Re: [PATCH net-next v7 3/5] ptp: clockmatrix: dco input-to-output delay is 20 FOD cycles + 8ns

[Horatiu Vultur]

The 05/01/2024 12:03, Min Li wrote:

Hi Min Li,

> 
> From: Min Li <min.li.xe@xxxxxxxxxxx>
> 
> Set dco input-to-output delay is 20 FOD cycles + 8ns

Can you explain why you are doing this change in the commit message?
Is this a fix or just improvement?

> 
> Signed-off-by: Min Li <min.li.xe@xxxxxxxxxxx>
> ---
>  drivers/ptp/ptp_clockmatrix.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/ptp/ptp_clockmatrix.c b/drivers/ptp/ptp_clockmatrix.c
> index d069b6e451ef..21f3a2c179f5 100644
> --- a/drivers/ptp/ptp_clockmatrix.c
> +++ b/drivers/ptp/ptp_clockmatrix.c
> @@ -2165,7 +2165,7 @@ static int configure_channel_pll(struct idtcm_channel *channel)
> 
>  /*
>   * Compensate for the PTP DCO input-to-output delay.
> - * This delay is 18 FOD cycles.
> + * This delay is 20 FOD cycles + 8ns.
>   */
>  static u32 idtcm_get_dco_delay(struct idtcm_channel *channel)
>  {
> @@ -2196,7 +2196,7 @@ static u32 idtcm_get_dco_delay(struct idtcm_channel *channel)
>         fodFreq = (u32)div_u64(m, n);
> 
>         if (fodFreq >= 500000000)
> -               return (u32)div_u64(18 * (u64)NSEC_PER_SEC, fodFreq);
> +               return (u32)div_u64(20 * (u64)NSEC_PER_SEC, fodFreq) + 8;
> 
>         return 0;
>  }
> --
> 2.39.2
> 
> 

-- 
/Horatiu