openwrt/staging/blogic.git
7 years agoselftests: Introduce a new script to generate tc batch file
Chris Mi [Fri, 27 Oct 2017 01:24:42 +0000 (18:24 -0700)]
selftests: Introduce a new script to generate tc batch file

  # ./tdc_batch.py -h
  usage: tdc_batch.py [-h] [-n NUMBER] [-o] [-s] [-p] device file

  TC batch file generator

  positional arguments:
    device                device name
    file                  batch file name

  optional arguments:
    -h, --help            show this help message and exit
    -n NUMBER, --number NUMBER
                          how many lines in batch file
    -o, --skip_sw         skip_sw (offload), by default skip_hw
    -s, --share_action    all filters share the same action
    -p, --prio            all filters have different prio

Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: Lucas Bates <lucasb@mojatatu.com>
Signed-off-by: Chris Mi <chrism@mellanox.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: fix call_rcu() race on act_sample module removal
Cong Wang [Fri, 27 Oct 2017 01:24:41 +0000 (18:24 -0700)]
net_sched: fix call_rcu() race on act_sample module removal

Similar to commit c78e1746d3ad
("net: sched: fix call_rcu() race on classifier module unloads"),
we need to wait for flying RCU callback tcf_sample_cleanup_rcu().

Cc: Yotam Gigi <yotamg@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: add rtnl assertion to tcf_exts_destroy()
Cong Wang [Fri, 27 Oct 2017 01:24:40 +0000 (18:24 -0700)]
net_sched: add rtnl assertion to tcf_exts_destroy()

After previous patches, it is now safe to claim that
tcf_exts_destroy() is always called with RTNL lock.

Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in tcindex filter
Cong Wang [Fri, 27 Oct 2017 01:24:39 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in tcindex filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in rsvp filter
Cong Wang [Fri, 27 Oct 2017 01:24:38 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in rsvp filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in route filter
Cong Wang [Fri, 27 Oct 2017 01:24:37 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in route filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in u32 filter
Cong Wang [Fri, 27 Oct 2017 01:24:36 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in u32 filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in matchall filter
Cong Wang [Fri, 27 Oct 2017 01:24:35 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in matchall filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in fw filter
Cong Wang [Fri, 27 Oct 2017 01:24:34 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in fw filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in flower filter
Cong Wang [Fri, 27 Oct 2017 01:24:33 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in flower filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in flow filter
Cong Wang [Fri, 27 Oct 2017 01:24:32 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in flow filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in cgroup filter
Cong Wang [Fri, 27 Oct 2017 01:24:31 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in cgroup filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in bpf filter
Cong Wang [Fri, 27 Oct 2017 01:24:30 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in bpf filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: use tcf_queue_work() in basic filter
Cong Wang [Fri, 27 Oct 2017 01:24:29 +0000 (18:24 -0700)]
net_sched: use tcf_queue_work() in basic filter

Defer the tcf_exts_destroy() in RCU callback to
tc filter workqueue and get RTNL lock.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: introduce a workqueue for RCU callbacks of tc filter
Cong Wang [Fri, 27 Oct 2017 01:24:28 +0000 (18:24 -0700)]
net_sched: introduce a workqueue for RCU callbacks of tc filter

This patch introduces a dedicated workqueue for tc filters
so that each tc filter's RCU callback could defer their
action destroy work to this workqueue. The helper
tcf_queue_work() is introduced for them to use.

Because we hold RTNL lock when calling tcf_block_put(), we
can not simply flush works inside it, therefore we have to
defer it again to this workqueue and make sure all flying RCU
callbacks have already queued their work before this one, in
other words, to ensure this is the last one to execute to
prevent any use-after-free.

On the other hand, this makes tcf_block_put() ugly and
harder to understand. Since David and Eric strongly dislike
adding synchronize_rcu(), this is probably the only
solution that could make everyone happy.

Please also see the code comments below.

Reported-by: Chris Mi <chrism@mellanox.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Jiri Pirko <jiri@resnulli.us>
Cc: John Fastabend <john.fastabend@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'sctp-endianness-fixes'
David S. Miller [Sun, 29 Oct 2017 09:03:25 +0000 (18:03 +0900)]
Merge branch 'sctp-endianness-fixes'

Xin Long says:

====================
sctp: a bunch of fixes for some sparse warnings

As Eric noticed, when running 'make C=2 M=net/sctp/', a plenty of
warnings or errors checked by sparse appear. They are all problems
about Endian and type cast.

Most of them are just warnings by which no issues could be caused
while some might be bugs.

This patchset fixes them with four patches basically according to
how they are introduced.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosctp: fix some type cast warnings introduced since very beginning
Xin Long [Sat, 28 Oct 2017 11:43:57 +0000 (19:43 +0800)]
sctp: fix some type cast warnings introduced since very beginning

These warnings were found by running 'make C=2 M=net/sctp/'.
They are there since very beginning.

Note after this patch, there still one warning left in
sctp_outq_flush():
  sctp_chunk_fail(chunk, SCTP_ERROR_INV_STRM)

Since it has been moved to sctp_stream_outq_migrate on net-next,
to avoid the extra job when merging net-next to net, I will post
the fix for it after the merging is done.

Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosctp: fix a type cast warnings that causes a_rwnd gets the wrong value
Xin Long [Sat, 28 Oct 2017 11:43:56 +0000 (19:43 +0800)]
sctp: fix a type cast warnings that causes a_rwnd gets the wrong value

These warnings were found by running 'make C=2 M=net/sctp/'.

Commit d4d6fb5787a6 ("sctp: Try not to change a_rwnd when faking a
SACK from SHUTDOWN.") expected to use the peers old rwnd and add
our flight size to the a_rwnd. But with the wrong Endian, it may
not work as well as expected.

So fix it by converting to the right value.

Fixes: d4d6fb5787a6 ("sctp: Try not to change a_rwnd when faking a SACK from SHUTDOWN.")
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosctp: fix some type cast warnings introduced by transport rhashtable
Xin Long [Sat, 28 Oct 2017 11:43:55 +0000 (19:43 +0800)]
sctp: fix some type cast warnings introduced by transport rhashtable

These warnings were found by running 'make C=2 M=net/sctp/'.

They are introduced by not aware of Endian for the port when
coding transport rhashtable patches.

Fixes: 7fda702f9315 ("sctp: use new rhlist interface on sctp transport rhashtable")
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosctp: fix some type cast warnings introduced by stream reconf
Xin Long [Sat, 28 Oct 2017 11:43:54 +0000 (19:43 +0800)]
sctp: fix some type cast warnings introduced by stream reconf

These warnings were found by running 'make C=2 M=net/sctp/'.

They are introduced by not aware of Endian when coding stream
reconf patches.

Since commit c0d8bab6ae51 ("sctp: add get and set sockopt for
reconf_enable") enabled stream reconf feature for users, the
Fixes tag below would use it.

Fixes: c0d8bab6ae51 ("sctp: add get and set sockopt for reconf_enable")
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet_sched: avoid matching qdisc with zero handle
Cong Wang [Sat, 28 Oct 2017 05:08:56 +0000 (22:08 -0700)]
net_sched: avoid matching qdisc with zero handle

Davide found the following script triggers a NULL pointer
dereference:

ip l a name eth0 type dummy
tc q a dev eth0 parent :1 handle 1: htb

This is because for a freshly created netdevice noop_qdisc
is attached and when passing 'parent :1', kernel actually
tries to match the major handle which is 0 and noop_qdisc
has handle 0 so is matched by mistake. Commit 69012ae425d7
tries to fix a similar bug but still misses this case.

Handle 0 is not a valid one, should be just skipped. In
fact, kernel uses it as TC_H_UNSPEC.

Fixes: 69012ae425d7 ("net: sched: fix handling of singleton qdiscs with qdisc_hash")
Fixes: 59cc1f61f09c ("net: sched:convert qdisc linked list to hashtable")
Reported-by: Davide Caratti <dcaratti@redhat.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosctp: reset owner sk for data chunks on out queues when migrating a sock
Xin Long [Fri, 27 Oct 2017 18:13:29 +0000 (02:13 +0800)]
sctp: reset owner sk for data chunks on out queues when migrating a sock

Now when migrating sock to another one in sctp_sock_migrate(), it only
resets owner sk for the data in receive queues, not the chunks on out
queues.

It would cause that data chunks length on the sock is not consistent
with sk sk_wmem_alloc. When closing the sock or freeing these chunks,
the old sk would never be freed, and the new sock may crash due to
the overflow sk_wmem_alloc.

syzbot found this issue with this series:

  r0 = socket$inet_sctp()
  sendto$inet(r0)
  listen(r0)
  accept4(r0)
  close(r0)

Although listen() should have returned error when one TCP-style socket
is in connecting (I may fix this one in another patch), it could also
be reproduced by peeling off an assoc.

This issue is there since very beginning.

This patch is to reset owner sk for the chunks on out queues so that
sk sk_wmem_alloc has correct value after accept one sock or peeloff
an assoc to one sock.

