Uptime is never less than one on Vero 4k

The system load, as reported by “uptime”, is never less than 1.00 on my Vero 4k device:

osmc@vero4k:~$ uptime
12:47:31 up 30 min, 1 user, load average: 1.21, 1.18, 1.00

Is this normal for all Vero 4k devices or it is just my device?

osmc@vero4k:~$ uname -a
Linux vero4k 3.14.29-156-osmc #1 SMP Tue Oct 15 20:21:25 UTC 2019 aarch64 GNU/Linux

I’ve been thinking it was high load as well, it’s almost like there is a decimal pointer wrong, but it’s normal:

18:56:55 up 8 days, 4:56, 2 users, load average: 2.28, 2.31, 2.39

that’s my load haveing a couple of services running plus a irc-session, Not a heavy load at all.

Some additional data from “top”:

      PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
      585 osmc      20   0  502892  87448  26516 S  20.5  4.9   7:15.68 kodi.bin
     1951 osmc      20   0    7040   1532   1036 R   0.7  0.1   0:01.26 top
     1906 root      20   0       0      0      0 S   0.3  0.0   0:01.41 kworker/1:0
        1 root      20   0   26360   3564   2516 S   0.0  0.2   0:01.10 systemd
        2 root      20   0       0      0      0 S   0.0  0.0   0:00.00 kthreadd

And one additional piece, sorry for the multiple postings: the same OSMC release running on a Raspberry PI 3 B does not have this issue. Uptime reports low system load:

osmc@osmcf1:~$ uptime
 13:10:33 up 7 min,  1 user,  load average: 0.14, 0.22, 0.15
osmc@osmcf1:~$ uname -a
Linux osmcf1 4.19.55-6-osmc #1 SMP PREEMPT Sun Nov 3 22:15:28 UTC 2019 armv7l GNU/Linux
osmc@osmcf1:~$ cat /etc/os-release
PRETTY_NAME="Open Source Media Center"
VERSION="November 2019"


Notably, OSMC on my Raspberry PI 3B runs on kernel 4.19, while my Vero 4k device runs on kernel 3.14. “sudo apt-get dist-upgrade” on my Vero 4k device says that there is nothing to upgrade. Is that normal?

It doesn’t matter – it’s just caused by a mutex in the video decoder.


@joakim_s : AFAIK uptime for an idle system – the one that does not perform any active tasks for the moment - should almost always be less than one. Uptime of one or greater may mean that some application or a kernel thread is always active, which is unusual and is often a sign of a misconfigured or buggy application or kernel driver.

Kodi is always active. Shud down Kodi and the load average will fall. Here’s uptime from my laptop running Linux Mint with chromium and a few other apps running:

brian@auton ~ $ uptime
13:25:28 up 17 days, 16:27,  1 user,  load average: 1.10, 1.20, 0.98

@bmillham : how is it possible then that Raspberry PI 3B has Kodi running all the time, but its uptime while Kodi is idling is around 0.15?

@sam_nazarko : thank you for clarification. It looks to me like that mutex might be causing runaway system interrupts. Is there a way for me to install kernel 4.19 on Vero 4k to see if it makes any difference?

You are comparing two different pieces of hardware.

What issues are you experiencing, do you have some traces to suggest this?
There shouldn’t be any issues here.

Not at this time. The kernel would need to be adapted for the SoC. There is work underway for a 4.9 kernel update which will be released in the near future.


@sam_nazarko : thank you again. No, I am not experiencing any visible performance issues with this Vero 4k unit. It is just in my experience with various Linuxes the uptime of 1.0 or more is usually a sign of some problem. I do look forward to testing 4.19 kernel on this unit when the upgrade becomes available. There has been many improvements and bugfixes of the kernel between 3.14 and 4.19. Chances are that Vero 4k won’t have this issue anymore with 4.19.

The 4.19 kernel wouldn’t have this problem, but it wouldn’t have support for hardware video acceleration either, which is quite important.

The next kernel will be 4.9 based as it will be supported until 2023 (LTS).


Ouch. Hardware video acceleration is important. What about 4.9 on Vero 4k, is it going to support hardware video acceleration?


Awesome! Thank you very much @sam_nazarko. Your effort is very much appreciated.

After looking at this, this is an artificial load caused by a semaphore in the vdec code as suspected.

Nothing to worry about - but it may cause some confusion as it did here.

The CPU usage also appears a little higher because we are governing it until we do some real work.

I want to make one more comment about this. A load average of 1.21 on a quad core processor like the Vero has is not bad at all. You don’t have to worry about load average until it hits about 2.8.

To get a better idea of the LA, you need to divide the LA by the number of processors, so a LA of 1.21 on a quad core is like .3 on a single core system.

This page has a good explanation of LA: Understanding Linux CPU Load - when should you be worried? | Scout APM Blog