vrf: Do not attempt to create IPv6 mcast rule if IPv6 is disabled

A user reported that vrf create fails when IPv6 is disabled at boot using
'ipv6.disable=1':
   https://bugzilla.kernel.org/show_bug.cgi?id=204903

The failure is adding fib rules at create time. Add RTNL_FAMILY_IP6MR to
the check in vrf_fib_rule if ipv6_mod_enabled is disabled.

Fixes: e4a38c0c4b27 ("ipv6: add vrf table handling code for ipv6 mcast")
Signed-off-by: David Ahern <dsahern@gmail.com>
Cc: Patrick Ruddy <pruddy@vyatta.att-mail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge git://git./pub/scm/linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
pull-request: bpf 2019-09-27

The following pull-request contains BPF updates for your *net* tree.

The main changes are:

1) Fix libbpf's BTF dumper to not skip anonymous enum definitions, from Andrii.

2) Fix BTF verifier issues when handling the BTF of vmlinux, from Alexei.

3) Fix nested calls into bpf_event_output() from TCP sockops BPF
   programs, from Allan.

4) Fix NULL pointer dereference in AF_XDP's xsk map creation when
   allocation fails, from Jonathan.

5) Remove unneeded 64 byte alignment requirement of the AF_XDP UMEM
   headroom, from Bjorn.

6) Remove unused XDP_OPTIONS getsockopt() call which results in an error
   on older kernels, from Toke.

7) Fix a client/server race in tcp_rtt BPF kselftest case, from Stanislav.

8) Fix indentation issue in BTF's btf_enum_check_kflag_member(), from Colin.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'qdisc-destroy'

Vlad Buslov says:

====================
Fix Qdisc destroy issues caused by adding fine-grained locking to filter API

TC filter API unlocking introduced several new fine-grained locks. The
change caused sleeping-while-atomic BUGs in several Qdiscs that call cls
APIs which need to obtain new mutex while holding sch tree spinlock. This
series fixes affected Qdiscs by ensuring that cls API that became sleeping
is only called outside of sch tree lock critical section.
====================

Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: sched: sch_sfb: don't call qdisc_put() while holding tree lock

Recent changes that removed rtnl dependency from rules update path of tc
also made tcf_block_put() function sleeping. This function is called from
ops->destroy() of several Qdisc implementations, which in turn is called by
qdisc_put(). Some Qdiscs call qdisc_put() while holding sch tree spinlock,
which results sleeping-while-atomic BUG.

Steps to reproduce for sfb:

tc qdisc add dev ens1f0 handle 1: root sfb
tc qdisc add dev ens1f0 parent 1:10 handle 50: sfq perturb 10
tc qdisc change dev ens1f0 root handle 1: sfb

Resulting dmesg:

[ 7265.938717] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:909
[ 7265.940152] in_atomic(): 1, irqs_disabled(): 0, pid: 28579, name: tc
[ 7265.941455] INFO: lockdep is turned off.
[ 7265.942744] CPU: 11 PID: 28579 Comm: tc Tainted: G        W         5.3.0-rc8+ #721
[ 7265.944065] Hardware name: Supermicro SYS-2028TP-DECR/X10DRT-P, BIOS 2.0b 03/30/2017
[ 7265.945396] Call Trace:
[ 7265.946709]  dump_stack+0x85/0xc0
[ 7265.947994]  ___might_sleep.cold+0xac/0xbc
[ 7265.949282]  __mutex_lock+0x5b/0x960
[ 7265.950543]  ? tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 7265.951803]  ? tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 7265.953022]  tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 7265.954248]  tcf_block_put_ext.part.0+0x21/0x50
[ 7265.955478]  tcf_block_put+0x50/0x70
[ 7265.956694]  sfq_destroy+0x15/0x50 [sch_sfq]
[ 7265.957898]  qdisc_destroy+0x5f/0x160
[ 7265.959099]  sfb_change+0x175/0x330 [sch_sfb]
[ 7265.960304]  tc_modify_qdisc+0x324/0x840
[ 7265.961503]  rtnetlink_rcv_msg+0x170/0x4b0
[ 7265.962692]  ? netlink_deliver_tap+0x95/0x400
[ 7265.963876]  ? rtnl_dellink+0x2d0/0x2d0
[ 7265.965064]  netlink_rcv_skb+0x49/0x110
[ 7265.966251]  netlink_unicast+0x171/0x200
[ 7265.967427]  netlink_sendmsg+0x224/0x3f0
[ 7265.968595]  sock_sendmsg+0x5e/0x60
[ 7265.969753]  ___sys_sendmsg+0x2ae/0x330
[ 7265.970916]  ? ___sys_recvmsg+0x159/0x1f0
[ 7265.972074]  ? do_wp_page+0x9c/0x790
[ 7265.973233]  ? __handle_mm_fault+0xcd3/0x19e0
[ 7265.974407]  __sys_sendmsg+0x59/0xa0
[ 7265.975591]  do_syscall_64+0x5c/0xb0
[ 7265.976753]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 7265.977938] RIP: 0033:0x7f229069f7b8
[ 7265.979117] Code: 89 02 48 c7 c0 ff ff ff ff eb bb 0f 1f 80 00 00 00 00 f3 0f 1e fa 48 8d 05 65 8f 0c 00 8b 00 85 c0 75 17 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 58 c3 0f 1f 80 00 00 00 00 48 83 ec 28 89 5
4
[ 7265.981681] RSP: 002b:00007ffd7ed2d158 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
[ 7265.983001] RAX: ffffffffffffffda RBX: 000000005d813ca1 RCX: 00007f229069f7b8
[ 7265.984336] RDX: 0000000000000000 RSI: 00007ffd7ed2d1c0 RDI: 0000000000000003
[ 7265.985682] RBP: 0000000000000000 R08: 0000000000000001 R09: 000000000165c9a0
[ 7265.987021] R10: 0000000000404eda R11: 0000000000000246 R12: 0000000000000001
[ 7265.988309] R13: 000000000047f640 R14: 0000000000000000 R15: 0000000000000000

In sfb_change() function use qdisc_purge_queue() instead of
qdisc_tree_flush_backlog() to properly reset old child Qdisc and save
pointer to it into local temporary variable. Put reference to Qdisc after
sch tree lock is released in order not to call potentially sleeping cls API
in atomic section. This is safe to do because Qdisc has already been reset
by qdisc_purge_queue() inside sch tree lock critical section.

Reported-by: syzbot+ac54455281db908c581e@syzkaller.appspotmail.com
Fixes: c266f64dbfa2 ("net: sched: protect block state with mutex")
Suggested-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: sched: multiq: don't call qdisc_put() while holding tree lock

Recent changes that removed rtnl dependency from rules update path of tc
also made tcf_block_put() function sleeping. This function is called from
ops->destroy() of several Qdisc implementations, which in turn is called by
qdisc_put(). Some Qdiscs call qdisc_put() while holding sch tree spinlock,
which results sleeping-while-atomic BUG.

Steps to reproduce for multiq:

tc qdisc add dev ens1f0 root handle 1: multiq
tc qdisc add dev ens1f0 parent 1:10 handle 50: sfq perturb 10
ethtool -L ens1f0 combined 2
tc qdisc change dev ens1f0 root handle 1: multiq

Resulting dmesg:

[ 5539.419344] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:909
[ 5539.420945] in_atomic(): 1, irqs_disabled(): 0, pid: 27658, name: tc
[ 5539.422435] INFO: lockdep is turned off.
[ 5539.423904] CPU: 21 PID: 27658 Comm: tc Tainted: G        W         5.3.0-rc8+ #721
[ 5539.425400] Hardware name: Supermicro SYS-2028TP-DECR/X10DRT-P, BIOS 2.0b 03/30/2017
[ 5539.426911] Call Trace:
[ 5539.428380]  dump_stack+0x85/0xc0
[ 5539.429823]  ___might_sleep.cold+0xac/0xbc
[ 5539.431262]  __mutex_lock+0x5b/0x960
[ 5539.432682]  ? tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 5539.434103]  ? __nla_validate_parse+0x51/0x840
[ 5539.435493]  ? tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 5539.436903]  tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 5539.438327]  tcf_block_put_ext.part.0+0x21/0x50
[ 5539.439752]  tcf_block_put+0x50/0x70
[ 5539.441165]  sfq_destroy+0x15/0x50 [sch_sfq]
[ 5539.442570]  qdisc_destroy+0x5f/0x160
[ 5539.444000]  multiq_tune+0x14a/0x420 [sch_multiq]
[ 5539.445421]  tc_modify_qdisc+0x324/0x840
[ 5539.446841]  rtnetlink_rcv_msg+0x170/0x4b0
[ 5539.448269]  ? netlink_deliver_tap+0x95/0x400
[ 5539.449691]  ? rtnl_dellink+0x2d0/0x2d0
[ 5539.451116]  netlink_rcv_skb+0x49/0x110
[ 5539.452522]  netlink_unicast+0x171/0x200
[ 5539.453914]  netlink_sendmsg+0x224/0x3f0
[ 5539.455304]  sock_sendmsg+0x5e/0x60
[ 5539.456686]  ___sys_sendmsg+0x2ae/0x330
[ 5539.458071]  ? ___sys_recvmsg+0x159/0x1f0
[ 5539.459461]  ? do_wp_page+0x9c/0x790
[ 5539.460846]  ? __handle_mm_fault+0xcd3/0x19e0
[ 5539.462263]  __sys_sendmsg+0x59/0xa0
[ 5539.463661]  do_syscall_64+0x5c/0xb0
[ 5539.465044]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 5539.466454] RIP: 0033:0x7f1fe08177b8
[ 5539.467863] Code: 89 02 48 c7 c0 ff ff ff ff eb bb 0f 1f 80 00 00 00 00 f3 0f 1e fa 48 8d 05 65 8f 0c 00 8b 00 85 c0 75 17 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 58 c3 0f 1f 80 00 00 00 00 48 83 ec 28 89 5
4
[ 5539.470906] RSP: 002b:00007ffe812de5d8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
[ 5539.472483] RAX: ffffffffffffffda RBX: 000000005d8135e3 RCX: 00007f1fe08177b8
[ 5539.474069] RDX: 0000000000000000 RSI: 00007ffe812de640 RDI: 0000000000000003
[ 5539.475655] RBP: 0000000000000000 R08: 0000000000000001 R09: 000000000182e9b0
[ 5539.477203] R10: 0000000000404eda R11: 0000000000000246 R12: 0000000000000001
[ 5539.478699] R13: 000000000047f640 R14: 0000000000000000 R15: 0000000000000000

Rearrange locking in multiq_tune() in following ways:

- In loop that removes Qdiscs from disabled queues, call
  qdisc_purge_queue() instead of qdisc_tree_flush_backlog() on Qdisc that
  is being destroyed. Save the Qdisc in temporary allocated array and call
  qdisc_put() on each element of the array after sch tree lock is released.
  This is safe to do because Qdiscs have already been reset by
  qdisc_purge_queue() inside sch tree lock critical section.

