Write barrier elimination applies when there is no potential Dart call between a StoreFieldInstr and an AllocateObjectInstr. But the control flow between all potentially throwing instructions in a try block and the corresponding catch block is not explicitly repreresented in the flow graph, so the writer barrier elimination pass needs to handle catch entries specially.
TEST=vm/dart/write_barrier_catch_entry_test
Bug: https://github.com/dart-lang/sdk/issues/60836
Change-Id: I96d5999433320e87da095f85e4a7438bcf5e134f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/432880
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Adds a Mach-O loader, similar to the existing ELF loader, that
can be used on platforms that do not support loading Mach-O dynamic
libraries via dlopen().
Other changes:
* Renames the --force-load-elf-from-memory command line argument
to --force-load-from-memory.
* Use CPU_TYPE_ANY and CPU_SUBTYPE_ANY for architectures that do
not have specific constants in <mach/machine.h> (e.g., RISCV),
as the snapshot header check after loading also catches architecture
mismatches.
* Emit the unwinding information at the end of the text segment
for Windows Mach-O snapshots as is done for ELF ones.
TEST=vm/dart/unobfuscated_static_symbols_test
vm/dart/use_dwarf_stack_traces_flag_test
Issue: https://github.com/dart-lang/sdk/issues/60307
Change-Id: I34a2a334f47d18d5c4f4a712956e71fd0ac94024
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-mac-release-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-mac-debug-arm64-try,vm-fuchsia-release-x64-try,vm-fuchsia-release-arm64-try,vm-linux-debug-ia32-try,vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv32-try,vm-aot-linux-debug-simriscv64-try,vm-aot-linux-release-simarm_x64-try,vm-gcc-linux-try,vm-ubsan-linux-release-arm64-try,vm-aot-win-release-arm64-try,vm-aot-win-release-x64-try,vm-win-release-x64-try,vm-win-release-arm64-try,vm-aot-win-debug-arm64-try,vm-win-debug-arm64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/430100
Reviewed-by: Slava Egorov <vegorov@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Use a combination of gen_snapshot's new ability to generate dylibs directly and the copy of llvm in buildtools to avoid using XCode, which is not available in shards.
Add definitions for debug builders now that they can complete without timing out.
Cq-Include-Trybots: luci.dart.try:vm-aot-mac-product-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try
Change-Id: I79924128f11aafab4cd74ea6e9ab5848c64557da
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/431700
Reviewed-by: Tess Strickland <sstrickl@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
This is a reland of commit 38ef28a058
Fixes:
* Fix comparisons in ASSERT_EQUAL statements on 32-bit architectures.
* Take simulated architectures into account when deciding whether
to use dlopen() for native shared object formats.
* Fix struct/field name collision for GCC.
* Use CPU_TYPE_ANY/CPU_SUBTYPE_ANY for architectures that do not
have more specific cpu_type_t/cpu_subtype_t constants defined.
Original change's description:
> [vm, gen_snapshot] Add app-aot-macho-dylib option for AOT snapshots.
>
> This is the initial framework for creating snapshots as Mach-O dynamic
> libraries. Note that this framework is not 100% feature complete
> compared to generating Mach-O snapshots via assembly. In particular,
> the directly-compiled Mach-O dylib does not yet contain compact
> unwinding information.
>
> Other changes:
>
> * Adds UuidCommand to the native_stack_traces package's Mach-O reader,
> which now appropriately returns the UUID as the build ID for Mach-O
> shared objects.
>
> * Adds Utils::Basename(path) for portably retrieving the basename
> from a path. (Returns nullptr for all arguments where it is not
> currently implemented on Fuchsia or Windows.)
>
> * Adjusts vm/timeline.h to avoid pulling in <mach_o/loader.h> on MacOS,
> as that interferes with uses of the namespaced Mach-O definitions
> in platform/mach_o.h.
>
> * Only attempt to dlopen() a snapshot if ELF is the native format
> for the host platform or the snapshot is not an ELF shared object.
> If dlopen() is used, report the error message if it fails rather
> than attempting to manually load the snapshot as an ELF shared object.
>
> * Fix the magic number stored in DylibAppSnapshot for loaded non-ELF
> dynamic libraries.
