Re: [PATCH] cpufreq: Fix get_cpu_device() failed in add_cpu_dev_symlink()

[Rafael J. Wysocki]

On Mon, Nov 29, 2021 at 10:10 AM Viresh Kumar <viresh.kumar@xxxxxxxxxx> wrote:
>
> On 29-11-21, 16:02, Xiongfeng Wang wrote:
> > When I hot added a CPU, I found 'cpufreq' directory is not created below
> > /sys/devices/system/cpu/cpuX/. It is because get_cpu_device() failed in
> > add_cpu_dev_symlink().
> >
> > cpufreq_add_dev() is the .add_dev callback of a CPU subsys interface. It
> > will be called when the CPU device registered into the system. The stack
> > is as follows.
> >   register_cpu()
> >   ->device_register()
> >    ->device_add()
> >     ->bus_probe_device()
> >      ->cpufreq_add_dev()
> >
> > But only after the CPU device has been registered, we can get the CPU
> > device by get_cpu_device(), otherwise it will return NULL. Since we
> > already have the CPU device in cpufreq_add_dev(), pass it to
> > add_cpu_dev_symlink(). I noticed that the 'kobj' of the cpu device has
> > been added into the system before cpufreq_add_dev().
> >
> > Signed-off-by: Xiongfeng Wang <wangxiongfeng2@xxxxxxxxxx>
> > ---
> >  drivers/cpufreq/cpufreq.c | 9 ++++-----
> >  1 file changed, 4 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
> > index e338d2f010fe..22aa2793e4d2 100644
> > --- a/drivers/cpufreq/cpufreq.c
> > +++ b/drivers/cpufreq/cpufreq.c
> > @@ -1004,10 +1004,9 @@ static struct kobj_type ktype_cpufreq = {
> >       .release        = cpufreq_sysfs_release,
> >  };
> >
> > -static void add_cpu_dev_symlink(struct cpufreq_policy *policy, unsigned int cpu)
> > +static void add_cpu_dev_symlink(struct cpufreq_policy *policy, unsigned int cpu,
> > +                             struct device *dev)
> >  {
> > -     struct device *dev = get_cpu_device(cpu);
> > -
> >       if (unlikely(!dev))
> >               return;
> >
> > @@ -1391,7 +1390,7 @@ static int cpufreq_online(unsigned int cpu)
> >       if (new_policy) {
> >               for_each_cpu(j, policy->related_cpus) {
> >                       per_cpu(cpufreq_cpu_data, j) = policy;
> > -                     add_cpu_dev_symlink(policy, j);
> > +                     add_cpu_dev_symlink(policy, j, get_cpu_device(j));
> >               }
> >
> >               policy->min_freq_req = kzalloc(2 * sizeof(*policy->min_freq_req),
> > @@ -1565,7 +1564,7 @@ static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
> >       /* Create sysfs link on CPU registration */
> >       policy = per_cpu(cpufreq_cpu_data, cpu);
> >       if (policy)
> > -             add_cpu_dev_symlink(policy, cpu);
> > +             add_cpu_dev_symlink(policy, cpu, dev);
> >
> >       return 0;
> >  }
>
> Interesting that I never hit it earlier despite doing rigorous testing of
> hotplug stuff :(

This is the real hot-add path which isn't tested on a regular basis.

> Anyway the patch is okay,

It would be good to add a Fixes: tag to it, though.  Any idea about
the commit this should point to?

> Acked-by: Viresh Kumar <viresh.kumar@xxxxxxxxxx>