The CallMarshaller is used not only from the compiler, but also
from the FfiCall runtime entry used by the interpreter. Since it
only has one use of the thread's CompilerState, looking up the
TypedData class, move the storage of that class from the compiler
state to the object store and remove this dependency.
TEST=ffi/address_of_array_generated_test
ffi/address_of_cast_test
fii/address_of_struct_generated_test
ffi/address_of_typeddata_generated_test
Fixes: https://github.com/dart-lang/sdk/issues/61913
Cq-Include-Trybots: luci.dart.try:vm-dyn-linux-debug-x64-try
Change-Id: Ia4fc5d9ecef370aa9476b998e37cc4ae94ee447f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/498563
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
- Break false dependency of non-VM platforms on VM platform
- Break false dependency of bootstrap gen_kernel on VM platform
- Extend precompile_tools to gen_kernel steps
- Rename intermediate kernel files so JIT versus AOT is visible in ninjatracing
TEST=ci
Change-Id: I07011abe8303597af61d2b8c73e788b661482cc0
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/510060
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Implicit setters of covariant fields (both covariant by declaration and
covariant by class aka generic-covariant-impl) need to perform a type
check of their argument.
It means that inferred type of the field cannot be used as inferred
type of parameter of such setter. This change removes such uses of
an inferred type of the field, which restores parameter type check
which was previously incorrectly optimized out.
TEST=runtime/tests/vm/dart/regress_63419_test.dart
Fixes https://github.com/dart-lang/sdk/issues/63419
Change-Id: Ie313dbaab51dff15d60ce4390e7e41bdc66ad59d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/509020
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
- Declare OS::SafeReadMemory in os.h
- Implement using pread64(/proc/self/mem) on Linux and Android
- Add _readNativeMemory VM Service RPC with pre-checks for null
and address overflow
TEST=runtime/vm/service_test.cc
Cq-Include-Trybots: dart/try:vm-asan-linux-release-x64-try,vm-asan-mac-release-arm64-try,vm-asan-win-release-x64-try,vm-dyn-linux-debug-x64-try,vm-dyn-mac-debug-arm64-try,vm-ffi-qemu-linux-release-arm-try,vm-ffi-qemu-linux-release-riscv64-try,vm-linux-debug-arm64-try,vm-linux-debug-ia32-try,vm-linux-debug-simriscv32-try,vm-linux-debug-simriscv64-try,vm-linux-debug-x64-try,vm-linux-debug-x64c-try,vm-linux-release-simarm-try,vm-mac-debug-arm64-try,vm-mac-debug-x64-try,vm-msan-linux-release-x64-try,vm-reload-linux-debug-x64-try,vm-tsan-linux-release-x64-try,vm-tsan-mac-release-arm64-try,vm-ubsan-linux-release-x64-try,vm-ubsan-mac-release-arm64-try,vm-ubsan-win-release-x64-try,vm-win-debug-arm64-try,vm-win-debug-x64-try,vm-win-debug-x64c-try
Change-Id: Id15a82bf478bc4822c08d7fdf0a5c8bfd71a1fe0
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/505781
Auto-Submit: Nourhan Hasan <nourhan.m.hasan@gmail.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
The only actual change required to the profiler service is appropriately
retrieving the source position for interpreted frames during processing.
Otherwise, most of the tests just need small changes to account for the
slightly different stack trace generated by interpreted code when
walking the stack as code instead of functions:
* Interpreted code calls the runtime entry DRT_AllocateObject directly
for slow-path allocation instead of via the AllocateObjectSlow stub.
* The current name for bytecode frames when walked as code starts with
"[Bytecode]" instead of "[Unoptimized]".
The test vm/cc/Profiler_GetSourceReport is skipped since:
* It tests compiled code-specific features like classifying source
positions.
* Inserting fake samples for interpreted functions is messier since
many PC offsets may map to the same source position.
The other skipped tests are specific to inlining and/or optimized
compiled code, so there's no point in running them on interpreted code.
