Package Base Details: linux-rt

Git Clone URL: (read-only, click to copy)
Submitter: schivmeister
Maintainer: jhernberg (dvzrv, sangy)
Last Packager: dvzrv
Votes: 176
Popularity: 2.24
First Submitted: 2011-08-09 20:03
Last Updated: 2021-05-18 20:55

Pinned Comments

dvzrv commented on 2021-01-12 21:52

The repository for linux-rt and linux-rt-lts has moved to a new location.

If you use the custom repository, please update your pacman.conf accordingly, as the other one is going away by the end of the month (January 2021)!

Latest Comments

« First ‹ Previous ... 14 15 16 17 18 19 20 21 22 23 24 ... Next › Last »

Ralf_Mardorf commented on 2017-12-22 05:36

I posted that much, including jackd messages, in reply to smoge's requests. However, I mentioned to continue at, but since we now have got an audio mailing list again, I recommend to join Btw. I posted my results of cyclictest to the mailinglist, see

jhernberg commented on 2017-12-22 00:32

@Gimmeapill: Well I've also had that impression, another reason that I've been reluctant to upgrade it, but it was getting impossible to sit on it so long..;)

But keep in mind that the -rt version is really the experimental branch of the rt patchset. I plan to upgrade linux-rt-lts to 4.9-rt once I know that there are no big issues with this kernel. That is still a rock solid kernel AFAIK, and possibly the one most people should to use. That is a LTS release of the kernel and a stable version of the rt patchset.

@kflak: I also wonder what you use to build with, because that's certainly not normal.. Maybe you get confused because it's a split package and it creates 3 different packages.

My recommendation would be to install devtools, and use the extra-x86_64-build script to build the kernel. Follow dvzrv instructions, but replace the makepkg command with extra-x86_64-build which will create a chroot, install all needed dependencies and finally build the packages. That is how most TUs and devs build the distro packages.

@dvzrv: That would be a stone off my chest and great for the community!

dvzrv commented on 2017-12-21 22:02

@jhernberg: Thanks so much! We'll hopefully see to it, that this kernel can be included in [community] one day!

@kflak: How are you building? With

  git clone 
  cd linux-rt 
  makepkg -L 

this should do fine.

kflak commented on 2017-12-21 21:25

Newbie question here, I preemptively apologize for my stupidity...: When I install the kernel the whole build process runs three times, once for linux-rt, once for linux-rt-headers and once for linux-rt-docs. However, it seems that it is pretty much the same thing happening all three times. Am I crazy, or is there a way of reducing the build time to only once, and still get all the headers/docs installed?

Gimmeapill commented on 2017-12-21 09:10

@jhernberg: I forgot to mention that with the stock kernel I get around 35 micro secs worst case scenario. So yes, there is still a very tangible benefit with the RT patch.

Regarding the perf degradation across versions of the RT kernel: I could also empirically notice more xruns with 4.14 RT than 4.11 RT, as I always run at the limits.

It could be the RT patch being not very mature yet for this version, or the kernel itself (which seems to have grown significantly bigger). I'll try to rebuild with localmodconfig see if that helps (although in theory it shouldn't).

jhernberg commented on 2017-12-21 07:45

@Gimmeapill: Thanks! It's good to know that the work is appreciated!

Regarding the cyclictest results, I've had the same impression. I guess this is due to the kernel or patch itself, on the other hand a difference of 12us vs 15us doesn't really seem like it would be a huge problem ;) This is already very respectable results compare to other kernels.

Gimmeapill commented on 2017-12-20 22:17

@jhernberg: Thanks a lot for your work during all this years, the burnout is quite understandable.

I just hope you won't drop the towel completely ;-)

Anyway I gave 4.14 a very brief test, and performance seems to be a bit worse than 4.11. (which was itself a bit worse than 4.9).

More specifically: 'cyclictest -S -m -p98' with governor in performance mode shows on my system a max latency of 15 micro secs with 4.14.6-rt7-1-rt, vs 13 micro secs with 4.11.12-rt14-1-rt, vs 12 micro secs with 4.9.40-rt30-1-rt

On the good news: no stability issue so far.

jhernberg commented on 2017-12-20 16:28

@Ralf_Mardorf: Please restrain your pastes in the comments, as reading the comments is really getting a little bit difficult with everything you have pasted. For instance there really is no need to paste the output of jack starting up, this is not a mailing list where one can just move on to the next message...

jhernberg commented on 2017-12-20 16:02

Hello everyone,

Sorry that I've been missing in action and have neglected my packages. I haven't had much time at all due to real life and other engagements. In fact I've come to realize that I've burnt out on packaging... I think I've found someone to take over this and some other packages, so hopefully things will be a lot better in the future!

I just updated this package to the latest upstream version, in fact due to a lot changing in Archlinux upstream, this is more or less a new package. I'm not on my normal machine to test it, so not 100% sure that everything is running at it's best, but I think everything is ok. Please give it a good test and let me know about any issues.

Ralf_Mardorf commented on 2017-12-03 06:02


I didn't update to 4.14.3-rt5 since building vbox modules failed for 4.14 from testing as well as for the local build of 4.14-rt1.

$ pacman -Q linux{,-rt{,-cornflower,-pussytoes}}|awk '{print $2}'

I booted

[rocketmouse@archlinux ~]$ uname -r


PS: I neither upgraded to 4.14.3-1, the current version available by testing.