Reason for revert:
This patch actually seemed to cause a further GameBoy regression! Reverting it seems to address the regression.
Original issue's description:
> Restore per-TypedArray-class length accessors as a perf workaround
>
> This patch is a workaround to the performance regression caused by
> implementing the ES2015 TypedArray prototype chain: Include a
> per-TypedArray-subclass length getter so that the superclass getter does
> not become polymorphic. The patch appears to fix a regression in the
> Gameboy Octane benchmark.
>
> BUG=chromium:579905
> R=adamk
> LOG=Y
>
> Committed: https://crrev.com/03ce7711e474a0ef74f723b30ae1527c89dec010
> Cr-Commit-Position: refs/heads/master@{#33501}
R=adamk@chromium.org
BUG=chromium:579905,chromium:593634
LOG=Y
Review URL: https://codereview.chromium.org/1812143004
Cr-Commit-Position: refs/heads/master@{#34906}
API change: This adds a new flag skip_html_comments to v8::ScriptOriginOptions. This flag controls whether V8 will attempt to honour HTML-style comments in JS sources.
(That is: Gracefully ignore <!-- ... ---> in JS sources, which was a popular technique in the early days of JavaScript, to prevent non-JS-enabled browsers from displaying script sources to uses.)
The flag defaults to 'true' when using v8::ScriptOrigin constructor, which preserves the existing behaviour. Embedders which are happy with the existing behaviour will thus not need any changes.
BUG=chromium:573887
LOG=Y
Review URL: https://codereview.chromium.org/1801203002
Cr-Commit-Position: refs/heads/master@{#34904}
There are multiple ways we may construct an instruction sequence.
Moved the split edge form validation where they must all eventually
chokepoint to - the register allocation pipeline, which is also the main
and first consumer of this property.
BUG=
Review URL: https://codereview.chromium.org/1810293003
Cr-Commit-Position: refs/heads/master@{#34903}
This moves the call-sites that ensure we have a feedback vector present
before kicking off a compiler into the actual compilation pipeline. The
backends no longer need to worry about the feedback vector.
R=mvstanton@chromium.org
Review URL: https://codereview.chromium.org/1811973006
Cr-Commit-Position: refs/heads/master@{#34900}
Since we clear invalid slots eagerly, there is no need to check if a slot is a valid slot.
BUG=chromium:578883
LOG=NO
Review URL: https://codereview.chromium.org/1817523002
Cr-Commit-Position: refs/heads/master@{#34898}
This is an experiment to see if it affects crash rates on Nexus 9 devices.
A small performance impact (3-4% on Octane) is expected.
BUG=chromium:524337
LOG=n
R=hablich@chromium.org
Review URL: https://codereview.chromium.org/1808243002
Cr-Commit-Position: refs/heads/master@{#34896}
This is a pure refactoring CL and should not contain any functional
changes to the code. The following has been done:
- Group compiler.cc into sections for each component.
- Surround local helper methods by anonymous namespace.
- Move implementation of Compiler (API class) together.
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/1809403002
Cr-Commit-Position: refs/heads/master@{#34891}
The higher word (bits 32 - 63) of FPU register is set on zero before
storing result.
TEST=cctest/test-macro-assembler-mips64/Cvt_s_uw_Trunc_uw_s
BUG=
Review URL: https://codereview.chromium.org/1812193003
Cr-Commit-Position: refs/heads/master@{#34889}
Using std::sort on a FixedArray with HeapNumbers is not very GC friendly.
Adding a separate step to properly record possible HeapNumbers.
BUG=chromium:594954
LOG=N
Review URL: https://codereview.chromium.org/1807173002
Cr-Commit-Position: refs/heads/master@{#34888}
The trigger point in question is by now obsolete. The optimized compile
job will itself ensure that deoptimization support is present on the
incoming SharedFunctionInfo, this will make sure to produce baseline
code when necessary. The ScopeInfo is also installed at that point in
time.
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/1816513002
Cr-Commit-Position: refs/heads/master@{#34887}
This changes the compilation pipeline so that SharedFunctionInfo objects
are always allocated before the various compilers are invoked. It is a
preparation towards having that object available during compile time and
hence reducing the dependency on FunctionLiteral and the need to copy a
lot of the information into the CompilationInfo.
Optimizing compilers already assume the SharedFunctionInfo is present
and the baseline compilers have other heap accesses sprinkled throughout
the compilation process. Duplicating statically available information
from the SharedFunctionInfo within the CompilationInfo has no benefit.
