Closed Bug 570965 Opened 14 years ago Closed 6 years ago

need statistical garbage reclamation selftest

Categories

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

x86
macOS
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WONTFIX
Future

People

(Reporter: pnkfelix, Unassigned)

References

Details

Lars has convinced me that trying to make a selftest (of e.g. weak references) that has 100% precision is not reasonable. Lars suggests a statistical approach instead: rather than check that all known garbage is reclaimed, instead check that some large percentage of it is. This sounds like a promising tack, at least for the near term. To my knowledge, none of the existing tests employ such a strategy. It would be good to have at least one that gets reviewed by the group, to serve as a model template. Question: what is a reasonable minimal percentage to guarantee gets reclaimed? I suppose the answer depends on where the float retention is coming from. My (unconfirmed) hypothesis so far as been that some of the retention is due to the OS adding non-zeroed padding on the stack (to keep certain points at aligned addresses, an issue I've wrestled with in the past for FFI stuff), this would indicate that the expected accidental retention is a function of what we expect the stack frame size to be. But I should really confirm that hypothesis before making calculations based on it.
See Also: → 568615
Target Milestone: --- → Future
Assignee: fklockii → nobody
Flags: flashplayer-qrb+
Tamarin isn't maintained anymore. WONTFIX remaining bugs.
Status: NEW → RESOLVED
Closed: 6 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.