Dead Code Elimination Regression at -O2 (trunk vs. 13.0.0) #53320
Description
cat case.c #5986
case.txt
struct {
signed b;
char c;
unsigned : 9;
unsigned d : 1;
signed k: 7;
} e;
void foo(void);
long(a)(long f, long g) { return f > 9223372036854775807 - g ? f : g; }
int main() {
a(0, 0);
if (e.d)
e.b = 0;
if (!a(2, e.c ^ 128))
foo();
}clang-f29256a64ac11cf59cea878c8d1ba9537db4f523 (trunk) -O2 can not eliminate foo but clang-llvmorg-13.0.0 -O2 can.
Target: x86_64-unknown-linux-gnu
clang-f29256a64ac11cf59cea878c8d1ba9537db4f523 (trunk) -O2 [-emit-llvm] -S -o /dev/stdout case.c
Emitted IR
; ModuleID = 'case.c'
source_filename = "case.c"
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
%struct.anon = type { i32, i8, [3 x i8] }
@e = dso_local local_unnamed_addr global %struct.anon zeroinitializer, align 4
; Function Attrs: mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn
define dso_local i64 @a(i64 noundef %0, i64 noundef %1) local_unnamed_addr #0 {
%3 = sub nsw i64 9223372036854775807, %1
%4 = icmp slt i64 %3, %0
%5 = select i1 %4, i64 %0, i64 %1
ret i64 %5
}
; Function Attrs: nounwind uwtable
define dso_local i32 @main() local_unnamed_addr #1 {
%1 = load i24, i24* bitcast ([3 x i8]* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 2) to i24*), align 1
%2 = and i24 %1, 512
%3 = icmp eq i24 %2, 0
br i1 %3, label %5, label %4
4: ; preds = %0
store i32 0, i32* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 0), align 4, !tbaa !3
br label %5
5: ; preds = %4, %0
%6 = load i8, i8* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 1), align 4, !tbaa !8
%7 = sext i8 %6 to i64
%8 = xor i64 %7, 128
%9 = and i64 %8, -2
%10 = icmp ne i64 %9, 9223372036854775806
%11 = icmp eq i64 %8, 0
%12 = and i1 %11, %10
br i1 %12, label %13, label %14
13: ; preds = %5
tail call void @foo() #3
br label %14
14: ; preds = %13, %5
ret i32 0
}
declare dso_local void @foo() local_unnamed_addr #2
attributes #0 = { mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #1 = { nounwind uwtable "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #2 = { "frame-pointer"="none" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #3 = { nounwind }
!llvm.module.flags = !{!0, !1}
!llvm.ident = !{!2}
!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{i32 7, !"uwtable", i32 1}
!2 = !{!"clang version 14.0.0 (https://github.com/llvm/llvm-project.git f29256a64ac11cf59cea878c8d1ba9537db4f523)"}
!3 = !{!4, !5, i64 0}
!4 = !{!"", !5, i64 0, !6, i64 4, !5, i64 6, !5, i64 6}
!5 = !{!"int", !6, i64 0}
!6 = !{!"omnipotent char", !7, i64 0}
!7 = !{!"Simple C/C++ TBAA"}
!8 = !{!4, !6, i64 4}Reduced assembly
main: # @main
.cfi_startproc
# %bb.0:
pushq %rax
.cfi_def_cfa_offset 16
movzwl e+5(%rip), %eax
testl $512, %eax # imm = 0x200
je .LBB1_2
# %bb.1:
movl $0, e(%rip)
.LBB1_2:
movsbq e+4(%rip), %rax
xorq $128, %rax
jne .LBB1_5
# %bb.3:
andq $-2, %rax
movabsq $9223372036854775806, %rcx # imm = 0x7FFFFFFFFFFFFFFE
cmpq %rcx, %rax
je .LBB1_5
# %bb.4:
callq foo
.LBB1_5:
xorl %eax, %eax
popq %rcx
.cfi_def_cfa_offset 8
retq
.Lfunc_end1:
.size main, .Lfunc_end1-mainclang-llvmorg-13.0.0 -O2 [-emit-llvm] -S -o /dev/stdout case.c
