tg3: Fix phylib locking strategy
Subsystems (2)
PATCH v2 1/1 tg3: Fix phylib locking strategy
O.K. Here is the latest version. Felix, can you verify your problem is solved with this patch?
---
Felix Radensky noted that chip resets were generating stack trace dumps. This is because the driver is attempting to acquire the mdio bus mutex while holding the tp->lock spinlock. The fix is to change the code such that every phy access takes the tp->lock spinlock instead.
--- drivers/net/tg3.c | 41 ++++++++++++++--------------------------- drivers/net/tg3.h | 1 - 2 files changed, 14 insertions(+), 28 deletions(-)
drivers/net/tg3.c +14 -27
@@ -902,11 +902,12 @@ static int tg3_mdio_read(struct mii_bus *bp, int mii_id, int reg)struct tg3 *tp = bp->priv;u32 val; -if (tp->tg3_flags3 &TG3_FLG3_MDIOBUS_PAUSED ) -return -EAGAIN ; +spin_lock_bh (&tp->lock);if (tg3_readphy (tp, reg, &val)) -return -EIO ; + val = -EIO ; + +spin_unlock_bh (&tp->lock);return val; }
@@ -914,14 +915,16 @@ static int tg3_mdio_read(struct mii_bus *bp, int mii_id, int reg)static int tg3_mdio_write (struct mii_bus *bp,int mii_id,int reg,u16 val) {struct tg3 *tp = bp->priv; +u32 ret =0 ; -if (tp->tg3_flags3 &TG3_FLG3_MDIOBUS_PAUSED ) -return -EAGAIN ; +spin_lock_bh (&tp->lock);if (tg3_writephy (tp, reg, val)) -return -EIO ; + ret = -EIO ; -return 0 ; +spin_unlock_bh (&tp->lock); + +return ret; }static int tg3_mdio_reset (struct mii_bus *bp)
@@ -1011,12 +1014,6 @@ static void tg3_mdio_config_5785(struct tg3 *tp)static void tg3_mdio_start (struct tg3 *tp) { -if (tp->tg3_flags3 &TG3_FLG3_MDIOBUS_INITED ) { -mutex_lock (&tp->mdio_bus->mdio_lock); - tp->tg3_flags3 &= ~TG3_FLG3_MDIOBUS_PAUSED ; -mutex_unlock (&tp->mdio_bus->mdio_lock); - } - tp->mi_mode &= ~MAC_MI_MODE_AUTO_POLL ;tw32_f (MAC_MI_MODE , tp->mi_mode);udelay (80 );
@@ -1041,15 +1038,6 @@ static void tg3_mdio_start(struct tg3 *tp)tg3_mdio_config_5785 (tp); } -static void tg3_mdio_stop (struct tg3 *tp) -{ -if (tp->tg3_flags3 &TG3_FLG3_MDIOBUS_INITED ) { -mutex_lock (&tp->mdio_bus->mdio_lock); - tp->tg3_flags3 |=TG3_FLG3_MDIOBUS_PAUSED ; -mutex_unlock (&tp->mdio_bus->mdio_lock); - } -} -static int tg3_mdio_init (struct tg3 *tp) {int i;
@@ -1141,7 +1129,6 @@ static void tg3_mdio_fini(struct tg3 *tp) tp->tg3_flags3 &= ~TG3_FLG3_MDIOBUS_INITED ;mdiobus_unregister (tp->mdio_bus);mdiobus_free (tp->mdio_bus); - tp->tg3_flags3 &= ~TG3_FLG3_MDIOBUS_PAUSED ; } }
@@ -1363,7 +1350,7 @@ static void tg3_adjust_link(struct net_device *dev)struct tg3 *tp =netdev_priv (dev);struct phy_device *phydev = tp->mdio_bus->phy_map[PHY_ADDR ]; -spin_lock (&tp->lock); +spin_lock_bh (&tp->lock); mac_mode = tp->mac_mode & ~(MAC_MODE_PORT_MODE_MASK |MAC_MODE_HALF_DUPLEX );
@@ -1431,7 +1418,7 @@ static void tg3_adjust_link(struct net_device *dev) tp->link_config.active_speed = phydev->speed; tp->link_config.active_duplex = phydev->duplex; -spin_unlock (&tp->lock); +spin_unlock_bh (&tp->lock);if (linkmesg)tg3_link_report (tp);
@@ -6392,8 +6379,6 @@ static int tg3_chip_reset(struct tg3 *tp)tg3_nvram_lock (tp); -tg3_mdio_stop (tp); -tg3_ape_lock (tp,TG3_APE_LOCK_GRC );/* No matching tg3_nvram_unlock() after this because
