Flashing a Orin NX 8GB?

[AustinSchuh]

I've been trying to adapt the p3509-a02-p3767-0000 config to flash an Orin NX 8GB module rather than the 16 GB module it appears designed for. I'm currently getting the following error occasionally over the serial console:

��70016FW Boot Done
I> Binary@ device:3/0 block-24832 (partition size: 0x380000), name: A_cpu-bootloader
��
��I> Partition name: A_secure-os
I> Size of partition: 4194304
I> Binary@ device:3/0 block-32000 (partition size: 0x400000), name: A_secure-os
I> MB2-params @ 0x40060000
I> cpubl: Authentication Finalize Done
I> Binary cpubl loaded successfully at 0x268200000
I> tos: Authentication Finalize Done
I> Binary tos loaded successfully at 0x277dfb000
I> Relocating OP-TEE dtb from: 0x277eff770 to 0x268630000, size: 0x1976
I> [0] START: 0x80000000, SIZE: 0x1e8670000
I> [1] START: 0x26e000000, SIZE: 0x2000000
I> Setting NS memory ranges to OP-TEE dtb finished.
I> Partition name: A_eks
I> Size of partition: 262144
I> Binary@ device:3/0 block-44288 (partition size: 0x40000), name: A_eks
E> Heap free list corrupted !!!

The following patch (I didn't make a new config, just hacked up the current one for now) is how far I've gotten.

diff --git a/conf/machine/p3509-a02-p3767-0000.conf b/conf/machine/p3509-a02-p3767-0000.conf
index 66beda93..769f6e91 100644
--- a/conf/machine/p3509-a02-p3767-0000.conf
+++ b/conf/machine/p3509-a02-p3767-0000.conf
@@ -3,16 +3,16 @@
 #@DESCRIPTION: Nvidia Jetson Orin NX module in P3509 carrier
 
 TEGRA_BOARDID ?= "3767"
-TEGRA_FAB ?= "ES1"
-TEGRA_BOARDSKU ?= "0000"
-TEGRA_BOARDREV ?= "A.3"
+TEGRA_FAB ?= "300"
+TEGRA_BOARDSKU ?= "0001"
+TEGRA_BOARDREV ?= "N.2"
 TEGRA_CHIPREV ?= "1"
-NVPMODEL ?= "nvpmodel_p3767_0000"
+NVPMODEL ?= "nvpmodel_p3767_0001"
 NVFANCONTROL ?= "nvfancontrol_p3767_0000"
 
-TNSPEC_BOOTDEV ?= "nvme0n1p1"
+TNSPEC_BOOTDEV_DEFAULT ?= "nvme0n1p1"
 
-TEGRAFLASH_NO_INTERNAL_STORAGE = "1"
+#TEGRAFLASH_NO_INTERNAL_STORAGE = "0"
 
 require conf/machine/include/tegra234.inc
 require conf/machine/include/devkit-wifi.inc
@@ -28,7 +28,7 @@ TEGRA_BUPGEN_SPECS ?= "fab=000;boardsku=0000;boardrev=;bup_type=bl \
                        fab=000;boardsku=0005;boardrev=;bup_type=bl \
                        fab=000;boardsku=0005;boardrev=;bup_type=kernel"
 
-KERNEL_DEVICETREE ?= "tegra234-p3767-0000-p3509-a02.dtb"
+KERNEL_DEVICETREE ?= "tegra234-p3767-0001-p3509-a02.dtb"
 KERNEL_ARGS ?= "mminit_loglevel=4 console=ttyTCU0,115200 firmware_class.path=/etc/firmware fbcon=map:0 net.ifnames=0"
 MACHINE_EXTRA_RRECOMMENDS += "kernel-module-r8168 kernel-module-r8169 kernel-module-realtek"
 MACHINE_EXTRA_RDEPENDS += "linux-firmware-rtl8168"
@@ -38,12 +38,12 @@ BOOTPART_SIZE ?= "8388608"
 # 55GiB default rootfs size
 ROOTFSPART_SIZE ?= "59055800320"
 ODMDATA ?= "gbe-uphy-config-8,hsstp-lane-map-3,hsio-uphy-config-0"
-EMMC_BCT ?= "tegra234-p3767-0000-sdram-l4t.dts"
+EMMC_BCT ?= "tegra234-p3767-0001-sdram-l4t.dts"
 NVIDIA_BOARD ?= "t186ref"
-PARTITION_LAYOUT_TEMPLATE ?= "flash_t234_qspi_sd.xml"
+PARTITION_LAYOUT_TEMPLATE ?= "flash_t234_qspi_nvme.xml"
 TEGRA_AUDIO_DEVICE ?= "tegra-hda-p3767-p3509"
 
 OTABOOTDEV ?= "/dev/mtdblock0"
 OTAGPTDEV ?= "/dev/mtdblock0"
 
