Skip to content

Commit

Permalink
[AMDGPU][GlobalIsel] Use isRegisterClassType for G_FREEZE and G_IMPLI…
Browse files Browse the repository at this point in the history 
…CIT_DEF (#101331)

G_FREEZE was legal for <13 x S32> which caused an infinite loop in the
combiner
  • Loading branch information
@sstipano
sstipano authored Sep 24, 2024
1 parent cde7b30 commit 63b2595
Show file tree
Hide file tree
Showing 8 changed files with 1,968 additions and 157 deletions.
8 changes: 5 additions & 3 deletions llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -292,6 +292,7 @@ static const LLT S160 = LLT::scalar(160);
static const LLT S224 = LLT::scalar(224);
static const LLT S256 = LLT::scalar(256);
static const LLT S512 = LLT::scalar(512);
static const LLT S1024 = LLT::scalar(1024);
static const LLT MaxScalar = LLT::scalar(MaxRegisterSize);

static const LLT V2S8 = LLT::fixed_vector(2, 8);
Expand Down Expand Up @@ -332,8 +333,8 @@ static const LLT V16S64 = LLT::fixed_vector(16, 64);
static const LLT V2S128 = LLT::fixed_vector(2, 128);
static const LLT V4S128 = LLT::fixed_vector(4, 128);

static std::initializer_list<LLT> AllScalarTypes = {S32, S64, S96, S128,
S160, S224, S256, S512};
static std::initializer_list<LLT> AllScalarTypes = {
S32, S64, S96, S128, S160, S224, S256, S512, S1024};

static std::initializer_list<LLT> AllS16Vectors{
V2S16, V4S16, V6S16, V8S16, V10S16, V12S16, V16S16, V2S128, V4S128};
Expand Down Expand Up @@ -889,10 +890,11 @@ AMDGPULegalizerInfo::AMDGPULegalizerInfo(const GCNSubtarget &ST_,
.clampScalar(0, S16, S64);

