Rolling v8/build: 15f3aed..aebd733
Rolling v8/buildtools: 734683a..18d22db
Rolling v8/buildtools/third_party/libc++/trunk: 84f0693..369cbf3
Rolling v8/third_party/depot_tools: 8cdc635..eb16430
Rolling v8/third_party/fuchsia-sdk/sdk: version:9.20220831.2.1..version:9.20220902.0.1
Rolling v8/tools/clang: e0c2881..0a22859R=v8-waterfall-sheriff@grotations.appspotmail.com,mtv-sf-v8-sheriff@grotations.appspotmail.com
Change-Id: I01a1df8f6fe1db74fad773480edb5801f2851219
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3872510
Commit-Queue: v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com>
Bot-Commit: v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com>
Cr-Commit-Position: refs/heads/main@{#82957}
This CL makes SelectGarbageCollector choose MinorMC to finalize
minor incremental marking if the allocation space is new space.
Bug: v8:13012
Change-Id: I0125d956842d83683cf3b1c5999de0e10a43dee5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3871074
Commit-Queue: Leon Bettscheider <bettscheider@google.com>
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82956}
This CL exclusively activates ScavengeTaskObserver, used by Scavenger
and atomic MinorMC, or MinorMCTaskObserver, used by concurrent
MinorMC.
Bug: v8:13012
Change-Id: I313b0cbe250828cc76d7d4b25e9abb457315112e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3871295
Commit-Queue: Leon Bettscheider <bettscheider@google.com>
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82955}
When internalizing cached external string that are not in-place
internalizable, there is an optimization that avoids copying the
contents of the string. This optimization doesn't work when the string
table is shared, as it leaves strings in a partly initialized state that
can be accessed in a concurrent environment.
Bug: v8:12007
Change-Id: I49fcbb232893c87d065af114546a6f1a15471016
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3870469
Auto-Submit: Patrick Thier <pthier@chromium.org>
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Shu-yu Guo <syg@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82953}
When aborting compaction on a page, slots need to be recorded in the
non-evacuated part. Now that maps might be recorded as well, slots
need to be recorded in the map word as well.
Bug: chromium:1359294, v8:12578
Change-Id: I91e4a98ae2951d814d4b50e527b34d8e54d55434
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3871297
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Auto-Submit: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82951}
DoubleToInteger, which corresponds to the ToIntegerOrInfinity AO in
ecma262, never returns -0. Currently there's a bug as std::ceil can
return -0.
Bug: v8:10271
Change-Id: Id5e7d040ef9d186462022dc96052d7920be6ebed
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3869196
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82950}
A minor refactoring to call a static method instead of just creating an
instance in order to do the work.
This also makes it easier to later add methods to process all feedback
before dumping profile information.
R=jkummerow@chromium.org
Bug: v8:13209
Change-Id: I157357753b3a1cff83df426660501c43959937b9
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3870468
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82949}
The `target` field of call feedback may be stored in the feedback
vector without protection by generated code (see
TryInitializeAsMonomorphic). We thus can't assume a memory fence
exists when creating the ref, switch to TryMakeRef instead.
Bug: v8:7790,v8:12876
Change-Id: I428b00b19a417e818c315f1cf9ee62d19f0747d7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3867728
Auto-Submit: Jakob Linke <jgruber@chromium.org>
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82947}
Add BitcastWordToTagged and BitcastWordToTagged to
CannotObserveStoreField in StoreStoreElimination to
allow more stores to be eliminated.
Also add the use properties in CannotObserveStoreField
so that future nodes don't block opportunities.
Improves Speedometer2 by 0.16% and 0.12% on a Cortex-A55
and a Neoverse-N1 machine respectively, with Preact seeing
+0.8% on both.
Change-Id: I0cfe887ec4c0c435dbb0f9d8dc3da4bfcdabc265
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865560
Reviewed-by: Patrick Thier <pthier@chromium.org>
Commit-Queue: George Wort <george.wort@arm.com>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82946}
Fixes the use of v8_enable_fast_mksnapshot on platforms that
aren't ia32 or x64. This was introduced by
https://chromium-review.googlesource.com/c/v8/v8/+/3863281
Change-Id: I0d18c2d16dae4937baa40ee336d8c246a8b77889
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3867408
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: George Wort <george.wort@arm.com>
Cr-Commit-Position: refs/heads/main@{#82945}
This makes it clear that the constructors are to be used for a basic block that starts a loop or a basic block that starts an exception handler.
Bug: v8:7700
Change-Id: Ic58dd80f223e45b4e5cf5da6e78e989ba18d55a3
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3870916
Auto-Submit: Victor Gomes <victorgomes@chromium.org>
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82944}
Move the logic to negate a flag properly if the name starts with '!' to
the general {FlagName} helper. This fixes an otherwise weird formatting.
