On Thu, Mar 21, 2019 at 01:35:05PM -0700, Jakub Kicinski wrote:

On Thu, 21 Mar 2019 14:40:21 +0100 (CET), Michal Kubecek wrote:

quoted

Permanent hardware address of a network device was traditionally provided
via ethtool ioctl interface but as Jiri Pirko pointed out in a review of
ethtool netlink interface, rtnetlink is much more suitable for it so let's
add it to the RTM_NEWLINK message.

As permanent address is not modifiable, reject userspace requests
containing IFLA_PERM_ADDRESS attribute.

Note: we already provide permanent hardware address for bond slaves;
unfortunately we cannot drop that attribute for backward compatibility
reasons.

Signed-off-by: Michal Kubecek <redacted>
---
 include/uapi/linux/if_link.h | 1 +
 net/core/rtnetlink.c         | 4 ++++
 2 files changed, 5 insertions(+)

diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h
index 5b225ff63b48..351ef746b8b0 100644
--- a/include/uapi/linux/if_link.h
+++ b/include/uapi/linux/if_link.h

@@ -167,6 +167,7 @@ enum {
 	IFLA_NEW_IFINDEX,
 	IFLA_MIN_MTU,
 	IFLA_MAX_MTU,
+	IFLA_PERM_ADDRESS,
 	__IFLA_MAX
 };
 

diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index a51cab95ba64..a72e8f4d777b 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c

@@ -1026,6 +1026,7 @@ static noinline size_t if_nlmsg_size(const struct net_device *dev,
 	       + nla_total_size(4)  /* IFLA_CARRIER_DOWN_COUNT */
 	       + nla_total_size(4)  /* IFLA_MIN_MTU */
 	       + nla_total_size(4)  /* IFLA_MAX_MTU */
+	       + nla_total_size(MAX_ADDR_LEN) /* IFLA_PERM_ADDRESS */
 	       + 0;
 }
 

@@ -1683,6 +1684,8 @@ static int rtnl_fill_ifinfo(struct sk_buff *skb,
 	    nla_put_s32(skb, IFLA_NEW_IFINDEX, new_ifindex) < 0)
 		goto nla_put_failure;
 
+	if (nla_put(skb, IFLA_PERM_ADDRESS, dev->addr_len, dev->perm_addr))

Should we check the perm_addr is non-zero, i.e. it has been filled in
by the driver?

It makes sense logically but there is IMHO one practical problem: if we
omit the attribute when the address is zero (i.e. not set or maybe
a device which does not have permanent address), userspace won't be able
to distinguish this case from an older kernel not implementing the
attribute.

For the iproute2 patch, I plan to show the permanent address in the
output of "ip link show" only if it differs from current address.
I think it would be desirable to distinguish "current address is the
permanent one" from "no/unknown permanent address" but absence of the
attribute could also mean "kernel does not provide the information".

Michal