TX2 not booting up with HDMI connected

I am attempting to reinstall Jetpack and reflash.
At the end of the download and installation of Jetpack in the host, I looked at the Terminal_installation.log.

There’s a write error with some tar file, yet things end with “success”:

Copying USB device mode filesystem image to /home/raphael/Jetpack/64_TX2/Linux_for_Tegra/rootfs
Disabling NetworkManager-wait-online.service
Disable the ondemand service by changing the runlevels to 'K'
Extracting the firmwares and kernel modules to /home/raphael/Jetpack/64_TX2/Linux_for_Tegra/rootfs
Extracting the kernel headers to /home/raphael/Jetpack/64_TX2/Linux_for_Tegra/rootfs/usr/src
tar: write error
Adding symlink /home/raphael/Jetpack/64_TX2/Linux_for_Tegra/rootfs/lib/modules/4.4.38-tegra/build --> /usr/src/linux-headers-4.4.38-tegra
Installing Image into /boot in target rootfs
Installing the board *.dtb files into /boot in target rootfs
Success!

Is this normal?

If that can help, see attached:

  • complete Jetpack download/installation log
  • log of the gtkterm when it started to spit out things from the Jetson, i.e, after the install process asked to put it in Force Recovery Mode.
  • flash_os_tx2.log
    Terminal_Installation.log (428 KB)
    gtkterm_log2.log (105 KB)
    flash_os_tx2.log (417 KB)

I’m not where I can check right now, but I recall there is a tar write error on some stage which doesn’t actually get in the way…will know more tomorrow.

The Xfinity modems can be logged in to, I’ll check tomorrow as well to see if I can remember where to find DHCP request information.

Normally, to clear JetPack and start clean the only thing you need to do is what you already did to delete everything…there is no special uninstall needed for flashing the Jetson.

Note that if you saw anything readable from serial console during recovery mode, then it also means you had it correct for regular boot as well. Lack of any output even during U-Boot stage when you know the serial console is otherwise set correctly tends to make me believe hardware issues are possible…but being able to also flash says the opposite, that things should work in hardware and perhaps it is just software. I’ll look closer tomorrow and see if anything shows up.

Thank you Linuxdev, I really appreciate.

It was not a problem to log into the Xfinity modem (10.0.0.1) but i just explored everywhere and didn’t find any sort of log or history of DHCP request.

Through this thread if we rule out software issues i guess i’ll contact the support team that takes care of warranty stuff and beg for a new unit. This unit is for our new A.I project at Nasa/GSFC and not being able to advertise how the Jetson would support our project has been rather frustrating. So, thanks again for your effort in helping out.

if you can not find any dhcp request you may possibly still define a static ip somehow. On the other hand they may not have the dhcp sever at all. [though I did not work with the xfinity modems and just assume they are like a typical router]

I guess I’m at an advantage that I use my host as a DHCP server on a second private network card (this makes it easy to look up DHCP requests).

On Xfinity I can see under “Connected Devices” the MAC addresses currently connected, as well as known MAC addresses which are not currently connected. Unfortunately, this doesn’t help since it has no time stamp to show when that last address assignment occurred. When I went to actual logs it only logs router DHCP information (Jetsons would be client, not router), and thus you can’t see any history of the devices connected/disconnected.

So far as the “tar: write error” issue goes I cannot find this listed in my logs. However, I am fairly sure I’ve seen this before when manually unpacking some part of the flash software without JetPack such that this was not an issue. I cannot confirm the error message has no effect, but my feeling is that this is not really an issue (there is a small chance the error is significant). My JetPack logs are incomplete since I normally use command line flash. For anyone wanting to see this look at the “Terminal_installation.log” attached in #41.

What was the operation being performed during the “gtkterm_log2.log”? What I see is a successful boot and serial console was working as expected. If you can get this same setup to log anything at all from a normal cold boot you might be in business. If this has become completely missing when you know serial console setup is otherwise valid, then I’m inclined to go with hardware failure. I realize you’ve flashed many times now, but if you flash again and can get to this point after flash completes, then you know the hardware itself is working (and perhaps it was the tar error). Log excerpt from your log:

tegra-ubuntu login: nvidia (automatic login)

Welcome to Ubuntu 16.04 LTS (GNU/Linux 4.4.38-tegra aarch64)

 * Documentation:  https://help.ubuntu.com/

618 packages can be updated.
301 updates are security updates.

The programs included with the Ubuntu system are free software;
the exact distribution terms for each program are described in the
individual files in /usr/share/doc/*/copyright.

Ubuntu comes with ABSOLUTELY NO WARRANTY, to the extent permitted by
applicable law.

