Hi,
This seems newish, maybe 4.16 debug kernels were doing it also, but I've seen if for all of 4.17. But what's new, I think, with 4.17.0-0.rc5.git1.1.fc29.x86_64 is I'm seeing a 4 kidle_inject threads each using around 20% CPU per top, and the everything is sluggish, the mouse arrow jerks around on screen, and apps aren't responsive.
I'm not sure where to start with troubleshooting; my first thought is to just wait for rc6 and see if it's still happening.
[ 698.266514] intel_powerclamp: Start idle injection to reduce power [ 699.234406] NOHZ: local_softirq_pending 80 [ 699.235553] NOHZ: local_softirq_pending 282 [ 699.236533] NOHZ: local_softirq_pending 282 [ 699.237543] NOHZ: local_softirq_pending 282 [ 699.238537] NOHZ: local_softirq_pending 282 [ 699.239533] NOHZ: local_softirq_pending 282 [ 699.954395] NOHZ: local_softirq_pending 01 [ 699.955375] NOHZ: local_softirq_pending 03 [ 699.956528] NOHZ: local_softirq_pending 03 [ 699.956620] NOHZ: local_softirq_pending 03 [ 770.446697] intel_powerclamp: Stop forced idle injection [ 3431.071390] intel_powerclamp: Start idle injection to reduce power [ 3499.240452] intel_powerclamp: Stop forced idle injection [ 3559.338108] intel_powerclamp: Start idle injection to reduce power [ 3603.420097] intel_powerclamp: Stop forced idle injection [ 3715.571884] intel_powerclamp: Start idle injection to reduce power [ 3755.677148] intel_powerclamp: Stop forced idle injection
Thanks,
On Thu, May 17, 2018 at 4:25 PM, Chris Murphy lists@colorremedies.com wrote:
Hi,
This seems newish, maybe 4.16 debug kernels were doing it also, but I've seen if for all of 4.17. But what's new, I think, with 4.17.0-0.rc5.git1.1.fc29.x86_64 is I'm seeing a 4 kidle_inject threads each using around 20% CPU per top, and the everything is sluggish, the mouse arrow jerks around on screen, and apps aren't responsive.
I'm not sure where to start with troubleshooting; my first thought is to just wait for rc6 and see if it's still happening.
Is this only visible in debug kernels, or do non debug kernels do it as
well?
Justin
[ 698.266514] intel_powerclamp: Start idle injection to reduce power [ 699.234406] NOHZ: local_softirq_pending 80 [ 699.235553] NOHZ: local_softirq_pending 282 [ 699.236533] NOHZ: local_softirq_pending 282 [ 699.237543] NOHZ: local_softirq_pending 282 [ 699.238537] NOHZ: local_softirq_pending 282 [ 699.239533] NOHZ: local_softirq_pending 282 [ 699.954395] NOHZ: local_softirq_pending 01 [ 699.955375] NOHZ: local_softirq_pending 03 [ 699.956528] NOHZ: local_softirq_pending 03 [ 699.956620] NOHZ: local_softirq_pending 03 [ 770.446697] intel_powerclamp: Stop forced idle injection [ 3431.071390] intel_powerclamp: Start idle injection to reduce power [ 3499.240452] intel_powerclamp: Stop forced idle injection [ 3559.338108] intel_powerclamp: Start idle injection to reduce power [ 3603.420097] intel_powerclamp: Stop forced idle injection [ 3715.571884] intel_powerclamp: Start idle injection to reduce power [ 3755.677148] intel_powerclamp: Stop forced idle injection
Thanks,
-- Chris Murphy _______________________________________________ kernel mailing list -- kernel@lists.fedoraproject.org To unsubscribe send an email to kernel-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/kernel@ lists.fedoraproject.org/message/7HWFMYOQPBHOTH3WEJ3NP6DAELIJ5OJ4/
On Fri, May 18, 2018 at 7:12 AM, Justin Forbes jforbes@redhat.com wrote:
On Thu, May 17, 2018 at 4:25 PM, Chris Murphy lists@colorremedies.com wrote:
Hi,
This seems newish, maybe 4.16 debug kernels were doing it also, but I've seen if for all of 4.17. But what's new, I think, with 4.17.0-0.rc5.git1.1.fc29.x86_64 is I'm seeing a 4 kidle_inject threads each using around 20% CPU per top, and the everything is sluggish, the mouse arrow jerks around on screen, and apps aren't responsive.
