bitcoind crashes when opening wallet.dat second time without multiwallet mode #14657

Describe the issue

When bitcoind is loaded with the default wallet:

1. run loadwallet wallet.dat rpc command

The behavior can not be reproduced if wallet= directive exist in the config file (one entry is enough). In that case Wallet file verification failed: Error loading wallet wallet.dat. Duplicate -wallet filename specified. error message is correctly returned via RPC.

What behavior did you expect?

Bitcoind should not crash and return error.

What was the actual behavior (provide screenshots if the issue is GUI-related)?

bitcoin-qt crashes with (tested with bitcoind as well): bitcoin-qt: wallet/db.cpp:236: BerkeleyEnvironment::VerifyResult BerkeleyEnvironment::Verify(const string&, BerkeleyEnvironment::recoverFunc_type, std::__cxx11::string&): Assertion `mapFileUseCount.count(strFile) == 0' failed. Aborted (core dumped)

How reliably can you reproduce the issue, what are the steps to do so?

Could reproduce on 0.17.0 and 0.17.0.1. I would need someone to test 0.16.3, I recall that version did not have this issue. (that version has no dynamic wallet loading, as it does not know the loadwallet rpc)

What version of Bitcoin Core are you using, where did you get it (website, self-compiled, etc)?

0.17.0.1 -> self compiled

What type of machine are you observing the error on (OS/CPU and disk type)?

OS Ubuntu 18.04.1 LTS

I've been able to recreate. I compiled 0.17.0.1 on Ubuntu 18.04.1. Then did src/bitcoind, followed by src/bitcoin-cli loadwallet wallet.dat. This results in the same crash as above:

2018-11-05T13:52:48Z Using BerkeleyDB version Berkeley DB 4.8.30: (April 9, 2010)
2018-11-05T13:52:48Z Using wallet wallet.dat
bitcoind: wallet/db.cpp:236: BerkeleyEnvironment::VerifyResult BerkeleyEnvironment::Verify(const string&, BerkeleyEnvironment::recoverFunc_type, std::__cxx11::string&): Assertion `mapFileUseCount.count(strFile) == 0' failed.
Aborted (core dumped)

Duplicate of #14538, fixed by #14552.

Thanks @promag. Closing as a duplicate #14538.