Previously, we sent a critical low memory pressure notification when
attempting to commit WASM code when the code manager was near the
limit for total amount of code allocated.
https://chromium-review.googlesource.com/c/v8/v8/+/1073412 "fixed" that,
but it causes OOMs on Windows.
Since we no longer have the isolate on the code manager, and thus cannot
send this notification on commit, send the notification upon the next
module creation.
This is still not optimal, but should fix OOM issues for lots of
small modules on Windows.
BUG=v8:7845
R=mstarzinger@chromium.orgCC=clemensh@chromium.org
Change-Id: I6e20d0c1ee9bc6926a83e0c2fbdc9e9e453588ec
Reviewed-on: https://chromium-review.googlesource.com/1098921
Commit-Queue: Ben Titzer <titzer@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53700}
No need to have a separately defined CallDescriptor.
R=titzer@chromium.org
Change-Id: Ic7c0ee87d458fa8e55bef4d750aa7f61a763237f
Reviewed-on: https://chromium-review.googlesource.com/1098927
Reviewed-by: Ben Titzer <titzer@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53699}
The {WasmSharedModuleData} struct was introduced to hold data common to
all wasm instances belonging to the same module. The idea was to keep
"internal state" separate from the JS-facing {WasmModuleObject}. Since
this objective has no real value, and we already store some internal
data on the {WasmModuleObject}, this CL merges these two objects.
R=titzer@chromium.org, mstarzinger@chromium.org
Bug: v8:7754
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng
Change-Id: I04f6d07bf5d812bc4717af26f0f64231345861f9
Reviewed-on: https://chromium-review.googlesource.com/1097491
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53698}
Replace all uses of Deoptimizer::BailoutType and CodeEventListener::DeoptKind
with DeoptimizeKind from src/globals.h.
Change-Id: I5b9002583a69bc43d995cacc7619b018e5a70727
Reviewed-on: https://chromium-review.googlesource.com/1097331
Commit-Queue: Georgia Kouveli <georgia.kouveli@arm.com>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53695}
The often used construct {OFStream(stdout)} does not work on Android.
This CL introduces an {StdoutStream} which behaves exactly like
{OFStream(stdout)} on non-android platforms, and redirects to the
Android log on appropriate systems and configurations.
R=mstarzinger@chromium.org
Bug: v8:7820
Change-Id: Ia682fdf6d064e37c605c19b032f5a10b96ac825b
Reviewed-on: https://chromium-review.googlesource.com/1088911
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53692}
Mark values if their corresponding key was already marked. Also
take into account the size for backing store.
Bug: chromium:844008
Change-Id: I58570f935c69fb39b686fd22dc93caba309f83b2
Reviewed-on: https://chromium-review.googlesource.com/1097400
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@google.com>
Cr-Commit-Position: refs/heads/master@{#53690}
We assumed that if the ErrorThrower is empty after instantiation, then
instantiation succeeded and an instance exists which we can return.
However, if the start function throws, no instance exists, which caused
a crash. With this CL we handle execeptions thrown by the start
function correctly.
R=clemensh@chromium.org
Bug: chromium:848966
Change-Id: I51dc94e6bc563aa4a4b88c44a14e831af913fbd8
Reviewed-on: https://chromium-review.googlesource.com/1092234
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53687}
For serialization, we do not need the {WasmCompiledModule}. All we need
is the {NativeModule}.
R=mstarzinger@chromium.org
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng
Change-Id: Iff7dc9cde3b1ab777028dbafcc1c870e45e28a93
Reviewed-on: https://chromium-review.googlesource.com/1097480
Reviewed-by: Ben Titzer <titzer@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53686}
Mutable globals are now included in the wasm v1 spec.
Bug: v8:7625
Change-Id: Ib9b92d8348102f99a3b92820d0057b2c11a1e49a
Reviewed-on: https://chromium-review.googlesource.com/1095650
Commit-Queue: Ben Smith <binji@chromium.org>
Reviewed-by: Ben Titzer <titzer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53683}
This is a reland of a229e12101
The original commit broke a layout test in Blink. The test in Blink
has been marked to be skipped and will be updated once this patch
lands.
See https://chromium-review.googlesource.com/c/chromium/src/+/1097455
Original change's description:
> [builtins] set DataView.length to 1
>
> Refs: https://github.com/tc39/ecma262/pull/1131
> Test: test262/built-ins/DataView/length
> Bug: v8:7816
> Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
> Change-Id: I66a06734bd32cd2043a8d04728b2185f6093bd69
> Reviewed-on: https://chromium-review.googlesource.com/1094980
> Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
> Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#53649}
Bug: v8:7816
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: If63be80523a68d3a2b515fe1d55a243d2dd2a9b2
Reviewed-on: https://chromium-review.googlesource.com/1097568
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53682}
Mutable globals are stored in the instance differently, so they must be
exported differently as well.
