This flag is intended as a staging flag for TurboFan. It serves as a
single flag that always enables a most recent configuration of TurboFan
for test suites and benchmarks, without needing to update test drivers.
R=titzer@chromium.org,machenbach@chromium.org
Review URL: https://codereview.chromium.org/1094573002
Cr-Commit-Position: refs/heads/master@{#27896}
Port 13459c1ae3
Original commit message:
Calling new Array(JSObject::kInitialMaxFastElementArray) in optimized code
makes a stub call that bails out due to the length. Currently, the bailout
code a) doesn't have the allocation site, and b) wouldn't use it if it did
because the length is perceived to be too high.
This CL passes the allocation site to the stub call (rather than undefined),
and alters the bailout code to utilize the feedback.
R=mvstanton@chromium.org, michael_dawson@ca.ibm.com, dstence@us.ibm.com
BUG=
Review URL: https://codereview.chromium.org/1089913002
Cr-Commit-Position: refs/heads/master@{#27895}
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.
BUG=
Review URL: https://codereview.chromium.org/1053843003
Cr-Commit-Position: refs/heads/master@{#27757}
Review URL: https://codereview.chromium.org/1083413003
Cr-Commit-Position: refs/heads/master@{#27893}
The previous restrictions were overshooting (didn't allow a class to refer to a
later class under any circumstances); after this CL we're undershooting (allow
referring to any class from inside a method).
Implementing the correct checks (allow referring only if the class declarations
are in a consecutive block and if there's no dependency cycle) will be
implemented as a follow up.
BUG=v8:3956
LOG=N
Review URL: https://codereview.chromium.org/1087543004
Cr-Commit-Position: refs/heads/master@{#27888}
Implements the strong mode proposal's restrictions on the syntax of the
switch statement. Also fixes a minor bug with empty statements in strong
mode and improves StrongUndefinedArrow parser synch tests.
BUG=v8:3956
LOG=N
Review URL: https://codereview.chromium.org/1084983002
Cr-Commit-Position: refs/heads/master@{#27885}
Will be used for calculating changes between versions
BUG=
NOTRY=true
Review URL: https://codereview.chromium.org/1095483002
Cr-Commit-Position: refs/heads/master@{#27882}
DoParseProgram doesn't appear to need to receive toplevel scopes as
arguments; it can properly set the end_position of the scopes to the
scanner's position after parsing is complete.
R=marja@chromium.org
BUG=
LOG=N
Review URL: https://codereview.chromium.org/1091743002
Cr-Commit-Position: refs/heads/master@{#27880}
port 0179ec5797 (r27846).
original commit message:
The cells are stored on prototypes (in their map's PrototypeInfo). When a
prototype object changes its map, then both its own validity cell and those
of all "downsstream" prototypes are invalidated; handlers for a given receiver
embed the currently valid cell for that receiver's prototype during their
compilation and check it on execution.
BUG=
Review URL: https://codereview.chromium.org/1090803002
Cr-Commit-Position: refs/heads/master@{#27877}
port 13459c1ae3 (r27857)
original commit message:
Array() in optimized code can create with wrong ElementsKind in corner cases.
Calling new Array(JSObject::kInitialMaxFastElementArray) in optimized code
makes a stub call that bails out due to the length. Currently, the bailout
code a) doesn't have the allocation site, and b) wouldn't use it if it did
because the length is perceived to be too high.
This CL passes the allocation site to the stub call (rather than undefined),
and alters the bailout code to utilize the feedback.
BUG=
Review URL: https://codereview.chromium.org/1088423002
Cr-Commit-Position: refs/heads/master@{#27875}
This fixes a race where
- mutator changes the fixed array length by trimming it,
- sweeper thread reads the length of the fixed array.
Also rename FROM_GC and FROM_MUTATOR to be more precise.
BUG=chromium:462908
LOG=NO
Review URL: https://codereview.chromium.org/1034163002
Cr-Commit-Position: refs/heads/master@{#27869}
This breaks the (evacuation -> slots buffer overflow -> abort -> new GC -> evacuation) cycle for popular pages.
BUG=
Review URL: https://codereview.chromium.org/1037433002
Cr-Commit-Position: refs/heads/master@{#27868}
Reason for revert:
Seems to cause performance regressions.
Original issue's description:
> [x64] Use xorl to materialize smi zero.
>
> Before we always loaded smi zero via a movabs with a 64-bit immediate,
> which is pretty expensive compared to the xorl.
>
> R=jarin@chromium.org
>
> Committed: https://crrev.com/f236777bfe6e080ff1ead6baf847cc9b6bb4f9cb
> Cr-Commit-Position: refs/heads/master@{#27829}
TBR=jarin@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:477592
LOG=n
Review URL: https://codereview.chromium.org/1059543004
Cr-Commit-Position: refs/heads/master@{#27867}
Split interface and implementation of ControlEquivalence and add a
dedicated trace flag --trace-turbo-ceq to make it reusable outside the
scheduler.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1056093005
Cr-Commit-Position: refs/heads/master@{#27862}