>
> * Remove the detection of reverse-endian Mach-O magic numbers in
> DartUtils::SniffForMagicNumber(), since all our Mach-O related code
> assumes host-endian Mach-O files and so there's no point other than
> to give a slightly better error message when failing.
>
> TEST=vm/dart/exported_symbols_test
> vm/dart/unobfuscated_static_symbols_test
> vm/dart/use_dwarf_stack_traces_flag_test
> vm/cc/CanDetectMachOFiles
>
> Issue: https://github.com/dart-lang/sdk/issues/60307
> Change-Id: Idf5b49d6c6d035ab033509613212b95520d65965
> Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-mac-release-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-mac-debug-arm64-try,vm-fuchsia-release-x64-try,vm-fuchsia-release-arm64-try
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/415020
> Reviewed-by: Slava Egorov <vegorov@google.com>
> Commit-Queue: Tess Strickland <sstrickl@google.com>
TEST=vm/dart/exported_symbols_test
vm/dart/unobfuscated_static_symbols_test
vm/dart/use_dwarf_stack_traces_flag_test
vm/cc/CanDetectMachOFiles
ci on trybots that failed on the original CL
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-mac-release-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-mac-debug-arm64-try,vm-fuchsia-release-x64-try,vm-fuchsia-release-arm64-try,vm-linux-debug-ia32-try,vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv32-try,vm-aot-linux-debug-simriscv64-try,vm-aot-linux-release-simarm_x64-try,vm-gcc-linux-try,vm-ubsan-linux-release-arm64-try
Change-Id: Iaffea0ddc6173100c8b5b2a9fe46c45f4f611a2e
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/431240
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
This reverts commit 38ef28a058.
Reason for revert: Breaks build on debug mode 32-bit arches,
release mode simarm_x64
Issue: https://github.com/dart-lang/sdk/issues/60307
Original change's description:
> [vm, gen_snapshot] Add app-aot-macho-dylib option for AOT snapshots.
>
> This is the initial framework for creating snapshots as Mach-O dynamic
> libraries. Note that this framework is not 100% feature complete
> compared to generating Mach-O snapshots via assembly. In particular,
> the directly-compiled Mach-O dylib does not yet contain compact
> unwinding information.
>
> Other changes:
>
> * Adds UuidCommand to the native_stack_traces package's Mach-O reader,
> which now appropriately returns the UUID as the build ID for Mach-O
> shared objects.
>
> * Adds Utils::Basename(path) for portably retrieving the basename
> from a path. (Returns nullptr for all arguments where it is not
> currently implemented on Fuchsia or Windows.)
>
> * Adjusts vm/timeline.h to avoid pulling in <mach_o/loader.h> on MacOS,
> as that interferes with uses of the namespaced Mach-O definitions
> in platform/mach_o.h.
>
> * Only attempt to dlopen() a snapshot if ELF is the native format
> for the host platform or the snapshot is not an ELF shared object.
> If dlopen() is used, report the error message if it fails rather
> than attempting to manually load the snapshot as an ELF shared object.
>
> * Fix the magic number stored in DylibAppSnapshot for loaded non-ELF
> dynamic libraries.
>
> * Remove the detection of reverse-endian Mach-O magic numbers in
> DartUtils::SniffForMagicNumber(), since all our Mach-O related code
> assumes host-endian Mach-O files and so there's no point other than
> to give a slightly better error message when failing.
>
> TEST=vm/dart/exported_symbols_test
> vm/dart/unobfuscated_static_symbols_test
> vm/dart/use_dwarf_stack_traces_flag_test
> vm/cc/CanDetectMachOFiles
>
> Issue: https://github.com/dart-lang/sdk/issues/60307
> Change-Id: Idf5b49d6c6d035ab033509613212b95520d65965
> Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-mac-release-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-mac-debug-arm64-try,vm-fuchsia-release-x64-try,vm-fuchsia-release-arm64-try
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/415020
> Reviewed-by: Slava Egorov <vegorov@google.com>
> Commit-Queue: Tess Strickland <sstrickl@google.com>
Issue: https://github.com/dart-lang/sdk/issues/60307
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-mac-release-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-mac-debug-arm64-try,vm-fuchsia-release-x64-try,vm-fuchsia-release-arm64-try
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Change-Id: Iff2ab4c84a513a184784129f456bd357ae8e3a67
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/431220
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Auto-Submit: Tess Strickland <sstrickl@google.com>
This is the initial framework for creating snapshots as Mach-O dynamic
libraries. Note that this framework is not 100% feature complete
compared to generating Mach-O snapshots via assembly. In particular,
the directly-compiled Mach-O dylib does not yet contain compact
unwinding information.
