153897 interrupts a second?

[John Klos <john%ziaspace.com@localhost>]

Hi,

I'm seeing something strange going on with an Xserve G4 running NetBSD-5 
from today's sources. Output from top shows something like:

load averages:  0.02,  0.22,  0.26;               up 0+00:33:01        19:41:42
20 processes: 19 sleeping, 1 on CPU
CPU states:  0.0% user,  0.0% nice,  0.0% system, 56.7% interrupt, 43.3% idle
Memory: 274M Act, 864K Wired, 5636K Exec, 261M File, 1577M Free
Swap: 2048M Total, 2048M Free

I was wondering what was causing all of the interrupt load, so I looked at 
sysstat's output:

    2 users    Load  0.00  0.18  0.24                  Mon Aug 30 19:42:39

Proc:r  d  s  w     Csw    Trp    Sys   Int   Sof    Flt      PAGING   SWAPPING
           7         52      1        *****    23             in  out   in  out
                                                        ops
   0.0% Sy   0.0% Us   0.0% Ni  56.8% In  43.2% Id    pages
|    |    |    |    |    |    |    |    |    |    |
%%%%%%%%%%%%%%%%%%%%%%%%%%%%                                              forks
                                                                          fkppw
           memory totals (in kB)          154186 Interrupts               fksvm
          real  virtual     free             100 cpu0 clock               pwait
Active  280588   280588  1614484                 openpic irq 22           relck
All     416152   416152  3711632                 openpic irq 19           rlkok
                                                 openpic irq 27           noram
Namei         Sys-cache     Proc-cache           openpic irq 28           ndcpy
    Calls     hits    %     hits     %           openpic irq 52           fltcp
        6        6  100                   154084 openpic irq 1            zfod
                                                 openpic irq 40           cow
Disks:   cd0   st0   wd0   md0                 1 openpic irq 41       256 fmin
 seeks                                         1 gem0 interrupts      341 ftarg
 xfers                                                                    itarg
 bytes                                                                216 wired
 %busy                                                                    pdfre


Hmmm... 154084 interrupts a second from the IDE bus?

pdcide0 at pci1 dev 21 function 0
pdcide0: Promise Ultra133TX2v2/ATA Bus Master IDE Accelerator (rev. 0x03)
pdcide0: bus-master DMA support present
pdcide0: primary channel configured to native-PCI mode
pdcide0: using irq 1 for native-PCI interrupt

Just for testing purposes, I loaded the drive into another Xserve G4 which 
showed the same problem (although it was strange that the first Xserve let 
me boot using hd:,\ofwboot.xcf, but the second only after replacing that 
with ultra0:,\ofwboot.xcf).

Both Xserves are the recent version with DDR memory. One has a Sonnet 1.8 
GHz 7447a dual processor card with a uniprocessor kernel (the magic smoke 
escaped from the power supply components for the second processor, but it 
works fine as uniprocessor), and the other an Apple original 1.42 GHz 
single 7455 processor. Interesting picture:

http://www.ziaspace.com/~john/processor_no_smoke.jpg

Does anyone have any ideas what's up with the IDE bus on these machines?

Thanks,
John Klos