Note that when resetting owner sk for chunks on outqueue, it has to
sctp_clear_owner_w/skb_orphan chunks before changing assoc->base.sk
first and then sctp_set_owner_w them after changing assoc->base.sk,
due to that sctp_wfree and it's callees are using assoc->base.sk.

Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'sockmap-fixes'
David S. Miller [Sun, 29 Oct 2017 02:18:49 +0000 (11:18 +0900)]
Merge branch 'sockmap-fixes'

John Fastabend says:

====================
net: sockmap fixes

Last two fixes (as far as I know) for sockmap code this round.

First, we are using the qdisc cb structure when making the data end
calculation. This is really just wrong so, store it with the other
metadata in the correct tcp_skb_cb sturct to avoid breaking things.

Next, with recent work to attach multiple programs to a cgroup a
specific enumeration of return codes was agreed upon. However,
I wrote the sk_skb program types before seeing this work and used
a different convention. Patch 2 in the series aligns the return
codes to avoid breaking with this infrastructure and also aligns
with other programming conventions to avoid being the odd duck out
forcing programs to remember SK_SKB programs are different. Pusing
to net because its a user visible change. With this SK_SKB program
return codes are the same as other cgroup program types.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: rename sk_actions to align with bpf infrastructure
John Fastabend [Fri, 27 Oct 2017 16:45:53 +0000 (09:45 -0700)]
bpf: rename sk_actions to align with bpf infrastructure

Recent additions to support multiple programs in cgroups impose
a strict requirement, "all yes is yes, any no is no". To enforce
this the infrastructure requires the 'no' return code, SK_DROP in
this case, to be 0.

To apply these rules to SK_SKB program types the sk_actions return
codes need to be adjusted.

This fix adds SK_PASS and makes 'SK_DROP = 0'. Finally, remove
SK_ABORTED to remove any chance that the API may allow aborted
program flows to be passed up the stack. This would be incorrect
behavior and allow programs to break existing policies.

Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: bpf_compute_data uses incorrect cb structure
John Fastabend [Fri, 27 Oct 2017 16:45:34 +0000 (09:45 -0700)]
bpf: bpf_compute_data uses incorrect cb structure

SK_SKB program types use bpf_compute_data to store the end of the
packet data. However, bpf_compute_data assumes the cb is stored in the
qdisc layer format. But, for SK_SKB this is the wrong layer of the
stack for this type.

It happens to work (sort of!) because in most cases nothing happens
to be overwritten today. This is very fragile and error prone.
Fortunately, we have another hole in tcp_skb_cb we can use so lets
put the data_end value there.

Note, SK_SKB program types do not use data_meta, they are failed by
sk_skb_is_valid_access().

Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotap: reference to KVA of an unloaded module causes kernel panic
Girish Moodalbail [Fri, 27 Oct 2017 07:00:16 +0000 (00:00 -0700)]
tap: reference to KVA of an unloaded module causes kernel panic

The commit 9a393b5d5988 ("tap: tap as an independent module") created a
separate tap module that implements tap functionality and exports
interfaces that will be used by macvtap and ipvtap modules to create
create respective tap devices.

However, that patch introduced a regression wherein the modules macvtap
and ipvtap can be removed (through modprobe -r) while there are
applications using the respective /dev/tapX devices. These applications
cause kernel to hold reference to /dev/tapX through 'struct cdev
macvtap_cdev' and 'struct cdev ipvtap_dev' defined in macvtap and ipvtap
modules respectively. So,  when the application is later closed the
kernel panics because we are referencing KVA that is present in the
unloaded modules.

----------8<------- Example ----------8<----------
$ sudo ip li add name mv0 link enp7s0 type macvtap
$ sudo ip li show mv0 |grep mv0| awk -e '{print $1 $2}'
  14:mv0@enp7s0:
$ cat /dev/tap14 &
$ lsmod |egrep -i 'tap|vlan'
macvtap                16384  0
macvlan                24576  1 macvtap
tap                    24576  3 macvtap
$ sudo modprobe -r macvtap
$ fg
cat /dev/tap14
^C

<...system panics...>
BUG: unable to handle kernel paging request at ffffffffa038c500
IP: cdev_put+0xf/0x30
----------8<-----------------8<----------

The fix is to set cdev.owner to the module that creates the tap device
(either macvtap or ipvtap). With this set, the operations (in
fs/char_dev.c) on char device holds and releases the module through
cdev_get() and cdev_put() and will not allow the module to unload
prematurely.

Fixes: 9a393b5d5988ea4e (tap: tap as an independent module)
Signed-off-by: Girish Moodalbail <girish.moodalbail@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: refresh tp timestamp before tcp_mtu_probe()
Eric Dumazet [Fri, 27 Oct 2017 04:21:40 +0000 (21:21 -0700)]
tcp: refresh tp timestamp before tcp_mtu_probe()

In the unlikely event tcp_mtu_probe() is sending a packet, we
want tp->tcp_mstamp being as accurate as possible.

This means we need to call tcp_mstamp_refresh() a bit earlier in
tcp_write_xmit().

Fixes: 385e20706fac ("tcp: use tp->tcp_mstamp in output path")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotuntap: properly align skb->head before building skb
Jason Wang [Fri, 27 Oct 2017 03:05:44 +0000 (11:05 +0800)]
tuntap: properly align skb->head before building skb

An unaligned alloc_frag->offset caused by previous allocation will
result an unaligned skb->head. This will lead unaligned
skb_shared_info and then unaligned dataref which requires to be
aligned for accessing on some architecture. Fix this by aligning
alloc_frag->offset before the frag refilling.

Fixes: 0bbd7dad34f8 ("tun: make tun_build_skb() thread safe")
Cc: Eric Dumazet <edumazet@google.com>
Cc: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Cc: Wei Wei <dotweiba@gmail.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Reported-by: Wei Wei <dotweiba@gmail.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net...
David S. Miller [Fri, 27 Oct 2017 15:05:34 +0000 (00:05 +0900)]
Merge branch '40GbE' of git://git./linux/kernel/git/jkirsher/net-queue

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates 2017-10-26

This series contains fixes to e1000, igb, ixgbe and i40e.

Vincenzo Maffione fixes a potential race condition which would result in
the interface being up but transmits are disabled in the hardware.

Colin Ian King fixes a possible NULL pointer dereference in e1000, which
was found by Coverity.

Jean-Philippe Brucker fixes a possible kernel panic when a driver cannot
map a transmit buffer, which is caused by an erroneous test.

Alex provides a fix for ixgbe, which is a partial revert of the commit
ffed21bcee7a ("ixgbe: Don't bother clearing buffer memory for descriptor rings")
because the previous commit messed up the exception handling path by
adding the count back in when we did not need to.  Also fixed a typo,
where the transmit ITR setting was being used to determine if we were
using adaptive receive interrupt moderation or not.  Lastly, fixed a
memory leak by including programming descriptors in the cleaned count.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoip6_gre: update dst pmtu if dev mtu has been updated by toobig in __gre6_xmit
Xin Long [Thu, 26 Oct 2017 11:27:17 +0000 (19:27 +0800)]
ip6_gre: update dst pmtu if dev mtu has been updated by toobig in __gre6_xmit

When receiving a Toobig icmpv6 packet, ip6gre_err would just set
tunnel dev's mtu, that's not enough. For skb_dst(skb)'s pmtu may
still be using the old value, it has no chance to be updated with
tunnel dev's mtu.

Jianlin found this issue by reducing route's mtu while running
netperf, the performance went to 0.

ip6ip6 and ip4ip6 tunnel can work well with this, as they lookup
the upper dst and update_pmtu it's pmtu or icmpv6_send a Toobig
to upper socket after setting tunnel dev's mtu.

We couldn't do that for ip6_gre, as gre's inner packet could be
any protocol, it's difficult to handle them (like lookup upper
dst) in a good way.

So this patch is to fix it by updating skb_dst(skb)'s pmtu when
dev->mtu < skb_dst(skb)'s pmtu in tx path. It's safe to do this
update there, as usually dev->mtu <= skb_dst(skb)'s pmtu and no
performance regression can be caused by this.

Fixes: c12b395a4664 ("gre: Support GRE over IPv6")
Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoip6_gre: only increase err_count for some certain type icmpv6 in ip6gre_err
Xin Long [Thu, 26 Oct 2017 11:23:27 +0000 (19:23 +0800)]
ip6_gre: only increase err_count for some certain type icmpv6 in ip6gre_err

The similar fix in patch 'ipip: only increase err_count for some
certain type icmp in ipip_err' is needed for ip6gre_err.

In Jianlin's case, udp netperf broke even when receiving a TooBig
icmpv6 packet.

Fixes: c12b395a4664 ("gre: Support GRE over IPv6")
Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoipip: only increase err_count for some certain type icmp in ipip_err
Xin Long [Thu, 26 Oct 2017 11:19:56 +0000 (19:19 +0800)]
ipip: only increase err_count for some certain type icmp in ipip_err

t->err_count is used to count the link failure on tunnel and an err
will be reported to user socket in tx path if t->err_count is not 0.
udp socket could even return EHOSTUNREACH to users.

Since commit fd58156e456d ("IPIP: Use ip-tunneling code.") removed
the 'switch check' for icmp type in ipip_err(), err_count would be
increased by the icmp packet with ICMP_EXC_FRAGTIME code. an link
failure would be reported out due to this.

