Add a test to verify correctness of remembered set update for Heap::CopyJSObject.

Add a test to verify correctness of remembered set update for Heap::CopyJSObject.

Committed: http://code.google.com/p/v8/source/detail?r=4286

[antonm, 2010-03-24 14:47:55]

Mads, Kasper,

may you have a look?

It's monstrous and still doesn't guarantee to check the stuff all the time.  Any
suggestions are most appreciated.

[Mads Ager (chromium), 2010-03-25 09:42:08]

Anton, can you force a full compacting GC at the beginning of this test to make
sure that you will not get free list allocation?  It seems that we should be
able to set up the heap so we can get a deterministic test here.

http://codereview.chromium.org/1256002/diff/1/4
File test/cctest/test-heap.cc (right):

http://codereview.chromium.org/1256002/diff/1/4#newcode840
test/cctest/test-heap.cc:840: // just enough room to allocate JSbject and thus
fill the newspace.
JSbject -> JSObject

[antonm, 2010-03-25 14:42:45]

<quote>
Anton, can you force a full compacting GC at the beginning of this test to make
sure that you will not get free list allocation?  It seems that we should be
able to set up the heap so we can get a deterministic test here.
</quote>

I thought about that, however decided against as VM  is freshly initialized.

Anyway, as I was in doubt and you're pro that, that's added :)

http://codereview.chromium.org/1256002/diff/1/4
File test/cctest/test-heap.cc (right):

http://codereview.chromium.org/1256002/diff/1/4#newcode840
test/cctest/test-heap.cc:840: // just enough room to allocate JSbject and thus
fill the newspace.
On 2010/03/25 09:42:09, Mads Ager wrote:
> JSbject -> JSObject

Done.

[Mads Ager (chromium), 2010-03-25 15:02:59]

LGTM

This is hairy, but it is good to have a test for this.  You have comments
explaining what is going on so we will have some info to go on if this one
starts failing.