Protect SerializedData from copying.

Protect SerializedData from copying.

Compiler-generated copy constructor does not generate
correct code for this class, so make it move-only type.

Review-Url: https://codereview.chromium.org/2781993005
Cr-Commit-Position: refs/heads/master@{#44266}
Committed: https://chromium.googlesource.com/v8/v8/+/b6912850dffa2e6ba52d2c254cc5b72761e63f6d

[Slava Chigrin, 2017-03-29 16:41:36]

Description was changed from

==========
Protect SerializedData from copying.

Compiler-generated copy constructor does not generate
correct code for this class, so make it move-only type.
==========

to

==========
Protect SerializedData from copying.

Compiler-generated copy constructor does not generate
correct code for this class, so make it move-only type.
==========

[Slava Chigrin, 2017-03-29 16:41:37]

vchigrin@yandex-team.ru changed reviewers:
+ yangguo@chromium.org

[Slava Chigrin, 2017-03-29 16:45:26]

Before that it was possible to write code like
...
SnapshotData snapshot_data(&serializer);
SnapshotData copy = snapshot_data;
...
Program will crash after destruction both these objects.
I think it is good idea to protect programmer from such errors at compilation
time, when possible.

[Yang, 2017-03-29 17:06:59]

thanks! One comment

https://codereview.chromium.org/2781993005/diff/1/src/snapshot/serializer-com...
File src/snapshot/serializer-common.h (right):

https://codereview.chromium.org/2781993005/diff/1/src/snapshot/serializer-com...
src/snapshot/serializer-common.h:254: 
We have a macro called DISALLOW_COPY_AND_ASSIGN or similar for this.

[Slava Chigrin, 2017-03-30 09:40:06]

I also have proposition about
ExternalReferenceTable::kDeoptTableSerializeEntryCount constant. Seems it value
is set by kind of heuristic, and it already was increased several times in CLs
like https://codereview.chromium.org/196953003.

In our browser we added a code to serialize references to arbitrary deoptimizer
entries by means of adding special flags to top bits of external reference ID.
At serialization time we get deoptimization entry ID by code like this
https://cs.chromium.org/chromium/src/v8/src/disassembler.cc?q=Deoptimizer::Ge...

I'll be happy to upstream these changes to main V8 repository - thus, this
heuristic constant can be removed, and it will be possible to use serializer in
more scenarios. What is your opinion about these changes? Is there chance they
will be accepted? If it worth to see it I will build CL in the nearest time.

Thanks!

https://codereview.chromium.org/2781993005/diff/1/src/snapshot/serializer-com...
File src/snapshot/serializer-common.h (right):

https://codereview.chromium.org/2781993005/diff/1/src/snapshot/serializer-com...
src/snapshot/serializer-common.h:254: 
On 2017/03/29 17:06:59, Yang wrote:
> We have a macro called DISALLOW_COPY_AND_ASSIGN or similar for this.

Done. Thanks - didn't notice that V8 also have this macro.

[Yang, 2017-03-30 09:50:50]

On 2017/03/30 09:40:06, Slava Chigrin wrote:
> I also have proposition about
> ExternalReferenceTable::kDeoptTableSerializeEntryCount constant. Seems it
value
> is set by kind of heuristic, and it already was increased several times in CLs
> like https://codereview.chromium.org/196953003.
> 
> In our browser we added a code to serialize references to arbitrary
deoptimizer
> entries by means of adding special flags to top bits of external reference ID.
> At serialization time we get deoptimization entry ID by code like this
>
https://cs.chromium.org/chromium/src/v8/src/disassembler.cc?q=Deoptimizer::Ge...
> 
> I'll be happy to upstream these changes to main V8 repository - thus, this
> heuristic constant can be removed, and it will be possible to use serializer
in
> more scenarios. What is your opinion about these changes? Is there chance they
> will be accepted? If it worth to see it I will build CL in the nearest time.
> 
> Thanks!
> 
>
https://codereview.chromium.org/2781993005/diff/1/src/snapshot/serializer-com...
> File src/snapshot/serializer-common.h (right):
> 
>
https://codereview.chromium.org/2781993005/diff/1/src/snapshot/serializer-com...
> src/snapshot/serializer-common.h:254: 
> On 2017/03/29 17:06:59, Yang wrote:
> > We have a macro called DISALLOW_COPY_AND_ASSIGN or similar for this.
> 
> Done. Thanks - didn't notice that V8 also have this macro.

lgtm.

Regarding deoptimization entries. I'd be happy to take patches! I guess you are
suggesting to introduce a new serializer bytecode to represent deoptimizer
entries, and upon deserialization ask for them? That SGTM.

[Slava Chigrin, 2017-03-30 10:35:55]

The CQ bit was checked by vchigrin@yandex-team.ru

[commit-bot: I haz the power, 2017-03-30 10:35:59]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-03-30 11:44:15]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1490870155275030,
"parent_rev": "d389d473a73bde75e2df6b260a113f2e6da15dc7", "commit_rev":
"b6912850dffa2e6ba52d2c254cc5b72761e63f6d"}

[commit-bot: I haz the power, 2017-03-30 11:44:21]

Description was changed from

==========
Protect SerializedData from copying.

Compiler-generated copy constructor does not generate
correct code for this class, so make it move-only type.
==========

to

==========
Protect SerializedData from copying.

Compiler-generated copy constructor does not generate
correct code for this class, so make it move-only type.

Review-Url: https://codereview.chromium.org/2781993005
Cr-Commit-Position: refs/heads/master@{#44266}
Committed:
https://chromium.googlesource.com/v8/v8/+/b6912850dffa2e6ba52d2c254cc5b72761e...
==========

[commit-bot: I haz the power, 2017-03-30 11:44:22]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/v8/v8/+/b6912850dffa2e6ba52d2c254cc5b72761e...