It is not always safe to allocate new heap objects in the JSGraph. We
might have to revisit this later once we do the canonicalization for
HeapConstants.
R=jarin@chromium.org
BUG=chromium:545364
LOG=n
Review URL: https://codereview.chromium.org/1413373002
Cr-Commit-Position: refs/heads/master@{#31421}
This patch refactors array methods to have the
InnerArray{Map,Filter} methods convert to a GlobalArray
rather than the callers.
BUG=chromium:544991
R=yangguo,adamk
CC=mstarzinger,jochen
LOG=Y
Review URL: https://codereview.chromium.org/1408213004
Cr-Commit-Position: refs/heads/master@{#31403}
This introduces an explicit lazy bailout. It is wrapped in the call
node, mostly because the lazy deoptimization processing is married
to the call processing in the instruction selector and the code generator.
It is still a terrible hack.
R=bmeurer@chromium.org,mstarzinger@chromium.org
BUG=chromium:543994,v8:4195
LOG=n
Review URL: https://codereview.chromium.org/1412443003
Cr-Commit-Position: refs/heads/master@{#31353}
This is in preparation to enabling --turbo-inlining by default, fixing
various issues when general purpose inlining is running against our
entire test suite.
R=bmeurer@chromium.org
BUG=v8:4493
LOG=n
Review URL: https://codereview.chromium.org/1407533004
Cr-Commit-Position: refs/heads/master@{#31294}
For an edge case like the following in sloppy mode
with ({}) function f() {}
a lexical scope needs to be created around the body in order
to hold the function declaration, so it can be hoisted according
to a loose interpretation of Annex B 3.3 (loose because ES2015
does not permit a function declaration as the body of a with
statement). This patch adds that scope.
BUG=chromium:542100
LOG=N
R=adamk
Review URL: https://codereview.chromium.org/1404803002
Cr-Commit-Position: refs/heads/master@{#31269}
This patch fixes an obscure edge case for functions defined as the
direct body of a for-of/for-in loop, such as the following:
for (foo in []) function foo() { return foo; }
Here, the first occurrence of foo should point to the outer scope;
however, before this patch, it pointed to the inner foo in an
invalid way which caused an assertion about the scope chain to fail.
This patch fixes the scope chain by inserting an extra scope for
the body of the loop, not including the header.
BUG=chromium:542099
LOG=N
R=rossberg
Review URL: https://codereview.chromium.org/1396663004
Cr-Commit-Position: refs/heads/master@{#31268}
The flag for deactivating break points also affects stepping, since both
are implemented via debug break slots. Fixing this by introducing a new
flag solely responsible for deactivating actual break points.
R=mvstanton@chromium.org
BUG=chromium:119800
LOG=N
Review URL: https://codereview.chromium.org/1402913002
Cr-Commit-Position: refs/heads/master@{#31236}
Looking up 'name' and 'message' properties at the same time and loading
the properties later can cause assertion failure if one of the properties
is an accessor and calling it changes the holder map. That may invalidate
the other lookup.
R=jkummerow@chromium.org
BUG=chromium:542101
LOG=N
Review URL: https://codereview.chromium.org/1403923002
Cr-Commit-Position: refs/heads/master@{#31229}
Clusterfuzz testing discovered that sloppy-mode block-scoped function
declarations introduce lexically-scoped variables in scopes that were
thrown away under the expectation that no lexically-scoped variables
were introduced. These cases are:
for (;;) function foo() {}
for (x in y) function foo() {}
This patch ensures that a block is created in those cases to hold the
lexically scoped variable. Usually, scope analysis should discover that
that block is not important, and it should not have a runtime
representation.
BUG=chromium:536750,chromium:536751
LOG=Y
R=adamk
Review URL: https://codereview.chromium.org/1382123002
Cr-Commit-Position: refs/heads/master@{#31109}
This name makes it clear that the flag (also the variant in the Compiler)
is talking about specializing to the function context instead of i.e. the
native context.
R=mstarzinger@chromium.org
Review URL: https://codereview.chromium.org/1372513003
Cr-Commit-Position: refs/heads/master@{#30934}
Whenever a generalization is computed, the inputs must be checked for being cleared, and if they are, the generalization must be Type::Any.
Hopefully this fixes Chromium issue 527994 as well.
BUG=v8:4325,chromium:527994
LOG=n
Review URL: https://codereview.chromium.org/1361103002
Cr-Commit-Position: refs/heads/master@{#30887}
This CL also renames wrongly named test for v8:4173.
