platform/kernel/linux-amlogic.git
5 years agosiphash: add cryptographically secure PRF
Jason A. Donenfeld [Fri, 16 Aug 2019 23:01:12 +0000 (00:01 +0100)]
siphash: add cryptographically secure PRF

commit 2c956a60778cbb6a27e0c7a8a52a91378c90e1d1 upstream.

SipHash is a 64-bit keyed hash function that is actually a
cryptographically secure PRF, like HMAC. Except SipHash is super fast,
and is meant to be used as a hashtable keyed lookup function, or as a
general PRF for short input use cases, such as sequence numbers or RNG
chaining.

For the first usage:

There are a variety of attacks known as "hashtable poisoning" in which an
attacker forms some data such that the hash of that data will be the
same, and then preceeds to fill up all entries of a hashbucket. This is
a realistic and well-known denial-of-service vector. Currently
hashtables use jhash, which is fast but not secure, and some kind of
rotating key scheme (or none at all, which isn't good). SipHash is meant
as a replacement for jhash in these cases.

There are a modicum of places in the kernel that are vulnerable to
hashtable poisoning attacks, either via userspace vectors or network
vectors, and there's not a reliable mechanism inside the kernel at the
moment to fix it. The first step toward fixing these issues is actually
getting a secure primitive into the kernel for developers to use. Then
we can, bit by bit, port things over to it as deemed appropriate.

While SipHash is extremely fast for a cryptographically secure function,
it is likely a bit slower than the insecure jhash, and so replacements
will be evaluated on a case-by-case basis based on whether or not the
difference in speed is negligible and whether or not the current jhash usage
poses a real security risk.

For the second usage:

A few places in the kernel are using MD5 or SHA1 for creating secure
sequence numbers, syn cookies, port numbers, or fast random numbers.
SipHash is a faster and more fitting, and more secure replacement for MD5
in those situations. Replacing MD5 and SHA1 with SipHash for these uses is
obvious and straight-forward, and so is submitted along with this patch
series. There shouldn't be much of a debate over its efficacy.

Dozens of languages are already using this internally for their hash
tables and PRFs. Some of the BSDs already use this in their kernels.
SipHash is a widely known high-speed solution to a widely known set of
problems, and it's time we catch-up.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Jean-Philippe Aumasson <jeanphilippe.aumasson@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Eric Biggers <ebiggers3@gmail.com>
Cc: David Laight <David.Laight@aculab.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 4.9 as dependency of commits df453700e8d8 "inet: switch
 IP ID generator to siphash" and 3c79107631db "netfilter: ctnetlink: don't
 use conntrack/expect object addresses as id"]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agovhost: scsi: add weight support
Jason Wang [Fri, 16 Aug 2019 23:01:01 +0000 (00:01 +0100)]
vhost: scsi: add weight support

commit c1ea02f15ab5efb3e93fc3144d895410bf79fcf2 upstream.

This patch will check the weight and exit the loop if we exceeds the
weight. This is useful for preventing scsi kthread from hogging cpu
which is guest triggerable.

This addresses CVE-2019-3900.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Fixes: 057cbf49a1f0 ("tcm_vhost: Initial merge for vhost level target fabric driver")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
[bwh: Backported to 4.9:
 - Drop changes in vhost_scsi_ctl_handle_vq()
 - Adjust context]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agovhost_net: fix possible infinite loop
Jason Wang [Fri, 16 Aug 2019 23:00:53 +0000 (00:00 +0100)]
vhost_net: fix possible infinite loop

commit e2412c07f8f3040593dfb88207865a3cd58680c0 upstream.

When the rx buffer is too small for a packet, we will discard the vq
descriptor and retry it for the next packet:

while ((sock_len = vhost_net_rx_peek_head_len(net, sock->sk,
      &busyloop_intr))) {
...
/* On overrun, truncate and discard */
if (unlikely(headcount > UIO_MAXIOV)) {
iov_iter_init(&msg.msg_iter, READ, vq->iov, 1, 1);
err = sock->ops->recvmsg(sock, &msg,
 1, MSG_DONTWAIT | MSG_TRUNC);
pr_debug("Discarded rx packet: len %zd\n", sock_len);
continue;
}
...
}

This makes it possible to trigger a infinite while..continue loop
through the co-opreation of two VMs like:

1) Malicious VM1 allocate 1 byte rx buffer and try to slow down the
   vhost process as much as possible e.g using indirect descriptors or
   other.
2) Malicious VM2 generate packets to VM1 as fast as possible

Fixing this by checking against weight at the end of RX and TX
loop. This also eliminate other similar cases when:

- userspace is consuming the packets in the meanwhile
- theoretical TOCTOU attack if guest moving avail index back and forth
  to hit the continue after vhost find guest just add new buffers

This addresses CVE-2019-3900.

Fixes: d8316f3991d20 ("vhost: fix total length when packets are too short")
Fixes: 3a4d5c94e9593 ("vhost_net: a kernel-level virtio server")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
[bwh: Backported to 4.9:
 - Both Tx modes are handled in one loop in handle_tx()
 - Adjust context]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agovhost: introduce vhost_exceeds_weight()
Jason Wang [Fri, 16 Aug 2019 23:00:44 +0000 (00:00 +0100)]
vhost: introduce vhost_exceeds_weight()

commit e82b9b0727ff6d665fff2d326162b460dded554d upstream.

We used to have vhost_exceeds_weight() for vhost-net to:

- prevent vhost kthread from hogging the cpu
- balance the time spent between TX and RX

This function could be useful for vsock and scsi as well. So move it
to vhost.c. Device must specify a weight which counts the number of
requests, or it can also specific a byte_weight which counts the
number of bytes that has been processed.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
[bwh: Backported to 4.9:
 - In vhost_net, both Tx modes are handled in one loop in handle_tx()
 - Adjust context]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agovhost_net: introduce vhost_exceeds_weight()
Jason Wang [Fri, 16 Aug 2019 23:00:36 +0000 (00:00 +0100)]
vhost_net: introduce vhost_exceeds_weight()

commit 272f35cba53d088085e5952fd81d7a133ab90789 upstream.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 4.9: adjust context]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agovhost_net: use packet weight for rx handler, too
Paolo Abeni [Fri, 16 Aug 2019 23:00:28 +0000 (00:00 +0100)]
vhost_net: use packet weight for rx handler, too

commit db688c24eada63b1efe6d0d7d835e5c3bdd71fd3 upstream.

Similar to commit a2ac99905f1e ("vhost-net: set packet weight of
tx polling to 2 * vq size"), we need a packet-based limit for
handler_rx, too - elsewhere, under rx flood with small packets,
tx can be delayed for a very long time, even without busypolling.

The pkt limit applied to handle_rx must be the same applied by
handle_tx, or we will get unfair scheduling between rx and tx.
Tying such limit to the queue length makes it less effective for
large queue length values and can introduce large process
scheduler latencies, so a constant valued is used - likewise
the existing bytes limit.

The selected limit has been validated with PVP[1] performance
test with different queue sizes:

queue size 256 512 1024

baseline 366 354 362
weight 128 715 723 670
weight 256 740 745 733
weight 512 600 460 583
weight 1024 423 427 418

A packet weight of 256 gives peek performances in under all the
tested scenarios.

No measurable regression in unidirectional performance tests has
been detected.

[1] https://developers.redhat.com/blog/2017/06/05/measuring-and-comparing-open-vswitch-performance/

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agovhost-net: set packet weight of tx polling to 2 * vq size
haibinzhang(张海斌) [Fri, 16 Aug 2019 23:00:19 +0000 (00:00 +0100)]
vhost-net: set packet weight of tx polling to 2 * vq size

commit a2ac99905f1ea8b15997a6ec39af69aa28a3653b upstream.

handle_tx will delay rx for tens or even hundreds of milliseconds when tx busy
polling udp packets with small length(e.g. 1byte udp payload), because setting
VHOST_NET_WEIGHT takes into account only sent-bytes but no single packet length.

Ping-Latencies shown below were tested between two Virtual Machines using
netperf (UDP_STREAM, len=1), and then another machine pinged the client:

vq size=256
Packet-Weight   Ping-Latencies(millisecond)
                   min      avg       max
Origin           3.319   18.489    57.303
64               1.643    2.021     2.552
128              1.825    2.600     3.224
256              1.997    2.710     4.295
512              1.860    3.171     4.631
1024             2.002    4.173     9.056
2048             2.257    5.650     9.688
4096             2.093    8.508    15.943

vq size=512
Packet-Weight   Ping-Latencies(millisecond)
                   min      avg       max
Origin           6.537   29.177    66.245
64               2.798    3.614     4.403
128              2.861    3.820     4.775
256              3.008    4.018     4.807
512              3.254    4.523     5.824
1024             3.079    5.335     7.747
2048             3.944    8.201    12.762
4096             4.158   11.057    19.985

Seems pretty consistent, a small dip at 2 VQ sizes.
Ring size is a hint from device about a burst size it can tolerate. Based on
benchmarks, set the weight to 2 * vq size.

To evaluate this change, another tests were done using netperf(RR, TX) between
two machines with Intel(R) Xeon(R) Gold 6133 CPU @ 2.50GHz, and vq size was
tweaked through qemu. Results shown below does not show obvious changes.

vq size=256 TCP_RR                vq size=512 TCP_RR
size/sessions/+thu%/+normalize%   size/sessions/+thu%/+normalize%
   1/       1/  -7%/        -2%      1/       1/   0%/        -2%
   1/       4/  +1%/         0%      1/       4/  +1%/         0%
   1/       8/  +1%/        -2%      1/       8/   0%/        +1%
  64/       1/  -6%/         0%     64/       1/  +7%/        +3%
  64/       4/   0%/        +2%     64/       4/  -1%/        +1%
  64/       8/   0%/         0%     64/       8/  -1%/        -2%
 256/       1/  -3%/        -4%    256/       1/  -4%/        -2%
 256/       4/  +3%/        +4%    256/       4/  +1%/        +2%
 256/       8/  +2%/         0%    256/       8/  +1%/        -1%

vq size=256 UDP_RR                vq size=512 UDP_RR
size/sessions/+thu%/+normalize%   size/sessions/+thu%/+normalize%
   1/       1/  -5%/        +1%      1/       1/  -3%/        -2%
   1/       4/  +4%/        +1%      1/       4/  -2%/        +2%
   1/       8/  -1%/        -1%      1/       8/  -1%/         0%
  64/       1/  -2%/        -3%     64/       1/  +1%/        +1%
  64/       4/  -5%/        -1%     64/       4/  +2%/         0%
  64/       8/   0%/        -1%     64/       8/  -2%/        +1%
 256/       1/  +7%/        +1%    256/       1/  -7%/         0%
 256/       4/  +1%/        +1%    256/       4/  -3%/        -4%
 256/       8/  +2%/        +2%    256/       8/  +1%/        +1%

vq size=256 TCP_STREAM            vq size=512 TCP_STREAM
size/sessions/+thu%/+normalize%   size/sessions/+thu%/+normalize%
  64/       1/   0%/        -3%     64/       1/   0%/         0%
  64/       4/  +3%/        -1%     64/       4/  -2%/        +4%
  64/       8/  +9%/        -4%     64/       8/  -1%/        +2%
 256/       1/  +1%/        -4%    256/       1/  +1%/        +1%
 256/       4/  -1%/        -1%    256/       4/  -3%/         0%
 256/       8/  +7%/        +5%    256/       8/  -3%/         0%
 512/       1/  +1%/         0%    512/       1/  -1%/        -1%
 512/       4/  +1%/        -1%    512/       4/   0%/         0%
 512/       8/  +7%/        -5%    512/       8/  +6%/        -1%
1024/       1/   0%/        -1%   1024/       1/   0%/        +1%
1024/       4/  +3%/         0%   1024/       4/  +1%/         0%
1024/       8/  +8%/        +5%   1024/       8/  -1%/         0%
2048/       1/  +2%/        +2%   2048/       1/  -1%/         0%
2048/       4/  +1%/         0%   2048/       4/   0%/        -1%
2048/       8/  -2%/         0%   2048/       8/   5%/        -1%
4096/       1/  -2%/         0%   4096/       1/  -2%/         0%
4096/       4/  +2%/         0%   4096/       4/   0%/         0%
4096/       8/  +9%/        -2%   4096/       8/  -5%/        -1%

Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Haibin Zhang <haibinzhang@tencent.com>
Signed-off-by: Yunfang Tai <yunfangtai@tencent.com>
Signed-off-by: Lidong Chen <lidongchen@tencent.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agobpf: add bpf_jit_limit knob to restrict unpriv allocations
Daniel Borkmann [Fri, 16 Aug 2019 23:00:08 +0000 (00:00 +0100)]
bpf: add bpf_jit_limit knob to restrict unpriv allocations

commit ede95a63b5e84ddeea6b0c473b36ab8bfd8c6ce3 upstream.

Rick reported that the BPF JIT could potentially fill the entire module
space with BPF programs from unprivileged users which would prevent later
attempts to load normal kernel modules or privileged BPF programs, for
example. If JIT was enabled but unsuccessful to generate the image, then
before commit 290af86629b2 ("bpf: introduce BPF_JIT_ALWAYS_ON config")
we would always fall back to the BPF interpreter. Nowadays in the case
where the CONFIG_BPF_JIT_ALWAYS_ON could be set, then the load will abort
with a failure since the BPF interpreter was compiled out.

Add a global limit and enforce it for unprivileged users such that in case
of BPF interpreter compiled out we fail once the limit has been reached
or we fall back to BPF interpreter earlier w/o using module mem if latter
was compiled in. In a next step, fair share among unprivileged users can
be resolved in particular for the case where we would fail hard once limit
is reached.

Fixes: 290af86629b2 ("bpf: introduce BPF_JIT_ALWAYS_ON config")
Fixes: 0a14842f5a3c ("net: filter: Just In Time compiler for x86-64")
Co-Developed-by: Rick Edgecombe <rick.p.edgecombe@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Jann Horn <jannh@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: LKML <linux-kernel@vger.kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
[bwh: Backported to 4.9: adjust context]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agobpf: restrict access to core bpf sysctls
Daniel Borkmann [Fri, 16 Aug 2019 22:59:56 +0000 (23:59 +0100)]
bpf: restrict access to core bpf sysctls

commit 2e4a30983b0f9b19b59e38bbf7427d7fdd480d98 upstream.

Given BPF reaches far beyond just networking these days, it was
never intended to allow setting and in some cases reading those
knobs out of a user namespace root running without CAP_SYS_ADMIN,
thus tighten such access.

Also the bpf_jit_enable = 2 debugging mode should only be allowed
if kptr_restrict is not set since it otherwise can leak addresses
to the kernel log. Dump a note to the kernel log that this is for
debugging JITs only when enabled.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
[bwh: Backported to 4.9:
 - We don't have bpf_dump_raw_ok(), so drop the condition based on it. This
   condition only made it a bit harder for a privileged user to do something
   silly.
 - Drop change to bpf_jit_kallsyms]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agobpf: get rid of pure_initcall dependency to enable jits
Daniel Borkmann [Fri, 16 Aug 2019 22:59:20 +0000 (23:59 +0100)]
bpf: get rid of pure_initcall dependency to enable jits

commit fa9dd599b4dae841924b022768354cfde9affecb upstream.

Having a pure_initcall() callback just to permanently enable BPF
JITs under CONFIG_BPF_JIT_ALWAYS_ON is unnecessary and could leave
a small race window in future where JIT is still disabled on boot.
Since we know about the setting at compilation time anyway, just
initialize it properly there. Also consolidate all the individual
bpf_jit_enable variables into a single one and move them under one
location. Moreover, don't allow for setting unspecified garbage
values on them.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
[bwh: Backported to 4.9 as dependency of commit 2e4a30983b0f
 "bpf: restrict access to core bpf sysctls":
 - Drop change in arch/mips/net/ebpf_jit.c
 - Drop change to bpf_jit_kallsyms
 - Adjust filenames, context]
Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomm/memcontrol.c: fix use after free in mem_cgroup_iter()
Miles Chen [Tue, 13 Aug 2019 22:37:28 +0000 (15:37 -0700)]
mm/memcontrol.c: fix use after free in mem_cgroup_iter()

commit 54a83d6bcbf8f4700013766b974bf9190d40b689 upstream.

This patch is sent to report an use after free in mem_cgroup_iter()
after merging commit be2657752e9e ("mm: memcg: fix use after free in
mem_cgroup_iter()").

