Do not enter wfi when SBA access would not work

[bugadani]

In a recent probe-rs PR I've fixed a bunch of issues that caused RISC-V CPUs to be halted when they otherwise don't need to be. This, however, caused some issues with RTT: on a C3, if the CPU entered WFI, probe-rs was no longer able to read the RTT buffer (originall reported here).

Turns out this issue has been worked around in esp-idf long ago. I'm applying that workaround to esp-hal here.

All RISC-V CPUs are affected, it just doesn't look like that, because we accidentally have SYSTEM_CPU_WAIT_MODE_FORCE_ON enabled on some of them.

[Copilot]

Pull request overview

This PR addresses an issue where RISC-V ESP32 CPUs entering WFI (Wait For Interrupt) prevented debuggers from reading memory via SBA (System Bus Access) when SYSTEM_CPU_WAIT_MODE_FORCE_ON is disabled. The solution applies a workaround from esp-idf that checks if a debugger is connected before entering WFI mode.

Changes:

• Created a unified wait_for_interrupt() function in esp_hal::interrupt for both RISC-V and Xtensa architectures
• Removed architecture-specific idle_hook() implementations from task modules and replaced them with a unified version that calls wait_for_interrupt()
• Added RISC-V-specific logic to skip WFI when a debugger is connected and cpu_wait_mode_force_on is disabled

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
esp-hal/src/interrupt/riscv.rs	Added wait_for_interrupt() function with debugger detection logic and cpu_wait_mode_on() helper
esp-hal/src/interrupt/xtensa.rs	Added simple wait_for_interrupt() function wrapping waiti 0 instruction
esp-rtos/src/task/mod.rs	Added unified idle_hook() using esp_hal::interrupt::wait_for_interrupt()
esp-rtos/src/task/riscv.rs	Removed RISC-V-specific idle_hook() implementation
esp-rtos/src/task/xtensa.rs	Removed Xtensa-specific idle_hook() implementation
esp-hal/CHANGELOG.md	Added changelog entry for the new function
esp-rtos/CHANGELOG.md	Added changelog entry for the idle hook fix

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[bjoernQ]

Interesting - I remember I stumbled upon esp_cpu_wait_for_intr (and SYSTEM_CPU_WAIT_MODE_FORCE_ON) a couple of times but never realized what is really happening there (and why) 🤷‍♂️

Nice find!

[Copilot]

Pull request overview

Copilot reviewed 7 out of 7 changed files in this pull request and generated no new comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[bjoernQ]

we accidentally have SYSTEM_CPU_WAIT_MODE_FORCE_ON enabled on some of them

should we align that?

[bugadani]

Eventually yes, I imagine saving power is better than not saving power 😅 But this code comes from depths of C hell.

[bjoernQ]

LGTM

We probably want to create an issue to set SYSTEM_CPU_WAIT_MODE_FORCE_ON to a defined state?

[bugadani]

@MabezDev I'd like this backported as latest probe-rs needs it

[github-actions]

Created backport PR for esp-hal-1.0.x:

• [Backport esp-hal-1.0.x] Do not enter wfi when SBA access would not work #4892 with remaining conflicts!

Please cherry-pick the changes locally and resolve any conflicts.

git fetch origin backport-4782-to-esp-hal-1.0.x
git worktree add --checkout .worktree/backport-4782-to-esp-hal-1.0.x backport-4782-to-esp-hal-1.0.x
cd .worktree/backport-4782-to-esp-hal-1.0.x
git reset --hard HEAD^
git cherry-pick -x 3d68e27485e56aeeadc76ebe494a8bfd89579545
git push --force-with-lease