Previously, for the various customisation points of String builtins
(like String.prototype.replace), we skipped the customisation symbol
lookup (like for Symbol.replace) for Smis.
But, we do need to do the lookup for Smis in case Number.prototype or
Object.prototype have the Symbol. This missing lookup was creating an
observable difference between Smis and HeapNumbers.
Bug: chromium:1092896
Change-Id: I8928d237fa74abeaa2aa81318b8903087c507f0d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2238030
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Auto-Submit: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68285}
The reference types wasm proposal dropped all subtyping. Subsequently,
the 'anyref' type was renamed to externref.
This changes all references of the *type* anyref to externref.
Additionally, the flag that permits this extension is renamed to
"reftypes" to mirror the proposal name.
Bug: v8:7748
Change-Id: Icf323f13b9660fd10540e65125af053fca3a03f9
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2232941
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Kim-Anh Tran <kimanh@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68270}
A previous CL removed the kNoThrow flags from both
SpeculativeBigIntAdd and SpeculativeBigIntSubtract. This introduced a
bug, because the JSTypeHintLowering phase, where these operators are
introduced during inlining, does not support the generation of throwing
operators.
Since these operators always deoptimize in case of an error, instead of
throwing the exception directly, it is safe to mark them as kNoThrow.
Bug: chromium:1091461
No-Try: true
No-Tree-Checks: true
Change-Id: I551616b0c462647574e5af8824d9ed7b3252659d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2235113
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68254}
Speculative BigInt addition fails to throw the expected exception
when called with non-BigInt inputs when the result of the computation
is unused. In paricular, this CL does:
- Remove kNoThrow on speculative BigInt operators
- Fix AddWithFeedback to not lose type feedback if builtin throws
to elide existing deopt loops
- Add handling of TypeCheckKind in RepresentationChanger where this
was previously ignored
Bug: chromium:1073440
Change-Id: I953a5b790fc3b37a6824f0b6546a0488c51fbb3b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2228493
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Mythri Alle <mythria@chromium.org>
Auto-Submit: Nico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68181}
We use StoreOwnIC to initialize the object after creating a new object
with CreateObjectLiteral. CreateObjectLiteral stores kHoleNaNInt64
to indicate an uninitialized double field. When we actually try
to store a NaN value into that field later using StoreOwnIC, IC avoids
actually storing the new value since the existing value is "same as"
the value we try to write. The float comparison treats all NaNs as
equal. In this particular case, we should actually store the new value
since kHoleNaNInt64 value is used to represent an uninitialized field.
This cl just stores the new value even when the existing value is same
as the new value for double fields. The check is still required to
correctly track const fields.
Bug: chromium:1082293
Change-Id: Ib37061802f2403545cffa6d6fef08be074b0825d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2228886
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68167}
All subtyping has been removed from the reference-types proposal. This
CL implements this proposal change now in V8.
R=manoskouk@chromium.org
Bug: v8:10556
Change-Id: I08ef064952278e03ea655461fa9f0c96426157c7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2222345
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68152}
Even in unreachable code, the targets of br_table have to have matching
types.
R=thibaudm@chromium.org
Bug: v8:10556
Change-Id: I2e85df3cb92f7910a6bcb5ac03927c424194660d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2218062
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Thibaud Michaud <thibaudm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68148}
With recent changes to the anyref proposal, null refs now have a type
immediate which declares the type of a null ref constant. Likewise,
the RefIsNull instruction is type aware now. This CL addresses these
proposal changes now.
R=jkummerow@chromium.org
Bug: v8:10556
Change-Id: I810dfa3a4ab4389afc9639f897cee5d43e9b62cb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2215172
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68141}
This reverts commit 76debfda32.
Reason for revert: Nullptr access in new test: https://ci.chromium.org/p/v8/builders/ci/V8%20Linux/37265
Original change's description:
> [wasm-simd][liftoff] Fix I64x2Mul
>
> The I64x2Mul overwrote the lhs/rhs if they are the same as dst. So when
> deciding if we need temporaries, we should not only check the
> cache_state, but whether they alias dst or not.
>
> Bug: chromium:1088273
> Change-Id: I82efa9b45e0a3d321a06efde60971ce95b21490f
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2225796
> Commit-Queue: Zhi An Ng <zhin@chromium.org>
> Reviewed-by: Clemens Backes <clemensb@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#68114}
TBR=clemensb@chromium.org,zhin@chromium.org
Change-Id: I5fd337b71d82d262d36ff410077a11c17b50036b
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:1088273
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2226756
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68117}
The I64x2Mul overwrote the lhs/rhs if they are the same as dst. So when
deciding if we need temporaries, we should not only check the
cache_state, but whether they alias dst or not.
