You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Storing dataAddr pointers as internal pointers in temp vars is disabled or not yet coded in multiple areas as further work is needed. This issue explain and lists what's needed for it to be enabled.
One known issue with enabling it is intrinsic CAS calls, where a child has the destination object and another has the offset. For OffHeap the destination object is the aloadi <contiguousArrayDataAddrField> of the destination array, or a temp/aRegLoad of the dataAddrPtr, but that child is also used for card-marking address calculation which needs the base object value. If The child is aloadi <contiguousArrayDataAddrField> then we can access the base object as its child, if its a temp/aRegLoad then its more complex to get the value of the base object.
Given issues when storing dataAddrPtr in temps and the dual purpose of
using that as the destination address (correct) and destination object
(incorrect) from some evaluators, the store of dataAddrPtr in temps
is disabled until its fixed (Issue: OMR eclipse-omr#7560)
Signed-off-by: Abdulrahman Alattas <[email protected]>
Storing dataAddr pointers as internal pointers in temp vars is disabled or not yet coded in multiple areas as further work is needed. This issue explain and lists what's needed for it to be enabled.
One known issue with enabling it is intrinsic CAS calls, where a child has the destination object and another has the offset. For OffHeap the destination object is the
aloadi <contiguousArrayDataAddrField>
of the destination array, or a temp/aRegLoad of the dataAddrPtr, but that child is also used for card-marking address calculation which needs the base object value. If The child isaloadi <contiguousArrayDataAddrField>
then we can access the base object as its child, if its a temp/aRegLoad then its more complex to get the value of the base object.#7562
#7368 (comment)
The text was updated successfully, but these errors were encountered: