887: executor: miri fixes r=Dirbaio a=Dirbaio
Fixes a few MIRI errors due to loosely mixing `&TaskStorage<F>` and `&TaskHeader`. References "downgrade" the provenance. `TaskHeader` is smaller, so once you have a `&TaskHeader` you can't use pointer casts to access the whole `TaskStorage<F>`. This fixes it by always keeping the raw pointer around, which doesn't downgrade provenance.
The error was:
```
[dirbaio@mars std]$ MIRIFLAGS=-Zmiri-disable-isolation cargo miri run --bin tick
Finished dev [unoptimized + debuginfo] target(s) in 0.05s
Running `/home/dirbaio/.rustup/toolchains/nightly-2022-07-13-x86_64-unknown-linux-gnu/bin/cargo-miri target/miri/x86_64-unknown-linux-gnu/debug/tick`
error: Undefined Behavior: trying to reborrow <12349> for SharedReadWrite permission at alloc2[0x30], but that tag does not exist in the borrow stack for this location
--> /home/dirbaio/embassy/embassy/embassy-executor/src/executor/raw/mod.rs:162:20
|
162 | let this = &*(p.as_ptr() as *const TaskStorage<F>);
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
| |
| trying to reborrow <12349> for SharedReadWrite permission at alloc2[0x30], but that tag does not exist in the borrow stack for this location
| this error occurs as part of a reborrow at alloc2[0x30..0x40]
|
= help: this indicates a potential bug in the program: it performed an invalid operation, but the Stacked Borrows rules it violated are still experimental
= help: see https://github.com/rust-lang/unsafe-code-guidelines/blob/master/wip/stacked-borrows.md for further information
help: <12349> was created by a retag at offsets [0x0..0x30]
--> src/bin/tick.rs:15:1
|
15 | #[embassy_executor::main]
| ^^^^^^^^^^^^^^^^^^^^^^^^^
= note: backtrace:
= note: inside `embassy_executor::executor::raw::TaskStorage::<std::future::from_generator::GenFuture<[static generator@src/bin/tick.rs:15:1: 15:26]>>::poll` at /home/dirbaio/embassy/embassy/embassy-executor/src/executor/raw/mod.rs:162:20
= note: inside closure at /home/dirbaio/embassy/embassy/embassy-executor/src/executor/raw/mod.rs:390:13
= note: inside `embassy_executor::executor::raw::run_queue::RunQueue::dequeue_all::<[closure@embassy_executor::executor::raw::Executor::poll::{closure#1}]>` at /home/dirbaio/embassy/embassy/embassy-executor/src/executor/raw/run_queue.rs:69:13
= note: inside `embassy_executor::executor::raw::Executor::poll` at /home/dirbaio/embassy/embassy/embassy-executor/src/executor/raw/mod.rs:373:9
= note: inside `embassy_executor::executor::Executor::run::<[closure@src/bin/tick.rs:15:1: 15:26]>` at /home/dirbaio/embassy/embassy/embassy-executor/src/executor/arch/std.rs:52:22
```
Co-authored-by: Dario Nieuwenhuis <dirbaio@dirbaio.net>
882: Remove separation of stable vs. nightly implementations in executor/raw r=Dirbaio a=jannic
Since rust 1.61, the required const features are availabe on stable rust
Co-authored-by: Jan Niehusmann <jan@gondor.com>
879: Improve ADC configuration options r=Dirbaio a=chemicstry
This smooths out a few rough edges with ADC:
- Make `Resolution::to_max_count()` pub for all ADC versions. Useful if performing conversion manually.
- `VREF` and `VREFINT` were convoluted. `VREF` is an external voltage reference used for all ADC conversions and it is exposed as a separate pin on large chips, while on smaller ones it is connected to `VDDA` internally. `VREFINT` is an internal voltage reference connected to one of the ADC channels and can be used to calculate unknown `VREF` voltage.
- Add a separate `VREF_DEFAULT_MV`, equal to 3.3V, which is the usual supply voltage and should work for most simple cases.
