linux-stable.git/net/ipv4, branch v6.0.6

netfilter: rpfilter/fib: Set ->flowic_uid correctly for user namespaces.

2022-10-29T08:08:34+00:00

[ Upstream commit 1fcc064b305a1aadeff0d4bff961094d27660acd ]

Currently netfilter's rpfilter and fib modules implicitely initialise
->flowic_uid with 0. This is normally the root UID. However, this isn't
the case in user namespaces, where user ID 0 is mapped to a different
kernel UID. By initialising ->flowic_uid with sock_net_uid(), we get
the root UID of the user namespace, thus keeping the same behaviour
whether or not we're running in a user namepspace.

Note, this is similar to commit 8bcfd0925ef1 ("ipv4: add missing
initialization for flowi4_uid"), which fixed the rp_filter sysctl.

Fixes: 622ec2c9d524 ("net: core: add UID to flows, rules, and routes")
Signed-off-by: Guillaume Nault 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: rpfilter/fib: Populate flowic_l3mdev field

2022-10-29T08:08:34+00:00

[ Upstream commit acc641ab95b66b813c1ce856c377a2bbe71e7f52 ]

Use the introduced field for correct operation with VRF devices instead
of conditionally overwriting flowic_oif. This is a partial revert of
commit b575b24b8eee3 ("netfilter: Fix rpfilter dropping vrf packets by
mistake"), implementing a simpler solution.

Signed-off-by: Phil Sutter 
Reviewed-by: David Ahern 
Reviewed-by: Guillaume Nault 
Signed-off-by: Florian Westphal 
Stable-dep-of: 1fcc064b305a ("netfilter: rpfilter/fib: Set ->flowic_uid correctly for user namespaces.")
Signed-off-by: Sasha Levin

udp: Update reuse->has_conns under reuseport_lock.

2022-10-29T08:08:33+00:00

[ Upstream commit 69421bf98482d089e50799f45e48b25ce4a8d154 ]

When we call connect() for a UDP socket in a reuseport group, we have
to update sk->sk_reuseport_cb->has_conns to 1.  Otherwise, the kernel
could select a unconnected socket wrongly for packets sent to the
connected socket.

However, the current way to set has_conns is illegal and possible to
trigger that problem.  reuseport_has_conns() changes has_conns under
rcu_read_lock(), which upgrades the RCU reader to the updater.  Then,
it must do the update under the updater's lock, reuseport_lock, but
it doesn't for now.

For this reason, there is a race below where we fail to set has_conns
resulting in the wrong socket selection.  To avoid the race, let's split
the reader and updater with proper locking.

 cpu1                               cpu2
+----+                             +----+

__ip[46]_datagram_connect()        reuseport_grow()
.                                  .
|- reuseport_has_conns(sk, true)   |- more_reuse = __reuseport_alloc(more_socks_size)
|  .                               |
|  |- rcu_read_lock()
|  |- reuse = rcu_dereference(sk->sk_reuseport_cb)
|  |
|  |                               |  /* reuse->has_conns == 0 here */
|  |                               |- more_reuse->has_conns = reuse->has_conns
|  |- reuse->has_conns = 1         |  /* more_reuse->has_conns SHOULD BE 1 HERE */
|  |                               |
|  |                               |- rcu_assign_pointer(reuse->socks[i]->sk_reuseport_cb,
|  |                               |                     more_reuse)
|  `- rcu_read_unlock()            `- kfree_rcu(reuse, rcu)
|
|- sk->sk_state = TCP_ESTABLISHED

Note the likely(reuse) in reuseport_has_conns_set() is always true,
but we put the test there for ease of review.  [0]

For the record, usually, sk_reuseport_cb is changed under lock_sock().
The only exception is reuseport_grow() & TCP reqsk migration case.

  1) shutdown() TCP listener, which is moved into the latter part of
     reuse->socks[] to migrate reqsk.

  2) New listen() overflows reuse->socks[] and call reuseport_grow().

  3) reuse->max_socks overflows u16 with the new listener.

  4) reuseport_grow() pops the old shutdown()ed listener from the array
     and update its sk->sk_reuseport_cb as NULL without lock_sock().

shutdown()ed TCP sk->sk_reuseport_cb can be changed without lock_sock(),
but, reuseport_has_conns_set() is called only for UDP under lock_sock(),
so likely(reuse) never be false in reuseport_has_conns_set().

[0]: https://lore.kernel.org/netdev/CANn89iLja=eQHbsM_Ta2sQF0tOGU8vAGrh_izRuuHjuO1ouUag@mail.gmail.com/

Fixes: acdcecc61285 ("udp: correct reuseport selection with connected sockets")
Signed-off-by: Kuniyuki Iwashima 
Link: https://lore.kernel.org/r/20221014182625.89913-1-kuniyu@amazon.com
Signed-off-by: Paolo Abeni 
Signed-off-by: Sasha Levin

net: flag sockets supporting msghdr originated zerocopy

2022-10-26T10:22:56+00:00

commit e993ffe3da4bcddea0536b03be1031bf35cd8d85 upstream.

