how to avoid CPU watchdog error during debug
Hello, While debugging a kernel module, I get this error and the OS crashes. [39068.833652] rcu: INFO: rcu_sched self-detected stall on CPU [39068.834739] watchdog: BUG: soft lockup - CPU#3 stuck for 164s! [systemd-journal:399] [39068.834854] watchdog: BUG: soft lockup - CPU#0 stuck for 164s! [test_axpu_app:2896] This seems because I stopped the processor for debugging. How can I avoid this watchdog error during debug? Any help will be appreciated. Thanks! Chan Kim
On 8/25/21 3:06 AM, ckim@etri.re.kr wrote:
Hello,
While debugging a kernel module, I get this error and the OS crashes.
[39068.833652] rcu: INFO: rcu_sched self-detected stall on CPU
[39068.834739] watchdog: BUG: soft lockup - CPU#3 stuck for 164s! [systemd-journal:399]
[39068.834854] watchdog: BUG: soft lockup - CPU#0 stuck for 164s! [test_axpu_app:2896]
This seems because I stopped the processor for debugging.
How can I avoid this watchdog error during debug?
Any help will be appreciated. Thanks!
Chan Kim
_______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Hi Kim, since many things are time depending in the kernel it is difficult to stop parts of the kernel. The question is if it is really absolutely necessary to do this? Why you can not do it by logging with printk? Debugging is described in: https://www.doc-developpement-durable.org/file/Projets-informatiques/cours-&... See Chapter 18 Debugging Thanks! Bye Philipp
since many things are time depending in the kernel it is difficult to stop parts of the kernel. The question is if it is really absolutely necessary to do this? Why you can not do it by logging with printk?
Debugging is described in: https://www.doc-developpement-durable.org/file/Projets- informatiques/cours-&-manuels- informatiques/Linux/Linux%20Kernel%20Development,%203rd%20Edition.pdf
See Chapter 18 Debugging
Thanks! Bye Philipp
Hi, Philipp, I've found your answer too late. Thanks for the info(I learned something about the early_prink from the book). Actually, shortly after I sent my question, I learned how to make printk appear in the console (not in the dmesg buffer) and was using it. (I added the line "kernel.prink = 5 4 1 7" in the /etc/sysctl.conf file. So I've set the default console log level to 5 making default printk appear in the console.) Thanks! Chan
_______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
On Sun, Aug 29, 2021 at 7:46 PM <ckim@etri.re.kr> wrote:
since many things are time depending in the kernel it is difficult to stop parts of the kernel. The question is if it is really absolutely necessary to do this? Why you can not do it by logging with printk?
Debugging is described in: https://www.doc-developpement-durable.org/file/Projets- informatiques/cours-&-manuels- informatiques/Linux/Linux%20Kernel%20Development,%203rd%20Edition.pdf
See Chapter 18 Debugging
Thanks! Bye Philipp
Hi, Philipp, I've found your answer too late. Thanks for the info(I learned something about the early_prink from the book). Actually, shortly after I sent my question, I learned how to make printk appear in the console (not in the dmesg buffer) and was using it. (I added the line "kernel.prink = 5 4 1 7" in the /etc/sysctl.conf file. So I've set the default console log level to 5 making default printk appear in the console.) Thanks!
Chan
thanks for the followup
participants (3)
-
ckim@etri.re.kr -
jim.cromie@gmail.com -
Philipp Hortmann