The page cache implementation uses a pre-allocated vector (`entries`)
with fixed capacity, along with a custom hash map and freelist. This
design requires expensive upfront allocation when creating a new
connection, which severely impacted performance in workloads that open
many short-lived connections (e.g., our concurrent write benchmarks that
create a new connection per transaction).
Therefore, replace the pre-allocated vector with an intrusive doubly-
linked list. This eliminates the page cache initialization overhead from
connection establishment, but also reduces memory usage to entries that
are actually used. Furthermore, the approach allows us to grow the page
cache with much less overhead.
The patch improves concurrent write throughput benchmark by 4x for
single-threaded performance.
Before:
```
$ write-throughput --threads 1 --batch-size 100 -i 1000 --mode concurrent
Running write throughput benchmark with 1 threads, 100 batch size, 1000 iterations, mode: Concurrent
Database created at: write_throughput_test.db
Thread 0: 100000 inserts in 3.82s (26173.63 inserts/sec)
```
After:
```
$ write-throughput --threads 1 --batch-size 100 -i 1000 --mode concurrent
Running write throughput benchmark with 1 threads, 100 batch size, 1000 iterations, mode: Concurrent
Database created at: write_throughput_test.db
Thread 0: 100000 inserts in 0.90s (110848.46 inserts/sec)
```
Closes#3456
The page cache implementation uses a pre-allocated vector (`entries`)
with fixed capacity, along with a custom hash map and freelist. This
design requires expensive upfront allocation when creating a new
connection, which severely impacted performance in workloads that open
many short-lived connections (e.g., our concurrent write benchmarks that
create a new connection per transaction).
Therefore, replace the pre-allocated vector with an intrusive
doubly-linked list. This eliminates the page cache initialization
overhead from connection establishment, but also reduces memory usage to
entries that are actually used. Furthermore, the approach allows us to
grow the page cache with much less overhead.
The patch improves concurrent write throughput benchmark by 4x for
single-threaded performance.
Before:
```
$ write-throughput --threads 1 --batch-size 100 -i 1000 --mode concurrent
Running write throughput benchmark with 1 threads, 100 batch size, 1000 iterations, mode: Concurrent
Database created at: write_throughput_test.db
Thread 0: 100000 inserts in 3.82s (26173.63 inserts/sec)
```
After:
```
$ write-throughput --threads 1 --batch-size 100 -i 1000 --mode concurrent
Running write throughput benchmark with 1 threads, 100 batch size, 1000 iterations, mode: Concurrent
Database created at: write_throughput_test.db
Thread 0: 100000 inserts in 0.90s (110848.46 inserts/sec)
```
not sure how these would even work with mvcc - either way, an ephemeral
table use an ephemeral database file and pager so i don't think putting
its writes into MV store makes sense
TBH i have no idea if there are any weird interactions here but the code
we have now for sure does not work
Closes#3486
Reviewed-by: Nikita Sivukhin (@sivukhin)
Closes#3490
Consolidates the `exec_trim`, `exec_rtrim`, `exec_ltrim` code and only
pattern matches on whitespace character.
Fixes#3319
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3437
There were 2 problems:
1. The SELECT wasn't propagating which register it used for its results,
so sometimes the INSERT read bad data.
2. `TableReferences::contains_table` was only checking the top-level
tables, not the nested tables in FROM queries. This condition is used to
emit "template 4", the bytecode template for self-inserts.
Closes https://github.com/tursodatabase/turso/issues/3312
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3436
Sqlite has a crazy easter egg where a 1 Gib file offset, it creates a
`PENDING_BYTE_PAGE` that is used only by the VFS layer, and is never
read or written into.
To properly test this, I took inspiration from SQLITE testing framework,
and defined a helper method, that is conditionally compiled with the
`test_helper` feature enabled.
https://github.com/sqlite/sqlite/blob/7e38287da43ea3b661da3d8c1f431aa907
d648c9/src/main.c#L4327
As the `PENDING_BYTE` is normally at the 1 Gib mark, I created a
function that modifies the static `PENDING_BYTE` atomic to whatever
value we want. This means we can test this unusual behaviours at any DB
file size we want.
`fuzz_pending_byte_database` is the test that fuzzes different pending
byte offsets and does an integrity check at the end to confirm, we are
compatible with SQLITE
Closes#2749
<img width="1100" height="740" alt="image" src="https://github.com/user-
attachments/assets/06eb258f-b4b4-47bf-85f9-df1cf411e1df" />
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3431
**Handle table ID / rootpages properly for both checkpointed and non-
checkpointed tables**
Table ID is an opaque identifier that is only meaningful to the MV
store.
Each checkpointed MVCC table corresponds to a single B-tree on the
pager,
which naturally has a root page.
**We cannot use root page as the MVCC table ID directly because:**
- We assign table IDs during MVCC commit, but
- we commit pages to the pager only during checkpoint
which means the root page is not easily knowable ahead of time.
**Hence:**
- MVCC table ids are always negative
- sqlite_schema rows will have a negative rootpage column if the
table has not been checkpointed yet.
- on checkpoint when the table is allocated a real root page, we update
the row in sqlite_schema and in MV store's internal mapping
**On recovery:**
- All sqlite_schema tables are read directly from disk and assigned
`table_id = -1 * root_page` -- root_page on disk must be positive
- Logical log is deserialized and inserted into MV store
- Schema changes from logical_log are captured into the DB's global
schema
**Note about recovery:**
I changed MVCC recovery to happen on DB initialization which should
prevent any races, so no need for `recover_lock`, right @pereman2 ?
Closes#3419
This PR implements support for `ON CONFLICT` clause chain, e.g.
```
INSERT INTO ct(id, x, y) VALUES (4, 'x', 'y1'), (5, 'a1', 'b'), (3, '_', '_')
ON CONFLICT(x) DO UPDATE SET x = excluded.x || '-' || x, y = excluded.y || '@' || y, z = 'x'
ON CONFLICT(y) DO UPDATE SET x = excluded.x || '+' || x, y = excluded.y || '!' || y, z = 'y'
ON CONFLICT DO UPDATE SET x = excluded.x || '#' || x, y = excluded.y || '%' || y, z = 'fallback';
```
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Reviewed-by: Preston Thorpe <preston@turso.tech>
Closes#3453
As per SQLite API, sqlite3_reset() does *not* clear bind parameters.
Instead they're persistent across statement reset and only cleared with
sqlite3_clear_bindings().
Reviewed-by: Avinash Sajjanshetty (@avinassh)
Closes#3466
As per SQLite API, sqlite3_reset() does *not* clear bind parameters.
Instead they're persistent across statement reset and only cleared with
sqlite3_clear_bindings().
This PR bundles 2 fixes:
1. Index search must skip NULL values
2. UPDATE must avoid using index which column is used in the SET clause
* This was an optimization to not do full scan in case of `UPDATE t
SET ... WHERE col = ?` but instead of doing this hacks we must properly
load updated row set to the ephemeral index and flush it after update
will be finished instead of modifying BTree inplace
* So, for now we completely remove this optimization and quitely
wait for proper optimization to land
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3459
We had encryption feature behind a compiler flag. However, it wasn't
enabled by default. This patch:
- enables compiler flag by default
- it also adds an opt in runtime flag `experimental-encryption`
- the runtime flag is disabled by default
Closes#3457
Table ID is an opaque identifier that is only meaningful to the MV store.
Each checkpointed MVCC table corresponds to a single B-tree on the pager,
which naturally has a root page.
We cannot use root page as the MVCC table ID directly because:
- We assign table IDs during MVCC commit, but
- we commit pages to the pager only during checkpoint
which means the root page is not easily knowable ahead of time.
Hence, we:
- store the mapping between table id and btree rootpage
- sqlite_schema rows will have a negative rootpage column if the
table has not been checkpointed yet.
This PR auto-assign ids for anonymous variables straight into parser.
Otherwise - it's pretty easy to mess up with traversal order in the core
code and assign ids incorrectly.
For example, before the fix, following code worked incorrectly because
parameter values were assigned first to conflict clause instead of
values:
```rs
let mut stmt = conn.prepare("INSERT INTO test VALUES (?, ?), (?, ?) ON CONFLICT DO UPDATE SET v = ?")?;
stmt.bind_at(1.try_into()?, Value::Integer(1));
stmt.bind_at(2.try_into()?, Value::Integer(20));
stmt.bind_at(3.try_into()?, Value::Integer(3));
stmt.bind_at(4.try_into()?, Value::Integer(40));
stmt.bind_at(5.try_into()?, Value::Integer(66));
```
Closes#3455
- even if index search will return only 1 row - it will call next in the loop - and we incorrecty can process same row values multiple times
- the following query failed with this optimization:
turso> CREATE TABLE t (id INTEGER PRIMARY KEY AUTOINCREMENT, k TEXT, c0 INT);
turso> CREATE UNIQUE INDEX idx_p1_0 ON t(c0);
turso> insert into t values (null, 'uu', -1);
turso> insert into t values (null, 'uu', -2);
turso> UPDATE t SET c0 = NULL WHERE c0 = -1;
turso> SELECT * FROM t
┌────┬────┬────┐
│ id │ k │ c0 │
├────┼────┼────┤
│ 1 │ uu │ │
├────┼────┼────┤
│ 2 │ uu │ │
└────┴────┴────┘