I work with android kernel tree (4.9 & 4.14), and commit be2657752e9e
("mm: memcg: fix use after free in mem_cgroup_iter()") has been merged
to the trees.  However, I can still observe use after free issues
addressed in the commit be2657752e9e.  (on low-end devices, a few times
this month)

backtrace:
        css_tryget <- crash here
        mem_cgroup_iter
        shrink_node
        shrink_zones
        do_try_to_free_pages
        try_to_free_pages
        __perform_reclaim
        __alloc_pages_direct_reclaim
        __alloc_pages_slowpath
        __alloc_pages_nodemask

To debug, I poisoned mem_cgroup before freeing it:

  static void __mem_cgroup_free(struct mem_cgroup *memcg)
        for_each_node(node)
        free_mem_cgroup_per_node_info(memcg, node);
        free_percpu(memcg->stat);
  +     /* poison memcg before freeing it */
  +     memset(memcg, 0x78, sizeof(struct mem_cgroup));
        kfree(memcg);
  }

The coredump shows the position=0xdbbc2a00 is freed.

  (gdb) p/x ((struct mem_cgroup_per_node *)0xe5009e00)->iter[8]
  $13 = {position = 0xdbbc2a00, generation = 0x2efd}

  0xdbbc2a00:     0xdbbc2e00      0x00000000      0xdbbc2800      0x00000100
  0xdbbc2a10:     0x00000200      0x78787878      0x00026218      0x00000000
  0xdbbc2a20:     0xdcad6000      0x00000001      0x78787800      0x00000000
  0xdbbc2a30:     0x78780000      0x00000000      0x0068fb84      0x78787878
  0xdbbc2a40:     0x78787878      0x78787878      0x78787878      0xe3fa5cc0
  0xdbbc2a50:     0x78787878      0x78787878      0x00000000      0x00000000
  0xdbbc2a60:     0x00000000      0x00000000      0x00000000      0x00000000
  0xdbbc2a70:     0x00000000      0x00000000      0x00000000      0x00000000
  0xdbbc2a80:     0x00000000      0x00000000      0x00000000      0x00000000
  0xdbbc2a90:     0x00000001      0x00000000      0x00000000      0x00100000
  0xdbbc2aa0:     0x00000001      0xdbbc2ac8      0x00000000      0x00000000
  0xdbbc2ab0:     0x00000000      0x00000000      0x00000000      0x00000000
  0xdbbc2ac0:     0x00000000      0x00000000      0xe5b02618      0x00001000
  0xdbbc2ad0:     0x00000000      0x78787878      0x78787878      0x78787878
  0xdbbc2ae0:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2af0:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b00:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b10:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b20:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b30:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b40:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b50:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b60:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b70:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2b80:     0x78787878      0x78787878      0x00000000      0x78787878
  0xdbbc2b90:     0x78787878      0x78787878      0x78787878      0x78787878
  0xdbbc2ba0:     0x78787878      0x78787878      0x78787878      0x78787878

In the reclaim path, try_to_free_pages() does not setup
sc.target_mem_cgroup and sc is passed to do_try_to_free_pages(), ...,
shrink_node().

In mem_cgroup_iter(), root is set to root_mem_cgroup because
sc->target_mem_cgroup is NULL.  It is possible to assign a memcg to
root_mem_cgroup.nodeinfo.iter in mem_cgroup_iter().

        try_to_free_pages
         struct scan_control sc = {...}, target_mem_cgroup is 0x0;
        do_try_to_free_pages
        shrink_zones
        shrink_node
          mem_cgroup *root = sc->target_mem_cgroup;
          memcg = mem_cgroup_iter(root, NULL, &reclaim);
        mem_cgroup_iter()
         if (!root)
         root = root_mem_cgroup;
         ...

         css = css_next_descendant_pre(css, &root->css);
         memcg = mem_cgroup_from_css(css);
         cmpxchg(&iter->position, pos, memcg);

My device uses memcg non-hierarchical mode.  When we release a memcg:
invalidate_reclaim_iterators() reaches only dead_memcg and its parents.
If non-hierarchical mode is used, invalidate_reclaim_iterators() never
reaches root_mem_cgroup.

  static void invalidate_reclaim_iterators(struct mem_cgroup *dead_memcg)
  {
        struct mem_cgroup *memcg = dead_memcg;

        for (; memcg; memcg = parent_mem_cgroup(memcg)
        ...
  }

So the use after free scenario looks like:

  CPU1 CPU2

  try_to_free_pages
  do_try_to_free_pages
  shrink_zones
  shrink_node
  mem_cgroup_iter()
      if (!root)
       root = root_mem_cgroup;
      ...
      css = css_next_descendant_pre(css, &root->css);
      memcg = mem_cgroup_from_css(css);
      cmpxchg(&iter->position, pos, memcg);

         invalidate_reclaim_iterators(memcg);
         ...
         __mem_cgroup_free()
         kfree(memcg);

  try_to_free_pages
  do_try_to_free_pages
  shrink_zones
  shrink_node
  mem_cgroup_iter()
      if (!root)
       root = root_mem_cgroup;
      ...
      mz = mem_cgroup_nodeinfo(root, reclaim->pgdat->node_id);
      iter = &mz->iter[reclaim->priority];
      pos = READ_ONCE(iter->position);
      css_tryget(&pos->css) <- use after free

To avoid this, we should also invalidate root_mem_cgroup.nodeinfo.iter
in invalidate_reclaim_iterators().

[cai@lca.pw: fix -Wparentheses compilation warning]
Link: http://lkml.kernel.org/r/1564580753-17531-1-git-send-email-cai@lca.pw
Link: http://lkml.kernel.org/r/20190730015729.4406-1-miles.chen@mediatek.com
Fixes: 5ac8fb31ad2e ("mm: memcontrol: convert reclaim iterator to simple css refcounting")
Signed-off-by: Miles Chen <miles.chen@mediatek.com>
Signed-off-by: Qian Cai <cai@lca.pw>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomm/usercopy: use memory range to be accessed for wraparound check
Isaac J. Manjarres [Tue, 13 Aug 2019 22:37:37 +0000 (15:37 -0700)]
mm/usercopy: use memory range to be accessed for wraparound check

commit 951531691c4bcaa59f56a316e018bc2ff1ddf855 upstream.

Currently, when checking to see if accessing n bytes starting at address
"ptr" will cause a wraparound in the memory addresses, the check in
check_bogus_address() adds an extra byte, which is incorrect, as the
range of addresses that will be accessed is [ptr, ptr + (n - 1)].

This can lead to incorrectly detecting a wraparound in the memory
address, when trying to read 4 KB from memory that is mapped to the the
last possible page in the virtual address space, when in fact, accessing
that range of memory would not cause a wraparound to occur.

Use the memory range that will actually be accessed when considering if
accessing a certain amount of bytes will cause the memory address to
wrap around.

Link: http://lkml.kernel.org/r/1564509253-23287-1-git-send-email-isaacm@codeaurora.org
Fixes: f5509cc18daa ("mm: Hardened usercopy")
Signed-off-by: Prasad Sodagudi <psodagud@codeaurora.org>
Signed-off-by: Isaac J. Manjarres <isaacm@codeaurora.org>
Co-developed-by: Prasad Sodagudi <psodagud@codeaurora.org>
Reviewed-by: William Kucharski <william.kucharski@oracle.com>
Acked-by: Kees Cook <keescook@chromium.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Trilok Soni <tsoni@codeaurora.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[kees: backport to v4.9]
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agosh: kernel: hw_breakpoint: Fix missing break in switch statement
Gustavo A. R. Silva [Sat, 10 Aug 2019 04:43:56 +0000 (23:43 -0500)]
sh: kernel: hw_breakpoint: Fix missing break in switch statement

commit 1ee1119d184bb06af921b48c3021d921bbd85bac upstream.

Add missing break statement in order to prevent the code from falling
through to case SH_BREAKPOINT_WRITE.

Fixes: 09a072947791 ("sh: hw-breakpoints: Add preliminary support for SH-4A UBC.")
Cc: stable@vger.kernel.org
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoscsi: mpt3sas: Use 63-bit DMA addressing on SAS35 HBA
Suganath Prabu [Tue, 30 Jul 2019 07:43:57 +0000 (03:43 -0400)]
scsi: mpt3sas: Use 63-bit DMA addressing on SAS35 HBA

commit df9a606184bfdb5ae3ca9d226184e9489f5c24f7 upstream.

Although SAS3 & SAS3.5 IT HBA controllers support 64-bit DMA addressing, as
per hardware design, if DMA-able range contains all 64-bits
set (0xFFFFFFFF-FFFFFFFF) then it results in a firmware fault.

E.g. SGE's start address is 0xFFFFFFFF-FFFF000 and data length is 0x1000
bytes. when HBA tries to DMA the data at 0xFFFFFFFF-FFFFFFFF location then
HBA will fault the firmware.

Driver will set 63-bit DMA mask to ensure the above address will not be
used.

Cc: <stable@vger.kernel.org> # 5.1.20+
Signed-off-by: Suganath Prabu <suganath-prabu.subramani@broadcom.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoiwlwifi: don't unmap as page memory that was mapped as single
Emmanuel Grumbach [Sun, 21 Jul 2019 11:02:27 +0000 (14:02 +0300)]
iwlwifi: don't unmap as page memory that was mapped as single

commit 87e7e25aee6b59fef740856f4e86d4b60496c9e1 upstream.

In order to remember how to unmap a memory (as single or
as page), we maintain a bit per Transmit Buffer (TBs) in
the meta data (structure iwl_cmd_meta).
We maintain a bitmap: 1 bit per TB.
If the TB is set, we will free the memory as a page.
This bitmap was never cleared. Fix this.

Cc: stable@vger.kernel.org
Fixes: 3cd1980b0cdf ("iwlwifi: pcie: introduce new tfd and tb formats")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomwifiex: fix 802.11n/WPA detection
Brian Norris [Wed, 24 Jul 2019 19:46:34 +0000 (12:46 -0700)]
mwifiex: fix 802.11n/WPA detection

commit df612421fe2566654047769c6852ffae1a31df16 upstream.

Commit 63d7ef36103d ("mwifiex: Don't abort on small, spec-compliant
vendor IEs") adjusted the ieee_types_vendor_header struct, which
inadvertently messed up the offsets used in
mwifiex_is_wpa_oui_present(). Add that offset back in, mirroring
mwifiex_is_rsn_oui_present().

As it stands, commit 63d7ef36103d breaks compatibility with WPA (not
WPA2) 802.11n networks, since we hit the "info: Disable 11n if AES is
not supported by AP" case in mwifiex_is_network_compatible().

Fixes: 63d7ef36103d ("mwifiex: Don't abort on small, spec-compliant vendor IEs")
Cc: <stable@vger.kernel.org>
Signed-off-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agosmb3: send CAP_DFS capability during session setup
Steve French [Thu, 25 Jul 2019 23:13:10 +0000 (18:13 -0500)]
smb3: send CAP_DFS capability during session setup

commit 8d33096a460d5b9bd13300f01615df5bb454db10 upstream.

We had a report of a server which did not do a DFS referral
because the session setup Capabilities field was set to 0
(unlike negotiate protocol where we set CAP_DFS).  Better to
send it session setup in the capabilities as well (this also
more closely matches Windows client behavior).

Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-off-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
CC: Stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoSMB3: Fix deadlock in validate negotiate hits reconnect
Pavel Shilovsky [Mon, 22 Jul 2019 18:34:59 +0000 (11:34 -0700)]
SMB3: Fix deadlock in validate negotiate hits reconnect

commit e99c63e4d86d3a94818693147b469fa70de6f945 upstream.

Currently we skip SMB2_TREE_CONNECT command when checking during
reconnect because Tree Connect happens when establishing
an SMB session. For SMB 3.0 protocol version the code also calls
validate negotiate which results in SMB2_IOCL command being sent
over the wire. This may deadlock on trying to acquire a mutex when
checking for reconnect. Fix this by skipping SMB2_IOCL command
when doing the reconnect check.

Signed-off-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
CC: Stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomac80211: don't WARN on short WMM parameters from AP
Brian Norris [Fri, 26 Jul 2019 22:47:58 +0000 (15:47 -0700)]
mac80211: don't WARN on short WMM parameters from AP

commit 05aaa5c97dce4c10a9e7eae2f1569a684e0c5ced upstream.

In a very similar spirit to commit c470bdc1aaf3 ("mac80211: don't WARN
on bad WMM parameters from buggy APs"), an AP may not transmit a
fully-formed WMM IE. For example, it may miss or repeat an Access
Category. The above loop won't catch that and will instead leave one of
the four ACs zeroed out. This triggers the following warning in
drv_conf_tx()

  wlan0: invalid CW_min/CW_max: 0/0

and it may leave one of the hardware queues unconfigured. If we detect
such a case, let's just print a warning and fall back to the defaults.

Tested with a hacked version of hostapd, intentionally corrupting the
IEs in hostapd_eid_wmm().

Cc: stable@vger.kernel.org
Signed-off-by: Brian Norris <briannorris@chromium.org>
Link: https://lore.kernel.org/r/20190726224758.210953-1-briannorris@chromium.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoALSA: hda - Don't override global PCM hw info flag
Takashi Iwai [Tue, 6 Aug 2019 12:03:56 +0000 (14:03 +0200)]
ALSA: hda - Don't override global PCM hw info flag

commit c1c6c877b0c79fd7e05c931435aa42211eaeebaf upstream.

The commit bfcba288b97f ("ALSA - hda: Add support for link audio time
reporting") introduced the conditional PCM hw info setup, but it
overwrites the global azx_pcm_hw object.  This will cause a problem if
any other HD-audio controller, as it'll inherit the same bit flag
although another controller doesn't support that feature.

Fix the bug by setting the PCM hw info flag locally.

Fixes: bfcba288b97f ("ALSA - hda: Add support for link audio time reporting")
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoALSA: firewire: fix a memory leak bug
Wenwen Wang [Thu, 8 Aug 2019 05:50:58 +0000 (00:50 -0500)]
ALSA: firewire: fix a memory leak bug

commit 1be3c1fae6c1e1f5bb982b255d2034034454527a upstream.

In iso_packets_buffer_init(), 'b->packets' is allocated through
kmalloc_array(). Then, the aligned packet size is checked. If it is
larger than PAGE_SIZE, -EINVAL will be returned to indicate the error.
However, the allocated 'b->packets' is not deallocated on this path,
leading to a memory leak.

To fix the above issue, free 'b->packets' before returning the error code.

Fixes: 31ef9134eb52 ("ALSA: add LaCie FireWire Speakers/Griffin FireWave Surround driver")
Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
Reviewed-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Cc: <stable@vger.kernel.org> # v2.6.39+
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agohwmon: (nct7802) Fix wrong detection of in4 presence
Guenter Roeck [Fri, 26 Jul 2019 15:00:49 +0000 (08:00 -0700)]
hwmon: (nct7802) Fix wrong detection of in4 presence

commit 38ada2f406a9b81fb1249c5c9227fa657e7d5671 upstream.

The code to detect if in4 is present is wrong; if in4 is not present,
the in4_input sysfs attribute is still present.

In detail:

- Ihen RTD3_MD=11 (VSEN3 present), everything is as expected (no bug).
- If we have RTD3_MD!=11 (no VSEN3), we unexpectedly have a in4_input
  file under /sys and the "sensors" command displays in4_input.
  But as expected, we have no in4_min, in4_max, in4_alarm, in4_beep.

Fix is_visible function to detect and report in4_input visibility
as expected.

Reported-by: Gilles Buloz <Gilles.Buloz@kontron.com>
Cc: Gilles Buloz <Gilles.Buloz@kontron.com>
Cc: stable@vger.kernel.org
Fixes: 3434f37835804 ("hwmon: Driver for Nuvoton NCT7802Y")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agocan: peak_usb: pcan_usb_fd: Fix info-leaks to USB devices
Tomas Bortoli [Wed, 31 Jul 2019 14:54:47 +0000 (10:54 -0400)]
can: peak_usb: pcan_usb_fd: Fix info-leaks to USB devices

commit 30a8beeb3042f49d0537b7050fd21b490166a3d9 upstream.

Uninitialized Kernel memory can leak to USB devices.

Fix by using kzalloc() instead of kmalloc() on the affected buffers.

