When tracing etm data of multiple threads on multiple cpus through
perf interface, each cpu has a unique etr_perf_buffer while sharing
the same etr device. There is no guarantee that the last cpu starts
etm tracing also stops last. So the perf_data check is no longer valid.
Signed-off-by: Yabin Cui <[email protected]>
---
drivers/hwtracing/coresight/coresight-tmc-etr.c | 9 ---------
drivers/hwtracing/coresight/coresight-tmc.h | 2 --
2 files changed, 11 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-tmc-etr.c b/drivers/hwtracing/coresight/coresight-tmc-etr.c
index 17006705287a..0418440e0141 100644
--- a/drivers/hwtracing/coresight/coresight-tmc-etr.c
+++ b/drivers/hwtracing/coresight/coresight-tmc-etr.c
@@ -1484,20 +1484,12 @@ tmc_update_etr_buffer(struct coresight_device *csdev,
goto out;
}
- if (WARN_ON(drvdata->perf_data != etr_perf)) {
- lost = true;
- spin_unlock_irqrestore(&drvdata->spinlock, flags);
- goto out;
- }
-
CS_UNLOCK(drvdata->base);
tmc_flush_and_stop(drvdata);
tmc_sync_etr_buf(drvdata);
CS_LOCK(drvdata->base);
- /* Reset perf specific data */
- drvdata->perf_data = NULL;
spin_unlock_irqrestore(&drvdata->spinlock, flags);
size = etr_buf->len;
@@ -1556,7 +1548,6 @@ static int tmc_enable_etr_sink_perf(struct coresight_device *csdev, void *data)
}
etr_perf->head = PERF_IDX2OFF(handle->head, etr_perf);
- drvdata->perf_data = etr_perf;
/*
* No HW configuration is needed if the sink is already in
diff --git a/drivers/hwtracing/coresight/coresight-tmc.h b/drivers/hwtracing/coresight/coresight-tmc.h
index 1ed50411cc3c..3881a9ee565a 100644
--- a/drivers/hwtracing/coresight/coresight-tmc.h
+++ b/drivers/hwtracing/coresight/coresight-tmc.h
@@ -178,7 +178,6 @@ struct etr_buf {
* device configuration register (DEVID)
* @idr: Holds etr_bufs allocated for this ETR.
* @idr_mutex: Access serialisation for idr.
- * @perf_data: PERF buffer for ETR.
* @sysfs_data: SYSFS buffer for ETR.
*/
struct tmc_drvdata {
@@ -202,7 +201,6 @@ struct tmc_drvdata {
struct idr idr;
struct mutex idr_mutex;
struct etr_buf *sysfs_buf;
- void *perf_data;
};
struct etr_buf_operations {
--
2.22.0.770.g0f2c4a37fd-goog
Hi Yabin,
Thank you for the analysis and the patch.
On 08/08/2019 20:31, Yabin Cui wrote:
> When tracing etm data of multiple threads on multiple cpus through
> perf interface, each cpu has a unique etr_perf_buffer while sharing
> the same etr device. There is no guarantee that the last cpu starts
> etm tracing also stops last. So the perf_data check is no longer valid.
>
> Signed-off-by: Yabin Cui <[email protected]>
> ---
> drivers/hwtracing/coresight/coresight-tmc-etr.c | 9 ---------
> drivers/hwtracing/coresight/coresight-tmc.h | 2 --
> 2 files changed, 11 deletions(-)
>
> diff --git a/drivers/hwtracing/coresight/coresight-tmc-etr.c b/drivers/hwtracing/coresight/coresight-tmc-etr.c
> index 17006705287a..0418440e0141 100644
> --- a/drivers/hwtracing/coresight/coresight-tmc-etr.c
> +++ b/drivers/hwtracing/coresight/coresight-tmc-etr.c
> @@ -1484,20 +1484,12 @@ tmc_update_etr_buffer(struct coresight_device *csdev,
> goto out;
> }
>
> - if (WARN_ON(drvdata->perf_data != etr_perf)) {
> - lost = true;
> - spin_unlock_irqrestore(&drvdata->spinlock, flags);
> - goto out;
> - }
> -
I think some sort of sanity check is a good idea to make sure we don't loose the
context. Even when different CPUs have different etr_perf buffer, the underlying
etr_buf should be the same. So, we should be able to simply convert the check
to, something like :
struct etr_perf_buffer *perf_buf = drvdata->perf_data;
...
if (WARN_ON(perf_buf->etr_buf != etr_perf->buf)) {
....
}
Does that solve the problem for you ?
Suzuki
I totally agree that checking etr_buf is a better way.
It solves my problem.
I will upload a new patch soon.