[PlanetCCRMA] fedora 17 support: kernel 3.4.0-rt7 is available

[Fernando Lopez-Lezcano]

Hi all,

I have started the long process of building for Fedora 17. You can find 
planetccrma-repo in the usual place (I updated the web page a little bit 
with pointers to all the stuff). For those that venture into the new 
land of Fedora 17 there is now a Planet CCRMA low latency patched kernel 
available based on 3.4.0-rt7. Brand new. Seems to work very well (only 
lightly tested on x86_64).

Some notes:

= rtirq
As usual installing planetccrma-core (or planetccrma-core-PAE on i386) 
should bring in all the goodies. Rtirq has some patches, by default it 
does not include usb in the cards that are processed at boot time but 
will up the priority of the irq thread when a usb card is inserted (this 
means dynamically inserted cards will be automatically recognized[*]). 
It will also remember the increased priorities when the computer is set 
to sleep and later wakes up.

= Realtime scheduling priorities:
All the priorities match the new scheme in Fedora 17. Max realtime 
scheduling priority is now 70 so soundcards and other related stuff are 
located between 65 and 70, udev soundcards always land at 69 and the 
rest goes to 50 which is the default. Jack runs at 60 and jack clients 
run at 55 so all is well (but, for example, jconvolver decrements its 
thread priority by two for each thread so the lowest one goes lower than 
50... maybe we need a patch. This works whether you boot into the Planet 
CCRMA kernel or the normal Fedora kernel with "threadirqs" in the boot 
line.

= Realtime scheduling access:
You need to add your user to the group jackuser (and logout and login 
again) to get access to realtime scheduling. You can also install 
planetccrma-rt-permissions-all to give rt access to all users (with an 
increased security risk, of course)

Anyway, the new kernel seems to run fine in my test laptop with an 
x86_64 install.

My build server is busy right now building more packages...
More tomorrow.

Enjoy!
-- Fernando

[*] it will also remove the high priority if udev issues a "remove" 
event for the soundcard. I tested this again today with a usb soundcard 
and it did not work (ie: no "remove" event). Will have to debug this. I 
swear this was working a week ago... Computers... Sigh.