Support weak heap references in the ConstantPool to support IsWeakObjectInOptimizedCode objects.

Support weak heap references in the ConstantPool to support IsWeakObjectInOptimizedCode objects.

[rmcilroy, 2014-03-18 19:40:50]

PTAL.

[ulan, 2014-03-19 08:49:59]

https://codereview.chromium.org/203583007/diff/1/src/objects-visiting-inl.h
File src/objects-visiting-inl.h (right):

https://codereview.chromium.org/203583007/diff/1/src/objects-visiting-inl.h#n...
src/objects-visiting-inl.h:501: StaticVisitor::VisitPointer(heap,
IIUC, each heap object pointer in constant pool is also visited via reloc-info
in
StaticMarkingVisitor<StaticVisitor>::VisitEmbeddedPointer and
StaticMarkingVisitor<StaticVisitor>::VisitCell.
Those visitors already handle weak pointers.

It would be cleaner to avoid copying that logic here and in assembler-arm.

I think it is safe to only record slot for a heap pointer and not mark it (i.e.
replace VisitPointer with RecordSlot) since we will mark it via reloc-info
anyway. If we do that, then there is no need to track weak pointers separately
from heap pointers. WDYT?

[rmcilroy, 2014-03-20 09:37:10]

On 2014/03/19 08:49:59, ulan wrote:
> https://codereview.chromium.org/203583007/diff/1/src/objects-visiting-inl.h
> File src/objects-visiting-inl.h (right):
> 
>
https://codereview.chromium.org/203583007/diff/1/src/objects-visiting-inl.h#n...
> src/objects-visiting-inl.h:501: StaticVisitor::VisitPointer(heap,
> IIUC, each heap object pointer in constant pool is also visited via reloc-info
> in
> StaticMarkingVisitor<StaticVisitor>::VisitEmbeddedPointer and
> StaticMarkingVisitor<StaticVisitor>::VisitCell.
> Those visitors already handle weak pointers.
> 
> It would be cleaner to avoid copying that logic here and in assembler-arm.
> 
> I think it is safe to only record slot for a heap pointer and not mark it
(i.e.
> replace VisitPointer with RecordSlot) since we will mark it via reloc-info
> anyway. If we do that, then there is no need to track weak pointers separately
> from heap pointers. WDYT?

Ahh yeah, this would probably work (I harbor a wish to avoid the GC having to
interpret relocinfo at all for entries in the constant pool eventually, so I was
wanting to have the ConstantPoolArray do the right thing, but we can work on
that in the future once OOL constant pool is enabled.  I will abandon this CL
and send you another one with the change you suggest.