Other changes:
* Adds UuidCommand to the native_stack_traces package's Mach-O reader,
which now appropriately returns the UUID as the build ID for Mach-O
shared objects.
* Adds Utils::Basename(path) for portably retrieving the basename
from a path. (Returns nullptr for all arguments where it is not
currently implemented on Fuchsia or Windows.)
* Adjusts vm/timeline.h to avoid pulling in <mach_o/loader.h> on MacOS,
as that interferes with uses of the namespaced Mach-O definitions
in platform/mach_o.h.
* Only attempt to dlopen() a snapshot if ELF is the native format
for the host platform or the snapshot is not an ELF shared object.
If dlopen() is used, report the error message if it fails rather
than attempting to manually load the snapshot as an ELF shared object.
* Fix the magic number stored in DylibAppSnapshot for loaded non-ELF
dynamic libraries.
* Remove the detection of reverse-endian Mach-O magic numbers in
DartUtils::SniffForMagicNumber(), since all our Mach-O related code
assumes host-endian Mach-O files and so there's no point other than
to give a slightly better error message when failing.
TEST=vm/dart/exported_symbols_test
vm/dart/unobfuscated_static_symbols_test
vm/dart/use_dwarf_stack_traces_flag_test
vm/cc/CanDetectMachOFiles
Issue: https://github.com/dart-lang/sdk/issues/60307
Change-Id: Idf5b49d6c6d035ab033509613212b95520d65965
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-mac-release-arm64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-mac-debug-arm64-try,vm-fuchsia-release-x64-try,vm-fuchsia-release-arm64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/415020
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
In release builds most `Field` objects are tree shaken,
but if an instance field is e.g. a `late final` field then the `Field`
object is retained in the snapshot for the `LateInitializationError`
slow path. Though `Field` will not appear in `Class.fields()`, it
will only be referenced in the `ObjectPool`.
=> Instead of dumping `Class.fields()` we collect all `Field`s with
the same owner and dump those (a superset of `Class.fields()`)
We dump more information for a `Field`: Whether it's a reference field
or unboxed field (in which case we dump what unboxed type it is). We
also dump field flags (e.g. `late`, `final`, `static`).
We also dump now a `instance_slots` property on the class which
describes every slot in the instance. If a slot in an instance
* has a corresponding `Field` we emit `instance_field` type and
the id of the `Field` object.
* has no corresponding `Field` we emit a `unknown_slot` type
* we emit whether the slot is a reference field
This brings the information the `analyze_snapshot` tool dumps closer to
what Blutter uses, but some differences remain, e.g.:
* Blutter seems to scan constant instances of classes (if there's any)
and tries to determine based on those constants what unboxed slots
could mean.
It may look at 8 bytes and "guess" whether it's more likely to be an
integer or a double. (Side note: It doesn't consider
`Float32x4`/`Float64x2`)
=> The guess can be incorrect, so we do not do that.
* Blutter seems to scan constant instances of classes (if there's any)
and tries to guess the type of a reference field based on what that
constant's field points to.
=> The guess can be incorrect, so we do not do that.
go/dart-ama
TEST=vm/dart/analyze_snapshot_binary_test
Change-Id: I116560ba5e5f89d4349f2227362b8494e3af7a1c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/425261
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
`package:native_assets_cli` has been split up and
`package:native_assets_builder` has been renamed.
