Many classes in graphics/java and elsewhere deallocate native memory
in a finalizer on the assumption that instance methods can no longer
be called once the finalizer has been called. This is incorrect if
the object can be used, possibly indirectly, from another finalizer,
possibly one in the application.
This is the initial installment of a patch to cause such post-finalization
uses to at least see a null pointer rather than causing memory corruption
by accessing deallocated native memory. This should make it possible to
identify and fix such finalization ordering issues.
There are more graphics classes that need this treatment, and probably
many more in other subsystems.
This solution is < 100% effective if finalizers can be invoked
concurrently. We currently promise that they aren't.
(In my opinion, the real cause here is a language spec bug. But that ship
has sailed.)
Bug: 18178237
Change-Id: I844cf1e0fbb190407389c4f8e8f072752cca6198
Changes in this patch include
[x] Long is used to store native pointers as they can
be 64-bit.
[x] Some minor changes have been done to conform with
standard JNI practice (e.g. use of jint instead of int
in JNI function prototypes)
[x] AssetAtlasManager is not completely 64-bit compatible
yet. Specifically mAtlasMap member has to be converted
to hold native pointer using long. Added a TODO to
AssetAtlasManager.java to indicate the change required.
Change-Id: I940433f601c6db998c1a8ffff338f5361200d5ed
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
Signed-off-by: Craig Barber <craig.barber@arm.com>
Signed-off-by: Kévin PETIT <kevin.petit@arm.com>
Signed-off-by: Marcus Oakland <marcus.oakland@arm.com>