Before:
# Contradictory flag implications from --!liftoff and --wasm-speculative-inlining for flag --wasm-dynamic-tiering.
After:
# Contradictory flag implications from --no-liftoff and --wasm-speculative-inlining for flag --wasm-dynamic-tiering.
R=tebbi@chromium.org
Change-Id: I21236b4ff338aa4d2ddd0872f85e2362ef8dc813
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3870915
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82943}
We used to scan the full heap of all clients for pointers into the
shared heap. Now that the OLD_TO_SHARED remembered set is considered
complete, we don't need to scan objects in the old generation anymore.
Instead we just need to check slots in the OLD_TO_SHARED remembered
set.
Since we don't have write barriers for objects guaranteed to be in
new space, we still need to scan objects in the young generation.
Bug: v8:11708
Change-Id: I1121f90ee63521c9141431f4cb31973796f1a67c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865561
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82941}
Add logic to drop cyclic contradictory flags from
correctness-fuzzing command lines. Add the currently known
biggest offenders.
Without this, the correctness fuzzing harness runs into a CHECK
failure during smoke testing, when attempting to pass cyclic flags
to d8. It fails fast, but uselessly burns fuzzing time.
This change drops one of the known cyclic flags instead to make the
test run still useful. The precedence is right to left like in the
V8 test framework.
Additionally on Clusterfuzz, all crashes during smoke testing are
deduped as one crash report. We don't know if there are other
problems before this one is fixed/hidden.
No-Try: true
Bug: chromium:1330303
Change-Id: I06cbb4655cd3cf467f5cce6f84dba653834ca72e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865562
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Reviewed-by: Alexander Schulze <alexschulze@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82939}
The intention is to be restrictive for now: modules should not
start to depend on this subtyping while the stringref type
hierarchy question is being settled (see
https://github.com/WebAssembly/stringref/issues/3 for details).
Bug: v8:12868
Change-Id: I0140e72f92550c88393dc84bb1fa3ce65840a048
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865019
Commit-Queue: Matthias Liedtke <mliedtke@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82937}
At the start of the graph builder, we add merge states to exception
handlers basic block with ExceptionPhis (normal phis with no input,
but with an interpreter register "owner").
Every Node that can throw, can also lazy deopt, so we use the
lazy deopt IFS to recover the exception phi values in a trampoline
before jumping to the exception catch block.
Bug: v8:7700
Change-Id: I62fe7f19ce5e89c3df645224ea62f9fc2798207c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865865
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Victor Gomes <victorgomes@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82935}
This change adds support for defining globals whose value is imported
and not defined inline. This was already possible for importing globals
from other modules, now it is also supported for non-global values, e.g.
values created by a wasm function and exported to JS.
Bug: v8:7748
Change-Id: I4fe22a7ab33b431cb731458900c0f332dff8b8f7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865554
Reviewed-by: Manos Koukoutos <manoskouk@chromium.org>
Commit-Queue: Matthias Liedtke <mliedtke@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82934}
Instead of implementing our own shared mutex, use std::shared_mutex,
which does not have the problem of deadlocking when interrupted by
signals (see https://crbug.com/v8/12037).
This is on Mac only, for now. If this fixes the regressions, we can
switch all systems to std::shared_mutex.
R=ishell@chromium.orgCC=dmercadier@chromium.org
Bug: v8:12037, v8:13256, chromium:1358856
Change-Id: Ie4be7cc5431905ca1e4f74809168eb6a9f584d28
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3870465
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82933}
ResumeGenerator is semantically a successor of SuspendGenerator (for
reasoning about liveness), but operationally it's a successor of
SwitchOnGeneratorState. This means that the jump to ResumeGenerator will
always create a new basic block, even if the SuspendGenerator was dead.
This causes problems if we made other assumptions on liveness based on
the semantics; in particular, we assume that JumpLoop is dead if the
loop header is dead (thanks to loop irreducibility).
SwitchOnGeneratorState breaks irreducibility, and this manifests as the
JumpLoop being alive and trying to jump to a dead header.
Since this is a special case, and loops are otherwise irreducible, we
can also fix it with a special case; namely, MarkBytecodeDead now has a
special case for SuspendGenerator which manually advances the iterator
and kills the ResumeGenerator.
Bug: v8:7700
Change-Id: Ice162f061e7ba1dda7ceb4f6fe9234889655b417
Fixed: v8:13250
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865556
Reviewed-by: Victor Gomes <victorgomes@chromium.org>
Commit-Queue: Victor Gomes <victorgomes@chromium.org>
Auto-Submit: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82931}
Check the function for TF code before finalizing a maglev compilation,
so that we don't accidentally overwrite the higher tier.
