[dpdk-dev] [PATCH v4 4/7] move rte_eth_dev_check_mq_mode() logic to driver

[Wu, Jingjing]

> -----Original Message-----
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Pawel Wodkowski
> Sent: Thursday, February 19, 2015 11:55 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH v4 4/7] move rte_eth_dev_check_mq_mode()
> logic to driver
> 
> Function rte_eth_dev_check_mq_mode() is driver specific. It should be
> done in PF configuration phase. This patch move igb/ixgbe driver specific mq
> check and SRIOV configuration code to driver part. Also rewriting log
> messages to be shorter and more descriptive.
> 
> Signed-off-by: Pawel Wodkowski <pawelx.wodkowski at intel.com>
> ---
>  lib/librte_ether/rte_ethdev.c       | 197 -----------------------------------
>  lib/librte_pmd_e1000/igb_ethdev.c   |  43 ++++++++
>  lib/librte_pmd_ixgbe/ixgbe_ethdev.c | 105 ++++++++++++++++++-
>  lib/librte_pmd_ixgbe/ixgbe_ethdev.h |   5 +-
>  lib/librte_pmd_ixgbe/ixgbe_pf.c     | 202
> +++++++++++++++++++++++++++++++-----
>  5 files changed, 327 insertions(+), 225 deletions(-)
> 
> diff --git a/lib/librte_pmd_ixgbe/ixgbe_ethdev.c
> b/lib/librte_pmd_ixgbe/ixgbe_ethdev.c
> index 02b9cda..8e9da3b 100644
> --- a/lib/librte_pmd_ixgbe/ixgbe_ethdev.c
> +++ b/lib/librte_pmd_ixgbe/ixgbe_ethdev.c
> @@ -863,7 +863,8 @@ eth_ixgbe_dev_init(__attribute__((unused)) struct
> eth_driver *eth_drv,
>  			"Failed to allocate %u bytes needed to store "
>  			"MAC addresses",
>  			ETHER_ADDR_LEN * hw->mac.num_rar_entries);
> -		return -ENOMEM;
> +		diag = -ENOMEM;
> +		goto error;
>  	}
>  	/* Copy the permanent MAC address */
>  	ether_addr_copy((struct ether_addr *) hw->mac.perm_addr, @@ -
> 876,7 +877,8 @@ eth_ixgbe_dev_init(__attribute__((unused)) struct
> eth_driver *eth_drv,
>  		PMD_INIT_LOG(ERR,
>  			"Failed to allocate %d bytes needed to store MAC
> addresses",
>  			ETHER_ADDR_LEN * IXGBE_VMDQ_NUM_UC_MAC);
> -		return -ENOMEM;
> +		diag = -ENOMEM;
> +		goto error;
>  	}
> 
>  	/* initialize the vfta */
> @@ -886,7 +888,13 @@ eth_ixgbe_dev_init(__attribute__((unused)) struct
> eth_driver *eth_drv,
>  	memset(hwstrip, 0, sizeof(*hwstrip));
> 
>  	/* initialize PF if max_vfs not zero */
> -	ixgbe_pf_host_init(eth_dev);
> +	diag = ixgbe_pf_host_init(eth_dev);
> +	if (diag < 0) {
> +		PMD_INIT_LOG(ERR,
> +			"Failed to allocate %d bytes needed to store MAC
> addresses",
> +			ETHER_ADDR_LEN * IXGBE_VMDQ_NUM_UC_MAC);
> +		goto error;
> +	}
> 
>  	ctrl_ext = IXGBE_READ_REG(hw, IXGBE_CTRL_EXT);
