I don't have a repro, but it seems that when we are close to the stack limit,
CollectEvacuationCandidates' local variables can blow the limit. To avoid
this, instead of always allocating a 2000-pointer-sizes array on the stack,
use a std::vector (which stores its elements on the heap), and while we're
at it, only allocate as many elements as we actually need.
Review URL: https://codereview.chromium.org/1073123002
Cr-Commit-Position: refs/heads/master@{#27742}
Currently we always generate a diamond in the graph builder for every
legacy const context slot, which we cannot get rid of until late control
reduction, even if we know after context specialization that the slot is
already initialized.
Now we generate a select instead, which the CommonOperatorReducer
happily removes during typed lowering. This greatly speeds up asm.js
code generated by Emscripten with the new POINTER_MASKING mode.
R=mstarzinger@chromium.org
Review URL: https://codereview.chromium.org/1072353002
Cr-Commit-Position: refs/heads/master@{#27739}
This makes sure Crankshaft respects interceptors in the global object
even when they shadow a constant global property.
R=verwaest@chromium.org
TEST=cctest/test-api-interceptors/PrePropertyHandler
Review URL: https://codereview.chromium.org/1070803002
Cr-Commit-Position: refs/heads/master@{#27733}
Ideally we would not need the StoreWriteBarrier instructions at all,
but represent the RecordWrite functionality as machine subgraph, but
that'll take some time to get there. In the mean time we can have a
shorter instruction sequence on Intel platforms by recognizing immediate
indices here.
R=svenpanne@chromium.org
Review URL: https://codereview.chromium.org/1075103002
Cr-Commit-Position: refs/heads/master@{#27731}
Port 35a67b745d
Original commit message:
Thanks to some careful assumptions, we can examine the object found at
vector[slot] and trust it's a heap object where the second field is
either a map if it's a WeakCell, or definitely not a map if it's a
Symbol, String or FixedArray. Use this to save a memory read.
R=mbrandy@us.ibm.com
BUG=
Review URL: https://codereview.chromium.org/1059133005
Cr-Commit-Position: refs/heads/master@{#27728}
Port 9af9f1d026
Original commit message:
These operators compute the absolute floating point value of some
arbitrary input, and are implemented without any branches (i.e. using
vabs on arm, and andps/andpd on x86).
R=mbrandy@us.ibm.com
BUG=
Review URL: https://codereview.chromium.org/1072963002
Cr-Commit-Position: refs/heads/master@{#27727}
Port 2d281e71ac
Original commit message:
This enables eager optimization of top-level code with TurboFan and
extends test coverage by triggering it with the --always-opt flag.
Script contexts are now also properly allocated in TurboFan.
R=titzer@chromium.org, mstarzinger@chromium.org,mbrandy@us.ibm.com
BUG=
Review URL: https://codereview.chromium.org/1052123005
Cr-Commit-Position: refs/heads/master@{#27726}
Always use the same branch name (the old version leaked
branches). Always try to delete a possibly existing branch
on start-up and also clean up in the end.
NOTRY=true
TBR=tandrii@chromium.org
Review URL: https://codereview.chromium.org/1077633002
Cr-Commit-Position: refs/heads/master@{#27722}
gcc rejects the following snippet, clang rejects it in -std=c++11 mode:
namespace A { template<class T> class C {}; }
namespace B { template class A::C<int>; }
Indeed, the C++ standard says in 14.7.2p2 "An explicit instantiation shall
appear in an enclosing namespace of its template", so cl.exe is incorrect to
allow this.
Just move the instantiation out of the v8 namespace to fix. No intended
behavior change. Fixes building with clang-cl on Windows.
BUG=chromium:475643
LOG=N
TBR=svenpanne@chromium.org
Review URL: https://codereview.chromium.org/1073903002
Cr-Commit-Position: refs/heads/master@{#27721}
When ObjectToString is installed on Object.prototype twice (once in v8natives.js, and once in harmony-tostring.js), this pollutes old code spaces on some devices. To prevent this, the function is only installed once, preventing test failures when the --harmony-tostring flag is flipped on by default.
BUG=v8:3502
LOG=N
R=arv@chromium.org
Review URL: https://codereview.chromium.org/1072083002
Cr-Commit-Position: refs/heads/master@{#27720}
Port 146598f44a
Original commit message:
Optimistically pushing a lot of arguments can run into the stack limit of the process, at least on operating systems where this limit is close to the limit that V8 sets for itself.
