Commit Graph

691 Commits

Author SHA1 Message Date
mstarzinger@chromium.org
eeca7c7759 Add GCTracer metrics for a scavenger GC for DOM wrappers
This patch adds the following three metrics for the --trace_gc_nvp option.

nodes_died_in_new_space_; // Number of died nodes in the new space.
nodes_copied_in_new_space_; // Number of copied nodes to the new space.
nodes_promoted; // Number of promoted nodes to the old space.

BUG=
TEST=Manually confirmed that the "--trace_gc --trace_gc_nvp" option prints the metrics

Review URL: https://codereview.chromium.org/11365146
Patch from Kentaro Hara <haraken@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13159 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-12-07 09:44:10 +00:00
mstarzinger@chromium.org
a4c4862ed8 Unify object groups iteration in global handles.
This unifies the iteration logic of objects groups iteration in global
handles. The scavenger as well as mark-and-compact now rely on the same
underlying logic.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/11299248

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13128 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-12-04 10:23:43 +00:00
mstarzinger@chromium.org
ff5e63761e Revert "Allow incremental marking when expose_gc is turned on."
This reverts r13072 from bleeding edge.

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/11316268

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13099 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-30 09:42:20 +00:00
svenpanne@chromium.org
5a4e0f1c79 Simplify and fix code aging.
Making the code size predictable is hard, and to make things even more
complicated, the start of a function can contain various stuff like calls to a
profiling hook, receiver adjustment or dynamic frame alignment. Instead of
tackling all these problems separately, we now simply record the offset where
patching should happen later in the Code object itself.

Review URL: https://codereview.chromium.org/11316218

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13081 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-29 07:38:00 +00:00
mstarzinger@chromium.org
50dcf96e63 Remove bogus assertion from idle notification.
Callbacks that run after garbage collection might fire a context dispose
notification and hence increse the counter before we actually reach this
position.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/11280214

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13080 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-28 17:56:17 +00:00
hpayer@chromium.org
c772d649cf Allow incremental marking when expose_gc is turned on.
BUG=

Review URL: https://codereview.chromium.org/11299154

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13072 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-28 08:43:10 +00:00
mstarzinger@chromium.org
d8922dd6f4 Move assertion about stopped marking up (again).
Weak persistent handle callbacks might trigger nested GCs that in turn
can cause marking to be restarted. The assertion needs to hold before
firing callbacks only.

R=ulan@chromium.org
TEST=cctest/test-api --stress-compaction

Review URL: https://codereview.chromium.org/11413188

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13069 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-27 17:03:12 +00:00
mstarzinger@chromium.org
3b291a0f48 Fix incremental marking to be off during heap snapshots.
The heap snapshot generator uses the UnreachableObjectsFilter which in
turn messes with marking bits. This requires incremental marking to be
turned off while the snapshot is being generated.

R=verwaest@chromium.org
TEST=cctest/test-heap-profiler --stress-compaction

Review URL: https://codereview.chromium.org/11413183

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13066 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-27 13:18:55 +00:00
ulan@chromium.org
735d052f4c Remove unused heap debug utils code.
This functionality is already implemented by PathTracer.

R=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/11412162

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13052 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-26 10:23:16 +00:00
verwaest@chromium.org
2c1ac55213 - Initialize the result array with holes if we concat a double array into an object array, since it may cause a marking step while boxing a double.
- Ensure we go holey if we are concatting any holey array.

Review URL: https://chromiumcodereview.appspot.com/11413142

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13038 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-22 16:22:57 +00:00
yangguo@chromium.org
72db2287c3 Add parallel recompilation time to histogram and plot execution pause times.
BUG=

Review URL: https://chromiumcodereview.appspot.com/11412125

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13036 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-22 13:04:11 +00:00
rossberg@chromium.org
ce05280bfc Get rid of static module allocation, do it in code.
Modules now have their own local scope, represented by their own context.
Module instance objects have an accessor for every export that forwards
access to the respective slot from the module's context. (Exports that are
modules themselves, however, are simple data properties.)

All modules have a _hosting_ scope/context, which (currently) is the
(innermost) enclosing global scope. To deal with recursion, nested modules
are hosted by the same scope as global ones.

