[heap] Black is encoded with 11, grey with 10.

[heap] Black is encoded with 11, grey with 10.

This CL changes the color for encoding black and grey. Moreover, it introduces a higher level live object iterator.

BUG=chromium:561449
LOG=n

Committed: https://crrev.com/67f99ee1026f7832b477b956beb394485be372f7
Cr-Commit-Position: refs/heads/master@{#33208}

[Hannes Payer (out of office), 2015-12-21 13:53:30]

Description was changed from

==========
[heap] Black is encoded with 11, grey with 10.

BUG=chromium:561449
==========

to

==========
[heap] Black is encoded with 11, grey with 10.

BUG=chromium:561449
LOG=n
==========

[Hannes Payer (out of office), 2016-01-05 12:32:49]

hpayer@chromium.org changed reviewers:
+ mlippautz@chromium.org, ulan@chromium.org

[Hannes Payer (out of office), 2016-01-05 12:42:43]

Description was changed from

==========
[heap] Black is encoded with 11, grey with 10.

BUG=chromium:561449
LOG=n
==========

to

==========
[heap] Black is encoded with 11, grey with 10.

This CL changes the color for encoding black and grey. Moreover, it introduces a
higher level live object iterator.

BUG=chromium:561449
LOG=n
==========

[ulan, 2016-01-05 14:45:23]

Beautiful clean-up :)

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (left):

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2962: if ((cells[start_index] & index_in_cell) != 0) {
As discussed offline, we still need special handling for left-trimming here.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:1539: if (marking_deque()->IsFull()) return;
yay!

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2948: // The object start is before the slot index.
Hence, in this case the
The comment is obsolete.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2950: CHECK(index_in_cell > 0);
index_in_cell will always be > 0 by definition.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2952: slot_mask = index_in_cell + index_in_cell - 1;
let's add explicit parens index_in_cell + (index_in_cell - 1) to not think about
potential overflow.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact.h
File src/heap/mark-compact.h (right):

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:161: INLINE(static bool TransferColor(HeapObject* from,
HeapObject* to)) {
Let's assert that the color of the _to_ is white.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:166: // Looks black so far.
This comment doesn't help anymore. Let's just remove it.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:870: while (!it_.Done()) {
IUC, this condition will cause early return for the last cell, i.e. not all
objects of the last cell will be found.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:891: clear_overlapping_bit_ = true;
We would need to fetch the next cell here anyway.

After fetching and checking the LSB, we could clear it immediately instead of
setting clear_overlapping_bit_.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:910: return mode_ == kAllLiveObjects ||
As discussed we can templatize this.

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
File src/x64/macro-assembler-x64.cc (left):

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
src/x64/macro-assembler-x64.cc:5298: // Note that we are using a 4-byte aligned
8-byte load.
// This change here should be sufficient to convert the function to HasColor.
if (first_bit == 0 && second_bit == 0) {
  movp(mask_scratch, Immediate(0));
} else if (first_bit == 1 && second_bit == 1) {
  movp(mask_scratch, rcx);
} else {
  DCHECK(first_bit == 1 && second_bit == 0);
}

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
File src/x64/macro-assembler-x64.cc (right):

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
src/x64/macro-assembler-x64.cc:5285: Label::Distance on_black_near) {
let's keep the old name for on_black_distance.

[Hannes Payer (out of office), 2016-01-07 10:30:32]

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (left):

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2962: if ((cells[start_index] & index_in_cell) != 0) {
On 2016/01/05 14:45:23, ulan wrote:
> As discussed offline, we still need special handling for left-trimming here.

Done.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:1539: if (marking_deque()->IsFull()) return;
On 2016/01/05 14:45:23, ulan wrote:
> yay!

yay!

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2948: // The object start is before the slot index.
Hence, in this case the
On 2016/01/05 14:45:23, ulan wrote:
> The comment is obsolete.

Done.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2950: CHECK(index_in_cell > 0);
On 2016/01/05 14:45:23, ulan wrote:
> index_in_cell will always be > 0 by definition.

Done.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.cc:2952: slot_mask = index_in_cell + index_in_cell - 1;
On 2016/01/05 14:45:23, ulan wrote:
> let's add explicit parens index_in_cell + (index_in_cell - 1) to not think
about
> potential overflow.

Done.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact.h
File src/heap/mark-compact.h (right):

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:161: INLINE(static bool TransferColor(HeapObject* from,
HeapObject* to)) {
On 2016/01/05 14:45:23, ulan wrote:
> Let's assert that the color of the _to_ is white.

Done.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:166: // Looks black so far.
On 2016/01/05 14:45:23, ulan wrote:
> This comment doesn't help anymore. Let's just remove it.

