Correctly hook up back references to internalized strings in code deserializer.

Correctly hook up back references to internalized strings in code deserializer.

R=mvstanton@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=22540

[mvstanton, 2014-07-22 13:31:32]

LGTM, couple of comments.

https://codereview.chromium.org/411483002/diff/40001/src/objects-inl.h
File src/objects-inl.h (right):

https://codereview.chromium.org/411483002/diff/40001/src/objects-inl.h#newcod...
src/objects-inl.h:6593: ASSERT(!HasHashCode());
How about words along this line?

"Setting a tagged value (canonical) in the hash code causes the hash code to be
interpreted as uninitialized. We use this fact to recognize that we have a
forwarded string."

https://codereview.chromium.org/411483002/diff/40001/src/serialize.cc
File src/serialize.cc (right):

https://codereview.chromium.org/411483002/diff/40001/src/serialize.cc#newcode985
src/serialize.cc:985: if (deserializing_user_code()) {                          
            \
Just a thought, take it or leave it: what about checking
deserializing_user_code() inside of ProcessBackRefInSerializedCode()?

I say that because it looks reasonable to consider it as a standard thing to do
when dealing with backreferences. It's a detail that it only needs to be done in
user code.

[Yang, 2014-07-23 06:55:58]

https://codereview.chromium.org/411483002/diff/40001/src/objects-inl.h
File src/objects-inl.h (right):

https://codereview.chromium.org/411483002/diff/40001/src/objects-inl.h#newcod...
src/objects-inl.h:6593: ASSERT(!HasHashCode());
On 2014/07/22 13:31:31, mvstanton wrote:
> How about words along this line?
> 
> "Setting a tagged value (canonical) in the hash code causes the hash code to
be
> interpreted as uninitialized. We use this fact to recognize that we have a
> forwarded string."

Done.

https://codereview.chromium.org/411483002/diff/40001/src/serialize.cc
File src/serialize.cc (right):

https://codereview.chromium.org/411483002/diff/40001/src/serialize.cc#newcode985
src/serialize.cc:985: if (deserializing_user_code()) {                          
            \
On 2014/07/22 13:31:31, mvstanton wrote:
> Just a thought, take it or leave it: what about checking
> deserializing_user_code() inside of ProcessBackRefInSerializedCode()?
> 
> I say that because it looks reasonable to consider it as a standard thing to
do
> when dealing with backreferences. It's a detail that it only needs to be done
in
> user code.

I'd like to leave it this way, since it mirrors the call to
ProcessNewObjectFromSerializedCode.

[Yang, 2014-07-23 07:16:41]

Committed patchset #4 manually as r22540 (presubmit successful).