Rust ESP32 Community Meeting 2022-01-04

[georgik]

Please add topics and links to this discussion.

The meeting is scheduled 4th January 2022 - 5pm CET/GMT+1 via Google Meet. The meeting invite is automatically sent to attendees of previous meetings. If you'd like to join, please notify us at https://matrix.to/#/#esp-rs-community-meetins:matrix.org

You can find information from previous meeting at https://github.com/esp-rs/rust/discussions

[georgik]

Important

• ESP-IDF master is now 5.0 - may contain breaking changes
• recommendation: setup build with release/4.4 and master in CI to validate compatibility

rustzx-esp32 project - research results

LCD performance related topic tested with C based project:

• Kaluga and ESP32-S3-USB-OTG - sending whole screen to the display is slow - less than 1 FPS
• turn on Double or Triple buffer using HAGL - https://github.com/tuupola/hagl_esp_mipi - speed about 20, up to 60 FPS in 120x120x16bit
• Triple buffer might not fit into RAM for whole display
• another sophisticated driver for ILI9341 in C++: https://github.com/juj/fbcp-ili9341

Kaluga LCD interference with SD card transfers

• discovered weird behavior of display when ESP32S2 reads from connected SD card - dots appear on the display from top left
• reason not clear yet, probably problem with SPI or DMA
• reading from spiffs works without problem

Complex input handling in Rust:

• Bluetooth and USB keyboard is not available in Rust yet
• alternative approach PS/2 keyboard

[ivmarkov]

• esp-idf-hal

  • ADC driver implemented
  • Experimental support for critical sections and interrupt enable/disable
  • Upcoming?: migration from embedded-hal V0.2 to embedded-hal V1.0.alpha6 (driven by @andresv)
  • Justification:
    • Prio A: TWAI driver by @andresv. TWAI Traits only available in embedded-hal V1.0
    • Prio B: ADC support for degraded pins (GpioPin) only possible with V1.0
    • Prio C: Other goodies not available in V0.2, like Watchdog traits

• esp-idf-sys/embuild

  • In progress: utilizing user-supplied ESP-IDF (via $IDF_PATH) as well as re-using an already activated ESP-IDF environment (#41 and #32)

• ESP-IDF realpath in master

  • Tested, looks good, pls merge in 4.4

• Next issues to address

  • Issues accumulating with our Rust STD support for filesystem operations (SPIFFS and VFAT). This was never properly tested. Needs some bugfixing
    • Q: can we have an implementation of ftruncate in ESP-IDF? There is a truncate implementation, but Rust STD needs ftruncate (i.e. the one based on a file descriptor rather than on a file path)

[MabezDev]

• Documentation for our esp-idf-* crates
  • Unless we a copy of vendor esp-idf in the repo we won't be able to build on docs.rs
  • Host them on GH pages like the book?
  • Host different idf versions? i.e is there value in hosting v4.3, v4.4 v5.0 etc
    • Could be useful as more build testing
• Allocator issue reported: Default allocator doesn't guarantee allocation bigger than 4? #99
  • I think we'll need to pass more HEAP_CAP options to the idf allocator to tune alignment etc.

[ivmarkov]

* Could be useful as more build testing

• Allocator issue reported: Default allocator doesn't guarantee allocation bigger than 4? #99

  • I think we'll need to pass more HEAP_CAP options to the idf allocator to tune alignment etc.

This is the relevant alloc code in Rust STD that calls into ESP-IDF. It calls either libc::malloc or libc::memalign. Also note that implementation of libc::posix_memalign landed just now to ESP-IDF 4.4, but not sure we really need to use it, given that libc::memalign is already supported?

In addition to that we need to fix the allocator in esp-idf-sys for the no_std case.

[ivmarkov]

• Documentation for our esp-idf-* crates

  • Unless we a copy of vendor esp-idf in the repo we won't be able to build on docs.rs
  • Host them on GH pages like the book?
  • Host different idf versions? i.e is there value in hosting v4.3, v4.4 v5.0 etc

Here the relevant issue with some thoughts from myself and @N3xed. It is in the context of esp-idf-sys only, but once we solve the problem for esp-idf-sys, the other crates are easier.

[N3xed]

• embuild
  • Fixes
    • git: Shallow submodules clone embuild#31
    • esp-idf-sys fails to build esp-idf-sys#49: embuild::pio::Framework failed to parse
    • Building in workspace, the .embuild folder is placed outside of the workspace dir esp-idf-sys#44:
      Invalid embuild::cargo::workspace_dir() when $HOST==$TARGET
  • General library improvements (e.g. upgrade to 2021 edition)
  • New minor release pending
• esp-idf-sys
  • ESP_ERR* constants should be i32 esp-idf-sys#46
  • Better sdkconfig.defaults handling
    • ESP_IDF_SDKCONFIG_DEFAULTS & ESP_IDF_SDKCONFIG should have default values, so to say esp-idf-sys#45
    • esp-rs/esp-idf-sys@db657cd
  • Upgraded to 2021 edition
  • Save esp-idf build info JSON file for cargo-idf
• cargo-idf
  • Tool for menuconfig, monitor, size, (flash, erase-flash) using the native build.
  • Similar to cargo pio espidf
  • Work in progress