overloads.h is a header but has no include guard / #pragma once. This can lead to repeated processing and makes it inconsistent with the rest of the core headers (which all use #pragma once). Add an include guard or #pragma once at the top.

analogReadResolution() and analogWriteResolution() getters are declared unconditionally, but their definitions are compiled only when CONFIG_ADC and CONFIG_PWM || CONFIG_DAC are enabled respectively. This can cause undefined references for C++ sketches built without those Kconfig options. Wrap these declarations in matching #if guards (or provide always-available stubs that return a default).

The clockCyclesToMicroseconds(a) / microsecondsToClockCycles(a) macros don't parenthesize a, so expressions like x + 1 will be evaluated with unexpected precedence. Also clockCyclesPerMicrosecond() can evaluate to 0 on low-frequency targets due to integer division, which would make clockCyclesToMicroseconds() divide by zero. Parenthesize macro arguments and consider computing cycles/us with 64-bit math and rounding so it never returns 0.

analogWriteResolution(int bits) accepts any value, but later BIT(_analog_write_resolution) is used to compute maxInput. If bits <= 0 this yields maxInput == 0 and causes a division-by-zero in map64(). If bits is negative or too large, the shift inside BIT() is undefined. Clamp/validate bits to a safe range (e.g., 1..31) and consider storing it as an unsigned type.

Adding analogWrite(enum dacPins, int) introduces an overload that can make existing calls like analogWrite(3, value) ambiguous on C++ (an int can convert to both pin_size_t and an unscoped enum). Consider avoiding the overload (e.g., use pin_size_t for DAC pins too, or provide a differently named API / a scoped enum class with explicit conversion).

analogReference() updates the separate channel_cfg[] array, but analogRead() still calls adc_channel_setup() with &arduino_adc[idx].channel_cfg. That means changes made by analogReference() are never applied to actual reads. Use channel_cfg[idx] in adc_channel_setup() (and ensure the indexing matches), or update arduino_adc[idx].channel_cfg instead of maintaining a separate array.

analogReadResolution(int bits) stores bits directly, but analogRead() later shifts by (seq.resolution - read_resolution) / (read_resolution - seq.resolution). If bits is negative or very large, this can trigger undefined shifts and/or signed overflow in the return expression. Validate/clamp bits to a sensible range (e.g., 1..16 or 1..31) and do the scaling using an unsigned wider type before converting to int.

The scaling logic in analogRead() does shifts based on user-controlled read_resolution. Without clamping, buf >> (seq.resolution - read_resolution) / buf << (read_resolution - seq.resolution) can shift by a negative or excessively large amount, and the left-shift is done on an int after promotion which can overflow (UB). Clamp read_resolution to a safe range and perform the shift on a wider unsigned type before returning.