[dpdk-dev] [PATCH v6 5/6] enicpmd: DPDK-ENIC PMD interface
Hi, ssujith > -Original Message- > From: Sujith Sankar (ssujith) [mailto:ssujith at cisco.com] > Sent: Tuesday, December 30, 2014 12:46 PM > To: Wu, Jingjing; dev at dpdk.org > Cc: Prasad Rao (prrao) > Subject: Re: [dpdk-dev] [PATCH v6 5/6] enicpmd: DPDK-ENIC PMD interface > > > > On 29/12/14 1:45 pm, "Wu, Jingjing" wrote: > > >Hi, ssujith > > > >> + .tx_queue_release = enicpmd_dev_tx_queue_release, > >> + .dev_led_on = NULL, > >> + .dev_led_off = NULL, > >> + .flow_ctrl_get= NULL, > >> + .flow_ctrl_set= NULL, > >> + .priority_flow_ctrl_set = NULL, > >> + .mac_addr_add = enicpmd_add_mac_addr, > >> + .mac_addr_remove = enicpmd_remove_mac_addr, > >> + .fdir_add_signature_filter= NULL, > >> + .fdir_update_signature_filter = NULL, > >> + .fdir_remove_signature_filter = NULL, > >> + .fdir_infos_get = enicpmd_fdir_info_get, > >> + .fdir_add_perfect_filter = enicpmd_fdir_add_perfect_filter, > >> + .fdir_update_perfect_filter = enicpmd_fdir_add_perfect_filter, > >> + .fdir_remove_perfect_filter = enicpmd_fdir_remove_perfect_filter, > >> + .fdir_set_masks = NULL, > >> +}; > >> + > > > >I found that in perfect fdir is also supported in enic driver. > > > >During the R1.8 development, we defined a new dev_ops call filter_ctrl, > >which can be used to control kinds of filters, flow director is > >included too. Which is mentioned in > >http://www.dpdk.org/ml/archives/dev/2014-September/005179.html . > >In R1.8, filter_ctrl is only used by i40e driver. And we also planned > >use it in the existing ixgbe/e1000 driver in the next days. The old > >APIs such as fdir_add_perfect_filter, fdir_remove_perfect_filter can be > >replaced then. > > > Hi Jingjing, > Thanks for the info and the link. I shall take a look at it. > It looks like bringing in one interface for all filter related operations. > I believe ENIC should also move to it. > > Thanks, > -Sujith > Just let you know that, I already sent the patch to migrate the flow director in ixgbe driver to new filter_ctrl API. http://www.dpdk.org/ml/archives/dev/2015-January/011830.html To avoid compile error and influence to enic driver. I didn't remove the old APIs and structures in rte_ethdev. I think they can be removed when the migration in enic driver is done. Do you have any plan for this? Thank u! Jingjing > > > >So, do you have any plan to migrate the fdir in enic to the filter_ctrl > >API? > > > >Jingjing > > > >Thanks! > > > >> +struct enic *enicpmd_list_head = NULL; > >> +/* Initialize the driver > >> + * It returns 0 on success. > >> + */ > >> +static int eth_enicpmd_dev_init( > >> + __attribute__((unused))struct eth_driver *eth_drv, > >> + struct rte_eth_dev *eth_dev) > >> +{ > >> + struct rte_pci_device *pdev; > >> + struct rte_pci_addr *addr; > >> + struct enic *enic = pmd_priv(eth_dev); > >> + > >> + ENICPMD_FUNC_TRACE(); > >> + > >> + enic->rte_dev = eth_dev; > >> + eth_dev->dev_ops = _eth_dev_ops; > >> + eth_dev->rx_pkt_burst = _recv_pkts; > >> + eth_dev->tx_pkt_burst = _xmit_pkts; > >> + > >> + pdev = eth_dev->pci_dev; > >> + enic->pdev = pdev; > >> + addr = >addr; > >> + > >> + snprintf(enic->bdf_name, ENICPMD_BDF_LENGTH, > >> "%04x:%02x:%02x.%x", > >> + addr->domain, addr->bus, addr->devid, addr->function); > >> + > >> + return enic_probe(enic); > >> +} > >> + > >> +static struct eth_driver rte_enic_pmd = { > >> + { > >> + .name = "rte_enic_pmd", > >> + .id_table = pci_id_enic_map, > >> + .drv_flags = RTE_PCI_DRV_NEED_MAPPING, > >> + }, > >> + .eth_dev_init = eth_enicpmd_dev_init, > >> + .dev_private_size = sizeof(struct enic), }; > >> + > >> +/* Driver initialization routine. > >> + * Invoked once at EAL init time. > >> + * Register as the [Poll Mode] Driver of Cisco ENIC device. > >> + */ > >> +int rte_enic_pmd_init(const char *name __rte_unused, > >> + const char *params __rte_unused) > >> +{ > >> + ENICPMD_FUNC_TRACE(); > >> + > >> + rte_eth_driver_register(_enic_pmd); > >> + return 0; > >> +} > >> + > >> +static struct rte_driver rte_enic_driver = { > >> + .type = PMD_PDEV, > >> + .init = rte_enic_pmd_init, > >> +}; > >> + > >> +PMD_REGISTER_DRIVER(rte_enic_driver); > >> + > >> -- > >> 1.9.1 > >
