New option can be used instead of --aot to produce a tree-shaken
minimal .dill file which can run on the VM in JIT mode.
Can be used along with --no-link-platform and --gen-bytecode to further
decrease size of the .dill file.
Change-Id: I35e8d81263a5e223decf160d79322c06cfd27321
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/140740
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
A combination of `runZoned` and `runZonedGuarded` where only the latter allows an `onError` parameter,
and only that has a nullable return type.
Retains the `onError` parameter on `runZoned` for now because it's too breaking to remove it until packages have been migrated off of it.
It will be removed in a follow-up CL.
Change-Id: If0e86c8d14e13fa089c66f4af975aeacb2616cf6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/137302
Reviewed-by: Jake Macdonald <jakemac@google.com>
Reviewed-by: Nate Bosch <nbosch@google.com>
In bare instructions mode, we do not have a per-Code object pool,
but rather a global shared object pool. When generating instructions
that access the object pool, we use the index in the (eventually
created) global object pool.
Recently, when the V8 snapshot profile writer is active, we began
attaching an per-Code object pool so that each object in the global
object pool can be attributed to the Code objects that caused its
inclusion. This pool only has those objects attributable to this
Code object, and not all objects in the global pool up to the point
of compiling the Code object.
However, the disassembler currently only looks at whether the current
Code object has an object pool when decoding object pool accesses.
This was safe when there were no per-Code object pools in bare
instruction mode, but since the change described above, it attempts
to access the object in the Code-local pool at the same index as the
global pool and crashes.
This change fixes the disassembler so that we only try to decode the
object being accessed in the object pool when in non-bare instructions
mode.
Fixes https://github.com/dart-lang/sdk/issues/41149
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-linux-release-simarm_x64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-precomp-win-release-x64-try
Change-Id: I96fd61d05c657a5ee1d86565cc39c83ebee587f1
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/140603
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Commit-Queue: Teagan Strickland <sstrickl@google.com>
When performing protobuf-aware treeshaking, we may generate a kernel
file which will later be used in conjunction with a non-transformed
platform kernel. In this case signature shaking is unsafe.
The original revision is in patchset 1.
This reverts commit b59dad38df.
Change-Id: I1a85e42f245fcdbd821fa1f07b2e9966b8c2d958
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/139100
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
So far we relied on our normal Dart tests to exercise and test TTS.
This CL adds concrete TTS unit tests, which directly construct the
values the TTS accepts via the calling convention (instance,
dst_type, instantiator tav, function tav), and allows us to test
* whether TTS handled the type test
* whether TTS was falling back to STC which handled the type test
* whether TTS (and STC) preserve the necessary registers
This CL also gets rid of TMP usage in TTS related code.
This CL also adds documentation about TTS to
runtime/docs/compiler/type_testing_stubs.md.
This CL also fixes two issues:
* handle all-dynamic null vector `as Foo<..., T, ...>`
* compare the actual type parameter value in `as Foo<..., T, ...>`
(instead of using type_class_id>)
There seem to be no major performance changes, some benchmarks change
positive/negative but many of them seem to be bi-modal.
Issue https://github.com/dart-lang/sdk/issues/40964
Issue https://github.com/dart-lang/sdk/issues/40736
Change-Id: Iad41e405fe693c67be6221a18e3dad03cc0a5f24
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/139461
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Régis Crelier <regis@google.com>
Static methods and constructors may receive unboxed parameters.
This change adds a bitmap to the RawFunction object indicating
which parameters are unboxed and their types. Also, the stack frame
bitmap was updated to include new bits for those parameters.
The ArgumentsDescriptor object was updated to store not only the number
of arguments, but also their size, since 32-bit platforms would use two
words for each int64 and double value.
In addition, static methods that return non-nullable integers
and doubles may return these values unboxed.
Dart AOT x64:
TypedData.Uint8ClampedListViewVarBench +11.73%
DeltaBlue +9.32%
Meteor (Intel Xeon) +4.64%
Flutter release android-armv8:
velocity_tracker_bench -2.32% (less is better)
flutter_gallery_readonlydata_size +0.18%
flutter_gallery_total_size +0.13%
flutter_gallery_instructions_size +0.12%
Issue https://github.com/dart-lang/sdk/issues/33549
Change-Id: I97e310f8f30213920af2333fbdedf193d03db5c3
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/135313
Commit-Queue: Victor Agnez Lima <victoragnez@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
The code previously used the target offset directly instead of the
JumpLabel which could cause it to miss multiple indirections, causing
a -1 offset.
This would manifest as a jump to an entry which would be
off-by-minus-one and thus hit the pre-code int3 barrier.
