No other display server (or app?) hangs and/or crashes waiting for entropy. It's a bad user experience and obviously unacceptable.
I don't care if it "would fail audit", that misses the point. The bigger problem is that we currently fail basic usability. Mir needs to start up quickly and without crashing. Presently it does not do that.
Furthermore, Mir needs to keep responding quickly and without crashing. So to defer entropy collection only defers the problem without solving it.
This branch is the only way forward I can tell. If we've made heavy crypto a priority over a usable system, we've failed. It's not that the current design needs to pass security audit, but the current design is a failure if it can't pass a security audit without hanging indefinitely or crashing.
The logic is simple:
No other display server (or app?) hangs and/or crashes waiting for entropy. It's a bad user experience and obviously unacceptable.
I don't care if it "would fail audit", that misses the point. The bigger problem is that we currently fail basic usability. Mir needs to start up quickly and without crashing. Presently it does not do that.
Furthermore, Mir needs to keep responding quickly and without crashing. So to defer entropy collection only defers the problem without solving it.
This branch is the only way forward I can tell. If we've made heavy crypto a priority over a usable system, we've failed. It's not that the current design needs to pass security audit, but the current design is a failure if it can't pass a security audit without hanging indefinitely or crashing.