We need an efficient way in io_uring to check whether a socket supports
zerocopy with msghdr provided ubuf_info. Add a new flag into the struct
socket flags fields.

Cc:  # 6.0
Signed-off-by: Pavel Begunkov 
Acked-by: Jakub Kicinski 
Link: https://lore.kernel.org/r/3dafafab822b1c66308bb58a0ac738b1e3f53f74.1666346426.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

net-next: Fix IP_UNICAST_IF option behavior for connected sockets

2022-10-21T10:39:07+00:00

[ Upstream commit 0e4d354762cefd3e16b4cff8988ff276e45effc4 ]

The IP_UNICAST_IF socket option is used to set the outgoing interface
for outbound packets.

The IP_UNICAST_IF socket option was added as it was needed by the
Wine project, since no other existing option (SO_BINDTODEVICE socket
option, IP_PKTINFO socket option or the bind function) provided the
needed characteristics needed by the IP_UNICAST_IF socket option. [1]
The IP_UNICAST_IF socket option works well for unconnected sockets,
that is, the interface specified by the IP_UNICAST_IF socket option
is taken into consideration in the route lookup process when a packet
is being sent. However, for connected sockets, the outbound interface
is chosen when connecting the socket, and in the route lookup process
which is done when a packet is being sent, the interface specified by
the IP_UNICAST_IF socket option is being ignored.

This inconsistent behavior was reported and discussed in an issue
opened on systemd's GitHub project [2]. Also, a bug report was
submitted in the kernel's bugzilla [3].

To understand the problem in more detail, we can look at what happens
for UDP packets over IPv4 (The same analysis was done separately in
the referenced systemd issue).
When a UDP packet is sent the udp_sendmsg function gets called and
the following happens:

1. The oif member of the struct ipcm_cookie ipc (which stores the
output interface of the packet) is initialized by the ipcm_init_sk
function to inet->sk.sk_bound_dev_if (the device set by the
SO_BINDTODEVICE socket option).

2. If the IP_PKTINFO socket option was set, the oif member gets
overridden by the call to the ip_cmsg_send function.

3. If no output interface was selected yet, the interface specified
by the IP_UNICAST_IF socket option is used.

4. If the socket is connected and no destination address is
specified in the send function, the struct ipcm_cookie ipc is not
taken into consideration and the cached route, that was calculated in
the connect function is being used.

Thus, for a connected socket, the IP_UNICAST_IF sockopt isn't taken
into consideration.

This patch corrects the behavior of the IP_UNICAST_IF socket option
for connect()ed sockets by taking into consideration the
IP_UNICAST_IF sockopt when connecting the socket.

In order to avoid reconnecting the socket, this option is still
ignored when applied on an already connected socket until connect()
is called again by the Richard Gobert.

Change the __ip4_datagram_connect function, which is called during
socket connection, to take into consideration the interface set by
the IP_UNICAST_IF socket option, in a similar way to what is done in
the udp_sendmsg function.

[1] https://lore.kernel.org/netdev/1328685717.4736.4.camel@edumazet-laptop/T/
[2] https://github.com/systemd/systemd/issues/11935#issuecomment-618691018
[3] https://bugzilla.kernel.org/show_bug.cgi?id=210255

Signed-off-by: Richard Gobert 
Reviewed-by: David Ahern 
Link: https://lore.kernel.org/r/20220829111554.GA1771@debian
Signed-off-by: Jakub Kicinski 
Signed-off-by: Sasha Levin

tcp: annotate data-race around tcp_md5sig_pool_populated

2022-10-21T10:39:06+00:00

[ Upstream commit aacd467c0a576e5e44d2de4205855dc0fe43f6fb ]

tcp_md5sig_pool_populated can be read while another thread
changes its value.

The race has no consequence because allocations
are protected with tcp_md5sig_mutex.

This patch adds READ_ONCE() and WRITE_ONCE() to document
the race and silence KCSAN.

Reported-by: Abhishek Shah 
Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller 
Signed-off-by: Sasha Levin

once: add DO_ONCE_SLOW() for sleepable contexts

2022-10-21T10:38:21+00:00

[ Upstream commit 62c07983bef9d3e78e71189441e1a470f0d1e653 ]

Christophe Leroy reported a ~80ms latency spike
happening at first TCP connect() time.

This is because __inet_hash_connect() uses get_random_once()
to populate a perturbation table which became quite big
after commit 4c2c8f03a5ab ("tcp: increase source port perturb table to 2^16")

get_random_once() uses DO_ONCE(), which block hard irqs for the duration
of the operation.

