Add a protector cell for string bounds checks that is being used to
protect speculative bounds for String.prototype.charCodeAt and
String.prototype.charAt in TurboFan (and Crankshaft). This way we don't
have the diamond in optimized code, which stands in the way of other
optimizations for charCodeAt that are currently being worked on by
petermarshall@.
BUG=v8:6391
TBR=mlippautz@chromium.orgR=petermarshall@chromium.org
Review-Url: https://codereview.chromium.org/2905623003
Cr-Commit-Position: refs/heads/master@{#45514}
Add a sequential string type to the compiler, and transform
charCodeAt on SeqString into SeqStringCharCodeAt.
SeqStringCharCodeAt can handle one and two byte strings.
Bug: v8:6391
Change-Id: I2785257522c28f3b268c9833f5313e9630cb982a
Reviewed-on: https://chromium-review.googlesource.com/509573
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Peter Marshall <petermarshall@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45508}
We're now using explicit APIs.
Bug:
Change-Id: I4a4248e44543f6e7dfcbdc66456e610fb98ff5ee
Reviewed-on: https://chromium-review.googlesource.com/513406
Commit-Queue: Brad Nelson <bradnelson@chromium.org>
Reviewed-by: Brad Nelson <bradnelson@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45500}
Previously the inlining of accessors into try-blocks (i.e. try/catch,
try/finally, for-of, etc.) was disabled in JSNativeContextSpecialization,
which prevented a couple of interesting optimizations, i.e. we end up
with a LOAD_IC in optimized code for this simple example:
class A { get x() { return 1; } }
function foo(a) {
try {
return a.x;
} catch (e) {
return 0;
}
}
foo(new A)
This is now fixed and the accessors are properly rewired into the
handler chain.
BUG=v8:6278,v8:6344,v8:6424
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2902533003
Cr-Commit-Position: refs/heads/master@{#45485}
This makes sure that property lookups on the provided imports object are
non-observable to JavaScript. It allows instantiation failures to fall
back to JavaScript proper without accidentally calling accessors twice.
Also accessors might invalidate previous checks done during linking or
throw exceptions.
R=clemensh@chromium.org
TEST=mjsunit/regress/regress-crbug-719384
BUG=chromium:719384
Change-Id: I3db2672d2a496110f705d02b82878e70cd5d701f
Reviewed-on: https://chromium-review.googlesource.com/509552
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45481}
Validation normally happens while generating the turbofan graph of a
wasm function. For lazy compilation (behind the flag
--wasm-lazy-compilation), we skip this graph generation step during
module generation. Thus we need to validate explicitely.
R=ahaas@chromium.org
BUG=chromium:724851
Change-Id: Ic70887c0d823460a272d0bb636dc98b2b7a7e55e
Reviewed-on: https://chromium-review.googlesource.com/509574
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45478}
The validation of utf8 strings in WebAssembly modules used the character
kBadChar = 0xFFFD to indicate a validation error. However, this
character can appear in a valid utf8 string. This CL fixes this problem
by duplicating some of the code in {Utf8::CalculateValue} and inlining
it directly into Utf8::Validate. Note that Utf8::Validate is used only
for WebAssembly.
Tests for this change are in the WebAssembly spec tests, which I will
update in a separate CL.
R=vogelheim@chromium.org
Change-Id: I8697b9299f3e98a8eafdf193bff8bdff90efd7dc
Reviewed-on: https://chromium-review.googlesource.com/509534
Reviewed-by: Daniel Vogelheim <vogelheim@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45476}
This is to avoid ClusterFuzz picking up and using those calls.
With the proper syntax (no whitespace), they are recognized as runtime
calls and will be checked against a whitelist.
R=mstarzinger@chromium.org
BUG=chromium:724459
Change-Id: I5533f066feeb66f622230b12f79f9d227e2b2465
Reviewed-on: https://chromium-review.googlesource.com/509575
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45475}
If the maximum number of memory pages is raised using
--wasm-max-mem-pages, we might allocate more than kMaxInt bytes for
wasm memory. The byte length is stored as int in JSArrayBuffer, hence
this can lead to failures.
