This change extends/fixes the exiting "pattern recognition" which tries
to recognize the pattern
v2 <- LoadClassIdInstr(v1)
BranchIf v2 == IntegerConstant(cid)
Furthermore we start inlining the recognized `ClassID.getID` method very
early in the pipeline. This allows the VM to recognize the above
pattern and insert redefinitions before the actual inlining pass.
Furthermore we special-case two very hot methods in utf8 decoding by
manually having two loops, one of which is guarded by a class-id check
against the _Uint8ArrayView class, which is most common. (In the future
we would like to unify the typed data layouts so we no longer need to
use `ClassId.getID`, thereby also allowing non core library code to use
this).
This improves dart-aot by
* 31%+ for a protobuf decoding benchmark we care about
Issue https://github.com/dart-lang/sdk/issues/31954
Change-Id: I7181bbf096aabe303634fd3b2bff9cc96d69719c
Reviewed-on: https://dart-review.googlesource.com/c/85443
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
This reverts commit d7cf6959bb.
Reason for revert: This causes crashes on the vm-obfuscate builder at https://ci.chromium.org/p/dart/builders/luci.dart.ci.sandbox/vm-kernel-precomp-obfuscate-linux-release-x64
like
FAILED: dartkp-dart_precompiled release_x64 corelib_2/linked_hash_map_test
Expected: Pass
Actual: Crash
Unexpected compile error.
--- Command "vm_compile_to_kernel" (took 11.000189s):
DART_CONFIGURATION=ReleaseX64 /b/s/w/ir/cache/builder/sdk/pkg/vm/tool/gen_kernel --aot --platform=out/ReleaseX64/vm_platform_strong.dill -o /b/s/w/ir/cache/builder/sdk/out/ReleaseX64/generated_compilations/dartkp/tests_corelib_2_linked_hash_map_test/out.dill /b/s/w/ir/cache/builder/sdk/tests/corelib_2/linked_hash_map_test.dart --packages=/b/s/w/ir/cache/builder/sdk/.packages -Ddart.developer.causal_async_stacks=true
exit code:
0
--- Command "precompiler" (took 583ms):
DART_CONFIGURATION=ReleaseX64 out/ReleaseX64/gen_snapshot --snapshot-kind=app-aot-assembly --assembly=/b/s/w/ir/cache/builder/sdk/out/ReleaseX64/generated_compilations/dartkp/tests_corelib_2_linked_hash_map_test/out.S --obfuscate --sync-async --ignore-unrecognized-flags --packages=/b/s/w/ir/cache/builder/sdk/.packages /b/s/w/ir/cache/builder/sdk/out/ReleaseX64/generated_compilations/dartkp/tests_corelib_2_linked_hash_map_test/out.dill
exit code:
-6
stderr:
Warning: This VM has been configured to obfuscate symbol information which violates the Dart standard.
See dartbug.com/30524 for more information.
===== CRASH =====
version=2.2.0-edge.83712405657ff736033380cb24b0bd5a62fc3692 (Tue Nov 27 16:56:07 2018 +0000) on "linux_x64"
si_signo=Segmentation fault(11), si_code=1, si_addr=0xd1400050e5f
Dumping native stack trace for thread 2b71
[0x000056001fc426c7] Unknown symbol
[0x000056001fc426c7] Unknown symbol
[0x000056001fe5763d] Unknown symbol
[0x000056001fe4c99c] Unknown symbol
[0x000056001fe6166a] Unknown symbol
[0x000056001fe59c0c] Unknown symbol
Original change's description:
> [VM] Inline ClassID.getID() eagerly, extend pattern matching logic to recognize it, use it to special case ascii decoding
>
> This change extends/fixes the exiting "pattern recognition" which tries
> to recognize the pattern
>
> v2 <- LoadClassIdInstr(v1)
> BranchIf v2 == IntegerConstant(cid)
>
> Furthermore we start inlining the recognized `ClassID.getID` method very
> early in the pipeline. This allows the VM to recognize the above
> pattern and insert redefinitions before the actual inlining pass.
>
> Furthermore we special-case two very hot methods in utf8 decoding by
> manually having two loops, one of which is guarded by a class-id check
> against the _Uint8ArrayView class, which is most common. (In the future
> we would like to unify the typed data layouts so we no longer need to
> use `ClassId.getID`, thereby also allowing non core library code to use
> this).
