[dpdk-dev,v2] fm10k: enable PCIe port level Loopback Suppression

Message ID 1454589801-691-1-git-send-email-shaopeng.he@intel.com (mailing list archive)
State Accepted, archived
Delegated to: Bruce Richardson
Headers

Commit Message

He, Shaopeng Feb. 4, 2016, 12:43 p.m. UTC
  In FM10K, a single PCIe port can derive out a few logical ports,
like SRIOV PF/VF devices, VMDQ objects. To better manage them, FM10K
silicon assigned a Unique GLORT ID to each logical ports.
When a logical port sends a broadcast packet, the silicon will flood
it to all Logical ports, including the one sent the broadcast packet.
To prevent this, silicon has a rxq register to fill the glort id of
the logical port that queue binds to.
FM10K has a switch core inside, which has another loopback suppression
mechanism in the switch level. Switch level loopback suppression mostly
works for the ether port traffic.
This patch assigns a SGLORT for each RX queue, and enables PCIe port
level Loopback Suppression.

Signed-off-by: Shaopeng He <shaopeng.he@intel.com>
---
 drivers/net/fm10k/fm10k_ethdev.c | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)
  

Comments

Chen, Jing D Feb. 5, 2016, 9:25 a.m. UTC | #1
Hi, 

Best Regards,
Mark


> -----Original Message-----
> From: He, Shaopeng
> Sent: Thursday, February 04, 2016 8:43 PM
> To: dev@dpdk.org
> Cc: Chen, Jing D; Wang, Xiao W; He, Shaopeng
> Subject: [PATCH v2] fm10k: enable PCIe port level Loopback Suppression
> 
> In FM10K, a single PCIe port can derive out a few logical ports,
> like SRIOV PF/VF devices, VMDQ objects. To better manage them, FM10K
> silicon assigned a Unique GLORT ID to each logical ports.
> When a logical port sends a broadcast packet, the silicon will flood
> it to all Logical ports, including the one sent the broadcast packet.
> To prevent this, silicon has a rxq register to fill the glort id of
> the logical port that queue binds to.
> FM10K has a switch core inside, which has another loopback suppression
> mechanism in the switch level. Switch level loopback suppression mostly
> works for the ether port traffic.
> This patch assigns a SGLORT for each RX queue, and enables PCIe port
> level Loopback Suppression.
> 
> Signed-off-by: Shaopeng He <shaopeng.he@intel.com>
Acked-by : Jing Chen <jing.d.chen@intel.com>
  
Bruce Richardson Feb. 25, 2016, 9:45 p.m. UTC | #2
On Fri, Feb 05, 2016 at 09:25:02AM +0000, Chen, Jing D wrote:
> Hi, 
> 
> Best Regards,
> Mark
> 
> 
> > -----Original Message-----
> > From: He, Shaopeng
> > Sent: Thursday, February 04, 2016 8:43 PM
> > To: dev@dpdk.org
> > Cc: Chen, Jing D; Wang, Xiao W; He, Shaopeng
> > Subject: [PATCH v2] fm10k: enable PCIe port level Loopback Suppression
> > 
> > In FM10K, a single PCIe port can derive out a few logical ports,
> > like SRIOV PF/VF devices, VMDQ objects. To better manage them, FM10K
> > silicon assigned a Unique GLORT ID to each logical ports.
> > When a logical port sends a broadcast packet, the silicon will flood
> > it to all Logical ports, including the one sent the broadcast packet.
> > To prevent this, silicon has a rxq register to fill the glort id of
> > the logical port that queue binds to.
> > FM10K has a switch core inside, which has another loopback suppression
> > mechanism in the switch level. Switch level loopback suppression mostly
> > works for the ether port traffic.
> > This patch assigns a SGLORT for each RX queue, and enables PCIe port
> > level Loopback Suppression.
> > 
> > Signed-off-by: Shaopeng He <shaopeng.he@intel.com>
> Acked-by : Jing Chen <jing.d.chen@intel.com>
> 
Applied to dpdk-next-net/rel_16_04

/Bruce
  

Patch

diff --git a/drivers/net/fm10k/fm10k_ethdev.c b/drivers/net/fm10k/fm10k_ethdev.c
index f6eb05d..60f821a 100644
--- a/drivers/net/fm10k/fm10k_ethdev.c
+++ b/drivers/net/fm10k/fm10k_ethdev.c
@@ -690,12 +690,15 @@  static int
 fm10k_dev_rx_init(struct rte_eth_dev *dev)
 {
 	struct fm10k_hw *hw = FM10K_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+	struct fm10k_macvlan_filter_info *macvlan;
 	int i, ret;
 	struct fm10k_rx_queue *rxq;
 	uint64_t base_addr;
 	uint32_t size;
 	uint32_t rxdctl = FM10K_RXDCTL_WRITE_BACK_MIN_DELAY;
+	uint32_t logic_port = hw->mac.dglort_map;
 	uint16_t buf_size;
+	uint16_t queue_stride = 0;
 
 	/* Disable RXINT to avoid possible interrupt */
 	for (i = 0; i < hw->mac.max_queues; i++)
@@ -735,7 +738,8 @@  fm10k_dev_rx_init(struct rte_eth_dev *dev)
 		buf_size -= FM10K_RX_DATABUF_ALIGN;
 
 		FM10K_WRITE_REG(hw, FM10K_SRRCTL(i),
-				buf_size >> FM10K_SRRCTL_BSIZEPKT_SHIFT);
+				(buf_size >> FM10K_SRRCTL_BSIZEPKT_SHIFT) |
+				FM10K_SRRCTL_LOOPBACK_SUPPRESS);
 
 		/* It adds dual VLAN length for supporting dual VLAN */
 		if ((dev->data->dev_conf.rxmode.max_rx_pkt_len +
@@ -762,6 +766,18 @@  fm10k_dev_rx_init(struct rte_eth_dev *dev)
 	/* Decide the best RX function */
 	fm10k_set_rx_function(dev);
 
+	/* update RX_SGLORT for loopback suppress*/
+	if (hw->mac.type != fm10k_mac_pf)
+		return 0;
+	macvlan = FM10K_DEV_PRIVATE_TO_MACVLAN(dev->data->dev_private);
+	if (macvlan->nb_queue_pools)
+		queue_stride = dev->data->nb_rx_queues / macvlan->nb_queue_pools;
+	for (i = 0; i < dev->data->nb_rx_queues; ++i) {
+		if (i && queue_stride && !(i % queue_stride))
+			logic_port++;
+		FM10K_WRITE_REG(hw, FM10K_RX_SGLORT(i), logic_port);
+	}
+
 	return 0;
 }