net: fec: add initial XDP support
Shenwei Wang
reviewerPATCH v3 1/1 net: fec: add initial XDP support
This patch adds the initial XDP support to Freescale driver. It supports XDP_PASS, XDP_DROP and XDP_REDIRECT actions. Upcoming patches will add support for XDP_TX and Zero Copy features.
As the patch is rather large, the part of codes to collect the statistics is separated and will prepare a dedicated patch for that part.
I just tested with the application of xdpsock.
-- Native here means running command of "xdpsock -i eth0" -- SKB-Mode means running command of "xdpsock -S -i eth0"
The following are the testing result relating to XDP mode:
root@imx8qxpc0mek:~/bpf# ./xdpsock -i eth0
sock0@eth0:0 rxdrop xdp-drv
pps pkts 1.00
rx 371347 2717794
tx 0 0root@imx8qxpc0mek:~/bpf# ./xdpsock -S -i eth0
sock0@eth0:0 rxdrop xdp-skb
pps pkts 1.00
rx 202229 404528
tx 0 0root@imx8qxpc0mek:~/bpf# ./xdp2 eth0 proto 0: 496708 pkt/s proto 0: 505469 pkt/s proto 0: 505283 pkt/s proto 0: 505443 pkt/s proto 0: 505465 pkt/s
root@imx8qxpc0mek:~/bpf# ./xdp2 -S eth0 proto 0: 0 pkt/s proto 17: 118778 pkt/s proto 17: 118989 pkt/s proto 0: 1 pkt/s proto 17: 118987 pkt/s proto 0: 0 pkt/s proto 17: 118943 pkt/s proto 17: 118976 pkt/s proto 0: 1 pkt/s proto 17: 119006 pkt/s proto 0: 0 pkt/s proto 17: 119071 pkt/s proto 17: 119092 pkt/s
---
changes in V3:
- remove the codes to update the RX ring size to avoid potential risk.
- update the testing data based on the new implementation.
changes in V2:
- Get rid of the expensive fec_net_close/open function calls during XDP/Normal Mode switch.
- Update the testing data on i.mx8qxp mek board.
- fix the compile issue reported by kernel_test_robot
drivers/net/ethernet/freescale/fec.h | 4 +- drivers/net/ethernet/freescale/fec_main.c | 224 +++++++++++++++++++++- 2 files changed, 226 insertions(+), 2 deletions(-)
drivers/net/ethernet/freescale/fec.h +3 -1
@@ -348,7 +348,6 @@ struct bufdesc_ex {*/ #define FEC_ENET_XDP_HEADROOM (XDP_PACKET_HEADROOM) -#define FEC_ENET_RX_PAGES 256#define FEC_ENET_RX_FRSIZE (PAGE_SIZE - FEC_ENET_XDP_HEADROOM \ - SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))
@@ -663,6 +662,9 @@ struct fec_enet_private {struct imx_sc_ipc *ipc_handle; +/* XDP BPF Program */ +struct bpf_prog *xdp_prog; +u64 ethtool_stats[]; };
drivers/net/ethernet/freescale/fec_main.c +223 -1
284 lines not on this page · raw message
--
2.34.1
Shenwei Wang
reviewerA kind ping. 😊
Regards,
Shenwei
Quoted 409 lines
-----Original Message-----
From: Shenwei Wang <[email protected]>
Sent: Monday, October 31, 2022 1:54 PM
To: David S. Miller <[email protected]>; Eric Dumazet <[email protected]>; Jakub Kicinski <[email protected]>; Paolo Abeni <[email protected]>
Cc: Alexei Starovoitov <[email protected]>; Daniel Borkmann <[email protected]>; Jesper Dangaard Brouer <[email protected]>; John Fastabend <[email protected]>; [email protected]; linux- [email protected]; [email protected]; Shenwei Wang <[email protected]>; kernel test robot <[email protected]>
Subject: [PATCH v3 1/1] net: fec: add initial XDP supportThis patch adds the initial XDP support to Freescale driver. It supports
XDP_PASS,XDP_DROPandXDP_REDIRECTactions. Upcoming patches will add support forXDP_TXand Zero Copy features.As the patch is rather large, the part of codes to collect the statistics is separated and will prepare a dedicated patch for that part.
I just tested with the application of xdpsock.
