platform/kernel/linux-rpi3.git
5 years agomlxsw: pci: Ring CQ's doorbell before RDQ's
Ido Schimmel [Fri, 18 Jan 2019 15:57:55 +0000 (15:57 +0000)]
mlxsw: pci: Ring CQ's doorbell before RDQ's

When a packet should be trapped to the CPU the device consumes a WQE
(work queue element) from an RDQ (receive descriptor queue) and copies
the packet to the address specified in the WQE. The device then tries to
post a CQE (completion queue element) that contains various metadata
(e.g., ingress port) about the packet to a CQ (completion queue).

In case the device managed to consume a WQE, but did not manage to post
the corresponding CQE, it will get stuck. This unlikely situation can be
triggered due to the scheme the driver is currently using to process
CQEs.

The driver will consume up to 512 CQEs at a time and after processing
each corresponding WQE it will ring the RDQ's doorbell, letting the
device know that a new WQE was posted for it to consume. Only after
processing all the CQEs (up to 512), the driver will ring the CQ's
doorbell, letting the device know that new ones can be posted.

Fix this by having the driver ring the CQ's doorbell for every processed
CQE, but before ringing the RDQ's doorbell. This guarantees that
whenever we post a new WQE, there is a corresponding CQE available. Copy
the currently processed CQE to prevent the device from overwriting it
with a new CQE after ringing the doorbell.

Note that the driver still arms the CQ only after processing all the
pending CQEs, so that interrupts for this CQ will only be delivered
after the driver finished its processing.

Before commit 8404f6f2e8ed ("mlxsw: pci: Allow to use CQEs of version 1
and version 2") the issue was virtually impossible to trigger since the
number of CQEs was twice the number of WQEs and the number of CQEs
processed at a time was equal to the number of available WQEs.

Fixes: 8404f6f2e8ed ("mlxsw: pci: Allow to use CQEs of version 1 and version 2")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reported-by: Semion Lisyansky <semionl@mellanox.com>
Tested-by: Semion Lisyansky <semionl@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomlxsw: spectrum_fid: Update dummy FID index
Nir Dotan [Fri, 18 Jan 2019 15:57:59 +0000 (15:57 +0000)]
mlxsw: spectrum_fid: Update dummy FID index

[ Upstream commit a11dcd6497915ba79d95ef4fe2541aaac27f6201 ]

When using a tc flower action of egress mirred redirect, the driver adds
an implicit FID setting action. This implicit action sets a dummy FID to
the packet and is used as part of a design for trapping unmatched flows
in OVS.  While this implicit FID setting action is supposed to be a NOP
when a redirect action is added, in Spectrum-2 the FID record is
consulted as the dummy FID index is an 802.1D FID index and the packet
is dropped instead of being redirected.

Set the dummy FID index value to be within 802.1Q range. This satisfies
both Spectrum-1 which ignores the FID and Spectrum-2 which identifies it
as an 802.1Q FID and will then follow the redirect action.

Fixes: c3ab435466d5 ("mlxsw: spectrum: Extend to support Spectrum-2 ASIC")
Signed-off-by: Nir Dotan <nird@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: ipv4: Fix memory leak in network namespace dismantle
Ido Schimmel [Wed, 9 Jan 2019 09:57:39 +0000 (09:57 +0000)]
net: ipv4: Fix memory leak in network namespace dismantle

[ Upstream commit f97f4dd8b3bb9d0993d2491e0f22024c68109184 ]

IPv4 routing tables are flushed in two cases:

1. In response to events in the netdev and inetaddr notification chains
2. When a network namespace is being dismantled

In both cases only routes associated with a dead nexthop group are
flushed. However, a nexthop group will only be marked as dead in case it
is populated with actual nexthops using a nexthop device. This is not
the case when the route in question is an error route (e.g.,
'blackhole', 'unreachable').

Therefore, when a network namespace is being dismantled such routes are
not flushed and leaked [1].

To reproduce:
# ip netns add blue
# ip -n blue route add unreachable 192.0.2.0/24
# ip netns del blue

Fix this by not skipping error routes that are not marked with
RTNH_F_DEAD when flushing the routing tables.

To prevent the flushing of such routes in case #1, add a parameter to
fib_table_flush() that indicates if the table is flushed as part of
namespace dismantle or not.