Emitted IR
; ModuleID = 'case.c'
source_filename = "case.c"
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
%struct.anon = type { i32, i8, [3 x i8] }
@e = dso_local local_unnamed_addr global %struct.anon zeroinitializer, align 4
; Function Attrs: mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn
define dso_local i64 @a(i64 %0, i64 %1) local_unnamed_addr #0 {
%3 = sub nsw i64 9223372036854775807, %1
%4 = icmp slt i64 %3, %0
%5 = select i1 %4, i64 %0, i64 %1
ret i64 %5
}
; Function Attrs: nounwind uwtable
define dso_local i32 @main() local_unnamed_addr #1 {
%1 = load i24, i24* bitcast ([3 x i8]* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 2) to i24*), align 1
%2 = and i24 %1, 512
%3 = icmp eq i24 %2, 0
br i1 %3, label %5, label %4
4: ; preds = %0
store i32 0, i32* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 0), align 4, !tbaa !3
br label %5
5: ; preds = %0, %4
ret i32 0
}
attributes #0 = { mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #1 = { nounwind uwtable "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
!llvm.module.flags = !{!0, !1}
!llvm.ident = !{!2}
!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{i32 7, !"uwtable", i32 1}
!2 = !{!"clang version 13.0.0 (https://github.com/llvm/llvm-project.git d7b669b3a30345cfcdb2fde2af6f48aa4b94845d)"}
!3 = !{!4, !5, i64 0}
!4 = !{!"", !5, i64 0, !6, i64 4, !5, i64 6, !5, i64 6}
!5 = !{!"int", !6, i64 0}
!6 = !{!"omnipotent char", !7, i64 0}
!7 = !{!"Simple C/C++ TBAA"}Reduced assembly
main: # @main
.cfi_startproc
# %bb.0:
movzwl e+5(%rip), %eax
testl $512, %eax # imm = 0x200
je .LBB1_2
# %bb.1:
movl $0, e(%rip)
.LBB1_2:
xorl %eax, %eax
retq
.Lfunc_end1:
.size main, .Lfunc_end1-mainBisection
Bisected to: b554e41
Committed by: @LebedevRI
clang-b554e41e2d15b78679231b390e543153e05c9efe -O2 [-emit-llvm] -S -o /dev/stdout case.c
Emitted IR
; ModuleID = 'case.c'
source_filename = "case.c"
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
%struct.anon = type { i32, i8, [3 x i8] }
@e = dso_local local_unnamed_addr global %struct.anon zeroinitializer, align 4
; Function Attrs: mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn
define dso_local i64 @a(i64 %0, i64 %1) local_unnamed_addr #0 {
%3 = sub nsw i64 9223372036854775807, %1
%4 = icmp slt i64 %3, %0
%5 = select i1 %4, i64 %0, i64 %1
ret i64 %5
}
; Function Attrs: nounwind uwtable
define dso_local i32 @main() local_unnamed_addr #1 {
%1 = load i24, i24* bitcast ([3 x i8]* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 2) to i24*), align 1
%2 = and i24 %1, 512
%3 = icmp eq i24 %2, 0
br i1 %3, label %5, label %4
4: ; preds = %0
store i32 0, i32* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 0), align 4, !tbaa !3
br label %5
5: ; preds = %4, %0
%6 = load i8, i8* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 1), align 4, !tbaa !8
%7 = sext i8 %6 to i64
%8 = xor i64 %7, 128
%9 = and i64 %8, -2
%10 = icmp ne i64 %9, 9223372036854775806
%11 = icmp eq i64 %8, 0
%12 = and i1 %11, %10
br i1 %12, label %13, label %14
13: ; preds = %5
tail call void @foo() #3
br label %14
14: ; preds = %13, %5
ret i32 0
}
