Reset opt count on optimized code map lookup failure.

Reset opt count on optimized code map lookup failure.

R=jkummerow@chromium.org
BUG=309723

[Jakob Kummerow, 2013-11-05 14:35:04]

LGTM with comments.

https://codereview.chromium.org/59823002/diff/1/src/code-stubs-hydrogen.cc
File src/code-stubs-hydrogen.cc (right):

https://codereview.chromium.org/59823002/diff/1/src/code-stubs-hydrogen.cc#ne...
src/code-stubs-hydrogen.cc:1201: int32_t opt_count_mask =
BitField has a kMask which you can use. Note that you'll want the
DisabledOptimizationReasonBits::kMask, though, so that opt_count is zeroed out.

https://codereview.chromium.org/59823002/diff/1/src/runtime-profiler.cc
File src/runtime-profiler.cc (right):

https://codereview.chromium.org/59823002/diff/1/src/runtime-profiler.cc#newco...
src/runtime-profiler.cc:309: if (shared->opt_count() >= FLAG_max_opt_count) {
This changes behavior, and could easily have unintended side effects. As
discussed, I'd prefer to make this change in a separate CL and carefully watch
the performance impact.

[Yang, 2013-11-05 18:31:14]

On 2013/11/05 14:35:04, Jakob wrote:
> LGTM with comments.
> 
> https://codereview.chromium.org/59823002/diff/1/src/code-stubs-hydrogen.cc
> File src/code-stubs-hydrogen.cc (right):
> 
>
https://codereview.chromium.org/59823002/diff/1/src/code-stubs-hydrogen.cc#ne...
> src/code-stubs-hydrogen.cc:1201: int32_t opt_count_mask =
> BitField has a kMask which you can use. Note that you'll want the
> DisabledOptimizationReasonBits::kMask, though, so that opt_count is zeroed
out.
> 
> https://codereview.chromium.org/59823002/diff/1/src/runtime-profiler.cc
> File src/runtime-profiler.cc (right):
> 
>
https://codereview.chromium.org/59823002/diff/1/src/runtime-profiler.cc#newco...
> src/runtime-profiler.cc:309: if (shared->opt_count() >= FLAG_max_opt_count) {
> This changes behavior, and could easily have unintended side effects. As
> discussed, I'd prefer to make this change in a separate CL and carefully watch
> the performance impact.

Updated according to comment.

I also changed the hydrogen stub. Apparently we store the opt count and bailout
reason as a smi on 32bit systems, but as a pseudo smi on x64. That's why the
mask is still wrong. Please take another look.

[danno, 2013-11-06 08:15:52]

https://codereview.chromium.org/59823002/diff/100001/src/heap.cc
File src/heap.cc (right):

https://codereview.chromium.org/59823002/diff/100001/src/heap.cc#newcode5729
src/heap.cc:5729: contexts_disposed_ = 0;
DBC: Why is this change needed? It seems unrelated to the title of the CL

[Jakob Kummerow, 2013-11-06 09:07:23]

lgtm