Argus::ICaptureMetadata::getSensorTimestamp clock domain in L4T 32.4.4

[Denisss]

Hello,
When looking at Argus API, the function to get the camera capture timestamp is:

Argus::ICaptureMetadata::getSensorTimestamp

The help says:
Returns the start timestamp for the sensor (in nanoseconds).
This is the time that the first data from this capture arrives from the sensor.

But we don’t know the clock domain.
Is this CLOCK_MONOTONIC, CLOCK_MONOTONIC_RAW or another?
We use L4T 32.4.4.

thank you

[ShaneCCC]

Have reference to below topic.

Argus Timestamp Domain

hello j2innet, sorry for late reply, I did not receive system notification. instead of asking a follow-up question in the discussion thread which already mark as solved. also, this one is an old thread consider to closed since Nov/2019. I would suggest you initial another new discussion thread next time, for the better supports. below capture the start-of-frame timestamp of the sensor signaling, Argus::ICaptureMetadata::getSensorTimestamp while EGL stream capture the timestamp when it rend…

[Denisss]

Hello,
I have read this topic and you say that Argus::ICaptureMetadata::getSensorTimestamp give the kernel timestamp.
What is the clock domain for kernel timestamp: is it Wall Clock or MONOTONIC_CLOCK or another? the topic given doesn’t say it.
I need to compare this timestamp to another timestamp based on MONOTONIC_CLOCK. How can i do it?
thank you

[ShaneCCC]

Looks like it’s MONOTONIC_RAW

Argus Timestamp Domain

When getting a timestamp from libargus with Argus::ICaptureMetadata::getSensorTimestamp or EGLStream::IFrame::getTime, what clock domain are the timestamps in? can they be compared to linux’s CLOCK_MONOTONIC system timestamps, for instance? if they’re in a different clock domain, how can I rectify these timestamps against wall time?

[Denisss]

Hello,
excuse me but “looks like it’s MONOTONIC_RAW” seems a little imprecise.
I need to be certain that it is “MONOTONIC_RAW” or not. And if not then which clock it is.
thanks

[JerryChang]

hello Denisss,

since you’re now working with l4r-r32.4.4. please note that there’ll be timestamp modification for next public release.
please refer to the statement in Topic 78821, recap as below…

we had some changes for Xavier to plumbed up the hardware timestamps into camera software stack.
there’ll be start-of-frame and end-of-frame sensor timestamps include in metadata from TSC hardware, by using new interfaces for TSC HW timestamp; you may expect those changes will be include to next public release, i.e. JetPack-4.5 / l4t-r32.5.

iCaptureMetadata->getSensorTimestamp() return the kernel timestamp (i.e. MONOTONIC_RAW) of sensor start-of-frame.
as you can see, there’s formula to calculate the HW capture timestamp (TSC) from system time.

clock_gettime(MONOTONIC_RAW) = cycle_ns(TSC) - offset_ns

please refer to below example for offset_ns to compensate the capture timestamp.
for example,

root@tegra-ubuntu:~# cat /sys/devices/system/clocksource/clocksource0/offset_ns
9045955168

diff --git a/drivers/media/platform/tegra/camera/vi/vi5_fops.c b/drivers/media/platform/tegra/camera/vi/vi5_fops.c
index f8c0df5..6a7c370 100644
--- a/drivers/media/platform/tegra/camera/vi/vi5_fops.c
+++ b/drivers/media/platform/tegra/camera/vi/vi5_fops.c
@@ -427,6 +427,9 @@ static void vi5_capture_dequeue(struct tegra_channel *chan,
    /* Read SOF from capture descriptor */
    ts = ns_to_timespec((s64)descr->status.sof_timestamp);
... 
+   pr_err("Jerry %s timestamp= %ld.%ld Line(%d)\n", __func__, ts.tv_sec, ts.tv_nsec, __LINE__);

kernel log
[  128.924693] Jerry vi5_capture_dequeue timestamp= 137.483546848 

which means… 128.924693 + 9045955168(ns) ~= 137.483546848

VI drivers to capture frame and store the timestamp with TSC.
you may also adding some kernel APIs for checking timestamp difference within the VI drivers.
for example,

*ts = ns_to_timespec((s64)status.sof_ts);
+ printk("VI: RTCPU SOF timestamp %lu.%lu s\n", ts->tv_sec, ts->tv_nsec);
+ sys_clock_gettime(CLOCK_MONOTONIC, &monotonic);
+ printk("VI: Monotonic Raw timestamp %lu.%lu s\n", monotonic.tv_sec, monotonic.tv_nsec);

[JerryChang]

hi all,

for the r32.5 release
you are able to fetch start-of-frame and end-of-frame timing via TSC HW timestamp.

it returns the VI HW (SoF/EoF) timestamp based-on tegra wide timestamp system counter (TSC), it’s the timestamp for the sensor (unit in nanoseconds).
here’s code snippet by modify Argus sample code (i.e. userAutoExposure) to use the interface to get TSC HW timestamp.
for example,

+++ b/public/samples/userAutoExposure/main.cpp
@@ -32,6 +32,7 @@
 #include <stdlib.h>
 #include <Argus/Argus.h>
 #include <Argus/Ext/BayerAverageMap.h>
+#include <Argus/Ext/SensorTimestampTsc.h>
 #include <EGLStream/EGLStream.h>
 #include "PreviewConsumer.h"
 #include "CommonOptions.h"
@@ -298,6 +299,14 @@ static bool execute(const UserAutoExposureSampleOptions& options)
                 uint64_t frameExposureTime = iMetadata->getSensorExposureTime();
                 float frameGain = iMetadata->getSensorAnalogGain();

+                const Ext::ISensorTimestampTsc *iSensorTimestampTsc = interface_cast<const Ext::ISensorTimestampTsc>(metaData);
+                if (iSensorTimestampTsc)
+                {
+                        printf("sof= %lu, eof= %lu\n",
+                               iSensorTimestampTsc->getSensorSofTimestampTsc(),
+                               iSensorTimestampTsc->getSensorEofTimestampTsc());
+                }