-- Native here means running command of "xdpsock -i eth0" -- SKB-Mode means running command of "xdpsock -S -i eth0"The following are the testing result relating to XDP mode:
root@imx8qxpc0mek:~/bpf# ./xdpsock -i eth0 sock0@eth0:0 rxdrop xdp-drv pps pkts 1.00 rx 371347 2717794 tx 0 0root@imx8qxpc0mek:~/bpf# ./xdpsock -S -i eth0 sock0@eth0:0 rxdrop xdp-skb pps pkts 1.00 rx 202229 404528 tx 0 0root@imx8qxpc0mek:~/bpf# ./xdp2 eth0 proto 0: 496708 pkt/s proto 0: 505469 pkt/s proto 0: 505283 pkt/s proto 0: 505443 pkt/s proto 0: 505465 pkt/sroot@imx8qxpc0mek:~/bpf# ./xdp2 -S eth0 proto 0: 0 pkt/s proto 17: 118778 pkt/s proto 17: 118989 pkt/s proto 0: 1 pkt/s proto 17: 118987 pkt/s proto 0: 0 pkt/s proto 17: 118943 pkt/s proto 17: 118976 pkt/s proto 0: 1 pkt/s proto 17: 119006 pkt/s proto 0: 0 pkt/s proto 17: 119071 pkt/s proto 17: 119092 pkt/sSigned-off-by Shenwei Wang <[email protected]>Reported-by kernel test robot <[email protected]>---
changes in V3:
- remove the codes to update the RX ring size to avoid potential risk.
- update the testing data based on the new implementation.changes in V2:
- Get rid of the expensive fec_net_close/open function calls during XDP/Normal Mode switch.
- Update the testing data on i.mx8qxp mek board.
- fix the compile issue reported bykernel_test_robotdrivers/net/ethernet/freescale/fec.h | 4 +- drivers/net/ethernet/freescale/fec_main.c | 224 +++++++++++++++++++++- 2 files changed, 226 insertions(+), 2 deletions(-)drivers/net/ethernet/freescale/fec.h +4
-1@@ -348,7 +348,6 @@ struct bufdesc_ex {*/ #define FEC_ENET_XDP_HEADROOM (XDP_PACKET_HEADROOM) -#define FEC_ENET_RX_PAGES 256#define FEC_ENET_RX_FRSIZE (PAGE_SIZE - FEC_ENET_XDP_HEADROOM \ - SKB_DATA_ALIGN(sizeof(struct skb_shared_info))) @@ -663,6 +662,9 @@ struct fec_enet_private {struct imx_sc_ipc *ipc_handle; +/* XDP BPF Program */ +struct bpf_prog *xdp_prog; +u64 ethtool_stats[]; };drivers/net/ethernet/freescale/fec_main.c +5
-0@@ -89,6 +89,11 @@ static const u16 fec_enet_vlan_pri_to_queue[8] = {0, 0, 1, 1, 1, 2, 2, 2};#define FEC_ENET_OPD_V 0xFFF0#define FEC_MDIO_PM_TIMEOUT 100/* ms */ +#define FEC_ENET_XDP_PASS 0 +#define FEC_ENET_XDP_CONSUMED BIT(0) +#define FEC_ENET_XDP_TX BIT(1) +#define FEC_ENET_XDP_REDIR BIT(2) +struct fec_devinfo {u32 quirks;};
drivers/net/ethernet/freescale/fec_main.c +55
-1@@ -418,13 +423,14 @@ static int fec_enet_create_page_pool(struct fec_enet_private *fep, struct fec_enet_priv_rx_q *rxq, int size) { + struct bpf_prog *xdp_prog = READ_ONCE(fep->xdp_prog);struct page_pool_params pp_params = { .order =0 , .flags =PP_FLAG_DMA_MAP |PP_FLAG_DMA_SYNC_DEV , .pool_size = size, .nid =dev_to_node (&fep->pdev->dev), .dev = &fep->pdev->dev, - .dma_dir =DMA_FROM_DEVICE , + .dma_dir = xdp_prog ?DMA_BIDIRECTIONAL :DMA_FROM_DEVICE , .offset =FEC_ENET_XDP_HEADROOM , .max_len =FEC_ENET_RX_FRSIZE , };@@ -1499,6 +1505,59 @@ static void fec_enet_update_cbd(struct fec_enet_priv_rx_q *rxq, bdp->cbd_bufaddr =cpu_to_fec32 (phys_addr); } +static u32 +fec_enet_run_xdp (struct fec_enet_private *fep,struct bpf_prog *prog, +struct xdp_buff *xdp,struct fec_enet_priv_rx_q *rxq,int index) { +unsigned int sync, len = xdp->data_end - xdp->data; +u32 ret =FEC_ENET_XDP_PASS ; +struct page *page; +int err; +u32 act; + + act =bpf_prog_run_xdp (prog, xdp); + +/* Due xdp_adjust_tail: DMA sync for_device cover max len CPU touch */ + sync = xdp->data_end - xdp->data_hard_start -FEC_ENET_XDP_HEADROOM ; + sync =max (sync, len); + +switch (act) { +case XDP_PASS : + ret =FEC_ENET_XDP_PASS ; +break ; + +case XDP_REDIRECT : + err =xdp_do_redirect (fep->netdev, xdp, prog); +if (!err) { + ret =FEC_ENET_XDP_REDIR ; + }else { + ret =FEC_ENET_XDP_CONSUMED ; + page =virt_to_head_page (xdp->data); +page_pool_put_page (rxq->page_pool, page, sync, true); + } +break ; + +default : +bpf_warn_invalid_xdp_action (fep->netdev, prog, act); + fallthrough; + +case XDP_TX : +bpf_warn_invalid_xdp_action (fep->netdev, prog, act); + fallthrough; + +case XDP_ABORTED : + fallthrough;/* handle aborts by dropping packet */ + +case XDP_DROP : + ret =FEC_ENET_XDP_CONSUMED ; + page =virt_to_head_page (xdp->data); +page_pool_put_page (rxq->page_pool, page, sync, true); +break ; + } + +return ret; +} +/* During a receive, the bd_rx.cur points to the current incoming buffer. * When we update through the ring, if the next incoming buffer has * not been given to the system, we just set the empty indicator, @@ -1520,6 +1579,9 @@ fec_enet_rx_queue(struct net_device *ndev, int budget, u16
queue_id)drivers/net/ethernet/freescale/fec_main.c +4
-0u16 vlan_tag;int index =0 ;bool need_swap = fep->quirks &FEC_QUIRK_SWAP_FRAME ; +struct bpf_prog *xdp_prog =READ_ONCE (fep->xdp_prog); +u32 ret, xdp_result =FEC_ENET_XDP_PASS ; +struct xdp_buff xdp;struct page *page;#ifdef CONFIG_M532x@@ -1531,6 +1593,7 @@ fec_enet_rx_queue(struct net_device *ndev, intbudget, u16 queue_id) * These get messed up if we get called due to a busy condition. */ bdp = rxq->bd.cur; +xdp_init_buff (&xdp,PAGE_SIZE , &rxq->xdp_rxq);while (!((status =fec16_to_cpu (bdp->cbd_sc)) &BD_ENET_RX_EMPTY )){
drivers/net/ethernet/freescale/fec_main.c +11
-0@@ -1580,6 +1643,17 @@ fec_enet_rx_queue(struct net_device *ndev, int budget, u16 queue_id)prefetch (page_address (page));fec_enet_update_cbd (rxq, bdp, index); +if (xdp_prog) { +xdp_buff_clear_frags_flag (&xdp); +xdp_prepare_buff (&xdp,page_address (page), +FEC_ENET_XDP_HEADROOM , pkt_len, false); + + ret =fec_enet_run_xdp (fep, xdp_prog, &xdp, rxq, index); + xdp_result |= ret; +if (ret !=FEC_ENET_XDP_PASS ) + goto rx_processing_done; + } +drivers/net/ethernet/freescale/fec_main.c +147
-1/* The packet length includes FCS, but we don't want to * include that when passing upstream as it messes up * bridging applications. @@ -1675,6 +1749,10 @@ fec_enet_rx_queue(struct net_device *ndev, int budget, u16 queue_id)writel (0 , rxq->bd.reg_desc_active); } rxq->bd.cur = bdp; + +if (xdp_result &FEC_ENET_XDP_REDIR ) +xdp_do_flush_map (); +return pkt_received; }@@ -3476,6 +3554,148 @@ static u16 fec_enet_select_queue(struct net_device *ndev, struct sk_buff *skb, return fec_enet_vlan_pri_to_queue[vlan_tag >> 13]; } +static int fec_enet_bpf (struct net_device *dev,struct netdev_bpf *bpf) +{ +struct fec_enet_private *fep =netdev_priv (dev); +bool is_run =netif_running (dev); +struct bpf_prog *old_prog; + +switch (bpf->command) { +case XDP_SETUP_PROG : +if (is_run) { +napi_disable (&fep->napi); +netif_tx_disable (dev); + } + + old_prog =xchg (&fep->xdp_prog, bpf->prog); +fec_restart (dev); + +if (is_run) { +napi_enable (&fep->napi); +netif_tx_start_all_queues (dev); + } + +if (old_prog) +bpf_prog_put (old_prog); + +return 0 ; + +case XDP_SETUP_XSK_POOL : +return -EOPNOTSUPP ; + +default : +return -EOPNOTSUPP ; + } +} + +static int +fec_enet_xdp_get_tx_queue (struct fec_enet_private *fep,int cpu) { +int index = cpu; + +if (unlikely (index <0 )) + index =0 ; + +while (index >= fep->num_tx_queues) + index -= fep->num_tx_queues; + +return index; +} + +static int fec_enet_txq_xmit_frame (struct fec_enet_private *fep, +struct fec_enet_priv_tx_q *txq, +struct xdp_frame *frame) +{ +unsigned int index, status, estatus; +struct bufdesc *bdp, *last_bdp; +dma_addr_t dma_addr; +int entries_free; + + entries_free =fec_enet_get_free_txdesc_num (txq); +if (entries_free <MAX_SKB_FRAGS +1 ) { +netdev_err (fep->netdev,"NOT enough BD for SG!\n" ); +return NETDEV_TX_OK ; + } + +/* Fill in a Tx ring entry */ + bdp = txq->bd.cur; + last_bdp = bdp; + status =fec16_to_cpu (bdp->cbd_sc); + status &= ~BD_ENET_TX_STATS ; + + index =fec_enet_get_bd_index (bdp, &txq->bd); + + dma_addr =dma_map_single (&fep->pdev->dev, frame->data, + frame->len,DMA_TO_DEVICE ); +if (dma_mapping_error (&fep->pdev->dev, dma_addr)) +return FEC_ENET_XDP_CONSUMED ; + + status |= (BD_ENET_TX_INTR |BD_ENET_TX_LAST ); +if (fep->bufdesc_ex) + estatus =BD_ENET_TX_INT ; + + bdp->cbd_bufaddr =cpu_to_fec32 (dma_addr); + bdp->cbd_datlen =cpu_to_fec16 (frame->len); + +if (fep->bufdesc_ex) { +struct bufdesc_ex *ebdp = (struct bufdesc_ex *)bdp; + +if (fep->quirks &FEC_QUIRK_HAS_AVB ) + estatus |=FEC_TX_BD_FTYPE (txq->bd.qid); + + ebdp->cbd_bdu =0 ; + ebdp->cbd_esc =cpu_to_fec32 (estatus); + } + + index =fec_enet_get_bd_index (last_bdp, &txq->bd); + txq->tx_skbuff[index] =NULL ; + +/* Send it on its way. Tell FEC it's ready, interrupt when done, +* it's the last BD of the frame, and to put the CRC on the end. +*/ + status |= (BD_ENET_TX_READY |BD_ENET_TX_TC ); + bdp->cbd_sc =cpu_to_fec16 (status); + +/* If this was the last BD in the ring, start at the beginning again. */ + bdp =fec_enet_get_nextdesc (last_bdp, &txq->bd); + + txq->bd.cur = bdp; + +return 0 ; +} + +static int fec_enet_xdp_xmit (struct net_device *dev, +int num_frames, +struct xdp_frame **frames, +u32 flags) +{ +struct fec_enet_private *fep =netdev_priv (dev); +struct fec_enet_priv_tx_q *txq; +int cpu =smp_processor_id (); +struct netdev_queue *nq; +unsigned int queue; +int i; + + queue =fec_enet_xdp_get_tx_queue (fep, cpu); + txq = fep->tx_queue[queue]; + nq =netdev_get_tx_queue (fep->netdev, queue); + +__netif_tx_lock (nq, cpu); + +for (i =0 ; i < num_frames; i++) +fec_enet_txq_xmit_frame (fep, txq, frames[i]); + +/* Make sure the update to bdp and tx_skbuff are performed. */ +wmb (); + +/* Trigger transmission start */ +writel (0 , txq->bd.reg_desc_active); + +__netif_tx_unlock (nq); + +return num_frames; +} +static const struct net_device_ops fec_netdev_ops = { .ndo_open = fec_enet_open, .ndo_stop = fec_enet_close,@@ -3490,6 +3710,8 @@ static const struct net_device_ops fec_netdev_ops = { .ndo_poll_controller = fec_poll_controller, #endif .ndo_set_features = fec_set_features, + .ndo_bpf = fec_enet_bpf, + .ndo_xdp_xmit = fec_enet_xdp_xmit, }; static const unsigned short offset_des_active_rxq[] = { --2.34.1
Jakub Kicinski
maintainerQuoted 1 line · On Thu, 3 Nov 2022 02:07:19 +0000 Shenwei Wang wrote:
A kind ping. 😊
Pong.
