Re: [patch] Real-Time Preemption, -RT-2.6.10-rc1-mm3-V0.7.25-1

[Shane Shrybman <shrybman@aei.ca>]

On Fri, 2004-11-12 at 13:23, Mark_H_Johnson@RAYTHEON.COM wrote:
> >Typical example of the error message:
> >
> >kernel: hde: dma_timer_expiry: dma status == 0x24
> >kernel: ALSA sound/core/pcm_native.c:1424: playback drain error (DMA or
> IRQ trouble?)
> >kernel: PDC202XX: Primary channel reset.
> >kernel: hde: DMA interrupt recovery
> >kernel: hde: lost interrupt
> >
> >This was on a Promise TX2 133 ide card with one IDE disk. The problem
> >would show itself if using the RT patches and APIC. But the problem seems
> >to have been resolved now.
> 
> I had errors like that one when the IDE IRQ was at a priority less than
> the real time task. Since then, I run with all the IRQ's at max RT priority
> and will continue to do so until I get a better assessment of what my real
> application (not these audio tests...) needs for IRQ priorities.
> 

Ok, I wasn't comparing apples to apples. I forgot I had to remove the sb live
card from this machine a few days ago. So the hardware config wasn't exactly
the same. I have reinstalled the sb live card now and I am retesting on 0.7.25.
The sb live shares an irq with ide2(promise card)

           CPU0       
  0:     835791    IO-APIC-edge  timer  0/35791
  1:       2207    IO-APIC-edge  i8042  1/2207
  8:          4    IO-APIC-edge  rtc  0/4
  9:          0   IO-APIC-level  acpi  0/0
 15:         11    IO-APIC-edge  ide1  1/9
 16:      70527   IO-APIC-level  ide2, EMU10K1  0/70527
 17:       1093   IO-APIC-level  eth0  0/1092
 18:      37440   IO-APIC-level  bttv0, Bt87x audio  173/37439
 19:      39147   IO-APIC-level  aic7xxx, ivtv0  340/39143
 21:      25091   IO-APIC-level  uhci_hcd, uhci_hcd, uhci_hcd, uhci_hcd  52/25091
 22:      85425   IO-APIC-level  VIA8237  494/71991
NMI:     836519 
LOC:     836340 
ERR:          0
MIS:          1

With the sb live card back in use the system has hung once with the sound looping.
It hung after I started playing a video in a second instance of the mythfrontend
application. I had the nmi_watchdog on and netconsole logging to another machine
but there was nothing in the logs.

I have rebooted and I am trying to verify the dma_timer_expiry issue is still gone
with the sb live in use.

> This may have been fixed as a side effect of Ingo setting the IRQ threads
> at
> RT priorities in the 40's.
> 

I had originally thought this might be the cause as well so I jacked
ide2 priority but it didn't help. However it does share the irq so maybe
that is a factor here.

> --Mark H Johnson
>   <mailto:Mark_H_Johnson@raytheon.com>
> 

shane