| Message ID | 20190122181144.26645-4-mathieu.poirier@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | perf: Communicate sink via event::attr:config2 | expand |
On 01/22/2019 06:11 PM, Mathieu Poirier wrote: > This patch uses the information conveyed by perf_event::attr::config2 > to select a sink to use for the session. That way a sink can easily be > selected to be used by more than one source, something that isn't currently > possible with the sysfs implementation. > > Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org> > --- > .../hwtracing/coresight/coresight-etm-perf.c | 24 ++++++------ > drivers/hwtracing/coresight/coresight-priv.h | 1 + > drivers/hwtracing/coresight/coresight.c | 39 +++++++++++++++++++ > 3 files changed, 52 insertions(+), 12 deletions(-) > > diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c > index c68a0036532c..ea031eb673b3 100644 > --- a/drivers/hwtracing/coresight/coresight-etm-perf.c > +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c > @@ -31,11 +31,14 @@ static DEFINE_PER_CPU(struct coresight_device *, csdev_src); > PMU_FORMAT_ATTR(cycacc, "config:" __stringify(ETM_OPT_CYCACC)); > PMU_FORMAT_ATTR(timestamp, "config:" __stringify(ETM_OPT_TS)); > PMU_FORMAT_ATTR(retstack, "config:" __stringify(ETM_OPT_RETSTK)); > +/* Sink ID - same for all ETMs */ > +PMU_FORMAT_ATTR(sinkid, "config2:0-31"); > > static struct attribute *etm_config_formats_attr[] = { > &format_attr_cycacc.attr, > &format_attr_timestamp.attr, > &format_attr_retstack.attr, > + &format_attr_sinkid.attr, > NULL, > }; > > @@ -191,6 +194,7 @@ static void etm_free_aux(void *data) > static void *etm_setup_aux(struct perf_event *event, void **pages, > int nr_pages, bool overwrite) > { > + u32 id; > int cpu = event->cpu; > cpumask_t *mask; > struct coresight_device *sink; > @@ -201,18 +205,14 @@ static void *etm_setup_aux(struct perf_event *event, void **pages, > return NULL; > INIT_WORK(&event_data->work, free_event_data); > > - /* > - * In theory nothing prevent tracers in a trace session from being > - * associated with different sinks, nor having a sink per tracer. But > - * until we have HW with this kind of topology we need to assume tracers > - * in a trace session are using the same sink. Therefore go through > - * the coresight bus and pick the first enabled sink. > - * > - * When operated from sysFS users are responsible to enable the sink > - * while from perf, the perf tools will do it based on the choice made > - * on the cmd line. As such the "enable_sink" flag in sysFS is reset. > - */ > - sink = coresight_get_enabled_sink(true); > + /* First get the selected sink from user space. */ > + if (event->attr.config2) { > + id = (u32)event->attr.config2; > + sink = coresight_get_sink_by_id(id); > + } else { > + sink = coresight_get_enabled_sink(true); > + } > + > if (!sink || !sink_ops(sink)->alloc_buffer) > goto err; > > diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h > index 579f34943bf1..b936c6d7e13f 100644 > --- a/drivers/hwtracing/coresight/coresight-priv.h > +++ b/drivers/hwtracing/coresight/coresight-priv.h > @@ -147,6 +147,7 @@ void coresight_disable_path(struct list_head *path); > int coresight_enable_path(struct list_head *path, u32 mode, void *sink_data); > struct coresight_device *coresight_get_sink(struct list_head *path); > struct coresight_device *coresight_get_enabled_sink(bool reset); > +struct coresight_device *coresight_get_sink_by_id(u32 id); > struct list_head *coresight_build_path(struct coresight_device *csdev, > struct coresight_device *sink); > void coresight_release_path(struct list_head *path); > diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c > index d7fa90be6f42..c5f2df186a19 100644 > --- a/drivers/hwtracing/coresight/coresight.c > +++ b/drivers/hwtracing/coresight/coresight.c > @@ -11,6 +11,7 @@ > #include <linux/err.h> > #include <linux/export.h> > #include <linux/slab.h> > +#include <linux/stringhash.h> > #include <linux/mutex.h> > #include <linux/clk.h> > #include <linux/coresight.h> > @@ -541,6 +542,44 @@ struct coresight_device *coresight_get_enabled_sink(bool deactivate) > return dev ? to_coresight_device(dev) : NULL; > } > > +static int coresight_sink_by_id(struct device *dev, void *data) > +{ > + struct coresight_device *csdev = to_coresight_device(dev); > + u32 hash; > + > + if (csdev->type == CORESIGHT_DEV_TYPE_SINK || > + csdev->type == CORESIGHT_DEV_TYPE_LINKSINK) { > + /* > + * See function etm_perf_sink_name_show() to know where this > + * comes from. > + */ > + hash = hashlen_hash(hashlen_string(NULL, dev_name(dev))); > + > + if (hash == (*(u32 *)data)) > + return 1; > + } > + > + return 0; > +} > + > +/** > + * coresight_get_sink_by_id - returns the sink that matches the id > + * @id: Id of the sink to match > + * > + * The name of a sink is unique, whether it is found on the AMBA bus or > + * otherwise. As such the hash of that name can easily be used to identify > + * a sink. > + */ > +struct coresight_device *coresight_get_sink_by_id(u32 id) > +{ > + struct device *dev = NULL; > + > + dev = bus_find_device(&coresight_bustype, NULL, &id, > + coresight_sink_by_id); > + > + return dev ? to_coresight_device(dev) : NULL; > +} > + The patch looks good to me. As mentioned in the previous patch, we may be able to optimize the look up a bit using dev_ext_attribute. Either way, Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