Bug: chromium:1088273
Change-Id: I82efa9b45e0a3d321a06efde60971ce95b21490f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2225796
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68114}
Allocating a new feedback vector happens in two steps: We create an
empty structure and then initialize the array based on the
FeedbackMetadata.When allocating a new feedback array we could trigger
a GC which might flush the bytecode and associated feedback metadata.
This shouldn't happen in normal cases, because we either allocate
feedback vector after compilation or when we reach the expected budget.
In both cases, the age of the feedback vector should be 0 and hence
bytecode shouldn't be flushed. However, with debugger enabled we may
allocate feedback vectors even when the bytecode array is old
for example: when we enable precise invocation counters. This also
causes issues in tests with --stress-flush-bytecode. In the stress mode
we flush bytecode without considering the age. Holding on to the
feedback metadata prevents crashes in such cases.
Bug: v8:10560
Change-Id: Ie806ff4102cb5fcf257c8683d5ca957853e38c05
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2218066
Commit-Queue: Mythri Alle <mythria@chromium.org>
Auto-Submit: Mythri Alle <mythria@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68052}
Temporarily disable stress-bytecode-flush on
mjsunit/regress/regress-786784 while we investigate failures related
to bytecode flushing.
Bug: v8:10560
Change-Id: Ieb5cc7ba87da04133e98c6be25c9a499d79543e0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2218038
Reviewed-by: Marja Hölttä <marja@chromium.org>
Auto-Submit: Mythri Alle <mythria@chromium.org>
Commit-Queue: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68046}
In BinaryOpAssembler::Generate_BinaryOperationWithFeedback, the
feedback is stored only after the respective builtin/runtime call.
If this call throws an exception, the feedback is lost, leading
to a deopt loop in some cases. This CL fixes that issue by writing
the gathered feedback before passing control to the builtin.
Bug: chromium:1077197, v8:9441
Change-Id: I20e4b14815520224e2c6f8af1af6a89f754ccddf
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2202904
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Auto-Submit: Nico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68038}
This is a reland of 6204768bab
The original issue exposed the problem that NumberEqual performs
implicit conversion of oddballs to numbers, which is incorrect for
abstract equality comparison (i.e. 0 == null must not be true).
This reland fixes this by applying the following steps:
* Introduced a new kNumberOrBoolean value for CompareOperationFeedback,
CompareOperationHint, TypeCheckKind and CheckedTaggedInputMode.
* In CodeStubAssembler::Equal: Further distinguish between boolean and
non-boolean oddballs and set feedback accoringly.
* In JSTypedLowering: Construct [Speculative]NumberEqual operator with
CompareOperationHint::kNumberOrBoolean, when this feedback is present.
JSOperatorBuilder and operator cache are extended accordingly.
* In SimplifiedLowering: Propagate a UseInfo with new
TypeCheckKind::kNumberOrBoolean.
* This leads to the generation of CheckedTaggedToFloat64 in
RepresentationChanger with new CheckedTaggedInputMode::kNumberOrBoolean.
* In EffectControlLinearizer: Handle this new mode. Accept and convert
number and boolean and deopt for rest.
Original change's description:
> [turbofan] Improve equality on NumberOrOddball
>
> This CL cleans up CompareOperationFeedback by replacing it with a
> composable set of flags. The interpreter is changed to collect
> more specific feedback for abstract equality, especially if oddballs
> are involved.
>
> TurboFan is changed to construct SpeculativeNumberEqual operator
> instead of the generic JSEqual in many more cases. This change has
> shown a local speedup of a factor of 3-10, because the specific
> operator is way faster than calling into the generic builtin, but
> it also enables additional optimizations, further improving
> runtime performance.
>
> Bug: v8:5660
> Change-Id: I856752caa707e9a4f742c6e7a9c75552fb431d28
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2162854
> Reviewed-by: Mythri Alle <mythria@chromium.org>
> Reviewed-by: Georg Neis <neis@chromium.org>
> Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#67645}
TBR: tebbi@chromium.org
Bug: v8:5660
Change-Id: I12e733149a1d2773cafb781a1d4b10aa1eb242a7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2193713
Commit-Queue: Georg Neis <neis@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68037}
There was a legacy place in map code that wasn't fully ported to use
the strong, new SloppyArgumentsElements type because of code that used
hard-coded constants.
