Uh oh!
There was an error while loading. Please reload this page.
Destroy internals if created during Py_Finalize() - #892
Merged
Conversation
jagerman
commented
Jun 7, 2017
MemberAuthor
Cc @dean0x7d for confirmation that the fix looks good. |
jagermanforce-pushed
the
internals-destroy-fix
branch
from
June 7, 2017 18:25
f484d95 to
dfa4238Comparedean0x7d
reviewed
Jun 8, 2017
| // didn't exist above. (But we still use the above because if nothing calls get_internals() we | ||
| // might not have the static pointer set in get_internals_ptr()) | ||
| if (!internals_ptr_ptr) | ||
| internals_ptr_ptr = &detail::get_internals_ptr(); |
Member
There was a problem hiding this comment.
Good catch! I'd suggest moving up get_internals_ptr() to mirror the code from detail:: get_internals(). This is invariant with respect to Py_Finalize():
auto **internals_ptr_ptr = &detail::get_internals_ptr();
if (builtins.contains(id) && isinstance<capsule>(builtins[id]))
internals_ptr_ptr = capsule(builtins[id]);Edit: Originally suggested internals &*, but edited back to internals **.
Py_Finalize could potentially invoke code that calls `get_internals()`, which could create a new internals object if one didn't exist. `finalize_interpreter()` didn't catch this because it only used the pre-finalize interpreter pointer status; if this happens, it results in the internals pointer not being properly destroyed with the interpreter, which leaks, and also causes a `get_internals()` under a future interpreter to return an internals object that is wrong in various ways.
jagermanforce-pushed
the
internals-destroy-fix
branch
from
June 8, 2017 18:35
dfa4238 to
4238defCompare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Py_Finalizecould potentially invoke code that callsget_internals(),which could create a new internals object if one didn't exist.
finalize_interpreter()didn't catch this because it only used thepre-finalize interpreter pointer status; if this happens, it results in
the internals pointer not being properly destroyed with the interpreter,
which leaks, and also causes a
get_internals()under a futureinterpreter to return an internals object that is wrong in various ways.
The two lines added to
embed.hare the fix; the rest is just adding a test case that triggers and detects the issue.