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>
Right now `utf8.encode()` has a static return type of `List<int>`
due to extending `Encoding` (which extends `Codec<String, List<int>>`).
We cannot easily change `Encoding` to extend `Codec<String, Uint8List>`
because that would also change `utf8.decode()` to require `Uint8List`
which would be a breaking change.
So instead we override `utf8.encode()` to have more precise return type.
Some parts of our SDK are run using the checked-in SDK, so it cannot
rely on the changed return type yet (until checked-in SDK is rolled).
So we use `const Utf8Encoder().convert()` as a temporary change, as
that already has `Uint8List` return type.
Issue https://github.com/dart-lang/sdk/issues/52801
TEST=ci
CoreLibraryReviewExempt: More precise return type for existing API
Change-Id: I2861d1f0eb3d292d8e3ec8437c0d441a2d2bd193
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/254903
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Lasse Nielsen <lrn@google.com>
So, that we don't have to copy all class members for 99% cases when
there are no augmentaitons.
Change-Id: I5c225905a0460ce5162543553e207a2d4412ecb9
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/313082
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Samuel Rawlins <srawlins@google.com>
Print type in `dart_object_printer.dart`.
Use `assertDartObjectText` instead of whatever was being used before.
Change-Id: Ifdfb6343013bb5fe654bcff870a9ff83b155402f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312885
Commit-Queue: Kallen Tu <kallentu@google.com>
Reviewed-by: Bob Nystrom <rnystrom@google.com>
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
This is for an internal debug adapter to extend the test debug adapter and provide a custom URI converter to provide internal local paths during breakpoints.
Change-Id: I3b61b70fe07e14f08ff37443ef87facc523bf7fb
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/313000
Reviewed-by: Ben Konyi <bkonyi@google.com>
Commit-Queue: Helin Shiah <helinx@google.com>
TEST=Loaded the timeline in DevTools on a debug build with the recorder
set to ring, endless, and startup. Recorded traces on a debug build with
the recorder set to systrace, file, and perfettofile. Ran the
vm/cc/TimelineRingRecorderRace test on ASAN and TSAN.
Issue: https://github.com/flutter/flutter/issues/129258
Change-Id: I39920ac3e4a70e5714915fcd554ae171ed82504a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312721
Reviewed-by: Ben Konyi <bkonyi@google.com>
Commit-Queue: Derek Xu <derekx@google.com>
TEST=Recorded traces using the file recorder on ASAN and TSAN builds.
Change-Id: I2251a8373d952d6e3ef1803e9a79de85522f518a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312720
Reviewed-by: Ben Konyi <bkonyi@google.com>
We used to generate our own thread ID starting at 1 and counting up. There was always a 1:1 mapping from a DAP thread to a VM Isolate. With this change, we always use the Isolate number as the thread ID which makes it easier for tools using both VM Service and DAP at the same time.
Change-Id: Id8d82f6fd5134987f2ecfeaa761765d55999405d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312906
Reviewed-by: Ben Konyi <bkonyi@google.com>
Commit-Queue: Ben Konyi <bkonyi@google.com>
Currently this throws a Wasm RuntimeError, which is difficult to
recover from.
Change-Id: I747682d6959b19746d6c98f2ea1b3fefd1ed2d03
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312891
Reviewed-by: Joshua Litt <joshualitt@google.com>
Commit-Queue: Srujan Gaddam <srujzs@google.com>
Makes the TSAN skips apply to
`test.py -mrelease --sanitizer=tsan`
and not just
`test.py -n vm-tsan-linux-release-x64`.
Also makes the timeouts agree.
Change-Id: I10315e754a4ebb3020f3c2f6cecfac6b77e77a9b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311828
Reviewed-by: Alexander Thomas <athom@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
The timeout multipiler for linux-arm64 was originally 4, then lowered to
1 (see [0]), then increased again to 2 (see [1]).
Though it seems that service tests are still flakily timing out, so
let's try restoring the original multilier.
[0] https://dart-review.googlesource.com/c/sdk/+/306662
[1] https://dart-review.googlesource.com/c/sdk/+/307972
Also special case `ia32` in timeout calculations due to not using
an AppJIT trained `kernel-isolate` snapshot and therefore being
very slow, especially in ia32-debug mode.
Issue https://github.com/dart-lang/sdk/issues/52589
TEST=ci
Change-Id: Iab8c768866aec9e77bb83c7a3242cc5de8fb4e2f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312905
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
This is a revert of b6bbfff8c0c6c44cdab0d7d78 which temporarily disabled
this.