Bug: chromium:1086470
Change-Id: Ieba152e4bd92c89125f831949c2efb4f4219f95c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2215059
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Daniel Clifford <danno@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67984}
We can't consistently rewire the successor blocks of an unreachable node to
disconnect them from the graph when we are trying to maintain the schedule.
Instead simply leave the code there. As a future optimization we could add a
proper scheduled dead code elimination phase which can deal with this.
As a side-effect, one of the tests sees a int64 DeadValue, so add support for that
in the instruction selector.
BUG=chromium:1083272,chromium:1083763,chromium:1084953,v8:9684
Change-Id: I69a6feaeef4eae62110392e27ea848b28bccf787
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2209061
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Auto-Submit: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67953}
The proposal uses the lane shape, e.g. i64x2.anytrue, and we were using
s1x2.anytrue in our opcodes. This was a legacy naming, because we were
trying to bitpack the booleans. Now that we aren't doing that, rename
these to be more consistent with the proposal.
This was done with a straightforward sed script, changing both cpp code
and also some comments in mjsunit test files.
Bug: v8:10506
Change-Id: If077ed805de23520d8580d6b3b1906c80f67b94f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2207915
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67945}
This was introduced by https://crrev.com/c/2207137.
Load offsets can be negative.
Drive-by: Add a helper function to wrap the verbose static casts in
bounds checks.
Bug: chromium:1084872,chromium:1083450
Change-Id: I48934d04a8ab15a8fc347465064b190e32c00716
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2209066
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Auto-Submit: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67924}
Registers cannot be used as a merge destination if they have more than
one use, otherwise the merge will unexpectedly affect other uses of that
register.
R=ahaas@chromium.org,clemensb@chromium.org
Bug: chromium:1084151
Change-Id: I0d6ad97c585920357a37d95361e0320d32c71f4b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2208851
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Thibaud Michaud <thibaudm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67904}
So far this is mainly a readability improvement to specify
expectations on the packed argument. In the future we should also
check signedness during bytecode generation.
Drive-by: Update DCHECK to allow signed args to
CHECK_CURRENT_POSITION.
Bug: chromium:1083450
Change-Id: I9376ec691b51eb251c972309ad65dd6c04eec3ae
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2207137
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67880}
The lowering for anytrue was assuming that the input nodes are all
integers. The regression test added in https://crrev.com/c/2194471 calls
anytrue with float operands, this was causing the lowering to generate
cmpl instructions with a float register and an immediate, which is
wrong.
The fix is to use GetReplacementsWithType on the input nodes, but
only if the input were floats, since we use Word32Equal.
Drive-by clean up of comments in the aforementioned regression test.
Bug: v8:10535
Change-Id: I4de89516c178e9003a4c745808d831be87918381
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2203400
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67878}
The scheduler could schedule unreachable nodes on two basic blocks that
later merge. Update DCHECK in graph-assembler's basic block updater to
only check for the self-containedness of unreachable basic blocks
removed from the schedule after all the blocks have been re-written to
allow for this case.
BUG=chromium:1079446,v8:9684
Change-Id: I91899dbf389e4425542dbd2b1ca95c3f6ad79c05
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2196354
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67812}
The AVX implementation does not have dst == input(0), so the vminps call
was wrong. The intention is to compare the 2 input operands.
Bug: chromium:1081030
Change-Id: Id54074327a6aca4b75988fc9d85beccfeabfc791
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2194471
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67786}
... when one of the receivers is a JSArray that may have a read-only
length.
Bug: chromium:1069530
Change-Id: Idbaf1a9030bb5a0f9c25e30925f18f603a99832f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2196353
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67783}
Prior to this CL we still implemented a HasProperty-GetProperty
sequence when accessing named captures in GetSubstitution. This was
briefly part of the spec (we also threw an exception when the property
was not present), but since late 2017 the GetProperty call has been
unconditional.
See https://tc39.es/ecma262/#sec-getsubstitution.
Bug: v8:10513
Change-Id: Id82c06958b0b0feffc6eede580b99ab8676a0dae
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2195821
Auto-Submit: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67733}
... when the element is read-only in one of the prototypes:
* the length should not be updated,
* in strict mode the store operation should throw TypeError.
Bug: chromium:1055138
Change-Id: I7fc08e22c83f8a9848053cfe20851dc1b82f0e3d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2172090
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67717}
Scripts aren't callable functions. Even though internally they were for a
while, they aren't anymore. We shouldn't return them to users as if they were.
We already remove strict-mode functions from CallSites, so we now do the same
for internal functions that are created for scripts.
