Any update of PCIe driver and/or USB3.0 driver for TX2's L4T ?

I use four USB3.0 cameras, Basler/daA1600-60us, with TX1/TX2, but some errors occur with TX2.
The place and the timing where the error occurs are not fixed. It does not always happen.
e.g.
at opening camera device
at setting parameter of camera device
at starting grabbing image
at grabbing data

And, errors rarely occur when using with TX1.

My environment is as follows.

  • Jetson: TX1 or TX2.
  • JetPack: 3.1 (L4T R28.1)
  • USB3.0: Jetson's PCIe x4 - PEX8608 - four uPD720202 - four USB3.0 cameras.
  • Basler camera: daA1600-60uc
  • Basler Pylon: 5.0.11
  • Basler's Max Transfer size: 2000000 (as 2GB)
  • /sys/module/usbcore/parameters/usbfs_memory_mb: 2000 (as 2GB)

For example, a Basler’s Pylon library outputs following message when failed to open a camera device.

[b]
Pylon::GenericException:

Failed to open device '2676:ba03:10:1:2' for XML file download. Error: 'Failed to read the first register (maximum device response time). This can be caused by problems with the USB hardware used, i.e., the USB host controller and its corresponding driver.[/b]

So, I guess there is any difference between TX1’s L4T and TX2’s about PCIe and/or USB3.0 drivers.
If there is any update of PCIe driver and/or USB3.0 driver for TX2’s L4T, please give me some information.

Best Regards,

1 Like

Hi nanoto,
So the issue is seen on developer kit + PICe x4 card with 4 USB3.0 camera connected?

Please also try the patch at
[url]https://devtalk.nvidia.com/default/topic/1027100/jetson-tx2/stream-4-cameras-with-gstreamer/post/5225245/#5225245[/url]

Hi DaneLLL,

Thanks for your information. I try the patch.

Four USB3.0 cameras are connected our carrier board.
It has a PEX8608, four uPD720202s, and USB3.0 ports, and the carrier board works well with TX1.

Best Regards,

Hi nanoto,
for your custom board, which is your HW config number per oem design guide?
https://developer.nvidia.com/embedded/dlc/jetson-tx2-oem-product-designguide

Have you applied the patch at Jetson/TX2 USB - eLinux.org ?

Hi DaneLLL

Thank you for your additional comment.

I tried patch for xhci-pci.c, but it was no effect unfortunately.
Is there any thing I should try in addition?

for your custom board, which is your HW config number per oem design guide?

I use default of L4T R28.1, so HW config number is 2.

Have you applied the patch at Jetson/TX2 USB - eLinux.org ?

I don’t use the USB3.0 port of TX2. But, should I apply this patch?

Best Regards,

Hi nanoto,

You need it if you see the error:

[    4.404116] xhci-tegra 3530000.xhci: can't get usb3-0 phy (-517)

Hi DaneLLL,

Thank you for your message. And I’m sorry for my late message.

The error message you told does not occur.

It seems that my problem is caused by difference of between TX1 and TX2.
So, I tried to find the difference of TX1 and TX2 regarding PCIe and/or USB.
Then, I found a difference the PCIe’s ASPM. TX1 is Disable, and TX2 is L0s and L1 Enabled.
I changed TX2’s ASPM to disable by setting “performance” to “/sys/module/pcie_aspm/parameters/policy”.
By this setting, the TX2’s ASPM becomes Disabled. But, the problem is not solved.

Could you please give me any advice to solve this problem?
I don’t know why the TX2 is more unstable than TX1 when capturing by USB3.0, because TX 2 should have better performance in CPU power and CPU cores than TX1.

Best Regards,

Hi nanoto, does it work fine with one or two or three USB3 cameras? Looks like you run 4 cameras. Please try other cases also.

And does the camera use uvcvideo driver?

Hi DaneLLL,

Thank you for your message.

In any case, the problem occurs with TX2.
No, I don’t use uvcvideo driver.
I use the Pylon library and driver provided by Basler.

Best Regards,

nanoto,
When using 3 cameras configuration, does the issue occur the same way as that of existing 4 cameras case? This simply is a cross reference purpose. Thanks.

Hi chijen,

Thank you for your reply.
Yes, it does.
It often occurs when stopping capture and restarting capture.

Best Regards,

nanoto,
we will collect our USB 3.0 testing outcome with 3 or 4 connections and let you know mostly under TX2.

"errors rarely occur when using with TX1.
=> so is the error observed using Jetson TX1 or not?

Hi chijen,

"errors rarely occur when using with TX1.
=> so is the error observed using Jetson TX1 or not?

First I heard that it also happen in TX 1, but as a result of reconfirmation I confirmed that it does not occur in TX1.

Best Regards,

Can you please share the kernel log when the issue occurs?
Also, by error you mean, the system is still accessible after it occurs or does it hang completely?
Do you see error occurring even with one camera? with and without switch used?

nanoto,

"Do you see error occurring even with one camera? with and without switch used?
=>
What vidyas stated above is crucial for the issue debugging so please help to provide the outcome. Thanks.

hi vidyas and chijen,

Thank you for your reply, and sorry for my late response.

Can you please share the kernel log when the issue occurs?

Yes, I can share it.
But, I shipped camera’s with TX1 because TX1 works stable.
I need to prepare camera’s. Please give me a time.

Also, by error you mean, the system is still accessible after it occurs or does it hang completely?

System is still accessible. But, a basler camera which occurred error can’t be accessible.
If the power supply of error camera is turned off and then turned on, the camera can be accessible again.

Do you see error occurring even with one camera? with and without switch used?

Yes. Sometimes it does not occur easily, and sometimes it occurs frequently.

Best Regards,

hi vidyas and chijen,

I got log files from the engineer for the moment. I attached them.

If you need other information, please do not hesitate to say.

Best Regards,
dmesg@NG_180122-1.txt (77.3 KB)
dmesg@NG_180122-2.txt (80.5 KB)
dmesg@OK_180122-1.txt (73 KB)
dmesg@OK_180122-2.txt (73 KB)

Hi,

This issue should be fixed after rel-28.2. Please try latest Jetpack3.2 if possible.

Thanks.

Hi WayneWWW,

Thank you for letting me know. I will try it.
Could you please tell me a cause of this issue?

Best,

Hi WayneWWW,

We tested with R28.2, but this problem wasn’t solved.
I attached the log file.

Best Regards,
dmesg@NG_180315-2.txt (96.1 KB)