Implement committed physical memory stats for Linux.

Implement committed physical memory stats for Linux.

The patch introduces CommittedPhysicalMemory function to
the Heap class that reports committed *physical* memory acquired
for the heap from the OS.
It is important because some OSes may defer actual committment on e.g.
first access to the region.
So reporting just plain committed size led to various weird artifacts
like showing V8 allocated memory higher than the whole process
private size.

BUG=v8:2191
Committed: https://code.google.com/p/v8/source/detail?r=12793

[alph, 2012-10-11 13:17:23]

Michael,

I've implemented the committed physical size approximation without use of
mincore which happened to be prohibited in the sandbox. PTAL.

[Michael Starzinger, 2012-10-16 13:01:06]

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc
File src/spaces.cc (right):

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc#newcode882
src/spaces.cc:882: MemoryChunk::UpdateHighWaterMark(allocation_info_.top);
This function is called by PagedSpace::SlowAllocateRaw which uses the free-list
allocator. So the updating should be covered by PagedSpace::SetTop(). Can we
drop this?

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc#newcode1142
src/spaces.cc:1142: MemoryChunk::UpdateHighWaterMark(allocation_info_.top);
I think this one is better bottle-necked in NewSpace::UpdateAllocationInfo()
which is called after the two semi-spaces have been flipped. Can we move it
there?

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc#newcode1227
src/spaces.cc:1227: MemoryChunk::UpdateHighWaterMark(allocation_info_.top);
This one will also be covered by NewSpace::UpdateAllocationInfo(). Can we drop
this?

[alph, 2012-10-16 13:44:41]

Thanks for the review. PTAL.

P.S. Looks like I rebaselined the patch at some point so diff with the previous
patch is screwed. I'm sorry.

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc
File src/spaces.cc (right):

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc#newcode882
src/spaces.cc:882: MemoryChunk::UpdateHighWaterMark(allocation_info_.top);
On 2012/10/16 13:01:06, Michael Starzinger wrote:
> This function is called by PagedSpace::SlowAllocateRaw which uses the
free-list
> allocator. So the updating should be covered by PagedSpace::SetTop(). Can we
> drop this?

Done.

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc#newcode1142
src/spaces.cc:1142: MemoryChunk::UpdateHighWaterMark(allocation_info_.top);
On 2012/10/16 13:01:06, Michael Starzinger wrote:
> I think this one is better bottle-necked in NewSpace::UpdateAllocationInfo()
> which is called after the two semi-spaces have been flipped. Can we move it
> there?

Done.

https://chromiumcodereview.appspot.com/11066118/diff/1/src/spaces.cc#newcode1227
src/spaces.cc:1227: MemoryChunk::UpdateHighWaterMark(allocation_info_.top);
On 2012/10/16 13:01:06, Michael Starzinger wrote:
> This one will also be covered by NewSpace::UpdateAllocationInfo(). Can we drop
> this?

Done.

[Michael Starzinger, 2012-10-22 16:22:46]

LGTM. I'll land that for you.