>  	/* let hardware know driver is loaded */ @@ -918,6 +926,11 @@
> eth_ixgbe_dev_init(__attribute__((unused)) struct eth_driver *eth_drv,
>  	ixgbe_enable_intr(eth_dev);
> 
>  	return 0;
> +
> +error:
> +	rte_free(eth_dev->data->hash_mac_addrs);
> +	rte_free(eth_dev->data->mac_addrs);
> +	return diag;
>  }
> 
> 
> @@ -1434,7 +1447,93 @@ ixgbe_dev_configure(struct rte_eth_dev *dev)
>  	struct ixgbe_interrupt *intr =
>  		IXGBE_DEV_PRIVATE_TO_INTR(dev->data->dev_private);
> 
> +	struct rte_eth_conf *dev_conf = &dev->data->dev_conf;
> +	struct rte_eth_dev_info dev_info;
> +	int retval;
> +
>  	PMD_INIT_FUNC_TRACE();
> +	retval = ixgbe_pf_configure_mq_sriov(dev);
> +	if (retval <= 0)
> +		return retval;
> +
> +	uint16_t nb_rx_q = dev->data->nb_rx_queues;
> +	uint16_t nb_tx_q = dev->data->nb_rx_queues;
> +
> +	/* For DCB we need to obtain maximum number of queues
> dinamically,
> +	 * as this depends on max VF exported in PF. */
> +	if ((dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB) ||
> +		(dev_conf->txmode.mq_mode == ETH_MQ_TX_DCB)) {
> +		/* Use dev_infos_get field as this might be pointer to PF or
> VF. */
> +		(*dev->dev_ops->dev_infos_get)(dev, &dev_info);
Why not call ixgbe_dev_info_get directly? And it looks only max_rx_queues and max_tx_queues
are used below, maybe hw->mac.max_rx_queues and hw->mac.max_tx_queues can be used
below instead of calling a function.

> +	}
> +
> +	/* For vmdq+dcb mode check our configuration before we go further
> */
> +	if (dev_conf->rxmode.mq_mode == ETH_MQ_RX_VMDQ_DCB) {
> +		const struct rte_eth_vmdq_dcb_conf *conf;
> +
> +		if (nb_rx_q != ETH_VMDQ_DCB_NUM_QUEUES) {
> +			PMD_INIT_LOG(ERR, " VMDQ+DCB,
> nb_rx_q != %d\n",
> +					ETH_VMDQ_DCB_NUM_QUEUES);
> +			return (-EINVAL);
> +		}
> +		conf = &(dev_conf->rx_adv_conf.vmdq_dcb_conf);
> +		if (conf->nb_queue_pools != ETH_16_POOLS &&
> +				conf->nb_queue_pools != ETH_32_POOLS) {
> +			PMD_INIT_LOG(ERR, " VMDQ+DCB selected, "
> +					"number of RX queue pools must
> be %d or %d\n",
> +					ETH_16_POOLS, ETH_32_POOLS);
> +			return (-EINVAL);
> +		}
> +	} else if (dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB) {
> +		/* For DCB mode check out configuration before we go
> further */
> +		const struct rte_eth_dcb_rx_conf *conf;
> +
> +		if (nb_rx_q != dev_info.max_rx_queues) {
> +			PMD_INIT_LOG(ERR, " DCB, number of RX
> queues != %d\n",
> +					ETH_DCB_NUM_QUEUES);
The check is using dev_info.max_rx_queues, while ETH_DCB_NUM_QUEUES in log.