This patch adds DO_ONCE_SLOW() which uses a mutex instead of a spinlock
for operations where we prefer to stay in process context.

Then __inet_hash_connect() can use get_random_slow_once()
to populate its perturbation table.

Fixes: 4c2c8f03a5ab ("tcp: increase source port perturb table to 2^16")
Fixes: 190cc82489f4 ("tcp: change source port randomizarion at connect() time")
Reported-by: Christophe Leroy 
Link: https://lore.kernel.org/netdev/CANn89iLAEYBaoYajy0Y9UmGFff5GPxDUoG-ErVB2jDdRNQ5Tug@mail.gmail.com/T/#t
Signed-off-by: Eric Dumazet 
Cc: Willy Tarreau 
Tested-by: Christophe Leroy 
Signed-off-by: David S. Miller 
Signed-off-by: Sasha Levin

tcp: fix tcp_cwnd_validate() to not forget is_cwnd_limited

2022-10-21T10:38:19+00:00

[ Upstream commit f4ce91ce12a7c6ead19b128ffa8cff6e3ded2a14 ]

This commit fixes a bug in the tracking of max_packets_out and
is_cwnd_limited. This bug can cause the connection to fail to remember
that is_cwnd_limited is true, causing the connection to fail to grow
cwnd when it should, causing throughput to be lower than it should be.

The following event sequence is an example that triggers the bug:

 (a) The connection is cwnd_limited, but packets_out is not at its
     peak due to TSO deferral deciding not to send another skb yet.
     In such cases the connection can advance max_packets_seq and set
     tp->is_cwnd_limited to true and max_packets_out to a small
     number.

(b) Then later in the round trip the connection is pacing-limited (not
     cwnd-limited), and packets_out is larger. In such cases the
     connection would raise max_packets_out to a bigger number but
     (unexpectedly) flip tp->is_cwnd_limited from true to false.

This commit fixes that bug.

One straightforward fix would be to separately track (a) the next
window after max_packets_out reaches a maximum, and (b) the next
window after tp->is_cwnd_limited is set to true. But this would
require consuming an extra u32 sequence number.

Instead, to save space we track only the most important
information. Specifically, we track the strongest available signal of
the degree to which the cwnd is fully utilized:

(1) If the connection is cwnd-limited then we remember that fact for
the current window.

(2) If the connection not cwnd-limited then we track the maximum
number of outstanding packets in the current window.

In particular, note that the new logic cannot trigger the buggy
(a)/(b) sequence above because with the new logic a condition where
tp->packets_out > tp->max_packets_out can only trigger an update of
tp->is_cwnd_limited if tp->is_cwnd_limited is false.

This first showed up in a testing of a BBRv2 dev branch, but this
buggy behavior highlighted a general issue with the
tcp_cwnd_validate() logic that can cause cwnd to fail to increase at
the proper rate for any TCP congestion control, including Reno or
CUBIC.

Fixes: ca8a22634381 ("tcp: make cwnd-limited checks measurement-based, and gentler")
Signed-off-by: Neal Cardwell 
Signed-off-by: Kevin(Yudong) Yang 
Signed-off-by: Yuchung Cheng 
Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller 
Signed-off-by: Sasha Levin

netfilter: nft_fib: Fix for rpath check with VRF devices

2022-10-21T10:38:18+00:00

[ Upstream commit 2a8a7c0eaa8747c16aa4a48d573aa920d5c00a5c ]

Analogous to commit b575b24b8eee3 ("netfilter: Fix rpfilter
dropping vrf packets by mistake") but for nftables fib expression:
Add special treatment of VRF devices so that typical reverse path
filtering via 'fib saddr . iif oif' expression works as expected.

Fixes: f6d0cbcf09c50 ("netfilter: nf_tables: add fib expression")
Signed-off-by: Phil Sutter 
Signed-off-by: Florian Westphal 
Signed-off-by: Sasha Levin

esp: choose the correct inner protocol for GSO on inter address family tunnels

2022-10-21T10:38:04+00:00

[ Upstream commit 26dbd66eab8080be51759e48280da04015221e22 ]

Commit 23c7f8d7989e ("net: Fix esp GSO on inter address family
tunnels.") is incomplete. It passes to skb_eth_gso_segment the
protocol for the outer IP version, instead of the inner IP version, so
we end up calling inet_gso_segment on an inner IPv6 packet and
ipv6_gso_segment on an inner IPv4 packet and the packets are dropped.

This patch completes the fix by selecting the correct protocol based
on the inner mode's family.

Fixes: c35fe4106b92 ("xfrm: Add mode handlers for IPsec on layer 2")
Signed-off-by: Sabrina Dubroca 
Signed-off-by: Steffen Klassert 
Signed-off-by: Sasha Levin