net: disable fragment reassembly if high_thresh is set to zero

commit 30759219f562cfaaebe7b9c1d1c0e6b5445c69b0 upstream.

Before commit 6d7b857d541e ("net: use lib/percpu_counter API for
fragmentation mem accounting"), setting high threshold to 0 prevented
fragment reassembly as first fragment would be always evicted before
second could be added to the queue. While inefficient, some users
apparently relied on it.

Since the commit mentioned above, a percpu counter is used for
reassembly memory accounting and high batch size avoids taking slow path
in most common scenarios. As a result, a whole full sized packet can be
reassembled without the percpu counter's main counter changing its
value so that even with high_thresh set to 0, fragmented packets can be
still reassembled and processed.

Add explicit checks preventing reassembly if high threshold is zero.

[mk] backport to 3.12

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Willy Tarreau <w@1wt.eu>

mISDN: Fixing missing validation in base_sock_bind()

commit b821646826e22f0491708768fccce58eef3f5704 upstream.

Add validation code into mISDN/socket.c

Signed-off-by: Emrah Demir <ed@abdsec.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

mISDN: Support DR6 indication in mISDNipac driver

commit 1e1589ad8b5cb5b8a6781ba5850cf710ada0e919 upstream.

According to figure 39 in PEB3086 data sheet, version 1.4 this indication
replaces DR when layer 1 transition source state is F6.

This fixes mISDN layer 1 getting stuck in F6 state in TE mode on
Dialogic Diva 2.02 card (and possibly others) when NT deactivates it.

Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
Acked-by: Karsten Keil <keil@b1-systems.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

net: ratelimit warnings about dst entry refcount underflow or overflow

commit 8bf4ada2e21378816b28205427ee6b0e1ca4c5f1 upstream.

Kernel generates a lot of warnings when dst entry reference counter
overflows and becomes negative. That bug was seen several times at
machines with outdated 3.10.y kernels. Most like it's already fixed
in upstream. Anyway that flood completely kills machine and makes
further debugging impossible.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

bonding: Fix bonding crash

commit 24b27fc4cdf9e10c5e79e5923b6b7c2c5c95096c upstream.

Following few steps will crash kernel -

  (a) Create bonding master
      > modprobe bonding miimon=50
  (b) Create macvlan bridge on eth2
      > ip link add link eth2 dev mvl0 address aa:0:0:0:0:01 \
   type macvlan
  (c) Now try adding eth2 into the bond
      > echo +eth2 > /sys/class/net/bond0/bonding/slaves
      <crash>

Bonding does lots of things before checking if the device enslaved is
busy or not.

In this case when the notifier call-chain sends notifications, the
bond_netdev_event() assumes that the rx_handler /rx_handler_data is
registered while the bond_enslave() hasn't progressed far enough to
register rx_handler for the new slave.

This patch adds a rx_handler check that can be performed right at the
beginning of the enslave code to avoid getting into this situation.

Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tcp: take care of truncations done by sk_filter()

commit ac6e780070e30e4c35bd395acfe9191e6268bdd3 upstream.

With syzkaller help, Marco Grassi found a bug in TCP stack,
crashing in tcp_collapse()

Root cause is that sk_filter() can truncate the incoming skb,
but TCP stack was not really expecting this to happen.
It probably was expecting a simple DROP or ACCEPT behavior.

We first need to make sure no part of TCP header could be removed.
Then we need to adjust TCP_SKB_CB(skb)->end_seq

Many thanks to syzkaller team and Marco for giving us a reproducer.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Marco Grassi <marco.gra@gmail.com>
Reported-by: Vladis Dronov <vdronov@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tcp: fix wrong checksum calculation on MTU probing

commit 2fe664f1fcf7c4da6891f95708a7a56d3c024354 upstream.

With TCP MTU probing enabled and offload TX checksumming disabled,
tcp_mtu_probe() calculated the wrong checksum when a fragment being copied
into the probe's SKB had an odd length. This was caused by the direct use
of skb_copy_and_csum_bits() to calculate the checksum, as it pads the
fragment being copied, if needed. When this fragment was not the last, a
subsequent call used the previous checksum without considering this
padding.

The effect was a stale connection in one way, as even retransmissions
wouldn't solve the problem, because the checksum was never recalculated for
the full SKB length.

Signed-off-by: Douglas Caetano dos Santos <douglascs@taghos.com.br>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tcp: fix overflow in __tcp_retransmit_skb()

commit ffb4d6c8508657824bcef68a36b2a0f9d8c09d10 upstream.

If a TCP socket gets a large write queue, an overflow can happen
in a test in __tcp_retransmit_skb() preventing all retransmits.

The flow then stalls and resets after timeouts.

Tested:

sysctl -w net.core.wmem_max=1000000000
netperf -H dest -- -s 1000000000

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tcp: properly scale window in tcp_v[46]_reqsk_send_ack()

commit 20a2b49fc538540819a0c552877086548cff8d8d upstream.

When sending an ack in SYN_RECV state, we must scale the offered
window if wscale option was negotiated and accepted.

Tested:
 Following packetdrill test demonstrates the issue :

0.000 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
+0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0

+0 bind(3, ..., ...) = 0
+0 listen(3, 1) = 0

// Establish a connection.
+0 < S 0:0(0) win 20000 <mss 1000,sackOK,wscale 7, nop, TS val 100 ecr 0>
+0 > S. 0:0(0) ack 1 win 28960 <mss 1460,sackOK, TS val 100 ecr 100, nop, wscale 7>

+0 < . 1:11(10) ack 1 win 156 <nop,nop,TS val 99 ecr 100>
// check that window is properly scaled !
+0 > . 1:1(0) ack 1 win 226 <nop,nop,TS val 200 ecr 100>

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tcp: fix use after free in tcp_xmit_retransmit_queue()

commit bb1fceca22492109be12640d49f5ea5a544c6bb4 upstream.

When tcp_sendmsg() allocates a fresh and empty skb, it puts it at the
tail of the write queue using tcp_add_write_queue_tail()

Then it attempts to copy user data into this fresh skb.

If the copy fails, we undo the work and remove the fresh skb.

Unfortunately, this undo lacks the change done to tp->highest_sack and
we can leave a dangling pointer (to a freed skb)

Later, tcp_xmit_retransmit_queue() can dereference this pointer and
access freed memory. For regular kernels where memory is not unmapped,
this might cause SACK bugs because tcp_highest_sack_seq() is buggy,
returning garbage instead of tp->snd_nxt, but with various debug
features like CONFIG_DEBUG_PAGEALLOC, this can crash the kernel.

This bug was found by Marco Grassi thanks to syzkaller.

Fixes: 6859d49475d4 ("[TCP]: Abstract tp->highest_sack accessing & point to next skb")
Reported-by: Marco Grassi <marco.gra@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

net/irda: handle iriap_register_lsap() allocation failure

commit 5ba092efc7ddff040777ae7162f1d195f513571b upstream.

If iriap_register_lsap() fails to allocate memory, self->lsap is
set to NULL. However, none of the callers handle the failure and
irlmp_connect_request() will happily dereference it:

    iriap_register_lsap: Unable to allocated LSAP!
    ================================================================================
    UBSAN: Undefined behaviour in net/irda/irlmp.c:378:2
    member access within null pointer of type 'struct lsap_cb'
    CPU: 1 PID: 15403 Comm: trinity-c0 Not tainted 4.8.0-rc1+ #81
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.9.3-0-ge2fc41e-prebuilt.qemu-project.org
    04/01/2014
     0000000000000000 ffff88010c7e78a8 ffffffff82344f40 0000000041b58ab3
     ffffffff84f98000 ffffffff82344e94 ffff88010c7e78d0 ffff88010c7e7880
     ffff88010630ad00 ffffffff84a5fae0 ffffffff84d3f5c0 000000000000017a
    Call Trace:
     [<ffffffff82344f40>] dump_stack+0xac/0xfc
     [<ffffffff8242f5a8>] ubsan_epilogue+0xd/0x8a
     [<ffffffff824302bf>] __ubsan_handle_type_mismatch+0x157/0x411
     [<ffffffff83b7bdbc>] irlmp_connect_request+0x7ac/0x970
     [<ffffffff83b77cc0>] iriap_connect_request+0xa0/0x160
     [<ffffffff83b77f48>] state_s_disconnect+0x88/0xd0
     [<ffffffff83b78904>] iriap_do_client_event+0x94/0x120
     [<ffffffff83b77710>] iriap_getvaluebyclass_request+0x3e0/0x6d0
     [<ffffffff83ba6ebb>] irda_find_lsap_sel+0x1eb/0x630
     [<ffffffff83ba90c8>] irda_connect+0x828/0x12d0
     [<ffffffff833c0dfb>] SYSC_connect+0x22b/0x340
     [<ffffffff833c7e09>] SyS_connect+0x9/0x10
     [<ffffffff81007bd3>] do_syscall_64+0x1b3/0x4b0
     [<ffffffff845f946a>] entry_SYSCALL64_slow_path+0x25/0x25
    ================================================================================

The bug seems to have been around since forever.

There's more problems with missing error checks in iriap_init() (and
indeed all of irda_init()), but that's a bigger problem that needs
very careful review and testing. This patch will fix the most serious
bug (as it's easily reached from unprivileged userspace).

I have tested my patch with a reproducer.

Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ip6_tunnel: disable caching when the traffic class is inherited

commit b5c2d49544e5930c96e2632a7eece3f4325a1888 upstream.

If an ip6 tunnel is configured to inherit the traffic class from
the inner header, the dst_cache must be disabled or it will foul
the policy routing.

The issue is apprently there since at leat Linux-2.6.12-rc2.

Reported-by: Liam McBirnie <liam.mcbirnie@boeing.com>
Cc: Liam McBirnie <liam.mcbirnie@boeing.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ip6_tunnel: Clear IP6CB in ip6tunnel_xmit()

commit 23f4ffedb7d751c7e298732ba91ca75d224bc1a6 upstream.

skb->cb may contain data from previous layers. In the observed scenario,
the garbage data were misinterpreted as IP6CB(skb)->frag_max_size, so
that small packets sent through the tunnel are mistakenly fragmented.

This patch unconditionally clears the control buffer in ip6tunnel_xmit(),
which affects ip6_tunnel, ip6_udp_tunnel and ip6_gre. Currently none of
these tunnels set IP6CB(skb)->flags, otherwise it needs to be done earlier.

Cc: stable@vger.kernel.org
Signed-off-by: Eli Cooper <elicooper@gmx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: dccp: add missing bind_conflict to dccp_ipv6_mapped

commit 990ff4d84408fc55942ca6644f67e361737b3d8e upstream.

While fuzzing kernel with syzkaller, Andrey reported a nasty crash
in inet6_bind() caused by DCCP lacking a required method.

Fixes: ab1e0a13d7029 ("[SOCK] proto: Add hashinfo member to struct proto")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Tested-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: dccp: fix out of bound access in dccp_v6_err()

commit 1aa9d1a0e7eefcc61696e147d123453fc0016005 upstream.

dccp_v6_err() does not use pskb_may_pull() and might access garbage.

We only need 4 bytes at the beginning of the DCCP header, like TCP,
so the 8 bytes pulled in icmpv6_notify() are more than enough.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: correctly add local routes when lo goes up

commit a220445f9f4382c36a53d8ef3e08165fa27f7e2c upstream.

The goal of the patch is to fix this scenario:
 ip link add dummy1 type dummy
 ip link set dummy1 up
 ip link set lo down ; ip link set lo up

After that sequence, the local route to the link layer address of dummy1 is
not there anymore.

When the loopback is set down, all local routes are deleted by
addrconf_ifdown()/rt6_ifdown(). At this time, the rt6_info entry still
exists, because the corresponding idev has a reference on it. After the rcu
grace period, dst_rcu_free() is called, and thus ___dst_free(), which will
set obsolete to DST_OBSOLETE_DEAD.

In this case, init_loopback() is called before dst_rcu_free(), thus
obsolete is still sets to something <= 0. So, the function doesn't add the
route again. To avoid that race, let's check the rt6 refcnt instead.

Fixes: 25fb6ca4ed9c ("net IPv6 : Fix broken IPv6 routing table after loopback down-up")
Fixes: a881ae1f625c ("ipv6: don't call addrconf_dst_alloc again when enable lo")
Fixes: 33d99113b110 ("ipv6: reallocate addrconf router for ipv6 address when lo device up")
Reported-by: Francesco Santoro <francesco.santoro@6wind.com>
Reported-by: Samuel Gauthier <samuel.gauthier@6wind.com>
CC: Balakumaran Kannan <Balakumaran.Kannan@ap.sony.com>
CC: Maruthi Thotad <Maruthi.Thotad@ap.sony.com>
CC: Sabrina Dubroca <sd@queasysnail.net>
CC: Hannes Frederic Sowa <hannes@stressinduktion.org>
CC: Weilong Chen <chenweilong@huawei.com>
CC: Gao feng <gaofeng@cn.fujitsu.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ip6_gre: fix flowi6_proto value in ip6gre_xmit_other()

commit db32e4e49ce2b0e5fcc17803d011a401c0a637f6 upstream.