Also fix a counting bug that occurred when a module contained a global
and an imported mutable global (CalculateGlobalOffsets is called
twice).
Bug: v8:7625
Change-Id: I1cd7ef5d6ff7cb7e09239035f89d7b36d0436063
Reviewed-on: https://chromium-review.googlesource.com/1096673
Commit-Queue: Ben Smith <binji@chromium.org>
Reviewed-by: Ben Titzer <titzer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53681}
Port 13fc4a1b3d
Original Commit Message:
External references located within the isolate can be loaded as a
fixed offset from the root register. This avoids one load from memory
over the default indirect method (which reads from the external
reference table).
R=jgruber@chromium.org, joransiu@ca.ibm.com, michael_dawson@ca.ibm.com
BUG=
LOG=N
Change-Id: I4a02973a0279517ab5247f4f9519b0e26168c444
Reviewed-on: https://chromium-review.googlesource.com/1097628
Reviewed-by: Joran Siu <joransiu@ca.ibm.com>
Commit-Queue: Junliang Yan <jyan@ca.ibm.com>
Cr-Commit-Position: refs/heads/master@{#53680}
The isolate is mainly used for accounting purposes. As such, it
doesn't need a field in the WasmCodeManager, and cannot have one
if it is to be made isolate independent. Instead, pass the isolate
explicitly in the appropriate cases.
R=mstarzinger@chromium.org
BUG=v8:7424
Change-Id: I539c2b33692e57605a280530bd704ef25269ad0f
Reviewed-on: https://chromium-review.googlesource.com/1073412
Commit-Queue: Ben Titzer <titzer@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53676}
This CL estimates the sizes of the important managed objects in WASM:
the decoded module {WasmModule}, the native module that contains code
{NativeModule}, and the natively-allocated indirect and import tables
{WasmInstanceNativeAllocations}.
Since Managed<T> updates the isolate's external allocated memory,
it is no longer necessary to do so upon committing or releasing a
native module's memory.
R=mstarzinger@chromium.orgCC=ulan@chromium.org
Bug: v8:7424
Change-Id: Iff4e07d0d328383a925febd654ccbfc95f0930e9
Reviewed-on: https://chromium-review.googlesource.com/1079067
Commit-Queue: Ben Titzer <titzer@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53675}
As specified in https://tc39.github.io/ecma262/#sec-atomics-wait, the
critical section must occur before the load and comparison.
This slightly changes the `AtomicsWaitCallback` API, but in a
direction that arguably makes it more consistent.
As a drive-by fix, reset `node->waiting_` in case there
was an exception from the first callback.
Refs: https://chromium-review.googlesource.com/c/v8/v8/+/1095814
Bug: v8:7836
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng
Change-Id: I577cdf76cedfe39bc61f783203b543c7c68fc238
Reviewed-on: https://chromium-review.googlesource.com/1096236
Reviewed-by: Yang Guo <yangguo@chromium.org>
Commit-Queue: Ben Smith <binji@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53673}
This CL inserts NOP instructions a little bit earlier into empty
blocks; this ensures that instructions keep their initial position.
Bug: v8:7327
Change-Id: Idee5269f4fd7fc15c44bda83a2be74e8cff62df8
Reviewed-on: https://chromium-review.googlesource.com/1097078
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53672}
This is a necessary cleanup before porting Array[No,Single,N]ArgumentsConstructor
builtins to CSA.
Bug: v8:5269, v8:7703
Change-Id: I40a2dd83faab1f8c3c180d461ef62fa4d8578f5e
Reviewed-on: https://chromium-review.googlesource.com/1097079
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53671}
These tests are also affected by duplicate builtin copies (on- and
off-heap).
TBR=yangguo@chromium.org
Bug: v8:6666, v8:7718
Change-Id: I8adfe8b8b63fb5f539cdff5e709e9358a47dfc7e
Reviewed-on: https://chromium-review.googlesource.com/1097088
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53668}
This makes sure debug code in WasmCode traps immediately instead of
embedding a call to the Abort builtin, which would be Isolate specific.
R=clemensh@chromium.org
BUG=v8:7424
Change-Id: I08799d41b5e8d21b9f7af97d65f80a1594d9a613
Reviewed-on: https://chromium-review.googlesource.com/1095113
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53667}
Indirections for external references are now handled completely in
macro assemblers, which can produce more efficient code.
Bug: v8:6666
Change-Id: I158f2e2517e2d97a1006cec6e3b4ac3a7ef3fab6
Reviewed-on: https://chromium-review.googlesource.com/1095307
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53666}
This limits the call targets of {WasmCode} to only reference a limited
set of builtins via the {RelocInfo::CODE_TARGET} relocation mode. By now
most calls have been made independent of the Isolate.
