Re: [PATCH v5 08/14] coresight: cti: Enable CTI associated with devices.

[Suzuki Kuruppassery Poulose <suzuki.poulose@arm.com>]

On 19/11/2019 23:19, Mike Leach wrote:
> The CoreSight subsystem enables a path of devices from source to sink.
> Any CTI devices associated with the path devices must be enabled at the
> same time.
> 
> This patch adds an associated coresight_device element to the main
> coresight device structure, and uses this to create associations between
> the CTI and other devices based on the device tree data. The associated
> device element is used to enable CTI in conjunction with the path elements.
> 
> CTI devices are reference counted so where a single CTI is associated with
> multiple elements on the path, it will be enabled on the first associated
> device enable, and disabled with the last associated device disable.
> 
> Signed-off-by: Mike Leach <mike.leach@linaro.org>
> ---
>   drivers/hwtracing/coresight/coresight-cti.c   | 87 +++++++++++++++++++
>   .../hwtracing/coresight/coresight-platform.c  | 23 +++++
>   drivers/hwtracing/coresight/coresight-priv.h  |  6 ++
>   drivers/hwtracing/coresight/coresight.c       | 58 +++++++++++--
>   include/linux/coresight.h                     |  5 ++
>   5 files changed, 173 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/hwtracing/coresight/coresight-cti.c b/drivers/hwtracing/coresight/coresight-cti.c
> index 369488dd7b8e..cf116463149a 100644
> --- a/drivers/hwtracing/coresight/coresight-cti.c
> +++ b/drivers/hwtracing/coresight/coresight-cti.c
> @@ -440,6 +440,90 @@ int cti_channel_setop(struct device *dev, enum cti_chan_set_op op,
>   	return err;
>   }
>   
> +/*
> + * Look for a matching connection device name in the list of
> + * connections. If found then swap in the csdev name and return
> + * found.
> + */
> +static bool
> +cti_match_con_name(struct cti_device *ctidev, const char *node_name,
> +		   const char *csdev_name)

Here we actually fixup the name of the connection, rather than simply 
matching it. So it may be apt to rename this to cti_match_fixup_name()

> +{
> +	struct cti_trig_con *trig_con;
> +
> +	list_for_each_entry(trig_con, &ctidev->trig_cons, node) {
> +		if (trig_con->con_dev_name) {
> +			if (!strcmp(node_name, trig_con->con_dev_name)) {

Can there be duplicate node_name's ? Does it make sense to store the 
fwhandle along with the "temporary node_name" to match it later while
fixing up ?

> +				/* match: so swap in csdev name */
> +				kfree(trig_con->con_dev_name);
> +				trig_con->con_dev_name =
> +					kstrdup(csdev_name, GFP_KERNEL);
> +				return true;
> +			}
> +		}
> +	}
> +	return false;
> +}


> +/*
> + * Search the cti list to add an associated CTI into the supplied CS device
> + * This will set the association if CTI declared before the CS device
> + */
> +void cti_add_assoc_to_csdev(struct coresight_device *csdev)
> +{

..

 > +	struct cti_drvdata *ect_item;
 > +	struct cti_device *ctidev;
 > +	const char *node_name = NULL, *csdev_name;
 > +
 > +	/* protect the list */
 > +	mutex_lock(&ect_mutex);
 > +
 > +	/* exit if current is an ECT device.*/
 > +	if ((csdev->type == CORESIGHT_DEV_TYPE_ECT) || list_empty(&ect_net))
 > +		goto cti_add_done;
 > +
 > +	/* if we didn't find the csdev previously we used the fwnode name */
 > +	node_name = coresight_get_fwnode_name(csdev->dev.parent);

We used "cti_plat_get_node_name()" when we added the name in the
absence of csdev in patch 7, could we not reuse the function here ?