Similar to commit 3be07244b733 ("ip6_gre: fix flowi6_proto value in
xmit path"), set flowi6_proto to IPPROTO_GRE for output route lookup.

Up until now, ip6gre_xmit_other() has set flowi6_proto to a bogus value.
This affected output route lookup for packets sent on an ip6gretap device
in cases where routing was dependent on the value of flowi6_proto.

Since the correct proto is already set in the tunnel flowi6 template via
commit 252f3f5a1189 ("ip6_gre: Set flowi6_proto as IPPROTO_GRE in xmit
path."), simply delete the line setting the incorrect flowi6_proto value.

Suggested-by: Jiri Benc <jbenc@redhat.com>
Fixes: c12b395a4664 ("gre: Support GRE over IPv6")
Reviewed-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Signed-off-by: Lance Richardson <lrichard@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: fix rtnl locking in setsockopt for anycast and multicast

commit a9ed4a2986e13011fcf4ed2d1a1647c53112f55b upstream.

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST
triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to
take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with
ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before
calling ipv6_dev_mc_inc/dec.

This patch moves ASSERT_RTNL() up a level in the call stack.

Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Reported-by: Tommi Rantala <tt.rantala@gmail.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: <stable@vger.kernel.org> # 3.10.y: b7b1bfce: ipv6: split dad and rs timers
Cc: <stable@vger.kernel.org> # 3.10.y: c15b1cca: ipv6: move dad to workqueue
Cc: <stable@vger.kernel.org> # 3.10.y
[Mike Manning <mmanning@brocade.com>: resolved minor conflicts in addrconf.c]
Signed-off-by: Mike Manning <mmanning@brocade.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: addrconf: fix dev refcont leak when DAD failed

commit 751eb6b6042a596b0080967c1a529a9fe98dac1d upstream.

In general, when DAD detected IPv6 duplicate address, ifp->state
will be set to INET6_IFADDR_STATE_ERRDAD and DAD is stopped by a
delayed work, the call tree should be like this:

ndisc_recv_ns
  -> addrconf_dad_failure        <- missing ifp put
     -> addrconf_mod_dad_work
       -> schedule addrconf_dad_work()
         -> addrconf_dad_stop()  <- missing ifp hold before call it

addrconf_dad_failure() called with ifp refcont holding but not put.
addrconf_dad_work() call addrconf_dad_stop() without extra holding
refcount. This will not cause any issue normally.

But the race between addrconf_dad_failure() and addrconf_dad_work()
may cause ifp refcount leak and netdevice can not be unregister,
dmesg show the following messages:

IPv6: eth0: IPv6 duplicate address fe80::XX:XXXX:XXXX:XX detected!
...
unregister_netdevice: waiting for eth0 to become free. Usage count = 1

Cc: stable@vger.kernel.org
Fixes: c15b1ccadb32 ("ipv6: move DAD and addrconf_verify processing
to workqueue")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: <stable@vger.kernel.org> # 3.10.y
Signed-off-by: Mike Manning <mmanning@brocade.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: move DAD and addrconf_verify processing to workqueue

commit c15b1ccadb323ea50023e8f1cca2954129a62b51 upstream.

addrconf_join_solict and addrconf_join_anycast may cause actions which
need rtnl locked, especially on first address creation.

A new DAD state is introduced which defers processing of the initial
DAD processing into a workqueue.

To get rtnl lock we need to push the code paths which depend on those
calls up to workqueues, specifically addrconf_verify and the DAD
processing.

(v2)
addrconf_dad_failure needs to be queued up to the workqueue, too. This
patch introduces a new DAD state and stop the DAD processing in the
workqueue (this is because of the possible ipv6_del_addr processing
which removes the solicited multicast address from the device).

addrconf_verify_lock is removed, too. After the transition it is not
needed any more.

As we are not processing in bottom half anymore we need to be a bit more
careful about disabling bottom half out when we lock spin_locks which are also
used in bh.

Relevant backtrace:
[  541.030090] RTNL: assertion failed at net/core/dev.c (4496)
[  541.031143] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G           O 3.10.33-1-amd64-vyatta #1
[  541.031145] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[  541.031146]  ffffffff8148a9f0 000000000000002f ffffffff813c98c1 ffff88007c4451f8
[  541.031148]  0000000000000000 0000000000000000 ffffffff813d3540 ffff88007fc03d18
[  541.031150]  0000880000000006 ffff88007c445000 ffffffffa0194160 0000000000000000
[  541.031152] Call Trace:
[  541.031153]  <IRQ>  [<ffffffff8148a9f0>] ? dump_stack+0xd/0x17
[  541.031180]  [<ffffffff813c98c1>] ? __dev_set_promiscuity+0x101/0x180
[  541.031183]  [<ffffffff813d3540>] ? __hw_addr_create_ex+0x60/0xc0
[  541.031185]  [<ffffffff813cfe1a>] ? __dev_set_rx_mode+0xaa/0xc0
[  541.031189]  [<ffffffff813d3a81>] ? __dev_mc_add+0x61/0x90
[  541.031198]  [<ffffffffa01dcf9c>] ? igmp6_group_added+0xfc/0x1a0 [ipv6]
[  541.031208]  [<ffffffff8111237b>] ? kmem_cache_alloc+0xcb/0xd0
[  541.031212]  [<ffffffffa01ddcd7>] ? ipv6_dev_mc_inc+0x267/0x300 [ipv6]
[  541.031216]  [<ffffffffa01c2fae>] ? addrconf_join_solict+0x2e/0x40 [ipv6]
[  541.031219]  [<ffffffffa01ba2e9>] ? ipv6_dev_ac_inc+0x159/0x1f0 [ipv6]
[  541.031223]  [<ffffffffa01c0772>] ? addrconf_join_anycast+0x92/0xa0 [ipv6]
[  541.031226]  [<ffffffffa01c311e>] ? __ipv6_ifa_notify+0x11e/0x1e0 [ipv6]
[  541.031229]  [<ffffffffa01c3213>] ? ipv6_ifa_notify+0x33/0x50 [ipv6]
[  541.031233]  [<ffffffffa01c36c8>] ? addrconf_dad_completed+0x28/0x100 [ipv6]
[  541.031241]  [<ffffffff81075c1d>] ? task_cputime+0x2d/0x50
[  541.031244]  [<ffffffffa01c38d6>] ? addrconf_dad_timer+0x136/0x150 [ipv6]
[  541.031247]  [<ffffffffa01c37a0>] ? addrconf_dad_completed+0x100/0x100 [ipv6]
[  541.031255]  [<ffffffff8105313a>] ? call_timer_fn.isra.22+0x2a/0x90
[  541.031258]  [<ffffffffa01c37a0>] ? addrconf_dad_completed+0x100/0x100 [ipv6]

Hunks and backtrace stolen from a patch by Stephen Hemminger.

Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: <stable@vger.kernel.org> # 3.10.y: b7b1bfce: ipv6: split dad and rs timers
Cc: <stable@vger.kernel.org> # 3.10.y
[Mike Manning <mmanning@brocade.com>: resolved minor conflicts in addrconf.c]
Signed-off-by: Mike Manning <mmanning@brocade.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: split duplicate address detection and router solicitation timer

commit b7b1bfce0bb68bd8f6e62a28295922785cc63781 upstream.

This patch splits the timers for duplicate address detection and router
solicitations apart. The router solicitations timer goes into inet6_dev
and the dad timer stays in inet6_ifaddr.

The reason behind this patch is to reduce the number of unneeded router
solicitations send out by the host if additional link-local addresses
are created. Currently we send out RS for every link-local address on
an interface.

If the RS timer fires we pick a source address with ipv6_get_lladdr. This
change could hurt people adding additional link-local addresses and
specifying these addresses in the radvd clients section because we
no longer guarantee that we use every ll address as source address in
router solicitations.

Cc: Flavio Leitner <fleitner@redhat.com>
Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: David Stevens <dlstevens@us.ibm.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Reviewed-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: <stable@vger.kernel.org> # 3.10.y
[Mike Manning <mmanning@brocade.com>: resolved conflicts with 36bddb]
Signed-off-by: Mike Manning <mmanning@brocade.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ipv6: don't call fib6_run_gc() until routing is ready

commit 2c861cc65ef4604011a0082e4dcdba2819aa191a upstream.

When loading the ipv6 module, ndisc_init() is called before
ip6_route_init(). As the former registers a handler calling
fib6_run_gc(), this opens a window to run the garbage collector
before necessary data structures are initialized. If a network
device is initialized in this window, adding MAC address to it
triggers a NETDEV_CHANGEADDR event, leading to a crash in
fib6_clean_all().

Take the event handler registration out of ndisc_init() into a
separate function ndisc_late_init() and move it after
ip6_route_init().

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: <stable@vger.kernel.org> # 3.10.y
Signed-off-by: Mike Manning <mmanning@brocade.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

stddef.h: move offsetofend inside #ifndef/#endif guard, neaten

commit 8c7fbe5795a016259445a61e072eb0118aaf6a61 upstream.

Commit 3876488444e7 ("include/stddef.h: Move offsetofend() from vfio.h
to a generic kernel header") added offsetofend outside the normal
include #ifndef/#endif guard.  Move it inside.

Miscellanea:

o remove unnecessary blank line
o standardize offsetof macros whitespace style

Signed-off-by: Joe Perches <joe@perches.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[wt: backported only for ipv6 out-of-bounds fix]

Signed-off-by: Willy Tarreau <w@1wt.eu>

include/stddef.h: Move offsetofend() from vfio.h to a generic kernel header

commit 3876488444e71238e287459c39d7692b6f718c3e upstream.

Suggested by Andy.

Suggested-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Alexei Starovoitov <ast@plumgrid.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Will Drewry <wad@chromium.org>
Link: http://lkml.kernel.org/r/1425912738-559-1-git-send-email-dvlasenk@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
[wt: backported only for ipv6 out-of-bounds fix]

Signed-off-by: Willy Tarreau <w@1wt.eu>

drivers/vfio: Rework offsetofend()

commit b13460b92093b29347e99d6c3242e350052b62cd upstream.

The macro offsetofend() introduces unnecessary temporary variable
"tmp". The patch avoids that and saves a bit memory in stack.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
[wt: backported only for ipv6 out-of-bounds fix]

Signed-off-by: Willy Tarreau <w@1wt.eu>

vt: clear selection before resizing

commit 009e39ae44f4191188aeb6dfbf661b771dbbe515 upstream.

When resizing a vt its selection may exceed the new size, resulting in
an invalid memory access [1]. Clear the selection before resizing.

[1] http://lkml.kernel.org/r/CACT4Y+acDTwy4umEvf5ROBGiRJNrxHN4Cn5szCXE5Jw-d1B=Xw@mail.gmail.com

Reported-and-tested-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Scot Doyle <lkml14@scotdoyle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tty: vt, fix bogus division in csi_J

commit 42acfc6615f47e465731c263bee0c799edb098f2 upstream.

In csi_J(3), the third parameter of scr_memsetw (vc_screenbuf_size) is
divided by 2 inappropriatelly. But scr_memsetw expects size, not
count, because it divides the size by 2 on its own before doing actual
memset-by-words.

So remove the bogus division.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Petr PísaÅ\99 <ppisar@redhat.com>
Fixes: f8df13e0a9 (tty: Clean console safely)
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tty: limit terminal size to 4M chars

commit 32b2921e6a7461fe63b71217067a6cf4bddb132f upstream.

Size of kmalloc() in vc_do_resize() is controlled by user.
Too large kmalloc() size triggers WARNING message on console.
Put a reasonable upper bound on terminal size to prevent WARNINGs.

Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
CC: David Rientjes <rientjes@google.com>
Cc: One Thousand Gnomes <gnomes@lxorguk.ukuu.org.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jiri Slaby <jslaby@suse.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: linux-kernel@vger.kernel.org
Cc: syzkaller@googlegroups.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

tty: Prevent ldisc drivers from re-using stale tty fields

commit dd42bf1197144ede075a9d4793123f7689e164bc upstream.

Line discipline drivers may mistakenly misuse ldisc-related fields
when initializing. For example, a failure to initialize tty->receive_room
in the N_GIGASET_M101 line discipline was recently found and fixed [1].
Now, the N_X25 line discipline has been discovered accessing the previous
line discipline's already-freed private data [2].

Harden the ldisc interface against misuse by initializing revelant
tty fields before instancing the new line discipline.

