[PATCH 0 of 6 v2] xen: sched_credit: fix picking & tickling and also add some tracing

[Dario Faggioli <dario.faggioli@citrix.com>]

Hello everyone,

This is v2 of my previously submitted series about fixing scheduling anomalies
and introducing some tracing in the credit scheduler (with a couple of other
side effects).

All comments v1 got have been addressed and the series grew a couple of more
patches as I found some other issues, still falling under the broad description
given in the above paragraph. Details are given in the single changelogs but,
trying to make review as easy as possible, here it comes a short overview.

 [1 of 6] xen: sched_credit: improve picking up the idlal CPU for a VCPU
 [2 of 6] xen: sched_credit: improve tickling of idle CPUs

Are the fixes to the scheduling anomalies, happening during PCPU picking and
tickling, respectively. The latter has already been extensively discussed (by
me and George, mainly); the former is a new --small but nasty-- thing I
discovered during a couple of heavy tracing sessions. :-)

All the benchmarks have been rerun. No big changes in trends or anything, what
held true for v1 still does here (although, honestly, numbers looks even a
little bit better).

 [3 of 6] xen: sched_credit: use current_on_cpu() when appropriate

Is just (an attempt) to improve code readability.

 [4 of 6] xen: tracing: report where a VCPU wakes up

Is just (an attempt) to improve trace readability.

 [5 of 6] xen: tracing: introduce per-scheduler trace event IDs
 [6 of 6] xen: sched_credit: add some tracing

Finally, are what enables per-scheduler trace record generation already
discussed (again, mostly by me and George) and reworked as suggested and
requested during review of v1 of this series.

Thanks and Regards,
Dario

--
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)