[dpdk-dev] [PATCH v6 5/6] enicpmd: DPDK-ENIC PMD interface
2014-12-30 04:45, Sujith Sankar: > On 29/12/14 1:45 pm, "Wu, Jingjing" wrote: > >I found that in perfect fdir is also supported in enic driver. > > > >During the R1.8 development, we defined a new dev_ops call filter_ctrl, > >which can be used to control kinds of filters, flow director is included > >too. Which is mentioned in > >http://www.dpdk.org/ml/archives/dev/2014-September/005179.html . > >In R1.8, filter_ctrl is only used by i40e driver. And we also planned use > >it in the existing ixgbe/e1000 driver in the next days. The old APIs such > >as fdir_add_perfect_filter, fdir_remove_perfect_filter can be replaced > >then. > > Hi Jingjing, > Thanks for the info and the link. I shall take a look at it. > It looks like bringing in one interface for all filter related operations. > I believe ENIC should also move to it. Yes please. It is planned to remove or deprecate old flow director API. -- Thomas
[dpdk-dev] [PATCH v6 5/6] enicpmd: DPDK-ENIC PMD interface
On 29/12/14 1:45 pm, "Wu, Jingjing" wrote: >Hi, ssujith > >> +.tx_queue_release = enicpmd_dev_tx_queue_release, >> +.dev_led_on = NULL, >> +.dev_led_off = NULL, >> +.flow_ctrl_get= NULL, >> +.flow_ctrl_set= NULL, >> +.priority_flow_ctrl_set = NULL, >> +.mac_addr_add = enicpmd_add_mac_addr, >> +.mac_addr_remove = enicpmd_remove_mac_addr, >> +.fdir_add_signature_filter= NULL, >> +.fdir_update_signature_filter = NULL, >> +.fdir_remove_signature_filter = NULL, >> +.fdir_infos_get = enicpmd_fdir_info_get, >> +.fdir_add_perfect_filter = enicpmd_fdir_add_perfect_filter, >> +.fdir_update_perfect_filter = enicpmd_fdir_add_perfect_filter, >> +.fdir_remove_perfect_filter = enicpmd_fdir_remove_perfect_filter, >> +.fdir_set_masks = NULL, >> +}; >> + > >I found that in perfect fdir is also supported in enic driver. > >During the R1.8 development, we defined a new dev_ops call filter_ctrl, >which can be used to control kinds of filters, flow director is included >too. Which is mentioned in >http://www.dpdk.org/ml/archives/dev/2014-September/005179.html . >In R1.8, filter_ctrl is only used by i40e driver. And we also planned use >it in the existing ixgbe/e1000 driver in the next days. The old APIs such >as fdir_add_perfect_filter, fdir_remove_perfect_filter can be replaced >then. Hi Jingjing, Thanks for the info and the link. I shall take a look at it. It looks like bringing in one interface for all filter related operations. I believe ENIC should also move to it. Thanks, -Sujith > >So, do you have any plan to migrate the fdir in enic to the filter_ctrl >API? > >Jingjing > >Thanks! > >> +struct enic *enicpmd_list_head = NULL; >> +/* Initialize the driver >> + * It returns 0 on success. >> + */ >> +static int eth_enicpmd_dev_init( >> +__attribute__((unused))struct eth_driver *eth_drv, >> +struct rte_eth_dev *eth_dev) >> +{ >> +struct rte_pci_device *pdev; >> +struct rte_pci_addr *addr; >> +struct enic *enic = pmd_priv(eth_dev); >> + >> +ENICPMD_FUNC_TRACE(); >> + >> +enic->rte_dev = eth_dev; >> +eth_dev->dev_ops = _eth_dev_ops; >> +eth_dev->rx_pkt_burst = _recv_pkts; >> +eth_dev->tx_pkt_burst = _xmit_pkts; >> + >> +pdev = eth_dev->pci_dev; >> +enic->pdev = pdev; >> +addr = >addr; >> + >> +snprintf(enic->bdf_name, ENICPMD_BDF_LENGTH, >> "%04x:%02x:%02x.