Thread deadlock in 9.83

[vgough]

Encfs recently changed to use easylogging++, and a thread deadlock has been reported in the logging library: vgough/encfs#244

The referenced issue has the relevant stack traces from ThreadSanitizer. Can you confirm that this is an issue w/ easylogging++?

WARNING: ThreadSanitizer: lock-order-inversion (potential deadlock) (pid=3933)
Cycle in lock order graph: M11 (0x7d480000be90) => M13 (0x7d4400009d90) => M17 (0x7d580000fc10) => M11

[abumq]

This is possibly fixed in v9.91.

See dd12e89

Please re-open if it is not with updated ThreadSanitizer's output. Please update to latest version.

[abumq]

Re-opening as i found an issue with this in the fix

[abumq]

@aparajita do you have time to look in to this? NOTE: Please use develop branch and use pthread.cpp sample for re-producing point (unless you want to write your own)

[aparajita]

@mkhan3189 Without looking deeply at the code, I wonder if you might be a recursive lock on the mutex within the same thread, in which case you need to use a recursive mutex.