TEST=vm/cc/Profiler
Cq-Include-Trybots: luci.dart.try:vm-dyn-linux-debug-x64-try
Change-Id: I3e73f42b2562eb6e3e73fe8d5c3f14389039ee04
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507082
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
The debugger assumes a null suspend state variable in an asynchronous
function means that the function is still in the prologue prior
to setting up the suspend state. However, the interpreter clears the
suspend state variable before returning, and the debugger needs to be
able to pause before returning when single stepping, so earlier a hack
was added to the debugger that detects being at the direct call of the
async return method and/or the return instruction with a null suspend
state variable.
However, there's a much simpler way of ensuring the debugger pauses
before returning: just emit the source position for the return prior to
clearing the suspend state variable. This also ensures that the debugger
still has access to the function's suspend state when pausing before
the return, instead of waiting until it has been cleared and thus is no
longer accessible.
TEST=ci (should not change the result of any current tests)
Cq-Include-Trybots: luci.dart.try:vm-dyn-linux-debug-x64-try
Change-Id: I401cceb169d8692ac379cdc5a531e07cafbe9a65
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/500740
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
dartfuzz has been taking ~35 min to finish the build and upload steps even with 100% RBE cache hit rate, causing it to fail to finish before the first wave of nightly builders trigger.
Change-Id: I932facd77bc08f4d7236ae8c760013406a69fb0e
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507267
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
The original iteration of adding coverage arrays originally eagerly
allocated coverage arrays for bytecode methods with recorded coverage.
Now that these coverage arrays are lazily allocated,
Function::GetCoverageArray() needs to ensure that the appropriate
coverage array has been allocated. Otherwise, no misses within the
function are recorded in cases where no RecordCoverage instruction in
the method has yet been executed.
TEST=vm/cc/SourceReport_Coverage_UnusedClass_ForceCompile
pkg/vm_service/test/branch_coverage_test
pkg/vm_service/test/coverage_async_test
pkg/vm_service/test/coverage_closure_call_after_optimization_test
pkg/vm_service/test/coverage_closure_call_test
pkg/vm_service/test/coverage_instance_call_after_optimization_test
pkg/vm_service/test/coverage_static_call_after_optimization_test
Cq-Include-Trybots: luci.dart.try:vm-dyn-linux-debug-x64-try
Change-Id: Iff4ddaf583d0f9583bc415142bdee52c046a6712
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507440
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Extend --check-clean (enabled on bots) to verify build commands don't include absolute paths.
The non-RBE GCC and MSVC builds still have absolute paths. GCC doesn't have -ffile-compilation-dir so it uses -fdebug-prefix-map, which is not fixable. The MSVC build has absolute path in the toolchain wrappers, which might be fixable.
TEST=ci
Change-Id: I3b984aaab7aefa7ff527f0a039ca42281224a09d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/506505
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Avoid a race condition between async setenv/unsetenv calls on the VM thread and concurrent getenv/environ reads on the main thread.
In glibc, setenv and getenv are not thread-safe against each other. VmInteropHandler.setEnvironmentVariable previously sent a message to the VM thread to call setenv/unsetenv asynchronously, while the main thread immediately proceeded to access Platform.environment or Platform.localeName (which calls getenv). This could cause a crash (SIGSEGV) in getenv.
This CL makes VmInteropHandler.setEnvironmentVariable synchronous by awaiting a reply from the VM thread before returning.
TEST=pkg/dartdev/test/environment_test.dart
Fixes https://github.com/dart-lang/sdk/issues/63460
Change-Id: Ic8211897ce26ffbdc142fa594cd397189a61f061
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/506800
Commit-Queue: Ben Konyi <bkonyi@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Auto-Submit: Ben Konyi <bkonyi@google.com>
The previous attempt was reverted because it broke unoptimized JIT
on ARM 32. This reland force-optimizes the two getters.
Stacks on top of the int.{trailingZeroBitCount,oneBitCount} API CL
(commit 754239b077). Both getters route through OTHER_RECOGNIZED_LIST
when a hardware fast path is available; otherwise the newly added
Dart bodies inline at call sites via vm:prefer-inline. The C++
natives are removed.
