Thank you very much Dr. Chubb! Yes I think you are right. I got access to their FTP and did the same thing now I can boot original Linux. Will flash the U-boot so I can boot seL4. Thank you for confirming it, and thank you for testing out for me.

Best Regards
-Daniel Wang



On Oct 11, 2018, at 7:10 PM, <Peter.Chubb@data61.csiro.au> <Peter.Chubb@data61.csiro.au> wrote:

Hi Daniel,
  There must be something else funny about your system.  I just took
  a CEI-TK1_SOM, flashed it with Colorado's images from their FTP
  site (if you don't already have access to these, you should ask
  Colorado Engineering --- they give it freely to customers), and
  checked it could boot.

  I then flashed the current upstream U-Boot, and attempted to boot,
  with bootm_boot_mode set to `sec' -- and  Linux Ubuntu booted.
  I then set the variable to `nonsec' and see:
...
 Jetson-TK1 eMMC boot options                                                    
    Enter choice: 1                                                                 
 1:      primary kernel                                                          
 Retrieving file: /boot/zImage                                                   
 5995744 bytes read in 156 ms (36.7 MiB/s)                                       
 append: console=ttyS0,115200n8 console=tty1 no_console_suspend=1 lp0_vec=2064@0t
 Retrieving file: /boot/tegra124-tk1-som-pm375-000-c00-00.dtb                    
 59720 bytes read in 11 ms (5.2 MiB/s)                                           
 ## Flattened Device Tree blob at 82000000                                       
    Booting using the fdt blob at 0x82000000                                     
    Using Device Tree in place at 82000000, end 82011947                         
 couldn't find /cpus                                                             
 ERROR: arch-specific fdt fixup failed                                           
   - must RESET the board to recover.                                             
...
which is what I expect.


My guess is that something on the eMMC has become corrupted, and you
may be best off reflashing the entire board using Colorado's image.
--
Dr Peter Chubb         Tel: +61 2 9490 5852      http://ts.data61.csiro.au/
Trustworthy Systems Group                     Data61, CSIRO (formerly NICTA)