[1]
    commit fd98e9419d8d622a4de91f76b306af6aa627aa9c
    Author: Tilman Schmidt <tilman@imap.cc>
    Date:   Tue Jul 14 00:37:13 2015 +0200

    isdn/gigaset: reset tty->receive_room when attaching ser_gigaset

[2] Report from Sasha Levin <sasha.levin@oracle.com>
    [  634.336761] ==================================================================
    [  634.338226] BUG: KASAN: use-after-free in x25_asy_open_tty+0x13d/0x490 at addr ffff8800a743efd0
    [  634.339558] Read of size 4 by task syzkaller_execu/8981
    [  634.340359] =============================================================================
    [  634.341598] BUG kmalloc-512 (Not tainted): kasan: bad access detected
    ...
    [  634.405018] Call Trace:
    [  634.405277] dump_stack (lib/dump_stack.c:52)
    [  634.405775] print_trailer (mm/slub.c:655)
    [  634.406361] object_err (mm/slub.c:662)
    [  634.406824] kasan_report_error (mm/kasan/report.c:138 mm/kasan/report.c:236)
    [  634.409581] __asan_report_load4_noabort (mm/kasan/report.c:279)
    [  634.411355] x25_asy_open_tty (drivers/net/wan/x25_asy.c:559 (discriminator 1))
    [  634.413997] tty_ldisc_open.isra.2 (drivers/tty/tty_ldisc.c:447)
    [  634.414549] tty_set_ldisc (drivers/tty/tty_ldisc.c:567)
    [  634.415057] tty_ioctl (drivers/tty/tty_io.c:2646 drivers/tty/tty_io.c:2879)
    [  634.423524] do_vfs_ioctl (fs/ioctl.c:43 fs/ioctl.c:607)
    [  634.427491] SyS_ioctl (fs/ioctl.c:622 fs/ioctl.c:613)
    [  634.427945] entry_SYSCALL_64_fastpath (arch/x86/entry/entry_64.S:188)

Cc: Tilman Schmidt <tilman@imap.cc>
Cc: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[wt: adjust context]

Signed-off-by: Willy Tarreau <w@1wt.eu>

perf: Tighten (and fix) the grouping condition

commit c3c87e770458aa004bd7ed3f29945ff436fd6511 upstream.

The fix from 9fc81d87420d ("perf: Fix events installation during
moving group") was incomplete in that it failed to recognise that
creating a group with events for different CPUs is semantically
broken -- they cannot be co-scheduled.

Furthermore, it leads to real breakage where, when we create an event
for CPU Y and then migrate it to form a group on CPU X, the code gets
confused where the counter is programmed -- triggered in practice
as well by me via the perf fuzzer.

Fix this by tightening the rules for creating groups. Only allow
grouping of counters that can be co-scheduled in the same context.
This means for the same task and/or the same cpu.

Fixes: 9fc81d87420d ("perf: Fix events installation during moving group")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/20150123125834.090683288@infradead.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

perf symbols: Fixup symbol sizes before picking best ones

commit 432746f8e0b6a82ba832b771afe31abd51af6752 upstream.

When we call symbol__fixup_duplicate() we use algorithms to pick the
"best" symbols for cases where there are various functions/aliases to an
address, and those check zero size symbols, which, before calling
symbol__fixup_end() are _all_ symbols in a just parsed kallsyms file.

So first fixup the end, then fixup the duplicates.

Found while trying to figure out why 'perf test vmlinux' failed, see the
output of 'perf test -v vmlinux' to see cases where the symbols picked
as best for vmlinux don't match the ones picked for kallsyms.

Cc: Anton Blanchard <anton@samba.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Fixes: 694bf407b061 ("perf symbols: Add some heuristics for choosing the best duplicate symbol")
Link: http://lkml.kernel.org/n/tip-rxqvdgr0mqjdxee0kf8i2ufn@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

mtd: nand: davinci: Reinitialize the HW ECC engine in 4bit hwctl

commit f6d7c1b5598b6407c3f1da795dd54acf99c1990c upstream.

This fixes subpage writes when using 4-bit HW ECC.

There has been numerous reports about ECC errors with devices using this
driver for a while.  Also the 4-bit ECC has been reported as broken with
subpages in [1] and with 16 bits NANDs in the driver and in mach* board
files both in mainline and in the vendor BSPs.

What I saw with 4-bit ECC on a 16bits NAND (on an LCDK) which got me to
try reinitializing the ECC engine:
- R/W on whole pages properly generates/checks RS code
- try writing the 1st subpage only of a blank page, the subpage is well
  written and the RS code properly generated, re-reading the same page
  the HW detects some ECC error, reading the same page again no ECC
  error is detected

Note that the ECC engine is already reinitialized in the 1-bit case.

Tested on my LCDK with UBI+UBIFS using subpages.
This could potentially get rid of the issue workarounded in [1].

[1] 28c015a9daab ("mtd: davinci-nand: disable subpage write for keystone-nand")

Fixes: 6a4123e581b3 ("mtd: nand: davinci_nand, 4-bit ECC for smallpage")
Signed-off-by: Karl Beldan <kbeldan@baylibre.com>
Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

mtd: pmcmsp-flash: Allocating too much in init_msp_flash()

commit 79ad07d45743721010e766e65dc004ad249bd429 upstream.

There is a cut and paste issue here.  The bug is that we are allocating
more memory than necessary for msp_maps.  We should be allocating enough
space for a map_info struct (144 bytes) but we instead allocate enough
for an mtd_info struct (1840 bytes).  It's a small waste.

The other part of this is not harmful but when we allocated msp_flash
then we allocated enough space fro a map_info pointer instead of an
mtd_info pointer.  But since pointers are the same size it works out
fine.

Anyway, I decided to clean up all three allocations a bit to make them
a bit more consistent and clear.

Fixes: 68aa0fa87f6d ('[MTD] PMC MSP71xx flash/rootfs mappings')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

mtd: blkdevs: fix potential deadlock + lockdep warnings

commit f3c63795e90f0c6238306883b6c72f14d5355721 upstream.

Commit 073db4a51ee4 ("mtd: fix: avoid race condition when accessing
mtd->usecount") fixed a race condition but due to poor ordering of the
mutex acquisition, introduced a potential deadlock.

The deadlock can occur, for example, when rmmod'ing the m25p80 module, which
will delete one or more MTDs, along with any corresponding mtdblock
devices. This could potentially race with an acquisition of the block
device as follows.

 -> blktrans_open()
    ->  mutex_lock(&dev->lock);
    ->  mutex_lock(&mtd_table_mutex);

 -> del_mtd_device()
    ->  mutex_lock(&mtd_table_mutex);
    ->  blktrans_notify_remove() -> del_mtd_blktrans_dev()
       ->  mutex_lock(&dev->lock);

This is a classic (potential) ABBA deadlock, which can be fixed by
making the A->B ordering consistent everywhere. There was no real
purpose to the ordering in the original patch, AFAIR, so this shouldn't
be a problem. This ordering was actually already present in
del_mtd_blktrans_dev(), for one, where the function tried to ensure that
its caller already held mtd_table_mutex before it acquired &dev->lock:

        if (mutex_trylock(&mtd_table_mutex)) {
                mutex_unlock(&mtd_table_mutex);
                BUG();
        }

So, reverse the ordering of acquisition of &dev->lock and &mtd_table_mutex so
we always acquire mtd_table_mutex first.

Snippets of the lockdep output follow:

  # modprobe -r m25p80
  [   53.419251]
  [   53.420838] ======================================================
  [   53.427300] [ INFO: possible circular locking dependency detected ]
  [   53.433865] 4.3.0-rc6 #96 Not tainted
  [   53.437686] -------------------------------------------------------
  [   53.444220] modprobe/372 is trying to acquire lock:
  [   53.449320]  (&new->lock){+.+...}, at: [<c043fe4c>] del_mtd_blktrans_dev+0x80/0xdc
  [   53.457271]
  [   53.457271] but task is already holding lock:
  [   53.463372]  (mtd_table_mutex){+.+.+.}, at: [<c0439994>] del_mtd_device+0x18/0x100
  [   53.471321]
  [   53.471321] which lock already depends on the new lock.
  [   53.471321]
  [   53.479856]
  [   53.479856] the existing dependency chain (in reverse order) is:
  [   53.487660]
  -> #1 (mtd_table_mutex){+.+.+.}:
  [   53.492331]        [<c043fc5c>] blktrans_open+0x34/0x1a4
  [   53.497879]        [<c01afce0>] __blkdev_get+0xc4/0x3b0
  [   53.503364]        [<c01b0bb8>] blkdev_get+0x108/0x320
  [   53.508743]        [<c01713c0>] do_dentry_open+0x218/0x314
  [   53.514496]        [<c0180454>] path_openat+0x4c0/0xf9c
  [   53.519959]        [<c0182044>] do_filp_open+0x5c/0xc0
  [   53.525336]        [<c0172758>] do_sys_open+0xfc/0x1cc
  [   53.530716]        [<c000f740>] ret_fast_syscall+0x0/0x1c
  [   53.536375]
  -> #0 (&new->lock){+.+...}:
  [   53.540587]        [<c063f124>] mutex_lock_nested+0x38/0x3cc
  [   53.546504]        [<c043fe4c>] del_mtd_blktrans_dev+0x80/0xdc
  [   53.552606]        [<c043f164>] blktrans_notify_remove+0x7c/0x84
  [   53.558891]        [<c04399f0>] del_mtd_device+0x74/0x100
  [   53.564544]        [<c043c670>] del_mtd_partitions+0x80/0xc8
  [   53.570451]        [<c0439aa0>] mtd_device_unregister+0x24/0x48
  [   53.576637]        [<c046ce6c>] spi_drv_remove+0x1c/0x34
  [   53.582207]        [<c03de0f0>] __device_release_driver+0x88/0x114
  [   53.588663]        [<c03de19c>] device_release_driver+0x20/0x2c
  [   53.594843]        [<c03dd9e8>] bus_remove_device+0xd8/0x108
  [   53.600748]        [<c03dacc0>] device_del+0x10c/0x210
  [   53.606127]        [<c03dadd0>] device_unregister+0xc/0x20
  [   53.611849]        [<c046d878>] __unregister+0x10/0x20
  [   53.617211]        [<c03da868>] device_for_each_child+0x50/0x7c
  [   53.623387]        [<c046eae8>] spi_unregister_master+0x58/0x8c
  [   53.629578]        [<c03e12f0>] release_nodes+0x15c/0x1c8
  [   53.635223]        [<c03de0f8>] __device_release_driver+0x90/0x114
  [   53.641689]        [<c03de900>] driver_detach+0xb4/0xb8
  [   53.647147]        [<c03ddc78>] bus_remove_driver+0x4c/0xa0
  [   53.652970]        [<c00cab50>] SyS_delete_module+0x11c/0x1e4
  [   53.658976]        [<c000f740>] ret_fast_syscall+0x0/0x1c
  [   53.664621]
  [   53.664621] other info that might help us debug this:
  [   53.664621]
  [   53.672979]  Possible unsafe locking scenario:
  [   53.672979]
  [   53.679169]        CPU0                    CPU1
  [   53.683900]        ----                    ----
  [   53.688633]   lock(mtd_table_mutex);
  [   53.692383]                                lock(&new->lock);
  [   53.698306]                                lock(mtd_table_mutex);
  [   53.704658]   lock(&new->lock);
  [   53.707946]
  [   53.707946]  *** DEADLOCK ***

Fixes: 073db4a51ee4 ("mtd: fix: avoid race condition when accessing mtd->usecount")
Reported-by: Felipe Balbi <balbi@ti.com>
Tested-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

IB/cm: Mark stale CM id's whenever the mad agent was unregistered

commit 9db0ff53cb9b43ed75bacd42a89c1a0ab048b2b0 upstream.

When there is a CM id object that has port assigned to it, it means that
the cm-id asked for the specific port that it should go by it, but if
that port was removed (hot-unplug event) the cm-id was not updated.
In order to fix that the port keeps a list of all the cm-id's that are
planning to go by it, whenever the port is removed it marks all of them
as invalid.

This commit fixes a kernel panic which happens when running traffic between
guests and we force reboot a guest mid traffic, it triggers a kernel panic:

 Call Trace:
  [<ffffffff815271fa>] ? panic+0xa7/0x16f
  [<ffffffff8152b534>] ? oops_end+0xe4/0x100
  [<ffffffff8104a00b>] ? no_context+0xfb/0x260
  [<ffffffff81084db2>] ? del_timer_sync+0x22/0x30
  [<ffffffff8104a295>] ? __bad_area_nosemaphore+0x125/0x1e0
  [<ffffffff81084240>] ? process_timeout+0x0/0x10
  [<ffffffff8104a363>] ? bad_area_nosemaphore+0x13/0x20
  [<ffffffff8104aabf>] ? __do_page_fault+0x31f/0x480
  [<ffffffff81065df0>] ? default_wake_function+0x0/0x20
  [<ffffffffa0752675>] ? free_msg+0x55/0x70 [mlx5_core]
  [<ffffffffa0753434>] ? cmd_exec+0x124/0x840 [mlx5_core]
  [<ffffffff8105a924>] ? find_busiest_group+0x244/0x9f0
  [<ffffffff8152d45e>] ? do_page_fault+0x3e/0xa0
  [<ffffffff8152a815>] ? page_fault+0x25/0x30
  [<ffffffffa024da25>] ? cm_alloc_msg+0x35/0xc0 [ib_cm]
  [<ffffffffa024e821>] ? ib_send_cm_dreq+0xb1/0x1e0 [ib_cm]
  [<ffffffffa024f836>] ? cm_destroy_id+0x176/0x320 [ib_cm]
  [<ffffffffa024fb00>] ? ib_destroy_cm_id+0x10/0x20 [ib_cm]
  [<ffffffffa034f527>] ? ipoib_cm_free_rx_reap_list+0xa7/0x110 [ib_ipoib]
  [<ffffffffa034f590>] ? ipoib_cm_rx_reap+0x0/0x20 [ib_ipoib]
  [<ffffffffa034f5a5>] ? ipoib_cm_rx_reap+0x15/0x20 [ib_ipoib]
  [<ffffffff81094d20>] ? worker_thread+0x170/0x2a0
  [<ffffffff8109b2a0>] ? autoremove_wake_function+0x0/0x40
  [<ffffffff81094bb0>] ? worker_thread+0x0/0x2a0
  [<ffffffff8109aef6>] ? kthread+0x96/0xa0
  [<ffffffff8100c20a>] ? child_rip+0xa/0x20
  [<ffffffff8109ae60>] ? kthread+0x0/0xa0
  [<ffffffff8100c200>] ? child_rip+0x0/0x20

Fixes: a977049dacde ("[PATCH] IB: Add the kernel CM implementation")
Signed-off-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

IB/uverbs: Fix leak of XRC target QPs

commit 5b810a242c28e1d8d64d718cebe75b79d86a0b2d upstream.

