Re: [PATCH net-next v3 1/9] net: sunhme: Just restart autonegotiation if we can't bring the link up

[Simon Horman]

On Mon, Mar 13, 2023 at 08:36:05PM -0400, Sean Anderson wrote:
> If we've tried regular autonegotiation and forcing the link mode, just
> restart autonegotiation instead of reinitializing the whole NIC.
> 
> Signed-off-by: Sean Anderson <seanga2@xxxxxxxxx>

...

> @@ -606,6 +604,124 @@ static int is_lucent_phy(struct happy_meal *hp)
>  	return ret;
>  }
>  
> +/* hp->happy_lock must be held */
> +static void
> +happy_meal_begin_auto_negotiation(struct happy_meal *hp,
> +				  void __iomem *tregs,
> +				  const struct ethtool_link_ksettings *ep)
> +{
> +	int timeout;
> +
> +	/* Read all of the registers we are interested in now. */
> +	hp->sw_bmsr      = happy_meal_tcvr_read(hp, tregs, MII_BMSR);
> +	hp->sw_bmcr      = happy_meal_tcvr_read(hp, tregs, MII_BMCR);
> +	hp->sw_physid1   = happy_meal_tcvr_read(hp, tregs, MII_PHYSID1);
> +	hp->sw_physid2   = happy_meal_tcvr_read(hp, tregs, MII_PHYSID2);
> +
> +	/* XXX Check BMSR_ANEGCAPABLE, should not be necessary though. */
> +
> +	hp->sw_advertise = happy_meal_tcvr_read(hp, tregs, MII_ADVERTISE);
> +	if (!ep || ep->base.autoneg == AUTONEG_ENABLE) {
> +		/* Advertise everything we can support. */
> +		if (hp->sw_bmsr & BMSR_10HALF)
> +			hp->sw_advertise |= (ADVERTISE_10HALF);
> +		else
> +			hp->sw_advertise &= ~(ADVERTISE_10HALF);
> +
> +		if (hp->sw_bmsr & BMSR_10FULL)
> +			hp->sw_advertise |= (ADVERTISE_10FULL);
> +		else
> +			hp->sw_advertise &= ~(ADVERTISE_10FULL);
> +		if (hp->sw_bmsr & BMSR_100HALF)
> +			hp->sw_advertise |= (ADVERTISE_100HALF);
> +		else
> +			hp->sw_advertise &= ~(ADVERTISE_100HALF);
> +		if (hp->sw_bmsr & BMSR_100FULL)
> +			hp->sw_advertise |= (ADVERTISE_100FULL);
> +		else
> +			hp->sw_advertise &= ~(ADVERTISE_100FULL);
> +		happy_meal_tcvr_write(hp, tregs, MII_ADVERTISE, hp->sw_advertise);
> +
> +		/* XXX Currently no Happy Meal cards I know off support 100BaseT4,
> +		 * XXX and this is because the DP83840 does not support it, changes
> +		 * XXX would need to be made to the tx/rx logic in the driver as well
> +		 * XXX so I completely skip checking for it in the BMSR for now.
> +		 */
> +
> +		ASD("Advertising [ %s%s%s%s]\n",
> +		    hp->sw_advertise & ADVERTISE_10HALF ? "10H " : "",
> +		    hp->sw_advertise & ADVERTISE_10FULL ? "10F " : "",
> +		    hp->sw_advertise & ADVERTISE_100HALF ? "100H " : "",
> +		    hp->sw_advertise & ADVERTISE_100FULL ? "100F " : "");
> +
> +		/* Enable Auto-Negotiation, this is usually on already... */
> +		hp->sw_bmcr |= BMCR_ANENABLE;
> +		happy_meal_tcvr_write(hp, tregs, MII_BMCR, hp->sw_bmcr);
> +
> +		/* Restart it to make sure it is going. */
> +		hp->sw_bmcr |= BMCR_ANRESTART;
> +		happy_meal_tcvr_write(hp, tregs, MII_BMCR, hp->sw_bmcr);
> +
> +		/* BMCR_ANRESTART self clears when the process has begun. */
> +
> +		timeout = 64;  /* More than enough. */
> +		while (--timeout) {
> +			hp->sw_bmcr = happy_meal_tcvr_read(hp, tregs, MII_BMCR);
> +			if (!(hp->sw_bmcr & BMCR_ANRESTART))
> +				break; /* got it. */
> +			udelay(10);

nit: Checkpatch tells me that usleep_range() is preferred over udelay().
     Perhaps it would be worth looking into that for a follow-up patch.

> +		}
> +		if (!timeout) {
> +			netdev_err(hp->dev,
> +				   "Happy Meal would not start auto negotiation BMCR=0x%04x\n",
> +				   hp->sw_bmcr);
> +			netdev_notice(hp->dev,
> +				      "Performing force link detection.\n");
> +			goto force_link;
> +		} else {
> +			hp->timer_state = arbwait;
> +		}
> +	} else {
> +force_link:
> +		/* Force the link up, trying first a particular mode.
> +		 * Either we are here at the request of ethtool or
> +		 * because the Happy Meal would not start to autoneg.
> +		 */
> +
> +		/* Disable auto-negotiation in BMCR, enable the duplex and
> +		 * speed setting, init the timer state machine, and fire it off.
> +		 */
> +		if (!ep || ep->base.autoneg == AUTONEG_ENABLE) {
> +			hp->sw_bmcr = BMCR_SPEED100;
> +		} else {
> +			if (ep->base.speed == SPEED_100)
> +				hp->sw_bmcr = BMCR_SPEED100;
> +			else
> +				hp->sw_bmcr = 0;
> +			if (ep->base.duplex == DUPLEX_FULL)
> +				hp->sw_bmcr |= BMCR_FULLDPLX;
> +		}
> +		happy_meal_tcvr_write(hp, tregs, MII_BMCR, hp->sw_bmcr);
> +
> +		if (!is_lucent_phy(hp)) {
> +			/* OK, seems we need do disable the transceiver for the first
> +			 * tick to make sure we get an accurate link state at the
> +			 * second tick.
> +			 */
> +			hp->sw_csconfig = happy_meal_tcvr_read(hp, tregs,
> +							       DP83840_CSCONFIG);
> +			hp->sw_csconfig &= ~(CSCONFIG_TCVDISAB);
> +			happy_meal_tcvr_write(hp, tregs, DP83840_CSCONFIG,
> +					      hp->sw_csconfig);
> +		}
> +		hp->timer_state = ltrywait;
> +	}
> +
> +	hp->timer_ticks = 0;
> +	hp->happy_timer.expires = jiffies + (12 * HZ)/10;  /* 1.2 sec. */

nit: as a follow-up perhaps you could consider something like this.
     (* completely untested! * )

	hp->happy_timer.expires = jiffies + msecs_to_jiffies(1200);

> +	add_timer(&hp->happy_timer);
> +}
> +

...