Nvidia pegasus flashing failed

Please provide the following info:
Hardware Platform: DRIVE AGX Pegasus™ Developer Kit
Software Version: DRIVE Software 10
Host Machine Version: native Ubuntu 18.04
SDK Manager Version: Example: 1.3.1.7110

I’m trying the installation/flashing on the pegasus for the first time.
Installation finished successfully, but flashing failed.

The only relevant info I was able to find points me to https://developer.download.nvidia.com/driveworks/secure/docs/DRIVE_8.0_Release_Docs/NVIDIA_DRIVE_AGX_Flashing_Issues_Technical_Bulletin.pdf?cXaBps-at1iCJP5Xv-bq-SYzg9w1IempAu4HOiLBYtIBqtGJ_2PDSUK_E-vBsB20nEvbY-TNkX2MKcgx8U8GUMrXa1ZwgJncqc83UGW25A6oU-YSynqmTT7szPN1Lmm70qf66JHdoDNqL2Racsr2vrMmIYvlOUG15JbsNWriJpL-cE5FoqSpQjqU7VoOPO8cGXsSgRAsuO4hpCC7Ukt4kMtruWY

Section 2.3 described a similar situation.

However, I’m not sure (1) if that doc (for 8.0) is still applicable here and (2) the solution it gave was to “flash each SoC sequentially” but I’m not sure how to do that.

Error log attached below.

16:21:47 INFO: Flash Xavier A+B in parallel: ------------ Stack Trace ------------

16:21:47 INFO: Flash Xavier A+B in parallel: stack frame 0 - 329 AbnormalTermination /home/linyang/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/drive-t186ref-foundation/tools/host/flashtools/bootburn_t19x/bootburn_lib.sh

16:21:47 INFO: Flash Xavier A+B in parallel: stack frame 1 - 1286 GetTargetECID /home/linyang/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/drive-t186ref-foundation/tools/host/flashtools/bootburn_t19x/bootburn_lib.sh

16:21:47 INFO: Flash Xavier A+B in parallel: stack frame 2 - 378 source /home/linyang/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/drive-t186ref-foundation/tools/host/flashtools/bootburn_t19x/bootburn_active.sh

16:21:47 INFO: Flash Xavier A+B in parallel: stack frame 3 - 1270 main /home/linyang/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/drive-t186ref-foundation//tools/host/flashtools/bootburn_t19x/bootburn.sh

16:21:47 INFO: Flash Xavier A+B in parallel: -------------------------------------

16:21:47 INFO: Flash Xavier A+B in parallel:

16:21:47 INFO: Flash Xavier A+B in parallel: error-tool-tegrarcm-chipinfo – Failed to get UID of chip

16:21:47 INFO: Flash Xavier A+B in parallel:

16:21:47 INFO: Flash Xavier A+B in parallel: Flashing process exited with error 50

16:21:47 INFO: Flash Xavier A+B in parallel: Error in flashing!

16:21:47 INFO: Flash Xavier A+B in parallel: pdkinstaller cannot continue! Check the log file for details: /home/linyang/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/pdkinstaller/pdkinstaller.log

16:21:47 INFO: Flash Xavier A+B in parallel: [ Component Install Finished with Error ]

16:21:47 INFO: Flash Xavier A+B in parallel: [host] [ 706.40 MB used. Disk Avail: 737.67 GB ]

16:21:47 INFO: Flash Xavier A+B in parallel: [ NV_FLASH_XAVIER_PDKFLASH_PARALLEL_COMP Install took 15m40s ]

16:21:47 ERROR: Flash Xavier A+B in parallel: [error]: : [exec_command]: /bin/bash -c cd pdkinstaller && ./pdkinstaller --installtype bindflash --boardtype ES --pdkinstalledpath /home/linyang/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS --configjson ./manifests/install_pdk_config/av_ddp_pegasus_hyp71_linux_tegraAB_install_config.json --consoleout enable; [error]:

16:21:47 ERROR: Flash Xavier A+B in parallel: command terminated with error

16:21:48 SUMMARY: Flash Xavier A+B in parallel: Failed to execute commands with GenericInstaller

Hi @yuzhang,

Please tar ~/.nvsdkm and share it for our checking. At the same time, you can try to install on another host system. Thanks!

Please also provide the pdkinstaller.log. Thanks!