The real QP is destroyed in case of the ref count reaches zero, but
for XRC target QPs this call was missed and caused to QP leaks.

Let's call to destroy for all flows.

Fixes: 0e0ec7e0638e ('RDMA/core: Export ib_open_qp() to share XRC...')
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

IB/mlx4: Fix create CQ error flow

commit 593ff73bcfdc79f79a8a0df55504f75ad3e5d1a9 upstream.

Currently, if ib_copy_to_udata fails, the CQ
won't be deleted from the radix tree and the HW (HW2SW).

Fixes: 225c7b1feef1 ('IB/mlx4: Add a driver Mellanox ConnectX InfiniBand adapters')
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

IB/mlx4: Fix incorrect MC join state bit-masking on SR-IOV

commit e5ac40cd66c2f3cd11bc5edc658f012661b16347 upstream.

Because of an incorrect bit-masking done on the join state bits, when
handling a join request we failed to detect a difference between the
group join state and the request join state when joining as send only
full member (0x8). This caused the MC join request not to be sent.
This issue is relevant only when SRIOV is enabled and SM supports
send only full member.

This fix separates scope bits and join states bits a nibble each.

Fixes: b9c5d6a64358 ('IB/mlx4: Add multicast group (MCG) paravirtualization for SR-IOV')
Signed-off-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

IB/ipoib: Don't allow MC joins during light MC flush

commit 344bacca8cd811809fc33a249f2738ab757d327f upstream.

This fix solves a race between light flush and on the fly joins.
Light flush doesn't set the device to down and unset IPOIB_OPER_UP
flag, this means that if while flushing we have a MC join in progress
and the QP was attached to BC MGID we can have a mismatches when
re-attaching a QP to the BC MGID.

The light flush would set the broadcast group to NULL causing an on
the fly join to rejoin and reattach to the BC MCG as well as adding
the BC MGID to the multicast list. The flush process would later on
remove the BC MGID and detach it from the QP. On the next flush
the BC MGID is present in the multicast list but not found when trying
to detach it because of the previous double attach and single detach.

[18332.714265] ------------[ cut here ]------------
[18332.717775] WARNING: CPU: 6 PID: 3767 at drivers/infiniband/core/verbs.c:280 ib_dealloc_pd+0xff/0x120 [ib_core]
...
[18332.775198] Hardware name: Red Hat KVM, BIOS Bochs 01/01/2011
[18332.779411]  0000000000000000 ffff8800b50dfbb0 ffffffff813fed47 0000000000000000
[18332.784960]  0000000000000000 ffff8800b50dfbf0 ffffffff8109add1 0000011832f58300
[18332.790547]  ffff880226a596c0 ffff880032482000 ffff880032482830 ffff880226a59280
[18332.796199] Call Trace:
[18332.798015]  [<ffffffff813fed47>] dump_stack+0x63/0x8c
[18332.801831]  [<ffffffff8109add1>] __warn+0xd1/0xf0
[18332.805403]  [<ffffffff8109aebd>] warn_slowpath_null+0x1d/0x20
[18332.809706]  [<ffffffffa025d90f>] ib_dealloc_pd+0xff/0x120 [ib_core]
[18332.814384]  [<ffffffffa04f3d7c>] ipoib_transport_dev_cleanup+0xfc/0x1d0 [ib_ipoib]
[18332.820031]  [<ffffffffa04ed648>] ipoib_ib_dev_cleanup+0x98/0x110 [ib_ipoib]
[18332.825220]  [<ffffffffa04e62c8>] ipoib_dev_cleanup+0x2d8/0x550 [ib_ipoib]
[18332.830290]  [<ffffffffa04e656f>] ipoib_uninit+0x2f/0x40 [ib_ipoib]
[18332.834911]  [<ffffffff81772a8a>] rollback_registered_many+0x1aa/0x2c0
[18332.839741]  [<ffffffff81772bd1>] rollback_registered+0x31/0x40
[18332.844091]  [<ffffffff81773b18>] unregister_netdevice_queue+0x48/0x80
[18332.848880]  [<ffffffffa04f489b>] ipoib_vlan_delete+0x1fb/0x290 [ib_ipoib]
[18332.853848]  [<ffffffffa04df1cd>] delete_child+0x7d/0xf0 [ib_ipoib]
[18332.858474]  [<ffffffff81520c08>] dev_attr_store+0x18/0x30
[18332.862510]  [<ffffffff8127fe4a>] sysfs_kf_write+0x3a/0x50
[18332.866349]  [<ffffffff8127f4e0>] kernfs_fop_write+0x120/0x170
[18332.870471]  [<ffffffff81207198>] __vfs_write+0x28/0xe0
[18332.874152]  [<ffffffff810e09bf>] ? percpu_down_read+0x1f/0x50
[18332.878274]  [<ffffffff81208062>] vfs_write+0xa2/0x1a0
[18332.881896]  [<ffffffff812093a6>] SyS_write+0x46/0xa0
[18332.885632]  [<ffffffff810039b7>] do_syscall_64+0x57/0xb0
[18332.889709]  [<ffffffff81883321>] entry_SYSCALL64_slow_path+0x25/0x25
[18332.894727] ---[ end trace 09ebbe31f831ef17 ]---

Fixes: ee1e2c82c245 ("IPoIB: Refresh paths instead of flushing them on SM change events")
Signed-off-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

IB/core: Fix use after free in send_leave function

commit 68c6bcdd8bd00394c234b915ab9b97c74104130c upstream.

The function send_leave sets the member: group->query_id
(group->query_id = ret) after calling the sa_query, but leave_handler
can be executed before the setting and it might delete the group object,
and will get a memory corruption.

Additionally, this patch gets rid of group->query_id variable which is
not used.

Fixes: faec2f7b96b5 ('IB/sa: Track multicast join/leave requests')
Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

IB/ipoib: Fix memory corruption in ipoib cm mode connect flow

commit 546481c2816ea3c061ee9d5658eb48070f69212e upstream.

When a new CM connection is being requested, ipoib driver copies data
from the path pointer in the CM/tx object, the path object might be
invalid at the point and memory corruption will happened later when now
the CM driver will try using that data.

The next scenario demonstrates it:
neigh_add_path --> ipoib_cm_create_tx -->
queue_work (pointer to path is in the cm/tx struct)
#while the work is still in the queue,
#the port goes down and causes the ipoib_flush_paths:
ipoib_flush_paths --> path_free --> kfree(path)
#at this point the work scheduled starts.
ipoib_cm_tx_start --> copy from the (invalid)path pointer:
(memcpy(&pathrec, &p->path->pathrec, sizeof pathrec);)
 -> memory corruption.

To fix that the driver now starts the CM/tx connection only if that
specific path exists in the general paths database.
This check is protected with the relevant locks, and uses the gid from
the neigh member in the CM/tx object which is valid according to the ref
count that was taken by the CM/tx.

Fixes: 839fcaba35 ('IPoIB: Connected mode experimental support')
Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

EDAC: Increment correct counter in edac_inc_ue_error()

commit 993f88f1cc7f0879047ff353e824e5cc8f10adfc upstream.

Fix typo in edac_inc_ue_error() to increment ue_noinfo_count instead of
ce_noinfo_count.

Signed-off-by: Emmanouil Maroudas <emmanouil.maroudas@gmail.com>
Cc: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Cc: linux-edac <linux-edac@vger.kernel.org>
Fixes: 4275be635597 ("edac: Change internal representation to work with layers")
Link: http://lkml.kernel.org/r/1461425580-5898-1-git-send-email-emmanouil.maroudas@gmail.com
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Willy Tarreau <w@1wt.eu>

timers: Use proper base migration in add_timer_on()

commit 22b886dd1018093920c4250dee2a9a3cb7cff7b8 upstream.

Regardless of the previous CPU a timer was on, add_timer_on()
currently simply sets timer->flags to the new CPU.  As the caller must
be seeing the timer as idle, this is locally fine, but the timer
leaving the old base while unlocked can lead to race conditions as
follows.

Let's say timer was on cpu 0.

  cpu 0 cpu 1
  -----------------------------------------------------------------------------
  del_timer(timer) succeeds
del_timer(timer)
  lock_timer_base(timer) locks cpu_0_base
  add_timer_on(timer, 1)
    spin_lock(&cpu_1_base->lock)
    timer->flags set to cpu_1_base
    operates on @timer   operates on @timer

This triggered with mod_delayed_work_on() which contains
"if (del_timer()) add_timer_on()" sequence eventually leading to the
following oops.

  BUG: unable to handle kernel NULL pointer dereference at           (null)
  IP: [<ffffffff810ca6e9>] detach_if_pending+0x69/0x1a0
  ...
  Workqueue: wqthrash wqthrash_workfunc [wqthrash]
  task: ffff8800172ca680 ti: ffff8800172d0000 task.ti: ffff8800172d0000
  RIP: 0010:[<ffffffff810ca6e9>]  [<ffffffff810ca6e9>] detach_if_pending+0x69/0x1a0
  ...
  Call Trace:
   [<ffffffff810cb0b4>] del_timer+0x44/0x60
   [<ffffffff8106e836>] try_to_grab_pending+0xb6/0x160
   [<ffffffff8106e913>] mod_delayed_work_on+0x33/0x80
   [<ffffffffa0000081>] wqthrash_workfunc+0x61/0x90 [wqthrash]
   [<ffffffff8106dba8>] process_one_work+0x1e8/0x650
   [<ffffffff8106e05e>] worker_thread+0x4e/0x450
   [<ffffffff810746af>] kthread+0xef/0x110
   [<ffffffff8185980f>] ret_from_fork+0x3f/0x70

Fix it by updating add_timer_on() to perform proper migration as
__mod_timer() does.

Mike: apply tglx backport

Reported-and-tested-by: Jeff Layton <jlayton@poochiereds.net>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Chris Worley <chris.worley@primarydata.com>
Cc: bfields@fieldses.org
Cc: Michael Skralivetsky <michael.skralivetsky@primarydata.com>
Cc: Trond Myklebust <trond.myklebust@primarydata.com>
Cc: Shaohua Li <shli@fb.com>
Cc: Jeff Layton <jlayton@poochiereds.net>
Cc: kernel-team@fb.com
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/20151029103113.2f893924@tlielax.poochiereds.net
Link: http://lkml.kernel.org/r/20151104171533.GI5749@mtj.duckdns.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Mike Galbraith <mgalbraith@suse.de>
Signed-off-by: Willy Tarreau <w@1wt.eu>

cdc-acm: fix wrong pipe type on rx interrupt xfers

commit add125054b8727103631dce116361668436ef6a7 upstream.

This fixes the "BOGUS urb xfer" warning logged by usb_submit_urb().

Signed-off-by: Gavin Li <git@thegavinli.com>
Acked-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

thermal: hwmon: Properly report critical temperature in sysfs

commit f37fabb8643eaf8e3b613333a72f683770c85eca upstream.

In the critical sysfs entry the thermal hwmon was returning wrong
temperature to the user-space.  It was reporting the temperature of the
first trip point instead of the temperature of critical trip point.

For example:
/sys/class/hwmon/hwmon0/temp1_crit:50000
/sys/class/thermal/thermal_zone0/trip_point_0_temp:50000
/sys/class/thermal/thermal_zone0/trip_point_0_type:active
/sys/class/thermal/thermal_zone0/trip_point_3_temp:120000
/sys/class/thermal/thermal_zone0/trip_point_3_type:critical

Since commit e68b16abd91d ("thermal: add hwmon sysfs I/F") the driver
have been registering a sysfs entry if get_crit_temp() callback was
provided.  However when accessed, it was calling get_trip_temp() instead
of the get_crit_temp().

Fixes: e68b16abd91d ("thermal: add hwmon sysfs I/F")
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
[wt: s/thermal_hwmon.c/thermal_core.c in 3.10]

Signed-off-by: Willy Tarreau <w@1wt.eu>

iio: accel: kxsd9: Fix scaling bug

commit 307fe9dd11ae44d4f8881ee449a7cbac36e1f5de upstream.

All the scaling of the KXSD9 involves multiplication with a
fraction number < 1.