Signed-off-by: Tomas Bortoli <tomasbortoli@gmail.com>
Reported-by: syzbot+513e4d0985298538bf9b@syzkaller.appspotmail.com
Fixes: 0a25e1f4f185 ("can: peak_usb: add support for PEAK new CANFD USB adapters")
Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agocan: peak_usb: pcan_usb_pro: Fix info-leaks to USB devices
Tomas Bortoli [Wed, 31 Jul 2019 14:54:47 +0000 (10:54 -0400)]
can: peak_usb: pcan_usb_pro: Fix info-leaks to USB devices

commit ead16e53c2f0ed946d82d4037c630e2f60f4ab69 upstream.

Uninitialized Kernel memory can leak to USB devices.

Fix by using kzalloc() instead of kmalloc() on the affected buffers.

Signed-off-by: Tomas Bortoli <tomasbortoli@gmail.com>
Reported-by: syzbot+d6a5a1a3657b596ef132@syzkaller.appspotmail.com
Fixes: f14e22435a27 ("net: can: peak_usb: Do not do dma on the stack")
Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoperf/core: Fix creating kernel counters for PMUs that override event->cpu
Leonard Crestez [Wed, 24 Jul 2019 12:53:24 +0000 (15:53 +0300)]
perf/core: Fix creating kernel counters for PMUs that override event->cpu

[ Upstream commit 4ce54af8b33d3e21ca935fc1b89b58cbba956051 ]

Some hardware PMU drivers will override perf_event.cpu inside their
event_init callback. This causes a lockdep splat when initialized through
the kernel API:

 WARNING: CPU: 0 PID: 250 at kernel/events/core.c:2917 ctx_sched_out+0x78/0x208
 pc : ctx_sched_out+0x78/0x208
 Call trace:
  ctx_sched_out+0x78/0x208
  __perf_install_in_context+0x160/0x248
  remote_function+0x58/0x68
  generic_exec_single+0x100/0x180
  smp_call_function_single+0x174/0x1b8
  perf_install_in_context+0x178/0x188
  perf_event_create_kernel_counter+0x118/0x160

Fix this by calling perf_install_in_context with event->cpu, just like
perf_event_open

Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Mark Rutland <mark.rutland@arm.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Frank Li <Frank.li@nxp.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will@kernel.org>
Link: https://lkml.kernel.org/r/c4ebe0503623066896d7046def4d6b1e06e0eb2e.1563972056.git.leonard.crestez@nxp.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agotty/ldsem, locking/rwsem: Add missing ACQUIRE to read_failed sleep loop
Peter Zijlstra [Thu, 18 Jul 2019 13:03:15 +0000 (15:03 +0200)]
tty/ldsem, locking/rwsem: Add missing ACQUIRE to read_failed sleep loop

[ Upstream commit 952041a8639a7a3a73a2b6573cb8aa8518bc39f8 ]

While reviewing rwsem down_slowpath, Will noticed ldsem had a copy of
a bug we just found for rwsem.

  X = 0;

  CPU0 CPU1

  rwsem_down_read()
    for (;;) {
      set_current_state(TASK_UNINTERRUPTIBLE);

                        X = 1;
                        rwsem_up_write();
                          rwsem_mark_wake()
                            atomic_long_add(adjustment, &sem->count);
                            smp_store_release(&waiter->task, NULL);

      if (!waiter.task)
        break;

      ...
    }

  r = X;

Allows 'r == 0'.

Reported-by: Will Deacon <will@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Will Deacon <will@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Fixes: 4898e640caf0 ("tty: Add timed, writer-prioritized rw semaphore")
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: scsi_dh_alua: always use a 2 second delay before retrying RTPG
Hannes Reinecke [Fri, 12 Jul 2019 06:53:47 +0000 (08:53 +0200)]
scsi: scsi_dh_alua: always use a 2 second delay before retrying RTPG

[ Upstream commit 20122994e38aef0ae50555884d287adde6641c94 ]

Retrying immediately after we've received a 'transitioning' sense code is
pretty much pointless, we should always use a delay before retrying.  So
ensure the default delay is applied before retrying.

Signed-off-by: Hannes Reinecke <hare@suse.com>
Tested-by: Zhangguanghui <zhang.guanghui@h3c.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: ibmvfc: fix WARN_ON during event pool release
Tyrel Datwyler [Wed, 17 Jul 2019 19:48:27 +0000 (14:48 -0500)]
scsi: ibmvfc: fix WARN_ON during event pool release

[ Upstream commit 5578257ca0e21056821e6481bd534ba267b84e58 ]

While removing an ibmvfc client adapter a WARN_ON like the following
WARN_ON is seen in the kernel log:

WARNING: CPU: 6 PID: 5421 at ./include/linux/dma-mapping.h:541
ibmvfc_free_event_pool+0x12c/0x1f0 [ibmvfc]
CPU: 6 PID: 5421 Comm: rmmod Tainted: G            E     4.17.0-rc1-next-20180419-autotest #1
NIP:  d00000000290328c LR: d00000000290325c CTR: c00000000036ee20
REGS: c000000288d1b7e0 TRAP: 0700   Tainted: G            E      (4.17.0-rc1-next-20180419-autotest)
MSR:  800000010282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE,TM[E]>  CR: 44008828  XER: 20000000
CFAR: c00000000036e408 SOFTE: 1
GPR00: d00000000290325c c000000288d1ba60 d000000002917900 c000000289d75448
GPR04: 0000000000000071 c0000000ff870000 0000000018040000 0000000000000001
GPR08: 0000000000000000 c00000000156e838 0000000000000001 d00000000290c640
GPR12: c00000000036ee20 c00000001ec4dc00 0000000000000000 0000000000000000
GPR16: 0000000000000000 0000000000000000 00000100276901e0 0000000010020598
GPR20: 0000000010020550 0000000010020538 0000000010020578 00000000100205b0
GPR24: 0000000000000000 0000000000000000 0000000010020590 5deadbeef0000100
GPR28: 5deadbeef0000200 d000000002910b00 0000000000000071 c0000002822f87d8
NIP [d00000000290328c] ibmvfc_free_event_pool+0x12c/0x1f0 [ibmvfc]
LR [d00000000290325c] ibmvfc_free_event_pool+0xfc/0x1f0 [ibmvfc]
Call Trace:
[c000000288d1ba60] [d00000000290325c] ibmvfc_free_event_pool+0xfc/0x1f0 [ibmvfc] (unreliable)
[c000000288d1baf0] [d000000002909390] ibmvfc_abort_task_set+0x7b0/0x8b0 [ibmvfc]
[c000000288d1bb70] [c0000000000d8c68] vio_bus_remove+0x68/0x100
[c000000288d1bbb0] [c0000000007da7c4] device_release_driver_internal+0x1f4/0x2d0
[c000000288d1bc00] [c0000000007da95c] driver_detach+0x7c/0x100
[c000000288d1bc40] [c0000000007d8af4] bus_remove_driver+0x84/0x140
[c000000288d1bcb0] [c0000000007db6ac] driver_unregister+0x4c/0xa0
[c000000288d1bd20] [c0000000000d6e7c] vio_unregister_driver+0x2c/0x50
[c000000288d1bd50] [d00000000290ba0c] cleanup_module+0x24/0x15e0 [ibmvfc]
[c000000288d1bd70] [c0000000001dadb0] sys_delete_module+0x220/0x2d0
[c000000288d1be30] [c00000000000b284] system_call+0x58/0x6c
Instruction dump:
e8410018 e87f0068 809f0078 e8bf0080 e8df0088 2fa30000 419e008c e9230200
2fa90000 419e0080 894d098a 794a07e0 <0b0a0000e9290008 2fa90000 419e0028

This is tripped as a result of irqs being disabled during the call to
dma_free_coherent() by ibmvfc_free_event_pool(). At this point in the code path
we have quiesced the adapter and its overly paranoid anyways to be holding the
host lock.

Reported-by: Abdul Haleem <abdhalee@linux.vnet.ibm.com>
Signed-off-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: megaraid_sas: fix panic on loading firmware crashdump
Junxiao Bi [Mon, 22 Jul 2019 16:15:24 +0000 (09:15 -0700)]
scsi: megaraid_sas: fix panic on loading firmware crashdump

[ Upstream commit 3b5f307ef3cb5022bfe3c8ca5b8f2114d5bf6c29 ]

While loading fw crashdump in function fw_crash_buffer_show(), left bytes
in one dma chunk was not checked, if copying size over it, overflow access
will cause kernel panic.

Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com>
Acked-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoARM: davinci: fix sleep.S build error on ARMv4
Arnd Bergmann [Mon, 22 Jul 2019 14:51:50 +0000 (16:51 +0200)]
ARM: davinci: fix sleep.S build error on ARMv4

[ Upstream commit d64b212ea960db4276a1d8372bd98cb861dfcbb0 ]

When building a multiplatform kernel that includes armv4 support,
the default target CPU does not support the blx instruction,
which leads to a build failure:

arch/arm/mach-davinci/sleep.S: Assembler messages:
arch/arm/mach-davinci/sleep.S:56: Error: selected processor does not support `blx ip' in ARM mode

Add a .arch statement in the sources to make this file build.

Link: https://lore.kernel.org/r/20190722145211.1154785-1-arnd@arndb.de
Acked-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoACPI/IORT: Fix off-by-one check in iort_dev_find_its_id()
Lorenzo Pieralisi [Mon, 22 Jul 2019 16:25:48 +0000 (17:25 +0100)]
ACPI/IORT: Fix off-by-one check in iort_dev_find_its_id()

[ Upstream commit 5a46d3f71d5e5a9f82eabc682f996f1281705ac7 ]

Static analysis identified that index comparison against ITS entries in
iort_dev_find_its_id() is off by one.

Update the comparison condition and clarify the resulting error
message.

Fixes: 4bf2efd26d76 ("ACPI: Add new IORT functions to support MSI domain handling")
Link: https://lore.kernel.org/linux-arm-kernel/20190613065410.GB16334@mwanda/
Reviewed-by: Hanjun Guo <guohanjun@huawei.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Will Deacon <will@kernel.org>
Cc: Hanjun Guo <guohanjun@huawei.com>
Cc: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agodrbd: dynamically allocate shash descriptor
Arnd Bergmann [Mon, 22 Jul 2019 12:26:34 +0000 (14:26 +0200)]
drbd: dynamically allocate shash descriptor

[ Upstream commit 77ce56e2bfaa64127ae5e23ef136c0168b818777 ]

Building with clang and KASAN, we get a warning about an overly large
stack frame on 32-bit architectures:

drivers/block/drbd/drbd_receiver.c:921:31: error: stack frame size of 1280 bytes in function 'conn_connect'
      [-Werror,-Wframe-larger-than=]

We already allocate other data dynamically in this function, so
just do the same for the shash descriptor, which makes up most of
this memory.

Link: https://lore.kernel.org/lkml/20190617132440.2721536-1-arnd@arndb.de/
Reviewed-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Roland Kammerer <roland.kammerer@linbit.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf probe: Avoid calling freeing routine multiple times for same pointer
Arnaldo Carvalho de Melo [Thu, 18 Jul 2019 14:28:37 +0000 (11:28 -0300)]
perf probe: Avoid calling freeing routine multiple times for same pointer

[ Upstream commit d95daf5accf4a72005daa13fbb1d1bd8709f2861 ]

When perf_add_probe_events() we call cleanup_perf_probe_events() for the
pev pointer it receives, then, as part of handling this failure the main
'perf probe' goes on and calls cleanup_params() and that will again call
cleanup_perf_probe_events()for the same pointer, so just set nevents to
zero when handling the failure of perf_add_probe_events() to avoid the
double free.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-x8qgma4g813z96dvtw9w219q@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoALSA: compress: Be more restrictive about when a drain is allowed
Charles Keepax [Mon, 22 Jul 2019 09:24:36 +0000 (10:24 +0100)]
ALSA: compress: Be more restrictive about when a drain is allowed

[ Upstream commit 3b8179944cb0dd53e5223996966746cdc8a60657 ]

Draining makes little sense in the situation of hardware overrun, as the
hardware will have consumed all its available samples. Additionally,
draining whilst the stream is paused would presumably get stuck as no
data is being consumed on the DSP side.

Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Acked-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoALSA: compress: Don't allow paritial drain operations on capture streams
Charles Keepax [Mon, 22 Jul 2019 09:24:35 +0000 (10:24 +0100)]
ALSA: compress: Don't allow paritial drain operations on capture streams

[ Upstream commit a70ab8a8645083f3700814e757f2940a88b7ef88 ]

Partial drain and next track are intended for gapless playback and
don't really have an obvious interpretation for a capture stream, so
makes sense to not allow those operations on capture streams.

Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Acked-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoALSA: compress: Prevent bypasses of set_params
Charles Keepax [Mon, 22 Jul 2019 09:24:34 +0000 (10:24 +0100)]
ALSA: compress: Prevent bypasses of set_params

[ Upstream commit 26c3f1542f5064310ad26794c09321780d00c57d ]

Currently, whilst in SNDRV_PCM_STATE_OPEN it is possible to call
snd_compr_stop, snd_compr_drain and snd_compr_partial_drain, which
allow a transition to SNDRV_PCM_STATE_SETUP. The stream should
only be able to move to the setup state once it has received a
SNDRV_COMPRESS_SET_PARAMS ioctl. Fix this issue by not allowing
those ioctls whilst in the open state.

Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Acked-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoALSA: compress: Fix regression on compressed capture streams
Charles Keepax [Mon, 22 Jul 2019 09:24:33 +0000 (10:24 +0100)]
ALSA: compress: Fix regression on compressed capture streams

[ Upstream commit 4475f8c4ab7b248991a60d9c02808dbb813d6be8 ]

A previous fix to the stop handling on compressed capture streams causes
some knock on issues. The previous fix updated snd_compr_drain_notify to
set the state back to PREPARED for capture streams. This causes some
issues however as the handling for snd_compr_poll differs between the
two states and some user-space applications were relying on the poll
failing after the stream had been stopped.

To correct this regression whilst still fixing the original problem the
patch was addressing, update the capture handling to skip the PREPARED
state rather than skipping the SETUP state as it has done until now.

Fixes: 4f2ab5e1d13d ("ALSA: compress: Fix stop handling on compressed capture streams")
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Acked-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agos390/qdio: add sanity checks to the fast-requeue path
Julian Wiedmann [Thu, 11 Jul 2019 16:17:36 +0000 (18:17 +0200)]
s390/qdio: add sanity checks to the fast-requeue path

[ Upstream commit a6ec414a4dd529eeac5c3ea51c661daba3397108 ]

If the device driver were to send out a full queue's worth of SBALs,
current code would end up discovering the last of those SBALs as PRIMED
and erroneously skip the SIGA-w. This immediately stalls the queue.

Add a check to not attempt fast-requeue in this case. While at it also
make sure that the state of the previous SBAL was successfully extracted
before inspecting it.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agocpufreq/pasemi: fix use-after-free in pas_cpufreq_cpu_init()
Wen Yang [Wed, 17 Jul 2019 03:55:04 +0000 (11:55 +0800)]
cpufreq/pasemi: fix use-after-free in pas_cpufreq_cpu_init()

[ Upstream commit e0a12445d1cb186d875410d093a00d215bec6a89 ]

The cpu variable is still being used in the of_get_property() call
after the of_node_put() call, which may result in use-after-free.

Fixes: a9acc26b75f6 ("cpufreq/pasemi: fix possible object reference leak")
Signed-off-by: Wen Yang <wen.yang99@zte.com.cn>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agohwmon: (nct6775) Fix register address and added missed tolerance for nct6106
Björn Gerhart [Mon, 15 Jul 2019 16:33:55 +0000 (18:33 +0200)]
hwmon: (nct6775) Fix register address and added missed tolerance for nct6106

[ Upstream commit f3d43e2e45fd9d44ba52d20debd12cd4ee9c89bf ]

Fixed address of third NCT6106_REG_WEIGHT_DUTY_STEP, and
added missed NCT6106_REG_TOLERANCE_H.

Fixes: 6c009501ff200 ("hwmon: (nct6775) Add support for NCT6102D/6106D")
Signed-off-by: Bjoern Gerhart <gerhart@posteo.de>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agomac80211: don't warn about CW params when not using them
Brian Norris [Thu, 18 Jul 2019 01:57:12 +0000 (18:57 -0700)]
mac80211: don't warn about CW params when not using them

[ Upstream commit d2b3fe42bc629c2d4002f652b3abdfb2e72991c7 ]

ieee80211_set_wmm_default() normally sets up the initial CW min/max for
each queue, except that it skips doing this if the driver doesn't
support ->conf_tx. We still end up calling drv_conf_tx() in some cases
(e.g., ieee80211_reconfig()), which also still won't do anything
useful...except it complains here about the invalid CW parameters.

