How to disable DRAM ECC when flashing an image on the NVIDIA Jetson AGX Xavier/Orin Industrial?

[JoostBaars]

In the config file: tegra194-mb1-bct-misc-flash-jaxi.cfg I found settings regarding the ECC (carveout.misc.ecc_protected, carveout.os.ecc_protected, carveout.cpubl.ecc_protected, etc..) but disabling them causes the following flashing errors:
flash PC side:

[   4.0073 ] Sending bct_bootrom
[   4.0077 ] [................................................] 100%
[   4.0088 ] Sending bct_mb1
[   4.0136 ] [................................................] 100%
[   4.0179 ] Sending bct_mem
[   4.0659 ] [................................................] 100%
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

Jetson serial:

[0004.411] W> RATCHET: MB1 binary ratchet value 4 is larger than ratchet level 2 from HW fuses.
[0004.419] I> MB1 (prd-version: 2.6.0.0-t194-41334769-cab45716)
[0004.425] I> Boot-mode: RCM
[0004.427] I> Platform: Silicon
[0004.430] I> Chip revision : A02P
[0004.433] I> Bootrom patch version : 15 (correctly patched)
[0004.438] I> ATE fuse revision : 0x200
[0004.442] I> Ram repair fuse : 0x0
[0004.445] I> Ram Code : 0x1
[0004.447] I> rst_source: 0xb, rst_level: 0x1
[0004.453] I> USB configuration success
[0006.444] I> bct_bootrom image downloaded
[0006.453] W> PROD_CONFIG: device prod data is empty in MB1 BCT.
[0006.461] I> Temperature = 30500
[0006.464] W> Skipping boost for clk: BPMP_CPU_NIC
[0006.468] W> Skipping boost for clk: BPMP_APB
[0006.472] W> Skipping boost for clk: AXI_CBB
[0006.476] W> Skipping boost for clk: AON_CPU_NIC
[0006.480] W> Skipping boost for clk: CAN1
[0006.484] W> Skipping boost for clk: CAN2
[0006.488] I> Boot-device: QSPI (instance: 0)
[0006.492] I> Qspi flash params source = mb1bct
[0006.496] I> bct_mb1 image downloaded
[0006.507] I> ECC region[0]: Start:0x80000000, End:0x100000000
[0006.515] E> CARVEOUT: Failed to allocate memory of size 0x20000000 for CO:38.
[0006.522] C> Task 0x42 failed (err: 0x1f1f0603)
[0006.526] E> Top caller module: CARVEOUT, error module: CARVEOUT, reason: 0x03, aux_info: 0x06
[0006.534] C> Error: 0x1f1f0603
[0006.537] C> MB1(2.6.0.0-t194-41334769-cab45716) BIT boot status dump :
0000000000111111111001000000000001111011111100000011111111111111000000000000000000000000000000000000000000000000000000000000000010100000000001010000000000000101000000000010100000000000000000000000000000000000000000000000000000000000000000000000000000000000
[0006.567] I> Reset to recovery mode

And NVIDIA mentioned the use of a different config on their forums: https://forums.developer.nvidia.com/t/disabling-dram-ecc-on-jaxi/280970/3, but I tried to use the following config file for the BPFDTB: BPFDTB_FILE="tegra194-a02-bpmp-p2888-0008-b01-jaxi-noecc_lz4.dtb" which didn't seem to have any effect.

I want to disable DRAM ECC to keep the memory contents after a reboot (while power stays available) so we can use ramoops to log kernel errors after a reboot.

[madisongh]

Can you successfully flash using the stock L4T/JetPack kit and their jetson-agx-xavier-ind-noecc.conf configuration? That's what that dev forum post recommended (although they mentioned it as p2822-0000+p2888-0008-noecc.conf, which is the same file).

If that does work, then from comparing that file against the normal (ECC) L4T config file for JAXi, I see that besides the BPFDTB file, the kernel (and bootloader) DTB file is also different, the SCEFILE setting is removed, and the EMMC_BCT, MISC_CONFIG, and MISC_COLD_BOOT_CONFIG files go back to normal for the AGX Xavier. It also looks like they dropped one of the DTB overlays. You'd need to make corresponding changes to your machine config/flashvars settings.