However the scaling value returned from IIO_INFO_SCALE was
unpredictable as only the micros of the value was assigned, and
not the integer part, resulting in scaling like this:

$cat in_accel_scale
-1057462640.011978

Fix this by assigning zero to the integer part.

Tested-by: Jonathan Cameron <jic23@kernel.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Jonathan Cameron <jic23@kernel.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

iio: accel: kxsd9: Fix raw read return

commit 7ac61a062f3147dc23e3f12b9dfe7c4dd35f9cb8 upstream.

Any readings from the raw interface of the KXSD9 driver will
return an empty string, because it does not return
IIO_VAL_INT but rather some random value from the accelerometer
to the caller.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Jonathan Cameron <jic23@kernel.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

i2c: at91: fix write transfers by clearing pending interrupt first

commit 6f6ddbb09d2a5baded0e23add3ad2d9e9417ab30 upstream.

In some cases a NACK interrupt may be pending in the Status Register (SR)
as a result of a previous transfer. However at91_do_twi_transfer() did not
read the SR to clear pending interruptions before starting a new transfer.
Hence a NACK interrupt rose as soon as it was enabled again at the I2C
controller level, resulting in a wrong sequence of operations and strange
patterns of behaviour on the I2C bus, such as a clock stretch followed by
a restart of the transfer.

This first issue occurred with both DMA and PIO write transfers.

Also when a NACK error was detected during a PIO write transfer, the
interrupt handler used to wrongly start a new transfer by writing into the
Transmit Holding Register (THR). Then the I2C slave was likely to reply
with a second NACK.

This second issue is fixed in atmel_twi_interrupt() by handling the TXRDY
status bit only if both the TXCOMP and NACK status bits are cleared.

Tested with a at24 eeprom on sama5d36ek board running a linux-4.1-at91
kernel image. Adapted to linux-next.

Reported-by: Peter Rosin <peda@lysator.liu.se>
Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Tested-by: Peter Rosin <peda@lysator.liu.se>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Fixes: 93563a6a71bb ("i2c: at91: fix a race condition when using the DMA controller")
Signed-off-by: Willy Tarreau <w@1wt.eu>

i2c: core: fix NULL pointer dereference under race condition

commit 147b36d5b70c083cc76770c47d60b347e8eaf231 upstream.

Race condition between registering an I2C device driver and
deregistering an I2C adapter device which is assumed to manage that
I2C device may lead to a NULL pointer dereference due to the
uninitialized list head of driver clients.

The root cause of the issue is that the I2C bus may know about the
registered device driver and thus it is matched by bus_for_each_drv(),
but the list of clients is not initialized and commonly it is NULL,
because I2C device drivers define struct i2c_driver as static and
clients field is expected to be initialized by I2C core:

  i2c_register_driver()             i2c_del_adapter()
    driver_register()                 ...
      bus_add_driver()                ...
        ...                           bus_for_each_drv(..., __process_removed_adapter)
      ...                               i2c_do_del_adapter()
    ...                                   list_for_each_entry_safe(..., &driver->clients, ...)
    INIT_LIST_HEAD(&driver->clients);

To solve the problem it is sufficient to do clients list head
initialization before calling driver_register().

The problem was found while using an I2C device driver with a sluggish
registration routine on a bus provided by a physically detachable I2C
master controller, but practically the oops may be reproduced under
the race between arbitraty I2C device driver registration and managing
I2C bus device removal e.g. by unbinding the latter over sysfs:

% echo 21a4000.i2c > /sys/bus/platform/drivers/imx-i2c/unbind
  Unable to handle kernel NULL pointer dereference at virtual address 00000000
  Internal error: Oops: 17 [#1] SMP ARM
  CPU: 2 PID: 533 Comm: sh Not tainted 4.9.0-rc3+ #61
  Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
  task: e5ada400 task.stack: e4936000
  PC is at i2c_do_del_adapter+0x20/0xcc
  LR is at __process_removed_adapter+0x14/0x1c
  Flags: NzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
  Control: 10c5387d  Table: 35bd004a  DAC: 00000051
  Process sh (pid: 533, stack limit = 0xe4936210)
  Stack: (0xe4937d28 to 0xe4938000)
  Backtrace:
  [<c0667be0>] (i2c_do_del_adapter) from [<c0667cc0>] (__process_removed_adapter+0x14/0x1c)
  [<c0667cac>] (__process_removed_adapter) from [<c0516998>] (bus_for_each_drv+0x6c/0xa0)
  [<c051692c>] (bus_for_each_drv) from [<c06685ec>] (i2c_del_adapter+0xbc/0x284)
  [<c0668530>] (i2c_del_adapter) from [<bf0110ec>] (i2c_imx_remove+0x44/0x164 [i2c_imx])
  [<bf0110a8>] (i2c_imx_remove [i2c_imx]) from [<c051a838>] (platform_drv_remove+0x2c/0x44)
  [<c051a80c>] (platform_drv_remove) from [<c05183d8>] (__device_release_driver+0x90/0x12c)
  [<c0518348>] (__device_release_driver) from [<c051849c>] (device_release_driver+0x28/0x34)
  [<c0518474>] (device_release_driver) from [<c0517150>] (unbind_store+0x80/0x104)
  [<c05170d0>] (unbind_store) from [<c0516520>] (drv_attr_store+0x28/0x34)
  [<c05164f8>] (drv_attr_store) from [<c0298acc>] (sysfs_kf_write+0x50/0x54)
  [<c0298a7c>] (sysfs_kf_write) from [<c029801c>] (kernfs_fop_write+0x100/0x214)
  [<c0297f1c>] (kernfs_fop_write) from [<c0220130>] (__vfs_write+0x34/0x120)
  [<c02200fc>] (__vfs_write) from [<c0221088>] (vfs_write+0xa8/0x170)
  [<c0220fe0>] (vfs_write) from [<c0221e74>] (SyS_write+0x4c/0xa8)
  [<c0221e28>] (SyS_write) from [<c0108a20>] (ret_fast_syscall+0x0/0x1c)

Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Willy Tarreau <w@1wt.eu>

em28xx-i2c: rt_mutex_trylock() returns zero on failure

commit e44c153b30c9a0580fc2b5a93f3c6d593def2278 upstream.

The code is checking for negative returns but it should be checking for
zero.

Fixes: aab3125c43d8 ('[media] em28xx: add support for registering multiple i2c buses')

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

i2c-eg20t: fix race between i2c init and interrupt enable

commit 371a015344b6e270e7e3632107d9554ec6d27a6b upstream.

the eg20t driver call request_irq() function before the pch_base_address,
base address of i2c controller's register, is assigned an effective value.

there is one possible scenario that an interrupt which isn't inside eg20t
arrives immediately after request_irq() is executed when i2c controller
shares an interrupt number with others. since the interrupt handler
pch_i2c_handler() has already active as shared action, it will be called
and read its own register to determine if this interrupt is from itself.

At that moment, since base address of i2c registers is not remapped
in kernel space yet,so the INT handler will access an illegal address
and then a error occurs.

Signed-off-by: Yadi.hu <yadi.hu@windriver.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Willy Tarreau <w@1wt.eu>

hwrng: omap - Only fail if pm_runtime_get_sync returns < 0

commit ad8529fde9e3601180a839867a8ab041109aebb5 upstream.

Currently omap-rng checks the return value of pm_runtime_get_sync and
reports failure if anything is returned, however it should be checking
if ret < 0 as pm_runtime_get_sync return 0 on success but also can return
1 if the device was already active which is not a failure case. Only
values < 0 are actual failures.

Fixes: 61dc0a446e5d ("hwrng: omap - Fix assumption that runtime_get_sync will always succeed")
Signed-off-by: Dave Gerlach <d-gerlach@ti.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Willy Tarreau <w@1wt.eu>

hwrng: omap - Fix assumption that runtime_get_sync will always succeed

commit 61dc0a446e5d08f2de8a24b45f69a1e302bb1b1b upstream.

pm_runtime_get_sync does return a error value that must be checked for
error conditions, else, due to various reasons, the device maynot be
enabled and the system will crash due to lack of clock to the hardware
module.

Before:
12.562784] [00000000] *pgd=fe193835
12.562792] Internal error: : 1406 [#1] SMP ARM
[...]
12.562864] CPU: 1 PID: 241 Comm: modprobe Not tainted 4.7.0-rc4-next-20160624 #2
12.562867] Hardware name: Generic DRA74X (Flattened Device Tree)
12.562872] task: ed51f140 ti: ed44c000 task.ti: ed44c000
12.562886] PC is at omap4_rng_init+0x20/0x84 [omap_rng]
12.562899] LR is at set_current_rng+0xc0/0x154 [rng_core]
[...]

After the proper checks:
[   94.366705] omap_rng 48090000.rng: _od_fail_runtime_resume: FIXME:
missing hwmod/omap_dev info
[   94.375767] omap_rng 48090000.rng: Failed to runtime_get device -19
[   94.382351] omap_rng 48090000.rng: initialization failed.

Fixes: 665d92fa85b5 ("hwrng: OMAP: convert to use runtime PM")
Cc: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
[wt: adjusted context for pre-3.12-rc1 kernels]

Signed-off-by: Willy Tarreau <w@1wt.eu>

hwrng: exynos - Disable runtime PM on probe failure

commit 48a61e1e2af8020f11a2b8f8dc878144477623c6 upstream.

Add proper error path (for disabling runtime PM) when registering of
hwrng fails.

Fixes: b329669ea0b5 ("hwrng: exynos - Add support for Exynos random number generator")
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Willy Tarreau <w@1wt.eu>

Input: ili210x - fix permissions on "calibrate" attribute

commit b27c0d0c3bf3073e8ae19875eb1d3755c5e8c072 upstream.

"calibrate" attribute does not provide "show" methods and thus we should
not mark it as readable.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

Input: i8042 - set up shared ps2_cmd_mutex for AUX ports

commit 47af45d684b5f3ae000ad448db02ce4f13f73273 upstream.

The commit 4097461897df ("Input: i8042 - break load dependency ...")
correctly set up ps2_cmd_mutex pointer for the KBD port but forgot to do
the same for AUX port(s), which results in communication on KBD and AUX
ports to clash with each other.

Fixes: 4097461897df ("Input: i8042 - break load dependency ...")
Reported-by: Bruno Wolff III <bruno@wolff.to>
Tested-by: Bruno Wolff III <bruno@wolff.to>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

Input: i8042 - break load dependency between atkbd/psmouse and i8042

commit 4097461897df91041382ff6fcd2bfa7ee6b2448c upstream.

As explained in 1407814240-4275-1-git-send-email-decui@microsoft.com we
have a hard load dependency between i8042 and atkbd which prevents
keyboard from working on Gen2 Hyper-V VMs.

> hyperv_keyboard invokes serio_interrupt(), which needs a valid serio
> driver like atkbd.c.  atkbd.c depends on libps2.c because it invokes
> ps2_command().  libps2.c depends on i8042.c because it invokes
> i8042_check_port_owner().  As a result, hyperv_keyboard actually
> depends on i8042.c.
>
> For a Generation 2 Hyper-V VM (meaning no i8042 device emulated), if a
> Linux VM (like Arch Linux) happens to configure CONFIG_SERIO_I8042=m
> rather than =y, atkbd.ko can't load because i8042.ko can't load(due to
> no i8042 device emulated) and finally hyperv_keyboard can't work and
> the user can't input: https://bugs.archlinux.org/task/39820
> (Ubuntu/RHEL/SUSE aren't affected since they use CONFIG_SERIO_I8042=y)

To break the dependency we move away from using i8042_check_port_owner()
and instead allow serio port owner specify a mutex that clients should use
to serialize PS/2 command stream.

Reported-by: Mark Laws <mdl@60hz.org>
Tested-by: Mark Laws <mdl@60hz.org>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

qxl: check for kmap failures

commit f4cceb2affcd1285d4ce498089e8a79f4cd2fa66 upstream.

If kmap fails, it leads to memory corruption.

Fixes: f64122c1f6ad ('drm: add new QXL driver. (v1.4)')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20160711084633.GA31411@mwanda
Signed-off-by: Willy Tarreau <w@1wt.eu>

drm/radeon: Ensure vblank interrupt is enabled on DPMS transition to on

NOTE: This patch only applies to 4.5.y or older kernels. With newer
kernels, this problem cannot happen because the driver now uses
drm_crtc_vblank_on/off instead of drm_vblank_pre/post_modeset[0]. I
consider this patch safer for older kernels than backporting the API
change, because drm_crtc_vblank_on/off had various issues in older
kernels, and I'm not sure all fixes for those have been backported to
all stable branches where this patch could be applied.

    ---------------------

Fixes the vblank interrupt being disabled when it should be on, which
can cause at least the following symptoms:

* Hangs when running 'xset dpms force off' in a GNOME session with
  gnome-shell using DRI2.
* RandR 1.4 slave outputs freezing with garbage displayed using
  xf86-video-ati 7.8.0 or newer.

[0] See upstream commit:

commit 777e3cbc791f131806d9bf24b3325637c7fc228d
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Thu Jan 21 11:08:57 2016 +0100

    drm/radeon: Switch to drm_vblank_on/off

Reported-and-Tested-by: Max Staudt <mstaudt@suse.de>
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

drm: Reject page_flip for !DRIVER_MODESET

commit 6f00975c619064a18c23fd3aced325ae165a73b9 upstream.

