The weekly driver initiative

[japaric]

There's now a HAL published on crates.io: embedded-hal (Introduction blog post). 🎉

To put it to test and to expand the crates.io ecosystem I'm starting the weekly driver initiative.

The goal is to release a new no_std, generic, embedded-hal based driver crate every one or two
weeks. Driver crate here means a library crate that lets you interface an external component like a
digital sensor or a wireless transceiver. A driver release consists of (a) publishing a library
crate on crates.io (*) and (b) presenting your driver to the community via any means: a blog post,
a tweet, a reddit post, a u.r-l.o https://users.rust-lang.org post, etc.

(*) Make sure you tag your crate with the embedded-hal-driver keyword!

If you'd like to participate leave a comment here stating the driver you'll be working on. If you
can post a link to the repo where the (WIP) driver code is or will be, even better! If you don't
have a personal blog to present your driver to the community you can do a guest post on my blog.

For a list of released and WIP drivers check the awesome-embedded-rust list. If you'd like your crate to be added to that list simply send a PR to that repo!

TODO (potentially):

These are components that @japaric has at hand:

• DHT22 - temperature + humidity sensor (1-wire)
• ENC28J60 - ethernet controller (SPI)
• HC-06 - bluetooth (AT)
• HC-12 - wireless transceiver (AT)
• HC-SR04 - ultrasonic sensor
• L298N - Motor driver
• NEO-6M - GPS (USART)
• NEO-M8N - GPS (USART)
• PCA9685 - 16 channel Servo driver (I2C)
• PCF8574 (+ 16x2 LCD) - 8-bit I/O expander (I2C)
• PN532 - NFC (SPI)
• RA-02 - Long-range Radio (SPI)
• SN65HVD230 - CAN
• SSD1306 - 128x64 OLED (I2C)
• Servomotor (?)
• WS2812B - smart LED
• nRF24L01 - wireless transceiver (SPI)

[niklasad1]

I can probably help porting some NRF51/NRF52 drivers from TockOS that I have been involved in. I think the BLE driver could be interesting to have in a separate crate.

Let me know if you think it is a good idea!

[thejpster]

I'm going to look at my Raspberry Pi SenseHat crate and my LM4F120 chip crate, as used by the Stellaris Launchpad, as soon as I can. If TockOS starts using this HAL internally then new chip support should be easier to add in the future. I found the LM4F120 port fairly hard going.

[dbrgn]

If I find time I could try to do a DS18B20 (dallas one-wire) temperature sensor driver implementation.

Current stumbling blocks so far (I'm developing on the LPC11UXX):

• "overloaded" registers svd2rust#16
• ICE when trying to cross compile with xargo rust-lang/rust#47040

(In both cases, what @rnestler wrote)

[therealprof]

I've tons of devices at hand and I might write drivers for them but first I need to get the device crates and HALs up to snuff.

Apart from the usual Serial/I2C/SPI suspects what I'd like to see is whether the HAL approach is also viable for bit-banged GPIO, potentially even timing critical devices, some of which are also in @japaric's list, like the Neopixels. Something that would be really killer to have is support for multiplexed and charlieplexed LED matrices.

[alevy]

If TockOS starts using this HAL internally then new chip support should be easier to add in the future. I found the LM4F120 port fairly hard going.

I'd personally be highly supportive of that, but would need some convincing that it would fit the capsule architecture (that hasn't seemed to be the case for previous iterations). Perhaps @niklasad1 can report some insights if he ports his BLE stack from Tock.

[therealprof]

@japaric Is there going to be a registry for embedded stuff and especially drivers? crates.io is becoming more and more useless to find interesting crates...

[thejpster]

The LM4F120 crate now implements embedded_hal::serial::{Read, Write}. It was pretty straightforward.

[japaric]

@niklasad1

Let me know if you think it is a good idea!

More drivers is a good idea, IMO. :-)

porting some NRF51/NRF52 drivers

Just to clarify: Is this driver about interfacing an external pre-flashed nRF5x device, or is it some BleRadio trait on top of which a generic BLE stack con be built? In the later case you implement the BleRadio trait for, say, a nrf51 device crate and you get a BLE stack for free (ideally).

@therealprof

I'd like to see is whether the HAL approach is also viable for bit-banged GPIO

At the very least we'll want to add a digital::OutputPort trait whose API changes the state of a whole port (8 or 16 pins) atomically (single register op). And maybe another trait to change the state of N pins that belong to the same port atomically.

potentially even timing critical devices, ..., like the Neopixels

The words Neopixels and bit banging reminded me of an Arduino library I saw. It implemented the protocol using assembly and was fully blocking (nop delays) to meet the timing constraints. I hope that our neopixels driver provides a DMA based (asynchronous) mode, apart from a bit banged mode.

Is there going to be a registry for embedded stuff and especially drivers?

I meant to include this in the issue description: I'm using the embedded-hal keyword for my driver crates. That makes them easier to find. Perhaps we can just use crates.io and keywords instead of creating another registry / list. We can bikeshed the keywords to use in this thread: perhaps embedded-hal (or embedded-hal-driver) for the driver crates, embedded-hal-impl for the HAL impl crates, like the stm32f30x-hal crate, and svd2rust for the svd2rust generated device crates?

[therealprof]

@japaric

Just to clarify: Is this driver about interfacing an external pre-flashed nRF5x device, or is it some BleRadio trait on top of which a generic BLE stack con be built? In the later case you implement the BleRadio trait for, say, a nrf51 device crate and you get a BLE stack for free (ideally).

Coincidentally I'm working on a micro:bit crate (which is based on the NRF51822, as I'm pretty sure you're aware of 😉). I planned on adapting the BLE code from the Zephyr project, since Nordic is quite active there providing a BLE implementation for their devices without the humongous softdevice blob.

It's not quite ready for primetime yet since the documentation is somewhat lacking and reversing Zephyr somewhat time consuming due to the many layers of abstraction and indirection.

But your recent work and post kind of beat me to adapting what I have to the new singleton approach and also use the embedded-hal (which I previously haven't), so I'm trying to put some more time in.

At the very least we'll want to add a digital::OutputPort trait whose API changes the state of a whole port (8 or 16 pins) atomically (single register op). And maybe another trait to change the state of N pins that belong to the same port atomically.

Yeah, that's going to be interesting. But the reason I mentioned this is that there somehow (and I'm not sure what this will might look like just yet) should be a way to facility the required high speed updates.

The words Neopixels and bit banging reminded me of an Arduino library I saw. It implemented the protocol using assembly and was fully blocking (nop delays) to meet the timing constraints. I hope that our neopixels driver provides a DMA based (asynchronous) mode, apart from a bit banged mode.

Some of those WS controllers can be driven by abused SPI peripherals provided by some MCUs but that can't be taken for granted. AFAIR that doesn't work on the RasPi and that one also cannot do the required timing via bitbanging the GPIOs.

I meant to include this in the issue description: I'm using the embedded-hal keyword for my driver crates. That makes them easier to find. Perhaps we can just use crates.io and keywords instead of creating another registry / list. We can bikeshed the keywords to use in this thread: perhaps embedded-hal (or embedded-hal-driver) for the driver crates, embedded-hal-impl for the HAL impl crates, like the stm32f30x-hal crate, and svd2rust for the svd2rust generated device crates?

That might work but has quite a number of disadvantages:

• It relies on people knowing how to correctly tag the crates and actually do it
• crates.io doesn't tracking of crates so one doesn't get notified about new crates of interest or updates
• The quality of the entries varies a lot and investigating potentially relevant crates consumes a lot of time with a strong tendency to becoming even worse
• The amount of search results is often overwhelming and there's virtually no support narrow down the search. Just for fun I just tried this query and I didn't find the result very encouraging...

crates.io is okay for larger std crates but for the hundreds or even thousands of embedded crates we're going to have, this is going nowhere fast... Originally I was counting on japaric.io or rust-embedded becoming the authoritative source for everything embedded in Rust but somehow this didn't quite happen. ;)

My hopes are still that we can establish a lively community around those topics, with a comprehensive catalog of relevant crates (architectural, register mapping, MCU HAL, peripheral HAL, BSP and specific implementations), articles on the use of any of that and general development topics, guides for embedded Rust development and support (development tools and debugging software/hardware to use) and potentially even general MCU howtos (selecting and connecting peripherals to an MCU, developing your own hardware...)...

