When doing fine grained invalidation and reload we might end up losing
`is_implemented` bit and some other class-hierarchy related bits
on a class if we only reload the library where class is declared
but not transitive closure which depends on that library.
Consider:
library a;
class A {}
library b;
class B implements A {}
If we do a change to library `a` which does not impact the outline
then we do not need to reload `b`. This will cause class A to
be replaced with a new `Class` object in the class table - and lead
to us losing various CHA related bits.
This CL update ProgramReloadContext::RebuildDirectSubclasses to
handle this correctly, we also rename it to
RestoreClassHierarchyInvariants and make it share implementation
with ClassFinalizer (which contained almost identical function).
Fixes https://github.com/flutter/flutter/issues/151032
TEST=vm/cc/IsolateReload_IsImplementedBit
Change-Id: Ie620592befecb89897d6e8d46175ef07348bf11d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/374040
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
The [ReloadOperationScope] has [StackResource]s as fields and is itself
a [StackResource] which is problemantic if unwinding happens manually.
So we'll make it a macro that expands to the 3 fields instead.
TEST=ci
Change-Id: I3fb7bec7ca87193c83ec34908f9a43c5db005900
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/302201
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
The current hot-reload implementation [0] will perform a reload by
first sending OOB messages to all isolates and waiting until those OOB
messages are being handled. The handler of the OOB message will block
the thread (and unschedule isolate) and notify the thread performing
reload it's ready.
This requires that all isolates within a group can actually run & block.
This is the case for the VM implementation of isolates (as they are
run an unlimited size thread pool).
Though flutter seems to multiplex several engine isolates on the same OS
thread. Reloading can then result in one engine isolate performing
reload waiting for another to act on the OOB message (which it will not
do as it's multiplexed on the same thread as the former).
Now that we have a more flexible safepointing mechanism (introduced in
[1]) we can utilize for hot reloading by introducing a new "reloading"
safepoint level.
Reload safepoints
-----------------------
We introduce a new safepoint level (SafepointLevel::kGCAndDeoptAndReload).
Being at a "reload safepoint" implies being at a "deopt safepoint"
which implies being at a "gc safepoint".
Code has to explicitly opt-into making safepoint checks participate /
check into "reload safepoints" using [ReloadParticipationScope]. We do
that at certain well-defined places where reload is possible (e.g. event
loop boundaries, descheduling of isolates, OOM message processing, ...).
While running under [NoReloadScope] we disable checking into "reload
safepoints".
Initiator of hot-reload
-----------------------
When a mutator initiates a reload operation (e.g. as part of a
`ReloadSources` `vm-service` API call) it will use a
[ReloadSafepointOperationScope] to get all other mutators to a
safepoint.
For mutators that aren't already at a "reload safepoint", we'll
notify them via an OOB message (instead of scheduling kVMInterrupt).
While waiting for all mutators to check into a "reload safepoint", the
thread is itself at a safepoint (as other mutators may perform lower
level safepoint operations - e.g. GC, Deopt, ...)
Once all mutators are at a "reload safepoint" the thread will take
ownership of all safepoint levels.
Other mutators
-----------------------
Mutators can be at a "reload safepoint" already (e.g. isolate is not
scheduled). If they try to exit safepoint they will block until the
reload operation is finished.
Mutators that are not at a "reload safepoint" (e.g. executing Dart or VM
code) will be sent an OOB message indicating it should check into a
"reload safepoint". We assume mutators make progress until they can
process OOB message.
Mutators may run under a [NoReloadScope] when handling the OOM message.
In that case they will not check into the "reload safepoint" and simply
ignore the message. To ensure the thread will eventually check-in,
we'll make the destructor of [~NoReloadScope] check & send itself a new OOB
message indicating reload should happen. Eventually getting the mutator
to process the OOM message (which is a well-defined place where we can
check into the reload safepoint).
Non-isolate mutators such as the background compiler do not react to OOB
messages. This means that either those mutators have to be stopped (e.g.
bg compiler) before initiating a reload safepoint operation, the
threads have to explicitly opt-into participating in reload safepoints
or the threads have to deschedule themselves eventually.