- For an external voltage reference `set_vref()` can be used to set a known precise voltage.
- If no voltage reference is present, `VREFINT` calibration can be used to calculate unknown `VREF` voltage. If I understand correctly, this is only implemented for `adc_v3` (L4?), but is not currently exposed (private). If someone is interested, this can be fixed in separate PR.
Co-authored-by: chemicstry <chemicstry@gmail.com>
876: Add defmt support to embassy-embedded-hal errors r=Dirbaio a=matoushybl
`defmt::unwrap!()` should now work with shared buses. I tested it only with I2C as I don't have SPI in the target project.
Co-authored-by: Matous Hybl <hyblmatous@gmail.com>
856: embassy-stm32/pwm: Generalize channel selection r=Dirbaio a=bgamari
550da471be previously refactored the STM32
PWM logic in such a way to preclude use of non-contiguous channels (e.g.
channel 2 but not channel 1). Refactor it yet again to yet again allow
this sort of usage.
Co-authored-by: Ben Gamari <ben@smart-cactus.org>
Co-authored-by: Dario Nieuwenhuis <dirbaio@dirbaio.net>
550da471be previously refactored the STM32
PWM logic in such a way to preclude use of non-contiguous channels (e.g.
channel 2 but not channel 1). Refactor it yet again to yet again allow
this sort of usage.
842: WIP: Make unborrow safe to use r=Dirbaio a=GrantM11235
The basic idea is that `Unborrow::unborrow` is now safe to call and returns an `Unborrowed<'a, T>` which impls `Deref` and `DerefMut`
```rust
/// This is essentially a `&mut T`, but it is the size of `T` not the size
/// of a pointer. This is useful if T is a zero sized type.
pub struct Unborrowed<'a, T> {
inner: T,
_lifetime: PhantomData<&'a mut T>,
}
```
## Todo
- [x] Update other crates
- [x] embassy-cortex-m
- [x] embassy-hal-common
- [x] embassy-lora
- [x] embassy-nrf
- [x] embassy-rp
- [x] embassy-stm32
- [x] Remove usage of the unsafe `into_inner` method if possible
- [x] Review and amend docs for `Unborrow` and `Unborrowed`
Co-authored-by: Grant Miller <GrantM11235@gmail.com>
Co-authored-by: Dario Nieuwenhuis <dirbaio@dirbaio.net>
871: nrf/saadc: add type-erased AnyInput. r=Dirbaio a=Dirbaio
872: nrf/usb: prevent user code from constructing a PowerUsb directly. r=Dirbaio a=Dirbaio
PowerUsb must be constructed through `new()` so that it sets up the IRQ.
It must have at least one private field, otherwise user code can construct
it directly with `PowerUsb{}`.
Co-authored-by: Dario Nieuwenhuis <dirbaio@dirbaio.net>
874: Add inline attribute to embassy-rp async functions r=Dirbaio a=danbev
This commit adds the inline attribute to the recently added async gpio
functions. This is to enable cross-crate inlining and to be consistent
with the other functions implemented for Input and Flex.
Co-authored-by: Daniel Bevenius <daniel.bevenius@gmail.com>
This commit adds the inline attribute to the recently added async gpio
functions. This is to enable cross-crate inlining and to be consistent
with the other functions implemented for Input and Flex.
PowerUsb must be constructed through `new()` so that it sets up the IRQ.
It must have at least one private field, otherwise user code can construct
it directly with `PowerUsb{}`.
868: Rename XXBusDevice to XXDevice. r=Dirbaio a=Dirbaio
I think the current naming is a bit odd, the EH traits are organized as "Bus" vs "Device", and XxxBusDevice is a Device (not a Bus, not both)
`@kalkyl`
Co-authored-by: Dario Nieuwenhuis <dirbaio@dirbaio.net>
853: Add embedded_hal_async support for embassy-rp r=Dirbaio a=danbev
This commit adds support for embedded-hal-async to the Embassy
Raspberry PI crate.
Co-authored-by: Daniel Bevenius <daniel.bevenius@gmail.com>
