No need to patch, compile or otherwise change anything in code, you can adjust oom kill likelyhood per-process using a sysctl tunable, or better let systemd do it for you.
Of you search some more you will find the issue is not that simple though, as there are several different memory pressure systems in place on SFOS, (the kernel oom, the mce one, the Android one at least) and finding the right tuning for all of them isn’t trivial.
That being said, I have yet to see the proof that OOM killing has anything to do with the problem.
People seem to tend to attribute a lot of things to the OOM killer without really checking that is was to blame.
I’ve been trying to find out what causes this for a week now and:
Is this device related? Is there a phone I can get where Sailfish OS is usable? Is anyone working on fixing this?
Yes, never had this issue on all XA2 Platform, have on 10 III.
XA2 imo is best suitable, but phones are underpowered
Only some part of community. Jolla couldn’t reproduce this bug, like many others. Wouldn’t count on official/community fix anytime soon
Thank you for the answers. Should we offer Jolla an Xperia 10 III? Do Jolla employees use Sailfish OS phones?
This might be related:
Yes, it looks like the same issue.
For me it consistently happens in minutes with some apps.
Could it be worse with my carrier (PT NOS)? / 4G coverage? / the weather?
The place I live in is on the opposite side in Europe from Finland and (also) quite beautiful. Boat to the island every 10 minutes costing 1 Eur. Good beer for 50c.
Happening to me a lot of time on Xperia 10 on 4.5.0.19
Whatsapp voice notes doesn’t play until a reboot.
I think I accidentally found a way of keeping the audio from failing: Playing some music with the Unplayer music player immediately after a reboot. Then audio worked for a full 5 days without failure (I just had to reboot today because I removed the cards from the phone)
I had the same problem with XA2 and now 10III. Really annoying.
Is this Portugal beach. Seems really nice!
How do you play audio before the incident? Are you connecting headphones, bluetooth? What devices?
if you connected bluetooth headphones, do they show up in pactl list sinks? Also chekc if the volume is above 0 in the output.
If you do a pacmd info what sink is listed for the sink input? Also check volume of the sink-input.
Example:
[...snip...]
1 sink input(s) available.
index: 56
driver: <protocol-native.c>
flags: START_CORKED
state: RUNNING
sink: 6 <bluez_sink.78_2B_64_CD_39_3E.a2dp_sink>
[...snip...]
If there is nothing suspicious, we got a sink and a source connected to the sink. Both got volume. then i would expect the problem to be outside of pulseaudio
Maybe not related but anyway:
Since some version in 4.5 cycle my Bluetooth headphones are never listed there.
but they still work?
They do. But it’s not possible to boost to low max volume by:
pactl set-sink-volume ch# 200%
as ch# is not listed anymore.
Does anyone know, if this is fixed with latest version (4.5.0.21)?
I don’t think so. I still have it on .21 on my Xperia 10 III.
I’m on 4.5.0.21 and to me still occurs the bug after few days/weeks (just now I rebooted the phone 
), and I didn’t use GPS yesterday/today 
I’ve been traveling for the last 3 months and:
It’s not related specifically to GPS usage, I’ve been using a second phone for GPS.