I recently updated the GC's bytecode flushing logic to accept the
possibility that two SharedFunctionInfos refer to the same
BytecodeArray (see https://crrev.com/c/v8/v8/+/3916272 ). However, that
change was insufficient for baseline code flushing. This change updates
the flushing logic to also check for already decompiled bytecode held by
the baseline code.
Bug: v8:13473
Change-Id: I58694727c417d608de3f8bf26dfd765fda1ae4f7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4018272
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Cr-Commit-Position: refs/heads/main@{#84254}
Reland of https://crrev.com/c/3998633.
Each thread has its own MarkingBarrier instance for incremental
marking. A thread local variable is used to get the current thread's
instance on background threads.
However on main threads this thread local variable was always
set to nullptr. The main thread would get to its own instance through
the heap_ field in the host object's page header. This was solved this
way because setting current_marking_barrier on the main thread
seemed quite complex. Multiple isolates may be run on the same thread
and isolates may even be migrated between threads.
However, with --shared-space loading the heap_ field for a shared
object would return the main isolate's heap and we end up with
the wrong MarkingBarrier instance on client isolates. So this
CL makes main and background threads more uniform by setting the
thread local field also on the main thread. The field is set by
the already existing v8::Isolate::Scope API. Some embedders might have
to add these scopes if they don't use them properly already.
Bug: v8:13267
Change-Id: Idc257ecf6b6af09a379bdd7cd7c1d4a5e46689c9
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4016715
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/main@{#84237}
This constant wasn't really used anymore anymore, so we can drop
this constant and initialize the --concurrent-marking flag from
V8_ATOMIC_OBJECT_FIELD_WRITES instead.
Bug: v8:13267
Change-Id: I2533d80cd08fd799441a7dba764fd392bb9151d2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4020428
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#84232}
This reverts commit 39d387bb72.
Reason for revert: causes DCHECKS, timeouts, TSAN failures under stress
Original change's description:
> [heap] Invoke allocation observers before limiting the LAB
>
> Currently whenever we reach a step we get a small LAB the same size as
> the allocated object. This is becuase the remaining step size is smaller
> then the current allocation.
> Invoking observers before limiting the LAB, and thus updating step
> sizes, should eliminate the small LAB we get whenever we reach a step.
>
> Drive-by: remove redundant method arguments.
>
> Bug: v8:12612
> Change-Id: Ied92a947308368d3b289e246fdb4f40ac5e5981f
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4013683
> Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
> Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
> Commit-Queue: Omer Katz <omerkatz@chromium.org>
> Cr-Commit-Position: refs/heads/main@{#84157}
Bug: v8:12612, v8:13465
Change-Id: I74b412dc41bfc86ea7fb190ce3284349a9022057
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4018913
Auto-Submit: Adam Klein <adamk@chromium.org>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Cr-Commit-Position: refs/heads/main@{#84162}
Currently whenever we reach a step we get a small LAB the same size as
the allocated object. This is becuase the remaining step size is smaller
then the current allocation.
Invoking observers before limiting the LAB, and thus updating step
sizes, should eliminate the small LAB we get whenever we reach a step.
Drive-by: remove redundant method arguments.
Bug: v8:12612
Change-Id: Ied92a947308368d3b289e246fdb4f40ac5e5981f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4013683
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#84157}
This reverts commit 910def9edc.
Reason for revert: Speculative Revert https://ci.chromium.org/ui/p/v8/builders/ci/V8%20Linux64%20TSAN%20-%20stress-incremental-marking/9800/overview
Original change's description:
> [heap] Load MarkingBarrier from thread local on main thread
>
> Each thread has its own MarkingBarrier instance for incremental
> marking. A thread local variable is used to get the current thread's
> instance on background threads.
>
> However on main threads this thread local variable was always
> set to nullptr. The main thread would get to its own instance through
> the heap_ field in the host object's page header. This was solved this
> way because setting current_marking_barrier on the main thread
> seemed quite complex. Multiple isolates may be run on the same thread
> and isolates may even be migrated between threads.