In Jianlin's case, when receiving ICMP_EXC_FRAGTIME a icmp packet,
udp netperf failed with the err:
  send_data: data send error: No route to host (errno 113)

We expect this error reported from tunnel to socket when receiving
some certain type icmp, but not ICMP_EXC_FRAGTIME, ICMP_SR_FAILED
or ICMP_PARAMETERPROB ones.

This patch is to bring 'switch check' for icmp type back to ipip_err
so that it only reports link failure for the right type icmp, just as
in ipgre_err() and ipip6_err().

Fixes: fd58156e456d ("IPIP: Use ip-tunneling code.")
Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: First Queue must always be in DCB mode
Jose Abreu [Thu, 26 Oct 2017 09:07:12 +0000 (10:07 +0100)]
net: stmmac: First Queue must always be in DCB mode

According to DWMAC databook the first queue operating mode
must always be in DCB.

As MTL_QUEUE_DCB = 1, we need to always set the first queue
operating mode to DCB otherwise driver will think that queue
is in AVB mode (because MTL_QUEUE_AVB = 0).

Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: dwc-qos-eth: Fix typo in DT bindings parsing
Jose Abreu [Thu, 26 Oct 2017 08:51:33 +0000 (09:51 +0100)]
net: stmmac: dwc-qos-eth: Fix typo in DT bindings parsing

According to DT bindings documentation we are expecting a
property called "snps,read-requests" but we are parsing
instead a property called "read,read-requests".

This is clearly a typo. Fix it.

Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge tag 'mlx5-fixes-2017-10-26' of git://git.kernel.org/pub/scm/linux/kernel/git...
David S. Miller [Fri, 27 Oct 2017 13:23:41 +0000 (22:23 +0900)]
Merge tag 'mlx5-fixes-2017-10-26' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
Mellanox, mlx5 fixes 2017-10-26

The series includes some misc fixes for mlx5 core and etherent driver.
Please pull and let me know if there's any problem.

For -Stable:
net/mlx5e: Properly deal with encap flows add/del under neigh update (kernels >= 4.12)
net/mlx5: Fix health work queue spin lock to IRQ safe  (kernels >= 4.13)
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge tag 'mac80211-for-davem-2017-10-25' of git://git.kernel.org/pub/scm/linux/kerne...
David S. Miller [Fri, 27 Oct 2017 04:50:06 +0000 (13:50 +0900)]
Merge tag 'mac80211-for-davem-2017-10-25' of git://git./linux/kernel/git/jberg/mac80211

Johannes Berg says:

====================
pull-request: mac80211 2017-10-25

Here are:
 * follow-up fixes for the WoWLAN security issue, to fix a
   partial TKIP key material problem and to use crypto_memneq()
 * a change for better enforcement of FQ's memory limit
 * a disconnect/connect handling fix, and
 * a user rate mask validation fix
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoi40e: Add programming descriptors to cleaned_count
Alexander Duyck [Sun, 22 Oct 2017 01:12:29 +0000 (18:12 -0700)]
i40e: Add programming descriptors to cleaned_count

This patch updates the i40e driver to include programming descriptors in
the cleaned_count. Without this change it becomes possible for us to leak
memory as we don't trigger a large enough allocation when the time comes to
allocate new buffers and we end up overwriting a number of rx_buffers equal
to the number of programming descriptors we encountered.

Fixes: 0e626ff7ccbf ("i40e: Fix support for flow director programming status")
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Anders K. Pedersen <akp@cohaesio.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoi40e: Fix incorrect use of tx_itr_setting when checking for Rx ITR setup
Alexander Duyck [Fri, 20 Oct 2017 20:59:20 +0000 (13:59 -0700)]
i40e: Fix incorrect use of tx_itr_setting when checking for Rx ITR setup

It looks like there was either a copy/paste error or just a typo that
resulted in the Tx ITR setting being used to determine if we were using
adaptive Rx interrupt moderation or not.

This patch fixes the typo.

Fixes: 65e87c0398f5 ("i40evf: support queue-specific settings for interrupt moderation")
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: Fix Tx map failure path
Alexander Duyck [Thu, 19 Oct 2017 21:07:13 +0000 (17:07 -0400)]
ixgbe: Fix Tx map failure path

This patch is a partial revert of "ixgbe: Don't bother clearing buffer
memory for descriptor rings". Specifically I messed up the exception
handling path a bit and this resulted in us incorrectly adding the count
back in when we didn't need to.

In order to make this simpler I am reverting most of the exception handling
path change and instead just replacing the bit that was handled by the
unmap_and_free call.

Fixes: ffed21bcee7a ("ixgbe: Don't bother clearing buffer memory for descriptor rings")
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoigb: Fix TX map failure path
Jean-Philippe Brucker [Thu, 19 Oct 2017 19:07:36 +0000 (20:07 +0100)]
igb: Fix TX map failure path

When the driver cannot map a TX buffer, instead of rolling back
gracefully and retrying later, we currently get a panic:

[  159.885994] igb 0000:00:00.0: TX DMA map failed
[  159.886588] Unable to handle kernel paging request at virtual address ffff00000a08c7a8
               ...
[  159.897031] PC is at igb_xmit_frame_ring+0x9c8/0xcb8

Fix the erroneous test that leads to this situation.

Signed-off-by: Jean-Philippe Brucker <jean-philippe.brucker@arm.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoe1000: avoid null pointer dereference on invalid stat type
Colin Ian King [Fri, 22 Sep 2017 17:13:48 +0000 (18:13 +0100)]
e1000: avoid null pointer dereference on invalid stat type

Currently if the stat type is invalid then data[i] is being set
either by dereferencing a null pointer p, or it is reading from
an incorrect previous location if we had a valid stat type
previously.  Fix this by skipping over the read of p on an invalid
stat type.

Detected by CoverityScan, CID#113385 ("Explicit null dereferenced")

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoe1000: fix race condition between e1000_down() and e1000_watchdog
Vincenzo Maffione [Sat, 16 Sep 2017 16:00:00 +0000 (18:00 +0200)]
e1000: fix race condition between e1000_down() and e1000_watchdog

This patch fixes a race condition that can result into the interface being
up and carrier on, but with transmits disabled in the hardware.
The bug may show up by repeatedly IFF_DOWN+IFF_UP the interface, which
allows e1000_watchdog() interleave with e1000_down().

    CPU x                           CPU y
    --------------------------------------------------------------------
    e1000_down():
        netif_carrier_off()
                                    e1000_watchdog():
                                        if (carrier == off) {
                                            netif_carrier_on();
                                            enable_hw_transmit();
                                        }
        disable_hw_transmit();
                                    e1000_watchdog():
                                        /* carrier on, do nothing */

Signed-off-by: Vincenzo Maffione <v.maffione@gmail.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agotap: double-free in error path in tap_open()
Girish Moodalbail [Wed, 25 Oct 2017 07:23:04 +0000 (00:23 -0700)]
tap: double-free in error path in tap_open()

Double free of skb_array in tap module is causing kernel panic. When
tap_set_queue() fails we free skb_array right away by calling
skb_array_cleanup(). However, later on skb_array_cleanup() is called
again by tap_sock_destruct through sock_put(). This patch fixes that
issue.

Fixes: 362899b8725b35e3 (macvtap: switch to use skb array)
Signed-off-by: Girish Moodalbail <girish.moodalbail@oracle.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: call tcp_rate_skb_sent() when retransmit with unaligned skb->data
Yousuk Seung [Tue, 24 Oct 2017 23:44:42 +0000 (16:44 -0700)]
tcp: call tcp_rate_skb_sent() when retransmit with unaligned skb->data

Current implementation calls tcp_rate_skb_sent() when tcp_transmit_skb()
is called when it clones skb only. Not calling tcp_rate_skb_sent() is OK
for all such code paths except from __tcp_retransmit_skb() which happens
when skb->data address is not aligned. This may rarely happen e.g. when
small amount of data is sent initially and the receiver partially acks
odd number of bytes for some reason, possibly malicious.

Signed-off-by: Yousuk Seung <ysseung@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp/dccp: fix other lockdep splats accessing ireq_opt
Eric Dumazet [Tue, 24 Oct 2017 15:20:31 +0000 (08:20 -0700)]
tcp/dccp: fix other lockdep splats accessing ireq_opt

In my first attempt to fix the lockdep splat, I forgot we could
enter inet_csk_route_req() with a freshly allocated request socket,
for which refcount has not yet been elevated, due to complex
SLAB_TYPESAFE_BY_RCU rules.

We either are in rcu_read_lock() section _or_ we own a refcount on the
request.

Correct RCU verb to use here is rcu_dereference_check(), although it is
not possible to prove we actually own a reference on a shared
refcount :/

In v2, I added ireq_opt_deref() helper and use in three places, to fix other
possible splats.

