Introduce a dedicated Float64Log machine operator, that is either
implemented by a direct C call or by platform specific code, i.e.
using the FPU on x64 and ia32.
This operator is used to implement Math.log as a proper TurboFan
builtin on top of the CodeStubAssembler.
Also introduce a NumberLog simplified operator on top of Float64Log
and use that for the fast inline path of Math.log inside TurboFan
optimized code.
BUG=v8:5065
Review-Url: https://codereview.chromium.org/2029413005
Cr-Commit-Position: refs/heads/master@{#36703}
VC++ complains about truncation of integer constants despite use of
static_cast. This isn't very helpful as it gives no way of suppressing
the warning in code, so this change suppresses it on the command line.
Additionally, the linker complains about importing of locally defined
functions in component builds. Until this is fixed the warnings should
be suppressed.
NOTRY=true
Review-Url: https://codereview.chromium.org/2028353004
Cr-Commit-Position: refs/heads/master@{#36695}
Previously, CodeAssembler Variables declared in an explicit C++ scope would
continue to be merged into future labels beyond that scope, causing
asserts. This CL ensures that Variables are properly ignored when they go out of
scope.
Review-Url: https://codereview.chromium.org/2035683002
Cr-Commit-Position: refs/heads/master@{#36690}
Previously, turbofan selected the gap use from first predecessor block when
hinting a phi, unless that block was deferred, in which case the gap move from
the first non-deferred predecessor block was chosen.
This strategy didn't guarantee that an important invariant was maintained: the
predecessor blocks chosen for hinting phis must preceed the phi's block in the
rpo ordering. In most cases the strategy worked, since graphs generated by the
AstGraphBuilder and existing stubs just happened to always generate schedules
where this rpo ordering property for the first predecessor block, but it is
quite possible to generate a code stub by hand that doesn't have this property
(see included test case).
After this CL, the allocator chooses either the the first non-deferred
"rpo-preceeding" block to be the hinting block, or the first deferred
"rpo-preceeding" block if that doesn't exist. In all previously-existing code,
this behavior is the same as the original algorithm, but has the benefit of not
failing in the register allocator in hand-crafted stubs where all the
"rpo-preceeding" predecessors are all in deferred code.
Review-Url: https://codereview.chromium.org/2030463003
Cr-Commit-Position: refs/heads/master@{#36689}
... since CodeStubAssembler does not belong to v8::internal::compiler namespace anymore.
Review-Url: https://codereview.chromium.org/2035533003
Cr-Commit-Position: refs/heads/master@{#36683}
Most maps have a small code cache (often only one entry), so this patch
optimizes memory consumption of such cases by using plain FixedArrays,
only switching to CodeCacheHashTables when the number of cached entries
gets so large that linear-scan lookups get too slow.
On loading inbox.google.com, this gets the aggregate size of all maps'
code caches (there are about 13,600 of them) from 4300 KB to 970 KB.
Review-Url: https://codereview.chromium.org/2021373002
Cr-Commit-Position: refs/heads/master@{#36681}
This introduces optimized number operations based on type feedback.
Summary of changes:
1. Typed lowering produces SpeculativeNumberAdd/Subtract for JSAdd/Subtract if
there is suitable feedback. The speculative nodes are connected to both the
effect chain and the control chain and they retain the eager frame state.
2. Simplified lowering now executes in three phases:
a. Propagation phase computes truncations by traversing the graph from uses to
definitions until checkpoint is reached. It also records type-check decisions
for later typing phase, and computes representation.
b. The typing phase computes more precise types base on the speculative types (and recomputes
representation for affected nodes).
c. The lowering phase performs lowering and inserts representation changes and/or checks.
3. Effect-control linearization lowers the checks to machine graphs.
Notes:
- SimplifiedLowering will be refactored to have handling of each operation one place and
with clearer input/output protocol for each sub-phase. I would prefer to do this once
we have more operations implemented, and the pattern is clearer.
- The check operations (Checked<A>To<B>) should have some flags that would affect
the kind of truncations that they can handle. E.g., if we know that a node produces
a number, we can omit the oddball check in the CheckedTaggedToFloat64 lowering.
- In future, we want the typer to reuse the logic from OperationTyper.
BUG=v8:4583
LOG=n
Review-Url: https://codereview.chromium.org/1921563002
Cr-Commit-Position: refs/heads/master@{#36674}
Reason for revert:
Also failing with the new implementation:
https://build.chromium.org/p/chromium.gpu/builders/Linux%20Debug%20%28NVIDIA%29/builds/62646
Will do a local repro now.
Original issue's description:
> Track based on JSArrayBuffer addresses on pages instead of the attached
> backing store.
>
> Details of tracking:
> - Scavenge: New space pages are processes in bulk on the main thread
> - MC: Unswept pages are processed in bulk in parallel. All other pages
> are processed by the sweeper concurrently.
>
> BUG=chromium:611688
> LOG=N
> TEST=cctest/test-array-buffer-tracker/*
> CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_tsan_rel,v8_mac64_asan_rel
>
> Committed: https://crrev.com/279e274eccf95fbb4bd41d908b9153acf6ec118a
> Cr-Commit-Position: refs/heads/master@{#36653}
TBR=hpayer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:611688
Review-Url: https://codereview.chromium.org/2032973002
Cr-Commit-Position: refs/heads/master@{#36663}
Eliminating dead code in the bytecode array builder doesn't play nice
with the register elimination optimizer. We should move it to it's own
stage in the optimization pipeline, however doing so would require
refactoring of how we deal with jumps, so for now just remove the dead
code elimination optimization.
BUG=chromium:616064
Review-Url: https://codereview.chromium.org/2030583002
Cr-Commit-Position: refs/heads/master@{#36660}
Reason for revert:
There are crashes on Win32 and Win64 bots.
Original issue's description:
> Extend HasProperty stub with dictionary-mode, string wrapper and double-elements objects support.
>
> This CL also replaces some Branch() usages with GotoIf/GotoUnless.
>
> BUG=v8:2743
> LOG=Y
>
> Committed: https://crrev.com/24066b6df4259b302edfa1db884c479008776a7e
> Cr-Commit-Position: refs/heads/master@{#36657}
TBR=verwaest@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:2743
Review-Url: https://codereview.chromium.org/2028333002
Cr-Commit-Position: refs/heads/master@{#36659}
This CL also replaces some Branch() usages with GotoIf/GotoUnless.
BUG=v8:2743
LOG=Y
Review-Url: https://codereview.chromium.org/1995453002
Cr-Commit-Position: refs/heads/master@{#36657}
backing store.
Details of tracking:
- Scavenge: New space pages are processes in bulk on the main thread
- MC: Unswept pages are processed in bulk in parallel. All other pages
are processed by the sweeper concurrently.
BUG=chromium:611688
LOG=N
TEST=cctest/test-array-buffer-tracker/*
CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_tsan_rel,v8_mac64_asan_rel
Review-Url: https://codereview.chromium.org/2026633003
Cr-Commit-Position: refs/heads/master@{#36653}
This CL also updates the elements kind transition lookup logic:
1) First we go back to the root map,
2) Follow elements kind transitions,
3) Replay the property transitions.
BUG=v8:5009
LOG=Y
TBR=bmeurer@chromium.org
Review-Url: https://codereview.chromium.org/2015513002
Cr-Commit-Position: refs/heads/master@{#36652}
It happens that a scavenger runs during incremental marking. Currently scavenger does not care about MarkCompact's mark bits. When an object is alive and marked, and at least one scavenge happens during incremental marking, the object will be copied once to the other semispace in the new_space, and then once to the old_space. For surviving objects this is useless extra work.
In our current attempts (https://codereview.chromium.org/1988623002) to ensure marked objects are scavenged, all marked objects will survive therefore there will be many objects which will be uselessly copied.
This cl modifies our promotion logic so when incremental marking is in progress, and the object is marked, we promote it unconditionally.
BUG=
LOG=no
Review-Url: https://codereview.chromium.org/2005173003
Cr-Commit-Position: refs/heads/master@{#36643}
Async functions are built out of generators, but the
SharedFunctionInfo returns false for is_generator. is_resumable is
the broader query. This patch fixes many parts of V8 to refer
to is_resumable as appropriate.
One incidental change is to remove a check for generators extending
classes. This is part of a general check for constructors being the
only thing which can extend classes, so it is removed here and the
error message for the general case is made more accurate.
BUG=v8:4483
Review-Url: https://codereview.chromium.org/1996943002
Cr-Commit-Position: refs/heads/master@{#36621}
This prevents the compiler from optimizing
f64-to-tagged(tagged-to-f64(x)) ==> x
for non-number x (such as undefined).
Review-Url: https://codereview.chromium.org/2027593002
Cr-Commit-Position: refs/heads/master@{#36613}
Reason for revert:
Breaking
http://build.chromium.org/p/client.v8/builders/V8%20Linux64%20-%20avx2/builds/7972
Original issue's description:
> Track based on JSArrayBuffer addresses on pages instead of the attached
> backing store.
>
> Details of tracking:
> - Scavenge: New space pages are processes in bulk on the main thread
> - MC: Unswept pages are processed in bulk in parallel. All other pages
> are processed by the sweeper concurrently.
>
> BUG=chromium:611688
> LOG=N
> TEST=cctest/test-array-buffer-tracker/*
> CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_tsan_rel,v8_mac64_asan_rel
>
> Committed: https://crrev.com/089da007bb990b5b29aab257aa836fdd3f3b8ce0
> Cr-Commit-Position: refs/heads/master@{#36608}
TBR=hpayer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:611688
Review-Url: https://codereview.chromium.org/2028503003
Cr-Commit-Position: refs/heads/master@{#36609}
backing store.
Details of tracking:
- Scavenge: New space pages are processes in bulk on the main thread
- MC: Unswept pages are processed in bulk in parallel. All other pages
are processed by the sweeper concurrently.
BUG=chromium:611688
LOG=N
TEST=cctest/test-array-buffer-tracker/*
CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_tsan_rel,v8_mac64_asan_rel
Review-Url: https://codereview.chromium.org/2024063002
Cr-Commit-Position: refs/heads/master@{#36608}
Expose more or less the full functionality of the KeyAccumulator in the API:
- use the PropertyFilter introduced for GetOwnPropertyNames
- use KeyCollectionLimit for OWN_ONLY or INLCUDE_PROTOS
- use IndexFilter to eithe SKIP_INDICES or INCLUDE_INDICES
Rewire Object::GetOwnPropertyNames to use GetPropertyNames.
BUG=chromium:148757
Review-Url: https://codereview.chromium.org/2002203002
Cr-Commit-Position: refs/heads/master@{#36595}
Reason for revert:
Investigating new arm simulator failure:
https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm%20-%20sim/builds/851/steps/Check%20-%20novfp3/logs/box2d
Original issue's description:
> Reland "[heap] Fine-grained JSArrayBuffer tracking"
>
> Track based on JSArrayBuffer addresses on pages instead of the attached
> backing store.
>
> Details of tracking:
> - Scavenge: New space pages are processes in bulk on the main thread
> - MC: Unswept pages are processed in bulk in parallel. All other pages
> are processed by the sweeper concurrently.
>
> BUG=chromium:614730, chromium:611688
> LOG=N
> CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_tsan_rel,v8_mac64_asan_rel
>
> Committed: https://crrev.com/bc0fb6e83eb7351f52bfd6f780bb6c12bd87c58c
> Cr-Commit-Position: refs/heads/master@{#36592}
TBR=hpayer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:614730, chromium:611688
Review-Url: https://codereview.chromium.org/2021893002
Cr-Commit-Position: refs/heads/master@{#36593}
Track based on JSArrayBuffer addresses on pages instead of the attached
backing store.
Details of tracking:
- Scavenge: New space pages are processes in bulk on the main thread
- MC: Unswept pages are processed in bulk in parallel. All other pages
are processed by the sweeper concurrently.
BUG=chromium:614730, chromium:611688
LOG=N
CQ_EXTRA_TRYBOTS=tryserver.v8:v8_linux_arm64_gc_stress_dbg,v8_linux_gc_stress_dbg,v8_mac_gc_stress_dbg,v8_linux64_tsan_rel,v8_mac64_asan_rel
Review-Url: https://codereview.chromium.org/2026463002
Cr-Commit-Position: refs/heads/master@{#36592}
These builtins always call into C++ anyways and so there's no point in
having the JavaScript wrapper around them, but instead they can be
implemented as C++ builtins directly.
R=franzih@chromium.org
BUG=v8:5049
Review-Url: https://codereview.chromium.org/2018983002
Cr-Commit-Position: refs/heads/master@{#36569}
One of the top runtime calls in many workloads is Runtime_Call. Add an
intrinsic for the interpreter to inline this runtime call.
BUG=v8:4280
LOG=N
Review-Url: https://codereview.chromium.org/2019433003
Cr-Commit-Position: refs/heads/master@{#36565}
Annex B 3.3 applies only for ordinary FunctionDeclarations, not
GeneratorDeclarations or AsyncFunctionDeclarations. This patch
- Skips applying Annex B 3.3 to async functions
- Adds a flag to refrain from applying it to generators
- UseCounter for how often duplicate function in block occurs
with generators (unclear how to measure need for hoisting from block)
BUG=v8:4806
Review-Url: https://codereview.chromium.org/1995863002
Cr-Commit-Position: refs/heads/master@{#36557}
Adding optional operators for FNeg for WebAssembly, as the current implementation was significantly suboptimal for ARM.
Review-Url: https://codereview.chromium.org/2011303002
Cr-Commit-Position: refs/heads/master@{#36544}
Rewrite decodeURI and decodeURIComponent as builtin functions
and install them in the bootstrapper.
Delete unused runtime functions:
- TruncateString
- NewString
- OneByteSeqStringGetChar
- OneByteSeqStringSetChar
- TwoByteSeqStringGetChar
- TwoByteSeqStringSetChar
Add regression test for decoding large strings. Clusterfuzz detected
a problem with %TruncateString, see
https://bugs.chromium.org/p/chromium/issues/detail?id=612109#c6
This is automatically fixed by this rewrite because %TruncateString
is deleted anyways.
Crude benchmark on 585 decodeURI and decodeURIComponent tests
averaged over five runs:
* builtin functions
real 0m9.69s
user 2m39.8816s
sys 0m12.6398s
* JS functions calling into the runtime e.g., for %TruncateString
real 0m11.0598s
user 3m6.7026s
sys 0m13.5756s
By running:
$ time tools/run-tests.py --arch=x64 --mode=Release --buildbot
test262/built-ins/decodeURI* mjsunit/uri
>>> Running tests for x64.Release
BUG=v8:4912, chromium:612109
R=yangguo@chromium.org, bmeurer@chromium.org
Review-Url: https://codereview.chromium.org/1994733003
Cr-Commit-Position: refs/heads/master@{#36543}
This intrinsic (and the matching runtime entry) are no longer used by
now and can thereby be removed.
BUG=v8:5049
Review-Url: https://codereview.chromium.org/2016993003
Cr-Commit-Position: refs/heads/master@{#36540}
We get less "pollution" of type feedback if we have one vector per native
context, rather than one for the whole system. This CL moves the vector
appropriately.
BUG=
Review-Url: https://codereview.chromium.org/1906823002
Cr-Commit-Position: refs/heads/master@{#36539}
For the common use case of having a single resolve or reject callback,
the callbacks are stored directly. Only when an additional callback is
registered, we create an array to store these callbacks.
There are 3 possible states for the resolve, reject symbols when we add
a new callback --
1) UNDEFINED -- This is the zero state where there is no callback
registered. When we see this state, we directly attach the callbacks to
the symbol.
2) !IS_ARRAY -- There is a single callback directly attached to the
symbols. We need to create a new array to store additional callbacks.
3) IS_ARRAY -- There are multiple callbacks already registered,
therefore we can just push the new callback to the existing array.
Also, this change creates a new symbol for storing the deferred objects.
Previously the deferred objects were stored in the callback arrays, but
since we no longer create arrays for the initial case, we need this new
symbol. The cctest has been updated to account for this new symbol.
This patch results in a 19% improvement(over 5 runs) in the bluebird benchmark.
BUG=v8:5046
Review-Url: https://codereview.chromium.org/2007803002
Cr-Commit-Position: refs/heads/master@{#36536}
This patch does five things:
1. Extracts sampler as libsampler to provide sampling functionality support.
2. Makes SampleStack virtual so embedders can override the behaviour of sample collecting.
3. Removes sampler.[h|cc].
4. Moves sampling thread into log.cc as workaround to keep the --prof functionality.
5. Creates SamplerManager to manage the relationship between samplers and threads.
The reason we port hashmap.h is that in debug mode, STL containers are using
mutexes from a mutex pool, which may lead to deadlock when using asynchronously
signal handler.
Currently libsampler is used in V8 temporarily.
BUG=v8:4789
LOG=n
Committed: https://crrev.com/06cc9b7c176a6223971deaa9fbcafe1a05058c7b
Cr-Commit-Position: refs/heads/master@{#36527}
Review-Url: https://codereview.chromium.org/1922303002
Cr-Commit-Position: refs/heads/master@{#36532}
