[dpdk-dev] net/vmxnet3: restore correct filtering
Checks
Commit Message
We should only restore shadow_vfta when hw_vlan_filter is active.
Otherwise, we should restore the previous filtering behavior.
Fixes: f003fc383487("vmxnet3: enable vlan filtering")
Cc: stable@dpdk.org
Signed-off-by: Chas Williams <ciwillia@brocade.com>
---
drivers/net/vmxnet3/vmxnet3_ethdev.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
Comments
24/07/2017 16:22, Charles (Chas) Williams:
> We should only restore shadow_vfta when hw_vlan_filter is active.
> Otherwise, we should restore the previous filtering behavior.
>
> Fixes: f003fc383487("vmxnet3: enable vlan filtering")
> Cc: stable@dpdk.org
>
> Signed-off-by: Chas Williams <ciwillia@brocade.com>
Any review from VMware please?
On Sun, 30 Jul 2017, Thomas Monjalon wrote:
> 24/07/2017 16:22, Charles (Chas) Williams:
> > We should only restore shadow_vfta when hw_vlan_filter is active.
> > Otherwise, we should restore the previous filtering behavior.
> >
> > Fixes: f003fc383487("vmxnet3: enable vlan filtering")
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Chas Williams <ciwillia@brocade.com>
>
> Any review from VMware please?
I will reply back on this later today.
Thanks,
Shri
On Mon, 24 Jul 2017, Charles (Chas) Williams wrote:
> We should only restore shadow_vfta when hw_vlan_filter is active.
> Otherwise, we should restore the previous filtering behavior.
>
> Fixes: f003fc383487("vmxnet3: enable vlan filtering")
> Cc: stable@dpdk.org
>
> Signed-off-by: Chas Williams <ciwillia@brocade.com>
> ---
> drivers/net/vmxnet3/vmxnet3_ethdev.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.c b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> index ab5a824..3910991 100644
> --- a/drivers/net/vmxnet3/vmxnet3_ethdev.c
> +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> @@ -1219,7 +1219,10 @@ vmxnet3_dev_promiscuous_disable(struct rte_eth_dev *dev)
> struct vmxnet3_hw *hw = dev->data->dev_private;
> uint32_t *vf_table = hw->shared->devRead.rxFilterConf.vfTable;
>
> - memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
> + if (dev->data->dev_conf.rxmode.hw_vlan_filter)
> + memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
> + else
> + memset(vf_table, 0xff, VMXNET3_VFT_TABLE_SIZE);
From the device emulation standpoint, vf_table array set to all 1s means
that rx filtering is enabled but all vlan ID are allowed, while vf_table
array set to all 0s means that no vlan ID filtering is done.
So, it is better to go with all 0s rather than all 1s, it is also more
efficient for the emulation.
This is an existing problem in vmxnet3_dev_vlan_offload_set as well.
> vmxnet3_dev_set_rxmode(hw, VMXNET3_RXM_PROMISC, 0);
> VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD,
> VMXNET3_CMD_UPDATE_VLAN_FILTERS);
> --
> 2.1.4
>
>
On 08/01/2017 07:41 PM, Shrikrishna Khare wrote:
>
>
> On Mon, 24 Jul 2017, Charles (Chas) Williams wrote:
>
>> We should only restore shadow_vfta when hw_vlan_filter is active.
>> Otherwise, we should restore the previous filtering behavior.
>>
>> Fixes: f003fc383487("vmxnet3: enable vlan filtering")
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Chas Williams <ciwillia@brocade.com>
>> ---
>> drivers/net/vmxnet3/vmxnet3_ethdev.c | 5 ++++-
>> 1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.c b/drivers/net/vmxnet3/vmxnet3_ethdev.c
>> index ab5a824..3910991 100644
>> --- a/drivers/net/vmxnet3/vmxnet3_ethdev.c
>> +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.c
>> @@ -1219,7 +1219,10 @@ vmxnet3_dev_promiscuous_disable(struct rte_eth_dev *dev)
>> struct vmxnet3_hw *hw = dev->data->dev_private;
>> uint32_t *vf_table = hw->shared->devRead.rxFilterConf.vfTable;
>>
>> - memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
>> + if (dev->data->dev_conf.rxmode.hw_vlan_filter)
>> + memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
>> + else
>> + memset(vf_table, 0xff, VMXNET3_VFT_TABLE_SIZE);
>
> From the device emulation standpoint, vf_table array set to all 1s means
> that rx filtering is enabled but all vlan ID are allowed, while vf_table
> array set to all 0s means that no vlan ID filtering is done.