Bug: v8:10508
Change-Id: I270c714524439fba9ad90dd29826bed4811ba2b4
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2193716
Auto-Submit: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67709}
In the existing code we used a register of the UseScratchRegisterScope
for the destination address. However, this register is needed for the
ParallelRegisterMove as well. With this CL we use fixed registers for
the destination address and the offset as well. The CL also changes the
implementation of CalculateActualAddress to allow to set an explicit
register for the result.
R=clemensb@chromium.org
Bug: v8:10108, chromium:1079449
Change-Id: I39c11b9ffa5f3e937ce4820b9991482ad711b4b0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2192652
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67702}
This changes the existing implementation that creates an unresolved reference for those cases to look at exactly what scopes are relevant so it can correctly handle catch scopes and avoid re-resolving later.
Variable through with aren't marked as assigning since this information isn't relevant for the with itself; and if the with is passed through, there's no need to mark the outer variable as assigned since it's either initialized or it isn't.
The catch variable is assigned since it is relevant for the catch variable.
The CL uses LookupLocal which wouldn't work for deserialized scopes, but this isn't relevant because 1) eval scopes are declaration scopes, and 2) eval causes all outer variables to be maybe_assigned anyway.
Bug: chromium:1074737
Change-Id: I3febca479ddd1f3c62eae299190b06c0b4cd3746
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2187272
Commit-Queue: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Auto-Submit: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67683}
This reverts commit 6204768bab.
Reason for revert: A number of Clusterfuzz reports (e.g. https://bugs.chromium.org/p/chromium/issues/detail?id=1079474)
Original change's description:
> [turbofan] Improve equality on NumberOrOddball
>
> This CL cleans up CompareOperationFeedback by replacing it with a
> composable set of flags. The interpreter is changed to collect
> more specific feedback for abstract equality, especially if oddballs
> are involved.
>
> TurboFan is changed to construct SpeculativeNumberEqual operator
> instead of the generic JSEqual in many more cases. This change has
> shown a local speedup of a factor of 3-10, because the specific
> operator is way faster than calling into the generic builtin, but
> it also enables additional optimizations, further improving
> runtime performance.
>
> Bug: v8:5660
> Change-Id: I856752caa707e9a4f742c6e7a9c75552fb431d28
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2162854
> Reviewed-by: Mythri Alle <mythria@chromium.org>
> Reviewed-by: Georg Neis <neis@chromium.org>
> Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#67645}
TBR=rmcilroy@chromium.org,neis@chromium.org,mythria@chromium.org,nicohartmann@chromium.org
Change-Id: I3410310ed2b1ff2eaee70c1b91c3151d35866108
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:5660
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2190414
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67673}
This CL cleans up CompareOperationFeedback by replacing it with a
composable set of flags. The interpreter is changed to collect
more specific feedback for abstract equality, especially if oddballs
are involved.
TurboFan is changed to construct SpeculativeNumberEqual operator
instead of the generic JSEqual in many more cases. This change has
shown a local speedup of a factor of 3-10, because the specific
operator is way faster than calling into the generic builtin, but
it also enables additional optimizations, further improving
runtime performance.
Bug: v8:5660
Change-Id: I856752caa707e9a4f742c6e7a9c75552fb431d28
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2162854
Reviewed-by: Mythri Alle <mythria@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67645}
After sorting the work array but before writing the values back into
the actual receiver, we have an accessor check. This accessor check
needs to be stricter, in order to catch Array prototype protector
cell invalidations.
R=jgruber@chromium.org
Bug: chromium:1077508
Change-Id: I3c3bd4711f9019f9d4423701724319eee9d800a1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2187171
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Simon Zünd <szuend@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67638}
When the input to a speculative BigInt operation was an undefined
constant, no necessary type check was inserted by the
RepresentationChanger. This CL fixes this.
Bug: chromium:1077804
Change-Id: I3d4e15b1e018803d56e46c7b23b9d4b03832ba8a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2182455
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67610}
... to be in sync with KeyedLoadIC_SloppyArguments in handling OOB
accesses which may involve prototype chain walk.
Bug: chromium:1063796
Change-Id: I8421c19085dfd2f3b6360c64fd04f53b1351576c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2174504
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67541}
- Update opcode numbers, tests
- As the wasm-module-builder currently assumes opcode bytes, skip
the test that needs a multi-byte leb128 opcode
- Renumber post-MVP opcodes
Change-Id: I6531e954e63986dc6f7a3144ec054d16e6dc1b05
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2173952
Reviewed-by: Zhi An Ng <zhin@chromium.org>
Commit-Queue: Deepti Gandluri <gdeepti@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67517}
Spilling a register in Liftoff require a scratch register when the
offset of the stack slot from fp is greater than 2^12. This CL adds
a check to LiftoffAssembler::Spill on arm to check that a scratch
register is available. It also fixes one case where the scratch register
was not available.
