There may be a confused refcount issue and the drastic handling of GC at shutdown *is( a problem. (nsMsgService is shutdown before nsMsgDatabase and thus proper RemoveFromCache() cannot be done within nsMsgDatabase(), etc. Anyway, I could avoid heap-use-after-free with a patch in https://treeherder.mozilla.org/jobs?repo=try-comm-central&revision=756131c42270eb8398a046f139b1794f3b9436f4 So we can avoid the segfaulting and other nasty things that may happen due to the issue. But the build may cause leakage which may not be nice as in https://treeherder.mozilla.org/jobs?repo=try-comm-central&revision=756131c42270eb8398a046f139b1794f3b9436f4 The threshold of leak flagged as error is now meager 50KB. ( See https://thunderbird.topicbox.com/groups/developers/Tb6ce4cf69ec84306 ) I always prefer clean shutdown without segfault over a bit of known leak at shutdown time. However, I am doing a bit additional debugging and trying to see the leak can be reduced. https://treeherder.mozilla.org/jobs?repo=try-comm-central&revision=82d6e9a840f1aaf6bbca4e80c1f20c6a20f6d415
Bug 1677202 Comment 24 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
There may be a confused refcount issue and the drastic handling of GC at shutdown *is* a problem. (nsMsgService is shutdown before nsMsgDatabase and thus proper RemoveFromCache() cannot be done within nsMsgDatabase(), etc. Anyway, I could avoid heap-use-after-free with a patch in https://treeherder.mozilla.org/jobs?repo=try-comm-central&revision=756131c42270eb8398a046f139b1794f3b9436f4 So we can avoid the segfaulting and other nasty things that may happen due to the issue. But the build may cause leakage which may not be nice as in https://treeherder.mozilla.org/jobs?repo=try-comm-central&revision=756131c42270eb8398a046f139b1794f3b9436f4 The threshold of leak flagged as error is now meager 50KB. ( See https://thunderbird.topicbox.com/groups/developers/Tb6ce4cf69ec84306 ) I always prefer clean shutdown without segfault over a bit of known leak at shutdown time. However, I am doing a bit additional debugging and trying to see if the leak can be reduced. https://treeherder.mozilla.org/jobs?repo=try-comm-central&revision=82d6e9a840f1aaf6bbca4e80c1f20c6a20f6d415 EDIT: A few typos fixed.