>
> This improves dart-aot by
> * 31%+ for a protobuf decoding benchmark we care about
>
>
> Issue https://github.com/dart-lang/sdk/issues/31954
>
> Change-Id: Ia567b92b7e76ff28eda1726deaafda32732ed8f5
> Reviewed-on: https://dart-review.googlesource.com/c/85281
> Reviewed-by: Johnni Winther <johnniwinther@google.com>
> Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
> Reviewed-by: Jenny Messerly <jmesserly@google.com>
> Commit-Queue: Martin Kustermann <kustermann@google.com>
TBR=vegorov@google.com,kustermann@google.com,jmesserly@google.com,johnniwinther@google.com,sra@google.com
Change-Id: I912b0768c32cbb00297ce48db29dbdbea44c14fa
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://dart-review.googlesource.com/c/85441
Reviewed-by: William Hesse <whesse@google.com>
Commit-Queue: William Hesse <whesse@google.com>
This change extends/fixes the exiting "pattern recognition" which tries
to recognize the pattern
v2 <- LoadClassIdInstr(v1)
BranchIf v2 == IntegerConstant(cid)
Furthermore we start inlining the recognized `ClassID.getID` method very
early in the pipeline. This allows the VM to recognize the above
pattern and insert redefinitions before the actual inlining pass.
Furthermore we special-case two very hot methods in utf8 decoding by
manually having two loops, one of which is guarded by a class-id check
against the _Uint8ArrayView class, which is most common. (In the future
we would like to unify the typed data layouts so we no longer need to
use `ClassId.getID`, thereby also allowing non core library code to use
this).
This improves dart-aot by
* 31%+ for a protobuf decoding benchmark we care about
Issue https://github.com/dart-lang/sdk/issues/31954
Change-Id: Ia567b92b7e76ff28eda1726deaafda32732ed8f5
Reviewed-on: https://dart-review.googlesource.com/c/85281
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Reviewed-by: Jenny Messerly <jmesserly@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
This happened using APIs like `Stream.first` that set cancelOnError when
they called `Stream.listen()`. If an error is signaled, Stream internals
will call `cancel()` and wait for that future to complete before sending
the error to the resulting Future. However the `async*` implementation
class did not properly account for that possiblity, and never completed
the cancellation future in that case.
Change-Id: Ieaa35e56ee80208d7946904ea40bfd5f785ef9d6
Reviewed-on: https://dart-review.googlesource.com/c/83467
Commit-Queue: Alan Knight <alanknight@google.com>
Auto-Submit: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Alan Knight <alanknight@google.com>
This adds a method to the internal SDK runtime library that can be
called to clear state when a hot restart is desired.
This also refactors some of the logic around ignoring type errors,
as we need to ensure we can clear those caches. Also I noticed a bug
where the result of isSubtypeOf was assumed to be non-null, but this
was not the case. The new structure should make this more clear.
(Note: I think we may be able to remove a lot of the code for ignoring
type errors soon.)
Change-Id: Icf5072ffe21aefd85e9816ffc2fc29f679839bc4
Reviewed-on: https://dart-review.googlesource.com/c/78324
Reviewed-by: Jake Macdonald <jakemac@google.com>
Commit-Queue: Jenny Messerly <jmesserly@google.com>
mDNS queries (and many other local queries) should be allowed to set
their TTL to 255. This patch enables that on all platforms.
On macOS in particular, a system service reserves the mDNS port. The
socket option SO_REUSEPORT is required in addition to SO_REUSEADDR to
successfully bind there; the flag is also supported on Linux, so this
patch allows it to be used there as well (but prints an error message if
it's used on any other platform).
I could use some guidance as to tests for this - is it support availble
for running tests only on specific platforms?
Bug: https://github.com/dart-lang/sdk/issues/34799 and
Change-Id: I29b620d8ec04343f356a8171bae3d385ddfa9564
https://github.com/dart-lang/sdk/issues/34782
Reviewed-on: https://dart-review.googlesource.com/c/80082
Commit-Queue: Siva Annamalai <asiva@google.com>
Reviewed-by: Siva Annamalai <asiva@google.com>
This change expands typedefs into their underlying function type in the
compiler. Typedefs no longer exist at runtime in DDC, so they now have
similar equality and identity behavior as other function types.
The compiler used to preserve typedefs so they could have a better
toString. But over time that support has been almost entirely lost;
the caching for function/interface types eliminates typedefs, and
DDC+Kernel does not appear to get any TypedefTypes from the common
front end. So in practice typedef types were almost never present at
runtime. Because of this, we can remove the remaining support with very
little user visible effect. This also brings DDC's canonicalization
roughly in line with dart2js for compile-time typedefs, so frameworks
like Angular will be able to continue their use of `identical` for fast
type comparisons.
