This fixes a corner-case where extending an optimized code map left
stale entries in the abandoned copy. This can cause havoc not only in
the heap verifier but also in the GC, because stale entries have not
been recorded when being trated weakly.
Note that this also pre-tenures all optimized code maps into old-space
because their lifetime is coupled to the SharedFunctionInfo anyways.
R=hpayer@chromium.org
TEST=cctest/test-heap/Regress514122
BUG=chromium:514122
LOG=N
Review URL: https://codereview.chromium.org/1277873002
Cr-Commit-Position: refs/heads/master@{#30047}
Use the same path separator for test cases on the cmd-line
on all platforms as in testcase.GetLabel, which is used to
report failures.
BUG=chromium:511215
LOG=n
NOTRY=true
Review URL: https://codereview.chromium.org/1281453003
Cr-Commit-Position: refs/heads/master@{#30045}
We missed a check whether we can actually do incremental marking when starting
it on reaching the external allocation limit.
BUG=chromium:517195
LOG=N
Review URL: https://codereview.chromium.org/1274983002
Cr-Commit-Position: refs/heads/master@{#30043}
I explored gdb macros, but hit an issue quite early with instantiating and initializing an OFStream - "virtual baseclass botch".
Currently, I have a side-file that I include (and then remove before publishing CLs), which defines wrappers to the above operator<< APIs, but this is becoming quite awkward, and I believe the functionality to be quite useful to anyone working in this (regalloc) area, so it's worth having something better than local side-files. The gdb path seems overly-twisted for the problem at hand, and I've noticed elsewhere (e.g. Object) the presence of Print APIs - hence this change.
BUG=
Review URL: https://codereview.chromium.org/1280483002
Cr-Commit-Position: refs/heads/master@{#30039}
The code previously used a relative timeout and accumulated wait times to see
if the timeout was exceeded. Now we convert the timeout into an absolute time,
and always compare the current time against that.
BUG=v8:4357
R=machenbach@chromium.org
LOG=n
Review URL: https://codereview.chromium.org/1272733002
Cr-Commit-Position: refs/heads/master@{#30036}
These flags weren't doing any real work, since the decision of whether some
source code is a script or module is made outside the parser (currently,
by the V8 API).
The only behavior change in this patch is to always parse 'import' and
'export' as their Token values, which changes the error message from
"Unexpected reserved word" to "Unexpected token import" (which doesn't
seem particularly harmful).
Review URL: https://codereview.chromium.org/1262913003
Cr-Commit-Position: refs/heads/master@{#30034}
This is only an estimate since it counts objects that could be shared,
for example strings, cow arrays, heap numbers, etc.
It however ignores objects that could be shared, but may only be used
by the context to be measured, for example shared function infos,
script objects, scope infos, etc.
R=jochen@chromium.org
Review URL: https://codereview.chromium.org/1268333004
Cr-Commit-Position: refs/heads/master@{#30029}
With the recent changes to the incremental marking API we can now kick off
incremental marking while respecting callback flags.
Performance neutral for smoothness.image_decoding_cases on N9 (read: does not
crash) as long as we synchronously process phantom callbacks
(kGCCallbackFlagForced).
OORT single run:
"marksweep": {
"count": 5,
"pause_min": 7.5,
"pause_max": 158.8,
"pause_avg": 97.52000000000001,
"pause_gt_10ms": 4
}
--- vs ---
"marksweep": {
"count": 5,
"pause_min": 16.2,
"pause_max": 22.1,
"pause_avg": 19.32,
"pause_gt_10ms": 5
}
The number of actual full GCs varies. The improvement manifests in reduced
maximum and average pauses.
BUG=chromium:515795
LOG=N
Review URL: https://codereview.chromium.org/1271253002
Cr-Commit-Position: refs/heads/master@{#30028}
Previously, examples like (({a = x}, x) => {})({}, 0) did not throw a ReferenceError like they should. This CL
- Splits up DeclareFormalParameters such that the formals can be recorded first and declared later.
- Declaration then takes the complete parameter list into account. If it is not simple, temporaries are introduced for all parameters.
- BuildParameterInitializationBlock desugars all parameters from non-simple lists into let-bindings.
- Refactored Pre/ParserFormalParameters, so that the arity information is no longer duplicated in Parser.
- Rest is currently handled specially, until rest-via-destructuring has landed.
R=adamk@chromium.org, littledan@chromium.org
BUG=v8:811
LOG=N
Review URL: https://codereview.chromium.org/1259283002
Cr-Commit-Position: refs/heads/master@{#30025}
The heuristic can cause weird behavior when bootstrapping.
The memory savings is not worth this hassle.
Review URL: https://codereview.chromium.org/1265983006
Cr-Commit-Position: refs/heads/master@{#30019}
To avoid tanking context startup performance, only the actual installation of the
JS-exposed API is flag-guarded. The remainder of the implementation still
resides in the snapshot.
Review URL: https://codereview.chromium.org/1257063003
Cr-Commit-Position: refs/heads/master@{#30017}
This introduces a CopyFixedArrayAndGrow method on Factory that takes
the "grow amount" instead of the "new size" as an argument. The new
interface is safer because it allows for mutations by the GC that
potentially trim the source array.
This also fixes a bug in SharedFunctionInfo::AddToOptimizedCodeMap
where the aformentioned scenario led to unused entries within the
optimized code map.
Note that FixedArray::CopySize is hereby deprecated because it is
considered unsafe and should no longer be used.
R=hpayer@chromium.org
TEST=mjsunit/regress/regress-crbug-513507
BUG=chromium:513507
LOG=n
Review URL: https://codereview.chromium.org/1255173006
Cr-Commit-Position: refs/heads/master@{#30012}
Added a separate flag for this, since we intend to enable it for the linear allocator as well. Currently, the option is "on" for greedy, as a point in time to enable its testing (through the greedy allocator bots).
BUG=
Review URL: https://codereview.chromium.org/1256313003
Cr-Commit-Position: refs/heads/master@{#30005}
The race occurred when Workers were used. Since Workers call
Shell::ExecuteString from a different thread, TSAN (correctly) flags
this as a racy write. Solution would be to either synchronize the writes,
or to 'lift' the write higher up in the call stack and only write the flag
from the main thread. This implements this latter solution.
These methods call Shell::ExecuteString, but do *not* set script_executed:
- ExecuteInThread: Can only occur is JS has already been executed.
- Shell::Load: Callback for JS; so JS has already been executed when
we get there.
- Shell::RunShell: Interactive shell. We no longer need script_executed once
we're here.
BUG=v8:4330
LOG=N
Review URL: https://codereview.chromium.org/1258303004
Cr-Commit-Position: refs/heads/master@{#30003}