[   49.844590]  lockdep_rcu_suspicious+0xea/0xf3
[   49.846487]  inet_csk_route_req+0x53/0x14d
[   49.848334]  tcp_v4_route_req+0xe/0x10
[   49.850174]  tcp_conn_request+0x31c/0x6a0
[   49.851992]  ? __lock_acquire+0x614/0x822
[   49.854015]  tcp_v4_conn_request+0x5a/0x79
[   49.855957]  ? tcp_v4_conn_request+0x5a/0x79
[   49.858052]  tcp_rcv_state_process+0x98/0xdcc
[   49.859990]  ? sk_filter_trim_cap+0x2f6/0x307
[   49.862085]  tcp_v4_do_rcv+0xfc/0x145
[   49.864055]  ? tcp_v4_do_rcv+0xfc/0x145
[   49.866173]  tcp_v4_rcv+0x5ab/0xaf9
[   49.868029]  ip_local_deliver_finish+0x1af/0x2e7
[   49.870064]  ip_local_deliver+0x1b2/0x1c5
[   49.871775]  ? inet_del_offload+0x45/0x45
[   49.873916]  ip_rcv_finish+0x3f7/0x471
[   49.875476]  ip_rcv+0x3f1/0x42f
[   49.876991]  ? ip_local_deliver_finish+0x2e7/0x2e7
[   49.878791]  __netif_receive_skb_core+0x6d3/0x950
[   49.880701]  ? process_backlog+0x7e/0x216
[   49.882589]  __netif_receive_skb+0x1d/0x5e
[   49.884122]  process_backlog+0x10c/0x216
[   49.885812]  net_rx_action+0x147/0x3df

Fixes: a6ca7abe53633 ("tcp/dccp: fix lockdep splat in inet_csk_route_req()")
Fixes: c92e8c02fe66 ("tcp/dccp: fix ireq->opt races")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: kernel test robot <fengguang.wu@intel.com>
Reported-by: Maciej Żenczykowski <maze@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agords: Fix inaccurate accounting of unsignaled wrs
Håkon Bugge [Tue, 24 Oct 2017 14:16:28 +0000 (16:16 +0200)]
rds: Fix inaccurate accounting of unsignaled wrs

The number of unsignaled work-requests posted to the IB send queue is
tracked by a counter in the rds_ib_connection struct. When it reaches
zero, or the caller explicitly asks for it, the send-signaled bit is
set in send_flags and the counter is reset. This is performed by the
rds_ib_set_wr_signal_state() function.

However, this function is not always used which yields inaccurate
accounting. This commit fixes this, re-factors a code bloat related to
the matter, and makes the actual parameter type to the function
consistent.

Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agords: ib: Fix uninitialized variable
Håkon Bugge [Tue, 24 Oct 2017 16:17:18 +0000 (18:17 +0200)]
rds: ib: Fix uninitialized variable

send_flags needs to be initialized before calling
rds_ib_set_wr_signal_state().

Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge tag 'linux-can-fixes-for-4.14-20171024' of git://git.kernel.org/pub/scm/linux...
David S. Miller [Thu, 26 Oct 2017 08:17:45 +0000 (17:17 +0900)]
Merge tag 'linux-can-fixes-for-4.14-20171024' of git://git./linux/kernel/git/mkl/linux-can

Marc Kleine-Budde says:

====================
pull-request: can 2017-10-24

here's another pull request for net/master.

The patch by Gerhard Bertelsmann fixes the CAN_CTRLMODE_LOOPBACK in the
sun4i driver. Two patches by Jimmy Assarsson for the kvaser_usb driver
fix a print in the error path of the kvaser_usb_close() and remove a
wrong warning message with the Leaf v2 firmware version v4.1.844.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: mvpp2: do not sleep in set_rx_mode
Antoine Tenart [Tue, 24 Oct 2017 09:41:28 +0000 (11:41 +0200)]
net: mvpp2: do not sleep in set_rx_mode

This patch replaces GFP_KERNEL by GFP_ATOMIC to avoid sleeping in the
ndo_set_rx_mode() call which is called with BH disabled.

Fixes: 3f518509dedc ("ethernet: Add new driver for Marvell Armada 375 network unit")
Signed-off-by: Antoine Tenart <antoine.tenart@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: mvpp2: fix invalid parameters order when calling the tcam init
Antoine Tenart [Tue, 24 Oct 2017 09:41:27 +0000 (11:41 +0200)]
net: mvpp2: fix invalid parameters order when calling the tcam init

When calling mvpp2_prs_mac_multi_set() from mvpp2_prs_mac_init(), two
parameters (the port index and the table index) are inverted. Fixes
this.

Fixes: 3f518509dedc ("ethernet: Add new driver for Marvell Armada 375 network unit")
Signed-off-by: Antoine Tenart <antoine.tenart@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: mvpp2: fix typo in the tcam setup
Antoine Tenart [Tue, 24 Oct 2017 09:41:26 +0000 (11:41 +0200)]
net: mvpp2: fix typo in the tcam setup

This patch fixes a typo in the mvpp2_prs_tcam_data_cmp() function, as
the shift value is inverted with the data.

Fixes: 3f518509dedc ("ethernet: Add new driver for Marvell Armada 375 network unit")
Signed-off-by: Antoine Tenart <antoine.tenart@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet/mlx5e: DCBNL, Implement tc with ets type and zero bandwidth
Huy Nguyen [Thu, 28 Sep 2017 20:33:50 +0000 (15:33 -0500)]
net/mlx5e: DCBNL, Implement tc with ets type and zero bandwidth

Previously, tc with ets type and zero bandwidth is not accepted
by driver. This behavior does not follow the IEEE802.1qaz spec.

If there are tcs with ets type and zero bandwidth, these tcs are
assigned to the lowest priority tc_group #0. We equally distribute
100% bw of the tc_group #0 to these zero bandwidth ets tcs.
Also, the non zero bandwidth ets tcs are assigned to tc_group #1.

If there is no zero bandwidth ets tc, the non zero bandwidth ets tcs
are assigned to tc_group #0.

Fixes: cdcf11212b22 ("net/mlx5e: Validate BW weight values of ETS")
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
7 years agonet/mlx5e: Properly deal with encap flows add/del under neigh update
Or Gerlitz [Tue, 17 Oct 2017 10:33:43 +0000 (12:33 +0200)]
net/mlx5e: Properly deal with encap flows add/del under neigh update

Currently, the encap action offload is handled in the actions parse
function and not in mlx5e_tc_add_fdb_flow() where we deal with all
the other aspects of offloading actions (vlan, modify header) and
the rule itself.

When the neigh update code (mlx5e_tc_encap_flows_add()) recreates the
encap entry and offloads the related flows, we wrongly call again into
mlx5e_tc_add_fdb_flow(), this for itself would cause us to handle
again the offloading of vlans and header re-write which puts things
in non consistent state and step on freed memory (e.g the modify
header parse buffer which is already freed).

Since on error, mlx5e_tc_add_fdb_flow() detaches and may release the
encap entry, it causes a corruption at the neigh update code which goes
over the list of flows associated with this encap entry, or double free
when the tc flow is later deleted by user-space.

When neigh update (mlx5e_tc_encap_flows_del()) unoffloads the flows related
to an encap entry which is now invalid, we do a partial repeat of the eswitch
flow removal code which is wrong too.

To fix things up we do the following:

(1) handle the encap action offload in the eswitch flow add function
    mlx5e_tc_add_fdb_flow() as done for the other actions and the rule itself.

(2) modify the neigh update code (mlx5e_tc_encap_flows_add/del) to only
    deal with the encap entry and rules delete/add and not with any of
    the other offloaded actions.

Fixes: 232c001398ae ('net/mlx5e: Add support to neighbour update flow')
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Paul Blakey <paulb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
7 years agonet/mlx5: Delay events till mlx5 interface's add complete for pci resume
Huy Nguyen [Wed, 4 Oct 2017 22:58:21 +0000 (17:58 -0500)]
net/mlx5: Delay events till mlx5 interface's add complete for pci resume

mlx5_ib_add is called during mlx5_pci_resume after a pci error.
Before mlx5_ib_add completes, there are multiple events which trigger
function mlx5_ib_event. This cause kernel panic because mlx5_ib_event
accesses unitialized resources.

The fix is to extend Erez Shitrit's patch <97834eba7c19>
("net/mlx5: Delay events till ib registration ends") to cover
the pci resume code path.

