We use a Xavier NX 8GB production module with a custom board and want to flash to ssd and to boot directly from the ssd. To make things easier in this request I only use the defaults for the dev kit not special dtb, cfg etc. for our custom board. We’ve tested this with L4T 34.1.1.
When we use initrd_flash with following command and WITHOUT compiling the kernel (all to default for DevKit) everthing works as expected.
sudo ./tools/kernel_flash/l4t_initrd_flash.sh --external-device nvme0n1p1 -c ./tools/kernel_flash/flash_l4t_external.xml --showlogs jetson-xavier-nx-devkit-emmc nvme0n1p1
But now to the problem: When we compile the linux kernel (even with all in default) and copy all files as described in developer guide the same procedure fails.
This is the error we get:
With serial console (uart) from Xavier NX we get this:
[0062.898] I> Reset to recovery mode
[0280.468] W> RATCHET: MB1 binary ratchet value 4 is larger than ratchet level 2 from HW fuses.
[0280.476] I> MB1 (prd-version: 188.8.131.52-t194-41334769-3540ffaa)
[0280.482] I> Boot-mode: RCM
[0280.484] I> Platform: Silicon
[0280.487] I> Chip revision : A02P
[0280.490] I> Bootrom patch version : 15 (correctly patched)
[0280.496] I> ATE fuse revision : 0x200
[0280.499] I> Ram repair fuse : 0x0
[0280.502] I> Ram Code : 0x0
[0280.505] I> rst_source: 0xb, rst_level: 0x1
[0280.510] I> USB configuration success
[0282.652] I> bct_bootrom image downloaded
[0282.707] W> PROD_CONFIG: device prod data is empty in MB1 BCT.
[0282.715] I> Temperature = 30000
[0282.718] W> Skipping boost for clk: BPMP_CPU_NIC
[0282.722] W> Skipping boost for clk: BPMP_APB
[0282.726] W> Skipping boost for clk: AXI_CBB
[0282.730] W> Skipping boost for clk: AON_CPU_NIC
[0282.734] W> Skipping boost for clk: CAN1
[0282.738] W> Skipping boost for clk: CAN2
[0282.742] I> Boot-device: QSPI (instance: 0)
[0282.746] I> Qspi flash params source = mb1bct
[0282.750] I> bct_mb1 image downloaded
[0282.804] I> Non-ECC region: Start:0x80000000, End:0x100000000
[0282.812] W> Thermal config not found in BCT
[0282.820] W> MEMIO rail config not found in BCT
[0282.830] I> bct_mem image downloaded
[0283.523] E> RCM carveout size is smaller than blob size:0x40e1dcb
[0283.529] E> NV3P_SERVER: Failed to get load address for image blob from nv3p helper.
What could be the problem here? Is it because RCM carveout size is to small, but why? And how to increase this?
As we have found the “blob_boot0.img” has increased in size from 55031808 bytes to 61487104 bytes (marked in red). I don’t know why!? Maybe this could be a problem. So the questions here is why this size has increased (Kernel ‘Image’ files are almost same in size, what else is included in blob_boot0.img?). If this is okay , how could we handle with that.
This is the log at this point from flashing without compiling before:
Attached you find the complete logs for flashing with and without compiling kernel.
Thank you for any help in advance.