getActionDefinitionsBuilder({G_IMPLICIT_DEF, G_FREEZE})
.legalIf(isRegisterType(0))
.legalIf(isRegisterClassType(0))
// s1 and s16 are special cases because they have legal operations on
// them, but don't really occupy registers in the normal way.
.legalFor({S1, S16})
.clampNumElements(0, V16S32, V32S32)
.moreElementsIf(isSmallOddVector(0), oneMoreElement(0))
.clampScalarOrElt(0, S32, MaxScalar)
.widenScalarToNextPow2(0, 32)
Expand Down
18 changes: 8 additions & 10 deletions llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-unmerge-values.mir
View file
Original file line number Diff line number Diff line change
Expand Up @@ -171,11 +171,9 @@ body: |
; GCN-LABEL: name: test_unmerge_values_s_s64_s_s64_s64_s_s192
; GCN: liveins: $sgpr0_sgpr1_sgpr2_sgpr3
; GCN-NEXT: {{ $}}
; GCN-NEXT: [[DEF:%[0-9]+]]:sgpr_192 = IMPLICIT_DEF
; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY [[DEF]].sub0_sub1
; GCN-NEXT: [[COPY1:%[0-9]+]]:sreg_64 = COPY [[DEF]].sub2_sub3
; GCN-NEXT: [[COPY2:%[0-9]+]]:sreg_64 = COPY [[DEF]].sub4_sub5
; GCN-NEXT: S_ENDPGM 0, implicit [[COPY]], implicit [[COPY1]], implicit [[COPY2]]
; GCN-NEXT: [[DEF:%[0-9]+]]:sgpr(s192) = G_IMPLICIT_DEF
; GCN-NEXT: [[UV:%[0-9]+]]:sgpr(s64), [[UV1:%[0-9]+]]:sgpr(s64), [[UV2:%[0-9]+]]:sgpr(s64) = G_UNMERGE_VALUES [[DEF]](s192)
; GCN-NEXT: S_ENDPGM 0, implicit [[UV]](s64), implicit [[UV1]](s64), implicit [[UV2]](s64)
%0:sgpr(s192) = G_IMPLICIT_DEF
%1:sgpr(s64), %2:sgpr(s64), %3:sgpr(s64) = G_UNMERGE_VALUES %0
S_ENDPGM 0, implicit %1, implicit %2, implicit %3
Expand Down Expand Up @@ -294,11 +292,11 @@ body: |
; GCN-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:sgpr_384(<12 x s32>) = G_CONCAT_VECTORS [[COPY]](<3 x s32>), [[COPY1]](<3 x s32>), [[COPY2]](<3 x s32>), [[COPY3]](<3 x s32>)
; GCN-NEXT: [[COPY4:%[0-9]+]]:sgpr_96(<3 x s32>) = COPY [[CONCAT_VECTORS]].sub0_sub1_sub2(<12 x s32>)
; GCN-NEXT: [[COPY5:%[0-9]+]]:sgpr_96(<3 x s32>) = COPY [[CONCAT_VECTORS]].sub3_sub4_sub5(<12 x s32>)
; GCN-NEXT: [[UV:%[0-9]+]]:sgpr_96(<3 x s32>), [[UV1:%[0-9]+]]:sgpr_96(<3 x s32>), [[UV2:%[0-9]+]]:sgpr_96(<3 x s32>), [[UV3:%[0-9]+]]:sgpr_96(<3 x s32>) = G_UNMERGE_VALUES [[CONCAT_VECTORS]](<12 x s32>)
; GCN-NEXT: $sgpr0_sgpr1_sgpr2 = COPY [[UV]](<3 x s32>)
; GCN-NEXT: $sgpr4_sgpr5_sgpr6 = COPY [[UV1]](<3 x s32>)
; GCN-NEXT: $sgpr8_sgpr9_sgpr10 = COPY [[UV2]](<3 x s32>)
; GCN-NEXT: $sgpr12_sgpr13_sgpr14 = COPY [[UV3]](<3 x s32>)
; GCN-NEXT: [[COPY4:%[0-9]+]]:sgpr_96(<3 x s32>), [[COPY5:%[0-9]+]]:sgpr_96(<3 x s32>), [[UV:%[0-9]+]]:sgpr_96(<3 x s32>), [[UV1:%[0-9]+]]:sgpr_96(<3 x s32>) = G_UNMERGE_VALUES [[CONCAT_VECTORS]](<12 x s32>)
; GCN-NEXT: $sgpr0_sgpr1_sgpr2 = COPY [[COPY4]](<3 x s32>)
; GCN-NEXT: $sgpr4_sgpr5_sgpr6 = COPY [[COPY5]](<3 x s32>)
; GCN-NEXT: $sgpr8_sgpr9_sgpr10 = COPY [[UV]](<3 x s32>)
; GCN-NEXT: $sgpr12_sgpr13_sgpr14 = COPY [[UV1]](<3 x s32>)
%0:sgpr(<3 x s32>) = COPY $sgpr0_sgpr1_sgpr2
%1:sgpr(<3 x s32>) = COPY $sgpr4_sgpr5_sgpr6
%2:sgpr(<3 x s32>) = COPY $sgpr8_sgpr9_sgpr10
Expand Down
30 changes: 11 additions & 19 deletions llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-freeze.mir
View file
Original file line number Diff line number Diff line change
Expand Up @@ -171,12 +171,8 @@ body: |
