Commit Graph

21390 Commits

Author SHA1 Message Date
Alexander Aprelev 2bc1ace7fc [vm/compiler] Let SIMD values become constants.
Reland of https://github.com/dart-lang/sdk/commit/bdacbb8decbab527a43ff1fbe289c8e55b0f6ba6

Patchset 2 has fixes for tests failing in the reland.

TEST=ci

Change-Id: I1053ef653d4112d3998070cc4fc098d19d8ff264
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357440
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Alexander Aprelev <aam@google.com>
2024-03-25 17:33:01 +00:00
Jonas Termansen 54cd684549 Remove additional --no-goma options since it's the new default.
Bug: b/296994239
Tested: Commit queue
Change-Id: I04f61149c1466db965866e95ffc6dccc457460d9
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/359065
Reviewed-by: Alexander Thomas <athom@google.com>
Commit-Queue: Jonas Termansen <sortie@google.com>
Auto-Submit: Jonas Termansen <sortie@google.com>
2024-03-25 10:35:49 +00:00
Alexander Markov 35d8630176 [vm] Make String.codeUnitAt monomorphic
The new implementation of String.codeUnitAt is always inlined and
doesn't depend on the polymorphic inlining of recognized methods.

String.codeUnitAt now has a custom body in the flow graph builder
which  performs non-speculative bounds check and then branches between
OneByteString and TwoByteString. Corresponding graph intrinsic and
native method are removed.

This change also fixes passing of unboxed arguments to runtime
in the slow path of GenericCheckBound instruction in JIT mode
(when shared stubs are not used).

TEST=ci

Change-Id: Iab2805fc752df84c37089165f828e31aca5f043f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/359000
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
2024-03-24 14:50:17 +00:00
Tess Strickland 291051e02d [vm/compiler] Add an instruction to adjust unsafe untagged addresses.
ComputeElementAddress takes a base untagged address, an index, an index
scale, and an offset and returns base + (index * scale) + offset as an
untagged address.

This removes the last conversions between untagged and unboxed integers
for untagged GC-movable pointers. The only remaining conversions are
the following cases, which all involve non-GC-movable pointers:

1. Calls to the FFI resolver (on IA32 only).
2. Returning nullptr when a pointer to memory is expected in
   exceptional returns from FFI callbacks.
3. Converting user-provided integers to FFI pointer objects and back.

In addition, we no longer add the data field to materializations of
typed data views, but instead recompute the data field in
DeferredObject::Fill(). This removes the last source of unsafe untagged
pointers that may have arbitrary lifetimes in the flow graph.

Thus, we can now verify in the FlowGraphChecker that there are no
GC-triggering instructions between the creation of an untagged
GC-movable pointer and its use (including the use itself). To do this,
this CL adds a predicate MayCreateUnsafeUntaggedPointer to definitions,
which by default returns true for kUntagged results, false otherwise,
and should only be overwritten in cases where the result can be proven
to not be a untagged GC-movable pointer.

TEST=vm/dart/regress_54710_il_test
     vm/cc/AllocationSinking_NoViewDataMaterialization

Fixes: https://github.com/dart-lang/sdk/issues/54710
Cq-Include-Trybots: luci.dart.try:vm-aot-android-release-arm64c-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-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-linux-debug-ia32-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-release-ia32-try
Change-Id: Ie172a8bd0330a728a4f151478664a530f8d9b38a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/354862
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
2024-03-22 19:03:31 +00:00
Tess Strickland 1cebdcb2da [vm/ffi] Use untagged pointer representations for FFI pointers.
Previously, the FFI used unboxed integers as a native representation
for pointers in FFI code, as the compiler only handled very specific
uses of untagged pointers flowing between instructions. Since then,
this restriction has been removed for untagged pointers that do not
point to memory managed by the GC, like FFI pointers, so now they can
have a more precise representation.

By being precise about when untagged (untagged pointers to freshly
allocated Handles and the contents of Pointer data fields) and tagged
(TypedData objects constructed to hold the byte representation of
compound data) values are expected, we can remove the need to have
untagged pointers to GC-movable objects and/or having untagged
pointers escape as unboxed integers in the generated IL.

This CL also renames kUnboxedFfiIntPtr -> kUnboxedAddress and limits
its uses specifically to where the unboxed integer represents the
numeric representation of an untagged pointer.

This CL changes CCall to take Representations for the arguments and
return value instead of what looks like an arbitrary
NativeCallingConvention. However, the serializer and deserializers for
CCall, used in IL tests, originally assumed that the argument and return
representations were kUnboxedFfiIntPtr, so providing an arbitrary
NativeCallingConvention which didn't match that assumption would cause
failures during IL tests. That assumption came from the fact that
the only creator of CCall instructions was in kernel_to_il.cc, and there
that was the case.

Now CCall builds the native calling convention during construction
and deserialization from the argument and return representations and
stores both the representations and built native calling convention
internally. In the future, if we want to create CCall instructions with
more arbitrary native calling conventions, then we'll need to handle
serialization/deserialization of arbitrary native calling conventions,
and also add consistency checks that the provided representations
appropriately match the native calling convention.

TEST=ffi vm/dart/regress_306327173_il vm/dart/address_local_pointer_il

Issue: https://github.com/dart-lang/sdk/issues/54710
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-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-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-release-ia32-try
Change-Id: I34effe8fbdc80288b703e0152d5ba67ce2343400
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/353101
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
2024-03-22 10:20:48 +00:00
Tess Strickland a6a3434404 [vm/compiler] Use temps instead of writable registers in MemoryCopy.
This keeps the array registers unchanged across the instruction,
which can save instructions if there are unused registers to allocate
for the temporaries and the source or destination registers are live
after the copy (e.g., if the destination array is then used in a
LoadIndexed instruction post-copy).

