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>
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>
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>
This reverts commit e16bb210d2.
Reason for revert: Indication this caused engine test
failures of the kind:
```
[0;32m[ RUN ] [mEmbedderA11yTest.A11yTreeIsConsistentUsingV1Callbacks
../../flutter/shell/platform/embedder/tests/embedder_a11y_unittests.cc:639: Failure
Expected equality of these values:
std::strncmp(kTooltip, node->tooltip, sizeof(kTooltip) - 1)
Which is: 116
0
```
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=runtime/vm/compiler/backend/il_test.cc
> 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
> Closes: https://github.com/dart-lang/sdk/issues/51618
> Change-Id: Ic5d017005dedcedea40c455c4d24dbe774f91603
> CoreLibraryReviewExempt: Internal FFI implementation changes
> 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-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/+/284300
> Commit-Queue: Daco Harkes <dacoharkes@google.com>
> Reviewed-by: Alexander Markov <alexmarkov@google.com>
> Reviewed-by: Martin Kustermann <kustermann@google.com>
Change-Id: Icc87a6ca33bffecabb15c6b168a06ccc38c2fe5b
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-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
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/333840
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
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=runtime/vm/compiler/backend/il_test.cc
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
Closes: https://github.com/dart-lang/sdk/issues/51618
Change-Id: Ic5d017005dedcedea40c455c4d24dbe774f91603
CoreLibraryReviewExempt: Internal FFI implementation changes
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-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/+/284300
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Adds a `CachableIdempotentCallInstr` that can be invoked via
`@pragma('vm:cachable-idempotent')` if the call-sites is force
optimized.
The object pool is not visited by the scavenger. So, we store the
results as unboxed integers. Consequently, only dart functions that
return integers can be cached.
Cachable idempotent calls should never be inlined. After the first
call the function not be called again.
The call itself is on a slow path to avoid register spilling on the
fast path.
TEST=vm/cc/IRTest_CachableIdempotentCall
TEST=runtime/tests/vm/dart/cachable_idempotent_test.dart
Bug: https://github.com/dart-lang/sdk/issues/51618
Change-Id: I612e896f27add76f57796c060157e14cc687a0fd
Cq-Include-Trybots: luci.dart.try:vm-aot-android-release-arm64c-try,vm-aot-android-release-arm_x64-try,vm-aot-asan-linux-release-x64-try,vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv64-try,vm-aot-mac-release-arm64-try,vm-aot-mac-release-x64-try,vm-aot-msan-linux-release-x64-try,vm-aot-obfuscate-linux-release-x64-try,vm-aot-tsan-linux-release-x64-try,vm-aot-ubsan-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-kernel-linux-debug-x64-try,vm-kernel-precomp-linux-release-x64-try,vm-linux-debug-ia32-try,vm-linux-debug-simriscv64-try,vm-linux-debug-x64-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-tsan-linux-release-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/+/301601
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Follow up of:
* https://dart-review.googlesource.com/c/sdk/+/328861
* https://dart-review.googlesource.com/c/sdk/+/331087
* https://dart-review.googlesource.com/c/sdk/+/331120
The pool entries for switchable calls have their pool entries
transformed during snapshot loading in AOT. This behavior is now folded
into `SnapshotBehavior`.
Because snapshot behavior is set on the call sites, and the call sites
are already AOT specific, no conditionals inside app_snapshot.cc are
needed anymore.
The Arm64 implementation does not use the assembler for implicitly
adding pool entries, it does so explictly and loads two pool entries
in one load.
We could consider doing these explicit loads in the other architectures
as well. This would avoid adding the snapshot_behavior argument to
`LoadUnqiueObject` for the other architectures.
IA32 has no AOT support.
TEST=precomp bots
Change-Id: I75f70e00ca44f6545694d9b59e3ad08ad8bb3f8e
Cq-Include-Trybots: luci.dart.try:vm-aot-android-release-arm64c-try,vm-aot-android-release-arm_x64-try,vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv64-try,vm-aot-linux-debug-x64-try,vm-aot-linux-release-arm64-try,vm-aot-linux-release-simarm_x64-try,vm-ffi-qemu-linux-release-riscv64-try,vm-linux-debug-simriscv64-try,vm-aot-mac-product-arm64-try,vm-aot-win-debug-arm64-try,vm-ffi-android-debug-arm64c-try,vm-ffi-android-debug-arm-try,vm-ffi-qemu-linux-release-arm-try,vm-aot-win-debug-x64c-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/331101
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
`NativeCallInstr::EmitNativeCode` `link_lazily` code path uses the
new snapshotting behavior, and the non-`link_lazily` checks that it's
never snapshotted.
The snapshot behavior argument is threaded from the the EmitNativeCode
to the places where the pool entry is constructed.
Does not yet migrate `kSwitchableCallMissEntryPoint` and
`kMegamorphicCallEntryPoint`.
