Refactor-only, this moves interrupt scopes and stack guard code into
their own dedicated files.
Change-Id: I5723a04786a04bba31a0da54622f3cd0b926ef07
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655288
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62141}
{V8 Linux 64 - pointer compression} builder uploads dchecks enabled builds.
This CL creates a new builder that compiles V8 without dchecks enabled.
This CL uses the inverted naming predicate {without dchecks} to avoid renaming
the existing builder to {with dchecks} to avoid doing renames over multiple
repositories for a temporary builder that we'll remove after the ptr compression
merge to master.
R=sergiyb@chromium.orgCC=machenbach@chromium.org
Bug: v8:9345
Change-Id: I9e8cc1a9eb59325fd8eecc8fdcd2778b4da005c0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1657922
Commit-Queue: Tamer Tas <tmrts@chromium.org>
Commit-Queue: Sergiy Belozorov <sergiyb@chromium.org>
Auto-Submit: Tamer Tas <tmrts@chromium.org>
Reviewed-by: Sergiy Belozorov <sergiyb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62140}
As shown in the commit description of https://crrev.com/c/1619763, the JSON
format was supposed to refer to a list of "spaces" and not "pages", this was a
typo.
Bug: v8:9186
Change-Id: I1a674dac8af4b27b7ee46041e8c7a533bad8e68b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1657917
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#62139}
This check shouldn't have been there. Even with lazy feedback allocation
we still transition to pre-monomorphic from uninitialized. We could
remove pre-monomorphic states with lazy-feedback allocation but that
requires changes at several other places.
Change-Id: I8f878a83f0fe3200eb530a34a74811639dcdc153
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1634920
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62138}
Swap bits between bitfield2 and bitfield3 so that bitfield2 doesn't change
across named property transitions. This will allow us to share bf1/bf2 through
the descriptor array.
Change-Id: I3579ae89189ae0729cd492db1afb29cf90981f6a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1657908
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62134}
This is a reland of 823795fc2b, the reason
for the revert was flushing out a bug that was now fixed independently in
https://chromium-review.googlesource.com/c/v8/v8/+/1655307
Drive-by-fix: Correct wrong offset in CloneObjectIC fast-path.
Original change's description:
> [ic] Fix typo in Runtime_CloneObjectIC_Miss.
>
> https://chromium-review.googlesource.com/1649554 introduced a typo into
> Runtime_CloneObjectIC_Miss, where it wouldn't update the IC state UNLESS
> the source map is deprecated, which is the wrong way around of course.
>
> Bug: chromium:973045, v8:7611, v8:9114, v8:9183, v8:9343
> Change-Id: I7d6e0709e66ce4aaaf4a628d64ab801b84c8993c
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655291
> Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
> Commit-Queue: Toon Verwaest <verwaest@chromium.org>
> Reviewed-by: Toon Verwaest <verwaest@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#62106}
Bug: chromium:973045, v8:7611, v8:9114, v8:9183, v8:9343
Change-Id: I763d9eeab95043bed3bc4849fc3ddcda7787169a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655651
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62133}
StackGuard::HandleInterrupts used to take a lock for testing and
clearing each individual interrupt bit. This CL changes that to a
single read up front.
Slight behavioral changes:
1. A TERMINATE_EXECUTION interrupt is now handled first; we
immediately exit and preserve all other interrupts (in case V8 is
later resumed).
2. Since interrupts are read once, it is no longer possible to request
an interrupt *within* HandleInterrupts that will later be processed
within the same HandleInterrupts call.
3. Stack limits are reset immediately after reading the interrupt
bits, and prior to actually processing the interrupts.
Bug: v8:9328
Change-Id: I3048bb413213d11307df49e0014b64a2b43444e0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1653115
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62132}
ObjectPreventExtensions and ReflectPreventExtensions are now Torque builtins (previously CPP) and the Proxy path is implemented completely in Torque while everything else calls into runtime (and is thus a bit slower than previously).
