Re: [PATCH v6 37/37] firmware: arm_scmi: add dynamic scmi devices creation

[Cristian Marussi]

Hi

On Mon, Mar 15, 2021 at 08:33:27AM +0000, Sudeep Holla wrote:
> Hi Cristian,
> 
> Sorry for the delay.
> 

No worries.

> On Tue, Feb 02, 2021 at 10:15:55PM +0000, Cristian Marussi wrote:
> > Having added the support for SCMI protocols as modules in order to let
> > vendors extend the SCMI core with their own additions it seems odd to
> > then force SCMI drivers built on top to use a static device table to
> > declare their devices since this way any new SCMI drivers addition
> > would need the core SCMI device table to be updated too.
> >
> > Remove the static core device table and let SCMI drivers to simply declare
> > which device/protocol pair they need at initialization time: the core will
> > then take care to generate such devices dynamically during platform
> > initialization or at module loading time, as long as the requested
> > underlying protocol is defined in the DT.
> >
> > Signed-off-by: Cristian Marussi <cristian.marussi@xxxxxxx>
> > ---
> > v4 --> v5
> > - using klist instead of custom lists
> > v3 --> v4
> > - add a few comments
> > ---
> >  drivers/firmware/arm_scmi/bus.c    |  30 +++
> >  drivers/firmware/arm_scmi/common.h |   5 +
> >  drivers/firmware/arm_scmi/driver.c | 309 +++++++++++++++++++++++++----
> >  3 files changed, 310 insertions(+), 34 deletions(-)
> >
> > diff --git a/drivers/firmware/arm_scmi/bus.c b/drivers/firmware/arm_scmi/bus.c
> > index 88e5057f4e85..88149a46e6d9 100644
> > --- a/drivers/firmware/arm_scmi/bus.c
> > +++ b/drivers/firmware/arm_scmi/bus.c
> > @@ -51,6 +51,31 @@ static int scmi_dev_match(struct device *dev, struct device_driver *drv)
> >  	return 0;
> >  }
> >
> > +static int scmi_match_by_id_table(struct device *dev, void *data)
> > +{
> > +	struct scmi_device *sdev = to_scmi_dev(dev);
> > +	struct scmi_device_id *id_table = data;
> > +
> > +	return sdev->protocol_id == id_table->protocol_id &&
> > +		!strcmp(sdev->name, id_table->name);
> > +}
> > +
> > +struct scmi_device *scmi_find_child_dev(struct device *parent,
> > +					int prot_id, const char *name)
> > +{
> > +	struct scmi_device_id id_table;
> > +	struct device *dev;
> > +
> > +	id_table.protocol_id = prot_id;
> > +	id_table.name = name;
> > +
> > +	dev = device_find_child(parent, &id_table, scmi_match_by_id_table);
> > +	if (!dev)
> > +		return NULL;
> > +
> > +	return to_scmi_dev(dev);
> > +}
> > +
> >  const struct scmi_protocol *scmi_get_protocol(int protocol_id)
> >  {
> >  	const struct scmi_protocol *proto;
> > @@ -114,6 +139,10 @@ int scmi_driver_register(struct scmi_driver *driver, struct module *owner,
> >  {
> >  	int retval;
> >
> > +	retval = scmi_request_protocol_device(driver->id_table);
> > +	if (retval)
> > +		return retval;
> > +
> >  	driver->driver.bus = &scmi_bus_type;
> >  	driver->driver.name = driver->name;
> >  	driver->driver.owner = owner;
> > @@ -130,6 +159,7 @@ EXPORT_SYMBOL_GPL(scmi_driver_register);
> >  void scmi_driver_unregister(struct scmi_driver *driver)
> >  {
> >  	driver_unregister(&driver->driver);
> > +	scmi_unrequest_protocol_device(driver->id_table);
> >  }
> >  EXPORT_SYMBOL_GPL(scmi_driver_unregister);
> >
> > diff --git a/drivers/firmware/arm_scmi/common.h b/drivers/firmware/arm_scmi/common.h
> > index 1e2046c61d43..9a0519db4865 100644
> > --- a/drivers/firmware/arm_scmi/common.h
> > +++ b/drivers/firmware/arm_scmi/common.h
> > @@ -307,6 +307,11 @@ struct scmi_transport_ops {
> >  	bool (*poll_done)(struct scmi_chan_info *cinfo, struct scmi_xfer *xfer);
> >  };
> >
> > +int scmi_request_protocol_device(const struct scmi_device_id *id_table);
> > +void scmi_unrequest_protocol_device(const struct scmi_device_id *id_table);
> 
> Sorry for being pedantic, I don't like these names. I would prefer
> something like scmi_protocol_device_{create,destroy/delete}_request.
> The action the function does needs to be at the end of the function name.
> Atleast that is something I follow. I haven't checked all the previous
> patches, just this function made to look at both the name style and the
> name itself.
> 
Ok.

