Skip to contentSkip to conversation

2.4.22-pre5 interrupts problem ?

DISCUSSION Stefan Schmidt 1 participant 1 message Jul 13, 2003 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw
Quoted 15 lines · Ruben Puettmann

cat /proc/interrupts
CPU0

  0:     655923    IO-APIC-edge  timer
  1:      22221    IO-APIC-edge  keyboard
  2:          0          XT-PIC  cascade
  8:          4    IO-APIC-edge  rtc
  9:          0   IO-APIC-level  acpi
 14:      33084    IO-APIC-edge  ide0
 15:          3    IO-APIC-edge  ide1
 18:  690660829   IO-APIC-level  aic7xxx, es1371
 19:  686559334   IO-APIC-level  bttv, eth0
 20:  683012480   IO-APIC-level  ehci_hcd, usb-ohci
 22:     115606   IO-APIC-level  usb-ohci
NMI:          0 LOC:     655895 ERR:          0
MIS:          0

Hi, i am seeing the same behaviour of the Epox 8RDA3+ on 2.4.22-pre5-vanilla.
Please note that only level-APICs seem to be affected.
We don't see these high interrupt rates anymore when APIC is disabled in the BIOS.
I also noted that the POST LED on the board flips between 94 and 96 which it does not under WinXP where FF is displayed.
I have also a SATA Drive attached to the internal controller and under pre5 the box freezes if i do any serious amount of disk thoughput - for example using bonnie++. Maybe this is somehow related to this interrupt problem. I experience freezes with current 2.5 kernels too.
WinXP is installed on the SATA Drive and rock stable, so this actually seems not to be a hardware problem per se.

Quoted 7 lines

vmtstat 2
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----

 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0  0      0 680320  35816 278576    0    0    30    66 1995   313 38  4 58  0
 0  0      0 680316  35816 278576    0    0     0     0 322131     7  0  0 100  0

Is this normal?

If it were i would be desperate to hear why for my most frequented nameserver (>300queries/s) hardly even does 1000 interrupts/s. ;)

waves,
Stefan