The former contents of the VM isolate are now included into each isolate group. This makes each isolate group's heap independent, and in particular allows each heap to be allocated to a separate pointer cage (not done in this CL).
The duplicated stubs that allowed PC relative calls are removed, since the originals can now be the target of PC relative calls.
The bootstrapping needing to load an AppJIT or AppAOT snapshot is reduced to allocating the oddballs. The code is entirely dropped in the AOT runtime, but the JIT runtime still has it to allow for flags to affect the compilation of the stub code. Further refactoring might be able to remove this for the JIT runtime too, with only gen_snapshot knowing how to bootstrap.
Class serialization no longer distinguishes predefined classes.
The page containing null is marked as never-evacuate. null, false and true must not move because the compiler relies on their low bits having certain patterns for some optimizations. (Previously, the entire VM isolate heap never moved.)
Compaction is disabled for IA32. Due to register pressure, some stub calls must not use a scratch register and embed the address of Code.
The page containing the call-through-safepoint stub is frozen when running with --write-protect-code and the stub is created at runtime (instead of loaded from an AppJIT or AppAOT snapshot). This stub must remain executable even during a safepoint, as a foreign call might during return during a safepoint and only block after the stub directs it to the runtime.
The snapshot symbols are renamed to kDartSnapshotData and kDartSnapshotText. There is no need to distinguish the VM isolate's snapshot, and snaphots are per isolate group not per isolate. Aliases with the old names are added to ease migration.
Some global flags that were automatically set based on the VM isolate's snapshot are now isolate group flags and automatically set by the isolate group's snapshot.
TEST=ci
Change-Id: Iee82016057d609112e9b021d178fc3d4d18b5044
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/500621
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Tess Strickland <sstrickl@google.com>
SLSA-Policy-Verified: SLSA Policy Verification Service <devtools-gerritcodereview-exitgate@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Property-inducing augmentations were linked to the existing field or
top-level variable element before creating the getter and setter
fragments that the augmentation itself induces. This left augmentation
fragments without their corresponding accessor fragments.
Create induced getter and setter fragments for every field and top-level
variable fragment, including augmentations. When an accessor element
already exists, append the new fragment to it; otherwise, create the
accessor element independently so that augmenting getter-only and
setter-only properties works correctly.
Track the relationship at the fragment level in both directions, from
each property-inducing fragment to its induced accessors and from each
accessor fragment back to its inducing variable. Serialize these links
in summaries, expose them through the analyzer element API, and bump the
summary data version for the format change.
Change-Id: Iede43515564326182d8195bd872abf83345bbd72
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/504040
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
It is disabled when augmentations feature is enabled.
This is not a breaking change, because CFE did report this error.
The relevant reason is indirect, not a single sentence saying “on A
implements A is forbidden”.
Spec chain:
1. In dartLangSpec.tex:6019, a mixin declaration is modeled as:
`mixin N<T...> on T1, ..., Tn implements I1, ..., Ik { ... }`
2. In dartLangSpec.tex:6065, the mixin interface M_I is defined as if
by:
`abstract class N<T...> implements T1, ..., Tn, I1, ..., Ik { ... }`
and dartLangSpec.tex:6078 says it is a compile-time error for the mixin
if that synthetic class declaration would be a compile-time error.
3. In dartLangSpec.tex:5189, class superinterfaces say:
It is a compile-time error if two elements in the type list of the
IMPLEMENTS clause of a class C specifies the same type T.
So:
`mixin M on A implements A {}`
is checked like:
`abstract class M implements A, A {}`
which violates the duplicate implements type rule.
Change-Id: I29e5fa17c47d111346ef6a6e321ab5dbab985ad1
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503501
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Paul Berry <paulberry@google.com>
In the recent change,
https://dart-review.googlesource.com/c/sdk/+/503040, the asynchronous
nature of the change required adding _new_ hooks to track that the
plugin isolate is analyzing or not. I missed removing the old hooks.
The result is that an invocation of `dart analyze` will get a notification from the plugin isolate that it is analyzing, and then
_immediately_ a notification indicating that it is not.
So this change removes the old hooks.
Change-Id: Idc871d25c043ee1be94030729a1fd8226bdc753f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/504140
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Commit-Queue: Samuel Rawlins <srawlins@google.com>
This separates VariableDeclaration from Statement. VariableDeclaration no longer implements Statement and variable declared in a block or in a for-statement are now wrapped by a VariableStatement.
Currently there are two VariableStatement implementations; LegacyVariableStatement for variables in the current model, called LegacyVariable, and VariableInitialization for variables used in the new, still experimental, encoding that supports scope computation.
This CL is a step towards realigning the AST nodes to the new model in which each kind of variable has its own distinct subclass. (LocalVariable, PositionalParameter, NamedParameter, SyntheticVariable, etc.)
Note that it is not the intent to use VariableStatement in ForStatement going forward but that will be handled in a follow-up.
TEST=existing.
Change-Id: I5b309cd62c9b138f95b74fb054686edffa49a393
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/502681
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
Reviewed-by: Nicholas Shahan <nshahan@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Adds a benchmark exercising int.oneBitCount and int.trailingZeroBitCount
(landed in CL 498041) and a representative set of Int32x4 SIMD bitwise
kernels, through a small bit-array implementation built on a Uint32List
word store.
