This is an initial step towards clarifying the layout of the
instruction area. As follow-ups, we should remove additional
safepoint and handler table offset parameters, and perhaps alter
Code::safepoint_table_offset (handler_table) semantics to always
contain a real offset and avoid the magic 0 signifying nonexistent
tables.
Bug: v8:8758
Change-Id: I9f54629ff3ddad69904b0e1ce2a58e047397aa15
Reviewed-on: https://chromium-review.googlesource.com/c/1434036
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59202}
Procs return the result by increasing recursion through result_for.
This CL eliminates that mechanism from the Processor interface and uses boolen
return values for sending tests to signal success or the failure to load the
test into the execution queue.
R=machenbach@chromium.org
CC=yangguo@chromium.org,sergiyb@chromium.org
Bug: v8:8174,v8:8731
Change-Id: I073a86ca84bcf88da11132b90013d4c8455bc61e
Reviewed-on: https://chromium-review.googlesource.com/c/1439239
Commit-Queue: Tamer Tas <tmrts@chromium.org>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Reviewed-by: Sergiy Belozorov <sergiyb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59201}
Currently, in debug mode the snippets check that the result of decompression
equals to the full value stored in the field.
Bug: v8:7703
Change-Id: I43d20f15510de57582ee00ca23d676dfd4d06636
Reviewed-on: https://chromium-review.googlesource.com/c/1440049
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59200}
This implements preliminary handling of exceptions thrown by external
functions that have been imported into the module. Note that handling
internal function is still missing, but tests have been added.
R=clemensh@chromium.org
TEST=cctest/test-run-wasm-exceptions
BUG=v8:8091
Change-Id: I9d07739d8b4715a5643114fd7a868cdd8d72efd0
Reviewed-on: https://chromium-review.googlesource.com/c/1445751
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59199}
We can directly skip the write barrier in the case we set only a Smi.
Change-Id: Ic2f4f13f9edb851a18878bec14981a2ad7687433
Reviewed-on: https://chromium-review.googlesource.com/c/1444957
Reviewed-by: Marja Hölttä <marja@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59197}
We extracted the platform-specific API header into separate, platform-
specific header files to avoid OS-specific includes in include/v8.h.
As the platform-specific headers are only needed on their specific
platform, we did not make them available on other platforms in BUILD.gn.
Unfortunately, this is not possible, because the `include` validation in
`gn gen --check` checks the availability of header files without
considering `#ifdef` in the source files. Therefore the check fails on
Windows when the POSIX-specific header file is not available.
With this CL platform-specific headers will be generally available. This
is no problem because they can only be included on their specific
platform anyways.
R=adamk@chromium.org
Bug: chromium:921971
Change-Id: Iee7cc22063ce31a536f937d73799b7217ac1a94e
Reviewed-on: https://chromium-review.googlesource.com/c/1443057
Reviewed-by: Adam Klein <adamk@chromium.org>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59196}
This change lands a change to the register allocator. Other than
classical linear scan, we now take control flow events into account
when deciding spill descisions.
The basic idea is to restore the state of a predecessor on entry
of a successor. In the case of multiple predecessors, we use
heuristics to compute the new state based on the predecessors.
The main addition to the machinery is to support unspilling
live ranges and to undo live range splitting in certain cases.
Currently, disabled behind a flag.
Change-Id: I37a70c948be826a90d8b806a52856ad81f475573
Reviewed-on: https://chromium-review.googlesource.com/c/1426129
Commit-Queue: Stephan Herhut <herhut@chromium.org>
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59195}
In future, weak handles will be considered as independent and MarkActive() will
not be supported anymore. Users should switch to TracedGlobal, when relying on
special cases for using handles with v8::EmbedderHeapTracer.
Bug: chromium:923361, v8:8562
Change-Id: Ic6e01a1ab59a25c5fb0aa2ebfb8ddb02e454d72d
Reviewed-on: https://chromium-review.googlesource.com/c/1443064
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59194}
This removes the unnecessary helper GetNumberOfIncomingArguments
from several frames. Clients should use ComputeParametersCount
instead.
Change-Id: I54c261fd6ff0a8a5dc6d1832a0b397429a7ba726
Bug: v8:8757
Reviewed-on: https://chromium-review.googlesource.com/c/1442642
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59192}
The code reads whatever the stack pointer points at, and only
later decides whether to use that value for anything.
Change-Id: I9930cb046c4fcb3ba0c931058a9c3dd75df482f0
Reviewed-on: https://chromium-review.googlesource.com/c/1443503
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59187}
This is a speculative fix for issue 8744. I couldn't get it to
repro locally, but the stracktrace from the failing bot run points
at TickSample::Init, and according to code comments in that function
the value of {regs.sp} can be "arbitrary", so we must read from
that address using a method that's robust towards unalignment.