; CHECK-LABEL: name: test_freeze_s448
; CHECK: [[COPY:%[0-9]+]]:_(s512) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
; CHECK-NEXT: [[TRUNC:%[0-9]+]]:_(s448) = G_TRUNC [[COPY]](s512)
; CHECK-NEXT: [[FREEZE:%[0-9]+]]:_(s448) = G_FREEZE [[TRUNC]]
; CHECK-NEXT: [[UV:%[0-9]+]]:_(s64), [[UV1:%[0-9]+]]:_(s64), [[UV2:%[0-9]+]]:_(s64), [[UV3:%[0-9]+]]:_(s64), [[UV4:%[0-9]+]]:_(s64), [[UV5:%[0-9]+]]:_(s64), [[UV6:%[0-9]+]]:_(s64) = G_UNMERGE_VALUES [[FREEZE]](s448)
; CHECK-NEXT: [[DEF:%[0-9]+]]:_(s64) = G_IMPLICIT_DEF
; CHECK-NEXT: [[MV:%[0-9]+]]:_(s512) = G_MERGE_VALUES [[UV]](s64), [[UV1]](s64), [[UV2]](s64), [[UV3]](s64), [[UV4]](s64), [[UV5]](s64), [[UV6]](s64), [[DEF]](s64)
; CHECK-NEXT: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[MV]](s512)
; CHECK-NEXT: [[FREEZE:%[0-9]+]]:_(s512) = G_FREEZE [[COPY]]
; CHECK-NEXT: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[FREEZE]](s512)
%0:_(s512) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
%1:_(s448) = G_TRUNC %0
%2:_(s448) = G_FREEZE %1
Expand Down Expand Up @@ -399,14 +395,12 @@ body: |
bb.0:
; CHECK-LABEL: name: test_freeze_v33s32
; CHECK: [[DEF:%[0-9]+]]:_(<16 x s32>) = G_IMPLICIT_DEF
; CHECK: [[DEF:%[0-9]+]]:_(<32 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; CHECK-NEXT: [[FREEZE:%[0-9]+]]:_(<16 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[FREEZE1:%[0-9]+]]:_(<16 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[FREEZE2:%[0-9]+]]:_(s32) = G_FREEZE [[DEF1]]
; CHECK-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32), [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32), [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32), [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32), [[UV12:%[0-9]+]]:_(s32), [[UV13:%[0-9]+]]:_(s32), [[UV14:%[0-9]+]]:_(s32), [[UV15:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FREEZE]](<16 x s32>)
; CHECK-NEXT: [[UV16:%[0-9]+]]:_(s32), [[UV17:%[0-9]+]]:_(s32), [[UV18:%[0-9]+]]:_(s32), [[UV19:%[0-9]+]]:_(s32), [[UV20:%[0-9]+]]:_(s32), [[UV21:%[0-9]+]]:_(s32), [[UV22:%[0-9]+]]:_(s32), [[UV23:%[0-9]+]]:_(s32), [[UV24:%[0-9]+]]:_(s32), [[UV25:%[0-9]+]]:_(s32), [[UV26:%[0-9]+]]:_(s32), [[UV27:%[0-9]+]]:_(s32), [[UV28:%[0-9]+]]:_(s32), [[UV29:%[0-9]+]]:_(s32), [[UV30:%[0-9]+]]:_(s32), [[UV31:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FREEZE1]](<16 x s32>)
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<33 x s32>) = G_BUILD_VECTOR [[UV]](s32), [[UV1]](s32), [[UV2]](s32), [[UV3]](s32), [[UV4]](s32), [[UV5]](s32), [[UV6]](s32), [[UV7]](s32), [[UV8]](s32), [[UV9]](s32), [[UV10]](s32), [[UV11]](s32), [[UV12]](s32), [[UV13]](s32), [[UV14]](s32), [[UV15]](s32), [[UV16]](s32), [[UV17]](s32), [[UV18]](s32), [[UV19]](s32), [[UV20]](s32), [[UV21]](s32), [[UV22]](s32), [[UV23]](s32), [[UV24]](s32), [[UV25]](s32), [[UV26]](s32), [[UV27]](s32), [[UV28]](s32), [[UV29]](s32), [[UV30]](s32), [[UV31]](s32), [[FREEZE2]](s32)
; CHECK-NEXT: [[FREEZE:%[0-9]+]]:_(<32 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[FREEZE1:%[0-9]+]]:_(s32) = G_FREEZE [[DEF1]]
