There's just one case where we can't get the offset for Instructions
from somewhere else, and that's when we reset active_instructions_ to
instructions_ in Code::Enable(). Thus, we add a caching field for the
uncached offset to RawCode, but only when the active_instructions_
field is added (when not running in the precompiled runtime).
We also output the unchecked offset in snapshots for Code objects so
we can recalculate the entry point cache on loading. (For full JIT
snapshots, we output both the cached offset and the offset for
active_instructions_.)
Bug: https://github.com/dart-lang/sdk/issues/38451
Change-Id: I78572ec83d1574b00820df148bccb57e5c79eaaa
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/125400
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Teagan Strickland <sstrickl@google.com>
This reverts commit 20ec71d447.
Reason for revert: Performance regressions and seemingly broken TypedData.* benchmarks.
Original change's description:
> [vm/compiler] Fix TypeTestingStub -> SubtypeTestCache fallback code if dst_type = TypeParameter
>
> If the dst_type of an AssertAssignable is a type parameter, the AssertAssignable
> implementation will load the value of the type parameter using
> the (instantiator or function) type arguments. It will then call the
> type testing stub (TTS) of that type.
>
> If the TTS is not exhaustive (e.g. because `T = X<..>` wher `X` is
> implemented), it can fall back to the slower SubTypeTestCache implementation.
>
> Right now the STC fallback will get the loaded value of the type
> parameter for `dst_type` instead of the type parameter. Doing so is
> incorrect.
>
> => This CL ensures we preserve dst_type = TypeParameter for the STC
> fallback.
>
> Issue https://github.com/dart-lang/sdk/issues/39716
>
> Change-Id: Idea2405efbdc01c031ee68dbb345820e721533eb
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/127640
> Commit-Queue: Martin Kustermann <kustermann@google.com>
> Reviewed-by: Régis Crelier <regis@google.com>
TBR=vegorov@google.com,kustermann@google.com,regis@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: I79f2d7db8747743a778126088e375bb4d268c91a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/128020
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Régis Crelier <regis@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
When trying to inline SIMD shuffle operation in InlineSimdOp
(inliner.cc), FunctionEntry instruction was created and inherited
deopt_id and environment from call. Inheriting environment also
added FunctionEntry into env_use_list of other instructions.
If inlining didn't actualy happen due to invalid mask argument,
created FunctionEntry still left in the env_use_list.
Later optimization passes (for example AllocationSinking) may crash
while trying to update environment uses of instructions.
Fixes https://github.com/dart-lang/sdk/issues/39520
Change-Id: I6bdb7c6f7a7947259ea5ad097d05eebe5d73269b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/127802
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
If the dst_type of an AssertAssignable is a type parameter, the AssertAssignable
implementation will load the value of the type parameter using
the (instantiator or function) type arguments. It will then call the
type testing stub (TTS) of that type.
If the TTS is not exhaustive (e.g. because `T = X<..>` wher `X` is
implemented), it can fall back to the slower SubTypeTestCache implementation.
Right now the STC fallback will get the loaded value of the type
parameter for `dst_type` instead of the type parameter. Doing so is
incorrect.
=> This CL ensures we preserve dst_type = TypeParameter for the STC
fallback.
Issue https://github.com/dart-lang/sdk/issues/39716
Change-Id: Idea2405efbdc01c031ee68dbb345820e721533eb
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/127640
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Régis Crelier <regis@google.com>
Arguments to simd type constructors were being checked in the bootstrap
natives entrypoint but not elsewhere. Checking for null in Dart code
ensures that we don't accidentally miss these checks in AOT.
The changes to sdk_nnbd are required in order to ensure the number of
arguments passed to native code match the number of arguments from the
non-NNBD implementation (we needed to remove the implicit type args
parameter).
Fixes https://github.com/dart-lang/sdk/issues/39518
Change-Id: Iaf7d8790c154f1e85db613b6dc84004c8013df9a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/126905
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Ben Konyi <bkonyi@google.com>
Take full advantage of PushPair (stp) in code generation
visitors; use PushRegisterPair for ThrowErrorSlowPathCode.
This change improves Flutter Gallery code size by:
Instr: -0.24%
Total: -0.15%
Change-Id: Icba42cb956b07966861b52d3a8653d21eafab3b9
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/125971
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
Block R22 to hold a cached version of NullObject();
refresh it in the same way as for BARRIER_MASK register.
Alter assembler to avoid emitting load and to use NR
register directly instead.
This change improves Flutter Gallery code size by:
Instr: -1.81%.
Total: -1.16%.
Change-Id: Ifec654e799737527eec1d8a0e87b4a197ad0298a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/125662
Commit-Queue: Vyacheslav Egorov <vegorov@google.com>
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Rationale:
Loop analysis did not take outer indices used as bounds
(triangular loops) into account to enhance analysis. This
left some obvious BCE optimizations on the table (mainly AOT).
Observed improvements (performance and code size):
About 5-15% performance on golem benchmarks:
mainly armv7, and TypedData*Bench
Minor overall code size reduction on flutter gallery (arm32,arm64)
Comparing ./orig32.json (old) to ./new32.json (new)
Old : 6614912 bytes.
New : 6600336 bytes.