Note that this problem does not exist in IPv6 since error routes are
associated with the loopback device.

[1]
unreferenced object 0xffff888066650338 (size 56):
  comm "ip", pid 1206, jiffies 4294786063 (age 26.235s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 b0 1c 62 61 80 88 ff ff  ..........ba....
    e8 8b a1 64 80 88 ff ff 00 07 00 08 fe 00 00 00  ...d............
  backtrace:
    [<00000000856ed27d>] inet_rtm_newroute+0x129/0x220
    [<00000000fcdfc00a>] rtnetlink_rcv_msg+0x397/0xa20
    [<00000000cb85801a>] netlink_rcv_skb+0x132/0x380
    [<00000000ebc991d2>] netlink_unicast+0x4c0/0x690
    [<0000000014f62875>] netlink_sendmsg+0x929/0xe10
    [<00000000bac9d967>] sock_sendmsg+0xc8/0x110
    [<00000000223e6485>] ___sys_sendmsg+0x77a/0x8f0
    [<000000002e94f880>] __sys_sendmsg+0xf7/0x250
    [<00000000ccb1fa72>] do_syscall_64+0x14d/0x610
    [<00000000ffbe3dae>] entry_SYSCALL_64_after_hwframe+0x49/0xbe
    [<000000003a8b605b>] 0xffffffffffffffff
unreferenced object 0xffff888061621c88 (size 48):
  comm "ip", pid 1206, jiffies 4294786063 (age 26.235s)
  hex dump (first 32 bytes):
    6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b  kkkkkkkkkkkkkkkk
    6b 6b 6b 6b 6b 6b 6b 6b d8 8e 26 5f 80 88 ff ff  kkkkkkkk..&_....
  backtrace:
    [<00000000733609e3>] fib_table_insert+0x978/0x1500
    [<00000000856ed27d>] inet_rtm_newroute+0x129/0x220
    [<00000000fcdfc00a>] rtnetlink_rcv_msg+0x397/0xa20
    [<00000000cb85801a>] netlink_rcv_skb+0x132/0x380
    [<00000000ebc991d2>] netlink_unicast+0x4c0/0x690
    [<0000000014f62875>] netlink_sendmsg+0x929/0xe10
    [<00000000bac9d967>] sock_sendmsg+0xc8/0x110
    [<00000000223e6485>] ___sys_sendmsg+0x77a/0x8f0
    [<000000002e94f880>] __sys_sendmsg+0xf7/0x250
    [<00000000ccb1fa72>] do_syscall_64+0x14d/0x610
    [<00000000ffbe3dae>] entry_SYSCALL_64_after_hwframe+0x49/0xbe
    [<000000003a8b605b>] 0xffffffffffffffff

Fixes: 8cced9eff1d4 ("[NETNS]: Enable routing configuration in non-initial namespace.")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomlxsw: pci: Increase PCI SW reset timeout
Nir Dotan [Fri, 18 Jan 2019 15:57:56 +0000 (15:57 +0000)]
mlxsw: pci: Increase PCI SW reset timeout

[ Upstream commit d2f372ba0914e5722ac28e15f2ed2db61bcf0e44 ]

Spectrum-2 PHY layer introduces a calibration period which is a part of the
Spectrum-2 firmware boot process. Hence increase the SW timeout waiting for
the firmware to come out of boot. This does not increase system boot time
in cases where the firmware PHY calibration process is done quickly.

Fixes: c3ab435466d5 ("mlxsw: spectrum: Extend to support Spectrum-2 ASIC")
Signed-off-by: Nir Dotan <nird@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agovhost: log dirty page correctly
Jason Wang [Wed, 16 Jan 2019 08:54:42 +0000 (16:54 +0800)]
vhost: log dirty page correctly

[ Upstream commit cc5e710759470bc7f3c61d11fd54586f15fdbdf4 ]

Vhost dirty page logging API is designed to sync through GPA. But we
try to log GIOVA when device IOTLB is enabled. This is wrong and may
lead to missing data after migration.

To solve this issue, when logging with device IOTLB enabled, we will:

1) reuse the device IOTLB translation result of GIOVA->HVA mapping to
   get HVA, for writable descriptor, get HVA through iovec. For used
   ring update, translate its GIOVA to HVA
2) traverse the GPA->HVA mapping to get the possible GPA and log
   through GPA. Pay attention this reverse mapping is not guaranteed
   to be unique, so we should log each possible GPA in this case.

