last_100_frees.stp on aarch64 is crashing while doing sprint_ubacktrace()

[Mark Wielaard]

On Fri, 2015-06-19 at 17:01 -0400, William Cohen wrote:
> However, I can't seem to get the stap module to be built with
> debuginfo even with what seem like appropriate command line options.

Are you sure? The following should generate debuginfo for the stap.ko
you are creating:

stap -k -p4 -vv -B CONFIG_DEBUG_INFO=y -e 'probe begin {log("hello"); exit(); }'

If not, please post the -vv output so we can analyze what is going
wrong.

> > runtime/unwind.c:processCFI() works its way through the dwarf data
> > with some protection against e.g. bad data, but it may be
> > insufficient.
> 
> I recall Mark making some changes to make elf handling more robust in
> elfutils. I suspect similar robustification is missing somewhere in
> the systemtap unwinder code.

Maybe indeed. But I don't yet understand what is going in. Hopefully we
can get a module with debuginfo to better analyse.

Thanks,

Mark