From 5ce7f3f46f6b4a01dd2e2e7230351087034d0d1b Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Tue, 27 Oct 2020 23:54:41 +0100 Subject: [PATCH 1/3] net: neterion: s2io: Replace in_interrupt() for context detection wait_for_cmd_complete() uses in_interrupt() to detect whether it is safe to sleep or not. The usage of in_interrupt() in drivers is phased out and Linus clearly requested that code which changes behaviour depending on context should either be seperated or the context be conveyed in an argument passed by the caller, which usually knows the context. in_interrupt() also is only partially correct because it fails to chose the correct code path when just preemption or interrupts are disabled. Add an argument 'may_block' to both functions and adjust the callers to pass the context information. The following call chains which end up invoking wait_for_cmd_complete() were analyzed to be safe to sleep: s2io_card_up() s2io_set_multicast() init_nic() init_tti() s2io_close() do_s2io_delete_unicast_mc() do_s2io_add_mac() s2io_set_mac_addr() do_s2io_prog_unicast() do_s2io_add_mac() s2io_reset() do_s2io_restore_unicast_mc() do_s2io_add_mc() do_s2io_add_mac() s2io_open() do_s2io_prog_unicast() do_s2io_add_mac() The following call chains which end up invoking wait_for_cmd_complete() were analyzed to be safe to sleep: __dev_set_rx_mode() s2io_set_multicast() s2io_txpic_intr_handle() s2io_link() init_tti() Add a may_sleep argument to wait_for_cmd_complete(), s2io_set_multicast() and init_tti() and hand the context information in from the call sites. Signed-off-by: Sebastian Andrzej Siewior Cc: Jon Mason Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/neterion/s2io.c | 41 ++++++++++++++++------------ drivers/net/ethernet/neterion/s2io.h | 4 +-- 2 files changed, 25 insertions(+), 20 deletions(-) diff --git a/drivers/net/ethernet/neterion/s2io.c b/drivers/net/ethernet/neterion/s2io.c index d13d92bf7447..8f2f091bce89 100644 --- a/drivers/net/ethernet/neterion/s2io.c +++ b/drivers/net/ethernet/neterion/s2io.c @@ -1106,7 +1106,7 @@ static int s2io_print_pci_mode(struct s2io_nic *nic) * '-1' on failure */ -static int init_tti(struct s2io_nic *nic, int link) +static int init_tti(struct s2io_nic *nic, int link, bool may_sleep) { struct XENA_dev_config __iomem *bar0 = nic->bar0; register u64 val64 = 0; @@ -1166,7 +1166,7 @@ static int init_tti(struct s2io_nic *nic, int link) if (wait_for_cmd_complete(&bar0->tti_command_mem, TTI_CMD_MEM_STROBE_NEW_CMD, - S2IO_BIT_RESET) != SUCCESS) + S2IO_BIT_RESET, may_sleep) != SUCCESS) return FAILURE; } @@ -1659,7 +1659,7 @@ static int init_nic(struct s2io_nic *nic) */ /* Initialize TTI */ - if (SUCCESS != init_tti(nic, nic->last_link_state)) + if (SUCCESS != init_tti(nic, nic->last_link_state, true)) return -ENODEV; /* RTI Initialization */ @@ -3331,7 +3331,7 @@ static void s2io_updt_xpak_counter(struct net_device *dev) */ static int wait_for_cmd_complete(void __iomem *addr, u64 busy_bit, - int bit_state) + int bit_state, bool may_sleep) { int ret = FAILURE, cnt = 0, delay = 1; u64 val64; @@ -3353,7 +3353,7 @@ static int wait_for_cmd_complete(void __iomem *addr, u64 busy_bit, } } - if (in_interrupt()) + if (!may_sleep) mdelay(delay); else msleep(delay); @@ -4877,8 +4877,7 @@ static struct net_device_stats *s2io_get_stats(struct net_device *dev) * Return value: * void. */ - -static void s2io_set_multicast(struct net_device *dev) +static void s2io_set_multicast(struct net_device *dev, bool may_sleep) { int i, j, prev_cnt; struct netdev_hw_addr *ha; @@ -4903,7 +4902,7 @@ static void s2io_set_multicast(struct net_device *dev) /* Wait till command completes */ wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem, RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, - S2IO_BIT_RESET); + S2IO_BIT_RESET, may_sleep); sp->m_cast_flg = 1; sp->all_multi_pos = config->max_mc_addr - 1; @@ -4920,7 +4919,7 @@ static void s2io_set_multicast(struct net_device *dev) /* Wait till command completes */ wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem, RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, - S2IO_BIT_RESET); + S2IO_BIT_RESET, may_sleep); sp->m_cast_flg = 0; sp->all_multi_pos = 0; @@ -5000,7 +4999,7 @@ static void s2io_set_multicast(struct net_device *dev) /* Wait for command completes */ if (wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem, RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, - S2IO_BIT_RESET)) { + S2IO_BIT_RESET, may_sleep)) { DBG_PRINT(ERR_DBG, "%s: Adding Multicasts failed\n", dev->name); @@ -5030,7 +5029,7 @@ static void s2io_set_multicast(struct net_device *dev) /* Wait for command completes */ if (wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem, RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, - S2IO_BIT_RESET)) { + S2IO_BIT_RESET, may_sleep)) { DBG_PRINT(ERR_DBG, "%s: Adding Multicasts failed\n", dev->name); @@ -5041,6 +5040,12 @@ static void s2io_set_multicast(struct net_device *dev) } } +/* NDO wrapper for s2io_set_multicast */ +static void s2io_ndo_set_multicast(struct net_device *dev) +{ + s2io_set_multicast(dev, false); +} + /* read from CAM unicast & multicast addresses and store it in * def_mac_addr structure */ @@ -5127,7 +5132,7 @@ static int do_s2io_add_mac(struct s2io_nic *sp, u64 addr, int off) /* Wait till command completes */ if (wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem, RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, - S2IO_BIT_RESET)) { + S2IO_BIT_RESET, true)) { DBG_PRINT(INFO_DBG, "do_s2io_add_mac failed\n"); return FAILURE; } @@ -5171,7 +5176,7 @@ static u64 do_s2io_read_unicast_mc(struct s2io_nic *sp, int offset) /* Wait till command completes */ if (wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem, RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, - S2IO_BIT_RESET)) { + S2IO_BIT_RESET, true)) { DBG_PRINT(INFO_DBG, "do_s2io_read_unicast_mc failed\n"); return FAILURE; } @@ -7141,7 +7146,7 @@ static int s2io_card_up(struct s2io_nic *sp) } /* Setting its receive mode */ - s2io_set_multicast(dev); + s2io_set_multicast(dev, true); if (dev->features & NETIF_F_LRO) { /* Initialize max aggregatable pkts per session based on MTU */ @@ -7447,7 +7452,7 @@ static void s2io_link(struct s2io_nic *sp, int link) struct swStat *swstats = &sp->mac_control.stats_info->sw_stat; if (link != sp->last_link_state) { - init_tti(sp, link); + init_tti(sp, link, false); if (link == LINK_DOWN) { DBG_PRINT(ERR_DBG, "%s: Link down\n", dev->name); s2io_stop_all_tx_queue(sp); @@ -7604,7 +7609,7 @@ static int rts_ds_steer(struct s2io_nic *nic, u8 ds_codepoint, u8 ring) return wait_for_cmd_complete(&bar0->rts_ds_mem_ctrl, RTS_DS_MEM_CTRL_STROBE_CMD_BEING_EXECUTED, - S2IO_BIT_RESET); + S2IO_BIT_RESET, true); } static const struct net_device_ops s2io_netdev_ops = { @@ -7613,7 +7618,7 @@ static const struct net_device_ops s2io_netdev_ops = { .ndo_get_stats = s2io_get_stats, .ndo_start_xmit = s2io_xmit, .ndo_validate_addr = eth_validate_addr, - .ndo_set_rx_mode = s2io_set_multicast, + .ndo_set_rx_mode = s2io_ndo_set_multicast, .ndo_do_ioctl = s2io_ioctl, .ndo_set_mac_address = s2io_set_mac_addr, .ndo_change_mtu = s2io_change_mtu, @@ -7929,7 +7934,7 @@ s2io_init_nic(struct pci_dev *pdev, const struct pci_device_id *pre) writeq(val64, &bar0->rmac_addr_cmd_mem); wait_for_cmd_complete(&bar0->rmac_addr_cmd_mem, RMAC_ADDR_CMD_MEM_STROBE_CMD_EXECUTING, - S2IO_BIT_RESET); + S2IO_BIT_RESET, true); tmp64 = readq(&bar0->rmac_addr_data0_mem); mac_down = (u32)tmp64; mac_up = (u32) (tmp64 >> 32); diff --git a/drivers/net/ethernet/neterion/s2io.h b/drivers/net/ethernet/neterion/s2io.h index 6fa3159a977f..5a6032212c19 100644 --- a/drivers/net/ethernet/neterion/s2io.h +++ b/drivers/net/ethernet/neterion/s2io.h @@ -1066,7 +1066,7 @@ static void tx_intr_handler(struct fifo_info *fifo_data); static void s2io_handle_errors(void * dev_id); static void s2io_tx_watchdog(struct net_device *dev, unsigned int txqueue); -static void s2io_set_multicast(struct net_device *dev); +static void s2io_set_multicast(struct net_device *dev, bool may_sleep); static int rx_osm_handler(struct ring_info *ring_data, struct RxD_t * rxdp); static void s2io_link(struct s2io_nic * sp, int link); static void s2io_reset(struct s2io_nic * sp); @@ -1087,7 +1087,7 @@ static int s2io_set_swapper(struct s2io_nic * sp); static void s2io_card_down(struct s2io_nic *nic); static int s2io_card_up(struct s2io_nic *nic); static int wait_for_cmd_complete(void __iomem *addr, u64 busy_bit, - int bit_state); + int bit_state, bool may_sleep); static int s2io_add_isr(struct s2io_nic * sp); static void s2io_rem_isr(struct s2io_nic * sp); From dc5e8bfcd12eec8032be83b334896c103cdef1aa Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Tue, 27 Oct 2020 23:54:42 +0100 Subject: [PATCH 2/3] net: forcedeth: Replace context and lock check with a lockdep_assert() nv_update_stats() triggers a WARN_ON() when invoked from hard interrupt context because the locks in use are not hard interrupt safe. It also has an assert_spin_locked() which was the lock check before the lockdep era. Lockdep has way broader locking correctness checks and covers both issues, so replace the warning and the lock assert with lockdep_assert_held(). Signed-off-by: Sebastian Andrzej Siewior Cc: Rain River Cc: Zhu Yanjun Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/nvidia/forcedeth.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/drivers/net/ethernet/nvidia/forcedeth.c b/drivers/net/ethernet/nvidia/forcedeth.c index 2fc10a36afa4..7e85cf943be1 100644 --- a/drivers/net/ethernet/nvidia/forcedeth.c +++ b/drivers/net/ethernet/nvidia/forcedeth.c @@ -1666,11 +1666,7 @@ static void nv_update_stats(struct net_device *dev) struct fe_priv *np = netdev_priv(dev); u8 __iomem *base = get_hwbase(dev); - /* If it happens that this is run in top-half context, then - * replace the spin_lock of hwstats_lock with - * spin_lock_irqsave() in calling functions. */ - WARN_ONCE(in_irq(), "forcedeth: estats spin_lock(_bh) from top-half"); - assert_spin_locked(&np->hwstats_lock); + lockdep_assert_held(&np->hwstats_lock); /* query hardware */ np->estats.tx_bytes += readl(base + NvRegTxCnt); From beca92820dc4fd6f2413420c9828b586a1e77df5 Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Tue, 27 Oct 2020 23:54:44 +0100 Subject: [PATCH 3/3] net: tlan: Replace in_irq() usage The driver uses in_irq() to determine if the tlan_priv::lock has to be acquired in tlan_mii_read_reg() and tlan_mii_write_reg(). The interrupt handler acquires the lock outside of these functions so the in_irq() check is meant to prevent a lock recursion deadlock. But this check is incorrect when interrupt force threading is enabled because then the handler runs in thread context and in_irq() correctly returns false. The usage of in_*() in drivers is phased out and Linus clearly requested that code which changes behaviour depending on context should either be seperated or the context be conveyed in an argument passed by the caller, which usually knows the context. tlan_set_timer() has this conditional as well, but this function is only invoked from task context or the timer callback itself. So it always has to lock and the check can be removed. tlan_mii_read_reg(), tlan_mii_write_reg() and tlan_phy_print() are invoked from interrupt and other contexts. Split out the actual function body into helper variants which are called from interrupt context and make the original functions wrappers which acquire tlan_priv::lock unconditionally. Signed-off-by: Sebastian Andrzej Siewior Cc: Samuel