Misc
----
Owning a reload safepoint operation implies also owning the deopt &
gc safepoint operation. Yet some code would like to ensure it actually
runs under a [DeoptSafepointOperatoinScope]/[GCSafepointOperationScope].
=> The `Thread::OwnsGCSafepoint()` handles that.
While performing hot-reload we may exercise common code (e.g. kernel
loader, ...) that acquires safepoint locks. Normally it's disallows to
acquire safepoint locks while holding a safepoint operation (since
mutators may be stopped at places where they hold locks, creating
deadlock scenarios).
=> We explicitly opt code into participating in reload safepointing
requests. Those well-defined places aren't holding safepoint locks.
=> The `Thread::CanAcquireSafepointLocks()` will return `true` despite
owning a reload operation. (But if one also holds deopt/gc safepoint
operation it will return false)
Example where this matters: As part of hot-reload, we load kernel which
may create new symbols. The symbol creation code may acquire the symbol
lock and `InsertNewOrGet()` a symbol. This is safe as other mutators
don't hold the symbol lock at reload safepoints. The same cannot be said
for Deopt/GC safepoint operations - as they can interrupt code at many
more places where there's no guarantee that no locks are held.
[0] https://dart-review.googlesource.com/c/sdk/+/187461
[1] https://dart-review.googlesource.com/c/sdk/+/196927
Issue https://github.com/flutter/flutter/issues/124546
TEST=Newly added Reload_* tests.
Change-Id: I6842d7d2b284d043cc047fd702b7c5c7dd1fa3c5
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/296183
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Instead apply the same approach as we do in AOT: unbox based on the
static type information. There are no TFA results available in JIT,
but we could still unbox fields when running in sound null-safety.
TEST=ci
Cq-Include-Trybots: luci.dart.try:vm-kernel-reload-linux-release-x64-try,vm-kernel-reload-linux-debug-x64-try,vm-kernel-reload-rollback-linux-debug-x64-try,vm-kernel-reload-rollback-linux-release-x64-try,vm-kernel-precomp-linux-debug-x64-try,vm-kernel-precomp-linux-product-x64-try,vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-nnbd-linux-release-simarm64-try,vm-kernel-linux-debug-simriscv64-try,vm-kernel-precomp-linux-debug-simriscv64-try,vm-kernel-nnbd-linux-release-ia32-try,vm-kernel-nnbd-linux-debug-x64-try,vm-kernel-nnbd-linux-debug-ia32-try,vm-kernel-nnbd-linux-release-simarm-try,vm-kernel-nnbd-linux-release-simarm64-try
Change-Id: Ide2e78c6659261ef8d245a4586cf699ea0fbb459
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/256211
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
* Merge ClassTable and SharedClassTable back together;
* Simplify handling of multiple arrays growing in sync;
* Refactor how reload deals with ClassTable.
The last change is the most important because it makes it
much easier to reason about the code. We move away from
copying bits and pieces of the class table and shared
class table into reload contexts.
Having two class table fields in the isolate group makes
it easier to reason about. One field contains program
class table (one modified by kernel loader and accessed
by various program structure cid lookups) and heap
walk class table (used by GC visitors). Normally these
two fields point to the same class table, but during
hot reload we temporary split them apart: original
class table is kept as a heap walk class table, while
program class table is replaced by a clone and updated
by reload.
If reload succeeds we drop original class table and
set program class table as heap walk one.
If reload fails we drop the program class table and
restore original one from heap walk table.
TEST=ci
Cq-Include-Trybots: luci.dart.try:vm-kernel-reload-linux-release-x64-try,vm-kernel-reload-linux-debug-x64-try,vm-kernel-reload-rollback-linux-debug-x64-try,vm-kernel-reload-rollback-linux-release-x64-try,vm-kernel-linux-debug-x64-try,vm-kernel-precomp-tsan-linux-release-x64-try,vm-kernel-tsan-linux-release-x64-try,vm-kernel-precomp-asan-linux-release-x64-try,vm-kernel-asan-linux-release-x64-try
Change-Id: I8b66259fcc474dea7dd2af063e4772df99be06c4
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/258361
Commit-Queue: Slava Egorov <vegorov@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
The new implementation is based on suspend/resume stubs and doesn't
use desugaring of async functions on kernel AST.
