[3.11] gh-111942: Remove an extra incref in textiowrapper_change_encoding - #126542
[3.11] gh-111942: Remove an extra incref in textiowrapper_change_encoding#126542pablogsal merged 1 commit into
Conversation
Sorry, something went wrong.
…e_encoding There's an issue with the 3.11 backport commit e2421a3 The chane for the main branch was: ```diff + Py_INCREF(errors); ... Py_SETREF(self->encoding, encoding); - Py_SETREF(self->errors, Py_NewRef(errors)); + Py_SETREF(self->errors, errors); ``` but in 3.11 this became: ```diff + Py_INCREF(errors); ... Py_INCREF(errors); Py_SETREF(self->encoding, encoding); Py_SETREF(self->errors, errors); ``` i.e. there's an extra incref, but it looks -- at least to Git -- like the change that removes `Py_NewRef` was already applied. This was not caught because `test_io` refleaks tests were ineffective on 3.11 (see python#126478 (comment)). Remove the extraneous incref.
|
(stepping out for a bit, then I'll start the buildbots on this after GHA runs.) |
Sorry, something went wrong.
|
!buildbot AMD64 Windows11 Refleaks |
Sorry, something went wrong.
|
The failing builds are unstable (Alpine, Usan, Valgrind, FIPS, AIX), not meant for 3.10 (Android, iOS, WASI, NoGIL, bigmem/extralargefile), flaky (s390x RHEL8 Refleaks, PPC64LE Fedora), or show an unrelated problem (test_socket or Fedora). |
Sorry, something went wrong.
|
Thanks @encukou! I think it makes sense to land it so I went ahead |
Sorry, something went wrong.
|
Thank you! The buildbots have recovered :) |
Sorry, something went wrong.
A recent test fix uncovered a refleak that went in 3.11 on November due to a Git mis-merge.
So, @pablogsal, here's a RM decision for you. (Not time-sensitive, of course.)
This is not a security issue, but it should fix failing buildbots. (The other way to fix them would be to revert the change that uncovered the issue, which would mean we probably won't notice future refleaks in test_io.)
There's an issue with the 3.11 backport commit e2421a3 The change for the main branch was:
but in 3.11 this became:
+ Py_INCREF(errors); ... Py_INCREF(errors); Py_SETREF(self->encoding, encoding); Py_SETREF(self->errors, errors);i.e. there's an extra incref, but it looks -- at least to Git -- like the change that removes Py_NewRef was already applied.
This was not caught because test_io refleaks tests were ineffective on 3.11 (see #126478 (comment)).
Remove the extraneous incref.