R=clemensh@chromium.org
BUG=v8:7424
Change-Id: If4c643a7104e2ff0536ceb0f0de7fd30a923d854
Reviewed-on: https://chromium-review.googlesource.com/1092576
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53665}
In the process:
- Add strict ordering of Types so that name mangling is consistent
and build time. Previously, the UnionType stored the union's
types in a std::set<const Type*>, which did not have a consistent
ordering of the types in the set.
- Add a int31 type to enable consistency and correctness of
handling of 'constexpr int31' values on the C++ side.
- By removing the "implicit" keyword for operators, there is now
one less difference between operators and calls, another
incremental step in unifying operators and calls.
- Enable external (i.e. C++-defined) generic specializations
- Add CSA support for checking double ElementsKinds, including
tests.
- Clean up some constexpr/non-constexpr handling of ElementsKinds.
Bug: v8:7793
Change-Id: I27699aba70b98ebf5466e5b62b045d7b1dad62c8
Reviewed-on: https://chromium-review.googlesource.com/1091155
Commit-Queue: Daniel Clifford <danno@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53664}
Introduce debug-only non-isolate versions of oddball checks so that
we can check for oddballs in DCHECKs without requiring an isolate.
Bug: v8:7786
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng
Change-Id: I97ee22fb99b23dbfa5d2d0bc7c6b22392d16d694
Reviewed-on: https://chromium-review.googlesource.com/1094875
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53662}
External references located within the isolate can be loaded as a
fixed offset from the root register. This avoids one load from memory
over the default indirect method (which reads from the external
reference table).
Bug: v8:6666
Change-Id: I569308c03f1a8c3feb1be7ceb072178b5be17b25
Reviewed-on: https://chromium-review.googlesource.com/1092739
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53661}
This CL re-enables a mjsunit test for some simulator configurations
after Array.p.sort pivot selection improvements have landed.
R=jgruber@chromium.org
Bug: v8:7795
Change-Id: I03599c34e34a0bafd5f276b03aa4f059a8078026
Reviewed-on: https://chromium-review.googlesource.com/1096764
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Simon Zünd <szuend@google.com>
Cr-Commit-Position: refs/heads/master@{#53660}
This moves all accesses from Typer to the broker. This is policed by
DisallowHandleDereferenceScope in Typer::Visitor::Reduce.
Bug: v8:7790
Change-Id: Ic37d029261b3302eedb902a3b2249834a5dd0512
Reviewed-on: https://chromium-review.googlesource.com/1095305
Commit-Queue: Jaroslav Sevcik <jarin@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53659}
Code generation for stack checks and protected loads produces duplicate
loads of the stack pointer and memory base address. Without gvn these
turn into actual move operations and sign conversions. Even worse, as
turbofan seems to hoist the stack pointer loads to early blocks, I saw
a high register pressure and many spills.
While there is some compile time cost for gvn, I expect it to be rather
low. On the positive side, I see about a 5% performance inprovement for
some benchmarks. If compilation time turns out to be an issue, we can
still revert.
Change-Id: I13be2ace5b27c51c32430d0cb14fbbe8f31fbf6f
Reviewed-on: https://chromium-review.googlesource.com/1095335
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Ben Titzer <titzer@chromium.org>
Commit-Queue: Stephan Herhut <herhut@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53657}
Loading a root generates a kRootRegister-relative load on all
architectures but ia32 (where we currently don't have a root
register).
This is preferable to loading an external reference, which will
usually embed an absolute pointer into the generated code, or add an
indirection when within an embedded builtin.
Bug: v8:6666
Change-Id: I1a93b1325b375bfcf3547977760e1d9ada1d0ee4
Reviewed-on: https://chromium-review.googlesource.com/1095260
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53656}
This patch ports over the spec defined operation `GetOption` from
JavaScript to C++:
https://tc39.github.io/ecma402/#sec-getoption
The JS implementation will be deleted once all it's
users are migrated.
Refactors LocaleConstructor to use this method which fixes some test262
tests. The test262 test status file is updated to reflect this.
Bug: v8:5751, v8:7684
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: Ief5eae9b69dcea50062825163ca7658ed20bd0cf
Reviewed-on: https://chromium-review.googlesource.com/1094201
Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53652}
See: https://trac.webkit.org/changeset/232253/webkit
This is a defensive fix insofar as we were not observing any
issues, hence there is no regression test either.
Bug: v8:6791, v8:3770
Change-Id: Icdabe6a26aca9a5913fba278ae1fd0a79857dfd9
Reviewed-on: https://chromium-review.googlesource.com/1077660
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53651}