>
> So, it is better to go with all 0s rather than all 1s, it is also more
> efficient for the emulation.
>
> This is an existing problem in vmxnet3_dev_vlan_offload_set as well.
That may be the case, but this patch isn't the place to address that.
If this needs changed, there should be a follow on patch that makes
this change.
>
>> vmxnet3_dev_set_rxmode(hw, VMXNET3_RXM_PROMISC, 0);
>> VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD,
>> VMXNET3_CMD_UPDATE_VLAN_FILTERS);
>> --
>> 2.1.4
>>
>>
On Tue, 1 Aug 2017, Charles (Chas) Williams wrote:
>
>
> On 08/01/2017 07:41 PM, Shrikrishna Khare wrote:
> >
> >
> > On Mon, 24 Jul 2017, Charles (Chas) Williams wrote:
> >
> > > We should only restore shadow_vfta when hw_vlan_filter is active.
> > > Otherwise, we should restore the previous filtering behavior.
> > >
> > > Fixes: f003fc383487("vmxnet3: enable vlan filtering")
> > > Cc: stable@dpdk.org
> > >
> > > Signed-off-by: Chas Williams <ciwillia@brocade.com>
> > > ---
> > > drivers/net/vmxnet3/vmxnet3_ethdev.c | 5 ++++-
> > > 1 file changed, 4 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.c
> > > b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> > > index ab5a824..3910991 100644
> > > --- a/drivers/net/vmxnet3/vmxnet3_ethdev.c
> > > +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> > > @@ -1219,7 +1219,10 @@ vmxnet3_dev_promiscuous_disable(struct rte_eth_dev
> > > *dev)
> > > struct vmxnet3_hw *hw = dev->data->dev_private;
> > > uint32_t *vf_table = hw->shared->devRead.rxFilterConf.vfTable;
> > >
> > > - memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
> > > + if (dev->data->dev_conf.rxmode.hw_vlan_filter)
> > > + memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
> > > + else
> > > + memset(vf_table, 0xff, VMXNET3_VFT_TABLE_SIZE);
> >
> > From the device emulation standpoint, vf_table array set to all 1s means
> > that rx filtering is enabled but all vlan ID are allowed, while vf_table
> > array set to all 0s means that no vlan ID filtering is done.
> >
> > So, it is better to go with all 0s rather than all 1s, it is also more
> > efficient for the emulation.
> >
> > This is an existing problem in vmxnet3_dev_vlan_offload_set as well.
>
> That may be the case, but this patch isn't the place to address that.
> If this needs changed, there should be a follow on patch that makes
> this change.
Fair enough. You have my ack for this patch.
>
> >
> > > vmxnet3_dev_set_rxmode(hw, VMXNET3_RXM_PROMISC, 0);
> > > VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD,
> > > VMXNET3_CMD_UPDATE_VLAN_FILTERS);
> > > --
> > > 2.1.4
> > >
> > >
>
On Mon, 24 Jul 2017, Charles (Chas) Williams wrote:
> We should only restore shadow_vfta when hw_vlan_filter is active.
> Otherwise, we should restore the previous filtering behavior.
>
> Fixes: f003fc383487("vmxnet3: enable vlan filtering")
> Cc: stable@dpdk.org
>
> Signed-off-by: Chas Williams <ciwillia@brocade.com>
Acked-by: Shrikrishna Khare <skhare@vmware.com>
> > We should only restore shadow_vfta when hw_vlan_filter is active.
> > Otherwise, we should restore the previous filtering behavior.
> >
> > Fixes: f003fc383487("vmxnet3: enable vlan filtering")
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Chas Williams <ciwillia@brocade.com>
>
> Acked-by: Shrikrishna Khare <skhare@vmware.com>
Applied, thanks
@@ -1219,7 +1219,10 @@ vmxnet3_dev_promiscuous_disable(struct rte_eth_dev *dev)
struct vmxnet3_hw *hw = dev->data->dev_private;
uint32_t *vf_table = hw->shared->devRead.rxFilterConf.vfTable;
- memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
+ if (dev->data->dev_conf.rxmode.hw_vlan_filter)
+ memcpy(vf_table, hw->shadow_vfta, VMXNET3_VFT_TABLE_SIZE);
+ else
+ memset(vf_table, 0xff, VMXNET3_VFT_TABLE_SIZE);
vmxnet3_dev_set_rxmode(hw, VMXNET3_RXM_PROMISC, 0);
VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD,
VMXNET3_CMD_UPDATE_VLAN_FILTERS);