[llvm-bugs] [Bug 49379] New: "opt -enable-loop-distribute -loop-distribute" failing with "PHINode should have one entry for each predecessor of its parent basic block!"

via llvm-bugs llvm-bugs at lists.llvm.org
Mon Mar 1 06:24:23 PST 2021


https://bugs.llvm.org/show_bug.cgi?id=49379

            Bug ID: 49379
           Summary: "opt -enable-loop-distribute -loop-distribute" failing
                    with "PHINode should have one entry for each
                    predecessor of its parent basic block!"
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Scalar Optimizations
          Assignee: unassignedbugs at nondot.org
          Reporter: mikael.holmen at ericsson.com
                CC: llvm-bugs at lists.llvm.org

Created attachment 24578
  --> https://bugs.llvm.org/attachment.cgi?id=24578&action=edit
bbi-53389.ll

llvm commit 7b319df29bf
Reproduce with:
 opt -enable-loop-distribute -loop-distribute -S -o - bbi-53389.ll

Result:
PHINode should have one entry for each predecessor of its parent basic block!
  %1 = phi i16 [ undef, %for.end.lver.orig ], [ 0, %cleanup4.loopexit ]
LLVM ERROR: Broken module found, compilation aborted!

The offending basic block looks like:

cleanup4.lver.orig:                               ; preds = %for.end.lver.orig
  %1 = phi i16 [ undef, %for.end.lver.orig ], [ 0, %cleanup4.loopexit ]
  br i1 false, label %lbl1.lver.orig, label %return.us-lcssa.loopexit

and there is no branch from %cleanup4.loopexit to cleanup4.lver.orig.

cleanup4.loopexit looks like:

cleanup4.loopexit:                                ; No predecessors!
  br label %cleanup4

There are blocks in the input that are unreachable from entry. I suppose that
confuses loop-distribute.

-- 
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/20210301/849576e9/attachment.html>


More information about the llvm-bugs mailing list