Change-Id: Ic9d5c21bb3c7ecf7924fdee6b4281153ad37fac5
Cq-Include-Trybots: luci.dart.try:pkg-linux-debug-try,pkg-linux-release-arm64-try,pkg-linux-release-try,pkg-mac-release-arm64-try,pkg-win-release-arm64-try,pkg-mac-release-try,pkg-win-release-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/425504
Reviewed-by: Ivan Inozemtsev <iinozemtsev@google.com>
Commit-Queue: Daco Harkes <dacoharkes@google.com>
- Add `target-arch` option to `dart compile exe|aot-snapshot`
- Add artifacts cache in `~/.dart`, and download gen_snapshot and
dartaotruntime from cloud storage if necessary
I've manually built an aot snapshot and exe on macOS ARM64 for Linux
X64, copied them over to a linux machine, and made sure they work.
Change-Id: I74dd581ecb573ff4069a2fe19f9d04aaa352538e
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/417801
Reviewed-by: Tess Strickland <sstrickl@google.com>
Commit-Queue: Ivan Inozemtsev <iinozemtsev@google.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
When constant propagation evaluates EqualityCompare or StrictCompare,
it can unwrap phis (if they are redundant wrt current set of reachable
blocks), and calculate value if both lhs and rhs operands are the same.
Previously, phi was marked as unwrapped only if that calculation changes
the constant value of comparison.
This is not correct as calculation based on unwrapped phi
needs to be invalidated anyway if more reachable blocks are discovered
and unwrapped phi becomes non-redundant.
The fix is to mark the unwrapped phis regardless of whether the constant
value of comparison has been changed or not.
TEST=vm/dart/regress_60349_test
Fixes https://github.com/dart-lang/sdk/issues/60349
Change-Id: I01e46d6def67fdc5d4c0bdc3a73aaac94725e221
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/417021
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
We also disable the test in hotreload mode, because threads that are
stuck in ffi calls don't participate in hot-reload requests.
This is something we may want to look into. Right now it's conservative in
the sense that if an embedder did a `Dart_LookupClass()` in such a ffi call
and then a reload happened, such a handle would possibly hold pointer to
an old class object and weird things may happen.
This fixes test failures in hot reload / rollback mode of
`vm/dart/isolates/many_isolates_blocked_at_ffi_test`
and also make it more robust.
TEST=tests/ffi/dl_api_exit_enter_isolate_test
Change-Id: I823d07a6ab04e62e051e4d22ec80cbc3649762a3
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/409100
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
For entry-point pragma annotations, most of the time they are
used with either no argument or with an argument that evaluates
to either
* false to denote the annotation should not take effect, or
* null or true to denote the annotation should take effect.
However, the user can also specify that only part of the operations
on a member should be accessed from native code by using a string
argument that is either 'call', 'set', or 'get'.
The entry point verification in Invoke/InvokeGetter/InvokeSetter
assumes that for getters and setters, the only valid string argument
is 'get' or 'set', respectively. This is because those methods are
called via `Dart_GetField`[0] and `Dart_SetField`, respectively, as if
they were the getter or setter of a defined field.
However, the precompiler previously assumed that the string
argument 'call' was the only string argument that meant the link
to a function's code object should be saved. Similarly, it assumed the
string argument 'get' for functions meant that their implicit closure
function should be saved, which ends up including getters. Furthermore,
it did not do anything with setters annotated with the string argument
'set'. This means that the code link would not be saved for getters or
setters that were annotated with the string argument expected by the
entry point verifier.
This CL aligns the precompiler to match the expectations of other
parts of the codebase. It also changes TFA to report an error
if a getter or setter is marked with the string argument 'call'.
[0] `Dart_Invoke` can be called with the name of a getter that
returns a closure, but doing so is semantically equivalent to
calling `Dart_GetField` followed by `Dart_InvokeClosure`.
TEST=vm/dart/entrypoint_verification_test
Fixes: https://github.com/dart-lang/sdk/issues/59920
Change-Id: Ia2768bbaf9058bb14a1cdfb331eb85fa082a0e90
Cq-Include-Trybots: luci.dart.try:vm-aot-dwarf-linux-product-x64-try,vm-aot-linux-debug-x64-try,vm-aot-linux-product-x64-try,vm-aot-mac-product-arm64-try,vm-aot-obfuscate-linux-release-x64-try,vm-linux-debug-x64-try,vm-linux-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/404823
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Consider the following situation: member M1 is overridden by another
member M2; M2 is overridden in a dynamic module.
Members which can be overridden in a dynamic module (such as M2)
should be specified as 'can-be-overridden' in the dynamic interface.
Members which are overridden implicitly/transitively (such as M1)
are not required to be mentioned in the dynamic interface.
However, when determining possible targets for a call with
interface target M1, compiler should treat it as potentially
overridden in a dynamic module.
This change adds such handling to the VM/AOT. Dynamic interface
annotator now marks members such as M1 with
'dyn-module:can-be-overridden-implicitly' pragma, and
VM/AOT takes both can-be-overridden and can-be-overridden-implicitly
into account.
This change also simplifies handling of implicitly extenable classes
in the VM/AOT - now VM handles both extendable and implicitly-extendable
pragmas (from dynamic interface annotator) instead of recalculating
implicitly extendable classes on its own.
TEST=pkg/vm/test/transformations/dynamic_interface_annotator_test.dart
TEST=dynamic_modules_suite/implicitly_extendable
Fixes https://github.com/dart-lang/sdk/issues/59880
Change-Id: Id4570cc86303f8e45a061d696e9bca0d0b2b4b81
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/403951
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Nate Biggs <natebiggs@google.com>
Only insert Phi at JoinEntry and Parameter at CatchBlockEntry if
corresponding variable is alive into the block. This keeps insertion
consistent with environment pruning logic employed by RenameRecursive:
which prunes dead variables out environments when entering the block
based on contents of the live_in set.
Previously we would insert Parameter instructions for dead variables
which lead to an incorrect (unsound) set of catch entry moves
generated by the backend due to an inconsistency between Parameter
insertion and environment pruning logic and the fact that
catch entry moves generator would simply ignore all moves if the
source of the move was an optimized_out (aka constant_dead) marker.
The following situation was possible: if the variable was assigned
inside the try and partially alive inside of it (e.g. we had
blocks where it was alive and blocks where it was not) then we
would end up inserting Parameter for it, but some of the exceptional
edges (corresponding to blocks where the variable was dead) would
carry optimized_out value into that Parameter. However
`CatchEntryMoveFor` would not record these moves leading to
uninitialized garbage arriving on those exceptional edges instead.
Note that Phis were not susceptible to the same issue as Parameters
because regalloc / backend would not treat optimized_out value
specially.
This change converts silent treatment of optimized_out in the
catch entry moves generation into a release assert instead.
Reviewing other uses of constant_dead in the compiler revealed
a bug in canonicalization rule for `o.runtimeType` which
did not take exceptional edges into account. We add a
regression test for that and fix that issue as well.
Fixes https://github.com/dart-lang/sdk/issues/59822
TEST=vm/dart/regress_59822,vm/dart/regress_runtime_type_in_trycatch
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-aot-linux-debug-x64c-try,vm-aot-linux-product-x64-try,vm-aot-linux-release-arm64-try,vm-aot-linux-release-simarm_x64-try,vm-aot-linux-release-x64-try
Change-Id: I8dc397cb98d84048a1791b9dd6baa7586a2688c6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/403583
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
This is a reland of commit 982b9fad44
Original change's description:
> [vm] Turn on entry point checking in JIT mode.
>
> Now that Flutter tests that access entry points from native code
> have been annotated[1], we can turn on entry point checking in JIT
> mode.
>
> This CL also removes the A flag category from flag_list.h and the
> AOT_FLAG_MACRO definitions and uses from flags.[cc,h], as they were
> created as a temporary measure until this flag could be unconditionally
> defaulted to true.