Done.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:169: return true;  // Is black.
I also removed "// Is black." comment.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:870: while (!it_.Done()) {
On 2016/01/05 14:45:23, ulan wrote:
> IUC, this condition will cause early return for the last cell, i.e. not all
> objects of the last cell will be found.

Nope, last cell index is already invalid, so this should be fine.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:891: clear_overlapping_bit_ = true;
On 2016/01/05 14:45:23, ulan wrote:
> We would need to fetch the next cell here anyway.
> 
> After fetching and checking the LSB, we could clear it immediately instead of
> setting clear_overlapping_bit_.

Done.

https://codereview.chromium.org/1517993003/diff/140001/src/heap/mark-compact....
src/heap/mark-compact.h:910: return mode_ == kAllLiveObjects ||
On 2016/01/05 14:45:23, ulan wrote:
> As discussed we can templatize this.

Done.

[Michael Lippautz, 2016-01-07 14:02:54]

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact.h
File src/heap/mark-compact.h (right):

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:868: inline HeapObject* Next() {
Maybe move the implementation to to mark-compact-inl.h

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:870: HeapObject* object = NULL;
nit:nullptr

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:872: int trailing_zeros =
base::bits::CountTrailingZeros32(current_cell_);
nit: Maybe add
  DCHECK_NE(trailing_zeros, 32);

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:876: if (trailing_zeros < Bitmap::kBitIndexMask) {
You could do
  uint32_t bit_idx = 0;
  if (trailing_zeros < Bitmap::kBitIndexMask) {
    bit_idx = 1u << (trailing_zeros + 1);
    ...
  } else {
    bit_idx = 0x1;
    ...
  }
  current_cell_ &= ~bit_idx;

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:906: if (object != NULL) break;
nit: nullptr, likewise below

[Hannes Payer (out of office), 2016-01-07 16:26:16]

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact.h
File src/heap/mark-compact.h (right):

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:868: inline HeapObject* Next() {
On 2016/01/07 14:02:53, Michael Lippautz wrote:
> Maybe move the implementation to to mark-compact-inl.h

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:870: HeapObject* object = NULL;
On 2016/01/07 14:02:53, Michael Lippautz wrote:
> nit:nullptr

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:872: int trailing_zeros =
base::bits::CountTrailingZeros32(current_cell_);
On 2016/01/07 14:02:53, Michael Lippautz wrote:
> nit: Maybe add
>   DCHECK_NE(trailing_zeros, 32);

This is implied by current_cell_ != 0 and by the implementation of
CountTrailingZeros32.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:876: if (trailing_zeros < Bitmap::kBitIndexMask) {
On 2016/01/07 14:02:53, Michael Lippautz wrote:
> You could do
>   uint32_t bit_idx = 0;
>   if (trailing_zeros < Bitmap::kBitIndexMask) {
>     bit_idx = 1u << (trailing_zeros + 1);
>     ...
>   } else {
>     bit_idx = 0x1;
>     ...
>   }
>   current_cell_ &= ~bit_idx;

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:906: if (object != NULL) break;
On 2016/01/07 14:02:53, Michael Lippautz wrote:
> nit: nullptr, likewise below

Done.

[ulan, 2016-01-07 16:39:11]

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2930: if ((cells[start_index] & index_in_cell) != 0 &&
Let's make the two parts consistent (either treat them both as boolean or use !=
0, == 0 for both).

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2938: !(cells[start_index - 1] & 0x80000000)) {
Use Bitmap::kBitsPerCell instead of the constant or assert that the value is 32.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2978: if (leading_zeros == Bitmap::kBitIndexMask) {
Making offset a signed int, would allow to simply set

offset =  Bitmap::kBitIndexMask - leading_zeros - 1;
offset would range from -1 to 30.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2982:
CHECK(base::bits::CountLeadingZeros32(current_cell & slot_mask) == 0);
This check is subsumed by CHECK(Marking::IsBlack(Marking::MarkBitFrom(object)))
below.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact.h
File src/heap/mark-compact.h (right):

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:867: template <LiveObjectIterationMode mode>
The template parameter should be on the class, not method.

Otherwise, it allows to change the between calls.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:875: current_cell_ &= ~(1u << trailing_zeros);
unsigned color = 0;
if (trailing_zeros < Bitmap::kBitIndexMask) {
  color = (current_cell_ >> (trailing_zeros + 1)) & 1;
  current_cell_ = ~(3u << trailing_zeros);
} else {
  current_cell_ = ~(1u << trailing_zeros);
  DCHECK(!it_.Done());
  it_.Advance();
  cell_base_ = it_.CurrentCellBase();
  current_cell_ = *it_.CurrentCell();
  color = (current_cell_ & 1);
  current_cell_ &= 0xFFFFFFFE;
}

now use the color to check for object.