Don't top post.
Paolo Abeni
maintainerQuoted 264 lines · Shenwei Wang
This patch adds the initial XDP support to Freescale driver. It supports
XDP_PASS,XDP_DROPandXDP_REDIRECTactions. Upcoming patches will add support forXDP_TXand Zero Copy features.As the patch is rather large, the part of codes to collect the statistics is separated and will prepare a dedicated patch for that part.
I just tested with the application of xdpsock.
-- Native here means running command of "xdpsock -i eth0" -- SKB-Mode means running command of "xdpsock -S -i eth0"The following are the testing result relating to XDP mode:
root@imx8qxpc0mek:~/bpf# ./xdpsock -i eth0 sock0@eth0:0 rxdrop xdp-drv pps pkts 1.00 rx 371347 2717794 tx 0 0root@imx8qxpc0mek:~/bpf# ./xdpsock -S -i eth0 sock0@eth0:0 rxdrop xdp-skb pps pkts 1.00 rx 202229 404528 tx 0 0root@imx8qxpc0mek:~/bpf# ./xdp2 eth0 proto 0: 496708 pkt/s proto 0: 505469 pkt/s proto 0: 505283 pkt/s proto 0: 505443 pkt/s proto 0: 505465 pkt/sroot@imx8qxpc0mek:~/bpf# ./xdp2 -S eth0 proto 0: 0 pkt/s proto 17: 118778 pkt/s proto 17: 118989 pkt/s proto 0: 1 pkt/s proto 17: 118987 pkt/s proto 0: 0 pkt/s proto 17: 118943 pkt/s proto 17: 118976 pkt/s proto 0: 1 pkt/s proto 17: 119006 pkt/s proto 0: 0 pkt/s proto 17: 119071 pkt/s proto 17: 119092 pkt/sSigned-off-by Shenwei Wang <[email protected]>Reported-by kernel test robot <[email protected]>---
changes in V3:
- remove the codes to update the RX ring size to avoid potential risk.
- update the testing data based on the new implementation.changes in V2:
- Get rid of the expensive fec_net_close/open function calls during XDP/Normal Mode switch.
- Update the testing data on i.mx8qxp mek board.
