vti6: better validate user provided tunnel names

[ Upstream commit 537b361fbcbcc3cd6fe2bb47069fd292b9256d16 ]

Use valid_name() to make sure user does not provide illegal
device name.

Fixes: ed1efb2aefbb ("ipv6: Add support for IPsec virtual tunnel interfaces")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ip6_tunnel: better validate user provided tunnel names

[ Upstream commit db7a65e3ab78e5b1c4b17c0870ebee35a4ee3257 ]

Use valid_name() to make sure user does not provide illegal
device name.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ip6_gre: better validate user provided tunnel names

[ Upstream commit 5f42df013b8bc1b6511af7a04bf93b014884ae2a ]

Use dev_valid_name() to make sure user does not provide illegal
device name.

syzbot caught the following bug :

BUG: KASAN: stack-out-of-bounds in strlcpy include/linux/string.h:300 [inline]
BUG: KASAN: stack-out-of-bounds in ip6gre_tunnel_locate+0x334/0x860 net/ipv6/ip6_gre.c:339
Write of size 20 at addr ffff8801afb9f7b8 by task syzkaller851048/4466

CPU: 1 PID: 4466 Comm: syzkaller851048 Not tainted 4.16.0+ #1
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x1b9/0x29f lib/dump_stack.c:53
 print_address_description+0x6c/0x20b mm/kasan/report.c:256
 kasan_report_error mm/kasan/report.c:354 [inline]
 kasan_report.cold.7+0xac/0x2f5 mm/kasan/report.c:412
 check_memory_region_inline mm/kasan/kasan.c:260 [inline]
 check_memory_region+0x13e/0x1b0 mm/kasan/kasan.c:267
 memcpy+0x37/0x50 mm/kasan/kasan.c:303
 strlcpy include/linux/string.h:300 [inline]
 ip6gre_tunnel_locate+0x334/0x860 net/ipv6/ip6_gre.c:339
 ip6gre_tunnel_ioctl+0x69d/0x12e0 net/ipv6/ip6_gre.c:1195
 dev_ifsioc+0x43e/0xb90 net/core/dev_ioctl.c:334
 dev_ioctl+0x69a/0xcc0 net/core/dev_ioctl.c:525
 sock_ioctl+0x47e/0x680 net/socket.c:1015
 vfs_ioctl fs/ioctl.c:46 [inline]
 file_ioctl fs/ioctl.c:500 [inline]
 do_vfs_ioctl+0x1cf/0x1650 fs/ioctl.c:684
 ksys_ioctl+0xa9/0xd0 fs/ioctl.c:701
 SYSC_ioctl fs/ioctl.c:708 [inline]
 SyS_ioctl+0x24/0x30 fs/ioctl.c:706
 do_syscall_64+0x29e/0x9d0 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x42/0xb7

Fixes: c12b395a4664 ("gre: Support GRE over IPv6")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ipv6: sit: better validate user provided tunnel names

[ Upstream commit b95211e066fc3494b7c115060b2297b4ba21f025 ]

Use dev_valid_name() to make sure user does not provide illegal
device name.

syzbot caught the following bug :

BUG: KASAN: stack-out-of-bounds in strlcpy include/linux/string.h:300 [inline]
BUG: KASAN: stack-out-of-bounds in ipip6_tunnel_locate+0x63b/0xaa0 net/ipv6/sit.c:254
Write of size 33 at addr ffff8801b64076d8 by task syzkaller932654/4453

CPU: 0 PID: 4453 Comm: syzkaller932654 Not tainted 4.16.0+ #1
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x1b9/0x29f lib/dump_stack.c:53
 print_address_description+0x6c/0x20b mm/kasan/report.c:256
 kasan_report_error mm/kasan/report.c:354 [inline]
 kasan_report.cold.7+0xac/0x2f5 mm/kasan/report.c:412
 check_memory_region_inline mm/kasan/kasan.c:260 [inline]
 check_memory_region+0x13e/0x1b0 mm/kasan/kasan.c:267
 memcpy+0x37/0x50 mm/kasan/kasan.c:303
 strlcpy include/linux/string.h:300 [inline]
 ipip6_tunnel_locate+0x63b/0xaa0 net/ipv6/sit.c:254
 ipip6_tunnel_ioctl+0xe71/0x241b net/ipv6/sit.c:1221
 dev_ifsioc+0x43e/0xb90 net/core/dev_ioctl.c:334
 dev_ioctl+0x69a/0xcc0 net/core/dev_ioctl.c:525
 sock_ioctl+0x47e/0x680 net/socket.c:1015
 vfs_ioctl fs/ioctl.c:46 [inline]
 file_ioctl fs/ioctl.c:500 [inline]
 do_vfs_ioctl+0x1cf/0x1650 fs/ioctl.c:684
 ksys_ioctl+0xa9/0xd0 fs/ioctl.c:701
 SYSC_ioctl fs/ioctl.c:708 [inline]
 SyS_ioctl+0x24/0x30 fs/ioctl.c:706
 do_syscall_64+0x29e/0x9d0 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x42/0xb7

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ip_tunnel: better validate user provided tunnel names

[ Upstream commit 9cb726a212a82c88c98aa9f0037fd04777cd8fe5 ]

Use dev_valid_name() to make sure user does not provide illegal
device name.

syzbot caught the following bug :

BUG: KASAN: stack-out-of-bounds in strlcpy include/linux/string.h:300 [inline]
BUG: KASAN: stack-out-of-bounds in __ip_tunnel_create+0xca/0x6b0 net/ipv4/ip_tunnel.c:257
Write of size 20 at addr ffff8801ac79f810 by task syzkaller268107/4482

CPU: 0 PID: 4482 Comm: syzkaller268107 Not tainted 4.16.0+ #1
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x1b9/0x29f lib/dump_stack.c:53
 print_address_description+0x6c/0x20b mm/kasan/report.c:256
 kasan_report_error mm/kasan/report.c:354 [inline]
 kasan_report.cold.7+0xac/0x2f5 mm/kasan/report.c:412
 check_memory_region_inline mm/kasan/kasan.c:260 [inline]
 check_memory_region+0x13e/0x1b0 mm/kasan/kasan.c:267
 memcpy+0x37/0x50 mm/kasan/kasan.c:303
 strlcpy include/linux/string.h:300 [inline]
 __ip_tunnel_create+0xca/0x6b0 net/ipv4/ip_tunnel.c:257
 ip_tunnel_create net/ipv4/ip_tunnel.c:352 [inline]
 ip_tunnel_ioctl+0x818/0xd40 net/ipv4/ip_tunnel.c:861
 ipip_tunnel_ioctl+0x1c5/0x420 net/ipv4/ipip.c:350
 dev_ifsioc+0x43e/0xb90 net/core/dev_ioctl.c:334
 dev_ioctl+0x69a/0xcc0 net/core/dev_ioctl.c:525
 sock_ioctl+0x47e/0x680 net/socket.c:1015
 vfs_ioctl fs/ioctl.c:46 [inline]
 file_ioctl fs/ioctl.c:500 [inline]
 do_vfs_ioctl+0x1cf/0x1650 fs/ioctl.c:684
 ksys_ioctl+0xa9/0xd0 fs/ioctl.c:701
 SYSC_ioctl fs/ioctl.c:708 [inline]
 SyS_ioctl+0x24/0x30 fs/ioctl.c:706
 do_syscall_64+0x29e/0x9d0 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x42/0xb7

Fixes: c54419321455 ("GRE: Refactor GRE tunneling code.")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: fool proof dev_valid_name()

[ Upstream commit a9d48205d0aedda021fc3728972a9e9934c2b9de ]

We want to use dev_valid_name() to validate tunnel names,
so better use strnlen(name, IFNAMSIZ) than strlen(name) to make
sure to not upset KASAN.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bonding: process the err returned by dev_set_allmulti properly in bond_enslave

[ Upstream commit 9f5a90c107741b864398f4ac0014711a8c1d8474 ]

When dev_set_promiscuity(1) succeeds but dev_set_allmulti(1) fails,
dev_set_promiscuity(-1) should be done before going to the err path.
Otherwise, dev->promiscuity will leak.

Fixes: 7e1a1ac1fbaa ("bonding: Check return of dev_set_promiscuity/allmulti")
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bonding: move dev_mc_sync after master_upper_dev_link in bond_enslave

[ Upstream commit ae42cc62a9f07f1f6979054ed92606b9c30f4a2e ]

Beniamino found a crash when adding vlan as slave of bond which is also
the parent link:

  ip link add bond1 type bond
  ip link set bond1 up
  ip link add link bond1 vlan1 type vlan id 80
  ip link set vlan1 master bond1

The call trace is as below:

  [<ffffffffa850842a>] queued_spin_lock_slowpath+0xb/0xf
  [<ffffffffa8515680>] _raw_spin_lock+0x20/0x30
  [<ffffffffa83f6f07>] dev_mc_sync+0x37/0x80
  [<ffffffffc08687dc>] vlan_dev_set_rx_mode+0x1c/0x30 [8021q]
  [<ffffffffa83efd2a>] __dev_set_rx_mode+0x5a/0xa0
  [<ffffffffa83f7138>] dev_mc_sync_multiple+0x78/0x80
  [<ffffffffc084127c>] bond_enslave+0x67c/0x1190 [bonding]
  [<ffffffffa8401909>] do_setlink+0x9c9/0xe50
  [<ffffffffa8403bf2>] rtnl_newlink+0x522/0x880
  [<ffffffffa8403ff7>] rtnetlink_rcv_msg+0xa7/0x260
  [<ffffffffa8424ecb>] netlink_rcv_skb+0xab/0xc0
  [<ffffffffa83fe498>] rtnetlink_rcv+0x28/0x30
  [<ffffffffa8424850>] netlink_unicast+0x170/0x210
  [<ffffffffa8424bf8>] netlink_sendmsg+0x308/0x420
  [<ffffffffa83cc396>] sock_sendmsg+0xb6/0xf0

This is actually a dead lock caused by sync slave hwaddr from master when
the master is the slave's 'slave'. This dead loop check is actually done
by netdev_master_upper_dev_link. However, Commit 1f718f0f4f97 ("bonding:
populate neighbour's private on enslave") moved it after dev_mc_sync.

This patch is to fix it by moving dev_mc_sync after master_upper_dev_link,
so that this loop check would be earlier than dev_mc_sync. It also moves
if (mode == BOND_MODE_8023AD) into if (!bond_uses_primary) clause as an
improvement.

Note team driver also has this issue, I will fix it in another patch.

Fixes: 1f718f0f4f97 ("bonding: populate neighbour's private on enslave")
Reported-by: Beniamino Galvani <bgalvani@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bonding: fix the err path for dev hwaddr sync in bond_enslave

[ Upstream commit 5c78f6bfae2b10ff70e21d343e64584ea6280c26 ]

vlan_vids_add_by_dev is called right after dev hwaddr sync, so on
the err path it should unsync dev hwaddr. Otherwise, the slave
dev's hwaddr will never be unsync when this err happens.

Fixes: 1ff412ad7714 ("bonding: change the bond's vlan syncing functions with the standard ones")
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Reviewed-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Acked-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

vrf: Fix use after free and double free in vrf_finish_output

[ Upstream commit 82dd0d2a9a76fc8fa2b18d80b987d455728bf83a ]

Miguel reported an skb use after free / double free in vrf_finish_output
when neigh_output returns an error. The vrf driver should return after
the call to neigh_output as it takes over the skb on error path as well.

Patch is a simplified version of Miguel's patch which was written for 4.9,
and updated to top of tree.

Fixes: 8f58336d3f78a ("net: Add ethernet header for pass through VRF device")
Signed-off-by: Miguel Fadon Perlines <mfadon@teldat.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

vlan: also check phy_driver ts_info for vlan's real device

[ Upstream commit ec1d8ccb07deaf30fd0508af6755364ac47dc08d ]

Just like function ethtool_get_ts_info(), we should also consider the
phy_driver ts_info call back. For example, driver dp83640.

Fixes: 37dd9255b2f6 ("vlan: Pass ethtool get_ts_info queries to real device.")
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

vhost: correctly remove wait queue during poll failure

[ Upstream commit dc6455a71c7fc5117977e197f67f71b49f27baba ]

We tried to remove vq poll from wait queue, but do not check whether
or not it was in a list before. This will lead double free. Fixing
this by switching to use vhost_poll_stop() which zeros poll->wqh after
removing poll from waitqueue to make sure it won't be freed twice.

Cc: Darren Kenny <darren.kenny@oracle.com>
Reported-by: syzbot+c0272972b01b872e604a@syzkaller.appspotmail.com
Fixes: 2b8b328b61c79 ("vhost_net: handle polling errors when setting backend")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

sky2: Increase D3 delay to sky2 stops working after suspend

[ Upstream commit afb133637071be6deeb8b3d0e55593ffbf63c527 ]

The sky2 ethernet stops working after system resume from suspend:
[ 582.852065] sky2 0000:04:00.0: Refused to change power state, currently in D3

The current 150ms delay is not enough, change it to 200ms can solve the
issue.

BugLink: https://bugs.launchpad.net/bugs/1758507
Cc: Stable <stable@vger.kernel.org>
Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

sctp: sctp_sockaddr_af must check minimal addr length for AF_INET6

[ Upstream commit 81e98370293afcb58340ce8bd71af7b97f925c26 ]

Check must happen before call to ipv6_addr_v4mapped()

syzbot report was :

BUG: KMSAN: uninit-value in sctp_sockaddr_af net/sctp/socket.c:359 [inline]
BUG: KMSAN: uninit-value in sctp_do_bind+0x60f/0xdc0 net/sctp/socket.c:384
CPU: 0 PID: 3576 Comm: syzkaller968804 Not tainted 4.16.0+ #82
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x185/0x1d0 lib/dump_stack.c:53
 kmsan_report+0x142/0x240 mm/kmsan/kmsan.c:1067
 __msan_warning_32+0x6c/0xb0 mm/kmsan/kmsan_instr.c:676
 sctp_sockaddr_af net/sctp/socket.c:359 [inline]
 sctp_do_bind+0x60f/0xdc0 net/sctp/socket.c:384
 sctp_bind+0x149/0x190 net/sctp/socket.c:332
 inet6_bind+0x1fd/0x1820 net/ipv6/af_inet6.c:293
 SYSC_bind+0x3f2/0x4b0 net/socket.c:1474
 SyS_bind+0x54/0x80 net/socket.c:1460
 do_syscall_64+0x309/0x430 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x3d/0xa2
RIP: 0033:0x43fd49
RSP: 002b:00007ffe99df3d28 EFLAGS: 00000213 ORIG_RAX: 0000000000000031
RAX: ffffffffffffffda RBX: 00000000004002c8 RCX: 000000000043fd49
RDX: 0000000000000010 RSI: 0000000020000000 RDI: 0000000000000003
RBP: 00000000006ca018 R08: 00000000004002c8 R09: 00000000004002c8
R10: 00000000004002c8 R11: 0000000000000213 R12: 0000000000401670
R13: 0000000000401700 R14: 0000000000000000 R15: 0000000000000000

Local variable description: ----address@SYSC_bind
Variable was created at:
 SYSC_bind+0x6f/0x4b0 net/socket.c:1461
 SyS_bind+0x54/0x80 net/socket.c:1460

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Vlad Yasevich <vyasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

sctp: do not leak kernel memory to user space

[ Upstream commit 6780db244d6b1537d139dea0ec8aad10cf9e4adb ]

syzbot produced a nice report [1]

Issue here is that a recvmmsg() managed to leak 8 bytes of kernel memory
to user space, because sin_zero (padding field) was not properly cleared.