Thanks Vick. Files are attached.nvsdkm.tar.gz (2.8 MB) pdkinstaller.log (454.7 KB)

Please check below three related topics to see if helps. Maybe you can try with another host system or USB 2.0 A to A cable. Thanks!

@VickNV regarding the first fix, can you also provide a link on how to flash the two SoCs sequentially? I couldn’t find any instruction in the pdf doc. Thanks

I found “detected no board connected to host with USB cable” message even earlier in your log.
Please refer to the topic and make sure your USB 2.0 A to A cable connected correctly.

Hm… interesting. I’ll try it out and report back, but I’m pretty sure I did ‘lsusb’ and saw all the future technology lines showing up.

The usb connection looks good to me. I’m using the provided A-A cable with one end connected to the debug port, and I can see the future technology items by running “lsusb”.

However, I realized that I don’t have cuda installed (no nvcc).

I tried installing cuda 10.2 myself and now flashing fails with a different error.
The version I installed (via apt install) is 10.2.89 which > 10.2.19 so I’m sure why it’s complaining about it.

linyang@DPM-UBU005:~$ nvcc -V
nvcc: NVIDIA ® Cuda compiler driver
Copyright © 2005-2019 NVIDIA Corporation
Built on Wed_Oct_23_19:24:38_PDT_2019
Cuda compilation tools, release 10.2, V10.2.89

SDK manager terminal log below:

20:45:10 ERROR: Flash Xavier A+B in parallel: command terminated with error

20:45:10 SUMMARY: Flash Xavier A+B in parallel: Failed to execute commands with GenericInstaller

20:45:10 DEBUG: running command < true >

20:45:10 INFO: command finished successfully

20:45:10 DEBUG: CUDA on Host: using adapter to install NV_CUDA_HOST_COMP@host to /home/linyang/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_host

20:45:10 INFO: CUDA on Host: exec_command: dpkg -f /home/linyang/Downloads/nvidia/sdkm_downloads/cuda-repo-ubuntu1804-10-2-local-10.2.19-430.17_1.0-1_amd64.deb Package | tr -d ‘\n’

20:45:10 INFO: CUDA on Host: exec_command: dpkg -f /home/linyang/Downloads/nvidia/sdkm_downloads/cuda-repo-ubuntu1804-10-2-local-10.2.19-430.17_1.0-1_amd64.deb Version | tr -d ‘\n’

20:45:10 INFO: CUDA on Host: exec_command: sudo -E apt-get install -y --allow-downgrades ‘/home/linyang/Downloads/nvidia/sdkm_downloads/cuda-repo-ubuntu1804-10-2-local-10.2.19-430.17_1.0-1_amd64.deb’

20:45:10 INFO: CUDA on Host: Reading package lists…

20:45:10 INFO: CUDA on Host: Building dependency tree…

20:45:10 INFO: CUDA on Host: Reading state information…

20:45:11 INFO: CUDA on Host: 0 upgraded, 0 newly installed, 0 to remove and 22 not upgraded.

20:45:11 INFO: CUDA on Host: exec_command: dpkg -L cuda-repo-ubuntu1804-10-2-local-10.2.19-430.17 | grep ‘.pub$’ | tr -d ‘\n’

20:45:11 INFO: CUDA on Host: exec_command: sudo apt-key add /var/cuda-repo-10-2-local-10.2.19-430.17/7fa2af80.pub

20:45:11 : CUDA on Host: Warning: apt-key output should not be parsed (stdout is not a terminal)

20:45:11 INFO: CUDA on Host: OK

20:45:11 INFO: CUDA on Host: exec_command: sudo apt-get check

20:45:11 INFO: CUDA on Host: Reading package lists…

20:45:11 INFO: CUDA on Host: Building dependency tree…

20:45:11 INFO: CUDA on Host: Reading state information…

20:45:11 INFO: CUDA on Host: exec_command: sudo apt-get update; sleep 0.5; sudo apt-get -y --allow-downgrades install cuda-toolkit-10-2=10.2.19*

20:45:11 INFO: CUDA on Host: Get:1 file:/var/nv-tensorrt-repo-cuda10.2-trt5.1.4.2-ga-20190506 InRelease

20:45:11 INFO: CUDA on Host: Get:2 file:/var/nv-tensorrt-repo-cuda10.2-trt5.1.4.2-ga-20190506 Release [574 B]

20:45:11 INFO: CUDA on Host: Get:2 file:/var/nv-tensorrt-repo-cuda10.2-trt5.1.4.2-ga-20190506 Release [574 B]