[Hannes Payer (out of office), 2016-01-07 20:17:33]

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
File src/x64/macro-assembler-x64.cc (left):

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
src/x64/macro-assembler-x64.cc:5298: // Note that we are using a 4-byte aligned
8-byte load.
On 2016/01/05 14:45:23, ulan wrote:
> // This change here should be sufficient to convert the function to HasColor.
> if (first_bit == 0 && second_bit == 0) {
>   movp(mask_scratch, Immediate(0));
> } else if (first_bit == 1 && second_bit == 1) {
>   movp(mask_scratch, rcx);
> } else {
>   DCHECK(first_bit == 1 && second_bit == 0);
> }

Not sure what you mean here? Should I bring back the old code and patch it
accordingly? The current HasColor implementation is in sync with the other
platforms.

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
File src/x64/macro-assembler-x64.cc (right):

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
src/x64/macro-assembler-x64.cc:5285: Label::Distance on_black_near) {
On 2016/01/05 14:45:23, ulan wrote:
> let's keep the old name for on_black_distance.

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact.h
File src/heap/mark-compact.h (right):

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:867: template <LiveObjectIterationMode mode>
On 2016/01/07 16:39:11, ulan wrote:
> The template parameter should be on the class, not method.
> 
> Otherwise, it allows to change the between calls.

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.h:875: current_cell_ &= ~(1u << trailing_zeros);
On 2016/01/07 16:39:11, ulan wrote:
> unsigned color = 0;
> if (trailing_zeros < Bitmap::kBitIndexMask) {
>   color = (current_cell_ >> (trailing_zeros + 1)) & 1;
>   current_cell_ = ~(3u << trailing_zeros);
> } else {
>   current_cell_ = ~(1u << trailing_zeros);
>   DCHECK(!it_.Done());
>   it_.Advance();
>   cell_base_ = it_.CurrentCellBase();
>   current_cell_ = *it_.CurrentCell();
>   color = (current_cell_ & 1);
>   current_cell_ &= 0xFFFFFFFE;
> }
> 
> now use the color to check for object.

I refactored it the other way around.

[Michael Lippautz, 2016-01-08 07:46:18]

lgtm from my side

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact....
src/heap/mark-compact.cc:2918: unsigned int start_index = mark_bit_index >>
Bitmap::kBitsPerCellLog2;
nits: The naming here is really (really) weird. I had a hard time following with
the current naming of variables. Feel free to ignore.
 - s/start_index/cell_index/
 - s/index_in_cell/index_mask/
     with index_mask = 1u << Bitmap::IndexInCell(mark_bit_index);
 - s/cell_base/base_address/
 - s/cell_base_start_index/base_address_cell_index/

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact....
src/heap/mark-compact.cc:2931: !(cells[start_index] & (index_in_cell >> 1))) {
Please write out the negation part of the boolean condition, i.e.,
  !((cell[start_index] & (index_in_cell >> 1)) != 0)

(In general it would be nice to write this using MarkBit accessors, but we lack
at least Prev() for this to work.)

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact....
src/heap/mark-compact.cc:2938: !(cells[start_index - 1] & 0x80000000)) {
Same as above w/ the boolean condition.

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact....
src/heap/mark-compact.cc:2982:
CHECK(base::bits::CountLeadingZeros32(current_cell & slot_mask) == 0);
You are and'ing with all 1-s. The check should just be
  base::bits::CountLeadingZeros32(current_cell) == 0
?

[Hannes Payer (out of office), 2016-01-08 08:27:05]

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact....
src/heap/mark-compact.cc:2918: unsigned int start_index = mark_bit_index >>
Bitmap::kBitsPerCellLog2;
On 2016/01/08 07:46:18, Michael Lippautz wrote:
> nits: The naming here is really (really) weird. I had a hard time following
with
> the current naming of variables. Feel free to ignore.
>  - s/start_index/cell_index/
>  - s/index_in_cell/index_mask/
>      with index_mask = 1u << Bitmap::IndexInCell(mark_bit_index);
>  - s/cell_base/base_address/
>  - s/cell_base_start_index/base_address_cell_index/

The naming scheme was introduced earlier. Happy to change it. Done.

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact....
src/heap/mark-compact.cc:2931: !(cells[start_index] & (index_in_cell >> 1))) {
On 2016/01/08 07:46:18, Michael Lippautz wrote:
> Please write out the negation part of the boolean condition, i.e.,
>   !((cell[start_index] & (index_in_cell >> 1)) != 0)
> 
> (In general it would be nice to write this using MarkBit accessors, but we
lack
> at least Prev() for this to work.)