On IA32, we still use writable registers due to register pressure.

TEST=ci

Change-Id: Ia955f8db39e4aba0f01b9d06ea50992a7375a492
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-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-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-release-ia32-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357901
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
2024-03-22 10:17:11 +00:00
Tess Strickland 0d3ade255b [vm/compiler] Limit exposure of untagged pointers to managed memory.
After https://dart-review.googlesource.com/c/sdk/+/330600, there were
more chances for the optimizing compiler to introduce or move
GC-triggering instructions like allocations or boxings between the
retrieval of an untagged pointer to GC-moveable memory and its use.

To limit the chance of this happening, this CL removes the explicit
loading of the untagged payload address when building the initial
flow graph in most cases when the array is not known to be an external
array (an external string, an external typed data object, or an FFI
Pointer).

The remaining case is during view allocation, which extracts the
payload address of the base typed data object underlying the view
(which may be GC-movable) to calculate the payload address that should
be stored in the data field of the view object. See
https://github.com/dart-lang/sdk/issues/54884.

During canonicalization of LoadIndexed, StoreIndexed, and MemoryCopy
instructions, if the cid of an array input is an external array
(external string, external typed data object, or Pointer), then a
LoadField instruction that extracts the untagged payload address
is inserted before the instruction and the corresponding input is
rebound to that LoadField instruction.

Once all compiler passes that involve code motion have been performed,
a new pass looks for LoadIndexed, StoreIndexed, or MemoryCopy where
the cid stored in the instruction for the array is a typed data cid.
In these cases, if the array is not an internal typed data object,
then the payload address is extracted. Waiting until this point ensures
that no GC-triggering instructions are inserted between the extraction
of the payload address and the use. (Internal typed data objects are
left as-is because the payload address is inside the object itself
and doesn't require indirection through the data field of the object).

This CL also replaces code conditional on the array cid with code
that is instead conditional on the array element representation in
cases where it makes sense to do so, since this is a less brittle
check than checking the array cid (e.g., checking for kUnboxedInt8
to load, store, or copy an signed byte from an array instead of
listing all possible array cids that store signed bytes).

This CL also fixes an issue with the ARM64 assembler where calling
LoadFromOffset with an Address that has a non-Offset type would
silently generate bad code instead of triggering the ASSERT in
PrepareLargeOffset.

TEST=vm/dart/typed_list_index_checkbound_il_test

Issue: https://github.com/dart-lang/sdk/issues/54710
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-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-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-release-ia32-try
Change-Id: I25b5f314943e9254d3d28986d720a5d47f12feeb
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/352363
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
2024-03-22 10:12:39 +00:00
Ryan Macnak 4b5d3e05e9 [vm, gc] Use temporal coarsening when accounting for old-space bump pointer allocations.
To avoid accounting overhead during each bump pointer allocation, we add the size of the whole bump area when starting to use that area and
subtract the remaining size when switching to a new area.

TEST=ci
Change-Id: If0c429db47c27ca51aac5c055268e85f70bad9ac
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/359160
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
2024-03-21 22:22:42 +00:00
Ben Konyi 6eb85949fd [ VM / DDS ] Add --print-dtd-uri flag and launch DTD from the correct snapshot for AOT
This adds support for printing the DTD connection information to stdout
when --print-dtd-uri is passed.

This change also fixes an issue where DDS would fail to spawn an isolate
with the DTD snapshot when DDS was running in AOT mode. This means the
SDK must be shipped with both AppJIT and AOT DTD snapshots, at least
until dartdev is moved to run from AOT.

Fixes https://github.com/dart-lang/sdk/issues/55034

TEST=run_test.dart

Change-Id: I788ef9bfe76297a8d594992a2aac440ed9e2ecac
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/358541
Reviewed-by: Siva Annamalai <asiva@google.com>
Commit-Queue: Ben Konyi <bkonyi@google.com>
Reviewed-by: Kenzie Davisson <kenzieschmoll@google.com>
2024-03-21 19:50:46 +00:00
Vyacheslav Egorov 395b024685 [vm] Use codegen block order in regalloc in AOT.
Codegen block order keeps loop blocks together and reflecting that in
live ranges allows to produce better register allocation decisions.

Consider for example the loop:

    v = def();
    while (cond) {
      if (smth) { // (*)
        use(v);
        return true;
      }
    }

If block (*) is interspersed with loop blocks the register allocator
might decide to allocate it to the register in some of the loop blocks.
If the same block is "sunk" away from loop blocks - the register
allocator can clearly see that `v` does not necessarily have to live on
the register in the whole loop.

In JIT codegen block order is not topologically sorted and as such
is unsuitable for our linear scan.

TEST=ci

Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-aot-mac-release-arm64-try,vm-aot-linux-release-x64-try
Change-Id: I0726815db998b559267949e157cd2158f5dd55f7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/358448
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
2024-03-21 10:35:11 +00:00
Vyacheslav Egorov 3bc14606f9 [vm/compiler] Improve AOT block scheduler
Try to keep loop blocks together, previously AOT compiler would produce
bad block order for loops like these:

    while (true) {
      if (smth) {
        // (*)
        return;
      }
    }

And put `(*)` blocks in between loop header and other loop blocks.

TEST=ci

Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-aot-mac-release-arm64-try,vm-aot-linux-release-x64-try
Change-Id: I52304810b61ff9288fe9df648b21c0258299d61e
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/358447
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
2024-03-21 10:00:36 +00:00
Vyacheslav Egorov bebee99b23 [vm/compiler] Move reorder_blocks onto the graph.
Make FlowGraph constructor compute whether we are planning to reorder
blocks before code generation or not. This makes the state of the
graph at the end of the compilation more clear.

This is just a refactoring without any functional changes.