Trace:
mlx5_core 0001:01:00.6: mlx5_pci_resume was called
mlx5_core 0001:01:00.6: firmware version: 16.20.1011
mlx5_core 0001:01:00.6: mlx5_attach_interface:164:(pid 779):
mlx5_ib_event:2996:(pid 34777): warning: event on port 1
mlx5_ib_event:2996:(pid 34782): warning: event on port 1
Unable to handle kernel paging request for data at address 0x0001c104
Faulting instruction address: 0xd000000008f411fc
Oops: Kernel access of bad area, sig: 11 [#1]
...
...
Call Trace:
[c000000fff77bb70] [d000000008f4119c] mlx5_ib_event+0x64/0x470 [mlx5_ib] (unreliable)
[c000000fff77bc60] [d000000008e67130] mlx5_core_event+0xb8/0x210 [mlx5_core]
[c000000fff77bd10] [d000000008e4bd00] mlx5_eq_int+0x528/0x860[mlx5_core]

Fixes: 97834eba7c19 ("net/mlx5: Delay events till ib registration ends")
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Reviewed-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
7 years agonet/mlx5: Fix health work queue spin lock to IRQ safe
Moshe Shemesh [Thu, 19 Oct 2017 11:14:29 +0000 (14:14 +0300)]
net/mlx5: Fix health work queue spin lock to IRQ safe

spin_lock/unlock of health->wq_lock should be IRQ safe.
It was changed to spin_lock_irqsave since adding commit 0179720d6be2
("net/mlx5: Introduce trigger_health_work function") which uses
spin_lock from asynchronous event (IRQ) context.
Thus, all spin_lock/unlock of health->wq_lock should have been moved
to IRQ safe mode.
However, one occurrence on new code using this lock missed that
change, resulting in possible deadlock:
  kernel: Possible unsafe locking scenario:
  kernel:       CPU0
  kernel:       ----
  kernel:  lock(&(&health->wq_lock)->rlock);
  kernel:  <Interrupt>
  kernel:    lock(&(&health->wq_lock)->rlock);
  kernel: #012 *** DEADLOCK ***

Fixes: 2a0165a034ac ("net/mlx5: Cancel delayed recovery work when unloading the driver")
Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
7 years agotun: allow positive return values on dev_get_valid_name() call
Julien Gomes [Wed, 25 Oct 2017 18:50:50 +0000 (11:50 -0700)]
tun: allow positive return values on dev_get_valid_name() call

If the name argument of dev_get_valid_name() contains "%d", it will try
to assign it a unit number in __dev__alloc_name() and return either the
unit number (>= 0) or an error code (< 0).
Considering positive values as error values prevent tun device creations
relying this mechanism, therefor we should only consider negative values
as errors here.

Signed-off-by: Julien Gomes <julien@arista.com>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: refuse offloading filters that redirects to upper devices
Pieter Jansen van Vuuren [Wed, 25 Oct 2017 18:47:05 +0000 (11:47 -0700)]
nfp: refuse offloading filters that redirects to upper devices

Previously we did not ensure that a netdev is a representative netdev
before dereferencing its private data. This can occur when an upper netdev
is created on a representative netdev. This patch corrects this by first
ensuring that the netdev is a representative netdev before using it.
Checking only switchdev_port_same_parent_id is not sufficient to ensure
that we can safely use the netdev. Failing to check that the netdev is also
a representative netdev would result in incorrect dereferencing.

Fixes: 1a1e586f54bf ("nfp: add basic action capabilities to flower offloads")
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet/unix: don't show information about sockets from other namespaces
Andrei Vagin [Wed, 25 Oct 2017 17:16:42 +0000 (10:16 -0700)]
net/unix: don't show information about sockets from other namespaces

socket_diag shows information only about sockets from a namespace where
a diag socket lives.

But if we request information about one unix socket, the kernel don't
check that its netns is matched with a diag socket namespace, so any
user can get information about any unix socket in a system. This looks
like a bug.

v2: add a Fixes tag

Fixes: 51d7cccf0723 ("net: make sock diag per-namespace")
Signed-off-by: Andrei Vagin <avagin@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomac80211: don't compare TKIP TX MIC key in reinstall prevention
Johannes Berg [Tue, 24 Oct 2017 19:12:13 +0000 (21:12 +0200)]
mac80211: don't compare TKIP TX MIC key in reinstall prevention

For the reinstall prevention, the code I had added compares the
whole key. It turns out though that iwlwifi firmware doesn't
provide the TKIP TX MIC key as it's not needed in client mode,
and thus the comparison will always return false.

For client mode, thus always zero out the TX MIC key part before
doing the comparison in order to avoid accepting the reinstall
of the key with identical encryption and RX MIC key, but not the
same TX MIC key (since the supplicant provides the real one.)

Fixes: fdf7cb4185b6 ("mac80211: accept key reinstall without changing anything")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
7 years agonet: dsa: check master device before put
Vivien Didelot [Tue, 24 Oct 2017 20:37:19 +0000 (16:37 -0400)]
net: dsa: check master device before put

In the case of pdata, the dsa_cpu_parse function calls dev_put() before
making sure it isn't NULL. Fix this.

Fixes: 71e0bbde0d88 ("net: dsa: Add support for platform data")
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agostrparser: Use delayed work instead of timer for msg timeout
Tom Herbert [Fri, 20 Oct 2017 23:40:43 +0000 (16:40 -0700)]
strparser: Use delayed work instead of timer for msg timeout

Sock lock may be taken in the message timer function which is a
problem since timers run in BH. Instead of timers use delayed_work.

Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
Fixes: bbb03029a899 ("strparser: Generalize strparser")
Signed-off-by: Tom Herbert <tom@quantonium.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec
David S. Miller [Tue, 24 Oct 2017 11:17:38 +0000 (20:17 +0900)]
Merge branch 'master' of git://git./linux/kernel/git/klassert/ipsec

Steffen Klassert says:

====================
pull request (net): ipsec 2017-10-24

1) Fix a memleak when we don't find a inner_mode
   during bundle creation. From David Miller.

2) Fix a xfrm policy dump crash. We may crash
   on error when dumping policies via netlink.
   Fix this by initializing the policy walk
   with the cb->start method. This fix is a
   serious stable candidate. From Herbert Xu.

Please pull or let me know if there are problems.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agocan: kvaser_usb: Ignore CMD_FLUSH_QUEUE_REPLY messages
Jimmy Assarsson [Tue, 24 Oct 2017 10:23:29 +0000 (12:23 +0200)]
can: kvaser_usb: Ignore CMD_FLUSH_QUEUE_REPLY messages

To avoid kernel warning "Unhandled message (68)", ignore the
CMD_FLUSH_QUEUE_REPLY message for now.

As of Leaf v2 firmware version v4.1.844 (2017-02-15), flush tx queue is
synchronous. There is a capability bit indicating whether flushing tx
queue is synchronous or asynchronous.

A proper solution would be to query the device for capabilities. If the
synchronous tx flush capability bit is set, we should wait for
CMD_FLUSH_QUEUE_REPLY message, while flushing the tx queue.

Signed-off-by: Jimmy Assarsson <jimmyassarsson@gmail.com>
Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
7 years agocan: kvaser_usb: Correct return value in printout
Jimmy Assarsson [Tue, 24 Oct 2017 10:23:28 +0000 (12:23 +0200)]
can: kvaser_usb: Correct return value in printout

If the return value from kvaser_usb_send_simple_msg() was non-zero, the
return value from kvaser_usb_flush_queue() was printed in the kernel
warning.

Signed-off-by: Jimmy Assarsson <jimmyassarsson@gmail.com>
Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
7 years agocan: sun4i: fix loopback mode
Gerhard Bertelsmann [Thu, 17 Aug 2017 13:59:49 +0000 (15:59 +0200)]
can: sun4i: fix loopback mode

Fix loopback mode by setting the right flag and remove presume mode.

Signed-off-by: Gerhard Bertelsmann <info@gerhard-bertelsmann.de>
Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
7 years agosctp: full support for ipv6 ip_nonlocal_bind & IP_FREEBIND
Laszlo Toth [Mon, 23 Oct 2017 17:19:33 +0000 (19:19 +0200)]
sctp: full support for ipv6 ip_nonlocal_bind & IP_FREEBIND

Commit 9b9742022888 ("sctp: support ipv6 nonlocal bind")
introduced support for the above options as v4 sctp did,
so patched sctp_v6_available().

In the v4 implementation it's enough, because
sctp_inet_bind_verify() just returns with sctp_v4_available().
However sctp_inet6_bind_verify() has an extra check before that
for link-local scope_id, which won't respect the above options.

Added the checks before calling ipv6_chk_addr(), but
not before the validation of scope_id.

before (w/ both options):
 ./v6test fe80::10 sctp
 bind failed, errno: 99 (Cannot assign requested address)
 ./v6test fe80::10 tcp
 bind success, errno: 0 (Success)

after (w/ both options):
 ./v6test fe80::10 sctp
 bind success, errno: 0 (Success)

Signed-off-by: Laszlo Toth <laszlth@gmail.com>
Reviewed-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: mvpp2: do not call txq_done from the Tx path when Tx irqs are used
Antoine Tenart [Mon, 23 Oct 2017 13:24:31 +0000 (15:24 +0200)]
net: mvpp2: do not call txq_done from the Tx path when Tx irqs are used

When Tx IRQs are used, txq_bufs_free() can be called from both the Tx
path and from NAPI poll(). This led to CPU stalls as if these two tasks
(Tx and Poll) are scheduled on two CPUs at the same time, DMA unmapping
operations are done on the same txq buffers.

This patch adds a check not to call txq_done() from the Tx path if Tx
interrupts are used as it does not make sense to do so.

Fixes: edc660fa09e2 ("net: mvpp2: replace TX coalescing interrupts with hrtimer")
Signed-off-by: Antoine Tenart <antoine.tenart@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: mvpp2: do not unmap TSO headers buffers
Antoine Tenart [Mon, 23 Oct 2017 13:24:30 +0000 (15:24 +0200)]
net: mvpp2: do not unmap TSO headers buffers

The TSO header buffers are coming from a per cpu pool and should not
be unmapped as they are reused. The PPv2 driver was unmapping all
descriptors buffers unconditionally. This patch fixes this by checking
the buffers dma addresses before unmapping them, and by not unmapping
those who are located in the TSO header pool.