- Do the same change for second loop that initializes Qdiscs for newly
  enabled queues in multiq_tune() function. Since sch tree lock is obtained
  and released on each iteration of this loop, just call qdisc_put()
  directly outside of critical section. Don't verify that old Qdisc is not
  noop_qdisc before releasing reference to it because such check is already
  performed by qdisc_put*() functions.

Fixes: c266f64dbfa2 ("net: sched: protect block state with mutex")
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: sched: sch_htb: don't call qdisc_put() while holding tree lock

Recent changes that removed rtnl dependency from rules update path of tc
also made tcf_block_put() function sleeping. This function is called from
ops->destroy() of several Qdisc implementations, which in turn is called by
qdisc_put(). Some Qdiscs call qdisc_put() while holding sch tree spinlock,
which results sleeping-while-atomic BUG.

Steps to reproduce for htb:

tc qdisc add dev ens1f0 root handle 1: htb default 12
tc class add dev ens1f0 parent 1: classid 1:1 htb rate 100kbps ceil 100kbps
tc qdisc add dev ens1f0 parent 1:1 handle 40: sfq perturb 10
tc class add dev ens1f0 parent 1:1 classid 1:2 htb rate 100kbps ceil 100kbps

Resulting dmesg:

[ 4791.148551] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:909
[ 4791.151354] in_atomic(): 1, irqs_disabled(): 0, pid: 27273, name: tc
[ 4791.152805] INFO: lockdep is turned off.
[ 4791.153605] CPU: 19 PID: 27273 Comm: tc Tainted: G        W         5.3.0-rc8+ #721
[ 4791.154336] Hardware name: Supermicro SYS-2028TP-DECR/X10DRT-P, BIOS 2.0b 03/30/2017
[ 4791.155075] Call Trace:
[ 4791.155803]  dump_stack+0x85/0xc0
[ 4791.156529]  ___might_sleep.cold+0xac/0xbc
[ 4791.157251]  __mutex_lock+0x5b/0x960
[ 4791.157966]  ? console_unlock+0x363/0x5d0
[ 4791.158676]  ? tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 4791.159395]  ? tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 4791.160103]  tcf_chain0_head_change_cb_del.isra.0+0x1b/0xf0
[ 4791.160815]  tcf_block_put_ext.part.0+0x21/0x50
[ 4791.161530]  tcf_block_put+0x50/0x70
[ 4791.162233]  sfq_destroy+0x15/0x50 [sch_sfq]
[ 4791.162936]  qdisc_destroy+0x5f/0x160
[ 4791.163642]  htb_change_class.cold+0x5df/0x69d [sch_htb]
[ 4791.164505]  tc_ctl_tclass+0x19d/0x480
[ 4791.165360]  rtnetlink_rcv_msg+0x170/0x4b0
[ 4791.166191]  ? netlink_deliver_tap+0x95/0x400
[ 4791.166907]  ? rtnl_dellink+0x2d0/0x2d0
[ 4791.167625]  netlink_rcv_skb+0x49/0x110
[ 4791.168345]  netlink_unicast+0x171/0x200
[ 4791.169058]  netlink_sendmsg+0x224/0x3f0
[ 4791.169771]  sock_sendmsg+0x5e/0x60
[ 4791.170475]  ___sys_sendmsg+0x2ae/0x330
[ 4791.171183]  ? ___sys_recvmsg+0x159/0x1f0
[ 4791.171894]  ? do_wp_page+0x9c/0x790
[ 4791.172595]  ? __handle_mm_fault+0xcd3/0x19e0
[ 4791.173309]  __sys_sendmsg+0x59/0xa0
[ 4791.174024]  do_syscall_64+0x5c/0xb0
[ 4791.174725]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 4791.175435] RIP: 0033:0x7f0aa41497b8
[ 4791.176129] Code: 89 02 48 c7 c0 ff ff ff ff eb bb 0f 1f 80 00 00 00 00 f3 0f 1e fa 48 8d 05 65 8f 0c 00 8b 00 85 c0 75 17 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 58 c3 0f 1f 80 00 00 00 00 48 83 ec 28 89 5
4
[ 4791.177532] RSP: 002b:00007fff4e37d588 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
[ 4791.178243] RAX: ffffffffffffffda RBX: 000000005d8132f7 RCX: 00007f0aa41497b8
[ 4791.178947] RDX: 0000000000000000 RSI: 00007fff4e37d5f0 RDI: 0000000000000003
[ 4791.179662] RBP: 0000000000000000 R08: 0000000000000001 R09: 00000000020149a0
[ 4791.180382] R10: 0000000000404eda R11: 0000000000000246 R12: 0000000000000001
[ 4791.181100] R13: 000000000047f640 R14: 0000000000000000 R15: 0000000000000000

In htb_change_class() function save parent->leaf.q to local temporary
variable and put reference to it after sch tree lock is released in order
not to call potentially sleeping cls API in atomic section. This is safe to
do because Qdisc has already been reset by qdisc_purge_queue() inside sch
tree lock critical section.

Fixes: c266f64dbfa2 ("net: sched: protect block state with mutex")
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/rds: Check laddr_check before calling it

In rds_bind(), laddr_check is called without checking if it is NULL or
not.  And rs_transport should be reset if rds_add_bound() fails.

Fixes: c5c1a030a7db ("net/rds: An rds_sock is added too early to the hash table")
Reported-by: syzbot+fae39afd2101a17ec624@syzkaller.appspotmail.com
Signed-off-by: Ka-Cheong Poon <ka-cheong.poon@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'SO_PRIORITY'

Eric Dumazet says:

====================
tcp: provide correct skb->priority

SO_PRIORITY socket option requests TCP egress packets
to contain a user provided value.

TCP manages to send most packets with the requested values,
notably for TCP_ESTABLISHED state, but fails to do so for
few packets.

These packets are control packets sent on behalf
of SYN_RECV or TIME_WAIT states.

Note that to test this with packetdrill, it is a bit
of a hassle, since packetdrill can not verify priority
of egress packets, other than indirect observations,
using for example sch_prio on its tunnel device.

The bad skb priorities cause problems for GCP,
as this field is one of the keys used in routing.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: honor SO_PRIORITY in TIME_WAIT state

ctl packets sent on behalf of TIME_WAIT sockets currently
have a zero skb->priority, which can cause various problems.

In this patch we :

- add a tw_priority field in struct inet_timewait_sock.

- populate it from sk->sk_priority when a TIME_WAIT is created.

- For IPv4, change ip_send_unicast_reply() and its two
  callers to propagate tw_priority correctly.
  ip_send_unicast_reply() no longer changes sk->sk_priority.

- For IPv6, make sure TIME_WAIT sockets pass their tw_priority
  field to tcp_v6_send_response() and tcp_v6_send_ack().

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: tcp: provide sk->sk_priority to ctl packets

We can populate skb->priority for some ctl packets
instead of always using zero.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: add priority parameter to ip6_xmit()

Currently, ip6_xmit() sets skb->priority based on sk->sk_priority

This is not desirable for TCP since TCP shares the same ctl socket
for a given netns. We want to be able to send RST or ACK packets
with a non zero skb->priority.

This patch has no functional change.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bpf: Fix bpf_event_output re-entry issue

BPF_PROG_TYPE_SOCK_OPS program can reenter bpf_event_output because it
can be called from atomic and non-atomic contexts since we don't have
bpf_prog_active to prevent it happen.

This patch enables 3 levels of nesting to support normal, irq and nmi
context.

We can easily reproduce the issue by running netperf crr mode with 100
flows and 10 threads from netperf client side.

