bpo-31746: Prevent segfaults with uninitialised sqlite3.Connection objects
#27431
Conversation
| if (!self->initialized) { | ||
| pysqlite_state *state = pysqlite_get_state(NULL); | ||
| PyErr_SetString(state->ProgrammingError, | ||
| "Base Connection.__init__ not called."); | ||
| return NULL; | ||
| } |
There was a problem hiding this comment.
We cannot use pysqlite_check_connection here, since it also checks self->db. Calling close() twice should still be a no-op.
|
Backport to 3.9 as well? |
|
Thanks @erlend-aasland for the PR, and @pablogsal for merging it |
|
Sorry, @erlend-aasland and @pablogsal, I could not cleanly backport this to |
|
Thanks @erlend-aasland for the PR, and @pablogsal for merging it |
|
Sorry @erlend-aasland and @pablogsal, I had trouble checking out the |
|
Oh, seems that the backport party crashed :( @erlend-aasland can you do them manually? |
Yes, I was expecting it to fail. I'll fix it right away. Should we backport to 3.9 as well? |
|
@erlend-aasland apparently we have a buildbot failure: https://buildbot.python.org/all/#/builders/596/builds/703/ We may need to revert this meanwhile we understand what failed |
…alised (pythonGH-27431). (cherry picked from commit 7e311e4) Co-authored-by: Erlend Egeberg Aasland <erlend.aasland@innova.no>
|
Yes, please revert it. |
|
GH-27465 is a backport of this pull request to the 3.9 branch. |
|
We are missing the 3.10 backport still |
Doh, I'll get to it after morning coffee :) |
…ialised (pythonGH-27431). (cherry picked from commit 7e311e4) Co-authored-by: Erlend Egeberg Aasland <erlend.aasland@innova.no>
|
GH-27472 is a backport of this pull request to the 3.10 branch. |
Sync with main bco. pythonGH-27431
sqlite3.Connection.initializedwhen the connection objectis initialised
close()isolation_level[1] This may happen during connection init, so we cannot use
pysqlite_connection_commit, since it requires an initialisedconnection object. Previously, this was not a problem, since the
initialised flag was incorrectly set before the connection object was
initialised.
https://bugs.python.org/issue31746