is nRF54 able to use CTRL-AP to reset itself?

[Kelso]

I'm aware of https://devzone.nordicsemi.com/f/nordic-q-a/125252/reset-over-ctrl-ap/555253?focus=true

Is it possible for an nRF54 to safely use the CPU access to the AP registers to induce a pin reset on its own as an alternative to triggering a WDR?

I'm looking for software ways to cause the chip to reset to handle the edgecase of a poorly configured watchdog that is locked out.

TAMPC, WDR, and CTRL-AP resets appear to be sufficient to correct this based on the datasheet.  I would not be able to drive a pin reset externally in this situation.



The results from your AI support the idea that this connection exists and can be used, but caveats the safety question:

Based on the knowledge sources, yes — the nRF54L series does expose a CPU-side CTRL-AP RESET register (at peripheral address offset 0x520) that can be written directly by the CPU (not just by an external debugger).

The CPU-side RESET register supports the following values:

Value ID	Value	Description
NoReset	0	No reset is generated
SoftReset	1	Perform a device soft reset
HardReset	2	Perform a device hard reset
PinReset	4	Perform a device pin reset

[CPU-side RESET register]

So writing 4 (PinReset) to the RESET register at base address 0x50052000 (secure) or 0x40052000 (non-secure) + offset 0x520 should trigger a pin reset from software, without needing an external signal or a watchdog.

A few important caveats the knowledge sources do not fully address, so I want to be transparent about limitations:

• The knowledge sources confirm the register exists and is CPU-writable (W access), but do not explicitly describe any side effects, restrictions, or security/TrustZone gating that might prevent a non-secure or unprivileged context from writing it. You should verify whether TrustZone configuration restricts access to the secure (SCTRLAP) vs. non-secure (NS) instance.
• The instance table shows both a secure (0x50052000) and non-secure (0x40052000) base address, so access from non-secure code may be possible depending on your configuration. [Register overview]
• The knowledge sources do not confirm whether a CPU-initiated PinReset via this register is treated identically to a hardware pin reset in terms of reset domain behavior (e.g., whether it clears the same state as a true pin reset). You should cross-reference the full reset domain documentation in the datasheet to confirm it meets your requirements for clearing a locked watchdog configuration.

In summary: yes, writing PinReset (value 4) to the CPU-side CTRL-AP RESET register appears to be a valid software-initiated pin reset mechanism on the nRF54L15/L10/L05, and is a reasonable alternative to a WDT reset for your described edge case — but verify TrustZone access permissions and reset domain behavior in the full datasheet.

[Justin012]

Interesting use case. I'd be cautious about relying on a CPU-initiated CTRL-AP pin reset until Nordic confirms that it's supported for this purpose and has the same reset-domain effects as an actual pin reset. The register description alone doesn't necessarily guarantee that behavior in every security configuration. free online games

If the watchdog can be locked into a bad configuration, I'd also check whether the reset mechanism remains accessible under those conditions and whether any relevant state survives the reset. TrustZone permissions could be another factor.

Have you tested writing PinReset from firmware on the actual nRF54 target, or is this still based on the register documentation? A test confirming which reset sources are asserted and what state gets cleared would be useful. I'd be interested in Nordic's take on whether this is a supported recovery mechanism rather than just something that appears to work.