Per discussion with Eric, we were meant to also collect data on slow scripts in the parent, but we don't right now.

We'd want to record to the event telemetry added in bug 1652613, but from the code in the nsGlobalWindowInner slow script handler, only in the parent process. But we want to collect this information even when the dialog is disabled (the new default), so we'd need to adjust this code: https://searchfox.org/mozilla-central/rev/89d33e1c3b0a57a9377b4815c2f4b58d933b7c32/js/xpconnect/src/XPCJSContext.cpp#639-642

However... we'd be adding telemetry on the main thread. But the main thread might never recover, if it's permanently hung. So we wouldn't get any information about permanent hangs of this sort, only about ones that are bad, but where we eventually recover (at least enough to send a telemetry ping) - not bad enough that the user or OS ends up just killing the process. Eric, with that caveat, are you still interested in this data? (I'm pretty sure the answer will be "yes" but I figured I should be clear from the start...)

Yes, I'm still interested in the data.

Gijs and I Slack'd about the blind spot and approaches to address. Our consensus for the next step is to ensure this use case is incorporated into the upcoming architecture discussion for the BHR.

https://hg.mozilla.org/mozilla-central/rev/ea2172da20bd