aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorJoshua Watt <jpewhacker@gmail.com>2018-12-03 21:42:30 -0600
committerRichard Purdie <richard.purdie@linuxfoundation.org>2018-12-07 12:38:14 +0000
commite8b9d3f534ef404780be23b601d5a4bb9cec928a (patch)
tree159f982eadeac04237e14d5174b28da64879722f
parentb5686fc07d06e1f0a55161b5ebcec6552d74ceff (diff)
downloadbitbake-e8b9d3f534ef404780be23b601d5a4bb9cec928a.tar.gz
persist_data: Fix leaking cursors causing deadlock
The original implementation of persistent data executed all SQL statements via sqlite3.Connection.execute(). Behind the scenes, this function created a sqlite3 Cursor object, executed the statement, then returned the cursor. However, the implementation did not account for this and failed to close the cursor object when it was done. The cursor would eventually be closed when the garbage collector got around to destroying it. However, sqlite has a limit on the number of cursors that can exist at any given time, and once this limit is reached it will block a query to wait for a cursor to be destroyed. Under heavy database queries, this can result in Python deadlocking with itself, since the SQL query will block waiting for a free cursor, but Python can no longer run garbage collection (as it is blocked) to free one. This restructures the SQLTable class to use two decorators to aid in performing actions correctly. The first decorator (@retry) wraps a member function in the retry logic that automatically restarts the function in the event that the database is locked. The second decorator (@transaction) wraps the function so that it occurs in a database transaction, which will automatically COMMIT the changes on success and ROLLBACK on failure. This function additionally creates an explicit cursor, passes it to the wrapped function, and cleans it up when the function is finished. Note that it is still possible to leak cursors when iterating. This is much less frequent, but can still be mitigated by wrapping the iteration in a `with` statement: with db.iteritems() as it: for (k, v) in it: ... As a side effect, since most statements are wrapped in a transaction, setting the isolation_level when the connection is created is no longer necessary. [YOCTO #13030] Signed-off-by: Joshua Watt <JPEWhacker@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
-rw-r--r--lib/bb/persist_data.py188
1 files changed, 135 insertions, 53 deletions
diff --git a/lib/bb/persist_data.py b/lib/bb/persist_data.py
index bef701861..1a6319f94 100644
--- a/lib/bb/persist_data.py
+++ b/lib/bb/persist_data.py
@@ -29,6 +29,7 @@ import warnings
from bb.compat import total_ordering
from collections import Mapping
import sqlite3
+import contextlib
sqlversion = sqlite3.sqlite_version_info
if sqlversion[0] < 3 or (sqlversion[0] == 3 and sqlversion[1] < 3):
@@ -45,75 +46,154 @@ if hasattr(sqlite3, 'enable_shared_cache'):
@total_ordering
class SQLTable(collections.MutableMapping):
+ class _Decorators(object):
+ @staticmethod
+ def retry(f):
+ """
+ Decorator that restarts a function if a database locked sqlite
+ exception occurs.
+ """
+ def wrap_func(self, *args, **kwargs):
+ count = 0
+ while True:
+ try:
+ return f(self, *args, **kwargs)
+ except sqlite3.OperationalError as exc:
+ if 'is locked' in str(exc) and count < 500:
+ count = count + 1
+ self.connection.close()
+ self.connection = connect(self.cachefile)
+ continue
+ raise
+ return wrap_func
+
+ @staticmethod
+ def transaction(f):
+ """
+ Decorator that starts a database transaction and creates a database
+ cursor for performing queries. If no exception is thrown, the
+ database results are commited. If an exception occurs, the database
+ is rolled back. In all cases, the cursor is closed after the
+ function ends.
+
+ Note that the cursor is passed as an extra argument to the function
+ after `self` and before any of the normal arguments
+ """
+ def wrap_func(self, *args, **kwargs):
+ # Context manager will COMMIT the database on success,
+ # or ROLLBACK on an exception
+ with self.connection:
+ # Automatically close the cursor when done
+ with contextlib.closing(self.connection.cursor()) as cursor:
+ return f(self, cursor, *args, **kwargs)
+ return wrap_func
+
"""Object representing a table/domain in the database"""
def __init__(self, cachefile, table):
self.cachefile = cachefile
self.table = table
- self.cursor = connect(self.cachefile)
-
- self._execute("CREATE TABLE IF NOT EXISTS %s(key TEXT, value TEXT);"
- % table)
-
- def _execute(self, *query):
- """Execute a query, waiting to acquire a lock if necessary"""
- count = 0
- while True:
- try:
- return self.cursor.execute(*query)
- except sqlite3.OperationalError as exc:
- if 'database is locked' in str(exc) and count < 500:
- count = count + 1
+ self.connection = connect(self.cachefile)
+
+ self._execute_single("CREATE TABLE IF NOT EXISTS %s(key TEXT, value TEXT);" % table)
+
+ @_Decorators.retry
