There are problems calling WaitForRunningWorkers when we
call quit().
First, suppose the main thread first calls quit(),
and the worker thread calls quit() after the main
thread calls quit(), then sched_yield to wait for
quit_once_ updated to ONCE_STATE_DONE. However
the main thread is WaitForRunningWorkers to wait
for the worker thread to join, thus causing deadlock.
Second, suppose the worker thread calls quit() and empty
the running_workers_ by WaitForRunningWorkers, then
the main thread calls `onExit(isolate, true)` to dispose
the platform and other global data, which will crash other
running workers.
Bug: v8:12219
Change-Id: I333e5aad431daefb1c163f69e66d8e9d5e9bf754
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3518908
Reviewed-by: Marja Hölttä <marja@chromium.org>
Commit-Queue: Marja Hölttä <marja@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79525}
This implements the last bits of basic concurrent Maglev compilation.
When jobs have been processed, schedule an interrupt to trigger codegen
and building the Code object on the main thread.
Bug: v8:7700
Change-Id: I348ade4777ddddf7c3a6b0575d9f51e5fa00c9fb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3528494
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79516}
.. into new virtual subclass TurbofanCompilationJob. Update all TF code
to derive from this class. Specifically, the OptimizedCompilationInfo is
TF-specific and now lives in TurbofanCompilationJob.
The motivation behind this is that Maglev now also uses this
infrastructure.
Drive-by: Replace CompilationMode with ConcurrencyMode.
Bug: v8:7700
Change-Id: Iae6d1ffd1c810e2e45cad6c9b4e43d4c82ac54a7
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3528493
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Auto-Submit: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79515}
In JSCallReducer::ReduceArrayPrototypeShift, add Unsigned32
TypeGuard for index Node used in fast path, avoid representing
kRepFloat64 (Range(1, inf)) to kRepWord64 when converting
input for kLoadElement.
Bug: v8:12632
Change-Id: I2e4b00840dc5462e4351e13a372c33b6272b9ea1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3528373
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79514}
This started out as a minor code move of early-osr logic, but
became a more general refactor of the tiering decisions.
Early-OSR: the intent here is to trigger OSR as soon as possible
when matching OSR'd code is cached. Move this out of ShouldOptimize
(since it has side effects), and into a dedicated function that's
called early in the decision process.
Note that with this change, we no longer trigger normal TF optimization
along with the OSR request - TF tiering heuristics are already complex
enough, let's not add yet another special case right now.
Other refactors:
- Clarify terminology around OSR. None of the functions in TM actually
perform OSR; instead, they only increase the OSR urgency, effectively
increasing the set of loops that will trigger OSR compilation.
- Clarify the control flow through the tiering decisions. Notably,
we only increment OSR urgency when normal tierup has previously been
requested. Also, there is a bytecode size limit involved. These
conditions were previously hidden inside other functions.
Bug: v8:12161
Change-Id: I8f58b4332bd9851c6b299655ce840555fb7efa92
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3529448
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79512}
- Restructure the runtime function implementation.
- Rename osr_loop_nesting_level to osr_urgency and add helpers.
The motivation for the latter: I've always struggled with the
`osr_loop_nesting_level` term; it neither matches terminology of
what it's compared against (= the loop depth), nor implies what it's
used for (= osr is triggered when `loop depth < osr nesting level`).
In this CL it's renamed to `osr_urgency` to reflect that as urgency
rises, we consider more and more loops as OSR candidates.
Bug: v8:12161
Change-Id: I194ec5a3f1f02526641af1c7796ee0956b6fd3a1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3528735
Reviewed-by: Patrick Thier <pthier@chromium.org>
Reviewed-by: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79509}
CL https://chromium-review.googlesource.com/c/v8/v8/+/3514072 tried to
improve code generation for patterns like "x >> n == k" when n and k
are constant, and to generate instead "x == k << n" (with "k << n"
being computed at compile time).
However, this was also done when "x >> n" was reused later, which
caused "x" to be kept alive longer that it could have, which could
increase register pressure.
This CL thus ensures that this optimization is done only if "x >> n"
has a single use.
Bug: chromium:1305389
Change-Id: I377e120c4825e2a0deb4a5478138da838bcebc77
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3528987
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Darius Mercadier <dmercadier@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79502}
We check if debugger needs to hook into every call when we enqueue
the function to batch and when we compile, but we do not check it
when we install it (that is, set_baseline_code), which is done in
the main thread.
Bug: v8:12713
Change-Id: I81ba221caed1060976e8865174d392a861f2ab24
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3528988
Commit-Queue: Victor Gomes <victorgomes@chromium.org>
Auto-Submit: Victor Gomes <victorgomes@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79500}
Please revert this CL if it causes any trouble!
Bug: v8:11111
Change-Id: I6061fdc0aed82952f093ede9ecd252be1ac99519
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3528495
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Marja Hölttä <marja@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79499}
In text, we should use the properly capitalized "Wasm" instead of
"WASM". In particular for frame output, other frames typically use
CamelCase (like "JsToWasmFrame", "ExitFrame", "InternalFrame"), so Wasm
should do the same.