Here is the whole stack dump:

[  515.228898] WARNING: CPU: 20 PID: 14686 at kernel/trace/bpf_trace.c:549 bpf_event_output+0x1f9/0x220
[  515.228903] CPU: 20 PID: 14686 Comm: tcp_crr Tainted: G        W        4.15.0-smp-fixpanic #44
[  515.228904] Hardware name: Intel TBG,ICH10/Ikaria_QC_1b, BIOS 1.22.0 06/04/2018
[  515.228905] RIP: 0010:bpf_event_output+0x1f9/0x220
[  515.228906] RSP: 0018:ffff9a57ffc03938 EFLAGS: 00010246
[  515.228907] RAX: 0000000000000012 RBX: 0000000000000001 RCX: 0000000000000000
[  515.228907] RDX: 0000000000000000 RSI: 0000000000000096 RDI: ffffffff836b0f80
[  515.228908] RBP: ffff9a57ffc039c8 R08: 0000000000000004 R09: 0000000000000012
[  515.228908] R10: ffff9a57ffc1de40 R11: 0000000000000000 R12: 0000000000000002
[  515.228909] R13: ffff9a57e13bae00 R14: 00000000ffffffff R15: ffff9a57ffc1e2c0
[  515.228910] FS:  00007f5a3e6ec700(0000) GS:ffff9a57ffc00000(0000) knlGS:0000000000000000
[  515.228910] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  515.228911] CR2: 0000537082664fff CR3: 000000061fed6002 CR4: 00000000000226f0
[  515.228911] Call Trace:
[  515.228913]  <IRQ>
[  515.228919]  [<ffffffff82c6c6cb>] bpf_sockopt_event_output+0x3b/0x50
[  515.228923]  [<ffffffff8265daee>] ? bpf_ktime_get_ns+0xe/0x10
[  515.228927]  [<ffffffff8266fda5>] ? __cgroup_bpf_run_filter_sock_ops+0x85/0x100
[  515.228930]  [<ffffffff82cf90a5>] ? tcp_init_transfer+0x125/0x150
[  515.228933]  [<ffffffff82cf9159>] ? tcp_finish_connect+0x89/0x110
[  515.228936]  [<ffffffff82cf98e4>] ? tcp_rcv_state_process+0x704/0x1010
[  515.228939]  [<ffffffff82c6e263>] ? sk_filter_trim_cap+0x53/0x2a0
[  515.228942]  [<ffffffff82d90d1f>] ? tcp_v6_inbound_md5_hash+0x6f/0x1d0
[  515.228945]  [<ffffffff82d92160>] ? tcp_v6_do_rcv+0x1c0/0x460
[  515.228947]  [<ffffffff82d93558>] ? tcp_v6_rcv+0x9f8/0xb30
[  515.228951]  [<ffffffff82d737c0>] ? ip6_route_input+0x190/0x220
[  515.228955]  [<ffffffff82d5f7ad>] ? ip6_protocol_deliver_rcu+0x6d/0x450
[  515.228958]  [<ffffffff82d60246>] ? ip6_rcv_finish+0xb6/0x170
[  515.228961]  [<ffffffff82d5fb90>] ? ip6_protocol_deliver_rcu+0x450/0x450
[  515.228963]  [<ffffffff82d60361>] ? ipv6_rcv+0x61/0xe0
[  515.228966]  [<ffffffff82d60190>] ? ipv6_list_rcv+0x330/0x330
[  515.228969]  [<ffffffff82c4976b>] ? __netif_receive_skb_one_core+0x5b/0xa0
[  515.228972]  [<ffffffff82c497d1>] ? __netif_receive_skb+0x21/0x70
[  515.228975]  [<ffffffff82c4a8d2>] ? process_backlog+0xb2/0x150
[  515.228978]  [<ffffffff82c4aadf>] ? net_rx_action+0x16f/0x410
[  515.228982]  [<ffffffff830000dd>] ? __do_softirq+0xdd/0x305
[  515.228986]  [<ffffffff8252cfdc>] ? irq_exit+0x9c/0xb0
[  515.228989]  [<ffffffff82e02de5>] ? smp_call_function_single_interrupt+0x65/0x120
[  515.228991]  [<ffffffff82e020e1>] ? call_function_single_interrupt+0x81/0x90
[  515.228992]  </IRQ>
[  515.228996]  [<ffffffff82a11ff0>] ? io_serial_in+0x20/0x20
[  515.229000]  [<ffffffff8259c040>] ? console_unlock+0x230/0x490
[  515.229003]  [<ffffffff8259cbaa>] ? vprintk_emit+0x26a/0x2a0
[  515.229006]  [<ffffffff8259cbff>] ? vprintk_default+0x1f/0x30
[  515.229008]  [<ffffffff8259d9f5>] ? vprintk_func+0x35/0x70
[  515.229011]  [<ffffffff8259d4bb>] ? printk+0x50/0x66
[  515.229013]  [<ffffffff82637637>] ? bpf_event_output+0xb7/0x220
[  515.229016]  [<ffffffff82c6c6cb>] ? bpf_sockopt_event_output+0x3b/0x50
[  515.229019]  [<ffffffff8265daee>] ? bpf_ktime_get_ns+0xe/0x10
[  515.229023]  [<ffffffff82c29e87>] ? release_sock+0x97/0xb0
[  515.229026]  [<ffffffff82ce9d6a>] ? tcp_recvmsg+0x31a/0xda0
[  515.229029]  [<ffffffff8266fda5>] ? __cgroup_bpf_run_filter_sock_ops+0x85/0x100
[  515.229032]  [<ffffffff82ce77c1>] ? tcp_set_state+0x191/0x1b0
[  515.229035]  [<ffffffff82ced10e>] ? tcp_disconnect+0x2e/0x600
[  515.229038]  [<ffffffff82cecbbb>] ? tcp_close+0x3eb/0x460
[  515.229040]  [<ffffffff82d21082>] ? inet_release+0x42/0x70
[  515.229043]  [<ffffffff82d58809>] ? inet6_release+0x39/0x50
[  515.229046]  [<ffffffff82c1f32d>] ? __sock_release+0x4d/0xd0
[  515.229049]  [<ffffffff82c1f3e5>] ? sock_close+0x15/0x20
[  515.229052]  [<ffffffff8273b517>] ? __fput+0xe7/0x1f0
[  515.229055]  [<ffffffff8273b66e>] ? ____fput+0xe/0x10
[  515.229058]  [<ffffffff82547bf2>] ? task_work_run+0x82/0xb0
[  515.229061]  [<ffffffff824086df>] ? exit_to_usermode_loop+0x7e/0x11f
[  515.229064]  [<ffffffff82408171>] ? do_syscall_64+0x111/0x130
[  515.229067]  [<ffffffff82e0007c>] ? entry_SYSCALL_64_after_hwframe+0x3d/0xa2

Fixes: a5a3a828cd00 ("bpf: add perf event notificaton support for sock_ops")
Signed-off-by: Allan Zhang <allanzhang@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Stanislav Fomichev <sdf@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20190925234312.94063-2-allanzhang@google.com

net: dsa: qca8k: Fix port enable for CPU port

The CPU port does not have a PHY connected to it. So calling
phy_support_asym_pause() results in an Opps. As with other DSA
drivers, add a guard that the port is a user port.

Reported-by: Michal Vokáč <michal.vokac@ysoft.com>
Fixes: 0394a63acfe2 ("net: dsa: enable and disable all ports")
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Michal Vokáč <michal.vokac@ysoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sch_netem: fix rcu splat in netem_enqueue()

qdisc_root() use from netem_enqueue() triggers a lockdep warning.

__dev_queue_xmit() uses rcu_read_lock_bh() which is
not equivalent to rcu_read_lock() + local_bh_disable_bh as far
as lockdep is concerned.

WARNING: suspicious RCU usage
5.3.0-rc7+ #0 Not tainted
-----------------------------
include/net/sch_generic.h:492 suspicious rcu_dereference_check() usage!

other info that might help us debug this:

rcu_scheduler_active = 2, debug_locks = 1
3 locks held by syz-executor427/8855:
 #0: 00000000b5525c01 (rcu_read_lock_bh){....}, at: lwtunnel_xmit_redirect include/net/lwtunnel.h:92 [inline]
 #0: 00000000b5525c01 (rcu_read_lock_bh){....}, at: ip_finish_output2+0x2dc/0x2570 net/ipv4/ip_output.c:214
 #1: 00000000b5525c01 (rcu_read_lock_bh){....}, at: __dev_queue_xmit+0x20a/0x3650 net/core/dev.c:3804
 #2: 00000000364bae92 (&(&sch->q.lock)->rlock){+.-.}, at: spin_lock include/linux/spinlock.h:338 [inline]
 #2: 00000000364bae92 (&(&sch->q.lock)->rlock){+.-.}, at: __dev_xmit_skb net/core/dev.c:3502 [inline]
 #2: 00000000364bae92 (&(&sch->q.lock)->rlock){+.-.}, at: __dev_queue_xmit+0x14b8/0x3650 net/core/dev.c:3838

stack backtrace:
CPU: 0 PID: 8855 Comm: syz-executor427 Not tainted 5.3.0-rc7+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0x172/0x1f0 lib/dump_stack.c:113
 lockdep_rcu_suspicious+0x153/0x15d kernel/locking/lockdep.c:5357
 qdisc_root include/net/sch_generic.h:492 [inline]
 netem_enqueue+0x1cfb/0x2d80 net/sched/sch_netem.c:479
 __dev_xmit_skb net/core/dev.c:3527 [inline]
 __dev_queue_xmit+0x15d2/0x3650 net/core/dev.c:3838
 dev_queue_xmit+0x18/0x20 net/core/dev.c:3902
 neigh_hh_output include/net/neighbour.h:500 [inline]
 neigh_output include/net/neighbour.h:509 [inline]
 ip_finish_output2+0x1726/0x2570 net/ipv4/ip_output.c:228
 __ip_finish_output net/ipv4/ip_output.c:308 [inline]
 __ip_finish_output+0x5fc/0xb90 net/ipv4/ip_output.c:290
 ip_finish_output+0x38/0x1f0 net/ipv4/ip_output.c:318
 NF_HOOK_COND include/linux/netfilter.h:294 [inline]
 ip_mc_output+0x292/0xf40 net/ipv4/ip_output.c:417
 dst_output include/net/dst.h:436 [inline]
 ip_local_out+0xbb/0x190 net/ipv4/ip_output.c:125
 ip_send_skb+0x42/0xf0 net/ipv4/ip_output.c:1555
 udp_send_skb.isra.0+0x6b2/0x1160 net/ipv4/udp.c:887
 udp_sendmsg+0x1e96/0x2820 net/ipv4/udp.c:1174
 inet_sendmsg+0x9e/0xe0 net/ipv4/af_inet.c:807
 sock_sendmsg_nosec net/socket.c:637 [inline]
 sock_sendmsg+0xd7/0x130 net/socket.c:657
 ___sys_sendmsg+0x3e2/0x920 net/socket.c:2311
 __sys_sendmmsg+0x1bf/0x4d0 net/socket.c:2413
 __do_sys_sendmmsg net/socket.c:2442 [inline]
 __se_sys_sendmmsg net/socket.c:2439 [inline]
 __x64_sys_sendmmsg+0x9d/0x100 net/socket.c:2439
 do_syscall_64+0xfd/0x6a0 arch/x86/entry/common.c:296
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

kcm: disable preemption in kcm_parse_func_strparser()

After commit a2c11b034142 ("kcm: use BPF_PROG_RUN")
syzbot easily triggers the warning in cant_sleep().

