Hi Beilei,
Yes , as per our requirement, we need to have vlan stripped disabled,
when we enable vlan filter and add vlan to the PF from the guest, so that the
vlan id comes upto the app in the packet itself without striping. We also tried
to call rte_vlan_insert in the packet receive side, but that has huge
performance impacts.
What we do from the DPDK app, is during the port configuration, we enable
vlan_filter and disable vlan_strip as a part of dev_conf.rxmode.offloads. Then,
when we have vlans from the app to be added to the PF, we call add_vlan with
the vlan id to be added.
As a part of the add_vlan, the kernel PF driver enables vlan_strip too by
default. Now in this case we cannot use the vlan_offload_set from the DPDK to
disable the vlan_strip.
As per the code in rte_eth_dev_set_vlan_offload() , in this case if we pass the
mask as ETH_VLAN_FILTER_OFFLOAD, and both cur and org will evaluate to be
equal(as DEV_RX_OFFLOAD_VLAN_STRIP is already disabled during port conf) and
VLAN_STRIP will not be disabled.
cur = !!(offload_mask & ETH_VLAN_STRIP_OFFLOAD);
org = !!(dev_offloads & DEV_RX_OFFLOAD_VLAN_STRIP);
if (cur != org) {
if (cur)
dev_offloads |= DEV_RX_OFFLOAD_VLAN_STRIP;
else
dev_offloads &= ~DEV_RX_OFFLOAD_VLAN_STRIP;
mask |= ETH_VLAN_STRIP_MASK;
}
The only possibility will be that we keep VLAN_STRIP enabled during port
configuration and then call set_vlan_offload to disable the VLAN_STRIP, by
passing the mask as ETH_VLAN_FILTER_OFFLOAD .
But this will also have issues, when we try to add the 2nd vlan_id. As the PF
support till 8/16(depends on PF driver version) vlan ids from the guest(without
trust mode on) , the 2nd add_vlan call will again enable the vlan_strip in the
PF.
This will happen as after the 1st call to the set_vlan_offload, on successful
vlan_offload_set() , we will update the dev_conf.rxmode.offloads with the
computed dev_offloads (dev_offloads &= ~DEV_RX_OFFLOAD_VLAN_STRIP).
dev->data->dev_conf.rxmode.offloads = dev_offloads;
So, we will face the same issue as descried above where cur and org will
evaluate to be equal(as VLAN_STRIP already disabled in 1st call) and
VLAN_STRIP will not be disabled again.
Instead of that, if we call i40evf_disable_vlan_strip() when
dev_conf.rxmode.offloads has VLAN_STRIP set to disabled after successful call
of add_vlan from the pmd itself that will help in every scenario.
This scenario works fine if we use DPDK at the PF side too as we don’t enable
VLAN_STRIP in add_vlan in the DPDK i40e PF.
--
Regards,
Souvik
From: Xing, Beilei <[email protected]>
Sent: Tuesday, April 14, 2020 2:37 AM
To: Dey, Souvik <[email protected]>; [email protected]; [email protected]
Cc: Yigit, Ferruh <[email protected]>; Zhang, Qi Z <[email protected]>
Subject: RE: i40eVF pmd vlan id handling
________________________________
NOTICE: This email was received from an EXTERNAL sender
________________________________
Hi Souvik,
With kernel PF + DPDK VF, enable vlan filter and disable vlan strip on VF side,
is this your requirement?
If yes, I think you can do . vlan_offload_set after doing . vlan_filter_set on
VF side.
BR,
Beilei
From: Dey, Souvik <[email protected]<mailto:[email protected]>>
Sent: Tuesday, April 14, 2020 4:04 AM
To: [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>;
Xing, Beilei <[email protected]<mailto:[email protected]>>
Cc: Yigit, Ferruh <[email protected]<mailto:[email protected]>>;
Zhang, Qi Z <[email protected]<mailto:[email protected]>>
Subject: RE: i40eVF pmd vlan id handling
On debugging further it looks like that linux driver enables vlan_striping by
default when VIRTCHNL_OP_ADD_VLAN is sent to the PF.
static int i40e_vc_add_vlan_msg(struct i40e_vf *vf, u8 *msg)
{
…..
i40e_vlan_stripping_enable(vsi);
….
