diff mbox

[dpdk-dev] i40e: add VEB switching support for i40e

Message ID 1453358984-20550-1-git-send-email-zhe.tao@intel.com (mailing list archive)
State Changes Requested, archived
Delegated to: Bruce Richardson
Headers

Commit Message

Zhe Tao Jan. 21, 2016, 6:49 a.m. UTC 
  VEB switching feature for i40e is used to enable the switching between the
 VSIs connect to the virtual bridge. The old implementation is setting the
 virtual bridge mode as VEPA which is port aggregation. Enable the switching 
 ability by setting the loop back mode for the specific VSIs which connect to PF
 or VFs. 

Signed-off-by: Zhe Tao <zhe.tao@intel.com>
---
 drivers/net/i40e/i40e_ethdev.c | 48 +++++++++++++++++++++++++++++++++++-------
 1 file changed, 40 insertions(+), 8 deletions(-)

Comments

Thomas Monjalon Jan. 22, 2016, 10:16 a.m. UTC | #1 
2016-01-21 14:49, Zhe Tao:
> VEB switching feature for i40e is used to enable the switching between the
>  VSIs connect to the virtual bridge. The old implementation is setting the
>  virtual bridge mode as VEPA which is port aggregation. Enable the switching 
>  ability by setting the loop back mode for the specific VSIs which connect to PF
>  or VFs. 
> 
> Signed-off-by: Zhe Tao <zhe.tao@intel.com>

Is VEB something specific to i40e?
Please explain the acronyms VEB, VSI and VEPA in the commit log or in a
doc dedicated to i40e. Having some design explanations in a doc would help
to understand the changes and to contribute.
Thanks
Xu, Qian Q Jan. 25, 2016, 6:22 a.m. UTC | #2 
I have a quick check, VEB/VEPA/VSI is not i40e's specific, they are IEEE EVB terms. See below. 

IEEE EVB tutorial: http://www.ieee802.org/802_tutorials/2009-11/evb-tutorial-draft-20091116_v09.pdf

Virtual Ethernet Bridge (VEB) - This is an IEEE EVB term. A VEB is a VLAN Bridge internal to Fortville that bridges the traffic of multiple VSIs over an internal virtual network. 
Virtual Ethernet Port Aggregator (VEPA) - This is an IEEE EVB term. A VEPA multiplexes the traffic of one or more VSIs onto a single Fortville Ethernet port. The biggest difference between a VEB and a VEPA is that a VEB can switch packets internally between VSIs, whereas a VEPA cannot. 
Virtual Station Interface (VSI) - This is an IEEE EVB term that defines the properties of a virtual machine's (or a physical machine's) connection to the network. Each downstream v-port on a Fortville VEB or VEPA defines a VSI. A standards-based definition of VSI properties enables network management tools to perform virtual machine migration and associated network re-configuration in a vendor-neutral manner.



Thanks
Qian

-----Original Message-----
From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Thomas Monjalon
Sent: Friday, January 22, 2016 6:16 PM
To: Tao, Zhe; Mcnamara, John
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH] i40e: add VEB switching support for i40e

2016-01-21 14:49, Zhe Tao:
> VEB switching feature for i40e is used to enable the switching between 
> the  VSIs connect to the virtual bridge. The old implementation is 
> setting the  virtual bridge mode as VEPA which is port aggregation. 
> Enable the switching  ability by setting the loop back mode for the 
> specific VSIs which connect to PF  or VFs.
> 
> Signed-off-by: Zhe Tao <zhe.tao@intel.com>

Is VEB something specific to i40e?
Please explain the acronyms VEB, VSI and VEPA in the commit log or in a doc dedicated to i40e. Having some design explanations in a doc would help to understand the changes and to contribute.
Thanks
Jingjing Wu Feb. 19, 2016, 5:17 a.m. UTC | #3 
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Zhe Tao
> Sent: Thursday, January 21, 2016 2:50 PM
> To: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH] i40e: add VEB switching support for i40e
> 
> VEB switching feature for i40e is used to enable the switching between the
>  VSIs connect to the virtual bridge. The old implementation is setting the
>  virtual bridge mode as VEPA which is port aggregation. Enable the switching
>  ability by setting the loop back mode for the specific VSIs which connect to PF
>  or VFs.

As I know, there is a known issue about the veb switch on older NVM version.
I would be better to add a NVM version verification, if version > xx.xx, then enable it?