This fix the failure of scp to guest during migration. In -next, we
will probably support passing GIOVA->GPA instead of GIOVA->HVA.

Fixes: 6b1e6cc7855b ("vhost: new device IOTLB API")
Reported-by: Jintack Lim <jintack@cs.columbia.edu>
Cc: Jintack Lim <jintack@cs.columbia.edu>
Signed-off-by: Jason Wang <jasowang@redhat.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>
5 years agoopenvswitch: Avoid OOB read when parsing flow nlattrs
Ross Lagerwall [Mon, 14 Jan 2019 09:16:56 +0000 (09:16 +0000)]
openvswitch: Avoid OOB read when parsing flow nlattrs

[ Upstream commit 04a4af334b971814eedf4e4a413343ad3287d9a9 ]

For nested and variable attributes, the expected length of an attribute
is not known and marked by a negative number.  This results in an OOB
read when the expected length is later used to check if the attribute is
all zeros. Fix this by using the actual length of the attribute rather
than the expected length.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet_sched: refetch skb protocol for each filter
Cong Wang [Sat, 12 Jan 2019 02:55:42 +0000 (18:55 -0800)]
net_sched: refetch skb protocol for each filter

[ Upstream commit cd0c4e70fc0ccfa705cdf55efb27519ce9337a26 ]

Martin reported a set of filters don't work after changing
from reclassify to continue. Looking into the code, it
looks like skb protocol is not always fetched for each
iteration of the filters. But, as demonstrated by Martin,
TC actions could modify skb->protocol, for example act_vlan,
this means we have to refetch skb protocol in each iteration,
rather than using the one we fetch in the beginning of the loop.

This bug is _not_ introduced by commit 3b3ae880266d
("net: sched: consolidate tc_classify{,_compat}"), technically,
if act_vlan is the only action that modifies skb protocol, then
it is commit c7e2b9689ef8 ("sched: introduce vlan action") which
introduced this bug.

Reported-by: Martin Olsson <martin.olsson+netdev@sentorsecurity.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.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>
5 years agonet/sched: act_tunnel_key: fix memory leak in case of action replace
Davide Caratti [Thu, 10 Jan 2019 19:21:02 +0000 (20:21 +0100)]
net/sched: act_tunnel_key: fix memory leak in case of action replace

[ Upstream commit 9174c3df1cd181c14913138d50ccbe539bb08335 ]

running the following TDC test cases:

 7afc - Replace tunnel_key set action with all parameters
 364d - Replace tunnel_key set action with all parameters and cookie

it's possible to trigger kmemleak warnings like:

  unreferenced object 0xffff94797127ab40 (size 192):
  comm "tc", pid 3248, jiffies 4300565293 (age 1006.862s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 c0 93 f9 8a ff ff ff ff  ................
    41 84 ee 89 ff ff ff ff 00 00 00 00 00 00 00 00  A...............
  backtrace:
    [<000000001e85b61c>] tunnel_key_init+0x31d/0x820 [act_tunnel_key]
    [<000000007f3f6ee7>] tcf_action_init_1+0x384/0x4c0
    [<00000000e89e3ded>] tcf_action_init+0x12b/0x1a0
    [<00000000c1c8c0f8>] tcf_action_add+0x73/0x170
    [<0000000095a9fc28>] tc_ctl_action+0x122/0x160
    [<000000004bebeac5>] rtnetlink_rcv_msg+0x263/0x2d0
    [<000000009fd862dd>] netlink_rcv_skb+0x4a/0x110
    [<00000000b55199e7>] netlink_unicast+0x1a0/0x250
    [<000000004996cd21>] netlink_sendmsg+0x2c1/0x3c0
    [<000000004d6a94b4>] sock_sendmsg+0x36/0x40
    [<000000005d9f0208>] ___sys_sendmsg+0x280/0x2f0
    [<00000000dec19023>] __sys_sendmsg+0x5e/0xa0
    [<000000004b82ac81>] do_syscall_64+0x5b/0x180
    [<00000000a0f1209a>] entry_SYSCALL_64_after_hwframe+0x44/0xa9
    [<000000002926b2ab>] 0xffffffffffffffff

when the tunnel_key action is replaced, the kernel forgets to release the
dst metadata: ensure they are released by tunnel_key_init(), the same way
it's done in tunnel_key_release().

