On 02/02/2016 10:54 PM, Viresh Kumar wrote:
On 02-02-16, 17:32, Saravana Kannan wrote:
But if we are expecting sched dvfs to come in, why make it worse for it. It
would be completely pointless to try and shoehorn sched dvfs to use
cpufreq_governor.c
We can move the common part to cpufreq core
On 02/02/2016 10:57 PM, Viresh Kumar wrote:
On 02-02-16, 20:03, Saravana Kannan wrote:
This is the hotplug case I mentioned. The sysfs file removals will happen
only for the last CPU in that policy (we thankfully optimized that part last
year). We also know that multiple CPUs can't be
On Wed, Feb 3, 2016 at 2:21 PM, Viresh Kumar wrote:
> On 03-02-16, 13:42, Rafael J. Wysocki wrote:
>> > +static ssize_t governor_show(struct kobject *kobj, struct attribute *attr,
>> > +char *buf)
>> > +{
>> > + struct dbs_data *dbs_data = to_dbs_data(kobj);
>> >
On 03-02-16, 13:42, Rafael J. Wysocki wrote:
> > +static ssize_t governor_show(struct kobject *kobj, struct attribute *attr,
> > +char *buf)
> > +{
> > + struct dbs_data *dbs_data = to_dbs_data(kobj);
> > + struct governor_attr *gattr = to_gov_attr(attr);
>
On Wed, Feb 3, 2016 at 7:58 AM, Viresh Kumar wrote:
> On 02-02-16, 22:23, Rafael J. Wysocki wrote:
>> On Tue, Feb 2, 2016 at 11:57 AM, Viresh Kumar
>> wrote:
>
>> "The ondemand and conservative governors use the global-attr or
>> freq-attr structures to represent sysfs attributes corresponding
On 03-02-16, 10:51, Juri Lelli wrote:
> I also think that sched-dvfs should not use cpufreq_governor.c. It is
> useful boilerplate code for ondemand and conservative, as they share lot
> of data structures and how they work, but it doesn't necessarily suit
> everybody's needs, IMHO.
>
> OTOH,
On 03/02/16 12:24, Viresh Kumar wrote:
> On 02-02-16, 17:32, Saravana Kannan wrote:
> > But if we are expecting sched dvfs to come in, why make it worse for it. It
> > would be completely pointless to try and shoehorn sched dvfs to use
> > cpufreq_governor.c
>
> We can move the common part to
On 03-02-16, 10:51, Juri Lelli wrote:
> I also think that sched-dvfs should not use cpufreq_governor.c. It is
> useful boilerplate code for ondemand and conservative, as they share lot
> of data structures and how they work, but it doesn't necessarily suit
> everybody's needs, IMHO.
>
> OTOH,
On 03/02/16 12:24, Viresh Kumar wrote:
> On 02-02-16, 17:32, Saravana Kannan wrote:
> > But if we are expecting sched dvfs to come in, why make it worse for it. It
> > would be completely pointless to try and shoehorn sched dvfs to use
> > cpufreq_governor.c
>
> We can move the common part to
On 03-02-16, 13:42, Rafael J. Wysocki wrote:
> > +static ssize_t governor_show(struct kobject *kobj, struct attribute *attr,
> > +char *buf)
> > +{
> > + struct dbs_data *dbs_data = to_dbs_data(kobj);
> > + struct governor_attr *gattr = to_gov_attr(attr);
>
On Wed, Feb 3, 2016 at 2:21 PM, Viresh Kumar wrote:
> On 03-02-16, 13:42, Rafael J. Wysocki wrote:
>> > +static ssize_t governor_show(struct kobject *kobj, struct attribute *attr,
>> > +char *buf)
>> > +{
>> > + struct dbs_data *dbs_data
On Wed, Feb 3, 2016 at 7:58 AM, Viresh Kumar wrote:
> On 02-02-16, 22:23, Rafael J. Wysocki wrote:
>> On Tue, Feb 2, 2016 at 11:57 AM, Viresh Kumar
>> wrote:
>
>> "The ondemand and conservative governors use the global-attr or
>> freq-attr
On 02/02/2016 10:57 PM, Viresh Kumar wrote:
On 02-02-16, 20:03, Saravana Kannan wrote:
This is the hotplug case I mentioned. The sysfs file removals will happen
only for the last CPU in that policy (we thankfully optimized that part last
year). We also know that multiple CPUs can't be
On 02/02/2016 10:54 PM, Viresh Kumar wrote:
On 02-02-16, 17:32, Saravana Kannan wrote:
But if we are expecting sched dvfs to come in, why make it worse for it. It
would be completely pointless to try and shoehorn sched dvfs to use
cpufreq_governor.c
We can move the common part to cpufreq core
On 02-02-16, 22:23, Rafael J. Wysocki wrote:
> On Tue, Feb 2, 2016 at 11:57 AM, Viresh Kumar wrote:
> "The ondemand and conservative governors use the global-attr or
> freq-attr structures to represent sysfs attributes corresponding to
> their tunables
> (which of them is actually used depends
On 02-02-16, 20:03, Saravana Kannan wrote:
> This is the hotplug case I mentioned. The sysfs file removals will happen
> only for the last CPU in that policy (we thankfully optimized that part last
> year). We also know that multiple CPUs can't be hotplugged at the same time.