20:45:12 INFO: CUDA on Host: Ign:4 http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 InRelease

20:45:12 INFO: CUDA on Host: Hit:5 http://packages.microsoft.com/repos/vscode stable InRelease

20:45:12 INFO: CUDA on Host: Ign:6 http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64 InRelease

20:45:12 INFO: CUDA on Host: Hit:7 http://dl.google.com/linux/chrome/deb stable InRelease

20:45:12 INFO: CUDA on Host: Get:8 https://download.docker.com/linux/ubuntu bionic InRelease [64.4 kB]

20:45:12 INFO: CUDA on Host: Hit:9 http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 Release

20:45:12 INFO: CUDA on Host: Hit:10 https://deb.nodesource.com/node_12.x bionic InRelease

20:45:12 INFO: CUDA on Host: Hit:11 http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64 Release

20:45:12 INFO: CUDA on Host: Hit:12 http://security.ubuntu.com/ubuntu bionic-security InRelease

20:45:17 INFO: CUDA on Host: Hit:16 http://us.archive.ubuntu.com/ubuntu bionic InRelease

20:45:17 INFO: CUDA on Host: Hit:17 http://us.archive.ubuntu.com/ubuntu bionic-updates InRelease

20:45:17 INFO: CUDA on Host: Hit:18 http://ppa.launchpad.net/danielrichter2007/grub-customizer/ubuntu bionic InRelease

20:45:17 INFO: CUDA on Host: Hit:19 http://us.archive.ubuntu.com/ubuntu bionic-backports InRelease

20:45:17 INFO: CUDA on Host: Hit:15 https://packagecloud.io/slacktechnologies/slack/debian jessie InRelease

20:45:17 INFO: CUDA on Host: Hit:20 http://ppa.launchpad.net/graphics-drivers/ppa/ubuntu bionic InRelease

20:45:17 INFO: CUDA on Host: Hit:21 http://ppa.launchpad.net/linrunner/tlp/ubuntu bionic InRelease

20:45:18 INFO: CUDA on Host: Fetched 64.4 kB in 7s (9,525 B/s)

20:45:19 INFO: CUDA on Host: Reading package lists…

20:45:20 INFO: CUDA on Host: Reading package lists…

20:45:20 INFO: CUDA on Host: Building dependency tree…

20:45:20 INFO: CUDA on Host: Reading state information…

20:45:20 ERROR: CUDA on Host: E: Version ‘10.2.19*’ for ‘cuda-toolkit-10-2’ was not found

20:45:20 ERROR: CUDA on Host: command terminated with error

20:45:21 SUMMARY: DRIVE AV App Documentation: Depends on failed component

Please check if the solution in below topic can also fix on your problem.

Thanks @VickNV I’ll take a look

Meanwhile I’m also gonna try my luck with a different host machine. A few things I want to confirm:
(1) the host machine doesn’t need to have a gpu card right? (as long as it has ubuntu 18.04 and can install cuda and nvidia drivers - which the SDK manager will take care of)
(2) I don’t really need to connect the host with the dev platform until the “flash” step
(3) on this page: https://docs.nvidia.com/drive/drive_os_5.1.6.1L/drive-qsg/get-started-drive-software/index.html it lists cuda 10.2.19 and graphics driver >= 418.39 as prerequisites, but I also see “CUDA” as one of the host components managed by SDK manager. Do I need to install cuda and/or nvidia drivers myself or will the SDK manager take care of it?
(4) on the same page it mentions that “2 ethernet connectors are required”, but the youtube video (https://www.youtube.com/watch?v=FhyjhC5rD5g&feature=emb_logo) only uses one. Do I need to connect ethernet cables to both ports A and B of the ethernet adapter?

I havne’t tried without dGPU.

It depends on SDK Manager version. You can just connect before installation.

SDK Manager will take care of it.

one is enough (the wrong statement will be removed in the next release) .

It’s a different issue on a different host… SDK manager wouldn’t be able to download many of the files. Either it will show checksum error or download just halts (showing 0B/s). However if I copy a url from the terminal, either a failed one or a halted one, and download it manually, I can do that, and the speed is very fast. I wonder if there is a way that I can get the list of files to download and do it manually, and copy the files to the download location.

Hi @yuzhang,

If it’s a different issue, please create another topic for it and we will help you there. Thanks!

I was able to flash it with a different host. Thanks for your help

1 Like