<br><tt><font size=2>linuxppc-dev-bounces+lukebrowning=us.ibm.com@ozlabs.org
wrote on 07/15/2006 07:07:58 PM:<br>
<br>
> <br>
> > Not sure I follow you. I would expect the PURR value to be restored
after<br>
> > a context switch, even if we continue on a different physical
CPU.<br>
> > <br>
> > The idea behind monitoring both PURR and timebase is that the
difference<br>
> > between the two tells you how long the partition was suspended
during<br>
> > the hcall.<br>
> <br>
> Sounds good, last time I looked at the patch I thought it was gathering<br>
> the PURR only. That on its own would make for some confusing results.<br>
></font></tt>
<br>
<br><tt><font size=2>It would be more efficient to have a separate trace
log for PHYP dispatches as </font></tt>
<br><tt><font size=2>there are many more hypervisor calls than PHYP dispatches.
I believe PHYP provides</font></tt>
<br><tt><font size=2>a trace records for dispatches, which could be seperately
written. Perf tools can </font></tt>
<br><tt><font size=2>assemble the information from multiple records. It
would be nice to have processor</font></tt>
<br><tt><font size=2>specific time function that reads either PURR or TB
based on the platform.</font></tt>
<br>
<br><tt><font size=2>Luke</font></tt>
<br><tt><font size=2><br>
</font></tt>