If enum values have used defined fields that use fromEnvironment constants
the enum values themselves are seen as unevaluated constants when
compiling with dart2js. This means that the enum value are not recognized
correctly in exhaustiveness checking.
This CL changes the enum value representation of the CFE to use the
enum class and the name of the enum element, derived either from the
instance constant or the unevaluated constant expression.
Change-Id: I7d5791a41349dacd20b588f5dbfca37d8755ef79
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/299100
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Jens Johansen <jensj@google.com>
When compiling "compile.dart" the type given to computeNonNull
is already non-nullable in ~42% of the cases, meaning it spends
time not doing anything.
Change-Id: I94598efceb3b46f0ac81ec36814c216d01a9e38b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/298823
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Commit-Queue: Jens Johansen <jensj@google.com>
When compiling compile.dart:
148,192 FunctionType are created via `computeThisFunctionType`.
133,085 of them have no named parameters; 141,507 of them have no type
parameters; 128,357 of them have neither.
This CL makes it not create new lists when not needed
(or create empty iterators and sort empty lists), and makes the lists
it do create not growable.
Change-Id: Ibd5fef458b8a1254aa2f1fab80d0449a2f0bfd94
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/298541
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
This CL splits up `Reference.node` in to the most-often fast-case and the
has-to-load case, making the procedure smaller which makes the VM inline
it. This was "verified" by looking at the VMs optimized flow graph for
`Reference.asClass`.
This CL furthermore reduces the number of "calls" to `Reference.node`
by almost half (a reduction of over 3 mio calls when compiling
`compile.dart`) by "caching" the `.node` call in the `as*` methods.
Change-Id: I7b5497397a11f05fdeaf05d6cc420072d98dc030
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/298101
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Commit-Queue: Jens Johansen <jensj@google.com>
There is an `includeSources` option today to skip writing both source bytes as well as the line starts. Dart2JS needs the line starts to generate source maps from its modified dill. But the actual bytes aren't used passed a certain point. This new option would allow us to exclude the information we don't need from the dill.
For large applications this can reduce the size of the dill by close to 50%.
Change-Id: I5e69370ee30b2fb856320c346b12662893cb595b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/293761
Reviewed-by: Jens Johansen <jensj@google.com>
This uses the required type instead of the lookup type for object
patterns to ensure that extension members are accessed correctly.
Change-Id: I05b4765f1fa3f85acd005a0484349b467cff7c9f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/292703
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
This flag is set when a variable declaration is moved earlier in
order to be available for subsequent code that couldn't contain
its declaration. For instance variables declared in patterns which
needs to be declared before the matching expressions that initialize
them.
TEST=existing expectations tests
Change-Id: I80ab1138f08f725f3e01905c1a57a1483f809328
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/291820
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
This changes the analyzer and CFE to use the real exhaustiveness
checking algorithm. The fallback algorithm is kept and will be
removed once we know that the real exhaustiveness algorithm sticks.
Change-Id: Ic9df92c1ca9f7dec4cbdfa138dc6ed39ef2d4df5
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/288703
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
This removes the "unset" values and uses nullable types instead.
The names and uses of the pattern fields are normalized, using
`matchedValueType`, `requiredType` and `lookupType` consistently
between [Pattern] subclasses.
TEST=existing
Change-Id: Ic79ce17075aa8ce252e1c223b59bef660f32bb7c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/288704
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
This removes the use of [SwitchInfo], [SwitchCaseInfo] and
[ExhaustivessInfo] to pass on information for exhaustiveness checking
and instead uses the AST nodes directly in the constant evaluator.
Change-Id: Ie8d72b03e38334d1be9857794c38ed142dba5783
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/288402
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
The key can only be an Expression but was unneedingly wrapped in
a ConstantPattern. Since the shared type analysis only handles
map pattern keys as expressions, the expected properties of
ConstantPattern were not set. Amongst these were the static type
of the key expression, which is know passed directly to the
handleMapPatternEntry method.
Change-Id: I705fc655e440d534ccc442c9c1359c377955b3b1
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/288282
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
This allows us to better track all kinds of accesses to record
implementation classes in dart2wasm, which generates separate
record implementation class per record shape.
This change also allows us to remove mutable dispatch targets
which were used to implement dynamic accesses to record fields,
and make tracking of record field types more accurate
(record fields are now versioned per shape).
This is also a step towards inferring actual record types.