Fixes: 186cd4d4e414 ("net: mvpp2: software tso support")
Signed-off-by: Antoine Tenart <antoine.tenart@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: mvpp2: fix TSO headers allocation and management
Yan Markman [Mon, 23 Oct 2017 13:24:29 +0000 (15:24 +0200)]
net: mvpp2: fix TSO headers allocation and management

TSO headers are managed with txq index and therefore should be aligned
with the txq size, not with the aggregated txq size.

Fixes: 186cd4d4e414 ("net: mvpp2: software tso support")
Reported-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Yan Markman <ymarkman@marvell.com>
Signed-off-by: Antoine Tenart <antoine.tenart@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agocdc_ether: flag the Huawei ME906/ME909 as WWAN
Aleksander Morgado [Mon, 23 Oct 2017 15:16:41 +0000 (17:16 +0200)]
cdc_ether: flag the Huawei ME906/ME909 as WWAN

The Huawei ME906 (12d1:15c1) comes with a standard ECM interface that
requires management via AT commands sent over one of the control TTYs
(e.g. connected with AT^NDISDUP).

Signed-off-by: Aleksander Morgado <aleksander@aleksander.es>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodrivers/net/usb: add device id for TP-LINK UE300 USB 3.0 Ethernet
Ran Wang [Mon, 23 Oct 2017 10:10:23 +0000 (18:10 +0800)]
drivers/net/usb: add device id for TP-LINK UE300 USB 3.0 Ethernet

This product is named 'TP-LINK USB 3.0 Gigabit Ethernet Network
Adapter (Model No.is UE300)'. It uses chip RTL8153 and works with
driver drivers/net/usb/r8152.c

Signed-off-by: Ran Wang <ran.wang_1@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoipsec: Fix aborted xfrm policy dump crash
Herbert Xu [Thu, 19 Oct 2017 12:51:10 +0000 (20:51 +0800)]
ipsec: Fix aborted xfrm policy dump crash

An independent security researcher, Mohamed Ghannam, has reported
this vulnerability to Beyond Security's SecuriTeam Secure Disclosure
program.

The xfrm_dump_policy_done function expects xfrm_dump_policy to
have been called at least once or it will crash.  This can be
triggered if a dump fails because the target socket's receive
buffer is full.

This patch fixes it by using the cb->start mechanism to ensure that
the initialisation is always done regardless of the buffer situation.

Fixes: 12a169e7d8f4 ("ipsec: Put dumpers on the dump list")
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
7 years agotcp/dccp: fix lockdep splat in inet_csk_route_req()
Eric Dumazet [Sun, 22 Oct 2017 19:33:57 +0000 (12:33 -0700)]
tcp/dccp: fix lockdep splat in inet_csk_route_req()

This patch fixes the following lockdep splat in inet_csk_route_req()

  lockdep_rcu_suspicious
  inet_csk_route_req
  tcp_v4_send_synack
  tcp_rtx_synack
  inet_rtx_syn_ack
  tcp_fastopen_synack_time
  tcp_retransmit_timer
  tcp_write_timer_handler
  tcp_write_timer
  call_timer_fn

Thread running inet_csk_route_req() owns a reference on the request
socket, so we have the guarantee ireq->ireq_opt wont be changed or
freed.

lockdep can enforce this invariant for us.

Fixes: c92e8c02fe66 ("tcp/dccp: fix ireq->opt races")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: do tcp_mstamp_refresh before retransmits on TSQ handler
Koichiro Den [Sun, 22 Oct 2017 04:13:16 +0000 (13:13 +0900)]
tcp: do tcp_mstamp_refresh before retransmits on TSQ handler

When retransmission on TSQ handler was introduced in the commit
f9616c35a0d7 ("tcp: implement TSQ for retransmits"), the retransmitted
skbs' timestamps were updated on the actual transmission. In the later
commit 385e20706fac ("tcp: use tp->tcp_mstamp in output path"), it stops
being done so. In the commit, the comment says "We try to refresh
tp->tcp_mstamp only when necessary", and at present tcp_tsq_handler and
tcp_v4_mtu_reduced applies to this. About the latter, it's okay since
it's rare enough.

About the former, even though possible retransmissions on the tasklet
comes just after the destructor run in NET_RX softirq handling, the time
between them could be nonnegligibly large to the extent that
tcp_rack_advance or rto rearming be affected if other (remaining) RX,
BLOCK and (preceding) TASKLET sofirq handlings are unexpectedly heavy.

So in the same way as tcp_write_timer_handler does, doing tcp_mstamp_refresh
ensures the accuracy of algorithms relying on it.

Fixes: 385e20706fac ("tcp: use tp->tcp_mstamp in output path")
Signed-off-by: Koichiro Den <den@klaipeden.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Sun, 22 Oct 2017 02:44:48 +0000 (22:44 -0400)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:
 "A little more than usual this time around. Been travelling, so that is
  part of it.

  Anyways, here are the highlights:

   1) Deal with memcontrol races wrt. listener dismantle, from Eric
      Dumazet.

   2) Handle page allocation failures properly in nfp driver, from Jaku
      Kicinski.

   3) Fix memory leaks in macsec, from Sabrina Dubroca.

   4) Fix crashes in pppol2tp_session_ioctl(), from Guillaume Nault.

   5) Several fixes in bnxt_en driver, including preventing potential
      NVRAM parameter corruption from Michael Chan.

   6) Fix for KRACK attacks in wireless, from Johannes Berg.

   7) rtnetlink event generation fixes from Xin Long.

   8) Deadlock in mlxsw driver, from Ido Schimmel.

   9) Disallow arithmetic operations on context pointers in bpf, from
      Jakub Kicinski.

  10) Missing sock_owned_by_user() check in sctp_icmp_redirect(), from
      Xin Long.

  11) Only TCP is supported for sockmap, make that explicit with a
      check, from John Fastabend.

  12) Fix IP options state races in DCCP and TCP, from Eric Dumazet.

  13) Fix panic in packet_getsockopt(), also from Eric Dumazet.

  14) Add missing locked in hv_sock layer, from Dexuan Cui.

  15) Various aquantia bug fixes, including several statistics handling
      cures. From Igor Russkikh et al.

  16) Fix arithmetic overflow in devmap code, from John Fastabend.

  17) Fix busted socket memory accounting when we get a fault in the tcp
      zero copy paths. From Willem de Bruijn.

  18) Don't leave opt->tot_len uninitialized in ipv6, from Eric Dumazet"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (106 commits)
  stmmac: Don't access tx_q->dirty_tx before netif_tx_lock
  ipv6: flowlabel: do not leave opt->tot_len with garbage
  of_mdio: Fix broken PHY IRQ in case of probe deferral
  textsearch: fix typos in library helpers
  rxrpc: Don't release call mutex on error pointer
  net: stmmac: Prevent infinite loop in get_rx_timestamp_status()
  net: stmmac: Fix stmmac_get_rx_hwtstamp()
  net: stmmac: Add missing call to dev_kfree_skb()
  mlxsw: spectrum_router: Configure TIGCR on init
  mlxsw: reg: Add Tunneling IPinIP General Configuration Register
  net: ethtool: remove error check for legacy setting transceiver type
  soreuseport: fix initialization race
  net: bridge: fix returning of vlan range op errors
  sock: correct sk_wmem_queued accounting on efault in tcp zerocopy
  bpf: add test cases to bpf selftests to cover all access tests
  bpf: fix pattern matches for direct packet access
  bpf: fix off by one for range markings with L{T, E} patterns
  bpf: devmap fix arithmetic overflow in bitmap_size calculation
  net: aquantia: Bad udp rate on default interrupt coalescing
  net: aquantia: Enable coalescing management via ethtool interface
  ...

7 years agostmmac: Don't access tx_q->dirty_tx before netif_tx_lock
Bernd Edlinger [Sat, 21 Oct 2017 06:51:30 +0000 (06:51 +0000)]
stmmac: Don't access tx_q->dirty_tx before netif_tx_lock

This is the possible reason for different hard to reproduce
problems on my ARMv7-SMP test system.

The symptoms are in recent kernels imprecise external aborts,
and in older kernels various kinds of network stalls and
unexpected page allocation failures.

My testing indicates that the trouble started between v4.5 and v4.6
and prevails up to v4.14.

Using the dirty_tx before acquiring the spin lock is clearly
wrong and was first introduced with v4.6.

Fixes: e3ad57c96715 ("stmmac: review RX/TX ring management")
Signed-off-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoipv6: flowlabel: do not leave opt->tot_len with garbage
Eric Dumazet [Sat, 21 Oct 2017 19:26:23 +0000 (12:26 -0700)]
ipv6: flowlabel: do not leave opt->tot_len with garbage

When syzkaller team brought us a C repro for the crash [1] that
had been reported many times in the past, I finally could find
the root cause.

