Closed Bug 622844 Opened 14 years ago Closed 6 years ago

Need optimized GCMember->GCMember move operation

Categories

(Tamarin Graveyard :: Garbage Collection (mmGC), defect, P2)

defect

Tracking

(Not tracked)

RESOLVED WONTFIX
Future

People

(Reporter: treilly, Assigned: treilly)

References

Details

(Whiteboard: safegc)

Some write barriers are elided manually when this occurs in a critical spot but we could support it without having to leave off GCMember. The optimization is to skip the ref count traffic involved.
Blocks: 623276
Flags: flashplayer-bug-
Priority: P2 → P3
Priority: P3 → P4
Priority: P4 → P2
There a couple scenarios that could be imagined here but the only one that matters is the one where one GCMember is being transfered to a newly constructed GCMember that's being constructed. The old code would just copy the pointer to the dest and null out the source pointer. A new contructor form is probably warranted, maybe GCMember(GCMember& incoming, GCMember::kMoveFlag)?
Looking more carefully at it there appears to be some design work required here, and the only thing we miss out on by not doing it now is that some uses of DRCWB et al cannot be converted to GCMember in the Serrano time frame. Is that correct? If so I vote that we defer the bug until Brannan, and make it a blocker for RC removal.
Yes this problem cannot be solved w/o a new API or going back to raw pointers.
Target Milestone: Q3 11 - Serrano → Q4 11 - Anza
Flags: flashplayer-qrb+
Flags: flashplayer-injection-
Target Milestone: Q4 11 - Anza → Q1 12 - Brannan
Target Milestone: Q1 12 - Brannan → Future
Status: ASSIGNED → NEW
Whiteboard: safegc
Tamarin is a dead project now. Mass WONTFIX.
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → WONTFIX
Tamarin isn't maintained anymore. WONTFIX remaining bugs.
You need to log in before you can comment on or make changes to this bug.