aboutsummaryrefslogtreecommitdiffstats
path: root/ractor.c
Commit message (Collapse)AuthorAgeFilesLines
* expose some C-APIs for ractorKoichi Sasada2021-01-061-0/+11
| | | | | | | | | | | | | expose some C-APIs to try to make ractor utilities on external gems. * add * rb_ractor_local_storage_value_lookup() to check availability * expose * rb_ractor_make_shareable() * rb_ractor_make_shareable_copy() * rb_proc_isolate() (not public) * rb_proc_isolate_bang() (not public) * rb_proc_ractor_make_shareable() (not public)
* introduce rb_ractor_atfork()Koichi Sasada2020-12-241-0/+18
| | | | to reset main ractor at fork().
* fix ractor-locking around rb_ractor_thread_list()Koichi Sasada2020-12-241-9/+21
| | | | | With locking a ractor, rb_ary_push() can call RB_VM_LOCK_ENTER() and it violates an assertion: should not acquire ractor-lock.
* Update rb_ractor_ensure_shareableNobuyoshi Nakada2020-12-231-2/+2
| | | | | * Fixed use of rb_ractor_shareable_p * Raise Ractor::IsolationError
* Changed shareable literal semantics [Feature #17397]Nobuyoshi Nakada2020-12-231-0/+11
| | | | | | When `literal`, check if the literal about to be assigned to a constant is ractor-shareable, otherwise raise `Ractor::Error` at runtime instead of `SyntaxError`.
* ruby_single_main_ractor should be clear before warnKoichi Sasada2020-12-231-2/+2
| | | | | rb_warn can produce T_HASA object and it should not use transient heap.
* Remove debugging codeMarc-Andre Lafortune2020-12-211-1/+0
|
* fix ractor's doc. [ci skip]Koichi Sasada2020-12-221-6/+14
|
* add Ractor#[]/#[]= for ractor local storageKoichi Sasada2020-12-221-0/+46
| | | | | This API is similar to plain old Thread#[]/Fiber#[] interface with symbol key.
* separate rb_ractor_pub from rb_ractor_tKoichi Sasada2020-12-221-43/+31
| | | | | | | | | separate some fields from rb_ractor_t to rb_ractor_pub and put it at the beggining of rb_ractor_t and declare it in vm_core.h so vm_core.h can access rb_ractor_pub fields. Now rb_ec_ractor_hooks() is a complete inline function and no MJIT related issue.
* TracePoint.new(&block) should be ractor-localKoichi Sasada2020-12-221-0/+8
| | | | | TracePoint should be ractor-local because the Proc can violate the Ractor-safe.
* Introduce Ractor::IsolationErrorKoichi Sasada2020-12-211-12/+9
| | | | | | | | | | | Ractor has several restrictions to keep each ractor being isolated and some operation such as `CONST="foo"` in non-main ractor raises an exception. This kind of operation raises an error but there is confusion (some code raises RuntimeError and some code raises NameError). To make clear we introduce Ractor::IsolationError which is raised when the isolation between ractors is violated.
* [DOC] Remove about FrozenError from Ractor::MovedObject [ci skip]Kazuhiro NISHIYAMA2020-12-211-9/+0
| | | | | Because unfreeze Ractor::MovedObject at https://github.com/ruby/ruby/commit/76e88480371469400346fca609efe67096813915
* should use owned_p instead of locked_pKoichi Sasada2020-12-211-1/+2
| | | | | It should raise an error if the same thread tris to call receive/receive_if.
* rename to rb_ractor_make_shareable_copy()Koichi Sasada2020-12-211-1/+1
| | | | from rb_ractor_make_copy_shareable().
* fix Ractor.make_shareable() with Class/ModuleKoichi Sasada2020-12-211-1/+1
| | | | | To check shareable-ness, rb_ractor_shareable_p() is needed for Class/Module objects isntead of checking flags.
* Tweak Ractor doc [doc] [ci skip]Marc-Andre Lafortune2020-12-191-4/+4
|
* Add documentation for Ractor (#3895)Victor Shepelev2020-12-191-0/+94
|
* unfreeze Ractor::MovedObjectKoichi Sasada2020-12-191-2/+0
| | | | | Matz prefers to unfreeze the class. [Feature #17401]
* add "copy: true" option for Ractor.make_shareableKoichi Sasada2020-12-191-0/+10
| | | | | | | | | | Ractor.make_shareable(obj) tries to make obj a shareable object by changing the attribute of obj and traversable objects from obj (mainly freeze them). "copy: true" option is more conservative approach by make deep copied object and make it sharable. It doesn't affect any existing objects.