; CHECK-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32), [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32), [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32), [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32), [[UV12:%[0-9]+]]:_(s32), [[UV13:%[0-9]+]]:_(s32), [[UV14:%[0-9]+]]:_(s32), [[UV15:%[0-9]+]]:_(s32), [[UV16:%[0-9]+]]:_(s32), [[UV17:%[0-9]+]]:_(s32), [[UV18:%[0-9]+]]:_(s32), [[UV19:%[0-9]+]]:_(s32), [[UV20:%[0-9]+]]:_(s32), [[UV21:%[0-9]+]]:_(s32), [[UV22:%[0-9]+]]:_(s32), [[UV23:%[0-9]+]]:_(s32), [[UV24:%[0-9]+]]:_(s32), [[UV25:%[0-9]+]]:_(s32), [[UV26:%[0-9]+]]:_(s32), [[UV27:%[0-9]+]]:_(s32), [[UV28:%[0-9]+]]:_(s32), [[UV29:%[0-9]+]]:_(s32), [[UV30:%[0-9]+]]:_(s32), [[UV31:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FREEZE]](<32 x s32>)
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<33 x s32>) = G_BUILD_VECTOR [[UV]](s32), [[UV1]](s32), [[UV2]](s32), [[UV3]](s32), [[UV4]](s32), [[UV5]](s32), [[UV6]](s32), [[UV7]](s32), [[UV8]](s32), [[UV9]](s32), [[UV10]](s32), [[UV11]](s32), [[UV12]](s32), [[UV13]](s32), [[UV14]](s32), [[UV15]](s32), [[UV16]](s32), [[UV17]](s32), [[UV18]](s32), [[UV19]](s32), [[UV20]](s32), [[UV21]](s32), [[UV22]](s32), [[UV23]](s32), [[UV24]](s32), [[UV25]](s32), [[UV26]](s32), [[UV27]](s32), [[UV28]](s32), [[UV29]](s32), [[UV30]](s32), [[UV31]](s32), [[FREEZE1]](s32)
; CHECK-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<33 x s32>)
%0:_(<33 x s32>) = G_IMPLICIT_DEF
%1:_(<33 x s32>) = G_FREEZE %0
Expand All @@ -419,12 +413,10 @@ body: |
bb.0:
; CHECK-LABEL: name: test_freeze_v64s32
; CHECK: [[DEF:%[0-9]+]]:_(<16 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[FREEZE:%[0-9]+]]:_(<16 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[FREEZE1:%[0-9]+]]:_(<16 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[FREEZE2:%[0-9]+]]:_(<16 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[FREEZE3:%[0-9]+]]:_(<16 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<64 x s32>) = G_CONCAT_VECTORS [[FREEZE]](<16 x s32>), [[FREEZE1]](<16 x s32>), [[FREEZE2]](<16 x s32>), [[FREEZE3]](<16 x s32>)
; CHECK: [[DEF:%[0-9]+]]:_(<32 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[FREEZE:%[0-9]+]]:_(<32 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[FREEZE1:%[0-9]+]]:_(<32 x s32>) = G_FREEZE [[DEF]]
; CHECK-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<64 x s32>) = G_CONCAT_VECTORS [[FREEZE]](<32 x s32>), [[FREEZE1]](<32 x s32>)
; CHECK-NEXT: S_NOP 0, implicit [[CONCAT_VECTORS]](<64 x s32>)
%0:_(<64 x s32>) = G_IMPLICIT_DEF
%1:_(<64 x s32>) = G_FREEZE %0
Expand Down
28 changes: 8 additions & 20 deletions llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-implicit-def.mir
View file
Original file line number Diff line number Diff line change
Expand Up @@ -135,8 +135,9 @@ body: |
bb.0:
; CHECK-LABEL: name: test_implicit_def_s448
; CHECK: [[DEF:%[0-9]+]]:_(s448) = G_IMPLICIT_DEF
; CHECK-NEXT: [[EXTRACT:%[0-9]+]]:_(s32) = G_EXTRACT [[DEF]](s448), 0
; CHECK: [[DEF:%[0-9]+]]:_(s512) = G_IMPLICIT_DEF
; CHECK-NEXT: [[TRUNC:%[0-9]+]]:_(s448) = G_TRUNC [[DEF]](s512)
; CHECK-NEXT: [[EXTRACT:%[0-9]+]]:_(s32) = G_EXTRACT [[TRUNC]](s448), 0
; CHECK-NEXT: $vgpr0 = COPY [[EXTRACT]](s32)
%0:_(s448) = G_IMPLICIT_DEF
%1:_(s32) = G_EXTRACT %0, 0
Expand Down Expand Up @@ -295,18 +296,6 @@ body: |
$vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %0
...

