Fix crash in PDO_ODBC statement dtor #17586
Conversation
Port of 2ae897f to PDO_ODBC.
|
(For reference, #17567 is the function I'll introduce in master, I'll rebase after this is merged) |
| // TODO: Factor this out; pg/mysql/firebird do the same thing | ||
| bool server_obj_usable = !Z_ISUNDEF(stmt->database_object_handle) | ||
| && IS_OBJ_VALID(EG(objects_store).object_buckets[Z_OBJ_HANDLE(stmt->database_object_handle)]) | ||
| && !(OBJ_FLAGS(Z_OBJ(stmt->database_object_handle)) & IS_OBJ_FREE_CALLED); |
There was a problem hiding this comment.
This is super ugly. Wouldn't it be cleaner if the dbh would keep a list of stmts, so it could null stmt->H when being destroyed?
There was a problem hiding this comment.
That's more convoluted and consumes more memory, and the dbh destruction (at the driver site) already destroys the driver statements too. Imo, this is good enough especially when hidden behind a function.
There was a problem hiding this comment.
Fair enough!
Could it make sense to move that into Zend/ to have a generic function? I think that not only PDO is affected by this; likely also other DB extensions, and even others (enchant broker/dictinonary comes to mind)?
There was a problem hiding this comment.
enchant shouldn't be affected as it does not create strong cycles and even does its own refcounting of the broker IIRC.
Other DB extension: could be affected indeed.
* PHP-8.3: Fix crash in PDO_ODBC statement dtor (#17586)
* PHP-8.4: Fix crash in PDO_ODBC statement dtor (#17586)
Port of 2ae897f to PDO_ODBC. (GH-17585 is this, but for master instead of 8.3, targeting against correct version)