R=mbrandy@us.ibm.com
BUG=
Review URL: https://codereview.chromium.org/1073893002
Cr-Commit-Position: refs/heads/master@{#27719}
This is a workaround to make the debugger happy about TurboFan frames
when the debugger causes frame inspection. Note that this can happen
because the debugger can be activated while there still are optimized
TurboFan activations on the stack.
R=ishell@chromium.org
BUG=chromium:465298
TEST=mjsunit/regress/regress-crbug-465298
LOG=N
Review URL: https://codereview.chromium.org/1074793003
Cr-Commit-Position: refs/heads/master@{#27717}
These comparisons are only meant to be done by AstValueFactory itself (in
its string_table_ operations), so make the Compare() function a private
member of AstValueFactory.
All other clients of AstRawStrings should compare them by pointer value.
There were only two clients which failed to abide by this rule, one
recently-added (in ModuleDescriptor) and the other in Literal::Match
(in ast.cc, added in https://code.google.com/p/v8/source/detail?r=24396).
Review URL: https://codereview.chromium.org/1069423003
Cr-Commit-Position: refs/heads/master@{#27715}
This fixes the d8 target which previously was getting both USING_V8_SHARED
and BUILDING_V8_SHARED defined at the same time.
Renames direct_dependent_configs to public_configs (new name, same thing).
Review URL: https://codereview.chromium.org/1065403002
Cr-Commit-Position: refs/heads/master@{#27713}
Reason for revert:
Causes test failures on ARM bots related to cells and write barriers.
Original issue's description:
> Merge cellspace into old pointer space
>
> BUG=
>
> Committed: https://crrev.com/4e7163ce05f135918205c7855ae60a48e5d46cc5
> Cr-Commit-Position: refs/heads/master@{#27707}
TBR=hpayer@chromium.org,balazs.kilvady@imgtec.com,yangguo@chromium.org,verwaest@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review URL: https://codereview.chromium.org/1053243003
Cr-Commit-Position: refs/heads/master@{#27711}
Remove references to android_webview_build now that we no longer support
that build configuration. This also removes use_system_stlport which was
only supported with android_webview_build.
BUG=chromium:440793
LOG=n
Review URL: https://codereview.chromium.org/1069663002
Cr-Commit-Position: refs/heads/master@{#27708}
Compiling v8 for X87 platform with clang compiler leads to some test case failure.
So we set the default compiler for X87 to GCC.
BUG=
Review URL: https://codereview.chromium.org/1031793002
Cr-Commit-Position: refs/heads/master@{#27706}
Escaping used to happen lazily, implemented in an accessor property.
However, native implementation of RegExp methods use .source as well.
This leads to performance regressions. Now we do it eagerly instead.
R=jkummerow@chromium.org
BUG=chromium:436447
LOG=N
Review URL: https://codereview.chromium.org/1070093002
Cr-Commit-Position: refs/heads/master@{#27705}
Port 725cdc533c
Original commit message:
This reduces the overhead of recursive calls when context specialization
is enabled. Based on this it might be possible to further reduce the
overhead by also specializing the call itself.
As a drive-by-fix, port the fast context materialization optimization to
arm and arm64, that was previously only supported on x64 and ia32.
BUG=
Review URL: https://codereview.chromium.org/1074743002
Cr-Commit-Position: refs/heads/master@{#27700}
This makes some tests in test-heap.cc more robust against flags, now
that top-level code can be optimized by the --always-opt flag.
R=hpayer@chromium.org
TEST=cctest/test-heap/TestInternalWeakLists
Review URL: https://codereview.chromium.org/1071063004
Cr-Commit-Position: refs/heads/master@{#27699}
This is preparatory work to have MachineTypes encoded in AllocatedOperands.
Review URL: https://codereview.chromium.org/1075863002
Cr-Commit-Position: refs/heads/master@{#27698}
This is a very simple dead store elimination that removes StoreField
nodes which are immediately followed by other StoreField nodes that
store to the same field. Ideally there should be a fully featured store
elimination, which walks over the effect graph starting from the end,
but there are some technical difficulties to solve before we can get to
that, esp. we need to think about "effect producing" operators like
ValueEffect first. Once we have that, it is trivial to remove this temporary
poor man's store elimination.
R=dcarney@chromium.org
Review URL: https://codereview.chromium.org/1070003002
Cr-Commit-Position: refs/heads/master@{#27697}
port 2d281e71ac (r27633)
original commit message:
Make --always-opt also optimize top-level code.
This enables eager optimization of top-level code with TurboFan and
extends test coverage by triggering it with the --always-opt flag.
Script contexts are now also properly allocated in TurboFan.
BUG=
Review URL: https://codereview.chromium.org/1077523002
Cr-Commit-Position: refs/heads/master@{#27693}