BUG=v8:4121
LOG=Y
Review URL: https://codereview.chromium.org/1353363002
Cr-Commit-Position: refs/heads/master@{#30879}
The ES6 specification says that "Built-in functions that are ECMAScript
function objects must be strict mode functions", which in particular
means that you can never test for them using the "caller" field of a
sloppy mode function.
CQ_INCLUDE_TRYBOTS=tryserver.v8:v8_linux_layout_dbg,v8_linux_nosnap_dbg
R=mstarzinger@chromium.org
BUG=v8:105
LOG=n
Review URL: https://codereview.chromium.org/1347663002
Cr-Commit-Position: refs/heads/master@{#30750}
The assumption that every function body produces a value does not hold
for functions that e.g. unconditionally throw or endlessly loop. This
fixes the inlining logic to handle such cases.
R=bmeurer@chromium.org
TEST=mjsunit/regress/regress-crbug-530598
BUG=chromium:530598
LOG=n
Review URL: https://codereview.chromium.org/1333193005
Cr-Commit-Position: refs/heads/master@{#30738}
When looking up a special accessor for known TypedArray fields
("length", "byteLength", "byteOffset"), consider the entire prototype
chain, not only the direct prototype.
This allows subclasses of TypedArrays to benefit from fast specialized
accesses.
Review URL: https://codereview.chromium.org/1313493005
Cr-Commit-Position: refs/heads/master@{#30678}
Update the consolidated load case to carefully chose the load mode
based on the consolidated elements kind.
BUG=v8:4380
LOG=N
Review URL: https://codereview.chromium.org/1329793003
Cr-Commit-Position: refs/heads/master@{#30659}
This makes the C++ fallback implementations for the two intrinsics,
%Arguments and %ArgumentsLength composable with respect to inlining.
Using deoptimization information gives us accurate data here.
R=bmeurer@chromium.org
TEST=mjsunit/regress/regress-4374
BUG=v8:4374
LOG=n
Review URL: https://codereview.chromium.org/1328363002
Cr-Commit-Position: refs/heads/master@{#30654}
This switches Isolate::ComputeLocation to use baseline code when
computing message locations. This unifies locations between optimized
and non-optimized code by always going through the FrameSummary for
location computation.
R=bmeurer@chromium.org
TEST=message/regress/regress-4266
BUG=v8:4266
LOG=n
Review URL: https://codereview.chromium.org/1331603002
Cr-Commit-Position: refs/heads/master@{#30635}
- Move fast paths from builtins.cc ArraySlice to ElementsAccessor
- Handle more argument types in the fast path
BUG=
Review URL: https://codereview.chromium.org/1321773002
Cr-Commit-Position: refs/heads/master@{#30471}
This patch changes the switch scope desugaring to create blocks which
propagate their 'return value' for eval.
BUG=v8:4399
R=adamk
LOG=Y
Review URL: https://codereview.chromium.org/1309303006
Cr-Commit-Position: refs/heads/master@{#30454}
Switch statements introduce their own scope for cases, but this scope
is not necessarily executed in order, as the following function shows:
switch (x) {
case 1:
let y = 1;
case 2:
y = 2;
case 3:
print(y);
}
If x = 2 or x = 3, the code should throw a ReferenceError. However,
FullCodeGen's hole check elimination used the simple algorithm of
assuming that if the initializer was in the same scope, then it was
reached before the use, and therefore the hole check could be
eliminated.
This patch adds an extra bit to scopes, to track if they may
nonlinearly. The parser marks the scope that switch introduces as
nonlinear. FullCodeGen does not eliminate the hole check from
a scope which is nonlinear. This patch refactors FullCodeGen to
put the hole check elimination in one place, rather than in each
backend.
BUG=v8:3926
LOG=Y
R=adamk
Review URL: https://codereview.chromium.org/1312613003
Cr-Commit-Position: refs/heads/master@{#30453}
This fixes broken dynamic hole-checks for the temporal dead zone of
non-initializing assignments to {let} and {const} declared variables.
Also note that this exemplifies a case where the dynamic check for such
assignments to {let} declared variables can no longer be elided as the
comment suggested.
R=rossberg@chromium.org
TEST=mjsunit/regress/regress-4388
BUG=v8:4388
LOG=n
Review URL: https://codereview.chromium.org/1318693002
Cr-Commit-Position: refs/heads/master@{#30375}
The previous hack with HInstanceOfKnownGlobal was not only slower,
but also very brittle and required a lot of weird hacks to support it. And
what's even more important it wasn't even correct (because a map check
on the lhs is never enough for instanceof).