> So, in the start of
On 02-02-16, 17:32, Saravana Kannan wrote:
> But if we are expecting sched dvfs to come in, why make it worse for it. It
> would be completely pointless to try and shoehorn sched dvfs to use
> cpufreq_governor.c
We can move the common part to cpufreq core and not make sched-dvfs
reuse
On 02-02-16, 14:21, Saravana Kannan wrote:
> I also don't like this patch because it forces governors to either implement
> their own macros and management of their attributes or force them to use the
> governor structs that come with cpufreq_governor.h. cpufreq_governor.h IMHO
> is very ondemand
On 02-02-16, 17:01, Juri Lelli wrote:
> Hi Rafael,
>
> On 02/02/16 17:35, Rafael J. Wysocki wrote:
> > On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
> > > This patch cleans things up a lot, that's good.
> > >
> > > One thing I'm still concerned about, though: don't we need some locking
> >
On 02/02/2016 05:52 PM, Rafael J. Wysocki wrote:
On Wed, Feb 3, 2016 at 2:32 AM, Saravana Kannan wrote:
On 02/02/2016 05:07 PM, Rafael J. Wysocki wrote:
On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki
wrote:
On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan
wrote:
On 02/02/2016 11:40
On Wed, Feb 3, 2016 at 2:32 AM, Saravana Kannan wrote:
> On 02/02/2016 05:07 PM, Rafael J. Wysocki wrote:
>>
>> On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki
>> wrote:
>>>
>>> On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan
>>> wrote:
On 02/02/2016 11:40 AM, Rafael J. Wysocki
On 02/02/2016 05:07 PM, Rafael J. Wysocki wrote:
On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan wrote:
On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
[cut]
I also don't
On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki wrote:
> On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan
> wrote:
>> On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
>>>
>>> On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
[cut]
>>
>> I also don't like this patch because it forces
On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan wrote:
> On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
>>
>> On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
>>>
>>> Hi Rafael,
>>>
>>> On 02/02/16 17:35, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
Hi Rafael,
On 02/02/16 17:35, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
Hi Viresh,
On 02/02/16 16:27, Viresh Kumar wrote:
Until now, governors
On Tue, Feb 2, 2016 at 11:57 AM, Viresh Kumar wrote:
First, the subject might be better. What about something like
"cpufreq: governor: New sysfs show/store callbacks for governor
tunables", for example?
> Until now, governors (ondemand/conservative) were using the
> 'global-attr' or
On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
> Hi Rafael,
>
> On 02/02/16 17:35, Rafael J. Wysocki wrote:
>> On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
>> > Hi Viresh,
>> >
>> > On 02/02/16 16:27, Viresh Kumar wrote:
>> >> Until now, governors (ondemand/conservative) were using the
Hi Rafael,
On 02/02/16 17:35, Rafael J. Wysocki wrote:
> On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
> > Hi Viresh,
> >
> > On 02/02/16 16:27, Viresh Kumar wrote:
> >> Until now, governors (ondemand/conservative) were using the
> >> 'global-attr' or 'freq-attr', depending on the sysfs
On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
> Hi Viresh,
>
> On 02/02/16 16:27, Viresh Kumar wrote:
>> Until now, governors (ondemand/conservative) were using the
>> 'global-attr' or 'freq-attr', depending on the sysfs location where we
>> want to create governor's directory.
>>
>> The
Hi Viresh,
On 02/02/16 16:27, Viresh Kumar wrote:
> Until now, governors (ondemand/conservative) were using the
> 'global-attr' or 'freq-attr', depending on the sysfs location where we
> want to create governor's directory.
>
> The problem is that, in case of 'freq-attr', we are forced to use
>
Until now, governors (ondemand/conservative) were using the
'global-attr' or 'freq-attr', depending on the sysfs location where we
want to create governor's directory.
The problem is that, in case of 'freq-attr', we are forced to use
show()/store() present in cpufreq.c, which always take
On Tue, Feb 2, 2016 at 11:57 AM, Viresh Kumar wrote:
First, the subject might be better. What about something like
"cpufreq: governor: New sysfs show/store callbacks for governor
tunables", for example?
