openwrt/staging/blogic.git
5 years agor8152: set RTL8152_UNPLUG only for real disconnection
Hayes Wang [Thu, 4 Jul 2019 09:36:32 +0000 (17:36 +0800)]
r8152: set RTL8152_UNPLUG only for real disconnection

Set the flag of RTL8152_UNPLUG if and only if the device is unplugged.
Some error codes sometimes don't mean the real disconnection of usb device.
For those situations, set the flag of RTL8152_UNPLUG causes the driver skips
some flows of disabling the device, and it let the device stay at incorrect
state.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'hsr-bug-fixes'
David S. Miller [Fri, 5 Jul 2019 22:22:28 +0000 (15:22 -0700)]
Merge branch 'hsr-bug-fixes'

Cong Wang says:

====================
hsr: a few bug fixes

This patchset contains 3 bug fixes for hsr triggered by a syzbot
reproducer, please check each patch for details.
====================

Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
5 years agohsr: fix a NULL pointer deref in hsr_dev_xmit()
Cong Wang [Thu, 4 Jul 2019 00:21:14 +0000 (17:21 -0700)]
hsr: fix a NULL pointer deref in hsr_dev_xmit()

hsr_port_get_hsr() could return NULL and kernel
could crash:

 BUG: kernel NULL pointer dereference, address: 0000000000000010
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 8000000074b84067 P4D 8000000074b84067 PUD 7057d067 PMD 0
 Oops: 0000 [#1] SMP PTI
 CPU: 0 PID: 754 Comm: a.out Not tainted 5.2.0-rc6+ #718
 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-2.fc30 04/01/2014
 RIP: 0010:hsr_dev_xmit+0x20/0x31
 Code: 48 8b 1b eb e0 5b 5d 41 5c c3 66 66 66 66 90 55 48 89 fd 48 8d be 40 0b 00 00 be 04 00 00 00 e8 ee f2 ff ff 48 89 ef 48 89 c6 <48> 8b 40 10 48 89 45 10 e8 6c 1b 00 00 31 c0 5d c3 66 66 66 66 90
 RSP: 0018:ffffb5b400003c48 EFLAGS: 00010246
 RAX: 0000000000000000 RBX: ffff9821b4509a88 RCX: 0000000000000000
 RDX: ffff9821b4509a88 RSI: 0000000000000000 RDI: ffff9821bc3fc7c0
 RBP: ffff9821bc3fc7c0 R08: 0000000000000000 R09: 00000000000c2019
 R10: 0000000000000000 R11: 0000000000000002 R12: ffff9821bc3fc7c0
 R13: ffff9821b4509a88 R14: 0000000000000000 R15: 000000000000006e
 FS:  00007fee112a1800(0000) GS:ffff9821bd800000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000000000000010 CR3: 000000006e9ce000 CR4: 00000000000406f0
 Call Trace:
  <IRQ>
  netdev_start_xmit+0x1b/0x38
  dev_hard_start_xmit+0x121/0x21e
  ? validate_xmit_skb.isra.0+0x19/0x1e3
  __dev_queue_xmit+0x74c/0x823
  ? lockdep_hardirqs_on+0x12b/0x17d
  ip6_finish_output2+0x3d3/0x42c
  ? ip6_mtu+0x55/0x5c
  ? mld_sendpack+0x191/0x229
  mld_sendpack+0x191/0x229
  mld_ifc_timer_expire+0x1f7/0x230
  ? mld_dad_timer_expire+0x58/0x58
  call_timer_fn+0x12e/0x273
  __run_timers.part.0+0x174/0x1b5
  ? mld_dad_timer_expire+0x58/0x58
  ? sched_clock_cpu+0x10/0xad
  ? mark_lock+0x26/0x1f2
  ? __lock_is_held+0x40/0x71
  run_timer_softirq+0x26/0x48
  __do_softirq+0x1af/0x392
  irq_exit+0x53/0xa2
  smp_apic_timer_interrupt+0x1c4/0x1d9
  apic_timer_interrupt+0xf/0x20
  </IRQ>

Cc: Arvid Brodin <arvid.brodin@alten.se>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agohsr: implement dellink to clean up resources
Cong Wang [Thu, 4 Jul 2019 00:21:13 +0000 (17:21 -0700)]
hsr: implement dellink to clean up resources

hsr_link_ops implements ->newlink() but not ->dellink(),
which leads that resources not released after removing the device,
particularly the entries in self_node_db and node_db.

So add ->dellink() implementation to replace the priv_destructor.
This also makes the code slightly easier to understand.

Reported-by: syzbot+c6167ec3de7def23d1e8@syzkaller.appspotmail.com
Cc: Arvid Brodin <arvid.brodin@alten.se>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agohsr: fix a memory leak in hsr_del_port()
Cong Wang [Thu, 4 Jul 2019 00:21:12 +0000 (17:21 -0700)]
hsr: fix a memory leak in hsr_del_port()

hsr_del_port() should release all the resources allocated
in hsr_add_port().

As a consequence of this change, hsr_for_each_port() is no
longer safe to work with hsr_del_port(), switch to
list_for_each_entry_safe() as we always hold RTNL lock.

Cc: Arvid Brodin <arvid.brodin@alten.se>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec
David S. Miller [Fri, 5 Jul 2019 21:58:22 +0000 (14:58 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/klassert/ipsec

Steffen Klassert says:

====================
pull request (net): ipsec 2019-07-05

1)  Fix xfrm selector prefix length validation for
    inter address family tunneling.
    From Anirudh Gupta.

2) Fix a memleak in pfkey.
   From Jeremy Sowden.

3) Fix SA selector validation to allow empty selectors again.
   From Nicolas Dichtel.

4) Select crypto ciphers for xfrm_algo, this fixes some
   randconfig builds. From Arnd Bergmann.

5) Remove a duplicated assignment in xfrm_bydst_resize.
   From Cong Wang.

6) Fix a hlist corruption on hash rebuild.
   From Florian Westphal.

7) Fix a memory leak when creating xfrm interfaces.
   From Nicolas Dichtel.

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

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobonding: validate ip header before check IPPROTO_IGMP
Cong Wang [Tue, 2 Jul 2019 03:40:24 +0000 (20:40 -0700)]
bonding: validate ip header before check IPPROTO_IGMP

bond_xmit_roundrobin() checks for IGMP packets but it parses
the IP header even before checking skb->protocol.

We should validate the IP header with pskb_may_pull() before
using iph->protocol.

Reported-and-tested-by: syzbot+e5be16aa39ad6e755391@syzkaller.appspotmail.com
Fixes: a2fd940f4cff ("bonding: fix broken multicast with round-robin mode")
Cc: Jay Vosburgh <j.vosburgh@gmail.com>
Cc: Veaceslav Falico <vfalico@gmail.com>
Cc: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf
David S. Miller [Wed, 3 Jul 2019 19:09:00 +0000 (12:09 -0700)]
Merge git://git./pub/scm/linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
pull-request: bpf 2019-07-03

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

The main changes are:

1) Fix the interpreter to properly handle BPF_ALU32 | BPF_ARSH
   on BE architectures, from Jiong.

2) Fix several bugs in the x32 BPF JIT for handling shifts by 0,
   from Luke and Xi.

3) Fix NULL pointer deref in btf_type_is_resolve_source_only(),
   from Stanislav.

4) Properly handle the check that forwarding is enabled on the device
   in bpf_ipv6_fib_lookup() helper code, from Anton.

5) Fix UAPI bpf_prog_info fields alignment for archs that have 16 bit
   alignment such as m68k, from Baruch.

6) Fix kernel hanging in unregister_netdevice loop while unregistering
   device bound to XDP socket, from Ilya.

7) Properly terminate tail update in xskq_produce_flush_desc(), from Nathan.

8) Fix broken always_inline handling in test_lwt_seg6local, from Jiri.

9) Fix bpftool to use correct argument in cgroup errors, from Jakub.

10) Fix detaching dummy prog in XDP redirect sample code, from Prashant.

11) Add Jonathan to AF_XDP reviewers, from Björn.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns: add support for vlan TSO
Yonglong Liu [Wed, 3 Jul 2019 11:12:30 +0000 (19:12 +0800)]
net: hns: add support for vlan TSO

The hip07 chip support vlan TSO, this patch adds NETIF_F_TSO
and NETIF_F_TSO6 flags to vlan_features to improve the
performance after adding vlan to the net ports.

Signed-off-by: Yonglong Liu <liuyonglong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agosctp: count data bundling sack chunk for outctrlchunks
Xin Long [Wed, 3 Jul 2019 10:20:20 +0000 (18:20 +0800)]
sctp: count data bundling sack chunk for outctrlchunks

Now all ctrl chunks are counted for asoc stats.octrlchunks and net
SCTP_MIB_OUTCTRLCHUNKS either after queuing up or bundling, other
than the chunk maked and bundled in sctp_packet_bundle_sack, which
caused 'outctrlchunks' not consistent with 'inctrlchunks' in peer.

This issue exists since very beginning, here to fix it by increasing
both net SCTP_MIB_OUTCTRLCHUNKS and asoc stats.octrlchunks when sack
chunk is maked and bundled in sctp_packet_bundle_sack.

Reported-by: Ja Ram Jeon <jajeon@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agor8152: move calling r8153b_rx_agg_chg_indicate()
Hayes Wang [Wed, 3 Jul 2019 07:11:56 +0000 (15:11 +0800)]
r8152: move calling r8153b_rx_agg_chg_indicate()

r8153b_rx_agg_chg_indicate() needs to be called after enabling TX/RX and
before calling rxdy_gated_en(tp, false). Otherwise, the change of the
settings of RX aggregation wouldn't work.

Besides, adjust rtl8152_set_coalesce() for the same reason. If
rx_coalesce_usecs is changed, restart TX/RX to let the setting work.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: don't warn in inet diag when IPV6 is disabled
Stephen Hemminger [Tue, 2 Jul 2019 22:20:21 +0000 (15:20 -0700)]
net: don't warn in inet diag when IPV6 is disabled

If IPV6 was disabled, then ss command would cause a kernel warning
because the command was attempting to dump IPV6 socket information.
The fix is to just remove the warning.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=202249
Fixes: 432490f9d455 ("net: ip, diag -- Add diag interface for raw sockets")
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoxdp: fix hang while unregistering device bound to xdp socket
Ilya Maximets [Fri, 28 Jun 2019 08:04:07 +0000 (11:04 +0300)]
xdp: fix hang while unregistering device bound to xdp socket

Device that bound to XDP socket will not have zero refcount until the
userspace application will not close it. This leads to hang inside
'netdev_wait_allrefs()' if device unregistering requested:

  # ip link del p1
  < hang on recvmsg on netlink socket >

  # ps -x | grep ip
  5126  pts/0    D+   0:00 ip link del p1

  # journalctl -b

  Jun 05 07:19:16 kernel:
  unregister_netdevice: waiting for p1 to become free. Usage count = 1

  Jun 05 07:19:27 kernel:
  unregister_netdevice: waiting for p1 to become free. Usage count = 1
  ...

Fix that by implementing NETDEV_UNREGISTER event notification handler
to properly clean up all the resources and unref device.

This should also allow socket killing via ss(8) utility.

Fixes: 965a99098443 ("xsk: add support for bind for Rx")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
5 years agoxdp: hold device for umem regardless of zero-copy mode
Ilya Maximets [Fri, 28 Jun 2019 08:04:06 +0000 (11:04 +0300)]
xdp: hold device for umem regardless of zero-copy mode

Device pointer stored in umem regardless of zero-copy mode,
so we heed to hold the device in all cases.

Fixes: c9b47cc1fabc ("xsk: fix bug when trying to use both copy and zero-copy on one queue id")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
5 years agoselftests: bpf: fix inlines in test_lwt_seg6local
Jiri Benc [Tue, 2 Jul 2019 17:40:31 +0000 (19:40 +0200)]
selftests: bpf: fix inlines in test_lwt_seg6local

Selftests are reporting this failure in test_lwt_seg6local.sh:

+ ip netns exec ns2 ip -6 route add fb00::6 encap bpf in obj test_lwt_seg6local.o sec encap_srh dev veth2
Error fetching program/map!
Failed to parse eBPF program: Operation not permitted

The problem is __attribute__((always_inline)) alone is not enough to prevent
clang from inserting those functions in .text. In that case, .text is not
marked as relocateable.

See the output of objdump -h test_lwt_seg6local.o:

Idx Name          Size      VMA               LMA               File off  Algn
  0 .text         00003530  0000000000000000  0000000000000000  00000040  2**3
                  CONTENTS, ALLOC, LOAD, READONLY, CODE

This causes the iproute bpf loader to fail in bpf_fetch_prog_sec:
bpf_has_call_data returns true but bpf_fetch_prog_relo fails as there's no
relocateable .text section in the file.