Bug: https://github.com/flutter/flutter/issues/51298
Change-Id: I0ef7d8ed578594bfb5fe9544f9b4aebaabc86275
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/137303
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Commit-Queue: Clement Skau <cskau@google.com>
When deduplicating instructions, multiple Code objects may end up
mapping to the same set of program counters, which affects decoding a
stack trace using PC addresses. Since the DWARF stack traces test
depends on the stack trace decoding a certain way, pick a test as a
canary for issues with instruction deduplication that _doesn't_ depend
on how the stack trace will decode.
In previous experience, the underlying test used is unimportant,
so we build off the existing hello world code. If we have an error
due to code inadvertently assuming non-deduplicated instructions,
this error tends to manifest in the VM failing to start up no matter
what code should eventually run.
Fixes https://github.com/dart-lang/sdk/issues/40635
Change-Id: Ife9048740b91dc98b50c8fe4b9ff3dc052dd0a62
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-linux-release-simarm_x64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-precomp-win-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/136122
Commit-Queue: Teagan Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Sets a threshold of five continuations determining if the old
if-else or the new igoto-based implementation will be used.
Informal benchmarking on x64 and arm_x64 point towards the overhead
of the igoto-based impl. dropping off around this point.
Benchmarks of this CL (threshold=5) show drastic improvement in
Calls.IterableManualIterablePolymorphicManyYields of about ~35-65%
across {dart,dart-aot}-{ia32,x64,armv7hf,armv8}.
Note: This is a second revision of a CL landed and reverted from:
https://dart-review.googlesource.com/c/sdk/+/133586
The fix for the previous revert was swapping the parameters to
BuildIndirectEntry(int, int) in kernel_binary_flowgraph.cc.
Bug: https://github.com/dart-lang/sdk/issues/37754
Change-Id: If035235ce8aa660af2ea7f8cfbc8066cde6329b7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/136126
Commit-Queue: Clement Skau <cskau@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Sets a threshold of five continuations determining if the old
if-else or the new igoto-based implementation will be used.
Informal benchmarking on x64 and arm_x64 point towards the overhead
of the igoto-based impl. dropping off around this point.
Benchmarks of this CL (threshold=5) show drastic improvement in
Calls.IterableManualIterablePolymorphicManyYields of about ~35-65%
across {dart,dart-aot}-{ia32,x64,armv7hf,armv8}.
Bug: https://github.com/dart-lang/sdk/issues/37754
Change-Id: I6e113f1f98e9ab0f994cf93004227d616e9e4d07
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133586
Commit-Queue: Clement Skau <cskau@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Now, when writing an AOT snapshot in bare instructions mode, only
the actual instructions in the RawInstructions payload are serialized
instead of the entire RawInstructions object.
Since there are no longer RawInstructions objects in these AOT
snapshots, we also change how Code objects are serialized. Instead
of just containing a reference to the RawInstructions object, we
serialize two pieces of information: where the instructions
payload for this Code object begins and whether there was a single
entry for the instructions payload. (To save space, the single
entry bit is serialized as the low order bit of the unchecked
offset, which was already being serialized).
While we also need the length of the instructions payload, we
approximate it for all but the last Code object by subtracting
the next Code object's payload start from this Code object's
payload start. For the last Code object, we assume it extends
to the end of the instructions image.
Changes on flutter gallery in release mode:
armv7: instructions size -2.70%, total size -1.73%
armv8: instructions size -6.04%, total size -3.63%
Fixes https://github.com/dart-lang/sdk/issues/38451.
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-linux-release-simarm-try,vm-kernel-precomp-linux-release-simarm64-try,vm-kernel-precomp-linux-release-simarm_x64-try,vm-kernel-precomp-android-release-arm64-try,vm-kernel-precomp-android-release-arm_x64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-precomp-win-release-x64-try
Change-Id: Ia0a5c4e5e47c956776dc62503da38ec55a143c04
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/134325
Commit-Queue: Teagan Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This reverts commit 7475c637c3.
Reason for revert: Breakage on simarm_x64 (and thus arm_x64)
Original change's description:
> [vm/aot] Remove object wrapping of bare instructions for AOT snapshots.
>
> Now, when writing an AOT snapshot in bare instructions mode, only
> the actual instructions in the RawInstructions payload are serialized
> instead of the entire RawInstructions object.
>
> Since there are no longer RawInstructions objects in these AOT
> snapshots, we also change how Code objects are serialized. Instead
> of just containing a reference to the RawInstructions object, we
> serialize two pieces of information: where the instructions
> payload for this Code object begins and whether there was a single
> entry for the instructions payload. (To save space, the single
> entry bit is serialized as the low order bit of the unchecked
> offset, which was already being serialized).