TEST=CI
Change-Id: I8dd1dc5aa525f888b493856c60ae385087463d02
Cq-Include-Trybots: luci.dart.try:vm-ffi-qemu-linux-release-riscv64-try,vm-linux-debug-simriscv64-try,vm-aot-linux-release-arm64-try,vm-aot-android-release-arm64c-try,vm-aot-win-debug-arm64-try,vm-aot-mac-product-arm64-try,vm-mac-debug-arm64-try,vm-ffi-android-debug-arm-try,vm-ffi-qemu-linux-release-arm-try,vm-linux-debug-ia32-try,vm-linux-debug-x64-try,vm-aot-linux-debug-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/328861
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This is a reland of commit 06d7a2352e
This version fixes an issue when a phi node has multiple inputs with
different unboxed integer representations. The original CL made a
change where only the representations were considered, not the range
of values for the Phi calculated by range analysis. The reland goes
back to the old behavior for this case.
Also fixes the new tests on 32-bit architectures.
Original change's description:
> [vm/compiler] Change MemoryCopy to also take untagged addresses.
>
> This CL adds the ability to pass the payload address of the source
> and destination directly to the MemoryCopy instruction as an untagged
> value.
>
> The new translation of the _TypedListBase._memMoveN methods use the new
> MemoryCopy constructor, retrieving the untagged value of the data field
> of both the source and destination. This way, if inlining exposes the
> allocation of the object from which the data field is being retrieved,
> then allocation sinking can remove the intermediate allocation if there
> are no escaping uses of the object.
>
> Since Pointer.asTypedList allocates such ExternalTypedData objects,
> this CL makes that method inlined if at all possible, which removes
> the intermediate allocation if the only use of the TypedData object
> is to call setRange for memory copying purposes.
>
> This CL also separates unboxed native slots into two groups: those
> that contain untagged addresses and those that do not. The former
> group now have the kUntagged representation, which mimics the old
> use of LoadUntagged for the PointerBase data field and also ensures
> that any arithmetic operations on untagged addresses must first be
> explicitly converted to an unboxed integer and then explicitly converted
> back to untagged before being stored in a slot that contains untagged
> addresses.
>
> When a unboxed native slot that contains untagged addresses is defined,
> the definition also includes a boolean which represents whether
> addresses that may be moved by the GC can be stored in this slot or not.
> The redundancy eliminator uses this to decide whether it is safe to
> eliminate a duplicate load, replace a load with the value originally
> stored in the slot, or lift a load out of a loop.
>
> In particular, the PointerBase data field may contain GC-moveable
> addresses, but only for internal TypedData objects and views, not
> for external TypedData objects or Pointers. To allow load optimizations
> involving the latter, the LoadField and StoreField instructions now
> take boolean flags for whether loads or stores from the slot are
> guaranteed to not be GC-moveable, to override the information from
> the slot argument.
>
> Notable benchmark changes on x64 (similar for other archs unless noted):
>
> JIT:
> * FfiMemory.PointerPointer: 250.7%
> * FfiStructCopy.Copy1Bytes: -26.73% (only x64)
> * FfiStructCopy.Copy32Bytes: -25.18% (only x64)
> * MemoryCopy.64.setRange.Pointer.Uint8: 19.36%
> * MemoryCopy.64.setRange.Pointer.Double: 18.96%
> * MemoryCopy.8.setRange.Pointer.Double: 17.59%
> * MemoryCopy.8.setRange.Pointer.Uint8: 19.46%
>
> AOT:
> * FfiMemory.PointerPointer: 323.5%
> * FfiStruct.FieldLoadStore: 483.3%
> * FileIO_readwrite_64kb: 15.39%
> * FileIO_readwrite_512kb (Intel Xeon): 46.22%
> * MemoryCopy.512.setRange.Pointer.Uint8: 35.20%
> * MemoryCopy.64.setRange.Pointer.Uint8: 55.40%
> * MemoryCopy.512.setRange.Pointer.Double: 29.45%
> * MemoryCopy.64.setRange.Pointer.Double: 60.37%
> * MemoryCopy.8.setRange.Pointer.Double: 59.54%
> * MemoryCopy.8.setRange.Pointer.Uint8: 55.40%
> * FfiStructCopy.Copy32Bytes: 398.3%
> * FfiStructCopy.Copy1Bytes: 1233%
>
> TEST=vm/dart/address_local_pointer, vm/dart/pointer_as_typed_list
>
> Issue: https://github.com/dart-lang/sdk/issues/42072
> Fixes: https://github.com/dart-lang/sdk/issues/53124
>
> Cq-Include-Trybots: luci.dart.try:vm-ffi-qemu-linux-release-arm-try,vm-eager-optimization-linux-release-x64-try,vm-linux-release-x64-try,vm-linux-debug-x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-debug-x64-try
> Change-Id: I563e0bfac5b1ac6cf1111649934067c12891b631
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/324820
> Reviewed-by: Alexander Markov <alexmarkov@google.com>
> Commit-Queue: Tess Strickland <sstrickl@google.com>
> Reviewed-by: Martin Kustermann <kustermann@google.com>
TEST=vm/dart/address_local_pointer, vm/dart/pointer_as_typed_list
Issue: https://github.com/dart-lang/sdk/issues/42072
Fixes: https://github.com/dart-lang/sdk/issues/53124
Change-Id: Iabb0e910f12636d0ff51e711c8c9c98ad40e5811
Cq-Include-Trybots: luci.dart.try:vm-ffi-qemu-linux-release-arm-try,vm-eager-optimization-linux-release-x64-try,vm-linux-release-x64-try,vm-linux-debug-x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-debug-x64-try,vm-aot-linux-release-simarm_x64-try,vm-aot-linux-debug-simarm_x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/330600
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This reverts commit 06d7a2352e.