To fix this, convert to 'static __always_inline'.

v2: Use 'static __always_inline' instead of 'static inline
    __attribute__((always_inline))'

Fixes: c99a84eac026 ("selftests/bpf: test for seg6local End.BPF action")
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
5 years agoselftests: bpf: add tests for shifts by zero
Luke Nelson [Sat, 29 Jun 2019 05:57:51 +0000 (22:57 -0700)]
selftests: bpf: add tests for shifts by zero

There are currently no tests for ALU64 shift operations when the shift
amount is 0. This adds 6 new tests to make sure they are equivalent
to a no-op. The x32 JIT had such bugs that could have been caught by
these tests.

Cc: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Luke Nelson <luke.r.nels@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
5 years agobpf, x32: Fix bug with ALU64 {LSH, RSH, ARSH} BPF_K shift by 0
Luke Nelson [Sat, 29 Jun 2019 05:57:50 +0000 (22:57 -0700)]
bpf, x32: Fix bug with ALU64 {LSH, RSH, ARSH} BPF_K shift by 0

The current x32 BPF JIT does not correctly compile shift operations when
the immediate shift amount is 0. The expected behavior is for this to
be a no-op.

The following program demonstrates the bug. The expexceted result is 1,
but the current JITed code returns 2.

  r0 = 1
  r1 = 1
  r1 <<= 0
  if r1 == 1 goto end
  r0 = 2
end:
  exit

This patch simplifies the code and fixes the bug.

Fixes: 03f5781be2c7 ("bpf, x86_32: add eBPF JIT compiler for ia32")
Co-developed-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Luke Nelson <luke.r.nels@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
5 years agobpf, x32: Fix bug with ALU64 {LSH, RSH, ARSH} BPF_X shift by 0
Luke Nelson [Sat, 29 Jun 2019 05:57:49 +0000 (22:57 -0700)]
bpf, x32: Fix bug with ALU64 {LSH, RSH, ARSH} BPF_X shift by 0

The current x32 BPF JIT for shift operations is not correct when the
shift amount in a register is 0. The expected behavior is a no-op, whereas
the current implementation changes bits in the destination register.

The following example demonstrates the bug. The expected result of this
program is 1, but the current JITed code returns 2.

  r0 = 1
  r1 = 1
  r2 = 0
  r1 <<= r2
  if r1 == 1 goto end
  r0 = 2
end:
  exit

The bug is caused by an incorrect assumption by the JIT that a shift by
32 clear the register. On x32 however, shifts use the lower 5 bits of
the source, making a shift by 32 equivalent to a shift by 0.

This patch fixes the bug using double-precision shifts, which also
simplifies the code.

Fixes: 03f5781be2c7 ("bpf, x86_32: add eBPF JIT compiler for ia32")
Co-developed-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Luke Nelson <luke.r.nels@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
5 years agoxfrm interface: fix memory leak on creation
Nicolas Dichtel [Tue, 2 Jul 2019 15:51:39 +0000 (17:51 +0200)]
xfrm interface: fix memory leak on creation

The following commands produce a backtrace and return an error but the xfrm
interface is created (in the wrong netns):
$ ip netns add foo
$ ip netns add bar
$ ip -n foo netns set bar 0
$ ip -n foo link add xfrmi0 link-netnsid 0 type xfrm dev lo if_id 23
RTNETLINK answers: Invalid argument
$ ip -n bar link ls xfrmi0
2: xfrmi0@lo: <NOARP,M-DOWN> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
    link/none 00:00:00:00:00:00 brd 00:00:00:00:00:00

Here is the backtrace:
[   79.879174] WARNING: CPU: 0 PID: 1178 at net/core/dev.c:8172 rollback_registered_many+0x86/0x3c1
[   79.880260] Modules linked in: xfrm_interface nfsv3 nfs_acl auth_rpcgss nfsv4 nfs lockd grace sunrpc fscache button parport_pc parport serio_raw evdev pcspkr loop ext4 crc16 mbcache jbd2 crc32c_generic ide_cd_mod ide_gd_mod cdrom ata_$
eneric ata_piix libata scsi_mod 8139too piix psmouse i2c_piix4 ide_core 8139cp mii i2c_core floppy
[   79.883698] CPU: 0 PID: 1178 Comm: ip Not tainted 5.2.0-rc6+ #106
[   79.884462] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
[   79.885447] RIP: 0010:rollback_registered_many+0x86/0x3c1
[   79.886120] Code: 01 e8 d7 7d c6 ff 0f 0b 48 8b 45 00 4c 8b 20 48 8d 58 90 49 83 ec 70 48 8d 7b 70 48 39 ef 74 44 8a 83 d0 04 00 00 84 c0 75 1f <0f> 0b e8 61 cd ff ff 48 b8 00 01 00 00 00 00 ad de 48 89 43 70 66
[   79.888667] RSP: 0018:ffffc900015ab740 EFLAGS: 00010246
[   79.889339] RAX: ffff8882353e5700 RBX: ffff8882353e56a0 RCX: ffff8882353e5710
[   79.890174] RDX: ffffc900015ab7e0 RSI: ffffc900015ab7e0 RDI: ffff8882353e5710
[   79.891029] RBP: ffffc900015ab7e0 R08: ffffc900015ab7e0 R09: ffffc900015ab7e0
[   79.891866] R10: ffffc900015ab7a0 R11: ffffffff82233fec R12: ffffc900015ab770
[   79.892728] R13: ffffffff81eb7ec0 R14: ffff88822ed6cf00 R15: 00000000ffffffea
[   79.893557] FS:  00007ff350f31740(0000) GS:ffff888237a00000(0000) knlGS:0000000000000000
[   79.894581] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   79.895317] CR2: 00000000006c8580 CR3: 000000022c272000 CR4: 00000000000006f0
[   79.896137] Call Trace:
[   79.896464]  unregister_netdevice_many+0x12/0x6c
[   79.896998]  __rtnl_newlink+0x6e2/0x73b
[   79.897446]  ? __kmalloc_node_track_caller+0x15e/0x185
[   79.898039]  ? pskb_expand_head+0x5f/0x1fe
[   79.898556]  ? stack_access_ok+0xd/0x2c
[   79.899009]  ? deref_stack_reg+0x12/0x20
[   79.899462]  ? stack_access_ok+0xd/0x2c
[   79.899927]  ? stack_access_ok+0xd/0x2c
[   79.900404]  ? __module_text_address+0x9/0x4f
[   79.900910]  ? is_bpf_text_address+0x5/0xc
[   79.901390]  ? kernel_text_address+0x67/0x7b
[   79.901884]  ? __kernel_text_address+0x1a/0x25
[   79.902397]  ? unwind_get_return_address+0x12/0x23
[   79.903122]  ? __cmpxchg_double_slab.isra.37+0x46/0x77
[   79.903772]  rtnl_newlink+0x43/0x56
[   79.904217]  rtnetlink_rcv_msg+0x200/0x24c

In fact, each time a xfrm interface was created, a netdev was allocated
by __rtnl_newlink()/rtnl_create_link() and then another one by
xfrmi_newlink()/xfrmi_create(). Only the second one was registered, it's
why the previous commands produce a backtrace: dev_change_net_namespace()
was called on a netdev with reg_state set to NETREG_UNINITIALIZED (the
first one).

CC: Lorenzo Colitti <lorenzo@google.com>
CC: Benedict Wong <benedictwong@google.com>
CC: Steffen Klassert <steffen.klassert@secunet.com>
CC: Shannon Nelson <shannon.nelson@oracle.com>
CC: Antony Antony <antony@phenome.org>
CC: Eyal Birger <eyal.birger@gmail.com>
Fixes: f203b76d7809 ("xfrm: Add virtual xfrm interfaces")
Reported-by: Julien Floret <julien.floret@6wind.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
5 years agoxfrm: policy: fix bydst hlist corruption on hash rebuild
Florian Westphal [Tue, 2 Jul 2019 10:46:00 +0000 (12:46 +0200)]
xfrm: policy: fix bydst hlist corruption on hash rebuild

syzbot reported following spat:

BUG: KASAN: use-after-free in __write_once_size include/linux/compiler.h:221
BUG: KASAN: use-after-free in hlist_del_rcu include/linux/rculist.h:455
BUG: KASAN: use-after-free in xfrm_hash_rebuild+0xa0d/0x1000 net/xfrm/xfrm_policy.c:1318
Write of size 8 at addr ffff888095e79c00 by task kworker/1:3/8066
Workqueue: events xfrm_hash_rebuild
Call Trace:
 __write_once_size include/linux/compiler.h:221 [inline]
 hlist_del_rcu include/linux/rculist.h:455 [inline]
 xfrm_hash_rebuild+0xa0d/0x1000 net/xfrm/xfrm_policy.c:1318
 process_one_work+0x814/0x1130 kernel/workqueue.c:2269
Allocated by task 8064:
 __kmalloc+0x23c/0x310 mm/slab.c:3669
 kzalloc include/linux/slab.h:742 [inline]
 xfrm_hash_alloc+0x38/0xe0 net/xfrm/xfrm_hash.c:21
 xfrm_policy_init net/xfrm/xfrm_policy.c:4036 [inline]
 xfrm_net_init+0x269/0xd60 net/xfrm/xfrm_policy.c:4120
 ops_init+0x336/0x420 net/core/net_namespace.c:130
 setup_net+0x212/0x690 net/core/net_namespace.c:316

The faulting address is the address of the old chain head,
free'd by xfrm_hash_resize().

In xfrm_hash_rehash(), chain heads get re-initialized without
any hlist_del_rcu:

 for (i = hmask; i >= 0; i--)
    INIT_HLIST_HEAD(odst + i);

Then, hlist_del_rcu() gets called on the about to-be-reinserted policy
when iterating the per-net list of policies.

hlist_del_rcu() will then make chain->first be nonzero again:

static inline void __hlist_del(struct hlist_node *n)
{
   struct hlist_node *next = n->next;   // address of next element in list
   struct hlist_node **pprev = n->pprev;// location of previous elem, this
                                        // can point at chain->first
        WRITE_ONCE(*pprev, next);       // chain->first points to next elem
        if (next)
                next->pprev = pprev;

Then, when we walk chainlist to find insertion point, we may find a
non-empty list even though we're supposedly reinserting the first
policy to an empty chain.

To fix this first unlink all exact and inexact policies instead of
zeroing the list heads.

Add the commands equivalent to the syzbot reproducer to xfrm_policy.sh,
without fix KASAN catches the corruption as it happens, SLUB poisoning
detects it a bit later.

Reported-by: syzbot+0165480d4ef07360eeda@syzkaller.appspotmail.com
Fixes: 1548bc4e0512 ("xfrm: policy: delete inexact policies from inexact list on hash rebuild")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
5 years agoselftests/net: skip psock_tpacket test if KALLSYMS was not enabled
Po-Hsu Lin [Mon, 1 Jul 2019 04:40:31 +0000 (12:40 +0800)]
selftests/net: skip psock_tpacket test if KALLSYMS was not enabled

The psock_tpacket test will need to access /proc/kallsyms, this would
require the kernel config CONFIG_KALLSYMS to be enabled first.

Apart from adding CONFIG_KALLSYMS to the net/config file here, check the
file existence to determine if we can run this test will be helpful to
avoid a false-positive test result when testing it directly with the
following commad against a kernel that have CONFIG_KALLSYMS disabled:
    make -C tools/testing/selftests TARGETS=net run_tests

Signed-off-by: Po-Hsu Lin <po-hsu.lin@canonical.com>
Acked-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agorxrpc: Fix oops in tracepoint
David Howells [Tue, 2 Jul 2019 15:04:19 +0000 (16:04 +0100)]
rxrpc: Fix oops in tracepoint

If the rxrpc_eproto tracepoint is enabled, an oops will be cause by the
trace line that rxrpc_extract_header() tries to emit when a protocol error
occurs (typically because the packet is short) because the call argument is
NULL.

Fix this by using ?: to assume 0 as the debug_id if call is NULL.

This can then be induced by:

echo -e '\0\0\0\0\0\0\0\0' | ncat -4u --send-only <addr> 20001

where addr has the following program running on it:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/socket.h>
#include <arpa/inet.h>
#include <linux/rxrpc.h>
int main(void)
{
struct sockaddr_rxrpc srx;
int fd;
memset(&srx, 0, sizeof(srx));
srx.srx_family = AF_RXRPC;
srx.srx_service = 0;
srx.transport_type = AF_INET;
srx.transport_len = sizeof(srx.transport.sin);
srx.transport.sin.sin_family = AF_INET;
srx.transport.sin.sin_port = htons(0x4e21);
fd = socket(AF_RXRPC, SOCK_DGRAM, AF_INET6);
bind(fd, (struct sockaddr *)&srx, sizeof(srx));
sleep(20);
return 0;
}

It results in the following oops.

