From d9f903f6af3dc6d13b21c0eca8f4a169aa70d80d Mon Sep 17 00:00:00 2001 From: Heiner Kallweit Date: Wed, 9 Jan 2019 20:34:56 +0100 Subject: net: phy: fix too strict check in phy_start_aneg When adding checks to detect wrong usage of the phylib API we added a check to phy_start_aneg() which is too strict. If the phylib state machine is in state PHY_HALTED we should allow reconfiguring and restarting aneg, and just don't touch the state. Fixes: 2b3e88ea6528 ("net: phy: improve phy state checking") Reported-by: Chris Wilson Signed-off-by: Heiner Kallweit Tested-by: Chris Wilson Reviewed-by: Florian Fainelli Signed-off-by: David S. Miller --- drivers/net/phy/phy.c | 19 +++++++------------ 1 file changed, 7 insertions(+), 12 deletions(-) (limited to 'drivers/net/phy/phy.c') diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c index d33e7b3caf03..189cd2048c3a 100644 --- a/drivers/net/phy/phy.c +++ b/drivers/net/phy/phy.c @@ -543,13 +543,6 @@ int phy_start_aneg(struct phy_device *phydev) mutex_lock(&phydev->lock); - if (!__phy_is_started(phydev)) { - WARN(1, "called from state %s\n", - phy_state_to_str(phydev->state)); - err = -EBUSY; - goto out_unlock; - } - if (AUTONEG_DISABLE == phydev->autoneg) phy_sanitize_settings(phydev); @@ -560,11 +553,13 @@ int phy_start_aneg(struct phy_device *phydev) if (err < 0) goto out_unlock; - if (phydev->autoneg == AUTONEG_ENABLE) { - err = phy_check_link_status(phydev); - } else { - phydev->state = PHY_FORCING; - phydev->link_timeout = PHY_FORCE_TIMEOUT; + if (__phy_is_started(phydev)) { + if (phydev->autoneg == AUTONEG_ENABLE) { + err = phy_check_link_status(phydev); + } else { + phydev->state = PHY_FORCING; + phydev->link_timeout = PHY_FORCE_TIMEOUT; + } } out_unlock: -- cgit v1.2.3 From b79555d5d8d32643e9d7193341dcaff13bf9ffcd Mon Sep 17 00:00:00 2001 From: Heiner Kallweit Date: Tue, 12 Feb 2019 19:56:15 +0100 Subject: net: phy: fix interrupt handling in non-started states phylib enables interrupts before phy_start() has been called, and if we receive an interrupt in a non-started state, the interrupt handler returns IRQ_NONE. This causes problems with at least one Marvell chip as reported by Andrew. Fix this by handling interrupts the same as in phy_mac_interrupt(), basically always running the phylib state machine. It knows when it has to do something and when not. This change allows to handle interrupts gracefully even if they occur in a non-started state. Fixes: 2b3e88ea6528 ("net: phy: improve phy state checking") Reported-by: Andrew Lunn Signed-off-by: Heiner Kallweit Reviewed-by: Andrew Lunn Reviewed-by: Florian Fainelli Signed-off-by: David S. Miller --- drivers/net/phy/phy.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'drivers/net/phy/phy.c') diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c index 189cd2048c3a..ca5e0c0f018c 100644 --- a/drivers/net/phy/phy.c +++ b/drivers/net/phy/phy.c @@ -762,9 +762,6 @@ static irqreturn_t phy_interrupt(int irq, void *phy_dat) { struct phy_device *phydev = phy_dat; - if (!phy_is_started(phydev)) - return IRQ_NONE; /* It can't be ours. */ - if (phydev->drv->did_interrupt && !phydev->drv->did_interrupt(phydev)) return IRQ_NONE; -- cgit v1.2.3 From a2fc9d7e36f6d484d9be4a0a204400aaf6059544 Mon Sep 17 00:00:00 2001 From: Heiner Kallweit Date: Wed, 13 Feb 2019 20:11:40 +0100 Subject: net: phy: don't use locking in phy_is_started Russell suggested to remove the locking from phy_is_started() because the read is atomic anyway and actually the locking may be more misleading. Fixes: 2b3e88ea6528 ("net: phy: improve phy state checking") Suggested-by: Russell King - ARM Linux admin Signed-off-by: Heiner Kallweit Reviewed-by: Florian Fainelli Signed-off-by: David S. Miller --- drivers/net/phy/phy.c | 11 +++++------ include/linux/phy.h | 15 +-------------- 2 files changed, 6 insertions(+), 20 deletions(-) (limited to 'drivers/net/phy/phy.c') diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c index ca5e0c0f018c..602816d70281 100644 --- a/drivers/net/phy/phy.c +++ b/drivers/net/phy/phy.c @@ -553,7 +553,7 @@ int phy_start_aneg(struct phy_device *phydev) if (err < 0) goto out_unlock; - if (__phy_is_started(phydev)) { + if (phy_is_started(phydev)) { if (phydev->autoneg == AUTONEG_ENABLE) { err = phy_check_link_status(phydev); } else { @@ -709,7 +709,7 @@ void phy_stop_machine(struct phy_device *phydev) cancel_delayed_work_sync(&phydev->state_queue); mutex_lock(&phydev->lock); - if (__phy_is_started(phydev)) + if (phy_is_started(phydev)) phydev->state = PHY_UP; mutex_unlock(&phydev->lock); } @@ -839,15 +839,14 @@ EXPORT_SYMBOL(phy_stop_interrupts); */ void phy_stop(struct phy_device *phydev) { - mutex_lock(&phydev->lock); - - if (!__phy_is_started(phydev)) { + if (!phy_is_started(phydev)) { WARN(1, "called from state %s\n", phy_state_to_str(phydev->state)); - mutex_unlock(&phydev->lock); return; } + mutex_lock(&phydev->lock); + if (phy_interrupt_is_valid(phydev)) phy_disable_interrupts(phydev); diff --git a/include/linux/phy.h b/include/linux/phy.h index ef20aeea10cc..127fcc9c3778 100644 --- a/include/linux/phy.h +++ b/include/linux/phy.h @@ -674,26 +674,13 @@ phy_lookup_setting(int speed, int duplex, const unsigned long *mask, size_t phy_speeds(unsigned int *speeds, size_t size, unsigned long *mask); -static inline bool __phy_is_started(struct phy_device *phydev) -{ - WARN_ON(!mutex_is_locked(&phydev->lock)); - - return phydev->state >= PHY_UP; -} - /** * phy_is_started - Convenience function to check whether PHY is started * @phydev: The phy_device struct */ static inline bool phy_is_started(struct phy_device *phydev) { - bool started; - - mutex_lock(&phydev->lock); - started = __phy_is_started(phydev); - mutex_unlock(&phydev->lock); - - return started; + return phydev->state >= PHY_UP; } void phy_resolve_aneg_linkmode(struct phy_device *phydev); -- cgit v1.2.3 From a20049071796691cf99eb6433968fc3c27632b95 Mon Sep 17 00:00:00 2001 From: Heiner Kallweit Date: Wed, 13 Feb 2019 20:12:54 +0100 Subject: net: phy: fix potential race in the phylib state machine Russell reported the following race in the phylib state machine (quoting from his mail): if (phy_polling_mode(phydev) && phy_is_started(phydev)) phy_queue_state_machine(phydev, PHY_STATE_TIME); state = PHY_UP thread 0 thread 1 phy_disconnect() +-phy_is_started() phy_is_started() | `-phy_stop() +-phydev->state = PHY_HALTED `-phy_stop_machine() `-cancel_delayed_work_sync() phy_queue_state_machine() `-mod_delayed_work() At this point, the phydev->state_queue() has been added back onto the system workqueue despite phy_stop_machine() having been called and cancel_delayed_work_sync() called on it. Fix this by protecting the complete operation in thread 0. Fixes: 2b3e88ea6528 ("net: phy: improve phy state checking") Reported-by: Russell King - ARM Linux admin Signed-off-by: Heiner Kallweit Reviewed-by: Florian Fainelli Signed-off-by: David S. Miller --- drivers/net/phy/phy.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/net/phy/phy.c') diff --git a/drivers/net/phy/phy.c b/drivers/net/phy/phy.c index 602816d70281..c5675df5fc6f 100644 --- a/drivers/net/phy/phy.c +++ b/drivers/net/phy/phy.c @@ -985,8 +985,10 @@ void phy_state_machine(struct work_struct *work) * state machine would be pointless and possibly error prone when * called from phy_disconnect() synchronously. */ + mutex_lock(&phydev->lock); if (phy_polling_mode(phydev) && phy_is_started(phydev)) phy_queue_state_machine(phydev, PHY_STATE_TIME); + mutex_unlock(&phydev->lock); } /** -- cgit v1.2.3