We’re developing both std and realtime applications that run on a RT-Linux.
question is, what would be an efficient way of logging application traces from both realtime and non-realtime processes?
By effecient, I mean that the process of logging application traces shouldn’t cause RT-perf hit by increasing latency, etc.
Traces should ideally be stored into a single file with timestamp, to make it easier to track interaction between processes.
Since you are on linux, you can use syslog() :
this means your program forwards messages to another program, which can be of low priority.
If you want something more fancy, then boost logging.