BUG: kernel NULL pointer dereference, address: 0000000000000340
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
...
RIP: 0010:trace_event_raw_event_rxrpc_rx_eproto+0x47/0xac
...
Call Trace:
 <IRQ>
 rxrpc_extract_header+0x86/0x171
 ? rcu_read_lock_sched_held+0x5d/0x63
 ? rxrpc_new_skb+0xd4/0x109
 rxrpc_input_packet+0xef/0x14fc
 ? rxrpc_input_data+0x986/0x986
 udp_queue_rcv_one_skb+0xbf/0x3d0
 udp_unicast_rcv_skb.isra.8+0x64/0x71
 ip_protocol_deliver_rcu+0xe4/0x1b4
 ip_local_deliver+0xf0/0x154
 __netif_receive_skb_one_core+0x50/0x6c
 netif_receive_skb_internal+0x26b/0x2e9
 napi_gro_receive+0xf8/0x1da
 rtl8169_poll+0x303/0x4c4
 net_rx_action+0x10e/0x333
 __do_softirq+0x1a5/0x38f
 irq_exit+0x54/0xc4
 do_IRQ+0xda/0xf8
 common_interrupt+0xf/0xf
 </IRQ>
 ...
 ? cpuidle_enter_state+0x23c/0x34d
 cpuidle_enter+0x2a/0x36
 do_idle+0x163/0x1ea
 cpu_startup_entry+0x1d/0x1f
 start_secondary+0x157/0x172
 secondary_startup_64+0xa4/0xb0

Fixes: a25e21f0bcd2 ("rxrpc, afs: Use debug_ids rather than pointers in traces")
Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: usb: asix: init MAC address buffers
Phong Tran [Tue, 2 Jul 2019 00:10:08 +0000 (07:10 +0700)]
net: usb: asix: init MAC address buffers

This is for fixing bug KMSAN: uninit-value in ax88772_bind

Tested by
https://groups.google.com/d/msg/syzkaller-bugs/aFQurGotng4/eB_HlNhhCwAJ

Reported-by: syzbot+8a3fc6674bbc3978ed4e@syzkaller.appspotmail.com
syzbot found the following crash on:

HEAD commit:    f75e4cfe kmsan: use kmsan_handle_urb() in urb.c
git tree:       kmsan
console output: https://syzkaller.appspot.com/x/log.txt?x=136d720ea00000
kernel config:
https://syzkaller.appspot.com/x/.config?x=602468164ccdc30a
dashboard link:
https://syzkaller.appspot.com/bug?extid=8a3fc6674bbc3978ed4e
compiler:       clang version 9.0.0 (/home/glider/llvm/clang
06d00afa61eef8f7f501ebdb4e8612ea43ec2d78)
syz repro:
https://syzkaller.appspot.com/x/repro.syz?x=12788316a00000
C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=120359aaa00000

==================================================================
BUG: KMSAN: uninit-value in is_valid_ether_addr
include/linux/etherdevice.h:200 [inline]
BUG: KMSAN: uninit-value in asix_set_netdev_dev_addr
drivers/net/usb/asix_devices.c:73 [inline]
BUG: KMSAN: uninit-value in ax88772_bind+0x93d/0x11e0
drivers/net/usb/asix_devices.c:724
CPU: 0 PID: 3348 Comm: kworker/0:2 Not tainted 5.1.0+ #1
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Workqueue: usb_hub_wq hub_event
Call Trace:
  __dump_stack lib/dump_stack.c:77 [inline]
  dump_stack+0x191/0x1f0 lib/dump_stack.c:113
  kmsan_report+0x130/0x2a0 mm/kmsan/kmsan.c:622
  __msan_warning+0x75/0xe0 mm/kmsan/kmsan_instr.c:310
  is_valid_ether_addr include/linux/etherdevice.h:200 [inline]
  asix_set_netdev_dev_addr drivers/net/usb/asix_devices.c:73 [inline]
  ax88772_bind+0x93d/0x11e0 drivers/net/usb/asix_devices.c:724
  usbnet_probe+0x10f5/0x3940 drivers/net/usb/usbnet.c:1728
  usb_probe_interface+0xd66/0x1320 drivers/usb/core/driver.c:361
  really_probe+0xdae/0x1d80 drivers/base/dd.c:513
  driver_probe_device+0x1b3/0x4f0 drivers/base/dd.c:671
  __device_attach_driver+0x5b8/0x790 drivers/base/dd.c:778
  bus_for_each_drv+0x28e/0x3b0 drivers/base/bus.c:454
  __device_attach+0x454/0x730 drivers/base/dd.c:844
  device_initial_probe+0x4a/0x60 drivers/base/dd.c:891
  bus_probe_device+0x137/0x390 drivers/base/bus.c:514
  device_add+0x288d/0x30e0 drivers/base/core.c:2106
  usb_set_configuration+0x30dc/0x3750 drivers/usb/core/message.c:2027
  generic_probe+0xe7/0x280 drivers/usb/core/generic.c:210
  usb_probe_device+0x14c/0x200 drivers/usb/core/driver.c:266
  really_probe+0xdae/0x1d80 drivers/base/dd.c:513
  driver_probe_device+0x1b3/0x4f0 drivers/base/dd.c:671
  __device_attach_driver+0x5b8/0x790 drivers/base/dd.c:778
  bus_for_each_drv+0x28e/0x3b0 drivers/base/bus.c:454
  __device_attach+0x454/0x730 drivers/base/dd.c:844
  device_initial_probe+0x4a/0x60 drivers/base/dd.c:891
  bus_probe_device+0x137/0x390 drivers/base/bus.c:514
  device_add+0x288d/0x30e0 drivers/base/core.c:2106
  usb_new_device+0x23e5/0x2ff0 drivers/usb/core/hub.c:2534
  hub_port_connect drivers/usb/core/hub.c:5089 [inline]
  hub_port_connect_change drivers/usb/core/hub.c:5204 [inline]
  port_event drivers/usb/core/hub.c:5350 [inline]
  hub_event+0x48d1/0x7290 drivers/usb/core/hub.c:5432
  process_one_work+0x1572/0x1f00 kernel/workqueue.c:2269
  process_scheduled_works kernel/workqueue.c:2331 [inline]
  worker_thread+0x189c/0x2460 kernel/workqueue.c:2417
  kthread+0x4b5/0x4f0 kernel/kthread.c:254
  ret_from_fork+0x35/0x40 arch/x86/entry/entry_64.S:355

Signed-off-by: Phong Tran <tranmanphong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'macsec-fix-some-bugs-in-the-receive-path'
David S. Miller [Tue, 2 Jul 2019 21:12:30 +0000 (14:12 -0700)]
Merge branch 'macsec-fix-some-bugs-in-the-receive-path'

Andreas Steinmetz says:

====================
macsec: fix some bugs in the receive path

This series fixes some bugs in the receive path of macsec. The first
is a use after free when processing macsec frames with a SecTAG that
has the TCI E bit set but the C bit clear. In the 2nd bug, the driver
leaves an invalid checksumming state after decrypting the packet.

This is a combined effort of Sabrina Dubroca <sd@queasysnail.net> and me.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomacsec: fix checksumming after decryption
Andreas Steinmetz [Sun, 30 Jun 2019 20:46:45 +0000 (22:46 +0200)]
macsec: fix checksumming after decryption

Fix checksumming after decryption.

Signed-off-by: Andreas Steinmetz <ast@domdv.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomacsec: fix use-after-free of skb during RX
Andreas Steinmetz [Sun, 30 Jun 2019 20:46:42 +0000 (22:46 +0200)]
macsec: fix use-after-free of skb during RX

Fix use-after-free of skb when rx_handler returns RX_HANDLER_PASS.

Signed-off-by: Andreas Steinmetz <ast@domdv.de>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agorxrpc: Fix send on a connected, but unbound socket
David Howells [Tue, 2 Jul 2019 14:59:12 +0000 (15:59 +0100)]
rxrpc: Fix send on a connected, but unbound socket

If sendmsg() or sendmmsg() is called on a connected socket that hasn't had
bind() called on it, then an oops will occur when the kernel tries to
connect the call because no local endpoint has been allocated.

Fix this by implicitly binding the socket if it is in the
RXRPC_CLIENT_UNBOUND state, just like it does for the RXRPC_UNBOUND state.

Further, the state should be transitioned to RXRPC_CLIENT_BOUND after this
to prevent further attempts to bind it.

This can be tested with:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/socket.h>
#include <arpa/inet.h>
#include <linux/rxrpc.h>
static const unsigned char inet6_addr[16] = {
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -1, -1, 0xac, 0x14, 0x14, 0xaa
};
int main(void)
{
struct sockaddr_rxrpc srx;
struct cmsghdr *cm;
struct msghdr msg;
unsigned char control[16];
int fd;
memset(&srx, 0, sizeof(srx));
srx.srx_family = 0x21;
srx.srx_service = 0;
srx.transport_type = AF_INET;
srx.transport_len = 0x1c;
srx.transport.sin6.sin6_family = AF_INET6;
srx.transport.sin6.sin6_port = htons(0x4e22);
srx.transport.sin6.sin6_flowinfo = htons(0x4e22);
srx.transport.sin6.sin6_scope_id = htons(0xaa3b);
memcpy(&srx.transport.sin6.sin6_addr, inet6_addr, 16);
cm = (struct cmsghdr *)control;
cm->cmsg_len = CMSG_LEN(sizeof(unsigned long));
cm->cmsg_level = SOL_RXRPC;
cm->cmsg_type = RXRPC_USER_CALL_ID;
*(unsigned long *)CMSG_DATA(cm) = 0;
msg.msg_name = NULL;
msg.msg_namelen = 0;
msg.msg_iov = NULL;
msg.msg_iovlen = 0;
msg.msg_control = control;
msg.msg_controllen = cm->cmsg_len;
msg.msg_flags = 0;
fd = socket(AF_RXRPC, SOCK_DGRAM, AF_INET);
connect(fd, (struct sockaddr *)&srx, sizeof(srx));
sendmsg(fd, &msg, 0);
return 0;
}

Leading to the following oops:

BUG: kernel NULL pointer dereference, address: 0000000000000018
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
...
RIP: 0010:rxrpc_connect_call+0x42/0xa01
...
Call Trace:
 ? mark_held_locks+0x47/0x59
 ? __local_bh_enable_ip+0xb6/0xba
 rxrpc_new_client_call+0x3b1/0x762
 ? rxrpc_do_sendmsg+0x3c0/0x92e
 rxrpc_do_sendmsg+0x3c0/0x92e
 rxrpc_sendmsg+0x16b/0x1b5
 sock_sendmsg+0x2d/0x39
 ___sys_sendmsg+0x1a4/0x22a
 ? release_sock+0x19/0x9e
 ? reacquire_held_locks+0x136/0x160
 ? release_sock+0x19/0x9e
 ? find_held_lock+0x2b/0x6e
 ? __lock_acquire+0x268/0xf73
 ? rxrpc_connect+0xdd/0xe4
 ? __local_bh_enable_ip+0xb6/0xba
 __sys_sendmsg+0x5e/0x94
 do_syscall_64+0x7d/0x1bf
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

Fixes: 2341e0775747 ("rxrpc: Simplify connect() implementation and simplify sendmsg() op")
Reported-by: syzbot+7966f2a0b2c7da8939b4@syzkaller.appspotmail.com
Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'bridge-stale-ptrs'
David S. Miller [Tue, 2 Jul 2019 18:53:03 +0000 (11:53 -0700)]
Merge branch 'bridge-stale-ptrs'

Nikolay Aleksandrov says:

====================
net: bridge: fix possible stale skb pointers

In the bridge driver we have a couple of places which call pskb_may_pull
but we've cached skb pointers before that and use them after which can
lead to out-of-bounds/stale pointer use. I've had these in my "to fix"
list for some time and now we got a report (patch 01) so here they are.
Patches 02-04 are fixes based on code inspection. Also patch 01 was
tested by Martin Weinelt, Martin if you don't mind please add your
tested-by tag to it by replying with Tested-by: name <email>.
I've also briefly tested the set by trying to exercise those code paths.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: bridge: stp: don't cache eth dest pointer before skb pull
Nikolay Aleksandrov [Tue, 2 Jul 2019 12:00:21 +0000 (15:00 +0300)]
net: bridge: stp: don't cache eth dest pointer before skb pull

Don't cache eth dest pointer before calling pskb_may_pull.

Fixes: cf0f02d04a83 ("[BRIDGE]: use llc for receiving STP packets")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: bridge: don't cache ether dest pointer on input
Nikolay Aleksandrov [Tue, 2 Jul 2019 12:00:20 +0000 (15:00 +0300)]
net: bridge: don't cache ether dest pointer on input

We would cache ether dst pointer on input in br_handle_frame_finish but
after the neigh suppress code that could lead to a stale pointer since
both ipv4 and ipv6 suppress code do pskb_may_pull. This means we have to
always reload it after the suppress code so there's no point in having
it cached just retrieve it directly.

Fixes: 057658cb33fbf ("bridge: suppress arp pkts on BR_NEIGH_SUPPRESS ports")
Fixes: ed842faeb2bd ("bridge: suppress nd pkts on BR_NEIGH_SUPPRESS ports")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: bridge: mcast: fix stale ipv6 hdr pointer when handling v6 query
Nikolay Aleksandrov [Tue, 2 Jul 2019 12:00:19 +0000 (15:00 +0300)]
net: bridge: mcast: fix stale ipv6 hdr pointer when handling v6 query

We get a pointer to the ipv6 hdr in br_ip6_multicast_query but we may
call pskb_may_pull afterwards and end up using a stale pointer.
So use the header directly, it's just 1 place where it's needed.

