shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
llvm-project/llvm/test/Transforms/LoopVectorize/load-of-struct-deref-pred.ll
Florian Hahn 351d398a37
[VPlan] Run final VPlan simplifications before codegen. 
Dissolving the hierarchical VPlan CFG and converting abstract to
concrete recipes can expose additional simplification opportunities.

Do a final run of simplifyRecipes before executing the VPlan.
2025-08-16 18:54:27 +01:00

328 lines
17 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S %s | FileCheck %s
target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
%struct.Foo = type { [32000 x i32], [32000 x i32] }
@foo = global %struct.Foo zeroinitializer, align 4
define void @accesses_to_struct_dereferenceable(ptr noalias %dst) {
; CHECK-LABEL: define void @accesses_to_struct_dereferenceable
; CHECK-SAME: (ptr noalias [[DST:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ult <4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr [[STRUCT_FOO:%.*]], ptr @foo, i64 0, i32 1, i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr [[STRUCT_FOO]], ptr @foo, i64 0, i32 0, i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP1]], <4 x i32> [[WIDE_LOAD2]], <4 x i32> [[WIDE_LOAD1]]
; CHECK-NEXT: store <4 x i32> [[PREDPHI]], ptr [[TMP0]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 32000
; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[EXIT:%.*]]
; CHECK: scalar.ph:
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT: [[GEP_DST:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IV]]
; CHECK-NEXT: [[D:%.*]] = load i32, ptr [[GEP_DST]], align 4
; CHECK-NEXT: [[CMP3:%.*]] = icmp ult i32 [[D]], 0
; CHECK-NEXT: br i1 [[CMP3]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then:
; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 0, i64 [[IV]]
; CHECK-NEXT: [[L_A:%.*]] = load i32, ptr [[GEP_A]], align 4
; CHECK-NEXT: br label [[LOOP_LATCH]]
; CHECK: if.else:
; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[IV]]
; CHECK-NEXT: [[L_B:%.*]] = load i32, ptr [[GEP_B]], align 4
; CHECK-NEXT: br label [[LOOP_LATCH]]
; CHECK: loop.latch:
; CHECK-NEXT: [[TMP_0:%.*]] = phi i32 [ [[L_A]], [[IF_THEN]] ], [ [[L_B]], [[IF_ELSE]] ]
; CHECK-NEXT: store i32 [[TMP_0]], ptr [[GEP_DST]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 32000
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
%gep.dst = getelementptr inbounds i32, ptr %dst, i64 %iv
%d = load i32, ptr %gep.dst, align 4
%cmp3 = icmp ult i32 %d, 0
br i1 %cmp3, label %if.then, label %if.else
if.then:
%gep.a = getelementptr inbounds %struct.Foo, ptr @foo, i64 0, i32 0, i64 %iv
%l.a = load i32, ptr %gep.a, align 4
br label %loop.latch
if.else:
%gep.b = getelementptr inbounds %struct.Foo, ptr @foo, i64 0, i32 1, i64 %iv
%l.b = load i32, ptr %gep.b, align 4
br label %loop.latch
loop.latch:
%tmp.0 = phi i32 [ %l.a, %if.then ], [ %l.b, %if.else ]
store i32 %tmp.0, ptr %gep.dst, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 32000
br i1 %exitcond.not, label %exit, label %loop.header
exit:
ret void
}
define void @accesses_to_struct_may_not_be_dereferenceable_due_to_loop_bound(ptr noalias %dst) {
; CHECK-LABEL: define void @accesses_to_struct_may_not_be_dereferenceable_due_to_loop_bound
; CHECK-SAME: (ptr noalias [[DST:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE6:%.*]] ]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ult <4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP2:%.*]] = xor <4 x i1> [[TMP1]], splat (i1 true)
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i1> [[TMP2]], i32 0
; CHECK-NEXT: br i1 [[TMP3]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
; CHECK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_FOO:%.*]], ptr @foo, i64 0, i32 1, i64 [[TMP4]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i32 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
; CHECK: pred.load.continue:
; CHECK-NEXT: [[TMP8:%.*]] = phi <4 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP7]], [[PRED_LOAD_IF]] ]
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP2]], i32 1
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP10:%.*]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[TMP10]]