Let's just skip the WARN if we weren't going to do anything useful with
the parameters.

Signed-off-by: Brian Norris <briannorris@chromium.org>
Link: https://lore.kernel.org/r/20190718015712.197499-1-briannorris@chromium.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoiscsi_ibft: make ISCSI_IBFT dependson ACPI instead of ISCSI_IBFT_FIND
Thomas Tai [Thu, 18 Jul 2019 18:37:34 +0000 (18:37 +0000)]
iscsi_ibft: make ISCSI_IBFT dependson ACPI instead of ISCSI_IBFT_FIND

[ Upstream commit 94bccc34071094c165c79b515d21b63c78f7e968 ]

iscsi_ibft can use ACPI to find the iBFT entry during bootup,
currently, ISCSI_IBFT depends on ISCSI_IBFT_FIND which is
a X86 legacy way to find the iBFT by searching through the
low memory. This patch changes the dependency so that other
arch like ARM64 can use ISCSI_IBFT as long as the arch supports
ACPI.

ibft_init() needs to use the global variable ibft_addr declared
in iscsi_ibft_find.c. A #ifndef CONFIG_ISCSI_IBFT_FIND is needed
to declare the variable if CONFIG_ISCSI_IBFT_FIND is not selected.
Moving ibft_addr into the iscsi_ibft.c does not work because if
ISCSI_IBFT is selected as a module, the arch/x86/kernel/setup.c won't
be able to find the variable at compile time.

Signed-off-by: Thomas Tai <thomas.tai@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agonetfilter: nfnetlink: avoid deadlock due to synchronous request_module
Florian Westphal [Tue, 2 Jul 2019 19:41:40 +0000 (21:41 +0200)]
netfilter: nfnetlink: avoid deadlock due to synchronous request_module

[ Upstream commit 1b0890cd60829bd51455dc5ad689ed58c4408227 ]

Thomas and Juliana report a deadlock when running:

(rmmod nf_conntrack_netlink/xfrm_user)

  conntrack -e NEW -E &
  modprobe -v xfrm_user

They provided following analysis:

conntrack -e NEW -E
    netlink_bind()
        netlink_lock_table() -> increases "nl_table_users"
            nfnetlink_bind()
            # does not unlock the table as it's locked by netlink_bind()
                __request_module()
                    call_usermodehelper_exec()

This triggers "modprobe nf_conntrack_netlink" from kernel, netlink_bind()
won't return until modprobe process is done.

"modprobe xfrm_user":
    xfrm_user_init()
        register_pernet_subsys()
            -> grab pernet_ops_rwsem
                ..
                netlink_table_grab()
                    calls schedule() as "nl_table_users" is non-zero

so modprobe is blocked because netlink_bind() increased
nl_table_users while also holding pernet_ops_rwsem.

"modprobe nf_conntrack_netlink" runs and inits nf_conntrack_netlink:
    ctnetlink_init()
        register_pernet_subsys()
            -> blocks on "pernet_ops_rwsem" thanks to xfrm_user module

both modprobe processes wait on one another -- neither can make
progress.

Switch netlink_bind() to "nowait" modprobe -- this releases the netlink
table lock, which then allows both modprobe instances to complete.

Reported-by: Thomas Jarosch <thomas.jarosch@intra2net.com>
Reported-by: Juliana Rodrigueiro <juliana.rodrigueiro@intra2net.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agocan: peak_usb: fix potential double kfree_skb()
Stephane Grosjean [Fri, 5 Jul 2019 13:32:16 +0000 (15:32 +0200)]
can: peak_usb: fix potential double kfree_skb()

commit fee6a8923ae0d318a7f7950c6c6c28a96cea099b upstream.

When closing the CAN device while tx skbs are inflight, echo skb could
be released twice. By calling close_candev() before unlinking all
pending tx urbs, then the internal echo_skb[] array is fully and
correctly cleared before the USB write callback and, therefore,
can_get_echo_skb() are called, for each aborted URB.

Fixes: bb4785551f64 ("can: usb: PEAK-System Technik USB adapters driver core")
Signed-off-by: Stephane Grosjean <s.grosjean@peak-system.com>
Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agousb: yurex: Fix use-after-free in yurex_delete
Suzuki K Poulose [Mon, 5 Aug 2019 11:15:28 +0000 (12:15 +0100)]
usb: yurex: Fix use-after-free in yurex_delete

commit fc05481b2fcabaaeccf63e32ac1baab54e5b6963 upstream.

syzbot reported the following crash [0]:

BUG: KASAN: use-after-free in usb_free_coherent+0x79/0x80
drivers/usb/core/usb.c:928
Read of size 8 at addr ffff8881b18599c8 by task syz-executor.4/16007

CPU: 0 PID: 16007 Comm: syz-executor.4 Not tainted 5.3.0-rc2+ #23
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Call Trace:
  __dump_stack lib/dump_stack.c:77 [inline]
  dump_stack+0xca/0x13e lib/dump_stack.c:113
  print_address_description+0x6a/0x32c mm/kasan/report.c:351
  __kasan_report.cold+0x1a/0x33 mm/kasan/report.c:482
  kasan_report+0xe/0x12 mm/kasan/common.c:612
  usb_free_coherent+0x79/0x80 drivers/usb/core/usb.c:928
  yurex_delete+0x138/0x330 drivers/usb/misc/yurex.c:100
  kref_put include/linux/kref.h:65 [inline]
  yurex_release+0x66/0x90 drivers/usb/misc/yurex.c:392
  __fput+0x2d7/0x840 fs/file_table.c:280
  task_work_run+0x13f/0x1c0 kernel/task_work.c:113
  tracehook_notify_resume include/linux/tracehook.h:188 [inline]
  exit_to_usermode_loop+0x1d2/0x200 arch/x86/entry/common.c:163
  prepare_exit_to_usermode arch/x86/entry/common.c:194 [inline]
  syscall_return_slowpath arch/x86/entry/common.c:274 [inline]
  do_syscall_64+0x45f/0x580 arch/x86/entry/common.c:299
  entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x413511
Code: 75 14 b8 03 00 00 00 0f 05 48 3d 01 f0 ff ff 0f 83 04 1b 00 00 c3 48
83 ec 08 e8 0a fc ff ff 48 89 04 24 b8 03 00 00 00 0f 05 <48> 8b 3c 24 48
89 c2 e8 53 fc ff ff 48 89 d0 48 83 c4 08 48 3d 01
RSP: 002b:00007ffc424ea2e0 EFLAGS: 00000293 ORIG_RAX: 0000000000000003
RAX: 0000000000000000 RBX: 0000000000000007 RCX: 0000000000413511
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000006
RBP: 0000000000000001 R08: 0000000029a2fc22 R09: 0000000029a2fc26
R10: 00007ffc424ea3c0 R11: 0000000000000293 R12: 000000000075c9a0
R13: 000000000075c9a0 R14: 0000000000761938 R15: ffffffffffffffff

Allocated by task 2776:
  save_stack+0x1b/0x80 mm/kasan/common.c:69
  set_track mm/kasan/common.c:77 [inline]
  __kasan_kmalloc mm/kasan/common.c:487 [inline]
  __kasan_kmalloc.constprop.0+0xbf/0xd0 mm/kasan/common.c:460
  kmalloc include/linux/slab.h:552 [inline]
  kzalloc include/linux/slab.h:748 [inline]
  usb_alloc_dev+0x51/0xf95 drivers/usb/core/usb.c:583
  hub_port_connect drivers/usb/core/hub.c:5004 [inline]
  hub_port_connect_change drivers/usb/core/hub.c:5213 [inline]
  port_event drivers/usb/core/hub.c:5359 [inline]
  hub_event+0x15c0/0x3640 drivers/usb/core/hub.c:5441
  process_one_work+0x92b/0x1530 kernel/workqueue.c:2269
  worker_thread+0x96/0xe20 kernel/workqueue.c:2415
  kthread+0x318/0x420 kernel/kthread.c:255
  ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352

Freed by task 16007:
  save_stack+0x1b/0x80 mm/kasan/common.c:69
  set_track mm/kasan/common.c:77 [inline]
  __kasan_slab_free+0x130/0x180 mm/kasan/common.c:449
  slab_free_hook mm/slub.c:1423 [inline]
  slab_free_freelist_hook mm/slub.c:1470 [inline]
  slab_free mm/slub.c:3012 [inline]
  kfree+0xe4/0x2f0 mm/slub.c:3953
  device_release+0x71/0x200 drivers/base/core.c:1064
  kobject_cleanup lib/kobject.c:693 [inline]
  kobject_release lib/kobject.c:722 [inline]
  kref_put include/linux/kref.h:65 [inline]
  kobject_put+0x171/0x280 lib/kobject.c:739
  put_device+0x1b/0x30 drivers/base/core.c:2213
  usb_put_dev+0x1f/0x30 drivers/usb/core/usb.c:725
  yurex_delete+0x40/0x330 drivers/usb/misc/yurex.c:95
  kref_put include/linux/kref.h:65 [inline]
  yurex_release+0x66/0x90 drivers/usb/misc/yurex.c:392
  __fput+0x2d7/0x840 fs/file_table.c:280
  task_work_run+0x13f/0x1c0 kernel/task_work.c:113
  tracehook_notify_resume include/linux/tracehook.h:188 [inline]
  exit_to_usermode_loop+0x1d2/0x200 arch/x86/entry/common.c:163
  prepare_exit_to_usermode arch/x86/entry/common.c:194 [inline]
  syscall_return_slowpath arch/x86/entry/common.c:274 [inline]
  do_syscall_64+0x45f/0x580 arch/x86/entry/common.c:299
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

The buggy address belongs to the object at ffff8881b1859980
  which belongs to the cache kmalloc-2k of size 2048
The buggy address is located 72 bytes inside of
  2048-byte region [ffff8881b1859980ffff8881b185a180)
The buggy address belongs to the page:
page:ffffea0006c61600 refcount:1 mapcount:0 mapping:ffff8881da00c000
index:0x0 compound_mapcount: 0
flags: 0x200000000010200(slab|head)
raw: 0200000000010200 0000000000000000 0000000100000001 ffff8881da00c000
raw: 0000000000000000 00000000000f000f 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
  ffff8881b1859880: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
  ffff8881b1859900: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
ffff8881b1859980: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
                                               ^
  ffff8881b1859a00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
  ffff8881b1859a80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================

A quick look at the yurex_delete() shows that we drop the reference
to the usb_device before releasing any buffers associated with the
device. Delay the reference drop until we have finished the cleanup.

[0] https://lore.kernel.org/lkml/0000000000003f86d8058f0bd671@google.com/

Fixes: 6bc235a2e24a5e ("USB: add driver for Meywa-Denki & Kayac YUREX")
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Tomoki Sekiyama <tomoki.sekiyama@gmail.com>
Cc: Oliver Neukum <oneukum@suse.com>
Cc: andreyknvl@google.com
Cc: gregkh@linuxfoundation.org
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: syzkaller-bugs@googlegroups.com
Cc: dtor@chromium.org
Reported-by: syzbot+d1fedb1c1fdb07fca507@syzkaller.appspotmail.com
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: stable <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20190805111528.6758-1-suzuki.poulose@arm.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoperf record: Fix module size on s390
Thomas Richter [Wed, 24 Jul 2019 12:27:02 +0000 (14:27 +0200)]
perf record: Fix module size on s390

commit 12a6d2940b5f02b4b9f71ce098e3bb02bc24a9ea upstream.

On s390 the modules loaded in memory have the text segment located after
the GOT and Relocation table. This can be seen with this output:

  [root@m35lp76 perf]# fgrep qeth /proc/modules
  qeth 151552 1 qeth_l2, Live 0x000003ff800b2000
  ...
  [root@m35lp76 perf]# cat /sys/module/qeth/sections/.text
  0x000003ff800b3990
  [root@m35lp76 perf]#

There is an offset of 0x1990 bytes. The size of the qeth module is
151552 bytes (0x25000 in hex).

The location of the GOT/relocation table at the beginning of a module is
unique to s390.

commit 203d8a4aa6ed ("perf s390: Fix 'start' address of module's map")
adjusts the start address of a module in the map structures, but does
not adjust the size of the modules. This leads to overlapping of module
maps as this example shows:

[root@m35lp76 perf] # ./perf report -D
     0 0 0xfb0 [0xa0]: PERF_RECORD_MMAP -1/0: [0x3ff800b3990(0x25000)
          @ 0]:  x /lib/modules/.../qeth.ko.xz
     0 0 0x1050 [0xb0]: PERF_RECORD_MMAP -1/0: [0x3ff800d85a0(0x8000)
          @ 0]:  x /lib/modules/.../ip6_tables.ko.xz

The module qeth.ko has an adjusted start address modified to b3990, but
its size is unchanged and the module ends at 0x3ff800d8990.  This end
address overlaps with the next modules start address of 0x3ff800d85a0.

When the size of the leading GOT/Relocation table stored in the
beginning of the text segment (0x1990 bytes) is subtracted from module
qeth end address, there are no overlaps anymore:

   0x3ff800d8990 - 0x1990 = 0x0x3ff800d7000

which is the same as

   0x3ff800b2000 + 0x25000 = 0x0x3ff800d7000.

To fix this issue, also adjust the modules size in function
arch__fix_module_text_start(). Add another function parameter named size
and reduce the size of the module when the text segment start address is
changed.

Output after:
     0 0 0xfb0 [0xa0]: PERF_RECORD_MMAP -1/0: [0x3ff800b3990(0x23670)
          @ 0]:  x /lib/modules/.../qeth.ko.xz
     0 0 0x1050 [0xb0]: PERF_RECORD_MMAP -1/0: [0x3ff800d85a0(0x7a60)
          @ 0]:  x /lib/modules/.../ip6_tables.ko.xz

Reported-by: Stefan Liebler <stli@linux.ibm.com>
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: stable@vger.kernel.org
Fixes: 203d8a4aa6ed ("perf s390: Fix 'start' address of module's map")
Link: http://lkml.kernel.org/r/20190724122703.3996-1-tmricht@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoperf db-export: Fix thread__exec_comm()
Adrian Hunter [Thu, 8 Aug 2019 06:48:23 +0000 (09:48 +0300)]
perf db-export: Fix thread__exec_comm()

commit 3de7ae0b2a1d86dbb23d0cb135150534fdb2e836 upstream.

Threads synthesized from /proc have comms with a start time of zero, and
not marked as "exec". Currently, there can be 2 such comms. The first is
created by processing a synthesized fork event and is set to the
parent's comm string, and the second by processing a synthesized comm
event set to the thread's current comm string.

In the absence of an "exec" comm, thread__exec_comm() picks the last
(oldest) comm, which, in the case above, is the parent's comm string.
For a main thread, that is very probably wrong. Use the second-to-last
in that case.

This affects only db-export because it is the only user of
thread__exec_comm().

