Resubmit bigint changes of r41817 that were later reverted.

Resubmit bigint changes of r41817 that were later reverted.
An inliner issue was fixed in r41840.
A missing length increment is added by this cl.
Minor improvements and comments.

R=zra@google.com

Committed: https://code.google.com/p/dart/source/detail?r=41873

[regis, 2014-11-20 22:37:23]

regis@google.com changed reviewers:
+ zra@google.com

[regis, 2014-11-20 22:37:24]

https://codereview.chromium.org/747483002/diff/1/runtime/lib/bigint.dart
File runtime/lib/bigint.dart (right):

https://codereview.chromium.org/747483002/diff/1/runtime/lib/bigint.dart#newc...
runtime/lib/bigint.dart:209: length++;  // Account for leading zero for 64-bit
processing.
This increment was dropped by mistake in patch #4 of the reverted cl, here and
in the C++ version.

https://codereview.chromium.org/747483002/diff/1/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://codereview.chromium.org/747483002/diff/1/runtime/vm/object.cc#newcode...
runtime/vm/object.cc:16443: length++;  // Account for leading zero for 64-bit
processing.
Here too.

[zra, 2014-11-20 22:56:16]

lgtm

https://codereview.chromium.org/747483002/diff/20001/runtime/lib/bigint.dart
File runtime/lib/bigint.dart (right):

https://codereview.chromium.org/747483002/diff/20001/runtime/lib/bigint.dart#...
runtime/lib/bigint.dart:61: static _Bigint ONE = new _Bigint()._setInt(1);
final?

https://codereview.chromium.org/747483002/diff/20001/runtime/lib/bigint.dart#...
runtime/lib/bigint.dart:260: int _nbits(int x) {
x.bitLength ?

Not sure how that's implemented, but it might be worth looking into.

[regis, 2014-11-20 23:17:40]

Thanks!

https://codereview.chromium.org/747483002/diff/20001/runtime/lib/bigint.dart
File runtime/lib/bigint.dart (right):

https://codereview.chromium.org/747483002/diff/20001/runtime/lib/bigint.dart#...
runtime/lib/bigint.dart:61: static _Bigint ONE = new _Bigint()._setInt(1);
On 2014/11/20 22:56:16, zra wrote:
> final?

No. See comment above.

https://codereview.chromium.org/747483002/diff/20001/runtime/lib/bigint.dart#...
runtime/lib/bigint.dart:260: int _nbits(int x) {
On 2014/11/20 22:56:16, zra wrote:
> x.bitLength ?
> 
> Not sure how that's implemented, but it might be worth looking into.

It's a call to Utils::HighestBit, which I remember improving. So it is identical
to this call, except that we avoid the call to runtime plumbing and also save
processing for one bit (32 instead of 64).

An intrinsic could help, especially since we could make use of an instruction,
but not a runtime call.

I'll keep this in mind in case it shows up on the profile.

[regis, 2014-11-20 23:18:09]

Committed patchset #2 (id:20001) manually as r41873 (presubmit successful).