On Mon, Feb 13, 2017 at 10:47 AM, Sebastian Lackner <sebastian@fds-team.de> wrote:

Doesn't this suffer from the same problem as before? We might not have unwinding information
for the unwind_rdi_rsi_handler function if it is not supported by all x86_64 compilers.
Actually, I believe the change itself would also be acceptable even without tests.

Correct, I was resubmitting the patch set primarily for the debug registers patch. I didn't change this patch, because I wasn't sure how to avoid the issue since the whole point of the test is to do unwinding of the stack.

When I resubmit the patch set, I will exclude this test. The primary purpose was just to show that this is an actual bug and the patch fixes it.

-Andrew