Somehow this one slipped through, which means drivers without modeset
support can be oopsed (since those also don't call
drm_mode_config_init, which means the crtc lookup will chase an
uninitalized idr).

Reported-by: Alexander Potapenko <glider@google.com>
Cc: Alexander Potapenko <glider@google.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

drm/radeon: fix radeon_move_blit on 32bit systems

commit 13f479b9df4e2bbf2d16e7e1b02f3f55f70e2455 upstream.

This bug seems to be present for a very long time.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

driver core: fix race between creating/querying glue dir and its cleanup

commit cebf8fd16900fdfd58c0028617944f808f97fe50 upstream.

The global mutex of 'gdp_mutex' is used to serialize creating/querying
glue dir and its cleanup. Turns out it isn't a perfect way because
part(kobj_kset_leave()) of the actual cleanup action() is done inside
the release handler of the glue dir kobject. That means gdp_mutex has
to be held before releasing the last reference count of the glue dir
kobject.

This patch moves glue dir's cleanup after kobject_del() in device_del()
for avoiding the race.

Cc: Yijing Wang <wangyijing@huawei.com>
Reported-by: Chandra Sekhar Lingutla <clingutla@codeaurora.org>
Signed-off-by: Ming Lei <ming.lei@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

driver core: Delete an unnecessary check before the function call "put_device"

commit 5f0163a5ee9cc7c59751768bdfd94a73186debba upstream.

The put_device() function tests whether its argument is NULL and then
returns immediately. Thus the test around the call is not needed.

This issue was detected by using the Coccinelle software.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[wt: backported only to ease next patch as suggested by Jiri]

Signed-off-by: Willy Tarreau <w@1wt.eu>

hostfs: Freeing an ERR_PTR in hostfs_fill_sb_common()

commit 8a545f185145e3c09348cd74326268ecfc6715a3 upstream.

We can't pass error pointers to kfree() or it causes an oops.

Fixes: 52b209f7b848 ('get rid of hostfs_read_inode()')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Willy Tarreau <w@1wt.eu>

isofs: Do not return EACCES for unknown filesystems

commit a2ed0b391dd9c3ef1d64c7c3e370f4a5ffcd324a upstream.

When isofs_mount() is called to mount a device read-write, it returns
EACCES even before it checks that the device actually contains an isofs
filesystem. This may confuse mount(8) which then tries to mount all
subsequent filesystem types in read-only mode.

Fix the problem by returning EACCES only once we verify that the device
indeed contains an iso9660 filesystem.

Fixes: 17b7f7cf58926844e1dd40f5eb5348d481deca6a
Reported-by: Kent Overstreet <kent.overstreet@gmail.com>
Reported-by: Karel Zak <kzak@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Willy Tarreau <w@1wt.eu>

fs/super.c: fix race between freeze_super() and thaw_super()

commit 89f39af129382a40d7cd1f6914617282cfeee28e upstream.

Change thaw_super() to check frozen != SB_FREEZE_COMPLETE rather than
frozen == SB_UNFROZEN, otherwise it can race with freeze_super() which
drops sb->s_umount after SB_FREEZE_WRITE to preserve the lock ordering.

In this case thaw_super() will wrongly call s_op->unfreeze_fs() before
it was actually frozen, and call sb_freeze_unlock() which leads to the
unbalanced percpu_up_write(). Unfortunately lockdep can't detect this,
so this triggers misc BUG_ON()'s in kernel/rcu/sync.c.

Reported-and-tested-by: Nikolay Borisov <kernel@kyup.com>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Willy Tarreau <w@1wt.eu>

fs/seq_file: fix out-of-bounds read

commit 088bf2ff5d12e2e32ee52a4024fec26e582f44d3 upstream.

seq_read() is a nasty piece of work, not to mention buggy.

It has (I think) an old bug which allows unprivileged userspace to read
beyond the end of m->buf.

I was getting these:

    BUG: KASAN: slab-out-of-bounds in seq_read+0xcd2/0x1480 at addr ffff880116889880
    Read of size 2713 by task trinity-c2/1329
    CPU: 2 PID: 1329 Comm: trinity-c2 Not tainted 4.8.0-rc1+ #96
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.9.3-0-ge2fc41e-prebuilt.qemu-project.org 04/01/2014
    Call Trace:
      kasan_object_err+0x1c/0x80
      kasan_report_error+0x2cb/0x7e0
      kasan_report+0x4e/0x80
      check_memory_region+0x13e/0x1a0
      kasan_check_read+0x11/0x20
      seq_read+0xcd2/0x1480
      proc_reg_read+0x10b/0x260
      do_loop_readv_writev.part.5+0x140/0x2c0
      do_readv_writev+0x589/0x860
      vfs_readv+0x7b/0xd0
      do_readv+0xd8/0x2c0
      SyS_readv+0xb/0x10
      do_syscall_64+0x1b3/0x4b0
      entry_SYSCALL64_slow_path+0x25/0x25
    Object at ffff880116889100, in cache kmalloc-4096 size: 4096
    Allocated:
    PID = 1329
      save_stack_trace+0x26/0x80
      save_stack+0x46/0xd0
      kasan_kmalloc+0xad/0xe0
      __kmalloc+0x1aa/0x4a0
      seq_buf_alloc+0x35/0x40
      seq_read+0x7d8/0x1480
      proc_reg_read+0x10b/0x260
      do_loop_readv_writev.part.5+0x140/0x2c0
      do_readv_writev+0x589/0x860
      vfs_readv+0x7b/0xd0
      do_readv+0xd8/0x2c0
      SyS_readv+0xb/0x10
      do_syscall_64+0x1b3/0x4b0
      return_from_SYSCALL_64+0x0/0x6a
    Freed:
    PID = 0
    (stack is not available)
    Memory state around the buggy address:
     ffff88011688a000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
     ffff88011688a080: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
    >ffff88011688a100: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
       ^
     ffff88011688a180: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
     ffff88011688a200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
    ==================================================================
    Disabling lock debugging due to kernel taint

This seems to be the same thing that Dave Jones was seeing here:

  https://lkml.org/lkml/2016/8/12/334

There are multiple issues here:

  1) If we enter the function with a non-empty buffer, there is an attempt
     to flush it. But it was not clearing m->from after doing so, which
     means that if we try to do this flush twice in a row without any call
     to traverse() in between, we are going to be reading from the wrong
     place -- the splat above, fixed by this patch.

  2) If there's a short write to userspace because of page faults, the
     buffer may already contain multiple lines (i.e. pos has advanced by
     more than 1), but we don't save the progress that was made so the
     next call will output what we've already returned previously. Since
     that is a much less serious issue (and I have a headache after
     staring at seq_read() for the past 8 hours), I'll leave that for now.

Link: http://lkml.kernel.org/r/1471447270-32093-1-git-send-email-vegard.nossum@oracle.com
Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

NFSv4: Open state recovery must account for file permission changes

commit 304020fe48c6c7fff8b5a38f382b54404f0f79d3 upstream.

If the file permissions change on the server, then we may not be able to
recover open state. If so, we need to ensure that we mark the file
descriptor appropriately.

Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Tested-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

NFS: Don't drop CB requests with invalid principals

commit a4e187d83d88eeaba6252aac0a2ffe5eaa73a818 upstream.

Before commit 778be232a207 ("NFS do not find client in NFSv4
pg_authenticate"), the Linux callback server replied with
RPC_AUTH_ERROR / RPC_AUTH_BADCRED, instead of dropping the CB
request. Let's restore that behavior so the server has a chance to
do something useful about it, and provide a warning that helps
admins correct the problem.

Fixes: 778be232a207 ("NFS do not find client in NFSv4 ...")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

NFSD: Using free_conn free connection

commit 3f42d2c428c724212c5f4249daea97e254eb0546 upstream.

Connection from alloc_conn must be freed through free_conn,
otherwise, the reference of svc_xprt will never be put.

Signed-off-by: Kinglong Mee <kinglongmee@gmail.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

NFSv4.x: Fix a refcount leak in nfs_callback_up_net

commit 98b0f80c2396224bbbed81792b526e6c72ba9efa upstream.

On error, the callers expect us to return without bumping
nn->cb_users[].

Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

UBI: fastmap: scrub PEB when bitflips are detected in a free PEB EC header

commit ecbfa8eabae9cd73522d1d3d15869703c263d859 upstream.

scan_pool() does not mark the PEB for scrubing when bitflips are
detected in the EC header of a free PEB (VID header region left to
0xff).
Make sure we scrub the PEB in this case.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Fixes: dbb7d2a88d2a ("UBI: Add fastmap core")
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ubifs: Fix regression in ubifs_readdir()

commit a00052a296e54205cf238c75bd98d17d5d02a6db upstream.

Commit c83ed4c9dbb35 ("ubifs: Abort readdir upon error") broke
overlayfs support because the fix exposed an internal error
code to VFS.

Reported-by: Peter Rosin <peda@axentia.se>
Tested-by: Peter Rosin <peda@axentia.se>
Reported-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Tested-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Fixes: c83ed4c9dbb35 ("ubifs: Abort readdir upon error")
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ubifs: Abort readdir upon error

commit c83ed4c9dbb358b9e7707486e167e940d48bfeed upstream.

If UBIFS is facing an error while walking a directory, it reports this
error and ubifs_readdir() returns the error code. But the VFS readdir
logic does not make the getdents system call fail in all cases. When the
readdir cursor indicates that more entries are present, the system call
will just return and the libc wrapper will try again since it also
knows that more entries are present.
This causes the libc wrapper to busy loop for ever when a directory is
corrupted on UBIFS.
A common approach do deal with corrupted directory entries is
skipping them by setting the cursor to the next entry. On UBIFS this
approach is not possible since we cannot compute the next directory
entry cursor position without reading the current entry. So all we can
do is setting the cursor to the "no more entries" position and make
getdents exit.

Signed-off-by: Richard Weinberger <richard@nod.at>
[wt: adjusted context]

Signed-off-by: Willy Tarreau <w@1wt.eu>

UBIFS: Fix possible memory leak in ubifs_readdir()

commit aeeb14f763917ccf639a602cfbeee6957fd944a2 upstream.

If ubifs_tnc_next_ent() returns something else than -ENOENT
we leak file->private_data.

Signed-off-by: Richard Weinberger <richard@nod.at>
Reviewed-by: David Gstir <david@sigma-star.at>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ubifs: Fix xattr_names length in exit paths

commit 843741c5778398ea67055067f4cc65ae6c80ca0e upstream.

When the operation fails we also have to undo the changes
we made to ->xattr_names. Otherwise listxattr() will report
wrong lengths.

Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ubifs: Fix assertion in layout_in_gaps()

commit c0082e985fdf77b02fc9e0dac3b58504dcf11b7a upstream.

An assertion in layout_in_gaps() verifies that the gap_lebs pointer is
below the maximum bound. When computing this maximum bound the idx_lebs
count is multiplied by sizeof(int), while C pointers arithmetic does take
into account the size of the pointed elements implicitly already. Remove
the multiplication to fix the assertion.

Fixes: 1e51764a3c2ac05a ("UBIFS: add new flash file system")
Signed-off-by: Vincent Stehlé <vincent.stehle@intel.com>
Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ocfs2: fix start offset to ocfs2_zero_range_for_truncate()

commit d21c353d5e99c56cdd5b5c1183ffbcaf23b8b960 upstream.

If we punch a hole on a reflink such that following conditions are met:

1. start offset is on a cluster boundary
2. end offset is not on a cluster boundary
3. (end offset is somewhere in another extent) or
   (hole range > MAX_CONTIG_BYTES(1MB)),

we dont COW the first cluster starting at the start offset.  But in this
case, we were wrongly passing this cluster to
ocfs2_zero_range_for_truncate() to zero out.  This will modify the
cluster in place and zero it in the source too.

Fix this by skipping this cluster in such a scenario.

To reproduce:

1. Create a random file of say 10 MB
     xfs_io -c 'pwrite -b 4k 0 10M' -f 10MBfile
2. Reflink  it
     reflink -f 10MBfile reflnktest
3. Punch a hole at starting at cluster boundary  with range greater that
1MB. You can also use a range that will put the end offset in another
extent.
     fallocate -p -o 0 -l 1048615 reflnktest
4. sync
5. Check the  first cluster in the source file. (It will be zeroed out).
    dd if=10MBfile iflag=direct bs=<cluster size> count=1 | hexdump -C

Link: http://lkml.kernel.org/r/1470957147-14185-1-git-send-email-ashish.samant@oracle.com
Signed-off-by: Ashish Samant <ashish.samant@oracle.com>
Reported-by: Saar Maoz <saar.maoz@oracle.com>
Reviewed-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Cc: Mark Fasheh <mfasheh@versity.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <joseph.qi@huawei.com>
Cc: Eric Ren <zren@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ocfs2/dlm: fix race between convert and migration

commit e6f0c6e6170fec175fe676495f29029aecdf486c upstream.

