Make progress in incremental marking if scavenge is delaying mark-sweep.

Make progress in incremental marking if scavenge is delaying mark-sweep.

R=mstarzinger@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=11213

[Michael Starzinger, 2012-04-02 09:01:46]

https://chromiumcodereview.appspot.com/9965054/diff/1/src/heap.cc
File src/heap.cc (left):

https://chromiumcodereview.appspot.com/9965054/diff/1/src/heap.cc#oldcode4821
src/heap.cc:4821: idle_notification_will_schedule_next_gc_ = true;
This was the last site that changed the above flag. Can we get rid of the flag
completely?

https://chromiumcodereview.appspot.com/9965054/diff/1/src/incremental-marking.h
File src/incremental-marking.h (right):

https://chromiumcodereview.appspot.com/9965054/diff/1/src/incremental-marking...
src/incremental-marking.h:49: enum Finalizer {
The name "finalizer" feels wrong. Can we change this to something like ...
enum CompletionAction {
  NO_COMPLETION_ACTION,
  GC_VIA_STACK_GUARD
};

https://chromiumcodereview.appspot.com/9965054/diff/1/src/incremental-marking...
src/incremental-marking.h:113: void Step(intptr_t allocated, Finalizer finalizer
= GC_VIA_STACK_GUARD);
There are only three other call-sites, could we get along without a default
argument?

[ulan, 2012-04-02 09:42:24]

Thanks, I uploaded new patch set.

https://chromiumcodereview.appspot.com/9965054/diff/1/src/heap.cc
File src/heap.cc (left):

https://chromiumcodereview.appspot.com/9965054/diff/1/src/heap.cc#oldcode4821
src/heap.cc:4821: idle_notification_will_schedule_next_gc_ = true;
On 2012/04/02 09:01:46, Michael Starzinger wrote:
> This was the last site that changed the above flag. Can we get rid of the flag
> completely?

Done.

https://chromiumcodereview.appspot.com/9965054/diff/1/src/incremental-marking.h
File src/incremental-marking.h (right):

https://chromiumcodereview.appspot.com/9965054/diff/1/src/incremental-marking...
src/incremental-marking.h:49: enum Finalizer {
On 2012/04/02 09:01:46, Michael Starzinger wrote:
> The name "finalizer" feels wrong. Can we change this to something like ...
> enum CompletionAction {
>   NO_COMPLETION_ACTION,
>   GC_VIA_STACK_GUARD
> };

Done.

https://chromiumcodereview.appspot.com/9965054/diff/1/src/incremental-marking...
src/incremental-marking.h:113: void Step(intptr_t allocated, Finalizer finalizer
= GC_VIA_STACK_GUARD);
On 2012/04/02 09:01:46, Michael Starzinger wrote:
> There are only three other call-sites, could we get along without a default
> argument?

Done.

[Michael Starzinger, 2012-04-02 09:48:55]

LGTM (with one final nit).

https://chromiumcodereview.appspot.com/9965054/diff/6/test/cctest/test-heap.cc
File test/cctest/test-heap.cc (left):

https://chromiumcodereview.appspot.com/9965054/diff/6/test/cctest/test-heap.c...
test/cctest/test-heap.cc:1530: // code below.
Can you preserve this comment above the call to Step() above?