Perf improvement in micro-benchmark JSTests/Proxies
Before:
PreventExtensionsWithoutTrap-Proxies(Score): 1978
PreventExtensionsWithTrap-Proxies(Score): 739
After:
PreventExtensionsWithoutTrap-Proxies(Score): 3017
PreventExtensionsWithTrap-Proxies(Score): 2044
Bug: v8:6664
Change-Id: I6505d730cea6b0d197f6f5d0540b39056c8b763d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1652688
Commit-Queue: Z Nguyen-Huu <duongn@microsoft.com>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62130}
JSModuleNamespace does not have well defined CreationContext: current
implementation of JSReceiver::GetCreationContext crashes on CHECK.
R=lushnikov@chromium.org,yangguo@chromium.org
Bug: none
Change-Id: Ie2c0bfa39117d42d81f9709c21376c177b18e5ba
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1652559
Reviewed-by: Yang Guo <yangguo@chromium.org>
Commit-Queue: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62128}
Port ea4206556e
Original Commit Message:
API calls made via the CallApiCallback builtin, which is used from the
ICs and optimized code, are currently misattributed to the wrong counter
InvokeFunctionCallback instead of FunctionCallback. In addition we don't
use the C trampoline when only runtime call stats are enabled, but the
Chrome DevTools profiler is not active, which means that these calls
will not be attrituted properly at all, and that had to be worked around
using all kinds of tricks (i.e. disabling fast-paths in ICs when RCS is
active and not inlining calls/property accesses into optimized code
depending on the state of RCS).
All of this was really brittle and only due to the fact that the central
builtin didn't properly check for RCS (in addition to checking for the
CDT profiler). With this fix it's now handled in a central place and
attributed to the correct category, so user code doesn't need to worry
about RCS anymore and can just call straight into the fast-path.
core hand-written native code with the API callback logic.
R=bmeurer@chromium.org, joransiu@ca.ibm.com, jyan@ca.ibm.com, michael_dawson@ca.ibm.com
BUG=
LOG=N
Change-Id: I2d200be4544cf62393330bb2891b6ba6f088db68
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655343
Reviewed-by: Junliang Yan <jyan@ca.ibm.com>
Reviewed-by: Joran Siu <joransiu@ca.ibm.com>
Commit-Queue: Junliang Yan <jyan@ca.ibm.com>
Cr-Commit-Position: refs/heads/master@{#62126}
This change adjusts object initialization order for a few classes so
that the GC can never see those objects in an invalid, partially-
initialized state.
AccessorInfo: Just zeros out a few fields upon construction. This is the
simplest case.
FunctionTemplateInfo: Slightly changes the order in which fields are
set, so that the Smi field is set ahead of the call to SetCallHandler,
which can GC. Also a pretty simple case.
JSListFormat, JSPluralRules, JSRelativeTimeFormat, JSSegmenter: The spec
requires that we start with OrdinaryCreateFromConstructor, which has
observable side effects (it fetches the prototype from the new.target).
So we split JSObject::New in half: the first half does all of the user-
visible things and returns a Map, which we can pass to the second half
when we're ready to actually allocate the object.
JSTypedArray: Extends the pattern from JSListFormat into Torque code:
start with a Map and don't allocate the object until we're ready to set
all of its properties.
Bug: v8:9311
Change-Id: Id7703e8a0727ec756c774cfbb56af787658a111a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1646844
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62123}
Make sure to use the callback passed to cleanupSome
Bug: v8:8179
Change-Id: Ia5d90b56edf80e05bdaf0dc520b555c29042b64c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655306
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62122}
I originally marked AllocationMemento::allocation_site as @noVerifier
out of an abundance of caution because AllocationMemento::IsValid checks
whether allocation_site is an AllocationSite. However, I can't find any
way for allocation_site to not be an AllocationSite.
Bug: v8:9311
Change-Id: I8b7ab51978c90d9baff42dace28d176a1cd2921a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1653520
Reviewed-by: Michael Stanton <mvstanton@chromium.org>
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#62120}
With bytecode flushing and lazy feedback allocation, we need to call
%PrepareForOptimization before we call %OptimizeFunctionOnNextCall,
ideally after declaring the function.
Bug: v8:8801, v8:8394, v8:9183
Change-Id: I3fb257282a30f6526a376a3afdedb44786320d34
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1648255
Commit-Queue: Mathias Bynens <mathias@chromium.org>
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62119}
This adds regexp-macro-assembler-arch.h which contains the arch-specific
include dispatch.