Thus, we now additially check against kMaxInt, and fail instantiation
if this check fails.
Drive-by: Add/fix more bounds checks.
R=ahaas@chromium.org
BUG=chromium:724846
Change-Id: Id8e1a1e13e15f4aa355ab9414b4b950510e5e88a
Reviewed-on: https://chromium-review.googlesource.com/509255
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45465}
This makes sure that function imports without a single call site within
the asm.js module are still preserved in the WebAssembly module, hence
preserving intended JavaScript semantics during module instantiation.
R=clemensh@chromium.org
TEST=mjsunit/regress/regress-crbug-722348
BUG=chromium:722348
Change-Id: I624d0e52b32b864c1e3002187a99a0a63834a4b0
Reviewed-on: https://chromium-review.googlesource.com/509450
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45452}
This adds a bunch of assertions to CSA, mostly about documenting and checking
parameter types.
Drive-by-change: Removed unused function.
BUG=v8:6325
Review-Url: https://codereview.chromium.org/2847923003
Cr-Original-Commit-Position: refs/heads/master@{#45398}
Committed: b14a981496
Review-Url: https://codereview.chromium.org/2847923003
Cr-Commit-Position: refs/heads/master@{#45443}
This fixes crashes during validation when trying to construct modules
with excessively large function tables. The {WasmModuleBuilder} now
gracefully checks against existing WebAssembly implementation limits.
R=clemensh@chromium.org
TEST=mjsunit/regress/regress-crbug-715455
BUG=chromium:715455
Change-Id: Ia9738cb0b49a1eb4caf073b75301c0303f295699
Reviewed-on: https://chromium-review.googlesource.com/509530
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45429}
As per spec, (https://github.com/WebAssembly/design/pull/1068), we
don't have compile/instantiate overloads anymore, instead, we
have explicitly named members.
This change introduces the new APIs, implements instantiateStreaming
based on compileStreaming, and uses the existing embedder mechanism.
It does not yet remove the functionality from compile/instantiate -
we do that after we adopt the new APIs on the blink side.
Also, it temporarily handles exceptions on the v8 side, which is also
something we'll move to the blink side.
Bug:
Change-Id: I77673b1c0d395dfcf13b2f25464fd5dfd99c8d82
Reviewed-on: https://chromium-review.googlesource.com/508852
Commit-Queue: Brad Nelson <bradnelson@chromium.org>
Reviewed-by: Brad Nelson <bradnelson@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45411}
Generators were previously treated as "top level" for preparsing purposes,
since all their variables are context-allocated. But doing so isn't quite
correct: the allocation of the "arguments" variable for a generator
depends on whether it's referenced, and so an inner arrow function
which references "arguments" won't properly trigger allocation of
"arguments" since the reference will not be noticed in the preparser.
The same problem exists for "this" since commit 68f0a47b28a96a4966e7b747bfa304b555e726d1;
before that commit, all generators implicitly referenced their "this" argument
as part of the desugaring. With that implicit reference gone, "this"
falls into the same problem as arguments.
This patch restricts the special "top level" treatment to modules,
which have only a trivial "this" binding (it's always undefined), and no
arguments binding. Moreover, all code inside modules is strict, meaning
that unresolved references to "this" will also result in undefined.
R=marja@chromium.org
Bug: chromium:723132
Change-Id: I814d145fb8f3f1a65abb48e4e35595428d063051
Reviewed-on: https://chromium-review.googlesource.com/508055
Reviewed-by: Marja Hölttä <marja@chromium.org>
Commit-Queue: Adam Klein <adamk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45399}
This adds a bunch of assertions to CSA, mostly about documenting and checking
parameter types.
Drive-by-change: Removed unused function.
