Make SetBestChain() atomic #2058
Conversation
|
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/57b7b8137e23d877bd4d6d8f750ed1a87b65681d for binaries and test log. |
|
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/11c2cb7a7f8ed2234ccbb7a1eb25135afc518eed for binaries and test log. |
|
ACK |
In case a reorganisation fails, the internal state could become inconsistent (memory only). Previously, a cache per block connect or disconnect action was used, so blocks could not be applied in a partial way. Extend this to a cache for the entire reorganisation, making it atomic entirely. This also simplifies the code a bit.
|
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/d33a9218abda80d32239789f6ceaec6d68a290ce for binaries and test log. |
Make SetBestChain() atomic
* move nStart closer to used and initialize it
* Quells warning displayed with `-Wmaybe-uninitialized` flag
* remove unused mnEntries vectors
* add devnet magic bytes to mininode.py
* remove unused governance-misc.h file
* remove old TODO comments
* replace map count/insert w/emplace in instantx.cpp
This feels like an optimization, something about:
```
if map.count(key)
return false
else
insert(pair)
return true
```
... just feels icky. Plus, `vote.GetMasternodeOutpoint()` is only called
once. The previous version might be slightly more readable, however.
* Revert "replace map count/insert w/emplace in instantx.cpp"
This reverts commit d7f876e.
3 participants
In case a reorganisation fails, the internal state could become inconsistent (memory only). Previously, a cache per block connect or disconnect action was used, so blocks could not be applied in a partial way. Extend this to a cache for the entire reorganisation, making it atomic entirely. This also simplifies the code a bit.
I haven't seen any reports of problems because of this, but it should be fixed.