This CL updates the parsing of "break" statements in the debug console
in observatory, as well as handling that the same script can be a part
of several libraries.
Before one could get stuff like this:
$ break package:front_end/src/fasta/scope.dart:389
Function 'package:front_end/src/fasta/scope.dart:389' not found
$ break scope.dart:389
Script 'scope.dart' is ambiguous
The first one is fixes by this CL: We now allow package uris and thus
no longer think it's a function. In this case it wouldn't have mattered
though, it would just say it was ambiguous, as the last case.
This is caused by the same script being found twice. By using a set
instead of a list, that's fixed too.
Change-Id: I3931c8f2abc0a3a5a77cff516da8767037196bf2
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133593
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
In the VM, a single script can belong to several libraries because of
mixins.
Previously, when setting a breakpoint, it found the function to set the
breakpoint in by taking the "scripts library" (script.FindLibrary) and
then tried to find a function in that library that matched.
Sometimes it found the "wrong" library and the breakpoint thus wasn't
set.
This CL changes that so it goes though all libraries, finds all maching
libraries and tries all of them.
This solves the issue at hand, but might not solve all corner-cases.
This added test demonstrates the resolved case where before it would
just say that it couldn't add the breakpoint on line 13.
Change-Id: Ie44b0dfb4ea3e8de767d1867ec432b2aef429b76
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133592
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
This is a preparation for landing the shared heap CL (it will
temporarily regress the isolate spawning time and memory use).
Once the shared heap CL is in, we can make a follow-up change to actually take
advantage of it by sharing the entire program structure (cl in
preparation by aam@). Once that lands we will enable the isolate groups
in AOT and reduce spawning latency by 10x :)
Regarding testing: We have already made all our isolate related tests in the
SDK run with/without isolate groups (irrespective of the default), so we
should continue to have test coverage in both modes.
Issue https://github.com/dart-lang/sdk/issues/36097
Change-Id: Ib7e5101dfe8527948cd9ceed09d50051b04fb874
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/134280
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
The fuzzer found an issue where clamping negative zero would result in
conflicting answers depending on whether or not the clamp operation had
been optimized in JIT mode.
The MINPS and MAXPS instructions will always return the second operand
if the values being compared are -0 and 0, which is the opposite of what
the C implementation of clamp was doing.
Fixes https://github.com/dart-lang/sdk/issues/40426.
Change-Id: I3afb725bd0c8643758dbe753d863ba93c86ad747
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/134093
Reviewed-by: Alexander Markov <alexmarkov@google.com>
This scope can sneak objects larger than kAllocatablePageSize directly into non-large heap pages.
This scope also holds a heap lock outside of the GC, which is likely to create issues when multiple isolates run on the same heap.
Change-Id: Ic56277c83ffccfe2e1396ed773dd30b8d172bbba
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/134040
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Siva Annamalai <asiva@google.com>
This is a Dart-tailored implementation of the "row-displacement
dispatch table" technique for closed-world instance calls:
All interface targets in the program are grouped into selectors such
that all targets that could potentially be called from the same call
site have the same selector (currently just grouped by name).
Each selector is assigned a selector offset such that offset + classid
is unique for all selector/classid combinations where the class
implements the selector.
At every instance call site that has an interface target (i.e. where
the static type of the receiver is not dynamic), the selector offset +
receiver classid is computed and used as index into a global table of
entry points.
If the receiver can be null (as determined by the front-end TFA and the
VM type propagation), a null check is inserted before the call.
An arguments descriptor is provided (only) for selectors that need it
(those which have type parameters or optional/named parameters).
The dispatch table calls don't need the monomorphic entry code, so for
functions that are only called via dispatch table calls (i.e. never
called dynamically), the monomorphic entry code is left out.
Some future improvements to the table dispatch implementation are
mentioned in https://github.com/dart-lang/sdk/issues/40188
The table dispatch flag is disabled by default in this commit. A
separate commit enables the flag.
Change-Id: Ic2911742b4a2c9a8d3bc7df60605454cbe4c0714
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/126648
Commit-Queue: Aske Simon Christensen <askesc@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
If generated code refers to a constant object (which are old and canonical), a pointer
to it will be embedded in the instruction stream on ia32.
If we now reload and morph this constant object and allocate the
morph'ed copy in new space, the become operation will make RX writable
and the slot in the instruction stream gets updated to point to a new
space object and the code gets added to the remembered set.
The next scavenge operation will scan the remembered set as roots, see
the poitner to a new object, move the object and updates the pointer in
the instruction stream.
=> We hit a SEGV trying to write the RX memory
We can avoid this problem by ensuring to morph old+canonical instances
always into old space on ia32.
Change-Id: Id63eb6ee1735bbb661dbeceb1e8d0c6ac92225ae
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133228
Auto-Submit: Martin Kustermann <kustermann@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
- Handle bytecode for e.g. dartkb-simarm64 does not have source position.
