Date:   Wed, 19 Dec 2018 21:13:59 +0100
From:   Heiner Kallweit <hkallweit1@...il.com>
To:     Florian Fainelli <f.fainelli@...il.com>,
        Andrew Lunn <andrew@...n.ch>,
        David Miller <davem@...emloft.net>
Cc:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next 1/2] net: phy: make PHY_HALTED a transition state
 to PHY_READY

On 19.12.2018 19:32, Florian Fainelli wrote:
> On 12/18/18 10:53 PM, Heiner Kallweit wrote:
>> PHY_HALTED and PHY_READY both are non-started states and quite similar.
>> Major difference is that phy_start() changes from PHY_HALTED to
>> PHY_RESUMING which doesn't reconfigure aneg (what PHY_UP does).
>>
>> There's no guarantee that PHY registers are completely untouched when
>> waking up from power-down, e.g. after system suspend. Therefore it's
>> safer to reconfigure aneg also when starting from PHY_HALTED. This can
>> be achieved and state machine made simpler by making PHY_HALTED going
>> to PHY_READY after having stopped everything. Then the only way up is
>> over PHY_UP. Also let's warn in phy_start() if it's called from a
>> state other than PHY_READY.
>> As part of the change PHY_HALTED is renamed to PHY_HALT to reflect that
>> it is a transition state.
> 
> Sorry for being uber nitpicky here, but a state machine is supposed to
> contain.. state names, PHY_HALT is more of an action.
> 
Right, maybe PHY_HALTING would be better? Because in this state we
bring the link down and then change to PHY_READY.

> The PHY library is not particularly optimized at the moment to avoid
> disrupting the link when there is not a need to, so if somehow the
> register contents are lost because of a low power mode that the PHY has
> entered, the PHY driver's resume function is responsible for bringing
> things back online.
> 
Yes, we could do things like reconfiguring aneg in the resume callback.
However to me this seems to be more complex and not as easy as just
going the path PHY_READY -> PHY_UP -> reconfig -> PHY_RUNNING.
Why should we treat a phy_start() after a phy_stop() different than the
first phy_start() after connecting the PHY?
In my perspective there's no good justification for having separate
states PHY_READY and PHY_HALTED.

Only small overhead of my proposal is that a phy_start after a phy_stop()
checks the PHY aneg registers whether everything is as expected.
So far a phy_start() after a phy_stop() doesn't care and just checks
whether the link is up.

I know, I removed half of the state machine already and one may wonder
whether all this code was actually redundant. But well, it still seems
to work.

