Add the rate of js invocations from the api as a signal of idleness

Add the rate of js invocations from the api as a signal of idleness
for starting major GC in the memory reducer.

BUG=

Committed: https://crrev.com/27d23eee1213329cd36edc983b67b381035bffc6
Cr-Commit-Position: refs/heads/master@{#31655}

[ulan, 2015-10-27 17:07:21]

ulan@chromium.org changed reviewers:
+ hpayer@chromium.org

[ulan, 2015-10-27 17:07:21]

PTAL

[Hannes Payer (out of office), 2015-10-28 14:58:08]

LGTM

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.cc
File src/heap/memory-reducer.cc (right):

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.cc#...
src/heap/memory-reducer.cc:26: void MemoryReducer::TimerTask::RunInternal() {
Can we leave a comment somewhere mentioning under which conditions the memory
reducer will trigger?

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.h
File src/heap/memory-reducer.h (right):

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.h#n...
src/heap/memory-reducer.h:112: js_calls_counter_(0),
I am wondering if the js call rate live on the isolate? WDYT? The memory reducer
is also asking the heap about the allocation rate. Logically it sounds like a
better home for this counter.
Hmm, but the isolate is not computing it, the memory reducer is taking care of
that.

https://codereview.chromium.org/1414483004/diff/1/test/cctest/heap-tester.h
File test/cctest/heap-tester.h (right):

https://codereview.chromium.org/1414483004/diff/1/test/cctest/heap-tester.h#n...
test/cctest/heap-tester.h:26: V(TestMemoryReducerSampleJsCalls)     \
alphabetic order

https://codereview.chromium.org/1414483004/diff/1/test/cctest/test-heap.cc
File test/cctest/test-heap.cc (right):

https://codereview.chromium.org/1414483004/diff/1/test/cctest/test-heap.cc#ne...
test/cctest/test-heap.cc:6279: void CheckEquals(double expected, double actual)
{
Let's move this one to cctest.h since it is a utility function.

[ulan, 2015-10-29 12:39:25]

Thanks! I uploaded new patch set.

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.cc
File src/heap/memory-reducer.cc (right):

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.cc#...
src/heap/memory-reducer.cc:26: void MemoryReducer::TimerTask::RunInternal() {
On 2015/10/28 14:58:08, Hannes Payer wrote:
> Can we leave a comment somewhere mentioning under which conditions the memory
> reducer will trigger?

Done.

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.h
File src/heap/memory-reducer.h (right):

https://codereview.chromium.org/1414483004/diff/1/src/heap/memory-reducer.h#n...
src/heap/memory-reducer.h:112: js_calls_counter_(0),
On 2015/10/28 14:58:08, Hannes Payer wrote:
> I am wondering if the js call rate live on the isolate? WDYT? The memory
reducer
> is also asking the heap about the allocation rate. Logically it sounds like a
> better home for this counter.
> Hmm, but the isolate is not computing it, the memory reducer is taking care of
> that.

I think this is very specific to memory reducer so as much as possible should
live here.

I would move the remaining counter from isolate to memory reducer, but that is
perf sensitive.

https://codereview.chromium.org/1414483004/diff/1/test/cctest/heap-tester.h
File test/cctest/heap-tester.h (right):

https://codereview.chromium.org/1414483004/diff/1/test/cctest/heap-tester.h#n...
test/cctest/heap-tester.h:26: V(TestMemoryReducerSampleJsCalls)     \
On 2015/10/28 14:58:08, Hannes Payer wrote:
> alphabetic order

Done.

https://codereview.chromium.org/1414483004/diff/1/test/cctest/test-heap.cc
File test/cctest/test-heap.cc (right):

https://codereview.chromium.org/1414483004/diff/1/test/cctest/test-heap.cc#ne...
test/cctest/test-heap.cc:6279: void CheckEquals(double expected, double actual)
{
On 2015/10/28 14:58:08, Hannes Payer wrote:
> Let's move this one to cctest.h since it is a utility function.

Done.

[ulan, 2015-10-29 13:24:25]

The CQ bit was checked by ulan@chromium.org

[ulan, 2015-10-29 13:24:26]

The patchset sent to the CQ was uploaded after l-g-t-m from hpayer@chromium.org
Link to the patchset: https://codereview.chromium.org/1414483004/#ps20001
(title: "Address comments from Hannes")

[commit-bot: I haz the power, 2015-10-29 13:24:30]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1414483004/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1414483004/20001

[commit-bot: I haz the power, 2015-10-29 13:59:39]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-10-29 14:00:13]

Patchset 2 (id:??) landed as
https://crrev.com/27d23eee1213329cd36edc983b67b381035bffc6
Cr-Commit-Position: refs/heads/master@{#31655}