For every (global or nested) module literal, the hosting context has an
internal slot that points directly to the respective module context. This
enables quick access to (statically resolved) module members by 2-dimensional
access through the hosting context. For example,

  module A {
    let x;
    module B { let y; }
  }
  module C { let z; }

allocates contexts as follows:

[header| .A | .B | .C | A | C ]  (global)
          |    |    |
          |    |    +-- [header| z ]  (module)
          |    |
          |    +------- [header| y ]  (module)
          |
          +------------ [header| x | B ]  (module)

Here, .A, .B, .C are the internal slots pointing to the hosted module
contexts, whereas A, B, C hold the actual instance objects (note that every
module context also points to the respective instance object through its
extension slot in the header).

To deal with arbitrary recursion and aliases between modules,
they are created and initialized in several stages. Each stage applies to
all modules in the hosting global scope, including nested ones.

1. Allocate: for each module _literal_, allocate the module contexts and
   respective instance object and wire them up. This happens in the
   PushModuleContext runtime function, as generated by AllocateModules
   (invoked by VisitDeclarations in the hosting scope).

2. Bind: for each module _declaration_ (i.e. literals as well as aliases),
   assign the respective instance object to respective local variables. This
   happens in VisitModuleDeclaration, and uses the instance objects created
   in the previous stage.
   For each module _literal_, this phase also constructs a module descriptor
   for the next stage. This happens in VisitModuleLiteral.

3. Populate: invoke the DeclareModules runtime function to populate each
   _instance_ object with accessors for it exports. This is generated by
   DeclareModules (invoked by VisitDeclarations in the hosting scope again),
   and uses the descriptors generated in the previous stage.

4. Initialize: execute the module bodies (and other code) in sequence. This
   happens by the separate statements generated for module bodies. To reenter
   the module scopes properly, the parser inserted ModuleStatements.

R=mstarzinger@chromium.org,svenpanne@chromium.org
BUG=

Review URL: https://codereview.chromium.org/11093074

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13033 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-22 10:25:22 +00:00
yangguo@chromium.org
36f5b6d41f Rename IsAsciiRepresentation
This is a straight rename:

IsAsciiRepresentation -> IsOneByteRepresentation
IsAsciiRepresentationUnderneath -> IsOneByteRepresentationUnderneath
AllocateRawAsciiString -> AllocateRawOneByteString
AllocateStringFromAscii -> AllocateStringFromOneByte

R=yangguo@chromium.org,
BUG=

Review URL: https://chromiumcodereview.appspot.com/11308066
Patch from Dan Carney <dcarney@google.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@13023 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-21 10:01:05 +00:00
yangguo@chromium.org
3699616609 Rename SeqAsciiString
This is a straight rename:

NewRawAsciiString -> NewRawOneByteString
SeqAscii -> SeqOneByte

SeqOneByteString cannot yet take non-ascii data.

R=yangguo@chromium.org,
BUG=

Review URL: https://chromiumcodereview.appspot.com/11411005
Patch from Dan Carney <dcarney@google.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12972 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-15 13:31:27 +00:00
verwaest@chromium.org
a08194c83a Support all fast elements kinds in the major array operations.
Currently missing support for unshift.

BUG=

Review URL: https://chromiumcodereview.appspot.com/11377132

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12969 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-15 12:19:14 +00:00
mstarzinger@chromium.org
d6a31725b9 Add missing initialization for low survival rate period.
R=verwaest@chromium.org
BUG=v8:2403

Review URL: https://codereview.chromium.org/11419002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12964 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-15 09:25:40 +00:00
jkummerow@chromium.org
1c086d1202 Lattice-based representation inference, powered by left/right specific type feedback for BinaryOps and comparisons
Review URL: https://chromiumcodereview.appspot.com/10837165

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12961 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-14 15:59:45 +00:00
ulan@chromium.org
a590fff76b Output max incremental step delay in trace_gc_nvp.
R=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/11377134

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12959 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-14 11:44:56 +00:00
svenpanne@chromium.org
673924413f Re-land rev. 12849 and 12868 (Heavy cleanup of the external pointer API + related fix).
Review URL: https://codereview.chromium.org/11365224

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12945 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-13 12:27:03 +00:00
svenpanne@chromium.org
09666b60b8 Rollback of r12868, r12849 on bleeding_edge.
Fixed visibility attribute for GetPointerFromInternalField

Heavy cleanup of the external pointer API.