Example:

  $ sudo perf record -a -o pt-a-sleep-1 -e intel_pt//u -- sleep 1
  $ sudo chown ahunter pt-a-sleep-1

Before:

  $ perf script -i pt-a-sleep-1 --itrace=bep -s tools/perf/scripts/python/export-to-sqlite.py pt-a-sleep-1.db branches calls
  $ sqlite3 -header -column pt-a-sleep-1.db 'select * from comm_threads_view'
  comm_id     command     thread_id   pid         tid
  ----------  ----------  ----------  ----------  ----------
  1           swapper     1           0           0
  2           rcu_sched   2           10          10
  3           kthreadd    3           78          78
  5           sudo        4           15180       15180
  5           sudo        5           15180       15182
  7           kworker/4:  6           10335       10335
  8           kthreadd    7           55          55
  10          systemd     8           865         865
  10          systemd     9           865         875
  13          perf        10          15181       15181
  15          sleep       10          15181       15181
  16          kworker/3:  11          14179       14179
  17          kthreadd    12          29376       29376
  19          systemd     13          746         746
  21          systemd     14          401         401
  23          systemd     15          879         879
  23          systemd     16          879         945
  25          kthreadd    17          556         556
  27          kworker/u1  18          14136       14136
  28          kworker/u1  19          15021       15021
  29          kthreadd    20          509         509
  31          systemd     21          836         836
  31          systemd     22          836         967
  33          systemd     23          1148        1148
  33          systemd     24          1148        1163
  35          kworker/2:  25          17988       17988
  36          kworker/0:  26          13478       13478

After:

  $ perf script -i pt-a-sleep-1 --itrace=bep -s tools/perf/scripts/python/export-to-sqlite.py pt-a-sleep-1b.db branches calls
  $ sqlite3 -header -column pt-a-sleep-1b.db 'select * from comm_threads_view'
  comm_id     command     thread_id   pid         tid
  ----------  ----------  ----------  ----------  ----------
  1           swapper     1           0           0
  2           rcu_sched   2           10          10
  3           kswapd0     3           78          78
  4           perf        4           15180       15180
  4           perf        5           15180       15182
  6           kworker/4:  6           10335       10335
  7           kcompactd0  7           55          55
  8           accounts-d  8           865         865
  8           accounts-d  9           865         875
  10          perf        10          15181       15181
  12          sleep       10          15181       15181
  13          kworker/3:  11          14179       14179
  14          kworker/1:  12          29376       29376
  15          haveged     13          746         746
  16          systemd-jo  14          401         401
  17          NetworkMan  15          879         879
  17          NetworkMan  16          879         945
  19          irq/131-iw  17          556         556
  20          kworker/u1  18          14136       14136
  21          kworker/u1  19          15021       15021
  22          kworker/u1  20          509         509
  23          thermald    21          836         836
  23          thermald    22          836         967
  25          unity-sett  23          1148        1148
  25          unity-sett  24          1148        1163
  27          kworker/2:  25          17988       17988
  28          kworker/0:  26          13478       13478

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: stable@vger.kernel.org
Fixes: 65de51f93ebf ("perf tools: Identify which comms are from exec")
Link: http://lkml.kernel.org/r/20190808064823.14846-1-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoperf record: Fix wrong size in perf_record_mmap for last kernel module
Thomas Richter [Thu, 3 Aug 2017 13:49:02 +0000 (15:49 +0200)]
perf record: Fix wrong size in perf_record_mmap for last kernel module

commit 9ad4652b66f19a60f07e63b942b80b5c2d7465bf upstream.

During work on perf report for s390 I ran into the following issue:

0 0x318 [0x78]: PERF_RECORD_MMAP -1/0:
        [0x3ff804d6990(0xfffffc007fb2966f) @ 0]:
        x /lib/modules/4.12.0perf1+/kernel/drivers/s390/net/qeth_l2.ko

This is a PERF_RECORD_MMAP entry of the perf.data file with an invalid
module size for qeth_l2.ko (the s390 ethernet device driver).

Even a mainframe does not have 0xfffffc007fb2966f bytes of main memory.

It turned out that this wrong size is created by the perf record
command.  What happens is this function call sequence from
__cmd_record():

  perf_session__new():
    perf_session__create_kernel_maps():
      machine__create_kernel_maps():
        machine__create_modules():   Creates map for all loaded kernel modules.
          modules__parse():   Reads /proc/modules and extracts module name and
                              load address (1st and last column)
            machine__create_module():   Called for every module found in /proc/modules.
                              Creates a new map for every module found and enters
                              module name and start address into the map. Since the
                              module end address is unknown it is set to zero.

This ends up with a kernel module map list sorted by module start
addresses.  All module end addresses are zero.

Last machine__create_kernel_maps() calls function map_groups__fixup_end().
This function iterates through the maps and assigns each map entry's
end address the successor map entry start address. The last entry of the
map group has no successor, so ~0 is used as end to consume the remaining
memory.

Later __cmd_record calls function record__synthesize() which in turn calls
perf_event__synthesize_kernel_mmap() and perf_event__synthesize_modules()
to create PERF_REPORT_MMAP entries into the perf.data file.

On s390 this results in the last module qeth_l2.ko
(which has highest start address, see module table:
        [root@s8360047 perf]# cat /proc/modules
        qeth_l2 86016 1 - Live 0x000003ff804d6000
        qeth 266240 1 qeth_l2, Live 0x000003ff80296000
        ccwgroup 24576 1 qeth, Live 0x000003ff80218000
        vmur 36864 0 - Live 0x000003ff80182000
        qdio 143360 2 qeth_l2,qeth, Live 0x000003ff80002000
        [root@s8360047 perf]# )
to be the last entry and its map has an end address of ~0.

When the PERF_RECORD_MMAP entry is created for kernel module qeth_l2.ko
its start address and length is written. The length is calculated in line:
    event->mmap.len   = pos->end - pos->start;
and results in 0xffffffffffffffff - 0x3ff804d6990(*) = 0xfffffc007fb2966f

(*) On s390 the module start address is actually determined by a __weak function
named arch__fix_module_text_start() in machine__create_module().

I think this improvable. We can use the module size (2nd column of /proc/modules)
to get each loaded kernel module size and calculate its end address.
Only for map entries which do not have a valid end address (end is still zero)
we can use the heuristic we have now, that is use successor start address or ~0.

Signed-off-by: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com>
Reviewed-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com>
Cc: Zvonko Kosic <zvonko.kosic@de.ibm.com>
LPU-Reference: 20170803134902.47207-2-tmricht@linux.vnet.ibm.com
Link: http://lkml.kernel.org/n/tip-nmoqij5b5vxx7rq2ckwu8iaj@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Daniel Daz <daniel.diaz@linaro.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomm/vmalloc: Sync unmappings in __purge_vmap_area_lazy()
Joerg Roedel [Fri, 19 Jul 2019 18:46:52 +0000 (20:46 +0200)]
mm/vmalloc: Sync unmappings in __purge_vmap_area_lazy()

commit 3f8fd02b1bf1d7ba964485a56f2f4b53ae88c167 upstream.

On x86-32 with PTI enabled, parts of the kernel page-tables are not shared
between processes. This can cause mappings in the vmalloc/ioremap area to
persist in some page-tables after the region is unmapped and released.

When the region is re-used the processes with the old mappings do not fault
in the new mappings but still access the old ones.

This causes undefined behavior, in reality often data corruption, kernel
oopses and panics and even spontaneous reboots.

Fix this problem by activly syncing unmaps in the vmalloc/ioremap area to
all page-tables in the system before the regions can be re-used.

References: https://bugzilla.suse.com/show_bug.cgi?id=1118689
Fixes: 5d72b4fba40ef ('x86, mm: support huge I/O mapping capability I/F')
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lkml.kernel.org/r/20190719184652.11391-4-joro@8bytes.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agox86/mm: Sync also unmappings in vmalloc_sync_all()
Joerg Roedel [Fri, 19 Jul 2019 18:46:51 +0000 (20:46 +0200)]
x86/mm: Sync also unmappings in vmalloc_sync_all()

commit 8e998fc24de47c55b47a887f6c95ab91acd4a720 upstream.

With huge-page ioremap areas the unmappings also need to be synced between
all page-tables. Otherwise it can cause data corruption when a region is
unmapped and later re-used.

Make the vmalloc_sync_one() function ready to sync unmappings and make sure
vmalloc_sync_all() iterates over all page-tables even when an unmapped PMD
is found.

Fixes: 5d72b4fba40ef ('x86, mm: support huge I/O mapping capability I/F')
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lkml.kernel.org/r/20190719184652.11391-3-joro@8bytes.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agox86/mm: Check for pfn instead of page in vmalloc_sync_one()
Joerg Roedel [Fri, 19 Jul 2019 18:46:50 +0000 (20:46 +0200)]
x86/mm: Check for pfn instead of page in vmalloc_sync_one()

commit 51b75b5b563a2637f9d8dc5bd02a31b2ff9e5ea0 upstream.

Do not require a struct page for the mapped memory location because it
might not exist. This can happen when an ioremapped region is mapped with
2MB pages.

Fixes: 5d72b4fba40ef ('x86, mm: support huge I/O mapping capability I/F')
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lkml.kernel.org/r/20190719184652.11391-2-joro@8bytes.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agosound: fix a memory leak bug
Wenwen Wang [Thu, 8 Aug 2019 05:15:21 +0000 (00:15 -0500)]
sound: fix a memory leak bug

commit c7cd7c748a3250ca33509f9235efab9c803aca09 upstream.

In sound_insert_unit(), the controlling structure 's' is allocated through
kmalloc(). Then it is added to the sound driver list by invoking
__sound_insert_unit(). Later on, if __register_chrdev() fails, 's' is
removed from the list through __sound_remove_unit(). If 'index' is not less
than 0, -EBUSY is returned to indicate the error. However, 's' is not
deallocated on this execution path, leading to a memory leak bug.

To fix the above issue, free 's' before -EBUSY is returned.

Signed-off-by: Wenwen Wang <wenwen@cs.uga.edu>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agousb: iowarrior: fix deadlock on disconnect
Oliver Neukum [Thu, 8 Aug 2019 09:27:28 +0000 (11:27 +0200)]
usb: iowarrior: fix deadlock on disconnect

commit c468a8aa790e0dfe0a7f8a39db282d39c2c00b46 upstream.

We have to drop the mutex before we close() upon disconnect()
as close() needs the lock. This is safe to do by dropping the
mutex as intfdata is already set to NULL, so open() will fail.

Fixes: 03f36e885fc26 ("USB: open disconnect race in iowarrior")
Reported-by: syzbot+a64a382964bf6c71a9c0@syzkaller.appspotmail.com
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Link: https://lore.kernel.org/r/20190808092728.23417-1-oneukum@suse.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agousb: usbfs: fix double-free of usb memory upon submiturb error
Gavin Li [Sun, 4 Aug 2019 23:50:44 +0000 (16:50 -0700)]
usb: usbfs: fix double-free of usb memory upon submiturb error

commit c43f28dfdc4654e738aa6d3fd08a105b2bee758d upstream.

Upon an error within proc_do_submiturb(), dec_usb_memory_use_count()
gets called once by the error handling tail and again by free_async().
Remove the first call.

Signed-off-by: Gavin Li <git@thegavinli.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20190804235044.22327-1-gavinli@thegavinli.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoLinux 4.9.189 v4.9.189
Greg Kroah-Hartman [Sun, 11 Aug 2019 10:22:20 +0000 (12:22 +0200)]
Linux 4.9.189

5 years agox86/speculation/swapgs: Exclude ATOMs from speculation through SWAPGS
Thomas Gleixner [Wed, 17 Jul 2019 19:18:59 +0000 (21:18 +0200)]
x86/speculation/swapgs: Exclude ATOMs from speculation through SWAPGS

commit f36cf386e3fec258a341d446915862eded3e13d8 upstream.

Intel provided the following information:

 On all current Atom processors, instructions that use a segment register
 value (e.g. a load or store) will not speculatively execute before the
 last writer of that segment retires. Thus they will not use a
 speculatively written segment value.

That means on ATOMs there is no speculation through SWAPGS, so the SWAPGS
entry paths can be excluded from the extra LFENCE if PTI is disabled.

Create a separate bug flag for the through SWAPGS speculation and mark all
out-of-order ATOMs and AMD/HYGON CPUs as not affected. The in-order ATOMs
are excluded from the whole mitigation mess anyway.

Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Tyler Hicks <tyhicks@canonical.com>
Reviewed-by: Josh Poimboeuf <jpoimboe@redhat.com>
[bwh: Backported to 4.4:
 - There's no whitelist entry (or any support) for Hygon CPUs
 - Adjust context, indentation]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agox86/entry/64: Use JMP instead of JMPQ
Josh Poimboeuf [Mon, 15 Jul 2019 16:51:39 +0000 (11:51 -0500)]
x86/entry/64: Use JMP instead of JMPQ

commit 64dbc122b20f75183d8822618c24f85144a5a94d upstream.

Somehow the swapgs mitigation entry code patch ended up with a JMPQ
instruction instead of JMP, where only the short jump is needed.  Some
assembler versions apparently fail to optimize JMPQ into a two-byte JMP
when possible, instead always using a 7-byte JMP with relocation.  For
some reason that makes the entry code explode with a #GP during boot.

Change it back to "JMP" as originally intended.

Fixes: 18ec54fdd6d1 ("x86/speculation: Prepare entry code for Spectre v1 swapgs mitigations")
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[bwh: Backported to 4.9: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agox86/speculation: Enable Spectre v1 swapgs mitigations
Josh Poimboeuf [Mon, 8 Jul 2019 16:52:26 +0000 (11:52 -0500)]
x86/speculation: Enable Spectre v1 swapgs mitigations

commit a2059825986a1c8143fd6698774fa9d83733bb11 upstream.

The previous commit added macro calls in the entry code which mitigate the
Spectre v1 swapgs issue if the X86_FEATURE_FENCE_SWAPGS_* features are
enabled.  Enable those features where applicable.

The mitigations may be disabled with "nospectre_v1" or "mitigations=off".

There are different features which can affect the risk of attack:

- When FSGSBASE is enabled, unprivileged users are able to place any
  value in GS, using the wrgsbase instruction.  This means they can
  write a GS value which points to any value in kernel space, which can
  be useful with the following gadget in an interrupt/exception/NMI
  handler:

if (coming from user space)
swapgs
mov %gs:<percpu_offset>, %reg1
// dependent load or store based on the value of %reg
// for example: mov %(reg1), %reg2

  If an interrupt is coming from user space, and the entry code
  speculatively skips the swapgs (due to user branch mistraining), it
  may speculatively execute the GS-based load and a subsequent dependent
  load or store, exposing the kernel data to an L1 side channel leak.

  Note that, on Intel, a similar attack exists in the above gadget when
  coming from kernel space, if the swapgs gets speculatively executed to
  switch back to the user GS.  On AMD, this variant isn't possible
  because swapgs is serializing with respect to future GS-based
  accesses.

  NOTE: The FSGSBASE patch set hasn't been merged yet, so the above case
doesn't exist quite yet.

- When FSGSBASE is disabled, the issue is mitigated somewhat because
  unprivileged users must use prctl(ARCH_SET_GS) to set GS, which
  restricts GS values to user space addresses only.  That means the
  gadget would need an additional step, since the target kernel address
  needs to be read from user space first.  Something like:

if (coming from user space)
swapgs
mov %gs:<percpu_offset>, %reg1
mov (%reg1), %reg2
// dependent load or store based on the value of %reg2
// for example: mov %(reg2), %reg3

  It's difficult to audit for this gadget in all the handlers, so while
  there are no known instances of it, it's entirely possible that it
  exists somewhere (or could be introduced in the future).  Without
  tooling to analyze all such code paths, consider it vulnerable.

  Effects of SMAP on the !FSGSBASE case:

  - If SMAP is enabled, and the CPU reports RDCL_NO (i.e., not
    susceptible to Meltdown), the kernel is prevented from speculatively
    reading user space memory, even L1 cached values.  This effectively
    disables the !FSGSBASE attack vector.

  - If SMAP is enabled, but the CPU *is* susceptible to Meltdown, SMAP
    still prevents the kernel from speculatively reading user space
    memory.  But it does *not* prevent the kernel from reading the
    user value from L1, if it has already been cached.  This is probably
    only a small hurdle for an attacker to overcome.

Thanks to Dave Hansen for contributing the speculative_smap() function.

Thanks to Andrew Cooper for providing the inside scoop on whether swapgs
is serializing on AMD.

[ tglx: Fixed the USER fence decision and polished the comment as suggested
   by Dave Hansen ]

Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Dave Hansen <dave.hansen@intel.com>
[bwh: Backported to 4.9:
 - Check for X86_FEATURE_KAISER instead of X86_FEATURE_PTI
 - mitigations= parameter is x86-only here
 - Adjust filename, context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agox86/speculation: Prepare entry code for Spectre v1 swapgs mitigations
Josh Poimboeuf [Mon, 8 Jul 2019 16:52:25 +0000 (11:52 -0500)]
x86/speculation: Prepare entry code for Spectre v1 swapgs mitigations

commit 18ec54fdd6d18d92025af097cd042a75cf0ea24c upstream.

Spectre v1 isn't only about array bounds checks.  It can affect any
conditional checks.  The kernel entry code interrupt, exception, and NMI
handlers all have conditional swapgs checks.  Those may be problematic in
the context of Spectre v1, as kernel code can speculatively run with a user
GS.