The "0xa" comment in some assemblers is also outdated, the current tag
we emit for StackFrame::WASM is 0x8 (0x4 shifted by one).
R=jkummerow@chromium.org
Bug: v8:12425
Change-Id: Ic3e00c401b219c28b5424c82efb0f1a9df51690f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3525195
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79498}
JSObject::InitializeBody now checks whether the instance type of the
object being initialized can have embedder data slots around the
initialization logic for these slots. This fixes a performance
regression on certain benchmarks.
To perform this check efficiently, a new instance type,
JSObjectWithEmbedderSlots, is introduced so that the check becomes a
simple range check.
Bug: chromium:1304139
Change-Id: I00c892bc2276e950b59602257ca1c2435c10e517
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3507712
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Samuel Groß <saelo@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79497}
This method now handles external pointers in HeapObjects during
serialization by encoding the representation of the external pointer
(sandboxed, raw), the origin (internal, api) and potentially the
external pointer tag. It is currently only used to handle
JSExternalObjects but could, in the future, be extended to handle all
external pointers that need special handling during
serialization/deserialization.
Bug: v8:12700
Change-Id: Ib0747d765ddc632e4ca4ee94521616d0271be0bc
Cq-Include-Trybots: luci.v8.try:v8_linux64_heap_sandbox_dbg_ng,v8_linux_arm64_sim_heap_sandbox_dbg_ng
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3521904
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Samuel Groß <saelo@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79496}
This CL implements most parts of the concurrent maglev pipeline.
- MaglevConcurrentDispatcher: controls concurrent jobs.
- MaglevCompilationInfo: holds job-global data, controls handle
fiddling between the main isolate and local isolates, owns
job-global state like the Zone.
- MaglevCompilationUnit: same as before, holds per-unit data.
Still missing: job finalization.
Bug: v8:7700
Change-Id: I281178d945e79a0ba97fa2ac7023285d84a16641
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3516036
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79489}
This is necessary due to the indirect call to memfd_create. The function
is only used for testing though.
Bug: v8:12682
Change-Id: Ifb9fd855213064165fa4c2125846ff592368b625
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3525534
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Samuel Groß <saelo@chromium.org>
Auto-Submit: Samuel Groß <saelo@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79487}
The use of memfd_create causes V8 to require glibc 2.27 which wasn't
previously needed. This CL rewrites the affected code to check at
runtime whether memfd_create is available and otherwise use mkstemp.
Bug: v8:12682
Change-Id: I84dc3f5ab7504cec2b599bc92501ddecc2ae22cf
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3516870
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Samuel Groß <saelo@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79486}
Port 8a0d1b6fe5
Original Commit Message:
Modernise the RegList interface to be a proper class, rather than a
typedef to an integer, and add proper methods onto it rather than ad-hoc
bit manipulation.
In particular, this makes RegList typesafe, adding a DoubleRegList for
DoubleRegisters.
The Arm64 CPURegList isn't updated to use (or extend) the new RegList
interface, because of its weird type-erasing semantics (it can store
Registers and VRegisters). Maybe in the future we'll want to get rid of
CPURegList entirely and use RegList/DoubleRegList directly.
R=leszeks@chromium.org, joransiu@ca.ibm.com, junyan@redhat.com, midawson@redhat.com
BUG=
LOG=N
Change-Id: I997156fe4f4f2ccc40b2631d5cb752efdc8a5ad2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3525084
Reviewed-by: Junliang Yan <junyan@redhat.com>
Commit-Queue: Milad Farazmand <mfarazma@redhat.com>
Cr-Commit-Position: refs/heads/main@{#79484}
It turned out that on arm and arm64 we over-estimated the code size of a
Wasm module quite a bit. This CL adds some more output for the
--trace-wasm-compilation-times flag, and adds a script to compute the
factors we use for code size estimates from that output.
I ran the script on a few benchmarks (an older Epic module, the current
Photoshop module, and the benchmark from the linked bug), and adjusted
the constants accordingly.
Also, simplify the API of {ReservationSize} to only return a single
number, and fail internally if we need to allocate more than the engine
supports (which would only fail for artificially large modules).
R=jkummerow@chromium.org
Bug: chromium:1302310
Change-Id: I5b2c27ff3e360fb6738cf5dd697bcee09e106b6d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3522067
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79482}
This CL adds a missing header that was implicitly included by
"test/cctest/wasm/wasm-run-utils.h", which resulted in test failure
with v8_enable_webassembly = true.
Drive-by: Improve traceability of the fastcall cctests.
Bug: chromium:1052746
Change-Id: I4cafbce2e390958befee896782ad13a9bdfc0f30
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3525113
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Maya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79481}
Format strings are the convention in Py3
R=machenbach@chromium.org,alexschulze@chromium.org
Bug: chromium:1298869
Change-Id: Ia660e933f07d2a592c2edadaa99df68c47adb80d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3521789
Auto-Submit: Tamer Tas <tmrts@chromium.org>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/main@{#79477}
