Re: [PATCH v6 07/10] proc: flush task dcache entries from all procfs instances

[Alexey Gladkov]

On Mon, Dec 30, 2019 at 02:03:29PM -0800, J Freyensee wrote:
> > +#ifdef CONFIG_PROC_FS
> > +static inline void pidns_proc_lock(struct pid_namespace *pid_ns)
> > +{
> > +	down_write(&pid_ns->rw_proc_mounts);
> > +}
> > +
> > +static inline void pidns_proc_unlock(struct pid_namespace *pid_ns)
> > +{
> > +	up_write(&pid_ns->rw_proc_mounts);
> > +}
> > +
> > +static inline void pidns_proc_lock_shared(struct pid_namespace *pid_ns)
> > +{
> > +	down_read(&pid_ns->rw_proc_mounts);
> > +}
> > +
> > +static inline void pidns_proc_unlock_shared(struct pid_namespace *pid_ns)
> > +{
> > +	up_read(&pid_ns->rw_proc_mounts);
> > +}
> > +#else /* !CONFIG_PROC_FS */
> > +
> Apologies for my newbie question. I couldn't help but notice all these
> function calls are assuming that the parameter struct pid_namespace *pid_ns
> will never be NULL.  Is that a good assumption?

These inline helpers are introduced to improve readability. They only make
sense inside procfs. I don't think that defensive programming is useful
here.

-- 
Rgrds, legion