For example:

if (coming from user space)
swapgs
mov %gs:<percpu_offset>, %reg
mov (%reg), %reg1

When coming from user space, the CPU can speculatively skip the swapgs, and
then do a speculative percpu load using the user GS value.  So the user can
speculatively force a read of any kernel value.  If a gadget exists which
uses the percpu value as an address in another load/store, then the
contents of the kernel value may become visible via an L1 side channel
attack.

A similar attack exists when coming from kernel space.  The CPU can
speculatively do the swapgs, causing the user GS to get used for the rest
of the speculative window.

The mitigation is similar to a traditional Spectre v1 mitigation, except:

  a) index masking isn't possible; because the index (percpu offset)
     isn't user-controlled; and

  b) an lfence is needed in both the "from user" swapgs path and the
     "from kernel" non-swapgs path (because of the two attacks described
     above).

The user entry swapgs paths already have SWITCH_TO_KERNEL_CR3, which has a
CR3 write when PTI is enabled.  Since CR3 writes are serializing, the
lfences can be skipped in those cases.

On the other hand, the kernel entry swapgs paths don't depend on PTI.

To avoid unnecessary lfences for the user entry case, create two separate
features for alternative patching:

  X86_FEATURE_FENCE_SWAPGS_USER
  X86_FEATURE_FENCE_SWAPGS_KERNEL

Use these features in entry code to patch in lfences where needed.

The features aren't enabled yet, so there's no functional change.

Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Dave Hansen <dave.hansen@intel.com>
[bwh: Backported to 4.9:
 - Assign the CPU feature bits from word 7
 - Add FENCE_SWAPGS_KERNEL_ENTRY to NMI entry, since it does not
   use paranoid_entry
 - Include <asm/cpufeatures.h> in calling.h
 - Adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agox86: cpufeatures: Sort feature word 7
Ben Hutchings [Thu, 8 Aug 2019 19:03:32 +0000 (20:03 +0100)]
x86: cpufeatures: Sort feature word 7

This will make it clearer which bits are allocated, in case we need to
assign more feature bits for later backports.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agospi: bcm2835: Fix 3-wire mode if DMA is enabled
Lukas Wunner [Wed, 3 Jul 2019 10:29:31 +0000 (12:29 +0200)]
spi: bcm2835: Fix 3-wire mode if DMA is enabled

commit 8d8bef50365847134b51c1ec46786bc2873e4e47 upstream.

Commit 6935224da248 ("spi: bcm2835: enable support of 3-wire mode")
added 3-wire support to the BCM2835 SPI driver by setting the REN bit
(Read Enable) in the CS register when receiving data.  The REN bit puts
the transmitter in high-impedance state.  The driver recognizes that
data is to be received by checking whether the rx_buf of a transfer is
non-NULL.

Commit 3ecd37edaa2a ("spi: bcm2835: enable dma modes for transfers
meeting certain conditions") subsequently broke 3-wire support because
it set the SPI_MASTER_MUST_RX flag which causes spi_map_msg() to replace
rx_buf with a dummy buffer if it is NULL.  As a result, rx_buf is
*always* non-NULL if DMA is enabled.

Reinstate 3-wire support by not only checking whether rx_buf is non-NULL,
but also checking that it is not the dummy buffer.

Fixes: 3ecd37edaa2a ("spi: bcm2835: enable dma modes for transfers meeting certain conditions")
Reported-by: Nuno Sá <nuno.sa@analog.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Cc: stable@vger.kernel.org # v4.2+
Cc: Martin Sperl <kernel@martin.sperl.org>
Acked-by: Stefan Wahren <wahrenst@gmx.net>
Link: https://lore.kernel.org/r/328318841455e505370ef8ecad97b646c033dc8a.1562148527.git.lukas@wunner.de
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoblock: blk_init_allocated_queue() set q->fq as NULL in the fail case
xiao jin [Mon, 30 Jul 2018 06:11:12 +0000 (14:11 +0800)]
block: blk_init_allocated_queue() set q->fq as NULL in the fail case

commit 54648cf1ec2d7f4b6a71767799c45676a138ca24 upstream.

We find the memory use-after-free issue in __blk_drain_queue()
on the kernel 4.14. After read the latest kernel 4.18-rc6 we
think it has the same problem.

Memory is allocated for q->fq in the blk_init_allocated_queue().
If the elevator init function called with error return, it will
run into the fail case to free the q->fq.

Then the __blk_drain_queue() uses the same memory after the free
of the q->fq, it will lead to the unpredictable event.

The patch is to set q->fq as NULL in the fail case of
blk_init_allocated_queue().

Fixes: commit 7c94e1c157a2 ("block: introduce blk_flush_queue to drive flush machinery")
Cc: <stable@vger.kernel.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: xiao jin <jin.xiao@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
[groeck: backport to v4.4.y/v4.9.y (context change)]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Alessio Balsini <balsini@android.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agobnx2x: Disable multi-cos feature.
Sudarsana Reddy Kalluru [Wed, 24 Jul 2019 02:32:41 +0000 (19:32 -0700)]
bnx2x: Disable multi-cos feature.

[ Upstream commit d1f0b5dce8fda09a7f5f04c1878f181d548e42f5 ]

Commit 3968d38917eb ("bnx2x: Fix Multi-Cos.") which enabled multi-cos
feature after prolonged time in driver added some regression causing
numerous issues (sudden reboots, tx timeout etc.) reported by customers.
We plan to backout this commit and submit proper fix once we have root
cause of issues reported with this feature enabled.

Fixes: 3968d38917eb ("bnx2x: Fix Multi-Cos.")
Signed-off-by: Sudarsana Reddy Kalluru <skalluru@marvell.com>
Signed-off-by: Manish Chopra <manishc@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoife: error out when nla attributes are empty
Cong Wang [Tue, 23 Jul 2019 04:43:00 +0000 (21:43 -0700)]
ife: error out when nla attributes are empty

[ Upstream commit c8ec4632c6ac9cda0e8c3d51aa41eeab66585bd5 ]

act_ife at least requires TCA_IFE_PARMS, so we have to bail out
when there is no attribute passed in.

Reported-by: syzbot+fbb5b288c9cb6a2eeac4@syzkaller.appspotmail.com
Fixes: ef6980b6becb ("introduce IFE action")
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoip6_tunnel: fix possible use-after-free on xmit
Haishuang Yan [Thu, 25 Jul 2019 16:40:17 +0000 (00:40 +0800)]
ip6_tunnel: fix possible use-after-free on xmit

[ Upstream commit 01f5bffad555f8e22a61f4b1261fe09cf1b96994 ]

ip4ip6/ip6ip6 tunnels run iptunnel_handle_offloads on xmit which
can cause a possible use-after-free accessing iph/ipv6h pointer
since the packet will be 'uncloned' running pskb_expand_head if
it is a cloned gso skb.

Fixes: 0e9a709560db ("ip6_tunnel, ip6_gre: fix setting of DSCP on encapsulated packets")
Signed-off-by: Haishuang Yan <yanhaishuang@cmss.chinamobile.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agocompat_ioctl: pppoe: fix PPPOEIOCSFWD handling
Arnd Bergmann [Tue, 30 Jul 2019 19:25:20 +0000 (21:25 +0200)]
compat_ioctl: pppoe: fix PPPOEIOCSFWD handling

[ Upstream commit 055d88242a6046a1ceac3167290f054c72571cd9 ]

Support for handling the PPPOEIOCSFWD ioctl in compat mode was added in
linux-2.5.69 along with hundreds of other commands, but was always broken
sincen only the structure is compatible, but the command number is not,
due to the size being sizeof(size_t), or at first sizeof(sizeof((struct
sockaddr_pppox)), which is different on 64-bit architectures.

Guillaume Nault adds:

  And the implementation was broken until 2016 (see 29e73269aa4d ("pppoe:
  fix reference counting in PPPoE proxy")), and nobody ever noticed. I
  should probably have removed this ioctl entirely instead of fixing it.
  Clearly, it has never been used.

Fix it by adding a compat_ioctl handler for all pppoe variants that
translates the command number and then calls the regular ioctl function.

All other ioctl commands handled by pppoe are compatible between 32-bit
and 64-bit, and require compat_ptr() conversion.

This should apply to all stable kernels.

Acked-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agotipc: compat: allow tipc commands without arguments
Taras Kondratiuk [Mon, 29 Jul 2019 22:15:07 +0000 (22:15 +0000)]
tipc: compat: allow tipc commands without arguments

[ Upstream commit 4da5f0018eef4c0de31675b670c80e82e13e99d1 ]

Commit 2753ca5d9009 ("tipc: fix uninit-value in tipc_nl_compat_doit")
broke older tipc tools that use compat interface (e.g. tipc-config from
tipcutils package):

% tipc-config -p
operation not supported

The commit started to reject TIPC netlink compat messages that do not
have attributes. It is too restrictive because some of such messages are
valid (they don't need any arguments):

% grep 'tx none' include/uapi/linux/tipc_config.h
#define  TIPC_CMD_NOOP              0x0000    /* tx none, rx none */
#define  TIPC_CMD_GET_MEDIA_NAMES   0x0002    /* tx none, rx media_name(s) */
#define  TIPC_CMD_GET_BEARER_NAMES  0x0003    /* tx none, rx bearer_name(s) */
#define  TIPC_CMD_SHOW_PORTS        0x0006    /* tx none, rx ultra_string */
#define  TIPC_CMD_GET_REMOTE_MNG    0x4003    /* tx none, rx unsigned */
#define  TIPC_CMD_GET_MAX_PORTS     0x4004    /* tx none, rx unsigned */
#define  TIPC_CMD_GET_NETID         0x400B    /* tx none, rx unsigned */
#define  TIPC_CMD_NOT_NET_ADMIN     0xC001    /* tx none, rx none */

This patch relaxes the original fix and rejects messages without
arguments only if such arguments are expected by a command (reg_type is
non zero).

Fixes: 2753ca5d9009 ("tipc: fix uninit-value in tipc_nl_compat_doit")
Cc: stable@vger.kernel.org
Signed-off-by: Taras Kondratiuk <takondra@cisco.com>
Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: sched: Fix a possible null-pointer dereference in dequeue_func()
Jia-Ju Bai [Mon, 29 Jul 2019 08:24:33 +0000 (16:24 +0800)]
net: sched: Fix a possible null-pointer dereference in dequeue_func()

[ Upstream commit 051c7b39be4a91f6b7d8c4548444e4b850f1f56c ]

In dequeue_func(), there is an if statement on line 74 to check whether
skb is NULL:
    if (skb)

When skb is NULL, it is used on line 77:
    prefetch(&skb->end);

Thus, a possible null-pointer dereference may occur.

To fix this bug, skb->end is used when skb is not NULL.

This bug is found by a static analysis tool STCheck written by us.

Fixes: 76e3cc126bb2 ("codel: Controlled Delay AQM")
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet/mlx5: Use reversed order when unregister devices
Mark Zhang [Tue, 9 Jul 2019 02:37:12 +0000 (05:37 +0300)]
net/mlx5: Use reversed order when unregister devices

[ Upstream commit 08aa5e7da6bce1a1963f63cf32c2e7ad434ad578 ]

When lag is active, which is controlled by the bonded mlx5e netdev, mlx5
interface unregestering must happen in the reverse order where rdma is
unregistered (unloaded) first, to guarantee all references to the lag
context in hardware is removed, then remove mlx5e netdev interface which
will cleanup the lag context from hardware.

Without this fix during destroy of LAG interface, we observed following
errors:
 * mlx5_cmd_check:752:(pid 12556): DESTROY_LAG(0x843) op_mod(0x0) failed,
   status bad parameter(0x3), syndrome (0xe4ac33)
 * mlx5_cmd_check:752:(pid 12556): DESTROY_LAG(0x843) op_mod(0x0) failed,
   status bad parameter(0x3), syndrome (0xa5aee8).

Fixes: a31208b1e11d ("net/mlx5_core: New init and exit flow for mlx5_core")
Reviewed-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Mark Zhang <markz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: fix ifindex collision during namespace removal
Jiri Pirko [Sun, 28 Jul 2019 12:56:36 +0000 (14:56 +0200)]
net: fix ifindex collision during namespace removal

[ Upstream commit 55b40dbf0e76b4bfb9d8b3a16a0208640a9a45df ]

Commit aca51397d014 ("netns: Fix arbitrary net_device-s corruptions
on net_ns stop.") introduced a possibility to hit a BUG in case device
is returning back to init_net and two following conditions are met:
1) dev->ifindex value is used in a name of another "dev%d"
   device in init_net.
2) dev->name is used by another device in init_net.

Under real life circumstances this is hard to get. Therefore this has
been present happily for over 10 years. To reproduce:

$ ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 86:89:3f:86:61:29 brd ff:ff:ff:ff:ff:ff
3: enp0s2: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 52:54:00:12:34:56 brd ff:ff:ff:ff:ff:ff
$ ip netns add ns1
$ ip -n ns1 link add dummy1ns1 type dummy
$ ip -n ns1 link add dummy2ns1 type dummy
$ ip link set enp0s2 netns ns1
$ ip -n ns1 link set enp0s2 name dummy0
[  100.858894] virtio_net virtio0 dummy0: renamed from enp0s2
$ ip link add dev4 type dummy
$ ip -n ns1 a
1: lo: <LOOPBACK> mtu 65536 qdisc noop state DOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: dummy1ns1: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 16:63:4c:38:3e:ff brd ff:ff:ff:ff:ff:ff
3: dummy2ns1: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether aa:9e:86:dd:6b:5d brd ff:ff:ff:ff:ff:ff
4: dummy0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 52:54:00:12:34:56 brd ff:ff:ff:ff:ff:ff
$ ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 86:89:3f:86:61:29 brd ff:ff:ff:ff:ff:ff
4: dev4: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether 5a:e1:4a:b6:ec:f8 brd ff:ff:ff:ff:ff:ff
$ ip netns del ns1
[  158.717795] default_device_exit: failed to move dummy0 to init_net: -17
[  158.719316] ------------[ cut here ]------------
[  158.720591] kernel BUG at net/core/dev.c:9824!
[  158.722260] invalid opcode: 0000 [#1] SMP KASAN PTI
[  158.723728] CPU: 0 PID: 56 Comm: kworker/u2:1 Not tainted 5.3.0-rc1+ #18
[  158.725422] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-2.fc30 04/01/2014
[  158.727508] Workqueue: netns cleanup_net
[  158.728915] RIP: 0010:default_device_exit.cold+0x1d/0x1f
[  158.730683] Code: 84 e8 18 c9 3e fe 0f 0b e9 70 90 ff ff e8 36 e4 52 fe 89 d9 4c 89 e2 48 c7 c6 80 d6 25 84 48 c7 c7 20 c0 25 84 e8 f4 c8 3e
[  158.736854] RSP: 0018:ffff8880347e7b90 EFLAGS: 00010282
[  158.738752] RAX: 000000000000003b RBX: 00000000ffffffef RCX: 0000000000000000
[  158.741369] RDX: 0000000000000000 RSI: ffffffff8128013d RDI: ffffed10068fcf64
[  158.743418] RBP: ffff888033550170 R08: 000000000000003b R09: fffffbfff0b94b9c
[  158.745626] R10: fffffbfff0b94b9b R11: ffffffff85ca5cdf R12: ffff888032f28000
[  158.748405] R13: dffffc0000000000 R14: ffff8880335501b8 R15: 1ffff110068fcf72
[  158.750638] FS:  0000000000000000(0000) GS:ffff888036000000(0000) knlGS:0000000000000000
[  158.752944] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  158.755245] CR2: 00007fe8b45d21d0 CR3: 00000000340b4005 CR4: 0000000000360ef0
[  158.757654] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  158.760012] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  158.762758] Call Trace:
[  158.763882]  ? dev_change_net_namespace+0xbb0/0xbb0
[  158.766148]  ? devlink_nl_cmd_set_doit+0x520/0x520
[  158.768034]  ? dev_change_net_namespace+0xbb0/0xbb0
[  158.769870]  ops_exit_list.isra.0+0xa8/0x150
[  158.771544]  cleanup_net+0x446/0x8f0
[  158.772945]  ? unregister_pernet_operations+0x4a0/0x4a0
[  158.775294]  process_one_work+0xa1a/0x1740
[  158.776896]  ? pwq_dec_nr_in_flight+0x310/0x310
[  158.779143]  ? do_raw_spin_lock+0x11b/0x280
[  158.780848]  worker_thread+0x9e/0x1060
[  158.782500]  ? process_one_work+0x1740/0x1740
[  158.784454]  kthread+0x31b/0x420
[  158.786082]  ? __kthread_create_on_node+0x3f0/0x3f0
[  158.788286]  ret_from_fork+0x3a/0x50
[  158.789871] ---[ end trace defd6c657c71f936 ]---
[  158.792273] RIP: 0010:default_device_exit.cold+0x1d/0x1f
[  158.795478] Code: 84 e8 18 c9 3e fe 0f 0b e9 70 90 ff ff e8 36 e4 52 fe 89 d9 4c 89 e2 48 c7 c6 80 d6 25 84 48 c7 c7 20 c0 25 84 e8 f4 c8 3e
[  158.804854] RSP: 0018:ffff8880347e7b90 EFLAGS: 00010282
[  158.807865] RAX: 000000000000003b RBX: 00000000ffffffef RCX: 0000000000000000
[  158.811794] RDX: 0000000000000000 RSI: ffffffff8128013d RDI: ffffed10068fcf64
[  158.816652] RBP: ffff888033550170 R08: 000000000000003b R09: fffffbfff0b94b9c
[  158.820930] R10: fffffbfff0b94b9b R11: ffffffff85ca5cdf R12: ffff888032f28000
[  158.825113] R13: dffffc0000000000 R14: ffff8880335501b8 R15: 1ffff110068fcf72
[  158.829899] FS:  0000000000000000(0000) GS:ffff888036000000(0000) knlGS:0000000000000000
[  158.834923] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  158.838164] CR2: 00007fe8b45d21d0 CR3: 00000000340b4005 CR4: 0000000000360ef0
[  158.841917] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  158.845149] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400

Fix this by checking if a device with the same name exists in init_net
and fallback to original code - dev%d to allocate name - in case it does.

