[LAU] rtirq

Len Ovens len at ovenwerks.net
Sun Mar 15 15:08:23 UTC 2015

On Sun, 15 Mar 2015, Harry van Haaren wrote:

> On Sun, Mar 15, 2015 at 8:29 AM, Jeremy Jongepier <jeremy at autostatic.com> wrote:
>> Which disk writes? From the audio software? Aren't those performed in
>> the audio thread and if so don't they inherit the higher rtprio assigned
>> by JACK?
> Nope, the audio data is copied into a ringbuffer, and then written to
> disk using a non-RT thread.
> The write() function call has unbounded execution time, so it must not
> be called in the RT thread in order to avoid Xruns.

That was my thought too, SW writes to memory buffers and then the OS 
writes the buffers to disk. Though it sounds like you are suggesting there 
is yet another layer before the data even makes it to memory buffers.

The reason I asked, is from reading about people using three drives for 
audio: OS, sample data, and audio streaming (record). And though I have 
not had any problems with disk writes (even when I was using only 1GB of 
ram and a PATA drive) I do tend to simpler things of fewer tracks and no 
synth. For the most part, I would think any OS parts would be sitting in 
memory anyway.

I know there are other applications that run several disks 
together using only the outer half tracks in a raid 0 (I think) setup for 
faster disk writes, but even there I would think the disk writes are still 
much slower than ram.

Len Ovens

More information about the Linux-audio-user mailing list