Introduce OffThreadFactory with initial string construction support.
The OffThreadFactory shares with Factory a new CRTP base class, called
FactoryBase. Methods in FactoryBase return a FactoryHandle<Factory, T>
alias, which is Handle<T> for normal Factory and a new OffThreadHandle<T>
for OffThreadFactory. OffThreadHandle<T> behaves like Handle<T>, except
it stores the object in-line rather than needing external storage.
Any shared factory methods are moved into FactoryBase, which uses CRTP
to call the sub-class's AllocateRaw method (plus a few more customization
points which need Isolate access on the main thread).
Methods that used to take an Isolate or Factory, and are needed off the
main thread, are now expected to be templated on the factory type and
to use the appropriate handle.
Once an OffThreadFactory has finished being used (e.g. off-thread
compilation completed) its pages are "Published" into the main-thread
Heap. To deal with string internalization without creating a bunch of
ThinStrings, this is done in two stages:
1. 'FinishOffThread': The off-thread pages are walked to
collect all slots pointing to "internalized" strings. After this is
called it is invalid to allocate any more objects with the factory.
2. 'Publish': On the main thread, we transform these slots into
<Handle to holder, offset> pairs, then for each saved slot
re-internalize its string and update the slot to point to the
internalized string.
Bug: chromium:1011762
Change-Id: I008a694da3c357de34362bd86fe7e1f46b535d5e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1992434
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65787}
The DCHECKs ensured that all on-stack handles removed when the embedder
notifies V8 of an empty stack are indeed below the current stack limit.
This is brittle, as the calls that are guaranteed to have no stack
above, e.g., non-nestable tasks executing GC, sometimes have larger
stack depth then previously registered on-stack handles. Resetting the
slot to avoid UAF is not possible/needed as it is guaranteed in such
cases that the stack is indeed different from the stack that was used
when registering an on-stack handle.
This CL removes the DCHECKs and trust the embedder on such calls,
similar to when the embedder tells V8 that there's no interesting C++
stack on top of a call to avoid conservative stack scanning.
Bug: chromium:1040038
Change-Id: I2e8c77d8080f2d888f773984646998bede59e19c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000753
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65786}
When reserving the requested virtual memory fails (due to address space
exhaustion), simply return nullptr to indicate allocation failure, which
callers must be prepared to handle anyway. That way, ClusterFuzz will
correctly classify OOM situations.
Bonus change: skip demo test on simulators to save time.
Drive-by cleanup: add a 'simulator_run' section to mjsunit.status
Bug: chromium:1042151,chromium:1042173
Change-Id: I8569f3c0d2a681fbf6f91b665dcb88a4ac3b901e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2002391
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65785}
This moves first parts of the wrapper generation to the GraphAssembler.
We should migrate more code in follow-up CLs, and think about also
computing the schedule in the GraphAssembler (once everything is
migrated).
This also removes the only uses of the controversial {HalfDiamond}
construct, hence this is also removed in this CL.
Plus a bug fix in the GraphAssembler::Call method, and a new method
in GraphAssembler to load heap number values.
R=jkummerow@chromium.org, tebbi@chromium.org
Bug: v8:10123
Change-Id: Iac4661cdd50049cb73a2f305e280c1af6200729a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000756
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65784}
This CL adds optimizations for Word64And, Word64Or and Word64Xor
to the MachineOperatorReducer. Some of these (esp. constant folding)
have previously been removed from CodeAssembler to streamline
the optimization pipeline.
Bug: v8:10021
Change-Id: I679f0b60589a84b2d92ca6d9083efaddfe0b6423
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1997131
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65781}
Some architectures used {kConstantStackSpace}, others used
{kInstanceOffset}. This CL unifies it to {kInstanceOffset} and uses that
constant consistently (in {GetInstanceOperand}).
R=zhin@chromium.org
Bug: v8:10019
Change-Id: Ia2b6908e289591e2dbc48e559e11407877b7c4ad
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000146
Reviewed-by: Zhi An Ng <zhin@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65780}
This CL attempts to fix a chrome crash seen in the wild. Without a
reproducer, the current working theory is that we hit a 'null' context
in some edge case, causing us to access an empty handle. This CL
prevents the empty context handle to be dereferenced.
TBR=yangguo@chromium.org
Bug: chromium:1038747
Change-Id: Icd6f4853a22ddbf1e504f0f0f90c065b3437f8ab
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000752
Reviewed-by: Simon Zünd <szuend@chromium.org>
Commit-Queue: Simon Zünd <szuend@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65776}
Without the type check, Code() may read OOB. Note that this is an
internal, test-only runtime function.
Bug: chromium:1041316
Change-Id: I8c0b21ce3c2aea8aa3d065b99d8ab45a8c9e754f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000749
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Auto-Submit: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65775}
Native C++ arrays cannot have size 0 and thus need a dummy element
when filled with variadic template args. std::array does not have this
limitation and makes related code easier to read.
