Skip to content

Commit 90ba330

Browse files
nikicnikic
authored
[InstCombine] Canonicalize constant GEPs to i8 source element type (#68882)
This patch canonicalizes getelementptr instructions with constant indices to use the `i8` source element type. This makes it easier for optimizations to recognize that two GEPs are identical, because they don't need to see past many different ways to express the same offset. This is a first step towards https://discourse.llvm.org/t/rfc-replacing-getelementptr-with-ptradd/68699. This is limited to constant GEPs only for now, as they have a clear canonical form, while we're not yet sure how exactly to deal with variable indices. The test llvm/test/Transforms/PhaseOrdering/switch_with_geps.ll gives two representative examples of the kind of optimization improvement we expect from this change. In the first test SimplifyCFG can now realize that all switch branches are actually the same. In the second test it can convert it into simple arithmetic. These are representative of common optimization failures we see in Rust. Fixes #69841.
1 parent 98509c7 commit 90ba330

File tree

176 files changed

+2520
-2532
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

176 files changed

+2520
-2532
lines changed

‎clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -25,13 +25,13 @@ void test1(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi
2525
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 0
2626
// CHECK-NEXT: store <16 x i8> [[TMP2]], ptr [[RESP:%.*]], align 16
2727
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 1
28-
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 1
28+
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 16
2929
// CHECK-NEXT: store <16 x i8> [[TMP3]], ptr [[TMP4]], align 16
3030
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 2
31-
// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 2
31+
// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 32
3232
// CHECK-NEXT: store <16 x i8> [[TMP5]], ptr [[TMP6]], align 16
3333
// CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 3
34-
// CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 3
34+
// CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 48
3535
// CHECK-NEXT: store <16 x i8> [[TMP7]], ptr [[TMP8]], align 16
3636
// CHECK-NEXT: ret void
3737
//
@@ -60,7 +60,7 @@ void test3(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi
6060
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 0
6161
// CHECK-NEXT: store <16 x i8> [[TMP2]], ptr [[RESP:%.*]], align 16
6262
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 1
63-
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 1
63+
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 16
6464
// CHECK-NEXT: store <16 x i8> [[TMP3]], ptr [[TMP4]], align 16
6565
// CHECK-NEXT: ret void
6666
//
@@ -1072,7 +1072,7 @@ void test76(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns
10721072
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 0
10731073
// CHECK-NEXT: store <16 x i8> [[TMP2]], ptr [[RESP:%.*]], align 16
10741074
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 1
1075-
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 1
1075+
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 16
10761076
// CHECK-NEXT: store <16 x i8> [[TMP3]], ptr [[TMP4]], align 16
10771077
// CHECK-NEXT: ret void
10781078
//

‎clang/test/CodeGen/aarch64-ls64-inline-asm.c

Lines changed: 9 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -16,8 +16,8 @@ void load(struct foo *output, void *addr)
1616