Reason for revert: everything crashes on vm-aot-linux-debug-simarm_x64
Original change's description:
> [vm/compiler] Change MemoryCopy to also take untagged addresses.
>
> This CL adds the ability to pass the payload address of the source
> and destination directly to the MemoryCopy instruction as an untagged
> value.
>
> The new translation of the _TypedListBase._memMoveN methods use the new
> MemoryCopy constructor, retrieving the untagged value of the data field
> of both the source and destination. This way, if inlining exposes the
> allocation of the object from which the data field is being retrieved,
> then allocation sinking can remove the intermediate allocation if there
> are no escaping uses of the object.
>
> Since Pointer.asTypedList allocates such ExternalTypedData objects,
> this CL makes that method inlined if at all possible, which removes
> the intermediate allocation if the only use of the TypedData object
> is to call setRange for memory copying purposes.
>
> This CL also separates unboxed native slots into two groups: those
> that contain untagged addresses and those that do not. The former
> group now have the kUntagged representation, which mimics the old
> use of LoadUntagged for the PointerBase data field and also ensures
> that any arithmetic operations on untagged addresses must first be
> explicitly converted to an unboxed integer and then explicitly converted
> back to untagged before being stored in a slot that contains untagged
> addresses.
>
> When a unboxed native slot that contains untagged addresses is defined,
> the definition also includes a boolean which represents whether
> addresses that may be moved by the GC can be stored in this slot or not.
> The redundancy eliminator uses this to decide whether it is safe to
> eliminate a duplicate load, replace a load with the value originally
> stored in the slot, or lift a load out of a loop.
>
> In particular, the PointerBase data field may contain GC-moveable
> addresses, but only for internal TypedData objects and views, not
> for external TypedData objects or Pointers. To allow load optimizations
> involving the latter, the LoadField and StoreField instructions now
> take boolean flags for whether loads or stores from the slot are
> guaranteed to not be GC-moveable, to override the information from
> the slot argument.
>
> Notable benchmark changes on x64 (similar for other archs unless noted):
>
> JIT:
> * FfiMemory.PointerPointer: 250.7%
> * FfiStructCopy.Copy1Bytes: -26.73% (only x64)
> * FfiStructCopy.Copy32Bytes: -25.18% (only x64)
> * MemoryCopy.64.setRange.Pointer.Uint8: 19.36%
> * MemoryCopy.64.setRange.Pointer.Double: 18.96%
> * MemoryCopy.8.setRange.Pointer.Double: 17.59%
> * MemoryCopy.8.setRange.Pointer.Uint8: 19.46%
>
> AOT:
> * FfiMemory.PointerPointer: 323.5%
> * FfiStruct.FieldLoadStore: 483.3%
> * FileIO_readwrite_64kb: 15.39%
> * FileIO_readwrite_512kb (Intel Xeon): 46.22%
> * MemoryCopy.512.setRange.Pointer.Uint8: 35.20%
> * MemoryCopy.64.setRange.Pointer.Uint8: 55.40%
> * MemoryCopy.512.setRange.Pointer.Double: 29.45%
> * MemoryCopy.64.setRange.Pointer.Double: 60.37%
> * MemoryCopy.8.setRange.Pointer.Double: 59.54%
> * MemoryCopy.8.setRange.Pointer.Uint8: 55.40%
> * FfiStructCopy.Copy32Bytes: 398.3%
> * FfiStructCopy.Copy1Bytes: 1233%
>
> TEST=vm/dart/address_local_pointer, vm/dart/pointer_as_typed_list
>
> Issue: https://github.com/dart-lang/sdk/issues/42072
> Fixes: https://github.com/dart-lang/sdk/issues/53124
>
> Cq-Include-Trybots: luci.dart.try:vm-ffi-qemu-linux-release-arm-try,vm-eager-optimization-linux-release-x64-try,vm-linux-release-x64-try,vm-linux-debug-x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-debug-x64-try
> Change-Id: I563e0bfac5b1ac6cf1111649934067c12891b631
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/324820
> Reviewed-by: Alexander Markov <alexmarkov@google.com>
> Commit-Queue: Tess Strickland <sstrickl@google.com>
> Reviewed-by: Martin Kustermann <kustermann@google.com>
Issue: https://github.com/dart-lang/sdk/issues/42072
Change-Id: I7c31434e01108487de69a32154bbefd1538c6f0f
Cq-Include-Trybots: luci.dart.try:vm-ffi-qemu-linux-release-arm-try,vm-eager-optimization-linux-release-x64-try,vm-linux-release-x64-try,vm-linux-debug-x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-debug-x64-try
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/330523
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Alexander Aprelev <aam@google.com>
Auto-Submit: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
This CL adds the ability to pass the payload address of the source
and destination directly to the MemoryCopy instruction as an untagged
value.