diff --git a/drivers/hwtracing/coresight/coresight-etm-perf.c b/drivers/hwtracing/coresight/coresight-etm-perf.c index c68a0036532c..ea031eb673b3 100644 --- a/drivers/hwtracing/coresight/coresight-etm-perf.c +++ b/drivers/hwtracing/coresight/coresight-etm-perf.c @@ -31,11 +31,14 @@ static DEFINE_PER_CPU(struct coresight_device *, csdev_src); PMU_FORMAT_ATTR(cycacc, "config:" __stringify(ETM_OPT_CYCACC)); PMU_FORMAT_ATTR(timestamp, "config:" __stringify(ETM_OPT_TS)); PMU_FORMAT_ATTR(retstack, "config:" __stringify(ETM_OPT_RETSTK)); +/* Sink ID - same for all ETMs */ +PMU_FORMAT_ATTR(sinkid, "config2:0-31"); static struct attribute *etm_config_formats_attr[] = { &format_attr_cycacc.attr, &format_attr_timestamp.attr, &format_attr_retstack.attr, + &format_attr_sinkid.attr, NULL, }; @@ -191,6 +194,7 @@ static void etm_free_aux(void *data) static void *etm_setup_aux(struct perf_event *event, void **pages, int nr_pages, bool overwrite) { + u32 id; int cpu = event->cpu; cpumask_t *mask; struct coresight_device *sink; @@ -201,18 +205,14 @@ static void *etm_setup_aux(struct perf_event *event, void **pages, return NULL; INIT_WORK(&event_data->work, free_event_data); - /* - * In theory nothing prevent tracers in a trace session from being - * associated with different sinks, nor having a sink per tracer. But - * until we have HW with this kind of topology we need to assume tracers - * in a trace session are using the same sink. Therefore go through - * the coresight bus and pick the first enabled sink. - * - * When operated from sysFS users are responsible to enable the sink - * while from perf, the perf tools will do it based on the choice made - * on the cmd line. As such the "enable_sink" flag in sysFS is reset. - */ - sink = coresight_get_enabled_sink(true); + /* First get the selected sink from user space. */ + if (event->attr.config2) { + id = (u32)event->attr.config2; + sink = coresight_get_sink_by_id(id); + } else { + sink = coresight_get_enabled_sink(true); + } + if (!sink || !sink_ops(sink)->alloc_buffer) goto err; diff --git a/drivers/hwtracing/coresight/coresight-priv.h b/drivers/hwtracing/coresight/coresight-priv.h index 579f34943bf1..b936c6d7e13f 100644 --- a/drivers/hwtracing/coresight/coresight-priv.h +++ b/drivers/hwtracing/coresight/coresight-priv.h @@ -147,6 +147,7 @@ void coresight_disable_path(struct list_head *path); int coresight_enable_path(struct list_head *path, u32 mode, void *sink_data); struct coresight_device *coresight_get_sink(struct list_head *path); struct coresight_device *coresight_get_enabled_sink(bool reset); +struct coresight_device *coresight_get_sink_by_id(u32 id); struct list_head *coresight_build_path(struct coresight_device *csdev, struct coresight_device *sink); void coresight_release_path(struct list_head *path); diff --git a/drivers/hwtracing/coresight/coresight.c b/drivers/hwtracing/coresight/coresight.c index d7fa90be6f42..c5f2df186a19 100644 --- a/drivers/hwtracing/coresight/coresight.c +++ b/drivers/hwtracing/coresight/coresight.c @@ -11,6 +11,7 @@ #include <linux/err.h> #include <linux/export.h> #include <linux/slab.h> +#include <linux/stringhash.h> #include <linux/mutex.h> #include <linux/clk.h> #include <linux/coresight.h> @@ -541,6 +542,44 @@ struct coresight_device *coresight_get_enabled_sink(bool deactivate) return dev ? to_coresight_device(dev) : NULL; } +static int coresight_sink_by_id(struct device *dev, void *data) +{ + struct coresight_device *csdev = to_coresight_device(dev); + u32 hash; + + if (csdev->type == CORESIGHT_DEV_TYPE_SINK || + csdev->type == CORESIGHT_DEV_TYPE_LINKSINK) { + /* + * See function etm_perf_sink_name_show() to know where this + * comes from. + */ + hash = hashlen_hash(hashlen_string(NULL, dev_name(dev))); + + if (hash == (*(u32 *)data)) + return 1; + } + + return 0; +} + +/** + * coresight_get_sink_by_id - returns the sink that matches the id + * @id: Id of the sink to match + * + * The name of a sink is unique, whether it is found on the AMBA bus or + * otherwise. As such the hash of that name can easily be used to identify + * a sink. + */ +struct coresight_device *coresight_get_sink_by_id(u32 id) +{ + struct device *dev = NULL; + + dev = bus_find_device(&coresight_bustype, NULL, &id, + coresight_sink_by_id); + + return dev ? to_coresight_device(dev) : NULL; +} + /* * coresight_grab_device - Power up this device and any of the helper * devices connected to it for trace operation. Since the helper devices
This patch uses the information conveyed by perf_event::attr::config2 to select a sink to use for the session. That way a sink can easily be selected to be used by more than one source, something that isn't currently possible with the sysfs implementation. Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org> --- .../hwtracing/coresight/coresight-etm-perf.c | 24 ++++++------ drivers/hwtracing/coresight/coresight-priv.h | 1 + drivers/hwtracing/coresight/coresight.c | 39 +++++++++++++++++++ 3 files changed, 52 insertions(+), 12 deletions(-)