Fixes: 08b202b67264 ("bridge br_multicast: IPv6 MLD support.")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Tested-by: Martin Weinelt <martin@linuxlounge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: bridge: mcast: fix stale nsrcs pointer in igmp3/mld2 report handling
Nikolay Aleksandrov [Tue, 2 Jul 2019 12:00:18 +0000 (15:00 +0300)]
net: bridge: mcast: fix stale nsrcs pointer in igmp3/mld2 report handling

We take a pointer to grec prior to calling pskb_may_pull and use it
afterwards to get nsrcs so record nsrcs before the pull when handling
igmp3 and we get a pointer to nsrcs and call pskb_may_pull when handling
mld2 which again could lead to reading 2 bytes out-of-bounds.

 ==================================================================
 BUG: KASAN: use-after-free in br_multicast_rcv+0x480c/0x4ad0 [bridge]
 Read of size 2 at addr ffff8880421302b4 by task ksoftirqd/1/16

 CPU: 1 PID: 16 Comm: ksoftirqd/1 Tainted: G           OE     5.2.0-rc6+ #1
 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
 Call Trace:
  dump_stack+0x71/0xab
  print_address_description+0x6a/0x280
  ? br_multicast_rcv+0x480c/0x4ad0 [bridge]
  __kasan_report+0x152/0x1aa
  ? br_multicast_rcv+0x480c/0x4ad0 [bridge]
  ? br_multicast_rcv+0x480c/0x4ad0 [bridge]
  kasan_report+0xe/0x20
  br_multicast_rcv+0x480c/0x4ad0 [bridge]
  ? br_multicast_disable_port+0x150/0x150 [bridge]
  ? ktime_get_with_offset+0xb4/0x150
  ? __kasan_kmalloc.constprop.6+0xa6/0xf0
  ? __netif_receive_skb+0x1b0/0x1b0
  ? br_fdb_update+0x10e/0x6e0 [bridge]
  ? br_handle_frame_finish+0x3c6/0x11d0 [bridge]
  br_handle_frame_finish+0x3c6/0x11d0 [bridge]
  ? br_pass_frame_up+0x3a0/0x3a0 [bridge]
  ? virtnet_probe+0x1c80/0x1c80 [virtio_net]
  br_handle_frame+0x731/0xd90 [bridge]
  ? select_idle_sibling+0x25/0x7d0
  ? br_handle_frame_finish+0x11d0/0x11d0 [bridge]
  __netif_receive_skb_core+0xced/0x2d70
  ? virtqueue_get_buf_ctx+0x230/0x1130 [virtio_ring]
  ? do_xdp_generic+0x20/0x20
  ? virtqueue_napi_complete+0x39/0x70 [virtio_net]
  ? virtnet_poll+0x94d/0xc78 [virtio_net]
  ? receive_buf+0x5120/0x5120 [virtio_net]
  ? __netif_receive_skb_one_core+0x97/0x1d0
  __netif_receive_skb_one_core+0x97/0x1d0
  ? __netif_receive_skb_core+0x2d70/0x2d70
  ? _raw_write_trylock+0x100/0x100
  ? __queue_work+0x41e/0xbe0
  process_backlog+0x19c/0x650
  ? _raw_read_lock_irq+0x40/0x40
  net_rx_action+0x71e/0xbc0
  ? __switch_to_asm+0x40/0x70
  ? napi_complete_done+0x360/0x360
  ? __switch_to_asm+0x34/0x70
  ? __switch_to_asm+0x40/0x70
  ? __schedule+0x85e/0x14d0
  __do_softirq+0x1db/0x5f9
  ? takeover_tasklets+0x5f0/0x5f0
  run_ksoftirqd+0x26/0x40
  smpboot_thread_fn+0x443/0x680
  ? sort_range+0x20/0x20
  ? schedule+0x94/0x210
  ? __kthread_parkme+0x78/0xf0
  ? sort_range+0x20/0x20
  kthread+0x2ae/0x3a0
  ? kthread_create_worker_on_cpu+0xc0/0xc0
  ret_from_fork+0x35/0x40

 The buggy address belongs to the page:
 page:ffffea0001084c00 refcount:0 mapcount:-128 mapping:0000000000000000 index:0x0
 flags: 0xffffc000000000()
 raw: 00ffffc000000000 ffffea0000cfca08 ffffea0001098608 0000000000000000
 raw: 0000000000000000 0000000000000003 00000000ffffff7f 0000000000000000
 page dumped because: kasan: bad access detected

 Memory state around the buggy address:
 ffff888042130180: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
 ffff888042130200: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
 > ffff888042130280: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
                                     ^
 ffff888042130300: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
 ffff888042130380: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
 ==================================================================
 Disabling lock debugging due to kernel taint

Fixes: bc8c20acaea1 ("bridge: multicast: treat igmpv3 report with INCLUDE and no sources as a leave")
Reported-by: Martin Weinelt <martin@linuxlounge.net>
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Tested-by: Martin Weinelt <martin@linuxlounge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoxfrm: remove a duplicated assignment
Cong Wang [Sat, 29 Jun 2019 19:17:14 +0000 (12:17 -0700)]
xfrm: remove a duplicated assignment

Fixes: 30846090a746 ("xfrm: policy: add sequence count to sync with hash resize")
Cc: Florian Westphal <fw@strlen.de>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
5 years agor8152: fix the setting of detecting the linking change for runtime suspend
Hayes Wang [Mon, 1 Jul 2019 07:53:19 +0000 (15:53 +0800)]
r8152: fix the setting of detecting the linking change for runtime suspend

1. Rename r8153b_queue_wake() to r8153_queue_wake().

2. Correct the setting. The enable bit should be 0xd38c bit 0. Besides,
   the 0xd38a bit 0 and 0xd398 bit 8 have to be cleared for both enabled
   and disabled.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet/tls: make sure offload also gets the keys wiped
Jakub Kicinski [Fri, 28 Jun 2019 23:11:39 +0000 (16:11 -0700)]
net/tls: make sure offload also gets the keys wiped

Commit 86029d10af18 ("tls: zero the crypto information from tls_context
before freeing") added memzero_explicit() calls to clear the key material
before freeing struct tls_context, but it missed tls_device.c has its
own way of freeing this structure. Replace the missing free.

Fixes: 86029d10af18 ("tls: zero the crypto information from tls_context before freeing")
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet/tls: reject offload of TLS 1.3
Jakub Kicinski [Fri, 28 Jun 2019 23:07:59 +0000 (16:07 -0700)]
net/tls: reject offload of TLS 1.3

Neither drivers nor the tls offload code currently supports TLS
version 1.3. Check the TLS version when installing connection
state. TLS 1.3 will just fallback to the kernel crypto for now.

Fixes: 130b392c6cd6 ("net: tls: Add tls 1.3 support")
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'idr-fix-overflow-cases-on-32-bit-CPU'
David S. Miller [Tue, 2 Jul 2019 02:15:46 +0000 (19:15 -0700)]
Merge branch 'idr-fix-overflow-cases-on-32-bit-CPU'

Cong Wang says:

====================
idr: fix overflow cases on 32-bit CPU

idr_get_next_ul() is problematic by design, it can't handle
the following overflow case well on 32-bit CPU:

u32 id = UINT_MAX;
idr_alloc_u32(&id);
while (idr_get_next_ul(&id) != NULL)
 id++;

when 'id' overflows and becomes 0 after UINT_MAX, the loop
goes infinite.

Fix this by eliminating external users of idr_get_next_ul()
and migrating them to idr_for_each_entry_continue_ul(). And
add an additional parameter for these iteration macros to detect
overflow properly.

Please merge this through networking tree, as all the users
are in networking subsystem.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoselftests: add a test case for cls_lower handle overflow
Davide Caratti [Fri, 28 Jun 2019 18:03:43 +0000 (11:03 -0700)]
selftests: add a test case for cls_lower handle overflow

Reported-by: Li Shuang <shuali@redhat.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoidr: introduce idr_for_each_entry_continue_ul()
Cong Wang [Fri, 28 Jun 2019 18:03:42 +0000 (11:03 -0700)]
idr: introduce idr_for_each_entry_continue_ul()

Similarly, other callers of idr_get_next_ul() suffer the same
overflow bug as they don't handle it properly either.

Introduce idr_for_each_entry_continue_ul() to help these callers
iterate from a given ID.

cls_flower needs more care here because it still has overflow when
does arg->cookie++, we have to fold its nested loops into one
and remove the arg->cookie++.

Fixes: 01683a146999 ("net: sched: refactor flower walk to iterate over idr")
Fixes: 12d6066c3b29 ("net/mlx5: Add flow counters idr")
Reported-by: Li Shuang <shuali@redhat.com>
Cc: Davide Caratti <dcaratti@redhat.com>
Cc: Vlad Buslov <vladbu@mellanox.com>
Cc: Chris Mi <chrism@mellanox.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Tested-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoidr: fix overflow case for idr_for_each_entry_ul()
Cong Wang [Fri, 28 Jun 2019 18:03:41 +0000 (11:03 -0700)]
idr: fix overflow case for idr_for_each_entry_ul()

idr_for_each_entry_ul() is buggy as it can't handle overflow
case correctly. When we have an ID == UINT_MAX, it becomes an
infinite loop. This happens when running on 32-bit CPU where
unsigned long has the same size with unsigned int.

There is no better way to fix this than casting it to a larger
integer, but we can't just 64 bit integer on 32 bit CPU. Instead
we could just use an additional integer to help us to detect this
overflow case, that is, adding a new parameter to this macro.
Fortunately tc action is its only user right now.

Fixes: 65a206c01e8e ("net/sched: Change act_api and act_xxx modules to use IDR")
Reported-by: Li Shuang <shuali@redhat.com>
Tested-by: Davide Caratti <dcaratti@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mi <chrism@mellanox.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'vsock-virtio-fixes'
David S. Miller [Tue, 2 Jul 2019 02:09:07 +0000 (19:09 -0700)]
Merge branch 'vsock-virtio-fixes'

Stefano Garzarella says:

====================
vsock/virtio: several fixes in the .probe() and .remove()

During the review of "[PATCH] vsock/virtio: Initialize core virtio vsock
before registering the driver", Stefan pointed out some possible issues
in the .probe() and .remove() callbacks of the virtio-vsock driver.

This series tries to solve these issues:
- Patch 1 adds RCU critical sections to avoid use-after-free of
  'the_virtio_vsock' pointer.
- Patch 2 stops workers before to call vdev->config->reset(vdev) to
  be sure that no one is accessing the device.
- Patch 3 moves the works flush at the end of the .remove() to avoid
  use-after-free of 'vsock' object.

v2:
- Patch 1: use RCU to protect 'the_virtio_vsock' pointer
- Patch 2: no changes
- Patch 3: flush works only at the end of .remove()
- Removed patch 4 because virtqueue_detach_unused_buf() returns all the buffers
  allocated.

v1: https://patchwork.kernel.org/cover/10964733/
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agovsock/virtio: fix flush of works during the .remove()
Stefano Garzarella [Fri, 28 Jun 2019 12:36:59 +0000 (14:36 +0200)]
vsock/virtio: fix flush of works during the .remove()

This patch moves the flush of works after vdev->config->del_vqs(vdev),
because we need to be sure that no workers run before to free the
'vsock' object.

Since we stopped the workers using the [tx|rx|event]_run flags,
we are sure no one is accessing the device while we are calling
vdev->config->reset(vdev), so we can safely move the workers' flush.

Before the vdev->config->del_vqs(vdev), workers can be scheduled
by VQ callbacks, so we must flush them after del_vqs(), to avoid
use-after-free of 'vsock' object.

Suggested-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agovsock/virtio: stop workers during the .remove()
Stefano Garzarella [Fri, 28 Jun 2019 12:36:58 +0000 (14:36 +0200)]
vsock/virtio: stop workers during the .remove()

Before to call vdev->config->reset(vdev) we need to be sure that
no one is accessing the device, for this reason, we add new variables
in the struct virtio_vsock to stop the workers during the .remove().

This patch also add few comments before vdev->config->reset(vdev)
and vdev->config->del_vqs(vdev).

Suggested-by: Stefan Hajnoczi <stefanha@redhat.com>
Suggested-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agovsock/virtio: use RCU to avoid use-after-free on the_virtio_vsock
Stefano Garzarella [Fri, 28 Jun 2019 12:36:57 +0000 (14:36 +0200)]
vsock/virtio: use RCU to avoid use-after-free on the_virtio_vsock

Some callbacks used by the upper layers can run while we are in the
.remove(). A potential use-after-free can happen, because we free
the_virtio_vsock without knowing if the callbacks are over or not.

To solve this issue we move the assignment of the_virtio_vsock at the
end of .probe(), when we finished all the initialization, and at the
beginning of .remove(), before to release resources.
For the same reason, we do the same also for the vdev->priv.