The new translation of the _TypedListBase._memMoveN methods use the new
MemoryCopy constructor, retrieving the untagged value of the data field
of both the source and destination. This way, if inlining exposes the
allocation of the object from which the data field is being retrieved,
then allocation sinking can remove the intermediate allocation if there
are no escaping uses of the object.
Since Pointer.asTypedList allocates such ExternalTypedData objects,
this CL makes that method inlined if at all possible, which removes
the intermediate allocation if the only use of the TypedData object
is to call setRange for memory copying purposes.
This CL also separates unboxed native slots into two groups: those
that contain untagged addresses and those that do not. The former
group now have the kUntagged representation, which mimics the old
use of LoadUntagged for the PointerBase data field and also ensures
that any arithmetic operations on untagged addresses must first be
explicitly converted to an unboxed integer and then explicitly converted
back to untagged before being stored in a slot that contains untagged
addresses.
When a unboxed native slot that contains untagged addresses is defined,
the definition also includes a boolean which represents whether
addresses that may be moved by the GC can be stored in this slot or not.
The redundancy eliminator uses this to decide whether it is safe to
eliminate a duplicate load, replace a load with the value originally
stored in the slot, or lift a load out of a loop.
In particular, the PointerBase data field may contain GC-moveable
addresses, but only for internal TypedData objects and views, not
for external TypedData objects or Pointers. To allow load optimizations
involving the latter, the LoadField and StoreField instructions now
take boolean flags for whether loads or stores from the slot are
guaranteed to not be GC-moveable, to override the information from
the slot argument.
Notable benchmark changes on x64 (similar for other archs unless noted):
JIT:
* FfiMemory.PointerPointer: 250.7%
* FfiStructCopy.Copy1Bytes: -26.73% (only x64)
* FfiStructCopy.Copy32Bytes: -25.18% (only x64)
* MemoryCopy.64.setRange.Pointer.Uint8: 19.36%
* MemoryCopy.64.setRange.Pointer.Double: 18.96%
* MemoryCopy.8.setRange.Pointer.Double: 17.59%
* MemoryCopy.8.setRange.Pointer.Uint8: 19.46%
AOT:
* FfiMemory.PointerPointer: 323.5%
* FfiStruct.FieldLoadStore: 483.3%
* FileIO_readwrite_64kb: 15.39%
* FileIO_readwrite_512kb (Intel Xeon): 46.22%
* MemoryCopy.512.setRange.Pointer.Uint8: 35.20%
* MemoryCopy.64.setRange.Pointer.Uint8: 55.40%
* MemoryCopy.512.setRange.Pointer.Double: 29.45%
* MemoryCopy.64.setRange.Pointer.Double: 60.37%
* MemoryCopy.8.setRange.Pointer.Double: 59.54%
* MemoryCopy.8.setRange.Pointer.Uint8: 55.40%
* FfiStructCopy.Copy32Bytes: 398.3%
* FfiStructCopy.Copy1Bytes: 1233%
TEST=vm/dart/address_local_pointer, vm/dart/pointer_as_typed_list
Issue: https://github.com/dart-lang/sdk/issues/42072
Fixes: https://github.com/dart-lang/sdk/issues/53124
Cq-Include-Trybots: luci.dart.try:vm-ffi-qemu-linux-release-arm-try,vm-eager-optimization-linux-release-x64-try,vm-linux-release-x64-try,vm-linux-debug-x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-debug-x64-try
Change-Id: I563e0bfac5b1ac6cf1111649934067c12891b631
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/324820
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
The library name cache is dead.
The library exported name cache is used indirectly through Dart_Invoke etc, and is practically only used for one-shot lookups like `main` or the embedder's startup hooks. The lookups that were worth caching now happen in the Dart frontend instead of the VM.
Cf. b9045af57b
TEST=ci
Change-Id: Ibd8bd30fbba3e9af32e3fbf3c5c4ced51a8d15d7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/324401
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Lookup in the library import can be very slow and it is not needed
most of the time. This change makes Library::Lookup{Class,Function,Field}[AllowPrivate]
lookups local in the library and cleans up related or unused code.
TEST=ci
Change-Id: If5a99bce7d712ce3a898395f74d4febac0311048
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/323221
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
The main purpose of the low-level [PortMap] is to coordinate between
ports being opened & closed and concurrent message senders. That's the
only thing it should do.
Each isolate owns [ReceivePort]s. Only the isolate mutator can create
ports, delete them or change their "keeps-isolate-alive" state. Right
now it requires going via [PortMap] (which acquires lock) and
[MessageHandler] (which acquires lock) to change the
"keeps-isolate-alive" state of a port.
We'll move information whether a dart [ReceivePort] is closed and
whether it keeps the isolate alive into the [ReceivePort] object itself.
=> Changing the "keeps-isolate-alive" state of the port no longer
requires any locks. We could even avoid the runtime call itself in a
future CL.
