Avoid spilling a live range that has a register use at the next instruction.
The register use position has to be after the end of the next instruction so
that we don't end up trying to split a live range at the beginning.
BUG=105112
Review URL: http://codereview.chromium.org/8743011
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10095 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
The debugger preparation did not take optimized functions - including
inlined function into account. This caused the full-code used for
deoptimization to be the "lazy compile" builtin which did not work and
caused V8 to crash.
R=yangguo@chromium.org
BUG=chromium:105375, v8:1782
TEST=test/mjsunit/debug-break-inline.js
Review URL: http://codereview.chromium.org//8728031
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10094 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
a mark-sweep. We have a soft limit on old space size, which is designed to
trigger an old-space collection when we hit it. Unfortunately although the
soft limit had already triggered an old space collection, the soft limit was
preventing objects from new space from being promoted. For every promotion
candidate we were checking 3 different ways to allocate in old space before
giving up and putting the object in the other semispace. This change allows
the promoted objects to go to old space and also makes us more eager to
sweep a page before trying other ways to find space for an object.
Review URL: http://codereview.chromium.org/8748005
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10092 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This shaves 416+ KB, just under 1% off the size of the debug d8 executable
on Linux (mostly because the CheckHelper functions for assertions were
getting separate copies for each compilation unit). The difference in
release builds is negligible---a size reduction of 0.1%.
Also, change namespace-level 'static const' variables to remove the static
storage class as it's the default.
R=danno@chromium.org
BUG=
TEST=
Review URL: http://codereview.chromium.org/8680013
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10083 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
The ES.next drafts require that source code that matches the productions for
let and const bindings outside the extended mode trigger early syntax
errors. This CL adapts the parser / preparser accordingly under the harmony
scoping flag.
Summary:
* Harmony scoping flag not set: Old semantics allowing const in classic mode
with function level scope. Const binding in strict mode and let bindings in
classic and strict mode trigger early syntax errors.
* Harmony scoping is set: Use new harmony const and let in
extended mode and old const in classic mode. This is to preserve
compatibility with current web pages that already use
non-standard implementations of const. An early syntax error is
thrown on const in strict mode and on let in classic and strict
mode.
This depends on:
http://codereview.chromium.org/8562002/
TEST=mjsunit/harmony/block-early-errors.js
Review URL: http://codereview.chromium.org/8564001
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10079 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
source code positions it gets from the program counter to recreate the scope
chain by reparsing the function or program.
This CL includes the following changes
* Adds source code positions for the assignment added by the rewriter.
* Run the preparser over global code first.
* Use the ScopeType from the ScopeInfo to determine if the code being debugged
is eval, function or global code instead of looking up the '.result' symbol.
TEST=mjsunit/debug-stepout-scope.js
Review URL: http://codereview.chromium.org/8590027
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10076 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
So far the parser had its own harmony flag to disable the harmony scoping
feature when parsing native functions. With the introduction of the extended
language mode this becomes unnecessary because native functions will never enter
the extended mode. The parser can thus track FLAG_harmony_scoping and the
language mode of the current scope to see if harmony features are allowed. The
scanner and preparser have to keep their flag, because they can't use
FLAG_harmony_scoping as it is not available for the preparser-process
executable.
This depends on:
http://codereview.chromium.org/8417035/
Review URL: http://codereview.chromium.org/8562002
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10063 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This CL introduces a third mode next to the non-strict
(henceforth called 'classic mode') and 'strict mode'
which is called 'extended mode' as in the current
ES.next specification drafts. The extended mode is based on
the 'strict mode' and adds new functionality to it. This
means that most of the semantics of these two modes
coincide.
The 'extended mode' is entered instead of the 'strict mode'
during parsing when using the 'strict mode' directive
"use strict" and when the the harmony-scoping flag is
active. This should be changed once it is fully specified how the 'extended mode' is entered.
This change introduces a new 3 valued enum LanguageMode
(see globals.h) corresponding to the modes which is mostly
used by the frontend code. This includes the following
components:
* (Pre)Parser
* Compiler
* SharedFunctionInfo, Scope and ScopeInfo
* runtime functions: StoreContextSlot,
ResolvePossiblyDirectEval, InitializeVarGlobal,
DeclareGlobals
The old enum StrictModeFlag is still used in the backend
when the distinction between the 'strict mode' and the 'extended mode' does not matter. This includes:
* SetProperty runtime function, Delete builtin
* StoreIC and KeyedStoreIC
* StubCache
Review URL: http://codereview.chromium.org/8417035
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10062 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This generates optimized code for deep-copying of nested object literal
boilerplates which are statically known. Most of the boilerplates have
already been generated at crankshaft time, so this optimization should
kick in for virtually every object literal. Only nested object literal
graphs up to a certain depth and containing up to a certain total number
of properties are considered for this optimization. This will prevent
explosion of code size due to large object literals (e.g. eval on JSON).
Improves splay performance because object literals are created often.
R=fschneider@chromium.org
Review URL: http://codereview.chromium.org/8640001
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10061 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Port r9977 (9aded78be4).