- fix the compile issue reported bykernel_test_robotdrivers/net/ethernet/freescale/fec.h | 4 +- drivers/net/ethernet/freescale/fec_main.c | 224 +++++++++++++++++++++- 2 files changed, 226 insertions(+), 2 deletions(-)drivers/net/ethernet/freescale/fec.h +3
-1@@ -348,7 +348,6 @@ struct bufdesc_ex {*/ #define FEC_ENET_XDP_HEADROOM (XDP_PACKET_HEADROOM) -#define FEC_ENET_RX_PAGES 256#define FEC_ENET_RX_FRSIZE (PAGE_SIZE - FEC_ENET_XDP_HEADROOM \ - SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))@@ -663,6 +662,9 @@ struct fec_enet_private {struct imx_sc_ipc *ipc_handle; +/* XDP BPF Program */ +struct bpf_prog *xdp_prog; +u64 ethtool_stats[]; };drivers/net/ethernet/freescale/fec_main.c +113
-1@@ -89,6 +89,11 @@ static const u16 fec_enet_vlan_pri_to_queue[8] = {0, 0, 1, 1, 1, 2, 2, 2};#define FEC_ENET_OPD_V 0xFFF0#define FEC_MDIO_PM_TIMEOUT 100/* ms */ +#define FEC_ENET_XDP_PASS 0 +#define FEC_ENET_XDP_CONSUMED BIT(0) +#define FEC_ENET_XDP_TX BIT(1) +#define FEC_ENET_XDP_REDIR BIT(2) +struct fec_devinfo {u32 quirks; };@@ -418,13 +423,14 @@ static int fec_enet_create_page_pool(struct fec_enet_private *fep, struct fec_enet_priv_rx_q *rxq, int size) { + struct bpf_prog *xdp_prog = READ_ONCE(fep->xdp_prog);struct page_pool_params pp_params = { .order =0 , .flags =PP_FLAG_DMA_MAP |PP_FLAG_DMA_SYNC_DEV , .pool_size = size, .nid =dev_to_node (&fep->pdev->dev), .dev = &fep->pdev->dev, - .dma_dir =DMA_FROM_DEVICE , + .dma_dir = xdp_prog ?DMA_BIDIRECTIONAL :DMA_FROM_DEVICE , .offset =FEC_ENET_XDP_HEADROOM , .max_len =FEC_ENET_RX_FRSIZE , };@@ -1499,6 +1505,59 @@ static void fec_enet_update_cbd(struct fec_enet_priv_rx_q *rxq, bdp->cbd_bufaddr =cpu_to_fec32 (phys_addr); } +static u32 +fec_enet_run_xdp (struct fec_enet_private *fep,struct bpf_prog *prog, +struct xdp_buff *xdp,struct fec_enet_priv_rx_q *rxq,int index) +{ +unsigned int sync, len = xdp->data_end - xdp->data; +u32 ret =FEC_ENET_XDP_PASS ; +struct page *page; +int err; +u32 act; + + act =bpf_prog_run_xdp (prog, xdp); + +/* Due xdp_adjust_tail: DMA sync for_device cover max len CPU touch */ + sync = xdp->data_end - xdp->data_hard_start -FEC_ENET_XDP_HEADROOM ; + sync =max (sync, len); + +switch (act) { +case XDP_PASS : + ret =FEC_ENET_XDP_PASS ; +break ; + +case XDP_REDIRECT : + err =xdp_do_redirect (fep->netdev, xdp, prog); +if (!err) { + ret =FEC_ENET_XDP_REDIR ; + }else { + ret =FEC_ENET_XDP_CONSUMED ; + page =virt_to_head_page (xdp->data); +page_pool_put_page (rxq->page_pool, page, sync, true); + } +break ; + +default : +bpf_warn_invalid_xdp_action (fep->netdev, prog, act); + fallthrough; + +case XDP_TX : +bpf_warn_invalid_xdp_action (fep->netdev, prog, act); + fallthrough; + +case XDP_ABORTED : + fallthrough;/* handle aborts by dropping packet */ + +case XDP_DROP : + ret =FEC_ENET_XDP_CONSUMED ; + page =virt_to_head_page (xdp->data); +page_pool_put_page (rxq->page_pool, page, sync, true); +break ; + } + +return ret; +} +/* During a receive, the bd_rx.cur points to the current incoming buffer. * When we update through the ring, if the next incoming buffer has * not been given to the system, we just set the empty indicator, @@ -1520,6 +1579,9 @@ fec_enet_rx_queue(struct net_device *ndev, int budget, u16 queue_id)u16 vlan_tag;int index =0 ;bool need_swap = fep->quirks &FEC_QUIRK_SWAP_FRAME ; +struct bpf_prog *xdp_prog =READ_ONCE (fep->xdp_prog); +u32 ret, xdp_result =FEC_ENET_XDP_PASS ; +struct xdp_buff xdp;struct page *page;#ifdef CONFIG_M532x@@ -1531,6 +1593,7 @@ fec_enet_rx_queue(struct net_device *ndev, int budget, u16 queue_id)* These get messed up if we get called due to a busy condition. */ bdp = rxq->bd.cur; +xdp_init_buff (&xdp,PAGE_SIZE , &rxq->xdp_rxq);while (!