[meta] Audit uses of ContentParent:: and ContentChild::mRemoteType
Categories
(Core :: DOM: Content Processes, defect, P1)
Tracking
()
Fission Milestone | M4.1 |
People
(Reporter: mccr8, Assigned: mccr8)
References
(Blocks 1 open bug)
Details
(Keywords: meta)
Kris pointed out that ChromeUtils::RequestProcInfo() ends up returning an uninitialized ProcType if the remoteType is webIsolated=. I'll make that case fail better, fix it for Fission, and also audit the other C++ uses of mRemoteType which may also not be properly handling webIsolated=. Maybe it could be turned into a real data type.
Anyways, the RequestProcInfo() thing is causing widget/tests/browser/browser_test_procinfo.js to fail intermittently.
Assignee | ||
Comment 1•5 years ago
|
||
I've now looked at all of the uses of ContentChild::mRemoteType (and the places that value flows out to). Bug 1579218 is the only issue I found.
Assignee | ||
Comment 2•5 years ago
|
||
ContentParent::TryToRecycle() looks at ContentParent::mRemoteType and compares it to DEFAULT_REMOTE_TYPE, but it uses that to determine if it should use the preallocated process, which I assume is just broken completely with Fission so I won't concern myself with it.
Updated•4 years ago
|
Assignee | ||
Comment 3•4 years ago
|
||
I think this is fixed enough.
Updated•4 years ago
|
Description
•