Math fast-path cannot drop arguments because their side-effects
must be preserved. For example, Math.imul(x) dropped x entirely,
because if x is convertible to an integer, the result is 0.
This, however, is not OK because converting x to an integer might
throw.
Bug: chromium:818070, v8:7250, v8:7240
Change-Id: I8363e6dcd3fc78c879395aacb636d5782c3b023e
Reviewed-on: https://chromium-review.googlesource.com/948523
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51736}
This is a cosmetic change only.
Change-Id: I9a0ea9a23b4fc1490759433153c7d47e492b853d
Reviewed-on: https://chromium-review.googlesource.com/936624
Commit-Queue: Georg Neis <neis@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51733}
Implement in-place weak reference handling in GC.
Turn FeedbackVector::optimized_code_or_smi into an in-place weak reference (this
is the only in-place weak reference at this point).
(See bug for design doc.)
BUG=v8:7308
TBR=yangguo@chromium.org
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng
Change-Id: I16d65dc768f10ed431252e23a0df07bee9063534
Reviewed-on: https://chromium-review.googlesource.com/948493
Commit-Queue: Marja Hölttä <marja@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51731}
This makes sure that the object can be identified by a unique instance
type and hence is not accidentally confused with other FixedArrays on
the heap.
R=clemensh@chromium.org
BUG=v8:7509
Change-Id: I20521cdcabbbddecd89ca8cd4bb203a47e1db0cd
Reviewed-on: https://chromium-review.googlesource.com/946253
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51730}
ArrayIndexOf is a monolithic builtin which does some checking and then handles
three groups of fast arrays: those with holey doubles, compacted doubles and
SMIs+objects. TF cannot reuse this efficiently, because calling the TFJ
ArrayIndexOf duplicates some checks and also does not allow passing arguments
through registers. Similarly for ArrayInclude.
This CL splits the three different types of fast array handling into a separate
TF stubs, and makes the parent TFJ as well as TurboFan itself use them where
appropriate.
The TODOs not tackled in this CL inculde:
* passing an empty context to spare a register when possible
* inlining the search loop if there is any performance gain to it
(This is the contiunation of http://crrev.com/2757853002, moved due to Rietveld
deprecation.)
BUG=v8:5985
Change-Id: I00c97b71be4892f8bc7e1ed6d72e02087618a9a6
Reviewed-on: https://chromium-review.googlesource.com/573020
Commit-Queue: Vaclav Brozek <vabr@chromium.org>
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51729}
OpParameter<int32_t> was still used for an operator after the operators
parameter changed from int32_t to a struct. Coincidentally, the first
field of the struct holds the value previously stored in that int32_t,
so correctness tests did not catch this.
Bug: chromium:818611, v8:7517
Change-Id: Ie46f084f7fa8117cd3493fc5ceafac11553dc55e
Reviewed-on: https://chromium-review.googlesource.com/948546
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51728}
Since we currently only fill the constants table if we're serializing,
we should only create & finalize the table in that case. Otherwise,
leave it initialized to empty_fixed_array.
Cq-Include-Trybots: luci.v8.try:v8_linux64_fyi_rel_ng
Bug: v8:6666
Change-Id: I6ecbfac9dc9a9dac7ff0f11331be09b1cbfb4c18
Reviewed-on: https://chromium-review.googlesource.com/948490
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51727}
This changes the JSArrayIterator to always have only a single instance
type, instead of the zoo of instance types that we had before, and
which became less useful with the specification update to when "next"
is loaded from the iterator now. This greatly simplifies the baseline
implementation of the array iterator, which now only looks at the
iterated object during %ArrayIteratorPrototype%.next invocations.
In TurboFan we introduce a new JSCreateArrayIterator operator, that
holds the IterationKind and get's the iterated object as input. When
optimizing %ArrayIteratorPrototype%.next in the JSCallReducer, we
check whether the receiver is a JSCreateArrayIterator, and if so,
we try to infer maps for the iterated object from there. If we find
any, we speculatively assume that these won't have changed during
iteration (as we did before with the previous approach), and generate
fast code for both JSArray and JSTypedArray iteration.
Drive-by-fix: Drop the fast_array_iteration protector, it's not
necessary anymore since we have the deoptimization guard bit in
the JSCallReducer now.
This addresses the performance cliff noticed in webpack 4. The minimal
repro on the tracking bug goes from
console.timeEnd: mono, 124.773000
console.timeEnd: poly, 670.353000
to
console.timeEnd: mono, 118.709000
console.timeEnd: poly, 141.393000
so that's a 4.7x improvement.
Also make presubmit happy by adding the missing #undef's.
Bug: v8:7510, v7:7514
Change-Id: I79a46bfa2cd0f0710e09365ef72519b1bbb667b5
Reviewed-on: https://chromium-review.googlesource.com/946098
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51725}
On ia32, support for vsqrtss and vsqrtsd was missing, so I add the
implementation of these instructions and disassembly support.
