[dpdk-dev] [PATCH 1/4] xen: allow choosing dom0 support at runtime
Liu, Jijiang
jijiang.liu at intel.com
Sun Feb 15 05:07:21 CET 2015
Hi Stephen,
What do you mean ' allow choosing dom0 support at runtime'?
If you mean user can choose DPDK to run Xen Dom0 or not on DOM0 by a runtime flag, I don't think your change can achieve this goal.
Thanks
Jijiang Liu
> -----Original Message-----
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Stephen Hemminger
> Sent: Sunday, February 15, 2015 2:07 AM
> To: dev at dpdk.org
> Cc: Stephen Hemminger
> Subject: [dpdk-dev] [PATCH 1/4] xen: allow choosing dom0 support at runtime
>
> The previous code would only allow building library and application so that it
> ran on Xen DOM0 or not on DOM0. This changes that to a runtime flag.
>
> Signed-off-by: Stephen Hemminger <stephen at networkplumber.org>
> ---
> lib/librte_eal/common/include/rte_memory.h | 4 +++
> lib/librte_eal/linuxapp/eal/eal_memory.c | 7 ++++
> lib/librte_ether/rte_ethdev.c | 22 ++++++++++++
> lib/librte_ether/rte_ethdev.h | 23 ++++++++++++
> lib/librte_mempool/rte_mempool.c | 26 +++++++-------
> lib/librte_pmd_e1000/em_rxtx.c | 30 +++-------------
> lib/librte_pmd_e1000/igb_rxtx.c | 52 +++++++++------------------
> lib/librte_pmd_ixgbe/ixgbe_rxtx.c | 58 +++++++++---------------------
> 8 files changed, 108 insertions(+), 114 deletions(-)
>
> diff --git a/lib/librte_eal/common/include/rte_memory.h
> b/lib/librte_eal/common/include/rte_memory.h
> index 7f8103f..ab6c1ff 100644
> --- a/lib/librte_eal/common/include/rte_memory.h
> +++ b/lib/librte_eal/common/include/rte_memory.h
> @@ -176,6 +176,10 @@ unsigned rte_memory_get_nchannel(void); unsigned
> rte_memory_get_nrank(void);
>
> #ifdef RTE_LIBRTE_XEN_DOM0
> +
> +/**< Internal use only - should DOM0 memory mapping be used */ extern
> +int is_xen_dom0_supported(void);
> +
> /**
> * Return the physical address of elt, which is an element of the pool mp.
> *
> diff --git a/lib/librte_eal/linuxapp/eal/eal_memory.c
> b/lib/librte_eal/linuxapp/eal/eal_memory.c
> index a67a1b0..4afda2a 100644
> --- a/lib/librte_eal/linuxapp/eal/eal_memory.c
> +++ b/lib/librte_eal/linuxapp/eal/eal_memory.c
> @@ -98,6 +98,13 @@
> #include "eal_filesystem.h"
> #include "eal_hugepages.h"
>
> +#ifdef RTE_LIBRTE_XEN_DOM0
> +int is_xen_dom0_supported(void)
> +{
> + return internal_config.xen_dom0_support; } #endif
> +
> /**
> * @file
> * Huge page mapping under linux
> diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c index
> ea3a1fb..457e0bc 100644
> --- a/lib/librte_ether/rte_ethdev.c
> +++ b/lib/librte_ether/rte_ethdev.c
> @@ -2825,6 +2825,27 @@ _rte_eth_dev_callback_process(struct rte_eth_dev
> *dev,
> }
> rte_spinlock_unlock(&rte_eth_dev_cb_lock);
> }
> +
> +const struct rte_memzone *
> +rte_eth_dma_zone_reserve(const struct rte_eth_dev *dev, const char
> *ring_name,
> + uint16_t queue_id, size_t size, unsigned align,
> + int socket_id)
> +{
> + char z_name[RTE_MEMZONE_NAMESIZE];
> + const struct rte_memzone *mz;
> +
> + snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
> + dev->driver->pci_drv.name, ring_name,
> + dev->data->port_id, queue_id);
> +
> + mz = rte_memzone_lookup(z_name);
> + if (mz)
> + return mz;
> +
> + return rte_memzone_reserve_bounded(z_name, size,
> + socket_id, 0, align,
> RTE_PGSIZE_2M); }
> +
> #ifdef RTE_NIC_BYPASS
> int rte_eth_dev_bypass_init(uint8_t port_id) { @@ -3003,6 +3024,7 @@
> rte_eth_dev_bypass_wd_reset(uint8_t port_id)
> (*dev->dev_ops->bypass_wd_reset)(dev);
> return 0;
> }
> +
> #endif
>
> int
> diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h index
> 1200c1c..747acb5 100644
> --- a/lib/librte_ether/rte_ethdev.h
> +++ b/lib/librte_ether/rte_ethdev.h
> @@ -3664,6 +3664,29 @@ int rte_eth_dev_filter_supported(uint8_t port_id,
> enum rte_filter_type filter_ty int rte_eth_dev_filter_ctrl(uint8_t port_id,
> enum rte_filter_type filter_type,
> enum rte_filter_op filter_op, void *arg);
>
> +/**
> + * Create memzone for HW rings.
