protect storage writes

[TheL1ne]

We are creating multiple processors in a short time in our tests and they are failing sometimes because of concurrent map writes. We think we traced the problem to this point.

[ubunatic]

LGTM

[frairon]

Protecting that makes sense absolutely, but if you only protect that one code branch we might only buy some time and the problem will reappear later. So I suggest using the lock everywhere mStorages is used. Use RLock/RUnlock when reading only and Lock/Unlock for writing.

[ubunatic]

Protecting that makes sense absolutely, but if you only protect that one code branch we might only buy some time and the problem will reappear later. So I suggest using the lock everywhere mStorages is used. Use RLock/RUnlock when reading only and Lock/Unlock for writing.

Sounds great!
@TheL1ne Please implement it as proposed.

[TheL1ne]

@frairon did you meant it the way it is now?

I also did a similar thing for the in-memory storage defined outside of the tester because I think a similar problem could occur.