To run a command as administrator (user "root"), use "sudo <command>".
See "man sudo_root" for details.

nvidia@tegra-ubuntu:~$ [   17.139461] fuse init (API version 7.23)

The flash log itself shows success. The excerpt I gave above would be from the serial console after the flash completed and the unit rebooted. It really seems like the system is running other than no network and no serial console (I keep hoping that serial console issues might be something other than hardware failure).

So here is what I’m going to recommend: Do a manual flash without JetPack to simplify things as much as possible. Keep the serial console going during the flash, but be careful to not accidentally hit any keystrokes (during reboot this could halt things). The following instructions are for R28.2, but other than packages now being R28.2.1 there is no difference:
https://devtalk.nvidia.com/default/topic/1031769/jetson-tx2/unable-install-any-jetpacks-jetson-tx2/post/5249710/#5249710

At the point where it says this on the flash.sh output the actual flash is done and the Jetson should reboot by itself (this is from the flash.sh output, you do not need to manually force reboot, but you could force this if nothing shows after several minutes):

*** The target t186ref has been flashed successfully. ***
Reset the board to boot from internal eMMC.

…any further serial console output is from boot…any previous serial console output is from flash. Any output there at all from serial console during flash says the serial console was correctly set up. Any output at all from after reboot begins says the Jetson is booting and is alive. What is needed is to determine if a known good serial console has any output at all during boot which follows flash.

Be sure to set your serial console terminal for logging. Note that the URL I gave above for command line flash shows an option for logging the flash itself…you will want to do this:

sudo ./flash.sh -S 29318MiB jetson-tx2 mmcblk0p1 2>&1 | tee log_flash.txt

…but I’m going to modify this in a way which will cut down on progress bar spamming the log (the progress bar is actually a major part of the log size):

sudo ./flash.sh -S 29318MiB jetson-tx2 mmcblk0p1 2>&1 <b>| gawk '{gsub("[0-9][0-9]+[/][0-9[0-9]+ bytes sent..",".");print}'</b> | tee log_flash.txt

(the “gawk” will just cut out parts of the progress bar and substitute a single “.”)

If you notice a tar error on command line flash I can look that up.

This gtkterm log was written during the flashing process, right after I hit enter on that xterm screen that instructs to put the Jetson on Force Recovery Mode. After the installation ended, I uploaded the log here. Once I power off, powering back on gives me a mute Jetson again. Even if I reboot in Force Recovery Mode, serial stays also mute.

I’ll try flashing without Jetson, following your procedure. Thanks again for helping.

I have followed the reflash procedure without Jetpack. The Jetson is still mute after power off. Like with Jetpack, serial console is verbose during flash and during the 1st boot that automatically follows the flash. Mute after power off and back on.

There were no write error this time though:

raphael@raphael-All-Series:~/Linux_for_Tegra$ sudo ./apply_binaries.sh 
Using rootfs directory of: /home/raphael/Linux_for_Tegra/rootfs
Extracting the NVIDIA user space components to /home/raphael/Linux_for_Tegra/rootfs
Extracting the BSP test tools to /home/raphael/Linux_for_Tegra/rootfs
Extracting the NVIDIA gst test applications to /home/raphael/Linux_for_Tegra/rootfs
Extracting the configuration files for the supplied root filesystem to /home/raphael/Linux_for_Tegra/rootfs
Creating a symbolic link nvgstplayer pointing to nvgstplayer-1.0
Creating a symbolic link nvgstcapture pointing to nvgstcapture-1.0
Extracting Weston to /home/raphael/Linux_for_Tegra/rootfs
Adding symlink libcuda.so --> libcuda.so.1.1 in target rootfs
Adding symlink libGL.so --> libGL.so.1 in target rootfs
Adding symlink libnvbuf_utils.so --> libnvbuf_utils.so.1.0.0 in target rootfs
Adding symlink libnvid_mapper.so --> libnvid_mapper.so.1.0.0 in target rootfs
Adding symlink libcuda.so --> tegra/libcuda.so in target rootfs
Adding symlink libEGL.so --> libEGL.so.1 in target rootfs
Adding symlink libGLESv2.so --> libGLESv2.so.2 in target rootfs
Adding symlink /home/raphael/Linux_for_Tegra/rootfs/usr/lib/aarch64-linux-gnu/libdrm_nvdc.so --> /home/raphael/Linux_for_Tegra/rootfs/usr/lib/aarch64-linux-gnu/tegra/libdrm.so.2
Adding symlink nvidia_icd.json --> /etc/vulkan/icd.d/nvidia_icd.json in target rootfs
Creating symlinks for weston, wayland-demos, libinput and wayland-ivi-extention modules
Adding symlinks for NVIDIA systemd services
Copying USB device mode filesystem image to /home/raphael/Linux_for_Tegra/rootfs
Disabling NetworkManager-wait-online.service
Disable the ondemand service by changing the runlevels to 'K'
Extracting the firmwares and kernel modules to /home/raphael/Linux_for_Tegra/rootfs
Extracting the kernel headers to /home/raphael/Linux_for_Tegra/rootfs/usr/src
Adding symlink /home/raphael/Linux_for_Tegra/rootfs/lib/modules/4.4.38-tegra/build --> /usr/src/linux-headers-4.4.38-tegra
Installing Image into /boot in target rootfs
Installing the board *.dtb files into /boot in target rootfs
Success!

