Jetson AGX Orin (R36.4.3) TI DP83867 PHY (RGMII) Not Working

Robotics & Edge Computing Jetson Systems Jetson AGX Orin
,
45

I had a few questions.
Which of the above addresses correspond to the interupt. We made the following changes as seen in our pinmux.dtsi
soc_gpio17_pg4 {
nvidia,pins = “soc_gpio17_pg4”;
nvidia,function = “rsvd0”;
nvidia,pull = <TEGRA_PIN_PULL_NONE>;
nvidia,tristate = <TEGRA_PIN_ENABLE>;
nvidia,enable-input = <TEGRA_PIN_ENABLE>;
nvidia,lpdr = <TEGRA_PIN_DISABLE>;
};

		soc_gpio18_pg5 {
			nvidia,pins = "soc_gpio18_pg5";
			nvidia,function = "rsvd0";
			nvidia,pull = <TEGRA_PIN_PULL_NONE>;
			nvidia,tristate = <TEGRA_PIN_DISABLE>;
			nvidia,enable-input = <TEGRA_PIN_DISABLE>;
			nvidia,lpdr = <TEGRA_PIN_DISABLE>;
		};

We have external pull up resistor of value 2.2k for reset and interrupt.
When we do sudo gpioinfo
line 39: “PG.04” unused input active-high
line 40: “PG.05” “phy_reset” output active-high [used]
We see the reset line to be low on our logic analyser.
We also see the error -19: Failed to attach to PHY
We have extensively tested our dp83867 TI PHY chip with MSP430 but it seems as though PHY link bringup fails with Orin. Please help. Thank you so much!

46

Not quite sure about your question for the address. I already told the address in just the post right before your reply.

47

Is this one after you insmod the ethernet driver?

image
image1035×190 17.3 KB

48

Hi!
Yes, it is after insmod

49

orin@orin:~$ sudo busybox devmem 0x02434070
0x00000050
orin@orin:~$ sudo busybox devmem 0x02434078
0x00000400
orin@orin:~$ sudo busybox devmem 0x02445018
0x00002400
Hi!
We see the following values associated upon making the changes in pinmux. Is this correct? The rest hold the same values.
Thanks!

50

Please read this.

We need the interrupt pin to have pull up.

51

In our board, we have external pullup of 2.2k for both reset and interrupt due to which we have selected as “Pull None” in our pinmux.dtsi.

52

What is the state of RST pin when you rmmod the driver? Is it going with continuously toggled and then goes high?

53

when I do rmmod the driver the reset toggles from low to High continuously until insmod the driver.

54

Did you ever observe your MDC got any toggled right after UEFI? without doing rmmod.

55

Yes, but it is very inconsistent. Sometimes upon reboot there is no behaviour across both lines, mostly we see MDIO behaviour but only once upon reboot we saw MDC & MDIO activity.

56

Hi,

MDC+MDIO should have activities after reset gots active. Please help confirm if this behavior would happen in kernel after uefi.

Please be aware that you have to make sure you observe the signal in kernel but not UEFI

57

Hi!
Yes, this activity is observed in Kernel and Userspace, but it is inconsistent as mentioned above. MDC is pulled stagnant and MDIO activity is seen most of the times.

58

I don’t really think there is anything to check with software side. We may need to review hardware part or if this TI DP83867 has any extra requirement.

We got many failure case on this PHY before while other kind of PHY could work fine with same configuration.

59

Hi!
We checked our PHY with TI recommended debug kit, and it was functioning. We also tested the Orin module. They both seem to be working. Would you know of any changes to make on the driver end, say nvethernet, or of any additional requirements.
Thank you.

60

Sorry that I don’t know if any additional requirements you need here.

Actually, I don’t know what you are talking about here.

We checked our PHY with TI recommended debug kit

So what does that mean here?

We also tested the Orin module. They both seem to be working

So what does that mean you test Orin module here? What is working?

61

Hi!
We isolated the PHY chip and used TI recommended MSP430 eval kit. We ran a scan using USB2MDIO tool and it gave us an available address at 10. But when we connect to orin, we get the errors as discussed above.
When we do
sudo busybox devmem 0x02434058 (MDC apparently)
we get 0x0000058
sudo busybox devmem 0x02434010 (MDIO apparently)
we get 0x0000416
we later changed MDC using sudo busybox devmem 0x02434058 w 0x400
and the output effectively is
0x00000400
Still we observe absolutely no changes in the behaviour.
Your insight is appreciated. Thank you