We use RCU to be sure that all callbacks that use the_virtio_vsock
ended before freeing it. This is not required for callbacks that
use vdev->priv, because after the vdev->config->del_vqs() we are sure
that they are ended and will no longer be invoked.

We also take the mutex during the .remove() to avoid that .probe() can
run while we are resetting the device.

Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agovxlan: do not destroy fdb if register_netdevice() is failed
Taehee Yoo [Fri, 28 Jun 2019 05:07:25 +0000 (14:07 +0900)]
vxlan: do not destroy fdb if register_netdevice() is failed

__vxlan_dev_create() destroys FDB using specific pointer which indicates
a fdb when error occurs.
But that pointer should not be used when register_netdevice() fails because
register_netdevice() internally destroys fdb when error occurs.

This patch makes vxlan_fdb_create() to do not link fdb entry to vxlan dev
internally.
Instead, a new function vxlan_fdb_insert() is added to link fdb to vxlan
dev.

vxlan_fdb_insert() is called after calling register_netdevice().
This routine can avoid situation that ->ndo_uninit() destroys fdb entry
in error path of register_netdevice().
Hence, error path of __vxlan_dev_create() routine can have an opportunity
to destroy default fdb entry by hand.

Test command
    ip link add bonding_masters type vxlan id 0 group 239.1.1.1 \
    dev enp0s9 dstport 4789

Splat looks like:
[  213.392816] kasan: GPF could be caused by NULL-ptr deref or user memory access
[  213.401257] general protection fault: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI
[  213.402178] CPU: 0 PID: 1414 Comm: ip Not tainted 5.2.0-rc5+ #256
[  213.402178] RIP: 0010:vxlan_fdb_destroy+0x120/0x220 [vxlan]
[  213.402178] Code: df 48 8b 2b 48 89 fa 48 c1 ea 03 80 3c 02 00 0f 85 06 01 00 00 4c 8b 63 08 48 b8 00 00 00 00 00 fc d
[  213.402178] RSP: 0018:ffff88810cb9f0a0 EFLAGS: 00010202
[  213.402178] RAX: dffffc0000000000 RBX: ffff888101d4a8c8 RCX: 0000000000000000
[  213.402178] RDX: 1bd5a00000000040 RSI: ffff888101d4a8c8 RDI: ffff888101d4a8d0
[  213.402178] RBP: 0000000000000000 R08: fffffbfff22b72d9 R09: 0000000000000000
[  213.402178] R10: 00000000ffffffef R11: 0000000000000000 R12: dead000000000200
[  213.402178] R13: ffff88810cb9f1f8 R14: ffff88810efccda0 R15: ffff88810efccda0
[  213.402178] FS:  00007f7f6621a0c0(0000) GS:ffff88811b000000(0000) knlGS:0000000000000000
[  213.402178] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  213.402178] CR2: 000055746f0807d0 CR3: 00000001123e0000 CR4: 00000000001006f0
[  213.402178] Call Trace:
[  213.402178]  __vxlan_dev_create+0x3a9/0x7d0 [vxlan]
[  213.402178]  ? vxlan_changelink+0x740/0x740 [vxlan]
[  213.402178]  ? rcu_read_unlock+0x60/0x60 [vxlan]
[  213.402178]  ? __kasan_kmalloc.constprop.3+0xa0/0xd0
[  213.402178]  vxlan_newlink+0x8d/0xc0 [vxlan]
[  213.402178]  ? __vxlan_dev_create+0x7d0/0x7d0 [vxlan]
[  213.554119]  ? __netlink_ns_capable+0xc3/0xf0
[  213.554119]  __rtnl_newlink+0xb75/0x1180
[  213.554119]  ? rtnl_link_unregister+0x230/0x230
[ ... ]

Fixes: 0241b836732f ("vxlan: fix default fdb entry netlink notify ordering during netdev create")
Suggested-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Acked-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agosctp: fix error handling on stream scheduler initialization
Marcelo Ricardo Leitner [Thu, 27 Jun 2019 22:48:10 +0000 (19:48 -0300)]
sctp: fix error handling on stream scheduler initialization

It allocates the extended area for outbound streams only on sendmsg
calls, if they are not yet allocated.  When using the priority
stream scheduler, this initialization may imply into a subsequent
allocation, which may fail.  In this case, it was aborting the stream
scheduler initialization but leaving the ->ext pointer (allocated) in
there, thus in a partially initialized state.  On a subsequent call to
sendmsg, it would notice the ->ext pointer in there, and trip on
uninitialized stuff when trying to schedule the data chunk.

The fix is undo the ->ext initialization if the stream scheduler
initialization fails and avoid the partially initialized state.

Although syzkaller bisected this to commit 4ff40b86262b ("sctp: set
chunk transport correctly when it's a new asoc"), this bug was actually
introduced on the commit I marked below.

Reported-by: syzbot+c1a380d42b190ad1e559@syzkaller.appspotmail.com
Fixes: 5bbbbe32a431 ("sctp: introduce stream scheduler foundations")
Tested-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: Neil Horman <nhorman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonetrom: fix a memory leak in nr_rx_frame()
Cong Wang [Thu, 27 Jun 2019 21:30:58 +0000 (14:30 -0700)]
netrom: fix a memory leak in nr_rx_frame()

When the skb is associated with a new sock, just assigning
it to skb->sk is not sufficient, we have to set its destructor
to free the sock properly too.

Reported-by: syzbot+d6636a36d3c34bd88938@syzkaller.appspotmail.com
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: don't set IPv6 only flags to IPv4 addresses
Matteo Croce [Mon, 1 Jul 2019 17:01:55 +0000 (19:01 +0200)]
ipv4: don't set IPv6 only flags to IPv4 addresses

Avoid the situation where an IPV6 only flag is applied to an IPv4 address:

    # ip addr add 192.0.2.1/24 dev dummy0 nodad home mngtmpaddr noprefixroute
    # ip -4 addr show dev dummy0
    2: dummy0: <BROADCAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default qlen 1000
        inet 192.0.2.1/24 scope global noprefixroute dummy0
           valid_lft forever preferred_lft forever

Or worse, by sending a malicious netlink command:

    # ip -4 addr show dev dummy0
    2: dummy0: <BROADCAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default qlen 1000
        inet 192.0.2.1/24 scope global nodad optimistic dadfailed home tentative mngtmpaddr noprefixroute stable-privacy dummy0
           valid_lft forever preferred_lft forever

Signed-off-by: Matteo Croce <mcroce@redhat.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet:gue.h:Fix shifting signed 32-bit value by 31 bits problem
Vandana BN [Mon, 1 Jul 2019 14:16:10 +0000 (19:46 +0530)]
net:gue.h:Fix shifting signed 32-bit value by 31 bits problem

Fix GUE_PFLAG_REMCSUM to use "U" cast to avoid shifting signed
32-bit value by 31 bits problem.

Signed-off-by: Vandana BN <bnvandana@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dst.h: Fix shifting signed 32-bit value by 31 bits problem
Vandana BN [Mon, 1 Jul 2019 11:55:39 +0000 (17:25 +0530)]
net: dst.h: Fix shifting signed 32-bit value by 31 bits problem

Fix DST_FEATURE_ECN_CA to use "U" cast to avoid shifting signed
32-bit value by 31 bits problem.

Signed-off-by: Vandana BN <bnvandana@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoDocumentation/networking: fix default_ttl typo in mpls-sysctl
Hangbin Liu [Mon, 1 Jul 2019 08:45:28 +0000 (16:45 +0800)]
Documentation/networking: fix default_ttl typo in mpls-sysctl

default_ttl should be integer instead of bool

Reported-by: Ying Xu <yinxu@redhat.com>
Fixes: a59166e47086 ("mpls: allow TTL propagation from IP packets to be configured")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: openvswitch: fix csum updates for MPLS actions
John Hurley [Thu, 27 Jun 2019 13:37:30 +0000 (14:37 +0100)]
net: openvswitch: fix csum updates for MPLS actions

Skbs may have their checksum value populated by HW. If this is a checksum
calculated over the entire packet then the CHECKSUM_COMPLETE field is
marked. Changes to the data pointer on the skb throughout the network
stack still try to maintain this complete csum value if it is required
through functions such as skb_postpush_rcsum.

The MPLS actions in Open vSwitch modify a CHECKSUM_COMPLETE value when
changes are made to packet data without a push or a pull. This occurs when
the ethertype of the MAC header is changed or when MPLS lse fields are
modified.

The modification is carried out using the csum_partial function to get the
csum of a buffer and add it into the larger checksum. The buffer is an
inversion of the data to be removed followed by the new data. Because the
csum is calculated over 16 bits and these values align with 16 bits, the
effect is the removal of the old value from the CHECKSUM_COMPLETE and
addition of the new value.

However, the csum fed into the function and the outcome of the
calculation are also inverted. This would only make sense if it was the
new value rather than the old that was inverted in the input buffer.

Fix the issue by removing the bit inverts in the csum_partial calculation.

The bug was verified and the fix tested by comparing the folded value of
the updated CHECKSUM_COMPLETE value with the folded value of a full
software checksum calculation (reset skb->csum to 0 and run
skb_checksum_complete(skb)). Prior to the fix the outcomes differed but
after they produce the same result.

Fixes: 25cd9ba0abc0 ("openvswitch: Add basic MPLS support to kernel")
Fixes: bc7cc5999fd3 ("openvswitch: update checksum in {push,pop}_mpls")
Signed-off-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'bnxt_en-Bug-fixes'
David S. Miller [Sun, 30 Jun 2019 23:00:26 +0000 (16:00 -0700)]
Merge branch 'bnxt_en-Bug-fixes'

Michael Chan says:

====================
bnxt_en: Bug fixes.

Miscellaneous bug fix patches, including two resource handling fixes for
the RDMA driver, a PCI shutdown patch to add pci_disable_device(), a patch
to fix ethtool selftest crash, and the last one suppresses an unnecessry
error message.

Please also queue patches 1, 2, and 3 for -stable.  Thanks.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Suppress error messages when querying DSCP DCB capabilities.
Michael Chan [Sat, 29 Jun 2019 15:16:48 +0000 (11:16 -0400)]
bnxt_en: Suppress error messages when querying DSCP DCB capabilities.

Some firmware versions do not support this so use the silent variant
to send the message to firmware to suppress the harmless error.  This
error message is unnecessarily alarming the user.

Fixes: afdc8a84844a ("bnxt_en: Add DCBNL DSCP application protocol support.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Cap the returned MSIX vectors to the RDMA driver.
Michael Chan [Sat, 29 Jun 2019 15:16:47 +0000 (11:16 -0400)]
bnxt_en: Cap the returned MSIX vectors to the RDMA driver.

In an earlier commit to improve NQ reservations on 57500 chips, we
set the resv_irqs on the 57500 VFs to the fixed value assigned by
the PF regardless of how many are actually used.  The current
code assumes that resv_irqs minus the ones used by the network driver
must be the ones for the RDMA driver.  This is no longer true and
we may return more MSIX vectors than requested, causing inconsistency.
Fix it by capping the value.

Fixes: 01989c6b69d9 ("bnxt_en: Improve NQ reservations.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Fix statistics context reservation logic for RDMA driver.
Michael Chan [Sat, 29 Jun 2019 15:16:46 +0000 (11:16 -0400)]
bnxt_en: Fix statistics context reservation logic for RDMA driver.

The current logic assumes that the RDMA driver uses one statistics
context adjacent to the ones used by the network driver.  This
assumption is not true and the statistics context used by the
RDMA driver is tied to its MSIX base vector.  This wrong assumption
can cause RDMA driver failure after changing ethtool rings on the
network side.  Fix the statistics reservation logic accordingly.

Fixes: 780baad44f0f ("bnxt_en: Reserve 1 stat_ctx for RDMA driver.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Fix ethtool selftest crash under error conditions.
Michael Chan [Sat, 29 Jun 2019 15:16:45 +0000 (11:16 -0400)]
bnxt_en: Fix ethtool selftest crash under error conditions.

After ethtool loopback packet tests, we re-open the nic for the next
IRQ test.  If the open fails, we must not proceed with the IRQ test
or we will crash with NULL pointer dereference.  Fix it by checking
the bnxt_open_nic() return code before proceeding.

Reported-by: Somasundaram Krishnasamy <somasundaram.krishnasamy@oracle.com>
Fixes: 67fea463fd87 ("bnxt_en: Add interrupt test to ethtool -t selftest.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Disable bus master during PCI shutdown and driver unload.
Michael Chan [Sat, 29 Jun 2019 15:16:44 +0000 (11:16 -0400)]
bnxt_en: Disable bus master during PCI shutdown and driver unload.

Some chips with older firmware can continue to perform DMA read from
context memory even after the memory has been freed.  In the PCI shutdown
method, we need to call pci_disable_device() to shutdown DMA to prevent
this DMA before we put the device into D3hot.  DMA memory request in
D3hot state will generate PCI fatal error.  Similarly, in the driver
remove method, the context memory should only be freed after DMA has
been shutdown for correctness.