I attached the log from gtkterm, as well as the gawk-modified log generated with your new flash command.

Another thing i’ve noticed, maybe not relevant but odd: any time I flash, with or without Jetson (I had tried to install without Jetpack the 1st time I got the board in fact), the ethernet connection from the host pc gets “disconnected” after Ubuntu starts on the Jetson, at the end of the automatic reboot that follows the flash.

Thanks
gtkterm_flash.log (110 KB)
log_flash.txt (419 KB)

A last observation…

I couldn’t find anything on the documentation about that little jumper placed on my board, between the power button and the rec button. I noticed it yesterday. I have attached a picture of my board with that jumper. I’m asking because I do not see it in other online photo of the dev board. Is it supposed to be here at all?

According to the logs the system flashed correctly, booted correctly, and was using serial console correctly. Nothing seems wrong with the board.

That J9 jumper though is a good catch! J9 is a normally open 2-pin header between ground and recovery. This header is intended to allow an external button (such as if the Jetson is in a case) to be used for recovery. With a jumper there the recovery button is held down 100% of the time! The only way to reboot normally would probably be via the tegraboot command. Your system seems to be working as expected…I would bet that if you remove that jumper it starts booting correctly and serial console and all else would work.

If that was left on from production, and if this was the case for a number of boards, then I’m going to expect to see a lot more help calls. NVIDIA might want to mention to the customer service reps to check if J9 has a jumper that shouldn’t be there when an otherwise working system does not seem to flash correctly.

Can we verify if the system works after removing the J9 jumper?

Houston, we have lift-off!

Good catch indeed… i will not comment on the amount of inner screaming this involved ;)

Nvidia must be notified of this and it woud be in everyone’s best interest to update things so it is one of the the first thing to check when we troubleshoot. I had ordered the Jetson from the last year discount for registered developers which followed the Nvidia GTC DC (purchase was November last year).

I suspect that other people must have been affected, clearly i can’t be the only one who got the privilege of having a jumper for free…
For the poor souls having that jumper issue unknowingly, If the warranty was used to get a new board, they may have received the new one without the jumper, leading them (user & Nvidia) to wrongly confirm a hardware failure where there were none. Regardless, our experience here shows how much time was taken on both sides.

Regardless of that jumper story, thank you very much for your effort in helping out. That did lead me to check the board up close in the end. Without your effort maybe i wouldn’t have done that check and we wouldn’t have been… “successfool”? (forgive the terrible pun)

For completeness, I am attaching the gtkterm log of the successful boot. And again for completeness, I’ll reinstall/reflash with Jetpack and check again if i also have a successful boot, with hdmi, gtkterm,… i.e. the whole chatty family.

Thanks
gtkterm_flash2.log (180 KB)

@WaaallEEE,

we meet the same problem when connecting tx2 to a projector. The packet version is r28.2.1.

I downloaded the kernel source according to “l4t-documentation”, but could not find “drivers/video/tegra/dc/hdmi2.0.c”.

Will you please give me some help, thanks!

Could you share what you saw in your kernel source? I think it is very fast to just use command “find -iname hdmi2.0.c” to find this file…

I downloaded the kernel source with the script “source_sync.sh”. The tag is “tegra-l4t-r28.2.1”. Then i got a fold of “Linux_for_Tegra/sources/kernel”, which has “display” “kernel-4.4” inside.
I can only find hdmi2.0.c in the fold of “display”,but there is no codes for the patch to replace in the file.

./display/drivers/video/tegra/dc/hdmi2.0.c

Sorry that this thread was almost a year ago so I am not sure which patch are you talking about.

Do you mean this one ?
https://devtalk.nvidia.com/default/topic/1003956/jetson-tx2/tx2-not-booting-up-with-hdmi-connected/post/5160211/#5160211

yes, is there any update about the patch?

The official fix is on rel-32.1.

ok, we will update to 32.1…
Thank you!