>
> While we also need the length of the instructions payload, we
> approximate it for all but the last Code object by subtracting
> the next Code object's payload start from this Code object's
> payload start. For the last Code object, we assume it extends
> to the end of the instructions image.
>
> Changes on flutter gallery in release mode:
> armv7: instructions size -2.66%, total size -1.68%
> armv8: instructions size -5.81%, total size -3.49%
>
> Fixes https://github.com/dart-lang/sdk/issues/38451.
>
> Change-Id: Ia458e8d99bae18f5c3b6e849df2519027f06f574
> Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-linux-release-simarm-try,vm-kernel-precomp-linux-release-simarm64-try,vm-kernel-precomp-linux-release-simarm_x64-try,vm-kernel-precomp-android-release-arm64-try,vm-kernel-precomp-android-release-arm_x64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-precomp-win-release-x64-try
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/131067
> Commit-Queue: Teagan Strickland <sstrickl@google.com>
> Reviewed-by: Martin Kustermann <kustermann@google.com>
TBR=kustermann@google.com,rmacnak@google.com,sstrickl@google.com
Change-Id: If2c8ca6b0993211b2509b275af7ff11a8fa2baa3
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try, vm-kernel-precomp-linux-release-simarm-try, vm-kernel-precomp-linux-release-simarm64-try, vm-kernel-precomp-linux-release-simarm_x64-try, vm-kernel-precomp-android-release-arm64-try, vm-kernel-precomp-android-release-arm_x64-try, vm-kernel-precomp-mac-release-simarm64-try, vm-kernel-precomp-win-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/134322
Reviewed-by: Teagan Strickland <sstrickl@google.com>
Commit-Queue: Teagan Strickland <sstrickl@google.com>
Now, when writing an AOT snapshot in bare instructions mode, only
the actual instructions in the RawInstructions payload are serialized
instead of the entire RawInstructions object.
Since there are no longer RawInstructions objects in these AOT
snapshots, we also change how Code objects are serialized. Instead
of just containing a reference to the RawInstructions object, we
serialize two pieces of information: where the instructions
payload for this Code object begins and whether there was a single
entry for the instructions payload. (To save space, the single
entry bit is serialized as the low order bit of the unchecked
offset, which was already being serialized).
While we also need the length of the instructions payload, we
approximate it for all but the last Code object by subtracting
the next Code object's payload start from this Code object's
payload start. For the last Code object, we assume it extends
to the end of the instructions image.
Changes on flutter gallery in release mode:
armv7: instructions size -2.66%, total size -1.68%
armv8: instructions size -5.81%, total size -3.49%
Fixes https://github.com/dart-lang/sdk/issues/38451.
Change-Id: Ia458e8d99bae18f5c3b6e849df2519027f06f574
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-linux-release-simarm-try,vm-kernel-precomp-linux-release-simarm64-try,vm-kernel-precomp-linux-release-simarm_x64-try,vm-kernel-precomp-android-release-arm64-try,vm-kernel-precomp-android-release-arm_x64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-precomp-win-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/131067
Commit-Queue: Teagan Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This is a Dart-tailored implementation of the "row-displacement
dispatch table" technique for closed-world instance calls:
All interface targets in the program are grouped into selectors such
that all targets that could potentially be called from the same call
site have the same selector (currently just grouped by name).
Each selector is assigned a selector offset such that offset + classid
is unique for all selector/classid combinations where the class
implements the selector.
At every instance call site that has an interface target (i.e. where
the static type of the receiver is not dynamic), the selector offset +
receiver classid is computed and used as index into a global table of
entry points.
If the receiver can be null (as determined by the front-end TFA and the
VM type propagation), a null check is inserted before the call.
An arguments descriptor is provided (only) for selectors that need it
(those which have type parameters or optional/named parameters).
The dispatch table calls don't need the monomorphic entry code, so for
functions that are only called via dispatch table calls (i.e. never
called dynamically), the monomorphic entry code is left out.
Some future improvements to the table dispatch implementation are
mentioned in https://github.com/dart-lang/sdk/issues/40188
The table dispatch flag is disabled by default in this commit. A
separate commit enables the flag.
Change-Id: Ic2911742b4a2c9a8d3bc7df60605454cbe4c0714
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/126648
Commit-Queue: Aske Simon Christensen <askesc@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
- Handle bytecode for e.g. dartkb-simarm64 does not have source position.
- Use existing GetCallerSp() instead of working across frames.
- Nit: Adds clarifying comments to test.
- Nit: Updates name of non-async-stack test to clarify flags used.
Tested:
- CQ with --lazy-async-stacks on by default.
- vm/dart/causal_stacks and language_2/vm/causal_async_exception_stack_test with current flags.