Isolates are kept alive if there's any open receive ports (that have not
been marked as "does not keep isolate alive"). This is a property of an
isolate not of the message handler. For native message handlers we do
have a 1<->1 correspondence between port and handler (i.e. there's no
"number of open ports" tracking needed).
=> We'll move the logic of counting open receive ports and ports that
keep the isolate alive to the [Isolate].
=> We'll also remove locking around incrementing/decrementing or
accessing the counts.
=> The [IsolateMessageHandler] will ask the [Isolate] whether there's
any open ports for determining whether to shut down.
=> For native ports, the `Dart_NewNativePort()` & `Dart_CloseNativePort()`
functions will manage the lifetime (as their name also suggests).
Overall this makes the [Isolate] responsible for creation of dart
[ReceivePort]s and tracking whether the isolate should be kept alive:
* Isolate::CreateReceivePort()
* Isolate::SetReceivePortKeepAliveState()
* Isolate::CloseReceivePort()
TEST=ci
Change-Id: I847ae357c26254d3810cc277962e05deca18a1de
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/317960
Reviewed-by: Alexander Aprelev <aam@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
All uses of `PortMap::Create()` would call `PortMap::SetPortState()`
after creating the port. To avoid acquiring the port map lock twice, we
can initialize the state already in `PortMap::Create()`.
Additionally we fix an existing issue by making the isolate's main port
a control port (instead of leaving it as a `kNewPort`).
As a side-effect we'll no longer have any allocated ports with state
`kNewPort`.
We also remove the unused `PortMap::IsLocalPort()`
TEST=ci
Change-Id: I6198792a8e1132580b60262f10a807c5b12f9eaf
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/317680
Reviewed-by: Alexander Aprelev <aam@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
When evaluating an expressino at a breakpoint that's inside a closure,
the closure may refer to anything it closes over. That includes the
`this` from an enclosing method.
So depending on whether a closure's parent chain is an instance method
or a static method, the expression evaluation function is going to be an
instance method or a static method.
=> We walk up the parent chain to determine the correct enclosing class.
=> This avoids making a Type object with a cid from a top-level class (see [0])
Handling this correctly will now try to get the `this` when invoking an
eval function that has an enclosing instance method. Though we may often
have "<optimized out>" the `this` (e.g. due to not capturing it in
closure context chain).
=> We still allow running the expression evaluation function in this
case, but only if the expression being evaluated doesn't access `this`.
A similar issue occurs when trying to use variables in the eval
expression that the closure didn't capture. This results in a confusing
CFE compile-time error. This is a separate issue and tracked in [1].
=> We update the test to distinuish the cases that this CL makes passing
and those that are failing due to [1].
Fixes [0] https://github.com/dart-lang/sdk/issues/53061
See also [1] https://github.com/dart-lang/sdk/issues/53087
TEST=Fixes part of service/evaluate_activation_test failures
Change-Id: I3bb24e7338c7b2f12d5340311d944cb59a455641
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/317540
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
This change is almost trivial. The closure is stored on the callback's
RawReceivePort, not in the VM. So we can basically just remove the CFE
check and it pretty much works. The only problem is that we can't set
function.FfiCallbackTarget anymore, so most of the CL is dealing with
that.
A few places were deciding whether an FFI trampoline was a call or a
callback based on whether function.FfiCallbackTarget() was null. But
now the target will be null for async callbacks. So instead I've added
a new value to the FfiCallbackKind enum (and renamed it), and changed
those checks.
Sync callback closures will be a separate CL, because they're more
complicated.
Bug: https://github.com/dart-lang/sdk/issues/52689
Change-Id: I8e5dfb557362e679f66195b735c3c382e6792840
TEST=async_void_function_callbacks_test.dart
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/316160
Commit-Queue: Liam Appelbe <liama@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
a) Remove 2 fields in our object representation:
* PatchClass::library_data_
* Library::kernel_data_
=> This saves O(#libraries + #patch-classes) which can amount up to
10+ MB for big apps, as we save not just the slots but the
[ExternalTypedData] objects they used to reference.
Instead we'll compute the KernelLibraryData when we need it by making a
[TypedDataView] of sub parts of the component.
b) We make a kernel binary be represented by a single [ExternalTypedData].
Whenever we need a sub-part of a kernel binary (e.g. one for each
component for concatinated kernels, or one for a library inside a
component) we use proper [TypedDataView]s for that.
c) As we sometimes need to create a view of only a particular library
within a component we need to find start/end of a library based on
index.
=> We store the library index instead of the library offset on
Library/PatchClass.
=> We can easily derive the start/end of a library from it's index by
looking at the kernel component encoding.
d) We make the [Reader] object work purely based on a pointer - instead
of making it have if/else when reading bytes (either from pointer or
from a view).
e) We make the [KernelProgramInfo] store the kernel_component and
various TD views into it.
TEST=ci
Change-Id: Ibe160881ff48635e834c3d647a977a144b5d0565
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/313561
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
** KernelProgramInfo
The VM can load multiple kernel files into an isolate. Each kernel file
is represented as a [KernelProgramInfo] object. Loading a kernel blob
will result in a tree of objects
Library
- TopLevelClass
+ Field
+ Function
+ Class
+ Field
+ Function
+ ... closure functions ...