Chessman Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/ti/tlan.c | 98 ++++++++++++++++++++-------------- 1 file changed, 57 insertions(+), 41 deletions(-) diff --git a/drivers/net/ethernet/ti/tlan.c b/drivers/net/ethernet/ti/tlan.c index 267c080ee084..0b2ce4bdc2c3 100644 --- a/drivers/net/ethernet/ti/tlan.c +++ b/drivers/net/ethernet/ti/tlan.c @@ -186,6 +186,7 @@ static void tlan_reset_adapter(struct net_device *); static void tlan_finish_reset(struct net_device *); static void tlan_set_mac(struct net_device *, int areg, char *mac); +static void __tlan_phy_print(struct net_device *); static void tlan_phy_print(struct net_device *); static void tlan_phy_detect(struct net_device *); static void tlan_phy_power_down(struct net_device *); @@ -201,9 +202,11 @@ static void tlan_phy_finish_auto_neg(struct net_device *); static int tlan_phy_dp83840a_check(struct net_device *); */ -static bool tlan_mii_read_reg(struct net_device *, u16, u16, u16 *); +static bool __tlan_mii_read_reg(struct net_device *, u16, u16, u16 *); +static void tlan_mii_read_reg(struct net_device *, u16, u16, u16 *); static void tlan_mii_send_data(u16, u32, unsigned); static void tlan_mii_sync(u16); +static void __tlan_mii_write_reg(struct net_device *, u16, u16, u16); static void tlan_mii_write_reg(struct net_device *, u16, u16, u16); static void tlan_ee_send_start(u16); @@ -242,23 +245,20 @@ static u32 tlan_handle_rx_eoc }; -static inline void +static void tlan_set_timer(struct net_device *dev, u32 ticks, u32 type) { struct tlan_priv *priv = netdev_priv(dev); unsigned long flags = 0; - if (!in_irq()) - spin_lock_irqsave(&priv->lock, flags); + spin_lock_irqsave(&priv->lock, flags); if (priv->timer.function != NULL && priv->timer_type != TLAN_TIMER_ACTIVITY) { - if (!in_irq()) - spin_unlock_irqrestore(&priv->lock, flags); + spin_unlock_irqrestore(&priv->lock, flags); return; } priv->timer.function = tlan_timer; - if (!in_irq()) - spin_unlock_irqrestore(&priv->lock, flags); + spin_unlock_irqrestore(&priv->lock, flags); priv->timer_set_at = jiffies; priv->timer_type = type; @@ -1703,22 +1703,22 @@ static u32 tlan_handle_status_check(struct net_device *dev, u16 host_int) dev->name, (unsigned) net_sts); } if ((net_sts & TLAN_NET_STS_MIRQ) && (priv->phy_num == 0)) { - tlan_mii_read_reg(dev, phy, TLAN_TLPHY_STS, &tlphy_sts); - tlan_mii_read_reg(dev, phy, TLAN_TLPHY_CTL, &tlphy_ctl); + __tlan_mii_read_reg(dev, phy, TLAN_TLPHY_STS, &tlphy_sts); + __tlan_mii_read_reg(dev, phy, TLAN_TLPHY_CTL, &tlphy_ctl); if (!(tlphy_sts & TLAN_TS_POLOK) && !(tlphy_ctl & TLAN_TC_SWAPOL)) { tlphy_ctl |= TLAN_TC_SWAPOL; - tlan_mii_write_reg(dev, phy, TLAN_TLPHY_CTL, - tlphy_ctl); + __tlan_mii_write_reg(dev, phy, TLAN_TLPHY_CTL, + tlphy_ctl); } else if ((tlphy_sts & TLAN_TS_POLOK) && (tlphy_ctl & TLAN_TC_SWAPOL)) { tlphy_ctl &= ~TLAN_TC_SWAPOL; - tlan_mii_write_reg(dev, phy, TLAN_TLPHY_CTL, - tlphy_ctl); + __tlan_mii_write_reg(dev, phy, TLAN_TLPHY_CTL, + tlphy_ctl); } if (debug) - tlan_phy_print(dev); + __tlan_phy_print(dev); } } @@ -2379,7 +2379,7 @@ ThunderLAN driver PHY layer routines /********************************************************************* - * tlan_phy_print + * __tlan_phy_print * * Returns: * Nothing @@ -2391,11 +2391,13 @@ ThunderLAN driver PHY layer routines * ********************************************************************/ -static void tlan_phy_print(struct net_device *dev) +static void __tlan_phy_print(struct net_device *dev) { struct tlan_priv *priv = netdev_priv(dev); u16 i, data0, data1, data2, data3, phy; + lockdep_assert_held(&priv->lock); + phy = priv->phy[priv->phy_num]; if (priv->adapter->flags & TLAN_ADAPTER_UNMANAGED_PHY) { @@ -2404,10 +2406,10 @@ static void tlan_phy_print(struct net_device *dev) netdev_info(dev, "PHY 0x%02x\n", phy); pr_info(" Off. +0 +1 +2 +3\n"); for (i = 0; i < 0x20; i += 