0,(%esp,1)
0000002a        flds    (%esp,1)
0000002d        addl    $0x04,%esp
00000030        ret

//===---------------------------------------------------------------------===//

When using fastcc abi, align stack slot of argument of type double on 8 byte
boundary to improve performance.

//===---------------------------------------------------------------------===//

Codegen:

int f(int a, int b) {
  if (a == 4 || a == 6)
    b++;
  return b;
}


as:

or eax, 2
cmp eax, 6
jz label

//===---------------------------------------------------------------------===//

GCC's ix86_expand_int_movcc function (in i386.c) has a ton of interesting
simplifications for integer "x cmp y ? a : b".  For example, instead of:

int G;
void f(int X, int Y) {
  G = X < 0 ? 14 : 13;
}

compiling to:

_f:
        movl $14, %eax
        movl $13, %ecx
        movl 4(%esp), %edx
        testl %edx, %edx
        cmovl %eax, %ecx
        movl %ecx, _G
        ret

it could be:
_f:
        movl    4(%esp), %eax
        sarl    $31, %eax
        notl    %eax
        addl    $14, %eax
        movl    %eax, _G
        ret

etc.

Another is:
int usesbb(unsigned int a, unsigned int b) {
       return (a < b ? -1 : 0);
}
to:
_usesbb:
	movl	8(%esp), %eax
	cmpl	%eax, 4(%esp)
	sbbl	%eax, %eax
	ret

instead of:
_usesbb:
	xorl	%eax, %eax
	movl	8(%esp), %ecx
	cmpl	%ecx, 4(%esp)
	movl	$4294967295, %ecx
	cmovb	%ecx, %eax
	ret

//===---------------------------------------------------------------------===//

Currently we don't have elimination of redundant stack manipulations. Consider
the code:

int %main() {
entry:
	call fastcc void %test1( )
	call fastcc void %test2( sbyte* cast (void ()* %test1 to sbyte*) )
	ret int 0
}

declare fastcc void %test1()

declare fastcc void %test2(sbyte*)


This currently compiles to:

	subl $16, %esp
	call _test5
	addl $12, %esp
	subl $16, %esp
	movl $_test5, (%esp)
	call _test6
	addl $12, %esp

The add\sub pair is really unneeded here.

//===---------------------------------------------------------------------===//

We currently compile sign_extend_inreg into two shifts:

long foo(long X) {
  return (long)(signed char)X;
}

becomes:

_foo:
        movl 4(%esp), %eax
        shll $24, %eax
        sarl $24, %eax
        ret

This could be:

_foo:
        movsbl  4(%esp),%eax
        ret

//===---------------------------------------------------------------------===//

Consider the expansion of:

uint %test3(uint %X) {
        %tmp1 = rem uint %X, 255
        ret uint %tmp1
}

Currently it compiles to:

...
        movl $2155905153, %ecx
        movl 8(%esp), %esi
        movl %esi, %eax
        mull %ecx
...

This could be "reassociated" into:

        movl $2155905153, %eax
        movl 8(%esp), %ecx
        mull %ecx

to avoid the copy.  In fact, the existing two-address stuff would do this
except that mul isn't a commutative 2-addr instruction.  I guess this has
to be done at isel time based on the #uses to mul?

//===---------------------------------------------------------------------===//

Make sure the instruction which starts a loop does not cross a cacheline
boundary. This requires knowning the exact length of each machine instruction.
That is somewhat complicated, but doable. Example 256.bzip2:

In the new trace, the hot loop has an instruction which crosses a cacheline
boundary.  In addition to potential cache misses, this can't help decoding as I
imagine there has to be some kind of complicated decoder reset and realignment
to grab the bytes from the next cacheline.

532  532 0x3cfc movb     (1809(%esp, %esi), %bl   <<<--- spans 2 64 byte lines
942  942 0x3d03 movl     %dh, (1809(%esp, %esi)                                                                          
937  937 0x3d0a incl     %esi                           
3    3   0x3d0b cmpb     %bl, %dl                                               
27   27  0x3d0d jnz      0x000062db <main+11707>

//===---------------------------------------------------------------------===//

In c99 mode, the preprocessor doesn't like assembly comments like #TRUNCATE.

//===---------------------------------------------------------------------===//

This could be a single 16-bit load.

int f(char *p) {
    if ((p[0] == 1) & (p[1] == 2)) return 1;
    return 0;
}

//===---------------------------------------------------------------------===//

We should inline lrintf and probably other libc functions.

