[PATCH v4 1/5] dm: clk: add stub when CONFIG_CLK is desactivated

[Patrick DELAUNAY]

Hi Simon,

> From: Simon Goldschmidt <simon.k.r.goldschmidt at gmail.com>
> Sent: mercredi 4 mars 2020 20:49
> 
> Am 18.02.2020 um 09:34 schrieb Patrick Delaunay:
> > Add stub for functions clk_...() when CONFIG_CLK is desactivated.
> >
> > This patch avoids compilation issues for driver using these API
> > without protection (#if CONFIG_IS_ENABLED(CLK))
> >
> > For example, before this patch we have undefined reference to
> > `clk_disable_bulk') for code:
> >   clk_disable_bulk(&priv->clks);
> >   clk_release_bulk(&priv->clks);
> >
> > Signed-off-by: Patrick Delaunay <patrick.delaunay at st.com>
> > ---
> >
> > Changes in v4:
> > - Add stub for all functions using 'struct clk' or 'struct clk_bulk'
> >   after remarks on v3
> >
> > Changes in v3:
> > - Add stub for clk_disable_bulk
> >
> > Changes in v2: None
> >
> >  include/clk.h | 101
> > +++++++++++++++++++++++++++++++++++++++++++-------
> >  1 file changed, 88 insertions(+), 13 deletions(-)
> >
> > diff --git a/include/clk.h b/include/clk.h index
> > 3336301815..1fb415ddc8 100644
> > --- a/include/clk.h
> > +++ b/include/clk.h
> > @@ -312,6 +312,7 @@ static inline int clk_release_bulk(struct clk_bulk *bulk)
> >  	return clk_release_all(bulk->clks, bulk->count);  }
> >
> > +#if CONFIG_IS_ENABLED(CLK)
> >  /**
> >   * clk_request - Request a clock by provider-specific ID.
> >   *
> > @@ -433,19 +434,6 @@ int clk_disable_bulk(struct clk_bulk *bulk);
> >   */
> >  bool clk_is_match(const struct clk *p, const struct clk *q);
> >
> > -int soc_clk_dump(void);
> > -
> > -/**
> > - * clk_valid() - check if clk is valid
> > - *
> > - * @clk:	the clock to check
> > - * @return true if valid, or false
> > - */
> > -static inline bool clk_valid(struct clk *clk) -{
> > -	return clk && !!clk->dev;
> > -}
> > -
> >  /**
> >   * clk_get_by_id() - Get the clock by its ID
> >   *
> > @@ -465,6 +453,93 @@ int clk_get_by_id(ulong id, struct clk **clkp);
> >   * @return true on binded, or false on no
> >   */
> >  bool clk_dev_binded(struct clk *clk);
> > +
> > +#else /* CONFIG_IS_ENABLED(CLK) */
> > +
> > +static inline int clk_request(struct udevice *dev, struct clk *clk) {
> > +	return -ENOSYS;
> > +}
> > +
> > +static inline int clk_free(struct clk *clk) {
> > +	return -ENOSYS;
> > +}
> > +
> > +static inline ulong clk_get_rate(struct clk *clk) {
> > +	return -ENOSYS;
> > +}
> > +
> > +static inline struct clk *clk_get_parent(struct clk *clk) {
> > +	return (struct clk *)-ENOSYS;
> 
> This should use ERR_PTR() to care for platforms defining
> CONFIG_ERR_PTR_OFFSET.

Yes I take the point for V5.

> > +}
> > +
> > +static inline long long clk_get_parent_rate(struct clk *clk) {
> > +	return -ENOSYS;
> > +}
> > +
> > +static inline ulong clk_set_rate(struct clk *clk, ulong rate) {
> > +	return -ENOSYS;
> > +}
> > +
> > +static inline int clk_set_parent(struct clk *clk, struct clk *parent)
> > +{
> > +	return -ENOSYS;
> > +}
> > +
> > +static inline int clk_enable(struct clk *clk) {
> > +	return -ENOSYS;
> > +}
> > +
> > +static inline int clk_enable_bulk(struct clk_bulk *bulk) {
> > +	return bulk && bulk->count == 0 ? 0 : -ENOSYS;
> 
> For this test to work, someone would need to set bulk->count to 0. This is normally
> done by clk_get_bulk(), but you defined it to only return an error.
> 
> I guess it works for you because all clk_bulk objects you use are from the heap
> (which is currently zeroed out in U-Boot) or if they are on the stack, you have
> if/else code that doesn't bring you here. Still it seems wrong.

Yes exactly, it is working for me as the bulk was in private date, so zero allocated.

I will update on V5

> Regards,
> Simon

Thanks for the review