Random Freezes and Reboots on Jetson AGX Orin Developer Kit (64GB) Running JetPack 6.2.1
I am working with five Jetson AGX Orin Developer Kit (64GB), all running JetPack 6.2.1. The devices are configured to operate in kiosk mode, launching a Chromium instance and a Python service that creates a server and opens a camera stream (OV5640).
Kiosk mode setup:
- Flow:
- GDM Reads AccountsService selects Xsession={{kiosk_user}}
- Loads /usr/share/xsessions/{{kiosk_user}}.desktop
- Executes /usr/local/bin/{{kiosk_user}}-sessions.sh as {{kiosk_user}}
- Script set up WM + compositor + browser
- Window Manager: matchbox-window-manager
- Compositor: picom
(Both window manager and compositor have been chose to have minimal footprint, solving glitches and wrong window managing with Chromium).
All devices experience random freezes followed by reboots. This occurs under varios conditions and configurations.
Troubleshooting steps taken:
- Disconnected/reconnected peripherals individually.
- Monitored temperature, RAM, SWAP, CPU via tegrastats (values normal: RAM ~4GB/62GB, temp ~40C)
- Tested different power modes
- Ran without kiosk mode
- Re-flashed devices from scratch using NVIDIA SDK
- Disabled peripheral power-saving features
Observations:
- Setting power mode to 50W without running ‘jetson_clocks’ seemed to increase freeze frequency, but overall behaviour is inconsistent.
- Logs provide little insight; often no trace at all
- Captured one kernel panic during reboot (/sys/fs/pstore/dmesg-ramoops-1):
<1>[ 480.506125] Unable to handle kernel paging request at virtual address 000030313065c7d9
<1>[ 480.506136] Mem abort info:
<1>[ 480.506137] ESR = 0x0000000096000004
<1>[ 480.506139] EC = 0x25: DABT (current EL), IL = 32 bits
<1>[ 480.506141] SET = 0, FnV = 0
<1>[ 480.506142] EA = 0, S1PTW = 0
<1>[ 480.506142] FSC = 0x04: level 0 translation fault
<1>[ 480.506144] Data abort info:
<1>[ 480.506144] ISV = 0, ISS = 0x00000004
<1>[ 480.506145] CM = 0, WnR = 0
<1>[ 480.506146] user pgtable: 4k pages, 48-bit VAs, pgdp=000000013bbbd000
<1>[ 480.506149] [000030313065c7d9] pgd=0000000000000000, p4d=0000000000000000
<0>[ 480.506155] Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP
<6>[ 480.506159] Modules linked in: nvidia_drm(O) nvidia_modeset(O) xt_conntrack xt_MASQUERADE ip6table_nat ip6table_filter ip6_tables xt_set ip_set iptable_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 libcrc32c xt_addrtype iptable_filter nfnetlink wireguard(E) ip6_udp_tunnel udp_tunnel lzo_rle lzo_compress zram zsmalloc nvme_fabrics uvcvideo videobuf2_vmalloc ramoops reed_solomon joydev bridge stp llc usb_f_ncm usb_f_mass_storage algif_hash algif_skcipher af_alg bnep usb_f_acm u_serial usb_f_rndis u_ether libcomposite rtk_btusb(O) btusb btrtl btintel btbcm snd_soc_tegra210_admaif(O) snd_soc_tegra186_asrc(O) snd_soc_tegra_pcm snd_soc_tegra210_adx(O) snd_soc_tegra210_mixer(O) snd_soc_tegra210_sfc(O) snd_soc_tegra186_arad(O) snd_soc_tegra210_afc(O) snd_soc_tegra210_ope(O) snd_soc_tegra210_amx(O) snd_soc_tegra186_dspk(O) snd_soc_tegra210_mvc(O) snd_soc_tegra210_dmic(O) snd_soc_tegra210_i2s(O) snd_soc_tegra210_ahub(O) tegra210_adma spidev nvvrs_pseq_rtc(O)