[1]
BUG: KMSAN: uninit-value in copy_to_user include/linux/uaccess.h:184 [inline]
BUG: KMSAN: uninit-value in move_addr_to_user+0x32e/0x530 net/socket.c:227
CPU: 1 PID: 3586 Comm: syzkaller481044 Not tainted 4.16.0+ #82
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x185/0x1d0 lib/dump_stack.c:53
 kmsan_report+0x142/0x240 mm/kmsan/kmsan.c:1067
 kmsan_internal_check_memory+0x164/0x1d0 mm/kmsan/kmsan.c:1176
 kmsan_copy_to_user+0x69/0x160 mm/kmsan/kmsan.c:1199
 copy_to_user include/linux/uaccess.h:184 [inline]
 move_addr_to_user+0x32e/0x530 net/socket.c:227
 ___sys_recvmsg+0x4e2/0x810 net/socket.c:2211
 __sys_recvmmsg+0x54e/0xdb0 net/socket.c:2313
 SYSC_recvmmsg+0x29b/0x3e0 net/socket.c:2394
 SyS_recvmmsg+0x76/0xa0 net/socket.c:2378
 do_syscall_64+0x309/0x430 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x3d/0xa2
RIP: 0033:0x4401c9
RSP: 002b:00007ffc56f73098 EFLAGS: 00000217 ORIG_RAX: 000000000000012b
RAX: ffffffffffffffda RBX: 00000000004002c8 RCX: 00000000004401c9
RDX: 0000000000000001 RSI: 0000000020003ac0 RDI: 0000000000000003
RBP: 00000000006ca018 R08: 0000000020003bc0 R09: 0000000000000010
R10: 0000000000000000 R11: 0000000000000217 R12: 0000000000401af0
R13: 0000000000401b80 R14: 0000000000000000 R15: 0000000000000000

Local variable description: ----addr@___sys_recvmsg
Variable was created at:
 ___sys_recvmsg+0xd5/0x810 net/socket.c:2172
 __sys_recvmmsg+0x54e/0xdb0 net/socket.c:2313

Bytes 8-15 of 16 are uninitialized

==================================================================
Kernel panic - not syncing: panic_on_warn set ...

CPU: 1 PID: 3586 Comm: syzkaller481044 Tainted: G    B            4.16.0+ #82
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x185/0x1d0 lib/dump_stack.c:53
 panic+0x39d/0x940 kernel/panic.c:183
 kmsan_report+0x238/0x240 mm/kmsan/kmsan.c:1083
 kmsan_internal_check_memory+0x164/0x1d0 mm/kmsan/kmsan.c:1176
 kmsan_copy_to_user+0x69/0x160 mm/kmsan/kmsan.c:1199
 copy_to_user include/linux/uaccess.h:184 [inline]
 move_addr_to_user+0x32e/0x530 net/socket.c:227
 ___sys_recvmsg+0x4e2/0x810 net/socket.c:2211
 __sys_recvmmsg+0x54e/0xdb0 net/socket.c:2313
 SYSC_recvmmsg+0x29b/0x3e0 net/socket.c:2394
 SyS_recvmmsg+0x76/0xa0 net/socket.c:2378
 do_syscall_64+0x309/0x430 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x3d/0xa2

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Vlad Yasevich <vyasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

r8169: fix setting driver_data after register_netdev

[ Upstream commit 19c9ea363a244f85f90a424f9936e6d56449e33c ]

pci_set_drvdata() is called only after registering the net_device,
therefore we could run into a NPE if one of the functions using
driver_data is called before it's set.

Fix this by calling pci_set_drvdata() before registering the
net_device.

This fix is a candidate for stable. As far as I can see the
bug has been there in kernel version 3.2 already, therefore
I can't provide a reference which commit is fixed by it.

The fix may need small adjustments per kernel version because
due to other changes the label which is jumped to if
register_netdev() fails has changed over time.

Reported-by: David Miller <davem@davemloft.net>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

pptp: remove a buggy dst release in pptp_connect()

[ Upstream commit bfacfb457b36911a10140b8cb3ce76a74883ac5a ]

Once dst has been cached in socket via sk_setup_caps(),
it is illegal to call ip_rt_put() (or dst_release()),
since sk_setup_caps() did not change dst refcount.

We can still dereference it since we hold socket lock.

Caugth by syzbot :

BUG: KASAN: use-after-free in atomic_dec_return include/asm-generic/atomic-instrumented.h:198 [inline]
BUG: KASAN: use-after-free in dst_release+0x27/0xa0 net/core/dst.c:185
Write of size 4 at addr ffff8801c54dc040 by task syz-executor4/20088

CPU: 1 PID: 20088 Comm: syz-executor4 Not tainted 4.16.0+ #376
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x1a7/0x27d lib/dump_stack.c:53
 print_address_description+0x73/0x250 mm/kasan/report.c:256
 kasan_report_error mm/kasan/report.c:354 [inline]
 kasan_report+0x23c/0x360 mm/kasan/report.c:412
 check_memory_region_inline mm/kasan/kasan.c:260 [inline]
 check_memory_region+0x137/0x190 mm/kasan/kasan.c:267
 kasan_check_write+0x14/0x20 mm/kasan/kasan.c:278
 atomic_dec_return include/asm-generic/atomic-instrumented.h:198 [inline]
 dst_release+0x27/0xa0 net/core/dst.c:185
 sk_dst_set include/net/sock.h:1812 [inline]
 sk_dst_reset include/net/sock.h:1824 [inline]
 sock_setbindtodevice net/core/sock.c:610 [inline]
 sock_setsockopt+0x431/0x1b20 net/core/sock.c:707
 SYSC_setsockopt net/socket.c:1845 [inline]
 SyS_setsockopt+0x2ff/0x360 net/socket.c:1828
 do_syscall_64+0x281/0x940 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x42/0xb7
RIP: 0033:0x4552d9
RSP: 002b:00007f4878126c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000036
RAX: ffffffffffffffda RBX: 00007f48781276d4 RCX: 00000000004552d9
RDX: 0000000000000019 RSI: 0000000000000001 RDI: 0000000000000013
RBP: 000000000072bea0 R08: 0000000000000010 R09: 0000000000000000
R10: 00000000200010c0 R11: 0000000000000246 R12: 00000000ffffffff
R13: 0000000000000526 R14: 00000000006fac30 R15: 0000000000000000

Allocated by task 20088:
 save_stack+0x43/0xd0 mm/kasan/kasan.c:447
 set_track mm/kasan/kasan.c:459 [inline]
 kasan_kmalloc+0xad/0xe0 mm/kasan/kasan.c:552
 kasan_slab_alloc+0x12/0x20 mm/kasan/kasan.c:489
 kmem_cache_alloc+0x12e/0x760 mm/slab.c:3542
 dst_alloc+0x11f/0x1a0 net/core/dst.c:104
 rt_dst_alloc+0xe9/0x540 net/ipv4/route.c:1520
 __mkroute_output net/ipv4/route.c:2265 [inline]
 ip_route_output_key_hash_rcu+0xa49/0x2c60 net/ipv4/route.c:2493
 ip_route_output_key_hash+0x20b/0x370 net/ipv4/route.c:2322
 __ip_route_output_key include/net/route.h:126 [inline]
 ip_route_output_flow+0x26/0xa0 net/ipv4/route.c:2577
 ip_route_output_ports include/net/route.h:163 [inline]
 pptp_connect+0xa84/0x1170 drivers/net/ppp/pptp.c:453
 SYSC_connect+0x213/0x4a0 net/socket.c:1639
 SyS_connect+0x24/0x30 net/socket.c:1620
 do_syscall_64+0x281/0x940 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x42/0xb7

Freed by task 20082:
 save_stack+0x43/0xd0 mm/kasan/kasan.c:447
 set_track mm/kasan/kasan.c:459 [inline]
 __kasan_slab_free+0x11a/0x170 mm/kasan/kasan.c:520
 kasan_slab_free+0xe/0x10 mm/kasan/kasan.c:527
 __cache_free mm/slab.c:3486 [inline]
 kmem_cache_free+0x83/0x2a0 mm/slab.c:3744
 dst_destroy+0x266/0x380 net/core/dst.c:140
 dst_destroy_rcu+0x16/0x20 net/core/dst.c:153
 __rcu_reclaim kernel/rcu/rcu.h:178 [inline]
 rcu_do_batch kernel/rcu/tree.c:2675 [inline]
 invoke_rcu_callbacks kernel/rcu/tree.c:2930 [inline]
 __rcu_process_callbacks kernel/rcu/tree.c:2897 [inline]
 rcu_process_callbacks+0xd6c/0x17b0 kernel/rcu/tree.c:2914
 __do_softirq+0x2d7/0xb85 kernel/softirq.c:285

The buggy address belongs to the object at ffff8801c54dc000
 which belongs to the cache ip_dst_cache of size 168
The buggy address is located 64 bytes inside of
 168-byte region [ffff8801c54dc000, ffff8801c54dc0a8)
The buggy address belongs to the page:
page:ffffea0007153700 count:1 mapcount:0 mapping:ffff8801c54dc000 index:0x0
flags: 0x2fffc0000000100(slab)
raw: 02fffc0000000100 ffff8801c54dc000 0000000000000000 0000000100000010
raw: ffffea0006b34b20 ffffea0006b6c1e0 ffff8801d674a1c0 0000000000000000
page dumped because: kasan: bad access detected

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net/sched: fix NULL dereference in the error path of tcf_bpf_init()

[ Upstream commit 3239534a79ee6f20cffd974173a1e62e0730e8ac ]

when tcf_bpf_init_from_ops() fails (e.g. because of program having invalid
number of instructions), tcf_bpf_cfg_cleanup() calls bpf_prog_put(NULL) or
bpf_prog_destroy(NULL). Unless CONFIG_BPF_SYSCALL is unset, this causes
the following error:

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000020
 PGD 800000007345a067 P4D 800000007345a067 PUD 340e1067 PMD 0
 Oops: 0000 [#1] SMP PTI
 Modules linked in: act_bpf(E) ip6table_filter ip6_tables iptable_filter binfmt_misc ext4 mbcache jbd2 crct10dif_pclmul crc32_pclmul ghash_clmulni_intel snd_hda_codec_generic pcbc snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm aesni_intel crypto_simd glue_helper cryptd joydev snd_timer snd virtio_balloon pcspkr soundcore i2c_piix4 nfsd auth_rpcgss nfs_acl lockd grace sunrpc ip_tables xfs libcrc32c ata_generic pata_acpi qxl drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops ttm virtio_blk drm virtio_net virtio_console i2c_core crc32c_intel serio_raw virtio_pci ata_piix libata virtio_ring floppy virtio dm_mirror dm_region_hash dm_log dm_mod [last unloaded: act_bpf]
 CPU: 3 PID: 5654 Comm: tc Tainted: G            E    4.16.0.bpf_test+ #408
 Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
 RIP: 0010:__bpf_prog_put+0xc/0xc0
 RSP: 0018:ffff9594003ef728 EFLAGS: 00010202
 RAX: 0000000000000000 RBX: ffff9594003ef758 RCX: 0000000000000024
 RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000000
 RBP: 0000000000000000 R08: 0000000000000001 R09: 0000000000000044
 R10: 0000000000000220 R11: ffff8a7ab9f17131 R12: 0000000000000000
 R13: ffff8a7ab7c3c8e0 R14: 0000000000000001 R15: ffff8a7ab88f1054
 FS:  00007fcb2f17c740(0000) GS:ffff8a7abfd80000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000000000000020 CR3: 000000007c888006 CR4: 00000000001606e0
 Call Trace:
  tcf_bpf_cfg_cleanup+0x2f/0x40 [act_bpf]
  tcf_bpf_cleanup+0x4c/0x70 [act_bpf]
  __tcf_idr_release+0x79/0x140
  tcf_bpf_init+0x125/0x330 [act_bpf]
  tcf_action_init_1+0x2cc/0x430
  ? get_page_from_freelist+0x3f0/0x11b0
  tcf_action_init+0xd3/0x1b0
  tc_ctl_action+0x18b/0x240
  rtnetlink_rcv_msg+0x29c/0x310
  ? _cond_resched+0x15/0x30
  ? __kmalloc_node_track_caller+0x1b9/0x270
  ? rtnl_calcit.isra.29+0x100/0x100
  netlink_rcv_skb+0xd2/0x110
  netlink_unicast+0x17c/0x230
  netlink_sendmsg+0x2cd/0x3c0
  sock_sendmsg+0x30/0x40
  ___sys_sendmsg+0x27a/0x290
  ? mem_cgroup_commit_charge+0x80/0x130
  ? page_add_new_anon_rmap+0x73/0xc0
  ? do_anonymous_page+0x2a2/0x560
  ? __handle_mm_fault+0xc75/0xe20
  __sys_sendmsg+0x58/0xa0
  do_syscall_64+0x6e/0x1a0
  entry_SYSCALL_64_after_hwframe+0x3d/0xa2
 RIP: 0033:0x7fcb2e58eba0
 RSP: 002b:00007ffc93c496c8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
 RAX: ffffffffffffffda RBX: 00007ffc93c497f0 RCX: 00007fcb2e58eba0
 RDX: 0000000000000000 RSI: 00007ffc93c49740 RDI: 0000000000000003
 RBP: 000000005ac6a646 R08: 0000000000000002 R09: 0000000000000000
 R10: 00007ffc93c49120 R11: 0000000000000246 R12: 0000000000000000
 R13: 00007ffc93c49804 R14: 0000000000000001 R15: 000000000066afa0
 Code: 5f 00 48 8b 43 20 48 c7 c7 70 2f 7c b8 c7 40 10 00 00 00 00 5b e9 a5 8b 61 00 0f 1f 44 00 00 0f 1f 44 00 00 41 54 55 48 89 fd 53 <48> 8b 47 20 f0 ff 08 74 05 5b 5d 41 5c c3 41 89 f4 0f 1f 44 00
 RIP: __bpf_prog_put+0xc/0xc0 RSP: ffff9594003ef728
 CR2: 0000000000000020

Fix it in tcf_bpf_cfg_cleanup(), ensuring that bpf_prog_{put,destroy}(f)
is called only when f is not NULL.