---
name: test_implicit_def_v17s32
body: |
bb.0:
; CHECK-LABEL: name: test_implicit_def_v17s32
; CHECK: [[DEF:%[0-9]+]]:_(<17 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: S_NOP 0, implicit [[DEF]](<17 x s32>)
%0:_(<17 x s32>) = G_IMPLICIT_DEF
S_NOP 0, implicit %0
...

---
name: test_implicit_def_v32s32
body: |
Expand All @@ -328,9 +317,9 @@ body: |
; CHECK-LABEL: name: test_implicit_def_v33s32
; CHECK: liveins: $vgpr0_vgpr1
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<16 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<32 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; CHECK-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32), [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32), [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32), [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32), [[UV12:%[0-9]+]]:_(s32), [[UV13:%[0-9]+]]:_(s32), [[UV14:%[0-9]+]]:_(s32), [[UV15:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32), [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32), [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32), [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32), [[UV12:%[0-9]+]]:_(s32), [[UV13:%[0-9]+]]:_(s32), [[UV14:%[0-9]+]]:_(s32), [[UV15:%[0-9]+]]:_(s32), [[UV16:%[0-9]+]]:_(s32), [[UV17:%[0-9]+]]:_(s32), [[UV18:%[0-9]+]]:_(s32), [[UV19:%[0-9]+]]:_(s32), [[UV20:%[0-9]+]]:_(s32), [[UV21:%[0-9]+]]:_(s32), [[UV22:%[0-9]+]]:_(s32), [[UV23:%[0-9]+]]:_(s32), [[UV24:%[0-9]+]]:_(s32), [[UV25:%[0-9]+]]:_(s32), [[UV26:%[0-9]+]]:_(s32), [[UV27:%[0-9]+]]:_(s32), [[UV28:%[0-9]+]]:_(s32), [[UV29:%[0-9]+]]:_(s32), [[UV30:%[0-9]+]]:_(s32), [[UV31:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[DEF]](<32 x s32>)
; CHECK-NEXT: [[COPY:%[0-9]+]]:_(p1) = COPY $vgpr0_vgpr1
; CHECK-NEXT: G_STORE [[UV]](s32), [[COPY]](p1) :: (volatile store (s32), addrspace 1)
; CHECK-NEXT: G_STORE [[DEF1]](s32), [[COPY]](p1) :: (volatile store (s32), addrspace 1)
Expand All @@ -348,10 +337,9 @@ body: |
bb.0:
; CHECK-LABEL: name: test_implicit_def_v64s32
; CHECK: [[DEF:%[0-9]+]]:_(<16 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<64 x s32>) = G_CONCAT_VECTORS [[DEF]](<16 x s32>), [[DEF]](<16 x s32>), [[DEF]](<16 x s32>), [[DEF]](<16 x s32>)
; CHECK-NEXT: [[CONCAT_VECTORS1:%[0-9]+]]:_(<32 x s32>) = G_CONCAT_VECTORS [[DEF]](<16 x s32>), [[DEF]](<16 x s32>)
; CHECK-NEXT: S_NOP 0, implicit [[CONCAT_VECTORS]](<64 x s32>), implicit [[CONCAT_VECTORS1]](<32 x s32>)
; CHECK: [[DEF:%[0-9]+]]:_(<32 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<64 x s32>) = G_CONCAT_VECTORS [[DEF]](<32 x s32>), [[DEF]](<32 x s32>)
; CHECK-NEXT: S_NOP 0, implicit [[CONCAT_VECTORS]](<64 x s32>), implicit [[DEF]](<32 x s32>)
%0:_(<64 x s32>) = G_IMPLICIT_DEF
%1:_(<32 x s32>), %2:_(<32 x s32>) = G_UNMERGE_VALUES %0
S_NOP 0, implicit %0, implicit %1
Expand Down
14 changes: 5 additions & 9 deletions llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-insert-vector-elt.mir
View file
Original file line number Diff line number Diff line change
Expand Up @@ -190,13 +190,11 @@ body: |
; CHECK-LABEL: name: insert_vector_elt_64_65_v64s32
; CHECK: liveins: $sgpr0_sgpr1, $vgpr0_vgpr1, $vgpr2_vgpr3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<16 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[DEF:%[0-9]+]]:_(<32 x s32>) = G_IMPLICIT_DEF
; CHECK-NEXT: [[COPY:%[0-9]+]]:_(p1) = COPY $vgpr0_vgpr1