%x", >> +addr->domain, addr->bus, addr->devid, addr->function); >> + >> +return enic_probe(enic); >> +} >> + >> +static struct eth_driver rte_enic_pmd = { >> +{ >> +.name = "rte_enic_pmd", >> +.id_table = pci_id_enic_map, >> +.drv_flags = RTE_PCI_DRV_NEED_MAPPING, >> +}, >> +.eth_dev_init = eth_enicpmd_dev_init, >> +.dev_private_size = sizeof(struct enic), }; >> + >> +/* Driver initialization routine. >> + * Invoked once at EAL init time. >> + * Register as the [Poll Mode] Driver of Cisco ENIC device. >> + */ >> +int rte_enic_pmd_init(const char *name __rte_unused, >> +const char *params __rte_unused) >> +{ >> +ENICPMD_FUNC_TRACE(); >> + >> +rte_eth_driver_register(_enic_pmd); >> +return 0; >> +} >> + >> +static struct rte_driver rte_enic_driver = { >> +.type = PMD_PDEV, >> +.init = rte_enic_pmd_init, >> +}; >> + >> +PMD_REGISTER_DRIVER(rte_enic_driver); >> + >> -- >> 1.9.1 >
[dpdk-dev] [PATCH v6 5/6] enicpmd: DPDK-ENIC PMD interface
Hi, ssujith > + .tx_queue_release = enicpmd_dev_tx_queue_release, > + .dev_led_on = NULL, > + .dev_led_off = NULL, > + .flow_ctrl_get= NULL, > + .flow_ctrl_set= NULL, > + .priority_flow_ctrl_set = NULL, > + .mac_addr_add = enicpmd_add_mac_addr, > + .mac_addr_remove = enicpmd_remove_mac_addr, > + .fdir_add_signature_filter= NULL, > + .fdir_update_signature_filter = NULL, > + .fdir_remove_signature_filter = NULL, > + .fdir_infos_get = enicpmd_fdir_info_get, > + .fdir_add_perfect_filter = enicpmd_fdir_add_perfect_filter, > + .fdir_update_perfect_filter = enicpmd_fdir_add_perfect_filter, > + .fdir_remove_perfect_filter = enicpmd_fdir_remove_perfect_filter, > + .fdir_set_masks = NULL, > +}; > + I found that in perfect fdir is also supported in enic driver. During the R1.8 development, we defined a new dev_ops call filter_ctrl, which can be used to control kinds of filters, flow director is included too. Which is mentioned in http://www.dpdk.org/ml/archives/dev/2014-September/005179.html . In R1.8, filter_ctrl is only used by i40e driver. And we also planned use it in the existing ixgbe/e1000 driver in the next days. The old APIs such as fdir_add_perfect_filter, fdir_remove_perfect_filter can be replaced then. So, do you have any plan to migrate the fdir in enic to the filter_ctrl API? Jingjing Thanks! > +struct enic *enicpmd_list_head = NULL; > +/* Initialize the driver > + * It returns 0 on success. > + */ > +static int eth_enicpmd_dev_init( > + __attribute__((unused))struct eth_driver *eth_drv, > + struct rte_eth_dev *eth_dev) > +{ > + struct rte_pci_device *pdev; > + struct rte_pci_addr *addr; > + struct enic *enic = pmd_priv(eth_dev); > + > + ENICPMD_FUNC_TRACE(); > + > + enic->rte_dev = eth_dev; > + eth_dev->dev_ops = _eth_dev_ops; > + eth_dev->rx_pkt_burst = _recv_pkts; > + eth_dev->tx_pkt_burst = _xmit_pkts; > + > + pdev = eth_dev->pci_dev; > + enic->pdev = pdev; > + addr = >addr; > + > + snprintf(enic->bdf_name, ENICPMD_BDF_LENGTH, > "%04x:%02x:%02x.%x", > + addr->domain, addr->bus, addr->devid, addr->function); > + > + return enic_probe(enic); > +} > + > +static struct eth_driver rte_enic_pmd = { > + { > + .name = "rte_enic_pmd", > + .id_table = pci_id_enic_map, > + .drv_flags = RTE_PCI_DRV_NEED_MAPPING, > + }, > + .eth_dev_init = eth_enicpmd_dev_init, > + .dev_private_size = sizeof(struct enic), }; > + > +/* Driver initialization routine. > + * Invoked once at EAL init time. > + * Register as the [Poll Mode] Driver of Cisco ENIC device. > + */ > +int rte_enic_pmd_init(const char *name __rte_unused, > + const char *params __rte_unused) > +{ > + ENICPMD_FUNC_TRACE(); > + > + rte_eth_driver_register(_enic_pmd); > + return 0; > +} > + > +static struct rte_driver rte_enic_driver = { > + .type = PMD_PDEV, > + .init = rte_enic_pmd_init, > +}; > + > +PMD_REGISTER_DRIVER(rte_enic_driver); > + > -- > 1.9.1
