DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] eth dev doesn't free mbufs for 82575EB
@ 2014-05-30 12:44 Helmut Sim
  2014-06-11 16:18 ` Thomas Monjalon
  0 siblings, 1 reply; 3+ messages in thread
From: Helmut Sim @ 2014-05-30 12:44 UTC (permalink / raw)
  To: dev

[-- Attachment #1: Type: text/plain, Size: 1276 bytes --]

Hi all,

I am trying to verify that the 82575EB NIC functions well.
The file rte_pci_dev_ids.h includes the following lines:

#ifdef RTE_PCI_DEV_USE_82575EB_COPPER
RTE_PCI_DEV_ID_DECL_IGB(PCI_VENDOR_ID_INTEL, E1000_DEV_ID_82575EB_COPPER)
#endif

so I defined the RTE_PCI_DEV_USE_82575EB_COPPER and hoped it will work well.

I have one simple thread that reads from the device using rte_eth_rx_burst
, then copy the content to another buffer, then free the rx buffer and
transmit the new buffer.
The systems function well (pinging with an external PC) untill the ring
buffer of the tx is getting full and then there are no buffer descriptors
to send the packets anymore.

the buffers are all allocated from one single mempool that includes:
rte_mempool_create("testing_mempool",
1000,
5000,
64,
sizeof(struct rte_pktmbuf_pool_private),
rte_pktmbuf_pool_init,
NULL,
rte_pktmbuf_init,
NULL,
rte_socket_id(),
0);

I assume the tx (using rte_eth_tx_burst) is supposed to free the buffers,
but it doesn't.
Attached is the dev configuration, though I tried a lot of options but all
brought me to the same result so i am pretty sure it is not the root cause.

I've invest few days in order to understand this but with no success.
Any idea what could be the cause?


Thanks in advance

[-- Attachment #2: eth dev configuration.txt --]
[-- Type: text/plain, Size: 2340 bytes --]


struct rte_fdir_conf fdir_conf = {
	.mode = RTE_FDIR_MODE_NONE,
	.pballoc = RTE_FDIR_PBALLOC_64K,
	.status = RTE_FDIR_REPORT_STATUS,
	.flexbytes_offset = 0x6,
	.drop_queue = 127,
};


struct rte_eth_rxmode rx_mode = {
	.max_rx_pkt_len = ETHER_MAX_LEN, /**< Default maximum frame length. */
	.split_hdr_size = 0,
	.header_split   = 0, /**< Header Split disabled. */
	.hw_ip_checksum = 0, /**< IP checksum offload disabled. */
	.hw_vlan_filter = 1, /**< VLAN filtering enabled. */
	.hw_vlan_strip  = 1, /**< VLAN strip enabled. */
	.hw_vlan_extend = 0, /**< Extended VLAN disabled. */
	.jumbo_frame    = 0, /**< Jumbo Frame Support disabled. */
	.hw_strip_crc   = 0, /**< CRC stripping by hardware disabled. */
};

static const struct rte_eth_conf port_conf = {
	.link_speed = ETH_LINK_SPEED_AUTONEG,
	.link_duplex = ETH_LINK_FULL_DUPLEX,

//	.lpbk_mode = 0,
	.rxmode = {
			.max_rx_pkt_len = ETHER_MAX_LEN, /**< Default maximum frame length. */
			.split_hdr_size = 0,
			.header_split   = 0, /**< Header Split disabled. */
			.hw_ip_checksum = 0, /**< IP checksum offload disabled. */
			.hw_vlan_filter = 1, /**< VLAN filtering enabled. */
			.hw_vlan_strip  = 1, /**< VLAN strip enabled. */
			.hw_vlan_extend = 0, /**< Extended VLAN disabled. */
			.jumbo_frame    = 0, /**< Jumbo Frame Support disabled. */
			.hw_strip_crc   = 0, /**< CRC stripping by hardware disabled. */
			.mq_mode = ETH_MQ_RX_NONE,
	},
	.txmode = {
		.mq_mode = ETH_MQ_TX_NONE,
	},

