[LAU] rtirq usage

[Patrick Shirkey]

On 09/05/2009 09:07 AM, James Cameron wrote:
> On Thu, Sep 03, 2009 at 04:06:29PM +1000, Patrick Shirkey wrote:
>    
>> What I am seeing is a massive spike in xruns when running a gui app taht
>> connects to jack (even qjackctl) but nothing when playing a track from
>> the commandline ie mplayer -ao jack. This only happens if the wifi
>> module is installed. Without it there are no xruns for days.
>>
>> Unfortunately the snd the video device are on the same irq.
>>
>> Does anyone have an idea for why the gui system would become unstable
>> when the wifi device is added?
>>      
> /me handwaves
>
> It implies a larger than before amount of time being spent inside the
> driver for either the wifi device or the video device.  The extra time
> or the temporary disabling of interrupts within that driver might be
> affecting the delivery and processing of interrupts for the other two
> devices.
>    


Thanks for taking some time to consider this.

The wifi device is connecting via pcie.

It turns out that to get the xruns to a minimum it's necessary to run 
jack at rtprio 89.


--
Patrick Shirkey
Boost Hardware Ltd.


> Consider how the wifi device is communicating with the CPU.  Is it via
> PCI or USB?  If USB, then the type of USB controller becomes
> interesting.  You may have a controller that meets multiple
> specifications, and if so you may get different results by manually
> choosing the controller driver.  Controller is often called host.
>
> Look at your modules installed.  Scan for modules with names ending in
> hcd.  Test them one at a time.  That is to say, remove all USB devices
> except the offending device, remove (rmmod) all of the -hcd modules,
> then manually insert one at a time from this set:
>
> ehci-hcd.ko
> isp116x-hcd.ko
> ohci-hcd.ko
> r8a66597-hcd.ko
> sl811-hcd.ko
> u132-hcd.ko
> uhci-hcd.ko
>
> ... to find which ones work with the hardware and the device (using
> lsusb), then compare your findings against what was originally
> installed.
>
>