Bug: v8:7700
Change-Id: I20eb4e25f3bf2710b6e65f9d866cad143e77943d
Fixed: chromium:1359114
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3870464
Reviewed-by: Victor Gomes <victorgomes@chromium.org>
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Auto-Submit: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Victor Gomes <victorgomes@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82929}
This CL refactors most of the cctests and unittests (22 out of 31) that
directly invoke heap GC, so that the corresponding internal heap methods
are called from a few specific places in boilerplate code. This will
facilitate impending changes to the interface of GC-related internal
heap methods.
Bug: v8:13257
Change-Id: Ia6773a7952501b0792b279b799171519620497d9
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3869264
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Commit-Queue: Nikolaos Papaspyrou <nikolaos@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82927}
Use the correct helper function to emit an immediate.
Fixed: chromium:1358909
Change-Id: I2f2ae7819f40009b3f9c22067cdf11885d3347ca
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3869265
Reviewed-by: Matthias Liedtke <mliedtke@chromium.org>
Auto-Submit: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Matthias Liedtke <mliedtke@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82924}
This can save memory in cases where multiple frames in a process use the
same script, with sufficient time between loads that the script's
top-level SharedFunctionInfo is no longer present in the compilation
cache. Merging is relatively fast; it generally takes about one tenth as
long as deserialization.
Bug: v8:12808
Change-Id: I7366a51f1d2ca6a9f551cdf2bdbe0441450cf1bb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3868088
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82920}
base::SharedMutex was implemented as an exclusive lock on Mac, because
of an OS issue on Mac (see https://crbug.com/v8/12037).
https://crrev.com/c/3855361 then introduced a custom implementation on
Mac, which caused performance regressions (see
https://crbug.com/1358856).
Since we rely on C++17 now, we can instead just use {std::shared_mutex},
which does not seem to have the deadlock issue of {pthread_rwlock_t}.
As a smoke test (and to check whether this actually fixes the
performance regressions), only switch one mutex in Wasm compilation to
std::shared_mutex. If this CL looks good, then other places can be
switched over as well.
R=ishell@chromium.org
Bug: chromium:1358856, v8:13256
Change-Id: Ia56efcb7747f191cc3ed7a381096c8f57142aff8
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3868954
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82916}
.. just until I can implement the fix.
Bug: v8:7700,v8:13251
Change-Id: I8ccbe8b08351472a1144db46fd8d9bcbd4188633
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865919
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Jakob Linke <jgruber@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Auto-Submit: Jakob Linke <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82914}
This CL renames ConcurrentMarking::JobTask to JobTaskMajor, adds
JobTaskMinor, and makes ScheduleJob branch to schedule the respective
JobTask depending on its GarbageCollector parameter.
Bug: v8:13012
Change-Id: Ic7ab15ba70f7d4e86c94a6824623c258aa8b739c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3850482
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Leon Bettscheider <bettscheider@google.com>
Cr-Commit-Position: refs/heads/main@{#82913}
Fold the CheckHeapObject check into the other checks (CheckMaps,
CheckString, etc), to avoid emitting a separate IR node with separate
deopt info and a separate actual deopt point.
Allow this check to be elided when we already know the node is a heap
object.
Bug: v8:7700
Change-Id: I981860a6522c082d86abc856cfe1b3ff5658ac59
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3867733
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Victor Gomes <victorgomes@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82912}
Add a simple forward check elimination based on a side hashmap of "known
node aspects", namely the node type and node map (if any). This set of
aspects is cloned when merge states are created, and destructively
merged when merged into existing merge states -- destructive cloning
here means removing any mismatching information. This allows information
in dominators to be preserved.
Maps are kept separate from node types because we want to distinguish
between stable and unstable maps, where the former need a dependency and
the latter must be flushed across side-effecting calls.
The representation of this known information is currently very
inefficient, and won't win us any compilation speed prizes -- just
ZoneMaps keyed on ValueNode*. We should optimize this to take into
account some sort of liveness information, and clear out nodes that
aren't reachable anymore. There is also a lot more information we could
store per Node, e.g. known loaded fields or alternative representations;
depending on what we want to store and how that has to be invalidated,
we likely might need an alternative way of representing it. This
implementation is good enough for now though, for measuring the impact
of check elimination.
Bug: v8:7700
Change-Id: I2f001dedf8ab5d86f8acaa22416617bd80701982
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3865160
Auto-Submit: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Victor Gomes <victorgomes@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Jakob Linke <jgruber@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82911}
This CL removes the marking_worklists parameter from the
ConcurrentMarking constructor, and instead sets marking_worklists_
in ScheduleJob based on the new GarbageCollector parameter.
We will use the ConcurrentMarking class for both major and minor
marking later, and this CL does preparatory work for that by allowing
to change the mode of operation (minor/major) through ScheduleJob.
Bug: v8:13012
Change-Id: I44a35155cf19e1df139a6a4e5bc5cbedbc3e00aa
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3850289
Commit-Queue: Leon Bettscheider <bettscheider@google.com>
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82910}