> + * malloc can't be used as the physical address is needed.
> + * If the memzone is already created, then this function returns a ptr
> + * to the old one.
> + *
> + * @param eth_dev
> + * The *eth_dev* pointer is the address of the *rte_eth_dev* structure
> + * @param name
> + * The name of the memory zone
> + * @param queue_id
> + * The index of the queue to add to name
> + * @param size
> + * The sizeof of the memory area
> + * @param align
> + * Alignment for resulting memzone. Must be a power of 2.
> + * @param socket_id
> + * The *socket_id* argument is the socket identifier in case of NUMA.
> + */
> +const struct rte_memzone *
> +rte_eth_dma_zone_reserve(const struct rte_eth_dev *eth_dev, const char
> *name,
> + uint16_t queue_id, size_t size,
> + unsigned align, int socket_id);
> #ifdef __cplusplus
> }
> #endif
> diff --git a/lib/librte_mempool/rte_mempool.c
> b/lib/librte_mempool/rte_mempool.c
> index 4cf6c25..5056a4f 100644
> --- a/lib/librte_mempool/rte_mempool.c
> +++ b/lib/librte_mempool/rte_mempool.c
> @@ -372,19 +372,21 @@ rte_mempool_create(const char *name, unsigned n,
> unsigned elt_size,
> int socket_id, unsigned flags)
> {
> #ifdef RTE_LIBRTE_XEN_DOM0
> - return (rte_dom0_mempool_create(name, n, elt_size,
> - cache_size, private_data_size,
> - mp_init, mp_init_arg,
> - obj_init, obj_init_arg,
> - socket_id, flags));
> -#else
> - return (rte_mempool_xmem_create(name, n, elt_size,
> - cache_size, private_data_size,
> - mp_init, mp_init_arg,
> - obj_init, obj_init_arg,
> - socket_id, flags,
> - NULL, NULL, MEMPOOL_PG_NUM_DEFAULT,
> MEMPOOL_PG_SHIFT_MAX));
> + if (is_xen_dom0_supported())
> + return (rte_dom0_mempool_create(name, n, elt_size,
> + cache_size, private_data_size,
> + mp_init, mp_init_arg,
> + obj_init, obj_init_arg,
> + socket_id, flags));
> + else
> #endif
> + return (rte_mempool_xmem_create(name, n, elt_size,
> + cache_size, private_data_size,
> + mp_init, mp_init_arg,
> + obj_init, obj_init_arg,
> + socket_id, flags,
> + NULL, NULL,
> MEMPOOL_PG_NUM_DEFAULT,
> + MEMPOOL_PG_SHIFT_MAX));
> }
>
> /*
> diff --git a/lib/librte_pmd_e1000/em_rxtx.c b/lib/librte_pmd_e1000/em_rxtx.c
> index aa0b88c..9e09cfa 100644
> --- a/lib/librte_pmd_e1000/em_rxtx.c
> +++ b/lib/librte_pmd_e1000/em_rxtx.c
> @@ -1104,28 +1104,6 @@ eth_em_recv_scattered_pkts(void *rx_queue,
> struct rte_mbuf **rx_pkts,
> #define EM_MAX_BUF_SIZE 16384
> #define EM_RCTL_FLXBUF_STEP 1024
>
> -static const struct rte_memzone *
> -ring_dma_zone_reserve(struct rte_eth_dev *dev, const char *ring_name,
> - uint16_t queue_id, uint32_t ring_size, int socket_id)
> -{
> - const struct rte_memzone *mz;
> - char z_name[RTE_MEMZONE_NAMESIZE];
> -
> - snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
> - dev->driver->pci_drv.name, ring_name, dev->data->port_id,
> - queue_id);
> -
> - if ((mz = rte_memzone_lookup(z_name)) != 0)
> - return (mz);
> -
> -#ifdef RTE_LIBRTE_XEN_DOM0
> - return rte_memzone_reserve_bounded(z_name, ring_size,
> - socket_id, 0, RTE_CACHE_LINE_SIZE, RTE_PGSIZE_2M);
> -#else
> - return rte_memzone_reserve(z_name, ring_size, socket_id, 0);
> -#endif
> -}
> -
> static void
> em_tx_queue_release_mbufs(struct em_tx_queue *txq) { @@ -1273,8
> +1251,8 @@ eth_em_tx_queue_setup(struct rte_eth_dev *dev,
> * resizing in later calls to the queue setup function.