1717
// CHECK-LABEL: @store(
1818
// CHECK-NEXT: entry:
19-
// CHECK-NEXT: [[TMP1:%.*]] = load i512, ptr [[INPUT:%.*]], align 8
20-
// CHECK-NEXT: tail call void asm sideeffect "st64b $0,[$1]", "r,r,~{memory}"(i512 [[TMP1]], ptr [[ADDR:%.*]]) #[[ATTR1]], !srcloc !3
19+
// CHECK-NEXT: [[TMP0:%.*]] = load i512, ptr [[INPUT:%.*]], align 8
20+
// CHECK-NEXT: tail call void asm sideeffect "st64b $0,[$1]", "r,r,~{memory}"(i512 [[TMP0]], ptr [[ADDR:%.*]]) #[[ATTR1]], !srcloc !3
2121
// CHECK-NEXT: ret void
2222
//
2323
void store(const struct foo *input, void *addr)
@@ -29,25 +29,25 @@ void store(const struct foo *input, void *addr)
2929
// CHECK-NEXT: entry:
3030
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[IN:%.*]], align 4, !tbaa [[TBAA4:![0-9]+]]
3131
// CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP0]] to i64
32-
// CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 1
32+
// CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 4
3333
// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4, !tbaa [[TBAA4]]
3434
// CHECK-NEXT: [[CONV2:%.*]] = sext i32 [[TMP1]] to i64
35-
// CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 4
35+
// CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 16
3636
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4, !tbaa [[TBAA4]]
3737
// CHECK-NEXT: [[CONV5:%.*]] = sext i32 [[TMP2]] to i64
38-
// CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 16
38+
// CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 64
3939
// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX7]], align 4, !tbaa [[TBAA4]]
4040
// CHECK-NEXT: [[CONV8:%.*]] = sext i32 [[TMP3]] to i64
41-
// CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 25
41+
// CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 100
4242
// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX10]], align 4, !tbaa [[TBAA4]]
4343
// CHECK-NEXT: [[CONV11:%.*]] = sext i32 [[TMP4]] to i64
44-
// CHECK-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 36
44+
// CHECK-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 144
4545
// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX13]], align 4, !tbaa [[TBAA4]]
4646
// CHECK-NEXT: [[CONV14:%.*]] = sext i32 [[TMP5]] to i64
47-
// CHECK-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 49
47+
// CHECK-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 196
4848
// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX16]], align 4, !tbaa [[TBAA4]]
4949
// CHECK-NEXT: [[CONV17:%.*]] = sext i32 [[TMP6]] to i64
50-
// CHECK-NEXT: [[ARRAYIDX19:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 64
50+
// CHECK-NEXT: [[ARRAYIDX19:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 256
5151
// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX19]], align 4, !tbaa [[TBAA4]]
5252
// CHECK-NEXT: [[CONV20:%.*]] = sext i32 [[TMP7]] to i64
5353
// CHECK-NEXT: [[S_SROA_10_0_INSERT_EXT:%.*]] = zext i64 [[CONV20]] to i512

‎clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c

Lines changed: 24 additions & 24 deletions
Original file line numberDiff line numberDiff line change
@@ -30,21 +30,21 @@ DEFINE_STRUCT(bool)
3030

3131
// CHECK-128-LABEL: @read_int64(
3232
// CHECK-128-NEXT: entry:
33-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
33+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
3434
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 16, !tbaa [[TBAA2:![0-9]+]]
3535
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP0]], i64 0)
3636
// CHECK-128-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]]
3737
//
3838
// CHECK-256-LABEL: @read_int64(
3939
// CHECK-256-NEXT: entry:
40-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
40+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
4141
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 16, !tbaa [[TBAA2:![0-9]+]]
4242
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v4i64(<vscale x 2 x i64> undef, <4 x i64> [[TMP0]], i64 0)
4343
// CHECK-256-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]]
4444
//
4545
// CHECK-512-LABEL: @read_int64(
4646
// CHECK-512-NEXT: entry:
47-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
47+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
4848
// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i64>, ptr [[Y]], align 16, !tbaa [[TBAA2:![0-9]+]]
4949
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v8i64(<vscale x 2 x i64> undef, <8 x i64> [[TMP0]], i64 0)
5050
// CHECK-512-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]]
@@ -56,21 +56,21 @@ svint64_t read_int64(struct struct_int64 *s) {
5656
// CHECK-128-LABEL: @write_int64(
5757
// CHECK-128-NEXT: entry:
5858
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i64> @llvm.vector.extract.v2i64.nxv2i64(<vscale x 2 x i64> [[X:%.*]], i64 0)
59-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
59+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
6060
// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
6161
// CHECK-128-NEXT: ret void
6262
//
6363
// CHECK-256-LABEL: @write_int64(
6464
// CHECK-256-NEXT: entry:
6565
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i64> @llvm.vector.extract.v4i64.nxv2i64(<vscale x 2 x i64> [[X:%.*]], i64 0)
66-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
66+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
6767
// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
6868
// CHECK-256-NEXT: ret void
6969
//
7070
// CHECK-512-LABEL: @write_int64(
7171
// CHECK-512-NEXT: entry:
7272
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i64> @llvm.vector.extract.v8i64.nxv2i64(<vscale x 2 x i64> [[X:%.*]], i64 0)
73-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
73+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
7474
// CHECK-512-NEXT: store <8 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
7575
// CHECK-512-NEXT: ret void
7676
//
@@ -84,21 +84,21 @@ void write_int64(struct struct_int64 *s, svint64_t x) {
8484