> Until now, governors (ondemand/conservative) were using the
>
On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
Hi Rafael,
On 02/02/16 17:35, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
Hi Viresh,
On 02/02/16 16:27, Viresh Kumar
On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan wrote:
> On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
>>
>> On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
>>>
>>> Hi Rafael,
>>>
>>> On 02/02/16 17:35, Rafael J. Wysocki wrote:
On Tue,
On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki wrote:
> On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan
> wrote:
>> On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
>>>
>>> On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
On 02/02/2016 05:07 PM, Rafael J. Wysocki wrote:
On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan wrote:
On 02/02/2016 11:40 AM, Rafael J. Wysocki wrote:
On Tue, Feb 2, 2016 at 6:01 PM,
On Wed, Feb 3, 2016 at 2:32 AM, Saravana Kannan wrote:
> On 02/02/2016 05:07 PM, Rafael J. Wysocki wrote:
>>
>> On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki
>> wrote:
>>>
>>> On Tue, Feb 2, 2016 at 11:21 PM, Saravana Kannan
On Tue, Feb 2, 2016 at 6:01 PM, Juri Lelli wrote:
> Hi Rafael,
>
> On 02/02/16 17:35, Rafael J. Wysocki wrote:
>> On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
>> > Hi Viresh,
>> >
>> > On 02/02/16 16:27, Viresh Kumar wrote:
>> >> Until now, governors
Hi Viresh,
On 02/02/16 16:27, Viresh Kumar wrote:
> Until now, governors (ondemand/conservative) were using the
> 'global-attr' or 'freq-attr', depending on the sysfs location where we
> want to create governor's directory.
>
> The problem is that, in case of 'freq-attr', we are forced to use
>
Hi Rafael,
On 02/02/16 17:35, Rafael J. Wysocki wrote:
> On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
> > Hi Viresh,
> >
> > On 02/02/16 16:27, Viresh Kumar wrote:
> >> Until now, governors (ondemand/conservative) were using the
> >> 'global-attr' or 'freq-attr',
Until now, governors (ondemand/conservative) were using the
'global-attr' or 'freq-attr', depending on the sysfs location where we
want to create governor's directory.
The problem is that, in case of 'freq-attr', we are forced to use
show()/store() present in cpufreq.c, which always take
On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
> Hi Viresh,
>
> On 02/02/16 16:27, Viresh Kumar wrote:
>> Until now, governors (ondemand/conservative) were using the
>> 'global-attr' or 'freq-attr', depending on the sysfs location where we
>> want to create governor's
On 02-02-16, 17:01, Juri Lelli wrote:
> Hi Rafael,
>
> On 02/02/16 17:35, Rafael J. Wysocki wrote:
> > On Tue, Feb 2, 2016 at 4:47 PM, Juri Lelli wrote:
> > > This patch cleans things up a lot, that's good.
> > >
> > > One thing I'm still concerned about, though: don't we
On 02-02-16, 14:21, Saravana Kannan wrote:
> I also don't like this patch because it forces governors to either implement
> their own macros and management of their attributes or force them to use the
> governor structs that come with cpufreq_governor.h. cpufreq_governor.h IMHO
> is very ondemand
On 02-02-16, 17:32, Saravana Kannan wrote:
> But if we are expecting sched dvfs to come in, why make it worse for it. It
> would be completely pointless to try and shoehorn sched dvfs to use
> cpufreq_governor.c
We can move the common part to cpufreq core and not make sched-dvfs
reuse
On 02-02-16, 20:03, Saravana Kannan wrote:
> This is the hotplug case I mentioned. The sysfs file removals will happen
> only for the last CPU in that policy (we thankfully optimized that part last
> year). We also know that multiple CPUs can't be hotplugged at the same time.
> So, in the start of
On 02-02-16, 22:23, Rafael J. Wysocki wrote:
> On Tue, Feb 2, 2016 at 11:57 AM, Viresh Kumar wrote:
> "The ondemand and conservative governors use the global-attr or
> freq-attr structures to represent sysfs attributes corresponding to
> their tunables
> (which of them
On 02/02/2016 05:52 PM, Rafael J. Wysocki wrote:
On Wed, Feb 3, 2016 at 2:32 AM, Saravana Kannan wrote:
On 02/02/2016 05:07 PM, Rafael J. Wysocki wrote:
On Wed, Feb 3, 2016 at 12:42 AM, Rafael J. Wysocki
wrote:
On Tue, Feb 2, 2016 at 11:21 PM,
48 matches
Mail list logo