> +			return (-EINVAL);
> +		}
> +		conf = &(dev_conf->rx_adv_conf.dcb_rx_conf);
> +		if (conf->nb_tcs != ETH_4_TCS &&
> +			   conf->nb_tcs != ETH_8_TCS) {
> +			PMD_INIT_LOG(ERR, " DCB, number of RX TC must
> be %d or %d\n",
> +					ETH_4_TCS, ETH_8_TCS);
> +			return (-EINVAL);
> +		}
> +	}
> +
> +	if (dev_conf->txmode.mq_mode == ETH_MQ_TX_VMDQ_DCB) {
> +		const struct rte_eth_vmdq_dcb_tx_conf *conf;
> +
> +		if (nb_tx_q != ETH_VMDQ_DCB_NUM_QUEUES) {
> +			PMD_INIT_LOG(ERR, " VMDQ+DCB, number of TX
> queues != %d\n",
> +					ETH_VMDQ_DCB_NUM_QUEUES);
> +			return (-EINVAL);
> +		}
> +		conf = &(dev_conf->tx_adv_conf.vmdq_dcb_tx_conf);
> +		if (conf->nb_queue_pools != ETH_16_POOLS &&
> +			   conf->nb_queue_pools != ETH_32_POOLS) {
> +			PMD_INIT_LOG(ERR, " VMDQ+DCB selected, "
> +					"number of TX qqueue pools must
Typo: qqueue->queue

> be %d or %d\n",
> +					ETH_16_POOLS, ETH_32_POOLS);
> +			return (-EINVAL);
> +		}
> +	} else if (dev_conf->txmode.mq_mode == ETH_MQ_TX_DCB) {
> +		const struct rte_eth_dcb_tx_conf *conf;
> +
> +		if (nb_tx_q != dev_info.max_tx_queues) {
> +			PMD_INIT_LOG(ERR, " DCB, number of queues must
> be %d\n",
> +					ETH_DCB_NUM_QUEUES);

The check is using dev_info.max_rx_queues, while ETH_DCB_NUM_QUEUES in log.

> +			return (-EINVAL);
> +		}
> +		conf = &(dev_conf->tx_adv_conf.dcb_tx_conf);
> +		if (conf->nb_tcs != ETH_4_TCS &&
> +			   conf->nb_tcs != ETH_8_TCS) {
> +			PMD_INIT_LOG(ERR, " DCB, number of TX TC must
> be %d or %d\n",
> +					ETH_4_TCS, ETH_8_TCS);
> +			return (-EINVAL);
> +		}
> +	}
> 
>  	/* set flag to update link status after init */
>  	intr->flags |= IXGBE_FLAG_NEED_LINK_UPDATE; diff --git
> a/lib/librte_pmd_ixgbe/ixgbe_ethdev.h
> b/lib/librte_pmd_ixgbe/ixgbe_ethdev.h
> index 1383194..e70a6e8 100644
> --- a/lib/librte_pmd_ixgbe/ixgbe_ethdev.h
> +++ b/lib/librte_pmd_ixgbe/ixgbe_ethdev.h
> @@ -348,11 +348,14 @@ void ixgbe_vlan_hw_strip_enable_all(struct
> rte_eth_dev *dev);
> 
>  void ixgbe_vlan_hw_strip_disable_all(struct rte_eth_dev *dev);
> 
> -void ixgbe_pf_host_init(struct rte_eth_dev *eth_dev);
> +int ixgbe_pf_host_init(struct rte_eth_dev *eth_dev);
> 
>  void ixgbe_pf_mbx_process(struct rte_eth_dev *eth_dev);
> 
> +int ixgbe_pf_configure_mq_sriov(struct rte_eth_dev *dev);
> +
>  int ixgbe_pf_host_configure(struct rte_eth_dev *eth_dev);
> 
>  uint32_t ixgbe_convert_vm_rx_mask_to_val(uint16_t rx_mask, uint32_t
> orig_val);
> +
>  #endif /* _IXGBE_ETHDEV_H_ */
> diff --git a/lib/librte_pmd_ixgbe/ixgbe_pf.c
> b/lib/librte_pmd_ixgbe/ixgbe_pf.c index 4103e97..a7b9333 100644
> --- a/lib/librte_pmd_ixgbe/ixgbe_pf.c
> +++ b/lib/librte_pmd_ixgbe/ixgbe_pf.c
> @@ -91,7 +91,7 @@ ixgbe_mb_intr_setup(struct rte_eth_dev *dev)
>  	return 0;
>  }
> 
> -void ixgbe_pf_host_init(struct rte_eth_dev *eth_dev)
