Open Bug 906944 Opened 11 years ago Updated 2 years ago

Use a heartbeat to detect hanging childs

Categories

(Core :: IPC, defect)

defect

Tracking

()

Tracking Status
e10s later ---

People

(Reporter: evilpie, Unassigned)

References

Details

Attachments

(1 file)

Attached patch lastmessagetimeSplinter Review
I have been working on a way to detect hanging children. The current approach mostly lives in ContentParent/ContentChild and I am thinking about even removing the change to RPCChannel. Of course we could also move everything into the ipc layer, but I feel it's actually simpler to do the way I did, because we have easier access to various functions. For example KillHard. It also avoids including some random things into the IPC layer like timers. One problem that dvander already pointed out that when the computer suspends or the chrome process hangs, we shouldn't kill the child.
Assignee: nobody → evilpies
Status: NEW → ASSIGNED
Won't this cause every content process to wake up regularly? In general we shouldn't cause wakeups unless we actually need the content process.
Yeah we probably don't want this. One problem is wake-ups and the other problem is handling computers going into the suspended state. We should always try to avoid killing a process when it's not necessary.
Not sure if we even want to do this. But I am not working on this.
Assignee: evilpies → nobody
Status: ASSIGNED → NEW
Mass tracking-e10s flag change. Filter bugmail on "2be0fcce-e36a-4e2c-aa80-0e3d33eb5406".
tracking-e10s: --- → +
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: