2978b20af4
This patch adds a check in validation for delinearization to ensure that the offset calculation does not overflow. If it overflows, different array accesses (e.g., `A[0][0]` and `A[1][0]`) could map to the same linear index, leading to incorrect behavior. For fixed-size arrays, the check is relatively straightforward. However, for dynamic-size arrays (i.e., arrays where the size is not known at compile time), it's difficult to prove this statically, and it going to fail for almost all cases. Maybe we need to add some runtime checks or reasoning based on `inbounds` like LAA does. Fixes the test cases added in #169048.
78 lines
3.3 KiB
LLVM
78 lines
3.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt < %s -passes='print<delinearization>' -disable-output 2>&1 | FileCheck %s
|
|
|
|
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
|
|
|
|
; Function Attrs: noinline nounwind uwtable
|
|
define void @mat_mul(ptr %C, ptr %A, ptr %B, i64 %N) !kernel_arg_addr_space !2 !kernel_arg_access_qual !3 !kernel_arg_type !4 !kernel_arg_base_type !4 !kernel_arg_type_qual !5 {
|
|
; CHECK-LABEL: 'mat_mul'
|
|
; CHECK-NEXT: Inst: %tmp = load float, ptr %arrayidx, align 4
|
|
; CHECK-NEXT: AccessFunction: {(4 * %N * %call),+,4}<%for.inc>
|
|
; CHECK-NEXT: Base offset: %A
|
|
; CHECK-NEXT: ArrayDecl[UnknownSize][%N] with elements of 4 bytes.
|
|
; CHECK-NEXT: ArrayRef[%call][{0,+,1}<nuw><nsw><%for.inc>]
|
|
; CHECK-NEXT: Delinearization validation: Failed
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: Inst: %tmp5 = load float, ptr %arrayidx4, align 4
|
|
; CHECK-NEXT: AccessFunction: {(4 * %call1),+,(4 * %N)}<%for.inc>
|
|
; CHECK-NEXT: Base offset: %B
|
|
; CHECK-NEXT: ArrayDecl[UnknownSize][%N] with elements of 4 bytes.
|
|
; CHECK-NEXT: ArrayRef[{0,+,1}<nuw><nsw><%for.inc>][%call1]
|
|
; CHECK-NEXT: Delinearization validation: Failed
|
|
;
|
|
entry:
|
|
br label %entry.split
|
|
|
|
entry.split: ; preds = %entry
|
|
%call = tail call i64 @_Z13get_global_idj(i32 0)
|
|
%call1 = tail call i64 @_Z13get_global_idj(i32 1)
|
|
%cmp1 = icmp sgt i64 %N, 0
|
|
%mul = mul nsw i64 %call, %N
|
|
br i1 %cmp1, label %for.inc.lr.ph, label %for.end
|
|
|
|
for.inc.lr.ph: ; preds = %entry.split
|
|
br label %for.inc
|
|
|
|
for.inc: ; preds = %for.inc.lr.ph, %for.inc
|
|
%acc.03 = phi float [ 0.000000e+00, %for.inc.lr.ph ], [ %tmp6, %for.inc ]
|
|
%m.02 = phi i64 [ 0, %for.inc.lr.ph ], [ %inc, %for.inc ]
|
|
%add = add nsw i64 %m.02, %mul
|
|
%arrayidx = getelementptr inbounds float, ptr %A, i64 %add
|
|
%tmp = load float, ptr %arrayidx, align 4
|
|
%mul2 = mul nsw i64 %m.02, %N
|
|
%add3 = add nsw i64 %mul2, %call1
|
|
%arrayidx4 = getelementptr inbounds float, ptr %B, i64 %add3
|
|
%tmp5 = load float, ptr %arrayidx4, align 4
|
|
%tmp6 = tail call float @llvm.fmuladd.f32(float %tmp, float %tmp5, float %acc.03)
|
|
%inc = add nuw nsw i64 %m.02, 1
|
|
%exitcond = icmp ne i64 %inc, %N
|
|
br i1 %exitcond, label %for.inc, label %for.cond.for.end_crit_edge
|
|
|
|
for.cond.for.end_crit_edge: ; preds = %for.inc
|
|
%.lcssa = phi float [ %tmp6, %for.inc ]
|
|
br label %for.end
|
|
|
|
for.end: ; preds = %for.cond.for.end_crit_edge, %entry.split
|
|
%acc.0.lcssa = phi float [ %.lcssa, %for.cond.for.end_crit_edge ], [ 0.000000e+00, %entry.split ]
|
|
%add7 = add nsw i64 %mul, %call1
|
|
%arrayidx8 = getelementptr inbounds float, ptr %C, i64 %add7
|
|
store float %acc.0.lcssa, ptr %arrayidx8, align 4
|
|
ret void
|
|
}
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i64 @_Z13get_global_idj(i32)
|
|
|
|
; Function Attrs: nounwind readnone speculatable
|
|
declare float @llvm.fmuladd.f32(float, float, float)
|
|
|
|
!llvm.module.flags = !{!0}
|
|
!llvm.ident = !{!1}
|
|
|
|
!0 = !{i32 1, !"wchar_size", i32 4}
|
|
!1 = !{!"clang version 5.0.0 (trunk 303846) (llvm/trunk 303834)"}
|
|
!2 = !{i32 1, i32 1, i32 1, i32 0}
|
|
!3 = !{!"none", !"none", !"none", !"none"}
|
|
!4 = !{!"float*", !"float*", !"float*", !"long"}
|
|
!5 = !{!"", !"", !"", !""}
|