//===---------------------------------------------------------------------===//

Start using the flags more.  For example, compile:

int add_zf(int *x, int y, int a, int b) {
     if ((*x += y) == 0)
          return a;
     else
          return b;
}

to:
       addl    %esi, (%rdi)
       movl    %edx, %eax
       cmovne  %ecx, %eax
       ret
instead of:

_add_zf:
        addl (%rdi), %esi
        movl %esi, (%rdi)
        testl %esi, %esi
        cmove %edx, %ecx
        movl %ecx, %eax
        ret

and:

int add_zf(int *x, int y, int a, int b) {
     if ((*x + y) < 0)
          return a;
     else
          return b;
}

to:

add_zf:
        addl    (%rdi), %esi
        movl    %edx, %eax
        cmovns  %ecx, %eax
        ret

instead of:

_add_zf:
        addl (%rdi), %esi
        testl %esi, %esi
        cmovs %edx, %ecx
        movl %ecx, %eax
        ret

//===---------------------------------------------------------------------===//

This:
#include <math.h>
int foo(double X) { return isnan(X); }

compiles to (-m64):

_foo:
        pxor %xmm1, %xmm1
        ucomisd %xmm1, %xmm0
        setp %al
        movzbl %al, %eax
        ret

the pxor is not needed, we could compare the value against itself.

//===---------------------------------------------------------------------===//

These two functions have identical effects:

unsigned int f(unsigned int i, unsigned int n) {++i; if (i == n) ++i; return i;}
unsigned int f2(unsigned int i, unsigned int n) {++i; i += i == n; return i;}

We currently compile them to:

_f:
        movl 4(%esp), %eax
        movl %eax, %ecx
        incl %ecx
        movl 8(%esp), %edx
        cmpl %edx, %ecx
        jne LBB1_2      #UnifiedReturnBlock
LBB1_1: #cond_true
        addl $2, %eax
        ret
LBB1_2: #UnifiedReturnBlock
        movl %ecx, %eax
        ret
_f2:
        movl 4(%esp), %eax
        movl %eax, %ecx
        incl %ecx
        cmpl 8(%esp), %ecx
        sete %cl
        movzbl %cl, %ecx
        leal 1(%ecx,%eax), %eax
        ret

both of which are inferior to GCC's:

_f:
        movl    4(%esp), %edx
        leal    1(%edx), %eax
        addl    $2, %edx
        cmpl    8(%esp), %eax
        cmove   %edx, %eax
        ret
_f2:
        movl    4(%esp), %eax
        addl    $1, %eax
        xorl    %edx, %edx
        cmpl    8(%esp), %eax
        sete    %dl
        addl    %edx, %eax
        ret

//===---------------------------------------------------------------------===//

This code:

void test(int X) {
  if (X) abort();
}

is currently compiled to:

_test:
        subl $12, %esp
        cmpl $0, 16(%esp)
        jne LBB1_1
        addl $12, %esp
        ret
LBB1_1:
        call L_abort$stub

It would be better to produce:

_test:
        subl $12, %esp
        cmpl $0, 16(%esp)
        jne L_abort$stub
        addl $12, %esp
        ret

This can be applied to any no-return function call that takes no arguments etc.
Alternatively, the stack save/restore logic could be shrink-wrapped, producing
something like this:

_test:
        cmpl $0, 4(%esp)
        jne LBB1_1
        ret
LBB1_1:
        subl $12, %esp
        call L_abort$stub

Both are useful in different situations.  Finally, it could be shrink-wrapped
and tail called, like this:

_test:
        cmpl $0, 4(%esp)
        jne LBB1_1
        ret
LBB1_1:
        pop %eax   # realign stack.
        call L_abort$stub

Though this probably isn't worth it.

//===---------------------------------------------------------------------===//

We need to teach the codegen to convert two-address INC instructions to LEA
when the flags are dead (likewise dec).  For example, on X86-64, compile:

int foo(int A, int B) {
  return A+1;
}

to:

_foo:
        leal    1(%edi), %eax
        ret

instead of:

_foo:
        incl %edi
        movl %edi, %eax
        ret

Another example is:

;; X's live range extends beyond the shift, so the register allocator
;; cannot coalesce it with Y.  Because of this, a copy needs to be
;; emitted before the shift to save the register value before it is
;; clobbered.  However, this copy is not needed if the register
;; allocator turns the shift into an LEA.  This also occurs for ADD.