This was found using syzkaller.

Fixes: aca51397d014 ("netns: Fix arbitrary net_device-s corruptions on net_ns stop.")
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: bridge: mcast: don't delete permanent entries when fast leave is enabled
Nikolay Aleksandrov [Tue, 30 Jul 2019 11:21:00 +0000 (14:21 +0300)]
net: bridge: mcast: don't delete permanent entries when fast leave is enabled

[ Upstream commit 5c725b6b65067909548ac9ca9bc777098ec9883d ]

When permanent entries were introduced by the commit below, they were
exempt from timing out and thus igmp leave wouldn't affect them unless
fast leave was enabled on the port which was added before permanent
entries existed. It shouldn't matter if fast leave is enabled or not
if the user added a permanent entry it shouldn't be deleted on igmp
leave.

Before:
$ echo 1 > /sys/class/net/eth4/brport/multicast_fast_leave
$ bridge mdb add dev br0 port eth4 grp 229.1.1.1 permanent
$ bridge mdb show
dev br0 port eth4 grp 229.1.1.1 permanent

< join and leave 229.1.1.1 on eth4 >

$ bridge mdb show
$

After:
$ echo 1 > /sys/class/net/eth4/brport/multicast_fast_leave
$ bridge mdb add dev br0 port eth4 grp 229.1.1.1 permanent
$ bridge mdb show
dev br0 port eth4 grp 229.1.1.1 permanent

< join and leave 229.1.1.1 on eth4 >

$ bridge mdb show
dev br0 port eth4 grp 229.1.1.1 permanent

Fixes: ccb1c31a7a87 ("bridge: add flags to distinguish permanent mdb entires")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: bridge: delete local fdb on device init failure
Nikolay Aleksandrov [Mon, 29 Jul 2019 09:28:41 +0000 (12:28 +0300)]
net: bridge: delete local fdb on device init failure

[ Upstream commit d7bae09fa008c6c9a489580db0a5a12063b97f97 ]

On initialization failure we have to delete the local fdb which was
inserted due to the default pvid creation. This problem has been present
since the inception of default_pvid. Note that currently there are 2 cases:
1) in br_dev_init() when br_multicast_init() fails
2) if register_netdevice() fails after calling ndo_init()

This patch takes care of both since br_vlan_flush() is called on both
occasions. Also the new fdb delete would be a no-op on normal bridge
device destruction since the local fdb would've been already flushed by
br_dev_delete(). This is not an issue for ports since nbp_vlan_init() is
called last when adding a port thus nothing can fail after it.

Reported-by: syzbot+88533dc8b582309bf3ee@syzkaller.appspotmail.com
Fixes: 5be5a2df40f0 ("bridge: Add filtering support for default_pvid")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoatm: iphase: Fix Spectre v1 vulnerability
Gustavo A. R. Silva [Wed, 31 Jul 2019 03:21:41 +0000 (22:21 -0500)]
atm: iphase: Fix Spectre v1 vulnerability

[ Upstream commit ea443e5e98b5b74e317ef3d26bcaea54931ccdee ]

board is controlled by user-space, hence leading to a potential
exploitation of the Spectre variant 1 vulnerability.

This issue was detected with the help of Smatch:

drivers/atm/iphase.c:2765 ia_ioctl() warn: potential spectre issue 'ia_dev' [r] (local cap)
drivers/atm/iphase.c:2774 ia_ioctl() warn: possible spectre second half.  'iadev'
drivers/atm/iphase.c:2782 ia_ioctl() warn: possible spectre second half.  'iadev'
drivers/atm/iphase.c:2816 ia_ioctl() warn: possible spectre second half.  'iadev'
drivers/atm/iphase.c:2823 ia_ioctl() warn: possible spectre second half.  'iadev'
drivers/atm/iphase.c:2830 ia_ioctl() warn: potential spectre issue '_ia_dev' [r] (local cap)
drivers/atm/iphase.c:2845 ia_ioctl() warn: possible spectre second half.  'iadev'
drivers/atm/iphase.c:2856 ia_ioctl() warn: possible spectre second half.  'iadev'

Fix this by sanitizing board before using it to index ia_dev and _ia_dev

Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].

[1] https://lore.kernel.org/lkml/20180423164740.GY17484@dhcp22.suse.cz/

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agolibceph: use kbasename() and kill ceph_file_part()
Ilya Dryomov [Fri, 19 May 2017 09:33:16 +0000 (11:33 +0200)]
libceph: use kbasename() and kill ceph_file_part()

commit 6f4dbd149d2a151b89d1a5bbf7530ee5546c7908 upstream.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoobjtool: Add rewind_stack_do_exit() to the noreturn list
Josh Poimboeuf [Thu, 4 Apr 2019 17:17:35 +0000 (12:17 -0500)]
objtool: Add rewind_stack_do_exit() to the noreturn list

commit 4fa5ecda2bf96be7464eb406df8aba9d89260227 upstream.

This fixes the following warning seen on GCC 7.3:

  arch/x86/kernel/dumpstack.o: warning: objtool: oops_end() falls through to next function show_regs()

Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lkml.kernel.org/r/3418ebf5a5a9f6ed7e80954c741c0b904b67b5dc.1554398240.git.jpoimboe@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoobjtool: Add machine_real_restart() to the noreturn list
Josh Poimboeuf [Tue, 19 Jun 2018 15:47:50 +0000 (10:47 -0500)]
objtool: Add machine_real_restart() to the noreturn list

commit 684fb246578b9e81fc7b4ca5c71eae22edb650b2 upstream.

machine_real_restart() is annotated as '__noreturn", so add it to the
objtool noreturn list.  This fixes the following warning with clang and
CONFIG_CC_OPTIMIZE_FOR_SIZE=y:

  arch/x86/kernel/reboot.o: warning: objtool: native_machine_emergency_restart() falls through to next function machine_power_off()

Reported-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Matthias Kaehlcke <mka@chromium.org>
Link: https://lkml.kernel.org/r/791712792aa4431bdd55bf1beb33a169ddf3b4a2.1529423255.git.jpoimboe@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoIB: directly cast the sockaddr union to aockaddr
Greg Kroah-Hartman [Wed, 7 Aug 2019 16:44:12 +0000 (18:44 +0200)]
IB: directly cast the sockaddr union to aockaddr

Like commit 641114d2af31 ("RDMA: Directly cast the sockaddr union to
sockaddr") we need to quiet gcc 9 from warning about this crazy union.
That commit did not fix all of the warnings in 4.19 and older kernels
because the logic in roce_resolve_route_from_path() was rewritten
between 4.19 and 5.2 when that change happened.

Cc: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoRDMA: Directly cast the sockaddr union to sockaddr
Jason Gunthorpe [Mon, 13 May 2019 00:57:57 +0000 (21:57 -0300)]
RDMA: Directly cast the sockaddr union to sockaddr

commit 641114d2af312d39ca9bbc2369d18a5823da51c6 upstream.

gcc 9 now does allocation size tracking and thinks that passing the member
of a union and then accessing beyond that member's bounds is an overflow.

Instead of using the union member, use the entire union with a cast to
get to the sockaddr. gcc will now know that the memory extends the full
size of the union.

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoHID: Add quirk for HP X1200 PIXART OEM mouse
Sebastian Parschauer [Wed, 24 Jul 2019 18:40:03 +0000 (20:40 +0200)]
HID: Add quirk for HP X1200 PIXART OEM mouse

commit 49869d2ea9eecc105a10724c1abf035151a3c4e2 upstream.

The PixArt OEM mice are known for disconnecting every minute in
runlevel 1 or 3 if they are not always polled. So add quirk
ALWAYS_POLL for this one as well.

Jonathan Teh (@jonathan-teh) reported and tested the quirk.
Reference: https://github.com/sriemer/fix-linux-mouse/issues/15

Signed-off-by: Sebastian Parschauer <s.parschauer@gmx.de>
CC: stable@vger.kernel.org
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoHID: wacom: fix bit shift for Cintiq Companion 2
Aaron Armstrong Skomra [Tue, 23 Jul 2019 18:09:15 +0000 (11:09 -0700)]
HID: wacom: fix bit shift for Cintiq Companion 2

commit 693c3dab4e50403f91bca4b52fc6d8562a3180f6 upstream.

The bit indicating BTN_6 on this device is overshifted
by 2 bits, resulting in the incorrect button being
reported.

Also fix copy-paste mistake in comments.

Signed-off-by: Aaron Armstrong Skomra <aaron.skomra@wacom.com>
Reviewed-by: Ping Cheng <ping.cheng@wacom.com>
Link: https://github.com/linuxwacom/xf86-input-wacom/issues/71
Fixes: c7f0522a1ad1 ("HID: wacom: Slim down wacom_intuos_pad processing")
Cc: <stable@vger.kernel.org> # v4.5+
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agotcp: be more careful in tcp_fragment()
Eric Dumazet [Tue, 6 Aug 2019 15:09:14 +0000 (17:09 +0200)]
tcp: be more careful in tcp_fragment()

[ Upstream commit b617158dc096709d8600c53b6052144d12b89fab ]

Some applications set tiny SO_SNDBUF values and expect
TCP to just work. Recent patches to address CVE-2019-11478
broke them in case of losses, since retransmits might
be prevented.

We should allow these flows to make progress.

This patch allows the first and last skb in retransmit queue
to be split even if memory limits are hit.

It also adds the some room due to the fact that tcp_sendmsg()
and tcp_sendpage() might overshoot sk_wmem_queued by about one full
TSO skb (64KB size). Note this allowance was already present
in stable backports for kernels < 4.15

Note for < 4.15 backports :
 tcp_rtx_queue_tail() will probably look like :

static inline struct sk_buff *tcp_rtx_queue_tail(const struct sock *sk)
{
struct sk_buff *skb = tcp_send_head(sk);

return skb ? tcp_write_queue_prev(sk, skb) : tcp_write_queue_tail(sk);
}

Fixes: f070ef2ac667 ("tcp: tcp_fragment() should apply sane memory limits")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Andrew Prout <aprout@ll.mit.edu>
Tested-by: Andrew Prout <aprout@ll.mit.edu>
Tested-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Tested-by: Michal Kubecek <mkubecek@suse.cz>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Christoph Paasch <cpaasch@apple.com>
Cc: Jonathan Looney <jtl@netflix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoarm64: cpufeature: Fix feature comparison for CTR_EL0.{CWG,ERG}
Will Deacon [Mon, 5 Aug 2019 17:13:55 +0000 (18:13 +0100)]
arm64: cpufeature: Fix feature comparison for CTR_EL0.{CWG,ERG}

commit 147b9635e6347104b91f48ca9dca61eb0fbf2a54 upstream.

If CTR_EL0.{CWG,ERG} are 0b0000 then they must be interpreted to have
their architecturally maximum values, which defeats the use of
FTR_HIGHER_SAFE when sanitising CPU ID registers on heterogeneous
machines.

Introduce FTR_HIGHER_OR_ZERO_SAFE so that these fields effectively
saturate at zero.

Fixes: 3c739b571084 ("arm64: Keep track of CPU feature registers")
Cc: <stable@vger.kernel.org> # 4.9.y only
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoarm64: cpufeature: Fix CTR_EL0 field definitions
Will Deacon [Mon, 5 Aug 2019 17:13:54 +0000 (18:13 +0100)]
arm64: cpufeature: Fix CTR_EL0 field definitions

commit be68a8aaf925aaf35574260bf820bb09d2f9e07f upstream.

Our field definitions for CTR_EL0 suffer from a number of problems:

  - The IDC and DIC fields are missing, which causes us to enable CTR
    trapping on CPUs with either of these returning non-zero values.

  - The ERG is FTR_LOWER_SAFE, whereas it should be treated like CWG as
    FTR_HIGHER_SAFE so that applications can use it to avoid false sharing.

  - [nit] A RES1 field is described as "RAO"

This patch updates the CTR_EL0 field definitions to fix these issues.

Cc: <stable@vger.kernel.org> # 4.9.y only
Cc: Shanker Donthineni <shankerd@codeaurora.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoARM: dts: logicpd-som-lv: Fix Audio Mute
Adam Ford [Tue, 1 May 2018 13:58:53 +0000 (08:58 -0500)]
ARM: dts: logicpd-som-lv: Fix Audio Mute

[ Upstream commit 95e59fc3c3fa3187a07a75f40b21637deb4bd12d ]

The Audio has worked, but the mute pin has a weak pulldown which alows
some of the audio signal to pass very quietly.  This patch fixes
that so the mute pin is actively driven high for mute or low for normal
operation.

Fixes: ab8dd3aed011 ("ARM: DTS: Add minimal Support for Logic
PD DM3730 SOM-LV")

Signed-off-by: Adam Ford <aford173@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoARM: dts: Add pinmuxing for i2c2 and i2c3 for LogicPD torpedo
Adam Ford [Thu, 15 Feb 2018 14:25:55 +0000 (08:25 -0600)]
ARM: dts: Add pinmuxing for i2c2 and i2c3 for LogicPD torpedo

[ Upstream commit a135a392acbec7ecda782981788e8c03767a1571 ]

Since I2C1 and I2C4 have explicit pinmuxing set, let's be on the
safe side and set the pin muxing for I2C2 and I2C3.

Signed-off-by: Adam Ford <aford173@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoARM: dts: Add pinmuxing for i2c2 and i2c3 for LogicPD SOM-LV
Adam Ford [Thu, 15 Feb 2018 14:25:56 +0000 (08:25 -0600)]
ARM: dts: Add pinmuxing for i2c2 and i2c3 for LogicPD SOM-LV

[ Upstream commit 5fe3c0fa0d54877c65e7c9b4442aeeb25cdf469a ]

Since I2C1 and I2C4 have explicit pinmuxing set, let's be on the
safe side and set the pin muxing for I2C2 and I2C3.

Signed-off-by: Adam Ford <aford173@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: fcoe: Embed fc_rport_priv in fcoe_rport structure
Hannes Reinecke [Wed, 24 Jul 2019 09:00:55 +0000 (11:00 +0200)]
scsi: fcoe: Embed fc_rport_priv in fcoe_rport structure

commit 023358b136d490ca91735ac6490db3741af5a8bd upstream.

