Create a new paged heap space for global property cells. The new...

Create a new paged heap space for global property cells.  The new
space is similar to map space in that it has fixed-size objects.  A
common superclass for a space with fixed size objects is used for the
map space and cell space.

Allocate all cells in cell space.  Handle it during all GCs.  Modify
the free-list node representation (so that the size is not at a fixed
offset in all cells) to allow two-pointer free-list nodes.  Clean up
some stuff in the MC collector.


Committed: http://code.google.com/p/v8/source/detail?r=2411

[Kasper Lund, 2009-07-08 11:31:16]

LGTM.

http://codereview.chromium.org/155211/diff/20/1016
File src/heap.cc (right):

http://codereview.chromium.org/155211/diff/20/1016#newcode3129
Line 3129: // Try to shrink global property cell, map, old, and code spaces.
Maybe just cell, map ...

http://codereview.chromium.org/155211/diff/20/1016#newcode3130
Line 3130: cell_space_->Shrink();
If the order isn't important here, maybe we should look into doing it in the
same order everywhere? Not a big deal.

http://codereview.chromium.org/155211/diff/20/1024
File src/mark-compact.cc (right):

http://codereview.chromium.org/155211/diff/20/1024#newcode1753
Line 1753: 
Two newlines here?

http://codereview.chromium.org/155211/diff/20/1017
File src/serialize.h (right):

http://codereview.chromium.org/155211/diff/20/1017#newcode324
Line 324: List<Page*> cell_pages_;  // All pages in the global property cell
space.
the cell space.

http://codereview.chromium.org/155211/diff/20/1022
File src/spaces.cc (right):

http://codereview.chromium.org/155211/diff/20/1022#newcode1460
Line 1460: Reset();
Weird indentation.

[bak, 2009-07-08 11:45:57]

I would eliminate the remember set code for the cell space and add mandatory
iterations of global cells before a commit.
Otherwise, LGTM,
  Lars

http://codereview.chromium.org/155211/diff/20/1026
File src/globals.h (right):

http://codereview.chromium.org/155211/diff/20/1026#newcode268
Line 268: LAST_SPACE = LO_SPACE  // <= 9 (see kSpaceBits and kLOSpacePointer)
Please check the extra bit is needed. 
AND please add an assert checking the LAST_SPACE can be encoded in kSpaceTagSize
bits.

http://codereview.chromium.org/155211/diff/20/1016
File src/heap.cc (right):

http://codereview.chromium.org/155211/diff/20/1016#newcode850
Line 850: cell_space_->ClearRSet();
I'm confused. Why build a remembered set for a space where we are about to
ignore it????????

http://codereview.chromium.org/155211/diff/20/1016#newcode2850
Line 2850: space == cell_space_ ||
Why do we iterate over cell_space_?

[Kevin Millikin (Chromium), 2009-07-08 13:17:33]

New changelist uploaded.  Take a quick look at the diff.

http://codereview.chromium.org/155211/diff/20/1026
File src/globals.h (right):

http://codereview.chromium.org/155211/diff/20/1026#newcode268
Line 268: LAST_SPACE = LO_SPACE  // <= 9 (see kSpaceBits and kLOSpacePointer)
The extra bits are used for two "pseudo" tags in the serializer.  I've changed
the serializer to use 4 bits, everywhere else (ie, RetryAfterGC failures) to use
3.

Added asserts for both places.

http://codereview.chromium.org/155211/diff/20/1016
File src/heap.cc (right):

http://codereview.chromium.org/155211/diff/20/1016#newcode850
Line 850: cell_space_->ClearRSet();
I wanted to put the allocation change in first, then change GC to stop reading
rset bits, then stop writing them.  It's asier to review, and if something goes
wrong, easier to diagnose.  (On the other hand, it's harder to back out if we
have to.)

http://codereview.chromium.org/155211/diff/20/1024
File src/mark-compact.cc (right):

http://codereview.chromium.org/155211/diff/20/1024#newcode1753
Line 1753: 
On 2009/07/08 11:31:16, Kasper Lund wrote:
> Two newlines here?

Done.

http://codereview.chromium.org/155211/diff/20/1017
File src/serialize.h (right):

http://codereview.chromium.org/155211/diff/20/1017#newcode324
Line 324: List<Page*> cell_pages_;  // All pages in the global property cell
space.
On 2009/07/08 11:31:16, Kasper Lund wrote:
> the cell space.

Done.

http://codereview.chromium.org/155211/diff/20/1022
File src/spaces.cc (right):

http://codereview.chromium.org/155211/diff/20/1022#newcode1460
Line 1460: Reset();
On 2009/07/08 11:31:16, Kasper Lund wrote:
> Weird indentation.

Done.

[bak, 2009-07-08 13:53:17]

LGTM,
  Lars