Implement fast literal support in Crankshaft.

Implement fast literal support in Crankshaft.

This extends the current support for nested object literals we already
have in Crankshaft, to also support nested array literals and mixed
nested literals containing arrays and objects. All three types are
generated by the unified HFastLiteral instruction.

All previous upper bounds on nested literal graphs remain unchanged,
keeping the size of generated code in check.

The main intention is to boost performance of two-dimensional array
literals containing constant elements (aka. matrices).

R=danno@chromium.org
TEST=mjsunit/compiler/literals-optimized

Committed: https://code.google.com/p/v8/source/detail?r=10734

[danno, 2012-02-16 16:09:36]

LGTM with comments. There are a few places I didn't comment where you could
share the code to generate copies a little more (single instructions), but it
might require a bit or reordering, your call.

http://codereview.chromium.org/9403018/diff/1/src/arm/lithium-codegen-arm.cc
File src/arm/lithium-codegen-arm.cc (right):

http://codereview.chromium.org/9403018/diff/1/src/arm/lithium-codegen-arm.cc#...
src/arm/lithium-codegen-arm.cc:4385: bool has_elements = elements->length() > 0
&&
The empty fixed array should always be used when the length is zero, and it's
also COW. You could optionally add some asserts.

http://codereview.chromium.org/9403018/diff/1/src/arm/lithium-codegen-arm.cc#...
src/arm/lithium-codegen-arm.cc:4403: __ str(r2, FieldMemOperand(result,
object_offset + i));
The store is the same and can be moved out of the if.

http://codereview.chromium.org/9403018/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

http://codereview.chromium.org/9403018/diff/1/src/hydrogen.cc#newcode3447
src/hydrogen.cc:3447: if (max_depth <= 0) return false;
Don't you want to return on 0 and assert that it's not negative?

http://codereview.chromium.org/9403018/diff/1/src/hydrogen.cc#newcode3451
src/hydrogen.cc:3451: elements->map() !=
boilerplate->GetHeap()->fixed_cow_array_map()) {
length check unnecessary.

http://codereview.chromium.org/9403018/diff/1/src/mips/lithium-codegen-mips.cc
File src/mips/lithium-codegen-mips.cc (right):

http://codereview.chromium.org/9403018/diff/1/src/mips/lithium-codegen-mips.c...
src/mips/lithium-codegen-mips.cc:4295: __ sw(a2, FieldMemOperand(result,
object_offset + i));
Factor our after if

http://codereview.chromium.org/9403018/diff/1/src/x64/lithium-codegen-x64.cc
File src/x64/lithium-codegen-x64.cc (right):

http://codereview.chromium.org/9403018/diff/1/src/x64/lithium-codegen-x64.cc#...
src/x64/lithium-codegen-x64.cc:3999: __ movq(FieldOperand(result, object_offset
+ i), rcx);
After if

[Michael Starzinger, 2012-02-16 17:34:37]

Added new patch set. Landed.

https://chromiumcodereview.appspot.com/9403018/diff/1/src/arm/lithium-codegen...
File src/arm/lithium-codegen-arm.cc (right):

https://chromiumcodereview.appspot.com/9403018/diff/1/src/arm/lithium-codegen...
src/arm/lithium-codegen-arm.cc:4385: bool has_elements = elements->length() > 0
&&
On 2012/02/16 16:09:37, danno wrote:
> The empty fixed array should always be used when the length is zero, and it's
> also COW. You could optionally add some asserts.

Actually adding an assertion that all non-COW arrays have length > 0 fails all
over the place. As discussed offline I will leave the current check in. I'll
investigate where those "empty non-COW arrays" come from.

https://chromiumcodereview.appspot.com/9403018/diff/1/src/arm/lithium-codegen...
src/arm/lithium-codegen-arm.cc:4403: __ str(r2, FieldMemOperand(result,
object_offset + i));
On 2012/02/16 16:09:37, danno wrote:
> The store is the same and can be moved out of the if.

Done.

https://chromiumcodereview.appspot.com/9403018/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

https://chromiumcodereview.appspot.com/9403018/diff/1/src/hydrogen.cc#newcode...
src/hydrogen.cc:3447: if (max_depth <= 0) return false;
On 2012/02/16 16:09:37, danno wrote:
> Don't you want to return on 0 and assert that it's not negative?

Done.

https://chromiumcodereview.appspot.com/9403018/diff/1/src/hydrogen.cc#newcode...
src/hydrogen.cc:3451: elements->map() !=
boilerplate->GetHeap()->fixed_cow_array_map()) {
On 2012/02/16 16:09:37, danno wrote:
> length check unnecessary.

See response to first comment.

https://chromiumcodereview.appspot.com/9403018/diff/1/src/mips/lithium-codege...
File src/mips/lithium-codegen-mips.cc (right):

https://chromiumcodereview.appspot.com/9403018/diff/1/src/mips/lithium-codege...
src/mips/lithium-codegen-mips.cc:4295: __ sw(a2, FieldMemOperand(result,
object_offset + i));
On 2012/02/16 16:09:37, danno wrote:
> Factor our after if

Done.

https://chromiumcodereview.appspot.com/9403018/diff/1/src/x64/lithium-codegen...
File src/x64/lithium-codegen-x64.cc (right):

https://chromiumcodereview.appspot.com/9403018/diff/1/src/x64/lithium-codegen...
src/x64/lithium-codegen-x64.cc:3999: __ movq(FieldOperand(result, object_offset
+ i), rcx);
On 2012/02/16 16:09:37, danno wrote:
> After if

Done.