- Use existing GetCallerSp() instead of working across frames.
- Nit: Adds clarifying comments to test.
- Nit: Updates name of non-async-stack test to clarify flags used.
Tested:
- CQ with --lazy-async-stacks on by default.
- vm/dart/causal_stacks and language_2/vm/causal_async_exception_stack_test with current flags.
Bug: https://github.com/dart-lang/sdk/issues/39525
Change-Id: Ie6581a734cdcafbd4fb641bd86bffc03ed241532
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133063
Commit-Queue: Clement Skau <cskau@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This reverts commit 9eb531bde4.
Reason for revert: Bots are red. Some tests are failing.
https://ci.chromium.org/p/dart/builders/ci.sandbox/vm-kernel-precomp-obfuscate-linux-release-x64/6039https://ci.chromium.org/p/dart/builders/ci.sandbox/vm-kernel-precomp-android-release-arm_x64/957
Original change's description:
> [vm] Add support for real unboxed floating point fields in AOT
>
> Non-nullable floating point fields (double, Float32x4, Float64x2)
> are fully unboxed in their classes.
>
> A bitmap for each class was added to the shared class table in order to keep
> track of the pointers of the classes. Since all classes in Flutter Gallery
> have less than 64 fields, the bitmap is represented by a 64 bit integer and
> fields whose offset is more than 64 words are not unboxed.
>
> The instance sizes and field offsets might change between target and host
> in cross-compilation, since the number of words used to store unboxed fields
> may differ.
>
> dart-aot Xeon
>
> SplayLatency -4.62%
> SplayHarderLatency -4.17%
> NavierStokes -2.20%
> Tracer 8.72%
> ParticleSystemPaint 2.90%
> NBodySIMD 8.35%
> NBody 25.59%
>
> With hack TFA to make doubles in Rect/Offset/Size classes in flutter non-nullable:
>
> flutter arm-v8:
>
> gallery total size: -1%
>
> matrix_utils_transform_rect_perspective -16.70% (less is better)
> matrix_utils_transform_rect_affine -31.82% (less is better)
> matrix_utils_transform_point_perspective -24.90% (less is better)
> matrix_utils_transform_point_affine) -27.26% (less is better)
> rrect_contains_bench -4.719% (less is better)
>
> Change-Id: I9ae09c9c3167d99f9efd071a92937aa51093fd1d
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/131824
> Commit-Queue: Victor Agnez Lima <victoragnez@google.com>
> Reviewed-by: Martin Kustermann <kustermann@google.com>
> Reviewed-by: Ryan Macnak <rmacnak@google.com>
> Reviewed-by: Samir Jindel <sjindel@google.com>
TBR=kustermann@google.com,rmacnak@google.com,sjindel@google.com,victoragnez@google.com
Change-Id: Ic73858f6adb7f55c4129d4f46ff4731b378cb634
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/134020
Reviewed-by: Zichang Guo <zichangguo@google.com>
Commit-Queue: Zichang Guo <zichangguo@google.com>
With concurrent sweeping, the amount of capacity freed by a major GC is not known at the time growth is evaluated. This consistently biases the policy into growing more than it would with an accurate capacity from stop-the-world sweeping.
This is more interesting for larger heaps and when there is a tighter memory limit.
For a very large dart2js compile
Max Heap Capacity: 13.057 -> 11.861 GB
Compile Time: 545 -> 472 seconds
For dart2js self-compile
Max Heap Capacity: 1.343 -> 1.385 GB
Compile Time: 43.4 -> 43.5 seconds
Bug: https://github.com/dart-lang/sdk/issues/21364
Change-Id: If047e22a9cd9da6b34c8e499a39fde97427a69b9
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133660
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Siva Annamalai <asiva@google.com>
The following issue can happen:
When we performa a type-test against a type parameter the TTS stub for
the type parameter will call the TTS stub for the type argument of the
parameter (by loading it from the instantiator type arguments).
The (instantiated) type argument does not yet have a specialized TTS but
rather the lazy TTS stub, which will got to the runtime.
The runtime will instantiate the type parameter using the instantiator tav,
which will give it a non-nullable 'String!'. Though the NNBD mode used
for instnatiation is kLegacy, which means the instantiation will
actually return 'String*'.
Now we generate a specialized TTS and install it on 'String*'.
=> The original 'String!' will still have the lazy TTS installed and we
will therefore always go to runtime.
Example:
This code is massively slowed down after 61b4bbcb6b "[vm] Set non_nullable_flag() to true"
main(List<String> args) {
args = args.toList();
while (args.isNotEmpty) {
args.removeAt(0);
}
}
The embedder creates the List<String> when calling main. The embedder
API uses effectively `ObjectStore::string_type`, which is initialized,
after 61b4bbcb6b to 'String!' (which might be a bug because we do
have a specific `ObjectStore::non_nullablestring_type`).
