No, you don’t… just follow what I said…
There is no IP when the board is in recovery mode…
I want to install components but on that menu, if I click install button, that warning messages appear.
It says I should connect board to install…
Ok, first question here.
Is your jetson booting up now?
If yes, then second question :
What is the “ifconfig “ shown on your jetson?
Hi,
Since there is no valid IP on your jetson… so host side is not able to install sdk…
Also, usb device mode is not detected either…
Then, you mean jetson is not booted properly?
I want to re-flash jeston to clean install L4T.
Do you have the type C port connected? This is the usb device mode port I am talking about.
This is not 100% the fault on jetson… because this usb network requires host side to support it too.
You can also use ethernet or any other interface that you can ping from your host…
For example, if you can ssh to jetson by using wifi, then you can use that wifi IP to sdkamanger too…
The screenshot right before the reply is from jetson terminal.
At that time, I did not connected C port with host.
I can ssh to jetson on host pc when jetson is booted, but if I connect with C port and make jetson on recovery mode, I could not.
I tried ipv4 with 192.168.0.247 which is on last screenshot, but same error happened…
I tried to flash other jetson agx board but I got same error so I think the host is problem.
I cannot find /dev/ttyUSB0 port but I can get Nvidia Corp. from lsusb.
How can I get ttyUSB0 port?
Hi,
There is totally no ttyUSB0 mentioned in the whole discussion. Why do you want to search for that node…
What do you want to do here?
When you connect “type C” usb port on your jetson with your host, if jetson side comes out a “l4tbr0” interface with 192.168.55.x in ifconfig, then you can put this ip to sdkmanager.
Also, make sure you can ping or ssh from host side with 192.168.55.x IP…
If the interface does not come out, then change the “Connection” from “USB” to “Ethernet” in sdkmanager GUI and then give any IP from jetson ethenret…
And again… you need to make sure your host needs to ping /ssh to jetson with that IP too.
Again…installing SDK does not need recovery mode. Only reflashing the whole board needs it…
Installing SDK is using scp and ssh to install file… so make sure ssh can work with whatever IP you want to use…
I thought ttyUSB0 should be recognized if jetson is properly connected and sorry for my misunderstanding.
I cannot ping 192.158.55.1 when I connect jetson on recovery mode.
I want to reflash whole board.
How can I connect with that ip?
Hi,
Let me clarify this again…
-
When your board is in recovery mode → You can only reflash the “OS”, you cannot install “SDK” in this situation. There is no “IP” in recovery mode. Jetson is totally a brick in this situation. So it is normal that it would not able to ping it…
-
When your board boots up into linux → you can install sdk in this situation… Give the IP from l4tbrt0 interface to your sdkmanager and it will install the SDK…
Please be awara that “SDK” only includes CUDA/TensorRT…etc. It does not include the Linux for tegra OS…
So, if you want to reflash the board, don’t care about the IP…
Also, above “OS” and “SDK” could be install separately… so if this really confused you, just click them separately in the SDKM GUI…
Then, if I use manual setup on SDKM, I can reflash my board?
I want to cleanly setup my board.
If I use manual setup, I got this error log.
SDKM_logs_JetPack_4.6.2_Linux_for_Jetson_AGX_Xavier_modules_2022-10-18_15-33-26.zip (123.8 KB)
Hi,
Could you clarify what method can flash your board previoulsy? I think the issues here are:
-
If you set “manual setup” in sdkmanager, then you cannot flash the board
-
You don’t know how to install SDK.
So please do not mix them up. Your problem has 2 issues. And they are totally different.
I used
$ sudo ./flash.sh jetson-xavier mmcblk0p1
command on terminal which was on ~/nvidia/nvidia_sdk/JetPack_4.6.2_Linux_JETSON_AGX_XAVIER_TARGETS/Linux_for_Tegra directory.
-
If I can’t flash by manual setup on sdkmanager, the how can I flash board on sdkm?
-
How can I install SDK? I thought I can install on sdkmanager.
-
If you want to check why sdkmanager cannot flash your board, then I need you to to dump the log from serial console and share the log…
But sorry to say that honestly I don’t think you have ability to dump the correct info I need…
So if you just want to flash the board, then just run " sudo ./flash.sh jetson-xavier mmcblk0p1"
BTW, sdkmanager is using “/flash.sh -C nv-auto-config jetson-agx-xavier-devkit mmcblk0p1” to flash your board… -
As I said… SDK could be install separately from OS… Just select the SDK you want to install in SDKM… and do not select the flash OS…
What I am talking about is “flash OS with flash.sh” → “flash done, device boots up fine” → “use SDKManager but only select to install SDK, do not install the OS again”
Thanks for explaining steps.
I tried “/flash.sh -C nv-auto-config jetson-agx-xavier-devkit mmcblk0p1” on other agx development kit.
I got
Error: Return value 8
Command tegrarcm_v2 --download bct_bootrom br_bct_BR.bct --download bct_mb1 mb1_bct_MB1_sigheader.bct.encrypt --download bct_mem mem_rcm_sigheader.bct.encrypt
Failed flashing t186ref.
error message and failed to flash.
Can you help me how to resolve this error?
Hi,
Please dump the serial console log when you run flash.sh.
If you don’t know what yourself is doing… then please just use the simplest flash.sh method to flash the board and forget about what happened from sdkmanager…
I want to get serial console log but the problem I got was I cannot find serial device of jetson.
On the link, it said I should connect to /ttyUSBn but I can’t see any USB serial devices.
This is my devices.
What usb connection are there between jetson and your host?