|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] [PATCH 1/4] xen: sched: avoid dumping duplicate information
On Thu, Jun 25, 2015 at 1:15 PM, Dario Faggioli
<dario.faggioli@xxxxxxxxxx> wrote:
> When dumping scheduling information (debug key 'r'), what
> we print as 'Idle cpupool' is pretty much the same of what
> we print immediately after as 'Cpupool0'. In fact, if there
> are no pCPUs outside of any cpupools, it is exactly the
> same.
>
> If there are free pCPUs, there is some valuable information,
> but still a lot of duplication:
>
> (XEN) Online Cpus: 0-15
> (XEN) Free Cpus: 8
> (XEN) Idle cpupool:
> (XEN) Scheduler: SMP Credit Scheduler (credit)
> (XEN) info:
> (XEN) ncpus = 13
> (XEN) master = 0
> (XEN) credit = 3900
> (XEN) credit balance = 45
> (XEN) weight = 1280
> (XEN) runq_sort = 11820
> (XEN) default-weight = 256
> (XEN) tslice = 30ms
> (XEN) ratelimit = 1000us
> (XEN) credits per msec = 10
> (XEN) ticks per tslice = 3
> (XEN) migration delay = 0us
> (XEN) idlers: 00000000,00006d29
> (XEN) active vcpus:
> (XEN) 1: [1.7] pri=-1 flags=0 cpu=15 credit=-116 [w=256,cap=0] (84+300)
> {a/i=22/21 m=18+5 (k=0)}
> (XEN) 2: [1.3] pri=0 flags=0 cpu=1 credit=-113 [w=256,cap=0] (87+300)
> {a/i=37/36 m=11+544 (k=0)}
> (XEN) 3: [0.15] pri=-1 flags=0 cpu=4 credit=95 [w=256,cap=0] (210+300)
> {a/i=127/126 m=108+9 (k=0)}
> (XEN) 4: [0.10] pri=-2 flags=0 cpu=12 credit=-287 [w=256,cap=0]
> (-84+300) {a/i=163/162 m=36+568 (k=0)}
> (XEN) 5: [0.7] pri=-2 flags=0 cpu=2 credit=-242 [w=256,cap=0] (-42+300)
> {a/i=129/128 m=16+50 (k=0)}
> (XEN) CPU[08] sort=5791, sibling=00000000,00000300, core=00000000,0000ff00
> (XEN) run: [32767.8] pri=-64 flags=0 cpu=8
> (XEN) Cpupool 0:
> (XEN) Cpus: 0-5,10-15
> (XEN) Scheduler: SMP Credit Scheduler (credit)
> (XEN) info:
> (XEN) ncpus = 13
> (XEN) master = 0
> (XEN) credit = 3900
> (XEN) credit balance = 45
> (XEN) weight = 1280
> (XEN) runq_sort = 11820
> (XEN) default-weight = 256
> (XEN) tslice = 30ms
> (XEN) ratelimit = 1000us
> (XEN) credits per msec = 10
> (XEN) ticks per tslice = 3
> (XEN) migration delay = 0us
> (XEN) idlers: 00000000,00006d29
> (XEN) active vcpus:
> (XEN) 1: [1.7] pri=-1 flags=0 cpu=15 credit=-116 [w=256,cap=0] (84+300)
> {a/i=22/21 m=18+5 (k=0)}
> (XEN) 2: [1.3] pri=0 flags=0 cpu=1 credit=-113 [w=256,cap=0] (87+300)
> {a/i=37/36 m=11+544 (k=0)}
> (XEN) 3: [0.15] pri=-1 flags=0 cpu=4 credit=95 [w=256,cap=0] (210+300)
> {a/i=127/126 m=108+9 (k=0)}
> (XEN) 4: [0.10] pri=-2 flags=0 cpu=12 credit=-287 [w=256,cap=0]
> (-84+300) {a/i=163/162 m=36+568 (k=0)}
> (XEN) 5: [0.7] pri=-2 flags=0 cpu=2 credit=-242 [w=256,cap=0] (-42+300)
> {a/i=129/128 m=16+50 (k=0)}
> (XEN) CPU[00] sort=11801, sibling=00000000,00000003, core=00000000,000000ff
> (XEN) run: [32767.0] pri=-64 flags=0 cpu=0
> ... ... ...
> (XEN) CPU[15] sort=11820, sibling=00000000,0000c000, core=00000000,0000ff00
> (XEN) run: [1.7] pri=-1 flags=0 cpu=15 credit=-116 [w=256,cap=0] (84+300)
> {a/i=22/21 m=18+5 (k=0)}
> (XEN) 1: [32767.15] pri=-64 flags=0 cpu=15
> (XEN) Cpupool 1:
> (XEN) Cpus: 6-7,9
> (XEN) Scheduler: SMP RTDS Scheduler (rtds)
> (XEN) CPU[06]
> (XEN) CPU[07]
> (XEN) CPU[09]
>
> With this change, we get rid of the redundancy, and retain
> only the information about the free pCPUs.
>
> (While there, turn a loop index variable from `int' to
> `unsigned int' in schedule_dump().)
>
> Signed-off-by: Dario Faggioli <dario.faggioli@xxxxxxxxxx>
Acked-by: George Dunlap <george.dunlap@xxxxxxxxxxxxx>
> ---
> Cc: Juergen Gross <jgross@xxxxxxxx>
> Cc: George Dunlap <george.dunlap@xxxxxxxxxxxxx>
> ---
> xen/common/cpupool.c | 6 +++---
> xen/common/schedule.c | 18 +++++++++++++-----
> 2 files changed, 16 insertions(+), 8 deletions(-)
>
> diff --git a/xen/common/cpupool.c b/xen/common/cpupool.c
> index 563864d..5471f93 100644
> --- a/xen/common/cpupool.c
> +++ b/xen/common/cpupool.c
> @@ -728,10 +728,10 @@ void dump_runq(unsigned char key)
>
> print_cpumap("Online Cpus", &cpu_online_map);
> if ( !cpumask_empty(&cpupool_free_cpus) )
> + {
> print_cpumap("Free Cpus", &cpupool_free_cpus);
> -
> - printk("Idle cpupool:\n");
> - schedule_dump(NULL);
> + schedule_dump(NULL);
> + }
>
> for_each_cpupool(c)
> {
> diff --git a/xen/common/schedule.c b/xen/common/schedule.c
> index ecf1545..4ffcd98 100644
> --- a/xen/common/schedule.c
> +++ b/xen/common/schedule.c
> @@ -1473,16 +1473,24 @@ void scheduler_free(struct scheduler *sched)
>
> void schedule_dump(struct cpupool *c)
> {
> - int i;
> + unsigned int i;
> struct scheduler *sched;
> cpumask_t *cpus;
>
> /* Locking, if necessary, must be handled withing each scheduler */
>
> - sched = (c == NULL) ? &ops : c->sched;
> - cpus = cpupool_scheduler_cpumask(c);
> - printk("Scheduler: %s (%s)\n", sched->name, sched->opt_name);
> - SCHED_OP(sched, dump_settings);
> + if ( c != NULL )
> + {
> + sched = c->sched;
> + cpus = c->cpu_valid;
> + printk("Scheduler: %s (%s)\n", sched->name, sched->opt_name);
> + SCHED_OP(sched, dump_settings);
> + }
> + else
> + {
> + sched = &ops;
> + cpus = &cpupool_free_cpus;
> + }
>
> for_each_cpu (i, cpus)
> {
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel
|
 |
Lists.xenproject.org is hosted with RackSpace, monitoring our |