Move deopt_id and related helpers/definitions from Isolate to Thread

Move deopt_id and related helpers/definitions from Isolate to Thread

BUG=
R=zra@google.com

Committed: https://github.com/dart-lang/sdk/commit/6f5335061106766c7acc0dd01a1ce4d15f58b934

[srdjan, 2015-10-12 22:44:47]

srdjan@google.com changed reviewers:
+ zra@google.com

[srdjan, 2015-10-12 23:10:33]

On 2015/10/12 22:44:48, srdjan wrote:

The reason for this change:

With background compilation, one isolate can have two compilations running in
parallel but deopt-id increment happens per compilation. There can be only one
compilation per thread.

[zra, 2015-10-13 06:49:37]

lgtm

[Florian Schneider, 2015-10-13 09:35:26]

fschneider@google.com changed reviewers:
+ fschneider@google.com

[Florian Schneider, 2015-10-13 09:35:27]

https://codereview.chromium.org/1390153004/diff/1/runtime/vm/intermediate_lan...
File runtime/vm/intermediate_language.h (right):

https://codereview.chromium.org/1390153004/diff/1/runtime/vm/intermediate_lan...
runtime/vm/intermediate_language.h:602: 
Wouldn't it make sense to make kNoDeoptId top-level here in this file? It's
tedious to write Thread::kNoDeoptId, plus it is strange to depend on including
thread.h for this constant - intermediate_language.h seems like a better place.

enum {
  kNoDeoptId = -1;
};

[srdjan, 2015-10-13 17:08:26]

Committed patchset #1 (id:1) manually as
6f5335061106766c7acc0dd01a1ce4d15f58b934 (presubmit successful).

[srdjan, 2015-10-13 19:23:15]

https://codereview.chromium.org/1390153004/diff/1/runtime/vm/intermediate_lan...
File runtime/vm/intermediate_language.h (right):

https://codereview.chromium.org/1390153004/diff/1/runtime/vm/intermediate_lan...
runtime/vm/intermediate_language.h:602: 
On 2015/10/13 09:35:27, Florian Schneider wrote:
> Wouldn't it make sense to make kNoDeoptId top-level here in this file? It's
> tedious to write Thread::kNoDeoptId, plus it is strange to depend on including
> thread.h for this constant - intermediate_language.h seems like a better
place.
> 
> enum {
>   kNoDeoptId = -1;
> };

Good point. Note that Thread::kNoDeoptId is shorter than Isolate::kNoDeoptId
;-). 
Talked about it with others in MTV, and the only consensus is that we need to
make it better, and that making it a global is not the best solution.
Other possibilities: In the thread, store a compiler state instead of CHA,
deopt_id, .... and move the deopt_id handling there.

Leaving it as it is for now.