If FlowLabel info is merged by fl6_merge_options(), we leave
part of the opt_space storage provided by udp/raw/l2tp with random value
in opt_space.tot_len, unless a control message was provided at sendmsg()
time.

Then ip6_setup_cork() would use this random value to perform a kzalloc()
call. Undefined behavior and crashes.

Fix is to properly set tot_len in fl6_merge_options()

At the same time, we can also avoid consuming memory and cpu cycles
to clear it, if every option is copied via a kmemdup(). This is the
change in ip6_setup_cork().

[1]
kasan: CONFIG_KASAN_INLINE enabled
kasan: GPF could be caused by NULL-ptr deref or user memory access
general protection fault: 0000 [#1] SMP KASAN
Dumping ftrace buffer:
   (ftrace buffer empty)
Modules linked in:
CPU: 0 PID: 6613 Comm: syz-executor0 Not tainted 4.14.0-rc4+ #127
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
task: ffff8801cb64a100 task.stack: ffff8801cc350000
RIP: 0010:ip6_setup_cork+0x274/0x15c0 net/ipv6/ip6_output.c:1168
RSP: 0018:ffff8801cc357550 EFLAGS: 00010203
RAX: dffffc0000000000 RBX: ffff8801cc357748 RCX: 0000000000000010
RDX: 0000000000000002 RSI: ffffffff842bd1d9 RDI: 0000000000000014
RBP: ffff8801cc357620 R08: ffff8801cb17f380 R09: ffff8801cc357b10
R10: ffff8801cb64a100 R11: 0000000000000000 R12: ffff8801cc357ab0
R13: ffff8801cc357b10 R14: 0000000000000000 R15: ffff8801c3bbf0c0
FS:  00007f9c5c459700(0000) GS:ffff8801db200000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000020324000 CR3: 00000001d1cf2000 CR4: 00000000001406f0
DR0: 0000000020001010 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000600
Call Trace:
 ip6_make_skb+0x282/0x530 net/ipv6/ip6_output.c:1729
 udpv6_sendmsg+0x2769/0x3380 net/ipv6/udp.c:1340
 inet_sendmsg+0x11f/0x5e0 net/ipv4/af_inet.c:762
 sock_sendmsg_nosec net/socket.c:633 [inline]
 sock_sendmsg+0xca/0x110 net/socket.c:643
 SYSC_sendto+0x358/0x5a0 net/socket.c:1750
 SyS_sendto+0x40/0x50 net/socket.c:1718
 entry_SYSCALL_64_fastpath+0x1f/0xbe
RIP: 0033:0x4520a9
RSP: 002b:00007f9c5c458c08 EFLAGS: 00000216 ORIG_RAX: 000000000000002c
RAX: ffffffffffffffda RBX: 0000000000718000 RCX: 00000000004520a9
RDX: 0000000000000001 RSI: 0000000020fd1000 RDI: 0000000000000016
RBP: 0000000000000086 R08: 0000000020e0afe4 R09: 000000000000001c
R10: 0000000000000000 R11: 0000000000000216 R12: 00000000004bb1ee
R13: 00000000ffffffff R14: 0000000000000016 R15: 0000000000000029
Code: e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85 ea 0f 00 00 48 8d 79 04 48 b8 00 00 00 00 00 fc ff df 45 8b 74 24 04 48 89 fa 48 c1 ea 03 <0f> b6 14 02 48 89 f8 83 e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85
RIP: ip6_setup_cork+0x274/0x15c0 net/ipv6/ip6_output.c:1168 RSP: ffff8801cc357550

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoof_mdio: Fix broken PHY IRQ in case of probe deferral
Geert Uytterhoeven [Wed, 18 Oct 2017 11:54:03 +0000 (13:54 +0200)]
of_mdio: Fix broken PHY IRQ in case of probe deferral

If an Ethernet PHY is initialized before the interrupt controller it is
connected to, a message like the following is printed:

    irq: no irq domain found for /interrupt-controller@e61c0000 !

However, the actual error is ignored, leading to a non-functional (POLL)
PHY interrupt later:

    Micrel KSZ8041RNLI ee700000.ethernet-ffffffff:01: attached PHY driver [Micrel KSZ8041RNLI] (mii_bus:phy_addr=ee700000.ethernet-ffffffff:01, irq=POLL)

Depending on whether the PHY driver will fall back to polling, Ethernet
may or may not work.

To fix this:
  1. Switch of_mdiobus_register_phy() from irq_of_parse_and_map() to
     of_irq_get().
     Unlike the former, the latter returns -EPROBE_DEFER if the
     interrupt controller is not yet available, so this condition can be
     detected.
     Other errors are handled the same as before, i.e. use the passed
     mdio->irq[addr] as interrupt.
  2. Propagate and handle errors from of_mdiobus_register_phy() and
     of_mdiobus_register_device().

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotextsearch: fix typos in library helpers
Randy Dunlap [Fri, 20 Oct 2017 19:15:52 +0000 (12:15 -0700)]
textsearch: fix typos in library helpers

Fix spellos (typos) in textsearch library helpers.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorxrpc: Don't release call mutex on error pointer
David Howells [Fri, 20 Oct 2017 16:01:22 +0000 (17:01 +0100)]
rxrpc: Don't release call mutex on error pointer

Don't release call mutex at the end of rxrpc_kernel_begin_call() if the
call pointer actually holds an error value.

Fixes: 540b1c48c37a ("rxrpc: Fix deadlock between call creation and sendmsg/recvmsg")
Reported-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'stmmac-hw-tstamp-fixes'
David S. Miller [Sun, 22 Oct 2017 01:50:40 +0000 (02:50 +0100)]
Merge branch 'stmmac-hw-tstamp-fixes'

Jose Abreu says:

====================
net: stmmac: Fix HW timestamping

Three fixes for HW timestamping feature, all of them for RX side.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: Prevent infinite loop in get_rx_timestamp_status()
Jose Abreu [Fri, 20 Oct 2017 13:37:36 +0000 (14:37 +0100)]
net: stmmac: Prevent infinite loop in get_rx_timestamp_status()

Prevent infinite loop by correctly setting the loop condition to
break when i == 10.

Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: Fix stmmac_get_rx_hwtstamp()
Jose Abreu [Fri, 20 Oct 2017 13:37:35 +0000 (14:37 +0100)]
net: stmmac: Fix stmmac_get_rx_hwtstamp()

When using GMAC4 the valid timestamp is from CTX next desc but
we are passing the previous desc to get_rx_timestamp_status()
callback.

Fix this and while at it rework a little bit the function logic.

Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: Add missing call to dev_kfree_skb()
Jose Abreu [Fri, 20 Oct 2017 13:37:34 +0000 (14:37 +0100)]
net: stmmac: Add missing call to dev_kfree_skb()

When RX HW timestamp is enabled and a frame is discarded we are
not freeing the skb but instead only setting to NULL the entry.

Add a call to dev_kfree_skb_any() so that skb entry is correctly
freed.

Signed-off-by: Jose Abreu <joabreu@synopsys.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Joao Pinto <jpinto@synopsys.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Sun, 22 Oct 2017 01:46:39 +0000 (21:46 -0400)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

Pull input fixes from Dmitry Torokhov:

 - joydev now implements a blacklist to avoid creating joystick nodes
   for accelerometers found in composite devices such as PlaStation
   controllers

 - assorted driver fixes

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: ims-psu - check if CDC union descriptor is sane
  Input: joydev - blacklist ds3/ds4/udraw motion sensors
  Input: allow matching device IDs on property bits
  Input: factor out and export input_device_id matching code
  Input: goodix - poll the 'buffer status' bit before reading data
  Input: axp20x-pek - fix module not auto-loading for axp221 pek
  Input: tca8418 - enable interrupt after it has been requested
  Input: stmfts - fix setting ABS_MT_POSITION_* maximum size
  Input: ti_am335x_tsc - fix incorrect step config for 5 wire touchscreen
  Input: synaptics - disable kernel tracking on SMBus devices

7 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sun, 22 Oct 2017 01:39:18 +0000 (21:39 -0400)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

Pull vfs fixes from Al Viro:
 "MS_I_VERSION fixes - Mimi's fix + missing bits picked from Matthew
  (his patch contained a duplicate of the fs/namespace.c fix as well,
  but by that point the original fix had already been applied)"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  Convert fs/*/* to SB_I_VERSION
  vfs: fix mounting a filesystem with i_version

7 years agoMerge branch 'mlxsw-fixes'
David S. Miller [Sun, 22 Oct 2017 01:19:03 +0000 (02:19 +0100)]
Merge branch 'mlxsw-fixes'

Jiri Pirko says:

====================
mlxsw: spectrum: Configure TTL of "inherit" for offloaded tunnels

Petr says:

Currently mlxsw only offloads tunnels that are configured with TTL of "inherit"
(which is the default). However, Spectrum defaults to 255 and the driver
neglects to change the configuration. Thus the tunnel packets from offloaded
tunnels always have TTL of 255, even though tunnels with explicit TTL of 255 are
never actually offloaded.