>>
>> Signed-off-by: Heiner Kallweit <hkallweit1@...il.com>
>> ---
>>  drivers/net/phy/phy.c | 41 +++++++++++++++++------------------------
>>  include/linux/phy.h   | 16 ++++++++--------
>>  2 files changed, 25 insertions(+), 32 deletions(-)
>>
>> diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c
>> index d33e7b3ca..2a69d947e 100644
>> --- a/drivers/net/phy/phy.c
>> +++ b/drivers/net/phy/phy.c
>> @@ -47,12 +47,12 @@ static const char *phy_state_to_str(enum phy_state st)
>>  	switch (st) {
>>  	PHY_STATE_STR(DOWN)
>>  	PHY_STATE_STR(READY)
>> +	PHY_STATE_STR(HALT)
>>  	PHY_STATE_STR(UP)
>>  	PHY_STATE_STR(RUNNING)
>>  	PHY_STATE_STR(NOLINK)
>>  	PHY_STATE_STR(FORCING)
>>  	PHY_STATE_STR(CHANGELINK)
>> -	PHY_STATE_STR(HALTED)
>>  	PHY_STATE_STR(RESUMING)
>>  	}
>>  
>> @@ -733,7 +733,7 @@ static void phy_error(struct phy_device *phydev)
>>  	WARN_ON(1);
>>  
>>  	mutex_lock(&phydev->lock);
>> -	phydev->state = PHY_HALTED;
>> +	phydev->state = PHY_HALT;
>>  	mutex_unlock(&phydev->lock);
>>  
>>  	phy_trigger_machine(phydev);
>> @@ -859,16 +859,11 @@ void phy_stop(struct phy_device *phydev)
>>  	if (phy_interrupt_is_valid(phydev))
>>  		phy_disable_interrupts(phydev);
>>  
>> -	phydev->state = PHY_HALTED;
>> +	phydev->state = PHY_HALT;
>>  
>>  	mutex_unlock(&phydev->lock);
>>  
>>  	phy_state_machine(&phydev->state_queue.work);
>> -
>> -	/* Cannot call flush_scheduled_work() here as desired because
>> -	 * of rtnl_lock(), but PHY_HALTED shall guarantee irq handler
>> -	 * will not reenable interrupts.
>> -	 */
>>  }
>>  EXPORT_SYMBOL(phy_stop);
>>  
>> @@ -888,29 +883,26 @@ void phy_start(struct phy_device *phydev)
>>  
>>  	mutex_lock(&phydev->lock);
>>  
>> -	switch (phydev->state) {
>> -	case PHY_READY:
>> -		phydev->state = PHY_UP;
>> -		break;
>> -	case PHY_HALTED:
>> +	if (phydev->state == PHY_READY) {
>>  		/* if phy was suspended, bring the physical link up again */
>>  		__phy_resume(phydev);
>>  
>>  		/* make sure interrupts are re-enabled for the PHY */
>>  		if (phy_interrupt_is_valid(phydev)) {
>>  			err = phy_enable_interrupts(phydev);
>> -			if (err < 0)
>> -				break;
>> +			if (err < 0) {
>> +				WARN_ON(1);
>> +				goto out;
>> +			}
>>  		}
>> -
>> -		phydev->state = PHY_RESUMING;
>> -		break;
>> -	default:
>> -		break;
>> +		phydev->state = PHY_UP;
>> +		phy_trigger_machine(phydev);
>> +	} else {
>> +		WARN(1, "called from state %s\n",
>> +		     phy_state_to_str(phydev->state));
>>  	}
>> +out:
>>  	mutex_unlock(&phydev->lock);
>> -
>> -	phy_trigger_machine(phydev);
>>  }
>>  EXPORT_SYMBOL(phy_start);
>>  
>> @@ -962,12 +954,13 @@ void phy_state_machine(struct work_struct *work)
>>  			phy_link_down(phydev, false);
>>  		}
>>  		break;
>> -	case PHY_HALTED:
>> +	case PHY_HALT:
>>  		if (phydev->link) {
>>  			phydev->link = 0;
>>  			phy_link_down(phydev, true);
>>  			do_suspend = true;
>>  		}
>> +		phydev->state = PHY_READY;
>>  		break;
>>  	}
>>  
>> @@ -990,7 +983,7 @@ void phy_state_machine(struct work_struct *work)
>>  	 * PHY, if PHY_IGNORE_INTERRUPT is set, then we will be moving
>>  	 * between states from phy_mac_interrupt().
>>  	 *
>> -	 * In state PHY_HALTED the PHY gets suspended, so rescheduling the
>> +	 * In state PHY_HALT the PHY gets suspended, so rescheduling the
>>  	 * state machine would be pointless and possibly error prone when
>>  	 * called from phy_disconnect() synchronously.
>>  	 */
>> diff --git a/include/linux/phy.h b/include/linux/phy.h
>> index da039f211..21e553f51 100644
>> --- a/include/linux/phy.h
>> +++ b/include/linux/phy.h
>> @@ -288,38 +288,38 @@ struct phy_device *mdiobus_scan(struct mii_bus *bus, int addr);
>>   *
>>   * NOLINK: PHY is up, but not currently plugged in.
>>   * - irq or timer will set RUNNING if link comes back
>> - * - phy_stop moves to HALTED
>> + * - phy_stop moves to HALT
>>   *
>>   * FORCING: PHY is being configured with forced settings
>>   * - if link is up, move to RUNNING
>>   * - If link is down, we drop to the next highest setting, and
>>   *   retry (FORCING) after a timeout
>> - * - phy_stop moves to HALTED
>> + * - phy_stop moves to HALT
>>   *
>>   * RUNNING: PHY is currently up, running, and possibly sending
>>   * and/or receiving packets
>>   * - irq or timer will set NOLINK if link goes down
>> - * - phy_stop moves to HALTED
>> + * - phy_stop moves to HALT
>>   *
>>   * CHANGELINK: PHY experienced a change in link state
>>   * - timer moves to RUNNING if link
>>   * - timer moves to NOLINK if the link is down
>> - * - phy_stop moves to HALTED
>> + * - phy_stop moves to HALT
>>   *
>> - * HALTED: PHY is up, but no polling or interrupts are done. Or
>> + * HALT: PHY is up, but no polling or interrupts are done. Or
>>   * PHY is in an error state.
>>   *
>> - * - phy_start moves to RESUMING
>> + * - moves to READY
>>   *
>>   * RESUMING: PHY was halted, but now wants to run again.
>>   * - If we are forcing, or aneg is done, timer moves to RUNNING
>>   * - If aneg is not done, timer moves to AN
>> - * - phy_stop moves to HALTED
>> + * - phy_stop moves to HALT
>>   */
>>  enum phy_state {
>>  	PHY_DOWN = 0,
>>  	PHY_READY,
>> -	PHY_HALTED,
>> +	PHY_HALT,
>>  	PHY_UP,
>>  	PHY_RUNNING,
>>  	PHY_NOLINK,
>>
> 
> 

Powered by blists - more mailing lists