Using netconsole netpoll_poll_dev could be called from interrupt
context, thus using disable_irq() would cause the following kernel
warning with CONFIG_DEBUG_ATOMIC_SLEEP enabled:
BUG: sleeping function called from invalid context at kernel/irq/manage.c:137
in_atomic(): 1, irqs_disabled(): 128, non_block: 0, pid: 10, name: ksoftirqd/0
CPU: 0 PID: 10 Comm: ksoftirqd/0 Tainted: G W 5.15.42-00075-g816b502b2298-dirty #117
Hardware name: aml (r1) (DT)
Call trace:
dump_backtrace+0x0/0x270
show_stack+0x14/0x20
dump_stack_lvl+0x8c/0xac
dump_stack+0x18/0x30
___might_sleep+0x150/0x194
__might_sleep+0x64/0xbc
synchronize_irq+0x8c/0x150
disable_irq+0x2c/0x40
stmmac_poll_controller+0x140/0x1a0
netpoll_poll_dev+0x6c/0x220
netpoll_send_skb+0x308/0x390
netpoll_send_udp+0x418/0x760
write_msg+0x118/0x140 [netconsole]
console_unlock+0x404/0x500
vprintk_emit+0x118/0x250
dev_vprintk_emit+0x19c/0x1cc
dev_printk_emit+0x90/0xa8
__dev_printk+0x78/0x9c
_dev_warn+0xa4/0xbc
ath10k_warn+0xe8/0xf0 [ath10k_core]
ath10k_htt_txrx_compl_task+0x790/0x7fc [ath10k_core]
ath10k_pci_napi_poll+0x98/0x1f4 [ath10k_pci]
__napi_poll+0x58/0x1f4
net_rx_action+0x504/0x590
_stext+0x1b8/0x418
run_ksoftirqd+0x74/0xa4
smpboot_thread_fn+0x210/0x3c0
kthread+0x1fc/0x210
ret_from_fork+0x10/0x20
Since [0] .ndo_poll_controller is only needed if driver doesn't or
partially use NAPI. Because stmmac does so, stmmac_poll_controller
can be removed fixing the above warning.
[0] commit ac3d9dd034e5 ("netpoll: make ndo_poll_controller() optional")
Cc: <[email protected]> # 5.15.x
Signed-off-by: Remi Pommarel <[email protected]>
---
.../net/ethernet/stmicro/stmmac/stmmac_main.c | 30 -------------------
1 file changed, 30 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 9a3182b9e767..8d76334fff49 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -5991,33 +5991,6 @@ static irqreturn_t stmmac_msi_intr_rx(int irq, void *data)
return IRQ_HANDLED;
}
-#ifdef CONFIG_NET_POLL_CONTROLLER
-/* Polling receive - used by NETCONSOLE and other diagnostic tools
- * to allow network I/O with interrupts disabled.
- */
-static void stmmac_poll_controller(struct net_device *dev)
-{
- struct stmmac_priv *priv = netdev_priv(dev);
- int i;
-
- /* If adapter is down, do nothing */
- if (test_bit(STMMAC_DOWN, &priv->state))
- return;
-
- if (priv->plat->flags & STMMAC_FLAG_MULTI_MSI_EN) {
- for (i = 0; i < priv->plat->rx_queues_to_use; i++)
- stmmac_msi_intr_rx(0, &priv->dma_conf.rx_queue[i]);
-
- for (i = 0; i < priv->plat->tx_queues_to_use; i++)
- stmmac_msi_intr_tx(0, &priv->dma_conf.tx_queue[i]);
- } else {
- disable_irq(dev->irq);
- stmmac_interrupt(dev->irq, dev);
- enable_irq(dev->irq);
- }
-}
-#endif
-
/**
* stmmac_ioctl - Entry point for the Ioctl
* @dev: Device pointer.