As explained in commit 6cab5e90ab2b ("bpf: run bpf programs
with preemption disabled") we need to disable preemption before
running bpf programs.

BUG: assuming atomic context at net/kcm/kcmsock.c:382
in_atomic(): 0, irqs_disabled(): 0, pid: 7, name: kworker/u4:0
3 locks held by kworker/u4:0/7:
 #0: ffff888216726128 ((wq_completion)kstrp){+.+.}, at: __write_once_size include/linux/compiler.h:226 [inline]
 #0: ffff888216726128 ((wq_completion)kstrp){+.+.}, at: arch_atomic64_set arch/x86/include/asm/atomic64_64.h:34 [inline]
 #0: ffff888216726128 ((wq_completion)kstrp){+.+.}, at: atomic64_set include/asm-generic/atomic-instrumented.h:855 [inline]
 #0: ffff888216726128 ((wq_completion)kstrp){+.+.}, at: atomic_long_set include/asm-generic/atomic-long.h:40 [inline]
 #0: ffff888216726128 ((wq_completion)kstrp){+.+.}, at: set_work_data kernel/workqueue.c:620 [inline]
 #0: ffff888216726128 ((wq_completion)kstrp){+.+.}, at: set_work_pool_and_clear_pending kernel/workqueue.c:647 [inline]
 #0: ffff888216726128 ((wq_completion)kstrp){+.+.}, at: process_one_work+0x88b/0x1740 kernel/workqueue.c:2240
 #1: ffff8880a989fdc0 ((work_completion)(&strp->work)){+.+.}, at: process_one_work+0x8c1/0x1740 kernel/workqueue.c:2244
 #2: ffff888098998d10 (sk_lock-AF_INET){+.+.}, at: lock_sock include/net/sock.h:1522 [inline]
 #2: ffff888098998d10 (sk_lock-AF_INET){+.+.}, at: strp_sock_lock+0x2e/0x40 net/strparser/strparser.c:440
CPU: 0 PID: 7 Comm: kworker/u4:0 Not tainted 5.3.0+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Workqueue: kstrp strp_work
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0x172/0x1f0 lib/dump_stack.c:113
 __cant_sleep kernel/sched/core.c:6826 [inline]
 __cant_sleep.cold+0xa4/0xbc kernel/sched/core.c:6803
 kcm_parse_func_strparser+0x54/0x200 net/kcm/kcmsock.c:382
 __strp_recv+0x5dc/0x1b20 net/strparser/strparser.c:221
 strp_recv+0xcf/0x10b net/strparser/strparser.c:343
 tcp_read_sock+0x285/0xa00 net/ipv4/tcp.c:1639
 strp_read_sock+0x14d/0x200 net/strparser/strparser.c:366
 do_strp_work net/strparser/strparser.c:414 [inline]
 strp_work+0xe3/0x130 net/strparser/strparser.c:423
 process_one_work+0x9af/0x1740 kernel/workqueue.c:2269

Fixes: a2c11b034142 ("kcm: use BPF_PROG_RUN")
Fixes: 6cab5e90ab2b ("bpf: run bpf programs with preemption disabled")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: stmmac: Fix signedness bug in ipq806x_gmac_of_parse()

The "gmac->phy_mode" variable is an enum and in this context GCC will
treat it as an unsigned int so the error handling will never be
triggered.

Fixes: b1c17215d718 ("stmmac: add ipq806x glue layer")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: nixge: Fix a signedness bug in nixge_probe()

The "priv->phy_mode" is an enum and in this context GCC will treat it
as an unsigned int so it can never be less than zero.

Fixes: 492caffa8a1a ("net: ethernet: nixge: Add support for National Instruments XGE netdev")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

of: mdio: Fix a signedness bug in of_phy_get_and_connect()

The "iface" variable is an enum and in this context GCC treats it as
an unsigned int so the error handling is never triggered.

Fixes: b78624125304 ("of_mdio: Abstract a general interface for phy connect")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: axienet: fix a signedness bug in probe

The "lp->phy_mode" is an enum but in this context GCC treats it as an
unsigned int so the error handling is never triggered.

Fixes: ee06b1728b95 ("net: axienet: add support for standard phy-mode binding")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Radhey Shyam Pandey <radhey.shyam.pandey@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: stmmac: dwmac-meson8b: Fix signedness bug in probe

The "dwmac->phy_mode" is an enum and in this context GCC treats it as
an unsigned int so the error handling is never triggered.

Fixes: 566e82516253 ("net: stmmac: add a glue driver for the Amlogic Meson 8b / GXBB DWMAC")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: socionext: Fix a signedness bug in ave_probe()

The "phy_mode" variable is an enum and in this context GCC treats it as
an unsigned int so the error handling is never triggered.

Fixes: 4c270b55a5af ("net: ethernet: socionext: add AVE ethernet driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

enetc: Fix a signedness bug in enetc_of_get_phy()

The "priv->if_mode" is type phy_interface_t which is an enum.  In this
context GCC will treat the enum as an unsigned int so this error
handling is never triggered.

Fixes: d4fd0404c1c9 ("enetc: Introduce basic PF and VF ENETC ethernet drivers")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: netsec: Fix signedness bug in netsec_probe()

The "priv->phy_interface" variable is an enum and in this context GCC
will treat it as an unsigned int so the error handling is never
triggered.

Fixes: 533dd11a12f6 ("net: socionext: Add Synquacer NetSec driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: broadcom/bcmsysport: Fix signedness in bcm_sysport_probe()

The "priv->phy_interface" variable is an enum and in this context GCC
will treat it as unsigned so the error handling will never be
triggered.

Fixes: 80105befdb4b ("net: systemport: add Broadcom SYSTEMPORT Ethernet MAC driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: hisilicon: Fix signedness bug in hix5hd2_dev_probe()

The "priv->phy_mode" variable is an enum and in this context GCC will
treat it as unsigned to the error handling will never trigger.

Fixes: 57c5bc9ad7d7 ("net: hisilicon: add hix5hd2 mac driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cxgb4: Signedness bug in init_one()

The "chip" variable is an enum, and it's treated as unsigned int by GCC
in this context so the error handling isn't triggered.

Fixes: e8d452923ae6 ("cxgb4: clean up init_one")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: aquantia: Fix aq_vec_isr_legacy() return value

The irqreturn_t type is an enum or an unsigned int in GCC.  That
creates to problems because it can't detect if the
self->aq_hw_ops->hw_irq_read() call fails and at the end the function
always returns IRQ_HANDLED.

drivers/net/ethernet/aquantia/atlantic/aq_vec.c:316 aq_vec_isr_legacy() warn: unsigned 'err' is never less than zero.
drivers/net/ethernet/aquantia/atlantic/aq_vec.c:329 aq_vec_isr_legacy() warn: always true condition '(err >= 0) => (0-u32max >= 0)'

Fixes: 970a2e9864b0 ("net: ethernet: aquantia: Vector operations")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

dimlib: make DIMLIB a hidden symbol

According to Tal Gilboa the only benefit from DIM comes from a driver
that uses it. So it doesn't make sense to make this symbol user visible,
instead all drivers that use it should select it (as is already the case
AFAICT).

Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'wireless-drivers-for-davem-2019-09-26' of https://git./linux/kernel/git/kvalo/wireless-drivers

Kalle Valo says:

====================
wireless-drivers fixes for 5.4

First set of fixes for 5.4 sent during the merge window. Most are
regressions fixes but the mt7615 problem has been since it was merged.

iwlwifi

* fix a build regression related CONFIG_THERMAL

* avoid using GEO_TX_POWER_LIMIT command on certain firmware versions

rtw88

* fixes for skb leaks

zd1211rw

* fix a compiler warning on 32 bit

mt76

* fix the firmware paths for mt7615 to match with linux-firmware

wil6210

* fix use of skb after free
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

bpf: Clean up indentation issue in BTF kflag processing

There is a statement that is indented one level too deeply, remove
the extraneous tab.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20190925093835.19515-1-colin.king@canonical.com

libbpf: Teach btf_dumper to emit stand-alone anonymous enum definitions

BTF-to-C converter previously skipped anonymous enums in an assumption
that those are embedded in struct's field definitions. This is not
always the case and a lot of kernel constants are defined as part of
anonymous enums. This change fixes the logic by eagerly marking all
types as either referenced by any other type or not. This is enough to
distinguish two classes of anonymous enums and emit previously omitted
enum definitions.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20190925203745.3173184-1-andriin@fb.com

ipv6: do not free rt if FIB_LOOKUP_NOREF is set on suppress rule

Commit 7d9e5f422150 removed references from certain dsts, but accounting
for this never translated down into the fib6 suppression code. This bug
was triggered by WireGuard users who use wg-quick(8), which uses the
"suppress-prefix" directive to ip-rule(8) for routing all of their
internet traffic without routing loops. The test case added here
causes the reference underflow by causing packets to evaluate a suppress
rule.

Fixes: 7d9e5f422150 ("ipv6: convert major tx path to use RT6_LOOKUP_F_DST_NOREF")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Acked-by: Wei Wang <weiwan@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

openvswitch: change type of UPCALL_PID attribute to NLA_UNSPEC

userspace openvswitch patch "(dpif-linux: Implement the API
functions to allow multiple handler threads read upcall)"
changes its type from U32 to UNSPEC, but leave the kernel
unchanged

and after kernel 6e237d099fac "(netlink: Relax attr validation
for fixed length types)", this bug is exposed by the below
warning

[   57.215841] netlink: 'ovs-vswitchd': attribute type 5 has an invalid length.

Fixes: 5cd667b0a456 ("openvswitch: Allow each vport to have an array of 'port_id's")
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

dt-bindings: net: ravb: Add support for r8a774b1 SoC

Document RZ/G2N (R8A774B1) SoC bindings.