Fixes: 98f04cf0f1fc ("bnxt_en: Check context memory requirements from firmware.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: mv88e6xxx: wait after reset deactivation
Baruch Siach [Thu, 27 Jun 2019 18:17:39 +0000 (21:17 +0300)]
net: dsa: mv88e6xxx: wait after reset deactivation

Add a 1ms delay after reset deactivation. Otherwise the chip returns
bogus ID value. This is observed with 88E6390 (Peridot) chip.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnx2x: Prevent ptp_task to be rescheduled indefinitely
Guilherme G. Piccoli [Thu, 27 Jun 2019 16:31:33 +0000 (13:31 -0300)]
bnx2x: Prevent ptp_task to be rescheduled indefinitely

Currently bnx2x ptp worker tries to read a register with timestamp
information in case of TX packet timestamping and in case it fails,
the routine reschedules itself indefinitely. This was reported as a
kworker always at 100% of CPU usage, which was narrowed down to be
bnx2x ptp_task.

By following the ioctl handler, we could narrow down the problem to
an NTP tool (chrony) requesting HW timestamping from bnx2x NIC with
RX filter zeroed; this isn't reproducible for example with ptp4l
(from linuxptp) since this tool requests a supported RX filter.
It seems NIC FW timestamp mechanism cannot work well with
RX_FILTER_NONE - driver's PTP filter init routine skips a register
write to the adapter if there's not a supported filter request.

This patch addresses the problem of bnx2x ptp thread's everlasting
reschedule by retrying the register read 10 times; between the read
attempts the thread sleeps for an increasing amount of time starting
in 1ms to give FW some time to perform the timestamping. If it still
fails after all retries, we bail out in order to prevent an unbound
resource consumption from bnx2x.

The patch also adds an ethtool statistic for accounting the skipped
TX timestamp packets and it reduces the priority of timestamping
error messages to prevent log flooding. The code was tested using
both linuxptp and chrony.

Reported-and-tested-by: Przemyslaw Hausman <przemyslaw.hausman@canonical.com>
Suggested-by: Sudarsana Reddy Kalluru <skalluru@marvell.com>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@canonical.com>
Acked-by: Sudarsana Reddy Kalluru <skalluru@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoigmp: fix memory leak in igmpv3_del_delrec()
Eric Dumazet [Thu, 27 Jun 2019 08:27:01 +0000 (01:27 -0700)]
igmp: fix memory leak in igmpv3_del_delrec()

im->tomb and/or im->sources might not be NULL, but we
currently overwrite their values blindly.

Using swap() will make sure the following call to kfree_pmc(pmc)
will properly free the psf structures.

Tested with the C repro provided by syzbot, which basically does :

 socket(PF_INET, SOCK_DGRAM, IPPROTO_IP) = 3
 setsockopt(3, SOL_IP, IP_ADD_MEMBERSHIP, "\340\0\0\2\177\0\0\1\0\0\0\0", 12) = 0
 ioctl(3, SIOCSIFFLAGS, {ifr_name="lo", ifr_flags=0}) = 0
 setsockopt(3, SOL_IP, IP_MSFILTER, "\340\0\0\2\177\0\0\1\1\0\0\0\1\0\0\0\377\377\377\377", 20) = 0
 ioctl(3, SIOCSIFFLAGS, {ifr_name="lo", ifr_flags=IFF_UP}) = 0
 exit_group(0)                    = ?

BUG: memory leak
unreferenced object 0xffff88811450f140 (size 64):
  comm "softirq", pid 0, jiffies 4294942448 (age 32.070s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 ff ff ff ff 00 00 00 00  ................
    00 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................
  backtrace:
    [<00000000c7bad083>] kmemleak_alloc_recursive include/linux/kmemleak.h:43 [inline]
    [<00000000c7bad083>] slab_post_alloc_hook mm/slab.h:439 [inline]
    [<00000000c7bad083>] slab_alloc mm/slab.c:3326 [inline]
    [<00000000c7bad083>] kmem_cache_alloc_trace+0x13d/0x280 mm/slab.c:3553
    [<000000009acc4151>] kmalloc include/linux/slab.h:547 [inline]
    [<000000009acc4151>] kzalloc include/linux/slab.h:742 [inline]
    [<000000009acc4151>] ip_mc_add1_src net/ipv4/igmp.c:1976 [inline]
    [<000000009acc4151>] ip_mc_add_src+0x36b/0x400 net/ipv4/igmp.c:2100
    [<000000004ac14566>] ip_mc_msfilter+0x22d/0x310 net/ipv4/igmp.c:2484
    [<0000000052d8f995>] do_ip_setsockopt.isra.0+0x1795/0x1930 net/ipv4/ip_sockglue.c:959
    [<000000004ee1e21f>] ip_setsockopt+0x3b/0xb0 net/ipv4/ip_sockglue.c:1248
    [<0000000066cdfe74>] udp_setsockopt+0x4e/0x90 net/ipv4/udp.c:2618
    [<000000009383a786>] sock_common_setsockopt+0x38/0x50 net/core/sock.c:3126
    [<00000000d8ac0c94>] __sys_setsockopt+0x98/0x120 net/socket.c:2072
    [<000000001b1e9666>] __do_sys_setsockopt net/socket.c:2083 [inline]
    [<000000001b1e9666>] __se_sys_setsockopt net/socket.c:2080 [inline]
    [<000000001b1e9666>] __x64_sys_setsockopt+0x26/0x30 net/socket.c:2080
    [<00000000420d395e>] do_syscall_64+0x76/0x1a0 arch/x86/entry/common.c:301
    [<000000007fd83a4b>] entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fixes: 24803f38a5c0 ("igmp: do not remove igmp souce list info when set link down")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Hangbin Liu <liuhangbin@gmail.com>
Reported-by: syzbot+6ca1abd0db68b5173a4f@syzkaller.appspotmail.com
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'Sub-ns-increment-fixes-in-Macb-PTP'
David S. Miller [Sat, 29 Jun 2019 18:09:18 +0000 (11:09 -0700)]
Merge branch 'Sub-ns-increment-fixes-in-Macb-PTP'

Harini Katakam says:

====================
Sub ns increment fixes in Macb PTP

The subns increment register fields are not captured correctly in the
driver. Fix the same and also increase the subns incr resolution.

Sub ns resolution was increased to 24 bits in r1p06f2 version. To my
knowledge, this PTP driver, with its current BD time stamp
implementation, is only useful to that version or above. So, I have
increased the resolution unconditionally. Please let me know if there
is any IP versions incompatible with this - there is no register to
obtain this information from.

Changes from RFC:
None
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: macb: Fix SUBNS increment and increase resolution
Harini Katakam [Thu, 27 Jun 2019 06:21:00 +0000 (11:51 +0530)]
net: macb: Fix SUBNS increment and increase resolution

The subns increment register has 24 bits as follows:
RegBit[15:0] = Subns[23:8]; RegBit[31:24] = Subns[7:0]

Fix the same in the driver and increase sub ns resolution to the
best capable, 24 bits. This should be the case on all GEM versions
that this PTP driver supports.

Signed-off-by: Harini Katakam <harini.katakam@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: macb: Add separate definition for PPM fraction
Harini Katakam [Thu, 27 Jun 2019 06:20:59 +0000 (11:50 +0530)]
net: macb: Add separate definition for PPM fraction

The scaled ppm parameter passed to _adjfine() contains a 16 bit
fraction. This just happens to be the same as SUBNSINCR_SIZE now.
Hence define this separately.

Signed-off-by: Harini Katakam <harini.katakam@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agopacket: Fix undefined behavior in bit shift
Jiunn Chang [Thu, 27 Jun 2019 03:25:30 +0000 (22:25 -0500)]
packet: Fix undefined behavior in bit shift

Shifting signed 32-bit value by 31 bits is undefined.  Changing most
significant bit to unsigned.

Changes included in v2:
  - use subsystem specific subject lines
  - CC required mailing lists

Signed-off-by: Jiunn Chang <c0d1n61at3@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: make skb_dst_force return true when dst is refcounted
Florian Westphal [Wed, 26 Jun 2019 18:40:45 +0000 (20:40 +0200)]
net: make skb_dst_force return true when dst is refcounted

netfilter did not expect that skb_dst_force() can cause skb to lose its
dst entry.

I got a bug report with a skb->dst NULL dereference in netfilter
output path.  The backtrace contains nf_reinject(), so the dst might have
been cleared when skb got queued to userspace.

Other users were fixed via
if (skb_dst(skb)) {
skb_dst_force(skb);
if (!skb_dst(skb))
goto handle_err;
}

But I think its preferable to make the 'dst might be cleared' part
of the function explicit.

In netfilter case, skb with a null dst is expected when queueing in
prerouting hook, so drop skb for the other hooks.

v2:
 v1 of this patch returned true in case skb had no dst entry.
 Eric said:
   Say if we have two skb_dst_force() calls for some reason
   on the same skb, only the first one will return false.

 This now returns false even when skb had no dst, as per Erics
 suggestion, so callers might need to check skb_dst() first before
 skb_dst_force().

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agosctp: not bind the socket in sctp_connect
Xin Long [Wed, 26 Jun 2019 08:31:39 +0000 (16:31 +0800)]
sctp: not bind the socket in sctp_connect

Now when sctp_connect() is called with a wrong sa_family, it binds
to a port but doesn't set bp->port, then sctp_get_af_specific will
return NULL and sctp_connect() returns -EINVAL.

Then if sctp_bind() is called to bind to another port, the last
port it has bound will leak due to bp->port is NULL by then.

sctp_connect() doesn't need to bind ports, as later __sctp_connect
will do it if bp->port is NULL. So remove it from sctp_connect().
While at it, remove the unnecessary sockaddr.sa_family len check
as it's already done in sctp_inet_connect.

Fixes: 644fbdeacf1d ("sctp: fix the issue that flags are ignored when using kernel_connect")
Reported-by: syzbot+079bf326b38072f849d9@syzkaller.appspotmail.com
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobpf: fix uapi bpf_prog_info fields alignment
Baruch Siach [Fri, 28 Jun 2019 04:08:45 +0000 (07:08 +0300)]
bpf: fix uapi bpf_prog_info fields alignment

Merge commit 1c8c5a9d38f60 ("Merge
git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next") undid the
fix from commit 36f9814a494 ("bpf: fix uapi hole for 32 bit compat
applications") by taking the gpl_compatible 1-bit field definition from
commit b85fab0e67b162 ("bpf: Add gpl_compatible flag to struct
bpf_prog_info") as is. That breaks architectures with 16-bit alignment
like m68k. Add 31-bit pad after gpl_compatible to restore alignment of
following fields.

Thanks to Dmitry V. Levin his analysis of this bug history.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Acked-by: Song Liu <songliubraving@fb.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
5 years agonet: mvpp2: prs: Don't override the sign bit in SRAM parser shift
Maxime Chevallier [Thu, 20 Jun 2019 09:42:45 +0000 (11:42 +0200)]
net: mvpp2: prs: Don't override the sign bit in SRAM parser shift

The Header Parser allows identifying various fields in the packet
headers, used for various kind of filtering and classification
steps.

This is a re-entrant process, where the offset in the packet header
depends on the previous lookup results. This offset is represented in
the SRAM results of the TCAM, as a shift to be operated.

This shift can be negative in some cases, such as in IPv6 parsing.

This commit prevents overriding the sign bit when setting the shift
value, which could cause instabilities when parsing IPv6 flows.

Fixes: 3f518509dedc ("ethernet: Add new driver for Marvell Armada 375 network unit")
Suggested-by: Alan Winkowski <walan@marvell.com>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: phylink: further documentation clarifications
Russell King [Tue, 25 Jun 2019 09:44:33 +0000 (10:44 +0100)]
net: phylink: further documentation clarifications

Clarify the validate() behaviour in a few cases which weren't mentioned
in the documentation, but which are necessary for users to get the
correct behaviour.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: ethernet: ti: cpsw: Assign OF node to slave devices
Marek Vasut [Sun, 23 Jun 2019 12:11:43 +0000 (14:11 +0200)]
net: ethernet: ti: cpsw: Assign OF node to slave devices

Assign OF node to CPSW slave devices, otherwise it is not possible to
bind e.g. DSA switch to them. Without this patch, the DSA code tries
to find the ethernet device by OF match, but fails to do so because
the slave device has NULL OF node.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: David S. Miller <davem@davemloft.net>
Cc: Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf
David S. Miller [Fri, 28 Jun 2019 20:36:43 +0000 (13:36 -0700)]
Merge git://git./pub/scm/linux/kernel/git/pablo/nf

Pablo Neira Ayuso says:

====================
Netfilter/IPVS fixes for net

The following patchset contains Netfilter fixes for net:

1) Fix memleak reported by syzkaller when registering IPVS hooks,
   patch from Julian Anastasov.

2) Fix memory leak in start_sync_thread, also from Julian.

3) Fix conntrack deletion via ctnetlink, from Felix Kaechele.