>
> [1] See the following PRs:
> * https://github.com/flutter/engine/pull/57158
> * https://github.com/flutter/flutter/pull/160158
> * https://github.com/flutter/flutter/pull/160421
>
> TEST=vm/dart/entrypoints_verification_test vm/cc/IRTest
> vm/cc/StreamingFlowGraphBuilder vm/cc/STC vm/cc/TTS
>
> Issue: https://github.com/dart-lang/sdk/issues/50649
> Issue: https://github.com/flutter/flutter/issues/118608
>
> Cq-Include-Trybots: luci.dart.try:vm-aot-linux-product-x64-try,vm-aot-linux-debug-x64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-product-arm64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-linux-release-x64-try,vm-appjit-linux-product-x64-try
> Change-Id: Ibe5b21bb74f1a6fb88824b71ff87b9e555216dbf
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/400301
> Reviewed-by: Martin Kustermann <kustermann@google.com>
> Commit-Queue: Tess Strickland <sstrickl@google.com>
TEST=vm/dart/entrypoints_verification_test vm/cc/IRTest
vm/cc/StreamingFlowGraphBuilder vm/cc/STC vm/cc/TTS
Change-Id: Ibd5f362f908b4aaa68cda870a387c081537bbc16
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-product-x64-try,vm-aot-linux-debug-x64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-product-arm64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-linux-release-x64-try,vm-appjit-linux-product-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/403360
Auto-Submit: Ivan Inozemtsev <iinozemtsev@google.com>
Commit-Queue: Ivan Inozemtsev <iinozemtsev@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
The `ffi_test_functions` shared library needs to access functions from
`dart_api.h` and `dart_native_api.h`, which are only available in
the Dart executable.
UNIX shared libraries can have undefined symbols, which are resolved at
runtime and can be found in the loading executable. Windows DLLs cannot
have undefined symbols, but they can be dynamically linked to an
executable (in this case `dart.exe`). This requires the DLL to be able
to find the executable at runtime.
A better solution is to include implementations for the Dart APIs in the
DLL itself, that use `GetModuleHandle(NULL)` to get a handle to the
executable and `GetProcAddress` to get the address of the function.
This is what `dart_api_win.c` does.
Fixes https://github.com/dart-lang/sdk/issues/40579
Fixes https://github.com/dart-lang/sdk/issues/59677
TEST=ci
Cq-Include-Trybots: luci.dart.try:vm-win-release-ia32-try,vm-win-debug-x64c-try,vm-win-debug-x64-try,vm-win-debug-arm64-try,vm-msvc-windows-try,vm-aot-win-debug-x64c-try,vm-aot-win-debug-x64-try,vm-aot-win-debug-arm64-try,vm-aot-win-product-x64-try,vm-aot-win-release-x64-try,vm-aot-win-release-arm64-try,pkg-win-release-try,pkg-win-release-arm64-try,dart-sdk-win-try,dart-sdk-win-arm64-try
Change-Id: I7f971a8ce21e03d18ed2967e74998f925c9236b2
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/400582
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Alexander Thomas <athom@google.com>
Loop information may be outdated by the time code is generated
(for example, if flow graph is serialized/deserialized).
Recalculate loop information during code generation if it is needed
in order to align loop headers.
TEST=vm/dart/align_loops_verify_alignment_test
Change-Id: I1edd63354cee9940bbd474dc95535b4b74081830
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/401863
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Add awaiter-links leading from callbacks attached to
the original Futures to Completer produced by wait.
This requires changing representation of _FutureResult
slightly because we do not support annotating arbitrary
fields as awaiter-links only variables can be annotated
Thus we need to capture onReady callback in a the context of
closures passed to Future.then.
Fixes https://github.com/dart-lang/sdk/issues/59730
TEST=pkg/vm_service/test/pause_on_unhandled_exceptions_future_extensions_test.dart,runtime/tests/vm/dart/awaiter_stacks/async_stacks_test.dart
CoreLibraryReviewExempt: No API or significant implementation changes
Change-Id: I904b441b27c4e06f08b7ea7ba066a2fb03504ce6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/401062
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Lasse Nielsen <lrn@google.com>
Previously catch blocks were hanging off the function entry blocks being effectively an alternative entry-point into the function. Now catch blocks are hanging off the try-entry blocks.
Previously all the variables that are used in the catch block and beyond were declared as parameters to catch block. Now only those that have their definitions not dominating catch entry will become parameters (for example, if a variable is assigned in the try-block, it becomes a parameter to catch block).
During OSR, if OSR target entry point is inside some try-blocks, then all of corresponding try-entry/catch-blocks are pulled up to the OSR entry forming a chain that ends with a jump to the OSR target entry.
TEST=vm/dart/trycatch*, ci
Change-Id: Iae20c6548d5d65c63be6d7a53c1b0d2adac7ac31
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356311
Commit-Queue: Alexander Aprelev <aam@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Ryan Macnak