Fixes: bbc09e7842a5 ("net/sched: fix idr leak on the error path of tcf_bpf_init()")
Reported-by: Lucas Bates <lucasb@mojatatu.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net sched actions: fix dumping which requires several messages to user space

[ Upstream commit 734549eb550c0c720bc89e50501f1b1e98cdd841 ]

Fixes a bug in the tcf_dump_walker function that can cause some actions
to not be reported when dumping a large number of actions. This issue
became more aggrevated when cookies feature was added. In particular
this issue is manifest when large cookie values are assigned to the
actions and when enough actions are created that the resulting table
must be dumped in multiple batches.

The number of actions returned in each batch is limited by the total
number of actions and the memory buffer size.  With small cookies
the numeric limit is reached before the buffer size limit, which avoids
the code path triggering this bug. When large cookies are used buffer
fills before the numeric limit, and the erroneous code path is hit.

For example after creating 32 csum actions with the cookie
aaaabbbbccccdddd

$ tc actions ls action csum
total acts 26

    action order 0: csum (tcp) action continue
    index 1 ref 1 bind 0
    cookie aaaabbbbccccdddd

    .....

    action order 25: csum (tcp) action continue
    index 26 ref 1 bind 0
    cookie aaaabbbbccccdddd
total acts 6

    action order 0: csum (tcp) action continue
    index 28 ref 1 bind 0
    cookie aaaabbbbccccdddd

    ......

    action order 5: csum (tcp) action continue
    index 32 ref 1 bind 0
    cookie aaaabbbbccccdddd

Note that the action with index 27 is omitted from the report.

Fixes: 4b3550ef530c ("[NET_SCHED]: Use nla_nest_start/nla_nest_end")"
Signed-off-by: Craig Dillabaugh <cdillaba@mojatatu.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

netlink: make sure nladdr has correct size in netlink_connect()

[ Upstream commit 7880287981b60a6808f39f297bb66936e8bdf57a ]

KMSAN reports use of uninitialized memory in the case when |alen| is
smaller than sizeof(struct sockaddr_nl), and therefore |nladdr| isn't
fully copied from the userspace.

Signed-off-by: Alexander Potapenko <glider@google.com>
Fixes: 1da177e4c3f41524 ("Linux-2.6.12-rc2")
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net/ipv6: Increment OUTxxx counters after netfilter hook

[ Upstream commit 71a1c915238c970cd9bdd5bf158b1279d6b6d55b ]

At the end of ip6_forward(), IPSTATS_MIB_OUTFORWDATAGRAMS and
IPSTATS_MIB_OUTOCTETS are incremented immediately before the NF_HOOK call
for NFPROTO_IPV6 / NF_INET_FORWARD.  As a result, these counters get
incremented regardless of whether or not the netfilter hook allows the
packet to continue being processed.  This change increments the counters
in ip6_forward_finish() so that it will not happen if the netfilter hook
chooses to terminate the packet, which is similar to how IPv4 works.

Signed-off-by: Jeff Barnhill <0xeffeff@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net/ipv6: Fix route leaking between VRFs

[ Upstream commit b6cdbc85234b072340b8923e69f49ec293f905dc ]

Donald reported that IPv6 route leaking between VRFs is not working.
The root cause is the strict argument in the call to rt6_lookup when
validating the nexthop spec.

ip6_route_check_nh validates the gateway and device (if given) of a
route spec. It in turn could call rt6_lookup (e.g., lookup in a given
table did not succeed so it falls back to a full lookup) and if so
sets the strict argument to 1. That means if the egress device is given,
the route lookup needs to return a result with the same device. This
strict requirement does not work with VRFs (IPv4 or IPv6) because the
oif in the flow struct is overridden with the index of the VRF device
to trigger a match on the l3mdev rule and force the lookup to its table.

The right long term solution is to add an l3mdev index to the flow
struct such that the oif is not overridden. That solution will not
backport well, so this patch aims for a simpler solution to relax the
strict argument if the route spec device is an l3mdev slave. As done
in other places, use the FLOWI_FLAG_SKIP_NH_OIF to know that the
RT6_LOOKUP_F_IFACE flag needs to be removed.

Fixes: ca254490c8df ("net: Add VRF support to IPv6 stack")
Reported-by: Donald Sharp <sharpd@cumulusnetworks.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: fix possible out-of-bound read in skb_network_protocol()

[ Upstream commit 1dfe82ebd7d8fd43dba9948fdfb31f145014baa0 ]

skb mac header is not necessarily set at the time skb_network_protocol()
is called. Use skb->data instead.

BUG: KASAN: slab-out-of-bounds in skb_network_protocol+0x46b/0x4b0 net/core/dev.c:2739
Read of size 2 at addr ffff8801b3097a0b by task syz-executor5/14242

CPU: 1 PID: 14242 Comm: syz-executor5 Not tainted 4.16.0-rc6+ #280
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x194/0x24d lib/dump_stack.c:53
 print_address_description+0x73/0x250 mm/kasan/report.c:256
 kasan_report_error mm/kasan/report.c:354 [inline]
 kasan_report+0x23c/0x360 mm/kasan/report.c:412
 __asan_report_load_n_noabort+0xf/0x20 mm/kasan/report.c:443
 skb_network_protocol+0x46b/0x4b0 net/core/dev.c:2739
 harmonize_features net/core/dev.c:2924 [inline]
 netif_skb_features+0x509/0x9b0 net/core/dev.c:3011
 validate_xmit_skb+0x81/0xb00 net/core/dev.c:3084
 validate_xmit_skb_list+0xbf/0x120 net/core/dev.c:3142
 packet_direct_xmit+0x117/0x790 net/packet/af_packet.c:256
 packet_snd net/packet/af_packet.c:2944 [inline]
 packet_sendmsg+0x3aed/0x60b0 net/packet/af_packet.c:2969
 sock_sendmsg_nosec net/socket.c:629 [inline]
 sock_sendmsg+0xca/0x110 net/socket.c:639
 ___sys_sendmsg+0x767/0x8b0 net/socket.c:2047
 __sys_sendmsg+0xe5/0x210 net/socket.c:2081

Fixes: 19acc327258a ("gso: Handle Trans-Ether-Bridging protocol in skb_network_protocol()")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Pravin B Shelar <pshelar@ovn.org>
Reported-by: Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

lan78xx: Crash in lan78xx_writ_reg (Workqueue: events lan78xx_deferred_multicast_write)

[ Upstream commit 2d2d99ec13f62d5d2cecb6169dfdb6bbe05356d0 ]

Description:
Crash was reported with syzkaller pointing to lan78xx_write_reg routine.

Root-cause:
Proper cleanup of workqueues and init/setup routines was not happening
in failure conditions.

Fix:
Handled the error conditions by cleaning up the queues and init/setup
routines.

Fixes: 55d7de9de6c3 ("Microchip's LAN7800 family USB 2/3 to 10/100/1000 Ethernet device driver")
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: Raghuram Chary J <raghuramchary.jallipalli@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ipv6: the entire IPv6 header chain must fit the first fragment

[ Upstream commit 10b8a3de603df7b96004179b1b33b1708c76d144 ]

While building ipv6 datagram we currently allow arbitrary large
extheaders, even beyond pmtu size. The syzbot has found a way
to exploit the above to trigger the following splat:

kernel BUG at ./include/linux/skbuff.h:2073!
invalid opcode: 0000 [#1] SMP KASAN
Dumping ftrace buffer:
    (ftrace buffer empty)
Modules linked in:
CPU: 1 PID: 4230 Comm: syzkaller672661 Not tainted 4.16.0-rc2+ #326
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
RIP: 0010:__skb_pull include/linux/skbuff.h:2073 [inline]
RIP: 0010:__ip6_make_skb+0x1ac8/0x2190 net/ipv6/ip6_output.c:1636
RSP: 0018:ffff8801bc18f0f0 EFLAGS: 00010293
RAX: ffff8801b17400c0 RBX: 0000000000000738 RCX: ffffffff84f01828
RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff8801b415ac18
RBP: ffff8801bc18f360 R08: ffff8801b4576844 R09: 0000000000000000
R10: ffff8801bc18f380 R11: ffffed00367aee4e R12: 00000000000000d6
R13: ffff8801b415a740 R14: dffffc0000000000 R15: ffff8801b45767c0
FS:  0000000001535880(0000) GS:ffff8801db300000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000000002000b000 CR3: 00000001b4123001 CR4: 00000000001606e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
  ip6_finish_skb include/net/ipv6.h:969 [inline]
  udp_v6_push_pending_frames+0x269/0x3b0 net/ipv6/udp.c:1073
  udpv6_sendmsg+0x2a96/0x3400 net/ipv6/udp.c:1343
  inet_sendmsg+0x11f/0x5e0 net/ipv4/af_inet.c:764
  sock_sendmsg_nosec net/socket.c:630 [inline]
  sock_sendmsg+0xca/0x110 net/socket.c:640
  ___sys_sendmsg+0x320/0x8b0 net/socket.c:2046
  __sys_sendmmsg+0x1ee/0x620 net/socket.c:2136
  SYSC_sendmmsg net/socket.c:2167 [inline]
  SyS_sendmmsg+0x35/0x60 net/socket.c:2162
  do_syscall_64+0x280/0x940 arch/x86/entry/common.c:287
  entry_SYSCALL_64_after_hwframe+0x42/0xb7
RIP: 0033:0x4404c9
RSP: 002b:00007ffdce35f948 EFLAGS: 00000217 ORIG_RAX: 0000000000000133
RAX: ffffffffffffffda RBX: 00000000004002c8 RCX: 00000000004404c9
RDX: 0000000000000003 RSI: 0000000020001f00 RDI: 0000000000000003
RBP: 00000000006cb018 R08: 00000000004002c8 R09: 00000000004002c8
R10: 0000000020000080 R11: 0000000000000217 R12: 0000000000401df0
R13: 0000000000401e80 R14: 0000000000000000 R15: 0000000000000000
Code: ff e8 1d 5e b9 fc e9 15 e9 ff ff e8 13 5e b9 fc e9 44 e8 ff ff e8 29
5e b9 fc e9 c0 e6 ff ff e8 3f f3 80 fc 0f 0b e8 38 f3 80 fc <0f> 0b 49 8d
87 80 00 00 00 4d 8d 87 84 00 00 00 48 89 85 20 fe
RIP: __skb_pull include/linux/skbuff.h:2073 [inline] RSP: ffff8801bc18f0f0
RIP: __ip6_make_skb+0x1ac8/0x2190 net/ipv6/ip6_output.c:1636 RSP:
ffff8801bc18f0f0

As stated by RFC 7112 section 5:

   When a host fragments an IPv6 datagram, it MUST include the entire
   IPv6 Header Chain in the First Fragment.

So this patch addresses the issue dropping datagrams with excessive
extheader length. It also updates the error path to report to the
calling socket nonnegative pmtu values.

The issue apparently predates git history.

v1 -> v2: cleanup error path, as per Eric's suggestion

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reported-by: syzbot+91e6f9932ff122fa4410@syzkaller.appspotmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

arp: fix arp_filter on l3slave devices

[ Upstream commit 58b35f27689b5eb514fc293c332966c226b1b6e4 ]

arp_filter performs an ip_route_output search for arp source address and
checks if output device is the same where the arp request was received,
if it is not, the arp request is not answered.

This route lookup is always done on main route table so l3slave devices
never find the proper route and arp is not answered.

Passing l3mdev_master_ifindex_rcu(dev) return value as oif fixes the
lookup for l3slave devices while maintaining same behavior for non
l3slave devices as this function returns 0 in that case.

Fixes: 613d09b30f8b ("net: Use VRF device index for lookups on TX")
Signed-off-by: Miguel Fadon Perlines <mfadon@teldat.com>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode: Fix CPU synchronization routine

commit bb8c13d61a629276a162c1d2b1a20a815cbcfbb7 upstream.

Emanuel reported an issue with a hang during microcode update because my
dumb idea to use one atomic synchronization variable for both rendezvous
- before and after update - was simply bollocks:

  microcode: microcode_reload_late: late_cpus: 4
  microcode: __reload_late: cpu 2 entered
  microcode: __reload_late: cpu 1 entered
  microcode: __reload_late: cpu 3 entered
  microcode: __reload_late: cpu 0 entered
  microcode: __reload_late: cpu 1 left
  microcode: Timeout while waiting for CPUs rendezvous, remaining: 1

CPU1 above would finish, leave and the others will still spin waiting for
it to join.

So do two synchronization atomics instead, which makes the code a lot more
straightforward.

Also, since the update is serialized and it also takes quite some time per
microcode engine, increase the exit timeout by the number of CPUs on the
system.

That's ok because the moment all CPUs are done, that timeout will be cut
short.

Furthermore, panic when some of the CPUs timeout when returning from a
microcode update: we can't allow a system with not all cores updated.

Also, as an optimization, do not do the exit sync if microcode wasn't
updated.

Reported-by: Emanuel Czirai <xftroxgpx@protonmail.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Emanuel Czirai <xftroxgpx@protonmail.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Link: https://lkml.kernel.org/r/20180314183615.17629-2-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode: Attempt late loading only when new microcode is present

commit 2613f36ed965d0e5a595a1d931fd3b480e82d6fd upstream.

Return UCODE_NEW from the scanning functions to denote that new microcode
was found and only then attempt the expensive synchronization dance.

Reported-by: Emanuel Czirai <xftroxgpx@protonmail.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Emanuel Czirai <xftroxgpx@protonmail.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Link: https://lkml.kernel.org/r/20180314183615.17629-1-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode: Synchronize late microcode loading

commit a5321aec6412b20b5ad15db2d6b916c05349dbff upstream.

Original idea by Ashok, completely rewritten by Borislav.

Before you read any further: the early loading method is still the
preferred one and you should always do that. The following patch is
improving the late loading mechanism for long running jobs and cloud use
cases.

Gather all cores and serialize the microcode update on them by doing it
one-by-one to make the late update process as reliable as possible and
avoid potential issues caused by the microcode update.