To fix this, introduce support for TIGCR, the register that keeps the related
bits of global tunnel configuration, and use it on first offload to properly
configure inheritance of TTL of tunnel packets from overlay packets.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum_router: Configure TIGCR on init
Petr Machata [Fri, 20 Oct 2017 07:16:16 +0000 (09:16 +0200)]
mlxsw: spectrum_router: Configure TIGCR on init

Spectrum tunnels do not default to ttl of "inherit" like the Linux ones
do. Configure TIGCR on router init so that the TTL of tunnel packets is
copied from the overlay packets.

Fixes: ee954d1a91b2 ("mlxsw: spectrum_router: Support GRE tunnels")
Signed-off-by: Petr Machata <petrm@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: reg: Add Tunneling IPinIP General Configuration Register
Petr Machata [Fri, 20 Oct 2017 07:16:15 +0000 (09:16 +0200)]
mlxsw: reg: Add Tunneling IPinIP General Configuration Register

The TIGCR register is used for setting up the IPinIP Tunnel
configuration.

Fixes: ee954d1a91b2 ("mlxsw: spectrum_router: Support GRE tunnels")
Signed-off-by: Petr Machata <petrm@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: ethtool: remove error check for legacy setting transceiver type
Niklas Söderlund [Thu, 19 Oct 2017 23:32:08 +0000 (01:32 +0200)]
net: ethtool: remove error check for legacy setting transceiver type

Commit 9cab88726929605 ("net: ethtool: Add back transceiver type")
restores the transceiver type to struct ethtool_link_settings and
convert_link_ksettings_to_legacy_settings() but forgets to remove the
error check for the same in convert_legacy_settings_to_link_ksettings().
This prevents older versions of ethtool to change link settings.

    # ethtool --version
    ethtool version 3.16

    # ethtool -s eth0 autoneg on speed 100 duplex full
    Cannot set new settings: Invalid argument
      not setting speed
      not setting duplex
      not setting autoneg

While newer versions of ethtool works.

    # ethtool --version
    ethtool version 4.10

    # ethtool -s eth0 autoneg on speed 100 duplex full
    [   57.703268] sh-eth ee700000.ethernet eth0: Link is Down
    [   59.618227] sh-eth ee700000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx

Fixes: 19cab88726929605 ("net: ethtool: Add back transceiver type")
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reported-by: Renjith R V <renjith.rv@quest-global.com>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosoreuseport: fix initialization race
Craig Gallek [Thu, 19 Oct 2017 19:00:29 +0000 (15:00 -0400)]
soreuseport: fix initialization race

Syzkaller stumbled upon a way to trigger
WARNING: CPU: 1 PID: 13881 at net/core/sock_reuseport.c:41
reuseport_alloc+0x306/0x3b0 net/core/sock_reuseport.c:39

There are two initialization paths for the sock_reuseport structure in a
socket: Through the udp/tcp bind paths of SO_REUSEPORT sockets or through
SO_ATTACH_REUSEPORT_[CE]BPF before bind.  The existing implementation
assumedthat the socket lock protected both of these paths when it actually
only protects the SO_ATTACH_REUSEPORT path.  Syzkaller triggered this
double allocation by running these paths concurrently.

This patch moves the check for double allocation into the reuseport_alloc
function which is protected by a global spin lock.

Fixes: e32ea7e74727 ("soreuseport: fast reuseport UDP socket selection")
Fixes: c125e80b8868 ("soreuseport: fast reuseport TCP socket selection")
Signed-off-by: Craig Gallek <kraig@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: bridge: fix returning of vlan range op errors
Nikolay Aleksandrov [Thu, 19 Oct 2017 17:17:32 +0000 (20:17 +0300)]
net: bridge: fix returning of vlan range op errors

When vlan tunnels were introduced, vlan range errors got silently
dropped and instead 0 was returned always. Restore the previous
behaviour and return errors to user-space.

Fixes: efa5356b0d97 ("bridge: per vlan dst_metadata netlink support")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Acked-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosock: correct sk_wmem_queued accounting on efault in tcp zerocopy
Willem de Bruijn [Thu, 19 Oct 2017 16:40:39 +0000 (12:40 -0400)]
sock: correct sk_wmem_queued accounting on efault in tcp zerocopy

Syzkaller hits WARN_ON(sk->sk_wmem_queued) in sk_stream_kill_queues
after triggering an EFAULT in __zerocopy_sg_from_iter.

On this error, skb_zerocopy_stream_iter resets the skb to its state
before the operation with __pskb_trim. It cannot kfree_skb like
datagram callers, as the skb may have data from a previous send call.

__pskb_trim calls skb_condense for unowned skbs, which adjusts their
truesize. These tcp skbuffs are owned and their truesize must add up
to sk_wmem_queued. But they match because their skb->sk is NULL until
tcp_transmit_skb.

Temporarily set skb->sk when calling __pskb_trim to signal that the
skbuffs are owned and avoid the skb_condense path.

Fixes: 52267790ef52 ("sock: add MSG_ZEROCOPY")
Signed-off-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'bpf-range-marking-fixes'
David S. Miller [Sat, 21 Oct 2017 23:56:10 +0000 (00:56 +0100)]
Merge branch 'bpf-range-marking-fixes'

Daniel Borkmann says:

====================
Two BPF fixes for range marking

The set contains two fixes for direct packet access range
markings and test cases for all direct packet access patterns
that the verifier matches on.

They are targeted for net tree, note that once net gets merged
into net-next, there will be a minor merge conflict due to
signature change of the function find_good_pkt_pointers() as
well as data_meta patterns present in net-next tree. You can
just add bool false to the data_meta patterns and I will
follow-up with properly converting the patterns for data_meta
in a similar way.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: add test cases to bpf selftests to cover all access tests
Daniel Borkmann [Sat, 21 Oct 2017 00:34:23 +0000 (02:34 +0200)]
bpf: add test cases to bpf selftests to cover all access tests

Lets add test cases to cover really all possible direct packet
access tests for good/bad access cases so we keep tracking them.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: fix pattern matches for direct packet access
Daniel Borkmann [Sat, 21 Oct 2017 00:34:22 +0000 (02:34 +0200)]
bpf: fix pattern matches for direct packet access

Alexander had a test program with direct packet access, where
the access test was in the form of data + X > data_end. In an
unrelated change to the program LLVM decided to swap the branches
and emitted code for the test in form of data + X <= data_end.
We hadn't seen these being generated previously, thus verifier
would reject the program. Therefore, fix up the verifier to
detect all test cases, so we don't run into such issues in the
future.

Fixes: b4e432f1000a ("bpf: enable BPF_J{LT, LE, SLT, SLE} opcodes in verifier")
Reported-by: Alexander Alemayhu <alexander@alemayhu.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: fix off by one for range markings with L{T, E} patterns
Daniel Borkmann [Sat, 21 Oct 2017 00:34:21 +0000 (02:34 +0200)]
bpf: fix off by one for range markings with L{T, E} patterns

During review I noticed that the current logic for direct packet
access marking in check_cond_jmp_op() has an off by one for the
upper right range border when marking in find_good_pkt_pointers()
with BPF_JLT and BPF_JLE. It's not really harmful given access
up to pkt_end is always safe, but we should nevertheless correct
the range marking before it becomes ABI. If pkt_data' denotes a
pkt_data derived pointer (pkt_data + X), then for pkt_data' < pkt_end
in the true branch as well as for pkt_end <= pkt_data' in the false
branch we mark the range with X although it should really be X - 1
in these cases. For example, X could be pkt_end - pkt_data, then
when testing for pkt_data' < pkt_end the verifier simulation cannot
deduce that a byte load of pkt_data' - 1 would succeed in this
branch.

Fixes: b4e432f1000a ("bpf: enable BPF_J{LT, LE, SLT, SLE} opcodes in verifier")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: devmap fix arithmetic overflow in bitmap_size calculation
John Fastabend [Thu, 19 Oct 2017 16:03:52 +0000 (09:03 -0700)]
bpf: devmap fix arithmetic overflow in bitmap_size calculation

An integer overflow is possible in dev_map_bitmap_size() when
calculating the BITS_TO_LONG logic which becomes, after macro
replacement,

(((n) + (d) - 1)/ (d))

where 'n' is a __u32 and 'd' is (8 * sizeof(long)). To avoid
overflow cast to u64 before arithmetic.

Reported-by: Richard Weinberger <richard@nod.at>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge tag 'dmaengine-fix-4.14-rc6' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Sat, 21 Oct 2017 18:32:46 +0000 (14:32 -0400)]
Merge tag 'dmaengine-fix-4.14-rc6' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine fix from Vinod Koul:
 "Late fix for altera driver which fixes the locking in driver"

* tag 'dmaengine-fix-4.14-rc6' of git://git.infradead.org/users/vkoul/slave-dma:
  dmaengine: altera: Use IRQ-safe spinlock calls in the error paths as well

7 years agoMerge branch 'aquantia-fixes'
David S. Miller [Sat, 21 Oct 2017 11:32:25 +0000 (12:32 +0100)]
Merge branch 'aquantia-fixes'

Igor Russkikh says:

====================
net: aquantia: Atlantic driver 10/2017 updates

This patchset fixes various issues in driver,
improves parameters for better performance on 10Gbit link
====================

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