>
> However, with --shared-space loading the heap_ field for a shared
> object would return the main isolate's heap and we end up with
> the wrong MarkingBarrier instance on client isolates. So this
> CL makes main and background threads more uniform by setting the
> thread local field also on the main thread. The field is set by
> the already existing v8::Isolate::Scope API. Some embedders might have
> to add these scopes if they don't use them properly already.
>
> Bug: v8:13267
> Change-Id: Idfdaf35073d04dd5e13ad6065ef42eae3ce6a259
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3998633
> Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
> Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
> Cr-Commit-Position: refs/heads/main@{#84144}
Bug: v8:13267
Change-Id: Id8493dfac03d789721ca30cd29b0dd4b67006881
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4017192
Owners-Override: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/main@{#84151}
Each thread has its own MarkingBarrier instance for incremental
marking. A thread local variable is used to get the current thread's
instance on background threads.
However on main threads this thread local variable was always
set to nullptr. The main thread would get to its own instance through
the heap_ field in the host object's page header. This was solved this
way because setting current_marking_barrier on the main thread
seemed quite complex. Multiple isolates may be run on the same thread
and isolates may even be migrated between threads.
However, with --shared-space loading the heap_ field for a shared
object would return the main isolate's heap and we end up with
the wrong MarkingBarrier instance on client isolates. So this
CL makes main and background threads more uniform by setting the
thread local field also on the main thread. The field is set by
the already existing v8::Isolate::Scope API. Some embedders might have
to add these scopes if they don't use them properly already.
Bug: v8:13267
Change-Id: Idfdaf35073d04dd5e13ad6065ef42eae3ce6a259
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3998633
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#84144}
Restructure such that we only either shrink or grow the space, and only
at the end of the GC.
Bug: v8:12612
Change-Id: I7671e139b6023cd04f438f48401cf392d4549b8e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3980253
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#84076}
This CL introduces the mechanism for conservative stack scanning (CSS).
Behind a compile-time flag, it also introduces a CSS pass which scans
the stack during the GC marking phase and marking verification. This
pass is now redundant, i.e., it is not needed for the correctness of
garbage collection. It will be used for experimenting with CSS and for
benchmarking.
Bug: v8:13257
Change-Id: If35bc24fde3bc08c5735d9e2f1b67724f7e31ef7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3968710
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Commit-Queue: Nikolaos Papaspyrou <nikolaos@chromium.org>
Cr-Commit-Position: refs/heads/main@{#84036}
Since crrev.com/c/3973310 which renamed the isolate scopes, the unit and
cctests for the object-start bitmap and the conservative stack visitor
have broken.
Bug: v8:13257
Change-Id: If8a498827f2085108cf0740a9c5c994145424fc3
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3980255
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Commit-Queue: Nikolaos Papaspyrou <nikolaos@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83960}
Update the free list implementation for new space to set a larger
minimum size and skip redundant step in the allocation logic.
Bug: v8:12612
Change-Id: I480fe99cf4cfad7c25d687540b7841cd56d41d47
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3976508
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83920}
This CL simplifies safepoint scopes, there are now three kinds of
safepoint scopes:
1) IsolateSafepointScope - performs an isolate local safepoint
2) GlobalSafepointScope - a global safepoint across multiple isolates
3) SafepointScope - chooses based on condition between local/global
This CL is not supposed to change current safepointing behavior in
any way. The CL renames the current SafepointScope to
IsolateSafepointScope and changes GlobalSafepointScope to always
perform a global safepoint. It then also introduces the new
SafepointScope and makes use of it for snapshotting and in heap.cc.
Bug: v8:13267
Change-Id: Ie7e1f81b6158c98d3d98552ba735cc73c9b869c5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3973310
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83912}
The implementation of bytecode flushing assumes that there is a single
owning SharedFunctionInfo for each bytecode. However, sometimes there
can be two, because a couple of code paths copy content from one
SharedFunctionInfo to another (BackgroundCompileTask::FinalizeFunction
and BackgroundMergeTask::CompleteMergeInForeground). Usually this works
out okay in practice, because we only copy content from a
SharedFunctionInfo when we're about to abandon all references to it.
