You are not logged in.
Pages: 1
Topic closed
Hello,
I have been trying to track down the reason my system runs a little hot, and
when running powertop, I noticed that there were (what seemed like) a ton of
interrupts from acpi:
Top causes for wakeups:
99.5% (20393.4) [acpi] <interrupt>
0.2% ( 33.0) [iwlwifi] <interrupt>
0.1% ( 22.4) [Rescheduling interrupts] <kernel IPI>
...
My questions are: a) Is this normal? And if not, b) does anyone know what I
can do about it? I came across a bug report that makes me think this should
not be happening: https://bugs.archlinux.org/task/28859
I'm not sure what the 20393 refers to, but I suspect it is the number per
second, since it doesn't seem to change much, irrespective of the polling
interval I use.
Thanks in advance for your help : )
Last edited by wes (2012-04-06 21:05:18)
Offline
Top causes for wakeups:
47.4% (400.9) [Rescheduling interrupts] <kernel IPI>
15.8% (133.7) [uhci_hcd:usb3, snd_hda_intel, radeon, b43] <interrupt>
9.1% ( 77.4) swapper/0
6.4% ( 53.9) USB device 2-1 : Microsoft USB Optical Mouse (PixArt)
5.4% ( 45.9) swapper/1
5.0% ( 42.1) [ehci_hcd:usb1, uhci_hcd:usb2, firewire_ohci] <interrupt>
mine is relatively less than yours. maybe it depends on hardware, I guess.
LENOVO Y 580 IVYBRIDGE 660M NVIDIA
Unix is user-friendly. It just isn't promiscuous about which users it's friendly with. - Steven King
Offline
*Edit:* I don't see acpi at all... you are not using acpi_cpufreq, right?
Last edited by wes (2012-03-24 21:21:36)
Offline
no. I dont have it.
Edit:
I loaded the module manually, but still
Top causes for wakeups:
42.3% (394.6) [Rescheduling interrupts] <kernel IPI>
10.3% ( 96.2) swapper/1
10.1% ( 93.8) [uhci_hcd:usb3, snd_hda_intel, radeon, b43] <interrupt>
10.0% ( 93.6) swapper/0
9.8% ( 91.8) [extra timer interrupt]
6.8% ( 63.2) firefox
Suggestion: Enable the ondemand cpu speed governor for all processors via:
echo ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
Q - Quit R - Refresh O - enable Ondemand governor
Last edited by hadrons123 (2012-03-25 01:04:19)
LENOVO Y 580 IVYBRIDGE 660M NVIDIA
Unix is user-friendly. It just isn't promiscuous about which users it's friendly with. - Steven King
Offline
Hmmm... : ( I did the opposite- I booted without loading that module, and
*still* had 20000 interrupts from acpi.
I've also tried acpi=off in the kernel parameters, which sent the average
number of wakeups down to around 3.0/sec., but this gave some strange behavior
in other areas: upon booting, the screen resolution never changed as it
usually does after processing udev events, and parts of my rc.local (which I
guess were acpi-related) seemed to have failed. Anticipating disaster, I
didn't even try starting X.
Thanks for getting back to me.
Offline
Does it produce any trouble for you?
Having so many interrupts.
Last edited by hadrons123 (2012-03-25 11:16:57)
LENOVO Y 580 IVYBRIDGE 660M NVIDIA
Unix is user-friendly. It just isn't promiscuous about which users it's friendly with. - Steven King
Offline
The main things are that the system seems to run a little hot, and that the
battery isn't lasting as long as when it was running windows (20% shorter
life, maybe worse). Not a complete disaster, but pretty annoying. Anyway,
having that many interrupts seems like it could have something to do with
these issues, no? I'm pretty ignorant, but it seemed plausible.
Thanks for your help.
Offline
Does it produce any trouble for you?
Having so many interrupts.
Yes, as a matter of fact it does:
https://bbs.archlinux.org/viewtopic.php?id=138780
I wonder how many other things are performing poorly as a result of this. For arithmetic it is a factor two... pretty ridiculous.
Offline
I know this thread is a little stale, but I thought I should give an update.
I have "solved" the issue by taking the following steps:
1. turn off computer.
2. remove battery.
3. boot computer, and note that you no longer have a ton of acpi interrupts.
4. turn off computer.
5. put the battery back.
6. boot computer and note that acpi interrupts are still gone.
Since I don't have much intuition for why this worked, I am worried it may break again at some point. But for now I will mark as solved.
Offline
I'm having the same problem.
I used
$watch cat /proc/interrupts
to find out that acpi interrupts were the problem.
I found this which helped me to understand how to investigate the problem.
I also found this which solved the problem, but only after put the computer in suspend mode.
I combined these result and I have an idea what the problem is.
I found out that before suspending the pc I had something like that(gpeF6 is the interrupt that gives me the problem):
$cat /sys/firmware/acpi/interrupts/gpe6F
146657 EN STS enabled unmasked
and after suspending it i had:
$cat /sys/firmware/acpi/interrupts/gpe6F
252083 EN enabled unmasked
I don't know what STS means, but without it the problem is solved.
Can anyone explain to me something more? Can anyone find out a solution other than suspending the computer every time I boot?
PS: I tried to insert in the crontab an echo on the file, but it didn't work.
Offline
This thread is 11 years old. Please do not necrobump, especially threads marked solved.
Instead, start a new topic and link back to this one if you think it still applies after 11 years.
Closing.
Offline
Pages: 1
Topic closed