Commit ac7cf246dfdb ("ocfs2/dlm: fix race between convert and recovery")
checks if lockres master has changed to identify whether new master has
finished recovery or not.  This will introduce a race that right after
old master does umount ( means master will change), a new convert
request comes.

In this case, it will reset lockres state to DLM_RECOVERING and then
retry convert, and then fail with lockres->l_action being set to
OCFS2_AST_INVALID, which will cause inconsistent lock level between
ocfs2 and dlm, and then finally BUG.

Since dlm recovery will clear lock->convert_pending in
dlm_move_lockres_to_recovery_list, we can use it to correctly identify
the race case between convert and recovery.  So fix it.

Fixes: ac7cf246dfdb ("ocfs2/dlm: fix race between convert and recovery")
Link: http://lkml.kernel.org/r/57CE1569.8010704@huawei.com
Signed-off-by: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Jun Piao <piaojun@huawei.com>
Cc: Mark Fasheh <mfasheh@versity.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

btrfs: ensure that file descriptor used with subvol ioctls is a dir

commit 325c50e3cebb9208009083e841550f98a863bfa0 upstream.

If the subvol/snapshot create/destroy ioctls are passed a regular file
with execute permissions set, we'll eventually Oops while trying to do
inode->i_op->lookup via lookup_one_len.

This patch ensures that the file descriptor refers to a directory.

Fixes: cb8e70901d (Btrfs: Fix subvolume creation locking rules)
Fixes: 76dda93c6a (Btrfs: add snapshot/subvolume destroy ioctl)
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

libxfs: clean up _calc_dquots_per_chunk

commit 58d789678546d46d7bbd809dd7dab417c0f23655 upstream.

The function xfs_calc_dquots_per_chunk takes a parameter in units
of basic blocks.  The kernel seems to get the units wrong, but
userspace got 'fixed' by commenting out the unnecessary conversion.
Fix both.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

xfs: fix superblock inprogress check

commit f3d7ebdeb2c297bd26272384e955033493ca291c upstream.

From inspection, the superblock sb_inprogress check is done in the
verifier and triggered only for the primary superblock via a
"bp->b_bn == XFS_SB_DADDR" check.

Unfortunately, the primary superblock is an uncached buffer, and
hence it is configured by xfs_buf_read_uncached() with:

bp->b_bn = XFS_BUF_DADDR_NULL;  /* always null for uncached buffers */

And so this check never triggers. Fix it.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
[wt: s/xfs_sb.c/xfs_mount.c in 3.10]

Signed-off-by: Willy Tarreau <w@1wt.eu>

reiserfs: Unlock superblock before calling reiserfs_quota_on_mount()

commit 420902c9d086848a7548c83e0a49021514bd71b7 upstream.

If we hold the superblock lock while calling reiserfs_quota_on_mount(), we can
deadlock our own worker - mount blocks kworker/3:2, sleeps forever more.

crash> ps|grep UN
    715      2   3  ffff880220734d30  UN   0.0       0      0  [kworker/3:2]
   9369   9341   2  ffff88021ffb7560  UN   1.3  493404 123184  Xorg
   9665   9664   3  ffff880225b92ab0  UN   0.0   47368    812  udisks-daemon
  10635  10403   3  ffff880222f22c70  UN   0.0   14904    936  mount
crash> bt ffff880220734d30
PID: 715    TASK: ffff880220734d30  CPU: 3   COMMAND: "kworker/3:2"
 #0 [ffff8802244c3c20] schedule at ffffffff8144584b
 #1 [ffff8802244c3cc8] __rt_mutex_slowlock at ffffffff814472b3
 #2 [ffff8802244c3d28] rt_mutex_slowlock at ffffffff814473f5
 #3 [ffff8802244c3dc8] reiserfs_write_lock at ffffffffa05f28fd [reiserfs]
 #4 [ffff8802244c3de8] flush_async_commits at ffffffffa05ec91d [reiserfs]
 #5 [ffff8802244c3e08] process_one_work at ffffffff81073726
 #6 [ffff8802244c3e68] worker_thread at ffffffff81073eba
 #7 [ffff8802244c3ec8] kthread at ffffffff810782e0
 #8 [ffff8802244c3f48] kernel_thread_helper at ffffffff81450064
crash> rd ffff8802244c3cc8 10
ffff8802244c3cc8:  ffffffff814472b3 ffff880222f23250   .rD.....P2."....
ffff8802244c3cd8:  0000000000000000 0000000000000286   ................
ffff8802244c3ce8:  ffff8802244c3d30 ffff880220734d80   0=L$.....Ms ....
ffff8802244c3cf8:  ffff880222e8f628 0000000000000000   (.."............
ffff8802244c3d08:  0000000000000000 0000000000000002   ................
crash> struct rt_mutex ffff880222e8f628
struct rt_mutex {
  wait_lock = {
    raw_lock = {
      slock = 65537
    }
  },
  wait_list = {
    node_list = {
      next = 0xffff8802244c3d48,
      prev = 0xffff8802244c3d48
    }
  },
  owner = 0xffff880222f22c71,
  save_state = 0
}
crash> bt 0xffff880222f22c70
PID: 10635  TASK: ffff880222f22c70  CPU: 3   COMMAND: "mount"
 #0 [ffff8802216a9868] schedule at ffffffff8144584b
 #1 [ffff8802216a9910] schedule_timeout at ffffffff81446865
 #2 [ffff8802216a99a0] wait_for_common at ffffffff81445f74
 #3 [ffff8802216a9a30] flush_work at ffffffff810712d3
 #4 [ffff8802216a9ab0] schedule_on_each_cpu at ffffffff81074463
 #5 [ffff8802216a9ae0] invalidate_bdev at ffffffff81178aba
 #6 [ffff8802216a9af0] vfs_load_quota_inode at ffffffff811a3632
 #7 [ffff8802216a9b50] dquot_quota_on_mount at ffffffff811a375c
 #8 [ffff8802216a9b80] finish_unfinished at ffffffffa05dd8b0 [reiserfs]
 #9 [ffff8802216a9cc0] reiserfs_fill_super at ffffffffa05de825 [reiserfs]
    RIP: 00007f7b9303997a  RSP: 00007ffff443c7a8  RFLAGS: 00010202
    RAX: 00000000000000a5  RBX: ffffffff8144ef12  RCX: 00007f7b932e9ee0
    RDX: 00007f7b93d9a400  RSI: 00007f7b93d9a3e0  RDI: 00007f7b93d9a3c0
    RBP: 00007f7b93d9a2c0   R8: 00007f7b93d9a550   R9: 0000000000000001
    R10: ffffffffc0ed040e  R11: 0000000000000202  R12: 000000000000040e
    R13: 0000000000000000  R14: 00000000c0ed040e  R15: 00007ffff443ca20
    ORIG_RAX: 00000000000000a5  CS: 0033  SS: 002b

Signed-off-by: Mike Galbraith <efault@gmx.de>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Mike Galbraith <mgalbraith@suse.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Willy Tarreau <w@1wt.eu>

reiserfs: fix "new_insert_key may be used uninitialized ..."

commit 0a11b9aae49adf1f952427ef1a1d9e793dd6ffb6 upstream.

new_insert_key only makes any sense when it's associated with a
new_insert_ptr, which is initialized to NULL and changed to a
buffer_head when we also initialize new_insert_key.  We can key off of
that to avoid the uninitialized warning.

Link: http://lkml.kernel.org/r/5eca5ffb-2155-8df2-b4a2-f162f105efed@suse.com
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Jan Kara <jack@suse.cz>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ext4: sanity check the block and cluster size at mount time

commit 8cdf3372fe8368f56315e66bea9f35053c418093 upstream.

If the block size or cluster size is insane, reject the mount.  This
is important for security reasons (although we shouldn't be just
depending on this check).

Ref: http://www.securityfocus.com/archive/1/539661
Ref: https://bugzilla.redhat.com/show_bug.cgi?id=1332506
Reported-by: Borislav Petkov <bp@alien8.de>
Reported-by: Nikolay Borisov <kernel@kyup.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ext4: allow DAX writeback for hole punch

commit cca32b7eeb4ea24fa6596650e06279ad9130af98 upstream.

Currently when doing a DAX hole punch with ext4 we fail to do a writeback.
This is because the logic around filemap_write_and_wait_range() in
ext4_punch_hole() only looks for dirty page cache pages in the radix tree,
not for dirty DAX exceptional entries.

Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ext4: reinforce check of i_dtime when clearing high fields of uid and gid

commit 93e3b4e6631d2a74a8cf7429138096862ff9f452 upstream.

Now, ext4_do_update_inode() clears high 16-bit fields of uid/gid
of deleted and evicted inode to fix up interoperability with old
kernels. However, it checks only i_dtime of an inode to determine
whether the inode was deleted and evicted, and this is very risky,
because i_dtime can be used for the pointer maintaining orphan inode
list, too. We need to further check whether the i_dtime is being
used for the orphan inode list even if the i_dtime is not NULL.

We found that high 16-bit fields of uid/gid of inode are unintentionally
and permanently cleared when the inode truncation is just triggered,
but not finished, and the inode metadata, whose high uid/gid bits are
cleared, is written on disk, and the sudden power-off follows that
in order.

Signed-off-by: Daeho Jeong <daeho.jeong@samsung.com>
Signed-off-by: Hobin Woo <hobin.woo@samsung.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ext4: use __GFP_NOFAIL in ext4_free_blocks()

commit adb7ef600cc9d9d15ecc934cc26af5c1379777df upstream.

This might be unexpected but pages allocated for sbi->s_buddy_cache are
charged to current memory cgroup. So, GFP_NOFS allocation could fail if
current task has been killed by OOM or if current memory cgroup has no
free memory left. Block allocator cannot handle such failures here yet.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ext4: avoid modifying checksum fields directly during checksum verification

commit b47820edd1634dc1208f9212b7ecfb4230610a23 upstream.

We temporally change checksum fields in buffers of some types of
metadata into '0' for verifying the checksum values. By doing this
without locking the buffer, some metadata's checksums, which are
being committed or written back to the storage, could be damaged.
In our test, several metadata blocks were found with damaged metadata
checksum value during recovery process. When we only verify the
checksum value, we have to avoid modifying checksum fields directly.

Signed-off-by: Daeho Jeong <daeho.jeong@samsung.com>
Signed-off-by: Youngjin Gil <youngjin.gil@samsung.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

ext4: validate that metadata blocks do not overlap superblock

commit 829fa70dddadf9dd041d62b82cd7cea63943899d upstream.

A number of fuzzing failures seem to be caused by allocation bitmaps
or other metadata blocks being pointed at the superblock.

This can cause kernel BUG or WARNings once the superblock is
overwritten, so validate the group descriptor blocks to make sure this
doesn't happen.

Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: arcmsr: Send SYNCHRONIZE_CACHE command to firmware

commit 2bf7dc8443e113844d078fd6541b7f4aa544f92f upstream.

The arcmsr driver failed to pass SYNCHRONIZE CACHE to controller
firmware. Depending on how drive caches are handled internally by
controller firmware this could potentially lead to data integrity
problems.

Ensure that cache flushes are passed to the controller.

[mkp: applied by hand and removed unused vars]

Signed-off-by: Ching Huang <ching2048@areca.com.tw>
Reported-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: scsi_debug: Fix memory leak if LBP enabled and module is unloaded

commit 4d2b496f19f3c2cfaca1e8fa0710688b5ff3811d upstream.

map_storep was not being vfree()'d in the module_exit call.

Signed-off-by: Ewan D. Milne <emilne@redhat.com>
Reviewed-by: Laurence Oberman <loberman@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: arcmsr: Buffer overflow in arcmsr_iop_message_xfer()

commit 7bc2b55a5c030685b399bb65b6baa9ccc3d1f167 upstream.

We need to put an upper bound on "user_len" so the memcpy() doesn't
overflow.

[js] no ARCMSR_API_DATA_BUFLEN defined, use the number

Reported-by: Marco Grassi <marco.gra@gmail.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: Fix use-after-free

commit bcd8f2e94808fcddf6ef3af5f060a36820dcc432 upstream.

This patch fixes one use-after-free report[1] by KASAN.

In __scsi_scan_target(), when a type 31 device is probed,
SCSI_SCAN_TARGET_PRESENT is returned and the target will be scanned
again.

Inside the following scsi_report_lun_scan(), one new scsi_device
instance is allocated, and scsi_probe_and_add_lun() is called again to
probe the target and still see type 31 device, finally
__scsi_remove_device() is called to remove & free the device at the end
of scsi_probe_and_add_lun(), so cause use-after-free in
scsi_report_lun_scan().

And the following SCSI log can be observed:

scsi 0:0:2:0: scsi scan: INQUIRY pass 1 length 36
scsi 0:0:2:0: scsi scan: INQUIRY successful with code 0x0
scsi 0:0:2:0: scsi scan: peripheral device type of 31, no device added
scsi 0:0:2:0: scsi scan: Sending REPORT LUNS to (try 0)
scsi 0:0:2:0: scsi scan: REPORT LUNS successful (try 0) result 0x0
scsi 0:0:2:0: scsi scan: REPORT LUN scan
scsi 0:0:2:0: scsi scan: INQUIRY pass 1 length 36
scsi 0:0:2:0: scsi scan: INQUIRY successful with code 0x0
scsi 0:0:2:0: scsi scan: peripheral device type of 31, no device added
BUG: KASAN: use-after-free in __scsi_scan_target+0xbf8/0xe40 at addr ffff88007b44a104

This patch fixes the issue by moving the putting reference at
the end of scsi_report_lun_scan().

