On Thu, 6 Dec 2012, Jukka Ruohonen wrote:
On Thu, Dec 06, 2012 at 07:20:56AM -0800, Paul Goyette wrote:Under heavy load (such as, build.sh -j16) my system shows a lot of "system" time (30-50%, sometimes even more). systat shows that I'm getting ~140K interrupts per second on ioapic0 pin 9. Unfortunately, dmesg doesn't seem to indicate what device might be connected to that pin! It does show acpi0 at mainbus0: Intel ACPICA 20110623 acpi0: X/RSDT: OemId <112311,XSDT2034,20111123>, AslId <MSFT,00000097> acpi0: SCI interrupting at int 9 but I can't imagine ACPI generating that quantity of interrupts!It may when configured incorrectly. Verify against hw.acpi.stat.sci -sysctl.
Hmm, interesting: # x="sysctl -n hw.acpi.stat.sci" ; y=`$x` ; sleep 5 ; z=`$x` ; echo $(($z - $y)) 832191 # x="sysctl -n hw.acpi.stat.sci" ; y=`$x` ; sleep 5 ; z=`$x` ; echo $(($z - $y)) 851205 So yeah, around 160K per second... ------------------------------------------------------------------------- | Paul Goyette | PGP Key fingerprint: | E-mail addresses: | | Customer Service | FA29 0E3B 35AF E8AE 6651 | paul at whooppee.com | | Network Engineer | 0786 F758 55DE 53BA 7731 | pgoyette at juniper.net | | Kernel Developer | | pgoyette at netbsd.org | -------------------------------------------------------------------------