Re: [PATCH net-next 1/2] dt-bindings: net: dsa: microchip: add reset timing properties

[Rodolfo Giometti]

On Sun, Oct 04, 2026 at 09:27:04AM +0200, Krzysztof Kozlowski wrote:
> On Fri, Oct 02, 2026 at 03:55:41PM +0200, Rodolfo Giometti wrote:
> > How long a KSZ switch needs before it answers on its management bus
> > after the reset line is released depends on the board it sits on, and a
> > board that needs longer than the driver waits has no way to say so.
>
> Not completely. This is implied by compatible. Individual phys/mdios
> have their own delays but not the entire device.

I see. Honestly speaking, what I measured is the switch itself needing
longer after a long reset: 95-120ms after a 10ms pulse, ~160ms after 1s
held in reset, on one board, while a second board gets through the same
sequence in under 100ms. So I cannot tell whether the difference comes
from the board or from the part.

I will drop the binding patch, then. For v2 I am thinking of keeping the
fixed wait and then polling the chip ID until the switch answers, with
an upper bound in the driver, so that no property is needed: a generous
bound costs nothing to a switch that answers early.

If that is not the right direction, I'd be glad to hear it before posting.

Ciao,

Rodolfo

-- 
GNU/Linux Solutions                  e-mail: giometti@xxxxxxxxxxxx
Linux Device Driver                          giometti@xxxxxxxx
Embedded Systems                     phone:  +39 349 2432127
UNIX programming