Add mistagging-readbarrier to weak cell

Add mistagging-readbarrier to weak cell

R=hpayer@chromium.org,ulan@chromium.org
BUG=

[Erik Corry Chromium.org, 2015-02-01 17:05:27]

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

[Erik Corry Chromium.org, 2015-02-01 17:07:14]

Perhaps it would be better to add a boolean field to the weak cell.  The nice
thing about this version is that if the read barrier is forgotten, then it will
immediately crash, but it seems a bit clumsy to put it in the field, which can
already be a valid pointer or a Smi zero.

[Erik Corry Chromium.org, 2015-02-02 12:21:31]

If we don't want to expand the size of a weak cell, an alternative way to do
this is to have the bit in the map field of the weak cell.  In this scenario
there would be two different immortal maps for weak cells.  This works best if
they are a power-of-2 number of bytes apart, which apparently is not always the
case at the moment:

 if (base::bits::IsPowerOfTwo32(Map::kSize)) {

[Erik Corry Chromium.org, 2015-02-03 16:13:24]

Alternative approach, PTAL: https://codereview.chromium.org/893073006/

[Hannes Payer (out of office), 2015-02-04 13:03:44]

https://codereview.chromium.org/892843002/diff/1/src/ia32/macro-assembler-ia3...
File src/ia32/macro-assembler-ia32.cc (right):

https://codereview.chromium.org/892843002/diff/1/src/ia32/macro-assembler-ia3...
src/ia32/macro-assembler-ia32.cc:2421: j(not_zero, &done);
Do we need the test? Maybe the cpu is more happy with just untagging. WDYT?