4) { - tlan_mii_read_reg(dev, phy, i, &data0); - tlan_mii_read_reg(dev, phy, i + 1, &data1); - tlan_mii_read_reg(dev, phy, i + 2, &data2); - tlan_mii_read_reg(dev, phy, i + 3, &data3); + __tlan_mii_read_reg(dev, phy, i, &data0); + __tlan_mii_read_reg(dev, phy, i + 1, &data1); + __tlan_mii_read_reg(dev, phy, i + 2, &data2); + __tlan_mii_read_reg(dev, phy, i + 3, &data3); pr_info(" 0x%02x 0x%04hx 0x%04hx 0x%04hx 0x%04hx\n", i, data0, data1, data2, data3); } @@ -2417,7 +2419,15 @@ static void tlan_phy_print(struct net_device *dev) } +static void tlan_phy_print(struct net_device *dev) +{ + struct tlan_priv *priv = netdev_priv(dev); + unsigned long flags; + spin_lock_irqsave(&priv->lock, flags); + __tlan_phy_print(dev); + spin_unlock_irqrestore(&priv->lock, flags); +} /********************************************************************* @@ -2795,7 +2805,7 @@ these routines are based on the information in chap. 2 of the /*************************************************************** - * tlan_mii_read_reg + * __tlan_mii_read_reg * * Returns: * false if ack received ok @@ -2819,7 +2829,7 @@ these routines are based on the information in chap. 2 of the **************************************************************/ static bool -tlan_mii_read_reg(struct net_device *dev, u16 phy, u16 reg, u16 *val) +__tlan_mii_read_reg(struct net_device *dev, u16 phy, u16 reg, u16 *val) { u8 nack; u16 sio, tmp; @@ -2827,15 +2837,13 @@ tlan_mii_read_reg(struct net_device *dev, u16 phy, u16 reg, u16 *val) bool err; int minten; struct tlan_priv *priv = netdev_priv(dev); - unsigned long flags = 0; + + lockdep_assert_held(&priv->lock); err = false; outw(TLAN_NET_SIO, dev->base_addr + TLAN_DIO_ADR); sio = dev->base_addr + TLAN_DIO_DATA + TLAN_NET_SIO; - if (!in_irq()) - spin_lock_irqsave(&priv->lock, flags); - tlan_mii_sync(dev->base_addr); minten = tlan_get_bit(TLAN_NET_SIO_MINTEN, sio); @@ -2881,15 +2889,19 @@ tlan_mii_read_reg(struct net_device *dev, u16 phy, u16 reg, u16 *val) *val = tmp; - if (!in_irq()) - spin_unlock_irqrestore(&priv->lock, flags); - return err; - } +static void tlan_mii_read_reg(struct net_device *dev, u16 phy, u16 reg, + u16 *val) +{ + struct tlan_priv *priv = netdev_priv(dev); + unsigned long flags; - + spin_lock_irqsave(&priv->lock, flags); + __tlan_mii_read_reg(dev, phy, reg, val); + spin_unlock_irqrestore(&priv->lock, flags); +} /*************************************************************** * tlan_mii_send_data @@ -2971,7 +2983,7 @@ static void tlan_mii_sync(u16 base_port) /*************************************************************** - * tlan_mii_write_reg + * __tlan_mii_write_reg * * Returns: * Nothing @@ -2991,19 +3003,17 @@ static void tlan_mii_sync(u16 base_port) **************************************************************/ static void -tlan_mii_write_reg(struct net_device *dev, u16 phy, u16 reg, u16 val) +__tlan_mii_write_reg(struct net_device *dev, u16 phy, u16 reg, u16 val) { u16 sio; int minten; - unsigned long flags = 0; struct tlan_priv *priv = netdev_priv(dev); + lockdep_assert_held(&priv->lock); + outw(TLAN_NET_SIO, dev->base_addr + TLAN_DIO_ADR); sio = dev->base_addr + TLAN_DIO_DATA + TLAN_NET_SIO; - if (!in_irq()) - spin_lock_irqsave(&priv->lock, flags); - tlan_mii_sync(dev->base_addr); minten = tlan_get_bit(TLAN_NET_SIO_MINTEN, sio); @@ -3024,12 +3034,18 @@ tlan_mii_write_reg(struct net_device *dev, u16 phy, u16 reg, u16 val) if (minten) tlan_set_bit(TLAN_NET_SIO_MINTEN, sio); - if (!in_irq()) - spin_unlock_irqrestore(&priv->lock, flags); - } +static void +tlan_mii_write_reg(struct net_device *dev, u16 phy, u16 reg, u16 val) +{ + struct tlan_priv *priv = netdev_priv(dev); + unsigned long flags; + spin_lock_irqsave(&priv->lock, flags); + __tlan_mii_write_reg(dev, phy, reg, val); + spin_unlock_irqrestore(&priv->lock, flags); +} /*****************************************************************************