[llvm-bugs] [Bug 43227] New: [X86] EltsFromConsecutiveLoads - incorrect folding of non-zero offset base loads

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=43227

            Bug ID: 43227
           Summary: [X86] EltsFromConsecutiveLoads - incorrect folding of
                    non-zero offset base loads
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Backend: X86
          Assignee: unassignedbugs at nondot.org
          Reporter: llvm-dev at redking.me.uk
                CC: craig.topper at gmail.com, llvm-bugs at lists.llvm.org,
                    llvm-dev at redking.me.uk, spatel+llvm at rotateright.com

As reported on D64551, rL366681 doesn't correctly handle cases where the
deerived byte offset of the base load is non-zero.

define void @foo(i32* %explicit_0, <8 x i32>* %explicit_1) {
  %1 = getelementptr i32, i32* %explicit_0, i64 63
  %2 = bitcast i32* %1 to <3 x i32>*
  %3 = load <3 x i32>, <3 x i32>* %2, align 1
  %4 = shufflevector <3 x i32> %3, <3 x i32> undef, <2 x i32> <i32 1, i32 2>
  %5 = shufflevector <2 x i32> %4, <2 x i32> undef, <8 x i32> <i32 0, i32 1,
i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %6 = shufflevector <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 undef, i32 0,
i32 undef, i32 0>, <8 x i32> %5, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8,
i32 5, i32 9, i32 7>
  %7 = getelementptr inbounds <8 x i32>, <8 x i32>* %explicit_1, i64 21
  store <8 x i32> %6, <8 x i32>* %7, align 32
  ret void
}

-- 
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/20190905/c83b1834/attachment.html>