The entire tree belongs to one [KernelProgramInfo] object. There are
two exceptions
* after a hot-reload we inject [PatchClass]es (as owners of
old [Field]/[Function] objects) that identify the old kernel data
* expression evaluation functions will get a special data array with
information about the expression evaluation kernel data
=> We attach the [KernelProgramInfo] to the root of the tree, namely
the [Library]
=> We introduce `{Field,Function,Class,Library}::KernelProgramInfo()`
that will get that object - by walking up the chain.
=> We introduce `UntaggedPatchClass::kernel_program_info_` field that
allows saving the old [KernelProgramInfo] (on which the old
[Field]/[Function] objects are based upon) in the [PatchClass].
=> We include the [KernelProgramInfo] in the expression evaluation
data array.
** Scripts
A [Script] object should represent a .dart source file. It contains the
source, line offsets, etc. of that dart source file. The contents of a
dart source file may be used by many classes/libraries or even many
kernel blobs (e.g. via our copying mixin application).
When we build a tree like above, we'll attach a reference to a
[Script] object to [Class] objects. But members of a class may come from
a different script, in which case we inject an intermediary [PatchClass]
that references that other script.
When the compiler frontend loads kernel, builds flow graph, etc we walk
down the kernel binary (the tree above) - all belonging to the same
[KernelProgramInfo] object, but sub-parts may belong to different
[Script]s.
=> We use the existing [ActiveClass] class that keeps track of the
current class & member inside the class we're operating on.
=> Whenever we need the script we get it from the active member
function/class.
** Misc
We make kernel related classes (e.g. [KernelReaderHelper]) independent
of [Script].
We make the [Script] objects hold on to the [KernelProgramInfo] they are
based upon (e.g. for lazy source reading). But since a given [Script]
doesn't belong to a particular kernel blob (multiple kernel blobs may
contain information about the same script) we don't expose this in the
public API.
=> In the future we could have a unique [Script] object per url (instead
of having several as we do now)
TEST=ci
Change-Id: I7bd698f497b0bd8ab8e546540bb1dad5b5a6dd48
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/313381
Reviewed-by: Jens Johansen <jensj@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Goals of this CL, a followup to 2f63acea22:
* Ensure that X::New() depends on the initialization guarantees of
Object::Allocate<X>(...):
* Ptr fields are guaranteed to be initialized to Object::null().
* Non-Ptr fields are guaranteed to be zero-initialized.
In cases where the only uses of the allocated object before return
was to perform unnecessary field assignments, X::New() just simply
returns the result of Object::Allocate<X>(...).
Otherwise, the old now-unnecessary assignments have been changed
into ASSERTs so that they will be checked in DEBUG mode.
* Ensure that NoSafepointScopes are entered in X::New() only when
necessary (e.g., to ensure fields used to calculate to(...)
are properly set before being seen by pointer visitors as the GC
may run when outside a NoSafepointScope).
In particular, the often occurring pattern:
auto& result = X::Handle();
{
auto raw = Object::Allocate<X>(...);
NoSafepointScope no_safepoint;
result = raw;
}
...
has been replaced with:
const auto& result = X::Handle(Object::Allocate<X>(...));
* If a handle was allocated, the only uses of that handle before
returning must be performed under a NoSafepointScope, and the same
operations can be done directly on the object pointer, then do so
and remove the unnecessary handle allocation.
Notable changes outside the above:
* Swapped ObjectPool::EntryType::{kImmediate,kTaggedObject} so that
kImmediate has value 0, since Object::Allocate<ObjectPool>(len)
zero-initializes the payload and without this change,
ObjectPool::New() must set the entry types manually.
* Removed the old static ArrayPtr cached_array_ field on
SubtypeTestCache as well as SubtypeTestCache::{Init,Cleanup} and
instead added Object::empty_subtype_test_cache_array().
* Removed the no-arg Closure::New() method, which is never used.
* Inlined the no-arg Script::New() method into its only caller,
one of the other Script::New() overloads.
TEST=ci
Issue: https://github.com/dart-lang/sdk/issues/52876
Change-Id: I079b4c9f73c7d2c0146c30cf2cd570b91a1ecf36
Cq-Include-Trybots: luci.dart.try:vm-linux-debug-x64-try,vm-linux-debug-x64c-try,vm-linux-release-x64-try,vm-aot-linux-product-x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-debug-x64-try,vm-ffi-qemu-linux-release-riscv64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/313120
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
There have been multiple cases in the past where using Object::null()
as the initializer, even for non-Ptr fields, has caused nondeterministic
snapshot creation. In particular, this can happen if a non-Ptr field
is only used in some versions of an object but not others, and thus
there is no reason to worry about what the value of the field is during
runtime. Only when snapshotting can it become relevant that the field
value is a portion of the address of Object::null(), which can vary
over different runs.
Instead of initializing the entirety of allocated objects (outside
of a few rare cases) with Object::null(), only initialize the portion
of the object corresponding to object pointer fields (in most cases,
between from() and to() inclusively) to Object::null() and initialize
the rest of the object to 0.