However, we shouldn't rely on this lucky timing, especially considering
a possible future where conservative stack scanning could retain the
copied-from SharedFunctionInfo for an arbitrarily long time due to
spurious stack references. This change updates the bytecode flushing
implementation to correctly handle the case of two SharedFunctionInfos
that point to the same BytecodeArray.
I don't know if this fixes the linked bug, but so far it's the only
semi-plausible explanation I've found.
Bug: chromium:1359773
Change-Id: Iaa2c6e4953afcb46df2ac4b17828271151d85e59
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3916272
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83728}
This CL aligns minor sweeping with full sweeping such that no
sweeping happens in the atomic pause. If concurrent sweeping is
disabled, sweeping will be either on allocation or when we start
the next GC.
This CL doesn't yet enable concurrent sweeping for MinorMC.
Regressions on benchmarks with MinorMC enabled are expected.
Bug: v8:12612
Change-Id: I6df808b94783559122ee78306af932809860703d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3934769
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83712}
MinorMC cannot shrink the space while concurrent sweeping is active.
This results in races between sweeping empty pages concurrently and
releasing them on the main thread.
Without concurrent sweeping, MinorMC sweeps all empty pages, then
preallocates more pages if needed, then shrinks by releasing some of
those empty/new pages.
MinorMC can instead shrink during sweeping/evacuation in the atomic
pause, before concurrent sweeping is active.
This resolves the conflict between concurrent sweeping and shrinking and
generally reduces the amount of work the GC has to do (i.e. don't
sweep/preallocate pages that will then be released by shrinking).
Bug: v8:12612
Change-Id: If898e5111c65b0b11d6ee9ee5dcd45751024650b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3933056
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83638}
This is needed to simplify concurrent sweeping for MinorMC.
Also: move evacuation verifiers to a separate file so that they can be
used from heap.cc as well.
Bug: v8:12612
Change-Id: I2a738a31e83a357f4fdded8a30ccb2ff6ba70553
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3932720
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83557}
Detach the marking state from the collectors and move them to heap.
Also update users to access via heap and reduce dependencies on the
collectors.
This is a prerequisite for moving sweeper to the heap, which is needed
for concurrent sweeping in MinorMC.
Bug: v8:12612
Change-Id: Ia0bb2b7566b24eeb0d75c911edbfd626f07dad0f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3925548
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83542}
This CL moves the code for requesting a GC from a non-main thread from
LocalHeap to Heap into CollectGarbageBackground().
The CL then makes use of this method in CollectGarbageShared() to
request a GC with --shared-space.
Bug: v8:13267
Change-Id: I2946cf5068ef8eb9eb99f9d396ac466d68abc7ec
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3916634
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83456}
This CL implements collection of garbage in the shared spaces in
the shared heap isolate. GC on such an isolate should now work
correctly without worker isolates. Support for worker isolates will
be implemented in a subsequent CL.
Bug: v8:13267
Change-Id: I30125ce3b791e2faa0504d065f23639d6106e6b6
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3904647
Reviewed-by: Jakob Linke <jgruber@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83371}
SimulateFullSpace starts with no LAB, iterates over pages and allocates
all free space on each page. After the first page, the LAB is empty but
is no longer null.
Bug: v8:12612
Change-Id: I2c00b9ba68fdd5f60eda086ea940cb6e211a986e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3891294
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83186}
FillCurrentPage assumed that everything after top is empty, which
doesn't work with MinorMC and sweeping. Revise FillCurrentPage based
SimulateFullSpace for MinorMC.
I similar implementation is provided both in unittests and cctest.
Migrating affected cctest to unittests is left a future work.
Bug: v8:12612
Change-Id: Ie29be2fc7aaee25e1fd5f66b1c0959c2a45f007f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3885888
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83158}
ShouldFinalize should only be called if major incremental marking is
active, and can crash if minor incremental marking is active, if
MajorMC's local_marking_worklists_ was reset.
