I looked again into this last Friday and I did notice some more details that I did miss the first time I was investigating the scenario of the crash form the pernosco session: AtomSet::Get is creating a new AtomSet but it gets destroyed before it gets to the caller because MatchPattern::Init is being called late in the shutdown (I did put some details about what I did notice in the comment included in the patch attached in comment 6). It looked a legit scenario that we may have to account for, even if I have not been able to "force trigger it" locally, and so I opted to prepare a patch to deal with it, and eventually mark the patch as obsolete if (based on a feedback/review from the reviewer) it turns out that I did misread some of the details of the triggering scenario from the pernosco session and we should use a completely different approach to handle it.
Bug 1699298 Comment 7 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
I looked again into this last Friday and I did notice some more details that I did miss the first time I was investigating the scenario of the crash from the pernosco session: AtomSet::Get is creating a new AtomSet but it gets destroyed before it gets to the caller because MatchPattern::Init is being called late in the shutdown (I did put some details about what I did notice in the comment included in the patch attached in comment 6). It looked a legit scenario that we may have to account for, even if I have not been able to "force trigger it" locally, and so I opted to prepare a patch to deal with it, and eventually mark the patch as obsolete if (based on a feedback/review from the reviewer) it turns out that I did misread some of the details of the triggering scenario from the pernosco session and we should use a completely different approach to handle it.