Open Bug 1365149 Opened 6 years ago Updated 4 months ago

improve the queuing strategy of ChannelEventQueue


(Core :: Networking, enhancement, P3)





(Reporter: schien, Unassigned)


(Whiteboard: [necko-triaged])

follow-up of bug 1354455 for improving ChannelEventQueue.

Three objectives:
1. replace mRunningMonitor with atomic conditional variable, so that we can use single lock to avoid deadlock internally.
2. avoid FlushQueue be invoked on multiple threads simultaneously. 
3. simplify and document the rule of what operation can be run in parallel, what should be atomic, and what should be pending.
Whiteboard: [necko-active]
Thanks for filing this!
Bulk priority update:
Priority: -- → P1
Priority: P1 → P3
Whiteboard: [necko-active]
Whiteboard: [necko-triaged]
not going to work on this recently so unassigned myself for now.
Assignee: schien → nobody
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.