After umount /sys/kernel/debug, jetson_clocks command error

after umount /sys/kernel/debug,jetson_clocks command error.error info:
cat: /sys/kernel/debug/bpmp/debug/clk/emc/max_rate: No such file or directory
/usr/bin/jetson_clocks: line 494: /sys/kernel/debug/bpmp/debug/clk/emc/mrq_rate_locked: No such file or directory
/usr/bin/jetson_clocks: line 495: /sys/kernel/debug/bpmp/debug/clk/emc/rate: No such file or directory

question1:Can the debug file system be uninstalled after system startup? If not, what are the impacts?
question2:After uninstalling the debug file system after system startup, it is necessary to enable maximum power consumption mode and CPU frequency locking. Does this error affect the performance?

jetson_clocks就是用debugfs調clock rate 你unmount debugfs當然跑不起來…
不懂你為什麼要這樣操作

Just so you know, /sys is not on any disk. This is a result of kernel drivers pretending to be files for various I/O purposes, e.g., some drivers accept commands to do things. You could have actual files of those names, but they would not do anything because they would not be the drivers. Any umount of this is essentially cutting the I/O path for control or monitoring of drivers.

Why would you want to remove and then add the /sys files?

Due to the kernel opening debugfs, exposing memory addresses, being attacked by DOS, and posing security issues, it was closed.
After uninstalling the debug file system after system startup, it is necessary to enable maximum power consumption mode and CPU frequency locking. Does this error affect the performance?

你可不可以說中文就好… 我看不懂你用英文想表達什麼

這句話在說什麼?

系统启动后卸载debug调试文件系统后,需要启用最大功耗模式和CPU频率锁定。这个错误会影响性能吗?错误如下:
执行jetson_clocks报错:
cat: /sys/kernel/debug/bpmp/debug/clk/emc/max_rate: No such file or directory
/usr/bin/jetson_clocks: line 494: /sys/kernel/debug/bpmp/debug/clk/emc/mrq_rate_locked: No such file or directory
/usr/bin/jetson_clocks: line 495: /sys/kernel/debug/bpmp/debug/clk/emc/rate: No such file or directory

If you have some sort of DOS going on, then this is the wrong way to deal with it. You’re essentially lobotomizing various parts of the kernel from the outside world in such a way that talking to those parts of the kernel is no longer possible. This is not a bug, it did as you told it to do. Imagine your car is expected to be stolen, so you remove the transmission to prevent it from being driven away…then you want to get in your car and drive away and wonder why the car won’t move. Wouldn’t it be better to lock the doors or add a car alarm?

1 Like

你現在就沒有debugfs了當然開不了… 是要怎麼鎖定CPU頻率
還是我說中文你也看不懂?
麻煩不要做一些我們文件上沒有提到的奇怪操作 報錯了再來問說這樣子為什麼不會動…

麻烦你不懂的话请不要回复,给自己留点尊严,每次回复的都和问题无关。请尊重每个英伟达用户提出来的问题,这些问题不是凭空而来的,是用户需求产生而来的。

你可以開機之後先開jetson_clocks再把debugfs關掉 這樣應該還是會生效的
寫個system service之類的可以做到
那個錯誤就是說你現在調不了頻率而已 影響就是沒辦法像正常用jetson_clocks一樣讓機器永遠跑在最高頻
這樣應該代表我懂?

/sys/kernel/debug卸载或内核关闭的需求是我们安全团队提出来的, /sys/kernel/debug会暴露寄存器地址。
如你所说,agx orin使用jetson_clocks命令锁频必须依赖/sys/kernel/debug。同时兼顾安全问题,采用“你可以開機之後先開jetson_clocks再把debugfs關掉 這樣應該還是會生效的”这种方法会不会影响agx orin最佳性能。
我们将nvpmode开启最大,jetson_clocks执行锁频,开启压测测试程序(CPU、内存、GPU的使用率都超过90%),发现压测过程中,cpu的频率不会一直保持在2188000HZ上,会出现低于1957000HZ情况,请问你给出的方案能否解决压测过程中降频问题。

請問你們有測試過開著debugfs跑壓力測試會不會也有降頻的問題嗎?
如果會的話就跟debugfs無關 建議你可以另外開一個topic問
如果不會的話我需要內部確認

還有請問一下你的BSP版本 還有你的AGX Orin是哪一個SKU

1、agx orin jetpack5.0.2,且未卸载 /sys/kernel/debug/。设置最大功耗模式,设置锁屏,采用stress-ng和gpu_burn压测,cpu主频稳定在2188000HZ未出现降频。
2、agx orin jetpack5.1.2,且未卸载 /sys/kernel/debug/。设置最大功耗模式,设置锁屏,正在测试
3、agx orin jetpack5.1.2,卸载 /sys/kernel/debug/。设置最大功耗模式,设置锁屏,CPU主频降频,降频记录如下
image
版本信息如下:
image

1 Like

你可以檢查一下降頻的時候dmesg有沒有throttling或是over current的問題
如果有的話那就是正常的 跟debugfs沒有關係

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.