2022-02-11 00:42:19

by Yury Norov

Subject: [PATCH 39/49] arch/s390: replace cpumask_weight with cpumask_weight_eq where appropriate

cfset_all_start() calls cpumask_weight() to compare the weight of cpumask
with a given number. We can do it more efficiently with
cpumask_weight_{eq, ...} because conditional cpumask_weight may stop
traversing the cpumask earlier, as soon as condition is (or can't be) met.

Signed-off-by: Yury Norov <[email protected]>
---
arch/s390/kernel/perf_cpum_cf.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c
index ee8707abdb6a..4d217f7f5ccf 100644
--- a/arch/s390/kernel/perf_cpum_cf.c
+++ b/arch/s390/kernel/perf_cpum_cf.c
@@ -975,7 +975,7 @@ static int cfset_all_start(struct cfset_request *req)
return -ENOMEM;
cpumask_and(mask, &req->mask, cpu_online_mask);
on_each_cpu_mask(mask, cfset_ioctl_on, &p, 1);
- if (atomic_read(&p.cpus_ack) != cpumask_weight(mask)) {
+ if (!cpumask_weight_eq(mask, atomic_read(&p.cpus_ack))) {
on_each_cpu_mask(mask, cfset_ioctl_off, &p, 1);
rc = -EIO;
debug_sprintf_event(cf_dbg, 4, "%s CPUs missing", __func__);
--
2.32.0

2022-02-11 09:29:59

by Sven Schnelle

Subject: Re: [PATCH 39/49] arch/s390: replace cpumask_weight with cpumask_weight_eq where appropriate

Hi Yury,

Yury Norov <[email protected]> writes:

> cfset_all_start() calls cpumask_weight() to compare the weight of cpumask
> with a given number. We can do it more efficiently with
> cpumask_weight_{eq, ...} because conditional cpumask_weight may stop
> traversing the cpumask earlier, as soon as condition is (or can't be) met.
>
> Signed-off-by: Yury Norov <[email protected]>
> ---
> arch/s390/kernel/perf_cpum_cf.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c
> index ee8707abdb6a..4d217f7f5ccf 100644
> --- a/arch/s390/kernel/perf_cpum_cf.c
> +++ b/arch/s390/kernel/perf_cpum_cf.c
> @@ -975,7 +975,7 @@ static int cfset_all_start(struct cfset_request *req)
> return -ENOMEM;
> cpumask_and(mask, &req->mask, cpu_online_mask);
> on_each_cpu_mask(mask, cfset_ioctl_on, &p, 1);
> - if (atomic_read(&p.cpus_ack) != cpumask_weight(mask)) {
> + if (!cpumask_weight_eq(mask, atomic_read(&p.cpus_ack))) {
> on_each_cpu_mask(mask, cfset_ioctl_off, &p, 1);
> rc = -EIO;
> debug_sprintf_event(cf_dbg, 4, "%s CPUs missing", __func__);

given that you're adding a bunch of these functions - gt,lt,eq and
others, i wonder whether it makes sense to also add cpumask_weight_ne(),
so one could just write:

if (cpumask_weight_ne(mask, atomic_read(&p.cpus_ack))) {
...
}

?

/Sven

2022-02-12 16:37:52

by Yury Norov

Subject: Re: [PATCH 39/49] arch/s390: replace cpumask_weight with cpumask_weight_eq where appropriate

On Fri, Feb 11, 2022 at 07:54:26AM +0100, Sven Schnelle wrote:
> Hi Yury,
>
> Yury Norov <[email protected]> writes:
>
> > cfset_all_start() calls cpumask_weight() to compare the weight of cpumask
> > with a given number. We can do it more efficiently with
> > cpumask_weight_{eq, ...} because conditional cpumask_weight may stop
> > traversing the cpumask earlier, as soon as condition is (or can't be) met.
> >
> > Signed-off-by: Yury Norov <[email protected]>
> > ---
> > arch/s390/kernel/perf_cpum_cf.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c
> > index ee8707abdb6a..4d217f7f5ccf 100644
> > --- a/arch/s390/kernel/perf_cpum_cf.c
> > +++ b/arch/s390/kernel/perf_cpum_cf.c
> > @@ -975,7 +975,7 @@ static int cfset_all_start(struct cfset_request *req)
> > return -ENOMEM;
> > cpumask_and(mask, &req->mask, cpu_online_mask);
> > on_each_cpu_mask(mask, cfset_ioctl_on, &p, 1);
> > - if (atomic_read(&p.cpus_ack) != cpumask_weight(mask)) {
> > + if (!cpumask_weight_eq(mask, atomic_read(&p.cpus_ack))) {
> > on_each_cpu_mask(mask, cfset_ioctl_off, &p, 1);
> > rc = -EIO;
> > debug_sprintf_event(cf_dbg, 4, "%s CPUs missing", __func__);
>
> given that you're adding a bunch of these functions - gt,lt,eq and
> others, i wonder whether it makes sense to also add cpumask_weight_ne(),
> so one could just write:
>
> if (cpumask_weight_ne(mask, atomic_read(&p.cpus_ack))) {
> ...
> }
>
> ?

It will have 3 users in cpumask + 1 in nodemask. I have no strong opinion
whether we need it or not. Let's see what people say.

Thanks,
Yury