[ Borislav: Rewrite completely. ]

Co-developed-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Arjan Van De Ven <arjan.van.de.ven@intel.com>
Link: https://lkml.kernel.org/r/20180228102846.13447-8-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode: Request microcode on the BSP

commit cfb52a5a09c8ae3a1dafb44ce549fde5b69e8117 upstream.

... so that any newer version can land in the cache and can later be
fished out by the application functions. Do that before grabbing the
hotplug lock.

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Arjan Van De Ven <arjan.van.de.ven@intel.com>
Link: https://lkml.kernel.org/r/20180228102846.13447-7-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode/intel: Look into the patch cache first

commit d8c3b52c00a05036e0a6b315b4b17921a7b67997 upstream.

The cache might contain a newer patch - look in there first.

A follow-on change will make sure newest patches are loaded into the
cache of microcode patches.

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Cc: Arjan Van De Ven <arjan.van.de.ven@intel.com>
Link: https://lkml.kernel.org/r/20180228102846.13447-6-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode: Do not upload microcode if CPUs are offline

commit 30ec26da9967d0d785abc24073129a34c3211777 upstream.

Avoid loading microcode if any of the CPUs are offline, and issue a
warning. Having different microcode revisions on the system at any time
is outright dangerous.

[ Borislav: Massage changelog. ]

Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Arjan Van De Ven <arjan.van.de.ven@intel.com>
Link: http://lkml.kernel.org/r/1519352533-15992-4-git-send-email-ashok.raj@intel.com
Link: https://lkml.kernel.org/r/20180228102846.13447-5-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode/intel: Writeback and invalidate caches before updating microcode

commit 91df9fdf51492aec9fed6b4cbd33160886740f47 upstream.

Updating microcode is less error prone when caches have been flushed and
depending on what exactly the microcode is updating. For example, some
of the issues around certain Broadwell parts can be addressed by doing a
full cache flush.

[ Borislav: Massage it and use native_wbinvd() in both cases. ]

Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Cc: Arjan Van De Ven <arjan.van.de.ven@intel.com>
Link: http://lkml.kernel.org/r/1519352533-15992-3-git-send-email-ashok.raj@intel.com
Link: https://lkml.kernel.org/r/20180228102846.13447-4-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode/intel: Check microcode revision before updating sibling threads

commit c182d2b7d0ca48e0d6ff16f7d883161238c447ed upstream.

After updating microcode on one of the threads of a core, the other
thread sibling automatically gets the update since the microcode
resources on a hyperthreaded core are shared between the two threads.

Check the microcode revision on the CPU before performing a microcode
update and thus save us the WRMSR 0x79 because it is a particularly
expensive operation.

[ Borislav: Massage changelog and coding style. ]

Signed-off-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Cc: Arjan Van De Ven <arjan.van.de.ven@intel.com>
Link: http://lkml.kernel.org/r/1519352533-15992-2-git-send-email-ashok.raj@intel.com
Link: https://lkml.kernel.org/r/20180228102846.13447-3-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode: Get rid of struct apply_microcode_ctx

commit 854857f5944c59a881ff607b37ed9ed41d031a3b upstream.

It is a useless remnant from earlier times. Use the ucode_state enum
directly.

No functional change.

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Cc: Arjan Van De Ven <arjan.van.de.ven@intel.com>
Link: https://lkml.kernel.org/r/20180228102846.13447-2-bp@alien8.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/CPU: Check CPU feature bits after microcode upgrade

commit 42ca8082e260dcfd8afa2afa6ec1940b9d41724c upstream.

With some microcode upgrades, new CPUID features can become visible on
the CPU. Check what the kernel has mirrored now and issue a warning
hinting at possible things the user/admin can do to make use of the
newly visible features.

Originally-by: Ashok Raj <ashok.raj@intel.com>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Ashok Raj <ashok.raj@intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180216112640.11554-4-bp@alien8.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/CPU: Add a microcode loader callback

commit 1008c52c09dcb23d93f8e0ea83a6246265d2cce0 upstream.

Add a callback function which the microcode loader calls when microcode
has been updated to a newer revision. Do the callback only when no error
was encountered during loading.

Tested-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Ashok Raj <ashok.raj@intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180216112640.11554-3-bp@alien8.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/microcode: Propagate return value from updating functions

commit 3f1f576a195aa266813cbd4ca70291deb61e0129 upstream.

... so that callers can know when microcode was updated and act
accordingly.

Tested-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Ashok Raj <ashok.raj@intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180216112640.11554-2-bp@alien8.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

crypto: arm64/aes-ce-cipher - move assembler code to .S file

commit 019cd46984d04703a39924178f503a98436ac0d7 upstream.

Most crypto drivers involving kernel mode NEON take care to put the code
that actually touches the NEON register file in a separate compilation
unit, to prevent the compiler from reordering code that preserves or
restores the NEON context with code that may corrupt it. This is
necessary because we currently have no way to express the restrictions
imposed upon use of the NEON in kernel mode in a way that the compiler
understands.

However, in the case of aes-ce-cipher, it did not seem unreasonable to
deviate from this rule, given how it does not seem possible for the
compiler to reorder cross object function calls with asm blocks whose
in- and output constraints reflect that it reads from and writes to
memory.

Now that LTO is being proposed for the arm64 kernel, it is time to
revisit this. The link time optimization may replace the function
calls to kernel_neon_begin() and kernel_neon_end() with instantiations
of the IR that make up its implementation, allowing further reordering
with the asm block.

So let's clean this up, and move the asm() blocks into a separate .S
file.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-By: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Matthias Kaehlcke <mka@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

objtool: Add Clang support

commit 3c1f05835cbf9fdfe60b81c718d82ceb94b6c55e upstream.

Since the ORC unwinder was made the default on x86_64, Clang-built
defconfig kernels have triggered some new objtool warnings:

  drivers/gpu/drm/i915/i915_gpu_error.o: warning: objtool: i915_error_printf()+0x6c: return with modified stack frame
  drivers/gpu/drm/i915/intel_display.o: warning: objtool: pipe_config_err()+0xa6: return with modified stack frame

The problem is that objtool has never seen clang-built binaries before.

Shockingly enough, objtool is apparently able to follow the code flow
mostly fine, except for one instruction sequence.  Instead of a LEAVE
instruction, clang restores RSP and RBP the long way:

   67c:   48 89 ec                mov    %rbp,%rsp
   67f:   5d                      pop    %rbp

Teach objtool about this new code sequence.

Reported-and-test-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthias Kaehlcke <mka@chromium.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/fce88ce81c356eedcae7f00ed349cfaddb3363cc.1521741586.git.jpoimboe@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thermal: int3400_thermal: fix error handling in int3400_thermal_probe()

[ Upstream commit 0be86969ae385c5c944286bd9f66068525de15ee ]

There are resources that are not dealocated on failure path
in int3400_thermal_probe().

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tcmu: release blocks for partially setup cmds

[ Upstream commit 810b8153c4243d2012a6ec002ddd3bbc9a9ae8c2 ]

If we cannot setup a cmd because we run out of ring space
or global pages release the blocks before sleeping. This
prevents a deadlock where dev0 has waiting_blocks set and
needs N blocks, but dev1 to devX have each allocated N / X blocks
and also hit the global block limit so they went to sleep.

find_free_blocks is not able to take the sleeping dev's
blocks becaause their waiting_blocks is set and even
if it was not the block returned by find_last_bit could equal
dbi_max. The latter will probably never happen because
DATA_BLOCK_BITS is so high but in the next patches
DATA_BLOCK_BITS and TCMU_GLOBAL_MAX_BLOCKS will be settable so
it might be lower and could happen.

Signed-off-by: Mike Christie <mchristi@redhat.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf tools: Fix copyfile_offset update of output offset

[ Upstream commit fa1195ccc0af2d121abe0fe266a1caee8c265eea ]

We need to increase output offset in each iteration, not decrease it as
we currently do.

I guess we were lucky to finish in most cases in first iteration, so the
bug never showed. However it shows a lot when working with big (~4GB)
size data.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Fixes: 9c9f5a2f1944 ("perf tools: Introduce copyfile_offset() function")
Link: http://lkml.kernel.org/r/20180109133923.25406-1-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

crypto: aes-generic - build with -Os on gcc-7+

[ Upstream commit 148b974deea927f5dbb6c468af2707b488bfa2de ]

While testing other changes, I discovered that gcc-7.2.1 produces badly
optimized code for aes_encrypt/aes_decrypt. This is especially true when
CONFIG_UBSAN_SANITIZE_ALL is enabled, where it leads to extremely
large stack usage that in turn might cause kernel stack overflows:

crypto/aes_generic.c: In function 'aes_encrypt':
crypto/aes_generic.c:1371:1: warning: the frame size of 4880 bytes is larger than 2048 bytes [-Wframe-larger-than=]
crypto/aes_generic.c: In function 'aes_decrypt':
crypto/aes_generic.c:1441:1: warning: the frame size of 4864 bytes is larger than 2048 bytes [-Wframe-larger-than=]

I verified that this problem exists on all architectures that are
supported by gcc-7.2, though arm64 in particular is less affected than
the others. I also found that gcc-7.1 and gcc-8 do not show the extreme
stack usage but still produce worse code than earlier versions for this
file, apparently because of optimization passes that generally provide
a substantial improvement in object code quality but understandably fail
to find any shortcuts in the AES algorithm.

Possible workarounds include

a) disabling -ftree-pre and -ftree-sra optimizations, this was an earlier
   patch I tried, which reliably fixed the stack usage, but caused a
   serious performance regression in some versions, as later testing
   found.

b) disabling UBSAN on this file or all ciphers, as suggested by Ard
   Biesheuvel. This would lead to massively better crypto performance in
   UBSAN-enabled kernels and avoid the stack usage, but there is a concern
   over whether we should exclude arbitrary files from UBSAN at all.

c) Forcing the optimization level in a different way. Similar to a),
   but rather than deselecting specific optimization stages,
   this now uses "gcc -Os" for this file, regardless of the
   CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE/SIZE option. This is a reliable
   workaround for the stack consumption on all architecture, and I've
   retested the performance results now on x86, cycles/byte (lower is
   better) for cbc(aes-generic) with 256 bit keys:

-O2     -Os
gcc-6.3.1 14.9 15.1
gcc-7.0.1 14.7 15.3
gcc-7.1.1 15.3 14.7
gcc-7.2.1 16.8 15.9
gcc-8.0.0 15.5 15.6

This implements the option c) by enabling forcing -Os on all compiler
versions starting with gcc-7.1. As a workaround for PR83356, it would
only be needed for gcc-7.2+ with UBSAN enabled, but since it also shows
better performance on gcc-7.1 without UBSAN, it seems appropriate to
use the faster version here as well.

Side note: during testing, I also played with the AES code in libressl,
which had a similar performance regression from gcc-6 to gcc-7.2,
but was three times slower overall. It might be interesting to
investigate that further and possibly port the Linux implementation
into that.

Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83356
Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83651
Cc: Richard Biener <rguenther@suse.de>
Cc: Jakub Jelinek <jakub@gcc.gnu.org>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mtd: mtd_oobtest: Handle bitflips during reads

[ Upstream commit 12663b442e5ac5aa3d6097cd3f287c71ba46d26e ]

Reads from NAND devices usually trigger bitflips, this is an expected
behavior. While bitflips are under a given threshold, the MTD core
returns 0. However, when the number of corrected bitflips is above this
same threshold, -EUCLEAN is returned to inform the upper layer that this
block is slightly dying and soon the ECC engine will be overtaken so
actions should be taken to move the data out of it.

This particular condition should not be treated like an error and the
test should continue.

Signed-off-by: Miquel Raynal <miquel.raynal@free-electrons.com>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Input: goodix - disable IRQs while suspended

[ Upstream commit faec44b6838312484d63e82286087cf2d5ebb891 ]

We should not try to do any i2c transfers before the controller is
resumed (which happens before our resume method gets called).

So we need to disable our IRQ while suspended to enforce this. The
code paths for devices with GPIOs for the int and reset pins already
disable the IRQ the through goodix_free_irq().

This commit also disables the IRQ while suspended for devices without
GPIOs for the int and reset pins.

This fixes the i2c bus sometimes getting stuck after a suspend/resume
causing the touchscreen to sometimes not work after a suspend/resume.
This has been tested on a GPD pocked device.

BugLink: https://github.com/nexus511/gpd-ubuntu-packages/issues/10
BugLink: https://www.reddit.com/r/GPDPocket/comments/7niut2/fix_for_broken_touch_after_resume_all_linux/
Tested-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Bastien Nocera <hadess@hadess.net>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ibmvnic: Don't handle RX interrupts when not up.

[ Upstream commit 09fb35ead58cd557aa9b20576d15816bc91a4deb ]

Initiating a kdump via the command line can cause a pending interrupt
to be handled by the ibmvnic driver when initializing the sub-CRQ
irqs during driver initialization.

NIP [d000000000ca34f0] ibmvnic_interrupt_rx+0x40/0xd0 [ibmvnic]
LR [c000000008132ef0] __handle_irq_event_percpu+0xa0/0x2f0
Call Trace:
[c000000047fcfde0] [c000000008132ef0] __handle_irq_event_percpu+0xa0/0x2f0
[c000000047fcfea0] [c00000000813317c] handle_irq_event_percpu+0x3c/0x90
[c000000047fcfee0] [c00000000813323c] handle_irq_event+0x6c/0xd0
[c000000047fcff10] [c0000000081385e0] handle_fasteoi_irq+0xf0/0x250
[c000000047fcff40] [c0000000081320a0] generic_handle_irq+0x50/0x80
[c000000047fcff60] [c000000008014984] __do_irq+0x84/0x1d0
[c000000047fcff90] [c000000008027564] call_do_irq+0x14/0x24
[c00000003c92af00] [c000000008014b70] do_IRQ+0xa0/0x120
[c00000003c92af50] [c000000008002594] hardware_interrupt_common+0x114/0x180

Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

sdhci: Advertise 2.0v supply on SDIO host controller

[ Upstream commit 2a609abe71ca59e4bd7139e161eaca2144ae6f2e ]

On Intel Edison the Broadcom Wi-Fi card, which is connected to SDIO,
requires 2.0v, while the host, according to Intel Merrifield TRM,
supports 1.8v supply only.

The card announces itself as

  mmc2: new ultra high speed DDR50 SDIO card at address 0001

Introduce a custom OCR mask for SDIO host controller on Intel Merrifield
and add a special case to sdhci_set_power_noreg() to override 2.0v supply
by enforcing 1.8v power choice.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/gart: Exclude GART aperture from vmcore

[ Upstream commit 2a3e83c6f96c513f43ce5a8c9034608ea584a255 ]

On machines where the GART aperture is mapped over physical RAM
/proc/vmcore contains the remapped range and reading it may cause hangs or
reboots.

