Port dbd8640813
Original commit message:
Note: This is currently only used by yield*, we still need to support it in
other places (such as for-of loops). It can be used manually of course.
(This CL does not touch the full-codegen implementation of yield* because that
code is already dead. The yield* desugaring already supports return and doesn't
need to be touched.)
R=neis@chromium.org, joransiu@ca.ibm.com, jyan@ca.ibm.com, michael_dawson@ca.ibm.com
BUG=v8:3566
LOG=y
Review URL: https://codereview.chromium.org/1664413002
Cr-Commit-Position: refs/heads/master@{#33754}
Port bb31db3ad6
Original commit message:
(RELAND: the problem before was a missing write barrier for adding the code
entry to the new closure. It's been addressed with a new macro instruction
and test. The only change to this CL is the addition of two calls to
__ RecordWriteCodeEntryField() in the platform CompileLazy builtin.)
We get less "pollution" of type feedback if we have one vector per native
context, rather than one for the whole system. This CL moves the vector
appropriately.
We rely more heavily on the Optimized Code Map in the SharedFunctionInfo. The
vector actually lives in the first slot of the literals array (indeed there is
great commonality between those arrays, they can be thought of as the same
thing). So we make greater effort to ensure there is a valid literals array
after compilation.
This meant, for performance reasons, that we needed to extend
FastNewClosureStub to support creating closures with literals. And ultimately,
it drove us to move the optimized code map lookup out of FastNewClosureStub
and into the compile lazy builtin.
The heap change is trivial so I TBR Hannes for it...
Also, Yang has had a look at the debugger changes already and approved 'em. So he is TBR style too.
And Benedikt reviewed it as well.
R=mvstanton@chromium.org, joransiu@ca.ibm.com, jyan@ca.ibm.com, michael_dawson@ca.ibm.com
BUG=
Review URL: https://codereview.chromium.org/1671553002
Cr-Commit-Position: refs/heads/master@{#33753}
There might be several ExternalCallbackScope's created
during the native callback. Remove the assert that is not
aligned with that.
Moreover this iterator must work for any kind of
stacks including corrupted ones.
BUG=v8:4705
LOG=N
Review URL: https://codereview.chromium.org/1663193003
Cr-Commit-Position: refs/heads/master@{#33751}
This bit was ostensibly being used to provide appropriate syntax
errors for invalid destructuring assignment patterns, but adding a
single call to RecordPatternError() (in place of
BindingPatternUnexpectedToken()) seems to have replaced the need for it.
Review URL: https://codereview.chromium.org/1665043002
Cr-Commit-Position: refs/heads/master@{#33750}
Also various related cleanup in ParseVariableDeclarations(). The only
changes in logic are explained below:
- We were redundantly checking for parenthesized binding patterns;
these are already ruled out by BindingPatternUnexpectedToken()
calls in the places where we hit an LPAREN.
- There's no need to default-initialize a LET-mode variable in a
for-each loop, just as there isn't for CONST or CONST_LEGACY
(ParseForStatement will take care of properly initializing all
of the above).
Review URL: https://codereview.chromium.org/1661193002
Cr-Commit-Position: refs/heads/master@{#33749}
Note: This is currently only used by yield*, we still need to support it in
other places (such as for-of loops). It can be used manually of course.
(This CL does not touch the full-codegen implementation of yield* because that
code is already dead. The yield* desugaring already supports return and doesn't
need to be touched.)
BUG=v8:3566
LOG=y
Review URL: https://codereview.chromium.org/1639343005
Cr-Commit-Position: refs/heads/master@{#33744}
The test currently only shows the JavaScript stack frames, I'll then add
support for interleaved WebAssembly stack frames and update the test.
R=titzer@chromium.org
Review URL: https://codereview.chromium.org/1661383002
Cr-Commit-Position: refs/heads/master@{#33742}
(RELAND: the problem before was a missing write barrier for adding the code
entry to the new closure. It's been addressed with a new macro instruction
and test. The only change to this CL is the addition of two calls to
__ RecordWriteCodeEntryField() in the platform CompileLazy builtin.)
We get less "pollution" of type feedback if we have one vector per native
context, rather than one for the whole system. This CL moves the vector
appropriately.
We rely more heavily on the Optimized Code Map in the SharedFunctionInfo. The
vector actually lives in the first slot of the literals array (indeed there is
great commonality between those arrays, they can be thought of as the same
thing). So we make greater effort to ensure there is a valid literals array
after compilation.
This meant, for performance reasons, that we needed to extend
FastNewClosureStub to support creating closures with literals. And ultimately,
it drove us to move the optimized code map lookup out of FastNewClosureStub
and into the compile lazy builtin.
The heap change is trivial so I TBR Hannes for it...
Also, Yang has had a look at the debugger changes already and approved 'em. So he is TBR style too.
And Benedikt reviewed it as well.
TBR=hpayer@chromium.org, yangguo@chromium.org, bmeurer@chromium.org
BUG=
Review URL: https://codereview.chromium.org/1668103002
Cr-Commit-Position: refs/heads/master@{#33741}
Unstructured control flow caused by excpetion handling leads to a wrong x87 stack
state. This patch is to reset the x87 state at the hanlder entry point.
Thanks for help from weiliang.lin@intel.com.
BUG=
Review URL: https://codereview.chromium.org/1668463006
Cr-Commit-Position: refs/heads/master@{#33738}
Synchronize calls in the heap iterator have been put there for the
serializer, which never actually made use of them. This CL fixes that.
R=vogelheim@chromium.org
Review URL: https://codereview.chromium.org/1667063002
Cr-Commit-Position: refs/heads/master@{#33736}
This CL deals with yield* by desugaring it in the parser. Hence the
full-codegen implementation of it becomes obsolete and can be removed in a
future CL.