TEST=ci

Change-Id: Ieefb02237cc1ebd69d5d2b217bdc8ebbfdbf15c7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/358446
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
2024-03-20 21:18:10 +00:00
Alexander Markov 17d6ba15b6 [vm] Remove external strings
This change removes support for external strings from the VM along with
Dart_NewExternalLatin1String, Dart_NewExternalUTF16String and
Dart_IsExternalString Dart C API functions.

External strings are not used by the VM nor any known embedder, but
Dart VM was paying the maintenance and performance price for
the external string implementation classes.

TEST=ci

Change-Id: I094cd2d2b7ec0840e9f09e1ca9e5a7acd4e78c28
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/358760
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Siva Annamalai <asiva@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
2024-03-20 20:08:13 +00:00
Daco Harkes 922065b627 [gardening] Skip slow test on reload bots
Closes: https://github.com/dart-lang/sdk/issues/55226
Change-Id: I17e60ea9725cc018c26963ba44c450cf95cba3be
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/358240
Reviewed-by: Liam Appelbe <liama@google.com>
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Commit-Queue: Liam Appelbe <liama@google.com>
Auto-Submit: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
2024-03-18 21:02:53 +00:00
Vyacheslav Egorov 2dae577c67 Reapply "[vm/compiler] Pass arguments in registers"
This reverts commit 63179aa49d.

See commit 1ab083ae48 for the original
description of the change.

The original change was reverted because it did not take
into account that in Dart it is legal to make some
parameters optional when you override a method. Such parameters
should not be passed in registers - which means for each
possible interface target we need to compute the minimum
number of direct parameters accepted by all overrides.

I have also added two comprehensive tests for different
combinations of direct, optional and named parameters.

TEST=vm/dart/calling_conventions_test.dart,vm/dart/calling_conventions2_test.dart

Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv64-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,vm-aot-optimization-level-linux-release-x64-try
Change-Id: Ib76e2512d3e4077a40265d315483ebe3b0f5551c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357862
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
2024-03-18 20:56:24 +00:00
Ryan Macnak f792cc1d61 [vm] Also set immutability bit when deserializing.
Currently the immutable bit is a of function of the class. If we start having some instances of a class be immutable and others not, we will need separate clusters like we do for the canonical bit.

TEST=vm/dart/snapshot_immutable_bit_test
Bug: https://github.com/dart-lang/sdk/issues/55136
Cq-Include-Trybots: luci.dart.try:vm-appjit-linux-debug-x64-try
Change-Id: I073f28cabca8293b766aa7c0b224934a62bf9cb4
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356280
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
2024-03-15 22:24:24 +00:00
Alexander Aprelev c5946a1195 [vm] Ensure CpuInfo Windows flag detection works.
Fixes https://github.com/flutter/flutter/issues/140138
TEST=manually on win x64 with `__ int3();` added in front of `roundsd` and forced `CpuId::sse41=false;`
Change-Id: I9e2dbf2615549dcc7c979036b5442276c2e0fb29
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357218
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Alexander Aprelev <aam@google.com>
2024-03-15 19:26:13 +00:00
Daco Harkes 51ed0e5b07 [vm/test] Fix imports
Reduces the number of warnings originating from runtime/tests/vm/dart.

Change-Id: I0f6f527145c9b389952b5fa09a9d4e03824b9bf3
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357900
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Tess Strickland <sstrickl@google.com>
Auto-Submit: Daco Harkes <dacoharkes@google.com>
2024-03-15 17:21:30 +00:00
Vyacheslav Egorov 63179aa49d Revert "[vm/compiler] Pass arguments in registers"
This reverts commit 1ab083ae48.


Revert "[vm] Unbreak build with -Wctad-maybe-unsupported"

This reverts commit 734e0735bd.


Revert "[vm, compiler] Account for pair locations in MoveArgumentInstr on riscv32."

This reverts commit 52f7908b27.

Reverting due to https://github.com/dart-lang/sdk/issues/55180: there is mismatch in the calling convention when invoking _IntListMixin.getRange.

R=kustermann@google.com
TEST=this is a revert

Change-Id: Ic3a3013706280ebd80b04d5e6a830e5f56987384
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357881
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
2024-03-15 14:18:06 +00:00
Alexander Aprelev 2347820b22 [vm/cpuid] Fix number of info bytes printed.
Bug: https://github.com/flutter/flutter/issues/140138
Change-Id: I2624451c8ef5e6f7af7a1855fc16a976cfe6dad3
TEST=manually
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357743
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Alexander Aprelev <aam@google.com>
2024-03-14 21:48:59 +00:00
Ryan Macnak c2a6645c01 [vm] Fix ShouldPauseOnException with an empty Dart stack.
TEST=vm/dart/gc
Change-Id: I435d7a33745c30738899eb1f1c08bd94185302c7
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357740
Reviewed-by: Derek Xu <derekx@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
2024-03-14 21:12:01 +00:00
Zijie He 476ed37a25 Reland "[Fuchsia] Remove the legacy gn build rules for fuchsia"
This change may require a codereview from flutter side before moving
forward. See https://github.com/flutter/engine/pull/51072.

This reverts commit 283051acba.

Reason for revert: Redo the change with the similar one in flutter.

After this change, the next dart -> flutter roll needs extra cares
in https://github.com/flutter/engine/pull/51072. It also contains
more details about the reason why it could not be two-way
compatible.

