Re: [Intel-wired-lan] [PATCH net v2] ice: Unregister netdev and devlink_port only once

[Maciej Fijalkowski]

On Mon, Jun 19, 2023 at 12:58:13PM +0200, Petr Oros wrote:
> Since commit 6624e780a577fc ("ice: split ice_vsi_setup into smaller
> functions") ice_vsi_release does things twice. There is unregister
> netdev which is unregistered in ice_deinit_eth also.
> 
> It also unregisters the devlink_port twice which is also unregistered
> in ice_deinit_eth(). This double deregistration is hidden because
> devl_port_unregister ignores the return value of xa_erase.
> 
> [   68.642167] Call Trace:
> [   68.650385]  ice_devlink_destroy_pf_port+0xe/0x20 [ice]
> [   68.655656]  ice_vsi_release+0x445/0x690 [ice]
> [   68.660147]  ice_deinit+0x99/0x280 [ice]
> [   68.664117]  ice_remove+0x1b6/0x5c0 [ice]
> 
> [  171.103841] Call Trace:
> [  171.109607]  ice_devlink_destroy_pf_port+0xf/0x20 [ice]
> [  171.114841]  ice_remove+0x158/0x270 [ice]
> [  171.118854]  pci_device_remove+0x3b/0xc0
> [  171.122779]  device_release_driver_internal+0xc7/0x170
> [  171.127912]  driver_detach+0x54/0x8c
> [  171.131491]  bus_remove_driver+0x77/0xd1
> [  171.135406]  pci_unregister_driver+0x2d/0xb0
> [  171.139670]  ice_module_exit+0xc/0x55f [ice]
> 
> Fixes: 6624e780a577 ("ice: split ice_vsi_setup into smaller functions")
> Signed-off-by: Petr Oros <poros@xxxxxxxxxx>
> ---
> v2: reword subject
> 
> v1: https://lore.kernel.org/netdev/20230619084948.360128-1-poros@xxxxxxxxxx/
> ---
>  drivers/net/ethernet/intel/ice/ice_lib.c | 27 ------------------------
>  1 file changed, 27 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c b/drivers/net/ethernet/intel/ice/ice_lib.c
> index 11ae0e41f518a1..284a1f0bfdb545 100644
> --- a/drivers/net/ethernet/intel/ice/ice_lib.c
> +++ b/drivers/net/ethernet/intel/ice/ice_lib.c
> @@ -3272,39 +3272,12 @@ int ice_vsi_release(struct ice_vsi *vsi)
>  		return -ENODEV;
>  	pf = vsi->back;
>  
> -	/* do not unregister while driver is in the reset recovery pending
> -	 * state. Since reset/rebuild happens through PF service task workqueue,
> -	 * it's not a good idea to unregister netdev that is associated to the
> -	 * PF that is running the work queue items currently. This is done to
> -	 * avoid check_flush_dependency() warning on this wq
> -	 */
> -	if (vsi->netdev && !ice_is_reset_in_progress(pf->state) &&
> -	    (test_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state))) {
> -		unregister_netdev(vsi->netdev);
> -		clear_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state);
> -	}
> -
> -	if (vsi->type == ICE_VSI_PF)
> -		ice_devlink_destroy_pf_port(pf);
> -
>  	if (test_bit(ICE_FLAG_RSS_ENA, pf->flags))
>  		ice_rss_clean(vsi);
>  
>  	ice_vsi_close(vsi);
>  	ice_vsi_decfg(vsi);
>  
> -	if (vsi->netdev) {
> -		if (test_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state)) {
> -			unregister_netdev(vsi->netdev);
> -			clear_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state);
> -		}
> -		if (test_bit(ICE_VSI_NETDEV_ALLOCD, vsi->state)) {
> -			free_netdev(vsi->netdev);
> -			vsi->netdev = NULL;
> -			clear_bit(ICE_VSI_NETDEV_ALLOCD, vsi->state);
> -		}
> -	}

Hmm it's a bit messy:D

I agree that ice_deinit_eth() should be the one that takes care of netdev
clean up as ice_init_eth() was the one to alloc and register it. I believe
that part of the split up work was to come up with entities that are
scoped to alloc/dealloc related resources...as this was even more messed
up before.

It would be worth to exercise other code paths where ice_vsi_release() is
used - like do a loopback test (ethtool -t $IFACE) or go through a reset.

Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@xxxxxxxxx>

> -
>  	/* retain SW VSI data structure since it is needed to unregister and
>  	 * free VSI netdev when PF is not in reset recovery pending state,\
>  	 * for ex: during rmmod.
> -- 
> 2.41.0
> 
> _______________________________________________
> Intel-wired-lan mailing list
> Intel-wired-lan@xxxxxxxxxx
> https://lists.osuosl.org/mailman/listinfo/intel-wired-lan