Gcc-9 complains for a memset across pointer boundaries, which happens as
the code tries to allocate a flexible array on the stack.  Turns out we
cannot do this without relying on gcc-isms, so with this patch we'll embed
the fc_rport_priv structure into fcoe_rport, can use the normal
'container_of' outcast, and will only have to do a memset over one
structure.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoLinux 4.9.188 v4.9.188
Greg Kroah-Hartman [Tue, 6 Aug 2019 16:29:42 +0000 (18:29 +0200)]
Linux 4.9.188

5 years agox86, mm, gup: prevent get_page() race with munmap in paravirt guest
Vlastimil Babka [Fri, 2 Aug 2019 16:06:14 +0000 (18:06 +0200)]
x86, mm, gup: prevent get_page() race with munmap in paravirt guest

The x86 version of get_user_pages_fast() relies on disabled interrupts to
synchronize gup_pte_range() between gup_get_pte(ptep); and get_page() against
a parallel munmap. The munmap side nulls the pte, then flushes TLBs, then
releases the page. As TLB flush is done synchronously via IPI disabling
interrupts blocks the page release, and get_page(), which assumes existing
reference on page, is thus safe.
However when TLB flush is done by a hypercall, e.g. in a Xen PV guest, there is
no blocking thanks to disabled interrupts, and get_page() can succeed on a page
that was already freed or even reused.

We have recently seen this happen with our 4.4 and 4.12 based kernels, with
userspace (java) that exits a thread, where mm_release() performs a futex_wake()
on tsk->clear_child_tid, and another thread in parallel unmaps the page where
tsk->clear_child_tid points to. The spurious get_page() succeeds, but futex code
immediately releases the page again, while it's already on a freelist. Symptoms
include a bad page state warning, general protection faults acessing a poisoned
list prev/next pointer in the freelist, or free page pcplists of two cpus joined
together in a single list. Oscar has also reproduced this scenario, with a
patch inserting delays before the get_page() to make the race window larger.

Fix this by removing the dependency on TLB flush interrupts the same way as the
generic get_user_pages_fast() code by using page_cache_add_speculative() and
revalidating the PTE contents after pinning the page. Mainline is safe since
4.13 where the x86 gup code was removed in favor of the common code. Accessing
the page table itself safely also relies on disabled interrupts and TLB flush
IPIs that don't happen with hypercalls, which was acknowledged in commit
9e52fc2b50de ("x86/mm: Enable RCU based page table freeing
(CONFIG_HAVE_RCU_TABLE_FREE=y)"). That commit with follups should also be
backported for full safety, although our reproducer didn't hit a problem
without that backport.

Reproduced-by: Oscar Salvador <osalvador@suse.de>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Juergen Gross <jgross@suse.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoobjtool: Support GCC 9 cold subfunction naming scheme
Josh Poimboeuf [Thu, 1 Nov 2018 02:57:30 +0000 (21:57 -0500)]
objtool: Support GCC 9 cold subfunction naming scheme

commit bcb6fb5da77c2a228adf07cc9cb1a0c2aa2001c6 upstream.

Starting with GCC 8, a lot of unlikely code was moved out of line to
"cold" subfunctions in .text.unlikely.

For example, the unlikely bits of:

  irq_do_set_affinity()

are moved out to the following subfunction:

  irq_do_set_affinity.cold.49()

Starting with GCC 9, the numbered suffix has been removed.  So in the
above example, the cold subfunction is instead:

  irq_do_set_affinity.cold()

Tweak the objtool subfunction detection logic so that it detects both
GCC 8 and GCC 9 naming schemes.

Reported-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/015e9544b1f188d36a7f02fa31e9e95629aa5f50.1541040800.git.jpoimboe@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoinclude/linux/module.h: copy __init/__exit attrs to init/cleanup_module
Miguel Ojeda [Fri, 2 Aug 2019 10:37:57 +0000 (12:37 +0200)]
include/linux/module.h: copy __init/__exit attrs to init/cleanup_module

commit a6e60d84989fa0e91db7f236eda40453b0e44afa upstream.

The upcoming GCC 9 release extends the -Wmissing-attributes warnings
(enabled by -Wall) to C and aliases: it warns when particular function
attributes are missing in the aliases but not in their target.

In particular, it triggers for all the init/cleanup_module
aliases in the kernel (defined by the module_init/exit macros),
ending up being very noisy.

These aliases point to the __init/__exit functions of a module,
which are defined as __cold (among other attributes). However,
the aliases themselves do not have the __cold attribute.

Since the compiler behaves differently when compiling a __cold
function as well as when compiling paths leading to calls
to __cold functions, the warning is trying to point out
the possibly-forgotten attribute in the alias.

In order to keep the warning enabled, we decided to silence
this case. Ideally, we would mark the aliases directly
as __init/__exit. However, there are currently around 132 modules
in the kernel which are missing __init/__exit in their init/cleanup
functions (either because they are missing, or for other reasons,
e.g. the functions being called from somewhere else); and
a section mismatch is a hard error.

A conservative alternative was to mark the aliases as __cold only.
However, since we would like to eventually enforce __init/__exit
to be always marked,  we chose to use the new __copy function
attribute (introduced by GCC 9 as well to deal with this).
With it, we copy the attributes used by the target functions
into the aliases. This way, functions that were not marked
as __init/__exit won't have their aliases marked either,
and therefore there won't be a section mismatch.

Note that the warning would go away marking either the extern
declaration, the definition, or both. However, we only mark
the definition of the alias, since we do not want callers
(which only see the declaration) to be compiled as if the function
was __cold (and therefore the paths leading to those calls
would be assumed to be unlikely).

Link: https://lore.kernel.org/lkml/259986242.BvXPX32bHu@devpool35/
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/lkml/20190123173707.GA16603@gmail.com/
Link: https://lore.kernel.org/lkml/20190206175627.GA20399@gmail.com/
Suggested-by: Martin Sebor <msebor@gcc.gnu.org>
Acked-by: Jessica Yu <jeyu@kernel.org>
Signed-off-by: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoBackport minimal compiler_attributes.h to support GCC 9
Miguel Ojeda [Fri, 2 Aug 2019 10:37:56 +0000 (12:37 +0200)]
Backport minimal compiler_attributes.h to support GCC 9

This adds support for __copy to v4.9.y so that we can use it in
init/exit_module to avoid -Werror=missing-attributes errors on GCC 9.

Link: https://lore.kernel.org/lkml/259986242.BvXPX32bHu@devpool35/
Cc: <stable@vger.kernel.org>
Suggested-by: Rolf Eike Beer <eb@emlix.com>
Signed-off-by: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoeeprom: at24: make spd world-readable again
Jean Delvare [Sun, 28 Jul 2019 16:41:38 +0000 (18:41 +0200)]
eeprom: at24: make spd world-readable again

commit 25e5ef302c24a6fead369c0cfe88c073d7b97ca8 upstream.

The integration of the at24 driver into the nvmem framework broke the
world-readability of spd EEPROMs. Fix it.

Signed-off-by: Jean Delvare <jdelvare@suse.de>
Cc: stable@vger.kernel.org
Fixes: 57d155506dd5 ("eeprom: at24: extend driver to plug into the NVMEM framework")
Cc: Andrew Lunn <andrew@lunn.ch>
Cc: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Bartosz Golaszewski <brgl@bgdev.pl>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
[Bartosz: backported the patch to older branches]
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agocoredump: fix race condition between collapse_huge_page() and core dumping
Andrea Arcangeli [Thu, 13 Jun 2019 22:56:11 +0000 (15:56 -0700)]
coredump: fix race condition between collapse_huge_page() and core dumping

commit 59ea6d06cfa9247b586a695c21f94afa7183af74 upstream.

When fixing the race conditions between the coredump and the mmap_sem
holders outside the context of the process, we focused on
mmget_not_zero()/get_task_mm() callers in 04f5866e41fb70 ("coredump: fix
race condition between mmget_not_zero()/get_task_mm() and core
dumping"), but those aren't the only cases where the mmap_sem can be
taken outside of the context of the process as Michal Hocko noticed
while backporting that commit to older -stable kernels.

If mmgrab() is called in the context of the process, but then the
mm_count reference is transferred outside the context of the process,
that can also be a problem if the mmap_sem has to be taken for writing
through that mm_count reference.

khugepaged registration calls mmgrab() in the context of the process,
but the mmap_sem for writing is taken later in the context of the
khugepaged kernel thread.

collapse_huge_page() after taking the mmap_sem for writing doesn't
modify any vma, so it's not obvious that it could cause a problem to the
coredump, but it happens to modify the pmd in a way that breaks an
invariant that pmd_trans_huge_lock() relies upon.  collapse_huge_page()
needs the mmap_sem for writing just to block concurrent page faults that
call pmd_trans_huge_lock().

Specifically the invariant that "!pmd_trans_huge()" cannot become a
"pmd_trans_huge()" doesn't hold while collapse_huge_page() runs.

The coredump will call __get_user_pages() without mmap_sem for reading,
which eventually can invoke a lockless page fault which will need a
functional pmd_trans_huge_lock().

So collapse_huge_page() needs to use mmget_still_valid() to check it's
not running concurrently with the coredump...  as long as the coredump
can invoke page faults without holding the mmap_sem for reading.

This has "Fixes: khugepaged" to facilitate backporting, but in my view
it's more a bug in the coredump code that will eventually have to be
rewritten to stop invoking page faults without the mmap_sem for reading.
So the long term plan is still to drop all mmget_still_valid().

Link: http://lkml.kernel.org/r/20190607161558.32104-1-aarcange@redhat.com
Fixes: ba76149f47d8 ("thp: khugepaged")
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Michal Hocko <mhocko@suse.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Jann Horn <jannh@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[Ajay: Just adjusted to apply on v4.9]
Signed-off-by: Ajay Kaher <akaher@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoinfiniband: fix race condition between infiniband mlx4, mlx5 driver and core dumping
Ajay Kaher [Sun, 4 Aug 2019 03:59:26 +0000 (09:29 +0530)]
infiniband: fix race condition between infiniband mlx4, mlx5 driver and core dumping

This patch is the extension of following upstream commit to fix
the race condition between get_task_mm() and core dumping
for IB->mlx4 and IB->mlx5 drivers:

commit 04f5866e41fb ("coredump: fix race condition between
mmget_not_zero()/get_task_mm() and core dumping")'

Thanks to Jason for pointing this.

Signed-off-by: Ajay Kaher <akaher@vmware.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agocoredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping
Andrea Arcangeli [Fri, 19 Apr 2019 00:50:52 +0000 (17:50 -0700)]
coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping

commit 04f5866e41fb70690e28397487d8bd8eea7d712a upstream.

The core dumping code has always run without holding the mmap_sem for
writing, despite that is the only way to ensure that the entire vma
layout will not change from under it.  Only using some signal
serialization on the processes belonging to the mm is not nearly enough.
This was pointed out earlier.  For example in Hugh's post from Jul 2017:

  https://lkml.kernel.org/r/alpine.LSU.2.11.1707191716030.2055@eggly.anvils

  "Not strictly relevant here, but a related note: I was very surprised
   to discover, only quite recently, how handle_mm_fault() may be called
   without down_read(mmap_sem) - when core dumping. That seems a
   misguided optimization to me, which would also be nice to correct"

In particular because the growsdown and growsup can move the
vm_start/vm_end the various loops the core dump does around the vma will
not be consistent if page faults can happen concurrently.

Pretty much all users calling mmget_not_zero()/get_task_mm() and then
taking the mmap_sem had the potential to introduce unexpected side
effects in the core dumping code.

Adding mmap_sem for writing around the ->core_dump invocation is a
viable long term fix, but it requires removing all copy user and page
faults and to replace them with get_dump_page() for all binary formats
which is not suitable as a short term fix.

For the time being this solution manually covers the places that can
confuse the core dump either by altering the vma layout or the vma flags
while it runs.  Once ->core_dump runs under mmap_sem for writing the
function mmget_still_valid() can be dropped.

Allowing mmap_sem protected sections to run in parallel with the
coredump provides some minor parallelism advantage to the swapoff code
(which seems to be safe enough by never mangling any vma field and can
keep doing swapins in parallel to the core dumping) and to some other
corner case.

In order to facilitate the backporting I added "Fixes: 86039bd3b4e6"
however the side effect of this same race condition in /proc/pid/mem
should be reproducible since before 2.6.12-rc2 so I couldn't add any
other "Fixes:" because there's no hash beyond the git genesis commit.

Because find_extend_vma() is the only location outside of the process
context that could modify the "mm" structures under mmap_sem for
reading, by adding the mmget_still_valid() check to it, all other cases
that take the mmap_sem for reading don't need the new check after
mmget_not_zero()/get_task_mm().  The expand_stack() in page fault
context also doesn't need the new check, because all tasks under core
dumping are frozen.

Link: http://lkml.kernel.org/r/20190325224949.11068-1-aarcange@redhat.com
Fixes: 86039bd3b4e6 ("userfaultfd: add new syscall to provide memory externalization")
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Jann Horn <jannh@google.com>
Suggested-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Jann Horn <jannh@google.com>
Acked-by: Jason Gunthorpe <jgg@mellanox.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[akaher@vmware.com: stable 4.9 backport
-  handle binder_update_page_range - mhocko@suse.com]
Signed-off-by: Ajay Kaher <akaher@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoIB/mlx5: Fix RSS Toeplitz setup to be aligned with the HW specification
Yishai Hadas [Tue, 23 Jul 2019 06:57:29 +0000 (09:57 +0300)]
IB/mlx5: Fix RSS Toeplitz setup to be aligned with the HW specification

commit b7165bd0d6cbb93732559be6ea8774653b204480 upstream.

The specification for the Toeplitz function doesn't require to set the key
explicitly to be symmetric. In case a symmetric functionality is required
a symmetric key can be simply used.

Wrongly forcing the algorithm to symmetric causes the wrong packet
distribution and a performance degradation.

Link: https://lore.kernel.org/r/20190723065733.4899-7-leon@kernel.org
Cc: <stable@vger.kernel.org> # 4.7
Fixes: 28d6137008b2 ("IB/mlx5: Add RSS QP support")
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Alex Vainman <alexv@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoxen/swiotlb: fix condition for calling xen_destroy_contiguous_region()
Juergen Gross [Fri, 14 Jun 2019 05:46:02 +0000 (07:46 +0200)]
xen/swiotlb: fix condition for calling xen_destroy_contiguous_region()

commit 50f6393f9654c561df4cdcf8e6cfba7260143601 upstream.

The condition in xen_swiotlb_free_coherent() for deciding whether to
call xen_destroy_contiguous_region() is wrong: in case the region to
be freed is not contiguous calling xen_destroy_contiguous_region() is
the wrong thing to do: it would result in inconsistent mappings of
multiple PFNs to the same MFN. This will lead to various strange
crashes or data corruption.

Instead of calling xen_destroy_contiguous_region() in that case a
warning should be issued as that situation should never occur.

Cc: stable@vger.kernel.org
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agodrivers/perf: arm_pmu: Fix failure path in PM notifier
Will Deacon [Mon, 29 Jul 2019 10:43:48 +0000 (11:43 +0100)]
drivers/perf: arm_pmu: Fix failure path in PM notifier

commit 0d7fd70f26039bd4b33444ca47f0e69ce3ae0354 upstream.

Handling of the CPU_PM_ENTER_FAILED transition in the Arm PMU PM
notifier code incorrectly skips restoration of the counters. Fix the
logic so that CPU_PM_ENTER_FAILED follows the same path as CPU_PM_EXIT.

Cc: <stable@vger.kernel.org>
Fixes: da4e4f18afe0f372 ("drivers/perf: arm_pmu: implement CPU_PM notifier")
Reported-by: Anders Roxell <anders.roxell@linaro.org>
Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agos390/dasd: fix endless loop after read unit address configuration
Stefan Haberland [Thu, 1 Aug 2019 11:06:30 +0000 (13:06 +0200)]
s390/dasd: fix endless loop after read unit address configuration

commit 41995342b40c418a47603e1321256d2c4a2ed0fb upstream.

After getting a storage server event that causes the DASD device driver
to update its unit address configuration during a device shutdown there is
the possibility of an endless loop in the device driver.

In the system log there will be ongoing DASD error messages with RC: -19.

The reason is that the loop starting the ruac request only terminates when
the retry counter is decreased to 0. But in the sleep_on function there are
early exit paths that do not decrease the retry counter.

Prevent an endless loop by handling those cases separately.

Remove the unnecessary do..while loop since the sleep_on function takes
care of retries by itself.

Fixes: 8e09f21574ea ("[S390] dasd: add hyper PAV support to DASD device driver, part 1")
Cc: stable@vger.kernel.org # 2.6.25+
Signed-off-by: Stefan Haberland <sth@linux.ibm.com>
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>