Change-Id: Ibc2be8059d54b57afeed9b7ce244229ce1bd79bc
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655296
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Auto-Submit: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Peter Marshall <petermarshall@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62118}
Make sure that we pop the async function promise from the catch
prediction stack when `await`ing inside an async function, and
push it back onto the stack when we're resuming execution. This
is to ensure that we don't leak memory when there are suspended
async functions while navigating away to a new page.
Bug: chromium:968603
Change-Id: I004715bc95d426958f1a89ce76c4856da1d4ceee
Cq-Include-Trybots: luci.chromium.try:linux-rel,win7-rel
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655652
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Yang Guo <yangguo@chromium.org>
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62117}
Previously the object cloning fast-path had a single loop which would
initialize the object _and_ at the same time clone MutableHeapNumbers.
But since that can trigger GCs, the heap verifier was a bit sad to see
double fields holding undefined values. This was flushed out by the CL
https://chromium-review.googlesource.com/1655291, which changed the GC
timing slightly and thus made the test crash in the verifier.
So instead of the one loop, we now have a second loop that takes care
of cloning any MutableHeapNumbers. This has the advantage that the first
loop can always run without write barriers.
Bug: chromium:964748, chromium:973045, v8:7611, v8:9114, v8:9183, v8:9343
Change-Id: I724a1c1e534243ce9ecde95bf0c07ca26363b515
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655307
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62114}
This is pre-work for a refactoring that changes how heap objects
are handled in the assembler: Currently, we put the handle location in
the constant pool, and replace these with the actual heap object when
we copy the code from the assembler's buffer to the heap.
In the future, we will put a small index in the constant pool, which
will ultimately enable 32bit constant pool slots for compressed heap
objects. This small index will be fixed up when we copy the code to
the heap.
This CL makes the assembler tests copy the code to the heap, which
ensures that the fix-up phase is actually run.
Change-Id: I80cd69dc57414a3bd0a27f8d558616aadcae05a2
Bug: v8:7703
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1647166
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62112}
As a component of the wider Turbofan logging scheme, it makes sense
for JSHeapBroker logging to come through flags specified in the
OptimizedCompilationInfo class, which uses --trace-turbo-filter
to control which functions are logged.
Bug: v8:7790
Change-Id: I3b068d8be78867ab0bd9607dda9eca4123b9d7b1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655297
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Commit-Queue: Michael Stanton <mvstanton@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62111}
API calls made via the CallApiCallback builtin, which is used from the
ICs and optimized code, are currently misattributed to the wrong counter
InvokeFunctionCallback instead of FunctionCallback. In addition we don't
use the C trampoline when only runtime call stats are enabled, but the
Chrome DevTools profiler is not active, which means that these calls
will not be attrituted properly at all, and that had to be worked around
using all kinds of tricks (i.e. disabling fast-paths in ICs when RCS is
active and not inlining calls/property accesses into optimized code
depending on the state of RCS).
All of this was really brittle and only due to the fact that the central
builtin didn't properly check for RCS (in addition to checking for the
CDT profiler). With this fix it's now handled in a central place and
attributed to the correct category, so user code doesn't need to worry
about RCS anymore and can just call straight into the fast-path.
Drive-by-fix: Do the same for AccessorInfo getter calls, which share the
core hand-written native code with the API callback logic.
Bug: v8:9183
Change-Id: Id0cd99d3dd676635fe3272b67cd76a19a9a9cea4
Cq-Include-Trybots: luci.chromium.try:linux-rel,win7-rel
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1651470
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62109}
The C++ standard says that template functions should be parsed immediately,
and only type-dependent things should be deferred.
cl.exe (MSVC's compiler) instead deferred parsing of all template functions
until the end of the translation unit, and unreferenced template functions
are not parsed at all. clang-cl emulates cl.exe's behavior.
Recently, cl.exe (and clang-cl) grew a /Zc:twoPhase flag that opts in to the
standards-conforming behavior, and system headers are now clean enough
to build with this flag set.
This cleans up v8 to also build with this flag. There was just a single issue:
RecyclingZoneAllocator() is unused and contains invalid code: It calls
the superclass ctor using `ZoneAllocator(nullptr, nullptr)`, when it should
be doing `ZoneAllocator<T>(nullptr, nullptr)`. With /Zc:twoPhase, this is
now a parsing error. However, since the RecyclingZoneAllocator() default
constructor isn't used anywhere, just delete it.
