https://bugs.winehq.org/show_bug.cgi?id=42695
--- Comment #42 from Paul Gofman gofmanp@gmail.com --- (In reply to Nikolay Sivov from comment #41)
(In reply to Paul Gofman from comment #40)
CoInitializeSecurity() is a part of WMI query call, and happens after CoInitializeEx(), so this change won't matter. I don't think CoInitializeSecurity() is restricted like that, because it sets per-process info that overrides default parameters.
Well, I don't have any ole enabled trace to see the sequence of inits and uninits. I just saw CoInitializeSecurity() stub call, and checked that Wine stub succeeds unconditionally, while it fails on Windows witn CO_E_UNITIALIZED in this case. My wild guess was that if noone initialized COM in this thread before and this function will fail with the same error as on Windows instead of succeeding it may trigger some different path in the application. In any case a log might be helpful, so much discussions in this bug report and not a single relevant log :)