Re: [ PATCH 1/1 2.6.25-rc9] bnx2: Add EEH support in bnx2x driver

["Michael Chan" <mchan@broadcom.com>]

On Mon, 2008-04-21 at 16:20 -0500, wendy xiong wrote:

> 
> diff -Nuarp linux-2.6.25-rc9.orig/drivers/net/bnx2.c linux-2.6.25-rc9.eeh/drivers/net/bnx2.c
> --- linux-2.6.25-rc9.orig/drivers/net/bnx2.c    2008-04-18 22:12:53.000000000 -0500
> +++ linux-2.6.25-rc9.eeh/drivers/net/bnx2.c     2008-04-18 22:28:42.000000000 -0500
> @@ -7108,6 +7108,7 @@ bnx2_init_board(struct pci_dev *pdev, st
>         }
> 
>         pci_set_master(pdev);
> +       pci_save_state(pdev);
> 
>         bp->pm_cap = pci_find_capability(pdev, PCI_CAP_ID_PM);
>         if (bp->pm_cap == 0) {
> @@ -7617,6 +7618,85 @@ bnx2_resume(struct pci_dev *pdev)
>         return 0;
>  }
> 
> +/**
> + * bnx2_io_error_detected - called when PCI error is detected
> + * @pdev: Pointer to PCI device
> + * @state: The current pci connection state
> + *
> + * This function is called after a PCI bus error affecting
> + * this device has been detected.
> + */
> +static pci_ers_result_t bnx2_io_error_detected(struct pci_dev *pdev,
> +                                               pci_channel_state_t state)
> +{
> +       struct net_device *dev = pci_get_drvdata(pdev);
> +       struct bnx2 *bp = netdev_priv(dev);

I think it is better to add rtnl_lock() here to prevent racing with
other functions such as ethtool and close that can be resetting the
chip.  The same should be done in the other 2 functions below.

> +
> +       netif_device_detach(dev);
> +
> +       if (netif_running(dev)) {
> +               bnx2_netif_stop(bp);
> +               del_timer_sync(&bp->timer);
> +               bnx2_reset_nic(bp, BNX2_DRV_MSG_CODE_RESET);
> +       }
> +
> +       pci_disable_device(pdev);
> +

Add rtnl_unlock() here.

> +       /* Request a slot slot reset. */
> +       return PCI_ERS_RESULT_NEED_RESET;
> +}
> +
> +/**
> + * bnx2_io_slot_reset - called after the pci bus has been reset.
> + * @pdev: Pointer to PCI device
> + *
> + * Restart the card from scratch, as if from a cold-boot.
> + */
> +static pci_ers_result_t bnx2_io_slot_reset(struct pci_dev *pdev)
> +{
> +       struct net_device *dev = pci_get_drvdata(pdev);
> +       struct bnx2 *bp = netdev_priv(dev);
> +
> +       if (pci_enable_device(pdev)) {
> +               dev_err(&pdev->dev,
> +                       "Cannot re-enable PCI device after reset.\n");
> +               return PCI_ERS_RESULT_DISCONNECT;
> +       }
> +       pci_set_master(pdev);
> +       pci_restore_state(pdev);
> +
> +       if (netif_running(dev)) {
> +               bnx2_set_power_state(bp, PCI_D0);
> +               bnx2_init_nic(bp);
> +       }
> +
> +       return PCI_ERS_RESULT_RECOVERED;
> +}
> +
> +/**
> + * bnx2_io_resume - called when traffic can start flowing again.
> + * @pdev: Pointer to PCI device
> + *
> + * This callback is called when the error recovery driver tells us that
> + * its OK to resume normal operation.
> + */
> +static void bnx2_io_resume(struct pci_dev *pdev)
> +{
> +       struct net_device *dev = pci_get_drvdata(pdev);
> +       struct bnx2 *bp = netdev_priv(dev);
> +
> +       if (netif_running(dev))
> +               bnx2_netif_start(bp);
> +
> +       netif_device_attach(dev);

Shouldn't we attach first before netif_start?

Thanks.