How to reboot to forced-recovery from UEFI shell?

[tbottom]

Hello

I am working on a remotely-located xavier devkit and have just upgraded the platform to use the new UEFI bootloader.

A change I was testing is causing a kernel panic on boot, resulting in infinite loop. I have a serial port attached and have been trying to figure out how to reboot the jetson in forced recovery mode from here.

I tried setting the value of L4TDefaultBootMode to 0x3 by performing the below steps based on the documentation here: VariableList · NVIDIA/edk2-nvidia Wiki (github.com) but it is not behaving as I had expected it to.

I eventually got someone to physically reset the devkit to recovery mode without issue, but it would be very helpful to do this without the need for physical intervention. Is there a way to force recovery mode from UEFI shell?

Shell> dmpstore -all L4TDefaultBootMode
Variable NV+RT+BS '781E084C-A330-417C-B678-38E696380CB9:L4TDefaultBootMode' Data
Size = 0x04
  00000000: FF 00 00 00  
  
Shell> setvar -guid 781E084C-A330-417C-B678-38E696380CB9 L4TDefaultBootMode =0x3

Shell> setvar -guid 781E084C-A330-417C-B678-38E696380CB9 L4TDefaultBootMode
781E084C-A330-417C-B678-38E696380CB9 - L4TDefaultBootMode - 0001 Bytes
03

Shell> reset

...

Jetson UEFI firmware (version 1.0-d7fb19b built on 2022-08-10T20:18:13-07:00)
Press ESCAPE for boot options **  WARNING: Test Key is used.  **
......
      L4TLauncher: Attempting Recovery Boot
Header not seen at either offset 0 or offset 0x1000
Failed to boot recovery:0 partition

[kayccc]

Sorry for the late response, is this still an issue to support? Thanks

[tbottom]

Hello,

Yes, I am still looking for a solution for this. I have tried a few more things:

I dug around in the EDK source see now that setting the L4TDefaultBootMode to 0x3 is trying to boot a recovery kernel partition that is not part of the yocto meta-tegra linux images.

Based on how the “forced-recovery” reboot argument is handled in the L4T code at drivers/soc/tegra/pmc.c I tried writing to TEGRA_PMC_SCRATCH0 register from uefi but this resulted in an exception and the bootloader resetting.

From device tree:
PMC base address = 0xc360000

offset from pmc.c:
[TEGRA_PMC_SCRATCH0] = 0x2000,

Exception trying to read:

Shell> mm 0xC362000 -w 4 -n
MEM

Synchronous Exception at 0x0000000857C88874

I created a related issue on the edk2-nvidia github: failing boot retry count requires hard flash · Issue #22 · NVIDIA/edk2-nvidia (github.com) in which I attempt to either reset the reboot retry count or enter forced recovery mode from the UEFI shell but have not been able to find a way to.

Any assistance greatly appreciated.

[kayccc]

We’re check with our internal team to see if can provide suggestions.

[WayneWWW]

Hi,

Currently there is no support for this.