declare dso_local void @foo() local_unnamed_addr #2
attributes #0 = { mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #1 = { nounwind uwtable "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #2 = { "frame-pointer"="none" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #3 = { nounwind }
!llvm.module.flags = !{!0, !1}
!llvm.ident = !{!2}
!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{i32 7, !"uwtable", i32 1}
!2 = !{!"clang version 14.0.0 (https://github.com/llvm/llvm-project.git b554e41e2d15b78679231b390e543153e05c9efe)"}
!3 = !{!4, !5, i64 0}
!4 = !{!"", !5, i64 0, !6, i64 4, !5, i64 6, !5, i64 6}
!5 = !{!"int", !6, i64 0}
!6 = !{!"omnipotent char", !7, i64 0}
!7 = !{!"Simple C/C++ TBAA"}
!8 = !{!4, !6, i64 4}Reduced assembly
main: # @main
.cfi_startproc
# %bb.0:
pushq %rax
.cfi_def_cfa_offset 16
movzwl e+5(%rip), %eax
testl $512, %eax # imm = 0x200
je .LBB1_2
# %bb.1:
movl $0, e(%rip)
.LBB1_2:
movsbq e+4(%rip), %rax
xorq $128, %rax
jne .LBB1_5
# %bb.3:
andq $-2, %rax
movabsq $9223372036854775806, %rcx # imm = 0x7FFFFFFFFFFFFFFE
cmpq %rcx, %rax
je .LBB1_5
# %bb.4:
callq foo
.LBB1_5:
xorl %eax, %eax
popq %rcx
.cfi_def_cfa_offset 8
retq
.Lfunc_end1:
.size main, .Lfunc_end1-mainPrevious commit: 81441cf
clang-81441cf44c145e1d90a10c743e0190a44fbf8fcb -O2 [-emit-llvm] -S -o /dev/stdout case.c
Emitted IR
; ModuleID = 'case.c'
source_filename = "case.c"
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
%struct.anon = type { i32, i8, [3 x i8] }
@e = dso_local local_unnamed_addr global %struct.anon zeroinitializer, align 4
; Function Attrs: mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn
define dso_local i64 @a(i64 %0, i64 %1) local_unnamed_addr #0 {
%3 = sub nsw i64 9223372036854775807, %1
%4 = icmp slt i64 %3, %0
%5 = select i1 %4, i64 %0, i64 %1
ret i64 %5
}
; Function Attrs: nounwind uwtable
define dso_local i32 @main() local_unnamed_addr #1 {
%1 = load i24, i24* bitcast ([3 x i8]* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 2) to i24*), align 1
%2 = and i24 %1, 512
%3 = icmp eq i24 %2, 0
br i1 %3, label %5, label %4
4: ; preds = %0
store i32 0, i32* getelementptr inbounds (%struct.anon, %struct.anon* @e, i64 0, i32 0), align 4, !tbaa !3
br label %5
5: ; preds = %0, %4
ret i32 0
}
attributes #0 = { mustprogress nofree norecurse nosync nounwind readnone uwtable willreturn "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
attributes #1 = { nounwind uwtable "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "tune-cpu"="generic" }
!llvm.module.flags = !{!0, !1}
!llvm.ident = !{!2}
!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{i32 7, !"uwtable", i32 1}
!2 = !{!"clang version 14.0.0 (https://github.com/llvm/llvm-project.git 81441cf44c145e1d90a10c743e0190a44fbf8fcb)"}
!3 = !{!4, !5, i64 0}
!4 = !{!"", !5, i64 0, !6, i64 4, !5, i64 6, !5, i64 6}
!5 = !{!"int", !6, i64 0}
!6 = !{!"omnipotent char", !7, i64 0}
!7 = !{!"Simple C/C++ TBAA"}Reduced assembly
main: # @main
.cfi_startproc
# %bb.0:
movzwl e+5(%rip), %eax
testl $512, %eax # imm = 0x200
je .LBB1_2
# %bb.1:
movl $0, e(%rip)
.LBB1_2:
xorl %eax, %eax
retq
.Lfunc_end1:
.size main, .Lfunc_end1-mainEdit: Added IR