Commit Graph

513 Commits

Author SHA1 Message Date
bmeurer@chromium.org
b09989ec58 Refactor transitioning stores.
There's actually no need to have the transition as part of the HStoreNamedField instruction. In fact, it is cleaner and faster to generate a separate HStoreNamedField for the transition map. This will also help to eliminate map stores with store elimination, as well as reduce register pressure for transitioning stores on ia32.

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/295743002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21383 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-20 11:06:23 +00:00
ishell@chromium.org
6e74578968 Use stability to only conditionally flush information from the map check table.
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/266083007

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21274 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-12 20:05:52 +00:00
bmeurer@chromium.org
c3b5f70414 Use HType::NonPrimitive() for unclassified HeapObjects (i.e. Maps).
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/279813002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21215 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-09 12:05:56 +00:00
jarin@chromium.org
3976ebef93 Make new space iterable for --log-gc and --heap-stats options
R=hpayer@chromium.org
BUG=370827
TEST=test/mjsunit/regress/regress-370827.js
LOG=N

Review URL: https://codereview.chromium.org/272503005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21209 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-09 09:23:10 +00:00
bmeurer@chromium.org
7c45d49861 Improve Array.shift() performance for small arrays.
TEST=mjsunit/array-shift,mjsunit/array-shift2,mjsunit/array-shift3
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/279743002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21203 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-09 08:28:25 +00:00
bmeurer@chromium.org
a3fed0af84 Fix constructors for HLoadNamedField.
Also try to determine an appropriate HType for tracked fields,
instead of (incorrectly) using HType::NonPrimitive() everywhere.

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/269353003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21186 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-07 08:41:35 +00:00
bmeurer@chromium.org
a051bc5762 Next bunch of fixes for check elimination.
- Canonicalize HCheckMapValue with constant map to
  HCheckMaps, and get rid of the special treatment
  during check elimination.
- Track only stable object maps for HConstants and
  add CHECK()s to verify state during code generation.

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/263923004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21154 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-06 07:05:07 +00:00
bmeurer@chromium.org
785bdf7c1e Don't add code dependencies eagerly for HCheckMaps.
Instead of adding code dependencies on stable during
graph creation, we now add them during code generation
for those HCheckMaps that survived dead code elimination.

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/264973013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@21139 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-05 11:03:14 +00:00
bmeurer@chromium.org
7c1036d424 Various cleanups in check elimination.
- Move IsMapAccess() to HObjectAccess.
- Collect the actual objects instead of HStoreNamedFields in
  effects, and also consider HTransitionElementsKind.
- Fix ownership of HCheckMaps::maps() and HLoadNamedField::maps().
- Avoid heavy copying of the same map sets all the time during
  check elimination, and do something useful with the memory
  instead by slightly bumping the maximum number of tracked
  objects.
- Slightly optimize UniqueSet::Contains().

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/264693011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@21133 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-05 06:53:19 +00:00
bmeurer@chromium.org
b7148380cc Drop obsolete HCheckMaps::HandleSideEffectsDominator().
In HCheckMaps::HandleSideEffectsDominator() we try to remove
redundant HCheckMaps instructions that are dominated by an
HStoreNamedField instruction with a transition. This is just
a special case of the check elimination and therefore obsolete
(it also gives some nice speedup since GVN just got cheaper).

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/261003002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@21108 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-05-02 05:59:07 +00:00
alexandre.rames@arm.com
67ea9e4b42 ARM64: Generate optimized code for Math.floor and Math.round with double outputs.
R=jkummerow@chromium.org, ulan@chromium.org

Review URL: https://codereview.chromium.org/258793002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@21091 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-30 13:38:00 +00:00
bmeurer@chromium.org
d4b533d41b Bulk update of Google copyright headers in source files.
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/259183002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@21035 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-29 06:42:26 +00:00
verwaest@chromium.org
d2179f2062 Don't adopt the AST id from previous if id is none, since previous may have mismatching expected stack height.
Additionally, harden merging of simulates after instructions with side effects and ensure there's a simulate before HEnterInlined.

