regression: esp_hal::init() resets interrupts

[kaspar030]

Bug description

Since #2091, esp_hal::init() unconditionally disables all interrupts.

This broke our threading, as we set up the interrupt handler before, start threading, and then in a thread call esp_hal::init(). With #2091, this now disabled our scheduler interrupt again.

Resetting all interrupts might be actually acceptable behavior (probably most users call esp_hal::init() very early), but the PR#2091 decription only mentions an API rename, and the new behavior wasn't mentioned in the changelog.

(I'm slightly grumpy because we had to track this down but of course please don't read any blame. 😉).

To Reproduce

1. ...
2. ...

Expected behavior

Environment

• Target device: any xtensa
• Crate name and version: esp-hal 0.21.1

[bugadani]

Thanks, this will be fixed by #2377