- Posts: 158
- Thank you received: 0
Built from source, can't get past "Using POSIX non-realtime" with Preempt_RT
6 years 5 months ago #168019
by jhandel
Howdy,
I am working on getting a local build going (so far so good--ish) Everything builds, all 230 tests pass.. However when I run the latency test I keep seeing "Using POSIX non-realtime" despite having an RT kernel..
here are the details on the setup:
OS: lubuntu 19.10
Processor: J3455
Machine: AWOW mini PC with 6gb memory and 128gb SSD ( www.amazon.com/Computer-AWOW-AK34-Ethern...d=1589680391&sr=8-18 )
I have attached a histogram from running rt-tests package cyclictest... I am no expert but that seems to be showing some pretty real-time type latency (to be fare I am a touch confused about the second peak... but way better than the 1,000,000 + I get from the latency-test running in POSIX non-realtime
Any help would be great, thanks all!
I am working on getting a local build going (so far so good--ish) Everything builds, all 230 tests pass.. However when I run the latency test I keep seeing "Using POSIX non-realtime" despite having an RT kernel..
here are the details on the setup:
OS: lubuntu 19.10
Code:
uname -a
Linux cnc-pc 5.6.10-rt5 #1 SMP PREEMPT_RT Sat May 16 14:17:03 CDT 2020 x86_64 x86_64 x86_64 GNU/Linux
Machine: AWOW mini PC with 6gb memory and 128gb SSD ( www.amazon.com/Computer-AWOW-AK34-Ethern...d=1589680391&sr=8-18 )
I have attached a histogram from running rt-tests package cyclictest... I am no expert but that seems to be showing some pretty real-time type latency (to be fare I am a touch confused about the second peak... but way better than the 1,000,000 + I get from the latency-test running in POSIX non-realtime
Any help would be great, thanks all!
Please Log in to join the conversation.
6 years 5 months ago #168020
by jhandel
Replied by jhandel on topic Built from source, can't get past "Using POSIX non-realtime" with Preempt_RT
Forgot to mention I was building the 2.8 branch..
Looks like a bug/choice on line 332 of uspace_common.h
PREEMPT_RT is actually the signature used..
Looks like it is fixed in 2.9... Any reason not to fix it in 2.8?
Looks like a bug/choice on line 332 of uspace_common.h
Code:
uname(&u);
crit1 = strcasestr (u.version, "PREEMPT RT") != 0;
PREEMPT_RT is actually the signature used..
Looks like it is fixed in 2.9... Any reason not to fix it in 2.8?
Please Log in to join the conversation.
Time to create page: 0.083 seconds
