Attention is currently required from: daniel, fixeria, laforge, osmith.
2 comments:
Patchset:
what I don't really understand here is why there is no code being removed. […]
The plan is to tackle this kind of task once these preliminary libosmo-netif changes have been submitted.
I guess printing the timestamp infront of the log should just be removed? from a quick glance this s […]
What we have observed was that sometimes, an arbitrary number of loop iterations is needed until the code section in question becomes active (each loop iteration contains one timestamp increment), and sometimes the code needs exactly the minimum number of loop needed.
I'm suspecting it's due to nondeterministic compiler behavior, probably caused by type punning (Vadim has also already seen a warning because of this). The reason for this assumption being that behavior never differed for the same binary and only changed between compilations, even for the same code (but with changed timestamps due to trying different commits etc. in between).
I will add a patch suggestion to get rid of the type punning, hoping we won't see this ever again.
To view, visit change 33198. To unsubscribe, or for help writing mail filters, visit settings.