Backend codegen
---------------
ARM64: NEON CNT + UADDLV (popcount); RBIT + CLZ (ctz).
ARM: NEON CNT + VPADDL chain (popcount); RBIT + CLZ on the
register pair (ctz).
x64: popcntq when TargetCPUFeatures::popcnt_supported();
LoadImmediate(64) + rep_bsfq for ctz (decodes as tzcnt
on BMI1+, preserves dest on zero otherwise).
RISC-V 64: cpop / ctz when RV_baseline includes Zbb.
Per-arch availability is encapsulated in
UnaryInt64OpInstr::IsSupported(Token::Kind).
Apple M-series ARM64, AOT (us/iter, lower is better):
cardinality.swar 371
cardinality.accelerated 154 (2.4x)
forEachSetBit.swar 19031
forEachSetBit.accelerated 4988 (3.8x)
select.swar 199
select.accelerated 77 (2.6x)
complementCardinality.swar 399
complementCardinality.accel 152 (2.6x)
Work towards https://github.com/dart-lang/sdk/issues/6486 (popcount
and ctz intrinsification).
Work towards https://github.com/dart-lang/sdk/issues/1053 (efficient
BitSet implementation).
Fixes https://github.com/dart-lang/sdk/issues/52673
Fixes https://github.com/dart-lang/sdk/issues/38346
Fixes https://github.com/dart-lang/sdk/issues/63436
Issue https://github.com/dart-lang/sdk/issues/10212
Issue https://github.com/dart-lang/sdk/issues/5798
TEST=tests/corelib/int_bit_count_test
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,vm-aot-linux-debug-simarm_x64-try,dart-sdk-linux-riscv64-try
Change-Id: Ib812cbaec6e371b9720df7a543411f78e524cac1
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/506060
Reviewed-by: Martin Kustermann <kustermann@google.com>
Auto-Submit: Modestas Valauskas <valauskasmodestas@gmail.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
The Integer_parse VM native has no Dart-side caller. int.parse is
implemented entirely in Dart in
sdk/lib/_internal/vm_shared/lib/integers_patch.dart (_tryParseSmi
and _parseRadix). Removes the native entry and drops a stale
reference in the comment on the shared ParseInteger helper, which
is still used by Integer_fromEnvironment.
TEST=ci
Change-Id: I8d034e3081c09357094abc6415fc709e3edbea07
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507322
Reviewed-by: Martin Kustermann <kustermann@google.com>
Auto-Submit: Modestas Valauskas <valauskasmodestas@gmail.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
Add `final` to PredecessorAt and PredecessorCount on the six concrete
BlockEntryInstr subclasses (GraphEntry, JoinEntry, TargetEntry,
FunctionEntry, OsrEntry, CatchBlockEntry), so the compiler can
devirtualize the calls.
Measured on a naive 49 KLOC generated lexer with --huge_method_cutoff
gates lifted and with synchronous compilation: JIT compile drops from
~30.6s to ~27.4s (about 10%).
Work towards https://github.com/dart-lang/sdk/issues/63230
TEST=ci
Change-Id: Ia2ef2761646a4c94748bcd7dfe0591262d93cd69
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/505401
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Auto-Submit: Modestas Valauskas <valauskasmodestas@gmail.com>
The test tries to keep fields alive by my making a
toString() => field1 + field2 + ...
Though some field values don't have a `+` operator (e.g. instances of `class C`).
Improving TFA precision can infer that some of these `+` operations will throw,
which then makes other `+` operations unreachable, which removes usages
of those other fields, which will tree shake those other fields, which
will make the test not test anymore what it's intended to.
=> Make the test more robust by using string concatenation instead.
Change-Id: I80788af516083ea3d78ad910eb394e4b7e122384
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507000
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
Implement api methods to create and shutdown isolates from IsolateGroup-bound callbacks (normally invoked from native code), run dart code on such isolates.
The rest of the api is not implemented yet.
