We have a heavily modularized application. When running with privileges to get sys-tracing, the query backlog can grow quite high.
The shutdown order looks something like:
Unload most modules.
Unload our profiling-tracy module (blocks until tracy::ShutdownProfiler() returns)
Unload the rest of our modules
Exit application
Unfortunately, if a SymbolCodeQuery query is processed for a module that unloaded, we usually end up in the CrashHandler and the application is hung at this point.
We have a heavily modularized application. When running with privileges to get sys-tracing, the query backlog can grow quite high.
The shutdown order looks something like:
tracy::ShutdownProfiler()returns)Unfortunately, if a SymbolCodeQuery query is processed for a module that unloaded, we usually end up in the CrashHandler and the application is hung at this point.