英伟达团队您好，请教个问题，使用rt内核情况下，以下这个内核参数配置，如果高负载情况下，会很容易引起rcu吗？

Robotics & Edge Computing Jetson Systems Jetson Thor
, ,
1
  1. 内核启动参数配置：rcu_nocb_poll isolcpus=3,8 nohz=on rcu_nocbs=1,2,3,4,5,6,7,8,9,10,11 nohz_full=1,2,3,4,5,6,7,8,9,10,11
  2. rcu的日志：[ 29.665309] debugfs: Directory ‘null’ with parent ‘/’ alrea!
    [ 29.665457] cnss: Failed to get cooling device node
    [ 29.688436] [Networ][0x1c4fa11][07:33:54.133637] wlan: [1727:E:MLO_MGR] mlo_d
    [ 29.688522] [Networ][0x1c4fa68][07:33:54.133724] wlan: [1727:E:PMO] pmo_regi0
    [ 29.691163] [Networ][0x1c504b8][07:33:54.136364] wlan: [1727:E:PMO] pmo_regi1
    [ 29.694199] WARNING: CPU: 11 PID: 1727 at kernel/softirq.c:210 __local_bh_enc
    [ 29.694404] —[ end trace 0000000000000000 ]—
    [ 29.694409] WARNING: CPU: 11 PID: 1727 at kernel/irq/irqdesc.c:673 handle_ir8
    [ 29.694568] WARNING: CPU: 11 PID: 1727 at drivers/irqchip/irq-gic-v3.c:790 gc
    [ 29.694715] —[ end trace 0000000000000000 ]—
    [ 29.694724] WARNING: CPU: 11 PID: 1727 at kernel/softirq.c:148 __local_bh_di0
    [ 29.694850] —[ end trace 0000000000000000 ]—
    [ 29.777508] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.777863] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.785790] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.790066] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.790088] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.790097] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.790105] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.791114] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.791240] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 29.791535] NOHZ tick-stop error: local softirq work is pending, handler #80!
    [ 50.702151] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
    [ 50.702154] rcu: Tasks blocked on level-0 rcu_node (CPUs 0-13): P1727/2:l
    [ 50.702160] rcu: (detected by 12, t=5252 jiffies, g=2989, q=7793 ncpus=1)
    [ 50.702840] rcu: rcu_preempt kthread timer wakeup didn’t happen for 2764 jif2
    [ 50.702842] rcu: Possible timer handling issue on cpu=0 timer-softirq=529
    [ 50.702843] rcu: rcu_preempt kthread starved for 2765 jiffies! g2989 f0x0 RC0
    [ 50.702844] rcu: Unless rcu_preempt kthread gets sufficient CPU time, OO.
    [ 50.702844] rcu: RCU grace-period kthread stack dump:
    [ 50.702869] NMI backtrace for cpu 0CU GP kthread last ran:
    [ 113.722150] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
    [ 113.722153] rcu: Tasks blocked on level-0 rcu_node (CPUs 0-13): P1727/2:l
    [ 113.722157] rcu: (detected by 13, t=21007 jiffies, g=2989, q=8117 ncpus=)
    [ 176.742154] rcu: INFOTasks blocked on level-0 rcu_node (CPUs 0-13): P1727/2:l
    [ 176.742158] rcu: (detected by 11, t=36762 jiffies, g=2989, q=8197 ncpus=)
    [ 239.762150] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
    [ 239.762153] rcu: Tasks blocked on level-0 rcu_node (CPUs 0-13): P1727/2:l
    [ 242.654158] INFO: task rcub/0:21 blocked for more than 120 seconds.64 ncpus=)
    [ 242.654161] Tainted: G W O 6.8.12-rt-tegra #1
    [ 242.654162] “echo 0 > /proc/sys/kernel/hung_task_timeout_secs” disables this.
    [ 242.654225] INFO: task kworker/u30:3:629 blocked for more than 120 seconds.
    [ 242.654227] Tainted: G W O 6.8.12-rt-tegra #1
    [ 242.654228] “echo 0 > /proc/sys/kernel/hung_task_timeout_secs” disables this.
    [ 242.654273] INFO: task systemd-network:983 blocked for more than 120 seconds.
    [ 242.654274] Tainted: G W O 6.8.12-rt-tegra #1
    [ 242.654275] “echo 0 > /proc/sys/kernel/hung_task_timeout_secs” disables this.
    [ 242.654326] INFO: task cups-browsed:2661 blocked for more than 120 seconds.
    [ 242.654326] Tainted: G W O 6.8.12-rt-tegra #1
    [ 242.654327] “echo 0 > /proc/sys/kernel/hung_task_timeout_secs” disables this.
4

Hi,
Do you use Jetpack 7.1 r38.4? Would like to make sure you use latest release for Jetson Thor.

5

yes, jetpack 7.1 r38.4

agi@localhost:~$ cat /etc/nv_tegra_release

R38 (release), REVISION: 4.0, GCID: 43443517, BOARD: generic, EABI: aarch64, DATE: Wed Dec 31 00:15:19 UTC 2025

KERNEL_VARIANT: oot

TARGET_USERSPACE_LIB_DIR=nvidia
TARGET_USERSPACE_LIB_DIR_PATH=usr/lib/aarch64-linux-gnu/nvidia
INSTALL_TYPE=openrm

6

Hi,
Please ensure you applied this to rebuild RT kernel:
Jetson AGX Thor FAQ

If the issue persists, please share the steps to replicate it on AGX Thor developer kit. So that we can set up and check.