Change: -14576 bytes.
Comparing ./orig64.json (old) to ./new64.json (new)
Old : 6459216 bytes.
New : 6458736 bytes.
Change: -480 bytes.
Similar on velocity_tracker_bench (arm32,arm64)
Comparing orig32.json (old) to new32.json (new)
Old : 804280 bytes.
New : 794552 bytes.
Change: -9728 bytes.
Comparing orig64.json (old) to new64.json (new)
Old : 705520 bytes.
New : 705312 bytes.
Change: -208 bytes.
Change-Id: Ifc3b3a378ae6fd98743c48944a86a9e90ce4da5c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/117200
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Aart Bik <ajcbik@google.com>
Right now `Script::yield_positions()` is an array mapping function start
token positions to arrays. Those arrays contain token positions and are
indexed by yield index.
=> This `Script::yield_positions()` is not available in AOT mode.
The fast async stack implementation will need to be able to find out
where an async closure was suspended. It does so by looking at the
":await_jump_var", which contains the yield index. It then needs to
associate the yield index to token position of the yield.
This CL adds an entry into the PcDescriptors for every yield in a
async/async* function and removes `Script::yield_positions()`.
The entry will associate the yield index with the token position.
Flutter gallery total size impact for flutter-release:
- armv7: +0.016%
- armv8: +0.045%
Issue https://github.com/dart-lang/sdk/issues/37668
Change-Id: I0b2ce41e85d8f5d590201bf2fb091578d7379890
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/125408
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Clement Skau <cskau@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Runtime functions do not yet implement the NNBD semantics.
In a next change, generated code will pass NNBD mode to the runtime.
Declare one new VM flags: --strong-non-nullable-type-checks
Change-Id: I050a468e4a6b665ce46c345bafba04a947cf7cb0
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/124105
Commit-Queue: Régis Crelier <regis@google.com>
Reviewed-by: Siva Annamalai <asiva@google.com>
DART_PRECOMPILER define is used when building gen_snapshot and
run_vm_tests. Both gen_snapshot and run_vm_tests support AOT and
JIT modes, selected by FLAG_precompiled_mode.
Before this change, DART_PRECOMPILER always used AOT exception
handling (catch_entry_moves_maps). This works fine even if code is
generated in JIT mode, but it is not compatible with JIT exception
handling. run_vm_tests uses kernel service which comes from app-jit
snapshot (on x64), which is generated in JIT mode with JIT exception
handling. This causes crash if there is a Dart exception in kernel
service.
This change adjusts exception handling in DART_PRECOMPILER mode to
support both AOT and JIT implementations and select one using
FLAG_precompiled_mode.
Fixes https://github.com/dart-lang/sdk/issues/39388
Change-Id: I6c9ce05552a98bd4eeaea18c082a98e9bdbe9460
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/125302
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
--code-comments should be passed both at compile/training-time and run-time to get
comments in the disassembled instructions.
The required --disassemble-* flags need only be passed at run-time. Flow-graph printing
is not yet supported because we don't save flow-graph like assembly comments.
This functionality is excluded from product builds.
Change-Id: If35fbb08cff6eae40069f916ecec9c6f589c0cad
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/124989
Commit-Queue: Samir Jindel <sjindel@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This reverts commit ea57b1e62c.
Reason for revert: Please see https://buganizer.corp.google.com/issues/144304690
Original change's description:
> [VM/nnbd] Pass nullability when creating Class::DeclarationType.
>
> Passing nullability rather than forcing kLegacy helps slightly reduce the number
> of Types in the VM and in snapshots.
> This allows to remove the kludge about declaration type of Null being kNullable.
> Also pass nullability when creating a simple type may help.
>
> Note that CFE is still producing super types and some interface types that are
> non-nullable, even with nnbd features disabled. But this should get fixed and
> further reduce the number of types in the snapshot.
>
> Change-Id: I397f00902026ee5bc91d36328f4156427702efdb
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/124480
> Commit-Queue: Régis Crelier <regis@google.com>
> Reviewed-by: Alexander Markov <alexmarkov@google.com>
TBR=rmacnak@google.com,alexmarkov@google.com,asiva@google.com,regis@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: Icae0a69cda6cc531bba4d5d12471e7267a277340
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/125102
Reviewed-by: Siva Annamalai <asiva@google.com>
Passing nullability rather than forcing kLegacy helps slightly reduce the number
of Types in the VM and in snapshots.
This allows to remove the kludge about declaration type of Null being kNullable.
Also pass nullability when creating a simple type may help.
Note that CFE is still producing super types and some interface types that are
non-nullable, even with nnbd features disabled. But this should get fixed and
further reduce the number of types in the snapshot.
Change-Id: I397f00902026ee5bc91d36328f4156427702efdb
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/124480
Commit-Queue: Régis Crelier <regis@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
This field is only used to implement partial instantiation of generic
closures - so it is irrelevant in non-generic ones.
Keeping it null instead of empty_type_arguments allows to completely
elide a store (which saves around 2 instructions per closure allocation)
Change-Id: I1039993a1cfd1bc4b476214712d90b2c443e7cc6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/121983
Commit-Queue: Vyacheslav Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Victor Lima