Commit Graph

35 Commits

Author SHA1 Message Date
Pedro J. Estébanez 5b5cdf2414 Fixup recent changes to threading concerns
ResourceLoader:
- Fix invalid tokens being returned.
- Remove no longer written `ThreadLoadTask::dependent_path` and the code reading from it.
- Clear deadlock hazard by keeping the mutex unlocked during userland polling.

WorkerThreadPool:
- Include thread call queue override in the thread state reset set, which allows to simplify the code that handled that (imperfectly) in the ResourceLoader.
- Handle the mutex type correctly on entering an allowance zone.

CommandQueueMT:
- Handle the additional possibility of command buffer reallocation that mutex unlock allowance introduces.
2024-07-16 11:03:02 +02:00
Pedro J. Estébanez 10b543f8a7 WorkerThreadPool: Fix wrong sync logic breaking task map integrity 2024-07-15 12:15:25 +02:00
Pedro J. Estébanez f2f9a6b8a1 WorkerThreadPool: Fix wrong pointer used in the case of BinaryMutex 2024-06-28 19:46:55 +02:00
Rémi Verschelde 6f8b90e412 Merge pull request #93032 from RandomShaper/wtp_antilock
GDScript: Avoid deadlock possibility in multi-threaded load
2024-06-28 14:42:48 +02:00
Pedro J. Estébanez 03d14e436b WorkerThreadPool: Refactor deadlock prevention collaboration into a generic mechanism
This is strictly beyond a refactor because it also changes when the mutexes are relocked,
but that's only for extra safety.
2024-06-19 14:01:54 +02:00
Pedro J. Estébanez 21c03d1956 WorkerThreadPool: Fix thread message queue not restored after overridden in a task
Also, simplifies the thread override teardown in MessageQueue.
2024-06-13 10:31:08 +02:00
Pedro J. Estébanez 8192d1cbc7 WorkerThreadPool: Fix yield-over for not-yet-started tasks 2024-04-18 19:25:19 +02:00
Pedro J. Estébanez 9653bcaa43 WorkerThreadPool: Polish yielding (fix corner case, remove misleading warning) 2024-04-17 18:32:14 +02:00
Pedro J. Estébanez 1b104ffcd8 WorkerThreadPool: Support daemon-like tasks (via yield semantics) 2024-04-10 18:47:42 +02:00
Pedro J. Estébanez 71facbaa88 WorkerThreadPool: Fix data race 2024-04-10 18:47:41 +02:00
Robert Yevdokimov 13e82094ee Remove word duplicates in comments and strings, and fix casing and punctuation 2024-02-23 17:28:28 -05:00
Adam Scott bd70b8e1f6 Add THREADS_ENABLED macro in order to compile Godot to run on the main thread 2024-01-17 13:58:29 -05:00
Pedro J. Estébanez a731774813 WorkerThreadPool: Avoid most runtime allocations
Just a little optimization.

**NOTE:**
With `RID_Owner` we could replace each pair of `PagedAllocator` and
`HashMap`-of-ids-to-pointers. However, that would force us to expose
`RID` as the task/group id, instead of `int`, which would break the
API. Too bad. Let's wait until Godot 5.0.
2024-01-08 12:46:31 +01:00
Pedro J. Estébanez ae418f9469 WorkerThreadPool: Avoid deadlocks when CommandQueueMT is involved
This commit lets CommandQueueMT play nicely with the WorkerThreadPool to avoid
non-progressable situations caused by an interdependence between both. While a
command queue is being flushed, it allows the WTP to release its lock while tasks
are being awaited so they can make progress in case they need in turn to post
to the command queue.
2024-01-08 12:45:43 +01:00
Pedro J. Estébanez 9444d297ed WorkerThreadPool: Overhaul scheduling and synchronization
This commits rewrites the sync logic in a way that the
`use_system_threads_for_low_priority_tasks` setting, which was added due to
the lack of a cross-platform wait-for-multiple-objects functionality, can be
removed (it's as if it was effectively hardcoded to `false`).

With the new implementation, we have the best of both worlds: threads don't
have to poll, plus no bespoke threads are used.

In addition, regarding deadlock prevention, since not every possible case of
wait-deadlock could be avoided, this commits removes the current best-effort
avoidance mechanisms and keeps only a simple, pessimistic way of detection.

It turns out that the only current user of deadlock prevention, ResourceLoader,
works fine with it and so every possible situation in resource loading is now
properly handled, with no possibilities of deadlocking. There's a comment in
the code with further details.

Lastly, a potential for load tasks never being awaited/disposed is cleared.
2024-01-08 12:45:42 +01:00
Pedro J. Estébanez cba15cd625 Add WorkerThreadPool::get_thread_index() 2023-12-01 16:42:26 +01:00
Pedro J. Estébanez f3e96a8548 Make languages bookkeeping thread-safe 2023-11-09 18:03:43 +01:00
kobewi 09b30be86d Add vararg call() method to C++ Callable 2023-10-05 11:35:29 +02:00
A Thousand Ships 893f889d74 [Core] Replace ERR_FAIL_COND with ERR_FAIL_NULL where applicable 2023-09-11 19:45:49 +02:00
Rémi Verschelde 862a2841c0 Merge pull request #78845 from RandomShaper/fix_wtp_zero_batch_lazy
Fix zero-sized WorkerThreadPool not processing group tasks
2023-07-07 08:27:43 +02:00
Pedro J. Estébanez 30925fcfb0 Fix management of safe-for-nodes flag in ResourceLoader and WorkerThreadPool 2023-07-03 12:27:57 +02:00
Pedro J. Estébanez 28d0d56a69 Fix zero-sized WorkerThreadPool not processing group tasks 2023-06-29 19:44:42 +02:00
Pedro J. Estébanez 2b001db76e Let user scripts disable thread safety checks 2023-06-09 18:24:01 +02:00
Pedro J. Estébanez a6e43f71b6 Avoid multiple possibilites of deadlock in resource loading 2023-05-17 01:52:39 +02:00
Rémi Verschelde 20ed51a912 Merge pull request #72704 from myaaaaaaaaa/single-main-thread
Partial fixes for `threading/worker_pool/max_threads` = 0
2023-05-12 15:09:50 +02:00