Bug: v8:8744
Change-Id: I7a45cc257e0eb557715ec67d9e66e54a6f2c1867
Reviewed-on: https://chromium-review.googlesource.com/c/1440463
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59186}
-fno-inline is unavailable on clang-cl nor cl.exe, and causes a build
failure on Windows bot.
This CL updates that flag to "/Ob0", which is msvc-equivalent of
-fno-inline.
Change-Id: Id9ffed03a855ac0b773ee28625472de664c3cb30
Reviewed-on: https://chromium-review.googlesource.com/c/1442013
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Taiju Tsuiki <tzik@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59185}
The motivation of this change was originally to preserve is_growable
flag over PostMessage in d8. Adding a more general constructor that
uses SharedArrayBuffer::Contents.
Change-Id: Ib8f6c36d659e91f6cfb6487f56de34fa7e8841a9
Bug: v8:8564
Reviewed-on: https://chromium-review.googlesource.com/c/1383093
Commit-Queue: Deepti Gandluri <gdeepti@chromium.org>
Reviewed-by: Ben Smith <binji@chromium.org>
Reviewed-by: Adam Klein <adamk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59184}
TracedGlobal integrates with the use case of EmbedderHeapTracer and replaces
regular weak Global or Persistent nodes for such cases. This allows to simplify
the case for regular weak handles in a sense that they follow regular weak
semantics (if the underlying object is otherwise unreachable the weak handle
will be reset).
TracedGlobal requires slightly different semantics in the sense that it can be
required to keep them alive on Scavenge garbage collections because there's a
transitive path that is only known when using the EmbedderHeapTracer.
TracedGlobal accomodates that use case.
TracedGlobal follows move semantics and can thus be used in regular std
containers without wrapping data structure.
The internal state uses 20% less memory and allows for only iterating those
nodes when necessary. The design trades the virtual call when iterating
interesting persistents in the GC prologue with calling out through the
EmbedderHeapTracer for each node which is also a virtual call. There is one less
iteration over the set of handles required though and the design is robust
against recursive GCs that mutate the embedder state during the prologue
callback.
Bug: chromium:923361
Change-Id: Idbacfbe4723cd12af9de21058a4792e51dc4df74
Reviewed-on: https://chromium-review.googlesource.com/c/1425523
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59183}
Enables WebAssembly's js-api module/customSection. The specification has
been updated; see https://github.com/WebAssembly/spec/issues/915. V8 was
already using DOMString.
Bug: v8:8633
Change-Id: I4c3e93c21594dbba84b3697e7e85069c3ff8b441
Cq-Include-Trybots: luci.chromium.try:linux-blink-rel
Reviewed-on: https://chromium-review.googlesource.com/c/1415554
Commit-Queue: Sven Sauleau <ssauleau@igalia.com>
Reviewed-by: Adam Klein <adamk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59182}
In the trap handler we validate the list of registered code objects
every time we register or de-register a new code object. The complexity
of this validation is O(num-code-objects * num-instructions). For big
WebAssembly modules with several hundred thousand code objects, this
validation causes significant overhead (we saw up to 10x) and makes
debugging very tedious. With this CL I mark the validation as slow.
Thereby it is still enabled in most tests on our bots, but it is
possible to disable validation when debugging large web applications.
The referenced bug issue was created by developers who had problems
with debugging because of this issue.
R=mark@chromium.org
Bug: v8:8536
Change-Id: If7ecb554eebcb04eb43a1f791b96c7a42a47e60f
Reviewed-on: https://chromium-review.googlesource.com/c/1442634
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59181}
This reverts commit b022e825bd.
Reason for revert: Fails nosnap: https://ci.chromium.org/p/v8/builders/luci.v8.ci/V8%20Linux%20-%20nosnap%20-%20debug/22761
Original change's description:
> [builtins] Verify Isolate compatibility with the embedded blob
>
> Embedded builtins (= the embedded blob) have a few dependencies on the
> snapshot state. For instance, they require that metadata stored on
> builtin Code objects as well as the builtins constant table remain
> unchanged from mksnapshot-time. Embedders may violate these
> assumptions by accident, e.g. by loading a snapshot generated with
> different build flags, leading to seemingly unrelated failures later
> on.
>
> This CL introduces an Isolate hash stored in the embedded blob which
> hashes relevant parts of builtin Code objects and the builtins
> constant table. It's verified in Isolate::Init in debug builds.
