[arch-proaudio] Changes to jack/jack2 and realtime privileges

[Ralf Mardorf]

On Tue, 21 Aug 2018 23:43:26 +0200, David Runge wrote:
>It is now an optional dependency to jack and jack2{,-dbus} and can of
>course also be dependend upon from other packages if needed.

Hi,

it's reasonable to change the group name from "audio" to "realtime",
since regular usage of CNC machines has less to do with audio and they
are even not much used for industrial music.

However, wasn't the original plan to tidy up Arch pro-audio step by
step and among other things, to stay away from esoteric?

What's the advantage of 'rtprio 98' over upstream's recommendation
'rtprio 99'?

The packages URL is even more confusing.



The package:

[rocketmouse at archlinux community-any]$ cat 99-realtime-privileges.conf 
@realtime - rtprio 98
@realtime - memlock unlimited
[rocketmouse at archlinux community-any]$ grep url PKGBUILD 
url="https://wiki.archlinux.org/index.php/Realtime_process_management"



Upstream:

"@realtime   -  rtprio     99
@realtime   -  memlock    unlimited" -
http://jackaudio.org/faq/linux_rt_config.html



The package's URL:

"An example for why one might want to alter these settings is to get
high-performance audio working. The defaults are permissive enough to
get jack-server running with hydrogen or ardour. However, for higher
performance audio applications it might be necessary to redefine the
values for rt_prio from 65 to 80 or even higher! The following settings
work well with ardour:

@audio          -       rtprio          70
@audio          -       memlock         250000" -
https://wiki.archlinux.org/index.php/Realtime_process_management#Configuring_PAM



I prefer to stay with my manual edited settings...

[rocketmouse at archlinux community-any]$ tail -2 /etc/security/limits.conf
@audio 	- rtprio 	99
@audio 	- memlock 	unlimited

...perhaps I'll migrate to a group "realtime", too.



What are the below rules good for?

[rocketmouse at archlinux community-any]$ cat 40-realtime-privileges.rules 
# rw access to /dev/cpu_dma_latency to prevent CPUs from going into idle state
KERNEL=="cpu_dma_latency", GROUP="realtime"
# rw access to /dev/rtc0
KERNEL=="rtc0", GROUP="realtime"
# rw access to /dev/hpet0
KERNEL=="hpet", GROUP="realtime"



For example, using hpet without such a rule seemingly works without
issues:

[rocketmouse at archlinux community-any]$ grep -i hpet /usr/lib/udev/rules.d/*; echo $?
1

Regards,
Ralf

-- 
pacman -Q linux{,-rt{-pussytoes,-cornflower,,-securityink}}|cut -d\  -f2
4.18.3.arch1-1
4.18_rc8_rt1-1
4.16.18_rt12-1
4.16.18_rt11-1
4.16.18_rt10-1