Add checks to runtime functions.

Add checks to runtime functions.

BUG=

Committed: https://code.google.com/p/v8/source/detail?r=12471

[Yang, 2012-09-04 14:41:06]

PTAL.

[Yang, 2012-09-05 14:32:25]

Please take another look.

[Erik Corry, 2012-09-07 09:22:33]

It seems to me that in lithium-ia32.cc in DoDateField the MarkAsCall call should
have a third argument CAN_DEOPTIMIZE_EAGERLY since now the instruction can
deopt.  But if your test is working then apparently that is not needed for some
reason.

LGTM

http://codereview.chromium.org/10915062/diff/3001/src/messages.js
File src/messages.js (right):

http://codereview.chromium.org/10915062/diff/3001/src/messages.js#newcode205
src/messages.js:205: "not_date_object",              ["this is not a Date
object."],
We normally use the word "Receiver", rather than 'this'.  That would also let
you capitalize it correctly.

http://codereview.chromium.org/10915062/diff/3001/src/runtime.cc
File src/runtime.cc (right):

http://codereview.chromium.org/10915062/diff/3001/src/runtime.cc#newcode2070
src/runtime.cc:2070: f->RemovePrototype();
I wonder why RemovePrototype does not return void?

http://codereview.chromium.org/10915062/diff/3001/test/mjsunit/date.js
File test/mjsunit/date.js (right):

http://codereview.chromium.org/10915062/diff/3001/test/mjsunit/date.js#newcod...
test/mjsunit/date.js:335: assertThrows(function() {
Date.prototype.getTime.call(""); }, TypeError);
Are you sure you are getting a deopt here?  There's an OptimizationState
predicate for testing this.

[Yang, 2012-09-07 13:02:56]

On 2012/09/07 09:22:33, Erik Corry wrote:
> It seems to me that in lithium-ia32.cc in DoDateField the MarkAsCall call
should
> have a third argument CAN_DEOPTIMIZE_EAGERLY since now the instruction can
> deopt.  But if your test is working then apparently that is not needed for
some
> reason.
> 
> LGTM
> 
> http://codereview.chromium.org/10915062/diff/3001/src/messages.js
> File src/messages.js (right):
> 
> http://codereview.chromium.org/10915062/diff/3001/src/messages.js#newcode205
> src/messages.js:205: "not_date_object",              ["this is not a Date
> object."],
> We normally use the word "Receiver", rather than 'this'.  That would also let
> you capitalize it correctly.
> 
> http://codereview.chromium.org/10915062/diff/3001/src/runtime.cc
> File src/runtime.cc (right):
> 
> http://codereview.chromium.org/10915062/diff/3001/src/runtime.cc#newcode2070
> src/runtime.cc:2070: f->RemovePrototype();
> I wonder why RemovePrototype does not return void?
> 
> http://codereview.chromium.org/10915062/diff/3001/test/mjsunit/date.js
> File test/mjsunit/date.js (right):
> 
>
http://codereview.chromium.org/10915062/diff/3001/test/mjsunit/date.js#newcod...
> test/mjsunit/date.js:335: assertThrows(function() {
> Date.prototype.getTime.call(""); }, TypeError);
> Are you sure you are getting a deopt here?  There's an OptimizationState
> predicate for testing this.

I guess it works without CAN_DEOPTIMIZE_EAGERLY since there are no side effects
here. Nevertheless I added this option.

I also changed RemovePrototype to void and added a check for optimization state
in the test case.