TEST=tests/ffi/threading_test.dart
BUG=https://github.com/dart-lang/sdk/issues/62407
CoreLibraryReviewExempt: vm-only change to isolate library
Change-Id: I0271ead8ba011dfe9d7953769415d6a88a962854
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/486522
Commit-Queue: Alexander Aprelev <aam@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Also load/store canonical hashes in the heap for non-empty TypedData
instances in the same manner as canonical hashes for Arrays.
TEST=ci (refactoring only)
Cq-Include-Trybots: luci.dart.try:vm-dyn-linux-debug-x64-try
Change-Id: I54274b558fa9f0c8e304198b18cb3f0e9c3e0dfb
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/504600
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
This adds serialization of the new local variables, VariableStatement and VariableDeclaration. This prepares for replacing LegacyVariable with the new variables in all backends.
TEST=existing
Change-Id: I9bbebfbfd372042d6b7027f0fabd24c165699832
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/506240
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Jens Johansen <jensj@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
On mac with two WorkSet classes, WorkSet destructor from lib/isolate.cc is invoked instead of WorkSet destructor from vm/object.cc. This results in consistent assertion failures on flutter dart-debug builds on mac.
Might help with https://github.com/flutter/flutter/issues/186134
TEST=flutter test with dart-debug build
Change-Id: I9a0a0c9d02ee013da900807b8726df26a3660cc7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/505661
Commit-Queue: Alexander Aprelev <aam@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
When the unoptimized code of a function exceeds 1 MB, start its
optimized compile with far_branch_level=1 instead of 0. Without this,
the assembler fails on out-of-range conditional branches at level=0,
longjmps out of the optimizing pipeline, and the whole pipeline reruns
at level=1 from scratch.
Measured on a naive 49 KLOC generated lexer with --huge_method_cutoff
gates lifted: JIT compile drops from ~26.6s to ~13.2s (about 50%).
Work towards https://github.com/dart-lang/sdk/issues/63230
TEST=ci
Change-Id: I2608a23ef899603a46b401b71be7322b7a8dd44b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/506120
Auto-Submit: Modestas Valauskas <valauskasmodestas@gmail.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
Stacks on top of the int.{trailingZeroBitCount,oneBitCount} API CL
(commit 754239b077). Both getters route through OTHER_RECOGNIZED_LIST
when a hardware fast path is available; otherwise the newly added
Dart bodies inline at call sites via vm:prefer-inline. The C++
natives are removed.
Backend codegen
---------------
ARM64: NEON CNT + UADDLV (popcount); RBIT + CLZ (ctz).
ARM: NEON CNT + VPADDL chain (popcount); RBIT + CLZ on the
register pair (ctz).
x64: popcntq when TargetCPUFeatures::popcnt_supported();
LoadImmediate(64) + rep_bsfq for ctz (decodes as tzcnt
on BMI1+, preserves dest on zero otherwise).
RISC-V 64: cpop / ctz when RV_baseline includes Zbb.
Per-arch availability is encapsulated in
UnaryInt64OpInstr::IsSupported(Token::Kind).
Apple M-series ARM64, AOT (us/iter, lower is better):
cardinality.swar 371
cardinality.accelerated 154 (2.4x)
forEachSetBit.swar 19031
forEachSetBit.accelerated 4988 (3.8x)
select.swar 199
select.accelerated 77 (2.6x)
complementCardinality.swar 399
complementCardinality.accel 152 (2.6x)
Work towards https://github.com/dart-lang/sdk/issues/6486 (popcount
and ctz intrinsification).
Work towards https://github.com/dart-lang/sdk/issues/1053 (efficient
BitSet implementation).
Fixes https://github.com/dart-lang/sdk/issues/52673
Fixes https://github.com/dart-lang/sdk/issues/38346
Issue https://github.com/dart-lang/sdk/issues/10212
Issue https://github.com/dart-lang/sdk/issues/5798
TEST=tests/corelib/int_bit_count_test
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-release-simarm_x64-try,vm-aot-linux-debug-simarm_x64-try,dart-sdk-linux-riscv64-try
Change-Id: I9c2c4225fd63d54e190398a0fb6649366317dd17
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/504221
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Auto-Submit: Modestas Valauskas <valauskasmodestas@gmail.com>
Alexander Markov