> This makes sense, however, I am still a bit confused. You are stating > that in Solaris 9 microstate accounting gets only updated when the lwp > is transitions from one state to another.
No, microstate data only gets updated when the lwp transitions from one state to another. This hasn't changed from one release to another. What is different is that on Solaris 9, procfs, which is the source that prstat uses to get its data about threads and processes, does not include the time spent in the current microstate when usage or lusage files are read. This means that if a lwp hasn't made a state transition recently, the numbers won't come close to matching 100%. In Solaris 10, procfs includes the time spent in the current microstate when generating its usage / lusage data, and thus all the numbers should converge upon 100 % or something very close. So, even though a process may not have made a state transition recently, we check the current state and the time that the lwp entered that state. The data about the current state is added to the usage. I hope this explanation makes a little more sense. -j _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