BUG=v8:6325
Review-Url: https://codereview.chromium.org/2847923003
Cr-Commit-Position: refs/heads/master@{#45398}
IC system does its best to properly mark stable transition source maps
as unstable (see https://chromium-review.googlesource.com/483442)
however an already recorded map can be deprecated later and the
optimizing compiler may try to generate an elements kind transition
from the updated version of deprecated map which can "become" stable
again.
Bug: chromium:723455
Change-Id: Ic0c392f153587c3cd7c7623a3a6ea85ec72ad5bd
Reviewed-on: https://chromium-review.googlesource.com/507887
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45384}
Original CL description:
[compiler] Delay allocation of heap numbers for deoptimization literals.
... until after the main bulk of code generation, which will soon run on a
different thread.
Bug: v8:6048, chromium:722978
Change-Id: I690c0b009211a2bac60cf06f577720a914c21000
Reviewed-on: https://chromium-review.googlesource.com/507207
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45371}
The underlying issue is that TF Nodes cannot handle input counts
outside the integer range. On an illegal br_table instruction, we
generated a switch node with a control output count >kMaxInt.
Operator::ControlOutputCount turned this into a negative integer later,
leading to a failing DCHECK.
Since such large numbers cannot occur in any valid wasm function anyway,
we just add an additional check to the br table count. There is already
a TODO in the code to change Operator::ControlOutputCount to size_t.
R=ahaas@chromium.org
BUG=chromium:722445
Change-Id: I1975072226e073dee6c8da3b9fa9a050a4695917
Reviewed-on: https://chromium-review.googlesource.com/505496
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45365}
The interpreter does not implement all asm.js specific opcodes. Thus
the combination of --validate-asm and --wasm-interpret-all might crash.
The interpreter does not need to execute asm.js modules, as they are
debugged by executing them in turbofan instead of the wasm interpreter.
This CL thus excludes asm.js modules from --wasm-interpret-all.
R=ahaas@chromium.org
BUG=chromium:719175
Change-Id: I14228ea11ee3ea8a229cfa6e4179338a442b6cca
Reviewed-on: https://chromium-review.googlesource.com/506160
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45364}
Migrate the Object.keys builtin to the CodeStubAssembler and
use the enum cache backing store whenever it is available. This
gives a nice speedup of 1.5x to 2x when using Object.keys on fast-mode
objects that have (or can have) an enum cache.
R=cbruni@chromium.org
BUG=v8:5269,v8:6405
Review-Url: https://codereview.chromium.org/2853393002
Cr-Commit-Position: refs/heads/master@{#45361}
Now that the optimized code hangs off the feedback vector, it is possible
to check whether a function has optimized code available every time it's
called in the interpreter entry trampoline. If optimized code exists, the
interpreter entry trampoline 'self-heals' the closure to point to the
optimized code and links the closure into the optimized code list.
BUG=v8:6246
Change-Id: I53b095db2a75ae4824c8195faf8649d766c86118
Reviewed-on: https://chromium-review.googlesource.com/501967
Reviewed-by: Michael Stanton <mvstanton@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45328}
This makes sure that the evaluation result of the first expression in
for-statements is properly dropped, to leave the stack in a balanced
state after the statement. It also makes sure validation failures in
said expression are handled correctly.
R=clemensh@chromium.org
TEST=mjsunit/regress/regress-crbug-721835
BUG=chromium:721835
Change-Id: I7e6cff4cea0bbf5aad6a3459e27a08ea814dbdbe
Reviewed-on: https://chromium-review.googlesource.com/506148
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45299}
With this CL SloppyArguments immediately go to dictionary elements on
deletion, keeping the arguments backing store packed.
Bug: v8:6251
Change-Id: I90d1972179447bf6810e7fe2b8e0bc8703b38d9d
Reviewed-on: https://chromium-review.googlesource.com/486921
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45286}
This makes sure that the order of exports as they appear in asm.js
modules is maintained globally (not just per function) while being
translated to a WASM module.