> */
> tsize = sizeof (txq->tx_ring[0]) * EM_MAX_RING_DESC;
> - if ((tz = ring_dma_zone_reserve(dev, "tx_ring", queue_idx, tsize,
> - socket_id)) == NULL)
> + if ((tz = rte_eth_dma_zone_reserve(dev, "tx_ring", queue_idx, tsize,
> + RTE_CACHE_LINE_SIZE, socket_id))
> == NULL)
> return (-ENOMEM);
>
> /* Allocate the tx queue data structure. */ @@ -1400,8 +1378,8 @@
> eth_em_rx_queue_setup(struct rte_eth_dev *dev,
>
> /* Allocate RX ring for max possible mumber of hardware descriptors.
> */
> rsize = sizeof (rxq->rx_ring[0]) * EM_MAX_RING_DESC;
> - if ((rz = ring_dma_zone_reserve(dev, "rx_ring", queue_idx, rsize,
> - socket_id)) == NULL)
> + if ((rz = rte_eth_dma_zone_reserve(dev, "rx_ring", queue_idx, rsize,
> + RTE_CACHE_LINE_SIZE, socket_id))
> == NULL)
> return (-ENOMEM);
>
> /* Allocate the RX queue data structure. */ diff --git
> a/lib/librte_pmd_e1000/igb_rxtx.c b/lib/librte_pmd_e1000/igb_rxtx.c index
> 5c394a9..d36469b 100644
> --- a/lib/librte_pmd_e1000/igb_rxtx.c
> +++ b/lib/librte_pmd_e1000/igb_rxtx.c
> @@ -1109,29 +1109,6 @@ eth_igb_recv_scattered_pkts(void *rx_queue,
> struct rte_mbuf **rx_pkts, #define IGB_MIN_RING_DESC 32 #define
> IGB_MAX_RING_DESC 4096
>
> -static const struct rte_memzone *
> -ring_dma_zone_reserve(struct rte_eth_dev *dev, const char *ring_name,
> - uint16_t queue_id, uint32_t ring_size, int socket_id)
> -{
> - char z_name[RTE_MEMZONE_NAMESIZE];
> - const struct rte_memzone *mz;
> -
> - snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
> - dev->driver->pci_drv.name, ring_name,
> - dev->data->port_id, queue_id);
> - mz = rte_memzone_lookup(z_name);
> - if (mz)
> - return mz;
> -
> -#ifdef RTE_LIBRTE_XEN_DOM0
> - return rte_memzone_reserve_bounded(z_name, ring_size,
> - socket_id, 0, IGB_ALIGN, RTE_PGSIZE_2M);
> -#else
> - return rte_memzone_reserve_aligned(z_name, ring_size,
> - socket_id, 0, IGB_ALIGN);
> -#endif
> -}
> -
> static void
> igb_tx_queue_release_mbufs(struct igb_tx_queue *txq) { @@ -1265,8
> +1242,8 @@ eth_igb_tx_queue_setup(struct rte_eth_dev *dev,
> * resizing in later calls to the queue setup function.
