There is a revertable quite easy way to replace the librtlsdr …
It’s been linked in this thread and fixes the issue no matter which kernel you run.
Apart from manually reverting, Raspbian increments the library version (and probably fixes the issues in librtlsdr by not using zerocopy), the manual intervention will also no longer be relevant.
After the upgrade i’ve realized a small CPU increase, mainly driven by “USB”. Are you aware of this? Nothing critical, just a side note (screenshot of the graph, change at 10:46)
Not that it matters, but there is an apparently harmless difference in your compiled binaries (thanks for doing that!). Below is my CPU utilization first on the older kernel with RPi compiled binaries, then the first blip is after reboot with still the old kernel with your binaries, the final blip is your binaries with the latest kernel from sudo apt dist-upgrade.
I have CPU scaling governor set to Performance on this RPi so ADS-B CPU utilization usually only changes from gain changes or aircraft traffic spikes.I double-check there were no gain changes. (This is using the FA Blue + stick)
I’ve been using the latest dump1090 dev branch on my secondary/test/I’m-bored station for a day and half without issues. The USB CPU utilization, on latest kernel, is down by 0.3%, to where it was prior to compiling my own librtlsdr .
Just want to say this worked for me as well, thank you. Exact same problem with RPi 3B+, FA Pro Plus stick, and latest install of Raspbian Buster Lite.
I came into the ADSB arena just last week with a Pi Zero W loaded with the FA Image plus FR24 and ADSBExchange.
Apparently with such configuration I’m in the edge of the Pi Zero limits. I experimented some clock inestability and applying the libsdr patch everything appears to work well now, but by reading this thread I’m now a little bit confused about what should be the proper dump1090-fa CPU usage.
I read some people saying their CPU usage (for dump1090-fa) climbed to 40% from 10% due to recent kernel updates.
My question is what CPU % is appropiate for the Pi Zero, I understand that because of it’s reduced specs, a standard 10% usage in a Pi4 maybe equivalent to a 40% usage in a PiZero.
I’m getting a 40% usage (top) on my PiZero, and I don’t know if it’s what it should be or if I should try to revert my kernel to lower it even more.
(BTW, I’m on: Linux piaware 5.4.51+ #1333 Mon Aug 10 16:38:02 BST 2020 armv6l GNU/Linux)
That sounds about right with those services running. So long as you are no longer experiencing timing/MLAT issues, you’re good to go. One thing to consider with the PiZ is the USB connection/power - that could also be an issue. I think this is still valid on the PiZero and may help on that end.
sudo sh -c "echo 'max_usb_current=1' >> /boot/config.txt"
.
This also may or may not also help (some will argue both ways):
I also tend to disable as many other services as possible (wifi/bt if not being used, etc)
I also always use htop instead of top - easier to read in my opinion and sortable.
Traffic and gain will also have an effect on CPU use, etc. It’s been quite awhile since I messed with the Zero with PiAware, but sounds like you are on the right track. As you get more comfortable, you can move to installing the Raspbian Lite (or whatever they choose to call the OS these days), strip it down some and then go the package install route for even more efficiency, but there is a learning curve involved.
Above all, have some fun - it’s great learning on the side.