| Message ID | 146231201072.25631.4767945168905205742.stgit@Solace.fritz.box (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 03/05/16 22:46, Dario Faggioli wrote: > All calculations that involve load_last_update uses quantities > shifted by LOADAVG_GRANULARITY_SHIFT, so make sure that this > is true even when the field is assigned a value for the first > time, during vcpu allocation. > > Also, during migration, while the loads of both the source and > destination runqueues certainly need changing, the vcpu being > moved does not change its running/non-running status, and its > calculated load should hence not be affected. > > Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com> Looks good, thanks. Reviewed-by: George Dunlap <george.dunlap@citrix.com> I suppose the code that sets 'svc->load_last_update = now' should probably have changed the variable name to make it clear that now != NOW(). Maybe next time I'm looking through that code I'll do a bit of renaming. > --- > Cc: George Dunlap <george.dunlap@citrix.com> > Cc: Wei Liu <wei.liu2@citrix.com> > --- > xen/common/sched_credit2.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/xen/common/sched_credit2.c b/xen/common/sched_credit2.c > index 803cc44..393364a 100644 > --- a/xen/common/sched_credit2.c > +++ b/xen/common/sched_credit2.c > @@ -867,7 +867,7 @@ csched2_alloc_vdata(const struct scheduler *ops, struct vcpu *vc, void *dd) > svc->weight = svc->sdom->weight; > /* Starting load of 50% */ > svc->avgload = 1ULL << (CSCHED2_PRIV(ops)->load_window_shift - 1); > - svc->load_last_update = NOW(); > + svc->load_last_update = NOW() >> LOADAVG_GRANULARITY_SHIFT; > } > else > { > @@ -1301,7 +1301,7 @@ static void migrate(const struct scheduler *ops, > if ( __vcpu_on_runq(svc) ) > { > __runq_remove(svc); > - update_load(ops, svc->rqd, svc, -1, now); > + update_load(ops, svc->rqd, NULL, -1, now); > on_runq=1; > } > __runq_deassign(svc); > @@ -1314,7 +1314,7 @@ static void migrate(const struct scheduler *ops, > __runq_assign(svc, trqd); > if ( on_runq ) > { > - update_load(ops, svc->rqd, svc, 1, now); > + update_load(ops, svc->rqd, NULL, 1, now); > runq_insert(ops, svc->vcpu->processor, svc); > runq_tickle(ops, svc->vcpu->processor, svc, now); > SCHED_STAT_CRANK(migrate_on_runq); >
diff --git a/xen/common/sched_credit2.c b/xen/common/sched_credit2.c index 803cc44..393364a 100644 --- a/xen/common/sched_credit2.c +++ b/xen/common/sched_credit2.c @@ -867,7 +867,7 @@ csched2_alloc_vdata(const struct scheduler *ops, struct vcpu *vc, void *dd) svc->weight = svc->sdom->weight; /* Starting load of 50% */ svc->avgload = 1ULL << (CSCHED2_PRIV(ops)->load_window_shift - 1); - svc->load_last_update = NOW(); + svc->load_last_update = NOW() >> LOADAVG_GRANULARITY_SHIFT; } else { @@ -1301,7 +1301,7 @@ static void migrate(const struct scheduler *ops, if ( __vcpu_on_runq(svc) ) { __runq_remove(svc); - update_load(ops, svc->rqd, svc, -1, now); + update_load(ops, svc->rqd, NULL, -1, now); on_runq=1; } __runq_deassign(svc); @@ -1314,7 +1314,7 @@ static void migrate(const struct scheduler *ops, __runq_assign(svc, trqd); if ( on_runq ) { - update_load(ops, svc->rqd, svc, 1, now); + update_load(ops, svc->rqd, NULL, 1, now); runq_insert(ops, svc->vcpu->processor, svc); runq_tickle(ops, svc->vcpu->processor, svc, now); SCHED_STAT_CRANK(migrate_on_runq);
All calculations that involve load_last_update uses quantities shifted by LOADAVG_GRANULARITY_SHIFT, so make sure that this is true even when the field is assigned a value for the first time, during vcpu allocation. Also, during migration, while the loads of both the source and destination runqueues certainly need changing, the vcpu being moved does not change its running/non-running status, and its calculated load should hence not be affected. Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com> --- Cc: George Dunlap <george.dunlap@citrix.com> Cc: Wei Liu <wei.liu2@citrix.com> --- xen/common/sched_credit2.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)