Re: [PATCH net-next 2/2] igb: offer a PTP Hardware Clock instead of the timecompare method

[Eric Dumazet <eric.dumazet@gmail.com>]

Le mardi 13 décembre 2011 à 04:00 +0100, Richard Cochran a écrit :
> This commit removes the legacy timecompare code from the igb driver and
> offers a tunable PHC instead.
> 
> Signed-off-by: Richard Cochran <richardcochran@gmail.com>
> ---

> +void igb_systim_to_hwtstamp(struct igb_adapter *adapter,
> +			    struct skb_shared_hwtstamps *hwtstamps,
> +			    u64 systim)
> +{
> +	u64 ns;
> +	unsigned long flags;
> +	unsigned int shift;
> +	int msb_set;
> +
> +	switch (adapter->hw.mac.type) {
> +	case e1000_i350:
> +	case e1000_82580:
> +		shift = OFL_SHIFT_82580;
> +		msb_set = (systim >> 32) & SYSTIMH_MSB_82580;
> +		break;
> +	case e1000_82576:
> +		shift = OFL_SHIFT_82576;
> +		msb_set = (systim >> 32) & SYSTIMH_MSB_82576;
> +		break;
> +	default:
> +		return;
> +	}
> +
> +	spin_lock_irqsave(&adapter->tmreg_lock, flags);
> +
> +	ns = igb_overflow_get(adapter, systim, msb_set, shift);
> +
> +	spin_unlock_irqrestore(&adapter->tmreg_lock, flags);
> +
> +	memset(hwtstamps, 0, sizeof(*hwtstamps));
> +	hwtstamps->hwtstamp = ns_to_ktime(ns);
> +}

Adding a (shared) spinlock on a multiqueue device is source of extra
delay (because of extra cache line trafic), I guess.

It seems current code doesnt need a spinlock, maybe it was a bug ?