AuroraMiddleware/v8
1
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity
11,643 Commits 1 Branch 0 Tags 839 MiB
643350c5d2
Commit Graph

514 Commits

Author SHA1 Message Date
yangguo@chromium.org
74556569d1 Reland "Enable map dependency to in-flight compilation info." 
BUG=248076
R=ulan@chromium.org

Review URL: https://chromiumcodereview.appspot.com/16782004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15077 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-12 09:43:22 +00:00
yangguo@chromium.org
6da97b1d4a Revert "Enable map dependency to in-flight compilation info." 
This includes r15032, r15030 and r15005.

R=ulan@chromium.org
BUG=248076

Review URL: https://chromiumcodereview.appspot.com/16482004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15061 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-11 11:55:56 +00:00
jkummerow@chromium.org
9447014780 Skip some conditional deopts for Div/Mul when all uses are truncating. 
- set "can be minus zero" flag properly so minus-zero checks are skipped
- skip "integer result?" check in division code when uses are truncating
- drive-by cleanup: consolidated computation of kCanOverflow flag for Add/Sub into range inference phase

BUG=v8:2132
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/16741002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15060 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-11 11:43:57 +00:00
jkummerow@chromium.org
57632e208a Increase sanity of integer division handling on ARM 
- In the INT32 BinaryOpStub, fix type feedback collection for DIV,
  bringing it in line with other platforms.
- In Lithium codegen, emit proper inlined code, don't call the stub.
- Drive-by fix: assert appropriate CpuFeaturesScope for SDIV.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/16082008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15057 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-11 10:47:44 +00:00
svenpanne@chromium.org
66bb4c5b8d Improve code for integral modulus calculation. 
Depending on what we know about the right operand, we basically do 3
different things (and the code is actually structured this way):

* If we statically know that the right operand is a power of 2, we do
  some bit fiddling instead of doing a "real" modulus calculation.
  This should actually be done on the Hydrogen level, not on the
  Lithium level, but this will be a separate CL.

* If type feedback tells us that the right operand is a power of 2, we
  do the same as above, but guarded by conditional deoptimization to
  make sure that the assumption is still valid. In the long run, we
  should make this guard visible on the Hydrogen level to make it
  visible for GVN and other optimizations.

* In the general case we only do the minimum steps necessary and don't
  try to be too clever, because cleverness actually slows us down on
  real-world code.

If we look at the code gerators for LModI, we actually see that we
basically have 3 (4 on ARM) fundamentally different translations. I
don't really like lumping them together, they should probably be
different Lithium instructions. For the time being, I restructured the
generators to make this crystal-clear, at the cost of some duplication
regarding the power-of-2 cases. This will go away when we do the
strength reduction on the Hydrogen level, so I'd like to keep it as it
is for now.

Note that the MIPS part was only slightly restructured, there is still
some work to do there.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/15769010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15034 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-10 12:05:54 +00:00
yangguo@chromium.org
b0afb77731 Fix parallel recompilation wrt transition maps dependency. 
R=ulan@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/15896038

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15032 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-10 11:33:23 +00:00
yangguo@chromium.org
1594eca8ed Fix parallel recompilation wrt initial object/array map dependency. 
R=ulan@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/16641002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15030 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-10 11:17:48 +00:00
yangguo@chromium.org
17cfe68015 Enable map dependency to in-flight compilation info. 
R=ulan@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/16542003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15005 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-07 13:27:03 +00:00
mvstanton@chromium.org
8a02fd3be3 Turn off allocation site info for crankshafted array constructor calls. 
Once we crankshaft a method, we should turn off allocation site info for
constructed arrays. Additionally, the semantics for doing this were
awkward because the constructed array code stubs get an
AllocationSiteMode as a minor key, but it's used as a permission to
determine the final mode locally based on ElementsKind. I refactored
this to a simpler boolean for override or local control.

BUG=
R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/16206007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14934 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-04 12:48:51 +00:00
mstarzinger@chromium.org
d7975dfdab Deprecate HAllocateObject in favor of HAllocate. 
This models the instantiation of an implicit receiver for CallNew nodes
in hydrogen using HAllocate together with generic stores instead of one
specialized HAllocateObject instruction, hence creating a single choking
point for inlined allocation in optimized code.

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/15714005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14930 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-04 08:28:33 +00:00
yangguo@chromium.org
7f8a3d803c Make assertion scopes thread safe. 
R=svenpanne@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/15691017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14919 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-06-03 15:32:22 +00:00
verwaest@chromium.org
1a4a904bef Replace DeoptimizeOnUndefined with whitelisting AllowUndefinedAsNan 
R=danno@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15952007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14894 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-30 09:11:06 +00:00
olivf@chromium.org
88755fb22a Convert ToBooleanStub to a HydrogenStub. Currently just using the existing HBranch instruction, which is still fully implemented in Lithium. Will refactor HBranch in a next CL. 
[
CL breaks mips, since the deoptimizer in Deoptimizer::DoTranslateCommand(...) does not convert smis passed to the stub via a0.
]

BUG=
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/15302004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14886 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-29 14:49:28 +00:00
yurys@chromium.org
665c644940 Fix two CPU profiler tests on ARM and MIPS simulators 
Use Simulator::get_pc instead of Simulator::get_register(pc) as the latter
will return value adjusted by Instruction::kPCReadOffset.