On x64, disassembly support for vsqrtss was missing, while vsqrtsd was
implemented. Now both are implemented.
The implementation of f32.sqrt and f64.sqrt is very straight-forward on
ia32 and x64, we can immediately emit the {v}sqrtss or {v}sqrtsd
instruction.
R=ahaas@chromium.org
Bug: v8:6600
Change-Id: Icf3ec05a97a23e94cdf70f4a72f30dd02fbddd13
Reviewed-on: https://chromium-review.googlesource.com/944221
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@{#51724}
Implement in-place weak reference handling in GC.
Turn FeedbackVector::optimized_code_or_smi into an in-place weak reference (this
is the only in-place weak reference at this point).
(See bug for design doc.)
BUG=v8:7308
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng;master.tryserver.chromium.linux:linux_chromium_rel_ng
Change-Id: I0f9f992cb4ee0457c40b7c868317dfb607bfb906
Reviewed-on: https://chromium-review.googlesource.com/873638
Commit-Queue: Marja Hölttä <marja@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Reviewed-by: Hannes Payer <hpayer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51722}
Only create spread-related feedback slots when the array literal
actually contains a spread.
Bug: v8:5940
Change-Id: I0afad81d4bf1a86ebc1bf81f1213f680eb22bc49
Reviewed-on: https://chromium-review.googlesource.com/947955
Commit-Queue: Georg Neis <neis@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51721}
The CHECK didn't account for the recent introduction of
StoreInArrayLiteralIC.
Bug: v8:5940, chromium:818438
Change-Id: I73b4120eb39b16d766f0b1a9cb82ba44804b09a3
Reviewed-on: https://chromium-review.googlesource.com/947950
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51719}
This enables the v8_enable_embedded_builtins gn flag on non-ia32 builds
and adds a new --stress-off-heap-code test mode to fyi bots.
v8_enable_embedded_builtins=true changes accesses to constants and
external references to go through the root list in builtins code.
--stress-off-heap-code copies builtins code off-heap on isolate
creation.
A few drive-by-fixes:
- ensure that we actually inspect the correct builtin during
isolate-independence testing.
- relax tests to decrease maintenance (now we only fail if a builtin
should be isolate-independent but isn't).
- switch to a different off-heap-trampoline register on arm due to
conflicts with custom stub linkages.
Cq-Include-Trybots: luci.v8.try:v8_linux64_fyi_rel_ng
Bug: v8:6666
Change-Id: I09ad3c75cb4342f4c548ea780f275993730896c8
Reviewed-on: https://chromium-review.googlesource.com/934281
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Hablich <hablich@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51717}
- Removes Reserve, Free (overload) and SetProtection methods.
- Updates comment on enum which we still need to distinguish
between allocated and reserved ArrayBuffers.
Bug: chromium:799573
Cq-Include-Trybots: master.tryserver.chromium.linux:linux_chromium_rel_ng
Change-Id: I1b4e08f97c22ae6b6af847fbcdde047be62fecf8
Reviewed-on: https://chromium-review.googlesource.com/924603
Reviewed-by: Hannes Payer <hpayer@chromium.org>
Reviewed-by: Eric Holk <eholk@chromium.org>
Commit-Queue: Bill Budge <bbudge@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51714}
Per spec, accesses to a DataView object must throw a TypeError if
the underlying ArrayBuffer has been detached/neutered. Since that
implies a length of 0, we used to detect this as an out-of-bounds
access and throw a RangeError. Adding a separate check for buffer
detachedness lets us distinguish both cases properly.
Bug: v8:4895
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: I1c1d4145dcd77dfb69f61062e14a6e8e538d45eb
Reviewed-on: https://chromium-review.googlesource.com/947585
Reviewed-by: Adam Klein <adamk@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51712}
When the multiplication steps fail, they have already thrown an
exception internally, so we should not throw another.
The power-of-two fast path erroneously did not throw at all for
a few input values.
Bug: chromium:818277
Change-Id: If90f6aa3e77fc72e3434daca3b898c77739933ab
Reviewed-on: https://chromium-review.googlesource.com/947254
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Adam Klein <adamk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51711}
- Merge new fail expectations that are dupes of existing issues
into the respective sections.
- Stop skipping tests we can run now.
Bug: v8:6791, v8:7511
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: I4e17ff8eb5d2596561a138e34c441b00b761d7d2
Reviewed-on: https://chromium-review.googlesource.com/947321
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51710}
... and use it in the implementation of array literal spreads,
replacing calls to %AppendElement.
Array spreads in destructuring will be taken care of in a separate CL.
