MIPS: ES6 symbols: turn symbols into a proper primitive type

MIPS: ES6 symbols: turn symbols into a proper primitive type

Port r14051 (e9140a92)

Original commit message:
(qua last week's TC39)

Specifically:
- Install Symbol constructor function on the global object.
- Adjust code generation for typeof.
- Remove IsSymbol built-in, IS_SYMBOL macro now defined using typeof.
- Remove hack that allowed symbols as constructor results, and some other special cases.
- Remove symbol_delegate and GetDelegate function.
- Extend ToBoolean stub to handle symbols.
- Extend ToNumber to return NaN on symbols.
- Poison symbol's toString function, and thereby ToString on symbols.

BUG=

Committed: https://code.google.com/p/v8/source/detail?r=14058

[Paul Lind, 2013-03-22 22:14:20]

lgtm. I will land this for you shortly. Please note the nit in the comments, we
will address this in a separate CL.

[Paul Lind, 2013-03-22 22:15:20]

Sorry here is the related comment on this code.

https://codereview.chromium.org/13007003/diff/1/src/mips/lithium-codegen-mips.cc
File src/mips/lithium-codegen-mips.cc (right):

https://codereview.chromium.org/13007003/diff/1/src/mips/lithium-codegen-mips...
src/mips/lithium-codegen-mips.cc:5831: __ GetObjectType(input, input, scratch);
There is some tricky stuff with delay slots here that I'm not certain is still
justified, and I'm not sure the comments line up 100% with the code. I'm going
to suggest we land this, to fix the build for mips. But I'm going to raise an
issue to come back and refactor this a bit for clarity.

[Paul Lind, 2013-03-23 01:23:40]

Made a small fix, now it really lgtm!

https://codereview.chromium.org/13007003/diff/1/src/mips/lithium-codegen-mips.cc
File src/mips/lithium-codegen-mips.cc (right):

https://codereview.chromium.org/13007003/diff/1/src/mips/lithium-codegen-mips...
src/mips/lithium-codegen-mips.cc:1907: __ Branch(true_label, eq, at,
Operand(SYMBOL_TYPE));
We cannot use 'at' register in this Branch(), as 'at' will be used to hold the
Operand(SYMBOL_TYPE) in the 'eq' case.

I made a small change here to use scratch1().

[palfia, 2013-03-23 01:49:45]

https://codereview.chromium.org/13007003/diff/1/src/mips/lithium-codegen-mips.cc
File src/mips/lithium-codegen-mips.cc (right):

https://codereview.chromium.org/13007003/diff/1/src/mips/lithium-codegen-mips...
src/mips/lithium-codegen-mips.cc:1907: __ Branch(true_label, eq, at,
Operand(SYMBOL_TYPE));
On 2013/03/23 01:23:40, Paul Lind wrote:
> We cannot use 'at' register in this Branch(), as 'at' will be used to hold the
> Operand(SYMBOL_TYPE) in the 'eq' case.
> 
> I made a small change here to use scratch1().

Thanks for fixing this! It escaped my attention, that we cannot use 'at' with
'eq'.

[palfia, 2013-03-23 02:39:43]

Committed manually as r14058 (presubmit successful).