| Message ID | 20241106101717.981850-1-jiawenwu@trustnetic.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] net: txgbe: fix lost GPIO interrupt | expand |
On Wed, Nov 06, 2024 at 06:17:17PM +0800, Jiawen Wu wrote: > Sometimes when clearing interrupts for workaround in txgbe_up_complete(), > the GPIO interrupt is lost once due to the interrupt polarity is > consistent with the GPIO state. It causes the SFP state cannot be updated > immediately. That is, SFP driver can only get into the correct state if > the GPIO state changes again. If i remember correctly, the basic issue is that your hardware cannot do IRQ_TYPE_EDGE_BOTH? You try to fake it by configuring for IRQ_TYPE_EDGE_RISING when the GPIO is low, IRQ_TYPE_EDGE_FALLING when the GPIO is high. And then hope you can handle the interrupt fast enough you don't miss an edge? Maybe you should just accept the interrupt controller hardware is broken, and let the SFP driver poll the GPIO? Do you really need to know in less than a second the SFP has been ejected, or changed is LOS status? Andrew
diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h index b54bffda027b..bf4285db605e 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h @@ -1078,6 +1078,7 @@ struct wx { bool wol_hw_supported; bool ncsi_enabled; bool gpio_ctrl; + bool gpio_trigger; raw_spinlock_t gpio_lock; /* Tx fast path data */ diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c index 93180225a6f1..0e6129dcff8f 100644 --- a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c +++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c @@ -82,7 +82,6 @@ static void txgbe_up_complete(struct wx *wx) { struct net_device *netdev = wx->netdev; - txgbe_reinit_gpio_intr(wx); wx_control_hw(wx, true); wx_configure_vectors(wx); @@ -97,6 +96,9 @@ static void txgbe_up_complete(struct wx *wx) rd32(wx, WX_PX_IC(1)); rd32(wx, WX_PX_MISC_IC); txgbe_irq_enable(wx, true); + wx->gpio_trigger = true; + txgbe_reinit_gpio_intr(wx); + wx->gpio_trigger = false; /* enable transmits */ netif_tx_start_all_queues(netdev); @@ -169,6 +171,7 @@ void txgbe_down(struct wx *wx) txgbe_disable_device(wx); txgbe_reset(wx); phylink_stop(wx->phylink); + txgbe_reinit_gpio_intr(wx); wx_clean_all_tx_rings(wx); wx_clean_all_rx_rings(wx); diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_phy.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_phy.c index 67b61afdde96..361dcb362d42 100644 --- a/drivers/net/ethernet/wangxun/txgbe/txgbe_phy.c +++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_phy.c @@ -401,17 +401,21 @@ static void txgbe_gpio_irq_unmask(struct irq_data *d) static void txgbe_toggle_trigger(struct gpio_chip *gc, unsigned int offset) { struct wx *wx = gpiochip_get_data(gc); - u32 pol, val; + u32 pol_r, pol_w, val; - pol = rd32(wx, WX_GPIO_POLARITY); + pol_r = rd32(wx, WX_GPIO_POLARITY); val = rd32(wx, WX_GPIO_EXT); if (val & BIT(offset)) - pol &= ~BIT(offset); + pol_w = pol_r & ~BIT(offset); else - pol |= BIT(offset); + pol_w = pol_r | BIT(offset); - wr32(wx, WX_GPIO_POLARITY, pol); + wr32(wx, WX_GPIO_POLARITY, pol_w); + + /* manually trigger the lost inpterrupt */ + if (wx->gpio_trigger) + wr32(wx, WX_GPIO_POLARITY, pol_r); } static int txgbe_gpio_set_type(struct irq_data *d, unsigned int type) @@ -560,6 +564,7 @@ static int txgbe_gpio_init(struct txgbe *txgbe) return ret; txgbe->gpio = gc; + wx->gpio_trigger = false; return 0; }
Sometimes when clearing interrupts for workaround in txgbe_up_complete(), the GPIO interrupt is lost once due to the interrupt polarity is consistent with the GPIO state. It causes the SFP state cannot be updated immediately. That is, SFP driver can only get into the correct state if the GPIO state changes again. So mannually trigger the GPIO interrupt while clearing it in txgbe_reinit_gpio_intr(), causing SFP driver to update to the correct state. And also clear GPIO interrupts in txgbe_down() to ensure that the interrupt polarity is consistent with the GPIO state. Fixes: b4a2496c17ed ("net: txgbe: fix GPIO interrupt blocking") Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com> --- drivers/net/ethernet/wangxun/libwx/wx_type.h | 1 + drivers/net/ethernet/wangxun/txgbe/txgbe_main.c | 5 ++++- drivers/net/ethernet/wangxun/txgbe/txgbe_phy.c | 15 ++++++++++----- 3 files changed, 15 insertions(+), 6 deletions(-)