In the past, the GART region was added into the resource map, implemented
by commit 56dd669a138c ("[PATCH] Insert GART region into resource map")

However, inserting the iomem_resource from the early GART code caused
resource conflicts with some AGP drivers (bko#72201), which got avoided by
reverting the patch in commit 707d4eefbdb3 ("Revert [PATCH] Insert GART
region into resource map"). This revert introduced the /proc/vmcore bug.

The vmcore ELF header is either prepared by the kernel (when using the
kexec_file_load syscall) or by the kexec userspace (when using the kexec_load
syscall). Since we no longer have the GART iomem resource, the userspace
kexec has no way of knowing which region to exclude from the ELF header.

Changes from v1 of this patch:
Instead of excluding the aperture from the ELF header, this patch
makes /proc/vmcore return zeroes in the second kernel when attempting to
read the aperture region. This is done by reusing the
gart_oldmem_pfn_is_ram infrastructure originally intended to exclude XEN
balooned memory. This works for both, the kexec_file_load and kexec_load
syscalls.

[Note that the GART region is the same in the first and second kernels:
regardless whether the first kernel fixed up the northbridge/bios setting
and mapped the aperture over physical memory, the second kernel finds the
northbridge properly configured by the first kernel and the aperture
never overlaps with e820 memory because the second kernel has a fake e820
map created from the crashkernel memory regions. Thus, the second kernel
keeps the aperture address/size as configured by the first kernel.]

register_oldmem_pfn_is_ram can only register one callback and returns an error
if the callback has been registered already. Since XEN used to be the only user
of this function, it never checks the return value. Now that we have more than
one user, I added a WARN_ON just in case agp, XEN, or any other future user of
register_oldmem_pfn_is_ram were to step on each other's toes.

Fixes: 707d4eefbdb3 ("Revert [PATCH] Insert GART region into resource map")
Signed-off-by: Jiri Bohac <jbohac@suse.cz>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Baoquan He <bhe@redhat.com>
Cc: Toshi Kani <toshi.kani@hpe.com>
Cc: David Airlie <airlied@linux.ie>
Cc: yinghai@kernel.org
Cc: joro@8bytes.org
Cc: kexec@lists.infradead.org
Cc: Borislav Petkov <bp@alien8.de>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Link: https://lkml.kernel.org/r/20180106010013.73suskgxm7lox7g6@dwarf.suse.cz
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

gpio: thunderx: fix error return code in thunderx_gpio_probe()

[ Upstream commit 76e28f5ffed82b1e81a86c4eb8d0420515765620 ]

Fix to return error code -ENOMEM from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: 5a2a30024d8c ("gpio: Add gpio driver support for ThunderX and OCTEON-TX")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Acked-by: David Daney <david.daney@cavium.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

RDMA/cma: Fix rdma_cm path querying for RoCE

[ Upstream commit 89838118a515847d3e5c904d2e022779a7173bec ]

The 'if' logic in ucma_query_path was broken with OPA was introduced
and started to treat RoCE paths as as OPA paths. Invert the logic
of the 'if' so only OPA paths are treated as OPA paths.

Otherwise the path records returned to rdma_cma users are mangled
when in RoCE mode.

Fixes: 57520751445b ("IB/SA: Add OPA path record type")
Signed-off-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: megaraid_sas: unload flag should be set after scsi_remove_host is called

[ Upstream commit f3f7920b3910171b2999c7dc2335eb9f583e44f2 ]

Issue - Driver returns DID_NO_CONNECT when unload is in progress,
indicated using instance->unload flag. In case of dynamic unload of
driver, this flag is set before calling scsi_remove_host(). While doing
manual driver unload, user will see lots of prints for Sync Cache
command with DID_NO_CONNECT status.

Fix - Set the instance->unload flag after scsi_remove_host(). Allow
device removal process to be completed and do not block any command
before that.  SCSI commands (like SYNC_CACHE) are received (as part of
scsi_remove_host) by driver during unload will be submitted further down
to the drives.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: megaraid_sas: Error handling for invalid ldcount provided by firmware in RAID map

[ Upstream commit 7ada701d0d5e5c6d357e157a72b841db3e8d03f4 ]

Currently driver does not validate ldcount provided by firmware.  If the
value is invalid, fail RAID map validation accordingly.  This issue is
rare to hit in field and is fixed as part of code review.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Shivasharan S <shivasharan.srikanteshwara@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cxgb4vf: Fix SGE FL buffer initialization logic for 64K pages

[ Upstream commit ea0a42109aee7b92e631c4eb3f2219fadf58acdd ]

We'd come in with SGE_FL_BUFFER_SIZE[0] and [1] both equal to 64KB and
the extant logic would flag that as an error. This was already fixed in
cxgb4 driver with "92ddcc7 cxgb4: Fix some small bugs in
t4_sge_init_soft() when our Page Size is 64KB".

Original Work by: Casey Leedom <leedom@chelsio.com>
Signed-off-by: Arjun Vynipadath <arjun@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

i40evf: don't rely on netif_running() outside rtnl_lock()

[ Upstream commit 44b034b406211fc103159f82b9e601e05675c739 ]

In i40evf_reset_task we use netif_running() to determine whether or not
the device is currently up. This allows us to properly free queue memory
and shut down things before we request the hardware reset.

It turns out that we cannot be guaranteed of netif_running() returning
false until the device is fully up, as the kernel core code sets
__LINK_STATE_START prior to calling .ndo_open. Since we're not holding
the rtnl_lock(), it's possible that the driver's i40evf_open handler
function is currently being called while we're resetting.

We can't simply hold the rtnl_lock() while checking netif_running() as
this could cause a deadlock with the i40evf_open() function.
Additionally, we can't avoid the deadlock by holding the rtnl_lock()
over the whole reset path, as this essentially serializes all resets,
and can cause massive delays if we have multiple VFs on a system.

Instead, lets just check our own internal state __I40EVF_RUNNING state
field. This allows us to ensure that the state is correct and is only
set after we've finished bringing the device up.

Without this change we might free data structures about device queues
and other memory before they've been fully allocated.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

uio_hv_generic: check that host supports monitor page

[ Upstream commit 06028d15177a1b406b7b075ea47c6a352732f23a ]

In order for userspace application to signal host, it needs the
host to support the monitor page property. Check for the flag
and fail if this is not supported.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

EDAC, mv64x60: Fix an error handling path

[ Upstream commit 68fa24f9121c04ef146b5158f538c8b32f285be5 ]

We should not call edac_mc_del_mc() if a corresponding call to
edac_mc_add_mc() has not been performed yet.

So here, we should go to err instead of err2 to branch at the right
place of the error handling path.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Cc: linux-edac <linux-edac@vger.kernel.org>
Link: http://lkml.kernel.org/r/20180107205400.14068-1-christophe.jaillet@wanadoo.fr
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

block, bfq: put async queues for root bfq groups too

[ Upstream commit 52257ffbfcaf58d247b13fb148e27ed17c33e526 ]

For each pair [device for which bfq is selected as I/O scheduler,
group in blkio/io], bfq maintains a corresponding bfq group. Each such
bfq group contains a set of async queues, with each async queue
created on demand, i.e., when some I/O request arrives for it.  On
creation, an async queue gets an extra reference, to make sure that
the queue is not freed as long as its bfq group exists.  Accordingly,
to allow the queue to be freed after the group exited, this extra
reference must released on group exit.

The above holds also for a bfq root group, i.e., for the bfq group
corresponding to the root blkio/io root for a given device. Yet, by
mistake, the references to the existing async queues of a root group
are not released when the latter exits. This causes a memory leak when
the instance of bfq for a given device exits. In a similar vein,
bfqg_stats_xfer_dead is not executed for a root group.

This commit fixes bfq_pd_offline so that the latter executes the above
missing operations for a root group too.

Reported-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Reported-by: Guoqing Jiang <gqjiang@suse.com>
Tested-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: Davide Ferrari <davideferrari8@gmail.com>
Signed-off-by: Paolo Valente <paolo.valente@linaro.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: n_gsm: Allow ADM response in addition to UA for control dlci

[ Upstream commit ea3d8465ab9b3e01be329ac5195970a84bef76c5 ]

Some devices have the control dlci stay in ADM mode instead of the UA
mode. This can seen at least on droid 4 when trying to open the ts
27.010 mux port. Enabling n_gsm debug mode shows the control dlci
always respond with DM to SABM instead of UA:

# modprobe n_gsm debug=0xff
# ldattach -d GSM0710 /dev/ttyS0 &
gsmld_output: 00000000: f9 03 3f 01 1c f9
--> 0) C: SABM(P)
gsmld_receive: 00000000: f9 03 1f 01 36 f9
<-- 0) C: DM(P)
...
$ minicom -D /dev/gsmtty1
minicom: cannot open /dev/gsmtty1: No error information
$ strace minicom -D /dev/gsmtty1
...
open("/dev/gsmtty1", O_RDWR|O_NOCTTY|O_NONBLOCK|O_LARGEFILE) = -1 EL2HLT

Note that this is different issue from other n_gsm -EL2HLT issues such
as timeouts when the control dlci does not respond at all.

The ADM mode seems to be a quite common according to "RF Wireless World"
article "GSM Issue-UE sends SABM and gets a DM response instead of
UA response":

  This issue is most commonly observed in GSM networks where in UE sends
  SABM and expects network to send UA response but it ends up receiving
  DM response from the network. SABM stands for Set asynchronous balanced
  mode, UA stands for Unnumbered Acknowledge and DA stands for
  Disconnected Mode.

  An RLP entity can be in one of two modes:
  - Asynchronous Balanced Mode (ABM)
  - Asynchronous Disconnected Mode (ADM)

Currently Linux kernel closes the control dlci after several retries
in gsm_dlci_t1() on DM. This causes n_gsm /dev/gsmtty ports to produce
error code -EL2HLT when trying to open them as the closing of control
dlci has already set gsm->dead.

Let's fix the issue by allowing control dlci stay in ADM mode after the
retries so the /dev/gsmtty ports can be opened and used. It seems that
it might take several attempts to get any response from the control
dlci, so it's best to allow ADM mode only after the SABM retries are
done.

Note that for droid 4 additional patches are needed to mux the ttyS0
pins and to toggle RTS gpio_149 to wake up the mdm6600 modem are also
needed to use n_gsm. And the mdm6600 modem needs to be powered on.

Cc: linux-serial@vger.kernel.org
Cc: Alan Cox <alan@llwyncelyn.cymru>
Cc: Jiri Prchal <jiri.prchal@aksignal.cz>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Marcel Partap <mpartap@gmx.net>
Cc: Michael Scott <michael.scott@linaro.org>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Russ Gorby <russ.gorby@intel.com>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Sebastian Reichel <sre@kernel.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

blk-mq: fix kernel oops in blk_mq_tag_idle()

[ Upstream commit 8ab0b7dc73e1b3e2987d42554b2bff503f692772 ]

HW queues may be unmapped in some cases, such as blk_mq_update_nr_hw_queues(),
then we need to check it before calling blk_mq_tag_idle(), otherwise
the following kernel oops can be triggered, so fix it by checking if
the hw queue is unmapped since it doesn't make sense to idle the tags
any more after hw queues are unmapped.

[  440.771298] Workqueue: nvme-wq nvme_rdma_del_ctrl_work [nvme_rdma]
[  440.779104] task: ffff894bae755ee0 ti: ffff893bf9bc8000 task.ti: ffff893bf9bc8000
[  440.788359] RIP: 0010:[<ffffffffb730e2b4>]  [<ffffffffb730e2b4>] __blk_mq_tag_idle+0x24/0x40
[  440.798697] RSP: 0018:ffff893bf9bcbd10  EFLAGS: 00010286
[  440.805538] RAX: 0000000000000000 RBX: ffff895bb131dc00 RCX: 000000000000011f
[  440.814426] RDX: 00000000ffffffff RSI: 0000000000000120 RDI: ffff895bb131dc00
[  440.823301] RBP: ffff893bf9bcbd10 R08: 000000000001b860 R09: 4a51d361c00c0000
[  440.832193] R10: b5907f32b4cc7003 R11: ffffd6cabfb57000 R12: ffff894bafd1e008
[  440.841091] R13: 0000000000000001 R14: ffff895baf770000 R15: 0000000000000080
[  440.849988] FS:  0000000000000000(0000) GS:ffff894bbdcc0000(0000) knlGS:0000000000000000
[  440.859955] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  440.867274] CR2: 0000000000000008 CR3: 000000103d098000 CR4: 00000000001407e0
[  440.876169] Call Trace:
[  440.879818]  [<ffffffffb7309d68>] blk_mq_exit_hctx+0xd8/0xe0
[  440.887051]  [<ffffffffb730dc40>] blk_mq_free_queue+0xf0/0x160
[  440.894465]  [<ffffffffb72ff679>] blk_cleanup_queue+0xd9/0x150
[  440.901881]  [<ffffffffc08a802b>] nvme_ns_remove+0x5b/0xb0 [nvme_core]
[  440.910068]  [<ffffffffc08a811b>] nvme_remove_namespaces+0x3b/0x60 [nvme_core]
[  440.919026]  [<ffffffffc08b817b>] __nvme_rdma_remove_ctrl+0x2b/0xb0 [nvme_rdma]
[  440.928079]  [<ffffffffc08b8237>] nvme_rdma_del_ctrl_work+0x17/0x20 [nvme_rdma]
[  440.937126]  [<ffffffffb70ab58a>] process_one_work+0x17a/0x440
[  440.944517]  [<ffffffffb70ac3a8>] worker_thread+0x278/0x3c0
[  440.951607]  [<ffffffffb70ac130>] ? manage_workers.isra.24+0x2a0/0x2a0
[  440.959760]  [<ffffffffb70b352f>] kthread+0xcf/0xe0
[  440.966055]  [<ffffffffb70b3460>] ? insert_kthread_work+0x40/0x40
[  440.973715]  [<ffffffffb76d8658>] ret_from_fork+0x58/0x90
[  440.980586]  [<ffffffffb70b3460>] ? insert_kthread_work+0x40/0x40
[  440.988229] Code: 5b 41 5c 5d c3 66 90 0f 1f 44 00 00 48 8b 87 20 01 00 00 f0 0f ba 77 40 01 19 d2 85 d2 75 08 c3 0f 1f 80 00 00 00 00 55 48 89 e5 <f0> ff 48 08 48 8d 78 10 e8 7f 0f 05 00 5d c3 0f 1f 00 66 2e 0f
[  441.011620] RIP  [<ffffffffb730e2b4>] __blk_mq_tag_idle+0x24/0x40
[  441.019301]  RSP <ffff893bf9bcbd10>
[  441.024052] CR2: 0000000000000008

Reported-by: Zhang Yi <yizhan@redhat.com>
Tested-by: Zhang Yi <yizhan@redhat.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: libsas: initialize sas_phy status according to response of DISCOVER

[ Upstream commit affc67788fe5dfffad5cda3d461db5cf2b2ff2b0 ]

The status of SAS PHY is in sas_phy->enabled. There is an issue that the
status of a remote SAS PHY may be initialized incorrectly: if disable
remote SAS PHY through sysfs interface (such as echo 0 >
/sys/class/sas_phy/phy-1:0:0/enable), then reboot the system, and we
will find the status of remote SAS PHY which is disabled before is
1 (cat /sys/class/sas_phy/phy-1:0:0/enable). But actually the status of
remote SAS PHY is disabled and the device attached is not found.