[dpdk-dev] [PATCH v6 5/6] enicpmd: DPDK-ENIC PMD interface
Signed-off-by: Sujith Sankar --- lib/librte_pmd_enic/enic_etherdev.c | 613 1 file changed, 613 insertions(+) create mode 100644 lib/librte_pmd_enic/enic_etherdev.c diff --git a/lib/librte_pmd_enic/enic_etherdev.c b/lib/librte_pmd_enic/enic_etherdev.c new file mode 100644 index 000..d0577aa --- /dev/null +++ b/lib/librte_pmd_enic/enic_etherdev.c @@ -0,0 +1,613 @@ +/* + * Copyright 2008-2014 Cisco Systems, Inc. All rights reserved. + * Copyright 2007 Nuova Systems, Inc. All rights reserved. + * + * Copyright (c) 2014, Cisco Systems, Inc. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS + * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE + * COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, + * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; + * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER + * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN + * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + * + */ +#ident "$Id$" + +#include +#include + +#include +#include +#include +#include + +#include "vnic_intr.h" +#include "vnic_cq.h" +#include "vnic_wq.h" +#include "vnic_rq.h" +#include "vnic_enet.h" +#include "enic.h" + +#define ENICPMD_FUNC_TRACE() \ + RTE_LOG(DEBUG, PMD, "ENICPMD trace: %s\n", __func__) + +/* + * The set of PCI devices this driver supports + */ +static struct rte_pci_id pci_id_enic_map[] = { +#define RTE_PCI_DEV_ID_DECL_ENIC(vend, dev) {RTE_PCI_DEVICE(vend, dev)}, +#ifndef PCI_VENDOR_ID_CISCO +#define PCI_VENDOR_ID_CISCO0x1137 +#endif +#include "rte_pci_dev_ids.h" +RTE_PCI_DEV_ID_DECL_ENIC(PCI_VENDOR_ID_CISCO, PCI_DEVICE_ID_CISCO_VIC_ENET) +RTE_PCI_DEV_ID_DECL_ENIC(PCI_VENDOR_ID_CISCO, PCI_DEVICE_ID_CISCO_VIC_ENET_VF) +{.vendor_id = 0, /* Sentinal */}, +}; + +static int enicpmd_fdir_remove_perfect_filter(struct rte_eth_dev *eth_dev, + struct rte_fdir_filter *fdir_filter, + uint16_t soft_id) +{ + struct enic *enic = pmd_priv(eth_dev); + + ENICPMD_FUNC_TRACE(); + return enic_fdir_del_fltr(enic, fdir_filter); +} + +static int enicpmd_fdir_add_perfect_filter(struct rte_eth_dev *eth_dev, + struct rte_fdir_filter *fdir_filter, uint16_t soft_id, + uint8_t queue, uint8_t drop) +{ + struct enic *enic = pmd_priv(eth_dev); + + ENICPMD_FUNC_TRACE(); + return enic_fdir_add_fltr(enic, fdir_filter, (uint16_t)queue, drop); +} + +static void enicpmd_fdir_info_get(struct rte_eth_dev *eth_dev, + struct rte_eth_fdir *fdir) +{ + struct enic *enic = pmd_priv(eth_dev); + + ENICPMD_FUNC_TRACE(); + *fdir = enic->fdir.stats; +} + +static void enicpmd_dev_tx_queue_release(void *txq) +{ + ENICPMD_FUNC_TRACE(); + enic_free_wq(txq); +} + +static int enicpmd_dev_setup_intr(struct enic *enic) +{ + int ret; + int index; + + ENICPMD_FUNC_TRACE(); + + /* Are we done with the init of all the queues? */ + for (index = 0; index < enic->cq_count; index++) { + if (!enic->cq[index].ctrl) + break; + } + + if (enic->cq_count != index) + return 0; + + ret = enic_alloc_intr_resources(enic); + if (ret) { + dev_err(enic, "alloc intr failed\n"); + return ret; + } + enic_init_vnic_resources(enic); + + ret = enic_setup_finish(enic); + if (ret) + dev_err(enic, "setup could not be finished\n"); + + return ret; +} + +static int enicpmd_dev_tx_queue_setup(struct rte_eth_dev *eth_dev, + uint16_t queue_idx, + uint16_t nb_desc, + unsigned int socket_id, + const struct rte_eth_txconf *tx_conf) +{ + int ret; + struct enic *enic = pmd_priv(eth_dev); + + ENICPMD_FUNC_TRACE(); + eth_dev->data->tx_queues[queue_idx] = (void *)>wq[queue_idx]; + + ret = enic_alloc_wq(enic, queue_idx, socket_id, nb_desc); + if (ret) { + dev_err(enic, "error in allocating wq\n"); +