Note: This is the port of the reapplied patch, not the original.
Original commit message:
To support deoptimization of exception handlers, the handler address in the
stack is converted to a pair of code object and an index into a separate
table of code offsets. The index part is invariant under deoptimization.
The index is packed into the handler state field so that handler size does
not change.
BUG=
TEST=
Review URL: http://codereview.chromium.org/8557003
Patch from Gergely Kis <gergely@homejinni.com>.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10039 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Port r10006 (4b344a03).
Original commit message:
Changes the way we do lazy deoptimization:
1. For side-effect instructions, we insert the lazy-deopt call at
the following LLazyBailout instruction.
CALL
GAP
LAZY-BAILOUT ==> lazy-deopt-call
2. For other instructions (StackCheck) we insert it right after the
instruction since the deopt targets an earlier deoptimization environment.
STACK-CHECK
GAP ==> lazy-deopt-call
The pc of the lazy-deopt call that will be patched in is recorded in the
deoptimization input data. Each Lithium instruction can have 0..n safepoints.
All safepoints get the deoptimization index of the associated LAZY-BAILOUT
instruction. On lazy deoptimization we use the return-pc to find the safepoint.
The safepoint tells us the deoptimization index, which in turn finds us the
PC where to insert the lazy-deopt-call.
Additional changes:
* RegExpLiteral marked it as having side-effects so that it
gets an explicitlazy-bailout instruction (instead of
treating it specially like stack-checks)
* Enable target recording CallFunctionStub to achieve
more inlining on optimized code.
BUG=
TEST=
Review URL: http://codereview.chromium.org/8587008
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10020 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
EmitIsObject is only called from one function, DoIsObjectAndBranch.
DoIsObjectAndBranch now passes the temp2 register to EmitIsObject
instead of simply relying on the two functions using the same
lithium scratch register.
BUG=
TEST=
Review URL: http://codereview.chromium.org/8588004
Patch from Gergely Kis <gergely@homejinni.com>.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10017 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Changes:
-separated the two code-paths (constant power of 2 divisor similar to ARM and everything else)
-replaced a bailout condition with faster handling (negative modulo result)
-removed a possibly useless mov instruction from one path
-replaced the IsConstantOperand condition with the more meaningful HasPowerOf2Divisor (although in this specific case they're equivalent)
BUG=
TEST=
Review URL: http://codereview.chromium.org/8591002
Patch from Gergely Kis <gergely@homejinni.com>.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10016 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Changes the way we do lazy deoptimization:
1. For side-effect instructions, we insert the lazy-deopt call at
the following LLazyBailout instruction.
CALL
GAP
LAZY-BAILOUT ==> lazy-deopt-call
2. For other instructions (StackCheck) we insert it right after the
instruction since the deopt targets an earlier deoptimization environment.
STACK-CHECK
GAP ==> lazy-deopt-call
The pc of the lazy-deopt call that will be patched in is recorded in the
deoptimization input data. Each Lithium instruction can have 0..n safepoints.
All safepoints get the deoptimization index of the associated LAZY-BAILOUT
instruction. On lazy deoptimization we use the return-pc to find the safepoint.
The safepoint tells us the deoptimization index, which in turn finds us the
PC where to insert the lazy-deopt-call.
Additional changes:
* RegExpLiteral marked it as having side-effects so that it
gets an explicitlazy-bailout instruction (instead of
treating it specially like stack-checks)
* Enable target recording CallFunctionStub to achieve
more inlining on optimized code.
BUG=v8:1789
TEST=jslint and uglify run without crashing, mjsunit/compiler/regress-lazy-deopt.js
Review URL: http://codereview.chromium.org/8492004
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10006 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
So far free variables references in eval code are not statically
resolved. For example in
function foo() { var x = 1; eval("y = x"); }
the variable x will get mode DYNAMIC and y will get mode DYNAMIC_GLOBAL,
i.e. free variable references trigger dynamic lookups with a fast case
handling for global variables.
The CL introduces static resolution of free variables references in eval
code. If possible variable references are resolved to bindings belonging to
outer scopes of the eval call site.
This is achieved by deserializing the outer scope chain using
Scope::DeserializeScopeChain prior to parsing the eval code similar to lazy
parsing of functions. The existing code for variable resolution is used,
however resolution starts at the first outer unresolved scope instead of
always starting at the root of the scope tree.
This is a prerequisite for statically checking validity of assignments in
the extended code as specified by the current ES.next draft which will be
introduced by a subsequent CL. More specifically section 11.13 of revision 4
of the ES.next draft reads:
* It is a Syntax Error if the AssignmentExpression is contained in extended
code and the LeftHandSideExpression is an Identifier that does not
statically resolve to a declarative environment record binding or if the
resolved binding is an immutable binding.
TEST=existing tests in mjsunit
Review URL: http://codereview.chromium.org/8508052
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9999 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Original commit message:
Add a level of indirection to exception handler addresses.
To support deoptimization of exception handlers, the handler address in the
stack is converted to a pair of code object and an index into a separate
table of code offsets. The index part is invariant under deoptimization.