BUG=v8:2628
R=loislo@chromium.org, svenpanne@chromium.org

Review URL: https://codereview.chromium.org/16154004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14880 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-29 12:09:03 +00:00
verwaest@chromium.org
7d32e7451f Don't explicitly pass requested representations to constants; implement ConstantS 
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15932011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14874 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-29 10:47:55 +00:00
verwaest@chromium.org
290909268a Update the gap resolver to support Smi constants. 
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15664005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14850 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-28 12:37:29 +00:00
hpayer@chromium.org
5b469a6f41 Added old data space allocation infrastructure for pretenuring. 
BUG=

Review URL: https://codereview.chromium.org/16099004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14843 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-28 09:38:28 +00:00
verwaest@chromium.org
cf9b1170d1 Support Smi in CompareIDAndBranch 
BUG=
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/16134003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14842 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-28 09:24:39 +00:00
verwaest@chromium.org
aa2444269b Fix the hole loading optimization. 
- Holes are only ever loaded as double or tagged.
- Change to tagged has to deoptimize on undefined (no implicit
  conversions from double the hole NaN -> tagged undefined).

BUG=
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/16099006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14829 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-27 17:33:14 +00:00
verwaest@chromium.org
a64b7f926f Remove HCheckSmi, LCheckSmi and rename LCheckSmiAndReturn to LCheckSmi. 
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15841007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14816 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-27 09:58:46 +00:00
verwaest@chromium.org
3cb8f95c91 Tag smi-constants as smi. This also fixes code that copies holes into arrays. 
BUG=
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15861009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14814 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-27 08:43:58 +00:00
verwaest@chromium.org
2a28f53b18 Fix hole handling, and ensure smi representation is handled properly 
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/16013003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14807 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-24 13:40:02 +00:00
titzer@chromium.org
77f85fa0d3 Remove offset() and is_in_object() from hydrogen and lithium LoadNamedField and StoreNamedField and use the appropriate methods on HObjectAccess instead 
BUG=

Review URL: https://codereview.chromium.org/15881003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14806 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-24 12:40:08 +00:00
verwaest@chromium.org
c9a22a3c8d Replace tagged keys for fast access by smi, and use smi in boundscheck. 
BUG=
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15763004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14794 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-24 08:52:35 +00:00
verwaest@chromium.org
632f591884 Tag length of FixedArrayBase and smi-array[x] as smi representation 
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15858006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14778 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-23 14:38:39 +00:00
yangguo@chromium.org
a1e18bdf3c Improve SeqStringSetChar implementation. 
R=jkummerow@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/15743006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14769 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-23 09:51:06 +00:00
verwaest@chromium.org
06b7bf60e1 Always require exact maps. 
R=danno@chromium.org

Review URL: https://chromiumcodereview.appspot.com/13923003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14768 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-23 09:19:18 +00:00
verwaest@chromium.org
308e69755b Implement HChange support for Smis and use it in Load/StoreNameField 
BUG=
R=verwaest@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15303004

Patch from Daniel Clifford <danno@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14765 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-23 08:32:07 +00:00
verwaest@chromium.org
d1b5a2518d Handle holes in smi-untag from LoadKeyed requiring hole handling. 
R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/15737003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14756 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-22 16:32:33 +00:00
rodolph.perfetta@gmail.com
45ec481659 ARM: Smi refactoring and improvements. 
Refactoring:
 * consistent use of SmiTag/Untag
 * added a few Smi macros and helpers
Improvements
 * small optimisations (e.g. merging untag and cmp #0)
 * added fixed point to double conversion instructions for simpler conversions

More on the last point: a Smi can be seen as a fixed point number with the
a one bit fractional part. Fixed to double instructions allow us to convert
a Smi to a double without untagging.

BUG=none
TEST=none

Review URL: https://chromiumcodereview.appspot.com/15085026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14724 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-17 15:38:14 +00:00
olivf@chromium.org
8bbe48c7b9 Add a HBreak instruction for debugging 
BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14997008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14691 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-15 14:24:47 +00:00
yurys@chromium.org
69c2f54d32 Skip samples where top function's stack frame is not setup properly 
Stack iterator takes return address based on the frame pointer (ebp) and detects JS frames based on value at fp + StandardFrameConstants::kMarkerOffset. So in order the iterator to work correctly this values should be already setup for the current function. Stack frame is constructed at the very beginning of JS function code and destroyed before return. If sample is taken before before the frame construction is completed or after it was destroyed the stack iterator will wrongly think that FP points at the current functions frame base and will skip callers frame. To avoid this we mark code ranges where  stack frame doesn't exist and completely ignore such samples.

