[llvm-bugs] [Bug 36116] New: After r299770, Assertion failed: (New->getType() == getType() && "replaceAllUses of value with new value of different type!"), function doRAUW, file lib/IR/Value.cpp, line 375.

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=36116

            Bug ID: 36116
           Summary: After r299770, Assertion failed: (New->getType() ==
                    getType() && "replaceAllUses of value with new value
                    of different type!"), function doRAUW, file
                    lib/IR/Value.cpp, line 375.
           Product: new-bugs
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: dimitry at andric.com
                CC: llvm-bugs at lists.llvm.org

As reported in https://bugs.freebsd.org/225471, building the FreeBSD
emulators/snes9x port for x86_64 with clang 6.0.0 results in an assertion:

Assertion failed: (New->getType() == getType() && "replaceAllUses of value with
new value of different type!"), function doRAUW, file
/poudriere/jails/head-amd64/usr/src/contrib/llvm/lib/IR/Value.cpp, line 404.

This assertion still occurs with trunk r322755, and bisection shows it has
regressed with http://reviews.llvm.org/rL299770 ("Reapply r298620: [LV]
Vectorize GEPs") by Matthew Simpson.  (Which was a reapplication of
https://reviews.llvm.org/rL298620.)

Minimized test case:

// clang -cc1 -triple x86_64 -S -O2 -vectorize-loops fxemu-minimized.cpp
struct {
  unsigned char *a[];
} b;
unsigned char c;
int d;
void e() {
  d = 0;
  for (; d < 32; d++)
    b.a[d] = &c + ((d & 16) << 1);
}

Note that -vectorize-loops is essential for reproducing the assertion.

As far as I can see, this is *not* a duplicate of bug 27969, bug 30783, bug
34783 or bug 34759, as those are all resolved, and this assertion still occurs
as of 2018-01-27.

-- 
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/20180126/37ef9ec5/attachment.html>