; Check that the shift gets turned into an LEA.
; RUN: llvm-upgrade < %s | llvm-as | llc -march=x86 -x86-asm-syntax=intel | \
; RUN:   not grep {mov E.X, E.X}

%G = external global int

int %test1(int %X, int %Y) {
        %Z = add int %X, %Y
        volatile store int %Y, int* %G
        volatile store int %Z, int* %G
        ret int %X
}

int %test2(int %X) {
        %Z = add int %X, 1  ;; inc
        volatile store int %Z, int* %G
        ret int %X
}

//===---------------------------------------------------------------------===//

Sometimes it is better to codegen subtractions from a constant (e.g. 7-x) with
a neg instead of a sub instruction.  Consider:

int test(char X) { return 7-X; }

we currently produce:
_test:
        movl $7, %eax
        movsbl 4(%esp), %ecx
        subl %ecx, %eax
        ret

We would use one fewer register if codegen'd as:

        movsbl 4(%esp), %eax
	neg %eax
        add $7, %eax
        ret

Note that this isn't beneficial if the load can be folded into the sub.  In
this case, we want a sub:

int test(int X) { return 7-X; }
_test:
        movl $7, %eax
        subl 4(%esp), %eax
        ret

//===---------------------------------------------------------------------===//

For code like:
phi (undef, x)

We get an implicit def on the undef side. If the phi is spilled, we then get:
implicitdef xmm1
store xmm1 -> stack

It should be possible to teach the x86 backend to "fold" the store into the
implicitdef, which just deletes the implicit def.

These instructions should go away:
#IMPLICIT_DEF %xmm1 
movaps %xmm1, 192(%esp) 
movaps %xmm1, 224(%esp) 
movaps %xmm1, 176(%esp)

//===---------------------------------------------------------------------===//

This is a "commutable two-address" register coallescing deficiency:

define <4 x float> @test1(<4 x float> %V) {
entry:
        %tmp8 = shufflevector <4 x float> %V, <4 x float> undef,
                                        <4 x i32> < i32 3, i32 2, i32 1, i32 0 >
        %add = add <4 x float> %tmp8, %V
        ret <4 x float> %add
}

this codegens to:

_test1:
        pshufd  $27, %xmm0, %xmm1
        addps   %xmm0, %xmm1
        movaps  %xmm1, %xmm0
        ret

instead of:

_test1:
        pshufd  $27, %xmm0, %xmm1
        addps   %xmm1, %xmm0
        ret

//===---------------------------------------------------------------------===//

Leaf functions that require one 4-byte spill slot have a prolog like this:

_foo:
        pushl   %esi
        subl    $4, %esp
...
and an epilog like this:
        addl    $4, %esp
        popl    %esi
        ret

It would be smaller, and potentially faster, to push eax on entry and to
pop into a dummy register instead of using addl/subl of esp.  Just don't pop 
into any return registers :)

//===---------------------------------------------------------------------===//

The X86 backend should fold (branch (or (setcc, setcc))) into multiple 
branches.  We generate really poor code for:

double testf(double a) {
       return a == 0.0 ? 0.0 : (a > 0.0 ? 1.0 : -1.0);
}

For example, the entry BB is:

_testf:
        subl    $20, %esp
        pxor    %xmm0, %xmm0
        movsd   24(%esp), %xmm1
        ucomisd %xmm0, %xmm1
        setnp   %al
        sete    %cl
        testb   %cl, %al
        jne     LBB1_5  # UnifiedReturnBlock
LBB1_1: # cond_true


it would be better to replace the last four instructions with:

	jp LBB1_1
	je LBB1_5
LBB1_1:

We also codegen the inner ?: into a diamond:

       cvtss2sd        LCPI1_0(%rip), %xmm2
        cvtss2sd        LCPI1_1(%rip), %xmm3
        ucomisd %xmm1, %xmm0
        ja      LBB1_3  # cond_true
LBB1_2: # cond_true
        movapd  %xmm3, %xmm2
LBB1_3: # cond_true
        movapd  %xmm2, %xmm0
        ret

We should sink the load into xmm3 into the LBB1_2 block.  This should
be pretty easy, and will nuke all the copies.