This fixes cctest/test-cpu-profiler/CollectCpuProfile flakiness.

BUG=v8:2628
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14253015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14670 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-14 22:51:33 +00:00
danno@chromium.org
6862babfd6 Remove HIsNilAndBranch (it's now unused) 
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/14971005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14660 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-14 12:56:53 +00:00
danno@chromium.org
b65b6d735f Ensure that soft-deopts don't count against opt_count 
This makes sure that Crankshaft doesn't disable optimization to early on hot functions that still contain unexecuted code without type information.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14738009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14658 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-14 11:45:33 +00:00
danno@chromium.org
05e8e0e7b4 Elide hole checks on KeyedLoads of holey double arrays 
Improves NavierStokes by about 5%

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/15014020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14630 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-13 07:35:26 +00:00
verwaest@chromium.org
df57747fc4 Track heap objects. 
R=danno@chromium.org

Review URL: https://chromiumcodereview.appspot.com/14996004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14625 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-10 17:17:50 +00:00
mstarzinger@chromium.org
8829a94c24 Remove obsolete HArrayLiteral instruction. 
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/14847013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14613 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-10 09:52:08 +00:00
mstarzinger@chromium.org
efe91b56f6 Remove obsolete HObjectLiteral instruction. 
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/14972008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14612 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-10 09:47:05 +00:00
mvstanton@chromium.org
ef5211d8af Parallel compilation bug: LCallNewArray dereferenced a handle unnecessarily. 
BUG=
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/14761010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14599 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-08 16:48:27 +00:00
verwaest@chromium.org
52008429b7 Use mutable heapnumbers to store doubles in fields. 
R=danno@chromium.org

Review URL: https://chromiumcodereview.appspot.com/14850006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14597 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-08 15:02:08 +00:00
titzer@chromium.org
dbcd0e925a Remove HLoadElements instruction and replace with use of more general HLoadNamedField. This removes the need for both the hydrogen and lithium instructions and allows both kinds of loads to be GVN'd. 
Review URL: https://codereview.chromium.org/14556020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14542 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-03 12:21:16 +00:00
verwaest@chromium.org
8b25614aa9 Always tag number candidate as smi if it does not come from load-keyed. 
Review URL: https://chromiumcodereview.appspot.com/14556007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14521 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-05-02 14:41:06 +00:00
svenpanne@chromium.org
f1e9ec2e12 Do not use repeated subtractions in mod-i, it hurts performance in general. 
ARM only: Check for -0 only when really necessary. Tiny code de-duplication.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14499011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14479 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-29 13:45:34 +00:00
verwaest@chromium.org
4b010329c9 Fix arm port of tracking field representations. Based on Rodolph Perfetta's comments 
BUG=

Review URL: https://chromiumcodereview.appspot.com/14522003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14470 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-26 19:21:54 +00:00
verwaest@chromium.org
99e17bb12a Track storage types of instance variables. 
Review URL: https://chromiumcodereview.appspot.com/14146005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14464 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-26 15:30:41 +00:00
mvstanton@chromium.org
84355ae411 Lithium: avoid registers for constants when possible. 
BUG=

Review URL: https://codereview.chromium.org/14364010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14460 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-26 14:04:07 +00:00
mvstanton@chromium.org
e751ad06d9 Constructed arrays can be created with Hydrogen code stubs. The feature is still off by default (--optimize-constructed-arrays). 
BUG=

Review URL: https://codereview.chromium.org/12385014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14441 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-25 16:00:32 +00:00
yangguo@chromium.org
28bbe46a98 Disallow dereferencing deferred handles when generating optimized code. 
R=mvstanton@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14403015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14388 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-23 09:23:07 +00:00
rodolph.perfetta@gmail.com
eed788a7bb ARM: VFP cleanup now that VFP2 is the baseline. 
BUG=none
TEST=none

Review URL: https://chromiumcodereview.appspot.com/14119008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14382 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-22 21:30:57 +00:00
svenpanne@chromium.org
07a2a9cd19 Various improvements regarding the way we print code code comments. 
* All Lithium instructions have an associated Hydrogen instruction now,
  simplifying things.

* Consistently print <Lithium instruction number,Hydrogen value id> prefixes.

* Do not print uninteresting Lithium instructions like empty gaps, jumps to the
  next instruction, etc.

* Removed special handling of HChange-like instructions, it is totally unclear
  why they had this special treatment. If we really want to print more
  information about Lithium instructions, we should do it in a totally way,
  anyway (e.g. by unifying things with the generation of hydrogen*.cfg files).

* Made deferred code and the jump table stand out a little bit more.

* Print info about special blocks like loop headers and OSR entries.

Review URL: https://codereview.chromium.org/14371005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14370 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2013-04-22 09:48:35 +00:00