Turns out that a non-TSAN gen_snapshot would - when emitting code via
`Assembler::TsanLoadAcquire` - use an incorrect `Thread`-offset:
It calculated the offset via `Thread::OffsetFromThread`. That function
took a `dart::RuntimeEntry*` and tried to find its offset.
We happen to have the following leaf runtime entries:
#define LEAF_RUNTIME_ENTRY_LIST(V)
...
V(void, MsanUnpoison, void*, size_t)
V(void, MsanUnpoisonParam, size_t)
V(void, TsanLoadAcquire, void*)
V(void, TsanStoreRelease, void*)
...
It loops over all runtime entries and finds the first one that has the
identical `dart::RuntimeEntry::function_` pointer.
Though all 4 of them are `nullptr` at `gen_snapshot` time, so when
searching for offset for
`Thread::OffsetFromThread(kTsanLoadAcquireRuntimeEntry)`
it looked for the first runtime entry with `nullptr` function pointer -
which turned out to be `MsanUnpoison` (instead of `TsanLoadAcquire`).
=> The obvious fix is to use the `dart::RuntimeEntry*` pointer for
comparison instead of it's `function_` member.
TEST=ci
Issue b/287638965
Change-Id: I85c06674927978ef8561e9e7bdfab4823c0a8e1c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312902
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
In computation of GUB both the left-hand and the right-hand sides of
the intersection should be updated as nullable when the other operator
to GUB is Null.
Closes https://github.com/dart-lang/sdk/issues/50431
Change-Id: I4b616a94a3e7bf149205ba1b90732453c19ace47
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/311845
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Commit-Queue: Chloe Stefantsova <cstefantsova@google.com>
This assertion hits flakily due to races between compilers
(e.g. mutator vs background compiler or multiple mutators racing
with itself).
The discovered races are benign, so we just going to remove
the assertion.
Closes https://github.com/dart-lang/sdk/issues/52806
TEST=ci
R=kustermann@google.com
Change-Id: I8c4202bd76acea0fbf3724f19dbbc69958937afc
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312903
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
This reverts commit 83616c9b56 as it was
not needed for flutter roll, which landed anyway.
TEST=ci
Change-Id: I38a85ab117cabd98420bc88e698a481b094d7345
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312886
Reviewed-by: Siva Annamalai <asiva@google.com>
Commit-Queue: Alexander Aprelev <aam@google.com>
Auto-Submit: Alexander Aprelev <aam@google.com>
The adjustments to g3 build rules that allow running non-TSAN
gen_snapshot targeting TSAN AOT runtime aren't quite fleshed out yet.
So we temporarily let `gen_snapshot` detect whether to target TSAN by
whether it is itself compiled in TSAN (instead of via
`-DTARGET_USES_THREAD_SANITIZER`)
TEST=ci
Issue b/287638965
Change-Id: I2cd744f959d8302ba81cb1feceedee2756e02656
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312901
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
We walk augmentations in the right order, so can update update
necessary data structures as we go.
Change-Id: Ibf3409658009b82a960caaacd538fcd6c779b007
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312883
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
So, we can read it more naturally, first the declaration, then its
augmentations.
Change-Id: I21805969bff6874660500d4477b39ce5aba7484a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312880
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
This is a continuation of a previous change that stops tests from verifying only individual files modified via a `WorkspaceEdit`. All verifications of these edits are done via a single string that includes all changes, with annotations for creates/deletes/renames.
Additionally, it migrates some additional tests to use TestCode instead of the old markers, and removes some extra indenting these tests had.
It also starts changing how capabilities are set for tests from having to nest function calls in `initialize()` calls to instead calling simple helpers prior to initialization.
Change-Id: I35d16a296c2125ab830685437e14eb3b29ea4704
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312841
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Commit-Queue: Brian Wilkerson <brianwilkerson@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
While in more recent SDK versions the 'patched_class_' field has
been replaced with 'wrapped_class_', this package can still be
used with the earlier SDK versions if we fall back to checking it.
Issue: https://github.com/flutter/flutter/issues/130009
Change-Id: Ifb1250393b72e58bbdc0764c7e2ce96269e659ea
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312802
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
This may simplify some g3 build rules as it allows passing this define
whenever we compile under TSAN - instead currently it can only be passed
if DART_PRECOPMILER is also passed.
TEST=ci
Change-Id: I73e51243716373cdeca7559b344408bb206e22ba
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/312801
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