R=clemensh@chromium.org
TEST=mjsunit/asm/asm-validation
BUG=chromium:720586
Change-Id: I8b26d717ae2f88467d41670bced901f196c7b3fc
Reviewed-on: https://chromium-review.googlesource.com/503708
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45277}
This makes sure that function variables as well as function table
variables are properly typed as immutable, hence assignments to them
should cause validation failures.
R=clemensh@chromium.org
TEST=mjsunit/asm/immutable
BUG=chromium:721271
Change-Id: Ia3f65fd0782ca571ffcf99520fdbd8fc5a359d16
Reviewed-on: https://chromium-review.googlesource.com/503209
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45256}
Clearing the pending exception is not enough - if we want to swallow an
exception while currently on top of an external handler (e.g. TryCatch),
we also need to clear external_caught_exception.
BUG=chromium:719380
Review-Url: https://codereview.chromium.org/2870423002
Cr-Commit-Position: refs/heads/master@{#45247}
This makes sure that the checking of stdlib values during module
instantiation is non-observable. It is needed to prevent observable
double evaluation of the involved property loads in case of failures
during instantiation and also fixes some issues with exceptions
happening during property loads.
R=clemensh@chromium.org
TEST=mjsunit/asm/asm-stdlib
BUG=v8:6297
Change-Id: I1d0c371e51bee8186d14fa794fb3f9b7f67e5944
Reviewed-on: https://chromium-review.googlesource.com/501887
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45237}
Since the feedback vector is itself a native context structure, why
not store optimized code for a function in there rather than in
a map from native context to code? This allows us to get rid of
the optimized code map in the SharedFunctionInfo, saving a pointer,
and making lookup of any optimized code quicker.
Original patch by Michael Stanton <mvstanton@chromium.org>
BUG=v8:6246,chromium:718891
TBR=yangguo@chromium.org,ulan@chromium.org
Change-Id: I3bb9ec0cfff32e667cca0e1403f964f33a6958a6
Reviewed-on: https://chromium-review.googlesource.com/500134
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45234}
This fixes cases where falling off the end of a non-void function was
accepted as a valid asm.js module. This led to translation failures in
the WASM code. Only void functions are allowed to fall off the end.
R=clemensh@chromium.org
TEST=mjsunit/asm/return-types
BUG=chromium:719286
Change-Id: I7b1c9ba5381b87a23cf0a2171bee5e5f5e8cd9de
Reviewed-on: https://chromium-review.googlesource.com/500312
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45221}
For small arrays, it's way faster to just move the elements instead of
doing the fairly complex and heavy-weight left-trimming. Crankshaft has
had this optimization for small arrays already; this CL more or less
ports this functionality to TurboFan, which yields a 4x speed-up when
using shift on small arrays (with up to 16 elements).
This should recover some of the regressions reported in the Node.js issues
https://github.com/nodejs/node/issues/12657
and discovered for the syncthrough module using
https://github.com/mcollina/syncthrough/blob/master/benchmarks/basic.js
as benchmark.
R=jarin@chromium.org
BUG=v8:6376
Review-Url: https://codereview.chromium.org/2874453002
Cr-Commit-Position: refs/heads/master@{#45216}
This tests and fixes validation failures caused by assignments to
variables holding functions references (which are all considered
immutable). Such references can come from "stdlib" or "foreign".
R=clemensh@chromium.org
TEST=mjsunit/asm/global-imports
BUG=chromium:719382
Change-Id: Ic02be765e0773a6cc74a54e11a09d42ffb683cb8
Reviewed-on: https://chromium-review.googlesource.com/500188
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45189}
Due to speculative optimizations, the compiler can run into situations
where it's asked perform impossible operations, like loading a tagged
element as a float64 instead. All of this is guaranteed to be in dead
code (unless there's a bug), but leads to confusion and violates
assumptions in the compiler (that make perfect sense for code that is
not dead). So teach LoadElimination not to mix up element accesses with
incompatible representations.
BUG=chromium:719479
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2866233002
Cr-Commit-Position: refs/heads/master@{#45185}