-TEGRA_PLUGIN_MANAGER_OVERLAYS ?= "tegra234-p3767-overlay.dtbo tegra234-p3767-0000-p3509-a02-hdr40.dtbo"
+TEGRA_PLUGIN_MANAGER_OVERLAYS ?= "tegra234-p3767-overlay.dtbo"
diff --git a/recipes-bsp/tegra-binaries/tegra-flashvars/p3509-a02-p3767-0000/flashvars b/recipes-bsp/tegra-binaries/tegra-flashvars/p3509-a02-p3767-0000/flashvars
index 087d5fde..cf485f85 100644
--- a/recipes-bsp/tegra-binaries/tegra-flashvars/p3509-a02-p3767-0000/flashvars
+++ b/recipes-bsp/tegra-binaries/tegra-flashvars/p3509-a02-p3767-0000/flashvars
@@ -1,6 +1,6 @@
 FLASHVARS="PINMUX_CONFIG BPFDTB_FILE DEVICE_CONFIG GPIOINT_CONFIG PMIC_CONFIG PMC_CONFIG DEVICEPROD_CONFIG DEV_PARAMS DEV_PARAMS_B PROD_CONFIG BR_CMD_CONFIG MB2BCT_CFG MINRATCHET_CONFIG UPHY_CONFIG BPF_FILE WB0SDRAM_BCT OVERLAY_DTB_FILE SCR_CONFIG EMC_FUSE_DEV_PARAMS"
 PINMUX_CONFIG="tegra234-mb1-bct-pinmux-p3767-hdmi-a03.dtsi"
-BPFDTB_FILE="tegra234-bpmp-3767-0000-@[email protected]"
+BPFDTB_FILE="tegra234-bpmp-3767-0001-3509-a02.dtb"
 DEVICE_CONFIG="tegra234-mb1-bct-device-p3767-0000.dts"
 GPIOINT_CONFIG="tegra234-mb1-bct-gpioint-p3767-0000.dts"
 PMIC_CONFIG="tegra234-mb1-bct-pmic-p3767-0000-@[email protected]"
@@ -15,7 +15,7 @@ MISC_CONFIG="tegra234-mb1-bct-misc-p3767-0000.dts"
 MINRATCHET_CONFIG="--minratchet_config tegra234-mb1-bct-ratchet-p3767-0000.dts"
 UPHY_CONFIG="--uphy tegra234-mb1-bct-uphylane-si.dtsi"
 BPF_FILE="bpmp_t234-TE990M-A1_prod.bin"
-WB0SDRAM_BCT="tegra234-p3767-0000-wb0sdram-l4t.dts"
+WB0SDRAM_BCT="tegra234-p3767-0001-wb0sdram-l4t.dts"
 OVERLAY_DTB_FILE="@OVERLAY_DTB_FILE@"
 SCR_CONFIG="tegra234-mb2-bct-scr-p3767-0000.dts"
 EMC_FUSE_DEV_PARAMS="tegra234-br-bct-diag-boot.dts"

Any ideas on what I'm missing, or strategies to debug this? I've been running both the NVIDIA stock flashing tools and doflash.sh and reading through the outputs and intermediates trying to find differences.

(For reference, I'm trying to flash https://wiki.seeedstudio.com/reComputer_J4012_Flash_Jetpack/ with an 8GB module)

[joekale-pp]

This has been an issue both @madisongh and I have seen. I saw it on the Orin Nanos. So far we haven’t had time to root cause it but the work around is to unplug for 30 seconds or so and it should clear this error and get you back to booting.

As far as root causing, the only correlation I have seen (and it’s not strong correlation), is that it typically happens after I have used nvpmodel or the nvfancontrol. If that doesn’t apply to you then we will have to discuss more how to correct it. I have not yet been able to recreate it when using JetPack to flash.

[joekale-pp]

For my learning/understanding, how does the “external device mechanism” differ from how meta-tegra?

[madisongh]

It's described on the initrd flashing wiki page.

[AustinSchuh]

Any ideas on how to test this theory out with minimal effort? My best idea is to do something like flash it once with the current script, then delete the nvme part of the XML file and flash it again, but I'm not sure if that will end up doing something different.

[madisongh]

Use these settings

PARTITION_LAYOUT_TEMPLATE = "flash_t234_qspi.xml"
PARTITION_LAYOUT_EXTERNAL = "flash_l4t_external.xml"

and use initrd-flash for flashing. I've been trying some tests with this combination on my p3767-0000 module in the p3768-0000 carrier, and after many flashings and reboots, I haven't run into that Heap free list corrupted !! error.

I'll try and run through some more tests and will probably switch to using this combination as the default setup for all of the Nano and NX modules going forward.

[AustinSchuh]

Oh, excellent, thanks! That worked pretty well for me with some tweaks. I used to fail 2-5 times per successful boot. I haven't seen a failure yet.

FYI, initrd-flash appears to not handle sparse images well. I had to shrink my APP partition to 4 GB to make it flash in a reasonable amount of time. It appeared to be trying to flash every byte.

Also, FYI, the initrd image doesn't have the pwm_fan module in it, so the fan stays off for programming. I added CONFIG_SENSORS_PWM_FAN=y to one of the files in recipes-kernel/linux/linux-tegra-5.10/ and it now programs with the fan on.

I'm happy to file real bugs if that is helpful, or convert the fan change to a patch if that is also helpful.

[madisongh]

Glad it's working better for you. As mentioned on the wiki page, if you have bmaptool installed on your host system, the script will use that instead of dd to write the partitions. That should speed up the flashing process significantly, since bmaptool does handle sparse files.

The fan driver shouldn't really be necessary during flashing, at least for the stock dev kits. If you're seeing over-temp warnings, or worse, thermal shutdowns, during flashing, then building the driver into your kernel, or adding kernel-module-pwm-fan to the tegra-initrd-flash-initramfs image for your builds is a good idea.