Bug: https://github.com/dart-lang/sdk/issues/39525
Change-Id: Ie6581a734cdcafbd4fb641bd86bffc03ed241532
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/133063
Commit-Queue: Clement Skau <cskau@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This reverts commit 9eb531bde4.
Reason for revert: Bots are red. Some tests are failing.
https://ci.chromium.org/p/dart/builders/ci.sandbox/vm-kernel-precomp-obfuscate-linux-release-x64/6039https://ci.chromium.org/p/dart/builders/ci.sandbox/vm-kernel-precomp-android-release-arm_x64/957
Original change's description:
> [vm] Add support for real unboxed floating point fields in AOT
>
> Non-nullable floating point fields (double, Float32x4, Float64x2)
> are fully unboxed in their classes.
>
> A bitmap for each class was added to the shared class table in order to keep
> track of the pointers of the classes. Since all classes in Flutter Gallery
> have less than 64 fields, the bitmap is represented by a 64 bit integer and
> fields whose offset is more than 64 words are not unboxed.
>
> The instance sizes and field offsets might change between target and host
> in cross-compilation, since the number of words used to store unboxed fields
> may differ.
>
> dart-aot Xeon
>
> SplayLatency -4.62%
> SplayHarderLatency -4.17%
> NavierStokes -2.20%
> Tracer 8.72%
> ParticleSystemPaint 2.90%
> NBodySIMD 8.35%
> NBody 25.59%
>
> With hack TFA to make doubles in Rect/Offset/Size classes in flutter non-nullable:
>
> flutter arm-v8:
>
> gallery total size: -1%
>
> matrix_utils_transform_rect_perspective -16.70% (less is better)
> matrix_utils_transform_rect_affine -31.82% (less is better)
> matrix_utils_transform_point_perspective -24.90% (less is better)
> matrix_utils_transform_point_affine) -27.26% (less is better)
> rrect_contains_bench -4.719% (less is better)
>
> Change-Id: I9ae09c9c3167d99f9efd071a92937aa51093fd1d
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/131824
> Commit-Queue: Victor Agnez Lima <victoragnez@google.com>
> Reviewed-by: Martin Kustermann <kustermann@google.com>
> Reviewed-by: Ryan Macnak <rmacnak@google.com>
> Reviewed-by: Samir Jindel <sjindel@google.com>
TBR=kustermann@google.com,rmacnak@google.com,sjindel@google.com,victoragnez@google.com
Change-Id: Ic73858f6adb7f55c4129d4f46ff4731b378cb634
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/134020
Reviewed-by: Zichang Guo <zichangguo@google.com>
Commit-Queue: Zichang Guo <zichangguo@google.com>
Non-nullable floating point fields (double, Float32x4, Float64x2)
are fully unboxed in their classes.
A bitmap for each class was added to the shared class table in order to keep
track of the pointers of the classes. Since all classes in Flutter Gallery
have less than 64 fields, the bitmap is represented by a 64 bit integer and
fields whose offset is more than 64 words are not unboxed.
The instance sizes and field offsets might change between target and host
in cross-compilation, since the number of words used to store unboxed fields
may differ.
dart-aot Xeon
SplayLatency -4.62%
SplayHarderLatency -4.17%
NavierStokes -2.20%
Tracer 8.72%
ParticleSystemPaint 2.90%
NBodySIMD 8.35%
NBody 25.59%
With hack TFA to make doubles in Rect/Offset/Size classes in flutter non-nullable:
flutter arm-v8:
gallery total size: -1%
matrix_utils_transform_rect_perspective -16.70% (less is better)
matrix_utils_transform_rect_affine -31.82% (less is better)
matrix_utils_transform_point_perspective -24.90% (less is better)
matrix_utils_transform_point_affine) -27.26% (less is better)
rrect_contains_bench -4.719% (less is better)
Change-Id: I9ae09c9c3167d99f9efd071a92937aa51093fd1d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/131824
Commit-Queue: Victor Agnez Lima <victoragnez@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Samir Jindel <sjindel@google.com>
For `foo() async* {}` frames we find the "caller" by finding out
what closure is registered as listener on the _AsyncStreamController.
There can be two cases:
a) The caller does a regular `foo().listen((_) {})`:
The stack trace will have the closure as the caller and unwinding stops.
b) The caller uses 'await for (... foo())':
In this case the listener will be a StreamIterator.
This CL changes our unwinding code to get the awaiter of `await it.moveNext()`.
Bug: https://github.com/dart-lang/sdk/issues/39525
Change-Id: I13f76025a15682aaf55fd968088fc2059982d842
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/132841
Commit-Queue: Clement Skau <cskau@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Alexander Markov