Remove sequential sweeping mode and perform lazy sweeping when no sweeper threads are active.

Remove sequential sweeping mode and perform lazy sweeping when no sweeper threads are active.

BUG=
R=jarin@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=22346

[Jarin, 2014-07-11 07:40:52]

I love this. Couple of comments below.

https://codereview.chromium.org/382793002/diff/80001/src/incremental-marking.cc
File src/incremental-marking.cc (right):

https://codereview.chromium.org/382793002/diff/80001/src/incremental-marking....
src/incremental-marking.cc:886: if
(heap_->mark_compact_collector()->sweeping_pending() &&
As discussed offline, perhaps we should only complete the sweeping only if there
are completed sweeping threads.

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc
File src/mark-compact.cc (right):

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc#newc...
src/mark-compact.cc:591: void MarkCompactCollector::WaitUntilSweepingCompleted()
{
Nit: WaitUntilSweepingCompleted -> EnsureSweepingCompleted?

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc#newc...
src/mark-compact.cc:619: bool MarkCompactCollector::IsSweepingCompleted() {
Nit: rename IsSweepingCompleted to IsConcurrentSweepingCompleted?

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc#newc...
src/mark-compact.cc:4211: if (FLAG_concurrent_sweeping) how_to_sweep =
CONCURRENT_CONSERVATIVE;
Is this is needed? CONCURRENT_CONSERVATIVE is the default anyway.

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.h
File src/mark-compact.h (right):

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.h#newco...
src/mark-compact.h:676: bool sweeping_pending() { return sweeping_pending_; }
Nit: so why is not this called IsSweepingInProgress (or some such)?

[Hannes Payer (out of office), 2014-07-11 09:36:13]

https://codereview.chromium.org/382793002/diff/80001/src/incremental-marking.cc
File src/incremental-marking.cc (right):

https://codereview.chromium.org/382793002/diff/80001/src/incremental-marking....
src/incremental-marking.cc:886: if
(heap_->mark_compact_collector()->sweeping_pending() &&
On 2014/07/11 07:40:52, jarin wrote:
> As discussed offline, perhaps we should only complete the sweeping only if
there
> are completed sweeping threads.

On our benchmark it seems to be fine. I left a comment in IsSweepingCompleted.

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc
File src/mark-compact.cc (right):

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc#newc...
src/mark-compact.cc:591: void MarkCompactCollector::WaitUntilSweepingCompleted()
{
On 2014/07/11 07:40:52, jarin wrote:
> Nit: WaitUntilSweepingCompleted -> EnsureSweepingCompleted?

Done.

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc#newc...
src/mark-compact.cc:619: bool MarkCompactCollector::IsSweepingCompleted() {
On 2014/07/11 07:40:52, jarin wrote:
> Nit: rename IsSweepingCompleted to IsConcurrentSweepingCompleted?

I left a comment for non-concurrent sweeping, instaed of renaming it since it is
also called in the non-concurrent case. WDYT?

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.cc#newc...
src/mark-compact.cc:4211: if (FLAG_concurrent_sweeping) how_to_sweep =
CONCURRENT_CONSERVATIVE;
On 2014/07/11 07:40:52, jarin wrote:
> Is this is needed? CONCURRENT_CONSERVATIVE is the default anyway.

I think it is good to initialize it.

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.h
File src/mark-compact.h (right):

https://codereview.chromium.org/382793002/diff/80001/src/mark-compact.h#newco...
src/mark-compact.h:676: bool sweeping_pending() { return sweeping_pending_; }
On 2014/07/11 07:40:52, jarin wrote:
> Nit: so why is not this called IsSweepingInProgress (or some such)?

It is a true getter now. Anyhow, I renamed the variable to
sweeping_in_progress_.

[Jarin, 2014-07-11 09:44:04]

lgtm

[Hannes Payer (out of office), 2014-07-11 09:51:45]

Committed patchset #6 manually as r22346 (tree was closed).