Re: [DynInst_API:] DYNINSTuserMessage() in a function called with oneTimeCode()

[Matthew LeGendre <legendre1@xxxxxxxx>]

On Mon, 11 Feb 2013, Josh Stone wrote:
> On 02/11/2013 01:47 PM, Matthew LeGendre wrote:
> > Also, your earlier message implied that you use DYNINSTUserMessage in all
> > of your probes.  As an FYI, DYNINSTUserMessage is fairly expensive and
> > involves pausing the mutatee and two context-switches.  It's not something
> > you'd ever want to put into an app's inner loop.  If you see any
> > performance problems you may want to consider replacing it with a named
> > pipe/socket or shared memory.
> Hmm, we do use shared memory between the mutator and mutatee already for
> other common state, and we were considering that as another way to add
> message passing.  If DYNINSTuserMessage is never expected to be
> performant, then perhaps we should try those other things after all.
I doubt DYNINSTuserMessage will be performant in the near future--Dyninst 
doesn't have any fast mutatee/mutator communication infrastructure.  One 
could build some (e.g, add a named pipe), but then you're talking about a 
significant project.
-Matt