On 11/03/18 18:54, Sven Eckelmann wrote:
On Sonntag, 11. März 2018 00:53:19 CET Linus Lüssing wrote:
This patchset adds netlink support for dumping DAT cache and multicast flags tables.
Changes in v2:
- Added missing includes, "#include <linux/netlink.h>" and "#include <uapi/linux/batman_adv.h>", to distributed-arp-table.c
- Added missing forward declaration for "struct netlink_callback" to distributed-arp-table.h
- Changed nla_put_u32() to nla_put_in_addr() for BATADV_ATTR_DC_ADDRESS, removed the then obsolete ntohl() conversion
- Added missing include, "#include <linux/netlink.h>", to multicast.c
- Changed nla_put_u8() to nla_put_u32() for BATADV_ATTR_MCAST_FLAGS and BATADV_ATTR_MCAST_FLAGS_PRIV to enhance extensibility
Thanks for the doing the requested changes.
@Antonio: Can you please check whether you are ok with the new netlink names and types.
[BATADV_ATTR_DC_ADDRESS] = { .type = NLA_U32 },[BATADV_ATTR_DC_HWADDRESS] = { .len = ETH_ALEN },[BATADV_ATTR_DC_VID] = { .type = NLA_U16 },
Personally I'd find "DAT_CACHE" easier to parse than "DC". I know we have the "batctl dc" command, but that's more to keep the CLI compact.
This said, DAT was meant to be extended in the future and not store only IPv4 addresses. This said, shouldn't we use a more specific name for the address?
Another observation: do these attributes need to be DAT specific? or can we just define (or reuse) a generic HWADDRESS attribute that we might already be using somewhere else (i.e. for TT)?
Regards,