Kernels:
- cardinality: popcount across the array.
- forEachSetBit: Brian-Kernighan iteration via ctz.
- select(k): position of the k-th set bit using popcount + ctz.
- complementCardinality: popcount of bitwise-NOT per word.
- totalBitLength: sum of int.bitLength per word.
- intersection / union / xor / difference: pair-op bitwise kernels using
Int32x4 SIMD with a scalar tail.
- complement: single-input bitwise NOT via Int32x4 XOR-with-all-ones.
Each kernel has a Swar baseline using portable bit-twiddle code and an
Accelerated variant that uses the new int getters or Int32x4 SIMD, so
the speedup attributable to the hardware/intrinsic path is visible
directly. main() first runs a correctness check that asserts both
implementations agree across a representative range of sizes and bit
densities, then reports timings via the standard BenchmarkBase harness.
Work towards https://github.com/dart-lang/sdk/issues/1053 (efficient
BitSet implementation).
Bug: https://github.com/dart-lang/sdk/issues/52673
Change-Id: I5ae5bc0b9d07f6de3e11907ac1fae80f02e77119
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503020
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Auto-Submit: Modestas Valauskas <valauskasmodestas@gmail.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Treat field fragments marked as augmentations like other augmented
members when collecting local member names. Do not add their getter and
setter fragments to the instance or static duplicate scopes, because
they augment existing declarations rather than introduce new local
members.
This avoids secondary duplicateDefinition and conflicting member
diagnostics when a field augmentation targets an existing declaration.
Non-augmenting declarations in augmentation blocks are still checked as
new declarations, so real duplicates continue to be reported.
We don't have yet checks for completed getter / setter from fields and
variables, so tests miss several errors.
Change-Id: I6901aea1770891917c966c11b59cb647aaf4ed86
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503840
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Paul Berry <paulberry@google.com>
This will allow use of Dart 3.13 features (such as primary
constructors) within the SDK itself.
Note that some generated files required benign updates due to
formatter behavior changes.
Change-Id: I595554e9cff3dbaa3f11da407d6c26036a6a6964
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503662
Auto-Submit: Paul Berry <paulberry@google.com>
Reviewed-by: Samuel Rawlins <srawlins@google.com>
Commit-Queue: Samuel Rawlins <srawlins@google.com>
Report new diagnostics when an introductory function, member, or factory
constructor is still incomplete after applying all augmentations.
Keep the existing missing-body diagnostics for declarations that have no
augmentations, but report augmentation-specific diagnostics when an
augmentation chain exists and none of the fragments provides a body or
factory redirection.
Move the checks into error verification so that extension and extension
type members can participate in augmentation completion before reporting
the existing abstract-member diagnostics. Also suppress the
corresponding shared parser diagnostic when it is reported by the
verifier.
Consolidate body-related tests by declaration shape instead of by
individual diagnostic. This keeps missing bodies, external bodies,
augmentation completion, and already-complete checks side by side,
making the interaction between these rules easier to review and extend.
Move the constructorAlreadyComplete coverage from its dedicated test
file into constructor_body_test.dart, and add the factory body
completeness cases there as well. Add executable_body_test.dart for
function and member body coverage, including top-level declarations,
static members, extension members, and extension type members.
Add diagnostic definitions and fix-status entries for the new
diagnostics.
Change-Id: I9ee803c1e767ff47a608c86413ef7ffc71518cfd
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503540
Reviewed-by: Paul Berry <paulberry@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Teach the analysis index and referenced-name computation to recognize
`// [diag.foo]` expectation comments embedded in string literals in
analyzer tests. Resolve `foo` through the analyzer diagnostic library
and record it as a qualified reference to the diagnostic variable.
This lets reference search find diagnostics that are used only in test
expectation strings, when developing analyzer itself.
Change-Id: I91fa020ccd2cfe49fa6e890c8f6796a5b8ca4d1f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503663
Reviewed-by: Paul Berry <paulberry@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
The correction producer implements a fix. Pending discussion we might
choose to extend it to produce an assist, either in addition to the fix
(if we keep the lint) or instead of a fix (if we drop the lint).
Whatever we choose, I'd like to do that in a separate CL because I want
this to be available for testing purposes as soon as possible.
Change-Id: Icdd17ba0130149fc785c196be69e9ef990529800
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503520
Reviewed-by: Keerti Parthasarathy <keertip@google.com>
Commit-Queue: Brian Wilkerson <brianwilkerson@google.com>
I'm honestly unsure why this is not reported by my IDE, or CI. But it
is reported internally in Google. I'm sure it has to do with the SDK
starting to add some extension methods in the core libs, which were
previously only available in package:collection. But this is still safe,
as analyzer's minimum SDK is Dart 3.11.0.
