Re: [PATCH v12 19/28] coresight: Save active path for system tracers

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

On 11/05/2026 12:11, Leo Yan wrote:
> This commit only set the path pointer for system tracers (e.g. STM) in
> coresight_{enable|disable}_source().
> 
> Later changes will set the path pointer locally for per-CPU sources.
> This is because the mode and path pointer must be set together, so that
> they are observed atomically by the CPU PM notifier.
> 
> Reviewed-by: Yeoreum Yun <yeoreum.yun@arm.com>
> Reviewed-by: James Clark <james.clark@linaro.org>
> Tested-by: James Clark <james.clark@linaro.org>
> Tested-by: Jie Gan <jie.gan@oss.qualcomm.com>
> Signed-off-by: Leo Yan <leo.yan@arm.com>
> ---
>   drivers/hwtracing/coresight/coresight-core.c | 23 ++++++++++++++++++++++-
>   include/linux/coresight.h                    |  2 ++
>   2 files changed, 24 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/hwtracing/coresight/coresight-core.c b/drivers/hwtracing/coresight/coresight-core.c
> index ff882769cab2bb20bc2cb768e86f111f8a09cc8d..f07f6f28b9162911cdc673a454702f3ac4dc29ad 100644
> --- a/drivers/hwtracing/coresight/coresight-core.c
> +++ b/drivers/hwtracing/coresight/coresight-core.c
> @@ -487,10 +487,28 @@ int coresight_enable_source(struct coresight_device *csdev,
>   			    struct perf_event *event, enum cs_mode mode,
>   			    struct coresight_path *path)
>   {
> +	int ret;
> +
>   	if (!coresight_is_device_source(csdev))
>   		return -EINVAL;
>   
> -	return source_ops(csdev)->enable(csdev, event, mode, path);
> +	ret = source_ops(csdev)->enable(csdev, event, mode, path);
> +	if (ret)
> +		return ret;
> +
> +	/*
> +	 * Update the path pointer until after the source is enabled to avoid
> +	 * races where multiple paths attempt to enable the same source.
> +	 *
> +	 * Do not set the path pointer here for per-CPU sources; set it locally
> +	 * on the CPU instead. Otherwise, there is a window where the path is
> +	 * enabled but the pointer is not yet set, causing CPU PM notifiers to
> +	 * miss PM operations due to reading a NULL pointer.
> +	 */
> +	if (!coresight_is_percpu_source(csdev))
> +		csdev->path = path;

Do we need to specail case this for non-percpu sources ? We could always 
let the driver save/clear the path leaving the core out of it. We can 
fix it separately from this series in a follow up patch.

Suzuki



> +
> +	return 0;
>   }
>   
>   void coresight_disable_source(struct coresight_device *csdev, void *data)
> @@ -498,6 +516,9 @@ void coresight_disable_source(struct coresight_device *csdev, void *data)
>   	if (!coresight_is_device_source(csdev))
>   		return;
>   
> +	if (!coresight_is_percpu_source(csdev))
> +		csdev->path = NULL;
> +
>   	source_ops(csdev)->disable(csdev, data);
>   }
>   EXPORT_SYMBOL_GPL(coresight_disable_source);
> diff --git a/include/linux/coresight.h b/include/linux/coresight.h
> index 5f9d7ea9f5941ab01eb6a084ca558a9417c7727f..58d474b269806d32cad6ed87da96550b06f1f30f 100644
> --- a/include/linux/coresight.h
> +++ b/include/linux/coresight.h
> @@ -257,6 +257,7 @@ struct coresight_trace_id_map {
>    *		by @coresight_ops.
>    * @access:	Device i/o access abstraction for this device.
>    * @dev:	The device entity associated to this component.
> + * @path:	Activated path pointer (only used for source device).
>    * @mode:	The device mode, i.e sysFS, Perf or disabled. This is actually
>    *		an 'enum cs_mode' but stored in an atomic type. Access is always
>    *		through atomic APIs, ensuring SMP-safe synchronisation between
> @@ -291,6 +292,7 @@ struct coresight_device {
>   	const struct coresight_ops *ops;
>   	struct csdev_access access;
>   	struct device dev;
> +	struct coresight_path *path;
>   	atomic_t mode;
>   	int refcnt;
>   	int cpu;
>