Re: [PATCH v3 3/5] arm64: dts: qcom: nord-embedded: Raise i2c18 pad drive strength
From: Abel Vesa
Date: Wed Nov 11 2026 - 09:45:35 EST
On 26-11-11 15:31:19, Shawn Guo wrote:
> The i2c18 pins (gpio154/155) are configured for 2 mA. That is not
> enough on nord-ride-embedded, where i2c18 is the audio control bus
> carrying the TCA6408 expander, ADAU1979 ADC and PCM1681 DAC behind
> TXS0102 level shifters. Transfers intermittently fail with arbitration
> loss and unexpected start/stop conditions:
>
> geni_i2c 890000.i2c: Bus arbitration lost, clock line undriveable
> geni_i2c 890000.i2c: Bus proto err, noisy/unexpected start/stop
> adau1977 0-0031: probe with driver adau1977 failed with error -11
> reg-fixed-voltage regulator-5p0-dac: setup of GPIO (default) failed: -11
>
> The failure became much more frequent with commit 8cb5c17be3a2
> ("i2c: qcom-geni: Restore the source-clock selection policy"), which
> moves the i2c18 serial engine clock from 32 MHz to 19.2 MHz. That
> commit only changes how often the marginal signalling shows up:
> reverting it, or lowering the bus to 100 kHz, still leaves occasional
> failures in a codec unbind/bind loop. With 4 mA the loop passes
> reliably at either source clock rate.
>
> nord-rrd uses the same level shifters and has not shown the failure
> at 2 mA, but that appears to be margin rather than a guarantee. Since
> this audio I2C arrangement is the reference design that customer
> boards will likely follow, raise the default in the shared dtsi
> instead of overriding it per board.
>
> Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@xxxxxxxxxxxxxxxx>
> Signed-off-by: Shawn Guo <shengchao.guo@xxxxxxxxxxxxxxxx>
Reviewed-by: Abel Vesa <abel.vesa@xxxxxxxxxxxxxxxx>