Fixes: d0f6dd8a914f4 ("net/sched: Introduce act_tunnel_key")
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: phy: mdio_bus: add missing device_del() in mdiobus_register() error handling
Thomas Petazzoni [Wed, 16 Jan 2019 09:53:58 +0000 (10:53 +0100)]
net: phy: mdio_bus: add missing device_del() in mdiobus_register() error handling

[ Upstream commit e40e2a2e78664fa90ea4b9bdf4a84efce2fea9d9 ]

The current code in __mdiobus_register() doesn't properly handle
failures returned by the devm_gpiod_get_optional() call: it returns
immediately, without unregistering the device that was added by the
call to device_register() earlier in the function.

This leaves a stale device, which then causes a NULL pointer
dereference in the code that handles deferred probing:

[    1.489982] Unable to handle kernel NULL pointer dereference at virtual address 00000074
[    1.498110] pgd = (ptrval)
[    1.500838] [00000074] *pgd=00000000
[    1.504432] Internal error: Oops: 17 [#1] SMP ARM
[    1.509133] Modules linked in:
[    1.512192] CPU: 1 PID: 51 Comm: kworker/1:3 Not tainted 4.20.0-00039-g3b73a4cc8b3e-dirty #99
[    1.520708] Hardware name: Xilinx Zynq Platform
[    1.525261] Workqueue: events deferred_probe_work_func
[    1.530403] PC is at klist_next+0x10/0xfc
[    1.534403] LR is at device_for_each_child+0x40/0x94
[    1.539361] pc : [<c0683fbc>]    lr : [<c0455d90>]    psr: 200e0013
[    1.545628] sp : ceeefe68  ip : 00000001  fp : ffffe000
[    1.550863] r10: 00000000  r9 : c0c66790  r8 : 00000000
[    1.556079] r7 : c0457d44  r6 : 00000000  r5 : ceeefe8c  r4 : cfa2ec78
[    1.562604] r3 : 00000064  r2 : c0457d44  r1 : ceeefe8c  r0 : 00000064
[    1.569129] Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
[    1.576263] Control: 18c5387d  Table: 0ed7804a  DAC: 00000051
[    1.582013] Process kworker/1:3 (pid: 51, stack limit = 0x(ptrval))
[    1.588280] Stack: (0xceeefe68 to 0xceef0000)
[    1.592630] fe60:                   cfa2ec78 c0c03c08 00000000 c0457d44 00000000 c0c66790
[    1.600814] fe80: 00000000 c0455d90 ceeefeac 00000064 00000000 0d7a542e cee9d494 cfa2ec78
[    1.608998] fea0: cfa2ec78 00000000 c0457d44 c0457d7c cee9d494 c0c03c08 00000000 c0455dac
[    1.617182] fec0: cf98ba44 cf926a00 cee9d494 0d7a542e 00000000 cf935a10 cf935a10 cf935a10
[    1.625366] fee0: c0c4e9b8 c0457d7c c0c4e80c 00000001 cf935a10 c0457df4 cf935a10 c0c4e99c
[    1.633550] ff00: c0c4e99c c045a27c c0c4e9c4 ced63f80 cfde8a80 cfdebc00 00000000 c013893c
[    1.641734] ff20: cfde8a80 cfde8a80 c07bd354 ced63f80 ced63f94 cfde8a80 00000008 c0c02d00
[    1.649936] ff40: cfde8a98 cfde8a80 ffffe000 c0139a30 ffffe000 c0c6624a c07bd354 00000000
[    1.658120] ff60: ffffe000 cee9e780 ceebfe00 00000000 ceeee000 ced63f80 c0139788 cf8cdea4
[    1.666304] ff80: cee9e79c c013e598 00000001 ceebfe00 c013e44c 00000000 00000000 00000000
[    1.674488] ffa0: 00000000 00000000 00000000 c01010e8 00000000 00000000 00000000 00000000
[    1.682671] ffc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[    1.690855] ffe0: 00000000 00000000 00000000 00000000 00000013 00000000 00000000 00000000
[    1.699058] [<c0683fbc>] (klist_next) from [<c0455d90>] (device_for_each_child+0x40/0x94)
[    1.707241] [<c0455d90>] (device_for_each_child) from [<c0457d7c>] (device_reorder_to_tail+0x38/0x88)
[    1.716476] [<c0457d7c>] (device_reorder_to_tail) from [<c0455dac>] (device_for_each_child+0x5c/0x94)
[    1.725692] [<c0455dac>] (device_for_each_child) from [<c0457d7c>] (device_reorder_to_tail+0x38/0x88)
[    1.734927] [<c0457d7c>] (device_reorder_to_tail) from [<c0457df4>] (device_pm_move_to_tail+0x28/0x40)
[    1.744235] [<c0457df4>] (device_pm_move_to_tail) from [<c045a27c>] (deferred_probe_work_func+0x58/0x8c)
[    1.753746] [<c045a27c>] (deferred_probe_work_func) from [<c013893c>] (process_one_work+0x210/0x4fc)
[    1.762888] [<c013893c>] (process_one_work) from [<c0139a30>] (worker_thread+0x2a8/0x5c0)
[    1.771072] [<c0139a30>] (worker_thread) from [<c013e598>] (kthread+0x14c/0x154)
[    1.778482] [<c013e598>] (kthread) from [<c01010e8>] (ret_from_fork+0x14/0x2c)
[    1.785689] Exception stack(0xceeeffb0 to 0xceeefff8)
[    1.790739] ffa0:                                     00000000 00000000 00000000 00000000
[    1.798923] ffc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[    1.807107] ffe0: 00000000 00000000 00000000 00000000 00000013 00000000
[    1.813724] Code: e92d47f0 e1a05000 e8900048 e1a00003 (e5937010)
[    1.819844] ---[ end trace 3c2c0c8b65399ec9 ]---

