Forward transition objects to grey if they are referenced by a large object

Forward transition objects to grey if they are referenced by a large object


BUG=

[Michael Starzinger, 2012-11-14 08:42:24]

https://codereview.chromium.org/11410080/diff/1/src/incremental-marking-inl.h
File src/incremental-marking-inl.h (right):

https://codereview.chromium.org/11410080/diff/1/src/incremental-marking-inl.h...
src/incremental-marking-inl.h:55: return false;
This will prevent the slot from being recorded when we do a forward
white-to-grey transition. Compaction might corrupt the pointer. In the case of a
white-to-grey transition we need to fall through to the code below.

[payer, 2012-11-14 09:12:37]

https://codereview.chromium.org/11410080/diff/1/src/incremental-marking-inl.h
File src/incremental-marking-inl.h (right):

https://codereview.chromium.org/11410080/diff/1/src/incremental-marking-inl.h...
src/incremental-marking-inl.h:55: return false;
On 2012/11/14 08:42:24, Michael Starzinger wrote:
> This will prevent the slot from being recorded when we do a forward
> white-to-grey transition. Compaction might corrupt the pointer. In the case of
a
> white-to-grey transition we need to fall through to the code below.

Done.

[Michael Starzinger, 2012-11-15 18:04:26]

https://codereview.chromium.org/11410080/diff/6002/src/incremental-marking-inl.h
File src/incremental-marking-inl.h (right):

https://codereview.chromium.org/11410080/diff/6002/src/incremental-marking-in...
src/incremental-marking-inl.h:44: // TODO(hpayer): make decision based on if
object has a progress bar
As discussed offline, can we rebase that to the brand new progress bar and check
for the HAS_PROGRESS_BAR page flag here?

https://codereview.chromium.org/11410080/diff/6002/src/incremental-marking-in...
src/incremental-marking-inl.h:49: } else if (Marking::IsBlack(obj_bit)) {
Can we rewrite the condition so that it does the Marking::IsBlack only once,
unfortunately GCC isn't smart enough to optimize it away.

[payer, 2012-11-16 13:43:05]

https://codereview.chromium.org/11410080/diff/6002/src/incremental-marking-inl.h
File src/incremental-marking-inl.h (right):

https://codereview.chromium.org/11410080/diff/6002/src/incremental-marking-in...
src/incremental-marking-inl.h:44: // TODO(hpayer): make decision based on if
object has a progress bar
On 2012/11/15 18:04:26, Michael Starzinger wrote:
> As discussed offline, can we rebase that to the brand new progress bar and
check
> for the HAS_PROGRESS_BAR page flag here?

Done.

I will add the optimization in the next CL.

https://codereview.chromium.org/11410080/diff/6002/src/incremental-marking-in...
src/incremental-marking-inl.h:49: } else if (Marking::IsBlack(obj_bit)) {
On 2012/11/15 18:04:26, Michael Starzinger wrote:
> Can we rewrite the condition so that it does the Marking::IsBlack only once,
> unfortunately GCC isn't smart enough to optimize it away.

Done.