For some reason, these assertions fire when snapshots are trained on RBE
inside a docker container. It's potentially a race condition if the
timer resolution is low enough or RBE is lying about the time. However,
there is no need to check if the start time is non-zero since there's
already a running_ variable protecting against stopping times that
haven't been started.
TEST=CQ, locally, and on RBE
Bug: b/296994239
Change-Id: I2a5b36d0b1cb9b939e94fe6aabb5f41fec077f2a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/342401
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Jonas Termansen <sortie@google.com>
Allow annotating top-level or static fields with `@Native` to create
fields backed by native memory.
By using the `_addressOf` operator implemented in the VM, these fields
can be implemented in the CFE by replacing them with accessors looking
up the pointer and then using existing methods to load and store the
value.
Closes https://github.com/dart-lang/sdk/issues/50551
TEST=tests/ffi/native_assets/asset_*_test.dart
TEST=pkg/analyzer/test/src/diagnostics/ffi_native_test.dart
CoreLibraryReviewExempt: VM & dart2wasm only feature
Change-Id: I61dccc88076723d6a6ba02d7fd848b18e4caf780
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/338020
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Reviewed-by: Lasse Nielsen <lrn@google.com>
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
STW marking is no longer O(new-space), so there's no longer a reason to delay finalizing marking hoping for a scavenge to make new-space mostly empty.
TEST=ci
Change-Id: Ie782e88852714d30e0c75aa9aecac62e56c434ce
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/319880
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
This reverts commit 49f6657c37.
Reason for revert: This caused a performance regression in Android startup time.
Original change's description:
> Call tzset() in OS::Init() on Android so localtime_r() has populated timezone information.
>
> POSIX and glibc do not guarantee that timezone setup is done by localtime_r().
>
> Android <= 7.1.1 did *not* do timezone setup in localtime_r(), which means that users will see their timezone as "GMT".
>
> macOS/iOS seem to guarantee that timezone setup is done in localtime_r() but the wording is vague:
>
> localtime_r() and gmtime_r() functions provide the same functionality
> as localtime() and gmtime()...
>
> N.B.: localtime() is guaranteed to do timezone setup
>
> Bug:https://github.com/dart-lang/sdk/issues/53276
> Change-Id: I8f6abfbfcf07a61cb341c1dbbd8323440cb8f581
> Tested: manually tested on Android 7.1.1
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/340281
> Reviewed-by: Siva Annamalai <asiva@google.com>
> Commit-Queue: Brian Quinlan <bquinlan@google.com>
Bug: https://github.com/dart-lang/sdk/issues/53276
Change-Id: I02ae79d5c3a3f5a3a9003fae71b81ff5a10ac272
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/341338
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Brian Quinlan <bquinlan@google.com>
Reviewed-by: Derek Xu <derekx@google.com>
Instead of implicitly creating closures for FFI
asFunction/lookupFunction APIs in the VM, now they are explicitly expressed in the kernel AST. That makes it possible to analyze
them in TFA.
FFI calls from Dart code into native are now performed in the following way:
```
block {
Pointer #ffiTarget0 = target;
@pragma('vm:ffi:call-closure', _FfiCall<Int32 Function(Int32)>(isLeaf: false))
#ffiClosure0(int arg1) {
_nativeEffect(arg1);
return _ffiCall<int>(#ffiTarget0);
}
} =>#ffiClosure0;
```
_ffiCall method is recognized by the VM and its call is replaced
directly with FFI calling sequence. _ffiCall uses closure
parameters implicitly. No extra trampolines are generated for FFI calls.
TEST=existing
Fixes https://github.com/dart-lang/sdk/issues/54172
Issue https://github.com/dart-lang/sdk/issues/39692
CoreLibraryReviewExempt: Implementation change only.
Change-Id: I92b3ff7391470686151ad0807e2cdbbf1a69d256
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/339662
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
- Initial and final marking no longer visit all of new-space, reducing the STW pause for major GC.
- A scavenge during concurrent marking must forward / filter objects in the marking worklist that are moved / collected, increasing the STW pause for minor GC.