> */
> size = sizeof(union e1000_adv_tx_desc) * IGB_MAX_RING_DESC;
> - tz = ring_dma_zone_reserve(dev, "tx_ring", queue_idx,
> - size, socket_id);
> + tz = rte_eth_dma_zone_reserve(dev, "tx_ring", queue_idx, size,
> + IGB_ALIGN, socket_id);
> if (tz == NULL) {
> igb_tx_queue_release(txq);
> return (-ENOMEM);
> @@ -1284,12 +1261,14 @@ eth_igb_tx_queue_setup(struct rte_eth_dev *dev,
> txq->port_id = dev->data->port_id;
>
> txq->tdt_reg_addr = E1000_PCI_REG_ADDR(hw, E1000_TDT(txq-
> >reg_idx)); -#ifndef RTE_LIBRTE_XEN_DOM0
> - txq->tx_ring_phys_addr = (uint64_t) tz->phys_addr;
> -#else
> - txq->tx_ring_phys_addr = rte_mem_phy2mch(tz->memseg_id, tz-
> >phys_addr);
> +#ifdef RTE_LIBRTE_XEN_DOM0
> + if (is_xen_dom0_supported())
> + txq->tx_ring_phys_addr = rte_mem_phy2mch(tz->memseg_id,
> tz->phys_addr);
> + else
> #endif
> - txq->tx_ring = (union e1000_adv_tx_desc *) tz->addr;
> + txq->tx_ring_phys_addr = (uint64_t) tz->phys_addr;
> +
> + txq->tx_ring = (union e1000_adv_tx_desc *) tz->addr;
> /* Allocate software ring */
> txq->sw_ring = rte_zmalloc("txq->sw_ring",
> sizeof(struct igb_tx_entry) * nb_desc, @@ -
> 1414,18 +1393,21 @@ eth_igb_rx_queue_setup(struct rte_eth_dev *dev,
> * resizing in later calls to the queue setup function.
> */
> size = sizeof(union e1000_adv_rx_desc) * IGB_MAX_RING_DESC;
> - rz = ring_dma_zone_reserve(dev, "rx_ring", queue_idx, size, socket_id);
> + rz = rte_eth_dma_zone_reserve(dev, "rx_ring", queue_idx, size,
> + IGB_ALIGN, socket_id);
> if (rz == NULL) {
> igb_rx_queue_release(rxq);
> return (-ENOMEM);
> }
> rxq->rdt_reg_addr = E1000_PCI_REG_ADDR(hw, E1000_RDT(rxq-
> >reg_idx));
> rxq->rdh_reg_addr = E1000_PCI_REG_ADDR(hw, E1000_RDH(rxq-
> >reg_idx)); -#ifndef RTE_LIBRTE_XEN_DOM0
> - rxq->rx_ring_phys_addr = (uint64_t) rz->phys_addr;
> -#else
> - rxq->rx_ring_phys_addr = rte_mem_phy2mch(rz->memseg_id, rz-
> >phys_addr);
> +#ifdef RTE_LIBRTE_XEN_DOM0
> + if (is_xen_dom0_supported())
> + rxq->rx_ring_phys_addr =
> + rte_mem_phy2mch(rz->memseg_id, rz->phys_addr);
> + else
> #endif
> + rxq->rx_ring_phys_addr = (uint64_t) rz->phys_addr;
> rxq->rx_ring = (union e1000_adv_rx_desc *) rz->addr;
>
> /* Allocate software ring. */
> diff --git a/lib/librte_pmd_ixgbe/ixgbe_rxtx.c
> b/lib/librte_pmd_ixgbe/ixgbe_rxtx.c
> index e6766b3..303144d 100644
> --- a/lib/librte_pmd_ixgbe/ixgbe_rxtx.c
> +++ b/lib/librte_pmd_ixgbe/ixgbe_rxtx.c
> @@ -1656,35 +1656,6 @@ ixgbe_recv_scattered_pkts(void *rx_queue, struct
> rte_mbuf **rx_pkts, #define IXGBE_MIN_RING_DESC 32 #define
> IXGBE_MAX_RING_DESC 4096
>
> -/*
> - * Create memzone for HW rings. malloc can't be used as the physical address
> is
> - * needed. If the memzone is already created, then this function returns a ptr
> - * to the old one.
