Re: [PATCH net-next 1/4] bnge: add steps in bnge_shutdown()

[Jakub Kicinski]

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.
---
--- Patch [1]: [PATCH net-next 1/4] bnge: add steps in bnge_shutdown() ---
commit 92ff52be8a6aa712b5331ce7118ada003154f2ee
Author: Vikas Gupta <vikas.gupta@xxxxxxxxxxxx>

bnge: add steps in bnge_shutdown()

This commit updates the shutdown callback to mirror the close sequence 
so the device is brought down cleanly during system shutdown.

> diff --git a/drivers/net/ethernet/broadcom/bnge/bnge_core.c b/drivers/net/ethernet/broadcom/bnge/bnge_core.c
> index 68b74eb2c3a29..9846d29b4fe3c 100644
> --- a/drivers/net/ethernet/broadcom/bnge/bnge_core.c
> +++ b/drivers/net/ethernet/broadcom/bnge/bnge_core.c
[ ... ]
> @@ -406,12 +408,34 @@ static void bnge_remove_one(struct pci_dev *pdev)
>  
>  static void bnge_shutdown(struct pci_dev *pdev)
>  {
[ ... ]
> +	if (bnge_hwrm_func_drv_unrgtr(bd)) {
> +		pcie_flr(pdev);
> +		goto shutdown_exit;
> +	}
> +	bnge_free_irqs(bd);

[Severity: Medium]
Will unconditionally freeing IRQs here trigger a kernel warning if the
RDMA auxiliary driver is currently bound?

When bnge_shutdown() calls bnge_free_irqs() without first tearing down
or unregistering the RDMA auxiliary device, the auxiliary driver may
still have active IRQ actions. This usually causes a WARN_ON for
irq_has_action() inside msi_domain_free_irqs() during system shutdown
or reboot.

Could the auxiliary device be cleanly shut down or removed before
freeing the interrupts?

>  	pci_disable_device(pdev);
>  
>  	if (system_state == SYSTEM_POWER_OFF) {
[ ... ]