[LLVMbugs] [Bug 19436] New: vectorizer will turn i1* -> <n x i1>*

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=19436

            Bug ID: 19436
           Summary: vectorizer will turn i1* -> <n x i1>*
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: Loop Optimizer
          Assignee: unassignedbugs at nondot.org
          Reporter: hfinkel at anl.gov
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Created attachment 12377
  --> http://llvm.org/bugs/attachment.cgi?id=12377&action=edit
test case

As demonstrated in the attached test case, the loop vectorizer will happily
turn a consecutive i1* load/store into a <n x i1>* load/store. However, my
understanding is that, although LLVM is currently inconsistent about this, the
desired memory representation of <n x i1> is packed, and so this transformation
is invalid.

The easiest thing will be to disable vectorizing i1*; alternatively, we could
vectorize by casting to <n x i8>* and back again.

-- 
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/20140415/95a74fbd/attachment.html>