Original change's description:
> Revert "[Fuchsia] Remove the legacy gn build rules for fuchsia"
>
> This reverts commit 094202bb91.
>
> Reason for revert: Break flutter, this change should happen after
> flutter being migrated to the same gn-sdk.
>
> Original change's description:
> > [Fuchsia] Remove the legacy gn build rules for fuchsia
> >
> > The updated test-scripts contains https://crrev.com/c/5341620 which
> > allows to generate fidl apis with an env var controlled location
> > rather than a hard-coded one. So the existing fuchsia gn build rules
> > in dart/sdk become obsolete and can be fully removed in favor of the
> > one in //third_party/fuchsia/gn-sdk/.
> >
> > Meanwhile the gn-sdk has been updated with
> > https://crrev.com/c/5325282 to use api-level version'ed idk but not
> > the ones in arch/.
> >
> > Cq-Include-Trybots: luci.dart.try:vm-fuchsia-release-arm64-try,vm-fuchsia-release-x64-try
> > Tested: ^^^^^
> > Bug: 40935282
> > Change-Id: I2ce958e6db1ff8221beef7b7ff953c32bb4525ba
> > Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/355283
> > Reviewed-by: Ryan Macnak <rmacnak@google.com>
> > Reviewed-by: Alexander Thomas <athom@google.com>
> > Commit-Queue: Zijie He <zijiehe@google.com>
>
> Bug: 40935282
> Change-Id: I7c455d1d362210523671c97d99ef018ede1743f4
> Cq-Include-Trybots: luci.dart.try:vm-fuchsia-release-arm64-try,vm-fuchsia-release-x64-try
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356307
> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
> Auto-Submit: Zijie He <zijiehe@google.com>
> Reviewed-by: Derek Xu <derekx@google.com>
> Commit-Queue: Zijie He <zijiehe@google.com>

Bug: 40935282
Change-Id: Id5a9a98013350359037b19f8506f24158c9a3120
Cq-Include-Trybots: luci.dart.try:vm-fuchsia-release-arm64-try,vm-fuchsia-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356924
Commit-Queue: Zijie He <zijiehe@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
2024-03-14 20:22:12 +00:00
Alexander Aprelev 70acceb4ca [vm/cpuid] Add trace_cpuid flag to troubleshoot cpu feature detection.
BUG=https://github.com/flutter/flutter/issues/140138
TEST=manually

Change-Id: I7beaee3901d9e530aa3edf350f37b919eea01d3e
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357662
Commit-Queue: Alexander Aprelev <aam@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
2024-03-14 19:44:07 +00:00
Ryan Macnak 52f7908b27 [vm, compiler] Account for pair locations in MoveArgumentInstr on riscv32.
TEST=dart-fuzz
Bug: https://github.com/dart-lang/sdk/issues/55181
Change-Id: I075f0b341858b8fcf408378dea1473bab07a87ec
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357214
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
2024-03-13 22:32:11 +00:00
Paul Berry fb210554eb Enable feature inference-update-3.
In discussion with the language team, we've decided to ship this
feature in Dart 3.4. This CL contains a minimal CHANGELOG entry for
the feature; I plan to expand on it once I'm finished with the feature
specification document.

Change-Id: Iac86ebed322feb8e478bf5a6780d50adb433ed7f
Tested: trybots
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/355900
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Nate Bosch <nbosch@google.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
Commit-Queue: Paul Berry <paulberry@google.com>
2024-03-13 22:04:23 +00:00
asiva 9f951f7fb5 [VM / vmservice] Run service isolate in sound null safe mode (JIT)
TEST=ci
CoreLibraryReviewExempt: No files in other platforms were changed.
Change-Id: I567dae64226867c2036809739a14c14fdf52077b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357211
Reviewed-by: Ben Konyi <bkonyi@google.com>
Commit-Queue: Siva Annamalai <asiva@google.com>
2024-03-13 17:56:49 +00:00
Ben Konyi 9c78c8f14b [ CLI ] Strip --observe and --enable-vm-service from dartdev argument list
The dart run argument parser was incorrectly parsing any argument following an
option that the VM treats as having a default (e.g., --observe and
--enable-vm-service), resulting in a "File not found ''" error and the
VM failing to start the user program. The following invocations would
fail due to 'foo.dart' being treated as the value for the option to the
left:

- dart run --observe foo.dart
- dart run --enable-vm-service foo.dart

This change strips these two options from the arguments sent to dartdev
as both are handled directly in the VM and are only included in the dart
run argument parser for the help message.

Fixes https://github.com/dart-lang/sdk/issues/55185

TEST=regression test added to run_test.dart.

Fixed: 55185
Change-Id: I1c89fec3ee46296ee4ebb847eb47a7845dfdb399
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357360
Reviewed-by: Derek Xu <derekx@google.com>
Commit-Queue: Ben Konyi <bkonyi@google.com>
2024-03-13 17:08:39 +00:00
Alexander Markov 8c1f86e5dd [vm/aot] Devirtualization of function calls
TFA now produces direct call metadata for function calls based on the
results of the analysis.

VM reads direct call metadata for function calls, but only uses it
if target is a tear-off, as it cannot handle inlining of an arbitrary
closure function yet (due to scope building dependencies between
compilation of a closure and its parent function).

TEST=pkg/vm/testcases/transformations/type_flow/transformer/closures.dart

Issue: https://github.com/dart-lang/sdk/issues/39692
Change-Id: Idf0b6988534bfca1a64c6661a82df8d453272abf
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357180
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
2024-03-13 16:48:34 +00:00
Vyacheslav Egorov 0fea05cf4c [vm] Unpoison spill slot area in FfiCall
With register calling conventions some parameters might arrive in
registers and later be spilled on the stack and addresses of those
stack slots passed as handles into native code.

This means we need to unpoison spill slot area otherwise
MSAN will complain.

