reference, declarationdefinition
definition → references, declarations, derived classes, virtual overrides
reference to multiple definitions → definitions
unreferenced
    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
; Test the handling of base + index + displacement addresses for large frames,
; in cases where both 12-bit and 20-bit displacements are allowed.
; The tests here assume z10 register pressure, without the high words
; being available.
;
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z10 | \
; RUN:   FileCheck -check-prefix=CHECK-NOFP %s
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z10 -frame-pointer=all | \
; RUN:   FileCheck -check-prefix=CHECK-FP %s

; This file tests what happens when a displacement is converted from
; being relative to the start of a frame object to being relative to
; the frame itself.  In some cases the test is only possible if two
; objects are allocated.
;
; Rather than rely on a particular order for those objects, the tests
; instead allocate two objects of the same size and apply the test to
; both of them.  For consistency, all tests follow this model, even if
; one object would actually be enough.

; First check the highest offset that is in range of the 12-bit form.
;
; The last in-range doubleword offset is 4088.  Since the frame has two
; emergency spill slots at 160(%r15), the amount that we need to allocate
; in order to put another object at offset 4088 is 4088 - 176 = 3912 bytes.
define void @f1(i8 %byte) {
; CHECK-NOFP-LABEL: f1:
; CHECK-NOFP: stc %r2, 4095(%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f1:
; CHECK-FP: stc %r2, 4095(%r11)
; CHECK-FP: br %r14
  %region1 = alloca [3912 x i8], align 8
  %region2 = alloca [3912 x i8], align 8
  %ptr1 = getelementptr inbounds [3912 x i8], [3912 x i8]* %region1, i64 0, i64 7
  %ptr2 = getelementptr inbounds [3912 x i8], [3912 x i8]* %region2, i64 0, i64 7
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Test the first offset that is out-of-range of the 12-bit form.
define void @f2(i8 %byte) {
; CHECK-NOFP-LABEL: f2:
; CHECK-NOFP: stcy %r2, 4096(%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f2:
; CHECK-FP: stcy %r2, 4096(%r11)
; CHECK-FP: br %r14
  %region1 = alloca [3912 x i8], align 8
  %region2 = alloca [3912 x i8], align 8
  %ptr1 = getelementptr inbounds [3912 x i8], [3912 x i8]* %region1, i64 0, i64 8
  %ptr2 = getelementptr inbounds [3912 x i8], [3912 x i8]* %region2, i64 0, i64 8
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Test the last offset that is in range of the 20-bit form.
;
; The last in-range doubleword offset is 524280, so by the same reasoning
; as above, we need to allocate objects of 524280 - 176 = 524104 bytes.
define void @f3(i8 %byte) {
; CHECK-NOFP-LABEL: f3:
; CHECK-NOFP: stcy %r2, 524287(%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f3:
; CHECK-FP: stcy %r2, 524287(%r11)
; CHECK-FP: br %r14
  %region1 = alloca [524104 x i8], align 8
  %region2 = alloca [524104 x i8], align 8
  %ptr1 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region1, i64 0, i64 7
  %ptr2 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region2, i64 0, i64 7
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Test the first out-of-range offset.  We can't use an index register here,
; and the offset is also out of LAY's range, so expect a constant load
; followed by an addition.
define void @f4(i8 %byte) {
; CHECK-NOFP-LABEL: f4:
; CHECK-NOFP: llilh %r1, 8
; CHECK-NOFP: stc %r2, 0(%r1,%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f4:
; CHECK-FP: llilh %r1, 8
; CHECK-FP: stc %r2, 0(%r1,%r11)
; CHECK-FP: br %r14
  %region1 = alloca [524104 x i8], align 8
  %region2 = alloca [524104 x i8], align 8
  %ptr1 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region1, i64 0, i64 8
  %ptr2 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region2, i64 0, i64 8
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Add 4095 to the previous offset, to test the other end of the STC range.
; The instruction will actually be STCY before frame lowering.
define void @f5(i8 %byte) {
; CHECK-NOFP-LABEL: f5:
; CHECK-NOFP: llilh %r1, 8
; CHECK-NOFP: stc %r2, 4095(%r1,%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f5:
; CHECK-FP: llilh %r1, 8
; CHECK-FP: stc %r2, 4095(%r1,%r11)
; CHECK-FP: br %r14
  %region1 = alloca [524104 x i8], align 8
  %region2 = alloca [524104 x i8], align 8
  %ptr1 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region1, i64 0, i64 4103
  %ptr2 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region2, i64 0, i64 4103
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Test the next offset after that, which uses STCY instead of STC.
define void @f6(i8 %byte) {
; CHECK-NOFP-LABEL: f6:
; CHECK-NOFP: llilh %r1, 8
; CHECK-NOFP: stcy %r2, 4096(%r1,%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f6:
; CHECK-FP: llilh %r1, 8
; CHECK-FP: stcy %r2, 4096(%r1,%r11)
; CHECK-FP: br %r14
  %region1 = alloca [524104 x i8], align 8
  %region2 = alloca [524104 x i8], align 8
  %ptr1 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region1, i64 0, i64 4104
  %ptr2 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region2, i64 0, i64 4104
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Now try an offset of 524287 from the start of the object, with the
; object being at offset 1048576 (1 << 20).  The backend prefers to create
; anchors 0x10000 bytes apart, so that the high part can be loaded using
; LLILH while still using STC in more cases than 0x40000 anchors would.
define void @f7(i8 %byte) {
; CHECK-NOFP-LABEL: f7:
; CHECK-NOFP: llilh %r1, 23
; CHECK-NOFP: stcy %r2, 65535(%r1,%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f7:
; CHECK-FP: llilh %r1, 23
; CHECK-FP: stcy %r2, 65535(%r1,%r11)
; CHECK-FP: br %r14
  %region1 = alloca [1048400 x i8], align 8
  %region2 = alloca [1048400 x i8], align 8
  %ptr1 = getelementptr inbounds [1048400 x i8], [1048400 x i8]* %region1, i64 0, i64 524287
  %ptr2 = getelementptr inbounds [1048400 x i8], [1048400 x i8]* %region2, i64 0, i64 524287
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Keep the object-relative offset the same but bump the size of the
; objects by one doubleword.
define void @f8(i8 %byte) {
; CHECK-NOFP-LABEL: f8:
; CHECK-NOFP: llilh %r1, 24
; CHECK-NOFP: stc %r2, 7(%r1,%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f8:
; CHECK-FP: llilh %r1, 24
; CHECK-FP: stc %r2, 7(%r1,%r11)
; CHECK-FP: br %r14
  %region1 = alloca [1048408 x i8], align 8
  %region2 = alloca [1048408 x i8], align 8
  %ptr1 = getelementptr inbounds [1048408 x i8], [1048408 x i8]* %region1, i64 0, i64 524287
  %ptr2 = getelementptr inbounds [1048408 x i8], [1048408 x i8]* %region2, i64 0, i64 524287
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Check a case where the original displacement is out of range.  The backend
; should force separate address logic from the outset.  We don't yet do any
; kind of anchor optimization, so there should be no offset on the STC itself.
;
; Before frame lowering this is an LA followed by the AGFI seen below.
; The LA then gets lowered into the LLILH/LA form.  The exact sequence
; isn't that important though.
define void @f9(i8 %byte) {
; CHECK-NOFP-LABEL: f9:
; CHECK-NOFP: llilh [[R1:%r[1-5]]], 16
; CHECK-NOFP: la [[R2:%r[1-5]]], 8([[R1]],%r15)
; CHECK-NOFP: agfi [[R2]], 524288
; CHECK-NOFP: stc %r2, 0([[R2]])
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f9:
; CHECK-FP: llilh [[R1:%r[1-5]]], 16
; CHECK-FP: la [[R2:%r[1-5]]], 8([[R1]],%r11)
; CHECK-FP: agfi [[R2]], 524288
; CHECK-FP: stc %r2, 0([[R2]])
; CHECK-FP: br %r14
  %region1 = alloca [1048408 x i8], align 8
  %region2 = alloca [1048408 x i8], align 8
  %ptr1 = getelementptr inbounds [1048408 x i8], [1048408 x i8]* %region1, i64 0, i64 524288
  %ptr2 = getelementptr inbounds [1048408 x i8], [1048408 x i8]* %region2, i64 0, i64 524288
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}