The new implementation provides a sane runtime implementation
for InstanceOf plus a fast case in the InstanceOfStub, combined with
a proper specialization in the case of a known global in CrankShaft,
which does only the prototype chain walk (coupled with a code
dependency on the known global).
As a drive-by-fix: Also fix the incorrect Object.prototype.isPrototypeOf
implementation.
BUG=v8:4376
LOG=y
Review URL: https://codereview.chromium.org/1304633002
Cr-Commit-Position: refs/heads/master@{#30342}
The ES2015 specification for switch statements 13.12.11 specifies that
they get their own lexical scope. This patch introduces such a scope
through a complex desugaring in terms of blocks, done so that Crankshaft
does not have to be updated to support multiple constructs providing
scopes.
Recommitting this patch after a bug fix in Crankshaft to allow a
desugaring with certain elements missing a source location:
https://codereview.chromium.org/1313443002
BUG=v8:4377
LOG=Y
R=adamk
Review URL: https://codereview.chromium.org/1309163003
Cr-Commit-Position: refs/heads/master@{#30340}
The simulator uses a separate JS stack, exhaustion of the C stack
however is not caught by JS limit checks. This change now lowers the
limit of the JS stack accordingly on function calls.
R=mvstanton@chromium.org
BUG=chromium:522380
TEST=mjsunit/regress/regress-crbug-522380
LOG=n
Review URL: https://codereview.chromium.org/1314623002
Cr-Commit-Position: refs/heads/master@{#30334}
Reason for revert:
Breaks cctest/test-cpu-profiler/SourceLocation on nosnap
Original issue's description:
> Add a separate scope for switch
>
> The ES2015 specification for switch statements 13.12.11 specifies that
> they get their own lexical scope. This patch introduces such a scope
> through a complex desugaring in terms of blocks, done so that Crankshaft
> does not have to be updated to support multiple constructs providing
> scopes.
>
> BUG=v8:4377
> LOG=Y
> R=adamk
>
> Committed: https://crrev.com/9edbc1f21eb1050cabbe3b8bc9aebf89ada7ebd7
> Cr-Commit-Position: refs/heads/master@{#30314}
TBR=adamk@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4377
Review URL: https://codereview.chromium.org/1309043004
Cr-Commit-Position: refs/heads/master@{#30316}
The parser has special behavior with respect to the bindings
of inner functions in sloppy mode which are not at the top
level of scopes. This behavior should be turned off when the
--harmony-sloppy-function flag is set, as lexical scoping
rules are used instead. Previously, the incorrect flag
--harmony-sloppy was used, resulting in a crashing bug.
BUG=chromium:520029
LOG=Y
R=adamk
Review URL: https://codereview.chromium.org/1303033003
Cr-Commit-Position: refs/heads/master@{#30315}
The ES2015 specification for switch statements 13.12.11 specifies that
they get their own lexical scope. This patch introduces such a scope
through a complex desugaring in terms of blocks, done so that Crankshaft
does not have to be updated to support multiple constructs providing
scopes.
BUG=v8:4377
LOG=Y
R=adamk
Review URL: https://codereview.chromium.org/1293283002
Cr-Commit-Position: refs/heads/master@{#30314}
This avoids many back-and-forth calls to the runtime.
This also slightly changes the way we avoid getters. Previously, we circumvent getting the name property of ReferenceError, SyntaxError and TypeError due to crbug/69187 (in order to avoid leaking information from those errors through a 'name' getter installed on their prototypes). Now we do that for all errors created by V8.
R=jkummerow@chromium.org, rossberg@chromium.org
BUG=crbug:513472, crbug:69187
LOG=N
Review URL: https://codereview.chromium.org/1281833002
Cr-Commit-Position: refs/heads/master@{#30105}
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}
The calculation now takes into account the size of the arguments object
if it is present in the optimized frame.
(Yang, many thanks for the awesome repro!)
BUG=chromium:514362
LOG=N
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/1264483008
Cr-Commit-Position: refs/heads/master@{#29973}
When the main thread terminates, it forcibly terminates all Worker threads.
When this happens, the threads objects were only half-created; they had a
JavaScript Worker object, but not a C++ worker object.
This CL fixes that bug, as well as some other fixes:
* Signatures on Worker methods
* Use SetAlignedPointerFromInternalField instead of using an External.
* Remove state_ from Worker. Simplify to atomic bool running_.
BUG=chromium:511880
R=jarin@chromium.org
LOG=n
Review URL: https://codereview.chromium.org/1255563002
Cr-Commit-Position: refs/heads/master@{#29911}
When a prototype object migrates from a slow to a fast map, where the slow map
was registered as a user of its own prototype, then the registration must be
transferred to the new map (just like MigrateToMap does for all other cases).
