Re: [PATCH] i40e: fix i40e_validate_mac_addr to permit multicast addresses

Hi Beilei,

-----Original Message-----
From: Xing, Beilei [mailto:beilei.xing@intel.com]

Hi,

quoted

-----Original Message-----
From: David Harton [mailto:dharton@cisco.com]

The i40e maintains a single MAC filter table for both unicast and
multicast addresses.  The i40e_validate_mac_addr function was
preventing multicast addresses from being added to the table via
i40evf_add_mac_addr.  Fixed the issue by removing the multicast
address check in i40e_validate_mac_addr.

Maybe fix line is needed here.

Please elaborate.  If feel this is a regression and you want 
me to add a fixline can you offer the offending commit?

quoted

Signed-off-by: David Harton <redacted>
---
 drivers/net/i40e/base/i40e_common.c | 12 +++++-------
 drivers/net/i40e/i40e_ethdev.c      |  3 ++-
 2 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/drivers/net/i40e/base/i40e_common.c

b/drivers/net/i40e/base/i40e_common.c
index 900d379..9779854 100644

--- a/drivers/net/i40e/base/i40e_common.c
+++ b/drivers/net/i40e/base/i40e_common.c

@@ -969,10 +969,10 @@ struct i40e_rx_ptype_decoded i40e_ptype_lookup[]

= {


 /**
- * i40e_validate_mac_addr - Validate unicast MAC address
+ * i40e_validate_mac_addr - Validate MAC address
  * @mac_addr: pointer to MAC address
  *
- * Tests a MAC address to ensure it is a valid Individual Address
+ * Tests a MAC address to ensure it is a valid Address
  **/
 enum i40e_status_code i40e_validate_mac_addr(u8 *mac_addr)  { @@ -
980,13 +980,11 @@ enum i40e_status_code i40e_validate_mac_addr(u8
*mac_addr)

 	DEBUGFUNC("i40e_validate_mac_addr");

-	/* Broadcast addresses ARE multicast addresses
-	 * Make sure it is not a multicast address
+	/*
 	 * Reject the zero address
 	 */
-	if (I40E_IS_MULTICAST(mac_addr) ||
-	    (mac_addr[0] == 0 && mac_addr[1] == 0 && mac_addr[2] == 0 &&
-	      mac_addr[3] == 0 && mac_addr[4] == 0 && mac_addr[5] == 0))
+	if (mac_addr[0] == 0 && mac_addr[1] == 0 && mac_addr[2] == 0 &&
+	    mac_addr[3] == 0 && mac_addr[4] == 0 && mac_addr[5] == 0)
 		status = I40E_ERR_INVALID_MAC_ADDR;

We'd better not to change the base code which is from kernel driver. How
about changing the i40evf_add_mac_addr function as following:
-	if (i40e_validate_mac_addr(addr->addr_bytes) != I40E_SUCCESS) {
+	if (is_zero_ether_addr (addr->addr_bytes) != I40E_SUCCESS) {

Then following modification in eth_i40e_dev_init function can be removed.
But we should also pay attention to other functions which call
i40evf_add_mac_addr.

Sure.  I'll post a new patch as suggested once the fixline question is answered.  

I was trying to fix the issue the way I did in case i40e passthru 
wanted to benefit.  Your suggestion will work for the VF as the only 
other caller for i40evf_add_mac_addr() performs its own macaddr validation.

Thanks,
Dave

Beilei

quoted

 	return status;

diff --git a/drivers/net/i40e/i40e_ethdev.c

b/drivers/net/i40e/i40e_ethdev.c index 5f26e24..00b6082 100644

--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c

@@ -1199,7 +1199,8 @@ eth_i40e_dev_init(struct rte_eth_dev *dev)

 	/* Get and check the mac address */
 	i40e_get_mac_addr(hw, hw->mac.addr);
-	if (i40e_validate_mac_addr(hw->mac.addr) != I40E_SUCCESS) {
+	if (i40e_validate_mac_addr(hw->mac.addr) != I40E_SUCCESS ||
+	    I40E_IS_MULTICAST(hw->mac.addr)) {
 		PMD_INIT_LOG(ERR, "mac address is not valid");
 		ret = -EIO;
 		goto err_get_mac_addr;
--

2.10.3.dirty