Zero-sized new-space for VM isolate.

Zero-sized new-space for VM isolate.

Also add unit test that ensures
a zero-sized Scavenger behaves correctly for all operations.

Reduces memory footprint by 32 MB.

R=iposva@google.com

Committed: https://code.google.com/p/dart/source/detail?r=35389

[koda, 2014-04-24 00:24:47]

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/heap.cc
File runtime/vm/heap.cc (right):

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/heap.cc#newco...
runtime/vm/heap.cc:351: isolate->heap()->GetMergedAddressRange(&start, &end);
Actually, why do we get explicitly get Isolate::Current()->heap() here, rather
than simply use 'this' (also below)?

[Ivan Posva, 2014-04-24 17:12:44]

LGTMwC -ip

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/heap.cc
File runtime/vm/heap.cc (right):

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/heap.cc#newco...
runtime/vm/heap.cc:35: Heap::Heap(intptr_t max_new_gen_words,
Thanks! This is also a good step towards being able to set different heap sizes
as we span isolates.

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/heap.cc#newco...
runtime/vm/heap.cc:351: isolate->heap()->GetMergedAddressRange(&start, &end);
On 2014/04/24 00:24:48, koda wrote:
> Actually, why do we get explicitly get Isolate::Current()->heap() here, rather
> than simply use 'this' (also below)?

Please use this.

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
File runtime/vm/scavenger_test.cc (right):

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
runtime/vm/scavenger_test.cc:16: EXPECT(false);
Please add a comment why you expect false here.

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
runtime/vm/scavenger_test.cc:24: EXPECT(false);
and here

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
runtime/vm/scavenger_test.cc:32: EXPECT(false);
and here

[koda, 2014-04-24 17:52:54]

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/heap.cc
File runtime/vm/heap.cc (right):

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/heap.cc#newco...
runtime/vm/heap.cc:351: isolate->heap()->GetMergedAddressRange(&start, &end);
On 2014/04/24 17:12:44, Ivan Posva wrote:
> On 2014/04/24 00:24:48, koda wrote:
> > Actually, why do we get explicitly get Isolate::Current()->heap() here,
rather
> > than simply use 'this' (also below)?
> 
> Please use this.

Actually, there was a reason. We need the isolate and a non-const heap pointer
to call the visitor.

I added an ASSERT (also in Verify below) that 'this' is indeed the current heap
and a TODO to beautify this code.

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
File runtime/vm/scavenger_test.cc (right):

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
runtime/vm/scavenger_test.cc:16: EXPECT(false);
On 2014/04/24 17:12:44, Ivan Posva wrote:
> Please add a comment why you expect false here.

Done.

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
runtime/vm/scavenger_test.cc:24: EXPECT(false);
On 2014/04/24 17:12:44, Ivan Posva wrote:
> and here

Done.

https://codereview.chromium.org/259483002/diff/20001/runtime/vm/scavenger_tes...
runtime/vm/scavenger_test.cc:32: EXPECT(false);
On 2014/04/24 17:12:44, Ivan Posva wrote:
> and here

Done.

[koda, 2014-04-24 18:13:40]

Committed patchset #3 manually as r35389 (presubmit successful).