; Repeat f4 in a case that needs the emergency spill slots (because all
; call-clobbered registers are live and no call-saved ones have been
; allocated).
define void @f10(i32 *%vptr, i8 %byte) {
; CHECK-NOFP-LABEL: f10:
; CHECK-NOFP: stg [[REGISTER:%r[1-9][0-4]?]], [[OFFSET:160|168]](%r15)
; CHECK-NOFP: llilh [[REGISTER]], 8
; CHECK-NOFP: stc %r3, 0([[REGISTER]],%r15)
; CHECK-NOFP: lg [[REGISTER]], [[OFFSET]](%r15)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f10:
; CHECK-FP: stg [[REGISTER:%r[1-9][0-4]?]], [[OFFSET:160|168]](%r11)
; CHECK-FP: llilh [[REGISTER]], 8
; CHECK-FP: stc %r3, 0([[REGISTER]],%r11)
; CHECK-FP: lg [[REGISTER]], [[OFFSET]](%r11)
; CHECK-FP: br %r14
  %i0 = load volatile i32, i32 *%vptr
  %i1 = load volatile i32, i32 *%vptr
  %i4 = load volatile i32, i32 *%vptr
  %i5 = load volatile i32, i32 *%vptr
  %region1 = alloca [524104 x i8], align 8
  %region2 = alloca [524104 x i8], align 8
  %ptr1 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region1, i64 0, i64 8
  %ptr2 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region2, i64 0, i64 8
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  store volatile i32 %i0, i32 *%vptr
  store volatile i32 %i1, i32 *%vptr
  store volatile i32 %i4, i32 *%vptr
  store volatile i32 %i5, i32 *%vptr
  ret void
}

