Spin polls the raw executor and never sleeps. It is useful for disabling
any power features associated with wfi/wfe-like instructions.
When implementing support for the CH32V30x MCU, the wfi instruction
had issues interacting with the USB OTG peripheral and appeared to be
non-spec-compliant.
1. When sending a USB Data-in packet, the USB peripheral appears to be
unable to read the system main memory while in WFI. This manifests in
the USB peripheral sending all or partially zeroed DATA packets.
Disabling WFI works around this issue.
2. The WFI instruction does not wake up the processor when MIE is
disabled. The MCU provides a WFITOWFE bit to emulate the WFE instruction
on arm, which, when enabled, ignores the MIE and allows the processor to
wake up. This works around the non-compliant WFI implementation.
Co-authored-by: Codetector <codetector@codetector.org>
Co-authored-by: Dummyc0m <y@types.moe>
Without that feature one will find import errors on opening the webpage, that are hard to debug.
The feature was indirectly enabled in the wasm example, however the reason wasn't documented and thus it was easy to miss.
* Feature auto-documentation
* Task arena sizes in a <details> list
* Non-documented comment explaining turbowakers with see-also link
Further improvements:
* Are the task-arena-size-* numbers sizes in bytes? or something else?
* Task arena section could benefit from advice about how to choose a
suitable size
This introduces a `Pender` struct with enum cases for thread-mode, interrupt-mode and
custom callback executors. This avoids calls through function pointers when using only
the thread or interrupt executors. Faster, and friendlier to `cargo-call-stack`.
`embassy-executor` now has `arch-xxx` Cargo features to select the arch and to enable
the builtin executors (thread and interrupt).
This is a `core` patch to make wakers 1 word (the task pointer) instead of 2 (task pointer + vtable). It allows having the "waker optimization" we had a while back on `WakerRegistration/AtomicWaker`, but EVERYWHERE, without patching all crates.
Advantages:
- Less memory usage.
- Faster.
- `AtomicWaker` can actually use atomics to load/store the waker, No critical section needed.
- No `dyn` call, which means `cargo-call-stack` can now see through wakes.
Disadvantages:
- You have to patch `core`...
- Breaks all executors and other things that create wakers, unless they opt in to using the new `from_ptr` API.
How to use:
- Run this shell script to patch `core`. https://gist.github.com/Dirbaio/c67da7cf318515181539122c9d32b395
- Enable `build-std`
- Enable `build-std-features = core/turbowakers`
- Enable feature `turbowakers` in `embassy-executor`, `embassy-sync`.
- Make sure you have no other crate creating wakers other than `embassy-executor`. These will panic at runtime.
Note that the patched `core` is equivalent to the unpached one when the `turbowakers` feature is not enabled, so it should be fine to leave it there.
1054: riscv fixes r=lulf a=swolix
With these changes I can run embassy on our RISC-V processor, please consider merging this, feedback is very welcome.
I don't fully understand the code in the executor, but I have implemented a critical section by globally disabling interrupts, which means the wfi inside the critical section will hang the whole thing.
Co-authored-by: Sijmen Woutersen <sijmen.woutersen@gmail.com>
