Allow the embedder to pass the virtual memory limit to v8

Allow the embedder to pass the virtual memory limit to v8

The getrlimit() call might be sandboxed, so it's not safe to use it.

BUG=none
R=mstarzinger@chromium.org
LOG=y

Committed: https://code.google.com/p/v8/source/detail?r=20615

[jochen (gone - plz use gerrit), 2014-04-08 17:42:13]

ptal

[jochen (gone - plz use gerrit), 2014-04-09 11:28:09]

updated to have less defaults all over the place

[Michael Starzinger, 2014-04-09 12:03:04]

LGTM, only nits left. Thanks for addressing my verbal comments already.

https://codereview.chromium.org/228923002/diff/40001/src/heap.cc
File src/heap.cc (right):

https://codereview.chromium.org/228923002/diff/40001/src/heap.cc#newcode6372
src/heap.cc:6372: // On 64-bit platform(s), we put all code objects in a 2 GB
range of
nit: The comment looks very outdated. Can we adapt it?

https://codereview.chromium.org/228923002/diff/40001/src/spaces.cc
File src/spaces.cc (right):

https://codereview.chromium.org/228923002/diff/40001/src/spaces.cc#newcode138
src/spaces.cc:138: size_t actual_requested = requested;
nit: Instead of having two variables "actual_requested" and "requested", can we
just mutate the latter in the condition below and drop the former?

https://codereview.chromium.org/228923002/diff/40001/test/cctest/test-mark-co...
File test/cctest/test-mark-compact.cc (right):

https://codereview.chromium.org/228923002/diff/40001/test/cctest/test-mark-co...
test/cctest/test-mark-compact.cc:79: heap->ConfigureHeap(2*256*KB, 1*MB, 1*MB,
i::kIs64BitArch ? 512 * i::MB : 0);
nit: Shouldn't passing "0" here do the trick now?

https://codereview.chromium.org/228923002/diff/40001/test/cctest/test-mark-co...
test/cctest/test-mark-compact.cc:104: heap->ConfigureHeap(2*256*KB, 1*MB, 1*MB,
i::kIs64BitArch ? 512 * i::MB : 0);
nit: Likewise.

[jochen (gone - plz use gerrit), 2014-04-09 12:44:56]

https://codereview.chromium.org/228923002/diff/40001/src/heap.cc
File src/heap.cc (right):

https://codereview.chromium.org/228923002/diff/40001/src/heap.cc#newcode6372
src/heap.cc:6372: // On 64-bit platform(s), we put all code objects in a 2 GB
range of
On 2014/04/09 12:03:04, Michael Starzinger wrote:
> nit: The comment looks very outdated. Can we adapt it?

Done.

https://codereview.chromium.org/228923002/diff/40001/src/spaces.cc
File src/spaces.cc (right):

https://codereview.chromium.org/228923002/diff/40001/src/spaces.cc#newcode138
src/spaces.cc:138: size_t actual_requested = requested;
On 2014/04/09 12:03:04, Michael Starzinger wrote:
> nit: Instead of having two variables "actual_requested" and "requested", can
we
> just mutate the latter in the condition below and drop the former?

Done.

https://codereview.chromium.org/228923002/diff/40001/test/cctest/test-mark-co...
File test/cctest/test-mark-compact.cc (right):

https://codereview.chromium.org/228923002/diff/40001/test/cctest/test-mark-co...
test/cctest/test-mark-compact.cc:79: heap->ConfigureHeap(2*256*KB, 1*MB, 1*MB,
i::kIs64BitArch ? 512 * i::MB : 0);
On 2014/04/09 12:03:04, Michael Starzinger wrote:
> nit: Shouldn't passing "0" here do the trick now?

Done.

[jochen (gone - plz use gerrit), 2014-04-09 12:46:05]

Committed patchset #4 manually as r20615 (presubmit successful).

[Paweł Hajdan Jr., 2014-04-17 13:59:53]

Drive-by with a suggestion.

https://codereview.chromium.org/228923002/diff/60001/include/v8.h
File include/v8.h (right):

https://codereview.chromium.org/228923002/diff/60001/include/v8.h#newcode3962
include/v8.h:3962: // Deprecated.
Shouldn't this use V8_DEPRECATED?

[jochen (gone - plz use gerrit), 2014-04-17 14:01:47]

https://codereview.chromium.org/228923002/diff/60001/include/v8.h
File include/v8.h (right):

https://codereview.chromium.org/228923002/diff/60001/include/v8.h#newcode3962
include/v8.h:3962: // Deprecated.
On 2014/04/17 13:59:53, Paweł Hajdan Jr. wrote:
> Shouldn't this use V8_DEPRECATED?

No, because then it wouldn't compile. We first have to introduce the new API and
switch chrome to the new API, before we can use V8_DEPRECATED on the old API.