Thanks
Jingjing

> Signed-off-by: Zhe Tao <zhe.tao@intel.com>
> ---
>  drivers/net/i40e/i40e_ethdev.c | 48 +++++++++++++++++++++++++++++++++++-------
>  1 file changed, 40 insertions(+), 8 deletions(-)
>
Zhe Tao Feb. 23, 2016, 3:48 a.m. UTC | #4 
On Fri, Feb 19, 2016 at 01:17:41PM +0800, Wu, Jingjing wrote:
> 
> 
> > -----Original Message-----
> > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Zhe Tao
> > Sent: Thursday, January 21, 2016 2:50 PM
> > To: dev@dpdk.org
> > Subject: [dpdk-dev] [PATCH] i40e: add VEB switching support for i40e
> > 
> > VEB switching feature for i40e is used to enable the switching between the
> >  VSIs connect to the virtual bridge. The old implementation is setting the
> >  virtual bridge mode as VEPA which is port aggregation. Enable the switching
> >  ability by setting the loop back mode for the specific VSIs which connect to PF
> >  or VFs.
> 
> As I know, there is a known issue about the veb switch on older NVM version.
> I would be better to add a NVM version verification, if version > xx.xx, then enable it?
> 
> Thanks
> Jingjing
> 
Great suggestion, i will add the check for it!
Thanks
Zhe Tao
> > Signed-off-by: Zhe Tao <zhe.tao@intel.com>
> > ---
> >  drivers/net/i40e/i40e_ethdev.c | 48 +++++++++++++++++++++++++++++++++++-------
> >  1 file changed, 40 insertions(+), 8 deletions(-)
> >
diff mbox

Patch

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index bf6220d..ba2ba9e 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -3822,6 +3822,39 @@  i40e_vsi_get_bw_config(struct i40e_vsi *vsi)
 	return I40E_SUCCESS;
 }
 
+/* i40e_enable_pf_lb
+ * @pf: pointer to the pf structure
+ *
+ * allow loopback on pf
+ */
+static inline void
+i40e_enable_pf_lb(struct i40e_pf *pf)
+{
+	struct i40e_hw *hw = I40E_PF_TO_HW(pf);
+	struct i40e_vsi_context ctxt;
+	int ret;
+
+	memset(&ctxt, 0, sizeof(ctxt));
+	ctxt.seid = pf->main_vsi_seid;
+	ctxt.pf_num = hw->pf_id;
+	ret = i40e_aq_get_vsi_params(hw, &ctxt, NULL);
+	if (ret) {
+		PMD_DRV_LOG(ERR, "cannot get pf vsi config, err %d, aq_err %d",
+			    ret, hw->aq.asq_last_status);
+		return;
+	}
+	ctxt.flags = I40E_AQ_VSI_TYPE_PF;
+	ctxt.info.valid_sections =
+		rte_cpu_to_le_16(I40E_AQ_VSI_PROP_SWITCH_VALID);
+	ctxt.info.switch_id |=
+		rte_cpu_to_le_16(I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB);
+
+	ret = i40e_aq_update_vsi_params(hw, &ctxt, NULL);
+	if (ret)
+		PMD_DRV_LOG(ERR, "update vsi switch failed, aq_err=%d\n",
+			    hw->aq.asq_last_status);
+}
+
 /* Setup a VSI */
 struct i40e_vsi *
 i40e_vsi_setup(struct i40e_pf *pf,
@@ -3857,6 +3890,8 @@  i40e_vsi_setup(struct i40e_pf *pf,
 			PMD_DRV_LOG(ERR, "VEB setup failed");
 			return NULL;
 		}
+		/* set ALLOWLOOPBACk on pf, when veb is created */
+		i40e_enable_pf_lb(pf);
 	}
 
 	vsi = rte_zmalloc("i40e_vsi", sizeof(struct i40e_vsi), 0);
@@ -4029,14 +4064,11 @@  i40e_vsi_setup(struct i40e_pf *pf,
 		ctxt.connection_type = 0x1;
 		ctxt.flags = I40E_AQ_VSI_TYPE_VF;
 
-		/**
-		 * Do not configure switch ID to enable VEB switch by
-		 * I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB. Because in Fortville,
-		 * if the source mac address of packet sent from VF is not
-		 * listed in the VEB's mac table, the VEB will switch the
-		 * packet back to the VF. Need to enable it when HW issue
-		 * is fixed.
-		 */
+		/* Configure switch ID */
+		ctxt.info.valid_sections |=
+			rte_cpu_to_le_16(I40E_AQ_VSI_PROP_SWITCH_VALID);
+		ctxt.info.switch_id =
+			rte_cpu_to_le_16(I40E_AQ_VSI_SW_ID_FLAG_ALLOW_LB);
 
 		/* Configure port/vlan */
 		ctxt.info.valid_sections |=