[llvm-bugs] [Bug 39073] New: "fatal error: error in backend: Unable to copy EFLAGS physical register!"

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=39073

            Bug ID: 39073
           Summary: "fatal error: error in backend: Unable to copy EFLAGS
                    physical register!"
           Product: new-bugs
           Version: 7.0
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: willdtz at gmail.com
                CC: llvm-bugs at lists.llvm.org

Occurs with LLVM 7 and trunk as of yesterday.
I've reproduced the problem using the prebuilt Ubuntu toolchain (in a docker
container, 16.04), as well.

Attached are the clang reproducers -- paths will need minor touchups to execute
on your system. Well I'll attach them immediately after submitting anyway.

Seems to be caused by using `-fPIC -mcmodel=large`, which I used with previous
versions of clang (4,5,6) when building code intended for the JIT.

If you're curious or in case it helps, the source file being compiled is this
one from musl:

https://raw.githubusercontent.com/dtzWill/musl/master/src/misc/a64l.c
(
https://raw.githubusercontent.com/dtzWill/musl/c50985d5c8e316c5c464f352e79eeebfed1121a9/src/misc/a64l.c
)

I'm happy to offer more details as needed, let me know if there's anything that
would help.

This is not personally a huge blocker, but since it's a fatal error and a
relatively simple test function, hopefully it helps pinpoint a problem that
maybe is a big deal for someone :).

-- 
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/20180925/4f3106ed/attachment-0001.html>