ASAN on Windows uses exceptions to manage its shadow memory. However,
this behavior can conflict with WebAssembly trap handler, because
WebAssembly trap handler are executed before the ASAN exception handler.
For some WebAssembly instructions we do not generate assembly code but
call to C functions instead. Since these functions are very simple, we
do not want to reset the thread-in-wasm flag before calling them.
However, when these functions trigger the ASAN exceptions, the
thread-in-wasm flag gets out-of-sync. This happened for the
memory_fill_wrapper function. Originally we thought that it's sufficient
to just mark the function with DISABLE_ASAN. However, this is not enough
because clang compiles the function to use memset, and memset gets
replaced by ASAN with asan_memset.
Therefore I decided now that just for sanitizer builds on Windows, we
reset the thread-in-wasm flag in memory_fill_wrapper. This is not ideal
because it's test-specific code within production code. However, the
alternatives also don't sound convincing.
Alternatives would be:
* Resetting the thread-in-wasm flag whenever we call a c-function
- This would be unnecessary performance overhead for production code
just to make a test work.
* Configure ASAN to not change memset.
- This would weaken ASAN also for other cases.
* Disable ASAN for trap handlers, or trap handlers in ASAN builds.
- This would reduce test coverage.
R=binji@chromium.org
Bug: chromium:957405
Change-Id: Ibd13c6fe7b898238f636db576552e3e4b278c04a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617671
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Ben Smith <binji@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61658}
The invariant is that Map::bit_field2 shouldn't change, and the
IsInRetainedMapListBit apparently changes when the map is held
weakly from optimized code. This causes TurboFan compilations to
change the Map::Hash() result, which in turn causes lookups on
the normalized map cache to miss (and maybe other bad consequences).
With this change we swap Map::IsInRetainedMapListBit (previously in
bit_field2) and Map::HasHiddenPrototypeBit (previously in bit_field3)
to address this problem.
Bug: chromium:963411, v8:9114, v8:9267
Change-Id: I040a27c37305fa602649750bd93bee40c91fca78
Cq-Include-Trybots: luci.chromium.try:linux-rel,win7-rel
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1619747
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61657}
Recently* a second ResolveRenames appeared in
typed-optimization.cc. In some extreme (non-default) jumbo
builds that file shared translation unit (and thus anonymous
namespace) with load-elimination.cc which also has a
ResolveRenames which is slightly different.
To avoid that name clash and failed compilation, this
renames ResolveRenames -> ResolveRenamesForOpt.
*) It appeared in
https://chromium-review.googlesource.com/c/v8/v8/+/1609803
Change-Id: Ia175468a8e978e93e95da1d28f74cee8dce253d2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1615465
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Auto-Submit: Daniel Bratell <bratell@opera.com>
Cr-Commit-Position: refs/heads/master@{#61655}
This fixes the error message generated for compile errors during
asynchronous instantiation. It shows "WebAssembly.instantiate()" now
instead of "WebAssembly.compile()".
R=mstarzinger@chromium.org
Bug: v8:9266
Change-Id: Ieae478d1c4f6843fbc17e15debb6c49f72059d99
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617940
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61654}
This adds three message tests to test the message stack trace we show
for
- synchronous compilation,
- asynchronous compilation, and
- asynchronous instantiation.
Note that the message for the asynchronous cases currently contain the
"WebAssembly.Module()" prefix, which will be fixed in a separate CL.
R=mstarzinger@chromium.org
Bug: v8:9266
Change-Id: I370f4211b5f577ea1b5da026a78b292b50c6a339
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617938
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61653}
This also fixes a bug in 'InitializeCallBuffer', where it wouldn't claim enough
slots for each parameter. This caused the Simd128 instruction selector test to
only claim 3 slots (rather than 4) and then perform an unnecessary padding poke.
v8_Default_embedded_blob_size from the generated file gen/embedded.S
Before: 4957056
After: 4954368
This gives a 0.05% size decrease.
Change-Id: Ic9bb998fb8a9111fb90e1c3e537ea0f2a5fa7b33
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617665
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Martyn Capewell <martyn.capewell@arm.com>
Cr-Commit-Position: refs/heads/master@{#61649}
Added the comparison of two decompresses to the DecompressionElimination
AdvancedReducer. Note that it works in the case that lhs is equal to rhs.
Also added tests for its implementation.
Cq-Include-Trybots: luci.v8.try:v8_linux64_pointer_compression_rel_ng
Cq-Include-Trybots: luci.v8.try:v8_linux64_arm64_pointer_compression_rel_ng
Bug: v8:8977, v8:7703
Change-Id: I21676a78b592859692768c3499ea11117d3bb5a5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1609793
Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61647}
Fixes the crash stated in the chromium bug.
We weren't decompressing before CheckMap which resulted in a crash.
Implemented the same for CompareMap.
Bug: chromium:963917
Change-Id: I1dcaf6845758c7266bbc7d9ecc9e43cf8d6b639a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617252
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61646}
Now that we introduced the Compressed representation, we can simply use
AccessBuilder::ForMap. AccessBuilder::ForCompressedMap was introduced
as a placeholder previously.
Cq-Include-Trybots: luci.v8.try:v8_linux64_pointer_compression_rel_ng
Cq-Include-Trybots: luci.v8.try:v8_linux64_arm64_pointer_compression_rel_ng
Bug: v8:8977, v8:7703
Change-Id: I15b559c27641e6a673862c86be176e259835308e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617664
Reviewed-by: Michael Stanton <mvstanton@chromium.org>
Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61645}
This CL also removes "HasComplexElements" helpers on JSReceiver
and NumberDictionary.
Drive-by: Remove unused SmiLexicographicCompare runtime function.
The C++ funtion is still used, but it is entered using a fast
C call from Array#sort.
Bug: v8:9183
Change-Id: Ibb5d54cca623486806587bc90506af8d97910dc2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617929
Commit-Queue: Simon Zünd <szuend@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Auto-Submit: Simon Zünd <szuend@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61644}
COMPLETE events are being deprecated and don't work well with perfetto
because we can't give out a handle into the buffer like we used to so
that the caller can update the existing event with the duration.
BEGIN/END pairs should be used instead to add two separate trace events
to the buffer which can be associated with one another by the trace
processor e.g. the trace viewer UI.
Cq-Include-Trybots: luci.v8.try:v8_linux64_perfetto_dbg_ng
Bug: v8:8339
Change-Id: Ib73c19d77ad58456ce23d15f0b658c26f3dc3d53
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1615257
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Peter Marshall <petermarshall@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61641}
These were leftover in another directory and can be incorporated into
our standard benchmark runner. We already had some Array slice cases
in js-perf-test so just add some of the important cases from the other
directory to the existing implementation.
Bug: v8:9254
Change-Id: I4cc235b8d3719ecd729f23fe9705ea36d445c340
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617258
Auto-Submit: Peter Marshall <petermarshall@chromium.org>
Commit-Queue: Yang Guo <yangguo@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61640}
This reverts commit 3dd5661204.
Reason for revert: causes leak tests in blink layout tests to fail: https://ci.chromium.org/p/chromium/builders/ci/WebKit%20Linux%20Trusty%20Leak/34295
Steps to reproduce:
- Build blink_tests with following GN args
is_component_build = false
is_debug = false
strip_absolute_paths_from_debug_symbols = true
- Run
third_party/blink/tools/run_web_tests.py --additional-expectations third_party/blink/web_tests/LeakExpectations --time-out-ms 48000 --enable-leak-detection external/wpt/wasm/webapi/body.any.html
Original change's description:
> [stack-trace] Include API functions in Error.stack stack trace
>
> This CL extends Error.stack to include frames of functions declared
> with the C++ FunctionTemplate API. For example, "print" in d8.
>
> Two changes are necessary:
> - HandleApiCall and friends need to go through an BUILTIN_EXIT frame
> instead of an EXIT frame. The existing stack-trace machinery will
> then pick up FunctionTemplate frames without additional changes.
> - Turbofan doesn't go through HandleApiCall, but instead uses an
> ASM builtin to enter FunctionTemplate functions. A "marker"
> frame state is needed to include these frames in the stack trace.
>
> Note: This CL only includes these frames in Error.stack,
> but not (yet) in the stack-trace API (v8.h).
>
> Bug: v8:8742,v8:6802
> Change-Id: Ic0631af883cf56e0d0122a2e0c54e36fed324d91
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1609835
> Commit-Queue: Simon Zünd <szuend@chromium.org>
> Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
> Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
> Reviewed-by: Jakob Gruber <jgruber@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#61602}
TBR=yangguo@chromium.org,sigurds@chromium.org,jgruber@chromium.org,bmeurer@chromium.org,szuend@chromium.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: v8:8742, v8:6802
Change-Id: I4942cd32c6ee5e249dae046eea6b9b2f7120b8ba
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617933
Reviewed-by: Yang Guo <yangguo@chromium.org>
Reviewed-by: Simon Zünd <szuend@chromium.org>
Commit-Queue: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61639}
Code that is being moved primarily deal with layout of a JSObject,
accessing properties and elements, and map transitions.
NOTREECHECKS=true
NOTRY=true
Bug: v8:9247
Change-Id: Ibce5d5926ac4021c8d40c4dd109948775ce1da58
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1613994
Commit-Queue: Yang Guo <yangguo@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61638}
This is a reland of e7e512da66
Original change's description:
> [turbofan] Add fast path for single-character String#startsWith()
>
> This CL adds a fast path to String#startsWith(s) if s is a
> single character string.
>
> Bug: v8:8400
> Change-Id: Ibd6a9d1e46d98f41c198d2b579208e25003eedb0
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1525362
> Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
> Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
> Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#61504}
Bug: v8:8400
Change-Id: Ic2d60ccb8fdeb51373fcd025a7e970fda0c14d79
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1618342
Reviewed-by: Sigurd Schneider <sigurds@chromium.org>
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61637}
We disabled the `delete` optimization, where `delete` on a fast-mode
object goes back in the transition tree, because that optimization
didn't pay attention to constant field tracking.
This change now does the proper fix, which is to invalidate the
constness and properly deoptimize all code that depends on it.
Drive-by-fix: Handlify the DeleteObjectPropertyFast helper.
Bug: chromium:962588, chromium:963999, v8:9233
Change-Id: I5978c32a48d1635b3ce42dc08b00bb2654baa36a
Cq-Include-Trybots: luci.chromium.try:linux-rel,win7-rel
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617251
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61636}
This CL also removes the ElementsAccessor "slice" implementation, as
the runtime function is the last use site.
R=verwaest@chromium.org
Bug: v8:9183
Change-Id: If268e20120e7c7bb4a58d9560482b35896b0992f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617662
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Commit-Queue: Simon Zünd <szuend@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61635}
If speculation is disallowed, this change lets us still do the
optimization if (a) the maps were already reliable or
(b) we are able to take stability dependencies.
Bug: v8:8820
Change-Id: I08340fc19ac87b80aa2b7ed77753dd642e89804f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617663
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61634}
The --log-timer-events doesn't work anymore, since it cannot be enabled
after the snapshot was created ever since the CallApiCallback stub was
turned into a builtin.
Bug: v8:9183
Change-Id: I6e0b92141f1986fc791943ffef42dd4961aea0a1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617931
Auto-Submit: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Yang Guo <yangguo@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61633}
deopt-unlinked.js: bytecode flushing destroys the information that
%GetDeoptCount relies on, so turn that off for this test.
cleanupsome-after-unregister.js: the function-local object {o} is
assumed to be alive throughout the function, so make sure its live
range (from the optimizing compiler's view) extends that far.
Drive-by cleanup: drop some unnecessary casting boilerplate from
Genesis::InitializeGlobal_harmony_intl_date_format_range().
Change-Id: I28617f842fe046dd0875a9a082cfc55a3a076bcf
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617674
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61621}
es6/classes is slow to run in stress mode and times out on PPC.
As a long term fix we should split this into multiple tests so
each individual test runs faster.
Bug: v8:9246
Change-Id: I95eed06d85f73cc66229dfbd83ac7521dcbbb54f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1615252
Commit-Queue: Mythri Alle <mythria@chromium.org>
Reviewed-by: Sergiy Belozorov <sergiyb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61619}
Fastpath failed to store the hole on the array left side.
Bug: chromium:940274
Change-Id: I1eca7b241030474cf5aed6c68f155a1d22ae553e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617255
Commit-Queue: Michael Stanton <mvstanton@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61618}
This adds a counter for the time it takes per single code GC. It is
sampled in all isolates whenever a GC finishes, and during regular
JavaScript GC to also get samples for GCs that never finish (which we
would need to fix).
R=mstarzinger@chromium.org, mpearson@chromium.org
Bug: v8:8217
Change-Id: I722d01b6bd7832c3af636362acfddeab7e8a485e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1615250
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61617}
This adds a constructor function to be used as the base class for all
exported functions. With type reflection enabled exported functions are
instances of this new constructor. Using the constructor directly as
well as the 'type' property is still missing.
R=jkummerow@chromium.org
TEST=mjsunit/wasm/type-reflection
BUG=v8:7742
Change-Id: Id58359cf7866d24be0745abf36a839d3d86e5922
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1617253
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61614}
We currently have the problem that we trigger too many code GCs since
{new_potentially_dead_code_size_} is never reset to zero.
This CL adds a counter which tells us how many GCs we ran per native
module. This counter is sampled on each code GC. It will give us a
good understanding of the amount of GC work we are executing in the
wild. The number should stay in the single-digits generally.
R=mstarzinger@chromium.org, mpearson@chromium.org
Bug: v8:8217
Change-Id: I978a98dff76e0f466ff51e067626886b58d52ded
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1615246
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61611}