	.fdir_conf = {
			.mode = RTE_FDIR_MODE_NONE,
			.pballoc = RTE_FDIR_PBALLOC_64K,
			.status = RTE_FDIR_REPORT_STATUS,
			.flexbytes_offset = 0x6,
			.drop_queue = 127,
	},

//	.intr_conf = {
//		.lsc = 1,			/**< link status interrupt feature enabled */
//	},
};


static const struct rte_eth_rxconf rx_conf = {
	.rx_thresh = {
		.pthresh = RX_PTHRESH,
		.hthresh = RX_HTHRESH,
		.wthresh = RX_WTHRESH,
	},
	.rx_free_thresh = 16, /**< Drives the freeing of RX descriptors. */
	.rx_drop_en = 1, /**< Drop packets if no descriptors are available. */
};


static const struct rte_eth_txconf tx_conf = {
	.tx_thresh = {
		.pthresh = TX_PTHRESH,
		.hthresh = TX_HTHRESH,
		.wthresh = TX_WTHRESH,
	},
	.tx_free_thresh = 16, 	/**< Drives the setting of RS bit on TXDs. */
	.tx_rs_thresh = 1, 		/**< Drives the setting of RS bit on TXDs. */
	.txq_flags = ETH_TXQ_FLAGS_NOMULTMEMP,
};
  

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [dpdk-dev] eth dev doesn't free mbufs for 82575EB
  2014-05-30 12:44 [dpdk-dev] eth dev doesn't free mbufs for 82575EB Helmut Sim
@ 2014-06-11 16:18 ` Thomas Monjalon
  2014-06-19 14:02   ` Helmut Sim
  0 siblings, 1 reply; 3+ messages in thread
From: Thomas Monjalon @ 2014-06-11 16:18 UTC (permalink / raw)
  To: Helmut Sim; +Cc: dev

2014-05-30 15:44, Helmut Sim:
> I am trying to verify that the 82575EB NIC functions well.
> The file rte_pci_dev_ids.h includes the following lines:
> 
> #ifdef RTE_PCI_DEV_USE_82575EB_COPPER
> RTE_PCI_DEV_ID_DECL_IGB(PCI_VENDOR_ID_INTEL, E1000_DEV_ID_82575EB_COPPER)
> #endif
> 
> so I defined the RTE_PCI_DEV_USE_82575EB_COPPER and hoped it will work well.

This commit could interest you:
	http://dpdk.org/browse/dpdk/commit/?id=cc333208d5658fea6420

-- 
Thomas

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [dpdk-dev] eth dev doesn't free mbufs for 82575EB
  2014-06-11 16:18 ` Thomas Monjalon
@ 2014-06-19 14:02   ` Helmut Sim
  0 siblings, 0 replies; 3+ messages in thread
From: Helmut Sim @ 2014-06-19 14:02 UTC (permalink / raw)
  To: Thomas Monjalon; +Cc: dev

Thanks,
it has the same effect as my changeee

i found the lost packets issue... it was related to my app implementation
(a bug with increasing the number of segments at the nb_segs variable).

anyway thanks.



On Wed, Jun 11, 2014 at 7:18 PM, Thomas Monjalon <thomas.monjalon@6wind.com>
wrote:

> 2014-05-30 15:44, Helmut Sim:
> > I am trying to verify that the 82575EB NIC functions well.
> > The file rte_pci_dev_ids.h includes the following lines:
> >
> > #ifdef RTE_PCI_DEV_USE_82575EB_COPPER
> > RTE_PCI_DEV_ID_DECL_IGB(PCI_VENDOR_ID_INTEL, E1000_DEV_ID_82575EB_COPPER)
> > #endif
> >
> > so I defined the RTE_PCI_DEV_USE_82575EB_COPPER and hoped it will work
> well.
>
> This commit could interest you:
>         http://dpdk.org/browse/dpdk/commit/?id=cc333208d5658fea6420
>
> --
> Thomas
>

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2014-06-19 14:02 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-05-30 12:44 [dpdk-dev] eth dev doesn't free mbufs for 82575EB Helmut Sim
2014-06-11 16:18 ` Thomas Monjalon
2014-06-19 14:02   ` Helmut Sim

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).