Merge branch 'rtnetlink-Add-support-for-rigid-checking-of-data-in-dump-request'

Merge branch 'rtnetlink-Add-support-for-rigid-checking-of-data-in-dump-request'

David Ahern says:

====================
rtnetlink: Add support for rigid checking of data in dump request

There are many use cases where a user wants to influence what is
returned in a dump for some rtnetlink command: one is wanting data
for a different namespace than the one the request is received and
another is limiting the amount of data returned in the dump to a
specific set of interest to userspace, reducing the cpu overhead of
both kernel and userspace. Unfortunately, the kernel has historically
not been strict with checking for the proper header or checking the
values passed in the header. This lenient implementation has allowed
iproute2 and other packages to pass any struct or data in the dump
request as long as the family is the first byte. For example, ifinfomsg
struct is used by iproute2 for all generic dump requests - links,
addresses, routes and rules when it is really only valid for link
requests.

There is 1 is example where the kernel deals with the wrong struct: link
dumps after VF support was added. Older iproute2 was sending rtgenmsg as
the header instead of ifinfomsg so a patch was added to try and detect
old userspace vs new:
e5eca6d41f53 ("rtnetlink: fix userspace API breakage for iproute2 < v3.9.0")

The latest example is Christian's patch set wanting to return addresses for
a target namespace. It guesses the header struct is an ifaddrmsg and if it
guesses wrong a netlink warning is generated in the kernel log on every
address dump which is unacceptable.

Another example where the kernel is a bit lenient is route dumps: iproute2
can send either a request with either ifinfomsg or a rtmsg as the header
struct, yet the kernel always treats the header as an rtmsg (see
inet_dump_fib and rtm_flags check). The header inconsistency impacts the
ability to add kernel side filters for route dumps - a necessary feature
for scale setups with 100k+ routes.

How to resolve the problem of not breaking old userspace yet be able to
move forward with new features such as kernel side filtering which are
crucial for efficient operation at high scale?

This patch set addresses the problem by adding a new socket flag,
NETLINK_DUMP_STRICT_CHK, that userspace can use with setsockopt to
request strict checking of headers and attributes on dump requests and
hence unlock the ability to use kernel side filters as they are added.

Kernel side, the dump handlers are updated to verify the message contains
at least the expected header struct:
    RTM_GETLINK:       ifinfomsg
    RTM_GETADDR:       ifaddrmsg
    RTM_GETMULTICAST:  ifaddrmsg
    RTM_GETANYCAST:    ifaddrmsg
    RTM_GETADDRLABEL:  ifaddrlblmsg
    RTM_GETROUTE:      rtmsg
    RTM_GETSTATS:      if_stats_msg
    RTM_GETNEIGH:      ndmsg
    RTM_GETNEIGHTBL:   ndtmsg
    RTM_GETNSID:       rtgenmsg
    RTM_GETRULE:       fib_rule_hdr
    RTM_GETNETCONF:    netconfmsg
    RTM_GETMDB:        br_port_msg

And then every field in the header struct should be 0 with the exception
of the family. There are a few exceptions to this rule where the kernel
already influences the data returned by values in the struct. Next the
message should not contain attributes unless the kernel implements
filtering for it. Any unexpected data causes the dump to fail with EINVAL.
If the new flag is honored by the kernel and the dump contents adjusted
by any data passed in the request, the dump handler can set the
NLM_F_DUMP_FILTERED flag in the netlink message header.

For old userspace on new kernel there is no impact as all checks are
wrapped in a check on the new strict flag. For new userspace on old
kernel, the data in the headers and any appended attributes are
silently ignored though the setsockopt failing is the clue to userspace
the feature is not supported. New userspace on new kernel gets the
requested data dump.

iproute2 patches can be found here:
    https://github.com/dsahern/iproute2 dump-enhancements

Major changes since v1
- inner header is supposed to be 4-bytes aligned. So for dumps that
  should not have attributes appended changed the check to use:
        if (nlmsg_attrlen(nlh, sizeof(hdr)))
  Only impacts patches with headers that are not multiples of 4-bytes
  (rtgenmsg, netconfmsg), but applied the change to all patches not
  calling nlmsg_parse for consistency.

- Added nlmsg_parse_strict and nla_parse_strict for tighter control on
  attribute parsing. There should be no unknown attribute types or extra
  bytes.

- Moved validation to a helper in most cases

Changes since rfc-v2
- dropped the NLM_F_DUMP_FILTERED flag from target nsid dumps per
  Jiri's objections
- changed the opt-in uapi from a netlink message flag to a socket
  flag. setsockopt provides an api for userspace to definitively
  know if the kernel supports strict checking on dumps.
- re-ordered patches to peel off the extack on dumps if needed to
  keep this set size within limits
- misc cleanups in patches based on testing
====================

Acked-by: Christian Brauner <christian@brauner.io>
Signed-off-by: David S. Miller <davem@davemloft.net>