Replaced empty handle checks with MaybeHandles

JSObject::DeepCopy and other functions returned an empty handle to
indicate an exception. All usages but one changed to return
MaybeHandles, and to use macros around the call.

The remaining work is in Compiler::GetUnoptimizedCode(), and when
that is turned into a MaybeHandle, then the macros dealing with
null handle returns can be eliminated.

R=yangguo@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=20877

[mvstanton, 2014-04-22 09:15:10]

Hi Yang, PTAL, thx!
--Michael

https://codereview.chromium.org/247013002/diff/40001/src/runtime.cc
File src/runtime.cc (right):

https://codereview.chromium.org/247013002/diff/40001/src/runtime.cc#newcode14852
src/runtime.cc:14852: initial_map = Map::AsElementsKind(initial_map, to_kind);
There isn't a way for Map::AsElementsKind to return a null handle.

[Yang, 2014-04-22 09:27:19]

LGTM!

https://codereview.chromium.org/247013002/diff/40001/src/jsregexp.cc
File src/jsregexp.cc (right):

https://codereview.chromium.org/247013002/diff/40001/src/jsregexp.cc#newcode347
src/jsregexp.cc:347: ASSERT(!last_match_info.is_null());
Kind of wonder why you have this. (Or at least, why you have this here and not
in IrregexpExec.

[mvstanton, 2014-04-22 09:30:18]

thanks very much, submitting...

https://codereview.chromium.org/247013002/diff/40001/src/jsregexp.cc
File src/jsregexp.cc (right):

https://codereview.chromium.org/247013002/diff/40001/src/jsregexp.cc#newcode347
src/jsregexp.cc:347: ASSERT(!last_match_info.is_null());
On 2014/04/22 09:27:20, Yang wrote:
> Kind of wonder why you have this. (Or at least, why you have this here and not
> in IrregexpExec.

Indeed, it's unnecessary, thx!

[mvstanton, 2014-04-22 09:32:52]

Committed patchset #3 manually as r20877 (presubmit successful).