Clock tree rework tracking issue

[bugadani]

This issue tracks the state of the clock tree rework.

• #4501
• Model clock trees of other chips
  • #4517
  • #4527
  • [ ] ESP32-C5
  • #4553
  • [ ] ESP32-C61
  • #4595
  • [ ] ESP32-P4
  • #4610
  • #4633
• [ ] Model all peripheral clock sources
• #4660
• [ ] Model missing options in clock trees:
  • ESP32:
    • [ ] XTAL32K external oscillator option
    • [ ] Configurable CK8M (RC_FAST_CLK)
    • [ ] LOW_POWER_CLK
  • ...
• [ ] Detect XTL frequency on demand to support clocking CPU from RC_FAST_CLK
• Optimize away unnecessary reference counts
  • [ ] Mark which clock nodes may be manually requested. Peripheral clocks should be, but out of system clocks probably only CPU.
  • #4512
  • [ ] ...?
• [ ] Add a bitmap that tracks active clocks (this can be postponed for upcoming low power rework)
• [x] Ensure code generator orders the initial configuration properly - #4536
• [x] Resolve #4440
• [ ] Decide whether caching the configuration is necessary or we can restore from hardware
• [ ] Decide whether we should cache active frequencies, or calculating them from the configurations is good enough
  • [ ] We should likely cache, but we need to implement this in a way that does not require a critical section to read current frequencies. Since we expect a clock to not be running when configured, maybe that is enough to remain consistent?
• [ ] Implement equation-based PLL configuration (for ESP32 APLL)
• [x] Remove ESP_HAL_CONFIG_XTAL_FREQUENCY - #4517
• [ ] Retain clock tree state between CPU-only resets
• [ ] Make configuration of always_on clocks and their dependencies mandatory
• [ ] Generate code directly into esp-hal, auto-update the list of handwritten functions
• [ ] Allow hand-written doc comments on clock nodes. Some types/config options need more explanation.
• [ ] Auto-generate on_changed callbacks
• #4665
• #4667
• [ ] Improve clock node request/release API - https://github.com/esp-rs/esp-hal/pull/4647#discussion_r2619314443
• [ ] Add the ability to iterate over a mux's variants - https://github.com/esp-rs/esp-hal/pull/4663/changes/1e1388dbd47f94987d8db525f704529509ad54aa#diff-f4823be2299dcbe2eea9219ec3ce12b28541a53c0f91e7c80428813abed2b309R183-R185

[Frostie314159]

I suppose, we can't make the RF clocks part of this due to the lack of info about them.

[bugadani]

What we can read out of esp-idf is definitely in scope, and there's plenty there still.