Previously, new implementation of async/async* was only supported in
AOT mode. This change adds all necessary bits for the JIT mode:
* Suspending variable-length frames (for unoptimized code).
* Handling of Code and pool pointers in Dart stack frames.
* OSR.
* Deoptimization.
* Hot reload.
* Debugger.
The new implementation is not enabled in JIT mode yet.
Design doc: go/compact-async-await.
TEST=ci
Issue: https://github.com/dart-lang/sdk/issues/48378
Change-Id: I477d6684bdce7cbc1edb179ae2271ff598b7dcc5
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/246081
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
In the codebase, we have several different interfaces for instance
methods that return a hash:
- uword Hash() const;
- intptr_t Hash() const;
- uint32_t Hash() const;
- intptr_t Hashcode() const;
This CL standardizes on `uword Hash() const` and adjusts any related
functions to match.
TEST=Existing test suites, as this is an internal refactoring.
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-debug-x64-try
Change-Id: If2cbce57f3fae0f0d24031b6e324f0323c965f41
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/195067
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
This is the initial implementation of hot reload with multi-isolate
groups.
Implementation:
As before, when a service API call triggers a reload it will be routed
as an OOB message to a specific isolate (**). As opposed to before, that
isolate has now to coordinate with all other isolates, ensuring that it
"owns" the reload and all other isolates are waiting in a state that
allows reload.
This is implemented as a [ReloadOperationScope] which first participates
in other reloads (if there are any) and then owns the reload. It will
send a new kind of service message to all other registered isolates. All
of them have to check in before reload can proceed. If a new isolate
is about to join the group, it will participate when registering the
isolate. If an old isolate wants to die, it will participate when
unregistering the isolate.
This means that in addition to the existing StackOverFlow checks that
can process OOB messages and therefore reload, we'll have isolate
registration and unregistration as well as a new
Isolate::kCheckForReload OOB message handler where an isolate can
participate in a reload.
We consider the isolate group to be reloadable if the main isolate has
loaded the program and set the root library. Helper isolates don't need
to load any more kernel code and only initialize core libraries, so it's
fine to reload them during this time.
(**) The reason we continue to send reload service API calls to any
isolate in an isolate group is that re-loading might involve calling out
to the embedder's tag handler. Doing so currently requires an active
isolate.
If we allowed a subset of dart_api.h (the subset needed by the tag
handler) to be used only with an active IsolateGroup instead of an
active Isolate we could remove this requirement.
Edge cases:
There's various edge cases to consider: The main edge case is, we currently
maintain an upper limit to the number of isolates executing in parallel
(to ensure each can have big enough chunk of new space, i.e. TLAB).
If there are more isolates with active work they are waiting until one
of the exiting ones "yields". To ensure progress, if any such actively
running isolate gets a request to participate in a reload, it will mark
its own thread as "blocked" and therefore "yields", so another isolate
can make progress until all isolates are participating and the reload
can start.
Marking an isolate as "blocked" happens by exiting that isolate. It will
free up it's TLAB, decrease active mutator count and (if running on VM's
thread pool) also temporarily increase the thread pool size.
The side-effect of this is that it will use one pthread per isolate
during reload. In the future we can extend this first implementation, by
specially handling isolates that don't have a message handler running.
Doing so would require careful consideration to avoid races.
Testing:
In order to test this we use a small helper framework for reload tests.
The helper framework will, similar to real world reload e.g. in flutter,
will spawn a subprocess. It will use the service API to trigger reloads
in this subproces.
To synchronize between the reload driver and the application being
reloaded it allows watching for events to be printed to stdout/stderr.
The reload test itself can be written - similar to multitests - with
annotations such as `// @include-in-relload-0` in them. The testing
framework will then generate multiple application versions that all get
compiled to kernel.
For simplicity we generate the kernel using the standalone VM with
`--snapshot-kind=kernel` and avoid using the incremental compiler.
