[linux-audio-user] Re: jackd "zombified - calling shutdown handler"

[Pieter Palmers]

grae wrote:
> On 2/24/07, Arnold Krille <arnold.krille at gmail.com> wrote:
>> 2007/2/23, Pieter Palmers <pieterp at joow.be>:
>> > Another tip: due to libfreebob internals, it is better to use 3 smaller
>> > periods instead of 2 larger, e.g. -n3 -p256 will probably work better
>> > than -n2 -p512.
>>
>> Hmm, using 2x128 and 2x256 proved to be more stable than 3x256 at my 
>> setup...

'theoretically' that is very strange, I think there is a problem elsewhere.

> Ok well I seem to still get the zombies.  I got the rtirq script and I
> *think* it's working.  I configured it just like the freebob site says
> 
> output from "rtirq status":
>  PID CLS RTPRIO  NI PRI %CPU STAT COMMAND
>  779 FF      80   - 120  0.0 S<   IRQ-8
> 2076 FF      75   - 115  0.0 S<   IRQ-16
> 2118 FF      64   - 104  0.0 S<   IRQ-17
> 2166 FF      63   - 103  0.0 S<   IRQ-18
> 2172 FF      62   - 102  0.0 S<   IRQ-19
>  815 FF      60   - 100  0.0 S<   IRQ-1
>  814 FF      59   -  99  0.2 S<   IRQ-12
>    4 FF      50   -  90  0.0 S    softirq-high/0
>    5 FF      50   -  90  0.0 S    softirq-timer/0
>    6 FF      50   -  90  0.0 S    softirq-net-tx/
>    7 FF      50   -  90  0.0 S    softirq-net-rx/
>    8 FF      50   -  90  0.0 S    softirq-block/0
>    9 FF      50   -  90  0.0 S    softirq-tasklet
>   10 FF      50   -  90  0.0 S    softirq-sched/0
>   11 FF      50   -  90  0.0 S    softirq-hrtimer
>   12 FF      50   -  90  0.0 S    softirq-rcu/0
>   42 FF      50   -  90  0.0 S<   IRQ-9
> 2215 FF      50   -  90  0.0 S<   IRQ-14
> 2216 FF      50   -  90  0.0 S<   IRQ-15
> 
> I used "/proc/interrupts" to show the irqs, specifically noted IRQ 16
> 16:        741   IO-APIC-fasteoi   uhci_hcd:usb1, ehci_hcd:usb5, 
> ohci1394, HDA
> Intel, i915 at pci:0000:00:02.0
Can you provide the complete /proc/interrupts?

> 
> It seems that my firewire device is sharing an IRQ with a couple usb
> ports, the internal soundcard and internal video.  My guess is that
> this isn't good.  Should I attempt the give the firewire it's own IRQ
> and if so, could you point me in the direction of how to do that?  I'm
> figuring this stuff out as I go here so it's hard to know how to
> proceed sometimes.
This is a common problem with laptops and i have been struggling with 
this myself, and AFAIK the only thing you can hope is that your system 
has an IO-APIC, and that it is routed in a favorable fashion (hardware 
that is). Try and enable the IO-APIC in the kernel and see if it comes 
up (some lines having the word IO-APIC should appear in /proc/interrupts.)

Note that intel mac mini's seem to have the same problem, but worse: 
almost everything shares one interrupt. Imagine: video + SATA controller 
+ audio card.

But as I said, this is all AFAIK, I really hope that someone stands up 
and explains how to fully split these interrupts...

> 
> Also, when setting the jack priority do I want it to be just under the
> firewire device?  for example:  if IRQ 16 was just the firewire
> (running at 75), then running jack at 74 would make sense?
You should take into account that libfreebob will also create threads to 
perform the packet handling, and those are running at jack_prio+10. It 
is therefore best to run jackd with a priority of less than ohci_prio - 10.