Signed-off-by: Biju Das <biju.das@bp.renesas.com>
Reviewed-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: stmmac: Fix page pool size

The size of individual pages in the page pool in given by an order. The
order is the binary logarithm of the number of pages that make up one of
the pages in the pool. However, the driver currently passes the number
of pages rather than the order, so it ends up wasting quite a bit of
memory.

Fix this by taking the binary logarithm and passing that in the order
field.

Fixes: 2af6106ae949 ("net: stmmac: Introducing support for Page Pool")
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

macsec: drop skb sk before calling gro_cells_receive

Fei Liu reported a crash when doing netperf on a topo of macsec
dev over veth:

  [  448.919128] refcount_t: underflow; use-after-free.
  [  449.090460] Call trace:
  [  449.092895]  refcount_sub_and_test+0xb4/0xc0
  [  449.097155]  tcp_wfree+0x2c/0x150
  [  449.100460]  ip_rcv+0x1d4/0x3a8
  [  449.103591]  __netif_receive_skb_core+0x554/0xae0
  [  449.108282]  __netif_receive_skb+0x28/0x78
  [  449.112366]  netif_receive_skb_internal+0x54/0x100
  [  449.117144]  napi_gro_complete+0x70/0xc0
  [  449.121054]  napi_gro_flush+0x6c/0x90
  [  449.124703]  napi_complete_done+0x50/0x130
  [  449.128788]  gro_cell_poll+0x8c/0xa8
  [  449.132351]  net_rx_action+0x16c/0x3f8
  [  449.136088]  __do_softirq+0x128/0x320

The issue was caused by skb's true_size changed without its sk's
sk_wmem_alloc increased in tcp/skb_gro_receive(). Later when the
skb is being freed and the skb's truesize is subtracted from its
sk's sk_wmem_alloc in tcp_wfree(), underflow occurs.

macsec is calling gro_cells_receive() to receive a packet, which
actually requires skb->sk to be NULL. However when macsec dev is
over veth, it's possible the skb->sk is still set if the skb was
not unshared or expanded from the peer veth.

ip_rcv() is calling skb_orphan() to drop the skb's sk for tproxy,
but it is too late for macsec's calling gro_cells_receive(). So
fix it by dropping the skb's sk earlier on rx path of macsec.

Fixes: 5491e7c6b1a9 ("macsec: enable GRO and RPS on macsec devices")
Reported-by: Xiumei Mu <xmu@redhat.com>
Reported-by: Fei Liu <feliu@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'mlx5-fixes-2019-09-24' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
Mellanox, mlx5 fixes 2019-09-24

This series introduces some fixes to mlx5 driver.
For more information please see tag log below.

Please pull and let me know if there is any problem.

For -stable v4.20:
 ('net/mlx5e: Fix traffic duplication in ethtool steering')

For -stable v4.19:
 ('net/mlx5: Add device ID of upcoming BlueField-2')

For -stable v5.3:
 ('net/mlx5e: Fix matching on tunnel addresses type')
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net: print proper warning on dst underflow

Proper warnings with stack traces make it much easier to figure out
what's doing the double free and create more meaningful bug reports from
users.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/sched: cbs: Fix not adding cbs instance to list

When removing a cbs instance when offloading is enabled, the crash
below can be observed.

The problem happens because that when offloading is enabled, the cbs
instance is not added to the list.

Also, the current code doesn't handle correctly the case when offload
is disabled without removing the qdisc: if the link speed changes the
credit calculations will be wrong. When we create the cbs instance
with offloading enabled, it's not added to the notification list, when
later we disable offloading, it's not in the list, so link speed
changes will not affect it.

The solution for both issues is the same, add the cbs instance being
created unconditionally to the global list, even if the link state
notification isn't useful "right now".

Crash log:

[518758.189866] BUG: kernel NULL pointer dereference, address: 0000000000000000
[518758.189870] #PF: supervisor read access in kernel mode
[518758.189871] #PF: error_code(0x0000) - not-present page
[518758.189872] PGD 0 P4D 0
[518758.189874] Oops: 0000 [#1] SMP PTI
[518758.189876] CPU: 3 PID: 4825 Comm: tc Not tainted 5.2.9 #1
[518758.189877] Hardware name: Gigabyte Technology Co., Ltd. Z390 AORUS ULTRA/Z390 AORUS ULTRA-CF, BIOS F7 03/14/2019
[518758.189881] RIP: 0010:__list_del_entry_valid+0x29/0xa0
[518758.189883] Code: 90 48 b8 00 01 00 00 00 00 ad de 55 48 8b 17 4c 8b 47 08 48 89 e5 48 39 c2 74 27 48 b8 00 02 00 00 00 00 ad de 49 39 c0 74 2d <49> 8b 30 48 39 fe 75 3d 48 8b 52 08 48 39 f2 75 4c b8 01 00 00 00
[518758.189885] RSP: 0018:ffffa27e43903990 EFLAGS: 00010207
[518758.189887] RAX: dead000000000200 RBX: ffff8bce69f0f000 RCX: 0000000000000000
[518758.189888] RDX: 0000000000000000 RSI: ffff8bce69f0f064 RDI: ffff8bce69f0f1e0
[518758.189890] RBP: ffffa27e43903990 R08: 0000000000000000 R09: ffff8bce69e788c0
[518758.189891] R10: ffff8bce62acd400 R11: 00000000000003cb R12: ffff8bce69e78000
[518758.189892] R13: ffff8bce69f0f140 R14: 0000000000000000 R15: 0000000000000000
[518758.189894] FS:  00007fa1572c8f80(0000) GS:ffff8bce6e0c0000(0000) knlGS:0000000000000000
[518758.189895] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[518758.189896] CR2: 0000000000000000 CR3: 000000040a398006 CR4: 00000000003606e0
[518758.189898] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[518758.189899] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[518758.189900] Call Trace:
[518758.189904]  cbs_destroy+0x32/0xa0 [sch_cbs]
[518758.189906]  qdisc_destroy+0x45/0x120
[518758.189907]  qdisc_put+0x25/0x30
[518758.189908]  qdisc_graft+0x2c1/0x450
[518758.189910]  tc_get_qdisc+0x1c8/0x310
[518758.189912]  ? get_page_from_freelist+0x91a/0xcb0
[518758.189914]  rtnetlink_rcv_msg+0x293/0x360
[518758.189916]  ? kmem_cache_alloc_node_trace+0x178/0x260
[518758.189918]  ? __kmalloc_node_track_caller+0x38/0x50
[518758.189920]  ? rtnl_calcit.isra.0+0xf0/0xf0
[518758.189922]  netlink_rcv_skb+0x48/0x110
[518758.189923]  rtnetlink_rcv+0x10/0x20
[518758.189925]  netlink_unicast+0x15b/0x1d0
[518758.189926]  netlink_sendmsg+0x1ea/0x380
[518758.189929]  sock_sendmsg+0x2f/0x40
[518758.189930]  ___sys_sendmsg+0x295/0x2f0
[518758.189932]  ? ___sys_recvmsg+0x151/0x1e0
[518758.189933]  ? do_wp_page+0x7e/0x450
[518758.189935]  __sys_sendmsg+0x48/0x80
[518758.189937]  __x64_sys_sendmsg+0x1a/0x20
[518758.189939]  do_syscall_64+0x53/0x1f0
[518758.189941]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[518758.189942] RIP: 0033:0x7fa15755169a
[518758.189944] Code: 48 c7 c0 ff ff ff ff eb be 0f 1f 80 00 00 00 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 18 b8 2e 00 00 00 c5 fc 77 0f 05 <48> 3d 00 f0 ff ff 77 5e c3 0f 1f 44 00 00 48 83 ec 28 89 54 24 1c
[518758.189946] RSP: 002b:00007ffda58b60b8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
[518758.189948] RAX: ffffffffffffffda RBX: 000055e4b836d9a0 RCX: 00007fa15755169a
[518758.189949] RDX: 0000000000000000 RSI: 00007ffda58b6128 RDI: 0000000000000003
[518758.189951] RBP: 00007ffda58b6190 R08: 0000000000000001 R09: 000055e4b9d848a0
[518758.189952] R10: 0000000000000000 R11: 0000000000000246 R12: 000000005d654b49
[518758.189953] R13: 0000000000000000 R14: 00007ffda58b6230 R15: 00007ffda58b6210
[518758.189955] Modules linked in: sch_cbs sch_etf sch_mqprio netlink_diag unix_diag e1000e igb intel_pch_thermal thermal video backlight pcc_cpufreq
[518758.189960] CR2: 0000000000000000
[518758.189961] ---[ end trace 6a13f7aaf5376019 ]---
[518758.189963] RIP: 0010:__list_del_entry_valid+0x29/0xa0
[518758.189964] Code: 90 48 b8 00 01 00 00 00 00 ad de 55 48 8b 17 4c 8b 47 08 48 89 e5 48 39 c2 74 27 48 b8 00 02 00 00 00 00 ad de 49 39 c0 74 2d <49> 8b 30 48 39 fe 75 3d 48 8b 52 08 48 39 f2 75 4c b8 01 00 00 00
[518758.189967] RSP: 0018:ffffa27e43903990 EFLAGS: 00010207
[518758.189968] RAX: dead000000000200 RBX: ffff8bce69f0f000 RCX: 0000000000000000
[518758.189969] RDX: 0000000000000000 RSI: ffff8bce69f0f064 RDI: ffff8bce69f0f1e0
[518758.189971] RBP: ffffa27e43903990 R08: 0000000000000000 R09: ffff8bce69e788c0
[518758.189972] R10: ffff8bce62acd400 R11: 00000000000003cb R12: ffff8bce69e78000
[518758.189973] R13: ffff8bce69f0f140 R14: 0000000000000000 R15: 0000000000000000
[518758.189975] FS:  00007fa1572c8f80(0000) GS:ffff8bce6e0c0000(0000) knlGS:0000000000000000
[518758.189976] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[518758.189977] CR2: 0000000000000000 CR3: 000000040a398006 CR4: 00000000003606e0
[518758.189979] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[518758.189980] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400

Fixes: e0a7683d30e9 ("net/sched: cbs: fix port_rate miscalculation")
Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers: net: Fix Kconfig indentation

Adjust indentation from spaces to tab (+optional two spaces) as in
coding style with command like:
    $ sed -e 's/^        /\t/' -i */Kconfig

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Kalle Valo <kvalo@codeaurora.org>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Fix Kconfig indentation

Adjust indentation from spaces to tab (+optional two spaces) as in
coding style with command like:
    $ sed -e 's/^        /\t/' -i */Kconfig

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

MAINTAINERS: add Yanjun to FORCEDETH maintainers list

Yanjun has been spending quite a lot of time fixing bugs
in FORCEDETH source code. I'd like to add Yanjun to maintainers
list.

