Hi. Can anyone confirm whether frozen threads (e.g. UWP apps) can hang a minifilter unload for very long times? We have a situation that looks like it, but no dump or WinDbg access to confirm. Regards, Dejan.
Anyone from MS, hopefully.
Minifilter unload hangs are usually reference count leaks. Have you tried with Filter Manager Verifier enabled? Should crash at some point and show you what FltMgr is waiting on.
Scott,
This is one in 20 million user machines (and even more unloads), so any
debugging is a no go 
It’s not a total hang, it proceeds after some 2-4 hours. I have no other
clue what might cause something like it (some oplock in a PreOp due to file
name query being in progress?). But we have seen in our testing frozen
threads timing us out, only nothing this high.
We have some logging that shows us objects we use before the unload
starts (PortDisconnect), and they are all clear. But due to timing/races
(when “stop filtering” signal is set, some Pre/PostOp could already be in
progress), some new ones can always pop in that aren’t listed.
Dejan.