Fixes https://github.com/dart-lang/sdk/issues/55179

TEST=existing msan configuration

R=kustermann@google.com

Fixed: 55179
Cq-Include-Trybots: luci.dart.try:vm-aot-msan-linux-release-x64-try,vm-msan-linux-release-x64-try
Change-Id: I592cde9cb623c2eabc848c7a39d0453e757e97f6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357340
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Auto-Submit: Slava Egorov <vegorov@google.com>
Commit-Queue: Daco Harkes <dacoharkes@google.com>
2024-03-13 16:32:54 +00:00
Alexander Markov 8e7137ea81 [vm/compiler] Enable CSE and LICM for tear-off closure allocations
Benchmarks:
JIT:
TearOff.NotInlined*
+32-33% (x64) +60-63% (arm32) +70-76% (arm64) +46-49% (arm64c)
TearOff.Inlined.InTry
+591% (x64) +431% (arm32) +810% (arm64) +509% (arm64c)
AOT:
TearOff.NotInlined
+49-65% (x64) +64-99% (arm32) +121-131% (arm64) +94-109% (arm64c)

TEST=runtime/tests/vm/dart/tearoff_optimizations_il_test.dart
Issue: https://github.com/dart-lang/sdk/issues/54808
Change-Id: I885e68c1ba0ce935c64882b7b91bcbdfe1615df9
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356900
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
2024-03-13 16:11:17 +00:00
Alexander Markov 1695da67c7 [vm] Move creation of closure functions to kernel loader
This change disentangles creation of closure functions from
flow graph building.

TEST=ci

Change-Id: Ia008c21202453062dd722f64be615e5e6af3bdc2
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357222
Reviewed-by: Slava Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
2024-03-13 14:33:48 +00:00
Tess Strickland ecf5abcbd3 [vm/compiler] Better handle small array element representations.
Moves RepresentationOfArrayElement out of LoadIndexed and StoreIndexed
into RepresentationUtils. Adjusts RepresentationOfArrayElement
to work for all concrete typed data classes. Changes
RepresentationOfArrayElement to return the precise representation of
the array element, which requires adding two new representations,
kUnboxedInt8 and kUnboxedInt16.

Remove Address::OperandSizeFor and its only (ignored) use in the
RISCV assembler and change Address::OperandSizeFor in the ARM and
ARM64 assemblers to use RepresentationUtils::OperandSize and RepresentationUtils::RepresentationOfArrayElement appropriately.

Add Boxing::NativeRepresentation, which converts representations for
unboxed integers smaller than 4 bytes to kUnboxedIntPtr and otherwise
returns the original representation. This mimics the current semantics
for LoadIndexed and StoreIndexed where returned values represented by
unboxed integer values smaller than 4 bytes are zero or sign extended
to word size appropriately, and received values represented by unboxed
integer values smaller than 4 bytes are appropriately truncated from a
word sized signed integer.

Create LoadIndexedInstr::ReturnRepresentation and
StoreIndexedInstr::ValueRepresentation, which use
Boxing::NativeRepresentation to convert the precise representation
returned by RepresentationUtils::RepresentationOfArrayElement to
the actual returned/expected representation for those instructions.
Also use Boxing::NativeRepresentation to appropriately adjust
the representations used for SIMD operations returning int8_t.

Add checks that unboxing and integer conversion never get
representations which result in a different representation when
passed to Boxing::NativeRepresentation, as those instructions do
not currently work with unboxed integer representations smaller than
4 bytes. (Adding such support can be done in a future CL.)

Inside the optimizing compiler, use the precise representation of the
array element where appropriate. That is:

* Range analysis
* Type propagation
* Redundancy elimination

This CL also cleans up how Representations are printed to C, making
that part of RepresentationUtils. It also removes the representation
printing and parsing utils from Location since they were only used by
the original IL serializer and deserializer, which has been removed.
It also renames RepresentationUtils::IsUnsigned to IsUnsignedInteger,
to make it clear that it only returns true for unboxed integer
representations.

TEST=standalone/bytedata_test

Fixes: https://github.com/dart-lang/sdk/issues/54622
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-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-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-release-ia32-try
Change-Id: I8c0dab911463ee4d0c55cd59ffea9d01e0f0bb7c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/354885
Reviewed-by: Slava Egorov <vegorov@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
2024-03-13 14:14:55 +00:00
Jens Johansen e8a1fe81be [service tests] Set batch mode expectations to ExtraSlow trying to avoid flakes
See also https://github.com/dart-lang/sdk/issues/53672

TEST=Existing tests.
Change-Id: I91ce7af3ae1766ff56e086c2574177c244651979
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357281
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: William Hesse <whesse@google.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
2024-03-13 12:42:12 +00:00
Vyacheslav Egorov 734e0735bd [vm] Unbreak build with -Wctad-maybe-unsupported
Without an explicit deduction guide compilation fails with

```
error: 'SimpleAllocator' may not intend to support class template argument deduction [-Werror,-Wctad-maybe-unsupported]
   53 |   SimpleAllocator cpu_allocator(DartCallingConvention::kCpuRegistersForArgs);
```

TEST=verified by manual build with a flag
R=aam@google.com

Change-Id: I858863f7febb44ab81635e2c8466f84055c6e8b2
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/357144
Auto-Submit: Slava Egorov <vegorov@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
2024-03-13 07:42:13 +00:00
Vyacheslav Egorov 1ab083ae48 [vm/compiler] Pass arguments in registers
Following the lead of unboxed calling conventions augment
out backend to support passing arguments in registers on
ARM, ARM64 and X64.

Similar to unboxing the calling convention change only applies
to functions which are invoked statically or through
global dispatch table. Functions which are invoked from native
code or directly invoked by dynamic call sites are not allowed
to use register calling convention.

This CL avoid messing too much with call instructions and
instead leans on adjusting existing machinery: MoveArgument
instructions corresponding to register arguments are simply
detached from the graph and are processed as input constraints
by the register allocator.

The computation of the calling convention is centralized in
`ComputeCallingConvention` in `flow_graph.cc` and is used
by callees and callers.