Signed-off-by: Rain River <rain.1986.08.12@gmail.com>
Acked-by: Zhu Yanjun <yanjun.zhu@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

selftests/bpf: adjust strobemeta loop to satisfy latest clang

Some recent changes in latest Clang started causing the following
warning when unrolling strobemeta test case main loop:

  progs/strobemeta.h:416:2: warning: loop not unrolled: the optimizer was
  unable to perform the requested transformation; the transformation might
  be disabled or specified as part of an unsupported transformation
  ordering [-Wpass-failed=transform-warning]

This patch simplifies loop's exit condition to depend only on constant
max iteration number (STROBE_MAX_MAP_ENTRIES), while moving early
termination logic inside the loop body. The changes are equivalent from
program logic standpoint, but fixes the warning. It also appears to
improve generated BPF code, as it fixes previously failing non-unrolled
strobemeta test cases.

Cc: Alexei Starovoitov <ast@fb.com>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

selftests/bpf: delete unused variables in test_sysctl

Remove no longer used variables and avoid compiler warnings.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

libbpf: fix false uninitialized variable warning

Some compilers emit warning for potential uninitialized next_id usage.
The code is correct, but control flow is too complicated for some
compilers to figure this out. Re-initialize next_id to satisfy
compiler.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

bpf/xskmap: Return ERR_PTR for failure case instead of NULL.

When kzalloc() failed, NULL was returned to the caller, which
tested the pointer with IS_ERR(), which didn't match, so the
pointer was used later, resulting in a NULL dereference.

Return ERR_PTR(-ENOMEM) instead of NULL.

Reported-by: syzbot+491c1b7565ba9069ecae@syzkaller.appspotmail.com
Fixes: 0402acd683c6 ("xsk: remove AF_XDP socket from map when the socket is released")
Signed-off-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

selftests/bpf: test_progs: fix client/server race in tcp_rtt

This is the same problem I found earlier in test_sockopt_inherit:
there is a race between server thread doing accept() and client
thread doing connect(). Let's explicitly synchronize them via
pthread conditional variable.

v2:
* don't exit from server_thread without signaling condvar,
  fixes possible issue where main() would wait forever (Andrii Nakryiko)

Fixes: b55873984dab ("selftests/bpf: test BPF_SOCK_OPS_RTT_CB")
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

net: macb: Remove dead code

macb_64b_desc is always called when HW_DMA_CAP_64B is defined.
So the return NULL can never be reached. Remove the dead code.

Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xilinx.com>
Reviewed-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: stmmac: selftests: Flow Control test can also run with ASYM Pause

The Flow Control selftest is also available with ASYM Pause. Lets add
this check to the test and fix eventual false positive failures.

Fixes: 091810dbded9 ("net: stmmac: Introduce selftests support")
Signed-off-by: Jose Abreu <Jose.Abreu@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

gianfar: Make reset_gfar static

Fix sparse warning:

drivers/net/ethernet/freescale/gianfar.c:2070:6:
 warning: symbol 'reset_gfar' was not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

atm: he: clean up an indentation issue

There is a statement that is indented one level too many, remove
the extraneous tab.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ppp: Fix memory leak in ppp_write

When ppp is closing, __ppp_xmit_process() failed to enqueue skb
and skb allocated in ppp_write() is leaked.

syzbot reported :
BUG: memory leak
unreferenced object 0xffff88812a17bc00 (size 224):
  comm "syz-executor673", pid 6952, jiffies 4294942888 (age 13.040s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<00000000d110fff9>] kmemleak_alloc_recursive include/linux/kmemleak.h:43 [inline]
    [<00000000d110fff9>] slab_post_alloc_hook mm/slab.h:522 [inline]
    [<00000000d110fff9>] slab_alloc_node mm/slab.c:3262 [inline]
    [<00000000d110fff9>] kmem_cache_alloc_node+0x163/0x2f0 mm/slab.c:3574
    [<000000002d616113>] __alloc_skb+0x6e/0x210 net/core/skbuff.c:197
    [<000000000167fc45>] alloc_skb include/linux/skbuff.h:1055 [inline]
    [<000000000167fc45>] ppp_write+0x48/0x120 drivers/net/ppp/ppp_generic.c:502
    [<000000009ab42c0b>] __vfs_write+0x43/0xa0 fs/read_write.c:494
    [<00000000086b2e22>] vfs_write fs/read_write.c:558 [inline]
    [<00000000086b2e22>] vfs_write+0xee/0x210 fs/read_write.c:542
    [<00000000a2b70ef9>] ksys_write+0x7c/0x130 fs/read_write.c:611
    [<00000000ce5e0fdd>] __do_sys_write fs/read_write.c:623 [inline]
    [<00000000ce5e0fdd>] __se_sys_write fs/read_write.c:620 [inline]
    [<00000000ce5e0fdd>] __x64_sys_write+0x1e/0x30 fs/read_write.c:620
    [<00000000d9d7b370>] do_syscall_64+0x76/0x1a0 arch/x86/entry/common.c:296
    [<0000000006e6d506>] entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fix this by freeing skb, if ppp is closing.

Fixes: 6d066734e9f0 ("ppp: avoid loop in xmit recursion detection code")
Reported-and-tested-by: syzbot+d9c8bf24e56416d7ce2c@syzkaller.appspotmail.com
Signed-off-by: Takeshi Misawa <jeliantsurux@gmail.com>
Reviewed-by: Guillaume Nault <gnault@redhat.com>
Tested-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'ibmvnic-serialization-fixes'

Juliet Kim says:

====================
net/ibmvnic: serialization fixes

This series includes two fixes. The first improves reset code to allow
linkwatch_event to proceed during reset. The second ensures that no more
than one thread runs in reset at a time.

v2:
- Separate change param reset from do_reset()
- Return IBMVNIC_OPEN_FAILED if __ibmvnic_open fails
- Remove setting wait_for_reset to false from __ibmvnic_reset(), this
  is done in wait_for_reset()
- Move the check for force_reset_recovery from patch 1 to patch 2

v3:
- Restore reset’s successful return in open failure case

v4:
- Change resetting flag access to atomic
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net/ibmvnic: prevent more than one thread from running in reset

The current code allows more than one thread to run in reset. This can
corrupt struct adapter data. Check adapter->resetting before performing
a reset, if there is another reset running delay (100 msec) before trying
again.

Signed-off-by: Juliet Kim <julietk@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/ibmvnic: unlock rtnl_lock in reset so linkwatch_event can run

Commit a5681e20b541 ("net/ibmnvic: Fix deadlock problem in reset")
made the change to hold the RTNL lock during a reset to avoid deadlock
but linkwatch_event is fired during the reset and needs the RTNL lock.
That keeps linkwatch_event process from proceeding until the reset
is complete. The reset process cannot tolerate the linkwatch_event
processing after reset completes, so release the RTNL lock during the
process to allow a chance for linkwatch_event to run during reset.
This does not guarantee that the linkwatch_event will be processed as
soon as link state changes, but is an improvement over the current code
where linkwatch_event processing is always delayed, which prevents
transmissions on the device from being deactivated leading transmit
watchdog timer to time-out.

Release the RTNL lock before link state change and re-acquire after
the link state change to allow linkwatch_event to grab the RTNL lock
and run during the reset.

Fixes: a5681e20b541 ("net/ibmnvic: Fix deadlock problem in reset")
Signed-off-by: Juliet Kim <julietk@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

wil6210: use after free in wil_netif_rx_any()

The debug code dereferences "skb" to print "skb->len" so we have to
print the message before we free "skb".

Fixes: f99fe49ff372 ("wil6210: add wil_netif_rx() helper function")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

skge: fix checksum byte order

Running old skge driver on PowerPC causes checksum errors
because hardware reported 1's complement checksum is in little-endian
byte order.

Reported-by: Benoit <benoit.sansoni@gmail.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

arcnet: provide a buffer big enough to actually receive packets

struct archdr is only big enough to hold the header of various types of
arcnet packets. So to provide enough space to hold the data read from
hardware provide a buffer large enough to hold a packet with maximal
size.

The problem was noticed by the stack protector which makes the kernel
oops.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

iwlwifi: fw: don't send GEO_TX_POWER_LIMIT command to FW version 36

The intention was to have the GEO_TX_POWER_LIMIT command in FW version
36 as well, but not all 8000 family got this feature enabled.  The
8000 family is the only one using version 36, so skip this version
entirely.  If we try to send this command to the firmwares that do not
support it, we get a BAD_COMMAND response from the firmware.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=204151.

Cc: stable@vger.kernel.org # 4.19+
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

mt76: mt7615: fix mt7615 firmware path definitions

mt7615 patch/n9/cr4 firmwares are available in mediatek folder in
linux-firmware repository. Because of this mt7615 won't work on regular
distributions like Ubuntu. Fix path definitions.  Moreover remove useless
firmware name pointers and use definitions directly

Fixes: 04b8e65922f6 ("mt76: add mac80211 driver for MT7615 PCIe-based chipsets")
Cc: stable@vger.kernel.org
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Merge branch 'check-CAP_NEW_RAW'

Greg Kroah-Hartman says:

====================
Raw socket cleanups

Ori Nimron pointed out that there are a number of places in the kernel
where you can create a raw socket, without having to have the
CAP_NET_RAW permission.

To resolve this, here's a short patch series to test these odd and old
protocols for this permission before allowing the creation to succeed

All patches are currently against the net tree.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

nfc: enforce CAP_NET_RAW for raw sockets

When creating a raw AF_NFC socket, CAP_NET_RAW needs to be checked
first.

