Re: [PATCH 01/30] drm: Introduce drm_bridge_init()

From: Daniel Vetter
Date: Tue Nov 26 2019 - 09:26:19 EST


On Tue, Nov 26, 2019 at 01:15:59PM +0000, Mihail Atanassov wrote:
> A simple convenience function to initialize the struct drm_bridge.
>
> Signed-off-by: Mihail Atanassov <mihail.atanassov@xxxxxxx>

The commit message here leaves figuring out why we need this to the
reader. Reading ahead the reasons seems to be to roll out bridge->dev
setting for everyone, so that we can set the device_link. Please explain
that in the commit message so the patch is properly motivated.

> ---
> drivers/gpu/drm/drm_bridge.c | 29 +++++++++++++++++++++++++++++
> include/drm/drm_bridge.h | 4 ++++
> 2 files changed, 33 insertions(+)
>
> diff --git a/drivers/gpu/drm/drm_bridge.c b/drivers/gpu/drm/drm_bridge.c
> index cba537c99e43..cbe680aa6eac 100644
> --- a/drivers/gpu/drm/drm_bridge.c
> +++ b/drivers/gpu/drm/drm_bridge.c
> @@ -89,6 +89,35 @@ void drm_bridge_remove(struct drm_bridge *bridge)
> }
> EXPORT_SYMBOL(drm_bridge_remove);
>
> +/**
> + * drm_bridge_init - initialise a drm_bridge structure
> + *
> + * @bridge: bridge control structure
> + * @funcs: control functions
> + * @dev: device
> + * @timings: timing specification for the bridge; optional (may be NULL)
> + * @driver_private: pointer to the bridge driver internal context (may be NULL)

Please also sprinkle some links to this new function to relevant places,
I'd add at least:

"Drivers should call drm_bridge_init() first." to the kerneldoc for
drm_bridge_add. drm_bridge_add should also mention drm_bridge_remove as
the undo function.

And perhaps a longer paragraph to &struct drm_bridge:

"Bridge drivers should call drm_bridge_init() to initialized a bridge
driver, and then register it with drm_bridge_add().

"Users of bridges link a bridge driver into their overall display output
pipeline by calling drm_bridge_attach()."

> + */
> +void drm_bridge_init(struct drm_bridge *bridge, struct device *dev,
> + const struct drm_bridge_funcs *funcs,
> + const struct drm_bridge_timings *timings,
> + void *driver_private)
> +{
> + WARN_ON(!funcs);
> +
> + bridge->dev = NULL;

Given that the goal here is to get bridge->dev set up, why not

WARN_ON(!dev);
bridge->dev = dev;

That should help us to really move forward with all this.
-Daniel

> + bridge->encoder = NULL;
> + bridge->next = NULL;
> +
> +#ifdef CONFIG_OF
> + bridge->of_node = dev->of_node;
> +#endif
> + bridge->timings = timings;
> + bridge->funcs = funcs;
> + bridge->driver_private = driver_private;
> +}
> +EXPORT_SYMBOL(drm_bridge_init);
> +
> /**
> * drm_bridge_attach - attach the bridge to an encoder's chain
> *
> diff --git a/include/drm/drm_bridge.h b/include/drm/drm_bridge.h
> index c0a2286a81e9..d6d9d5301551 100644
> --- a/include/drm/drm_bridge.h
> +++ b/include/drm/drm_bridge.h
> @@ -402,6 +402,10 @@ struct drm_bridge {
>
> void drm_bridge_add(struct drm_bridge *bridge);
> void drm_bridge_remove(struct drm_bridge *bridge);
> +void drm_bridge_init(struct drm_bridge *bridge, struct device *dev,
> + const struct drm_bridge_funcs *funcs,
> + const struct drm_bridge_timings *timings,
> + void *driver_private);
> struct drm_bridge *of_drm_find_bridge(struct device_node *np);
> int drm_bridge_attach(struct drm_encoder *encoder, struct drm_bridge *bridge,
> struct drm_bridge *previous);
> --
> 2.23.0
>

--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch