This reverts commit 890f694de5.
Previous commit was passing field_type_map_ down in a place where it was not passed down before.
This caused some handles to be used across zones, which caused crashes.
BUG=
Review URL: https://codereview.chromium.org/2453463006 .
i.e. #ifndef VM_WHATEVER -> #ifndef RUNTIME_VM_WHATEVER
This lets us remove a hack from the PRESUBMIT.py script that existed
for reasons that are no longer valid, and sets us up to add some
presubmit checks for the GN build.
R=asiva@google.com, rmacnak@google.com
Review URL: https://codereview.chromium.org/2450713004 .
Start by removing all get:runtimeType overrides in the patch files to have a single point computing the runtime type - Object.get:runtimeType. Handle string, double and integer types inside both intrinsic and runtime call to unify their handling and guarantee that code works even with intrinsifier disabled.
With overrides removed we can easily check that get:runtimeType is unique function name within the application that is being precompiled and use that to convert InstanceCall(get:runtimeType, ...) into StaticCall even nothing is known about the receiver.
This enables us to check if both left side and right side of comparison are StaticCall(Object.get:runtimeType, ...) when specializing InstanceCall(==, x, y). If they are we convert InstanceCall(==, StaticCall(get:runtimeType, a), StaticCall(get:runtimeType, b)) into StaticCall(Object._hasSameRuntimeType, a, b). A canonicalization rule will later delete unused get:runtimeType invocations.
Object._hasSameRuntimeType is implemented in C++ and intrinsified. It operates without creating new runtime types (except for Closures - where it does for simplicity). Cases of different class ids (i.e. a.[cid] != b.[cid]) and non-parameterized types are handled completely in the intrinsic. The rest is handled in the runtime code.
Microbenchmarking results:
Same parameterized classes: 15x improvement
Different parameterized classes: 300x improvement
Different/same non-parameterized classes: 2x improvement
BUG=
R=fschneider@google.com, regis@google.com
Review URL: https://codereview.chromium.org/2379733002 .
The functions where we have intrinsics were always inlined, so the intrinsic
code was actually not used. Move these functions to the inline black list instead to
ensure the intrinsic code is used instead. Inlining these in the generic inliner would
not help anyway since they are inlined as recognized methods by the optimizer.
BUG=
R=rmacnak@google.com
Review URL: https://codereview.chromium.org/2239903002 .
Annotate patch classes and top-level patch functions with @patch
instead of the pseudo-keyword patch. This allows the analyzer
to read patch files, and matches the syntax that dart2js uses.
The deprecated syntax is still supported, but a warning is printed when detected.
BUG=
Review URL: https://codereview.chromium.org/2220883004 .
We don't inline intrinsics to produce smaller code, but for small methods
(array length getters, etc.) we should do it because the code actually
gets smaller by inlining, plus we get better type/range information.
BUG=
R=rmacnak@google.com
Review URL: https://codereview.chromium.org/2147683002 .
This pattern ensures the generation of 'and' instructions on the Smi x Smi path of a polymorphic & operation.
There are two 'and' operations on most simple _InternalLinkedHashMap and _CompactLinkedHashSet operations. One of the inputs, the hashCode, is occasionally a _Bigint. This change ensures that the usual case path has an 'and' instruction instead of a call to _IntegerImplementation.&
R=regis@google.com
Review URL: https://codereview.chromium.org/1913663002 .
In Flutter, this caused
RangeError (index): Index out of range: index should be less than null: 0
Uint32List.[]= (dart:typed_data)
_HashVMBase&MapMixin&&_LinkedHashMapMixin._insert (dart:collection-patch/compact_hash.dart)
This reverts commit 0ae4ae5033.
TBR=fschneider@google.com
Review URL: https://codereview.chromium.org/2023423002 .
Added support for external string using flow graph based intrinsics
which helps with precompiled code, but also polymorphic calls in jitted code.
I also added support for the missing cases in the flow graph optimizer.
BUG=
R=vegorov@google.com
Review URL: https://codereview.chromium.org/1961393002 .
- Ensure the uniqueness of private keys without having to
search the existing key space.
- Pass a thread parameter where useful to library methods.
BUG=
Review URL: https://codereview.chromium.org/1947393003 .
1. Add fast path smi multiply
2. Recognize more operands as likely smi: If they are the result
another CheckedSmiOp, or the result of a phi that has a known smi.
3. Fix guessing receiver cids of getters/setters based.
4. Try to specialize calls based on propagated receiver type first before
using guessed cids (which generates a call with checks + megamorphic slow path)
BUG=
Review URL: https://codereview.chromium.org/1824023002 .
This removes the additional indirection to the VM's implementation
for all the factory constructors.
Additionally, it removes one class per typed data list type, thus reducing
snapshot and generated code size.
BUG=
R=asiva@google.com
Review URL: https://codereview.chromium.org/1790593002 .
not on the signature type hash anymore. This makes fingerprints insensitive to
implementation details of the type system. The signature type hash was included
because, in the early days, the token stream of a function started at the
opening brace and did not include tokens representing the signature.
R=srdjan@google.com
Review URL: https://codereview.chromium.org/1752503002 .
Super initializer calls get implicitly moved to the end of the initializer list, so we don't need the two-phase constructor protocol anymore.
Ryan, can you please look at the mirror changes and check wether I've missed something?
BUG=
R=asiva@google.com, rmacnak@google.com
Review URL: https://codereview.chromium.org/1569213003 .
Minimize the number of dependencies of compiler-related files.
This is needed to remove whole .cc/.h files from a compiler-less VM binary which
can be used for running precompiled code.
* removed dependencies of object.cc on flow_graph_compiler.h and flow_graph_builder.h
* removed unnecessary includes from object.h
BUG=
R=srdjan@google.com
Review URL: https://codereview.chromium.org/1411873005 .
They just redirect to the native typed data list constructor. Inlining
them makes type propagation more precise. There is no size cost with always
inlining these since they just contain a constructor call.
Also, remove dead code from the inliner: Unoptimized code is not accessed during
optimized compilation anymore.
BUG=
R=srdjan@google.com
Review URL: https://codereview.chromium.org/1416603002 .
Make the compact linked hash map, which is the default implementation for Map
(including map literals), a VM-internal class.
This makes it easy to have more efficient serialization of maps (although for
now, the implementation is straight-forward).
Refactor the compact hash class hierarchy to enable the VM-internal class and
the rest (also the Set classes) to share a maximal amount of code, by using
two different bases for implicit/explicit fields.
Remove existing proof-of-concept, C++-based internal VM-class.
BUG=http://dartbug.com/22982R=asiva@google.com
Review URL: https://codereview.chromium.org//1151523002
Vyacheslav Egorov