<6>[ 480.506230] snd_soc_tegra_machine_driver(O) crct10dif_ce tegra234_oc_event(O) snd_soc_tegra_utils(O) snd_soc_simple_card_utils tegra23x_perf_uncore(O) nvpmodel_clk_cap(O) tegra_mce(O) tegra234_aon(O) nvethernet(O) mttcan(O) rtl8822ce(O) tegra_cactmon_mc_all(O) snd_soc_rt5640 nvpps(O) can_dev snd_soc_rl6231 tegra_aconnect cfg80211 snd_hda_codec_hdmi at24 pwm_tegra_tachometer(O) snd_hda_tegra host1x_fence(O) snd_hda_codec spi_tegra114 snd_hda_core mc_hwpm(O) thermal_trip_event(O) nvidia(O) nvhost_isp5(O) tegra_pcie_dma_test(O) nvhost_vi5(O) tegra_pcie_edma(O) nvhost_nvcsi_t194(O) lm90 nvidia_vrs_pseq(O) i2c_nvvrs11(O) tegra_dce(O) tegra_camera(O) v4l2_dv_timings nvhost_nvcsi(O) tegra_camera_platform(O) capture_ivc(O) tegra_camera_rtcpu(O) ivc_bus(O) hsp_mailbox_client(O) ivc_ext(O) v4l2_fwnode governor_userspace v4l2_async videobuf2_dma_contig videobuf2_memops videobuf2_v4l2 videobuf2_common tegra_drm(O) videodev nvhost_pva(O) tegra_wmark(O) nvhost_nvdla(O) cec mc nvhwpm(O) tegra_se(O)
<6>[ 480.506296] nvhost_capture(O) drm_kms_helper tsecriscv(O) host1x_nvhost(O) crypto_engine tpm_ftpm_tee nvidia_p2p(O) ina3221 nvgpu(O) governor_pod_scaling(O) host1x(O) mc_utils(O) nvmap(O) nvsciipc(O) fuse drm ip_tables x_tables ipv6 pwm_fan pwm_tegra tegra_bpmp_thermal tegra_xudc ucsi_ccg typec_ucsi typec nvme nvme_core phy_tegra194_p2u pcie_tegra194
<6>[ 480.506326] CPU: 3 PID: 1418 Comm: sshd Tainted: G OE 5.15.148-tegra #1
<6>[ 480.506329] Hardware name: NVIDIA NVIDIA Jetson AGX Orin Developer Kit/Jetson, BIOS 36.4.7-gcid-42132812 09/18/2025
<6>[ 480.506331] pstate: 604000c9 (nZCv daIF +PAN -UAO -TCO -DIT -SSBS BTYPE=–)
<6>[ 480.506334] pc : percpu_ref_get_many+0x1c/0xe0
<6>[ 480.506344] lr : percpu_ref_get_many+0x1c/0xe0
<6>[ 480.506346] sp : ffff800010a0ba20
<6>[ 480.506347] x29: ffff800010a0ba20 x28: ffff00009876be00 x27: 000030313065c7d9
<6>[ 480.506350] x26: ffff000080bfb100 x25: fffffc0000000000 x24: 0001000000000000
<6>[ 480.506353] x23: ffff000098bace00 x22: 0000000000000000 x21: 000030313065c7d9
<6>[ 480.506356] x20: 000030313065c7d9 x19: 0000000000000001 x18: 0000000000000000
<6>[ 480.506358] x17: 0000000000000000 x16: ffffbcbde7106e50 x15: 0000000000000000
<6>[ 480.506361] x14: 0000000000000001 x13: 0000000000000000 x12: 0000000000000000
<6>[ 480.506363] x11: 0000000000000000 x10: 0000000000000050 x9 : 0000000000000000
<6>[ 480.506366] x8 : ffff800010a0bc18 x7 : 0000000000000000 x6 : ffff4351bae2a000
<6>[ 480.506369] x5 : 0000000000000108 x4 : ffff000095f74000 x3 : ffffbcbde7dcb008
<6>[ 480.506371] x2 : ffffbcbde5ee3458 x1 : ffff00009876be00 x0 : 0000000000000001
<6>[ 480.506374] Call trace:
<6>[ 480.506376] percpu_ref_get_many+0x1c/0xe0
<6>[ 480.506378] refill_obj_stock+0x70/0x1b0
<6>[ 480.506381] obj_cgroup_uncharge+0x34/0x50
<6>[ 480.506385] memcg_slab_free_hook+0x94/0x170
<6>[ 480.506388] kmem_cache_free+0x264/0x310
<6>[ 480.506391] reqsk_put+0xdc/0x170
<6>[ 480.506397] inet_csk_accept+0xc0/0x2a0
<6>[ 480.506400] inet_accept+0x64/0x1b0
<6>[ 480.506403] do_accept+0x128/0x1b0
<6>[ 480.506408] __sys_accept4_file+0x78/0xd0
<6>[ 480.506412] __sys_accept4+0x6c/0xb0
<6>[ 480.506415] __arm64_sys_accept+0x30/0x50
<6>[ 480.506419] invoke_syscall+0x5c/0x150
<6>[ 480.506424] el0_svc_common.constprop.0+0x64/0x120
<6>[ 480.506428] do_el0_svc+0x74/0xb0
<6>[ 480.506431] el0_svc+0x28/0x90
<6>[ 480.506435] el0t_64_sync_handler+0xac/0x130
<6>[ 480.506438] el0t_64_sync+0x1a4/0x1a8
<0>[ 480.506443] Code: a90153f3 aa0003f4 aa0103f3 97f7fa7e (f9400280)
<4>[ 480.506447] —[ end trace 80eb9602eb3d9e81 ]—
<0>[ 481.516861] Kernel panic - not syncing:
<6>[ 481.516864] Oops: Fatal exception
<2>[ 482.521798] SMP: stopping secondary CPUs
<0>[ 482.521807] Kernel Offset: 0x3cbdddeb0000 from 0xffff800008000000
<0>[ 482.521809] PHYS_OFFSET: 0x80000000
<0>[ 482.521810] CPU features: 0x0,000005c1,a3332e5a
<0>[ 482.521813] Memory Limit: none
Additional info
- Reboots occur both with and without kiosk mode, across different Jetsons, power modes, and fresh flashes.