//===---------------------------------------------------------------------===//

This:
        #include <algorithm>
        inline std::pair<unsigned, bool> full_add(unsigned a, unsigned b)
        { return std::make_pair(a + b, a + b < a); }
        bool no_overflow(unsigned a, unsigned b)
        { return !full_add(a, b).second; }

Should compile to:


        _Z11no_overflowjj:
                addl    %edi, %esi
                setae   %al
                ret

on x86-64, not:

__Z11no_overflowjj:
        addl    %edi, %esi
        cmpl    %edi, %esi
        setae   %al
        movzbl  %al, %eax
        ret


//===---------------------------------------------------------------------===//

Re-materialize MOV32r0 etc. with xor instead of changing them to moves if the
condition register is dead. xor reg reg is shorter than mov reg, #0.

//===---------------------------------------------------------------------===//

We aren't matching RMW instructions aggressively
enough.  Here's a reduced testcase (more in PR1160):

define void @test(i32* %huge_ptr, i32* %target_ptr) {
        %A = load i32* %huge_ptr                ; <i32> [#uses=1]
        %B = load i32* %target_ptr              ; <i32> [#uses=1]
        %C = or i32 %A, %B              ; <i32> [#uses=1]
        store i32 %C, i32* %target_ptr
        ret void
}

$ llvm-as < t.ll | llc -march=x86-64

_test:
        movl (%rdi), %eax
        orl (%rsi), %eax
        movl %eax, (%rsi)
        ret

That should be something like:

_test:
        movl (%rdi), %eax
        orl %eax, (%rsi)
        ret

//===---------------------------------------------------------------------===//

The following code:

bb114.preheader:		; preds = %cond_next94
	%tmp231232 = sext i16 %tmp62 to i32		; <i32> [#uses=1]
	%tmp233 = sub i32 32, %tmp231232		; <i32> [#uses=1]
	%tmp245246 = sext i16 %tmp65 to i32		; <i32> [#uses=1]
	%tmp252253 = sext i16 %tmp68 to i32		; <i32> [#uses=1]
	%tmp254 = sub i32 32, %tmp252253		; <i32> [#uses=1]
	%tmp553554 = bitcast i16* %tmp37 to i8*		; <i8*> [#uses=2]
	%tmp583584 = sext i16 %tmp98 to i32		; <i32> [#uses=1]
	%tmp585 = sub i32 32, %tmp583584		; <i32> [#uses=1]
	%tmp614615 = sext i16 %tmp101 to i32		; <i32> [#uses=1]
	%tmp621622 = sext i16 %tmp104 to i32		; <i32> [#uses=1]
	%tmp623 = sub i32 32, %tmp621622		; <i32> [#uses=1]
	br label %bb114

produces:

LBB3_5:	# bb114.preheader
	movswl	-68(%ebp), %eax
	movl	$32, %ecx
	movl	%ecx, -80(%ebp)
	subl	%eax, -80(%ebp)
	movswl	-52(%ebp), %eax
	movl	%ecx, -84(%ebp)
	subl	%eax, -84(%ebp)
	movswl	-70(%ebp), %eax
	movl	%ecx, -88(%ebp)
	subl	%eax, -88(%ebp)
	movswl	-50(%ebp), %eax
	subl	%eax, %ecx
	movl	%ecx, -76(%ebp)
	movswl	-42(%ebp), %eax
	movl	%eax, -92(%ebp)
	movswl	-66(%ebp), %eax
	movl	%eax, -96(%ebp)
	movw	$0, -98(%ebp)

This appears to be bad because the RA is not folding the store to the stack 
slot into the movl.  The above instructions could be:
	movl    $32, -80(%ebp)
...
	movl    $32, -84(%ebp)
...
This seems like a cross between remat and spill folding.

This has redundant subtractions of %eax from a stack slot. However, %ecx doesn't
change, so we could simply subtract %eax from %ecx first and then use %ecx (or
vice-versa).