The only caller is IsMarkingComplete. This CL changes the IsMarking
check to IsMajorMarking to solve this issue, and renames
IsMarkingComplete to IsMajorMarkingComplete.
Bug: v8:13012
Change-Id: Iba6bd5b7977ec8566c3ab0f047646d8cafd45038
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3879485
Commit-Queue: Leon Bettscheider <bettscheider@google.com>
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#83064}
Avoid the deprecated FLAG_* syntax, access flag values via the
{v8_flags} struct instead.
R=mlippautz@chromium.org
Bug: v8:12887
Change-Id: Id12f9d8270dd9fed651e1b3596c06e45aae9d6d1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3846151
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Auto-Submit: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82969}
This CL adds PKU support for V8 heap, but we will not enable
PKU by default before adding bots that are able to test the
PKU machinery.
Bug: v8:13023
Change-Id: I0465604d56900536ad63311f119ea0324ebe4f2f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3793944
Reviewed-by: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Wenqin Yang <wenqin.yang@intel.com>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82965}
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}
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}
Methods are now defined in heap-verifier.h in the HeapVerifier class.
Bug: v8:11708
Change-Id: I13e7f1760598f3659ad6aa31082840caf2e44038
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3857558
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Jakob Linke <jgruber@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82810}
Currently the ability to create shared isolates is partially exposed to
API. Instead of fully exposing it, this CL makes shared isolate and
shared heap handling transparent to the embedder.
If a flag that requires the shared heap is true (currently
--shared-string-table and --harmony-struct), the first isolate created
in the process will create and attach to a process-wide shared isolate.
Subsequent isolates will attach to that shared isolate. When that first isolate is deleted, the shared isolate is also deleted.
Bug: v8:12547
Change-Id: Idaf2947bc354066c44f2d10243e10162b1b7e4d6
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3848825
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Owners-Override: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82756}
This CL adds a soft limit (via AllocationObserver) to run
incremental marking for MinorMC.
Once the soft limit is triggered, roots are marked.
This a stepping stone for concurrent marking
(YoungGenerationConcurrentMarkingVisitor, go/YGCMV) integration.
Bug: v8:13012
Change-Id: I5bc9aeb80511159561845deb494023ade3fb7365
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3824339
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Leon Bettscheider <bettscheider@google.com>
Cr-Commit-Position: refs/heads/main@{#82695}
Concurrent markers could add work into the worklist before the CHECK.
Bug: v8:12775, v8:13223
Change-Id: I8ac252b0fec8e5acbcfec56dad04830e596c709d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3854496
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82690}
Users should rely on CppHeap which is the only supported way of using
v8::TracedReference in going forward.
Bug: v8:13207
Change-Id: Idd03f458167c74b06f285bb568e5c77ad46003fe
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3849037
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Simon Zünd <szuend@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82681}
IsRunning() and IsMarking() are now equivalent. So IsRunning() can be
removed in favor of IsMarking().
IsComplete() is also renamed to IsMarkingComplete().
Bug: v8:12775
Change-Id: Ife88be4d674af055590ba5178ec1e410f8fa89d9
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3849833
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82665}
- Swap() was not necessary as all uses merely required Merge()
- Remove unused empty Local ctor
- Use refrence for backref as it's always supposed to be non-null
Bug: v8:13193
Change-Id: Ide0a0de15185a67d028890371ae30528fd55a058
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3846863
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Anton Bikineev <bikineev@chromium.org>
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82648}
This is a reland of commit 9b0d5cb14b
The newly added check does not allow comparisons with stale or
invalid pointers because attempt to access the page header might
crash.
Original change's description:
> [ext-code-space] Fix Code vs non-Code comparisons
>
> When external code space is enabled comparing Code and non-Code objects
> by looking at compressed values is not always correct. Such an approach
> works only for comparing Code vs Code objects or non-Code vs non-Code
> objects.
>
> This CL instroduces SLOW_DCHECK into Object comparison operators to
> ensure that such a comparison is allowed. Also, this CL instroduces
> an Object::SafeEquals() method which compares uncompressed values
> and thus is safe to be used for comparing Code with non-Code objects.