There are 4 different tests exercising different aspects of
multi-isolate reload:
vm/dart_2/isolates/reload_active_stack_test:
Performs a reload while a fixed number of isolates have an active
stack, thereby ensuring e.g. that all frames of all isolate mutator
stacks get deoptimized, ...
vm/dart_2/isolates/reload_no_active_stack_test:
Similar to the test above, but instead of having an active stack the
isolates can yield to the event loop, possibly be even descheduled
vm/dart_2/isolates/reload_many_isolates_test:
Similar to the test above, but this test uses many more isolates.
vm/dart_2/isolates/reload_many_isolates_live_and_die_test:
Performs a reload where isolates get spawned and die all the time.
There are always P isolates alive at any given point in time, each
of them spawns children when their parent has died.
Performing a reload catches isolates as various stages of their
lifecycle and can therefore cover a lot of corner cases.
TEST=vm/dart_2/isolates/reload_*_test.dart
Issue https://github.com/dart-lang/sdk/issues/36097
Change-Id: I97039b4084de040b7f2e22f5832a40d57ba398d5
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/187461
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
This CL moves the [ProgramReloadContext] from [Isolate] to
[IsolateGroup] - since the program is shared across isolates. Amongst
other things, it also deoptimizes stacks of all isolate mutators.
There is still future work to do for hot-reloading multiple isolates
within a group, this is only the first change.
Issue https://github.com/dart-lang/sdk/issues/36097
TEST=Refactoring of existing implementation.
Change-Id: Ic394de265e3922f84657de2da25a46179b624647
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/183693
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
As part of making the compiler and other subsystems independent
of `Isolate` we have to move various state from `Isolate` to
`IsolateGroup` (or to another place).
Part of that is `Isolate::reload_context()`. In order to avoid name
conflict of what a `reload_context()` is, we separate it into
`reload_context()` and `program_reload_context()` where the former
contains GC information and the latter contains information about
changes in program structure.
Issue https://github.com/dart-lang/sdk/issues/36097
TEST=Pure refactoring - relying on existing test coverage.
Change-Id: I7e1b732ce8b34c8842718699d618be542bd59ff6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/177863
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Currently we have things called XPtr which are not what you get from ptr().
Old world:
handle->raw() returns RawObject* (tagged)
raw_obj->ptr() returns RawObject* (untagged)
After 6fe15f6df9:
handle->raw() returns ObjectPtr
obj_ptr->ptr() returns ObjectLayout*
New world:
handle->ptr() returns ObjectPtr
obj_ptr->untag() returns UntaggedObject*
TEST=ci
Change-Id: I6c7f34014cf20737607caaf84979838300d12df2
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/149367
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Siva Annamalai <asiva@google.com>
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
As part of making the compiler and other subsystems independent
of `Isolate` we have to move various state from `Isolate` to
`IsolateGroup`.
The class_table and object_store were already moved to `IsolateGroup`.
This CL only replaces usages of `Isolate::{object_store,class_table}`
with the equivalent in `IsolateGroup`.
Issue https://github.com/dart-lang/sdk/issues/36097
TEST=Pure refactoring - relying on existing test coverage.
Change-Id: I34a0682d715b054d6c5faff077a513980f59a348
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/177126
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Right now we assign class ids to top-level classes, abstract classes as
well as concrete classes. All of them have allocated from a 16-bit pool
of ids. The VM FATAL()s once it hits that limit.
Customers who run very large programs (significant amount of generated
code) on the Dart VM have started to hit this 16-bit class limit.
Concrete classes can have instances in the heap. Our current heap layout
only allows 16-bit class ids to be encoded in the header word. To avoid
increasing the size of heap objects or shrinking the size of the identity
hash code to 16-bit we keep class ids in object headers to be 16-bit.
Abstract classes cannot have instances in the heap. Though their class
ids are encoded in type objects. Furthermore we sort classes in
AOT/AppJIT mode to perform fast class-id range checks. To avoid impacting
this optimization we treat abstract classes the same way as concrete
classes.
Top-level classes cannot have instances in the heap. Their class ids are
only used in the runtime code, for example for hot-reload as well as
part of the service protocol.
=> We can allocate class ids outside the 16-bit range for top-level
classes, thereby freeing a significant amount of space in the 16-bit
range.
This CL does exactly that: We change classid_t to be int32_t. The
ClassLayout::id_ can now be assigned ids outside 16-bit range for
top-level classes. To do this we keep dart classes and top level classes
as separate arrays in the ClassTable.