I'm not sure where to start with troubleshooting; my first thought is to just wait for rc6 and see if it's still happening.
Is this only visible in debug kernels, or do non debug kernels do it as well?
Only debug kernels so far. I had been using 4.17.0-0.rc4.git4.1 most of this week with the powerclamp messages, but never the slow down and without the kinject threads running. 4.17.0-0.rc5.git1.1 is the first with powerclamp messages, slow downs, and kinject threads running. The slow down coincides with kinjects, and when those go away, the sluggishness goes away also.
I'm going to run 4.16.9 today.
On Fri, May 18, 2018 at 9:30 AM, Chris Murphy lists@colorremedies.com wrote:
On Fri, May 18, 2018 at 7:12 AM, Justin Forbes jforbes@redhat.com wrote:
On Thu, May 17, 2018 at 4:25 PM, Chris Murphy lists@colorremedies.com wrote:
Hi,
This seems newish, maybe 4.16 debug kernels were doing it also, but I've seen if for all of 4.17. But what's new, I think, with 4.17.0-0.rc5.git1.1.fc29.x86_64 is I'm seeing a 4 kidle_inject threads each using around 20% CPU per top, and the everything is sluggish, the mouse arrow jerks around on screen, and apps aren't responsive.
I'm not sure where to start with troubleshooting; my first thought is to just wait for rc6 and see if it's still happening.
Is this only visible in debug kernels, or do non debug kernels do it as well?
Only debug kernels so far. I had been using 4.17.0-0.rc4.git4.1 most of this week with the powerclamp messages, but never the slow down and without the kinject threads running. 4.17.0-0.rc5.git1.1 is the first with powerclamp messages, slow downs, and kinject threads running. The slow down coincides with kinjects, and when those go away, the sluggishness goes away also.
I'm going to run 4.16.9 today.
On 4.16.9 today. But no kinject threads spun up, no slow downs. Or possibly I just didn't notice it.
May 18 12:28:23 f28h.local kernel: intel_powerclamp: Start idle injection to reduce power May 18 12:28:24 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 01 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:26 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:26 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:26 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:27 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:29:20 f28h.local kernel: intel_powerclamp: Stop forced idle injection
On Fri, May 18, 2018 at 7:34 PM, Chris Murphy lists@colorremedies.com wrote:
On Fri, May 18, 2018 at 9:30 AM, Chris Murphy lists@colorremedies.com wrote:
On Fri, May 18, 2018 at 7:12 AM, Justin Forbes jforbes@redhat.com wrote:
On Thu, May 17, 2018 at 4:25 PM, Chris Murphy lists@colorremedies.com wrote:
Hi,
This seems newish, maybe 4.16 debug kernels were doing it also, but I've seen if for all of 4.17. But what's new, I think, with 4.17.0-0.rc5.git1.1.fc29.x86_64 is I'm seeing a 4 kidle_inject threads each using around 20% CPU per top, and the everything is sluggish, the mouse arrow jerks around on screen, and apps aren't responsive.
I'm not sure where to start with troubleshooting; my first thought is to just wait for rc6 and see if it's still happening.
Is this only visible in debug kernels, or do non debug kernels do it as well?
Only debug kernels so far. I had been using 4.17.0-0.rc4.git4.1 most of this week with the powerclamp messages, but never the slow down and without the kinject threads running. 4.17.0-0.rc5.git1.1 is the first with powerclamp messages, slow downs, and kinject threads running. The slow down coincides with kinjects, and when those go away, the sluggishness goes away also.
I'm going to run 4.16.9 today.
On 4.16.9 today. But no kinject threads spun up, no slow downs. Or possibly I just didn't notice it.
May 18 12:28:23 f28h.local kernel: intel_powerclamp: Start idle injection to reduce power May 18 12:28:24 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 01 May 18 12:28:25 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:26 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:26 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:26 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:28:27 f28h.local kernel: NOHZ: local_softirq_pending 80 May 18 12:29:20 f28h.local kernel: intel_powerclamp: Stop forced idle injection
Oh and this is 4.16.9-300.fc28.x86_64, no debug.
kernel@lists.fedoraproject.org