>
> Bug: v8:8723
> Change-Id: Ifc9bdbe6f56ea67d8984f162afa73a3572cfbba8
> Reviewed-on: https://chromium-review.googlesource.com/c/1442641
> Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#59177}
TBR=sigurds@chromium.org,jgruber@chromium.org
Change-Id: I027ae2585bd06290ea3245b1fc23efba544eb470
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:8723
Reviewed-on: https://chromium-review.googlesource.com/c/1443059
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59179}
Embedded builtins (= the embedded blob) have a few dependencies on the
snapshot state. For instance, they require that metadata stored on
builtin Code objects as well as the builtins constant table remain
unchanged from mksnapshot-time. Embedders may violate these
assumptions by accident, e.g. by loading a snapshot generated with
different build flags, leading to seemingly unrelated failures later
on.
This CL introduces an Isolate hash stored in the embedded blob which
hashes relevant parts of builtin Code objects and the builtins
constant table. It's verified in Isolate::Init in debug builds.
Bug: v8:8723
Change-Id: Ifc9bdbe6f56ea67d8984f162afa73a3572cfbba8
Reviewed-on: https://chromium-review.googlesource.com/c/1442641
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59177}
Registered callbacks should not keep the NativeModule alive. Otherwise,
tiering will always run to completion, even if the NativeModule is not
being used any more.
This change can cause the callback to not be called if the module dies
before it finishes top-tier compilation. This is the desired behaviour.
R=ahaas@chromium.orgCC=titzer@chromium.org, bbudge@chromium.org
Bug: v8:8689, chromium:719172
Change-Id: Ide9d639f465497c3ed3439c7ce25c76dceeb97eb
Reviewed-on: https://chromium-review.googlesource.com/c/1435937
Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59175}
This fixes RunWasmInterpreter_* tests on big endian mips/ppc/s390.
Change-Id: I4b9b767d0de45004ee1195ac225c6d1027c17a05
Reviewed-on: https://chromium-review.googlesource.com/c/1439517
Commit-Queue: Junliang Yan <jyan@ca.ibm.com>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59174}
This is not needed generally, and might prevent catching bugs where
labels are not bound correctly. Instead, only unuse labels on errors.
R=mstarzinger@chromium.org
Change-Id: I1dc77ea21ca80c5be801f17ba284908e1c04abe0
Reviewed-on: https://chromium-review.googlesource.com/c/1443051
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59173}
On Liftoff bailout, instead of binding all unbound labels (to avoid
triggering DCHECKS in their destructor), just Unuse them.
R=mstarzinger@chromium.org
Bug: chromium:924843
Change-Id: Icf581bca06eaa7369ab2bbd5d805112289d6a801
Reviewed-on: https://chromium-review.googlesource.com/c/1442645
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59172}
In some test-run-wasm-js cctests we want to call wasm code through the
js-to-wasm wrapper. This did not work if we wanted to run the wasm code
in the interpreter, because we never compiled a WasmInterpreterEntry,
and we did not patch the jump table.
With this CL we do both now.
R=clemensh@chromium.orgCC=fgm@chromium.org
Change-Id: I49971c5079053ed572928da7eab3e6ea3cb57c25
Reviewed-on: https://chromium-review.googlesource.com/c/1435941
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59170}
jumbo.gni will check whether use_jumbo_build is set or not and
if it's not set it will just check that |jumbo_excluded_sources|
lists do not include non-existing files. That means that there
is little/nothing to lose by always routing v8_source_set that
way, and it adds a bit in build file simplicity and local
checking of |jumbo_excluded_sources|.
Change-Id: I19ed91d304f4a2edbf10661d20e5f5d3180b3a35
Reviewed-on: https://chromium-review.googlesource.com/c/1430832
Commit-Queue: Daniel Bratell <bratell@opera.com>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59169}
During execution, the tier can be switch from Liftoff to TurboFan. We
already handle this in some locations by getting the tier before
execution, so that we later know which was the requested tier for that
unit.
In the case of the --wasm-tier-mask-for-testing flag, this accounting
was not right because the tier was already switched in the constructor.
This CL changes the compilation units to store both the requested and
the executed tier explicitly, so we know which counter to decrement
when the unit finishes.
R=ahaas@chromium.org
Bug: chromium:925671
Change-Id: I673463135e9b3ab17e40cfdfd5d3a526ad5a9b79
Reviewed-on: https://chromium-review.googlesource.com/c/1442639
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59168}
There is no linking stage any more, and all the {Link} method did was
to set the module to executable. Thus rename the method to reflect that.
