RE: [PATCH iwl-net 2/2] ice: skip the SW pin description on boards with generic DPLL pins

[Nowlin, Alexander]

> -----Original Message-----
> From: Petr Oros <poros@xxxxxxxxxx> 
> Sent: Thursday, September 17, 2026 12:03 PM
> To: netdev@xxxxxxxxxxxxxxx
> Cc: Oros, Petr <poros@xxxxxxxxxx>; Nguyen, Anthony L <anthony.l.nguyen@xxxxxxxxx>; Kitszel, Przemyslaw <przemyslaw.kitszel@xxxxxxxxx>; Andrew Lunn <andrew+netdev@xxxxxxx>; David S. Miller <davem@xxxxxxxxxxxxx>; Eric Dumazet <edumazet@xxxxxxxxxx>; Jakub Kicinski <kuba@xxxxxxxxxx>; Paolo Abeni <pabeni@xxxxxxxxxx>; Simon Horman <horms@xxxxxxxxxx>; Kubalewski, Arkadiusz <arkadiusz.kubalewski@xxxxxxxxx>; intel-wired-lan@xxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> Subject: [PATCH iwl-net 2/2] ice: skip the SW pin description on boards with generic DPLL pins
> 
> ice_dpll_init_info_sw_pins() describes the SMA and U.FL wrappers by pointing them at fixed positions of the input and output arrays, inputs
> 4 and 5 and outputs 0 and 1 on an E810-C SFP, and reads the ref-sync partner of the input it picked. The arrays are sized by the pin counts firmware reports. On a board that does not match the Intel reference layout the driver has already fallen back to generic pins, and when such a board reports fewer pins than the reference the reads run past the end of the arrays.
> 
> The wrappers are never registered on a generic board, but
> ice_dpll_init_pins() still calls ice_dpll_pin_ref_sync_register() on them, so a stray nonzero partner from the out of bounds read would pair NULL pins.
> 
> Describe the wrappers only when the static pin table is in use and keep their ref-sync registration under the same condition. The default programming of the PCA9575 clock mux stays on the generic path, a board that has the mux still needs its SMA connectors switched to inputs.
> 
> Fixes: 2dd5d03c77e2 ("ice: redesign dpll sma/u.fl pins control")
> Signed-off-by: Petr Oros <poros@xxxxxxxxxx>
> ---
>  drivers/net/ethernet/intel/ice/ice_dpll.c | 13 ++++++++-----
>  1 file changed, 8 insertions(+), 5 deletions(-)
 
Tested-by: Alexander Nowlin <alexander.nowlin@xxxxxxxxx>