4163093d63 wallet: use Mutex for g_sqlite_mutex instead of GlobalMutex (Vasil Dimov)
Pull request description:
Using `Mutex` provides stronger guarantee than `GlobalMutex` wrt Clang's
thread safety analysis. Thus it is better to reduce the usage of
`GlobalMutex` in favor of `Mutex`.
Using `Mutex` for `g_sqlite_mutex` is ok because its usage is limited in
`wallet/sqlite.cpp` and it does not require propagating the negative
annotations to not relevant code.
ACKs for top commit:
achow101:
ACK 4163093d63
hebasto:
re-ACK 4163093d63
TheCharlatan:
ACK 4163093d63
Tree-SHA512: 4913bcb8437ecf0e6b6cb781d02a6d24ffb4bf3e2e1899fa60785eab41c4c65dbdd9600bcb696290c873661b873ad61e5a4c4f205b7e66fdef2ae17c676cd12f
4aebd832a4 db: Change DatabaseCursor::Next to return status enum (Andrew Chow)
d79e8dcf29 wallet: Have cursor users use DatabaseCursor directly (Andrew Chow)
7a198bba0a wallet: Introduce DatabaseCursor RAII class for managing cursor (Andrew Chow)
69efbc011b Move SafeDbt out of BerkeleyBatch (Andrew Chow)
Pull request description:
Instead of having database cursors be tied to a particular `DatabaseBatch` object and requiring its setup and teardown be separate functions in that batch, we can have cursors be separate RAII classes. This makes it easier to create and destroy cursors as well as having cursors that have slightly different behaviors.
Additionally, since reading data from a cursor is a tri-state, this PR changes the return value of the `Next` function (formerly `ReadAtCursor`) to return an Enum rather than the current system of 2 booleans. This greatly simplifies and unifies the code that deals with cursors as now there is no confusion as to what the function returns when there are no records left to be read.
Extracted from #24914
ACKs for top commit:
furszy:
diff ACK 4aebd83
theStack:
Code-review ACK 4aebd832a4
Tree-SHA512: 5d0be56a18de5b08c777dd5a73ba5a6ef1e696fdb07d1dca952a88ded07887b7c5c04342f9a76feb2f6fe24a45dc31f094f1f5d9500e6bdf4a44f4edb66dcaa1
Next()'s result is a tri-state - failed, more to go, complete. Replace
the way that this is returned with an enum with values FAIL, MORE, and
DONE rather than with two booleans.
Instead of having DatabaseBatch deal with opening and closing database
cursors, have a separate RAII class that deals with those.
For now, DatabaseBatch manages DatabaseCursor, but this will change
later.
Using `Mutex` provides stronger guarantee than `GlobalMutex` wrt Clang's
thread safety analysis. Thus it is better to reduce the usage of
`GlobalMutex` in favor of `Mutex`.
Using `Mutex` for `g_sqlite_mutex` is ok because its usage is limited in
`wallet/sqlite.cpp` and it does not require propagating the negative
annotations to not relevant code.
No change in behavior. Just remove a little bit of code, reduce macro usage,
remove duplicative functions, and make BDB and SQLite implementations more
consistent with each other.
This commit does not change to any code and behavior. It it is easily reviewed
with the --color-moved=dimmed_zebra git diff option.
Motivation for this change is to:
- Consolidate redundant functions
IsBDBFile /ExistsBerkeleyDatabase / SplitWalletPath, and
IsSQLiteFile / ExistsSQLiteDatabase in the next commits
- Detect SQLite wallets consistently regardless whether bitcoin is built with
SQLite support in the next commits
- Avoid attempting to open SQLite databases with the BDB library when bitcoin
is built without SQLite support in the next commits
sqlite3 recommends that sqlite3_initialize be called when the
application starts, and sqlite3_shutdown when it stops. Since we don't
always use sqlite3, we initialize it when a SQLiteDatabse is constructed
(calling sqlite3_initialize after initialized is a no-op). We call
sqlite3_shutdown when we see that there are no databases opened. The
number of open databases is tracked by an atomic g_dbs_open.
