I believe this is just another instance of bug 1305298 or bug 1245239. Essentially those bugs are probably some various of the following - Linux overcommit behavior of shmems mean that we mmap in the shmem, and everything seems fine at the time. Then when we finally go to clear/initialize it, the kernel is finally forced to map the page and this causes it to OOM (which manifests as a SIGBUS). I am not sure we have any particularly good resolutions here... Looking through crash counts, it seems this behavior hit spikes on both 68.2esr, 68beta6, and 70.0 release. Looking at the release calendar, those happened about the same time. 68.2esr and 68b6 seem like a clue, though, since 68.1esr/68.0esr don't seem to report the same spike. Jeff, do you remember anything suspicious we did around that time?
Bug 1631680 Comment 5 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
I believe this is just another instance of bug 1305298 or bug 1245239. Essentially those bugs are probably some various of the following - Linux overcommit behavior of shmems mean that we mmap in the shmem, and everything seems fine at the time. Then when we finally go to clear/initialize it, the kernel is finally forced to map the page and this causes it to OOM (which manifests as a SIGBUS). I am not sure we have any particularly good resolutions here... Looking through crash counts, it seems this behavior hit spikes on both 68.2esr and and 70.0 release. Looking at the release calendar, those happened about the same time. 68.2esr seems like a clue, though, since 68.1esr/68.0esr don't seem to report the same spike. Jeff, do you remember anything suspicious we did around that time?