>
> Bug: v8:11880
> Change-Id: I7ccf1f90f927beb2bb9f45efb303e902b1838d02
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3838172
> Reviewed-by: Jakob Linke <jgruber@chromium.org>
> Reviewed-by: Camillo Bruni <cbruni@chromium.org>
> Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
> Commit-Queue: Igor Sheludko <ishell@chromium.org>
> Cr-Commit-Position: refs/heads/main@{#82611}
Bug: v8:11880
Change-Id: Iab3c8fe49cb954b2dc9171b3fc4b189e84763e73
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3842932
Reviewed-by: Jakob Linke <jgruber@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82624}
This reverts commit 9b0d5cb14b.
Reason for revert: Seems to fail on gc-stress bots (e.g. https://ci.chromium.org/ui/p/v8/builders/ci/V8%20Linux64%20GC%20Stress%20-%20custom%20snapshot/43472/overview)
Original change's description:
> [ext-code-space] Fix Code vs non-Code comparisons
>
> When external code space is enabled comparing Code and non-Code objects
> by looking at compressed values is not always correct. Such an approach
> works only for comparing Code vs Code objects or non-Code vs non-Code
> objects.
>
> This CL instroduces SLOW_DCHECK into Object comparison operators to
> ensure that such a comparison is allowed. Also, this CL instroduces
> an Object::SafeEquals() method which compares uncompressed values
> and thus is safe to be used for comparing Code with non-Code objects.
>
> Bug: v8:11880
> Change-Id: I7ccf1f90f927beb2bb9f45efb303e902b1838d02
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3838172
> Reviewed-by: Jakob Linke <jgruber@chromium.org>
> Reviewed-by: Camillo Bruni <cbruni@chromium.org>
> Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
> Commit-Queue: Igor Sheludko <ishell@chromium.org>
> Cr-Commit-Position: refs/heads/main@{#82611}
Bug: v8:11880
Change-Id: Ie34af0135625eff2975f78f4d2901a76b8517eb7
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3842930
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Owners-Override: Leszek Swirski <leszeks@chromium.org>
Auto-Submit: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82616}
When external code space is enabled comparing Code and non-Code objects
by looking at compressed values is not always correct. Such an approach
works only for comparing Code vs Code objects or non-Code vs non-Code
objects.
This CL instroduces SLOW_DCHECK into Object comparison operators to
ensure that such a comparison is allowed. Also, this CL instroduces
an Object::SafeEquals() method which compares uncompressed values
and thus is safe to be used for comparing Code with non-Code objects.
Bug: v8:11880
Change-Id: I7ccf1f90f927beb2bb9f45efb303e902b1838d02
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3838172
Reviewed-by: Jakob Linke <jgruber@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82611}
This CL removes the COMPLETE state from incremental marking. Since
then the only states left were STOPPED and MARKING, we can replace
the state with an is_running_ boolean field.
The state could change back-and-forth between MARKING and COMPLETE.
IsMarking() was already also checking for COMPLETE. So most code
already treated both states the same. IsComplete() now checks whether
marking is running and a transitive closure was reached already.
IncrementalMarking::Step() didn't process the marking queue when in
COMPLETE. This should be relatively rare though since it only
transitioned into COMPLETE when the stack guard was armed and the
allocation observer ran again before reaching a stack guard check.
Bug: v8:12775
Change-Id: Ied48d8c512ad3d1b3d2e29393d43b434b5fda8fe
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3835689
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82581}
Introducing IncrementalMarking::AdvanceForTesting as last bottleneck
for driving incremental marking in addition to AdvanceFromTask
and AdvanceOnAllocation.
Now that we have those 3 bottlenecks, Step() and AdvanceWithDeadline()
can become private methods in IncrementalMarking. We also don't need
the StepResult return value in Step() anymore, which allows us to
remove CombineStepResult.
Bug: v8:12775
Change-Id: I702714439ef7ea4b9abf2156387503d4d00a7a48
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3823131
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82552}
This is a reland of commit a19316d9d7
- Revert malloc_usable_size() changes temporarily to land them in
isolation.
