Expose gc(true) to JavaScript, which triggers a scavenger GC.

Expose gc(true) to JavaScript, which triggers a scavenger GC.

With the --expose_gc option, gc() is exposed to JavaScript. Currently gc() triggers a full GC.

To enable JavaScript to test the behavior of a scavenger GC, this patch exposes gc(true). If the first argument is true, gc(...) triggers a scavenger GC. Otherwise, gc(...) triggers a full GC.

BUG=
Test=Manually confirmed that gc() and gc(false) trigger a full GC and that gc(true) triggers a scavenger GC.

Committed: https://code.google.com/p/v8/source/detail?r=12815

[Michael Starzinger, 2012-10-25 11:47:29]

On a higher level note: I am not sure if we should use gc(true) to indicate a
minor GC, wouldn't gc(false) better convey the meaning of causing a minor GC? On
the other hand this would break the convention that an unspecified argument is
just set to undefined, which is a falsey value. This is just a remark, I am
happy to leave it as it is.

https://codereview.chromium.org/11232065/diff/2001/src/extensions/gc-extensio...
File src/extensions/gc-extension.cc (right):

https://codereview.chromium.org/11232065/diff/2001/src/extensions/gc-extensio...
src/extensions/gc-extension.cc:44: HEAP->PerformScavenge(true);
Instead of calling PerfromScavange(), which is really only intended to be used
by our cctests, I would prefer to use the following call instead.

HEAP->CollectGarbage(NEW_SPACE, "gc extension");

https://codereview.chromium.org/11232065/diff/2001/src/heap.cc
File src/heap.cc (right):

https://codereview.chromium.org/11232065/diff/2001/src/heap.cc#newcode674
src/heap.cc:674: void Heap::PerformScavenge(bool force_scavenge) {
This change becomes obsolete then.

https://codereview.chromium.org/11232065/diff/2001/src/heap.cc#newcode855
src/heap.cc:855: if (collector == SCAVENGER)
Looks like leftover debugging code.

https://codereview.chromium.org/11232065/diff/2001/src/heap.h
File src/heap.h (right):

https://codereview.chromium.org/11232065/diff/2001/src/heap.h#newcode1117
src/heap.h:1117: void PerformScavenge(bool force_scavenge = false);
This change becomes obsolete then.

[haraken, 2012-10-25 11:54:49]

> On a higher level note: I am not sure if we should use gc(true) to indicate a
> minor GC, wouldn't gc(false) better convey the meaning of causing a minor GC?
On
> the other hand this would break the convention that an unspecified argument is
> just set to undefined, which is a falsey value. This is just a remark, I am
> happy to leave it as it is.

Yeah, that's the reason I used gc(true) instead of gc(false).

https://codereview.chromium.org/11232065/diff/2001/src/extensions/gc-extensio...
File src/extensions/gc-extension.cc (right):

https://codereview.chromium.org/11232065/diff/2001/src/extensions/gc-extensio...
src/extensions/gc-extension.cc:44: HEAP->PerformScavenge(true);
On 2012/10/25 11:47:29, Michael Starzinger wrote:
> Instead of calling PerfromScavange(), which is really only intended to be used
> by our cctests, I would prefer to use the following call instead.
> 
> HEAP->CollectGarbage(NEW_SPACE, "gc extension");

Done.

https://codereview.chromium.org/11232065/diff/2001/src/heap.cc
File src/heap.cc (right):

https://codereview.chromium.org/11232065/diff/2001/src/heap.cc#newcode674
src/heap.cc:674: void Heap::PerformScavenge(bool force_scavenge) {
On 2012/10/25 11:47:29, Michael Starzinger wrote:
> This change becomes obsolete then.

Done.

https://codereview.chromium.org/11232065/diff/2001/src/heap.cc#newcode855
src/heap.cc:855: if (collector == SCAVENGER)
On 2012/10/25 11:47:29, Michael Starzinger wrote:
> Looks like leftover debugging code.

Done.

https://codereview.chromium.org/11232065/diff/2001/src/heap.h
File src/heap.h (right):

https://codereview.chromium.org/11232065/diff/2001/src/heap.h#newcode1117
src/heap.h:1117: void PerformScavenge(bool force_scavenge = false);
On 2012/10/25 11:47:29, Michael Starzinger wrote:
> This change becomes obsolete then.

Done.

[Michael Starzinger, 2012-10-25 12:18:24]

LGTM. Will land that for you.