[1] KASAN report
==================================================================
[    3.274597] PM: Adding info for serio:serio1
[    3.275127] BUG: KASAN: use-after-free in __scsi_scan_target+0xd87/0xdf0 at addr ffff880254d8c304
[    3.275653] Read of size 4 by task kworker/u10:0/27
[    3.275903] CPU: 3 PID: 27 Comm: kworker/u10:0 Not tainted 4.8.0 #2121
[    3.276258] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
[    3.276797] Workqueue: events_unbound async_run_entry_fn
[    3.277083]  ffff880254d8c380 ffff880259a37870 ffffffff94bbc6c1 ffff880078402d80
[    3.277532]  ffff880254d8bb80 ffff880259a37898 ffffffff9459fec1 ffff880259a37930
[    3.277989]  ffff880254d8bb80 ffff880078402d80 ffff880259a37920 ffffffff945a0165
[    3.278436] Call Trace:
[    3.278528]  [<ffffffff94bbc6c1>] dump_stack+0x65/0x84
[    3.278797]  [<ffffffff9459fec1>] kasan_object_err+0x21/0x70
[    3.279063] device: 'psaux': device_add
[    3.279616]  [<ffffffff945a0165>] kasan_report_error+0x205/0x500
[    3.279651] PM: Adding info for No Bus:psaux
[    3.280202]  [<ffffffff944ecd22>] ? kfree_const+0x22/0x30
[    3.280486]  [<ffffffff94bc2dc9>] ? kobject_release+0x119/0x370
[    3.280805]  [<ffffffff945a0543>] __asan_report_load4_noabort+0x43/0x50
[    3.281170]  [<ffffffff9507e1f7>] ? __scsi_scan_target+0xd87/0xdf0
[    3.281506]  [<ffffffff9507e1f7>] __scsi_scan_target+0xd87/0xdf0
[    3.281848]  [<ffffffff9507d470>] ? scsi_add_device+0x30/0x30
[    3.282156]  [<ffffffff94f7f660>] ? pm_runtime_autosuspend_expiration+0x60/0x60
[    3.282570]  [<ffffffff956ddb07>] ? _raw_spin_lock+0x17/0x40
[    3.282880]  [<ffffffff9507e505>] scsi_scan_channel+0x105/0x160
[    3.283200]  [<ffffffff9507e8a2>] scsi_scan_host_selected+0x212/0x2f0
[    3.283563]  [<ffffffff9507eb3c>] do_scsi_scan_host+0x1bc/0x250
[    3.283882]  [<ffffffff9507efc1>] do_scan_async+0x41/0x450
[    3.284173]  [<ffffffff941c1fee>] async_run_entry_fn+0xfe/0x610
[    3.284492]  [<ffffffff941a8954>] ? pwq_dec_nr_in_flight+0x124/0x2a0
[    3.284876]  [<ffffffff941d1770>] ? preempt_count_add+0x130/0x160
[    3.285207]  [<ffffffff941a9a84>] process_one_work+0x544/0x12d0
[    3.285526]  [<ffffffff941aa8e9>] worker_thread+0xd9/0x12f0
[    3.285844]  [<ffffffff941aa810>] ? process_one_work+0x12d0/0x12d0
[    3.286182]  [<ffffffff941bb365>] kthread+0x1c5/0x260
[    3.286443]  [<ffffffff940855cd>] ? __switch_to+0x88d/0x1430
[    3.286745]  [<ffffffff941bb1a0>] ? kthread_worker_fn+0x5a0/0x5a0
[    3.287085]  [<ffffffff956dde9f>] ret_from_fork+0x1f/0x40
[    3.287368]  [<ffffffff941bb1a0>] ? kthread_worker_fn+0x5a0/0x5a0
[    3.287697] Object at ffff880254d8bb80, in cache kmalloc-2048 size: 2048
[    3.288064] Allocated:
[    3.288147] PID = 27
[    3.288218]  [<ffffffff940b27ab>] save_stack_trace+0x2b/0x50
[    3.288531]  [<ffffffff9459f246>] save_stack+0x46/0xd0
[    3.288806]  [<ffffffff9459f4bd>] kasan_kmalloc+0xad/0xe0
[    3.289098]  [<ffffffff9459c07e>] __kmalloc+0x13e/0x250
[    3.289378]  [<ffffffff95078e5a>] scsi_alloc_sdev+0xea/0xcf0
[    3.289701]  [<ffffffff9507de76>] __scsi_scan_target+0xa06/0xdf0
[    3.290034]  [<ffffffff9507e505>] scsi_scan_channel+0x105/0x160
[    3.290362]  [<ffffffff9507e8a2>] scsi_scan_host_selected+0x212/0x2f0
[    3.290724]  [<ffffffff9507eb3c>] do_scsi_scan_host+0x1bc/0x250
[    3.291055]  [<ffffffff9507efc1>] do_scan_async+0x41/0x450
[    3.291354]  [<ffffffff941c1fee>] async_run_entry_fn+0xfe/0x610
[    3.291695]  [<ffffffff941a9a84>] process_one_work+0x544/0x12d0
[    3.292022]  [<ffffffff941aa8e9>] worker_thread+0xd9/0x12f0
[    3.292325]  [<ffffffff941bb365>] kthread+0x1c5/0x260
[    3.292594]  [<ffffffff956dde9f>] ret_from_fork+0x1f/0x40
[    3.292886] Freed:
[    3.292945] PID = 27
[    3.293016]  [<ffffffff940b27ab>] save_stack_trace+0x2b/0x50
[    3.293327]  [<ffffffff9459f246>] save_stack+0x46/0xd0
[    3.293600]  [<ffffffff9459fa61>] kasan_slab_free+0x71/0xb0
[    3.293916]  [<ffffffff9459bac2>] kfree+0xa2/0x1f0
[    3.294168]  [<ffffffff9508158a>] scsi_device_dev_release_usercontext+0x50a/0x730
[    3.294598]  [<ffffffff941ace9a>] execute_in_process_context+0xda/0x130
[    3.294974]  [<ffffffff9508107c>] scsi_device_dev_release+0x1c/0x20
[    3.295322]  [<ffffffff94f566f6>] device_release+0x76/0x1e0
[    3.295626]  [<ffffffff94bc2db7>] kobject_release+0x107/0x370
[    3.295942]  [<ffffffff94bc29ce>] kobject_put+0x4e/0xa0
[    3.296222]  [<ffffffff94f56e17>] put_device+0x17/0x20
[    3.296497]  [<ffffffff9505201c>] scsi_device_put+0x7c/0xa0
[    3.296801]  [<ffffffff9507e1bc>] __scsi_scan_target+0xd4c/0xdf0
[    3.297132]  [<ffffffff9507e505>] scsi_scan_channel+0x105/0x160
[    3.297458]  [<ffffffff9507e8a2>] scsi_scan_host_selected+0x212/0x2f0
[    3.297829]  [<ffffffff9507eb3c>] do_scsi_scan_host+0x1bc/0x250
[    3.298156]  [<ffffffff9507efc1>] do_scan_async+0x41/0x450
[    3.298453]  [<ffffffff941c1fee>] async_run_entry_fn+0xfe/0x610
[    3.298777]  [<ffffffff941a9a84>] process_one_work+0x544/0x12d0
[    3.299105]  [<ffffffff941aa8e9>] worker_thread+0xd9/0x12f0
[    3.299408]  [<ffffffff941bb365>] kthread+0x1c5/0x260
[    3.299676]  [<ffffffff956dde9f>] ret_from_fork+0x1f/0x40
[    3.299967] Memory state around the buggy address:
[    3.300209]  ffff880254d8c200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[    3.300608]  ffff880254d8c280: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[    3.300986] >ffff880254d8c300: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[    3.301408]                    ^
[    3.301550]  ffff880254d8c380: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[    3.301987]  ffff880254d8c400: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[    3.302396]
==================================================================

Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: ibmvfc: Fix I/O hang when port is not mapped

commit 07d0e9a847401ffd2f09bd450d41644cd090e81d upstream.

If a VFC port gets unmapped in the VIOS, it may not respond with a CRQ
init complete following H_REG_CRQ. If this occurs, we can end up having
called scsi_block_requests and not a resulting unblock until the init
complete happens, which may never occur, and we end up hanging I/O
requests.  This patch ensures the host action stay set to
IBMVFC_HOST_ACTION_TGT_DEL so we move all rports into devloss state and
unblock unless we receive an init complete.

Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Acked-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: megaraid_sas: fix macro MEGASAS_IS_LOGICAL to avoid regression

commit 5e5ec1759dd663a1d5a2f10930224dd009e500e8 upstream.

This patch will fix regression caused by commit 1e793f6fc0db ("scsi:
megaraid_sas: Fix data integrity failure for JBOD (passthrough)
devices").

The problem was that the MEGASAS_IS_LOGICAL macro did not have braces
and as a result the driver ended up exposing a lot of non-existing SCSI
devices (all SCSI commands to channels 1,2,3 were returned as
SUCCESS-DID_OK by driver).

[mkp: clarified patch description]

Fixes: 1e793f6fc0db920400574211c48f9157a37e3945
Reported-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com>
Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Tested-by: Sumit Saxena <sumit.saxena@broadcom.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Tested-by: Jens Axboe <axboe@fb.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: megaraid_sas: Fix data integrity failure for JBOD (passthrough) devices

commit 1e793f6fc0db920400574211c48f9157a37e3945 upstream.

Commit 02b01e010afe ("megaraid_sas: return sync cache call with
success") modified the driver to successfully complete SYNCHRONIZE_CACHE
commands without passing them to the controller. Disk drive caches are
only explicitly managed by controller firmware when operating in RAID
mode. So this commit effectively disabled writeback cache flushing for
any drives used in JBOD mode, leading to data integrity failures.

[mkp: clarified patch description]

Fixes: 02b01e010afeeb49328d35650d70721d2ca3fd59
Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com>
Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

mpt2sas: Fix secure erase premature termination

Problem:
This is a work around for a bug with LSI Fusion MPT SAS2 when
pefroming secure erase. Due to the very long time the operation
takes commands issued during the erase will time out and will trigger
execution of abort hook. Even though the abort hook is called for
the specific command which timed out this leads to entire device halt
(scsi_state terminated) and premature termination of the secured erase.

Fix:
Set device state to busy while erase in progress to reject any incoming
commands until the erase is done. The device is blocked any way during
this time and cannot execute any other command.
More data and logs can be found here -
https://drive.google.com/file/d/0B9ocOHYHbbS1Q3VMdkkzeWFkTjg/view

P.S
This is a backport from the same fix for mpt3sas driver intended
for pre-4.4 stable trees.

Signed-off-by: Andrey Grodzovsky <andrey2805@gmail.com>
Cc: Sreekanth Reddy <Sreekanth.Reddy@broadcom.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: PDL-MPT-FUSIONLINUX <MPT-FusionLinux.pdl@broadcom.com>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: mpt3sas: fix hang on ata passthrough commands

commit ffb58456589443ca572221fabbdef3db8483a779 upstream.

mpt3sas has a firmware failure where it can only handle one pass through
ATA command at a time.  If another comes in, contrary to the SAT
standard, it will hang until the first one completes (causing long
commands like secure erase to timeout).  The original fix was to block
the device when an ATA command came in, but this caused a regression
with

commit 669f044170d8933c3d66d231b69ea97cb8447338
Author: Bart Van Assche <bart.vanassche@sandisk.com>
Date:   Tue Nov 22 16:17:13 2016 -0800

    scsi: srp_transport: Move queuecommand() wait code to SCSI core

So fix the original fix of the secure erase timeout by properly
returning SAM_STAT_BUSY like the SAT recommends.  The original patch
also had a concurrency problem since scsih_qcmd is lockless at that
point (this is fixed by using atomic bitops to set and test the flag).

[mkp: addressed feedback wrt. test_bit and fixed whitespace]

Fixes: 18f6084a989ba1b (mpt3sas: Fix secure erase premature termination)
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@broadcom.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: Ingo Molnar <mingo@kernel.org>
Tested-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[wt: adjust context]
Signed-off-by: Willy Tarreau <w@1wt.eu>

scsi: mpt3sas: Unblock device after controller reset

commit 7ff723ad0f87feba43dda45fdae71206063dd7d4 upstream.

While issuing any ATA passthrough command to firmware the driver will
block the device. But it will unblock the device only if the I/O
completes through the ISR path. If a controller reset occurs before
command completion the device will remain in blocked state.

Make sure we unblock the device following a controller reset if an ATA
passthrough command was queued.

[mkp: clarified patch description]

Cc: <stable@vger.kernel.org> # v4.4+
Fixes: ac6c2a93bd07 ("mpt3sas: Fix for SATA drive in blocked state, after diag reset")
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[wt: adjust context]
Signed-off-by: Willy Tarreau <w@1wt.eu>