[llvm-bugs] [Bug 36424] New: input module is broken! (Intrinsic has incorrect argument type!)

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=36424

            Bug ID: 36424
           Summary: input module is broken!  (Intrinsic has incorrect
                    argument type!)
           Product: tools
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: bugpoint
          Assignee: unassignedbugs at nondot.org
          Reporter: paulsson at linux.vnet.ibm.com
                CC: llvm-bugs at lists.llvm.org

Created attachment 19887
  --> https://bugs.llvm.org/attachment.cgi?id=19887&action=edit
original test case

It seems to me that bugpoint is generating an incorrect program somehow and
fails to recognize this.

The original test case triggers an assert in NewGVN:

bin/opt -O3 -mcpu=z13 -S crash10.bc -enable-newgvn

opt: /home/ijonpan/llvm/llvm-dev-2/lib/Transforms/Scalar/NewGVN.cpp:3285: void
{anonymous}::NewGVN::verifyIterationSettled(llvm::Function&): Assertion
`BeforeCC->isEquivalentTo(AfterCC) && "Value number changed after main loop
completed!"' failed.
#0
...

I tried to reduce this, but bugpoint crashed with my script, in a way I have
not seen before:

...
Intrinsic has incorrect argument type!
void (i8*, i8*, i64, i1)* @llvm.memcpy.p0i8.p0i8.i64
/home/ijonpan/llvm/build/llvm-dev/bin/opt: bugpoint-input-d2e260a.bc: error:
input module is broken!
verify failed!

What is the reason for this? I don't know how bugpoint reduces the program, or
if the -enable-newgvn variable is somehow responsible?

-- 
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/20180218/d2d4ad87/attachment.html>