R=clemensb@chromium.orgCC=zhin@chromium.org
Bug: chromium:1075953
Change-Id: Idb2bc7e26e3d4fbd6bb0eb6c9a9b8cfd8b3c569e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2172424
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67494}
With this CL the registers in a register pair get ordered such that the
low word register always has a lower register code than the high word
register. This should allow easier reasoning about the register
allocation, and prevent some register allocation bugs.
Background: for many operations in Liftoff, input registers are reused
as output registers. With register pairs, input register pairs are
reused as output register pairs. Additional reasoning, and sometimes
even additional code is needed when the registers of the output register
pair are swapped, i.e. when the high word register of the input becomes
the low word register of the output. With this CL the additional
reasoning is not necessary anymore, as the high word and low word
registers would get swapped during register allocation.
Additionally this CL fixes the logic of the last_spilled_regs list. This
list stored the last spilled registers, but recorded only one of the two
registers of a register pair. With this CL, both registers get recorded.
This CL does not have a regression test. The regression test was more
than 9000 lines long, and quite slow. I was not able to minimize it
substantially. The test would be fragile by nature, as it has to create
a special register configuration or otherwise does not test anything
meaningful. All in all I think it's better not to add the test.
R=clemensb@chromium.org
Bug: chromium:1074586
Change-Id: I4b2475b0c6537c7ce2e51fee281388cdd85f2953
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2168875
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67473}
Having no value argument in DataView setters (e.g. setFloat64) caused
wrong behavior in compiled code.
Bug: chromium:1071190
Change-Id: I37ddba8555dafad321f8d4c1352da8a501a98453
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2170091
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Auto-Submit: Nico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67451}
In AtomicOp64 ClearRegister is called twice to clear the registers r8
and r9. Thereby new registers may get allocated. We forgot to add the
newly allocated registers to pinned after the first call to
ClearRegister, which caused the same registers to be allocated again in
the second ClearRegister, and thereby caused the bug.
R=clemensb@chromium.org
Change-Id: I0d069aea4c9438fe30c30c22406b4075ddf3e95c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2170088
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67445}
This removes the {RedirectToWasmInterpreter} runtime function and the
respective method from {WasmDebugInfo}.
Some tests test specifically the interaction between compiled code and
the interpreter. They are irrelevant now and are deleted.
R=thibaudm@chromium.org
Bug: v8:10389
Change-Id: I38330fcb523f7c65968fdf03abc60af3392bdcc8
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2164793
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Thibaud Michaud <thibaudm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67427}
The interpreter will be moved to be test-only, hence
--wasm-interpret-all also needs to be removed.
Since we don't have any non-compiling tier any more, we also remove the
implication from --jitless to --wasm-lazy-compilation. Instead, we add
another CHECK that we can't be in jitless mode if we trigger any wasm
compilation.
All tests that just ran other tests and additionally passed
--wasm-interpret-all become redundant and are deleted. Also all
regression tests that explicitly specify --wasm-interpret-all are not
needed any more.
R=thibaudm@chromium.org
Bug: v8:10389
Change-Id: I5ddf20a842117a6c05e277a5308f5cfe42e6bfa5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2164792
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Thibaud Michaud <thibaudm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67419}
This reverts the changes made in
https://chromium-review.googlesource.com/c/v8/v8/+/1695465https://chromium-review.googlesource.com/c/v8/v8/+/1776078
We originally moved this protector to the native context to avoid
cross-native-context pollution of protector state. Ideally,
invalidating a protector in one NC should not affect any other NC.
But as it turns out, having the protector on the NC causes more
problems than it solves since all affected callers now need to find
the correct native context to check. Sometimes (e.g. in CSA regexp
builtins) it is possible to blindly check the current NC, but the
reasoning behind this optimization is tricky to understand.
Sometimes, fetching the correct NC is not possible due to access
restrictions. These implementation complexities outweigh the (unknown)
potential performance benefits.
In the future we should attempt to move away from the protector
concept for these kinds of checks.
Bug: chromium:1069964,v8:9463
Change-Id: I2cbb2ec7266282165dae5e4a6c8bdbda520c50a9
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2157382
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67415}