; And again with maximum register pressure.  The only spill slots that the
; NOFP case needs are the emergency ones, so the offsets are the same as for f4.
; However, the FP case uses %r11 as the frame pointer and must therefore
; spill a second register.  This leads to an extra displacement of 8.
define void @f11(i32 *%vptr, i8 %byte) {
; CHECK-NOFP-LABEL: f11:
; CHECK-NOFP: stmg %r6, %r15,
; CHECK-NOFP: stg [[REGISTER:%r[1-9][0-4]?]], [[OFFSET:160|168]](%r15)
; CHECK-NOFP: llilh [[REGISTER]], 8
; CHECK-NOFP: stc %r3, 0([[REGISTER]],%r15)
; CHECK-NOFP: lg [[REGISTER]], [[OFFSET]](%r15)
; CHECK-NOFP: lmg %r6, %r15,
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f11:
; CHECK-FP: stmg %r6, %r15,
; CHECK-FP: stg [[REGISTER:%r[1-9][0-4]?]], [[OFFSET:160|168]](%r11)
; CHECK-FP: llilh [[REGISTER]], 8
; CHECK-FP: stc %r3, 8([[REGISTER]],%r11)
; CHECK-FP: lg [[REGISTER]], [[OFFSET]](%r11)
; CHECK-FP: lmg %r6, %r15,
; CHECK-FP: br %r14
  %i0 = load volatile i32, i32 *%vptr
  %i1 = load volatile i32, i32 *%vptr
  %i4 = load volatile i32, i32 *%vptr
  %i5 = load volatile i32, i32 *%vptr
  %i6 = load volatile i32, i32 *%vptr
  %i7 = load volatile i32, i32 *%vptr
  %i8 = load volatile i32, i32 *%vptr
  %i9 = load volatile i32, i32 *%vptr
  %i10 = load volatile i32, i32 *%vptr
  %i11 = load volatile i32, i32 *%vptr
  %i12 = load volatile i32, i32 *%vptr
  %i13 = load volatile i32, i32 *%vptr
  %i14 = load volatile i32, i32 *%vptr
  %region1 = alloca [524104 x i8], align 8
  %region2 = alloca [524104 x i8], align 8
  %ptr1 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region1, i64 0, i64 8
  %ptr2 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region2, i64 0, i64 8
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  store volatile i32 %i0, i32 *%vptr
  store volatile i32 %i1, i32 *%vptr
  store volatile i32 %i4, i32 *%vptr
  store volatile i32 %i5, i32 *%vptr
  store volatile i32 %i6, i32 *%vptr
  store volatile i32 %i7, i32 *%vptr
  store volatile i32 %i8, i32 *%vptr
  store volatile i32 %i9, i32 *%vptr
  store volatile i32 %i10, i32 *%vptr
  store volatile i32 %i11, i32 *%vptr
  store volatile i32 %i12, i32 *%vptr
  store volatile i32 %i13, i32 *%vptr
  store volatile i32 %i14, i32 *%vptr
  ret void
}

; Repeat f4 in a case where the index register is already occupied.
define void @f12(i8 %byte, i64 %index) {
; CHECK-NOFP-LABEL: f12:
; CHECK-NOFP: llilh %r1, 8
; CHECK-NOFP: agr %r1, %r15
; CHECK-NOFP: stc %r2, 0(%r3,%r1)
; CHECK-NOFP: br %r14
;
; CHECK-FP-LABEL: f12:
; CHECK-FP: llilh %r1, 8
; CHECK-FP: agr %r1, %r11
; CHECK-FP: stc %r2, 0(%r3,%r1)
; CHECK-FP: br %r14
  %region1 = alloca [524104 x i8], align 8
  %region2 = alloca [524104 x i8], align 8
  %index1 = add i64 %index, 8
  %ptr1 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region1, i64 0, i64 %index1
  %ptr2 = getelementptr inbounds [524104 x i8], [524104 x i8]* %region2, i64 0, i64 %index1
  store volatile i8 %byte, i8 *%ptr1
  store volatile i8 %byte, i8 *%ptr2
  ret void
}