TEST=pkg/vm/testcases/transformations/type_flow/transformer/records.dart
TEST=pkg/vm/testcases/transformations/type_flow/transformer/records_dart2wasm.dart
TEST=language/records/simple/dynamic_field_access_test
Issue https://github.com/dart-lang/sdk/issues/49719
Fixes https://github.com/dart-lang/sdk/issues/51363
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-nnbd-linux-release-x64-try
Change-Id: Icba62a7ca8cfd8ddbc7f2b7c38aeabbef5caec4b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/286950
Reviewed-by: Slava Egorov <vegorov@google.com>
Reviewed-by: Ömer Ağacan <omersa@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
This propagates information about reported error to the caller of
a shared type analysis. This is used to properly turn errors into
invalid expressions/patterns, as is normally done in the CFE in
face of errors.
Change-Id: Ibb8adedccb8314fabfe18ecaa3559e32ad7267ca
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/286145
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
This adds support for generic types in exhaustiveness checking.
There are two main obstacles:
1) Types that are not Dart subtypes might be related in the context
of exhaustiveness. For instance
sealed class A<X> {}
class B extends A<num> {}
method<T>(A<T> a) {
switch (a) {
case B(): ...
}
}
Here B is not a Dart subtype of A<T> but must still be a subtype
in the context of exhaustiveness since T could be num at runtime.
2) It is non-trivial to compute the subtypes of a generic sealed class
that is both sound and precise enough to be useful. For instance
sealed class A<X> {}
class B extends A<num> {}
class C<X> extends A<X> {}
class D<X extends num> extends A<X> {}
class E<X extends E<X>> extends A<X> {}
Computing the subtypes of A<F> for some type F, we would like to
exclude B (assuming F is not a supertype of num) because it
cannot be inhabit A<F>, and include C<F> so that we recognize
that C<F> exactly covers all C instances of A<F>. We would also
like to include D<F> but if we don't include the added `extends
num` bound, we might conclude that D<num> is _not_ sufficient to
cover D<F>. And what to do about E, can we even come up with a
type that represents the valid values?
The solution is threefold:
1) We add an 'overapproximate' function of types, which
replaces all type variables with their default types. This will compute
A<dynamic> for A<T>, A<int> for A<int>, A<List<dynamic>> for A<List<T>>,
and E<dynamic> for E<T extends E<T>>. This is similar to instantiate-
to-bounds, but is recursive.
We use this to test whether a type without type variables is a potential
subtype. For instance testing B <: overapproximate(A<F>) = A<F> shows
that B _cannot_ be a subtype of A<F>, and testing
B <: overapproximate(A<T>) = A<dynamic> shows that B _can_ be a subtype
of A<T>.
2) For finding subtypes of a sealed type, we recognize the case
when a type is a trivial subtype in which all type variables are passed
directly to the superclass, for instance like C<T> in the example. For
other cases we overapproximate the this type of the subclass.
3) To ensure that the [StaticType] can be subtype in the normal Dart
sense but also handle the overapproximation when computing sealed
subtypes, a new [WrappedStaticType] is added. This bridges the subtype
relation such that for instance B, when created as a sealed subtype
of A<T>, is both a subtype of A<num> (as it normally is) and of A<T>,
which it is by construction.
Change-Id: I9970c46009938ef15625e1193faf916b7544ce0b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/284681
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Paul Berry <paulberry@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
This replaces the [Library] parameter from bounds checks and
type inference methods with a `isNonNullableByDefault` parameter,
enabling usage of these algorithms in library-agnostic code, like
features only available post-nnbd.
Change-Id: Id565d00e1c20d25e7e3388badbb2d7f11c015802
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/284642
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Mayank Patke <fishythefish@google.com>
Before this CL kernel tags had specialized tags that was marked by the
single high bit, then using the three lowest bits for a value:
128 + value = 0b10000xxx
136 + value = 0b10001xxx
144 + value = 0b10010xxx
So the numbers from 128 to 151 is taken by this scheme, but because of
the high bit marking stuff being special we can't really use 152-256.
This CL shifts the specialized tags up so it instead uses the 3 highest
bits as a marker while still using the lower 3 bits for the value:
224 + value = 0b11100xxx
232 + value = 0b11101xxx
240 + value = 0b11110xxx
This takes up 224-247 and leave 248-255 unused. It would let us use
128-223 though.
(If we eventually need more we can probably remove one of the
specialized ranges (SpecializedVariableSet isn't used very much in
previously sampled dill files) and use 4 bits for tagging).
Additionally, a tool to print free tags has been added (via binary.md),
and the "binary version is in sync with VM" test has been prepared
for version > 99.
TEST=Existing tests.
Change-Id: If77b12cee6fc3801628dd67dc40afbb018ec8a61
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/284302
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Commit-Queue: Jens Johansen <jensj@google.com>
This makes it easier to navigate around in editors that can go to
matching braces/parenthesis (e.g. jump from end of constant section to
start).
TEST=Fixes kernel text printer & updates tests.