Review URL: https://codereview.chromium.org/11359125

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12911 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-09 10:07:54 +00:00
danno@chromium.org
bbcfa17134 Use movw/movt on ARM to load constant roots
BUG=none
TEST=none

Review URL: https://codereview.chromium.org/11307012
Patch from Jay Conrod <dconrod@codeaurora.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12895 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-08 10:32:39 +00:00
mstarzinger@chromium.org
762c36b6fb Allow collection of DOM objects in minor GC cycles.
A design document: https://docs.google.com/a/google.com/document/d/16DeHrzkm3cO9XCPT1aK3Y5qgUxXB3RFmueqQWYmN2rI/edit

Performance & memory results: https://docs.google.com/a/google.com/document/d/1h0-EsHu7T0sSMuZm5eE0r1e8sCAzY3weLvsDUpOSngE/edit

The WebKit side patch: https://bugs.webkit.org/show_bug.cgi?id=98725

At present no one is using the V8 APIs this patch is going to add. After this patch is landed, the WebKit side patch will use it.

BUG=

Review URL: https://codereview.chromium.org/11085015
Patch from Kentaro Hara <haraken@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12874 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-06 17:32:15 +00:00
mstarzinger@chromium.org
c54fc70f13 Store Object.observe state per-isolate rather than per-context
This requires adding a new JSObject to the strong root list and populating it from
object-observe.js. The main other change is that we now directly use ObjectHashTable
from JS rather than using WeakMap, since using the latter would end up leaking whichever
Context initialized that observation state.

Added a test via the API showing that different contexts all end up working on the same state.

Review URL: https://codereview.chromium.org/11274014
Patch from Adam Klein <adamk@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12873 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-06 16:47:15 +00:00
mstarzinger@chromium.org
03ba764f3e Put incremental code flushing behind a flag.
This is used to disable incremental code flushing by default for now
until we can stabilize it and make it ready for production.

R=verwaest@chromium.org
BUG=chromium:159140

Review URL: https://codereview.chromium.org/11367105

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12862 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-06 11:54:05 +00:00
svenpanne@chromium.org
f3807ca17b Heavy cleanup of the external pointer API.
Added highly efficient Object::SetAlignedPointerInInternalField and
Object::GetAlignedPointerFromInternalField functions for 2-byte-aligned
pointers. Their non-aligned counterparts Object::GetPointerFromInternalField and
Object::SetPointerInInternalField are now deprecated utility functions.

External is now a true Value again, with New/Value/Cast using a JSObject with an
internal field containing a Foreign. External::Wrap, and External::Unwrap are now
deprecated utility functions.

Added Context::GetEmbedderData and Context::SetEmbedderData. Deprecated
Context::GetData and Context::SetData, these are now only wrappers to access
internal field 0.

Added highly efficient Context::SetAlignedPointerInEmbedderData and
Context::GetAlignedPointerFromEmbedderData functions for 2-byte-aligned
pointers.

Review URL: https://codereview.chromium.org/11190050

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12849 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-05 10:25:32 +00:00
yangguo@chromium.org
64793b3f0d Correctly visit all external strings.
BUG=

Review URL: https://chromiumcodereview.appspot.com/11340010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12841 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-11-02 12:45:00 +00:00
mstarzinger@chromium.org
014f00fa51 Fix code flusher to process weak function links.
This fixes a corner case where weak function links of the code flushing
candidates list were destroyed by scavenges that happened during
incremental marking. Now those weak function links are updated while
scavenging happens.

R=ulan@chromium.org
TEST=cctest/test-heap/TestCodeFlushingIncrementalScavenge

Review URL: https://codereview.chromium.org/11271006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12825 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-26 09:44:34 +00:00
mstarzinger@chromium.org
b55988625d Get rid of obsolete unchecked accessors.
R=yangguo@chromium.org
BUG=v8:1490

Review URL: https://codereview.chromium.org/11271020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12813 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-25 11:52:37 +00:00
mstarzinger@chromium.org
cee59bf522 Enable incremental code flushing.
This enables code flushing even with incremental marking enabled and
fully shares the function link field in JSFunctions between candidates
for code flushing and the optimized functions list. If a candidate for
code flushing gets optimized, it will be evicted from the candidates
list.