- Add cosmetics from https://crrev.com/c/3827876
Original change's description:
> [heap] Rework Worklist base type
>
> Worklist uses a singly-linked list of segments to hold entries.
> Segment size was based on a compile-time constant but already stored
> in the segment itself.
>
> Rework the segments to query `malloc_usable_size()` on allocation and
> adjust the capacity properly. For PartitionAlloc, it turns out that
> there's ~20% more capacity available for the 64-element segments.
>
> This slows down actual allocation of the segments with the upside of
> improving utilization and requiring 20% less segments.
>
> Change-Id: Ib8595c3fb9fb75b02e4022f6c525bb59a2df7ab7
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3826047
> Commit-Queue: Anton Bikineev <bikineev@chromium.org>
> Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
> Reviewed-by: Anton Bikineev <bikineev@chromium.org>
> Cr-Commit-Position: refs/heads/main@{#82432}
Change-Id: Ic8c5257cfe3c347b11eea5c513ca7f62e09f637f
Bug: v8:13193
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3829475
Reviewed-by: Anton Bikineev <bikineev@chromium.org>
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82493}
This reverts commit a19316d9d7.
Reason for revert: https://ci.chromium.org/ui/p/v8/builders/ci/V8%20Linux64%20UBSan/22670/overview
Original change's description:
> [heap] Rework Worklist base type
>
> Worklist uses a singly-linked list of segments to hold entries.
> Segment size was based on a compile-time constant but already stored
> in the segment itself.
>
> Rework the segments to query `malloc_usable_size()` on allocation and
> adjust the capacity properly. For PartitionAlloc, it turns out that
> there's ~20% more capacity available for the 64-element segments.
>
> This slows down actual allocation of the segments with the upside of
> improving utilization and requiring 20% less segments.
>
> Change-Id: Ib8595c3fb9fb75b02e4022f6c525bb59a2df7ab7
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3826047
> Commit-Queue: Anton Bikineev <bikineev@chromium.org>
> Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
> Reviewed-by: Anton Bikineev <bikineev@chromium.org>
> Cr-Commit-Position: refs/heads/main@{#82432}
Change-Id: I14994e11ff5ffaba70b93d977d40dd2f6e9e5d35
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3829474
Owners-Override: Nico Hartmann <nicohartmann@chromium.org>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Auto-Submit: Nico Hartmann <nicohartmann@chromium.org>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Cr-Commit-Position: refs/heads/main@{#82438}
The existing version for paged spaces simply reset the freelist, which
doesn't work for tests that require actual objects in the space.
The version for new space also doesn't work because it assumes
everything after top is free space.
Fill the space with FixedArray by iterating over the freelist and
creating an object in place of each freelist entry.
This method actually fills the space, so that we can also use it to
force page promotion.
Bug: v8:12612
Change-Id: Ie0d73e846bbf688ea52030be29e0587b2f37ed4e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3823135
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82437}
Worklist uses a singly-linked list of segments to hold entries.
Segment size was based on a compile-time constant but already stored
in the segment itself.
Rework the segments to query `malloc_usable_size()` on allocation and
adjust the capacity properly. For PartitionAlloc, it turns out that
there's ~20% more capacity available for the 64-element segments.
This slows down actual allocation of the segments with the upside of
improving utilization and requiring 20% less segments.
Change-Id: Ib8595c3fb9fb75b02e4022f6c525bb59a2df7ab7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3826047
Commit-Queue: Anton Bikineev <bikineev@chromium.org>
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Anton Bikineev <bikineev@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82432}
This CL includes:
1) Using 1 full GC instead of 2 young GCs to force object promotion.
2) A couple of needed bailouts.
3) Using manual evacuation candidates in old space to ensure an object
is evacuted (moved to a different address) instead of relying on
Scavenger.
And some other minor tweaks
Bug: v8:12612
Change-Id: Idfd925ccdf30215998ab6e7cc632ce750fa2077a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3816661
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82327}
