Re: [Libunwind-devel] Libunwind support for NULL IP

[Arun Sharma]

On Fri, Mar 16, 2012 at 8:55 AM, Prabhat Verma
<address@hidden> wrote:
> We are trying to replace backtrace with libunwind (local unwinding) for our
> diagnostic purposes. A major motivation behind this decision was the
> inability of backtrace to handle stack frames containing NULL IP. In our
> case, this may happen if someone ends up doing this:

I pushed a few changes to the git repo. Could you please try them to
see if things work better for you?

In short: dwarf says NULL RBP is the end of the callchain. But
libunwind was terminating unwind on NULL RIP as well.

In the process I seem to have introduced a test failure
(run-ptrace-misc used to pass earlier). I'll dig into it when I get a
chance. On the face of it, it looks like unwinding from (ip == 0) used
to succeed earlier, but it fails now (due to lack of unwind info?).

 -Arun