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/252583004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20967 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-25 09:52:11 +00:00
verwaest@chromium.org
a55821eef2 Mark the simulate before EnterInlined with BailoutId::None(), and set ReturnId on EnterInlined. When merging simulates into the simulate before enter-inlined, adopt the last AST id that gets merged into it.
BUG=v8:3282
LOG=n
R=titzer@chromium.org

Review URL: https://codereview.chromium.org/257583004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20949 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-24 15:20:53 +00:00
bmeurer@chromium.org
9437b59a6c HBranch should also print its expected input types.
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/247153003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20881 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-22 11:00:27 +00:00
bmeurer@chromium.org
376bff2411 Optimize numeric comparison with known successors.
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/246133005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20880 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-22 10:49:28 +00:00
bmeurer@chromium.org
42c67d5fa2 Allow merging of monomorphic accesses to tracked fields.
Also add stability dependency only on maps that can transition,
and delay adding the dependencies until we are actually using
them, either in a HLoadNamedField or an HCheckMaps.

TEST=mjsunit/field-type-tracking
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/239923004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20796 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-16 11:41:09 +00:00
bmeurer@chromium.org
b5cec2b72f Handlify AddDependentCode(), AddDependentCompilationInfo() and AddDependentIC().
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/236193014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20756 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-15 10:14:50 +00:00
bmeurer@chromium.org
6b4d4b7287 Reland "Track field types.".
This is an initial step towards tracking the exact types instead of just
the representations of fields. It adds support to track up to one map of
heap object field values, eliminating various map checks on values
loaded from such fields, at the cost of making stores to such fields
slightly more expensive.

Issues with transitioning stores and fast object literals in Crankshaft
fixed.

TEST=mjsunit/field-type-tracking
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/238773002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20746 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-15 07:36:47 +00:00
jarin@chromium.org
c1a3ab6b4f Revert "Track field types."
Revert r20701.

TBR=bmeurer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/236843002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20704 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-14 08:24:15 +00:00
bmeurer@chromium.org
9cf3909975 Track field types.
This is an initial step towards tracking the exact types instead of just the representations of fields. It adds support to track up to one map of heap object field values, eliminating various map checks on values loaded from such fields, at the cost of making stores to such fields slightly more expensive.

TEST=mjsunit/field-type-tracking
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/167303005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20701 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-14 06:29:15 +00:00
svenpanne@chromium.org
c5231ccba6 Make sure that ranges are not accessed after range analysis. Remove HValue::PrintRangeTo.
The ranges are simply wrong after range analysis, and we should only rely on computed flags.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/234583005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20674 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-11 09:17:18 +00:00
jarin@chromium.org
fd988331ea There is no definition for HArgumentsObject, so LDummyUse confuses the register allocator. I have recently made similar fix for HCapturedObject (see https://codereview.chromium.org/222283002/).
BUG=
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/226613007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20663 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-11 06:29:51 +00:00
yangguo@chromium.org
4fcc06685f Handlify LookupSingleCharacterStringFromCode.
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/227593004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20559 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-08 06:45:53 +00:00
svenpanne@chromium.org
d4ff61058a Fix approximation of CanDeoptimize() and other small store elimination fixes.
R=titzer@chromium.org

Review URL: https://codereview.chromium.org/222263002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20468 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-03 10:39:04 +00:00
jarin@chromium.org
42d2d3cb9d Do not generate LDummyUse instruction for HCapturedObject
LDummyUse confuses the register allocator (since there is no definition
for the use).

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/222283002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20461 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-03 07:35:13 +00:00
verwaest@chromium.org
a99050d40e Set kTruncatingToSmi for HChange only when To rep is Smi
BUG=
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/218613002

Patch from Weiliang Lin <weiliang.lin@intel.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20385 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-04-01 10:32:25 +00:00
mvstanton@chromium.org
839588d411 HInnerAllocatedObject hydrogen trace printed offset incorrectly.
R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/210873003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20256 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-25 16:32:23 +00:00
yangguo@chromium.org
82f630a9f7 Reland "No longer OOM on invalid string length."
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/210683003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20225 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-25 09:09:24 +00:00
titzer@chromium.org
3c31102025 First implementation of store elimination.
BUG=
R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/100253004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20224 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-25 09:06:16 +00:00
yangguo@chromium.org
72932ae417 Revert "No longer OOM on invalid string length."
This reverts r20202.

