Quoting Sean Paul (2018-05-02 12:03:16)
> On Wed, May 02, 2018 at 10:01:59AM +0530, Sandeep Panda wrote:
>
> > + struct drm_display_mode curr_mode;
> > + struct mutex lock;
> > + unsigned int ctrl_ref_count;
> > +};
> > +
> > +static const struct regmap_range ti_sn_bridge_volatile_ranges[] = {
> > + { .range_min = 0, .range_max = 0xff },
> > +};
> > +
> > +static const struct regmap_access_table ti_sn_bridge_volatile_table = {
> > + .yes_ranges = ti_sn_bridge_volatile_ranges,
> > + .n_yes_ranges = ARRAY_SIZE(ti_sn_bridge_volatile_ranges),
> > +};
> > +
> > +static const struct regmap_config ti_sn_bridge_regmap_config = {
> > + .reg_bits = 8,
> > + .val_bits = 8,
> > + .volatile_table = &ti_sn_bridge_volatile_table,
> > + .cache_type = REGCACHE_NONE,
> > +};
> > +
> > +static int ti_sn_bridge_power_ctrl(struct ti_sn_bridge *pdata, bool enable)
> > +{
> > + int ret = 0;
> > +
> > + mutex_lock(&pdata->lock);
> > + if (enable)
> > + pdata->ctrl_ref_count++;
> > + else
> > + pdata->ctrl_ref_count--;
>
> I think you should use a kref instead of rolling your own ref_count. You can
> handle release by calling kref_put_mutex(), which will handle the reference and
> the lock. On the acquire side, you can use kref_get_unless_zero which will be
> fast if the reference is already active.
Why not use runtime PM?
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