In SAS protocol, NEGOTIATED LOGICAL LINK RATE field of DISCOVER response
is 0x1 when remote SAS PHY is disabled. So initialize sas_phy->enabled
according to the value of NEGOTIATED LOGICAL LINK RATE field.

Signed-off-by: chenxiang <chenxiang66@hisilicon.com>
Reviewed-by: John Garry <john.garry@huawei.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: libsas: fix error when getting phy events

[ Upstream commit 2b23d9509fd7174b362482cf5f3b5f9a2265bc33 ]

The intend purpose here was to goto out if smp_execute_task() returned
error. Obviously something got screwed up. We will never get these link
error statistics below:

~:/sys/class/sas_phy/phy-1:0:12 # cat invalid_dword_count
0
~:/sys/class/sas_phy/phy-1:0:12 # cat running_disparity_error_count
0
~:/sys/class/sas_phy/phy-1:0:12 # cat loss_of_dword_sync_count
0
~:/sys/class/sas_phy/phy-1:0:12 # cat phy_reset_problem_count
0

Obviously we should goto error handler if smp_execute_task() returns
non-zero.

Fixes: 2908d778ab3e ("[SCSI] aic94xx: new driver")
Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: John Garry <john.garry@huawei.com>
CC: chenqilin <chenqilin2@huawei.com>
CC: chenxiang <chenxiang66@hisilicon.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: libsas: fix memory leak in sas_smp_get_phy_events()

[ Upstream commit 4a491b1ab11ca0556d2fda1ff1301e862a2d44c4 ]

We've got a memory leak with the following producer:

while true;
do cat /sys/class/sas_phy/phy-1:0:12/invalid_dword_count >/dev/null;
done

The buffer req is allocated and not freed after we return. Fix it.

Fixes: 2908d778ab3e ("[SCSI] aic94xx: new driver")
Signed-off-by: Jason Yan <yanaijie@huawei.com>
CC: John Garry <john.garry@huawei.com>
CC: chenqilin <chenqilin2@huawei.com>
CC: chenxiang <chenxiang66@hisilicon.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bcache: segregate flash only volume write streams

[ Upstream commit 4eca1cb28d8b0574ca4f1f48e9331c5f852d43b9 ]

In such scenario that there are some flash only volumes
, and some cached devices, when many tasks request these devices in
writeback mode, the write IOs may fall to the same bucket as bellow:
| cached data | flash data | cached data | cached data| flash data|
then after writeback of these cached devices, the bucket would
be like bellow bucket:
| free | flash data | free | free | flash data |

So, there are many free space in this bucket, but since data of flash
only volumes still exists, so this bucket cannot be reclaimable,
which would cause waste of bucket space.

In this patch, we segregate flash only volume write streams from
cached devices, so data from flash only volumes and cached devices
can store in different buckets.

Compare to v1 patch, this patch do not add a additionally open bucket
list, and it is try best to segregate flash only volume write streams
from cached devices, sectors of flash only volumes may still be mixed
with dirty sectors of cached device, but the number is very small.

[mlyle: fixed commit log formatting, permissions, line endings]

Signed-off-by: Tang Junhui <tang.junhui@zte.com.cn>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bcache: stop writeback thread after detaching

[ Upstream commit 8d29c4426b9f8afaccf28de414fde8a722b35fdf ]

Currently, when a cached device detaching from cache, writeback thread is
not stopped, and writeback_rate_update work is not canceled. For example,
after the following command:
echo 1 >/sys/block/sdb/bcache/detach
you can still see the writeback thread. Then you attach the device to the
cache again, bcache will create another writeback thread, for example,
after below command:
echo  ba0fb5cd-658a-4533-9806-6ce166d883b9 > /sys/block/sdb/bcache/attach
then you will see 2 writeback threads.
This patch stops writeback thread and cancels writeback_rate_update work
when cached device detaching from cache.

Compare with patch v1, this v2 patch moves code down into the register
lock for safety in case of any future changes as Coly and Mike suggested.

[edit by mlyle: commit log spelling/formatting]

Signed-off-by: Tang Junhui <tang.junhui@zte.com.cn>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bcache: ret IOERR when read meets metadata error

[ Upstream commit b221fc130c49c50f4c2250d22e873420765a9fa2 ]

The read request might meet error when searching the btree, but the error
was not handled in cache_lookup(), and this kind of metadata failure will
not go into cached_dev_read_error(), finally, the upper layer will receive
bi_status=0.  In this patch we judge the metadata error by the return
value of bch_btree_map_keys(), there are two potential paths give rise to
the error:

1. Because the btree is not totally cached in memery, we maybe get error
   when read btree node from cache device (see bch_btree_node_get()), the
   likely errno is -EIO, -ENOMEM

2. When read miss happens, bch_btree_insert_check_key() will be called to
   insert a "replace_key" to btree(see cached_dev_cache_miss(), just for
   doing preparatory work before insert the missed data to cache device),
   a failure can also happen in this situation, the likely errno is
   -ENOMEM

bch_btree_map_keys() will return MAP_DONE in normal scenario, but we will
get either -EIO or -ENOMEM in above two cases. if this happened, we should
NOT recover data from backing device (when cache device is dirty) because
we don't know whether bkeys the read request covered are all clean.  And
after that happened, s->iop.status is still its initially value(0) before
we submit s->bio.bio, we set it to BLK_STS_IOERR, so it can go into
cached_dev_read_error(), and finally it can be passed to upper layer, or
recovered by reread from backing device.

[edit by mlyle: patch formatting, word-wrap, comment spelling,
commit log format]

Signed-off-by: Hua Rui <huarui.dev@gmail.com>
Reviewed-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Michael Lyle <mlyle@lyle.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: hns3: fix for changing MTU

[ Upstream commit 5bad95a1e55f4d5bb41e130db859d57eaf1b1549 ]

when changing MTU, The new MTU must need to be set to netdevice.

Fixes: a8e8b7ff3517 ("net: hns3: Add support to change MTU in HNS3 hardware")
Signed-off-by: Fuyun Liang <liangfuyun1@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: hns3: Fix an error macro definition of HNS3_TQP_STAT

[ Upstream commit 57ffee737b36dbb81e8e60a37e01791553157a5e ]

The member "stats_offset" was designed to indicate the offset
of each member of struct ring_stats in struct hns3_enet_ring,
but forgot to add the offset of the member in struct ring_stats.

Fixes: 496d03e960a ("net: hns3: Add Ethtool support to HNS3 driver")
Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: hns3: Fix a loop index error of tqp statistics query

[ Upstream commit 94bfaafac9d2a3c0bcca00d01e38f7597b741799 ]

An error loop index was used while querying statistics data
of tqps, which may cause call trace.

Fixes: 496d03e960ae ("net: hns3: Add Ethtool support to HNS3 driver")
Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: hns3: Fix an error of total drop packet statistics

[ Upstream commit d2a5dca8404871be683c6bbc175ebf9c56dd2865 ]

The dropped tx/rx packets number of each tqp should also
be counted into the total drop tx/rx packets numbers.

Fixes: 76ad4f0ee74 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net/mlx5: Fix race for multiple RoCE enable

[ Upstream commit 734dc065fc41f6143ff88225aa5d335cb1e0f6aa ]

There are two potential problems with the existing implementation.

1. Enable and disable can race after the atomic operations.
2. If a command fails the refcount is left in an inconsistent state.

Introduce a lock and perform error checking.

Fixes: a6f7d2aff623 ("net/mlx5: Add support for multiple RoCE enable")
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

wl1251: check return from call to wl1251_acx_arp_ip_filter

[ Upstream commit ac1181c60822292176ab96912208ec9f9819faf8 ]

Currently the less than zero error check on ret is incorrect
as it is checking a far earlier ret assignment rather than the
return from the call to wl1251_acx_arp_ip_filter. Fix this by
adding in the missing assginment.

Detected by CoverityScan, CID#1164835 ("Logically dead code")

Fixes: 204cc5c44fb6 ("wl1251: implement hardware ARP filtering")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

rt2x00: do not pause queue unconditionally on error path

[ Upstream commit 6dd80efd75ce7c2dbd9f117cf585ee2b33a42ee1 ]

Pausing queue without checking threshold is racy with txdone path.
Moreover we do not need pause queue on any error, but only if queue
is full - in case when we send RTS frame ( other cases of almost full
queue are already handled in rt2x00queue_write_tx_frame() ).

Patch fixes of theoretically possible problem of pausing empty
queue.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Tested-by: Enrico Mioso <mrkiko.rs@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

power: supply: axp288_charger: Properly stop work on probe-error / remove

[ Upstream commit 165c2357744e41391902a2a72dd170beb60c28d5 ]

Properly stop any work we may have queued on probe-errors / remove.

Rather then adding a remove driver callback for this, and goto style
error handling to probe, use a devm_action for this.

The devm_action gets registered before we register any of the extcon
notifiers which may queue the work, devm does cleanup in reverse order,
so this ensures that the notifiers are removed before we cancel the work.

Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.co.uk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ASoC: Intel: sst: Fix the return value of 'sst_send_byte_stream_mrfld()'

[ Upstream commit eaadb1caa966a91128297b754e90b7c92b350a00 ]

In some error handling paths, an error code is assiegned to 'ret'.
However, the function always return 0.

Fix it and return the error code if such an error paths is taken.

Fixes: 3d9ff34622ba ("ASoC: Intel: sst: add stream operations")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: disable preempt while sampling processor id.

[ Upstream commit dbeccabf5294e80f7cc9ee566746c42211bed736 ]

Calling smp_processor_id() without disabling preemption
triggers a warning (if CONFIG_DEBUG_PREEMPT).
I think the result of cfs_cpt_current() is only used as a hint for
load balancing, rather than as a precise and stable indicator of
the current CPU.  So it doesn't need to be called with
preemption disabled.

So disable preemption inside cfs_cpt_current() to silence the warning.

Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf report: Fix a no annotate browser displayed issue

[ Upstream commit 40c39e3046411f84bab82f66783ff3593e2bcd9b ]

When enabling '-b' option in perf record, for example,

  perf record -b ...
  perf report

and then browsing the annotate browser from perf report (press 'A'), it
would fail (annotate browser can't be displayed).

It's because the '.add_entry_cb' op of struct report is overwritten by
hist_iter__branch_callback() in builtin-report.c. But this function doesn't do
something like mapping symbols and sources. So next, do_annotate() will return
directly.

        notes = symbol__annotation(act->ms.sym);
        if (!notes->src)
                return 0;

This patch adds the lost code to hist_iter__branch_callback (refer to
hist_iter__report_callback).

v2:

Fix a crash bug when perform 'perf report --stdio'.

The reason is that we init the symbol annotation only in browser mode, it
doesn't allocate/init resources for stdio mode.

So now in hist_iter__branch_callback(), it will return directly if it's not in
browser mode.

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1514284963-18587-1-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tpm: return a TPM_RC_COMMAND_CODE response if command is not implemented

[ Upstream commit 095531f891e627e408606f2da4008d3d53e6748a ]

According to the TPM Library Specification, a TPM device must do a command
header validation before processing and return a TPM_RC_COMMAND_CODE code
if the command is not implemented.

So user-space will expect to handle that response as an error. But if the
in-kernel resource manager is used (/dev/tpmrm?), an -EINVAL errno code is
returned instead if the command isn't implemented. This confuses userspace
since it doesn't expect that error value.

This also isn't consistent with the behavior when not using TPM spaces and
accessing the TPM directly (/dev/tpm?). In this case, the command is sent
to the TPM even when not implemented and the TPM responds with an error.

Instead of returning an -EINVAL errno code when the tpm_validate_command()
function fails, synthesize a TPM command response so user-space can get a
TPM_RC_COMMAND_CODE as expected when a chip doesn't implement the command.

The TPM only sets 12 of the 32 bits in the TPM_RC response, so the TSS and
TAB specifications define that higher layers in the stack should use some
of the unused 20 bits to specify from which level of the stack the error
is coming from.

Since the TPM_RC_COMMAND_CODE response code is sent by the kernel resource
manager, set the error level to the TAB/RM layer so user-space is aware of
this.

Suggested-by: Jason Gunthorpe <jgg@ziepe.ca>
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: William Roberts <william.c.roberts@intel.com>
Reviewed-by: Philip Tricca <philip.b.tricca@intel.com>
Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Tested-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

nvme_fcloop: fix abort race condition

[ Upstream commit 278e096063f1914fccfc77a617be9fc8dbb31b0e ]

A test case revealed a race condition of an i/o completing on a thread
parallel to the delete_association generating the aborts for the
outstanding ios on the controller.  The i/o completion was freeing the
target fcloop context, thus the abort task referenced the just-freed
memory.

Correct by clearing the target/initiator cross pointers in the io
completion and abort tasks before calling the callbacks. On aborts
that detect already finished io's, ensure the complete context is
called.

Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

nvme_fcloop: disassocate local port structs

[ Upstream commit 6fda20283e55b9d288cd56822ce39fc8e64f2208 ]

The current fcloop driver gets its lport structure from the private
area co-allocated with the fc_localport. All is fine except the
teardown path, which wants to wait on the completion, which is marked
complete by the delete_localport callback performed after
unregister_localport.  The issue is, the nvme_fc transport frees the
localport structure immediately after delete_localport is called,
meaning the original routine is trying to wait on a complete that
was just freed.

Change such that a lport struct is allocated coincident with the
addition and registration of a localport. The private area of the
localport now contains just a backpointer to the real lport struct.
Now, the completion can be waited for, and after completing, the
new structure can be kfree'd.

Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

pinctrl: baytrail: Enable glitch filter for GPIOs used as interrupts

[ Upstream commit 9291c65b01d1c67ebd56644cb19317ad665c44b3 ]

On some systems, some PCB traces attached to GpioInts are routed in such
a way that they pick up enough interference to constantly (many times per
second) trigger.

