On Fri, 2007-12-21 at 11:29 -0500, Chad Mynhier wrote:
> On 12/21/07, Chad Mynhier <[EMAIL PROTECTED]> wrote:
> > On 12/21/07, Richard L. Hamilton <[EMAIL PROTECTED]> wrote:
> > > [...]
> > > > That, however, doesn't really help in solving the
> > > >  problem that I have
> > > > n my hands -- how do I determine the lifetime of the
> > > > process that
> > > > was started before the D script? Granted, I'm not a
> > > > kernel developer
> > > > so it could be that there's such a way and I would
> > > > love to be educated
> > > > on what it is.
> > >
> > > Is curpsinfo->pr_start of any use (assuming I read the docs right, never
> > > tried this)?   You may have to convert it to something more useful,
> > > i.e. multiply the tv_sec by a billion and then add the tv_nsec.
> > > See proc(4) for more info on struct psinfo (aka psinfo_t).
> >
> > It seems like that should be useful.  If you look at ptime, it does
> > something similar, but it's using the pr_create field of the prusage_t
> > structure (pr_term - pr_create (subtracted as timestruct_t's) for a
> > process started by ptime or gethrtime() (converted to a timestruct_t)
> > - pr_create for an extant process with the "-p <pid>" option which
> > should soon be putback into OpenSoalris.)
> 
> Doh!  I should have paid more attention to the start of this thread.
> Using curpsinfo->pr_start is what the OP (Roman) did.
> 
> I guess the ideal thing in this case would be if we exposed the
> prusage structure (curprusage?) so that we could use timestamp and
> curprusage->pr_create.  (I'm just throwing this out there, I haven't
> looked yet to see if there's some way to access that or if it would be
> reasonable or practical to expose that structure.)

  <me showing that I'm not a kernel engineer>
  My ideal scenario would be to get a hold of exactly the same
variable from which curpsinfo->pr_start gets initialized by the
kernel.
   </me showing that I'm not a kernel engineer>

  Thoughts?

Thanks,
Roman.

_______________________________________________
dtrace-discuss mailing list
[email protected]

Reply via email to