4) Fix reject for ICMP due to incorrect checksum handling, from
   He Zhe.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: b53: Disable all ports on setup
Benedikt Spranger [Fri, 28 Jun 2019 16:58:11 +0000 (18:58 +0200)]
net: dsa: b53: Disable all ports on setup

A b53 device may configured through an external EEPROM like the switch
device on the Lamobo R1 router board. The configuration of a port may
therefore differ from the reset configuration of the switch.

The switch configuration reported by the DSA subsystem is different until
the port is configured by DSA i.e. a port can be active, while the DSA
subsystem reports the port is inactive. Disable all ports and not only
the unused ones to put all ports into a well defined state.

Signed-off-by: Benedikt Spranger <b.spranger@linutronix.de>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonetfilter: Fix remainder of pseudo-header protocol 0
He Zhe [Mon, 24 Jun 2019 03:17:38 +0000 (11:17 +0800)]
netfilter: Fix remainder of pseudo-header protocol 0

Since v5.1-rc1, some types of packets do not get unreachable reply with the
following iptables setting. Fox example,

$ iptables -A INPUT -p icmp --icmp-type 8 -j REJECT
$ ping 127.0.0.1 -c 1
PING 127.0.0.1 (127.0.0.1) 56(84) bytes of data.
— 127.0.0.1 ping statistics —
1 packets transmitted, 0 received, 100% packet loss, time 0ms

We should have got the following reply from command line, but we did not.
From 127.0.0.1 icmp_seq=1 Destination Port Unreachable

Yi Zhao reported it and narrowed it down to:
7fc38225363d ("netfilter: reject: skip csum verification for protocols that don't support it"),

This is because nf_ip_checksum still expects pseudo-header protocol type 0 for
packets that are of neither TCP or UDP, and thus ICMP packets are mistakenly
treated as TCP/UDP.

This patch corrects the conditions in nf_ip_checksum and all other places that
still call it with protocol 0.

Fixes: 7fc38225363d ("netfilter: reject: skip csum verification for protocols that don't support it")
Reported-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
5 years agoMerge tag 'mac80211-for-davem-2019-06-28' of git://git.kernel.org/pub/scm/linux/kerne...
David S. Miller [Fri, 28 Jun 2019 16:49:56 +0000 (09:49 -0700)]
Merge tag 'mac80211-for-davem-2019-06-28' of git://git./linux/kernel/git/jberg/mac80211

Johannes Berg says:

====================
Just a single patch:
 * 1<<31 is undefined, use 1U<<31 in nl80211.h UAPI
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge tag 'batadv-net-for-davem-20190627' of git://git.open-mesh.org/linux-merge
David S. Miller [Fri, 28 Jun 2019 16:45:27 +0000 (09:45 -0700)]
Merge tag 'batadv-net-for-davem-20190627' of git://git.open-mesh.org/linux-merge

Simon Wunderlich says:

====================
Here are some batman-adv bugfixes:

 - fix a leaked TVLV handler which wasn't unregistered, by Jeremy Sowden

 - fix duplicated OGMs when interfaces are set UP, by Sven Eckelmann
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge tag 'wireless-drivers-for-davem-2019-06-28' of git://git.kernel.org/pub/scm...
David S. Miller [Fri, 28 Jun 2019 16:15:02 +0000 (09:15 -0700)]
Merge tag 'wireless-drivers-for-davem-2019-06-28' of git://git./linux/kernel/git/kvalo/wireless-drivers

Kalle Valo says:

====================
wireless-drivers fixes for 5.2

Hopefully the last set of fixes for 5.2. Nothing special this around,
only small fixes and support for new cards.

iwlwifi

* add new cards for 22000 series and smaller fixes

wl18xx

* fix a clang warning about unused variables

mwifiex

* properly handle small vendor IEs (a regression from the recent
  security fix)

ath

* fix few SPDX tags

mt76

* fix A-MSDU aggregation which got broken in v5.2-rc1
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonl80211: Fix undefined behavior in bit shift
Jiunn Chang [Thu, 27 Jun 2019 05:04:26 +0000 (00:04 -0500)]
nl80211: Fix undefined behavior in bit shift

Shifting signed 32-bit value by 31 bits is undefined.  Changing most
significant bit to unsigned.

Signed-off-by: Jiunn Chang <c0d1n61at3@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
5 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 28 Jun 2019 00:50:09 +0000 (08:50 +0800)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "A handful of clk driver fixes and one core framework fix

   - Do a DT/firmware lookup in clk_core_get() even when the DT index is
     a nonsensical value

   - Fix some clk data typos in the Amlogic DT headers/code

   - Avoid returning junk in the TI clk driver when an invalid clk is
     looked for

   - Fix dividers for the emac clks on Stratix10 SoCs

   - Fix default HDA rates on Tegra210 to correct distorted audio"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: socfpga: stratix10: fix divider entry for the emac clocks
  clk: Do a DT parent lookup even when index < 0
  clk: tegra210: Fix default rates for HDA clocks
  clk: ti: clkctrl: Fix returning uninitialized data
  clk: meson: meson8b: fix a typo in the VPU parent names array variable
  clk: meson: fix MPLL 50M binding id typo

5 years agoMerge tag 'for-5.2/dm-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/devic...
Linus Torvalds [Fri, 28 Jun 2019 00:48:21 +0000 (08:48 +0800)]
Merge tag 'for-5.2/dm-fixes-2' of git://git./linux/kernel/git/device-mapper/linux-dm

Pull device mapper fixes from Mike Snitzer:

 - Fix incorrect uses of kstrndup and DM logging macros in DM's early
   init code.

 - Fix DM log-writes target's handling of super block sectors so updates
   are made in order through use of completion.

 - Fix DM core's argument splitting code to avoid undefined behaviour
   reported as a side-effect of UBSAN analysis on ppc64le.

 - Fix DM verity target to limit the amount of error messages that can
   result from a corrupt block being found.

* tag 'for-5.2/dm-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm verity: use message limit for data block corruption message
  dm table: don't copy from a NULL pointer in realloc_argv()
  dm log writes: make sure super sector log updates are written in order
  dm init: remove trailing newline from calls to DMERR() and DMINFO()
  dm init: fix incorrect uses of kstrndup()

5 years agoMerge tag 'for-linus-20190627' of gitolite.kernel.org:pub/scm/linux/kernel/git/braune...
Linus Torvalds [Fri, 28 Jun 2019 00:41:18 +0000 (08:41 +0800)]
Merge tag 'for-linus-20190627' of gitolite.pub/scm/linux/kernel/git/brauner/linux

Pull pidfd fixes from Christian Brauner:
 "Userspace tools and libraries such as strace or glibc need a cheap and
  reliable way to tell whether CLONE_PIDFD is supported. The easiest way
  is to pass an invalid fd value in the return argument, perform the
  syscall and verify the value in the return argument has been changed
  to a valid fd.

  However, if CLONE_PIDFD is specified we currently check if pidfd == 0
  and return EINVAL if not.

  The check for pidfd == 0 was originally added to enable us to abuse
  the return argument for passing additional flags along with
  CLONE_PIDFD in the future.

  However, extending legacy clone this way would be a terrible idea and
  with clone3 on the horizon and the ability to reuse CLONE_DETACHED
  with CLONE_PIDFD there's no real need for this clutch. So remove the
  pidfd == 0 check and help userspace out.

  Also, accordig to Al, anon_inode_getfd() should only be used past the
  point of no failure and ksys_close() should not be used at all since
  it is far too easy to get wrong. Al's motto being "basically, once
  it's in descriptor table, it's out of your control". So Al's patch
  switches back to what we already had in v1 of the original patchset
  and uses a anon_inode_getfile() + put_user() + fd_install() sequence
  in the success path and a fput() + put_unused_fd() in the failure
  path.

  The other two changes should be trivial"

* tag 'for-linus-20190627' of gitolite.kernel.org:pub/scm/linux/kernel/git/brauner/linux:
  proc: remove useless d_is_dir() check
  copy_process(): don't use ksys_close() on cleanups
  samples: make pidfd-metadata fail gracefully on older kernels
  fork: don't check parent_tidptr with CLONE_PIDFD

5 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid
Linus Torvalds [Fri, 28 Jun 2019 00:39:18 +0000 (08:39 +0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/hid/hid

Pull HID fixes from Jiri Kosina:

 - fix for one corner case in HID++ protocol with respect to handling
   very long reports, from Hans de Goede

 - power management fix in Intel-ISH driver, from Hyungwoo Yang

 - use-after-free fix in Intel-ISH driver, from Dan Carpenter

 - a couple of new device IDs/quirks from Kai-Heng Feng, Kyle Godbey and
   Oleksandr Natalenko

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid:
  HID: intel-ish-hid: fix wrong driver_data usage
  HID: multitouch: Add pointstick support for ALPS Touchpad
  HID: logitech-dj: Fix forwarding of very long HID++ reports
  HID: uclogic: Add support for Huion HS64 tablet
  HID: chicony: add another quirk for PixArt mouse
  HID: intel-ish-hid: Fix a use after free in load_fw_from_host()

5 years agoMerge tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Fri, 28 Jun 2019 00:37:04 +0000 (08:37 +0800)]
Merge tag 'armsoc-fixes' of git://git./linux/kernel/git/soc/soc

Pull ARM SoC fixes from Olof Johansson:
 "A smaller batch of fixes, nothing that stands out as risky or scary.

  Mostly DTS tweaks for a few issues:

   - GPU fixlets for Meson

   - CPU idle fix for LS1028A

   - PWM interrupt fixes for i.MX6UL

  Also, enable a driver (FSL_EDMA) on arm64 defconfig, and a warning and
  two MAINTAINER tweaks"

* tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
  ARM: dts: imx6ul: fix PWM[1-4] interrupts
  ARM: omap2: remove incorrect __init annotation
  ARM: dts: gemini Fix up DNS-313 compatible string
  ARM: dts: Blank D-Link DIR-685 console
  arm64: defconfig: Enable FSL_EDMA driver
  arm64: dts: ls1028a: Fix CPU idle fail.
  MAINTAINERS: BCM53573: Add internal Broadcom mailing list
  MAINTAINERS: BCM2835: Add internal Broadcom mailing list
  ARM: dts: meson8b: fix the operating voltage of the Mali GPU
  ARM: dts: meson8b: drop undocumented property from the Mali GPU node
  ARM: dts: meson8: fix GPU interrupts and drop an undocumented property

5 years agoMerge tag 'afs-fixes-20190620' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowe...
Linus Torvalds [Fri, 28 Jun 2019 00:34:12 +0000 (08:34 +0800)]
Merge tag 'afs-fixes-20190620' of git://git./linux/kernel/git/dhowells/linux-fs

Pull AFS fixes from David Howells:
 "The in-kernel AFS client has been undergoing testing on opendev.org on
  one of their mirror machines. They are using AFS to hold data that is
  then served via apache, and Ian Wienand had reported seeing oopses,
  spontaneous machine reboots and updates to volumes going missing. This
  patch series appears to have fixed the problem, very probably due to
  patch (2), but it's not 100% certain.

  (1) Fix the printing of the "vnode modified" warning to exclude checks
      on files for which we don't have a callback promise from the
      server (and so don't expect the server to tell us when it
      changes).

      Without this, for every file or directory for which we still have
      an in-core inode that gets changed on the server, we may get a
      message logged when we next look at it. This can happen in bulk
      if, for instance, someone does "vos release" to update a R/O
      volume from a R/W volume and a whole set of files are all changed
      together.

      We only really want to log a message if the file changed and the
      server didn't tell us about it or we failed to track the state
      internally.

  (2) Fix accidental corruption of either afs_vlserver struct objects or
      the the following memory locations (which could hold anything).
      The issue is caused by a union that points to two different
      structs in struct afs_call (to save space in the struct). The call
      cleanup code assumes that it can simply call the cleanup for one
      of those structs if not NULL - when it might be actually pointing
      to the other struct.

      This means that every Volume Location RPC op is going to corrupt
      something.

  (3) Fix an uninitialised spinlock. This isn't too bad, it just causes
      a one-off warning if lockdep is enabled when "vos release" is
      called, but the spinlock still behaves correctly.

  (4) Fix the setting of i_block in the inode. This causes du, for
      example, to produce incorrect results, but otherwise should not be
      dangerous to the kernel"

* tag 'afs-fixes-20190620' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs:
  afs: Fix setting of i_blocks
  afs: Fix uninitialised spinlock afs_volume::cb_break_lock
  afs: Fix vlserver record corruption
  afs: Fix over zealous "vnode modified" warnings

5 years agoMerge tag 'csky-for-linus-5.2-fixup-gcc-unwind' of git://github.com/c-sky/csky-linux
Linus Torvalds [Fri, 28 Jun 2019 00:31:57 +0000 (08:31 +0800)]
Merge tag 'csky-for-linus-5.2-fixup-gcc-unwind' of git://github.com/c-sky/csky-linux

Pull arch/csky fixup from Guo Ren:
 "A fixup patch for rt_sigframe in signal.c"

* tag 'csky-for-linus-5.2-fixup-gcc-unwind' of git://github.com/c-sky/csky-linux:
  csky: Fixup libgcc unwind error

5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Fri, 28 Jun 2019 00:24:37 +0000 (08:24 +0800)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Fix ppp_mppe crypto soft dependencies, from Takashi Iawi.