Signed-off-by: Ori Nimron <orinimron123@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ieee802154: enforce CAP_NET_RAW for raw sockets

When creating a raw AF_IEEE802154 socket, CAP_NET_RAW needs to be
checked first.

Signed-off-by: Ori Nimron <orinimron123@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ax25: enforce CAP_NET_RAW for raw sockets

When creating a raw AF_AX25 socket, CAP_NET_RAW needs to be checked
first.

Signed-off-by: Ori Nimron <orinimron123@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

appletalk: enforce CAP_NET_RAW for raw sockets

When creating a raw AF_APPLETALK socket, CAP_NET_RAW needs to be checked
first.

Signed-off-by: Ori Nimron <orinimron123@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

mISDN: enforce CAP_NET_RAW for raw sockets

When creating a raw AF_ISDN socket, CAP_NET_RAW needs to be checked
first.

Signed-off-by: Ori Nimron <orinimron123@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: sched: fix possible crash in tcf_action_destroy()

If the allocation done in tcf_exts_init() failed,
we end up with a NULL pointer in exts->actions.

kasan: GPF could be caused by NULL-ptr deref or user memory access
general protection fault: 0000 [#1] PREEMPT SMP KASAN
CPU: 1 PID: 8198 Comm: syz-executor.3 Not tainted 5.3.0-rc8+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
RIP: 0010:tcf_action_destroy+0x71/0x160 net/sched/act_api.c:705
Code: c3 08 44 89 ee e8 4f cb bb fb 41 83 fd 20 0f 84 c9 00 00 00 e8 c0 c9 bb fb 48 89 d8 48 b9 00 00 00 00 00 fc ff df 48 c1 e8 03 <80> 3c 08 00 0f 85 c0 00 00 00 4c 8b 33 4d 85 f6 0f 84 9d 00 00 00
RSP: 0018:ffff888096e16ff0 EFLAGS: 00010246
RAX: 0000000000000000 RBX: 0000000000000000 RCX: dffffc0000000000
RDX: 0000000000040000 RSI: ffffffff85b6ab30 RDI: 0000000000000000
RBP: ffff888096e17020 R08: ffff8880993f6140 R09: fffffbfff11cae67
R10: fffffbfff11cae66 R11: ffffffff88e57333 R12: 0000000000000000
R13: 0000000000000000 R14: ffff888096e177a0 R15: 0000000000000001
FS:  00007f62bc84a700(0000) GS:ffff8880ae900000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000758040 CR3: 0000000088b64000 CR4: 00000000001426e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 tcf_exts_destroy+0x38/0xb0 net/sched/cls_api.c:3030
 tcindex_set_parms+0xf7f/0x1e50 net/sched/cls_tcindex.c:488
 tcindex_change+0x230/0x318 net/sched/cls_tcindex.c:519
 tc_new_tfilter+0xa4b/0x1c70 net/sched/cls_api.c:2152
 rtnetlink_rcv_msg+0x838/0xb00 net/core/rtnetlink.c:5214
 netlink_rcv_skb+0x177/0x450 net/netlink/af_netlink.c:2477
 rtnetlink_rcv+0x1d/0x30 net/core/rtnetlink.c:5241
 netlink_unicast_kernel net/netlink/af_netlink.c:1302 [inline]
 netlink_unicast+0x531/0x710 net/netlink/af_netlink.c:1328
 netlink_sendmsg+0x8a5/0xd60 net/netlink/af_netlink.c:1917
 sock_sendmsg_nosec net/socket.c:637 [inline]
 sock_sendmsg+0xd7/0x130 net/socket.c:657
 ___sys_sendmsg+0x3e2/0x920 net/socket.c:2311
 __sys_sendmmsg+0x1bf/0x4d0 net/socket.c:2413
 __do_sys_sendmmsg net/socket.c:2442 [inline]

Fixes: 90b73b77d08e ("net: sched: change action API to use array of pointers to actions")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Cc: Vlad Buslov <vladbu@mellanox.com>
Cc: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx5e: Fix matching on tunnel addresses type

In mlx5 parse_tunnel_attr() function dispatch on encap IP address type
is performed by directly checking flow_rule_match_key() on
FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS, and then on
FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS. However, since those are stored in
union, first check is always true if any type of encap address is set,
which leads to IPv6 tunnel encap address being parsed as IPv4 by mlx5.
Determine correct IP address type by checking control key first and if
it set, take address type from match.key->addr_type.

Fixes: d1bda7eecd88 ("net/mlx5e: Allow matching only enc_key_id/enc_dst_port for decapsulation action")
Signed-off-by: Dmytro Linkin <dmitrolin@mellanox.com>
Reviewed-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Eli Britstein <elibr@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

net/mlx5e: Fix traffic duplication in ethtool steering

Before this patch, when adding multiple ethtool steering rules with
identical classification, the driver used to append the new destination
to the already existing hw rule, which caused the hw to forward the
traffic to all destinations (rx queues).

Here we avoid this by setting the "no append" mlx5 fs core flag when
adding a new ethtool rule.

Fixes: 6dc6071cfcde ("net/mlx5e: Add ethtool flow steering support")
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

net/mlx5: Add device ID of upcoming BlueField-2

Add the device ID of upcoming BlueField-2 integrated ConnectX-6 Dx
network controller. Its VFs will be using the generic VF device ID:
0x101e "ConnectX Family mlx5Gen Virtual Function".

Fixes: 2e9d3e83ab82 ("net/mlx5: Update the list of the PCI supported devices")
Signed-off-by: Bodong Wang <bodong@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

net/mlx5: DR, Allow matching on vport based on vhca_id

In case source_eswitch_owner_vhca_id is given as a match,
the source_vport (vhca_id) will be set in case vhca_id_valid.

This will allow matching on peer vports, vports that belong
to the other pf.

Fixes: 26d688e33f88 ("net/mlx5: DR, Add Steering entry (STE) utilities")
Signed-off-by: Alaa Hleihel <alaa@mellanox.com>
Signed-off-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

net/mlx5: DR, Fix getting incorrect prev node in ste_free

When we free an STE and the STE is in the middle of collision
list, the prev_ste was obtained incorrectly from the list.
To avoid such issues list_entry calls replaced with standard list API.

Fixes: 26d688e33f88 ("net/mlx5: DR, Add Steering entry (STE) utilities")
Signed-off-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

net/mlx5: DR, Remove redundant vport number from action

The vport number is part of the vport_cap, there is no reason
to store in a separate variable on the vport.

Fixes: 9db810ed2d37 ("net/mlx5: DR, Expose steering action functionality")
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

net/mlx5: DR, Fix SW steering HW bits and definitions

Fix wrong reserved bits offsets.

Fixes: 97b5484ed608 ("net/mlx5: Add HW bits and definitions required for SW steering")
Signed-off-by: Yevgeny Kliteynik <kliteyn@mellanox.com>
Reviewed-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

rtw88: configure firmware after HCI started

After firmware has been downloaded, driver should send
some information to it through H2C commands. Those H2C
commands are transmitted through TX path.

But before HCI has been started, the TX path is not
working completely. Such as PCI interfaces, the interrupts
are not enabled, hence TX interrupts will not be issued
after H2C skb has been DMAed to the device. And the H2C
skbs will not be released until the device is powered off.

Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

net: dsa: Use the correct style for SPDX License Identifier

This patch corrects the SPDX License Identifier style
in header file for Distributed Switch Architecture drivers.
For C header files Documentation/process/license-rules.rst
mandates C-like comments (opposed to C source files where
C++ style should be used)

Changes made by using a script provided by Joe Perches here:
https://lkml.org/lkml/2019/2/7/46.

Suggested-by: Joe Perches <joe@perches.com>
Signed-off-by: Nishad Kamdar <nishadkamdar@gmail.com>
Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net: dsa: b53: Use the correct style for SPDX License Identifier

This patch corrects the SPDX License Identifier style
in header file for Broadcom BCM53xx managed switch driver.
For C header files Documentation/process/license-rules.rst
mandates C-like comments (opposed to C source files where
C++ style should be used)

Changes made by using a script provided by Joe Perches here:
https://lkml.org/lkml/2019/2/7/46.

Suggested-by: Joe Perches <joe@perches.com>
Signed-off-by: Nishad Kamdar <nishadkamdar@gmail.com>
Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

ionic: Fix an error code in ionic_lif_alloc()

We need to set the error code on this path.  Otherwise it probably
results in a NULL dereference down the line.

Fixes: aa3198819bea ("ionic: Add RSS support")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net: ena: Select DIMLIB for ENA_ETHERNET

If CONFIG_ENA_ETHERNET=y and CONFIG_DIMLIB=n,
below erros can be found:
drivers/net/ethernet/amazon/ena/ena_netdev.o: In function `ena_dim_work':
ena_netdev.c:(.text+0x21cc): undefined reference to `net_dim_get_rx_moderation'
ena_netdev.c:(.text+0x21cc): relocation truncated to
fit: R_AARCH64_CALL26 against undefined symbol `net_dim_get_rx_moderation'
drivers/net/ethernet/amazon/ena/ena_netdev.o: In function `ena_io_poll':
ena_netdev.c:(.text+0x7bd4): undefined reference to `net_dim'
ena_netdev.c:(.text+0x7bd4): relocation truncated to fit:
R_AARCH64_CALL26 against undefined symbol `net_dim'

After commit 282faf61a053 ("net: ena: switch to dim algorithm for rx adaptive
interrupt moderation"), it introduces dim algorithm, which configured by CONFIG_DIMLIB.
So, this patch is to select DIMLIB for ENA_ETHERNET.