((status =fec16_to_cpu (bdp->cbd_sc)) &BD_ENET_RX_EMPTY )) {@@ -1580,6 +1643,17 @@ fec_enet_rx_queue(struct net_device *ndev, int budget, u16 queue_id)prefetch (page_address (page));fec_enet_update_cbd (rxq, bdp, index); +if (xdp_prog) { +xdp_buff_clear_frags_flag (&xdp); +xdp_prepare_buff (&xdp,page_address (page), +FEC_ENET_XDP_HEADROOM , pkt_len, false); + + ret =fec_enet_run_xdp (fep, xdp_prog, &xdp, rxq, index); + xdp_result |= ret; +if (ret !=FEC_ENET_XDP_PASS ) + goto rx_processing_done; + } +/* The packet length includes FCS, but we don't want to * include that when passing upstream as it messes up * bridging applications. @@ -1675,6 +1749,10 @@ fec_enet_rx_queue(struct net_device *ndev, int budget, u16 queue_id)writel (0 , rxq->bd.reg_desc_active); } rxq->bd.cur = bdp; + +if (xdp_result &FEC_ENET_XDP_REDIR ) +xdp_do_flush_map (); +return pkt_received; }@@ -3476,6 +3554,148 @@ static u16 fec_enet_select_queue(struct net_device *ndev, struct sk_buff *skb,return fec_enet_vlan_pri_to_queue[vlan_tag >>13 ]; } +static int fec_enet_bpf (struct net_device *dev,struct netdev_bpf *bpf) +{ +struct fec_enet_private *fep =netdev_priv (dev); +bool is_run =netif_running (dev); +struct bpf_prog *old_prog; + +switch (bpf->command) { +case XDP_SETUP_PROG : +if (is_run) { +napi_disable (&fep->napi); +netif_tx_disable (dev); + } + + old_prog =xchg (&fep->xdp_prog, bpf->prog); +fec_restart (dev); + +if (is_run) { +napi_enable (&fep->napi); +netif_tx_start_all_queues (dev); + } + +if (old_prog) +bpf_prog_put (old_prog); + +return 0 ; + +case XDP_SETUP_XSK_POOL : +return -EOPNOTSUPP ; + +default : +return -EOPNOTSUPP ; + } +} +
drivers/net/ethernet/freescale/fec_main.c static u16 fec_enet_select_queue(struct net_device *ndev, struct sk_buff *skb, · line 3600
Quoted 10 lines
drivers/net/ethernet/freescale/fec_main.c +10
-0+static int +fec_enet_xdp_get_tx_queue (struct fec_enet_private *fep,int cpu) +{ +int index = cpu; + +if (unlikely (index <0 )) + index =0 ; + +while (index >= fep->num_tx_queues) + index -= fep->num_tx_queues;
Not a big deal, but I think kind of optimizations are not worthy and potentially dangerous since late '90 ;)
You could consider switching to a simpler '%', but IMHO it's not blocking.
Cheers,
Paolo
Shenwei Wang
reviewerQuoted 10 lines
-----Original Message-----
From: Paolo Abeni <[email protected]>drivers/net/ethernet/freescale/fec_main.c static u16 fec_enet_select_queue(struct net_device *ndev, struct sk_buff *skb, · line 3600
Quoted 5 lines
drivers/net/ethernet/freescale/fec_main.c +5
-0+if (unlikely (index <0 )) + index =0 ; + +while (index >= fep->num_tx_queues) + index -= fep->num_tx_queues;
Quoted 4 lines
Not a big deal, but I think kind of optimizations are not worthy and potentially dangerous since late '90 ;)
You could consider switching to a simpler '%', but IMHO it's not blocking.
Agree. Will optimize it in next patch of adding xdp statistics.
Thanks,
Shenwei
Quoted 5 lines
Cheers,
Paolo
This patch was applied to netdev/net-next.git (master) by Paolo Abeni <[email protected]>:
Hello:
This patch was applied to netdev/net-next.git (master) by Paolo Abeni <[email protected]>:
Quoted 9 lines · On Mon, 31 Oct 2022 13:53:50 -0500 you wrote:
This patch adds the initial XDP support to Freescale driver. It supports
XDP_PASS,XDP_DROPandXDP_REDIRECTactions. Upcoming patches will add support forXDP_TXand Zero Copy features.As the patch is rather large, the part of codes to collect the statistics is separated and will prepare a dedicated patch for that part.
[...]
Here is the summary with links:
- [v3,1/1] net: fec: add initial XDP support
https://git.kernel.org/netdev/net-next/c/6d6b39f180b8
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html