Bug: v8:9972
Change-Id: I70304b55525bd67d966fa69c663a71c202245d14
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000751
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Michael Stanton <mvstanton@chromium.org>
Auto-Submit: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Stanton <mvstanton@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65774}
With --stress-opt, the second run will share the NativeModule with the
first run, hence it's in a nondeterministic state and the test
expectations fail.
TBR=ahaas@chromium.org
CC=duongn@microsoft.com
No-Try: true
Bug: v8:10086, v8:9654
Change-Id: I74cf5e841ae2330b3b846ee742cc022305ec9636
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000750
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65773}
blendvpd should not be defined in the macro list, since the AVX version
has 4 operands, not 3.
Change-Id: Id020b460fa1a3510a91490f3b2286024cc6c5994
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1990139
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65771}
Add kWasmS128 to the list of supported types, and implement Fill for all
the architectures so that LocalGet works.
Add a new test file to contain tests that run only on Liftoff, and
assert that the code is indeed compiled by Liftoff.
We cannot rely on the nooptimization variant for testing
because by default, if Liftoff compilation fails, it will fall back to
Turbofan, and we accidentally get a test passing.
We skip these tests on mips architecture that don't support SIMD, since
there is no way to implement these, and we don't have a "lowering" phase
for Liftoff.
As we implement more of SIMD in Liftoff, we can add more
tests to this file and ensure correctness. Future patches will introduce
support for globals and params.
Bug: v8:9909
Change-Id: I7fc911f2d588d60c709ddb258b2efc1f22805fab
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1999470
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65768}
I also fixed one issue in the wasm interpreter.
R=clemensb@chromium.org
Bug: v8:10180
Change-Id: Ie30e908ad051a27fa611e8d36134b67aaf4c830c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000741
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65764}
Reuse logic in {CompileNativeModule} function in module-compiler.cc:
initialize parallel compile jobs, then wait for them to finish while
taking part in this compilation.
Bug: v8:9654
Change-Id: I9974d9f8b516e9faec716a592c7c0ee9c7077d8e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1977041
Commit-Queue: Z Nguyen-Huu <duongn@microsoft.com>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65763}
This Tnodifies the CombineFeedback and OverwriteFeedback methods and
changes the TaggedToWord32OrBigInt* ann TaggedToNumeric methods to take
TVariables.
Additionally it refactors bitwise binary operators in
intepreter-generator.cc and builtins-number-gen.cc and puts the common
code in NumberBuiltinsAssembler.
Bug: v8:10021
Change-Id: I3b15ecfadb42b50ffbfd0bd1114197e0fef42e99
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1995387
Commit-Queue: Dan Elphick <delphick@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65762}
The DCHECK was not correct in pointer compression mode.
Change-Id: Ifc00478df10962a8114f2d9cd1596ddaedc60d97
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000742
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65761}
This just removes the flag if it is not supported anyway. This avoids
fuzzers trapping over this.
The same was done for the --perf-prof flag in
https://crrev.com/c/1993969.
R=ahaas@chromium.org
Bug: chromium:1035233
Change-Id: I7b4b8fdd141df717cc62d795534f30435f7b38c1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1998083
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65760}
Tests which set the --perf-prof flag leave behind a file in the current
working directory every time they execute.
In order to avoid this, this CL introduces a --perf-prof-delete-file
flag, which removes this file right after creating it. This still allows
the process to write to it via the open handle, but the file will be
gone afterwards, even if the process crashes or gets killed while
executing.
R=ahaas@chromium.org
Bug: v8:10121
Change-Id: I99b159bb6d94255f77095ac78d98ba55106e94fc
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000738
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65759}
- Introduces a API to set top of the stack through
EmbedderHeapTracer::SetStackTop.
- Introduces a new API to inform V8 about an empty embedder stack.
- Switch internal representation of TracedReference
for on-stack handles to a proper stack that considers all
contained handles as roots.
- Handle garbage is avoided by cleaning up on handle creation or
GC.
Design doc: https://bit.ly/on-stack-traced-reference
Bug: chromium:1040038
Change-Id: I927ef0abb268fdb5853c9e17b1bc96e2491cf101
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1993973
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65757}
There is no particular reason that PropertyDescriptorObject should be a
subclass of FixedArray. By using a separate struct type, we get better
generated accessor functions, automatic verification, and runtime type
info, plus we save four bytes per instance.
Change-Id: If076782832aa9398806794e4ee6d019aea2f92b7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1999463
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#65756}
Currently the on-hold area is reset on incremental marking steps.
At the start of marking we conservatively assume that the on-hold area
spans the initially allocated linear allocation area, which may be large.
Bug: chromium:973627
Change-Id: I83f2d0e38a2a255c1e8d48549352e9303be89920
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000737
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65755}
The fix in https://crrev.com/c/1997135 didn't properly recurse the cache
scope after a with scope, passing the current scope rather than the
original cache scope up the recursion. Now the "use external cache" check
is done in LookupWith (and, analogously, LookupSloppyEval) while passing
the given cache scope through the Lookup recursion.
Fixed: chromium:1041210
Fixed: chromium:1041616
Change-Id: I5ac9ddc6c16d63b59aa034721fccec2f7781c4f8
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000133
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Auto-Submit: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65754}
Just a cleanup to remove an unused enum marker.
R=ahaas@chromium.org
Change-Id: I7b40f2389796f43d82d06eb161569ecea7892ef6
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000145
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65753}
This fixes local names that are the string representation of a valid
element index.
Even though both the Liftoff and the interpreter-based test are changed,
only the latter needs to be fixed right now since Liftoff does not use
the names currently. Modifying the test just ensures that we implement
this correctly once we use the name.
R=jkummerow@chromium.org
Bug: v8:10019
Change-Id: Ib7f7d6e244a344a85ab540b6c2c67f98b1f3078e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1998079
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65751}
TSan complains in "isolates" tests otherwise. Also further reduce
virtual memory requirements of the sample test to address flaky
allocation failures on 32-bit platforms.
Change-Id: I26c9a59965009d7083876b4ff4836ee879d33350
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000138
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65750}
This saves the addition when accessing the stack slot, and (more
importantly) will make it easier to access the stack slot for debugging,
since there is no platform-specific constant to be added any more.
R=zhin@chromium.org
Bug: v8:10019
Change-Id: I3eaf1838b78c2b7b343a435d7c8a32e7e71508ed
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1998082
Reviewed-by: Zhi An Ng <zhin@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65749}
It was missing the name field.
Bug: chromium:1036641
Change-Id: I686a46adfccfd656422cddef340ded58b7a9c9b1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000135
Commit-Queue: Georg Neis <neis@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65748}
This reverts commit 304e97d334.
Reason for revert: Last roll is failing - https://ci.chromium.org/p/chromium/builders/try/linux-rel/282356
Original change's description:
> [parser] Fix caching dynamic vars on wrong scope
>
> When looking up a variable in a deserialized WITH scope, we were
> unconditionally passing in the cache scope to the lookup, even if the
> with was inside the cache scope. This would lead to and outer scope of
> the with holding the generated dynamic variable. If the cache scope was
> the SCRIPT scope, the dynamic variable would be interpreted as a global
> object property.
>
> Now, we only store the WITH scope dynamic variables in the cache scope
> if it is an inner scope of the WITH scope, same as we do for 'normal'
> scope lookups.
>
> Fixed: chromium:1041210
> Change-Id: I4e8eb25bbb8ea58311355d13a9c7c97bf2fa3ec7
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1997135
> Reviewed-by: Toon Verwaest <verwaest@chromium.org>
> Commit-Queue: Toon Verwaest <verwaest@chromium.org>
> Auto-Submit: Leszek Swirski <leszeks@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#65732}
TBR=leszeks@chromium.org,verwaest@chromium.org
Change-Id: I7b6d77d03b603152a9a47541db466934f46b1176
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000140
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Commit-Queue: Maya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65747}
This CL fixes a negative capacity check up the call-chain to fix an
issue found a clusterfuzz. This is temporary since mid-term we want to
change the interface of allocation functions to take an unsigned C++
type, so implicit conversion errors like this one, can't happen.
Fixed: chromium:1041240
Change-Id: Ib344e5738d2648fbf9a2951ca943ff566ddd5f09
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2000134
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Simon Zünd <szuend@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65746}
The UpdateSharedWasmMemoryObjects function only creates a new
JSArrayBuffer when the the legths of old/new ArrayBuffer objects
are unequal, but the CHECK in the Grow() funciton assumes that a new
object is always created. Fix so that a new ArrayBuffer is always
allocated.
Bug: v8:10044, chromium:1040325
Change-Id: I66912bdc091e65a57e5b50f4ed63b0da5492dcc4
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1999603
Reviewed-by: Ben Smith <binji@chromium.org>
Commit-Queue: Deepti Gandluri <gdeepti@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65742}
Rather than explicitly requesting MAP_HUGETLB mappings, which requires
kernel configuration, we should rely on the "Transparent Hugepages"
feature, where eligible allocation requests are automatically fulfilled
with huge page mappings.
Bug: chromium:1041232
Change-Id: I5263da7a23290316aa7b99e63881ca88e65b4e34
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1997442
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65741}
To follow the style guide, and potentially improve performance, we
switch to {int} for all stack offsets. The size of the stack cannot grow
anywhere near {kMaxInt} anyway, and we have limits in place (for number
of locals, number of parameters, and function body size) which also
prevent this number from ever growing near that limit.
R=jkummerow@chromium.org
Change-Id: I05ca9124d2def79edd212464739bc12315f7b813
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1997445
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65740}
Note the tricky part in instruction-selector-x64, where we flip the
inputs given to the code generator. This is because the semantics we
want is: v128.andnot a b = a & !b, but the x64 instruction performs
andnps a b = !a & b. Therefore we flip the inputs, and combined with
g.DefineSameAsFirst, the output register will be the same as b, and we
can use andnps without any modifications in both SSE and AVX cases.
Bug: v8:10082
Change-Id: Iff98dc1dd944fbc642875f6306c6633d5d646615
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1980894
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65738}