8585
// CHECK-128-LABEL: @read_float64(
8686
// CHECK-128-NEXT: entry:
87-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
87+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
8888
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
8989
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v2f64(<vscale x 2 x double> undef, <2 x double> [[TMP0]], i64 0)
9090
// CHECK-128-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]]
9191
//
9292
// CHECK-256-LABEL: @read_float64(
9393
// CHECK-256-NEXT: entry:
94-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
94+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
9595
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
9696
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v4f64(<vscale x 2 x double> undef, <4 x double> [[TMP0]], i64 0)
9797
// CHECK-256-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]]
9898
//
9999
// CHECK-512-LABEL: @read_float64(
100100
// CHECK-512-NEXT: entry:
101-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
101+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
102102
// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x double>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
103103
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v8f64(<vscale x 2 x double> undef, <8 x double> [[TMP0]], i64 0)
104104
// CHECK-512-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]]
@@ -110,21 +110,21 @@ svfloat64_t read_float64(struct struct_float64 *s) {
110110
// CHECK-128-LABEL: @write_float64(
111111
// CHECK-128-NEXT: entry:
112112
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x double> @llvm.vector.extract.v2f64.nxv2f64(<vscale x 2 x double> [[X:%.*]], i64 0)
113-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
113+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
114114
// CHECK-128-NEXT: store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
115115
// CHECK-128-NEXT: ret void
116116
//
117117
// CHECK-256-LABEL: @write_float64(
118118
// CHECK-256-NEXT: entry:
119119
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x double> @llvm.vector.extract.v4f64.nxv2f64(<vscale x 2 x double> [[X:%.*]], i64 0)
120-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
120+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
121121
// CHECK-256-NEXT: store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
122122
// CHECK-256-NEXT: ret void
123123
//
124124
// CHECK-512-LABEL: @write_float64(
125125
// CHECK-512-NEXT: entry:
126126
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x double> @llvm.vector.extract.v8f64.nxv2f64(<vscale x 2 x double> [[X:%.*]], i64 0)
127-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
127+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
128128
// CHECK-512-NEXT: store <8 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
129129
// CHECK-512-NEXT: ret void
130130
//
@@ -138,21 +138,21 @@ void write_float64(struct struct_float64 *s, svfloat64_t x) {
138138