* Use category: :experimental in warnings that are related to experimental ↵Jeremy Evans2020-12-181-2/+3
| | | | | | | | | features This adds rb_category_compile_warn in order to emit compiler warnings with categories. Note that Ripper currently ignores the category for these warnings, but by default it ignores the warnings completely, so this shouldn't matter.
* fix timing bugKoichi Sasada2020-12-171-7/+21
| | | | | | | | | ractor_sleep() can remain wait.status by interrupt, so that this patch handles more correctly. This patch fixed this kind of assertion failures: Assertion Failed: ../src/ractor.c:1332:ractor_yield_atexit:cr->sync.wait.status == wait_none
* Ractor#receive_if to receive only matched messagesKoichi Sasada2020-12-161-32/+272
| | | | | | | Instead of Ractor.receive, Ractor.receive_if can provide a pattern by a block and you can choose the receiving message. [Feature #17378]
* Fixed a suspicious comparisonNobuyoshi Nakada2020-12-131-1/+1
|
* re-layout rb_ractor_tKoichi Sasada2020-12-091-112/+112
| | | | separate synchronization data and ractor local data.
* fix Thread's interrupt and Ractor#take issueKoichi Sasada2020-12-071-2/+4
| | | | | | | | | Thread's interrupt set Ractor's wakeup_status as interrupted, but the status remains next Ractor communication API. This patch makes to ignore the previous interrupt state. [Bug #17366] Also this patch solves the Thread#kill and Ractor#take issues.
* fix decl of ruby_single_main_ractorKoichi Sasada2020-12-071-7/+0
| | | | | | | On windows, MJIT doesn't work without this patch because of the declaration of ruby_single_main_ractor. This patch fix this issue and move the definition of it from ractor.c to vm.c to locate near place of ruby_current_vm_ptr.
* per-ractor object allocationKoichi Sasada2020-12-071-1/+1
| | | | | | | | Now object allocation requires VM global lock to synchronize objspace. However, of course, it introduces huge overhead. This patch caches some slots (in a page) by each ractor and use cached slots for object allocation. If there is no cached slots, acquire the global lock and get new cached slots, or start GC (marking or lazy sweeping).
* cancel theap on multi-ractorsKoichi Sasada2020-12-071-14/+25
| | | | | | | accessing theap needs complicating synchronization but it reduce performance on multi-ractor mode. So simply stop using theap on multi-ractor mode. In future, theap should be replaced with more cleaver memory strategy.
* ruby_single_main_ractor for single ractor modeKoichi Sasada2020-12-071-3/+6
| | | | | | | | ruby_multi_ractor was a flag that indicates the interpreter doesn't make any additional ractors (single ractor mode). Instead of boolean flag, ruby_single_main_ractor pointer is introduced which keeps main ractor's pointer if single ractor mode. If additional ractors are created, ruby_single_main_ractor becomes NULL.
* fix indexKoichi Sasada2020-12-041-2/+3
| | | | | `i` should not be used here because `i` will be `alen` when jumped here by "goto restart".
* Skip checking Ractor recursive lockingTakashi Kokubun2020-12-021-3/+7
| | | | | for an MJIT worker thread. We can't do it because its GET_EC() returns NULL.
