[llvm-bugs] [Bug 37296] New: "No live value at use" assertion failure for atomic store

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=37296

            Bug ID: 37296
           Summary: "No live value at use" assertion failure for atomic
                    store
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Common Code Generator Code
          Assignee: unassignedbugs at nondot.org
          Reporter: smeenai at fb.com
                CC: compnerd at compnerd.org, llvm-bugs at lists.llvm.org,
                    matze at braunis.de

Created attachment 20250
  --> https://bugs.llvm.org/attachment.cgi?id=20250&action=edit
Reduced source and IR

% bin/clang -cc1 -triple i686--windows-msvc19.11.0 -emit-obj -target-cpu
pentium4 -Os -fexceptions vni.mm
clang: ../../lib/CodeGen/MachineScheduler.cpp:1160: void
llvm::ScheduleDAGMILive::updatePressureDiffs(llvm::ArrayRef<llvm::RegisterMaskPair>):
Assertion `VNI && "No live value at use."' failed.

The IR is lightly reduced from the source file, and just running llc on it with
no further arguments should also trigger the assertion (since llc defaults to
-O2). Seems like an MIR level issue, but I'm not familiar enough with MIR to
attempt to reduce at that level.

The reduction is pretty sensitive in that changing any part of it seems to make
the assertion go away, but from running llc with -debug-only machine-scheduler,
the atomic store seems to be the culprit (and making that store non-atomic in
the IR indeed makes the assertion failure go away).

-- 
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/20180501/d64a72c5/attachment.html>