> +int ixgbe_pf_host_init(struct rte_eth_dev *eth_dev)
>  {
>  	struct ixgbe_vf_info **vfinfo =
>  		IXGBE_DEV_PRIVATE_TO_P_VFDATA(eth_dev->data-
> >dev_private);
> @@ -101,39 +101,31 @@ void ixgbe_pf_host_init(struct rte_eth_dev
> *eth_dev)
>          IXGBE_DEV_PRIVATE_TO_UTA(eth_dev->data->dev_private);
>  	struct ixgbe_hw *hw =
>  		IXGBE_DEV_PRIVATE_TO_HW(eth_dev->data->dev_private);
> +	int retval;
>  	uint16_t vf_num;
> -	uint8_t nb_queue;
> 
>  	PMD_INIT_FUNC_TRACE();
> 
> -	RTE_ETH_DEV_SRIOV(eth_dev).active = 0;
> -	if (0 == (vf_num = dev_num_vf(eth_dev)))
> -		return;
> +	/* Fill sriov structure using default configuration. */
> +	retval = ixgbe_pf_configure_mq_sriov(eth_dev);
> +	if (retval != 0) {
> +		if (retval < 0)
> +			PMD_INIT_LOG(ERR, " Setting up SRIOV with default
> device "
> +					"configuration should not fail. This is a
> BUG.");
> +		return 0;
> +	}
> 
> +	vf_num = dev_num_vf(eth_dev);
>  	*vfinfo = rte_zmalloc("vf_info", sizeof(struct ixgbe_vf_info) *
> vf_num, 0);
> -	if (*vfinfo == NULL)
> -		rte_panic("Cannot allocate memory for private VF data\n");
> +	if (*vfinfo == NULL) {
> +		PMD_INIT_LOG(ERR, "Cannot allocate memory for private VF
> data.");
> +		return (-ENOMEM);
> +	}
> 
>  	memset(mirror_info,0,sizeof(struct ixgbe_mirror_info));
>  	memset(uta_info,0,sizeof(struct ixgbe_uta_info));
>  	hw->mac.mc_filter_type = 0;
> 
> -	if (vf_num >= ETH_32_POOLS) {
> -		nb_queue = 2;
> -		RTE_ETH_DEV_SRIOV(eth_dev).active = ETH_64_POOLS;
> -	} else if (vf_num >= ETH_16_POOLS) {
> -		nb_queue = 4;
> -		RTE_ETH_DEV_SRIOV(eth_dev).active = ETH_32_POOLS;
> -	} else {
> -		nb_queue = 8;
> -		RTE_ETH_DEV_SRIOV(eth_dev).active = ETH_16_POOLS;
> -	}
> -
> -	RTE_ETH_DEV_SRIOV(eth_dev).nb_rx_q_per_pool = nb_queue;
> -	RTE_ETH_DEV_SRIOV(eth_dev).nb_tx_q_per_pool = nb_queue;
> -	RTE_ETH_DEV_SRIOV(eth_dev).def_vmdq_idx = vf_num;
> -	RTE_ETH_DEV_SRIOV(eth_dev).def_pool_q_idx =
> (uint16_t)(vf_num * nb_queue);
> -
>  	ixgbe_vf_perm_addr_gen(eth_dev, vf_num);
> 
>  	/* init_mailbox_params */
> @@ -142,7 +134,169 @@ void ixgbe_pf_host_init(struct rte_eth_dev
> *eth_dev)
>  	/* set mb interrupt mask */
>  	ixgbe_mb_intr_setup(eth_dev);
> 
> -	return;
> +	return 0;
> +}
> +
> +
> +/*
> + * Function that make SRIOV configuration, based on device
> +configuration,
> + * number of requested queues and number of VF created.
> + * Function returns:
> + * 1 - SRIOV is not enabled (no VF created)
> + * 0 - proper SRIOV configuration found.
> + * -EINVAL - no suitable SRIOV configuration found.
> + */
> +int
> +ixgbe_pf_configure_mq_sriov(struct rte_eth_dev *dev) {

If this function is called by ixgbe_pf_host_init. It is in the initialization process, the
dev_conf in data is meaningless. The following check is still necessary? Maybe it's
better to use different functions in configure and init phase.