=> For now we disable lazy specialization of non-nullable types - which
makes us fall back to subtype test cache searches in this case (which is
significantly slower).
See b/148528639
Issue https://github.com/dart-lang/sdk/issues/39755
Change-Id: I715fe2a9ee22932ae2d949b5cef93d6afb3176d5
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133981
Reviewed-by: Teagan Strickland <sstrickl@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
TypeRefs which are used to represent recursive types are now skipped
and removed when we need to change nullability of type after
instantiation. This change assumes that we only need to change
nullability of a top-level type and not inner types.
Change-Id: I5e92493991f0e2a1ebeb16535ef9d507054017b4
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133841
Reviewed-by: Siva Annamalai <asiva@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Non-nullable floating point fields (double, Float32x4, Float64x2)
are fully unboxed in their classes.
A bitmap for each class was added to the shared class table in order to keep
track of the pointers of the classes. Since all classes in Flutter Gallery
have less than 64 fields, the bitmap is represented by a 64 bit integer and
fields whose offset is more than 64 words are not unboxed.
The instance sizes and field offsets might change between target and host
in cross-compilation, since the number of words used to store unboxed fields
may differ.
dart-aot Xeon
SplayLatency -4.62%
SplayHarderLatency -4.17%
NavierStokes -2.20%
Tracer 8.72%
ParticleSystemPaint 2.90%
NBodySIMD 8.35%
NBody 25.59%
With hack TFA to make doubles in Rect/Offset/Size classes in flutter non-nullable:
flutter arm-v8:
gallery total size: -1%
matrix_utils_transform_rect_perspective -16.70% (less is better)
matrix_utils_transform_rect_affine -31.82% (less is better)
matrix_utils_transform_point_perspective -24.90% (less is better)
matrix_utils_transform_point_affine) -27.26% (less is better)
rrect_contains_bench -4.719% (less is better)
Change-Id: I9ae09c9c3167d99f9efd071a92937aa51093fd1d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/131824
Commit-Queue: Victor Agnez Lima <victoragnez@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Samir Jindel <sjindel@google.com>
Nullability of declared type parameters is changed to be undetermined or
non-nullable in opted-in libraries depending on the bound.
This fixes sharing of type arguments vectors with superclasses.
Also, while reading bounds of type parameters we may create new
TypeParameter types with different nullabilities, derived from
declared type parameter types. When those derived type parameters
are created, original type parameters may not have bound yet, so
derived type parameter types were getting null bounds.
This change fixes bounds of those derived type parameter types after
bounds of all type parameters were constructed.
Change-Id: I4bf50b383142ae70f194d103501bcbd10bf792e6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133745
Reviewed-by: Siva Annamalai <asiva@google.com>
Reviewed-by: Liam Appelbe <liama@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Dynamic invocation forwarders have the same recognized kind
as the method they are forwarding to.
That causes us to inline the recognized method and not the
dyn: forwarder itself.
It is not safe for user-implementable types. This is triggered
in some of the existing tests when unboxed integer fields are
allowed in AOT.
Issue https://github.com/dart-lang/sdk/issues/37737
Change-Id: Id518c6b13ccd08998ec78384e55c97c64546098a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133561
Commit-Queue: Victor Agnez Lima <victoragnez@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Work towards https://github.com/flutter/flutter/issues/46724
Background:
on the latest versions of iOS today, we have to use mdns to discover the observatory port and authentication code. For a variety of reasons this can fail, leaving us with no way to connect.
An alternative approach is to specify the observatory port and disable the authentication code. If binding to this port fails, however, we would still like to attempt connecting with mdns.
Overview:
This adds a new flag to the dart SDK, --enable-service-port-fallback. This amends the behavior of the vmservice with a specified port. After failing to bind 3 times to a non-zero port, update the port selection to 0.
Results:
Tested locally two dart VMs with the same specified port:
```
jonahwilliams@jonahwilliams0:~/Documents/engine/src/out/host_debug_unopt$ ./dart --enable-service-port-fallback --observe=8080 example.dart
Observatory server failed to start after 1 tries
Falling back to automatic port selection
vm-service: isolate(2785434094928835) 'main' has no debugger attached and is paused at exit. Connect to Observatory to debug.
Observatory listening on http://127.0.0.1:45663/62A5IHjmv9E=/
```
Change-Id: I582ab402a457330928e5b490f9718fcd52b56720
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133434
Reviewed-by: Ben Konyi <bkonyi@google.com>
Commit-Queue: Jonah Williams <jonahwilliams@google.com>
Previously large (>256k) typed arrays would be initialized 3 times:
- The OS initializing the new virtual memory to zero
- Object::InitializeObject initializing to object null
- TypedData::New reinitializing to zero
Change-Id: I432944919ab68aecb2038bba245594bbbaed9093
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/130928
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Vyacheslav Egorov