On Wed, Oct 28, 2026 at 10:45:04AM +0100, Stefano Radaelli wrote: > From: Stefano Radaelli <[email protected]> > > The optional reset control is deasserted without a delay, and probe then > proceeds directly to the first I2C register access. Since reset-gpio > initializes its line asserted, this provides no guaranteed minimum reset > pulse width or recovery interval before communication starts. > > Table 66 of the ST33KTPM-IDevID datasheet specifies a minimum reset pulse > width of 1 ms and up to 50 ms from reset release before the first command. > Choose a 2 ms assertion interval to provide margin above the minimum and > wait 50 ms after deassertion, before initializing the I2C guard times. > > Associate these timings with the existing st,st33ktpm2xi2c compatible using > match data. Apply the delays only when a reset control is present. Preserve > the existing deassert-only sequence and the behavior of devices without > timing data or a reset control, without adding Device Tree properties.
"Pull the timings from the matched device entry. If they don't exist, fall back to the legacy behavior without delays. Update of_tis_i2c_match with the timings for st33ktpm2xi2c." I.e. the last paragraph has the language of a white paper. Br, Jarkko