//===---------------------------------------------------------------------===//

For this code:

cond_next603:		; preds = %bb493, %cond_true336, %cond_next599
	%v.21050.1 = phi i32 [ %v.21050.0, %cond_next599 ], [ %tmp344, %cond_true336 ], [ %v.2, %bb493 ]		; <i32> [#uses=1]
	%maxz.21051.1 = phi i32 [ %maxz.21051.0, %cond_next599 ], [ 0, %cond_true336 ], [ %maxz.2, %bb493 ]		; <i32> [#uses=2]
	%cnt.01055.1 = phi i32 [ %cnt.01055.0, %cond_next599 ], [ 0, %cond_true336 ], [ %cnt.0, %bb493 ]		; <i32> [#uses=2]
	%byteptr.9 = phi i8* [ %byteptr.12, %cond_next599 ], [ %byteptr.0, %cond_true336 ], [ %byteptr.10, %bb493 ]		; <i8*> [#uses=9]
	%bitptr.6 = phi i32 [ %tmp5571104.1, %cond_next599 ], [ %tmp4921049, %cond_true336 ], [ %bitptr.7, %bb493 ]		; <i32> [#uses=4]
	%source.5 = phi i32 [ %tmp602, %cond_next599 ], [ %source.0, %cond_true336 ], [ %source.6, %bb493 ]		; <i32> [#uses=7]
	%tmp606 = getelementptr %struct.const_tables* @tables, i32 0, i32 0, i32 %cnt.01055.1		; <i8*> [#uses=1]
	%tmp607 = load i8* %tmp606, align 1		; <i8> [#uses=1]

We produce this:

LBB4_70:	# cond_next603
	movl	-20(%ebp), %esi
	movl	L_tables$non_lazy_ptr-"L4$pb"(%esi), %esi

However, ICC caches this information before the loop and produces this:

        movl      88(%esp), %eax                                #481.12

//===---------------------------------------------------------------------===//

This code:

	%tmp659 = icmp slt i16 %tmp654, 0		; <i1> [#uses=1]
	br i1 %tmp659, label %cond_true662, label %cond_next715

produces this:

	testw	%cx, %cx
	movswl	%cx, %esi
	jns	LBB4_109	# cond_next715

Shark tells us that using %cx in the testw instruction is sub-optimal. It
suggests using the 32-bit register (which is what ICC uses).

//===---------------------------------------------------------------------===//

rdar://5506677 - We compile this:

define i32 @foo(double %x) {
        %x14 = bitcast double %x to i64         ; <i64> [#uses=1]
        %tmp713 = trunc i64 %x14 to i32         ; <i32> [#uses=1]
        %tmp8 = and i32 %tmp713, 2147483647             ; <i32> [#uses=1]
        ret i32 %tmp8
}

to:

_foo:
        subl    $12, %esp
        fldl    16(%esp)
        fstpl   (%esp)
        movl    $2147483647, %eax
        andl    (%esp), %eax
        addl    $12, %esp
        #FP_REG_KILL
        ret

It would be much better to eliminate the fldl/fstpl by folding the bitcast 
into the load SDNode.  That would give us:

_foo:
        movl    $2147483647, %eax
        andl    4(%esp), %eax
        ret

//===---------------------------------------------------------------------===//

We compile this:

void compare (long long foo) {
  if (foo < 4294967297LL)
    abort();
}

to:

_compare:
        subl    $12, %esp
        cmpl    $0, 16(%esp)
        setne   %al
        movzbw  %al, %ax
        cmpl    $1, 20(%esp)
        setg    %cl
        movzbw  %cl, %cx
        cmove   %ax, %cx
        movw    %cx, %ax
        testb   $1, %al
        je      LBB1_2  # cond_true

(also really horrible code on ppc).  This is due to the expand code for 64-bit
compares.  GCC produces multiple branches, which is much nicer:

_compare:
        pushl   %ebp
        movl    %esp, %ebp
        subl    $8, %esp
        movl    8(%ebp), %eax
        movl    12(%ebp), %edx
        subl    $1, %edx
        jg     L5
L7:
        jl      L4
        cmpl    $0, %eax
        jbe      L4
L5:

//===---------------------------------------------------------------------===//

Tail call optimization improvements: Tail call optimization currently
pushes all arguments on the top of the stack (their normal place for
non-tail call optimized calls) before moving them to actual stack
slot. This is done to prevent overwriting of parameters (see example
below) that might be used, since the arguments of the callee
overwrites caller's arguments.

example:  

int callee(int32, int64); 
int caller(int32 arg1, int32 arg2) { 
  int64 local = arg2 * 2; 
  return callee(arg2, (int64)local); 
}

[arg1]          [!arg2 no longer valid since we moved local onto it]
[arg2]      ->  [(int64)
[RETADDR]        local  ]

Moving arg1 onto the stack slot of callee function would overwrite
arg2 of the caller.

Possible optimizations:

 - Only push those arguments to the top of the stack that are actual
   parameters of the caller function and have no local value in the
   caller.

