Updates all the PAC references for the `pygamer` BSP crate

[kyp44]

Summary

Updates PAC references for the pygamer BSP, which was necessary due to #576.

Checklist

• [x] CHANGELOG.md for the BSP or HAL updated
• [x] All new or modified code is well documented, especially public items
• [x] No new warnings or clippy suggestions have been introduced - CI will deny clippy warnings by default! You may #[allow] certain lints where reasonable, but ideally justify those with a short comment.

[jbeaurivage]

Thanks @kyp44! I'll keep this PR on standby for now, I want to make a new pygamer release before merging this to main, since this will mark the BSP's entry into Tier 1 status.

[jbeaurivage]

Closes #513.

[jbeaurivage]

[Follow-up of the conversation here]:

@jbeaurivage I did not update those dependencies, but it would definitely be nice to be able to use the latest versions of those. EDIT: I don't mind working on updating these (and perhaps any other outdated dependencies) if you like, either in this branch or in a new one. Just let me know.

I think that would be great. We should strive to provide the latest and greatest for T1 boards. I think the embedded-hal v1.0 SPI implementation is a bit buggy at the moment and has some fragile timing and race conditions. I'll work on trying to fix these. Hopefully you can help with testing, since I don't have a pygamer or an LCD display.

[kyp44]

Here are all the dependencies with major version status:

pygamer library

• cortex-m: Up to date at 0.7
• st7735-lcd: 0.8.1 -> 0.10.0
• cortex-m-rt: Up to date at 0.7
• usb-device: Up to date at 0.3.1 (latest is 0.3.2, which is just a minor change and so compatible), tied to usb feature

Examples only

• micromath: 0.5.1 -> 2.1.0
• embedded-sdmmc: Up to date at 0.8.0
• embedded-graphics: 0.7.1 -> 0.8.1
• smart-leds: 0.3 -> 0.4.0
• lis3dh: 0.1.0 -> 0.4.3
• tinybmp: 0.3.1 -> 0.6.0

Apparently unused (by the library or examples)

• cortex-m-rtic: 1.0 -> 1.1.4
• embedded-hal-02: Specific old version

Dependency questions / issues:

• micromath is listed as a feature then enabled as a dependency, but this seems to be for no particular reason. Nothing in the actual library uses this (it's not even re-exported), and it's only used in a couple of examples. It seems like this should just be moved to the dev dependencies.
• embedded-sdmmc is the same situation: tied to the sd-card feature, used in an example, but not used in the library at all, not even re-exported.

I imagine that we want to update all of the outdated library and examples dependencies, which I can do and test. I guess the question is whether I should create a separate branch and/or pull request for this, thoughts? Also, what are your thoughts about the issues above for micromath and embedded-sdmmc.

[jbeaurivage]

Thanks for writeup. I would suggest to work on updating the dependencies here (Could rename the PR to something along the lines of "Promote pygamer to Tier 1 status").

About your questions:

• micromath and embedded-sdmmc: Can definitely be moved to dev-dependencies
• cortex-m-rtic seems to be used in the button_rtic example. Most likely it shouldn't be too much effort to update the example to use RTIC v2
• I think we can yank out embedded-hal v0.2

Some additional thoughts

• Upgrading st7735-lcd to 0.10 will have the nice addition of using the embedded-hal v1.0 implementations of the HAL.

• I've recently been reading through our issues, and it seems there are a lot of reports of fragile timing/bad performance for SPI. Particularly, it affects timing-dependent drivers (e.g. neopixels #752), and might bring buffer overflows or significant slowdowns at high frequencies.

  I've just opened #772, which brings DMA transfers for embedded-hal SPI (and later on, I2C and UART) implementations. It would be great if you could test the new DMA impls with the pygamer display, and perhaps the recently removed neopixel examples.

  I don't expect the updated examples or any DMA stuff to make their way into this PR (depends on which will get merged first - probably this one), but it would definitely help build confidence in the implementation, and hopefully enable us to bring the neopixel examples back in a subsequent PR.

[kyp44]

Copy, I'll work on the updates just in this branch and rename the PR. I'll focus on the core library dependencies first, but it also shouldn't be a big deal to update the example dependencies too, unless this is being worked elsewhere.

I use the neopixels in my project, and they are generally problematic. Evidently they are driven with just a single pin that requires precise timing. The driver allows for various timing options. They actually seemed to work best with the old SpinTimer, which is now fully removed (was previously just deprecated). Currently I'm using a TimerCounter in critical sections because interrupts can screw up the timing for obvious reasons. This works alright but still glitches sometimes with incorrect colors on the LEDs. There is a version of the driver that uses a single SPI pin, but I have tried this, and it sounds like this is also problematic according #752.

Regarding the SPI implementation in general though, I have not experienced any issues or glitches with the display whatsoever. I can look into DMA to test with the display, but the display driver consumes a embedded_hal::spi::SpiDevice directly, so I'm not sure whether that would be compatible with using DMA. That's not so say I couldn't just send some display commands manually using DMA just for testing purposes.