R=ulan@chromium.org
BUG=v8:1609

Review URL: https://codereview.chromium.org/11140025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12796 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-23 08:25:04 +00:00
mstarzinger@chromium.org
37759f4d64 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

Review URL: https://codereview.chromium.org/11066118
Patch from Alexei Filippov <alph@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12793 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-22 16:33:10 +00:00
yangguo@chromium.org
3c5e899378 Fix two-char hash to use correct fallback for zero hashes.
R=verwaest@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/11228004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12782 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-22 10:38:40 +00:00
verwaest@chromium.org
d04a1fff1f Keep track of the first non-ascii word/char to avoid redoing the work.
Review URL: https://chromiumcodereview.appspot.com/11194053

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12762 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-18 15:08:11 +00:00
verwaest@chromium.org
72424b3987 Move DescriptorArray into the map.
Review URL: https://chromiumcodereview.appspot.com/11188031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12746 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-17 13:04:49 +00:00
ulan@chromium.org
c969afe137 Reland r12342: Flush monomorphic ICs on context disposal instead of context exit.
The crashes that caused r12342 to be reverted are fixed in r12563.

R=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/11092081

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12715 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-12 13:49:12 +00:00
mvstanton@chromium.org
b3c0ed8245 Enable --verify-heap in release mode
R=mstarzinger@chromium.org
BUG=v8:2120

Review URL: https://codereview.chromium.org/11118018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12713 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-12 11:41:14 +00:00
verwaest@chromium.org
6e1bafc808 Revert "Allow partial scanning of large arrays in order to avoid"
This reverts commit r12619.

BUG=

Review URL: https://chromiumcodereview.appspot.com/11029023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12660 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-04 11:09:17 +00:00
mstarzinger@chromium.org
5b64362706 Revert r12625 due to sandbox incompatibility.
Original message: Implement committed physical memory stats for Linux.
This had to be reverted because it crashes when compiled into Chromium
due to the sandbox not allowing the mincore call.

R=verwaest@chromium.org
BUG=v8:2191

Review URL: https://codereview.chromium.org/11023010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12640 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-10-01 12:11:06 +00:00
mstarzinger@chromium.org
974a2268bb Implement committed physical memory stats for Linux.
The patch introduces CommittedPhysicalMemory function to the Heap class
that reports committed *physical* memory acquired from the OS.
It is important because some OSes may postpone actual commitment on e.g.
first access to the previously committed region.
So reporting just plain committed size led to various weird artifacts
like DevTools showing V8 allocated memory higher than the whole process
private size.

BUG=v8:2191

Review URL: https://codereview.chromium.org/10961042
Patch from Alexei Filippov <alph@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12625 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-27 13:27:50 +00:00
erik.corry@gmail.com
aba09dcf2e Allow partial scanning of large arrays in order to avoid
mark stack overflow.  This is a reland of r12609 -
https://chromiumcodereview.appspot.com/10959011 - but
this time VisitPointers has been fixed (it used to assume
that the first slot was on the first page of a large object).
Review URL: https://chromiumcodereview.appspot.com/10996018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12619 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-26 12:17:24 +00:00
erik.corry@gmail.com
e8ffc2bebd Make the speed of incremental marking depend also on the rate
at which we are hitting expensive write barrier operations,
not just on the rate of allocation.
Review URL: https://chromiumcodereview.appspot.com/10974003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12618 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-26 11:35:42 +00:00
mstarzinger@chromium.org
1f13fca165 Improve --trace-gc-verbose to show sum of all spaces.
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/10974006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12597 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-24 13:03:11 +00:00
verwaest@chromium.org
947663aaa3 Use NumberOfOwnDescriptors/EnumLength for counting properties on fast objects.
Also split CNLT into small functions.

Review URL: https://chromiumcodereview.appspot.com/10950023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12545 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-19 10:06:02 +00:00
erik.corry@gmail.com
bafcfe5427 Fix misplaced assert in heap.cc.
Bug=2336
Review URL: https://chromiumcodereview.appspot.com/10911334

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12528 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-17 11:38:24 +00:00
erik.corry@gmail.com
8924052787 Fix compile errors on Win64.
Review URL: https://chromiumcodereview.appspot.com/10913273