@@ -8698,6 +8683,8 @@ static int tg3_close(struct net_device *dev)del_timer_sync (&tp->timer); +tg3_phy_stop (tp); +tg3_full_lock (tp,1 );#if 0 tg3_dump_state (tp);
drivers/net/tg3.h +0 -1
@@ -2748,7 +2748,6 @@ struct tg3 {#define TG3_FLG3_5701_DMA_BUG 0x00000008#define TG3_FLG3_USE_PHYLIB 0x00000010#define TG3_FLG3_MDIOBUS_INITED 0x00000020 -#define TG3_FLG3_MDIOBUS_PAUSED 0x00000040#define TG3_FLG3_PHY_CONNECTED 0x00000080#define TG3_FLG3_RGMII_STD_IBND_DISABLE 0x00000100#define TG3_FLG3_RGMII_EXT_IBND_RX_EN 0x00000200
Signature · 1 line
1.6.4.4
Hi, Matt
Quoted 4 lines · Matt Carlson
O.K. Here is the latest version. Felix, can you verify your problem is solved with this patch?
Thanks for the patch. I don't have the board with Broadcom NIC at the moment. Hopefully I'll get it back in a couple of days and will test your patch.
Felix.
David Miller
maintainerFrom: "Matt Carlson" <[email protected]>
Date: Mon, 5 Oct 2009 20:55:29 -0700
Quoted 11 lines
O.K. Here is the latest version. Felix, can you verify your problem is solved with this patch?
---
Felix Radensky noted that chip resets were generating stack trace dumps. This is because the driver is attempting to acquire the mdio bus mutex while holding the
tp->lockspinlock. The fix is to change the code such that every phy access takes thetp->lockspinlock instead.Signed-off-by Matt Carlson <[email protected]>
I'm going to apply this now to net-2.6, let me know if there are any updates after testing.
Hi, Matt
Quoted 11 lines · Matt Carlson
O.K. Here is the latest version. Felix, can you verify your problem is solved with this patch?
---
Felix Radensky noted that chip resets were generating stack trace dumps. This is because the driver is attempting to acquire the mdio bus mutex while holding the
tp->lockspinlock. The fix is to change the code such that every phy access takes thetp->lockspinlock instead.
Sorry for a very long delay, I didn't have access to hardware until today. Your patch fixes initial problem I've reported, but I have a different problem now, maybe it's related. It is 100% reproducible by the following sequence
modprobe tg3
ifconfig eth2 up
ifconfig eth2 down
ifconfig eth2 up
As a result, link is not recognized after third invocation of ifconfig, although network cable is plugged in. Below are relevant kernel messages. Reloading tg3 driver fixes the problem.
tg3.c:v3.102 (September 1, 2009)
tg3 0002:05:00.0: PME# disabled
tg3 mdio bus: probed
eth2: Tigon3 [partno(BCM57760) rev 57780001] (PCI Express) MAC address 00:10:18:00:00:00
eth2: attached PHY driver [Broadcom BCM57780] (mii_bus:phy_addr=500:01)
eth2: RXcsums[1] LinkChgREG[0] MIirq[0] ASF[0] TSOcap[1]
eth2: dma_rwctrl[76180000] dma_mask[64-bit]
tg3 0002:05:00.0: PME# disabled
tg3: eth2: Link is down.
tg3: eth2: Link is up at 1000 Mbps, full duplex.
tg3: eth2: Flow control is on for TX and on for RX.
tg3: eth2: Link is down.
tg3 0002:05:00.0: PME# disabled
tg3: eth2: Link is down.
Thanks.
Felix.