}
Due to this when ever we enable vlan on i40e the van is stripped by default
while sending it to the Guest. This could be handled by the below patch in DPDK
:
In file i40e_ethdev_vf.c
static int
i40evf_add_vlan(struct rte_eth_dev *dev, uint16_t vlanid)
{
struct i40e_vf *vf = I40EVF_DEV_PRIVATE_TO_VF(dev->data->dev_private);
struct virtchnl_vlan_filter_list *vlan_list;
uint8_t cmd_buffer[sizeof(struct virtchnl_vlan_filter_list) +
sizeof(uint16_t)];
int err;
struct vf_cmd_info args;
vlan_list = (struct virtchnl_vlan_filter_list *)cmd_buffer;
vlan_list->vsi_id = vf->vsi_res->vsi_id;
vlan_list->num_elements = 1;
vlan_list->vlan_id[0] = vlanid;
args.ops = VIRTCHNL_OP_ADD_VLAN;
args.in_args = (u8 *)&cmd_buffer;
args.in_args_size = sizeof(cmd_buffer);
args.out_buffer = vf->aq_resp;
args.out_size = I40E_AQ_BUF_SZ;
err = i40evf_execute_vf_cmd(dev, &args);
if (err)
PMD_DRV_LOG(ERR, "fail to execute command OP_ADD_VLAN");
+ if (!(dev_conf->rxmode.offloads & DEV_RX_OFFLOAD_VLAN_STRIP)) {
+ i40evf_disable_vlan_strip(dev);
+ }
return err;
}
Or we need to call . vlan_offload_set after doing . vlan_filter_set from the
DPDK app ?
--
Regards,
Souvik
From: Dey, Souvik
Sent: Monday, April 13, 2020 12:04 PM
To: [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>;
[email protected]<mailto:[email protected]>
Cc: [email protected]<mailto:[email protected]>;
[email protected]<mailto:[email protected]>
Subject: i40eVF pmd vlan id handling
Hi All,
I am using DPDK 18.11.2 and i40e PF linux driver on the host 2.4.6. I
see there, when I enable DEV_RX_OFFLOAD_VLAN_FILTER from the DPDK app, and then
configure the specific vlan id using rte_eth_dev_vlan_filter(). As per DPDK
code by default when we do dev_configure, we call i40evf_init_vlan() and if we
don’t enable rxmode.offloads with DEV_RX_OFFLOAD_VLAN_STRIP, then we should
send VIRTCHNL_OP_DISABLE_VLAN_STRIPPING command to the PF from the guest.
With more debugging enabled, when DPDK requests VLAN filtering by sending
VIRTCHNL_OP_ADD_VLAN to the PF, then we see that the VF stripping is enabled
also on Linux. If we don’t add VLAN ID and send VIRTCHNL_OP_ADD_VLAN message
down to the PF , then we do receive packets with VLAN ID set.
The same works fine in VmWare drivers, where we do receive VALN id in the
packets if STRIP is disabled.
In the linux case, when we receive frames with VLAN headers, the vlan id is
stripped at the PF, and the TCI will record the missing VLAN details when
handed up to the DPDK driver.
With i40e debug enabled, it's clear to see the difference being reported in
i40e_rxd_to_vlan_tci:
Example using VLAN on i40evf SR-IOV (vlan fails):
PMD: i40e_rxd_to_vlan_tci(): Mbuf vlan_tci: 8, vlan_tci_outer: 0
Port 0 pkt-len=60 nb-segs=1
ETH: src=00:10:E0:8D:A7:52 dst=FF:FF:FF:FF:FF:FF type=0x0806
ARP: hrd=1 proto=0x0800 hln=6 pln=4 op=1 (ARP Request)
sha=00:10:E0:8D:A7:52 sip=8.8.8.102<http://8.8.8.102>
tha=00:00:00:00:00:00 tip=8.8.8.3<http://8.8.8.3>
As the application requested tagging not be stripped, and the hardware driver
was not able to disable strip, in my opinion either DPDK or linux driver should
have a bug in handing this. Also as we see VmWare i40e driver working it could
be either a extra config required for linux driver or a bug in linux driver.
I also tried to add a call to rte_vlan_insert() to reinstate the VLAN header
using the data from TCI in i40e_rxd_to_vlan_tci(), but it is having performance
impacts as expected. Also as rte_vlan_insert() does not support QINQ, it will
have other issues too.
Can you please clarify if this is working for i40e vlan filter test and if we
are missing something in the DPDK, like sending
VIRTCHNL_OP_DISABLE_VLAN_STRIPPING after every VIRTCHNL_OP_ADD_VLAN if
DEV_RX_OFFLOAD_VLAN_STRIP is not set by the DPDk app.
--
Regards,
Souvik
________________________________
Notice: This e-mail together with any attachments may contain information of
Ribbon Communications Inc. that is confidential and/or proprietary for the sole
use of the intended recipient. Any review, disclosure, reliance or distribution
by others or forwarding without express permission is strictly prohibited. If
you are not the intended recipient, please notify the sender immediately and
then delete all copies, including any attachments.
________________________________
