Prospective fix for strange issue with invalid UTF-8 with test #494
Conversation
|
Test looking like it failed to write the thing I was writing to tmp over on CI. Figured something like that might occur. Will clean this up later presuming the general approach even seems sane here to fix the problem. |
|
I find your approach OK as it just allows listing opened files in additional (corner) use cases that you've provided. As we currently focus our efforts into the development of DNF5, I wouldn't develop the current solution further as it seems fixing your problematic cases. I guess that the CI failure could be related to passing the |
= changelog = msg: Avoid issue with garbage smaps chars in needs-restarting.py type: bugfix resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2212953 related: None
troglodyne
force-pushed
the
fix_utf8_smap_open
branch
from
475565a
to
50ffed7
Compare
|
Reworked commit, thanks for the input. Sorry to have missed the notification earlier. |
|
Hmm. The CI failures in the integration tests here don't really seem related to needs-restarting as far as I can tell. A reposync test failed, line 41563 of log, but nothing else that appears to be failing in the set. |
|
No, the failures of reposync are unrelated. I guess it's caused by some changes in createrepo_c which caused different set of metadata to be generated. Anyway, we need to fix the test suite. |
|
Everything looking good now, it seems :) |
For now this PR is only for purposes of showing the problem and one potential fix to it.
Related bug:
https://bugzilla.redhat.com/show_bug.cgi?id=2212953
I can fixup the commit later for formatting, etc. -- in the meantime I am going to try to gather more data from cPanel customers experiencing this to see just what the heck is in the smaps files for the failing cases.
Regardless, this appeared to me to be a real bug in the upstream code, so I figured I'd approach it in that manner.