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP12]], i32 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
; CHECK: pred.load.continue2:
; CHECK-NEXT: [[TMP14:%.*]] = phi <4 x i32> [ [[TMP8]], [[PRED_LOAD_CONTINUE]] ], [ [[TMP13]], [[PRED_LOAD_IF1]] ]
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP2]], i32 2
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP16:%.*]] = add i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[TMP16]]
; CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x i32> [[TMP14]], i32 [[TMP18]], i32 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
; CHECK: pred.load.continue4:
; CHECK-NEXT: [[TMP20:%.*]] = phi <4 x i32> [ [[TMP14]], [[PRED_LOAD_CONTINUE2]] ], [ [[TMP19]], [[PRED_LOAD_IF3]] ]
; CHECK-NEXT: [[TMP21:%.*]] = extractelement <4 x i1> [[TMP2]], i32 3
; CHECK-NEXT: br i1 [[TMP21]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP22:%.*]] = add i64 [[INDEX]], 3
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[TMP22]]
; CHECK-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x i32> [[TMP20]], i32 [[TMP24]], i32 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.continue6:
; CHECK-NEXT: [[TMP26:%.*]] = phi <4 x i32> [ [[TMP20]], [[PRED_LOAD_CONTINUE4]] ], [ [[TMP25]], [[PRED_LOAD_IF5]] ]
; CHECK-NEXT: [[TMP27:%.*]] = getelementptr [[STRUCT_FOO]], ptr @foo, i64 0, i32 0, i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP27]], align 4
; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP1]], <4 x i32> [[WIDE_LOAD7]], <4 x i32> [[TMP26]]
; CHECK-NEXT: store <4 x i32> [[PREDPHI]], ptr [[TMP0]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP28:%.*]] = icmp eq i64 [[INDEX_NEXT]], 32000
; CHECK-NEXT: br i1 [[TMP28]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 32000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT: [[GEP_DST:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IV]]
; CHECK-NEXT: [[D:%.*]] = load i32, ptr [[GEP_DST]], align 4
; CHECK-NEXT: [[CMP3:%.*]] = icmp ult i32 [[D]], 0
; CHECK-NEXT: br i1 [[CMP3]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then:
; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 0, i64 [[IV]]
; CHECK-NEXT: [[L_A:%.*]] = load i32, ptr [[GEP_A]], align 4
; CHECK-NEXT: br label [[LOOP_LATCH]]
; CHECK: if.else:
; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[IV]]
; CHECK-NEXT: [[L_B:%.*]] = load i32, ptr [[GEP_B]], align 4
; CHECK-NEXT: br label [[LOOP_LATCH]]
; CHECK: loop.latch:
; CHECK-NEXT: [[TMP_0:%.*]] = phi i32 [ [[L_A]], [[IF_THEN]] ], [ [[L_B]], [[IF_ELSE]] ]
; CHECK-NEXT: store i32 [[TMP_0]], ptr [[GEP_DST]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 32001
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[EXIT:%.*]], label [[LOOP_HEADER]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
%gep.dst = getelementptr inbounds i32, ptr %dst, i64 %iv
%d = load i32, ptr %gep.dst, align 4
%cmp3 = icmp ult i32 %d, 0
br i1 %cmp3, label %if.then, label %if.else
if.then:
%gep.a = getelementptr inbounds %struct.Foo, ptr @foo, i64 0, i32 0, i64 %iv
%l.a = load i32, ptr %gep.a, align 4
br label %loop.latch
if.else:
%gep.b = getelementptr inbounds %struct.Foo, ptr @foo, i64 0, i32 1, i64 %iv
%l.b = load i32, ptr %gep.b, align 4
br label %loop.latch
loop.latch:
%tmp.0 = phi i32 [ %l.a, %if.then ], [ %l.b, %if.else ]
store i32 %tmp.0, ptr %gep.dst, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 32001
br i1 %exitcond.not, label %exit, label %loop.header
exit:
ret void
}
define void @accesses_to_struct_may_not_be_dereferenceable_access_size(ptr noalias %dst) {
; CHECK-LABEL: define void @accesses_to_struct_may_not_be_dereferenceable_access_size
; CHECK-SAME: (ptr noalias [[DST:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE6:%.*]] ]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ult <4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP2:%.*]] = xor <4 x i1> [[TMP1]], splat (i1 true)
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i1> [[TMP2]], i32 0
; CHECK-NEXT: br i1 [[TMP3]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
; CHECK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_FOO:%.*]], ptr @foo, i64 0, i32 1, i64 [[TMP4]]