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/210143002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20203 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-24 15:36:15 +00:00
yangguo@chromium.org
531217502c No longer OOM on invalid string length.
R=ishell@chromium.org
BUG=v8:3060
LOG=Y

Review URL: https://codereview.chromium.org/207613005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20202 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-24 15:01:17 +00:00
jochen@chromium.org
2ce0bebba1 Rename A64 port to ARM64 port
BUG=354405
R=ulan@chromium.org, rodolph.perfetta@arm.com
LOG=y

Review URL: https://codereview.chromium.org/207823003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20148 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-21 09:28:26 +00:00
svenpanne@chromium.org
68237f6590 Implement flooring division by a constant via truncating division by a constant.
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/204583002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20123 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-20 13:10:23 +00:00
ishell@chromium.org
f77c51b0a6 Check elimination now sets known successor branch of HCompareObjectEqAndBranch (correctness fix).
BUG=chromium:352058
LOG=N
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/196383018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19964 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-17 09:11:38 +00:00
bmeurer@chromium.org
4da15df008 Remove uses of CanBeNegative() in HMod.
BUG=v8:3204
LOG=y
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/195793016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19878 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-13 08:17:44 +00:00
bmeurer@chromium.org
929313bd6b Remove uses of RangeCanInclude() in flooring division by power of 2.
Drive-By-Fix: Improve ARM code generation for flooring division by
power of 2.

BUG=v8:3204
LOG=y
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/196653009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19877 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-13 07:58:58 +00:00
bmeurer@chromium.org
750f2d98f8 Fix uses of range analysis results in HChange.
BUG=v8:3204
LOG=y
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/195023002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19872 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-13 06:11:52 +00:00
jkummerow@chromium.org
105c1e08b7 Fix HIsSmiAndBranch::KnownSuccessorBlock() by deleting it
Constants can still change their representation, so we cannot determine reachability of blocks based on their Smi-ness

BUG=chromium:351320
LOG=y
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/196943002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19836 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-12 10:14:29 +00:00
danno@chromium.org
ae1669b501 Fix handling of polymorphic array accesses with constant index
R=jkummerow@chromium.org
BUG=chromium:351319
LOG=Y

Review URL: https://codereview.chromium.org/196353004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19835 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-12 10:11:38 +00:00
jkummerow@chromium.org
8a1812f252 Fix lazy deopt after tagged binary ops
Also add policing code to ensure that optimized frames can in fact lazily deopt
at their respective current PC when we patch them for lazy bailout.

BUG=chromium:350434
LOG=y
R=jarin@chromium.org

Review URL: https://codereview.chromium.org/194703008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19834 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-12 09:59:36 +00:00
svenpanne@chromium.org
d6f673eba8 Fixed range analysis for HMathFloorOfDiv.
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/195133002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19831 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-12 06:49:28 +00:00
yangguo@chromium.org
6e1507331e Fix bug in constant folding object comparisons.
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/195063002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19798 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-11 13:34:01 +00:00
yangguo@chromium.org
e25d51cc85 Fix constant folding of %_IsMinusZero.
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/190793015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19762 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-10 15:06:54 +00:00
yangguo@chromium.org
78d23e5662 Implement KnownSuccessor method to some control instructions.
R=jkummerow@chromium.org
BUG=v8:3118
LOG=N

Review URL: https://codereview.chromium.org/174863002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19759 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-10 14:50:01 +00:00
bmeurer@chromium.org
c981914d4c Replace the recursion in PropagateMinusZeroChecks() with a loop and a worklist.
Also refactor the related code in preparation for fixing the
range analysis.

BUG=v8:3204
LOG=y
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/190713002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19737 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-10 05:52:03 +00:00
verwaest@chromium.org
8a3d715250 Revert "Use Representation::Integer32() for smi types on 32-bit-tagged systems."
Due to performance regression.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/189843006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19709 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-07 09:29:07 +00:00
jkummerow@chromium.org
1cc0bafc07 Fix HConstants with Smi-ranged HeapNumber values
BUG=chromium:349878
LOG=y
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/186123003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19693 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-06 16:21:09 +00:00
verwaest@chromium.org
7bf33c53eb Use Representation::Integer32() for smi types on 32-bit-tagged systems.
BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/187353005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19684 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-06 09:49:10 +00:00
bmeurer@chromium.org
766af5c2a8 Fix HStringAdd::PrintDataTo().
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/185403005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19646 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2014-03-04 12:34:04 +00:00