This CL also fixes DDC's type caching for generic function types
(typeHashCode was incorrect for them), and has some refactoring to
improve names/comments and make the SDK runtime code more clear.
Change-Id: I4e34c0c8f737a8535ba474a4683967ff41bcd3c8
Reviewed-on: https://dart-review.googlesource.com/74664
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Alan Knight <alanknight@google.com>
Fixes#34167. This implements the Dart 2 mixin proposal
(https://goo.gl/KEKQyv) for DDC. When the mixin is applied, a class
is created for the application that extends the correct superclass
and has all of the instance members, so `super` works correctly.
This also fixes a few minor issues in Analyzer's (mostly complete)
implementation:
- InterfaceType.isObject now returns false for Dart 2 mixins.
- Least upper bound calculation recognizes mixins are not Object.
- Interface of the mixin now implements its superclass constraints.
- Mixin superclass constraints are checked against the superclass and
all previously applied mixins (if any); this keeps it working with
the subtype fix above, and also prevents a not-yet-applied mixin
from satisfying the constraint
The language_2/mixin_declaration tests were updated with a few minor
fixes now that we can run Analyzer/dartdevc to test them.
This change implements super mixins for DDC's Kernel backend (DDK)
too. This will be enabled once Kernel adds a flag to recognize which
Class nodes are mixins (vs normal classes).
Change-Id: Ib3c4fcb12de9988345e52d92931196828d8227c3
Reviewed-on: https://dart-review.googlesource.com/74965
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Reviewed-by: Vijay Menon <vsm@google.com>
If a JS class is not found (because it's not loaded yet), dartdevc
should handle this gracefully. However the code assumed that the class
would not be nested in a module that itself has not been loaded. That's
fixed in this change.
Change-Id: I2de8f4b85e8063dea31ea257aca9721702c6ea7a
Reviewed-on: https://dart-review.googlesource.com/72065
Reviewed-by: Vijay Menon <vsm@google.com>
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Improves the error message when calling a null function, by suggesting
to the user that null may be the problem. Also prefixes the errors with
'NoSuchMethodError' to match Dart user expectations (the Dart type that
users can catch is indeed NoSuchMethodError).
Change-Id: Ia9bca5bc2a3e33e83f0c3728b48f7b17f98dcc6a
Reviewed-on: https://dart-review.googlesource.com/66228
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Bob Nystrom <rnystrom@google.com>
There appears to be some inconsistency in how "dart:" part-of URIs are
interpreted by different tools, so this goes back to the old style.
Change-Id: I02f9ce568c0b9859f2fa6239131b0232aa1f8ae4
Reviewed-on: https://dart-review.googlesource.com/66281
Reviewed-by: Bob Nystrom <rnystrom@google.com>
Commit-Queue: Jenny Messerly <jmesserly@google.com>
This is a reland of https://dart-review.googlesource.com/c/sdk/+/62484
with the following changes:
- _NativeSocket.connect now drops references to pending sockets on
an error or successful connection.
- eventhandlers are updated to ignore unset Dart ports on a close
command.
- Test updated to account for new SocketException.
This is the second part of https://dart-review.googlesource.com/c/sdk/+/62484
This CL adds a startConnect method to Socket types that returns
a ConnectionTask object that can be cancelled. Cancelling
a ConnectionTask closes any sockets that were opened for the
connection attempt that are not yet connected to the host.
This allows a closing HttpClient to close sockets for pending
requests whose sockets weren't fully connected yet.
related https://github.com/flutter/flutter/issues/18617
Change-Id: I47fe3564e41197d622079aad4bb644bbdfe0bfe8
Reviewed-on: https://dart-review.googlesource.com/63040
Reviewed-by: Zach Anderson <zra@google.com>
Commit-Queue: Zach Anderson <zra@google.com>
This was difficult to observe in practice, as it required a combination
of two unlikely things:
- an `async` function with no awaits
- a custom Zone with side-effecting scheduleMicrotask callback
The new implementation matches dart2js and the Dart VM's behavior.
Change-Id: I4ff275f37aa93d23643481f00c33aeacb7138b01
Reviewed-on: https://dart-review.googlesource.com/58980
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Vijay Menon <vsm@google.com>
This is the second part of https://dart-review.googlesource.com/c/sdk/+/62484
This CL adds a startConnect method to Socket types that returns
a ConnectionTask object that can be cancelled. Cancelling
a ConnectionTask closes any sockets that were opened for the
connection attempt that are not yet connected to the host.