- Unreachable intergenerational cycles and weak references are collected in the next mark-sweep instead of first requiring enough scavenges to promote the whole cycle or weak target into old-space.
- Artificial minor GCs are no longer needed to avoid memory leaks from back-to-back major GCs.
- reachabilityBarrier is now just a count of major GCs.
TEST=ci
Change-Id: Ic7754e8d972763654eae2b7faa8670735d9cda3f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/340644
Reviewed-by: Siva Annamalai <asiva@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Enables passing inner pointers to typed data in FFI leaf calls.
This works for typed data's in the Dart heap, external typed datas
(constructed from `Pointer.asTypedList`), and typed data views.
Notable implementation details:
1. The Dart signature is used in the Marshaller now. This means it
needs to keep track of whether there's a pointer argument in the
signature (`asFunction`) or not (`@Native`).
2. Unwrapping is done in `FfiCallInstr::EmitNativeCode` before moving
the arguments to their native location. This ensures we can use
the assembler logic to load the `TypedDataBase::data` field.
3. The `XXXList` user visible classes don't have predefined cids.
So the implementation uses symbols for comparison.
4. The type checking logic takes `isLeaf` as input to reject typed
data. This leads to an error message about the type not accepted.
Alternatively, we could consider adding an error message that
specifically says the function should be leaf.
5. To cover all calling convention variants, tests are generated with
up to 20 arguments.
TEST=pkg/analyzer/test/src/diagnostics/ffi_unwrap_typed_data_test.dart
TEST=tests/ffi/unwrap_typeddata_generated_native_test.dart
TEST=tests/ffi/unwrap_typeddata_generated_test.dart
TEST=tests/ffi/vmspecific_static_checks_typeddata_test.dart
Closes: https://github.com/dart-lang/sdk/issues/44589
Change-Id: Ia78f18bf3238d42ac6882929b441f6dc432fcefe
Cq-Include-Trybots: luci.dart.try:vm-aot-android-release-arm64c-try,vm-aot-android-release-arm_x64-try,vm-aot-linux-debug-x64-try,vm-aot-linux-debug-x64c-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-obfuscate-linux-release-x64-try,vm-aot-optimization-level-linux-release-x64-try,vm-aot-win-debug-arm64-try,vm-aot-win-debug-x64c-try,vm-aot-win-release-x64-try,vm-appjit-linux-debug-x64-try,vm-asan-linux-release-x64-try,vm-checked-mac-release-arm64-try,vm-eager-optimization-linux-release-ia32-try,vm-eager-optimization-linux-release-x64-try,vm-ffi-android-debug-arm-try,vm-ffi-android-debug-arm64c-try,vm-ffi-qemu-linux-release-arm-try,vm-ffi-qemu-linux-release-riscv64-try,vm-fuchsia-release-x64-try,vm-kernel-linux-debug-x64-try,vm-kernel-precomp-linux-release-x64-try,vm-linux-debug-ia32-try,vm-linux-debug-x64-try,vm-linux-debug-x64c-try,vm-mac-debug-arm64-try,vm-mac-debug-x64-try,vm-msan-linux-release-x64-try,vm-reload-linux-debug-x64-try,vm-reload-rollback-linux-debug-x64-try,vm-ubsan-linux-release-x64-try,vm-win-debug-arm64-try,vm-win-debug-x64-try,vm-win-debug-x64c-try,vm-win-release-ia32-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/338620
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
POSIX and glibc do not guarantee that timezone setup is done by localtime_r().
Android <= 7.1.1 did *not* do timezone setup in localtime_r(), which means that users will see their timezone as "GMT".
macOS/iOS seem to guarantee that timezone setup is done in localtime_r() but the wording is vague:
localtime_r() and gmtime_r() functions provide the same functionality
as localtime() and gmtime()...
N.B.: localtime() is guaranteed to do timezone setup
Bug:https://github.com/dart-lang/sdk/issues/53276
Change-Id: I8f6abfbfcf07a61cb341c1dbbd8323440cb8f581
Tested: manually tested on Android 7.1.1
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/340281
Reviewed-by: Siva Annamalai <asiva@google.com>
Commit-Queue: Brian Quinlan <bquinlan@google.com>
This adds the `Native.addressOf` function, which takes a constant tear-
off of a method annotated with `@Native` as a parameter and returns a
pointer to the underlying C function.
The CFE will resolve these calls in two steps: First, the existing
transformer for `@Native` methods adds a pragma describing the fully-
resolved native annotation (with the asset id inferred from the library
if not given explicitly). Then, the FFI use sites transformer rewrites
calls to `Native.addressOf` by searching for this pragma on the passed
function and passing the `Native` constants to `Native._addressOf`. The
latter method is implemented in the VM, which can re-use existing parts
already used for `@Native` calls.
An alternative implementation strategy would have been to forward
`addressOf` calls to `Native.ffi_resolver_function` directly without
any special consideration in the VM. However, the chosen approach makes
it easier to support static linking in the future, as this requires
unresolved symbols in the generated assembly.
Closes https://github.com/dart-lang/sdk/issues/50552
TEST=pkg/vm/testcases/transformations/ffi/ffinative.dart
TEST=tests/ffi/native_assets/asset_*_test.dart
TEST=tests/ffi/vmspecific_static_checks_ffinative_test.dart
TEST=pkg/analyzer/test/src/diagnostics/ffi_native_test.dart
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-aot-linux-debug-x64c-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-obfuscate-linux-release-x64-try,vm-aot-optimization-level-linux-release-x64-try,vm-aot-win-debug-arm64-try,vm-aot-win-debug-x64c-try,vm-aot-win-release-x64-try,vm-kernel-precomp-linux-release-x64-try
CoreLibraryReviewExempt: VM & dart2wasm only feature
Change-Id: Ic8e3a390146dffd44c95578f975a4472db79a0ee
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/333920
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Protect VM compiler code from crashing when incorrect
type arguments are passed to FFI _asFunctionInternal.
Also, improve front-end FFI checks to reject Function
as a Dart type corresponding to a NativeFunction<...>
(it should be an actual function type).
TEST=tests/ffi/vmspecific_static_checks_test.dart
Fixes https://github.com/dart-lang/sdk/issues/52730
Change-Id: I02c012bb76d5c94f551da2c725a476b68c0f9db4
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/339109
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Allows FFI to reuse FlowGraphCompiler::EmitMove in contexts where the delta between SP and FP isn't accurately described by StackSize.
TEST=dart2js codesize unchanged
Change-Id: Idf0643af3debfc77564672d6d16491990b87611c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/339040
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Conditional constant propagation doesn't update environments and may
also replace redefinition instructions with constants
without updating environment uses of their original definitions.
This causes discrepancy between arguments of a call and its environment.
It is harmless and corresponding check in the flow graph checker can be
relaxed.
TEST=runtime/tests/vm/dart/regress_54128_test.dart
Fixes https://github.com/dart-lang/sdk/issues/54128
Change-Id: I6e68e0525b7f110400435ac071934dd9a02eebd4
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/338941
Auto-Submit: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
Implementation of stores and their inlinings had some
stuff left over their since Dart 1:
* No need to insert null-checks (in sound null-safety mode).
* Since Dart 2 there is no need to insert (speculative)
cid checks. Inputs are guaranteed to be a value of a
supported implementation type. Inserting narrow speculative
checks for Smis is actually leads to worse code in JIT.
* There is no need to convert incomming integer values to
smaller representation - the store will take care of it. This was
left over from Dart 1 times when incomming integer could be _Bigint.
TEST=existing tests
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-release-simarm_x64-try,vm-aot-linux-release-arm64-try,vm-aot-linux-release-x64-try,vm-aot-mac-product-arm64-try,vm-aot-linux-product-x64-try,vm-aot-linux-debug-x64c-try,vm-aot-linux-debug-x64-try,vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv64-try,vm-aot-android-release-arm_x64-try,vm-aot-android-release-arm64c-try
Change-Id: I72cdaaecc524f1dccc63825df4f7b71241ab47a0
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/338600
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
When generating code for deferred units compiler
can't emit PC relative call to the shared Mint allocation
stub. This causes compiler to emit an indirect call
through a Code object. Such calls clobber CODE_REG
which is actually an allocatable register.