+ @_Decorators.transaction
+ def _execute_single(self, cursor, *query):
+ """
+ Executes a single query and discards the results. This correctly closes
+ the database cursor when finished
+ """
+ cursor.execute(*query)
+
+ @_Decorators.retry
+ def _row_iter(self, f, *query):
+ """
+ Helper function that returns a row iterator. Each time __next__ is
+ called on the iterator, the provided function is evaluated to determine
+ the return value
+ """
+ class CursorIter(object):
+ def __init__(self, cursor):
+ self.cursor = cursor
+
+ def __iter__(self):
+ return self
+
+ def __next__(self):
+ row = self.cursor.fetchone()
+ if row is None:
self.cursor.close()
- self.cursor = connect(self.cachefile)
- continue
- raise
+ raise StopIteration
+ return f(row)
+
+ def __enter__(self):
+ return self
+
+ def __exit__(self, typ, value, traceback):
+ self.cursor.close()
+ return False
+
+ cursor = self.connection.cursor()
+ try:
+ cursor.execute(*query)
+ return CursorIter(cursor)
+ except:
+ cursor.close()
def __enter__(self):
- self.cursor.__enter__()
+ self.connection.__enter__()
return self
def __exit__(self, *excinfo):
- self.cursor.__exit__(*excinfo)
-
- def __getitem__(self, key):
- data = self._execute("SELECT * from %s where key=?;" %
- self.table, [key])
- for row in data:
+ self.connection.__exit__(*excinfo)
+
+ @_Decorators.retry
+ @_Decorators.transaction
+ def __getitem__(self, cursor, key):
+ cursor.execute("SELECT * from %s where key=?;" % self.table, [key])
+ row = cursor.fetchone()
+ if row is not None:
return row[1]
raise KeyError(key)
- def __delitem__(self, key):
+ @_Decorators.retry
+ @_Decorators.transaction
+ def __delitem__(self, cursor, key):
if key not in self:
raise KeyError(key)
- self._execute("DELETE from %s where key=?;" % self.table, [key])
+ cursor.execute("DELETE from %s where key=?;" % self.table, [key])
- def __setitem__(self, key, value):
+ @_Decorators.retry
+ @_Decorators.transaction
+ def __setitem__(self, cursor, key, value):
if not isinstance(key, str):
raise TypeError('Only string keys are supported')
elif not isinstance(value, str):
raise TypeError('Only string values are supported')
- data = self._execute("SELECT * from %s where key=?;" %
- self.table, [key])
- exists = len(list(data))
- if exists:
- self._execute("UPDATE %s SET value=? WHERE key=?;" % self.table,
- [value, key])
+ cursor.execute("SELECT * from %s where key=?;" % self.table, [key])
+ row = cursor.fetchone()
+ if row is not None:
+ cursor.execute("UPDATE %s SET value=? WHERE key=?;" % self.table, [value, key])
else:
- self._execute("INSERT into %s(key, value) values (?, ?);" %
- self.table, [key, value])
-
- def __contains__(self, key):
- return key in set(self)
-
- def __len__(self):
- data = self._execute("SELECT COUNT(key) FROM %s;" % self.table)
- for row in data:
+ cursor.execute("INSERT into %s(key, value) values (?, ?);" % self.table, [key, value])
+
+ @_Decorators.retry
+ @_Decorators.transaction
+ def __contains__(self, cursor, key):
+ cursor.execute('SELECT * from %s where key=?;' % self.table, [key])
+ return cursor.fetchone() is not None
+
+ @_Decorators.retry
+ @_Decorators.transaction
+ def __len__(self, cursor):
+ cursor.execute("SELECT COUNT(key) FROM %s;" % self.table)
+ row = cursor.fetchone()
+ if row is not None:
return row[0]
def __iter__(self):
- data = self._execute("SELECT key FROM %s;" % self.table)
- return (row[0] for row in data)
+ return self._row_iter(lambda row: row[0], "SELECT key from %s;" % self.table)
def __lt__(self, other):
if not isinstance(other, Mapping):
@@ -122,25 +202,27 @@ class SQLTable(collections.MutableMapping):
return len(self) < len(other)
def get_by_pattern(self, pattern):
- data = self._execute("SELECT * FROM %s WHERE key LIKE ?;" %
- self.table, [pattern])
- return [row[1] for row in data]
+ return self._row_iter(lambda row: row[1], "SELECT * FROM %s WHERE key LIKE ?;" %
+ self.table, [pattern])
def values(self):
return list(self.itervalues())
def itervalues(self):
- data = self._execute("SELECT value FROM %s;" % self.table)
- return (row[0] for row in data)
+ return self._row_iter(lambda row: row[0], "SELECT value FROM %s;" %
+ self.table)
def items(self):
return list(self.iteritems())
def iteritems(self):
- return self._execute("SELECT * FROM %s;" % self.table)
+ return self._row_iter(lambda row: (row[0], row[1]), "SELECT * FROM %s;" %
+ self.table)
- def clear(self):
- self._execute("DELETE FROM %s;" % self.table)
+ @_Decorators.retry
+ @_Decorators.transaction
+ def clear(self, cursor):
+ cursor.execute("DELETE FROM %s;" % self.table)
def has_key(self, key):
return key in self
@@ -195,7 +277,7 @@ class PersistData(object):
del self.data[domain][key]
def connect(database):
- connection = sqlite3.connect(database, timeout=5, isolation_level=None)
+ connection = sqlite3.connect(database, timeout=5)
connection.execute("pragma synchronous = off;")
connection.text_factory = str
return connection