 2) Fix TX completion to be finite, from Sergej Benilov.

 3) Use register_pernet_device to avoid a dst leak in tipc, from Xin
    Long.

 4) Double free of TX cleanup in Dirk van der Merwe.

 5) Memory leak in packet_set_ring(), from Eric Dumazet.

 6) Out of bounds read in qmi_wwan, from Bjørn Mork.

 7) Fix iif used in mcast/bcast looped back packets, from Stephen
    Suryaputra.

 8) Fix neighbour resolution on raw ipv6 sockets, from Nicolas Dichtel.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (25 commits)
  af_packet: Block execution of tasks waiting for transmit to complete in AF_PACKET
  sctp: change to hold sk after auth shkey is created successfully
  ipv6: fix neighbour resolution with raw socket
  ipv6: constify rt6_nexthop()
  net: dsa: microchip: Use gpiod_set_value_cansleep()
  net: aquantia: fix vlans not working over bridged network
  ipv4: reset rt_iif for recirculated mcast/bcast out pkts
  team: Always enable vlan tx offload
  net/smc: Fix error path in smc_init
  net/smc: hold conns_lock before calling smc_lgr_register_conn()
  bonding: Always enable vlan tx offload
  net/ipv6: Fix misuse of proc_dointvec "skip_notify_on_dev_down"
  ipv4: Use return value of inet_iif() for __raw_v4_lookup in the while loop
  qmi_wwan: Fix out-of-bounds read
  tipc: check msg->req data len in tipc_nl_compat_bearer_disable
  net: macb: do not copy the mac address if NULL
  net/packet: fix memory leak in packet_set_ring()
  net/tls: fix page double free on TX cleanup
  net/sched: cbs: Fix error path of cbs_module_init
  tipc: change to use register_pernet_device
  ...

5 years agomt76: usb: fix rx A-MSDU support
Lorenzo Bianconi [Sat, 15 Jun 2019 14:03:32 +0000 (16:03 +0200)]
mt76: usb: fix rx A-MSDU support

Commit f8f527b16db5 ("mt76: usb: use EP max packet aligned buffer sizes
for rx") breaks A-MSDU support. When A-MSDU is enable the device can
receive frames up to q->buf_size but they will be discarded in
mt76u_process_rx_entry since there is no enough room for
skb_shared_info. Fix the issue reallocating the skb and copying in the
linear area the first 128B of the received frames and in the frag_list
the remaining part

Fixes: f8f527b16db5 ("mt76: usb: use EP max packet aligned buffer sizes for rx")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
5 years agoproc: remove useless d_is_dir() check
Christian Brauner [Thu, 27 Jun 2019 09:35:14 +0000 (11:35 +0200)]
proc: remove useless d_is_dir() check

Remove the d_is_dir() check from tgid_pidfd_to_pid().

It is pointless since you should never get &proc_tgid_base_operations
for f_op on a non-directory.

Suggested-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Christian Brauner <christian@brauner.io>
5 years agocopy_process(): don't use ksys_close() on cleanups
Al Viro [Thu, 27 Jun 2019 02:22:09 +0000 (22:22 -0400)]
copy_process(): don't use ksys_close() on cleanups

anon_inode_getfd() should be used *ONLY* in situations when we are
guaranteed to be past the last failure point (including copying the
descriptor number to userland, at that).  And ksys_close() should
not be used for cleanups at all.

anon_inode_getfile() is there for all nontrivial cases like that.
Just use that...

Fixes: b3e583825266 ("clone: add CLONE_PIDFD")
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Reviewed-by: Jann Horn <jannh@google.com>
Signed-off-by: Christian Brauner <christian@brauner.io>
5 years agoaf_packet: Block execution of tasks waiting for transmit to complete in AF_PACKET
Neil Horman [Tue, 25 Jun 2019 21:57:49 +0000 (17:57 -0400)]
af_packet: Block execution of tasks waiting for transmit to complete in AF_PACKET

When an application is run that:
a) Sets its scheduler to be SCHED_FIFO
and
b) Opens a memory mapped AF_PACKET socket, and sends frames with the
MSG_DONTWAIT flag cleared, its possible for the application to hang
forever in the kernel.  This occurs because when waiting, the code in
tpacket_snd calls schedule, which under normal circumstances allows
other tasks to run, including ksoftirqd, which in some cases is
responsible for freeing the transmitted skb (which in AF_PACKET calls a
destructor that flips the status bit of the transmitted frame back to
available, allowing the transmitting task to complete).

However, when the calling application is SCHED_FIFO, its priority is
such that the schedule call immediately places the task back on the cpu,
preventing ksoftirqd from freeing the skb, which in turn prevents the
transmitting task from detecting that the transmission is complete.

We can fix this by converting the schedule call to a completion
mechanism.  By using a completion queue, we force the calling task, when
it detects there are no more frames to send, to schedule itself off the
cpu until such time as the last transmitted skb is freed, allowing
forward progress to be made.

Tested by myself and the reporter, with good results

Change Notes:

V1->V2:
Enhance the sleep logic to support being interruptible and
allowing for honoring to SK_SNDTIMEO (Willem de Bruijn)

V2->V3:
Rearrage the point at which we wait for the completion queue, to
avoid needing to check for ph/skb being null at the end of the loop.
Also move the complete call to the skb destructor to avoid needing to
modify __packet_set_status.  Also gate calling complete on
packet_read_pending returning zero to avoid multiple calls to complete.
(Willem de Bruijn)

Move timeo computation within loop, to re-fetch the socket
timeout since we also use the timeo variable to record the return code
from the wait_for_complete call (Neil Horman)

V3->V4:
Willem has requested that the control flow be restored to the
previous state.  Doing so lets us eliminate the need for the
po->wait_on_complete flag variable, and lets us get rid of the
packet_next_frame function, but introduces another complexity.
Specifically, but using the packet pending count, we can, if an
applications calls sendmsg multiple times with MSG_DONTWAIT set, each
set of transmitted frames, when complete, will cause
tpacket_destruct_skb to issue a complete call, for which there will
never be a wait_on_completion call.  This imbalance will lead to any
future call to wait_for_completion here to return early, when the frames
they sent may not have completed.  To correct this, we need to re-init
the completion queue on every call to tpacket_snd before we enter the
loop so as to ensure we wait properly for the frames we send in this
iteration.

Change the timeout and interrupted gotos to out_put rather than
out_status so that we don't try to free a non-existant skb
Clean up some extra newlines (Willem de Bruijn)

Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Reported-by: Matteo Croce <mcroce@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agosctp: change to hold sk after auth shkey is created successfully
Xin Long [Mon, 24 Jun 2019 16:21:45 +0000 (00:21 +0800)]
sctp: change to hold sk after auth shkey is created successfully

Now in sctp_endpoint_init(), it holds the sk then creates auth
shkey. But when the creation fails, it doesn't release the sk,
which causes a sk defcnf leak,

Here to fix it by only holding the sk when auth shkey is created
successfully.

Fixes: a29a5bd4f5c3 ("[SCTP]: Implement SCTP-AUTH initializations.")
Reported-by: syzbot+afabda3890cc2f765041@syzkaller.appspotmail.com
Reported-by: syzbot+276ca1c77a19977c0130@syzkaller.appspotmail.com
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Neil Horman <nhorman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'ipv6-fix-neighbour-resolution-with-raw-socket'
David S. Miller [Wed, 26 Jun 2019 20:26:08 +0000 (13:26 -0700)]
Merge branch 'ipv6-fix-neighbour-resolution-with-raw-socket'

Nicolas Dichtel says:

====================
ipv6: fix neighbour resolution with raw socket

The first patch prepares the fix, it constify rt6_nexthop().
The detail of the bug is explained in the second patch.

v1 -> v2:
 - fix compilation warnings
 - split the initial patch
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: fix neighbour resolution with raw socket
Nicolas Dichtel [Mon, 24 Jun 2019 14:01:09 +0000 (16:01 +0200)]
ipv6: fix neighbour resolution with raw socket

The scenario is the following: the user uses a raw socket to send an ipv6
packet, destinated to a not-connected network, and specify a connected nh.
Here is the corresponding python script to reproduce this scenario:

 import socket
 IPPROTO_RAW = 255
 send_s = socket.socket(socket.AF_INET6, socket.SOCK_RAW, IPPROTO_RAW)
 # scapy
 # p = IPv6(src='fd00:100::1', dst='fd00:200::fa')/ICMPv6EchoRequest()
 # str(p)
 req = b'`\x00\x00\x00\x00\x08:@\xfd\x00\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01\xfd\x00\x02\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\xfa\x80\x00\x81\xc0\x00\x00\x00\x00'
 send_s.sendto(req, ('fd00:175::2', 0, 0, 0))

fd00:175::/64 is a connected route and fd00:200::fa is not a connected
host.

With this scenario, the kernel starts by sending a NS to resolve
fd00:175::2. When it receives the NA, it flushes its queue and try to send
the initial packet. But instead of sending it, it sends another NS to
resolve fd00:200::fa, which obvioulsy fails, thus the packet is dropped. If
the user sends again the packet, it now uses the right nh (fd00:175::2).

The problem is that ip6_dst_lookup_neigh() uses the rt6i_gateway, which is
:: because the associated route is a connected route, thus it uses the dst
addr of the packet. Let's use rt6_nexthop() to choose the right nh.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: constify rt6_nexthop()
Nicolas Dichtel [Mon, 24 Jun 2019 14:01:08 +0000 (16:01 +0200)]
ipv6: constify rt6_nexthop()

There is no functional change in this patch, it only prepares the next one.

rt6_nexthop() will be used by ip6_dst_lookup_neigh(), which uses const
variables.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Reported-by: kbuild test robot <lkp@intel.com>
Acked-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: dsa: microchip: Use gpiod_set_value_cansleep()
Marek Vasut [Sun, 23 Jun 2019 15:12:57 +0000 (17:12 +0200)]
net: dsa: microchip: Use gpiod_set_value_cansleep()

Replace gpiod_set_value() with gpiod_set_value_cansleep(), as the switch
reset GPIO can be connected to e.g. I2C GPIO expander and it is perfectly
fine for the kernel to sleep for a bit in ksz_switch_register().

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Andrew Lunn <andrew@lunn.ch>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Tristram Ha <Tristram.Ha@microchip.com>
Cc: Woojung Huh <Woojung.Huh@microchip.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: aquantia: fix vlans not working over bridged network
Dmitry Bogdanov [Sat, 22 Jun 2019 08:46:37 +0000 (08:46 +0000)]
net: aquantia: fix vlans not working over bridged network

In configuration of vlan over bridge over aquantia device
it was found that vlan tagged traffic is dropped on chip.

The reason is that bridge device enables promisc mode,
but in atlantic chip vlan filters will still apply.
So we have to corellate promisc settings with vlan configuration.

The solution is to track in a separate state variable the
need of vlan forced promisc. And also consider generic
promisc configuration when doing vlan filter config.

Fixes: 7975d2aff5af ("net: aquantia: add support of rx-vlan-filter offload")
Signed-off-by: Dmitry Bogdanov <dmitry.bogdanov@aquantia.com>
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: reset rt_iif for recirculated mcast/bcast out pkts
Stephen Suryaputra [Wed, 26 Jun 2019 06:21:16 +0000 (02:21 -0400)]
ipv4: reset rt_iif for recirculated mcast/bcast out pkts

Multicast or broadcast egress packets have rt_iif set to the oif. These
packets might be recirculated back as input and lookup to the raw
sockets may fail because they are bound to the incoming interface
(skb_iif). If rt_iif is not zero, during the lookup, inet_iif() function
returns rt_iif instead of skb_iif. Hence, the lookup fails.

v2: Make it non vrf specific (David Ahern). Reword the changelog to
    reflect it.
Signed-off-by: Stephen Suryaputra <ssuryaextr@gmail.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoteam: Always enable vlan tx offload
YueHaibing [Wed, 26 Jun 2019 16:03:39 +0000 (00:03 +0800)]
team: Always enable vlan tx offload

We should rather have vlan_tci filled all the way down
to the transmitting netdevice and let it do the hw/sw
vlan implementation.

Suggested-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'smc-fixes'
David S. Miller [Wed, 26 Jun 2019 17:10:16 +0000 (10:10 -0700)]
Merge branch 'smc-fixes'

Ursula Braun says:

====================
net/smc: fixes 2019-06-26

here are 2 small smc fixes for the net tree.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet/smc: Fix error path in smc_init
YueHaibing [Wed, 26 Jun 2019 15:47:50 +0000 (17:47 +0200)]
net/smc: Fix error path in smc_init

If register_pernet_subsys success in smc_init,
we should cleanup it in case any other error.

Fixes: 64e28b52c7a6 (net/smc: add pnet table namespace support")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>