139139
// CHECK-128-LABEL: @read_bfloat16(
140140
// CHECK-128-NEXT: entry:
141-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
141+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
142142
// CHECK-128-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
143143
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v8bf16(<vscale x 8 x bfloat> undef, <8 x bfloat> [[TMP0]], i64 0)
144144
// CHECK-128-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]]
145145
//
146146
// CHECK-256-LABEL: @read_bfloat16(
147147
// CHECK-256-NEXT: entry:
148-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
148+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
149149
// CHECK-256-NEXT: [[TMP0:%.*]] = load <16 x bfloat>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
150150
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v16bf16(<vscale x 8 x bfloat> undef, <16 x bfloat> [[TMP0]], i64 0)
151151
// CHECK-256-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]]
152152
//
153153
// CHECK-512-LABEL: @read_bfloat16(
154154
// CHECK-512-NEXT: entry:
155-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
155+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
156156
// CHECK-512-NEXT: [[TMP0:%.*]] = load <32 x bfloat>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
157157
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v32bf16(<vscale x 8 x bfloat> undef, <32 x bfloat> [[TMP0]], i64 0)
158158
// CHECK-512-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]]
@@ -164,21 +164,21 @@ svbfloat16_t read_bfloat16(struct struct_bfloat16 *s) {
164164
// CHECK-128-LABEL: @write_bfloat16(
165165
// CHECK-128-NEXT: entry:
166166
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x bfloat> @llvm.vector.extract.v8bf16.nxv8bf16(<vscale x 8 x bfloat> [[X:%.*]], i64 0)
167-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
167+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
168168
// CHECK-128-NEXT: store <8 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
169169
// CHECK-128-NEXT: ret void
170170
//
171171
// CHECK-256-LABEL: @write_bfloat16(
172172
// CHECK-256-NEXT: entry:
173173
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <16 x bfloat> @llvm.vector.extract.v16bf16.nxv8bf16(<vscale x 8 x bfloat> [[X:%.*]], i64 0)
174-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
174+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
175175
// CHECK-256-NEXT: store <16 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
176176
// CHECK-256-NEXT: ret void
177177
//
178178
// CHECK-512-LABEL: @write_bfloat16(
179179
// CHECK-512-NEXT: entry:
180180
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <32 x bfloat> @llvm.vector.extract.v32bf16.nxv8bf16(<vscale x 8 x bfloat> [[X:%.*]], i64 0)
181-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
181+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
182182
// CHECK-512-NEXT: store <32 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
183183
// CHECK-512-NEXT: ret void
184184
//
@@ -192,23 +192,23 @@ void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) {
192192

193193
// CHECK-128-LABEL: @read_bool(
194194
// CHECK-128-NEXT: entry:
195-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
195+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 2
196196
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr [[Y]], align 2, !tbaa [[TBAA2]]
197197
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v2i8(<vscale x 2 x i8> undef, <2 x i8> [[TMP0]], i64 0)
198198
// CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1>
199199
// CHECK-128-NEXT: ret <vscale x 16 x i1> [[TMP1]]
200200
//
201201
// CHECK-256-LABEL: @read_bool(
202202
// CHECK-256-NEXT: entry:
203-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
203+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 4
204204
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[Y]], align 2, !tbaa [[TBAA2]]
205205
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v4i8(<vscale x 2 x i8> undef, <4 x i8> [[TMP0]], i64 0)
206206
// CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1>
207207
// CHECK-256-NEXT: ret <vscale x 16 x i1> [[TMP1]]
208208
//
209209
// CHECK-512-LABEL: @read_bool(
210210
// CHECK-512-NEXT: entry:
211-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
211+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
212212
// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[Y]], align 2, !tbaa [[TBAA2]]
213213
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> undef, <8 x i8> [[TMP0]], i64 0)
214214
// CHECK-512-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1>
@@ -222,23 +222,23 @@ svbool_t read_bool(struct struct_bool *s) {
222222
// CHECK-128-NEXT: entry:
223223
// CHECK-128-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X:%.*]] to <vscale x 2 x i8>
224224
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i8> @llvm.vector.extract.v2i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0)
225-
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
225+
// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 2
226226
// CHECK-128-NEXT: store <2 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[TBAA2]]
227227
// CHECK-128-NEXT: ret void
228228
//
229229
// CHECK-256-LABEL: @write_bool(
230230
// CHECK-256-NEXT: entry:
231231
// CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X:%.*]] to <vscale x 2 x i8>
232232
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i8> @llvm.vector.extract.v4i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0)
233-
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
233+
// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 4
234234
// CHECK-256-NEXT: store <4 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[TBAA2]]
235235
// CHECK-256-NEXT: ret void
236236
//
237237
// CHECK-512-LABEL: @write_bool(
238238
// CHECK-512-NEXT: entry:
239239
// CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X:%.*]] to <vscale x 2 x i8>
240240
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0)
241-
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
241+
// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
242242
// CHECK-512-NEXT: store <8 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[TBAA2]]
243243
// CHECK-512-NEXT: ret void
244244
//

0 commit comments

Comments
 (0)