While working on this CL I have also discovered an existing bug
in catch move handling (see added test runtime/tests/vm/dart/try_catch_parameter_sync_test.dart).

On a large Flutter app this decreases the size of generated code
by 1.8% (380Kb) and there a mixed bag of improvements in runing time, e.g.

```
build_bench (Pixel 6a)
-3.536% (0.2 noise)423.16 438.68
animation_bench_close_first_frame (Pixel 6a)
-3.717% (0.0 noise)253.64 263.43
rrect_contains_bench (Pixel 6a)
-3.796% (0.0 noise)0.088700 0.092200
sync_star_sync_star (Pixel 6a)
-5.039% (0.0 noise)2.4310 2.5600
sync_star_list (Pixel 6a)
-5.329% (0.0 noise)1.0660 1.1260
velocity_tracker_bench (Pixel 6a)
-5.731% (0.1 noise)62.039 65.811
sync_star_generated (Pixel 6a)
-6.438% (0.0 noise)1.090 01.1650
matrix_utils_transform_point_perspective (Pixel 6a)
-6.844% (0.2 noise)0.012520 0.013439
matrix_utils_transform_rect_perspective (Pixel 6a)
-7.818% (0.2 noise)0.028117 0.030502
matrix_utils_transform_rect_affine
-27.05% (0.7 noise) 0.017494 0.023981
```

TEST=ci

Issue https://github.com/dart-lang/sdk/issues/54955

Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-simarm_x64-try,vm-aot-linux-debug-simriscv64-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,vm-aot-optimization-level-linux-release-x64-try
Change-Id: I9a134a764666a541d4693e6ca9f6813025ecd63e
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/353404
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
2024-03-12 20:44:49 +00:00
Ryan Macnak af700baa49 [test] Stop passing -nostdlib when building dynamic libraries on Mac.
Bug: https://github.com/dart-lang/sdk/issues/55105
Change-Id: I3f81875ad10f05deccf8440a8b163423e5446333
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356923
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Derek Xu <derekx@google.com>
2024-03-12 16:39:54 +00:00
Slava Egorov c2e9ceeb6b [vm/io] Cleanup long path handling on Windows
* Switch to `std::unique_ptr<wchar_t[]>` to represent a
dynamically strings instead of wrappers like
`StringRAII` and `Utf8ToWideScope`;
* Avoid back and forth conversion between UTF8 and UTF16:
convert to UTF16 first then work on that. This also simplifies
code - previously it tried to work with strings allocated in
different ways uniformly, which is actually unnecessary if
resulting string needs to be converted to UTF16 (and allocated
with `malloc`) anyway;
* Fix a bug in `File::CreateLink`: it was handling relative
links with long paths incorrectly. Change file_long_path_test
to cover this case and make it run on non-Windows systems as
well to ensure that all behavior that can match actually
matches.

TEST=ci

Change-Id: I1279aff1d2cdace5e2ce8633c2f7ea69a34fe41a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356680
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
2024-03-12 08:29:36 +00:00
Daco Harkes e7daade291 [gardening] Fix test
TEST=tests/ffi/deeply_immutable_c_api_finalizer_test.dart

Closes: https://github.com/dart-lang/sdk/issues/55165
Change-Id: I8e5925b978db4591442ca36d57a6a98f85501bbd
Cq-Include-Trybots: luci.dart.try:vm-win-debug-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356801
Reviewed-by: Ben Konyi <bkonyi@google.com>
Commit-Queue: Daco Harkes <dacoharkes@google.com>
2024-03-12 08:03:30 +00:00
Chris Bracken a735858e61 Update perfetto include directories
In https://dart-review.googlesource.com/c/sdk/+/356622 we updated the
perfetto include directory in libprotozero_config to use
`perfetto_root_path` rather than hardcoding `//third_party/perfetto`.

Some Perfetto headers, however, #include a generated header, which is
located at `$root_gen_dir/$perfetto_root_path/build_config`. See
declaration of perfetto_build_flags.h and buildflag_gen_dir_ at:
https://flutter.googlesource.com/third_party/perfetto/+/refs/heads/main/gn/BUILD.gn#26

This was missed in the first patch since while the default local engine
build of Flutter DOES do all gn processing, it DOES NOT do a full Dart
build, our presubmits however, revealed the gap.

This updates the include directories for the config to use the correct
value as declared in perfetto.

TEST=None. Refactor only; no semantic changes.

Issue: https://github.com/flutter/flutter/issues/144204
Change-Id: I44c1269af85e40eab8e397c3d099f5dc1bca1229
Part of: https://github.com/flutter/flutter/issues/67373
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356960
Commit-Queue: Chris Bracken <cbracken@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
2024-03-11 20:22:53 +00:00
Slava Egorov b0236fdde5 [vm] Fix DateTime.timeZoneName on Windows
It needs to look at the given moment to decide whether to use
summer time zone name or standard time zone name.

Previously it was looking at the current time to make this decision
which produced incorrect result: e.g. given
`DateTime.parse(2012-01-02T13:45:23)` its `timeZoneName` should be
returning standard name corresponding to the current time zone even
if we are currently running in summer time (e.g. it should
return PST if machine it is running on is in PDT).

This is revealed by a test which started to fail on Windows
because our Windows bots entered PDT.

Fixes https://github.com/dart-lang/sdk/issues/55159

TEST=corelib/date_time7_test.dart

CoreLibraryReviewExempt: No core library changes.
Change-Id: Ic938baf837c2f6130ec3f9604631701267369f87
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356681
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
2024-03-11 20:14:49 +00:00
Alexander Markov b9b341f4a7 [vm] Simplify and optimize method extractors
Previously, method extractors (getter functions which return
tear-offs) where implemented in the following way:

Method extractor body is an asm intrinsic which loads a few registers
and jumps to BuildMethodExtractor stubs.
BuildMethodExtractor stub loads more registers, calls AllocateClosure
stub and initializes a few fields of a closure (this stub also used
to call AllocateContext but it is no longer needed for tear-offs).
AllocateClosure allocates a closure object and initializes closure
fields.