With this change, the only special casing in Object::InitializeObject
that remains is to skip initialization for TypedData and Arrays when
the memory is guaranteed to be zero initialized and to initialize the
contents of Instructions objects with the break instruction, not 0.
Note that this behavior does not occur in the following cases, which
all involve copying an arbitrary object:
* Object::Clone(), which zero-initializes the contents of the object
in all cases. However, the contents of the original object are then
copied over before the thread can reach a safepoint.
* Inside the ObjectGraphCopier, which uses the old initialization
behavior. This is safe, as any GC-important fields are immediately
copied over, and the rest of the contents are eventually copied over
before the caller receives the root of the copied object graph.
TEST=ci
Issue: https://github.com/dart-lang/sdk/issues/52876
Change-Id: Ib09fc562a8b6af97b509b493eb2d64109230ec35
Cq-Include-Trybots: luci.dart.try:vm-linux-debug-x64-try,vm-linux-debug-x64c-try,vm-linux-release-x64-try,vm-aot-linux-product-x64-try,vm-aot-linux-release-x64-try,vm-aot-linux-debug-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312900
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Without this, compiling with MSVC fails with the error:
```
error C2397: conversion from 'int' to 'uint8_t' requires a narrowing conversion
```
TEST=vm-msvc-windows builds
Cq-Include-Trybots: luci.dart.try:vm-msvc-windows-try
Change-Id: I157b8ac9f6f6aaaf31b71d2c0a5f5469d6dc21ad
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312262
Auto-Submit: Tess Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
Previously, Class::RareType() returned a Type where the type arguments
were null (e.g., all dynamic). However, this is an invalid type for
classes that have at least one type parameter with a bound that is not a
top type, and could mean that comparisons (such as subtyping) against
the "rare" type could return incorrect answers.
Instead, use TypeParameters::defaults() to get the canonicalized
instantiated to bounds type argument vector and use that instead.
TEST=ci
Cq-Include-Trybots: luci.dart.try:vm-reload-linux-debug-x64-try,vm-reload-rollback-linux-debug-x64-try,vm-linux-debug-x64-try,vm-linux-release-x64-try
Change-Id: Iea591e93102f53713265b481476f9670cfb81c93
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312261
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
This reverts commit 1cbb5debeb.
Reason for revert: ../../runtime/vm/compiler/aot/precompiler.cc: 3113: error: expected: api_uses_.HasKey(*entry)
Original change's description:
> [vm] Improve Class::Hash by also hashing its library
>
> Class::Hash previously only hashed class name. This could cause
> hash collisions for top-level classes which have the same name.
>
> Function::Hash uses Class::Hash, so it could also
> suffer from hash collisions among top-level functions.
>
> This change adds hashing of library URI when calculating
> Class::Hash / Function::Hash.
>
> TEST=ci
>
> Change-Id: I8bed681ecb8b8a6b0fceb99866c551ff19a36ab7
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311930
> Reviewed-by: Martin Kustermann <kustermann@google.com>
> Commit-Queue: Alexander Markov <alexmarkov@google.com>
Change-Id: I0ec3746d2756cd681695cf4b4304733783df7192
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312060
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Class::Hash previously only hashed class name. This could cause
hash collisions for top-level classes which have the same name.
Function::Hash uses Class::Hash, so it could also
suffer from hash collisions among top-level functions.
This change adds hashing of library URI when calculating
Class::Hash / Function::Hash.
TEST=ci
Change-Id: I8bed681ecb8b8a6b0fceb99866c551ff19a36ab7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311930
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
The main contribution of this CL is unification of disparate
handling of various functions like `Future.timeout`,
`Future.wait`, `_SuspendState.createAsyncCallbacks` and
`_SuspendState._createAsyncStarCallback` into a single
`@pragma('vm:awaiter-link')` which allows Dart developers
to specify where awaiter unwinder should look for the next
awaiter.
For example this allows unwinding to succeed for the code like this:
Future<int> outer(Future<int> inner) {
@pragma('vm:awaiter-link')
final completer = Completer<int>();
inner.then((v) => completer.complete(v));
return completer.future;
}
This refactoring also ensures that we preserve information
(including Function & Code objects) required for awaiter
unwinding across all modes (JIT, AOT and AOT with DWARF stack
traces). This guarantees users will get the same information
no matter which mode they are running in. Previously
we have been disabling awaiter_stacks tests in some AOT
modes - which led to regressions in the quality of produced
stacks.
This CL also cleans up relationship between debugger and awaiter
stack returned by StackTrace.current - which makes stack trace
displayed by debugger (used for stepping out and determinining
whether exception is caught or not) and `StackTrace.current`
consistent.
Finally we make one user visible change to the stack trace:
awaiter stack will no always include intermediate listeners
created through `Future.then`. Previously we would sometimes
include these listeners at the tail of the stack trace,
which was inconsistent.
