On Wed, Sep 14, 2016 at 3:26 PM, Suzuki K Poulose <Suzuki.Poulose@xxxxxxx <mailto:Suzuki.Poulose@xxxxxxx>> wrote:
On 13/09/16 16:41, Mathieu Poirier wrote:
On 13 September 2016 at 06:20, Venkatesh Vivekanandan
<venkatesh.vivekanandan@xxxxxxxxxxxx <mailto:venkatesh.vivekanandan@xxxxxxxxxxxx>> wrote:
tmc_etb_dump_hw is never called in sysFS mode to collect trace from
hardware, because drvdata->mode is set to CS_MODE_DISABLED at
tmc_disable_etf/etr_sink
static void tmc_etb_disable_hw(struct tmc_drvdata *drvdata)
{
.
.
if (local_read(&drvdata->mode) == CS_MODE_SYSFS)
tmc_etb_dump_hw(drvdata);
.
.
}
static void tmc_disable_etf_sink(struct coresight_device *csdev)
{
.
.
val = local_xchg(&drvdata->mode, CS_MODE_DISABLED);
/* Disable the TMC only if it needs to */
if (val != CS_MODE_DISABLED)
tmc_etb_disable_hw(drvdata);
You are correct.
.
.
}
I think we should :
1) First switch the drvdata->mode to a normal type from local_t. Using an
atomic type for mode is completely unnecessary and comes with the overhead
of barriers/synchronisation instructions, while all accesses, including read/write
are performed under the drvdata->spinlock. I have a patch already for this, which
I plan to send it soon.
and
2) Do something like :
void tmc_disable_etX_sink()
{
if (drvdata->mode != CS_MODE_DISABLED) {
tmc_etX_disable_hw(drvdata);
drvdata->mode = CS_MODE_DISABLED;
}
}
You will fix this along with above changes?