The only change in semantics should be that the results of the iterator's next
and throw methods are checked to be objects, which didn't happen before but is
required by the spec.
BUG=
Review URL: https://codereview.chromium.org/1643903003
Cr-Commit-Position: refs/heads/master@{#33735}
Code compiled during snapshot are overwhelmingly for functions
that are only used for bootstrapping. It makes no sense to
include them in the startup snapshot, which bloats up the snapshot size
and slows down deserialization.
Snapshot sizes for comparison, for ia32:
w/o --ignition: 484k
w/ --ignition: 537k
bytecode removed: 489k
R=rmcilroy@chromium.org,mstarzinger@chromium.org
Review URL: https://codereview.chromium.org/1667693002
Cr-Commit-Position: refs/heads/master@{#33734}
This implements proper context switching while unwinding the stack due
to an exception being handled in interpreted code. The context under
which the handler is scoped is being preserved in a dedicated register
while the try-block is running. Both, the stack unwinding machinery as
well as the graph builder, restore the context from that register.
R=rmcilroy@chromium.org,bmeurer@chromium.org
BUG=v8:4674
LOG=n
Review URL: https://codereview.chromium.org/1665833002
Cr-Commit-Position: refs/heads/master@{#33733}
Removes skips for two tests in cctest that are no longer crashing with ignition.
BUG=v8:4680
LOG=N
Review URL: https://codereview.chromium.org/1668843003
Cr-Commit-Position: refs/heads/master@{#33732}
Moves the stack check from the function entry trampoline to instead be
after function activation using an explicit StackCheck bytecode. Also
add stack checks on back edges of loops.
BUG=v8:4280,v8:4678
LOG=N
Review URL: https://codereview.chromium.org/1665853002
Cr-Commit-Position: refs/heads/master@{#33730}
This is also a refactoring of the merge function in
escape analysis.
BUG=v8:4586
LOG=n
Review URL: https://codereview.chromium.org/1654163003
Cr-Commit-Position: refs/heads/master@{#33728}
This cleans up and simplifyies handling the bytes followin an opcode
with little helper structs that will be useful in the interpreter and
already have been in keeping OpcodeArity and OpcodeLength up to date
with the decoder.
R=bradnelson@chromium.org, ahaas@chromium.org
BUG=
Review URL: https://codereview.chromium.org/1664883002
Cr-Commit-Position: refs/heads/master@{#33723}
The goal of the Int64Reducer is to replace all int64 nodes in a tf graph
with a set of int32 nodes such that 64 bit tf functions can be executed
on 32 bit platforms. At the moment the Int64Reducer only replaces
Int64Constants, TruncateInt64ToInt32, and Word64And.
R=titzer@chromium.org
Review URL: https://codereview.chromium.org/1655883002
Cr-Commit-Position: refs/heads/master@{#33721}
The spec requires all Math functions to first call ToNumber on all
arguments before doing any other observable operation. So early
return in case of Infinity is not valid.
Drive-by-fix: Remove the use of %_Arguments / %_ArgumentsLength and
use (strict) arguments instead of allocating a temporary InternalArray
explicitly.
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/1669773002
Cr-Commit-Position: refs/heads/master@{#33717}
No need to micro-optimize here, and in almost all cases here, using
arguments should result in roughly the same code w/ Crankshaft anyway.
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/1664513007
Cr-Commit-Position: refs/heads/master@{#33716}
So far, we've been moving down gaps wholesale. This change moves
individual move operations instead. This improves some benchmarks,
and should overall reduce code size, because it improves the chance of
reducing the number of moves.
For example, there are improvements on x64 in Emscripten (Bullet, in
particular) , JetStream geomean, Embenchen (zlib).
In the process of making this change, I noticed we can separate the
tasks performed by the move optimizer, as follows:
- group gaps into 1
- push gaps down, jumping instructions (these 2 were together before)
- merge blocks (and then push gaps down)
- finalize
We can do without a finalization list. This avoids duplicating storage -
we already have the list of instructions; it also simplifies the logic, since,
with this change, we may process an instruction's gap twice.
Compile time doesn't regress much (see pathological cases), but we
may want to avoid the allocations of the few sets used in the new code.
I'll do that in a subsequent change.
BUG=
Review URL: https://codereview.chromium.org/1634093002
Cr-Commit-Position: refs/heads/master@{#33715}
Rolling v8/buildtools to e27b1f1459452013ce59c9d5dbc93c88982cb76e
Rolling v8/tools/clang to 6449b18afaa80290fd8930c3a42c80908505f41f
TBR=machenbach@chromium.org,vogelheim@chromium.org,hablich@chromium.org
Review URL: https://codereview.chromium.org/1660143007
Cr-Commit-Position: refs/heads/master@{#33714}
This removes --harmony-completion, --harmony-concat-spreadable, and
--harmony-tolength and moves the appropriate tests from harmony/ to es6/.
Review URL: https://codereview.chromium.org/1667453002
Cr-Commit-Position: refs/heads/master@{#33712}
Skips cctest/test-heap/NoWeakHashTableLeakWithIncrementalMarking
that is crashing on arm64.
BUG=v8:4280
LOG=N
NOTRY=true
Review URL: https://codereview.chromium.org/1660613005
Cr-Commit-Position: refs/heads/master@{#33708}
Unifies the meaning of kRegCount8 and kRegCount16 across bytecodes.
Call and CallJSRuntime had a slightly different use of the register
count operand. From this change forth, register count operands are
always based off of the previous register operand.
BUG=v8:4280,v8:4675
LOG=N
Review URL: https://codereview.chromium.org/1659023002
Cr-Commit-Position: refs/heads/master@{#33707}