> +	struct rte_eth_conf *dev_conf = &dev->data->dev_conf;
> +	struct rte_eth_dev_sriov *sriov = &RTE_ETH_DEV_SRIOV(dev);
> +	uint16_t vf_num;
> +
> +	vf_num = dev_num_vf(dev);
> +	if (vf_num == 0) {
> +		memset(sriov, 0, sizeof(*sriov));
> +		return 1;
> +	}
> +
> +	/* Check multi-queue mode. */
> +	if ((dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB) ||
> +		(dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB_RSS)
> ||
> +		(dev_conf->txmode.mq_mode == ETH_MQ_TX_DCB)) {
> +		/* SRIOV only works in VMDq enable mode */
> +		PMD_INIT_LOG(ERR, " SRIOV active, "
> +				"invlaid VMDQ rx mode (%u) or tx (%u)
> mode.",
> +				dev_conf->rxmode.mq_mode, dev_conf-
> >txmode.mq_mode);
> +		return (-EINVAL);
> +	}
> +
> +	switch (dev_conf->rxmode.mq_mode) {
> +	case ETH_MQ_RX_VMDQ_DCB:
> +		if (vf_num <= ETH_16_POOLS)
> +			sriov->nb_rx_q_per_pool = 8;
> +		else if (vf_num <= ETH_32_POOLS)
> +			sriov->nb_rx_q_per_pool = 4;
> +		else {
> +			PMD_INIT_LOG(ERR,
> +				"DCB (SRIOV active) - VF count (%d) must be
> less or equal 32.",
> +				vf_num);
> +			return (-EINVAL);
> +		}
> +
> +		if (dev->data->nb_rx_queues < sriov->nb_rx_q_per_pool) {
> +			PMD_INIT_LOG(WARNING,
> +				"DCB (SRIOV active) rx queues (%d) count is
> not equal %d.",
> +				dev->data->nb_rx_queues,
> +				sriov->nb_rx_q_per_pool);
> +		}
> +		break;
> +	case ETH_MQ_RX_RSS:
> +		PMD_INIT_LOG(INFO, "RSS (SRIOV active), "
> +				"rx mq mode is changed from: mq_mode %u
> into VMDQ mq_mode %u.",
> +				dev_conf->rxmode.mq_mode, dev->data-
> >dev_conf.rxmode.mq_mode);
> +		dev->data->dev_conf.rxmode.mq_mode =
> ETH_MQ_RX_VMDQ_RSS;
> +		/* falltrought */
> +	case ETH_MQ_RX_VMDQ_RSS:
> +		if (vf_num >= ETH_64_POOLS) {
> +			/* FIXME: Is vf_num > 64 realy supported by
> hardware? */
> +			PMD_INIT_LOG(ERR, "RSS (SRIOV active), "
> +					"VFs num must be less or equal 64.");
> +				return (-EINVAL);
> +		} else if (vf_num >= ETH_32_POOLS) {
> +			if (dev->data->nb_rx_queues != 1 && dev->data-
> >nb_rx_queues != 2) {
> +				PMD_INIT_LOG(ERR, "RSS (SRIOV active, VF
> count >= 32),"
> +						"invalid rx queues count %d.
> It must be 1 or 2.",
> +						dev->data->nb_rx_queues);
> +				return (-EINVAL);
> +			}
> +
> +			sriov->nb_rx_q_per_pool = dev->data-
> >nb_rx_queues;
> +		} else {
> +			/* FIXME: is VT(16) + RSS realy supported? */
> +			if (dev->data->nb_rx_queues != 4) {
> +				PMD_INIT_LOG(ERR, "RSS (SRIOV active, VFs
> count < 32), "
> +						"invalid rx queues count %d.