Fixes: 282faf61a053 ("net: ena: switch to dim algorithm for rx adaptive interrupt moderation")
Signed-off-by: Mao Wenan <maowenan@huawei.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

qede: qede_fp: simplify a bit 'qede_rx_build_skb()'

Use 'skb_put_data()' instead of rewritting it.
This improves readability.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net: stmmac: selftest: avoid large stack usage

Putting a struct stmmac_rss object on the stack is a bad idea,
as it exceeds the warning limit for a stack frame on 32-bit architectures:

drivers/net/ethernet/stmicro/stmmac/stmmac_selftests.c:1221:12: error: stack frame size of 1208 bytes in function '__stmmac_test_l3filt' [-Werror,-Wframe-larger-than=]
drivers/net/ethernet/stmicro/stmmac/stmmac_selftests.c:1338:12: error: stack frame size of 1208 bytes in function '__stmmac_test_l4filt' [-Werror,-Wframe-larger-than=]

As the object is the trivial empty case, change the called function
to accept a NULL pointer to mean the same thing and remove the
large variable in the two callers.

Fixes: 4647e021193d ("net: stmmac: selftests: Add selftest for L3/L4 Filters")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Jose Abreu <joabreu@synopsys.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net: dsa: sja1105: Add dependency for NET_DSA_SJA1105_TAS

If CONFIG_NET_DSA_SJA1105_TAS=y and CONFIG_NET_SCH_TAPRIO=n,
below error can be found:
drivers/net/dsa/sja1105/sja1105_tas.o: In function `sja1105_setup_tc_taprio':
sja1105_tas.c:(.text+0x318): undefined reference to `taprio_offload_free'
sja1105_tas.c:(.text+0x590): undefined reference to `taprio_offload_get'
drivers/net/dsa/sja1105/sja1105_tas.o: In function `sja1105_tas_teardown':
sja1105_tas.c:(.text+0x610): undefined reference to `taprio_offload_free'
make: *** [vmlinux] Error 1

sja1105_tas needs tc-taprio, so this patch add the dependency for it.

Fixes: 317ab5b86c8e ("net: dsa: sja1105: Configure the Time-Aware Scheduler via tc-taprio offload")
Signed-off-by: Mao Wenan <maowenan@huawei.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net_sched: add policy validation for action attributes

Similar to commit 8b4c3cdd9dd8
("net: sched: Add policy validation for tc attributes"), we need
to add proper policy validation for TC action attributes too.

Cc: David Ahern <dsahern@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net_sched: add max len check for TCA_KIND

The TCA_KIND attribute is of NLA_STRING which does not check
the NUL char. KMSAN reported an uninit-value of TCA_KIND which
is likely caused by the lack of NUL.

Change it to NLA_NUL_STRING and add a max len too.

Fixes: 8b4c3cdd9dd8 ("net: sched: Add policy validation for tc attributes")
Reported-and-tested-by: syzbot+618aacd49e8c8b8486bd@syzkaller.appspotmail.com
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net: remove netx ethernet driver

The ARM netx platform got removed in 5.3, so this driver
is now useless.

Reported-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net: qrtr: Stop rx_worker before freeing node

As the endpoint is unregistered there might still be work pending to
handle incoming messages, which will result in a use after free
scenario. The plan is to remove the rx_worker, but until then (and for
stable@) ensure that the work is stopped before the node is freed.

Fixes: bdabad3e363d ("net: Add Qualcomm IPC router")
Cc: stable@vger.kernel.org
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net/phy: fix DP83865 10 Mbps HDX loopback disable function

According to the DP83865 datasheet "the 10 Mbps HDX loopback can be
disabled in the expanded memory register 0x1C0.1". The driver erroneously
used bit 0 instead of bit 1.

Fixes: 4621bf129856 ("phy: Add file missed in previous commit.")
Signed-off-by: Peter Mamonov <pmamonov@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

usbnet: ignore endpoints with invalid wMaxPacketSize

Endpoints with zero wMaxPacketSize are not usable for transferring
data. Ignore such endpoints when looking for valid in, out and
status pipes, to make the drivers more robust against invalid and
meaningless descriptors.

The wMaxPacketSize of these endpoints are used for memory allocations
and as divisors in many usbnet minidrivers. Avoiding zero is therefore
critical.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

cdc_ncm: fix divide-by-zero caused by invalid wMaxPacketSize

Endpoints with zero wMaxPacketSize are not usable for transferring
data. Ignore such endpoints when looking for valid in, out and
status pipes, to make the driver more robust against invalid and
meaningless descriptors.

The wMaxPacketSize of the out pipe is used as divisor. So this change
fixes a divide-by-zero bug.

Reported-by: syzbot+ce366e2b8296e25d84f5@syzkaller.appspotmail.com
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

zd1211rw: zd_usb: Use "%zu" to format size_t

On 32-bit:

    drivers/net/wireless/zydas/zd1211rw/zd_usb.c: In function ‘check_read_regs’:
    drivers/net/wireless/zydas/zd1211rw/zd_def.h:18:25: warning: format ‘%ld’ expects argument of type ‘long int’, but argument 6 has type ‘size_t’ {aka ‘unsigned int’} [-Wformat=]
      dev_printk(level, dev, "%s() " fmt, __func__, ##args)
     ^~~~~~~
    drivers/net/wireless/zydas/zd1211rw/zd_def.h:22:4: note: in expansion of macro ‘dev_printk_f’
dev_printk_f(KERN_DEBUG, dev, fmt, ## args)
^~~~~~~~~~~~
    drivers/net/wireless/zydas/zd1211rw/zd_usb.c:1635:3: note: in expansion of macro ‘dev_dbg_f’
       dev_dbg_f(zd_usb_dev(usb),
       ^~~~~~~~~
    drivers/net/wireless/zydas/zd1211rw/zd_usb.c:1636:51: note: format string is defined here
 "error: actual length %d less than expected %ld\n",
     ~~^
     %d

Fixes: 84b0b66352470e64 ("zd1211rw: zd_usb: Use struct_size() helper")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

rtw88: pci: release tx skbs DMAed when stop

Interrupt is disabled to stop PCI, which means the skbs
queued for each TX ring will not be released via DMA
interrupt. To avoid those skbs remained being left in
the skb queue until PCI has been removed, driver needs
to release skbs by itself.

Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Reviewed-by: Brian Norris <briannorris@chromium.org>
Tested-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

rtw88: pci: extract skbs free routine for trx rings

These skbs free routines could be used when driver wants
to stop PCI bus, because some of the skbs remained in the
queue may not have been returned via DMA interrupt.

Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

ipv6: fix a typo in fib6_rule_lookup()

Yi Ren reported an issue discovered by syzkaller, and bisected
to the cited commit.

Many thanks to Yi, this trivial patch does not reflect the patient
work that has been done.

Fixes: d64a1f574a29 ("ipv6: honor RT6_LOOKUP_F_DST_NOREF in rule lookup logic")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Wei Wang <weiwan@google.com>
Bisected-and-reported-by: Yi Ren <c4tren@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

sch_netem: fix a divide by zero in tabledist()

syzbot managed to crash the kernel in tabledist() loading
an empty distribution table.

t = dist->table[rnd % dist->size];

Simply return an error when such load is attempted.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

dt-bindings: net: remove un-implemented property

The `adi,disable-energy-detect` property was implemented in an initial
version of the `adin` driver series, but after a review it was discarded in
favor of implementing the ETHTOOL_PHY_EDPD phy-tunable option.

With the ETHTOOL_PHY_EDPD control, it's possible to disable/enable
Energy-Detect-Power-Down for the `adin` PHY, so this device-tree is not
needed.

Fixes: 767078132ff9 ("dt-bindings: net: add bindings for ADIN PHY driver")
Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

ipv4: Revert removal of rt_uses_gateway

Julian noted that rt_uses_gateway has a more subtle use than 'is gateway
set':
    https://lore.kernel.org/netdev/alpine.LFD.2.21.1909151104060.2546@ja.home.ssi.bg/

Revert that part of the commit referenced in the Fixes tag.

Currently, there are no u8 holes in 'struct rtable'. There is a 4-byte hole
in the second cacheline which contains the gateway declaration. So move
rt_gw_family down to the gateway declarations since they are always used
together, and then re-use that u8 for rt_uses_gateway. End result is that
rtable size is unchanged.

Fixes: 1550c171935d ("ipv4: Prepare rtable for IPv6 gateway")
Reported-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

selftests: Update fib_nexthop_multiprefix to handle missing ping6

Some distributions (e.g., debian buster) do not install ping6. Re-use
the hook in pmtu.sh to detect this and fallback to ping.

Fixes: 735ab2f65dce ("selftests: Add test with multiple prefixes using single nexthop")
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

selftests: Update fib_tests to handle missing ping6

Some distributions (e.g., debian buster) do not install ping6. Re-use
the hook in pmtu.sh to detect this and fallback to ping.

Fixes: a0e11da78f48 ("fib_tests: Add tests for metrics on routes")
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

dt-bindings: net: dwmac: fix 'mac-mode' type

The 'mac-mode' property is similar to 'phy-mode' and 'phy-connection-type',
which are enums of mode strings.

The 'dwmac' driver supports almost all modes declared in the 'phy-mode'
enum (except for 1 or 2). But in general, there may be a case where
'mac-mode' becomes more generic and is moved as part of phylib or netdev.

In any case, the 'mac-mode' field should be made an enum, and it also makes
sense to just reference the 'phy-connection-type' from
'ethernet-controller.yaml'. That will also make it more future-proof for new
modes.

Fixes: 9c15d3597c62 ("dt-bindings: net: dwmac: document 'mac-mode' property")
Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>

net/sched: act_sample: don't push mac header on ip6gre ingress

current 'sample' action doesn't push the mac header of ingress packets if
they are received by a layer 3 tunnel (like gre or sit); but it forgot to
check for gre over ipv6, so the following script:

 # tc q a dev $d clsact
 # tc f a dev $d ingress protocol ip flower ip_proto icmp action sample \
 > group 100 rate 1
 # psample -v -g 100

dumps everything, including outer header and mac, when $d is a gre tunnel
over ipv6. Fix this adding a missing label for ARPHRD_IP6GRE devices.

Fixes: 5c5670fae430 ("net/sched: Introduce sample tc action")
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Reviewed-by: Yotam Gigi <yotam.gi@gmail.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>