On Mon, Nov 10, 2025 at 11:48:45AM +0000, James Clark wrote:
[...]
static void coresight_disable_path_from(struct coresight_path *path, - struct coresight_node *nd) + struct coresight_node *nd, + bool in_idle) { u32 type; struct coresight_device *csdev, *parent, *child; @@ -498,6 +499,10 @@ static void coresight_disable_path_from(struct coresight_path *path, CORESIGHT_DEV_TYPE_SINK : CORESIGHT_DEV_TYPE_LINK; + /* To reduce latency, CPU idle does not touch the sink */ + if (in_idle && type == CORESIGHT_DEV_TYPE_SINK) + continue;
switch (type) { case CORESIGHT_DEV_TYPE_SINK: coresight_disable_sink(csdev); @@ -527,7 +532,7 @@ static void coresight_disable_path_from(struct coresight_path *path, void coresight_disable_path(struct coresight_path *path) { - coresight_disable_path_from(path, NULL); + coresight_disable_path_from(path, NULL, false);
Something like a 'reason' enum would be more readable than a boolean. We might have more than two options in the future anyway.
I will add a "path->in_idle" flag instead.
} EXPORT_SYMBOL_GPL(coresight_disable_path); @@ -550,8 +555,9 @@ static int coresight_enable_helpers(struct coresight_device *csdev, return 0; } -int coresight_enable_path(struct coresight_path *path, enum cs_mode mode, - void *sink_data) +static int coresight_enable_path_internal(struct coresight_path *path, + enum cs_mode mode, void *sink_data, + bool in_idle) { int ret = 0; u32 type; @@ -564,10 +570,6 @@ int coresight_enable_path(struct coresight_path *path, enum cs_mode mode, csdev = nd->csdev; type = csdev->type; - /* Enable all helpers adjacent to the path first */ - ret = coresight_enable_helpers(csdev, mode, path); - if (ret) - goto err_disable_path; /* * ETF devices are tricky... They can be a link or a sink, * depending on how they are configured. If an ETF has been @@ -579,6 +581,15 @@ int coresight_enable_path(struct coresight_path *path, enum cs_mode mode, CORESIGHT_DEV_TYPE_SINK : CORESIGHT_DEV_TYPE_LINK; + /* To reduce latency, CPU idle does not touch the sink */ + if (in_idle && type == CORESIGHT_DEV_TYPE_SINK) + continue;
I can imagine that not being wanted for all sink types in the future, so we could let the sink decide this. But it's probably fine to leave in the core for now and it would be easy to change later.
I got to patch 12 and it adds a special condition in the core to do something different for TRBE. We already have pm_is_needed() so shouldn't the sinks be deciding this for themselves?
pm_is_needed() is related to pm_save_disable()/pm_restore_enable() pairs. OTOH, coresight_enable_path() invokes enable()/disable() callbacks. I don't want to mix up two different things.
But I will check if can use pm_is_needed() for sink device in cpu_pm_save()/cpu_pm_restore().
Thanks, Leo