> It must be 4.",
> +						dev->data->nb_rx_queues);
> +				return (-EINVAL);
> +			}
> +
> +			sriov->nb_rx_q_per_pool = 4;
> +		}
> +		break;
> +	default: /* ETH_MQ_RX_VMDQ_ONLY or ETH_MQ_RX_NONE */
> +		/* if nothing mq mode configure, use default scheme */
> +		if (dev->data->dev_conf.rxmode.mq_mode !=
> ETH_MQ_RX_VMDQ_ONLY) {
> +			PMD_INIT_LOG(INFO, "Rx mq mode changed
> from %u into VMDQ %u.",
> +					dev->data-
> >dev_conf.rxmode.mq_mode, ETH_MQ_RX_VMDQ_ONLY);
> +
> +			dev->data->dev_conf.rxmode.mq_mode =
> ETH_MQ_RX_VMDQ_ONLY;
> +		}
> +
> +		/* queue 0 of each pool is used. */
> +		sriov->nb_rx_q_per_pool = 1;
> +		break;
> +	}
> +
> +	switch (dev_conf->txmode.mq_mode) {
> +	case ETH_MQ_TX_VMDQ_DCB: /* DCB VMDQ in SRIOV mode*/
> +		if (vf_num <= ETH_16_POOLS)
> +			sriov->nb_tx_q_per_pool = 8;
> +		else if (vf_num <= ETH_32_POOLS)
> +			sriov->nb_tx_q_per_pool = 4;
> +		else if (vf_num <= ETH_64_POOLS)
> +			sriov->nb_tx_q_per_pool = 1;
> +		else {
> +			PMD_INIT_LOG(ERR, "DCB (SRIOV active), "
> +					"VF count (%d) must be less or equal
> 64.",
> +				vf_num);
> +			return (-EINVAL);
> +		}
> +		break;
> +	default: /* ETH_MQ_TX_VMDQ_ONLY or ETH_MQ_TX_NONE */
> +		/* if nothing mq mode configure, use default scheme */
> +		if (dev->data->dev_conf.txmode.mq_mode !=
> ETH_MQ_TX_VMDQ_ONLY) {
> +			PMD_INIT_LOG(INFO, "Tx mq mode is changed
> from %u into VMDQ %u.",
> +					dev->data-
> >dev_conf.txmode.mq_mode, ETH_MQ_TX_VMDQ_ONLY);
> +
> +			dev->data->dev_conf.txmode.mq_mode =
> ETH_MQ_TX_VMDQ_ONLY;
> +		}
> +
> +		/* queue 0 of each pool is used. */
> +		sriov->nb_tx_q_per_pool = 1;
> +		break;
> +	}
> +
> +	sriov->def_vmdq_idx = vf_num;
> +
> +	/*
> +	 * Pools starts at 2xN, 4xN or 8xN
> +	 */
> +	if (vf_num >= ETH_32_POOLS) {
> +		/* This must be vf_num <= ETH_64_POOLS */
> +		sriov->active = ETH_64_POOLS;
> +		sriov->def_pool_q_idx = vf_num * 2;
> +	} else if (vf_num >= ETH_16_POOLS) {
> +		sriov->active = ETH_32_POOLS;
> +		sriov->def_pool_q_idx = vf_num * 4;
> +	} else {
> +		sriov->active = ETH_16_POOLS;
> +		sriov->def_pool_q_idx = vf_num * 8;
> +	}
> +
> +	/* Check if available queus count is not less than allocated.*/
> +	if (dev->data->nb_rx_queues > sriov->nb_rx_q_per_pool) {
> +		PMD_INIT_LOG(ERR, "SRIOV active, rx queue count must
> less or equal %d.",
> +						sriov->nb_rx_q_per_pool);
> +				return (-EINVAL);
> +	}
> +
> +	if (dev->data->nb_rx_queues > sriov->nb_tx_q_per_pool) {
> +		PMD_INIT_LOG(ERR, "SRIOV active, tx queue count must
> less or equal %d.",
> +				sriov->nb_tx_q_per_pool);
> +		return (-EINVAL);
> +	}
> +
> +	return 0;
>  }
> 
>  int ixgbe_pf_host_configure(struct rte_eth_dev *eth_dev)
> --
> 1.9.1