- Frequency varies (sometimes after 20 minutes, sometimes after 4 days).
- It happens both when Im interacting with the application running and when Im not.
- Tried ethernet and wifi configurations.
- Logs are not consuming excessive memory.
- Using NVMe for flashing; internal storage has >50% free space:
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mmcblk0p1 59302580 25968060 30308124 47% /
tmpfs 32174532 854316 31320216 3% /dev/shm
tmpfs 12869816 2652 12867164 1% /run
tmpfs 5120 4 5116 1% /run/lock
/dev/mmcblk0p10 64511 110 64401 1% /boot/efi
tmpfs 6434904 92 6434812 1% /run/user/1001
tmpfs 6434904 80 6434824 1% /run/user/1000
Example tegrastas
30W
11-17-2025 09:25:03 RAM 3795/62841MB (lfb 17x4MB) SWAP 0/31420MB (cached 0MB) CPU [1@165116@165116@165116@165116@1651165%@1651165@1728,8@172872@172@1@@1@@1,@1%@1@126728,7@126@1@1@1@1@17%@17@12678,1@126@126@126@126@126@1267%@1267,2@126@126@126@126@1267%@1267,2@126@126@126@126@1267%@1267,@1267%@1267,off,off,off,off] GR3D_FREQ 19% cpu@42.187C soc2@38.75C soc0@39.375C gpu@38.562C tj@42.187C soc1@38.781C VDD_GPU_SOC 2386mW/2386mW VDD_CPU_CV 795mW/795mW VIN_SYS_5V0 3326mW/3326mW
50W with jetson_clocks
11-17-2025 11:25:18 RAM 3310/62841MB (lfb 16x4MB) SWAP 0/31420MB (cached 0MB) CPU [2%@1497,0%@1497,0%@1497,0%@1497,69%@1497,18%@1497,1%@1497,0%@1497,13%@1497,0%@1497,0%@1497,3%@1497] GR3D_FREQ 0% cpu@44.781C soc2@41.375C soc0@41.687C gpu@39.75C tj@44.781C soc1@41.125C VDD_GPU_SOC 4008mW/4008mW VDD_CPU_CV 1604mW/1604mW VIN_SYS_5V0 3962mW/3918mW
Despite extensive troubleshooting the issue persists. Any guidance or diagnositc steps would be greatly appreaciated.
A couple of additional questions from what I have seen so far:
- Is there any document relationship or dependency between setting a specific power mode and running jetson_clocks? Could not running jetson_clocks after chaning power mode lead to instability or kernel panics as i thought?
- Are there any recommend configurations or known issues when continously streaming from a usb camera using Python? Could this workload trigger memor or kernel-related issues under JetPack 6.2.1?
Final additional info
- chromium version (142.0.7444.59 snap): It could not be the latest versions because it did not work on jetson (apparently this is a known issue on the forums) and also it is only software accelerated…
- wireguard: Had to install it from https://github.com/MrVasquez96/wireguard-linux-compat.git since the default installation didnt work for this kernel.
- I am running DL model with onnx (for cross-platform I prefer rather than tensorrt). I need to use 50W powermode because otherwise I run into another problem ‘Free: invalid pointer (aborted dump error)’, wich I have also found common in forums. Therefore, I need the kernel to be stable in 50W mode (or even max mode).
Thank you