This is no issue. This print only tells you that you find the relevant PHC at /dev/ptp0.
You can try our fork of the AQtion driver: GitHub - ctu-vras/AQtion: Aquantia AQC multigigabit NIC linux driver (atlantic) - development preview , devel branch. I’m working on it right now. In a few minutes, this branch should contain all the relevant fixes needed for both HW and SW timestamping working and it should be buildable on kernels 4.9 - 6.9 . I mostly tested with AQC107, which is Gen 1 chipset, while AQC113 is Gen 2 chipset. However, there’s a good chance my changes could help even with AQC113.
If you want, you can try without jetson_clocks and pass --tx_timestamp_timeout 30 to ptp4l. You’ll see if the problem is that there’s not enough waiting time for the stamps (which jetson_clocks could help with), or if it is the case that timestamps get lost totally (I guess the 30 ms timeout should cover most proper cases).