@@ -6978,9 +6951,6 @@ static const struct net_device_ops stmmac_netdev_ops = {
.ndo_get_stats64 = stmmac_get_stats64,
.ndo_setup_tc = stmmac_setup_tc,
.ndo_select_queue = stmmac_select_queue,
-#ifdef CONFIG_NET_POLL_CONTROLLER
- .ndo_poll_controller = stmmac_poll_controller,
-#endif
.ndo_set_mac_address = stmmac_set_mac_address,
.ndo_vlan_rx_add_vid = stmmac_vlan_rx_add_vid,
.ndo_vlan_rx_kill_vid = stmmac_vlan_rx_kill_vid,
--
2.40.0
On Thu, Sep 07, 2023 at 11:23:16AM +0200, Paolo Abeni wrote:
> On Wed, 2023-09-06 at 11:13 +0200, Remi Pommarel wrote:
> > Using netconsole netpoll_poll_dev could be called from interrupt
> > context, thus using disable_irq() would cause the following kernel
> > warning with CONFIG_DEBUG_ATOMIC_SLEEP enabled:
> >
> > BUG: sleeping function called from invalid context at kernel/irq/manage.c:137
> > in_atomic(): 1, irqs_disabled(): 128, non_block: 0, pid: 10, name: ksoftirqd/0
> > CPU: 0 PID: 10 Comm: ksoftirqd/0 Tainted: G W 5.15.42-00075-g816b502b2298-dirty #117
> > Hardware name: aml (r1) (DT)
> > Call trace:
> > dump_backtrace+0x0/0x270
> > show_stack+0x14/0x20
> > dump_stack_lvl+0x8c/0xac
> > dump_stack+0x18/0x30
> > ___might_sleep+0x150/0x194
> > __might_sleep+0x64/0xbc
> > synchronize_irq+0x8c/0x150
> > disable_irq+0x2c/0x40
> > stmmac_poll_controller+0x140/0x1a0
> > netpoll_poll_dev+0x6c/0x220
> > netpoll_send_skb+0x308/0x390
> > netpoll_send_udp+0x418/0x760
> > write_msg+0x118/0x140 [netconsole]
> > console_unlock+0x404/0x500
> > vprintk_emit+0x118/0x250
> > dev_vprintk_emit+0x19c/0x1cc
> > dev_printk_emit+0x90/0xa8
> > __dev_printk+0x78/0x9c
> > _dev_warn+0xa4/0xbc
> > ath10k_warn+0xe8/0xf0 [ath10k_core]
> > ath10k_htt_txrx_compl_task+0x790/0x7fc [ath10k_core]
> > ath10k_pci_napi_poll+0x98/0x1f4 [ath10k_pci]
> > __napi_poll+0x58/0x1f4
> > net_rx_action+0x504/0x590
> > _stext+0x1b8/0x418
> > run_ksoftirqd+0x74/0xa4
> > smpboot_thread_fn+0x210/0x3c0
> > kthread+0x1fc/0x210
> > ret_from_fork+0x10/0x20
> >
> > Since [0] .ndo_poll_controller is only needed if driver doesn't or
> > partially use NAPI. Because stmmac does so, stmmac_poll_controller
> > can be removed fixing the above warning.
> >
> > [0] commit ac3d9dd034e5 ("netpoll: make ndo_poll_controller() optional")
> >
> > Cc: <[email protected]> # 5.15.x
> > Signed-off-by: Remi Pommarel <[email protected]>
>
> I'm sorry for the incremental feedback, but we also need a suitable
> Fixes tag, thanks!
I didn't include Fixes tag because it would go back up to the initial
driver support commit [0]. I can't be sure that this commit includes
necessary NAPI implementation to be able to get rid of
.ndo_poll_controller callback back then. And I am not able to test it on
older version than 5.15.x hence I only included the 5.15.x Cc tag
version prerequisite.
But I surely can add a Fixed tag if it is ok for it to be [0].
Also sorry for the long replying delay.
[0] commit 47dd7a540b8a ("net: add support for STMicroelectronics Ethernet controllers")
--
Remi
On Tue, 26 Sep 2023 11:47:56 +0200 Remi Pommarel wrote:
> > I'm sorry for the incremental feedback, but we also need a suitable
> > Fixes tag, thanks!
>
> I didn't include Fixes tag because it would go back up to the initial
> driver support commit [0]. I can't be sure that this commit includes
> necessary NAPI implementation to be able to get rid of
> .ndo_poll_controller callback back then. And I am not able to test it on
> older version than 5.15.x hence I only included the 5.15.x Cc tag
> version prerequisite.
>
> But I surely can add a Fixed tag if it is ok for it to be [0].
>
> Also sorry for the long replying delay.
>
> [0] commit 47dd7a540b8a ("net: add support for STMicroelectronics Ethernet controllers")
AFAIU the Fixes tag only indicates where the bug was present,
no guarantees on whether the fix can be backported as far back.
IOW I think [0] as Fixes tag will be perfectly correct, please
repost with it included?