HDMI and DisplayPort (Dual display mode) Bandwidth issue at Jetpack4.2(kernel 4.9.140 tegra)


I am using a carrier board designed for TX2. (no devkit EV board, board with DP port added)
The dts file I used was “tegra186-quill-p3310-1000-c03-00-dsi-hdmi-dp.dts” and enabled the DP port.

However, when using nvpmodel 0 (CPU 2GHz, GPU 1.3GHz with Cooler on) to use the full performance of the system, it sometimes happens that CPU gets locked or Ubuntu skin is broken.
It seems as if the memory arbitor is not scheduling properly due to a lack of bandwidth.

This phenomenon did not occur in Jetpack 3.2 (4.4.38 tegra kernel: same usage condition).

Have you made any special changes to Jetpack 4.2?

Symptom is reproducible when both HDMI and DP are connected.

Thank you.


Thanks for report.
Could you share a photo of such error?
Did you see any kernel error in dmesg?
Did you hit this error with any special application running?

The rate at which the issue is reproduced is much more that the CPU stops, and there is no dmesg in the kernel at that time.
When the problem is reproduced, the system is stopped and any input (mouse event or Keyboard etc …) is not available and the recovery to normal operation is not possible except HW reset or Power off and on method.

I will try to reproduce the phenomenon that the display screen is broken and try to catch the error message.

Tank you


Below is my dts diff data from original dts

— tegra186-quill-p3310-1000-c03-00-dsi-hdmi-dp-orig.dts 2019-05-30 11:49:48.361550183 +0900
+++ tegra186-quill-p3310-1000-c03-00-dsi-hdmi-dp.dts 2019-05-30 11:49:08.892610807 +0900
@@ -7867,7 +7867,7 @@
pinctrl-9 = <0x85>;
pinctrl-10 = <0x86>;
pinctrl-11 = <0x87>;

  •                   status = "disabled";
  •                   status = "okay";
                      fb_reserved = <0x8a>;
                      iommu-direct-regions = <0x88 0x89 0x8a>;
                      nvidia,dc-flags = <0x1>;

@@ -8182,7 +8182,7 @@
clock-names = “sor0_ref”, “sor_safe”, “sor0_pad_clkout”, “sor0”, “pll_dp”, “pllp_out0”, “maud”, “hda”, “hda2codec_2x”, “hda2hdmi”;
resets = <0x10 0x27 0x10 0xf 0x10 0x10 0x10 0x11>;
reset-names = “sor0”, “hda_rst”, “hda2codec_2x_rst”, “hda2hdmi_rst”;

  •                   status = "disabled";
  •                   status = "okay";
                      nvidia,ddc-i2c-bus = <0x97>;
                      nvidia,active-panel = <0x98>;
                      nvidia,hpd-gpio = <0x1b 0x78 0x1>;

@@ -8209,7 +8209,7 @@

                    dp-display {
                            compatible = "dp, display";
  •                           status = "disabled";
  •                           status = "okay";
                              nvidia,is_ext_dp_panel = <0x1>;
                              linux,phandle = <0x98>;
                              phandle = <0x98>;

@@ -8724,7 +8724,7 @@
resets = <0x10 0x5>;
reset-names = “dpaux”;
power-domains = <0x9c>;

  •                   status = "disabled";
  •                   status = "okay";
                      linux,phandle = <0x96>;
                      phandle = <0x96>;


Thanks. Actually we didn’t meet such issue before. Even our partners with custom carrier board (2 display usecase) don’t report such issue before. Thus, we need your information to show us how to reproduce this issue.

Also, are both display in 4k mode? Could you check what is the criteria (besides nvpmodel mode 0) to reproduce issue?

There is no change to display resolution (default value)
The monitors I connected are HDMI (2560x1080-21: 9) and DP (2560x1080-21: 9).

Removing one of the monitor connections also significantly reduces the problem.


It suddenly comes to my mind it may be a driver issue. I will check with our internal team.

Confirmed dual display should work on TX2+ jetpack4.2. Please share the full dts with us.

Please note that you cannot directly use the dts from jetpack3.3.

I did not use jetpack3.3 or jetpack 3.2’s dts file
I used jetpack4.2’s dts file

and I already sent diff file from jetpack4.2’s “tegra186-quill-p3310-1000-c03-00-dsi-hdmi-dp.dts”

you can see above message #3
tegra186-quill-p3310-1000-c03-00-dsi-hdmi-dp.dts.gz (58.2 KB)


Still questions here.

  1. Do both display work? I mean the gnome desktop shows or not?

If the gnome desktop shows up, it means the display driver has no issue. Otherwise it is (display) dts/driver issue.


  1. Is there any special application or setting that would hit error?
  1. Yes. both displays work!

  2. I did not do anything except set the DTS file to “okay” from “disable” to make the two displays work.

And I did not run any applications.

I only run the command “sudo nvpmodel -m 0”
The above command was executed when jetpack3.2, and there was no problem at that time.

How about other monitor with larger pixel data? For example, dual 4k monitors.

I tested just my monitor (21:9). I’ll test with other monitor

We failed to reproduce your issue on devkit with 2x 4k monitors. Is running “nvpmodel” a necessary step to reproduce issue?

I forgot one thing!

I ran “nvpmodel -m 0” and ran “jetson_clocks” shell to run fan.

Sorry that we still cannot reproduce issue after running above two application.

Could you share the reproduce rate? Are you sure there is nothing in dmesg/syslog when error happens?

It stops all, even the mouse and the keyboard do not work.

The most common occurrence is that ssh connections are often disconnected
And the system hangs more often than not

Is it possible that the HW settings are not correct?

Could the schematic of our newly designed carrier board go wrong?
If you give us your opinion, we will discuss with the HW person in charge of our company.

Ok I will reproduce the symptom tomorrow with debug(serial uart) cable and check it out!

symptom was reproduced (24 hour life test progress)
This time, the symptom is not the CPU
I got a GPU error symptom and got a log.

  • reproducing condition
    1. run : nvpmodel -m 0, /usr/bin/jetson_clocks
    2. youtube playing
    3. camera application playing (1280x720)
      ==> 6/12(10:00 AM) --> 6/13(8:00 AM) :Korean time

display blackout

Logged below.

[42671.334857] hrtimer: interrupt took 126944 ns
[67487.615996] nvgpu: 17000000.gp10b __nvgpu_timeout_expired_msg_cpu:94 [ERR] Timeout detected @ nvgpu_0
[67487.631760] nvgpu: 17000000.gp10b _nvgpu_timeout_expired_msg_cpu:94 [ERR] Timeout detected @ nvgpu
[67487.783563] nvgpu: 17000000.gp10b __nvgpu_timeout_expired_msg_cpu:94 [ERR] Timeout detected @ nvgpu_0
[67487.799198] nvgpu: 17000000.gp10b _nvgpu_timeout_expired_msg_cpu:94 [ERR] Timeout detected @ nvgpu
[67487.949303] nvgpu: 17000000.gp10b __nvgpu_timeout_expired_msg_cpu:94 [ERR] Timeout detected @ nvgpu_0
[67487.964954] nvgpu: 17000000.gp10b _nvgpu_timeout_expired_msg_cpu:94 [ERR] Timeout detected @ nvgpu

Hi bk1472,

Can you clarify what’s the issue now?
The original issue is “CPU gets locked or Ubuntu skin is broken”, is this issue still exist?
If you have extend or new issue, please file new topic.