Some more debugger-related refactorings.

Some more debugger-related refactorings.

R=ulan@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=21607

[Yang, 2014-05-27 15:24:05]

This depends on https://codereview.chromium.org/300553008/

[ulan, 2014-05-28 13:24:32]

https://codereview.chromium.org/297303006/diff/1/src/debug.cc
File src/debug.cc (right):

https://codereview.chromium.org/297303006/diff/1/src/debug.cc#newcode604
src/debug.cc:604: heap->CollectAllGarbage(Heap::kNoGCFlags,
"Debug::CreateScriptCache");
The comment is out of sync: there is only one GC and the heap is not iterable
afterwards.

If incremental marking is in progress then this will not trigger old space GC.

[Yang, 2014-06-02 12:19:31]

Updated

https://codereview.chromium.org/297303006/diff/1/src/debug.cc
File src/debug.cc (right):

https://codereview.chromium.org/297303006/diff/1/src/debug.cc#newcode604
src/debug.cc:604: heap->CollectAllGarbage(Heap::kNoGCFlags,
"Debug::CreateScriptCache");
On 2014/05/28 13:24:32, ulan wrote:
> The comment is out of sync: there is only one GC and the heap is not iterable
> afterwards.
> 
> If incremental marking is in progress then this will not trigger old space GC.

The comment actually makes sense:
Script wrappers are weak global handles. The first GC needs to trigger their
weak callbacks. The second one then removes dead script objects. Both have to be
done here because HeapIterator only does a GC if the heap is not yet iterable.

I'm not entirely sure whether the first GC has be to kMakeHeapIterableMask or
just kSweepPreciselyMask though.

[ulan, 2014-06-02 12:43:01]

I think using kMakeHeapIterableMask is correct. LGTM.

[Yang, 2014-06-02 12:51:34]

Committed patchset #2 manually as r21607 (presubmit successful).