 > +
 > +	if (!node_name)
 > +		goto cti_add_done;
 > +
 > +	/* this is the name we want to use for the association */
 > +	csdev_name = dev_name(&csdev->dev);


> +
> +	/* for each CTI in list... */
> +	list_for_each_entry(ect_item, &ect_net, node) {
> +		ctidev = &ect_item->ctidev;
> +		if (cti_match_con_name(ctidev, node_name, csdev_name)) {
> +			/*
> +			 * if we found a matching name then update the
> +			 * association pointers.
> +			 */
> +			csdev->ect_dev = ect_item->csdev;
> +			goto cti_add_done;

			break; instead ?

> +		}
> +	}
> +cti_add_done:
> +	mutex_unlock(&ect_mutex);
> +}
> +EXPORT_SYMBOL_GPL(cti_add_assoc_to_csdev);
> +
> +/*
> + * Update the cross references where the associated device was found
> + * while we were building the connection info. This will occur if the
> + * assoc device was registered before the CTI.
> + */
> +static void cti_update_conn_xrefs(struct cti_drvdata *drvdata)
> +{
> +	struct cti_trig_con *tc;
> +	struct cti_device *ctidev = &drvdata->ctidev;
> +
> +	list_for_each_entry(tc, &ctidev->trig_cons, node) {
> +		if (tc->con_dev)
> +			tc->con_dev->ect_dev = drvdata->csdev;
> +	}

Does this need to take the coresight_mutex to avoid racing against
a coresight_enable_path() ? Though this may be fine as long as the
CTI driver detects that that device was not enabled.

Also, it looks like we have a potential issue with perf vs sysfs mode.
The perf mode doesn't seem to take the coresight_mutex, for
build_path/enable_path operations. This is outside the scope of this
series though.

> +}
> +
>   /** cti ect operations **/
>   int cti_enable(struct coresight_device *csdev)
>   {
> @@ -574,6 +658,9 @@ static int cti_probe(struct amba_device *adev, const struct amba_id *id)
>   	drvdata->csdev_release = drvdata->csdev->dev.release;
>   	drvdata->csdev->dev.release = cti_device_release;
>   
> +	/* set any cross references */
> +	cti_update_conn_xrefs(drvdata);
> +
    	/* all done - dec pm refcount */
>   	pm_runtime_put(&adev->dev);
>   	dev_info(&drvdata->csdev->dev, "CTI initialized\n");
> diff --git a/drivers/hwtracing/coresight/coresight-platform.c b/drivers/hwtracing/coresight/coresight-platform.c
> index 3c5bee429105..6721cb1af5fe 100644
> --- a/drivers/hwtracing/coresight/coresight-platform.c
> +++ b/drivers/hwtracing/coresight/coresight-platform.c
> @@ -293,6 +293,12 @@ static int of_get_coresight_platform_data(struct device *dev,
>   
>   	return 0;
>   }
> +
> +static inline const char *of_coresight_get_node_name(struct device *dev)
> +{
> +	return dev->of_node->full_name;
> +}
> +
>   #else
>   static inline int
>   of_get_coresight_platform_data(struct device *dev,
> @@ -305,6 +311,11 @@ static inline int of_coresight_get_cpu(struct device *dev)
>   {
>   	return -ENODEV;
>   }
> +
> +static inline const char *of_coresight_get_node_name(struct device *dev)
> +{
> +	return NULL;
> +}
>   #endif
>   
>   #ifdef CONFIG_ACPI
> @@ -766,6 +777,18 @@ static inline int acpi_coresight_get_cpu(struct device *dev)
>   }
>   #endif
>   
> +const char *coresight_get_fwnode_name(struct device *dev)

As mentioned above, please could we reuse the name helper we used
during the insertion rather than introducing a new wrapper which
effectively does the same thing ?

> +{
> +	const char *node_name = NULL;
> +	struct fwnode_handle *fwnode = dev_fwnode(dev);
> +
> +	if (is_of_node(fwnode))
> +		node_name = of_coresight_get_node_name(dev);
> +
> +	return node_name;
> +}
> +EXPORT_SYMBOL_GPL(coresight_get_fwnode_name);

Why does this get exported ? If a following patch needs it, you may
always do that when you need it.


Cheers
Suzuki