Finally, improve the comment for ZoneAllocator's default constructor to
explain why it's needed on Windows.
Bug: chromium:969702
Change-Id: I7a516afde67fe090a512d7c7214a3c6932754aca
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1652503
Auto-Submit: Nico Weber <thakis@chromium.org>
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62108}
... in favor of Isolate*. It seems that it's better to be uniform in
using Isolate* or isolate root value, so if we decide to pass isolate
root value instead of Isolate* it should better be done everywhere and
it will be a separate CL anyway.
Regarding the "optionality" of the isolate parameter - C++ compilers
are smart enough to optimize it away during inlining.
Bug: v8:9353
Change-Id: Idf86a792476f49393041ced1c54b8671f5b1794a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1653121
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62107}
https://chromium-review.googlesource.com/1649554 introduced a typo into
Runtime_CloneObjectIC_Miss, where it wouldn't update the IC state UNLESS
the source map is deprecated, which is the wrong way around of course.
Bug: chromium:973045, v8:7611, v8:9114, v8:9183, v8:9343
Change-Id: I7d6e0709e66ce4aaaf4a628d64ab801b84c8993c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1655291
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62106}
Currently, in wasm-function stack traces, v8 displays the decimal offset
from the start of the function. However, the WebAssembly WebAPI
specification says that it should be a hex offset into the module.
This change makes the stack trace display with hex module offsets, as
well as fixing all the unit tests that depended on the old behaviour.
R=fgm@chromium.org, titzer@chromium.org, yangguo@chromium.org
Bug: v8:9172
Change-Id: I73737a319a42dd665521ab8a4b825199ae11c87f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1646846
Reviewed-by: Ben Titzer <titzer@chromium.org>
Reviewed-by: Simon Zünd <szuend@chromium.org>
Reviewed-by: Adam Klein <adamk@chromium.org>
Commit-Queue: Guanzhong Chen <gzchen@google.com>
Cr-Commit-Position: refs/heads/master@{#62103}
It was a good flag, but it's time to say goodbye. Let us take a moment
to remember the good times we've had during its short time on earth.
It shipped in Chrome 74.
BUG=v8:8523
R=adamk@chromium.org, mathias@chromium.org, gsathya@chromium.org
Change-Id: I37e58360614c0bb3582b8bbfac795d5ed3e5a149
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1641205
Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
Reviewed-by: Adam Klein <adamk@chromium.org>
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Reviewed-by: Mathias Bynens <mathias@chromium.org>
Auto-Submit: Caitlin Potter <caitp@igalia.com>
Cr-Commit-Position: refs/heads/master@{#62099}
elements.cc invokes ApplyAttributesToDictionary using NumberDictionary as its template
parameter. But the declaration of the template method is in js-object.cc, so nobody
can actually compile the version for number dictionary. This is fixed requesting
explicit instantiation for NumberDictionary.
This was breaking GCC build.
Bug: chromium:819294
Change-Id: I685ddc5b97e129d1a534dbdb04025c0932bc5ecd
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1649565
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: José Dapena Paz <jose.dapena@lge.com>
Cr-Commit-Position: refs/heads/master@{#62097}
This helps debugging incorrect usage of the SnapshotCreator API in
debug mode.
Change-Id: Ibd9db76a5f460cdf7ea6d14e865592ebaf69aeef
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1648240
Reviewed-by: Yang Guo <yangguo@chromium.org>
Commit-Queue: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62095}
Changes calls to conversion builtins in Wasm wrappers to use the builtin
address directly rather than going via the Code object which results in
the trampoline being unnecessarily executed.
Bug: v8:9338
Change-Id: I74593dbda3c8bd308be65bc98ce85213a98f78b5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1648274
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Dan Elphick <delphick@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62094}
This CL adds lint errors for unused Torque macros. To prevent lots of
noisy warnings, the check is rather narrow. Macros declared as "extern"
or marked with "@export" are ignored. Also macros starting with "Convert",
"Cast" or "FromConstexpr" are not checked.
Drive-by: Removing some unused macros.
Bug: v8:7793
Change-Id: Ie0d2e445f8882a9b0ebbda45876b342abf341248
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1645312
Commit-Queue: Simon Zünd <szuend@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62092}