Ultimately this means that code like this:
Future<int> inner() async {
await null; // asynchronous gap
print(StackTrace.current); // (*)
return 0;
}
Future<int> outer() async {
int process(int v) {
return v + 1;
}
return await inner().then(process);
}
void main() async {
await outer();
}
Produces stack trace like this:
inner
<asynchronous suspension>
outer.process
<asynchronous suspension>
outer
<asynchronous suspension>
main
<asynchronous suspension>
And when stepping out of `inner` execution will stop at `outer.process`
first and the next step out will bring execution to `outer` next.
Fixes https://github.com/dart-lang/sdk/issues/52797
Fixes https://github.com/dart-lang/sdk/issues/52203
Issue https://github.com/dart-lang/sdk/issues/47985
TEST=ci
Bug: b/279929839
CoreLibraryReviewExempt: CL just adds @pragma to facilitate unwinding
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-product-x64-try,vm-aot-linux-debug-x64-try,vm-aot-linux-release-x64-try,vm-aot-obfuscate-linux-release-x64-try,vm-aot-dwarf-linux-product-x64-try
Change-Id: If377d5329d6a11c86effb9369dc603a7ae616fe7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311680
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
The [PatchClass] no longer needs to refer to a original [Class] and a
patched [Class] objects: Since the CFE does handling of patches the VM
only has a single [Class] object, so those two fields are always the
same.
=> Change `PatchClass::{origin_class,patched_class} -> `PatchClass::wrapped_class`
We also remove the `Field::Origin()` / `Function::origin()` methods
as they return the same as the `Owner()` would return.
TEST=ci
Change-Id: Iec0849f6ffc2026760dad89a9bcf07e9469bc8b6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311840
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
Root cause is that `Function::Hash()` depends on the owner class
id. This causes two issues
a) Issue with reload:
During reload the reloader code will change `Function::owner()`
from a `Class` to a `PatchClass`. This in return will cause a change in
the function's hash, which means we won't find the function in the
closure cache anymore.
b) Issue with app-jit:
As part of app-jit we re-number class ids. This re-numbering would also
render the closure cache as invalid - as the function hash code would
change. So instead the closure cache was cleared entirely. While this
works for AOT, in app-jit we may serialize closure functions and their
code. Though at app-jit runtime we then fail to find the functions in
the closure function cache (as it was cleared during snapshoting)
(This will surface by us triggering an optimizing compile of a function
in the background compiler which will try to make a new `Function`
object for a closure, even though the app-jit snapshot already has one)
To fix these issues we make the `Function::Hash()` stable across reload
* we use `Function::Owner()` instead of `Function::owner()` (the
former will always return the actual class, the ladder can return
the `PatchClass`)
* we avoid re-hashing the closure cache by making `Function::Hash()`
use the name of the class instead of the class id
* we clear out the closure cache array during AOT snapshoting, as this
is not used in AOT runtime.
Closes https://github.com/dart-lang/sdk/issues/52803
TEST=Fixes various issues on CI.
Change-Id: I352d0a768df0f29d504cdd80e3533cbedc437b9b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311744
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
This field is now only used when reloading to check assignability
during `ProgramReloadContext::ReloadPhase4CommitFinish()` (via
`PostCommit()` -> `InvalidateWorld()` -> `InvalidateFields`).
However, the method `DeoptimizeDependentCode()` is called near the
start of the actual reloading process in `Reload` prior to the
commit/rollback phases. This calls `DeoptimizeTypeTestingStubs()`, which
walks the heap and resets the contents of all `SubtypeTestCaches`
(`STC`s).
That means that the `STC` information isn't actually kept between
different reloads, so all this does is cause a number of `STC` objects
to be allocated during reload whose backing arrays are not used
post-reload for anything and so are unnecessary garbage.
That, and the creation of the `STC`s during `InvalidateFields` creates
all-input `STC`s, as they don't just check assignability to a given field type but also for checking elements of `RecordTypes`. That means
the per-field `STC`s contain way more information needed to check for
valid assignments to the corresponding field type, even if the `STC`
was used between reloads.
This CL removes the unnecessary field and changes `InvalidateFields`
so that the `STC`s used during reload are instead lazily allocated once
per `FieldInvalidator` object and now both the `STC` and its backing
array are now garbage post-field invalidation.
TEST=reload ci
Change-Id: I3ffa199551ba69f0afd4eba6303c6ff73d7473a3
Cq-Include-Trybots: luci.dart.try:vm-reload-linux-debug-x64-try,vm-reload-linux-release-x64-try,vm-reload-rollback-linux-debug-x64-try,vm-reload-rollback-linux-release-x64-try,vm-aot-linux-debug-x64-try,vm-aot-linux-product-x64-try,vm-aot-linux-release-x64-try,vm-linux-release-x64-try,vm-linux-debug-x64-try,vm-aot-dwarf-linux-product-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311220
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
The round default value used in f7e26c50 means that we allocate
a new backing array shortly before hitting the cap, and once we
hit the cap the max load factor for the new backing array is only ~49%.
Instead, given a desired maximum, calculate an appropriate default
value that ensures the cap is only hit shortly before the array reaches
the max load factor that causes reallocation.
TEST=ci
Change-Id: I4c4fc3f5726aea3b260cf288cd00f886c2a1bd96
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311280
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Alexander Markov