R=ahaas@chromium.org
Bug: v8:8562
Change-Id: Ifc83b5cbbb1377312d7c318e50a0f9d1ede45aa6
Reviewed-on: https://chromium-review.googlesource.com/c/1439237
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59167}
This is a followup change for https://crrev.com/c/1437275 to keep
the core wasm spec tests in sync with the tests checked out in
test/wasm-js/data.
R=clemensh@chromium.org
Change-Id: I3c78c5db647e9c871b4c64efbc9656acd1266efb
Reviewed-on: https://chromium-review.googlesource.com/c/1442638
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59166}
Currently, errors somewhere deep in a promise chain are not reported as
test failures. We would print the error, but still return with exit
status 0. This CL fixes this to treat any uncaught exception as
failure, as we already do in {Shell::ExecuteString}.
R=yangguo@chromium.org
Change-Id: Idc4e58cc0d45f8fca3f7b2c5070ab68d317fc00b
Reviewed-on: https://chromium-review.googlesource.com/c/1439378
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59165}
This is a reland of d4cc132ec3
without changes.
Original change's description:
> [turbofan] Extend/change serializer list of bytecodes.
>
> In particular, don't clear environment for StackCheck.
> Clear it for all throw and generator-related operations.
>
> Also improve some printing.
>
> Bug: v8:7790
> Change-Id: I45d0994a06ba7ca430e78d631d85b2c6b2e9e323
> Reviewed-on: https://chromium-review.googlesource.com/c/1439376
> Commit-Queue: Georg Neis <neis@chromium.org>
> Reviewed-by: Maya Lekova <mslekova@chromium.org>
> Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#59157}
Tbr: mslekova@chromium.org
Bug: v8:7790
Change-Id: I86a72170eebff332ade08679f7fd895dd0652e1a
Reviewed-on: https://chromium-review.googlesource.com/c/1442633
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59164}
Also removes some obsolete gn definitions.
Bug: v8:8678
Change-Id: I418051177f059ab413b996a641ecf23f70d6c8bc
Reviewed-on: https://chromium-review.googlesource.com/c/1433789
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59163}
This is a reland of 3145505ad3,
after removing the clashing Register name binding.
Original change's description:
> [turbofan] Support new.target in the serializer.
>
> Bug: v8:7790
> Change-Id: Ie98cff6f8b1f184c8152952cc3d39e373c93565d
> Reviewed-on: https://chromium-review.googlesource.com/c/1435943
> Commit-Queue: Georg Neis <neis@chromium.org>
> Reviewed-by: Maya Lekova <mslekova@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#59118}
Tbr: mslekova@chromium.org
Bug: v8:7790
Change-Id: I354d7907342e1f040d01eecc8b51ffca6e83c28d
Reviewed-on: https://chromium-review.googlesource.com/c/1442233
Commit-Queue: Georg Neis <neis@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59160}
This reverts commit d4cc132ec3.
Reason for revert: Bad merge due to other reverted CL
Original change's description:
> [turbofan] Extend/change serializer list of bytecodes.
>
> In particular, don't clear environment for StackCheck.
> Clear it for all throw and generator-related operations.
>
> Also improve some printing.
>
> Bug: v8:7790
> Change-Id: I45d0994a06ba7ca430e78d631d85b2c6b2e9e323
> Reviewed-on: https://chromium-review.googlesource.com/c/1439376
> Commit-Queue: Georg Neis <neis@chromium.org>
> Reviewed-by: Maya Lekova <mslekova@chromium.org>
> Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#59157}
TBR=jarin@chromium.org,neis@chromium.org,mslekova@chromium.org
Change-Id: I349398e8ec1b00d2725032435374c0d4205982f3
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:7790
Reviewed-on: https://chromium-review.googlesource.com/c/1442632
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59159}
In particular, don't clear environment for StackCheck.
Clear it for all throw and generator-related operations.
Also improve some printing.
Bug: v8:7790
Change-Id: I45d0994a06ba7ca430e78d631d85b2c6b2e9e323
Reviewed-on: https://chromium-review.googlesource.com/c/1439376
Commit-Queue: Georg Neis <neis@chromium.org>
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59157}
This replaces InNewSpace with InYoungGeneration, which is
a prerequisite for young large objects.
Additional changes:
- FROM_SPACE, TO_SPACE flags are renamed to FROM_PAGE, TO_PAGE.
- A new LARGE_PAGE flag is added.
- The external string table is refactored to track young string
instead of new space strings.
Bug: chromium:924547
Change-Id: Ia4e3ba1b72995c3400257a1f98559f091533e811
Reviewed-on: https://chromium-review.googlesource.com/c/1437274
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59156}