; CHECK-NEXT: [[COPY1:%[0-9]+]]:_(p1) = COPY $vgpr2_vgpr3
; CHECK-NEXT: [[UV:%[0-9]+]]:_(<4 x s32>), [[UV1:%[0-9]+]]:_(<4 x s32>), [[UV2:%[0-9]+]]:_(<4 x s32>), [[UV3:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV4:%[0-9]+]]:_(<4 x s32>), [[UV5:%[0-9]+]]:_(<4 x s32>), [[UV6:%[0-9]+]]:_(<4 x s32>), [[UV7:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV8:%[0-9]+]]:_(<4 x s32>), [[UV9:%[0-9]+]]:_(<4 x s32>), [[UV10:%[0-9]+]]:_(<4 x s32>), [[UV11:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV12:%[0-9]+]]:_(<4 x s32>), [[UV13:%[0-9]+]]:_(<4 x s32>), [[UV14:%[0-9]+]]:_(<4 x s32>), [[UV15:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV:%[0-9]+]]:_(<4 x s32>), [[UV1:%[0-9]+]]:_(<4 x s32>), [[UV2:%[0-9]+]]:_(<4 x s32>), [[UV3:%[0-9]+]]:_(<4 x s32>), [[UV4:%[0-9]+]]:_(<4 x s32>), [[UV5:%[0-9]+]]:_(<4 x s32>), [[UV6:%[0-9]+]]:_(<4 x s32>), [[UV7:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<32 x s32>)
; CHECK-NEXT: [[UV8:%[0-9]+]]:_(<4 x s32>), [[UV9:%[0-9]+]]:_(<4 x s32>), [[UV10:%[0-9]+]]:_(<4 x s32>), [[UV11:%[0-9]+]]:_(<4 x s32>), [[UV12:%[0-9]+]]:_(<4 x s32>), [[UV13:%[0-9]+]]:_(<4 x s32>), [[UV14:%[0-9]+]]:_(<4 x s32>), [[UV15:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<32 x s32>)
; CHECK-NEXT: G_STORE [[UV]](<4 x s32>), [[COPY]](p1) :: (store (<4 x s32>), align 4, addrspace 1)
; CHECK-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 16
; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(p1) = G_PTR_ADD [[COPY]], [[C]](s64)
Expand Down Expand Up @@ -243,10 +241,8 @@ body: |
; CHECK-NEXT: [[C14:%[0-9]+]]:_(s64) = G_CONSTANT i64 240
; CHECK-NEXT: [[PTR_ADD14:%[0-9]+]]:_(p1) = G_PTR_ADD [[COPY]], [[C14]](s64)
; CHECK-NEXT: G_STORE [[UV15]](<4 x s32>), [[PTR_ADD14]](p1) :: (store (<4 x s32>) into unknown-address + 240, align 4, addrspace 1)
; CHECK-NEXT: [[UV16:%[0-9]+]]:_(<4 x s32>), [[UV17:%[0-9]+]]:_(<4 x s32>), [[UV18:%[0-9]+]]:_(<4 x s32>), [[UV19:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV20:%[0-9]+]]:_(<4 x s32>), [[UV21:%[0-9]+]]:_(<4 x s32>), [[UV22:%[0-9]+]]:_(<4 x s32>), [[UV23:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV24:%[0-9]+]]:_(<4 x s32>), [[UV25:%[0-9]+]]:_(<4 x s32>), [[UV26:%[0-9]+]]:_(<4 x s32>), [[UV27:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV28:%[0-9]+]]:_(<4 x s32>), [[UV29:%[0-9]+]]:_(<4 x s32>), [[UV30:%[0-9]+]]:_(<4 x s32>), [[UV31:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<16 x s32>)
; CHECK-NEXT: [[UV16:%[0-9]+]]:_(<4 x s32>), [[UV17:%[0-9]+]]:_(<4 x s32>), [[UV18:%[0-9]+]]:_(<4 x s32>), [[UV19:%[0-9]+]]:_(<4 x s32>), [[UV20:%[0-9]+]]:_(<4 x s32>), [[UV21:%[0-9]+]]:_(<4 x s32>), [[UV22:%[0-9]+]]:_(<4 x s32>), [[UV23:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<32 x s32>)
; CHECK-NEXT: [[UV24:%[0-9]+]]:_(<4 x s32>), [[UV25:%[0-9]+]]:_(<4 x s32>), [[UV26:%[0-9]+]]:_(<4 x s32>), [[UV27:%[0-9]+]]:_(<4 x s32>), [[UV28:%[0-9]+]]:_(<4 x s32>), [[UV29:%[0-9]+]]:_(<4 x s32>), [[UV30:%[0-9]+]]:_(<4 x s32>), [[UV31:%[0-9]+]]:_(<4 x s32>) = G_UNMERGE_VALUES [[DEF]](<32 x s32>)
; CHECK-NEXT: G_STORE [[UV16]](<4 x s32>), [[COPY1]](p1) :: (store (<4 x s32>), align 4, addrspace 1)
; CHECK-NEXT: [[PTR_ADD15:%[0-9]+]]:_(p1) = G_PTR_ADD [[COPY1]], [[C]](s64)
; CHECK-NEXT: G_STORE [[UV17]](<4 x s32>), [[PTR_ADD15]](p1) :: (store (<4 x s32>) into unknown-address + 16, align 4, addrspace 1)
Expand Down
Loading

0 comments on commit 63b2595

Please sign in to comment.