The index is packed into the handler state field so that handler size does
not change.
R=vegorov@chromium.org
BUG=
TEST=
Review URL: http://codereview.chromium.org/8538011
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9977 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
To support deoptimization of exception handlers, the handler address in the
stack is converted to a pair of code object and an index into a separate
table of code offsets. The index part is invariant under deoptimization.
The index is packed into the handler state field so that handler size does
not change.
R=vegorov@chromium.org,fschneider@chromium.org
BUG=
TEST=
Review URL: http://codereview.chromium.org/8462010
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9975 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Our Heap::FreeQueuedChunks generates fake inner chunks in large object
pages queued for freeing, so that StoreBuffer::Filter can recognize them
as pages to be freed. This also relies on MemoryChunk::Contains to work
properly, which is why the size field needs to be initialized as well.
R=vegorov@chromium.org
BUG=v8:1817
TEST=mozilla/js1_5/Regress/regress-360969-05
Review URL: http://codereview.chromium.org/8536009
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9964 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
r9871 (5baeaf57) changed the fill value for deleted entries in hash tables from null_value to the_hole_value.
This commit changes an assertion in MIPS code that expects this value.
BUG=
TEST=
Review URL: http://codereview.chromium.org/8479028
Patch from Gergely Kis <gergely@homejinni.com>.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9955 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Heap::Shrink is called from EnsureFromSpaceIsCommitted at the very start of the GC. At this moment live bytes counts on pages are in inconsistent states. Some pages might have been already swept but have not been yet reached by an incremental marker (or incremental marker is not in progress) and have live bytes count set to 0. Thus we can't rely only on LiveBytes to determine which pages can be released to the OS.
R=mstarzinger@chromium.org
BUG=100414
Review URL: http://codereview.chromium.org/8507038
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9953 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
The return value of pthread_create is now checked to be 0.
Tests on MIPS boards had some silent and hard to find timeouts and errors related to this.
This ensures a proper error message and shutdown if a thread could not be started.
BUG=
TEST=
Review URL: http://codereview.chromium.org/8497041
Patch from Gergely Kis <gergely@homejinni.com>.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9945 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
* Remove a couple of unused fields from the FunctionLiteral, ensure that all
the bools are packed.
* Rename SaveScope and LexicalScope in the parser.
* Use an enum to generate the numbers 0..N and the dependent count, rather
than static const ints. This is simpler to extend (coming in a future
change).
R=danno@chromium.org,keuchel@chromium.org
BUG=
TEST=
Review URL: http://codereview.chromium.org/8505012
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9933 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Replaced FIRST_PHANTOM_PROPERTY_TYPE by a predicate. Removed the (hopefully)
last default cases for switches on PropertyType. Benchmarks show that both
changes are performace-neutral.
Now every value of PropertyType should either be handled by an explicit case in
a switch or by an equality operator. Therefore, the C++ compiler should finally
be able to tell us which places to touch when changing PropertyType.
Review URL: http://codereview.chromium.org/8506004
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9930 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Each such string requires small amount of heap memory.
Heap snapshot generator copies all the strings into internal hash map for future use.
The each copy requires much more memory than original v8 string.
I made a workaround for this. The snapshot will copy only first 1024 symbols.
A simple drive-by fix was made for a performance problem in ToCString implementation.
BUG=v8:1816
TEST=none
Review URL: http://codereview.chromium.org/8509003
Patch from Ilya Tikhonovsky <loislo@chromium.org>.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9928 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Port r9909 (56c8728aa).
Original commit message:
Test at the bottom in the unwind loop. Eliminate the possibility of a
useless move to the eax/rax/r0 register (currently impossible because
this function has two call sites). Do not explicitly zero the context
because we've already saved 0 as the context in the handler.
BUG=
TEST=
Review URL: http://codereview.chromium.org/8507007
Patch from Gergely Kis <gergely@homejinni.com>.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9923 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This also refactors the way we set the length of an arrays' backing
store to use the new elements accessor interface. The actual fix is in
DictionaryElementsAccessor::SetLengthWithoutNormalize() where we first
search for non-deletable elements according to ES5 section 15.4.5.2
specifications.
Snippet from the specification: Attempting to set the length property of
an Array object to a value that is numerically less than or equal to the
largest numeric property name of an existing array indexed non-deletable
property of the array will result in the length being set to a numeric
value that is one greater than that largest numeric property name.
R=danno@chromium.org
TEST=test262/15.4.4.??-7-b-16
Review URL: http://codereview.chromium.org/8372064
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9911 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
In case of a degenerated CFG like in the example below processing
predecessors in the wrong order yields n^2 runtime.
do {
if (x) break;
if (x) break;
if (x) break;
if (x) break;
if (x) break;
if (x) break;
if (x) break;
if (x) break;
if (x) break;
if (x) break;
if (x) break;
// etc.
} while (false);
Reversing iteration order avoids this.
Review URL: http://codereview.chromium.org/8502012
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9905 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
