CV-Bowen opened a new pull request, #21385:
URL: https://github.com/apache/nuttx/pull/21385
## Summary
Add three work queue APIs that a driver owning its own queues needs:
- `work_queue_in_queue()`: whether the caller runs on one of the
queue's worker threads.
- `work_queue_dispatch()`: run the expired work of a queue from one of
its worker threads. `work_thread()`'s loop is moved into
`work_dispatch()` for that; the exit checks added by "harden custom
queue lifecycle" are kept.
- `work_queue_setpriority()`/`work_queue_setpriority_wq()`: the setter
matching `work_queue_priority()`.
The rpmsg rx work queues use all three: rx is handled on per device
queues, an rpmsg call that waits on a reply from a worker thread drains
its own queue with `work_queue_dispatch()` meanwhile, and a queue runs
at the priority of the endpoints it serves. That rpmsg change follows in
a separate PR.
## Impact
API additions only. Existing queues behave as before: the worker loop
runs the same steps, only from a function.
## Testing
Host: Ubuntu 22.04.5 LTS x86_64, QEMU version 10.2.1, sim: cc (Ubuntu
13.1.0-8ubuntu1~22.04) 13.1.0, qemu-armv8a: aarch64-none-elf-gcc (Arm GNU
Toolchain 12.2.Rel1 (Build arm-12.24)) 12.2.1 20221205
sim:smp (4 CPUs):
- `ostest wqueue`: all cases pass (HPWORK, custom queue with 1 and 2
threads, priority, periodic requeue, pending replacement, sync and
parallel cancel, teardown, self free).
- A test-only app (not in this PR) on a custom queue with one thread:
`work_queue_in_queue()` is false from main and true from a work item;
a work item queues a second item and calls `work_queue_dispatch()`,
and the second item has run when the call returns (with one thread it
can only run there); `work_queue_setpriority_wq(wq, 120)` makes
`work_queue_priority_wq()` return 120 and an out of range priority
returns -EINVAL. Output: `WQTEST PASS`.
sim:ostest: the full ostest (5 loops) runs the wqueue test in each loop
with every `call = 100, expect = 100`.
<details><summary>ostest wqueue on sim:smp</summary>
```
WQTEST PASS
wqueue_test: backend = flat
wqueue_test: HPWORK
wqueue_test: HPWORK done
wqueue_test: custom queue, threads = 1
wqueue_test: priority = 100, expect = 100
wqueue_test: API validation
wqueue_test: API validation done
wqueue_test: periodic requeue
wqueue_test: periodic calls = 3, expect = 3
wqueue_test: pending replacement
wqueue_test: replacement total = 2, expect = 2
wqueue_test: sync cancel finished = 1, expect = 1
wqueue_test: custom queue, threads = 1 done
wqueue_test: custom queue, threads = 2
wqueue_test: priority = 100, expect = 100
wqueue_test: sync cancel finished = 1, expect = 1
wqueue_test: parallel sync cancel
wqueue_test: parallel callbacks = 2, expect = 2
wqueue_test: custom queue, threads = 2 done
wqueue_test: multiple custom queues
wqueue_test: multiple queues loop 1/4 done
wqueue_test: multiple queues loop 2/4 done
wqueue_test: multiple queues loop 3/4 done
wqueue_test: multiple queues loop 4/4 done
wqueue_test: multiple custom queues done
wqueue_test: self free
wqueue_test: self free result = -35, expect = -35
wqueue_test: teardown
wqueue_test: pending callback = 0, expect = 0
wqueue_test: running callback = 1, expect = 1
wqueue_test: teardown requeue rejected
wqueue_test: teardown done
```
</details>
rpmsg baseline:
No regression in rpmsg: sim:rpserver/rpproxy and
qemu-armv8a:rpserver/rpproxy boot and pass the fixed rpmsg cases (dump, ping,
rpmsgfs; 15/15). qemu-armv8a was built with LIBC_ENVPATH and BINFMT_CHECKEXEC
disabled, which current master needs for nxinit to start nsh there; unrelated
to this change.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
