Rafael Vanoni wrote:
>> --- Comment #3 from Eric Saxe <eric.saxe at sun.com> 2008-12-02 >> 18:12:54 --- Incidentally... >> >> I looked into the cause of all the xcalls showing up in PowerTOP >> (when run on an idle system), and the vast majority of them were due >> to DTrace itself... >> >> *very* few of them were as a result of P-state changes.... >> > > DTrace uses xcalls to sync the per CPU buffers, so high aggregation > rates on multiprocessors generate lots of xcalls. Increasing > PowerTOP's interval (or using the default ;)) minimizes this effect. Can we improve the xcall dtrace probe by adding the hook function as a argument so that we can easily know what causes xcall? Currently we just know the xcall is called, not easy to know what causes it. I can come up with a patch for this if it's considered to be reasonable. Thanks, -Aubrey