Instead of doing 2 hops to the object allocation, method extractor
body now calls AllocateClosure stubs directly which initializes
all closure fields. This saves a jump and a handful of instructions
to load certain things on registers and set closure fields.

Assembler intrinsic for method extractor body and
architecture-specific BuildMethodExtractor stubs are now removed.

Also, as method extractors are really tiny, they are now always
inlined.

TEST=ci

Change-Id: Icfeed18414659a76c7d24d3f6f53a76c91e5bbd3
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356302
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
2024-03-11 15:56:33 +00:00
Daco Harkes 8682ca72ce [vm/ffi] Example with deeply-immutable Finalizable and dart_api_dl
A test exercising isolate group finalizers with `Finalizable` and
`pragma('vm:deeply-immutable')`.

Note that we cannot fully do without C code due to the signature of
`Dart_HandleFinalizer` which also passes a
`void* isolate_callback_data` in addition to the peer.

TEST=tests/ffi/deeply_immutable_c_api_finalizer_test.dart

Bug: https://github.com/dart-lang/sdk/issues/55062
Bug: https://github.com/dart-lang/sdk/issues/55120
Change-Id: Ibaf4899ca678ffb0c5d227ac4f10deb38d49fe6f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356720
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Hossein Yousefi <yousefi@google.com>
Commit-Queue: Daco Harkes <dacoharkes@google.com>
2024-03-11 15:51:46 +00:00
Natsu f4c8c9b1ba Fix incorrect definition and usage of REUSABLE_TYPE_PARAMETERS_HANDLESCOPE
Closes https://github.com/dart-lang/sdk/pull/55122

GitOrigin-RevId: c0358e16e5252ed658923345c461b5b6537f0df7
Change-Id: I5b01deb6d69bef94a437a4b3fdcd8de1fd8fe061
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356100
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
2024-03-11 14:05:12 +00:00
Zach Anderson 03101837d9 [gn] Deprecate copy_trees()
Part of https://github.com/flutter/flutter/issues/144430

This is a reland of https://dart-review.googlesource.com/c/sdk/+/356481 with the addition of dependencies to prevent races on directory creation.

Previously, this code did backflips to accomplish two goals
1. Collect the paths of all files that would be copied from one place to another after applying regex filters so that they could be listed as the "inputs" to a GN action.
2. Arrange so that the python script doing the above would only be invoked once during GN to reduce the cost of calling out to python.

However, this is exactly the use-case for the "depfile" parameter to a GN action. Instead of running the script during GN to populate the "inputs" list, when we run the copy_tree.py script, we can instead ask it to generate a depfile to collect all the input files that were actually copied after applying the regex filter. Using that, we don't have to run the python script at all during GN.

TEST=it builds

Change-Id: I593e2500544a9fff5dd9852d0d3370f97aafc464
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356620
Reviewed-by: Alexander Aprelev <aam@google.com>
Commit-Queue: Zach Anderson <zra@google.com>
2024-03-09 04:14:39 +00:00
Chris Bracken c1058b6203 Support overriding perfetto_root_path
As part of merging Flutter's flutter/buildroot and flutter/engine repos,
Flutter is migrating all third-party dependencies from //third_party to
//flutter/third_party.

This change allows consumers of the Dart SDK (notably Flutter), to
override the path of their third_party/perfetto directory by setting
perfetto_root_path in //build_overrides/build.gn. The Dart SDK does
not override perfetto's path, but does override other variables
(perfetto_build_with_embedder) in the equivalent overrides file secondary/build_overrides/build.gn.

TEST=None. Refactor only; no semantic changes.

Issue: https://github.com/flutter/flutter/issues/144204
Change-Id: Ie3835a5507a39bf2d68b1c86d42c626093358b17
Part of: https://github.com/flutter/flutter/issues/67373
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356622
Reviewed-by: Zach Anderson <zra@google.com>
Commit-Queue: Chris Bracken <cbracken@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
2024-03-09 01:01:29 +00:00
Zach Anderson 939ae55c00 Revert "[gn] Deprecate copy_trees()"
This reverts commit f60d32167f.

Reason for revert: build failures in post submit

Original change's description:
> [gn] Deprecate copy_trees()
>
> Part of https://github.com/flutter/flutter/issues/144430
>
> Previously, this code did backflips to accomplish two goals
> 1. Collect the paths of all files that would be copied from one place to another after applying regex filters so that they could be listed as the "inputs" to a GN action.
> 2. Arrange so that the python script doing the above would only be invoked once during GN to reduce the cost of calling out to python.
>
> However, this is exactly the use-case for the "depfile" parameter to a GN action. Instead of running the script during GN to populate the "inputs" list, when we run the copy_tree.py script, we can instead ask it to generate a depfile to collect all the input files that were actually copied after applying the regex filter. Using that, we don't have to run the python script at all during GN.
>
> TEST=it builds
>
> Change-Id: I41a251ce4659119cdc3997bb2d6fc7ee0831bb6d
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356481
> Reviewed-by: Siva Annamalai <asiva@google.com>
> Commit-Queue: Zach Anderson <zra@google.com>
> Reviewed-by: Alexander Aprelev <aam@google.com>

Change-Id: Ie4409ed93a3c990fbce375dba1f2d9a3c735ea3f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356308
Auto-Submit: Zach Anderson <zra@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Zach Anderson <zra@google.com>
2024-03-08 20:59:53 +00:00
Zijie He 283051acba Revert "[Fuchsia] Remove the legacy gn build rules for fuchsia"
This reverts commit 094202bb91.