Fix this by using non-allocatable register instead of
CODE_REG when generating indirect calls through
Code object in AOT mode. Callees don't expect
anything useful in CODE_REG anyway because AOT
calling convetion does not use it.
TEST=vm/cc/{BranchLinkPreservesRegisters,JumpAndLinkPreservesRegisters,CallCodePreservesRegisters}
Bug: b/242559057
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-release-arm64-try,vm-aot-linux-release-simarm_x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-product-x64-try,vm-aot-obfuscate-linux-release-x64-try,vm-aot-optimization-level-linux-release-x64-try,vm-aot-linux-debug-simriscv64-try,vm-ffi-qemu-linux-release-riscv64-try
Change-Id: Ib1fdc1c104d0269d41bb1ab9cbe292ad28c7cd49
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/338127
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
The value of best_token_pos was
exact_token_pos.Pos() - (requested_column - *best_column), but this
value is guaranteed to be the same as best_fit_pos.
TEST=CI
Change-Id: I96f21cc45ebbac94dfe35fa39af922a227e1fde6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/338080
Commit-Queue: Derek Xu <derekx@google.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
This test needs a larger backtracking stack then we provide
by default in AOT mode. Allow controlling stack size via
a command line flag regexp_backtrack_stack_size_kb.
Additionally cleanup types in the interpreter a bit: generated
code uses Int32List for both register and backtracking stack.
But the interpreter was using intptr_t for backtracking stack
and int32_t for registers. Align these types to use int32_t
everywhere and add a bit of sanity checking to ensure that
we don't try to use regexp engine with strings which are
too big to fit into int32_t range.
TEST=corelib/regexp/stack-overflow
Change-Id: I5c13c4b47d5d478b683f0450d3c0ce65f6961cff
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/338141
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Original change in patchset 1.
The reland contains a fix for handles. The objects passed in handles
must live on the stack (or in the heap) so that that memory location
can be used as a handle.
This CL introduces a new allocation policy `RequiresRegister` which
forces constants to be spilled to the stack during register
allocation.
Note: Parameters to FfiCall instructions are currently not used in a
way that can make the parameters assigned to registers. So only
constants are treated currently. If we have a way to reproduce params
being assigned to registers, then we can force spilling for
non-consts as well in the register allocator.
Original change's description:
> [vm/ffi] Optimize `@Native` calls
>
> This CL removes static fields for storing the `@Native`'s function
> addresses. Instead, the function addresses are stored in the object
> pool for all archs except for ia32. ia32 has no AOT and no AppJit
> snapshots, so the addresses are directly embedded in the code.
>
> This CL removes the closure wrapping for `@Native`s. Instead of
> `pointer.asFunctionInternal()()` where `asFunction` returns a closure
> which contains the trampoline, the function is compiled to a body
> which contains the trampoline `Native()`. This is possible for
> `@Native`s because the dylib and symbol names are known statically.
>
> Doing the compilation in kernel_to_il instead of a CFE transform
> enables supporting static linking later. (The alternative would have
> been to transform in the cfe to a `@pragma('vm:cachable-idempotent')`
> instead of constructing the IL in kernel_to_il.
>
> To enable running resolution in ia32 in kernel_to_il.cc, the
> resolution function has been made available via
> `runtime/lib/ffi_dynamic_library.h`.
>
> Because the new calls are simply static calls, the TFA can figure
> out const arguments flowing to these calls. This leads to constant
> locations in the parameters to FfiCalls. So, this CL also introduces
> logic to move constants into `NativeLocation`s.