Enabling glitch-filtering fixes this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

backlight: tdo24m: Fix the SPI CS between transfers

[ Upstream commit 2023b0524a6310e9ea80daf085f51c71bff9289f ]

Currently the LCD display (TD035S) on the cm-x300 platform is broken and
remains blank.

The TD0245S specification requires that the chipselect is toggled
between commands sent to the panel. This was also the purpose of the
former patch of commit f64dcac0b124 ("backlight: tdo24m: ensure chip
select changes between transfers").

Unfortunately, the "cs_change" field of a SPI transfer is
misleading. Its true meaning is that for a SPI message holding multiple
transfers, the chip select is toggled between each transfer, but for the
last transfer it remains asserted.

In this driver, all the SPI messages contain exactly one transfer, which
means that each transfer is the last of its message, and as a
consequence the chip select is never toggled.

Actually, there was a second bug hidding the first one, hence the
problem was not seen until v4.6. This problem was fixed by commit
a52db659c79c ("spi: pxa2xx: Fix cs_change management") for PXA based
boards.

This fix makes the TD035S work again on a cm-x300 board. The same
applies to other PXA boards, ie. corgi and tosa.

Fixes: a52db659c79c ("spi: pxa2xx: Fix cs_change management")
Reported-by: Andrea Adami <andrea.adami@gmail.com>
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Acked-by: Daniel Thompson <daniel.thompson@linaro.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

blk-mq: fix race between updating nr_hw_queues and switching io sched

[ Upstream commit fb350e0ad99359768e1e80b4784692031ec340e4 ]

In both elevator_switch_mq() and blk_mq_update_nr_hw_queues(), sched tags
can be allocated, and q->nr_hw_queue is used, and race is inevitable, for
example: blk_mq_init_sched() may trigger use-after-free on hctx, which is
freed in blk_mq_realloc_hw_ctxs() when nr_hw_queues is decreased.

This patch fixes the race be holding q->sysfs_lock.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: Yi Zhang <yi.zhang@redhat.com>
Tested-by: Yi Zhang <yi.zhang@redhat.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

blk-mq: avoid to map CPU into stale hw queue

[ Upstream commit 7d4901a90d02500c8011472a060f9b2e60e6e605 ]

blk_mq_pci_map_queues() may not map one CPU into any hw queue, but its
previous map isn't cleared yet, and may point to one stale hw queue
index.

This patch fixes the following issue by clearing the mapping table before
setting it up in blk_mq_pci_map_queues().

This patches fixes this following issue reported by Zhang Yi:

[  101.202734] BUG: unable to handle kernel NULL pointer dereference at 0000000094d3013f
[  101.211487] IP: blk_mq_map_swqueue+0xbc/0x200
[  101.216346] PGD 0 P4D 0
[  101.219171] Oops: 0000 [#1] SMP
[  101.222674] Modules linked in: sunrpc ipmi_ssif vfat fat intel_rapl sb_edac x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel intel_cstate intel_uncore mxm_wmi intel_rapl_perf iTCO_wdt ipmi_si ipmi_devintf pcspkr iTCO_vendor_support sg dcdbas ipmi_msghandler wmi mei_me lpc_ich shpchp mei acpi_power_meter dm_multipath ip_tables xfs libcrc32c sd_mod mgag200 i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops ttm drm ahci libahci crc32c_intel libata tg3 nvme nvme_core megaraid_sas ptp i2c_core pps_core dm_mirror dm_region_hash dm_log dm_mod
[  101.284881] CPU: 0 PID: 504 Comm: kworker/u25:5 Not tainted 4.15.0-rc2 #1
[  101.292455] Hardware name: Dell Inc. PowerEdge R730xd/072T6D, BIOS 2.5.5 08/16/2017
[  101.301001] Workqueue: nvme-wq nvme_reset_work [nvme]
[  101.306636] task: 00000000f2c53190 task.stack: 000000002da874f9
[  101.313241] RIP: 0010:blk_mq_map_swqueue+0xbc/0x200
[  101.318681] RSP: 0018:ffffc9000234fd70 EFLAGS: 00010282
[  101.324511] RAX: ffff88047ffc9480 RBX: ffff88047e130850 RCX: 0000000000000000
[  101.332471] RDX: ffffe8ffffd40580 RSI: ffff88047e509b40 RDI: ffff88046f37a008
[  101.340432] RBP: 000000000000000b R08: ffff88046f37a008 R09: 0000000011f94280
[  101.348392] R10: ffff88047ffd4d00 R11: 0000000000000000 R12: ffff88046f37a008
[  101.356353] R13: ffff88047e130f38 R14: 000000000000000b R15: ffff88046f37a558
[  101.364314] FS:  0000000000000000(0000) GS:ffff880277c00000(0000) knlGS:0000000000000000
[  101.373342] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  101.379753] CR2: 0000000000000098 CR3: 000000047f409004 CR4: 00000000001606f0
[  101.387714] Call Trace:
[  101.390445]  blk_mq_update_nr_hw_queues+0xbf/0x130
[  101.395791]  nvme_reset_work+0x6f4/0xc06 [nvme]
[  101.400848]  ? pick_next_task_fair+0x290/0x5f0
[  101.405807]  ? __switch_to+0x1f5/0x430
[  101.409988]  ? put_prev_entity+0x2f/0xd0
[  101.414365]  process_one_work+0x141/0x340
[  101.418836]  worker_thread+0x47/0x3e0
[  101.422921]  kthread+0xf5/0x130
[  101.426424]  ? rescuer_thread+0x380/0x380
[  101.430896]  ? kthread_associate_blkcg+0x90/0x90
[  101.436048]  ret_from_fork+0x1f/0x30
[  101.440034] Code: 48 83 3c ca 00 0f 84 2b 01 00 00 48 63 cd 48 8b 93 10 01 00 00 8b 0c 88 48 8b 83 20 01 00 00 4a 03 14 f5 60 04 af 81 48 8b 0c c8 <48> 8b 81 98 00 00 00 f0 4c 0f ab 30 8b 81 f8 00 00 00 89 42 44
[  101.461116] RIP: blk_mq_map_swqueue+0xbc/0x200 RSP: ffffc9000234fd70
[  101.468205] CR2: 0000000000000098
[  101.471907] ---[ end trace 5fe710f98228a3ca ]---
[  101.482489] Kernel panic - not syncing: Fatal exception
[  101.488505] Kernel Offset: disabled
[  101.497752] ---[ end Kernel panic - not syncing: Fatal exception

Reviewed-by: Christoph Hellwig <hch@lst.de>
Suggested-by: Christoph Hellwig <hch@lst.de>
Reported-by: Yi Zhang <yi.zhang@redhat.com>
Tested-by: Yi Zhang <yi.zhang@redhat.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

IB/rdmavt: Allocate CQ memory on the correct node

[ Upstream commit db9a2c6f9b6196b889b98e961cb9a37617b11ccf ]

CQ allocation does not ensure that completion queue entries
and the completion queue structure are allocated on the correct
numa node.

Fix by allocating the rvt_cq and kernel CQ entries on the device node,
leaving the user CQ entries on the default local node.  Also ensure
CQ resizes use the correct allocator when extending a CQ.

Reviewed-by: Sebastian Sanchez <sebastian.sanchez@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

powernv-cpufreq: Add helper to extract pstate from PMSR

[ Upstream commit ee1f4a7dafa997816ff3de96155c6f3edc21c1e6 ]

On POWERNV platform, the fields for pstates in the Power Management
Status Register (PMSR) and the Power Management Control Register
(PMCR) are 8-bits wide. On POWER8 the pstates are negatively numbered
while on POWER9 they are positively numbered.

The device-tree exports pstates as 32-bit entries. The device-tree
implementation sign-extends the 8-bit pstate values to obtain the
corresponding 32-bit entry.

Eg: On POWER8, a pstate value 0x82 [-126] is represented in the
device-tree as 0xfffffff82 while on POWER9, the same value 0x82 [130]
is represented in the device-tree as 0x00000082.

The powernv-cpufreq driver implementation represents pstates using the
integer type. In multiple places in the driver, the code interprets
the pstates extracted from the PMSR as a signed byte and assigns it to
a integer variable to get the sign-extention.

On POWER9 platforms which have greater than 128 pstates, this results
in the driver performing incorrect sign-extention, and thereby
treating a legitimate pstate (say 130) as an invalid pstates (since it
is interpreted as -126).

This patch fixes the issue by implementing a helper function to
extract Pstates from PMSR register, and correctly sign-extend it to be
consistent with the values provided by the device-tree.

Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Acked-by: Balbir Singh <bsingharora@gmail.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

gpio: label descriptors using the device name

[ Upstream commit 24e78079bf2250874e33da2e7cfbb6db72d3caf4 ]

Some GPIO lines appear named "?" in the lsgpio dump due to their
requesting drivers not passing a reasonable label.

Most typically this happens if a device tree node just defines
gpios = <...> and not foo-gpios = <...>, the former gets named
"foo" and the latter gets named "?".

However the struct device passed in is always valid so let's
just label the GPIO with dev_name() on the device if no proper
label was passed.

Cc: Reported-by: Jason Kridner <jkridner@beagleboard.org>
Reported-by: Jason Kridner <jkridner@beagleboard.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

vfb: fix video mode and line_length being set when loaded

[ Upstream commit 7b9faf5df0ac495a1a3d7cdb64921c179f9008ac ]

Currently, when loading the vfb module, the newly created fbdev
has a line_length of 0, and its video mode would be PSEUDOCOLOR
regardless of color depth. (The former could be worked around by
calling the FBIOPUT_VSCREENINFO ioctl with having the FBACTIVIATE_FORCE
flag set.) This patch automatically sets the line_length correctly,
and the video mode is derived from the bit depth now as well.

Thanks to Geert Uytterhoeven for confirming the bug and helping me with
the patch.

Output of `fbset -i' before the patch:
mode "1366x768-60"
    # D: 72.432 MHz, H: 47.403 kHz, V: 60.004 Hz
    geometry 1366 768 1366 768 32
    timings 13806 120 10 14 3 32 5
    rgba 8/0,8/8,8/16,8/24
endmode

Frame buffer device information:
    Name        : Virtual FB
    Address     : 0xffffaa1405d85000
    Size        : 4196352
    Type        : PACKED PIXELS
    Visual      : PSEUDOCOLOR
    XPanStep    : 1
    YPanStep    : 1
    YWrapStep   : 1
    LineLength  : 0                    <-- note this
    Accelerator : No

After:
mode "1366x768-60"
    # D: 72.432 MHz, H: 47.403 kHz, V: 60.004 Hz
    geometry 1366 768 1366 768 32
    timings 13806 120 10 14 3 32 5
    rgba 8/0,8/8,8/16,8/24
endmode

Frame buffer device information:
    Name        : Virtual FB
    Address     : 0xffffaa1405d85000
    Size        : 4196352
    Type        : PACKED PIXELS
    Visual      : TRUECOLOR
    XPanStep    : 1
    YPanStep    : 1
    YWrapStep   : 1
    LineLength  : 5464
    Accelerator : No

Signed-off-by: "Pieter \"PoroCYon\" Sluys" <pcy@national.shitposting.agency>
Reviewed-by: Geert Uytterhoeven <geert@linux-m68k.org>
[b.zolnierkie: minor fixups]
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mac80211: Fix setting TX power on monitor interfaces

[ Upstream commit 3a3713ec360138f806c6fc368d1de570f692b347 ]

Instead of calling ieee80211_recalc_txpower on monitor interfaces
directly, call it using the virtual monitor interface, if one exists.

In case of a single monitor interface given, reject setting TX power,
if no virtual monitor interface exists.

That being checked, don't warn in ieee80211_bss_info_change_notify,
after setting TX power on a monitor interface.

Fixes warning:
------------[ cut here ]------------
 WARNING: CPU: 0 PID: 2193 at net/mac80211/driver-ops.h:167
 ieee80211_bss_info_change_notify+0x111/0x190 Modules linked in: uvcvideo
 videobuf2_vmalloc videobuf2_memops videobuf2_v4l2 videobuf2_core
rndis_host cdc_ether usbnet mii tp_smapi(O) thinkpad_ec(O) ohci_hcd vboxpci(O)
 vboxnetadp(O) vboxnetflt(O) v boxdrv(O) x86_pkg_temp_thermal kvm_intel kvm
 irqbypass iwldvm iwlwifi ehci_pci ehci_hcd tpm_tis tpm_tis_core tpm CPU: 0
 PID: 2193 Comm: iw Tainted: G           O    4.12.12-gentoo #2 task:
 ffff880186fd5cc0 task.stack: ffffc90001b54000 RIP:
 0010:ieee80211_bss_info_change_notify+0x111/0x190 RSP: 0018:ffffc90001b57a10
 EFLAGS: 00010246 RAX: 0000000000000006 RBX: ffff8801052ce840 RCX:
 0000000000000064 RDX: 00000000fffffffc RSI: 0000000000040000 RDI:
 ffff8801052ce840 RBP: ffffc90001b57a38 R08: 0000000000000062 R09:
 0000000000000000 R10: ffff8802144b5000 R11: ffff880049dc4614 R12:
 0000000000040000 R13: 0000000000000064 R14: ffff8802105f0760 R15:
 ffffc90001b57b48 FS:  00007f92644b4580(0000) GS:ffff88021e200000(0000)
 knlGS:0000000000000000 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00007f9263c109f0 CR3: 00000001df850000 CR4: 00000000000406f0
 Call Trace:
  ieee80211_recalc_txpower+0x33/0x40
  ieee80211_set_tx_power+0x40/0x180
  nl80211_set_wiphy+0x32e/0x950

Reported-by: Peter Große <pegro@friiks.de>
Signed-off-by: Peter Große <pegro@friiks.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ACPI: EC: Fix debugfs_create_*() usage

[ Upstream commit 3522f867c13b63cf62acdf1b8ca5664c549a716a ]

acpi_ec.gpe is "unsigned long", hence treating it as "u32" would expose
the wrong half on big-endian 64-bit systems.  Fix this by changing its
type to "u32" and removing the cast, as all other code already uses u32
or sometimes even only u8.

Fixes: 1195a098168fcacf (ACPI: Provide /sys/kernel/debug/ec/...)
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

irqchip/gic-v3: Fix the driver probe() fail due to disabled GICC entry

[ Upstream commit ebe2f8718007d5a1238bb3cb8141b5bb2b4d5773 ]

The ACPI specification says OS shouldn't attempt to use GICC configuration
parameters if the flag ACPI_MADT_ENABLED is cleared. The ARM64-SMP code
skips the disabled GICC entries but not causing any issue. However the
current GICv3 driver probe bails out causing kernel panic() instead of
skipping the disabled GICC interfaces. This issue happens on systems
where redistributor regions are not in the always-on power domain and
one of GICC interface marked with ACPI_MADT_ENABLED=0.