Reason for revert: Break flutter, this change should happen after
flutter being migrated to the same gn-sdk.

Original change's description:
> [Fuchsia] Remove the legacy gn build rules for fuchsia
>
> The updated test-scripts contains https://crrev.com/c/5341620 which
> allows to generate fidl apis with an env var controlled location
> rather than a hard-coded one. So the existing fuchsia gn build rules
> in dart/sdk become obsolete and can be fully removed in favor of the
> one in //third_party/fuchsia/gn-sdk/.
>
> Meanwhile the gn-sdk has been updated with
> https://crrev.com/c/5325282 to use api-level version'ed idk but not
> the ones in arch/.
>
> Cq-Include-Trybots: luci.dart.try:vm-fuchsia-release-arm64-try,vm-fuchsia-release-x64-try
> Tested: ^^^^^
> Bug: 40935282
> Change-Id: I2ce958e6db1ff8221beef7b7ff953c32bb4525ba
> Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/355283
> Reviewed-by: Ryan Macnak <rmacnak@google.com>
> Reviewed-by: Alexander Thomas <athom@google.com>
> Commit-Queue: Zijie He <zijiehe@google.com>

Bug: 40935282
Change-Id: I7c455d1d362210523671c97d99ef018ede1743f4
Cq-Include-Trybots: luci.dart.try:vm-fuchsia-release-arm64-try,vm-fuchsia-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356307
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Auto-Submit: Zijie He <zijiehe@google.com>
Reviewed-by: Derek Xu <derekx@google.com>
Commit-Queue: Zijie He <zijiehe@google.com>
2024-03-08 20:50:31 +00:00
Zach Anderson f60d32167f [gn] Deprecate copy_trees()
Part of https://github.com/flutter/flutter/issues/144430

Previously, this code did backflips to accomplish two goals
1. Collect the paths of all files that would be copied from one place to another after applying regex filters so that they could be listed as the "inputs" to a GN action.
2. Arrange so that the python script doing the above would only be invoked once during GN to reduce the cost of calling out to python.

However, this is exactly the use-case for the "depfile" parameter to a GN action. Instead of running the script during GN to populate the "inputs" list, when we run the copy_tree.py script, we can instead ask it to generate a depfile to collect all the input files that were actually copied after applying the regex filter. Using that, we don't have to run the python script at all during GN.

TEST=it builds

Change-Id: I41a251ce4659119cdc3997bb2d6fc7ee0831bb6d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/356481
Reviewed-by: Siva Annamalai <asiva@google.com>
Commit-Queue: Zach Anderson <zra@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
2024-03-08 20:47:47 +00:00
Alexander Markov 97369d0903 [vm/compiler] Correct representation of shift rhs and speculative mode of BinaryUint32OpInstr
Speculative and non-speculative shift instructions have different
representations of RHS input, so canonicalization of multiplication
into a shift should account for that.

Also, mark BinaryUint32Op instruction as always non-speculative -
it never needs to speculate  as it is inserted by range analysis
after it can prove that it is valid to perform Uint32 op.
This prevents insertion of a speculative shift instruction in AOT
when BinaryUint32Op multiplication is canonicalized.

Fixes https://github.com/dart-lang/sdk/issues/55003
TEST=vm/dart/regress_55003_test

Change-Id: Ib1805ce2dc0e6f514e6f3b4a09be08808e75048b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/355820
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
2024-03-08 15:35:20 +00:00
Daco Harkes 8de00e2137 [vm] Introduce pragma vm:deeply-immutable
This CL introduces a way to mark all instances of a class as deeply
immutable.

In order to statically verify that all instances of a deeply immutable
class are immutable, a deeply immutable classes must have the following
properties:

1. All instance fields must
   1. have a deeply immutable type,
   2. be final, and
   3. be non-late.
2. The class must be `final` or `sealed`. This ensures no
   non-deeply-immutable subtypes are added by external code.
3. All subtypes must be deeply immutable. This ensures 1.1 can be
   trusted.
4. The super type must be deeply immutable (except for Object).

Note that instances of some classes in the VM are deeply immutable
while their class cannot be marked immutable.

* SendPort, Capability, RegExp, and StackTrace are not `final` and
  can be implemented by external code.
* UnmodifiableTypedDataViews do not have a public type. (It was
  recently deprecated.)

See runtime/docs/deeply_immutable.md for more details.

Use case:

This enables attaching a `Dart_FinalizableHandle` to a deeply immutable
object and the deeply immutable object with other isolates in the same
isolate group.

(Note that `NativeFinalizer`s live in an isolate, and not an isolate
group. So this should currently _not_ be used with `NativeFinalizer`s.
See https://github.com/dart-lang/sdk/issues/55062 for making a
`NativeFinalizer.shared(` that would live in an isolate group instead
of in an isolate.)

Implementation details:

Before this CL, the `ImmutableBit` in the object header was only ever
set to true for predefined class ids (and for const objects). After
this CL, the bit can also be set to true for non const instances of
user-defined classes. The object allocation and initialization code has
been changed to deal with this new case. The immutability of a class is
saved in the class state bits. On object allocation and initialization
the immutability bit is read from the class for non-predefined class
ids.

TEST=runtime/tests/vm/dart/isolates/fast_object_copy2_test.dart
TEST=runtime/vm/isolate_reload_test.cc
TEST=tests/lib/isolate/deeply_immutable_*

Bug: https://github.com/dart-lang/sdk/issues/55120
Bug: https://github.com/dart-lang/sdk/issues/54885
Change-Id: Ib97fe589cb4f81673cb928c93e3093838d82132d
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-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
Cq-Include-Trybots: dart-internal/g3.dart-internal.try:g3-cbuild-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/354902
Commit-Queue: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
2024-03-07 18:33:58 +00:00