TEST=tests/ffi/vmspecific_ffi_native_handles_test.dart
TEST=tests/ffi/function_*_native_(leaf_)test.dart
TEST=pkg/vm/testcases/transformations/ffi/ffinative_compound_return.dart
Closes: https://github.com/dart-lang/sdk/issues/47625
CoreLibraryReviewExempt: Internal FFI implementation changes
Change-Id: Ia1401b335524dcbf50c663a8770d9a02802923df
Cq-Include-Trybots: luci.dart.try:vm-aot-android-release-arm64c-try,vm-aot-android-release-arm_x64-try,vm-aot-linux-debug-x64-try,vm-aot-linux-debug-x64c-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-obfuscate-linux-release-x64-try,vm-aot-optimization-level-linux-release-x64-try,vm-aot-win-debug-arm64-try,vm-aot-win-debug-x64c-try,vm-aot-win-release-x64-try,vm-appjit-linux-debug-x64-try,vm-asan-linux-release-x64-try,vm-checked-mac-release-arm64-try,vm-eager-optimization-linux-release-ia32-try,vm-eager-optimization-linux-release-x64-try,vm-ffi-android-debug-arm-try,vm-ffi-android-debug-arm64c-try,vm-ffi-qemu-linux-release-arm-try,vm-ffi-qemu-linux-release-riscv64-try,vm-fuchsia-release-x64-try,vm-kernel-linux-debug-x64-try,vm-kernel-precomp-linux-release-x64-try,vm-linux-debug-ia32-try,vm-linux-debug-x64-try,vm-linux-debug-x64c-try,vm-mac-debug-arm64-try,vm-mac-debug-x64-try,vm-msan-linux-release-x64-try,vm-reload-linux-debug-x64-try,vm-reload-rollback-linux-debug-x64-try,vm-ubsan-linux-release-x64-try,vm-win-debug-arm64-try,vm-win-debug-x64-try,vm-win-debug-x64c-try,vm-win-release-ia32-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/333841
Reviewed-by: Slava Egorov <vegorov@google.com>
This language feature allows the user to declare a static type using
`extension type` syntax, for example:
extension type IdNumber(int i) {
operator <(IdNumber other) => i < other.i;
bool isValid(Some parameters) => ...;
}
This behaves similarly to a "wrapper" class:
class IdNumber {
final int i;
IdNumber(this.i);
operator <(IdNumber other) => i < other.i;
bool isValid(Some parameters) => ...;
}
However, at runtime, no wrapper objects are created; instead, an
instance of the extension type is represented directly by its
"representation type" (`int` in the above example), and methods like
`isValid` are resolved statically. This gives developers an
abstraction mechanism with the advantage of zero runtime performance
cost, since no extra heap space is required, and no extra instructions
are needed to convert between an extension type and its underlying
representation.
The disadvantage of using extension types as an abstraction mechanism
is that since no wrapper objects are created at runtime, the
abstraction can be bypassed using `dynamic`, runtime casts, or by
"laundering" the object through contravariant generic methods like
`List.add` (which are runtime checked in Dart). For example:
main() {
var id = IdNumber(1);
var list1 = <int>[];
List<Object> list2 = list1;
list2.add(id); // OK at compile time because `IdNumber` is a
// subtype of `Object`. OK at runtime because
// at runtime, `IdNumber` and `int` are
// indistinguishable.
int i = list1[0];
print(i);
}
Extension types are expected to be particularly useful for
low-overhead decoding of external data formats (such as JSON), and for
inter-operation with other languages (such as Javascript).
For additional information see the feature specification:
https://github.com/dart-lang/language/blob/main/accepted/future-releases/extension-types/feature-specification.md
Change-Id: I900a3a25dcfc38bfa9c9f9b5b9fa20f362883653
Tested: Standard trybots
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/335062
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Leaf Petersen <leafp@google.com>
Commit-Queue: Paul Berry <paulberry@google.com>
When disassembling functions, respect whether absolute addresses should
be used for PC offsets when printing stack map, PC descriptor and
exception handler information.
Also don't print out results for empty auxiliary information.
TEST=ci
Change-Id: Ieda990c1eabcce8be968981a81123a78746c6b6b
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/336663
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Replacing index checks with a single GenericCheckBound instruction
currently causes performance regressions on 32-bit ARM. Undo that
optimization there for now, instead inlining the original Dart code
into the indexing operations.
TEST=ci
Issue: https://github.com/flutter/flutter/issues/138689
Change-Id: I7e67966b576a9d1b5f510e17b44f853f23c98a91
Cq-Include-Trybots: luci.dart.try:vm-linux-release-simarm-try,vm-ffi-qemu-linux-release-arm-try,vm-aot-linux-release-simarm_x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/337480
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Jonas Termansen