[llvm-bugs] [Bug 31956] New: UNREACHABLE "Do not know how to widen this operator's operand!"

[via llvm-bugs]

http://bugs.llvm.org/show_bug.cgi?id=31956

            Bug ID: 31956
           Summary: UNREACHABLE "Do not know how to widen this operator's
                    operand!"
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: normal
          Priority: P
         Component: Common Code Generator Code
          Assignee: unassignedbugs at nondot.org
          Reporter: greg_bedwell at sn.scee.net
                CC: llvm-bugs at lists.llvm.org, mkuper at google.com

One of our random test generators found a case of "UNREACHABLE executed" in
SelectionDAG.

It's still reproducible as of r295055.  I've bisected its introduction to
r281402:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
33893419506150322836ecb439b9c23c986522eb is the first bad commit
commit 33893419506150322836ecb439b9c23c986522eb
Author: Michael Kuperstein <mkuper at google.com>
Date:   Tue Sep 13 21:53:32 2016 +0000

    [DAG] Allow build-to-shuffle combine to combine builds from two wide
vectors.

    This allows us to, in some cases, create a vector_shuffle out of a
build_vector, when
    the inputs to the build are extract_elements from two different vectors, at
least one
    of which is wider than the output. (E.g. a <8 x i16> being constructed out
of
    elements from a <16 x i16> and a <8 x i16>).

    Differential Revision: https://reviews.llvm.org/D24491


    git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@281402
91177308-0d34-0410-b5e6-96231b3b80d8
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Source-level test case:
// ============================================================================
typedef float __attribute__((ext_vector_type(8))) float8;
typedef float __attribute__((ext_vector_type(4))) float4;
typedef float __attribute__((ext_vector_type(2))) float2;
volatile float2 id36531;
float8 id36539;
float4 test173() {
  float8 id36530 =
      __builtin_shufflevector(id36531, id36531, 0, 0, 0, 0, 0, 0, 0, 0);
  return __builtin_shufflevector(id36530, id36539, 12, 10, 14, 4);
}
// ============================================================================

$ C:\work\public-git\upstream-llvm\build-bisect\Release\bin\clang.exe --version
clang version 5.0.0 (http://llvm.org/git/clang.git
0acb2562e4b8b9e2a351e3c335cd28afb2e11847) (http://llvm.org/git/llvm.git
90b785146d1146932624753ba497c8ddf0f1a1e6)
Target: x86_64-scei-ps4
Thread model: posix
InstalledDir: C:\work\public-git\upstream-llvm\build-bisect\Release\bin

$ C:\work\public-git\upstream-llvm\build-bisect\Release\bin\clang.exe -c 1.cpp
-target x86_64-scei-ps4 -O0

$ C:\work\public-git\upstream-llvm\build-bisect\Release\bin\clang.exe -c 1.cpp
-target x86_64-scei-ps4 -O1
WidenVectorOperand op #1: t24: v8f32 = insert_subvector undef:v8f32, t4,
Constant:i64<0>

Do not know how to widen this operator's operand!
UNREACHABLE executed at
C:\work\public-git\upstream-llvm\llvm\lib\CodeGen\SelectionDAG\LegalizeVectorTypes.cpp:2971!
<snip error output>

-- 
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/20170214/7f3d5526/attachment.html>