Re: [PATCH] kprobes: add kprobe_is_function_probed()

[Seth Jennings <sjenning@redhat.com>]

On Tue, Oct 21, 2014 at 09:40:32PM -0500, Josh Poimboeuf wrote:
> On Tue, Oct 21, 2014 at 11:25:56PM +0200, Jiri Kosina wrote:
> > On Tue, 21 Oct 2014, Josh Poimboeuf wrote:
> > > 
> > > I'm guessing kGraft doesn't have the address + length?  I think you
> > > could call kallsyms_lookup() to get both values.
> > > 
> > > Maybe we should see what our unified live patching code ends up looking
> > > like before deciding what interface(s) we need here?
> > 
> > Yes, that probably makes sense indeed. I am talking to David Miller wrt. 
> > mailinglist creation on vger.kernel.org as we speak, hopefully it'll 
> > materialize soon.
> 
> Ok, thanks!  Seth is currently slaving away on the code :-)

Yes, I am :)  Let me know if this impacts the information we need to
pass via the *_register() call to the core module.

Currently, I pass the old function name (char *), new function pointer
(void *), and the old_addr (optional, unsigned long).  The old_addr
serves to identify the old function by address instead of name if that
information is provided.

Thanks,
Seth