Change-Id: I6cbf34f5a62e11454e5de4d5fc9a313e0c6cf36c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/283641
Reviewed-by: Jens Johansen <jensj@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
This adds support for record types in the exhaustiveness algorithm.
The original algorithm was based on that record pattern would
match fields on all types, but that is no longer the case. Instead
record patterns only match corresponding to their own type. For this
reason the testing code is updated to create record spaces in relation
to a type. For instance, when the test create a record space {x: B}, it
is know create in relation to a type, say (x: A, y: A), and the create
space will therefore have (x: *, y: *) structure where the y: component
is implicitly Top, similar to how object patterns are used.
Unlike the Dart record types used for type checking and inference, the
record types used for exhaustiveness do not take the field types into
account for its subtype relation. This is avoid conclusions like
(int i, Object o) and (Object o, int i) having no values in common because
their corresponding types (int, Object) and (Object, int) are not subtypes
of each other. Instead, the subtype relation for record types used for
exhaustiveness only use the structure of the record to determine whether
two types are related.
Note though, that fields of a record type still know the type of the field.
This is used when expanded a record type into a space; the field spaces
will be derived from the field types in this case.
Change-Id: I84735d827494bcf384fd5f419d71933830ff5d15
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/283182
Reviewed-by: Paul Berry <paulberry@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
* Add an `targetOS` argument to `pkg/vm`'s `compileToKernel`,
that contains the target operating system's name.
* Add a new `--target-os` command line argument for all binaries
that use `compileToKernel` for clients to provide the target
operating system, if known.
* Add a new`"vm:platform:const"` annotation to certain field and
getters in the Platform class.
This annotation is used to annotate static getters and fields with
initializers where the getter body or field initializer must evaluate
to a constant value if the target operating system is known. This
annotation may be used outside the Platform class and in user code.
For example, this annotation can be used on a static `String` field
that is initialized with one value if `Platform.isWindows` is true
and to a different value if `Platform.isWindows` is false.
Note: If the const functions experimental flag is disabled, then
any annotated static methods can only contain a single expression
whose value is returned. If it is enabled, then the static method
is evaluated as if it is a const function with the special
handling of annotated static fields and getters above.
* Create a VM constant evaluator that evaluates uses of static getters
and fields marked with the above annotations when a target operating
system is provided.
* Use the new VM constant evaluator in the unreachable code elimination
transformer.
TEST=pkg/vm/test/transformations/platform_use_transformer
pkg/vm/test/transformations/unreachable_code_elimination
Change-Id: Ie381de70486a767fd7b1d515fd9e6bb58c6bf090
Bug: https://github.com/dart-lang/sdk/issues/31969
Cq-Include-Trybots: luci.dart.try:pkg-linux-release-try
CoreLibraryReviewExempt: Just adding vm-specific annotations.
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/274386
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
There's been some talks about weird failures on DDC and there was
ideas that this was caused by some sort of leak (of kernel libraries).
I haven't really found any "meaningful" leaks that should be able
to cause that.
What I have found is this:
* a temporary leak where the vm seemingly gets "one behind".
That's fixed here by creating a class for the ddc batch compiler,
having things in field variables etc. It shouldn't influence results.
* a leaks via https://github.com/dart-lang/sdk/issues/51317 --- but
that's just wasted memory, I can't see how that should influence
results.
* a leak via kernels dummy nodes --- again it shouldn't influence
results (unless we're leaving dummy nodes in the output somewhere
--- which, considering I'm removing that in my test by null'ing out
the parent pointer and stuff still working I'm guessing we're not).
* another leak via some ports map after a crash, but that seems
to go away on its own (the VM cleaning it up later than I think it
should?). Again it shouldn't influence anything other than
(temporary) wasted space.
I've wired up the leak testing stuff into the DDC batch mode so that
one can go into "leak test mode" by doing `export DDC_LEAK_TEST="true"`
on the terminal (I think `set DDC_LEAK_TEST="true"` on Windows).
Then one could run test.py, say
```
python3 tools/test.py -t10000 -c dartdevk --nnbd weak -m release \
-r none --enable-asserts --no-use-sdk -j 1 co19/LanguageFeatures/
```
and attach the leak tester via
```
out/ReleaseX64/dart \
pkg/front_end/test/vm_service_for_leak_detection.dart --dart-leak-test
```
Currently this finds - if I recall correctly - 3 leaks (I think all via
the ports map stuff), but it all goes away again on it's own, so likely
another case of the VM somehow getting "one behind".
Change-Id: Idbf057e3aedfe7b256370f90ddf72f1f8e6798a8
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/282027
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Nicholas Shahan <nshahan@google.com>
Johnni Winther