[dpdk-dev] [PATCH] e1000: fix setting of VF MAC address

[Lu, Wenzhuo]

Hi,

> > > Signed-off-by: Bernard Iremonger <bernard.iremonger at intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu at intel.com>

> > > ---
> > >  drivers/net/e1000/igb_ethdev.c | 12 +++++++++++-
> > >  drivers/net/e1000/igb_pf.c     |  8 +++++---
> > >  2 files changed, 16 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/drivers/net/e1000/igb_ethdev.c
> > > b/drivers/net/e1000/igb_ethdev.c index 4ed5e95..f1044b7 100644
> > > --- a/drivers/net/e1000/igb_ethdev.c
> > > +++ b/drivers/net/e1000/igb_ethdev.c
> > > @@ -1,7 +1,7 @@
> > >  /*-
> > >   *   BSD LICENSE
> > >   *
> > > - *   Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
> > > + *   Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
> > >   *   All rights reserved.
> > >   *
> > >   *   Redistribution and use in source and binary forms, with or without
> > > @@ -2819,6 +2819,7 @@ igbvf_dev_close(struct rte_eth_dev *dev)
> > >  	struct e1000_hw *hw = E1000_DEV_PRIVATE_TO_HW(dev->data-
> > > >dev_private);
> > >  	struct e1000_adapter *adapter =
> > >  		E1000_DEV_PRIVATE(dev->data->dev_private);
> > > +	struct ether_addr addr;
> > >
> > >  	PMD_INIT_FUNC_TRACE();
> > >
> > > @@ -2827,6 +2828,15 @@ igbvf_dev_close(struct rte_eth_dev *dev)
> > >  	igbvf_dev_stop(dev);
> > >  	adapter->stopped = 1;
> > >  	igb_dev_free_queues(dev);
> > > +
> > > +	/**
> > > +	 * reprogram the RAR with a zero mac address,
> > > +	 * to ensure that the VF traffic goes to the PF
> > > +	 * after stop, close and detach of the VF.
> > > +	 **/
> > > +
> > > +	memset(&addr, 0, sizeof(addr));
> > > +	igbvf_default_mac_addr_set(dev, &addr);
> > >  }
> > >
> > >  static int igbvf_set_vfta(struct e1000_hw *hw, uint16_t vid, bool
> > > on) diff --git a/drivers/net/e1000/igb_pf.c
> > > b/drivers/net/e1000/igb_pf.c index
> > > 1d00dda..95204e9 100644
> > > --- a/drivers/net/e1000/igb_pf.c
> > > +++ b/drivers/net/e1000/igb_pf.c
> > > @@ -1,7 +1,7 @@
> > >  /*-
> > >   *   BSD LICENSE
> > >   *
> > > - *   Copyright(c) 2010-2015 Intel Corporation. All rights reserved.
> > > + *   Copyright(c) 2010-2016 Intel Corporation. All rights reserved.
> > >   *   All rights reserved.
> > >   *
> > >   *   Redistribution and use in source and binary forms, with or without
> > > @@ -332,8 +332,10 @@ igb_vf_set_mac_addr(struct rte_eth_dev *dev,
> > > uint32_t vf, uint32_t *msgbuf)
> > >  	int rar_entry = hw->mac.rar_entry_count - (vf + 1);
> > >  	uint8_t *new_mac = (uint8_t *)(&msgbuf[1]);
> > >
> > > -	if (is_valid_assigned_ether_addr((struct ether_addr*)new_mac)) {
> > > -		rte_memcpy(vfinfo[vf].vf_mac_addresses, new_mac, 6);
> > > +	if (is_unicast_ether_addr((struct ether_addr *)new_mac)) {
> > > +		if (!is_zero_ether_addr((struct ether_addr *)new_mac))
> > > +			rte_memcpy(vfinfo[vf].vf_mac_addresses,
> > new_mac,
> > > +				sizeof(vfinfo[vf].vf_mac_addresses));
> > >  		hw->mac.ops.rar_set(hw, new_mac, rar_entry);
> > If the new mac is 0, after this, the rar is 0, but the address stored
> > in vfinfo is not changed and surely not 0. Right?
> > So, they're not align with each other. Could it cause some problem?
> >
> > >  		return 0;
> > >  	}
> > > --
> > > 2.6.3
> 
>  vfinfo[vf].vf_mac_addresses  contains the perm_addr MAC address.
> I do not want to overwrite this with a zero address. If the VF is attached again it
> uses the perm_addr  MAC address.
> The follow sequence works fine for me in testpmd, the VF is port 1:
> 
> Testpmd> port stop 1
> Testpmd> port close 1 		/* VF MAC address is set to zero here */
> Testpmd> port detach 1
> Testpmd> port attach 0000:04:10.0
> Testpmd> show port info 1	/* VF MAC address is perm_addr MAC address
> */
I think you're right. Only when the rar is 0, the vf_mac_addresses is different from rar. So, we can use it to keep the permanent address.

> 
> Regards,
> 
> Bernard.