netdev - Re: [PATCH net 4/7] ice: use correct xdp_ring with XDP

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20210123195219.55f6d4e1@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Sat, 23 Jan 2021 19:52:19 -0800
From:   Jakub Kicinski <kuba@...nel.org>
To:     Tony Nguyen <anthony.l.nguyen@...el.com>
Cc:     davem@...emloft.net, Piotr Raczynski <piotr.raczynski@...el.com>,
        netdev@...r.kernel.org, sassmann@...hat.com,
        George Kuruvinakunnel <george.kuruvinakunnel@...el.com>,
        Maciej Fijalkowski <maciej.fijalkowski@...el.com>,
        Björn Töpel <bjorn@...nel.org>,
        Magnus Karlsson <magnus.karlsson@...el.com>
Subject: Re: [PATCH net 4/7] ice: use correct xdp_ring with XDP_TX action

On Fri, 22 Jan 2021 15:57:31 -0800 Tony Nguyen wrote:
> From: Piotr Raczynski <piotr.raczynski@...el.com>
> 
> XDP queue number for XDP_TX action is used inconsistently
> and may result with no packets transmitted. Fix queue number
> used by the driver when doing XDP_TX, i.e. use receive queue
> number as in ice_finalize_xdp_rx.
> 
> Also, using smp_processor_id() is wrong here and won't
> work with less queues.
> 
> Fixes: efc2214b6047 ("ice: Add support for XDP")
> Signed-off-by: Piotr Raczynski <piotr.raczynski@...el.com>
> Tested-by: George Kuruvinakunnel <george.kuruvinakunnel@...el.com>
> Signed-off-by: Tony Nguyen <anthony.l.nguyen@...el.com>
> ---
>  drivers/net/ethernet/intel/ice/ice_txrx.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ethernet/intel/ice/ice_txrx.c b/drivers/net/ethernet/intel/ice/ice_txrx.c
> index b6fa83c619dd..7946a90b2da7 100644
> --- a/drivers/net/ethernet/intel/ice/ice_txrx.c
> +++ b/drivers/net/ethernet/intel/ice/ice_txrx.c
> @@ -546,7 +546,7 @@ ice_run_xdp(struct ice_ring *rx_ring, struct xdp_buff *xdp,
>  	case XDP_PASS:
>  		break;
>  	case XDP_TX:
> -		xdp_ring = rx_ring->vsi->xdp_rings[smp_processor_id()];
> +		xdp_ring = rx_ring->vsi->xdp_rings[rx_ring->q_index];

But then what protects you from one CPU trying to use the tx ring from
XDP_TX and another from ice_xdp_xmit() ?

Also why does this code not check queue_index < vsi->num_xdp_txq
like ice_xdp_xmit() does?

Let me CC your local XDP experts whose tags I'm surprised not to see on
this patch.

>  		result = ice_xmit_xdp_buff(xdp, xdp_ring);
>  		break;
>  	case XDP_REDIRECT: