[llvm-bugs] [Bug 44536] New: SLPVectorizer should drop nsw flags from add

via llvm-bugs llvm-bugs at lists.llvm.org
Mon Jan 13 08:53:20 PST 2020


https://bugs.llvm.org/show_bug.cgi?id=44536

            Bug ID: 44536
           Summary: SLPVectorizer should drop nsw flags from add
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Transformation Utilities
          Assignee: unassignedbugs at nondot.org
          Reporter: juneyoung.lee at sf.snu.ac.kr
                CC: llvm-bugs at lists.llvm.org

Created attachment 23014
  --> https://bugs.llvm.org/attachment.cgi?id=23014&action=edit
horizontal.ll

```
$ cat horizontal.ll # from excerpted from
test/Transforms/SLPVectorizer/X86/horizontal.ll 
@arr_i32 = global [32 x i32] zeroinitializer, align 16
declare i32 @foobar(i32)

define void @i32_red_call(i32 %val) {
entry:
  %0 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 0), align 16
  %1 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 1), align 4
  %add = add nsw i32 %1, %0
  %2 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 2), align 8
  %add.1 = add nsw i32 %2, %add
  %3 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 3), align 4
  %add.2 = add nsw i32 %3, %add.1
  %4 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 4), align 16
  %add.3 = add nsw i32 %4, %add.2
  %5 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 5), align 4
  %add.4 = add nsw i32 %5, %add.3
  %6 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 6), align 8
  %add.5 = add nsw i32 %6, %add.4
  %7 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr_i32,
i64 0, i64 7), align 4
  %add.6 = add nsw i32 %7, %add.5
  %res = call i32 @foobar(i32 %add.6)
  ret void
}
$ opt -slp-vectorizer -S -o - -mtriple=x86_64-apple-macosx -mcpu=corei7-avx
./horizontal.ll
@arr_i32 = global [32 x i32] zeroinitializer, align 16
declare i32 @foobar(i32) #0
define void @i32_red_call(i32 %val) #0 {
entry:
  %0 = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr_i32 to <8 x i32>*),
align 16
  %rdx.shuf = shufflevector <8 x i32> %0, <8 x i32> undef, <8 x i32> <i32 4,
i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add nsw <8 x i32> %0, %rdx.shuf
  %rdx.shuf1 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32>
<i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32
undef>
  %bin.rdx2 = add nsw <8 x i32> %bin.rdx, %rdx.shuf1
  %rdx.shuf3 = shufflevector <8 x i32> %bin.rdx2, <8 x i32> undef, <8 x i32>
<i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32
undef>
  %bin.rdx4 = add nsw <8 x i32> %bin.rdx2, %rdx.shuf3
  %1 = extractelement <8 x i32> %bin.rdx4, i32 0
  %res = call i32 @foobar(i32 %1)
  ret void
}

attributes #0 = { "target-cpu"="corei7-avx" }
```

SLPVectorizer reorders addition, so keeping nsw flag can result in introducing
undefined behavior.
As reassociate does, nsw flags should be dropped.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20200113/4837347b/attachment.html>


More information about the llvm-bugs mailing list