> 
> > +struct scmi_device *scmi_find_child_dev(struct device *parent,
> > +					int prot_id, const char *name);
> > +
> 
> scmi_child_dev_find based on what I mentioned above. Please change all
> other non-static functions even if I have not mentioned. Try to cover
> all the new functions introduced in this series, existing ones we can
> take up later.
> 

Ok I'll do.

> >  /**
> >   * struct scmi_desc - Description of SoC integration
> >   *
> > diff --git a/drivers/firmware/arm_scmi/driver.c b/drivers/firmware/arm_scmi/driver.c
> > index dcdfd94b47f7..9fc979e3b16f 100644
> > --- a/drivers/firmware/arm_scmi/driver.c
> > +++ b/drivers/firmware/arm_scmi/driver.c
> > @@ -21,6 +21,7 @@
> >  #include <linux/io.h>
> >  #include <linux/kernel.h>
> >  #include <linux/ktime.h>
> > +#include <linux/list.h>
> >  #include <linux/module.h>
> >  #include <linux/of_address.h>
> >  #include <linux/of_device.h>
> > @@ -56,6 +57,14 @@ static DEFINE_MUTEX(scmi_list_mutex);
> >  /* Track the unique id for the transfers for debug & profiling purpose */
> >  static atomic_t transfer_last_id;
> >
> > +static DEFINE_IDR(scmi_requested_devices);
> > +static DEFINE_MUTEX(scmi_requested_devices_mtx);
> > +
> > +struct scmi_requested_dev {
> > +	const struct scmi_device_id *id_table;
> > +	struct list_head node;
> > +};
> > +
> >  /**
> >   * struct scmi_xfers_info - Structure to manage transfer information
> >   *
> > @@ -113,6 +122,8 @@ struct scmi_protocol_instance {
> >   * @protocols_mtx: A mutex to protect protocols instances initialization.
> >   * @protocols_imp: List of protocols implemented, currently maximum of
> >   *	MAX_PROTOCOLS_IMP elements allocated by the base protocol
> > + * @active_protocols: IDR storing device_nodes for protocols actually defined
> > + *		      in the DT and confirmed as implemented by fw.
> >   * @notify_priv: Pointer to private data structure specific to notifications.
> >   * @node: List head
> >   * @users: Number of users of this instance
> > @@ -130,6 +141,7 @@ struct scmi_info {
> >  	/* Ensure mutual exclusive access to protocols instance array */
> >  	struct mutex protocols_mtx;
> >  	u8 *protocols_imp;
> > +	struct idr active_protocols;
> >  	void *notify_priv;
> >  	struct list_head node;
> >  	int users;
> > @@ -936,6 +948,13 @@ static void scmi_devm_put_protocol(struct scmi_device *sdev, u8 protocol_id)
> >  	WARN_ON(ret);
> >  }
> >
> > +static inline
> > +struct scmi_handle *scmi_handle_get_from_info(struct scmi_info *info)
> > +{
> > +	info->users++;
> 
> Doesn't it race with anything ? I have already forgotten how this is used
> and in what context this gets called.
> 
> > +	return &info->handle;
> > +}
> > +
> >  /**
> >   * scmi_handle_get() - Get the SCMI handle for a device
> >   *
> > @@ -957,8 +976,7 @@ struct scmi_handle *scmi_handle_get(struct device *dev)
> >  	list_for_each(p, &scmi_list) {
> >  		info = list_entry(p, struct scmi_info, node);
> >  		if (dev->parent == info->dev) {
> > -			handle = &info->handle;
> > -			info->users++;
> > +			handle = scmi_handle_get_from_info(info);
> 
> Ah here it is. Any particular reasons for moving it to separate function ?
> 

Answering both of the above, users++ is protected by scmi_list_mutex as it
was already and the reason for moving the get in a separate function is to
able to call it with mutex alrady acquired, so both here from scmi_handle_get(),
which acquires itself that mutex, BUT also from scmi_request_protocol_device()
which acquires already the mutex while scanning the list of all the existing
scmi instances; it is a sort of get_handle_unlocked but since I changed
also the param from dev to info (since I have it already) I named it get_from_info
instead. I could add a get_from_info_unlocked(0 to clarify the intent.

I'll rework all of the above and post the whole series on top of for-next-scmi +
Jonathan immutable branch locally merged.

Thanks

Cristian

> --
> Regards,
> Sudeep