>>MA-Tune finds the TCB not dispatched and >>looks at the OPSW in the RB, which points to PAUSE.
>I find it hard to believe that a sampling program looks at "RBOPSW" of >something that is not running. That has been pure guessing; I have no clue how sampling programs get to know what happens in the address space they're monitoring. If the sampling program finds the TCB is not dispatched, it wants to find out if the TCB is waiting to be dispatched (CPU delay sample). If not, the TCB is waiting for something. How would it find kind of wait this is, and where in the code the wait was triggered? That's where my guess came from. >As to the question/answer about disablement: while the application itself >is not disabled, almost all of pause processing is. >When sampling is done by external interrupt, there is often a misleading >result since the external interrupt would not have occurred during the >disabled time it will show up right after re-enablement, thus possibly >skewing the result. Interesting. But would you expect that external interrupt to be delayed for several hundredths of a second, indicating that PAUSE processing takes so long. I thought PAUSE / RESUME is a slim process. -- Peter Hunkeler ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