I propose to not double negate then: ((cell[start_index] & (index_in_cell >> 1))
== 0)
Done.

https://codereview.chromium.org/1517993003/diff/200001/src/heap/mark-compact....
src/heap/mark-compact.cc:2938: !(cells[start_index - 1] & 0x80000000)) {
On 2016/01/08 07:46:18, Michael Lippautz wrote:
> Same as above w/ the boolean condition.

Same as above. Done.

[Hannes Payer (out of office), 2016-01-11 12:24:32]

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
File src/x64/macro-assembler-x64.cc (left):

https://codereview.chromium.org/1517993003/diff/140001/src/x64/macro-assemble...
src/x64/macro-assembler-x64.cc:5298: // Note that we are using a 4-byte aligned
8-byte load.
On 2016/01/07 20:17:33, Hannes Payer wrote:
> On 2016/01/05 14:45:23, ulan wrote:
> > // This change here should be sufficient to convert the function to
HasColor.
> > if (first_bit == 0 && second_bit == 0) {
> >   movp(mask_scratch, Immediate(0));
> > } else if (first_bit == 1 && second_bit == 1) {
> >   movp(mask_scratch, rcx);
> > } else {
> >   DCHECK(first_bit == 1 && second_bit == 0);
> > }
> 
> Not sure what you mean here? Should I bring back the old code and patch it
> accordingly? The current HasColor implementation is in sync with the other
> platforms.

After discussing offline, I kept x64 special and simplified it.

[Hannes Payer (out of office), 2016-01-11 13:37:34]

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact.cc
File src/heap/mark-compact.cc (right):

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2930: if ((cells[start_index] & index_in_cell) != 0 &&
On 2016/01/07 16:39:11, ulan wrote:
> Let's make the two parts consistent (either treat them both as boolean or use
!=
> 0, == 0 for both).

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2938: !(cells[start_index - 1] & 0x80000000)) {
On 2016/01/07 16:39:11, ulan wrote:
> Use Bitmap::kBitsPerCell instead of the constant or assert that the value is
32.

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2978: if (leading_zeros == Bitmap::kBitIndexMask) {
On 2016/01/07 16:39:11, ulan wrote:
> Making offset a signed int, would allow to simply set
> 
> offset =  Bitmap::kBitIndexMask - leading_zeros - 1;
> offset would range from -1 to 30.
> 

Done.

https://codereview.chromium.org/1517993003/diff/160001/src/heap/mark-compact....
src/heap/mark-compact.cc:2982:
CHECK(base::bits::CountLeadingZeros32(current_cell & slot_mask) == 0);
On 2016/01/07 16:39:11, ulan wrote:
> This check is subsumed by
CHECK(Marking::IsBlack(Marking::MarkBitFrom(object)))
> below.
> 

Done.

[ulan, 2016-01-11 13:56:52]

lgtm!

[Hannes Payer (out of office), 2016-01-11 13:59:07]

The CQ bit was checked by hpayer@chromium.org

[Hannes Payer (out of office), 2016-01-11 13:59:07]

The patchset sent to the CQ was uploaded after l-g-t-m from
mlippautz@chromium.org, ulan@chromium.org
Link to the patchset: https://codereview.chromium.org/1517993003/#ps280001
(title: " ")

[commit-bot: I haz the power, 2016-01-11 13:59:16]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1517993003/280001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1517993003/280001

[commit-bot: I haz the power, 2016-01-11 14:32:52]

Description was changed from

==========
[heap] Black is encoded with 11, grey with 10.

This CL changes the color for encoding black and grey. Moreover, it introduces a
higher level live object iterator.

BUG=chromium:561449
LOG=n
==========

to

==========
[heap] Black is encoded with 11, grey with 10.

This CL changes the color for encoding black and grey. Moreover, it introduces a
higher level live object iterator.

BUG=chromium:561449
LOG=n
==========

[commit-bot: I haz the power, 2016-01-11 14:32:52]

Committed patchset #15 (id:280001)

[commit-bot: I haz the power, 2016-01-11 14:33:17]

Description was changed from

==========
[heap] Black is encoded with 11, grey with 10.

This CL changes the color for encoding black and grey. Moreover, it introduces a
higher level live object iterator.

BUG=chromium:561449
LOG=n
==========

to

==========
[heap] Black is encoded with 11, grey with 10.

This CL changes the color for encoding black and grey. Moreover, it introduces a
higher level live object iterator.

BUG=chromium:561449
LOG=n

Committed: https://crrev.com/67f99ee1026f7832b477b956beb394485be372f7
Cr-Commit-Position: refs/heads/master@{#33208}
==========

[commit-bot: I haz the power, 2016-01-11 14:33:18]

Patchset 15 (id:??) landed as
https://crrev.com/67f99ee1026f7832b477b956beb394485be372f7
Cr-Commit-Position: refs/heads/master@{#33208}