check_sanity() scoped mutex locked after assertion

[lioriz]

In check_sanity() function,
In the scenario where multiple processes share the same shared memory, if the shared memory is corrupted for some reason then when one of the processes calls check_sanity() the scoped mutex in the function locks but the process crashes with assertion without unlocking the mutex, causing any other process that will call check_sanity() to be stuck on the locked scoped mutex.

If there is a mutex in a shared memory object it must be unlocked before assertion, exception, or return.
I would expect the check_sanity() function to return false in the case of corrupted memory instead of assert.

[lioriz]

opened PR 178 for this issue.

[igaztanaga]

Many thanks for the PR!

[lioriz]

Many thanks for the PR!

My pleasure.