; CHECK-NEXT: [[TMP6:%.*]] = load i64, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i64> poison, i64 [[TMP6]], i32 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
; CHECK: pred.load.continue:
; CHECK-NEXT: [[TMP8:%.*]] = phi <4 x i64> [ poison, [[VECTOR_BODY]] ], [ [[TMP7]], [[PRED_LOAD_IF]] ]
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP2]], i32 1
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP10:%.*]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[TMP10]]
; CHECK-NEXT: [[TMP12:%.*]] = load i64, ptr [[TMP11]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i64> [[TMP8]], i64 [[TMP12]], i32 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
; CHECK: pred.load.continue2:
; CHECK-NEXT: [[TMP14:%.*]] = phi <4 x i64> [ [[TMP8]], [[PRED_LOAD_CONTINUE]] ], [ [[TMP13]], [[PRED_LOAD_IF1]] ]
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP2]], i32 2
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP16:%.*]] = add i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[TMP16]]
; CHECK-NEXT: [[TMP18:%.*]] = load i64, ptr [[TMP17]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x i64> [[TMP14]], i64 [[TMP18]], i32 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
; CHECK: pred.load.continue4:
; CHECK-NEXT: [[TMP20:%.*]] = phi <4 x i64> [ [[TMP14]], [[PRED_LOAD_CONTINUE2]] ], [ [[TMP19]], [[PRED_LOAD_IF3]] ]
; CHECK-NEXT: [[TMP21:%.*]] = extractelement <4 x i1> [[TMP2]], i32 3
; CHECK-NEXT: br i1 [[TMP21]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP22:%.*]] = add i64 [[INDEX]], 3
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[TMP22]]
; CHECK-NEXT: [[TMP24:%.*]] = load i64, ptr [[TMP23]], align 4
; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP24]], i32 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.continue6:
; CHECK-NEXT: [[TMP26:%.*]] = phi <4 x i64> [ [[TMP20]], [[PRED_LOAD_CONTINUE4]] ], [ [[TMP25]], [[PRED_LOAD_IF5]] ]
; CHECK-NEXT: [[TMP27:%.*]] = trunc <4 x i64> [[TMP26]] to <4 x i32>
; CHECK-NEXT: [[TMP28:%.*]] = getelementptr [[STRUCT_FOO]], ptr @foo, i64 0, i32 0, i64 [[INDEX]]
; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP1]], <4 x i32> [[WIDE_LOAD7]], <4 x i32> [[TMP27]]
; CHECK-NEXT: store <4 x i32> [[PREDPHI]], ptr [[TMP0]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], 32000
; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[EXIT:%.*]]
; CHECK: scalar.ph:
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT: [[GEP_DST:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[IV]]
; CHECK-NEXT: [[D:%.*]] = load i32, ptr [[GEP_DST]], align 4
; CHECK-NEXT: [[CMP3:%.*]] = icmp ult i32 [[D]], 0
; CHECK-NEXT: br i1 [[CMP3]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then:
; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 0, i64 [[IV]]
; CHECK-NEXT: [[L_A:%.*]] = load i32, ptr [[GEP_A]], align 4
; CHECK-NEXT: br label [[LOOP_LATCH]]
; CHECK: if.else:
; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr inbounds [[STRUCT_FOO]], ptr @foo, i64 0, i32 1, i64 [[IV]]
; CHECK-NEXT: [[L_B:%.*]] = load i64, ptr [[GEP_B]], align 4
; CHECK-NEXT: [[T:%.*]] = trunc i64 [[L_B]] to i32
; CHECK-NEXT: br label [[LOOP_LATCH]]
; CHECK: loop.latch:
; CHECK-NEXT: [[TMP_0:%.*]] = phi i32 [ [[L_A]], [[IF_THEN]] ], [ [[T]], [[IF_ELSE]] ]
; CHECK-NEXT: store i32 [[TMP_0]], ptr [[GEP_DST]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], 32000
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
%gep.dst = getelementptr inbounds i32, ptr %dst, i64 %iv
%d = load i32, ptr %gep.dst, align 4
%cmp3 = icmp ult i32 %d, 0
br i1 %cmp3, label %if.then, label %if.else
if.then:
%gep.a = getelementptr inbounds %struct.Foo, ptr @foo, i64 0, i32 0, i64 %iv
%l.a = load i32, ptr %gep.a, align 4
br label %loop.latch
if.else:
%gep.b = getelementptr inbounds %struct.Foo, ptr @foo, i64 0, i32 1, i64 %iv
%l.b = load i64, ptr %gep.b, align 4
%t = trunc i64 %l.b to i32
br label %loop.latch
loop.latch:
%tmp.0 = phi i32 [ %l.a, %if.then ], [ %t, %if.else ]
store i32 %tmp.0, ptr %gep.dst, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 32000
br i1 %exitcond.not, label %exit, label %loop.header
exit:
ret void
}
Reference in New Issue View Git Blame Copy Permalink