netdev - Re: [PATCH net-next 1/6] sfc: support encapsulation of xdp_frames in efx_tx

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]

Date:   Wed, 23 Oct 2019 00:19:17 +0200
From:   Jesper Dangaard Brouer <brouer@...hat.com>
To:     Charles McLachlan <cmclachlan@...arflare.com>
Cc:     <davem@...emloft.net>, <netdev@...r.kernel.org>,
        <linux-net-drivers@...arflare.com>, brouer@...hat.com
Subject: Re: [PATCH net-next 1/6] sfc: support encapsulation of xdp_frames
 in efx_tx_buffer.

On Tue, 22 Oct 2019 16:37:53 +0100
Charles McLachlan <cmclachlan@...arflare.com> wrote:

> Add a field to efx_tx_buffer so that we can track xdp_frames. Add a
> flag so that buffers that contain xdp_frames can be identified and
> passed to xdp_return_frame.
> 
> Signed-off-by: Charles McLachlan <cmclachlan@...arflare.com>
> ---
>  drivers/net/ethernet/sfc/net_driver.h | 10 ++++++++--
>  drivers/net/ethernet/sfc/tx.c         |  2 ++
>  2 files changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
> index 284a1b047ac2..7394d901e021 100644
> --- a/drivers/net/ethernet/sfc/net_driver.h
> +++ b/drivers/net/ethernet/sfc/net_driver.h
> @@ -27,6 +27,7 @@
>  #include <linux/i2c.h>
>  #include <linux/mtd/mtd.h>
>  #include <net/busy_poll.h>
> +#include <net/xdp.h>
>  
>  #include "enum.h"
>  #include "bitfield.h"
> @@ -136,7 +137,8 @@ struct efx_special_buffer {
>   * struct efx_tx_buffer - buffer state for a TX descriptor
>   * @skb: When @flags & %EFX_TX_BUF_SKB, the associated socket buffer to be
>   *	freed when descriptor completes
> - * @option: When @flags & %EFX_TX_BUF_OPTION, a NIC-specific option descriptor.
> + * @xdpf: When @flags & %EFX_TX_BUF_XDP, the XDP frame information; its @data
> + *	member is the associated buffer to drop a page reference on.
>   * @dma_addr: DMA address of the fragment.
>   * @flags: Flags for allocation and DMA mapping type
>   * @len: Length of this fragment.
> @@ -146,7 +148,10 @@ struct efx_special_buffer {
>   * Only valid if @unmap_len != 0.
>   */
>  struct efx_tx_buffer {
> -	const struct sk_buff *skb;
> +	union {
> +		const struct sk_buff *skb;
> +		struct xdp_frame *xdpf;
> +	};
>  	union {
>  		efx_qword_t option;
>  		dma_addr_t dma_addr;
> @@ -160,6 +165,7 @@ struct efx_tx_buffer {
>  #define EFX_TX_BUF_SKB		2	/* buffer is last part of skb */
>  #define EFX_TX_BUF_MAP_SINGLE	8	/* buffer was mapped with dma_map_single() */
>  #define EFX_TX_BUF_OPTION	0x10	/* empty buffer for option descriptor */
> +#define EFX_TX_BUF_XDP		0x20	/* buffer was sent with XDP */
>  
>  /**
>   * struct efx_tx_queue - An Efx TX queue
> diff --git a/drivers/net/ethernet/sfc/tx.c b/drivers/net/ethernet/sfc/tx.c
> index 65e81ec1b314..9905e8952a45 100644
> --- a/drivers/net/ethernet/sfc/tx.c
> +++ b/drivers/net/ethernet/sfc/tx.c
> @@ -95,6 +95,8 @@ static void efx_dequeue_buffer(struct efx_tx_queue *tx_queue,
>  		netif_vdbg(tx_queue->efx, tx_done, tx_queue->efx->net_dev,
>  			   "TX queue %d transmission id %x complete\n",
>  			   tx_queue->queue, tx_queue->read_count);
> +	} else if (buffer->flags & EFX_TX_BUF_XDP) {
> +		xdp_return_frame(buffer->xdpf);

Is this efx_dequeue_buffer() function always called under NAPI protection?
(So it could use the faster xdp_return_frame_rx_napi() ... ?)


>  	}
>  
>  	buffer->len = 0;



-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  LinkedIn: http://www.linkedin.com/in/brouer