[jacobrosenthal]

Greetings, Another open source nrf ble stack to check out is mynewt's nimble, which is the rtos im currently using in my daily work. https://github.com/apache/mynewt-core/tree/master/net/nimble Im hoping to try out rust for the first time, and ble on the microbit would be a great reason for me to finally dig in as 'full' functionality (bootloading over ble mainly) on nrf51822 devices is tough to come by with every other rtos out there due to the low memory. Curious how rust could stand up. This is much futher out there, but the bootloader zephyr and mynewt team are collaborating on is called mcuboot. It would be pretty cool to support that, allowing people to transition devices away from zephyr/mynewt into a rust rtos. https://github.com/runtimeco/mcuboot/ For now Im quietly lurking and watching you all. Keep up the great work

…

On Thu, Jan 18, 2018 at 6:03 PM, Daniel Egger ***@***.***> wrote: @japaric <https://github.com/japaric> Just to clarify: Is this driver about interfacing an external pre-flashed nRF5x device, or is it some BleRadio trait on top of which a generic BLE stack con be built? In the later case you implement the BleRadio trait for, say, a nrf51 device crate and you get a BLE stack for free (ideally). Coincidentally I'm working on a micro:bit crate (which is based on the NRF51822, as I'm pretty sure you're aware of 😉). I planned on adapting the BLE code from the Zephyr project, since Nordic is quite active there providing a BLE implementation for their devices without the humongous softdevice blob. It's not quite ready for primetime yet since the documentation is somewhat lacking and reversing Zephyr somewhat time consuming due to the many layers of abstraction and indirection. But your recent work and post kind of beat me to adapting what I have to the new singleton approach and also use the embedded-hal (which I previously haven't), so I'm trying to put some more time in. At the very least we'll want to add a digital::OutputPort trait whose API changes the state of a whole port (8 or 16 pins) atomically (single register op). And maybe another trait to change the state of N pins that belong to the same port atomically. Yeah, that's going to be interesting. But the reason I mentioned this is that there somehow (and I'm not sure what this will might look like just yet) should be a way to facility the required high speed updates. The words Neopixels and bit banging reminded me of an Arduino library I saw. It implemented the protocol using assembly and was fully blocking (nop delays) to meet the timing constraints. I hope that our neopixels driver provides a DMA based (asynchronous) mode, apart from a bit banged mode. Some of those WS controllers can be driven by abused SPI peripherals provided by some MCUs but that can't be taken for granted. AFAIR that doesn't work on the RasPi and that one also cannot do the required timing via bitbanging the GPIOs. I meant to include this in the issue description: I'm using the embedded-hal keyword for my driver crates. That makes them easier to find. Perhaps we can just use crates.io and keywords instead of creating another registry / list. We can bikeshed the keywords to use in this thread: perhaps embedded-hal (or embedded-hal-driver) for the driver crates, embedded-hal-impl for the HAL impl crates, like the stm32f30x-hal crate, and svd2rust for the svd2rust generated device crates? That might work but has quite a number of disadvantages: - It relies on people knowing how to correctly tag the crates and actually do it - crates.io doesn't tracking of crates so one doesn't get notified about new crates of interest or updates - The quality of the entries varies *a lot* and investigating potentially relevant crates consumes a lot of time with a strong tendency to becoming even worse - The amount of search results is often overwhelming and there's virtually *no* support narrow down the search. Just for fun I just tried this query <https://crates.io/search?q=embedded&sort=> and I didn't find the result very encouraging... crates.io is *okay* for larger std crates but for the hundreds or even thousands of embedded crates we're going to have, this is going nowhere fast... Originally I was counting on japaric.io <http://blog.japaric.io/> or rust-embedded <http://www.rust-embedded.org/> becoming the authoritative source for everything embedded in Rust but somehow this didn't quite happen. ;) My hopes are still that we can establish a lively community around those topics, with a comprehensive catalog of relevant crates (architectural, register mapping, MCU HAL, peripheral HAL, BSP and specific implementations), articles on the use of any of that and general development topics, guides for embedded Rust development and support (development tools and debugging software/hardware to use) and potentially even general MCU howtos (selecting and connecting peripherals to an MCU, developing your own hardware...)... — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#39 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAb0dPHuDDZgZCXBl3-PWRaJb2VPGVj_ks5tL-nugaJpZM4RiLZg> .

[therealprof]

@jacobrosenthal

Another open source nrf ble stack to check out is mynewt's nimble, which is
the rtos im currently using in my daily work.
https://github.com/apache/mynewt-core/tree/master/net/nimble

I am aware, I tried everything I could find for that MCU, including (but not limited to) Mynewt, RIOT and Zephyr. Most ecosystems rely on the softdevice while some others provide their own (but often very limited) BLE stack. The Zephyr stack is written by the vendor Nordic themself and has been certified so it is known to be feature complete and (somewhat) correct.

Im hoping to try out rust for the first time, and ble on the microbit would
be a great reason for me to finally dig in as 'full' functionality
(bootloading over ble mainly) on nrf51822 devices is tough to come by with
every other rtos out there due to the low memory. Curious how rust could
stand up.

Rust allows for ridiculously small firmwares on those MCUs (compared to other ecosystems that is) so I'm confident that there won't be any space issues per se. Of course if you add a BLE capable bootloader for DFU that whole situation changes a good deal because that will take a sizeable amount of flash by itself. I wouldn't hold my breath for that to materialize. ;)

https://github.com/runtimeco/mcuboot/

Bootloaders don't have a very high priority for me at the moment but adding support for mcuboot might be straight forward. Question is what the benefits would be, especially on the micro:bit but I'll have a look, thanks.

[whitequark]

Rust allows for ridiculously small firmwares on those MCUs (compared to other ecosystems that is) so I'm confident that there won't be any space issues per se.

Is that really true? The generated format! code is very bloated, for example, with its myriad of vtables, tables, floating point...

[therealprof]

@whitequark

Is that really true? The generated format! code is very bloated, for example, with its myriad of vtables, tables, floating point...

That's true, format! is relatively expensive due to the incapability to properly see the used types and monomorphize; however (unless there's another bug I'm not aware of, not using format! too much in my code) it should be a one-time overhead which is comparable to printf() formatting in C and no-one forces you to use formatters; outputting strings using write_str() as well as "manually" formatting numbers e.g. using the NumToA crate.

Here I have an example rendering a counter triggered by USART input onto an I2C driven SSD1306 OLED display:

File .text Size Name
0.2% 28.4% 604B USART1
0.2% 26.5% 564B main
0.1% 17.9% 380B [41 Others]
0.1% 11.3% 240B compiler_builtins::int::udiv::__udivsi3
0.0% 5.5% 116B i2c_ssd1306::ssd1306_print_bytes
0.0% 3.9% 84B stm32f042::peripherals::i2c::write_data
0.0% 3.0% 64B cortex_m_rt::reset_handler
0.0% 1.4% 30B __udivmodsi4
0.0% 0.8% 18B __aeabi_memcpy
0.0% 0.8% 18B __aeabi_uidivmod
0.0% 0.5% 10B __aeabi_uidiv
0.7% 100.0% 2.1KiB .text section size, the file size is 310.9KiB

This (a bit older example) includes write_str use and an implementation on the serial port and an somewhat similar implementation for the display rendering...

[fmckeogh]

I have both an HC05 and HC06 pair and a tricolour e-ink display. I should be able to start work on these pretty soon.