   In the above example local does not need to be pushed onto the top
   of the stack as it is definitely not a caller's function
   parameter.

 - Analyse the actual parameters of the callee to see which would
   overwrite a caller parameter which is used by the callee and only
   push them onto the top of the stack.

   int callee (int32 arg1, int32 arg2);
   int caller (int32 arg1, int32 arg2) {
       return callee(arg1,arg2);
   }

   Here we don't need to write any variables to the top of the stack
   since they don't overwrite each other.

   int callee (int32 arg1, int32 arg2);
   int caller (int32 arg1, int32 arg2) {
       return callee(arg2,arg1);
   }

   Here we need to push the arguments because they overwrite each
   other.


   Code for lowering directly onto callers arguments:
+  SmallVector<std::pair<unsigned, SDOperand>, 8> RegsToPass;
+  SmallVector<SDOperand, 8> MemOpChains;
+
+  SDOperand FramePtr;
+  SDOperand PtrOff;
+  SDOperand FIN;
+  int FI = 0;
+  // Walk the register/memloc assignments, inserting copies/loads.
+  for (unsigned i = 0, e = ArgLocs.size(); i != e; ++i) {
+    CCValAssign &VA = ArgLocs[i];
+    SDOperand Arg = Op.getOperand(5+2*VA.getValNo());
+    
+    ....
+    
+    if (VA.isRegLoc()) {
+      RegsToPass.push_back(std::make_pair(VA.getLocReg(), Arg));
+    } else {
+      assert(VA.isMemLoc());
+      // create frame index
+      int32_t Offset = VA.getLocMemOffset()+FPDiff;
+      uint32_t OpSize = (MVT::getSizeInBits(VA.getLocVT())+7)/8;
+      FI = MF.getFrameInfo()->CreateFixedObject(OpSize, Offset);
+      FIN = DAG.getFrameIndex(FI, MVT::i32);
+      // store relative to framepointer
+      MemOpChains.push_back(DAG.getStore(Chain, Arg, FIN, NULL, 0));
+    }
+  }
//===---------------------------------------------------------------------===//

main ()
{
  int i = 0;
  unsigned long int z = 0;

  do {
    z -= 0x00004000;
    i++;
    if (i > 0x00040000)
      abort ();
  } while (z > 0);
  exit (0);
}

gcc compiles this to:

_main:
	subl	$28, %esp
	xorl	%eax, %eax
	jmp	L2
L3:
	cmpl	$262144, %eax
	je	L10
L2:
	addl	$1, %eax
	cmpl	$262145, %eax
	jne	L3
	call	L_abort$stub
L10:
	movl	$0, (%esp)
	call	L_exit$stub

llvm:

_main:
	subl	$12, %esp
	movl	$1, %eax
	movl	$16384, %ecx
LBB1_1:	# bb
	cmpl	$262145, %eax
	jge	LBB1_4	# cond_true
LBB1_2:	# cond_next
	incl	%eax
	addl	$4294950912, %ecx
	cmpl	$16384, %ecx
	jne	LBB1_1	# bb
LBB1_3:	# bb11
	xorl	%eax, %eax
	addl	$12, %esp
	ret
LBB1_4:	# cond_true
	call	L_abort$stub

1. LSR should rewrite the first cmp with induction variable %ecx.
2. DAG combiner should fold
        leal    1(%eax), %edx
        cmpl    $262145, %edx
   =>
        cmpl    $262144, %eax

//===---------------------------------------------------------------------===//

define i64 @test(double %X) {
	%Y = fptosi double %X to i64
	ret i64 %Y
}

compiles to:

_test:
	subl	$20, %esp
	movsd	24(%esp), %xmm0
	movsd	%xmm0, 8(%esp)
	fldl	8(%esp)
	fisttpll	(%esp)
	movl	4(%esp), %edx
	movl	(%esp), %eax
	addl	$20, %esp
	#FP_REG_KILL
	ret

This should just fldl directly from the input stack slot.

//===---------------------------------------------------------------------===//

This code:
int foo (int x) { return (x & 65535) | 255; }

Should compile into:

_foo:
        movzwl  4(%esp), %eax
        orb     $-1, %al           ;; 'orl 255' is also fine :)
        ret

instead of:
_foo:
        movl    $255, %eax
        orl     4(%esp), %eax
        andl    $65535, %eax
        ret