BUG=chromium:513602
LOG=y
NOTREECHECKS=true
Review URL: https://codereview.chromium.org/1263543004
Cr-Commit-Position: refs/heads/master@{#29898}
Previous to this CL, ICs used a slightly different code idiom
to get to C++ code from generated code than runtime intrinsics,
using an IC_Utility class that in essence provided exactly
the same functionality as Runtime::FunctionForId, but in its
own quirky way.
This CL unifies the two mechanisms, folding IC_Utility
away by making all IC entry points in C++ code, e.g. IC
miss handlers, full-fledged runtime intrinsics. This makes
it possible to eliminate a bunch of ad-hoc declarations and
adapters that the IC system had to needlessly re-invent.
As a bonus and the original reason for this yak-shave:
IC-related C++ runtime functions are now callable from
TurboFan.
Review URL: https://codereview.chromium.org/1248303002
Cr-Commit-Position: refs/heads/master@{#29811}
A sloppy mode eval call that establishes strict mode will leak that strictness
into the sloppy surrounding scope on recompile. This changes the structure
of the type feedback vector for the function and crashes follow.
The fix is straightforward.
BUG=491536, 503565
LOG=N
Review URL: https://codereview.chromium.org/1231343003
Cr-Commit-Position: refs/heads/master@{#29671}
--debug-code causes full-codegen on arm64 to emit different number
of calls, which confuses the debugger when on-stack replacing code
with recompiled debug version on-stack.
BUG=chromium:507070
TBR=mstarzinger@chromium.org
LOG=N
Review URL: https://codereview.chromium.org/1228353004
Cr-Commit-Position: refs/heads/master@{#29638}
The new implemtation counts the number of calls (or continuations)
before the PC to find the corresponding PC in the new code.
R=mstarzinger@chromium.org
BUG=chromium:507070
LOG=N
Review URL: https://codereview.chromium.org/1235603002
Cr-Commit-Position: refs/heads/master@{#29636}
When quit() is called, d8 shell exits without cleanup. If a worker is running,
it might be holding the context_mutex_, which if destroyed will DCHECK.
BUG=4279
R=jarin@chromium.org
LOG=n
Review URL: https://codereview.chromium.org/1231473002
Cr-Commit-Position: refs/heads/master@{#29557}
This one occurred when serializing an object. When the property getter threw an
exception, that value was skipped, but the property count wasn't updated. The
deserializer then tried to deserialize the wrong value.
BUG=chromium:506549
R=jarin@chromium.org
LOG=n
Review URL: https://codereview.chromium.org/1220193004
Cr-Commit-Position: refs/heads/master@{#29541}
Smi immediates are not supported, so instructions with Smi representations need their constants in a register. LAddI has already been doing this. The manifestation of the bug was that an operation would compute 0 instead of the correct result.
BUG=chromium:478612
LOG=y
R=verwaest@chromium.org
Review URL: https://codereview.chromium.org/1224623017
Cr-Commit-Position: refs/heads/master@{#29529}
The context constant cannot be materialized from the frame when we are
compiling for OSR, because the context spill slot contains the current
instead of the outermost context in full-codegen.
R=bmeurer@chromium.org
Review URL: https://codereview.chromium.org/1220013003
Cr-Commit-Position: refs/heads/master@{#29472}
This is more consistent with the DOM API, and is clearer w.r.t. which values
are available in the lexical environment of the Worker.
BUG=chromium:497295
R=jarin@chromium.org
LOG=n
Review URL: https://codereview.chromium.org/1218553004
Cr-Commit-Position: refs/heads/master@{#29426}
v8::Internal::List will DCHECK when indexing out of the array, even if just to
get the address, and the value is never used. So this construct will fail:
memcpy(p, &data[0], length);
When data is empty and length is 0.
BUG=chromium:505778
R=mstarzinger@chromium.org
LOG=n
Review URL: https://codereview.chromium.org/1216853003
Cr-Commit-Position: refs/heads/master@{#29388}
This makes sure that the exit control flow that merges the try-block
with the catch-block after a try-catch-statement creates a new merge
node in cases where it has to. Otherwise dangling phi nodes might have
the wrong number of value inputs.
R=bmeurer@chromium.org
TEST=mjsunit/regress/regress-crbug-505354
BUG=chromium:505354
LOG=N
Review URL: https://codereview.chromium.org/1213183003
Cr-Commit-Position: refs/heads/master@{#29362}