This patch does the two things to fix the panic.
  - Don't return an error in gic_acpi_match_gicc() for disabled GICC entry.
  - No need to keep GICR region information for disabled GICC entry.

Observed kernel crash on QDF2400 platform GICC entry is disabled.
Kernel crash traces:
  Kernel panic - not syncing: No interrupt controller found.
  CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.13.5 #26
  [<ffff000008087770>] dump_backtrace+0x0/0x218
  [<ffff0000080879dc>] show_stack+0x14/0x20
  [<ffff00000883b078>] dump_stack+0x98/0xb8
  [<ffff0000080c5c14>] panic+0x118/0x26c
  [<ffff000008b62348>] init_IRQ+0x24/0x2c
  [<ffff000008b609fc>] start_kernel+0x230/0x394
  [<ffff000008b601e4>] __primary_switched+0x64/0x6c
  ---[ end Kernel panic - not syncing: No interrupt controller found.

Disabled GICC subtable example:
                   Subtable Type : 0B [Generic Interrupt Controller]
                          Length : 50
                        Reserved : 0000
            CPU Interface Number : 0000003D
                   Processor UID : 0000003D
           Flags (decoded below) : 00000000
               Processor Enabled : 0
 Performance Interrupt Trig Mode : 0
 Virtual GIC Interrupt Trig Mode : 0
        Parking Protocol Version : 00000000
           Performance Interrupt : 00000017
                  Parked Address : 0000000000000000
                    Base Address : 0000000000000000
        Virtual GIC Base Address : 0000000000000000
     Hypervisor GIC Base Address : 0000000000000000
           Virtual GIC Interrupt : 00000019
      Redistributor Base Address : 0000FFFF88F40000
                       ARM MPIDR : 000000000000000D
                Efficiency Class : 00
                        Reserved : 000000
Signed-off-by: Shanker Donthineni <shankerd@codeaurora.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: mpt3sas: Proper handling of set/clear of "ATA command pending" flag.

[ Upstream commit f49d4aed1315a7b766d855f1367142e682b0cc87 ]

1. In IO path, setting of "ATA command pending" flag early before device
   removal, invalid device handle etc., checks causes any new commands
   to be always returned with SAM_STAT_BUSY and when the driver removes
   the drive the SML issues SYNC Cache command and that command is
   always returned with SAM_STAT_BUSY and thus making SYNC Cache command
   to requeued.

2. If the driver gets an ATA PT command for a SATA drive then the driver
   set "ATA command pending" flag in device specific data structure not
   to allow any further commands until the ATA PT command is completed.
   However, after setting the flag if the driver decides to return the
   command back to upper layers without actually issuing to the firmware
   (i.e., returns from qcmd failure return paths) then the corresponding
   flag is not cleared and this prevents the driver from sending any new
   commands to the drive.

This patch fixes above two issues by setting of "ATA command pending"
flag after checking for whether device deleted, invalid device handle,
device busy with task management. And by setting "ATA command pending"
flag to false in all of the qcmd failure return paths after setting the
flag.

Signed-off-by: Chaitra P B <chaitra.basappa@broadcom.com>
Signed-off-by: Suganath Prabu S <suganath-prabu.subramani@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: libiscsi: Allow sd_shutdown on bad transport

[ Upstream commit d754941225a7dbc61f6dd2173fa9498049f9a7ee ]

If, for any reason, userland shuts down iscsi transport interfaces
before proper logouts - like when logging in to LUNs manually, without
logging out on server shutdown, or when automated scripts can't
umount/logout from logged LUNs - kernel will hang forever on its
sd_sync_cache() logic, after issuing the SYNCHRONIZE_CACHE cmd to all
still existent paths.

PID: 1 TASK: ffff8801a69b8000 CPU: 1 COMMAND: "systemd-shutdow"
 #0 [ffff8801a69c3a30] __schedule at ffffffff8183e9ee
 #1 [ffff8801a69c3a80] schedule at ffffffff8183f0d5
 #2 [ffff8801a69c3a98] schedule_timeout at ffffffff81842199
 #3 [ffff8801a69c3b40] io_schedule_timeout at ffffffff8183e604
 #4 [ffff8801a69c3b70] wait_for_completion_io_timeout at ffffffff8183fc6c
 #5 [ffff8801a69c3bd0] blk_execute_rq at ffffffff813cfe10
 #6 [ffff8801a69c3c88] scsi_execute at ffffffff815c3fc7
 #7 [ffff8801a69c3cc8] scsi_execute_req_flags at ffffffff815c60fe
 #8 [ffff8801a69c3d30] sd_sync_cache at ffffffff815d37d7
 #9 [ffff8801a69c3da8] sd_shutdown at ffffffff815d3c3c

This happens because iscsi_eh_cmd_timed_out(), the transport layer
timeout helper, would tell the queue timeout function (scsi_times_out)
to reset the request timer over and over, until the session state is
back to logged in state. Unfortunately, during server shutdown, this
might never happen again.

Other option would be "not to handle" the issue in the transport
layer. That would trigger the error handler logic, which would also need
the session state to be logged in again.

Best option, for such case, is to tell upper layers that the command was
handled during the transport layer error handler helper, marking it as
DID_NO_CONNECT, which will allow completion and inform about the
problem.

After the session was marked as ISCSI_STATE_FAILED, due to the first
timeout during the server shutdown phase, all subsequent cmds will fail
to be queued, allowing upper logic to fail faster.

Signed-off-by: Rafael David Tinoco <rafael.tinoco@canonical.com>
Reviewed-by: Lee Duncan <lduncan@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

spi: sh-msiof: Fix timeout failures for TX-only DMA transfers

[ Upstream commit 89434c3c35081439627baa2225622d5bd12242fe ]

When using RX (with or without TX), the DMA interrupt triggers
completion when the RX FIFO has been emptied, i.e. after the full
transfer has finished.

However, when using TX without RX, the DMA interrupt triggers completion
as soon as the DMA engine has filled the TX FIFO, i.e. before the full
transfer has finished.  Then sh_msiof_modify_ctr_wait() will spin until
the transfer has really finished and the TFSE bit is cleared, for at
most 1 ms.  For slow speeds and/or large transfers, this may cause
timeouts and transfer failures:

    spi_sh_msiof e6e10000.spi: failed to shut down hardware
    74x164 spi2.0: SPI transfer failed: -110
    spi_master spi2: failed to transfer one message from queue
    74x164 spi2.0: Failed writing: -110

Fix this by waiting explicitly until the TX FIFO has been emptied.

Based on a patch in the BSP by Hiromitsu Yamasaki.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ASoC: Intel: cht_bsw_rt5645: Analog Mic support

[ Upstream commit b70b309950418437bbd2a30afd169c4f09dee3e5 ]

Various Cherry Trail boards with a rt5645 codec have an analog mic
connected to IN2P + IN2N. The mic on this boards also needs micbias to
be enabled, on some boards micbias1 is used and on others micbias2, so
we enable both.

This commit adds a new "Int Analog Mic" DAPM widget for this, so that we
do not end up enabling micbias on boards with a digital mic which uses
the already present "Int Mic" widget. Some existing UCM files already
refer to "Int Mic" for their "Internal Analog Microphones" SectionDevice,
but these don't work anyways since they enable the RECMIX BST1 Switch
instead of the BST2 switch.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ASoC: Intel: Skylake: Disable clock gating during firmware and library download

[ Upstream commit d5cc0a1fcbb5ddbef9fdd4c4a978da3254ddbf37 ]

During firmware and library download, sometimes it is observed that
firmware and library download is timed-out resulting into probe failure.

This patch disables dynamic clock gating while firmware and library
download.

Signed-off-by: Pardha Saradhi K <pardha.saradhi.kesapragada@intel.com>
Signed-off-by: Sanyog Kale <sanyog.r.kale@intel.com>
Signed-off-by: Guneshwor Singh <guneshwor.o.singh@intel.com>
Acked-By: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

media: videobuf2-core: don't go out of the buffer range

[ Upstream commit df93dc61b0d8b19a5c9db545cf3fcc24f88dfde4 ]

Currently, there's no check if an invalid buffer range
is passed. However, while testing DVB memory mapped apps,
I got this:

   videobuf2_core: VB: num_buffers -2143943680, buffer 33, index -2143943647
   unable to handle kernel paging request at ffff888b773c0890
   IP: __vb2_queue_alloc+0x134/0x4e0 [videobuf2_core]
   PGD 4142c7067 P4D 4142c7067 PUD 0
   Oops: 0002 [#1] SMP
   Modules linked in: xt_CHECKSUM iptable_mangle ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_nat_ipv4 nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack tun bridge stp llc ebtable_filter ebtables ip6table_filter ip6_tables bluetooth rfkill ecdh_generic binfmt_misc rc_dvbsky sp2 ts2020 intel_rapl x86_pkg_temp_thermal dvb_usb_dvbsky intel_powerclamp dvb_usb_v2 coretemp m88ds3103 kvm_intel i2c_mux dvb_core snd_hda_codec_hdmi crct10dif_pclmul crc32_pclmul videobuf2_vmalloc videobuf2_memops snd_hda_intel ghash_clmulni_intel videobuf2_core snd_hda_codec rc_core mei_me intel_cstate snd_hwdep snd_hda_core videodev intel_uncore snd_pcm mei media tpm_tis tpm_tis_core intel_rapl_perf tpm snd_timer lpc_ich snd soundcore kvm irqbypass libcrc32c i915 i2c_algo_bit drm_kms_helper
   e1000e ptp drm crc32c_intel video pps_core
   CPU: 3 PID: 1776 Comm: dvbv5-zap Not tainted 4.14.0+ #78
   Hardware name:                  /NUC5i7RYB, BIOS RYBDWi35.86A.0364.2017.0511.0949 05/11/2017
   task: ffff88877c73bc80 task.stack: ffffb7c402418000
   RIP: 0010:__vb2_queue_alloc+0x134/0x4e0 [videobuf2_core]
   RSP: 0018:ffffb7c40241bc60 EFLAGS: 00010246
   RAX: 0000000080360421 RBX: 0000000000000021 RCX: 000000000000000a
   RDX: ffffb7c40241bcf4 RSI: ffff888780362c60 RDI: ffff888796d8e130
   RBP: ffffb7c40241bcc8 R08: 0000000000000316 R09: 0000000000000004
   R10: ffff888780362c00 R11: 0000000000000001 R12: 000000000002f000
   R13: ffff8887758be700 R14: 0000000000021000 R15: 0000000000000001
   FS:  00007f2849024740(0000) GS:ffff888796d80000(0000) knlGS:0000000000000000
   CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
   CR2: ffff888b773c0890 CR3: 000000043beb2005 CR4: 00000000003606e0
   Call Trace:
    vb2_core_reqbufs+0x226/0x420 [videobuf2_core]
    dvb_vb2_reqbufs+0x2d/0xc0 [dvb_core]
    dvb_dvr_do_ioctl+0x98/0x1d0 [dvb_core]
    dvb_usercopy+0x53/0x1b0 [dvb_core]
    ? dvb_demux_ioctl+0x20/0x20 [dvb_core]
    ? tty_ldisc_deref+0x16/0x20
    ? tty_write+0x1f9/0x310
    ? process_echoes+0x70/0x70
    dvb_dvr_ioctl+0x15/0x20 [dvb_core]
    do_vfs_ioctl+0xa5/0x600
    SyS_ioctl+0x79/0x90
    entry_SYSCALL_64_fastpath+0x1a/0xa5
   RIP: 0033:0x7f28486f7ea7
   RSP: 002b:00007ffc13b2db18 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
   RAX: ffffffffffffffda RBX: 000055b10fc06130 RCX: 00007f28486f7ea7
   RDX: 00007ffc13b2db48 RSI: 00000000c0086f3c RDI: 0000000000000007
   RBP: 0000000000000203 R08: 000055b10df1e02c R09: 000000000000002e
   R10: 0036b42415108357 R11: 0000000000000246 R12: 0000000000000000
   R13: 00007f2849062f60 R14: 00000000000001f1 R15: 00007ffc13b2da54
   Code: 74 0a 60 8b 0a 48 83 c0 30 48 83 c2 04 89 48 d0 89 48 d4 48 39 f0 75 eb 41 8b 42 08 83 7d d4 01 41 c7 82 ec 01 00 00 ff ff ff ff <4d> 89 94 c5 88 00 00 00 74 14 83 c3 01 41 39 dc 0f 85 f1 fe ff
   RIP: __vb2_queue_alloc+0x134/0x4e0 [videobuf2_core] RSP: ffffb7c40241bc60
   CR2: ffff888b773c0890

So, add a sanity check in order to prevent going past array.

Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

clk: sunxi-ng: a83t: Add M divider to TCON1 clock

[ Upstream commit 7dbc7f5f4904cfddc199af171ea095490a434f15 ]

TCON1 also has M divider, contrary to TCON0. And the mux is only
2 bits wide, instead of 3.

Fixes: 05359be1176b ("clk: sunxi-ng: Add driver for A83T CCU")
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
[wens@csie.org: Add description about mux width difference]
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

hwmon: (ina2xx) Make calibration register value fixed

[ Upstream commit 5d389b125186cf254ad5b8015763ac07c151aea4 ]

Calibration register is used for calculating current register in
hardware according to datasheet:
current = shunt_volt * calib_register / 2048 (ina 226)
current = shunt_volt * calib_register / 4096 (ina 219)

Fix calib_register value to 2048 for ina226 and 4096 for ina 219 in
order to avoid truncation error and provide best precision allowed
by shunt_voltage measurement. Make current scale value follow changes
of shunt_resistor from sysfs as calib_register value is now fixed.

Power_lsb value should also follow shunt_resistor changes as stated in
datasheet:
power_lsb = 25 * current_lsb (ina 226)
power_lsb = 20 * current_lsb (ina 219)

Signed-off-by: Maciej Purski <m.purski@samsung.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

RDMA/cma: Mark end of CMA ID messages

[ Upstream commit e48e5e198fb6ec77c91047a694022f0fefa45292 ]

The commit 1a1c116f3dcf ("RDMA/netlink: Simplify the put_msg and put_attr")
removes nlmsg_len calculation in ibnl_put_attr causing netlink messages and
caused to miss source and destination addresses.

Fixes: 1a1c116f3dcf ("RDMA/netlink: Simplify the put_msg and put_attr")
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>