Change-Id: Ibbb0078e463df03b31749296e1d2f474afb183ec
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503582
Auto-Submit: Samuel Rawlins <srawlins@google.com>
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Fixes https://github.com/dart-lang/sdk/issues/62649
This change overhauls what the PluginServer does when it receives requests from DAS like watch events, overlay changes, setting context roots, and setting priority files, to use the AnalysisDriver APIs. These APIs provide a much smarter, more fine-grained mechanism for determining which files need to be re-analyzed.
This requires a lot of changes that really don't seem obvious (or just,
to me, they are not intuitive), because they relate to Streams being
processed, drained, read, etc. So I'll summarize as much as I can:
* In order to track that results coming from AnalysisDriver are
definitely for the same AnalysisSession we are concerned with, at
any given time, we add `_filesBeingAnalyzed` and
`_filesBeingResolved`. These two maps are cleared any time the plugin
isolate is notified of changes, so there is not a memory concern.
* We add a `waitForIdle` method which is just used by tests
occasionally, in order to ensure analysis is complete.
* We tackle a related bug mentioned in the issue thread regarding
adding or removing overlays without changing the source contents.
There are now a few checks that old-content is not equal to
new-content.
* Due to the more async nature of reading events off of the analysis
driver, we have to be more particular in tests about sending
`AnalysisSetContextRootsParams` and reading from the notification
stream, so there are a lot of changes in tests to re-position these
calls.
This results in massive savings in re-analysis time. Using the example in https://github.com/dart-lang/sdk/issues/62649, typing in the
`build_runner/lib/src/logging/build_log_logger.dart` file, I see the
following:
* Before this fix, every keystroke results in 183 files being
re-analyzed by the plugin (always 183).
* After this fix:
* keystrokes that do not result in a summary change (or maybe
fine-grained deps analysis), like whitespace, results in 1 file
being re-analyzed (the file I'm editing).
* A keystroke that results in slightly changed summary, like changing
an import to something invalid, results 2 or 4 files being
re-analyzed.
* A keystroke that changes the name of class BuildLogLogger (large
change to summary) results in 9-10 files being re-analyzed.
Change-Id: Ie35053b0d90457b5b4b53bc8803188def0308d2a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503040
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Commit-Queue: Samuel Rawlins <srawlins@google.com>
This file is no longer published and has no other value. Removing it
removes one step from the process of writing a new lint, which improves
our velocity.
Change-Id: Ie81b914bfa59aa8ec2c8fb2921eec2e9963bd601
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/503620
Reviewed-by: Samuel Rawlins <srawlins@google.com>
Commit-Queue: Brian Wilkerson <brianwilkerson@google.com>
The session log normalizer replaces known paths/URIs in JSON but doesn't take into account different URI encoding between the client and the server. For example VS Code will encode ampersands whereas Dart does not:
```
file:///c:/uri&encoding&quirks
file:///c:/uri%26encoding%26quirks
```
This means not all file URIs are correctly normalized.
Adding additional groups for each potentially-encoded characters make the regex many times slower (the benchmark test here goes from around 25ms to over 1s per iteration), so instead this change has the normalizer accept the original JSON map and uses jsonEncode()s `toEncodable` option to normalize any URIs (by converting them to their file paths and then encoding using Dart's Uri class) so they will always be consistent before the replacement.
(I tried doing the replacement also in `toEncodable`, but invoking the regex many times also slowed things down a lot).
There is a small time increase (2-3ms) for a payload of 2MB. The "before" times quoted here are slightly higher than previously quoted, but that's because `jsonEncode()` was previously done inside `SessionLoggerFileSink` (and therefore excluded from the timings before), but is now done inside the normalizer to allow normalizing the URI escaping.
Replacing 250 paths in payload of 2097152 bytes
Iteration #1, First: 57ms, Rest: 40ms
Iteration #2, First: 49ms, Rest: 41ms
Iteration #3, First: 45ms, Rest: 41ms
Iteration #4, First: 47ms, Rest: 40ms
Iteration #5, First: 40ms, Rest: 40ms
Replacing 250 paths in payload of 2097152 bytes
Iteration #1, First: 59ms, Rest: 43ms
Iteration #2, First: 53ms, Rest: 44ms
Iteration #3, First: 52ms, Rest: 44ms
Iteration #4, First: 49ms, Rest: 43ms
Iteration #5, First: 49ms, Rest: 43ms
Fixes https://github.com/dart-lang/sdk/issues/63331
Change-Id: Ice2dc7ceceaa6c08e2ff634d7564efe9f0f7de44
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/502940
Reviewed-by: Keerti Parthasarathy <keertip@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Commit-Queue: Keerti Parthasarathy <keertip@google.com>
This updates the IOHandle type in eventhandler_fuchsia to use
zx_port_cancel_key to cancel outstanding asynchronous waits instead of
the (older) zx_port_cancel. The cancel_key operation requires only the
key used to register the initial wait and not the originating handle.
This means that the IOHandle operation no longer has to store the
original handle or worry about its lifetime.
Tested: Patched locally in a test embedder using this logic
Change-Id: Iec0ec632a7dd438a34e482a12b5b298be56ce476
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/501840
Reviewed-by: Zijie He <zijiehe@google.com>
Auto-Submit: James Robinson <jamesr@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Ryan Macnak