Bug: v8:5940, v8:7446
Change-Id: Idec52398902a7fd3c1244852cf73246f142404f0
Reviewed-on: https://chromium-review.googlesource.com/915364
Commit-Queue: Georg Neis <neis@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51709}
There are some unused bits in a BigInt's bit field. We never read their
their values explicitly, but whenever the entire object is moved around
(for serialization, or GC), this uninitialized memory is accessed. This
patch fixes that by initializing the entire field after allocation of a
BigInt, not just the bits we actually use.
Bug: chromium:818109
Change-Id: I5a4d24c3240242157b902c696fa9bb779799280d
Reviewed-on: https://chromium-review.googlesource.com/946676
Reviewed-by: Adam Klein <adamk@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51708}
This patch removes the ProxyConstructor_ConstructStub builtin,
merging its functionality into the refactored ProxyConstructor
TurboFan builtin.
This brings us closer to our goal of deprecating the `construct_stub`
field in `SharedFunctionInfo`.
Bug: v8:7503, v8:7518
Change-Id: Iee76ba1a116ba61a543da529ec55149df333dcca
Reviewed-on: https://chromium-review.googlesource.com/946488
Commit-Queue: Mathias Bynens <mathias@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51707}
Change-Id: I8f842187d9a02657474c47f5ea1c3257a7ec9317
Reviewed-on: https://chromium-review.googlesource.com/916143
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51706}
Unfortunately, different runtime libraries and/or compilers differ on
whether a class without any copy constructor, move constructor, copy
assignment and move assignment operator is considered trivially
copyable.
See discussion on https://crrev.com/c/941521.
This CL adds a comment about this, and deletes a test for this specific
case.
R=mstarzinger@chromium.orgCC=jyan@ca.ibm.com, ivica.bogosavljevic@mips.com
Change-Id: Ie07adda370e5e955b782e72356b50121477d4623
Reviewed-on: https://chromium-review.googlesource.com/944081
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51704}
This changes the BytecodeGraphBuilder to connect runtime calls that
don't return normally, but always throw exceptions, to End (via a Throw
node), instead of inserting Phis on the return values. This unblocks
the new optimization approach for array iteration.
Bug: v8:7510, v8:7514
Change-Id: Ic78216cc27034f191c4850e476f24e598c17deca
Reviewed-on: https://chromium-review.googlesource.com/946250
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51702}
Folding _ + NaN => NaN can widen type None to a constant type, which leads to floating DeadValue nodes. This CL fixes this by removing the optimization. Alternatively, we should consider removing all nodes of type None in simplified lowering.
Bug: chromium:817225
Change-Id: I2a126b360d70d3626f8a3c5e73ac72dc980ac8b3
Reviewed-on: https://chromium-review.googlesource.com/946129
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51699}
This CL removes the last method taking a raw callback as argument.
Bug: chromium:782550
Change-Id: Ia7f19120be7dcb2648f11cd3cdb8880f8818490d
Reviewed-on: https://chromium-review.googlesource.com/822198
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51695}
This patch removes the NumberConstructor_ConstructStub builtin,
merging its functionality into the refactored NumberConstructor
TurboFan builtin.
This brings us closer to our goal of deprecating the `construct_stub`
field in `SharedFunctionInfo`.
Bug: v8:7503, v8:7518
Change-Id: Ia407a2d43cda9fc7b26e13274fd79f110b6232ca
Reviewed-on: https://chromium-review.googlesource.com/946008
Commit-Queue: Mathias Bynens <mathias@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51694}
- Add B/KiB/MiB units to histograms
- Auto select fileReader to open dialog when pressing enter
Bug: v8:7266
Change-Id: I6fa56d2fa112f6ddbd541304cd26a1f6bd9322ce
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/946128
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51693}
This prevents us from accidentally overwriting the default map of the
empty_fixed_array since NewFixedArray might just return that.
Change-Id: Id2fd20db1c07ec0289905ce5b0e12b5530a60229
Reviewed-on: https://chromium-review.googlesource.com/942923
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51692}
This is a reland of b8bc26d099
Original change's description:
> [turbofan] Preserve order of compares in switches
>
> This CL makes sure that control flow optimization does
> not change the order of switches that ultimately get
> lowered to a series of comparisons anyway.
>
> Bug: v8:7326
> Change-Id: If004de6b71a7e9504d37754c847ca108a64e49db
> Reviewed-on: https://chromium-review.googlesource.com/941952
> Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
> Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
> Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#51679}
Bug: v8:7326
Change-Id: Ifbe61dece499c98bbd49fa3ae9b99ccf4e955ddc
Reviewed-on: https://chromium-review.googlesource.com/945770
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51691}
This makes sure that the object can be identified by a unique instance
type and hence is not accidentally confused with other FixedArrays on
the heap.
R=clemensh@chromium.org
BUG=v8:7509
Change-Id: Ib3dcdb4559821d0ad9ca8cced37754e0e3c1d578
Reviewed-on: https://chromium-review.googlesource.com/943781
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51687}