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/1813803002
Cr-Commit-Position: refs/heads/master@{#34885}
port 2aa070be4f (r34863)
original commit message:
Repair this to match what the runtime correctly does, by first checking if the
function is a constructor before we access the prototype.
BUG=
Review URL: https://codereview.chromium.org/1809333002
Cr-Commit-Position: refs/heads/master@{#34880}
port 992ae64de0 (r34866)
original commit message:
This new intrinsic is used by the desugared ES6 instanceof implementation for
the cases when the F[@@hasInstance] property is null or undefined.
BUG=
Review URL: https://codereview.chromium.org/1815623002
Cr-Commit-Position: refs/heads/master@{#34879}
It was never being set to false in production (though it was in test-parsing.cc,
due to that test having its own flag-setting logic).
Review URL: https://codereview.chromium.org/1815033002
Cr-Commit-Position: refs/heads/master@{#34878}
This patch fixes two bugs in Array.prototype.concat in conjunction with
subclassing Arrays:
- Create a new property rather than calling Set when adding elements to
the output array. This means setters are not called.
- If there is an exception thrown from DefineProperty, propagate it
outwards properly, rather than swallowing it. This can occur, e.g., with
a Proxy as the new output array.
R=adamk
LOG=Y
BUG=chromium:595319
Review URL: https://codereview.chromium.org/1814933002
Cr-Commit-Position: refs/heads/master@{#34876}
Previously, when a property was non-configurable or the object was
non-extensible, CreateDataProperty might just return false rather than
throwing, even if should_throw was on. This patch fixes that issue.
Tested by running the patch at https://codereview.chromium.org/1814933002
on top of this code and observing the tests to see an exception thrown.
R=adamk
BUG=chromium:595319
LOG=Y
Review URL: https://codereview.chromium.org/1809233002
Cr-Commit-Position: refs/heads/master@{#34875}
Functions with builtin ids can be compiled with Ignition, so it is no longer
an option to overlap the bytecode_array field with the builtin id on
the SharedFunctionInfo object. Instead overlap it with the
inferred_name, which is only used for debug and so shouldn't be required
for functions with builtin ids. This result in the inferred_name field
being renamed to function_identifier, and adding typed accessors for
inferred_name and builtin_function_id.
This is required to build the snapshot with --no-lazy.
BUG=v8:4280
LOG=N
Review URL: https://codereview.chromium.org/1801023002
Cr-Commit-Position: refs/heads/master@{#34867}
This new intrinsic is used by the desugared ES6 instanceof implementation for
the cases when the F[@@hasInstance] property is null or undefined.
R=mstarzinger@chromium.org
Review URL: https://codereview.chromium.org/1809993002
Cr-Commit-Position: refs/heads/master@{#34866}
Port 33c08596e1
Original commit message:
Int64Sub is lowered to a new turbofan operator, Int32SubPair. The new
operator takes 4 inputs an generates 2 outputs. The inputs are the low
word of the left input, high word of the left input, the low word of the
right input, and high word of the right input. The ouputs are the low
and high word of the result of the subtraction.
The implementation is very similar to the implementation of Int64Add.
R=ahaas@chromium.org, joransiu@ca.ibm.com, michael_dawson@ca.ibm.com, mbrandy@us.ibm.com
BUG=
Review URL: https://codereview.chromium.org/1807243002
Cr-Commit-Position: refs/heads/master@{#34865}
Repair this to match what the runtime correctly does, by first checking if the
function is a constructor before we access the prototype.
R=verwaest@chromium.org
BUG=
Review URL: https://codereview.chromium.org/1810953002
Cr-Commit-Position: refs/heads/master@{#34863}
Currently, if the size of two cmp or test operands is a byte or a word, we sign-extend or zero-extend each of them into a 32-bit register before doing the comparison, even when the conditions for the use of a memory operand are met.
This CL makes it possible to load only one of them into a register and address the other as a memory operand.
Meanwhile, comparisons between Uint8 values in the string relational comparison stubs are done with Uint32LessThan (previously we were always zero-extending the byte to a 32-bit value, so signed comparison was alright).
Found that Assembler::arithmetic_op_8(byte, Register, const Operand&) wasn't taking the Operand's rex_ field into account, so I fixed that too.
BUG=
Review URL: https://codereview.chromium.org/1780193003
Cr-Commit-Position: refs/heads/master@{#34862}