Issue https://github.com/dart-lang/sdk/issues/42533
See also b/160229360
Change-Id: I6710a644e7b0ab2d4f4c792bef8e1f91cb117421
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/153607
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
This CL:
* Moves [Heap]/[SharedClassTable] from [Isolate] to [IsolateGroup], which
will make all isolates in the group use the same heap. The GC will use
the shared class table for object size information.
* Adds support for entering/leaving an isolate group as a helper thread
(e.g. via [Thread::EnterIsolateGroupAsHelper]). The current active
isolate group can be accessed via TLS `IsolateGroup::Current()` or
`Thread::isolate_group_`. When entering as a helper thread there will be
no current isolate.
* Changes the GC to use the above mechanism and ensures GC works without
a currently active isolate. The GC will use information purely available via
[IsolateGroup]. The GC will iterate all isolates within an isolate
group e.g. for scanning roots.
* Makes spawning of new isolates start in their own isolate group.
Once the isolate is fully functional it's heap will be merged into
the original isolate group
* Moves ApiState, containing persistent and weak persistent handles,
from [Isolate] to [IsolateGroup], plus adds appropriate locking.
Issue https://github.com/dart-lang/sdk/issues/36097
Change-Id: Ia8e1d8aa78750e8400864200f4825395a182c004
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/126646
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
When creating a KernelProgramInfo, we create several logical views into the kernel buffer. These are fresh ExternalTypedDatas, rather than proper TypedDataViews, so they do not automically keep the original ExternalTypedData alive. Create an explicit reference to the orginal ExternalTypedData in the KernelProgramInfo. When creating snapshots, this reference is ignored/null'd and the views are turned into copies, effectively dropping the parts of the original buffer that do not have views.
Fixes a leak with reload and a use-after-free with eval.
Bug: https://github.com/dart-lang/sdk/issues/33973
Bug: https://github.com/dart-lang/sdk/issues/39610
Change-Id: I09d3830133314ccbaa0341d904127c2b6925c4ec
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/126825
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Similar to the split of ClassTable into ClassTable/SharedClassTable,
this CL splits up the IsolateReloadContext into:
* IsolateGroupReloadContext: Consists of reload-related information across all
isolates. The [Reload()] method is split up in phases that are
performed on all isolates before the next phase is started.
=> This allows each isolate to add reasons for rolling back, if no
reasons are found the reload will be accepted atomically.
* IsolateReloadContext: Constists of reload-related information for a
particular isolate (e.g. mappings of old to new classes)
The assumption is that all isolates have the same source (and therefore
the same libraries). For certain things, e.g. discovering which libraries
changed, it is necessary to examine the object store. We use the first
isolate in a group (but could use any of them) to do so, since the
isolate group does not have this information atm.
This is a preparation CL for supporting hot-reloading multiple isolates
within one isolate group. Though the support in this CL stays at having
only a single isolate in a group.
=> This CL turns off FLAG_enable_isolate_groups in JIT mode.
Issue https://github.com/dart-lang/sdk/issues/36097
Change-Id: I7f4d536d4f5ab4a2a73fb0c7618ba967c9b77234
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/123254
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Alexander Aprelev <aam@google.com>
This CL moves heap related information (namely instance sizes and
allocation stats) out of the [ClassTable] into a [SharedClassTable].
Both classes are always in sync (i.e. they have the same number of entries).
This CL also changes GC related code to start using the size information
from the new [SharedClassTable].
In a futher step we will move the heap as well as this shared class
table out of the [Isolate] and into [IsolateGroup].
Issue https://github.com/dart-lang/sdk/issues/36097
Change-Id: Id54a89c9251ad3bbc13e60d32dc4f7bcc7f1d805
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/116064
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Until now it was possible to register classes with a default size (16 bytes on 64-bit)
and later on change the size for the cid.
This CL changes this to ensure the size information in the class table
for a given cid is either 0 or the final instance size (and adds an
ASSERT for it)
Issue https://github.com/dart-lang/sdk/issues/36097
Change-Id: I94c61c6a1566c13dec7b9eb80c9ae0dadf0e6b6a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/115861
Commit-Queue: Martin Kustermann <kustermann@google.com>
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Ryan Macnak