> - */
> -static const struct rte_memzone *
> -ring_dma_zone_reserve(struct rte_eth_dev *dev, const char *ring_name,
> - uint16_t queue_id, uint32_t ring_size, int socket_id)
> -{
> - char z_name[RTE_MEMZONE_NAMESIZE];
> - const struct rte_memzone *mz;
> -
> - snprintf(z_name, sizeof(z_name), "%s_%s_%d_%d",
> - dev->driver->pci_drv.name, ring_name,
> - dev->data->port_id, queue_id);
> -
> - mz = rte_memzone_lookup(z_name);
> - if (mz)
> - return mz;
> -
> -#ifdef RTE_LIBRTE_XEN_DOM0
> - return rte_memzone_reserve_bounded(z_name, ring_size,
> - socket_id, 0, IXGBE_ALIGN, RTE_PGSIZE_2M);
> -#else
> - return rte_memzone_reserve_aligned(z_name, ring_size,
> - socket_id, 0, IXGBE_ALIGN);
> -#endif
> -}
> -
> static void
> ixgbe_tx_queue_release_mbufs(struct igb_tx_queue *txq) { @@ -1920,9
> +1891,9 @@ ixgbe_dev_tx_queue_setup(struct rte_eth_dev *dev,
> * handle the maximum ring size is allocated in order to allow for
> * resizing in later calls to the queue setup function.
> */
> - tz = ring_dma_zone_reserve(dev, "tx_ring", queue_idx,
> + tz = rte_eth_dma_zone_reserve(dev, "tx_ring", queue_idx,
> sizeof(union ixgbe_adv_tx_desc) *
> IXGBE_MAX_RING_DESC,
> - socket_id);
> + IXGBE_ALIGN, socket_id);
> if (tz == NULL) {
> ixgbe_tx_queue_release(txq);
> return (-ENOMEM);
> @@ -1950,11 +1921,14 @@ ixgbe_dev_tx_queue_setup(struct rte_eth_dev
> *dev,
> txq->tdt_reg_addr = IXGBE_PCI_REG_ADDR(hw,
> IXGBE_VFTDT(queue_idx));
> else
> txq->tdt_reg_addr = IXGBE_PCI_REG_ADDR(hw,
> IXGBE_TDT(txq->reg_idx));
> -#ifndef RTE_LIBRTE_XEN_DOM0
> - txq->tx_ring_phys_addr = (uint64_t) tz->phys_addr;
> -#else
> - txq->tx_ring_phys_addr = rte_mem_phy2mch(tz->memseg_id, tz-
> >phys_addr);
> +
> +#ifdef RTE_LIBRTE_XEN_DOM0
> + if (is_xen_dom0_supported())
> + txq->tx_ring_phys_addr =
> + rte_mem_phy2mch(tz->memseg_id, tz->phys_addr);
> + else
> #endif
> + txq->tx_ring_phys_addr = (uint64_t) tz->phys_addr;
> txq->tx_ring = (union ixgbe_adv_tx_desc *) tz->addr;
>
> /* Allocate software ring */
> @@ -2195,8 +2169,8 @@ ixgbe_dev_rx_queue_setup(struct rte_eth_dev *dev,
> * handle the maximum ring size is allocated in order to allow for
> * resizing in later calls to the queue setup function.
> */
> - rz = ring_dma_zone_reserve(dev, "rx_ring", queue_idx,
> - RX_RING_SZ, socket_id);
> + rz = rte_eth_dma_zone_reserve(dev, "rx_ring", queue_idx,
> + RX_RING_SZ, IXGBE_ALIGN, socket_id);
> if (rz == NULL) {
> ixgbe_rx_queue_release(rxq);
> return (-ENOMEM);
> @@ -2223,11 +2197,13 @@ ixgbe_dev_rx_queue_setup(struct rte_eth_dev
> *dev,
> rxq->rdh_reg_addr =
> IXGBE_PCI_REG_ADDR(hw, IXGBE_RDH(rxq->reg_idx));
> }
> -#ifndef RTE_LIBRTE_XEN_DOM0
> - rxq->rx_ring_phys_addr = (uint64_t) rz->phys_addr;
> -#else
> - rxq->rx_ring_phys_addr = rte_mem_phy2mch(rz->memseg_id, rz-
> >phys_addr);
> +#ifdef RTE_LIBRTE_XEN_DOM0
> + if (is_xen_dom0_supported())
> + rxq->rx_ring_phys_addr =
> + rte_mem_phy2mch(rz->memseg_id, rz->phys_addr);
> + else
> #endif
> + rxq->rx_ring_phys_addr = (uint64_t) rz->phys_addr;
> rxq->rx_ring = (union ixgbe_adv_rx_desc *) rz->addr;
>
> /*
> --
> 2.1.4
More information about the dev
mailing list