+primiano, haraken who are working on slow reports / background tracing and oom crash reports.
This is an ongoing project and will certainly include IndexedDb case as well since it is part of memory-infra. They will be able to provide more context and timeline.

There is a medium term plan to have traces in crash and in the background tracing (slow reports) infrastructure. But this is after Q2.

> We should update crash reports to tell us how much each module is allocating so that we can differentiate these cases. 
The problem of this approach is: when shall we poll these values? we cannot poll at the time of the crash, as it is too late then.

A short term solution while we get there would be updating crash keys setting memory requested by IDB every time it allocates more memory, but that feels to me something that IDB owners should do.

Not sure what do you expect ssid@ to do here.

Glad to hear that there's a roadmap to resolving this! No need to prioritize a short term solution if there's a medium term plan in place. I just wanted to file the bug to make sure the work item isn't lost :)

Yup, I don't think this problem is specific to IDB. 
There is the general problem of: which component is the culprit for OOM crashes, and we need a more general solution there.