llvm dev - Aug 2016 - Is there a sanitizer for "virtual call to a pure virtual function from a constructor..."?

I'm testing a library under a Raspberry Pi 3. Its ARMv8 device, and
its runs Raspbian OS. The library tests OK under other ARM platforms
and other architectures, like i686 and x86_64. The problem seems to be
specific to this platform.

The library's test suite is being aborted with the message:

    ...
    DLIES XOR ran 12 tests all ok
    [New Thread 0x765bf450 (LWP 3674)]
    pure virtual method called
    terminate called without an active exception

I'm thinking the "pure virtual method called" means this could be
the
problem: http://stackoverflow.com/questions/9185990. I have not been
able to zero-in further because I can't get a good back trace and the
tools I usually use are slightly broken in annoying ways. For example,
Valgrind unhandled instruction
(http://bugs.kde.org/show_bug.cgi?id=366464); and Clang 3.7 missing
UBsan libraries (http://bugs.launchpad.net/raspbian/+bug/1610502).

To confirm/deny the pure virtual call from a ctor in the test program,
I'd like to run the test program on x86_64 and see if it triggers a
finding.

Is there a sanitizer for "virtual call to a pure virtual function from
a constructor..."?

Thanks in advance.

ARMv8 but last I checked there is still no ability to run Aarch64 binaries
on the Pi3.

ARMv8 32 bit mode does have some differences from ARMv7, so it's definitely
possible something might work on Pi2 but not Pi3.


On Sat, Aug 6, 2016 at 5:34 PM, Jeffrey Walton via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> I'm testing a library under a Raspberry Pi 3. Its ARMv8 device, and
> its runs Raspbian OS. The library tests OK under other ARM platforms
> and other architectures, like i686 and x86_64. The problem seems to be
> specific to this platform.
>
> The library's test suite is being aborted with the message:
>
>     ...
>     DLIES XOR ran 12 tests all ok
>     [New Thread 0x765bf450 (LWP 3674)]
>     pure virtual method called
>     terminate called without an active exception
>
> I'm thinking the "pure virtual method called" means this
could be the
> problem: http://stackoverflow.com/questions/9185990. I have not been
> able to zero-in further because I can't get a good back trace and the
> tools I usually use are slightly broken in annoying ways. For example,
> Valgrind unhandled instruction
> (http://bugs.kde.org/show_bug.cgi?id=366464); and Clang 3.7 missing
> UBsan libraries (http://bugs.launchpad.net/raspbian/+bug/1610502).
>
> To confirm/deny the pure virtual call from a ctor in the test program,
> I'd like to run the test program on x86_64 and see if it triggers a
> finding.
>
> Is there a sanitizer for "virtual call to a pure virtual function from
> a constructor..."?
>
> Thanks in advance.
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160806/5735c8f2/attachment.html>