This allows a closing HttpClient to close sockets for pending
requests whose sockets weren't fully connected yet.
related https://github.com/flutter/flutter/issues/18617
Change-Id: I59c761b06e070d555fc514614079930b69c129dd
Reviewed-on: https://dart-review.googlesource.com/62780
Commit-Queue: Zach Anderson <zra@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Refactors async* implementation class to dart:async and ports it back to
Dart. Improves `await` to avoid zone registration at each await.
Change-Id: I15cac2a40bd56269c8ade1dc0810792c23459ef5
Reviewed-on: https://dart-review.googlesource.com/59260
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Bob Nystrom <rnystrom@google.com>
Reviewed-by: Vijay Menon <vsm@google.com>
This uses optional new/const and `=` in named argument defaults.
All changes are automated, except for:
- utils/dartdevc/BUILD.gn: run DDC build scripts with --preview-dart-2
- pkg/dev_compiler/tool/patch_sdk.dart: add a TODO that Analyzer doesn't
supporting implicit const in libraries.dart
- pkg/dev_compiler/tool/input_sdk/libraries.dart: was not formatted due
to the aforementioned Analyzer bug
- tools/bots/test_matrix.json: run DDC sourcemap suite in Dart 2 mode
- pkg/pkg.status: skip pkg/dev_compiler if running in Dart 1 mode
Change-Id: I9b80ccba0c2cc7b66efc662a0b16562e3660aee3
Reviewed-on: https://dart-review.googlesource.com/60402
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Bob Nystrom <rnystrom@google.com>
Currently we test dartdevk against SDK/packages that were compiled with
dartdevc (DDC+Analyzer). This CL migrates to having those files compiled
with DDC+Kernel. dartdevc testing is unchanged.
Most of the fixes are around things like special optimized annotations
used in our SDK code, that were not understood by DDK. Also some inline
JS type annoations were not correct.
Change-Id: Iccf4427e4b9beffd6d97a4be654253d91f6cb89e
Reviewed-on: https://dart-review.googlesource.com/57800
Reviewed-by: Vijay Menon <vsm@google.com>
String concat now uses + rather than tagged template strings.
Also refactors the JSTypeRep code into a shared location, and fixes a
bug in null inference for op assign expressions in Analyzer backend.
Change-Id: I74fb6950c1d739efb977a03bd639a07afe453eb6
Reviewed-on: https://dart-review.googlesource.com/57522
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Vijay Menon <vsm@google.com>
Also this removes the workaround for dart:html that used a RegExp to
attempt to match inline-JS code that might trigger the problem.
Change-Id: I74b447928aaeaca43b3a9241d42526c284dbeffa
Reviewed-on: https://dart-review.googlesource.com/55504
Reviewed-by: Vijay Menon <vsm@google.com>
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reason for revert: Breaking API change blocks Flutter rolls. Reverting
to reconsider, work around, etc..
Revert "[ VM ] Fixed issue where backing array was deleted from an ExternalTypedData before the ExternalTypedData was used."
This reverts commit 916b9da48d.
Revert "[ VM ] Fixed tests for non-UTF8 paths on Windows / Macos."
This reverts commit 6327fe91c2.
Revert "[ VM ] Fix failing tests on Windows bots after non-UTF-8 paths change."
This reverts commit b51f4b7956.
Revert "[ VM ] Updated non-UTF-8 tests to expect exceptions on OSX/iOS."
This reverts commit bcb36247d2.
Revert "Revert "[ VM ] Updated non-UTF-8 tests to expect exceptions on OSX/iOS.""
This reverts commit 24450c79a5.
Change-Id: I756f74f4acdde168d0984fe130f05f9252366334
Reviewed-on: https://dart-review.googlesource.com/55441
Reviewed-by: Todd Volkert <tvolkert@google.com>
Commit-Queue: Zach Anderson <zra@google.com>
Move implementation details into patch files, it does not belong in the interface.
Actually implement NoSuchMethod.withInvocation in dart2js.
Change-Id: I37049c258067b962d18eff42196e37aa127f0dea
Reviewed-on: https://dart-review.googlesource.com/55166
Commit-Queue: Lasse R.H. Nielsen <lrn@google.com>
Reviewed-by: Leaf Petersen <leafp@google.com>
Reland "Fix issues with FileSystemEntities which have non-UTF8 paths causing exceptions to be thrown when converting to/from a Dart VM String."
This reverts commit 4f549e2900.
Change-Id: Ia27c8c300beb178fa7b08ffe0bfcb2d61f65bbb5
Reviewed-on: https://dart-review.googlesource.com/55271
Reviewed-by: Zach Anderson <zra@google.com>
Commit-Queue: Ben Konyi <bkonyi@google.com>
Martin Kustermann