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12509 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-14 11:48:31 +00:00
erik.corry@gmail.com
5a8d1764bc Refactoring of snapshots. This simplifies and improves
the speed of deserializing code.  The current startup
time improvement for V8 is around 6%, but code deserialization
is speeded up disproportionately, and we will soon have more
code in the snapshot.
* Removed support for deserializing into large object space.
  The regular pages are 1Mbyte now and that is plenty.  This
  is a big simplification.
* Instead of reserving space for the snapshot we actually
  allocate it now.  This removes some special casing from
  the memory management and simplifies deserialization since
  we are just bumping a pointer rather than calling the
  normal allocation routines during deserialization.
* Record in the snapshot how much we need to boot up and
  allocate it instead of just assuming that allocations in
  a new VM will always be linear.
* In the snapshot we always address an object as a negative
  offset from the current allocation point.  We used to
  sometimes address from the start of the deserialized data,
  but this is less useful now that we have good support for
  roots and repetitions in the deserialization data.
* Code objects were previously deserialized (like other
  objects) by alternating raw data (deserialized with memcpy)
  and pointers (to external references, other objects, etc.).
  Now we deserialize code objects with a single memcpy,
  followed by a series of skips and pointers that partially
  overwrite the code we memcopied out of the snapshot.
  The skips are sometimes merged into the following
  instruction in the deserialization data to reduce dispatch
  time.
* Integers in the snapshot were stored in a variable length
  format that gives a compact representation for small positive
  integers.  This is still the case, but the new encoding can
  be decoded without branches or conditional instructions,
  which is faster on a modern CPU.
Review URL: https://chromiumcodereview.appspot.com/10918067

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12505 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-14 11:16:56 +00:00
verwaest@chromium.org
ebd3241b05 Sharing of descriptor arrays.
This CL adds multiple things:
Transition arrays do not directly point at their descriptor array anymore, but rather do so via an indirect pointer (a JSGlobalPropertyCell).

An ownership bit is added to maps indicating whether it owns its own descriptor array or not.

Maps owning a descriptor array can pass on ownership if a transition from that map is generated; but only if the descriptor array stays exactly the same; or if a descriptor is added.

Maps that don't have ownership get ownership back if their direct child to which ownership was passed is cleared in ClearNonLiveTransitions.

To detect which descriptors in an array are valid, each map knows its own NumberOfOwnDescriptors. Since the descriptors are sorted in order of addition, if we search and find a descriptor with index bigger than this number, it is not valid for the given map.

We currently still build up an enumeration cache (although this may disappear). The enumeration cache is always built for the entire descriptor array, even if not all descriptors are owned by the map. Once a descriptor array has an enumeration cache for a given map; this invariant will always be true, even if the descriptor array was extended. The extended array will inherit the enumeration cache from the smaller descriptor array. If a map with more descriptors needs an enumeration cache, it's EnumLength will still be set to invalid, so it will have to recompute the enumeration cache. This new cache will also be valid for smaller maps since they have their own enumlength; and use this to loop over the cache. If the EnumLength is still invalid, but there is already a cache present that is big enough; we just initialize the EnumLength field for the map.

When we apply ClearNonLiveTransitions and descriptor ownership is passed back to a parent map, the descriptor array is trimmed in-place and resorted. At the same time, the enumeration cache is trimmed in-place.

Only transition arrays contain descriptor arrays. If we transition to a map and pass ownership of the descriptor array along, the child map will not store the descriptor array it owns. Rather its parent will keep the pointer. So for every leaf-map, we find the descriptor array by following the back pointer, reading out the transition array, and fetching the descriptor array from the JSGlobalPropertyCell. If a map has a transition array, we fetch it from there. If a map has undefined as its back-pointer and has no transition array; it is considered to have an empty descriptor array.

When we modify properties, we cannot share the descriptor array. To accommodate this, the child map will get its own transition array; even if there are not necessarily any transitions leaving from the child map. This is necessary since it's the only way to store its own descriptor array.

Review URL: https://chromiumcodereview.appspot.com/10909007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12492 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-12 16:43:57 +00:00
yangguo@chromium.org
7276fa39ab Shorten code path in string allocation.
This was part of r12430 (http://codereview.chromium.org/10857030/) which has been reverted.

R=ulan@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/10913220

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12487 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-12 11:15:20 +00:00
yangguo@chromium.org
5ac2a5d22e Revert r12430, r12432, r12433 (basic support for Latin1).
BUG=

Review URL: https://chromiumcodereview.appspot.com/10905075

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12438 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-04 12:23:22 +00:00
yangguo@chromium.org
74f06b1f99 Add basic support for Latin1 to the API.
BUG=

Review URL: https://chromiumcodereview.appspot.com/10857030

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12430 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-09-03 15:06:36 +00:00
yangguo@chromium.org
5dd51bafef Cache results in SearchRegExpMultiple.
BUG=

Review URL: https://chromiumcodereview.appspot.com/10837290

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12416 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2012-08-31 09:28:01 +00:00