Package Details: linux-rt-headers

Git Clone URL: (read-only, click to copy)
Package Base: linux-rt
Description: Headers and scripts for building modules for the Linux RT kernel
Upstream URL:
Licenses: GPL2
Submitter: schivmeister
Maintainer: jhernberg (dvzrv)
Last Packager: dvzrv
Votes: 171
Popularity: 1.83
First Submitted: 2011-08-09 20:03
Last Updated: 2020-08-21 18:34

Pinned Comments

dvzrv commented on 2019-05-13 21:17

I have added a signed, unofficial user repository for linux-rt, backed by this PKGBUILD.

You can add it to your pacman.conf with:

Server =$arch

I'll try to keep it as up-to-date as possible and retain old versions.


Latest Comments

1 2 3 4 5 6 ... Next › Last »

dvzrv commented on 2020-09-03 19:03

@Ralf_Mardorf: Sorry for the late reply. I have been very busy and then on vacation.

The CONFIG_HZ=1000 setting is indeed thought to improve performance under certain workloads (and it doesn't seem to have much of a negative effect altogether). Note, that according to this article the effects of a higher resolution timer might be diametral on systems with high I/O.

If you encounter any issues let me know!

If you have time to do benchmarks between this setting and e.g. the default CONFIG_HZ=300, please do share!

Ralf_Mardorf commented on 2020-08-23 14:56

I missed #comment-745316, which probably is the reason to revert to 1000 Hz.

However, it's puzzling that when using USB MIDI devices, a faster USB polling rate does improve xrun issues of an audio device. Apart from this, I couldn't find information to which extend CONFIG_HZ is related to the USB polling rate.

Ralf_Mardorf commented on 2020-07-19 14:54

Hi, what's the reason for reverting to CONFIG_HZ=1000?

FWIW due to at least dkms building virtualbox-bin modules issues with 5+ kernels, I stay with 4.19+ kernels. At the moment I've got 4.19.127-rt55 and 4.19.132-rt59 installed, both with CONFIG_HZ=300.

aquilarubra commented on 2020-06-23 15:53

I still find issues with dkms modules. I tried now in 3 different computers, with one of them being AMD.

Results: In 1 Intel and one AMD I can load the nvidia dkms, but not the virtual box dkms. In the third one, I can load none. Errors may differ slightly, but basically disk detection is lost and the kernel cannot find the filesystem.

Marc.2377 commented on 2020-06-23 09:16

@habys: Would you happen to know if the xruns I'm getting on my USB soundcards can be expected to improve by raising the polling frequency in the same manner?

(edit) answering myself, after testing: no it doesn't. But I found the culprit. Running jack in duplex mode (default) results in clock drift between input and output. Running in playback only gets rid of xruns, and I can still manage to have capture by running 'alsa_in' on the same soundcard.

habys commented on 2020-05-15 00:16

Hey {jhernberg, dvzrv} would you consider

diff --git a/config b/config
index 381a044..7057c9c 100644
--- a/config
+++ b/config
@@ -457,8 +457,8 @@ CONFIG_EFI_MIXED=y
 # CONFIG_HZ_100 is not set
 # CONFIG_HZ_250 is not set
-# CONFIG_HZ_1000 is not set
+# CONFIG_HZ_300=y

The default relatively slow USB poll rate on a realtime audio box causes XRUNS when using USB MIDI devices.

aquilarubra commented on 2020-03-21 11:36

Tried virtualbox-bin, but same behavior. Loading modules on demand hangs system. I also have the nvidia dkms, but it shouldn't matter. The issue seems uefi-virtualbox related.

Ralf_Mardorf commented on 2020-03-21 08:24

Hi, I stopped building new 5.4 kernels and will continue with kernels > 5.4. I'm still building new 4.19 LTS releases and use those instead. The kernels are pretty much the same as the AUR kernels, just with minimal personalised changes and I'm using virtualbox-bin from AUR. Btw. I can boot into 5.4 kernels without efi and no issue at all, but virtualbox guests don't start. Ever since I can remember I load the modules on demand, not at startup.

$ pacman -Q linux-rt{,-pussytoes,-cornflower,-securityink} virtualbox-bin|cut -d\  -f2

aquilarubra commented on 2020-03-20 20:18

Any idea why virtualbox-host-dkms is incompatible? Initially it gave me a uefivars error at boot. I solved that by adding efi=runtime to the kernel. Then, as udev loads the virtualbox module, the system hangs with a ton of errors. The uefi disks are not recognized. I also tried disallowing the dkms module to load at boot. Then it boots, but as I do a modprobe vboxdrv, everything hangs. I think the incompatibility is between uefi and virtualbox. I see that Manjaro created a package with extra modules for virtualbox with the rt kernel and it is reported to work.

Ralf_Mardorf commented on 2020-03-14 17:39

Hi, the package contains the kernel in

$ pacman -Ql linux-rt | grep vmlinuz
linux-rt /usr/lib/modules/5.4.22-rt13-1.0/vmlinuz

Where the package is stored depends on your makepkg.conf

$ grep PKGDEST /etc/makepkg.conf

If you install the package, the kernel is installed to

$ diff /boot/vmlinuz-linux-rt /usr/lib/modules/5.4.22-rt13-1.0/vmlinuz; echo $?
$ ls -hl /boot/vmlinuz-linux-rt /usr/lib/modules/5.4.22-rt13-1.0/vmlinuz
-rw-r--r-- 1 root root 5.9M Mar  4 10:09 /boot/vmlinuz-linux-rt
-rw-r--r-- 1 root root 5.9M Mar  4 06:41 /usr/lib/modules/5.4.22-rt13-1.0/vmlinuz