* should not use rb_str_modify(), tooKoichi Sasada2020-12-011-1/+1
| | | | | | Same as 8247b8edde, should not use rb_str_modify() here. https://bugs.ruby-lang.org/issues/17343#change-88858
* rb_ext_ractor_safe() to declare ractor-safe extKoichi Sasada2020-12-011-0/+2
| | | | | | | | | | | C extensions can violate the ractor-safety, so only ractor-safe C extensions (C methods) can run on non-main ractors. rb_ext_ractor_safe(true) declares that the successive defined methods are ractor-safe. Otherwiwze, defined methods checked they are invoked in main ractor and raise an error if invoked at non-main ractors. [Feature #17307]
* should not use rb_ary_modify()Koichi Sasada2020-12-011-2/+3
| | | | | | | | | | | ractor_copy() used rb_ary_modify() to make sure this array is not sharing anything, but it also checks frozen flag. So frozen arrays raises an error. To solve this issue, this patch introduces new function rb_ary_cancel_sharing() which makes sure the array does not share another array and it doesn't check frozen flag. [Bug #17343] A test is quoted from https://github.com/ruby/ruby/pull/3817
* ractor local storage C-APIKoichi Sasada2020-12-011-23/+201
| | | | | | | | | | | | | | | | | | | | | | To manage ractor-local data for C extension, the following APIs are defined. * rb_ractor_local_storage_value_newkey * rb_ractor_local_storage_value * rb_ractor_local_storage_value_set * rb_ractor_local_storage_ptr_newkey * rb_ractor_local_storage_ptr * rb_ractor_local_storage_ptr_set At first, you need to create a key of storage by rb_ractor_local_(value|ptr)_newkey(). For ptr storage, it accepts the type of storage, how to mark and how to free with ractor's lifetime. rb_ractor_local_storage_value/set are used to access a VALUE and rb_ractor_local_storage_ptr/set are used to access a pointer. random.c uses this API.
* Fix `Ractor.make_shareable` for recursive structures with unfreezable componentsMarc-Andre Lafortune2020-11-301-1/+1
| | | | Followup to #3823
* Fixed Ractor.shareable? on cross-recursive objects [Bug #17344]Nobuyoshi Nakada2020-11-301-6/+32
|
* Use opaque struct pointer than voidNobuyoshi Nakada2020-11-271-3/+3
|
* mark default_randKoichi Sasada2020-11-271-1/+9
| | | | default_rand can points a Bignum seed, so it should be marked.
* per-ractor Random::DEFAULTKoichi Sasada2020-11-271-0/+16
| | | | | | | | | | Random generators are not Ractor-safe, so we need to prepare per-ractor default random genearators. This patch set `Random::DEFAULT = Randm` (not a Random instance, but the Random class) and singleton methods like `Random.rand()` use a per-ractor random generator. [Feature #17322]
* fix error messageKoichi Sasada2020-11-251-1/+1
|
* fix public interfaceKoichi Sasada2020-11-181-2/+3
| | | | | | | | | | | | | | | | | | To make some kind of Ractor related extensions, some functions should be exposed. * include/ruby/thread_native.h * rb_native_mutex_* * rb_native_cond_* * include/ruby/ractor.h * RB_OBJ_SHAREABLE_P(obj) * rb_ractor_shareable_p(obj) * rb_ractor_std*() * rb_cRactor and rm ractor_pub.h and rename srcdir/ractor.h to srcdir/ractor_core.h (to avoid conflict with include/ruby/ractor.h)
* ignore yield_atexit if outgoing port is closedKoichi Sasada2020-11-111-0/+4
| | | | | If outgoing_port is closed, Ractor.yield never successes. [Bug #17310]
* Threads in a ractor will be killed with the ractorKoichi Sasada2020-11-111-2/+2
| | | | | If a terminating ractor has child threads, then kill all child threads.
* a part of T_DATA object can Ractor#sendKoichi Sasada2020-11-061-0/+26
| | | | | | | | | T_DATA objects can refer unshareable objects and they should be copied recursively, however there is no way to replace with copied unshareable objects. However, if a T_DATA object refers only shareable objects, there is no need to replace. So this kind of T_DATA object (such as Time, Dir, File::Status and so on) can be sent by Ractor.send.
* strip trailing spaces [ci skip]Nobuyoshi Nakada2020-11-021-1/+1
|
* Copy for Ractor.send() without marshal.Koichi Sasada2020-11-021-32/+79
| | | | Now copying objects do not need marshal protocol.
* suppport Ractor.send(move: true) for more detaKoichi Sasada2020-11-021-124/+350
| | | | This patch allows to move more data types.
* Ractor's "will" doesn't need copying.Koichi Sasada2020-10-311-31/+63
| | | | | | | | | | `r = Ractor.new{ expr }` generates the block return value from `expr` and we can get this value by `r.take`. Ractor.yield and Ractor#take passing values by copying on default. However, the block return value (we named it "will" in the code) is not referred from the Ractor because the Ractor is already dead. So we can pass the reference of "will" to another ractor without copying. We can apply same story for the propagated exception.
generated by cgit v1.2.3 (git 2.39.1) at 2024-04-30 11:54:32 +0000