The actual error that we had from devm_gpiod_get_optional() was
-EPROBE_DEFER, due to the GPIO being provided by a driver that is
probed later than the Ethernet controller driver.

To fix this, we simply add the missing device_del() invocation in the
error path.

Fixes: 69226896ad636 ("mdio_bus: Issue GPIO RESET to PHYs")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: phy: marvell: Errata for mv88e6390 internal PHYs
Andrew Lunn [Thu, 10 Jan 2019 21:48:36 +0000 (22:48 +0100)]
net: phy: marvell: Errata for mv88e6390 internal PHYs

[ Upstream commit 8cbcdc1a51999ca81db2956608b917aacd28d837 ]

The VOD can be out of spec, unless some magic value is poked into an
undocumented register in an undocumented page.

Fixes: e4cf8a38fc0d ("net: phy: Marvell: Add mv88e6390 internal PHY")
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: Fix usage of pskb_trim_rcsum
Ross Lagerwall [Thu, 17 Jan 2019 15:34:38 +0000 (15:34 +0000)]
net: Fix usage of pskb_trim_rcsum

[ Upstream commit 6c57f0458022298e4da1729c67bd33ce41c14e7a ]

In certain cases, pskb_trim_rcsum() may change skb pointers.
Reinitialize header pointers afterwards to avoid potential
use-after-frees. Add a note in the documentation of
pskb_trim_rcsum(). Found by KASAN.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agonet: bridge: Fix ethernet header pointer before check skb forwardable
Yunjian Wang [Thu, 17 Jan 2019 01:46:41 +0000 (09:46 +0800)]
net: bridge: Fix ethernet header pointer before check skb forwardable

[ Upstream commit 28c1382fa28f2e2d9d0d6f25ae879b5af2ecbd03 ]

The skb header should be set to ethernet header before using
is_skb_forwardable. Because the ethernet header length has been
considered in is_skb_forwardable(including dev->hard_header_len
length).

To reproduce the issue:
1, add 2 ports on linux bridge br using following commands:
$ brctl addbr br
$ brctl addif br eth0
$ brctl addif br eth1
2, the MTU of eth0 and eth1 is 1500
3, send a packet(Data 1480, UDP 8, IP 20, Ethernet 14, VLAN 4)
from eth0 to eth1

So the expect result is packet larger than 1500 cannot pass through
eth0 and eth1. But currently, the packet passes through success, it
means eth1's MTU limit doesn't take effect.

Fixes: f6367b4660dd ("bridge: use is_skb_forwardable in forward path")
Cc: bridge@lists.linux-foundation.org
Cc: Nkolay Aleksandrov <nikolay@cumulusnetworks.com>
Cc: Roopa Prabhu <roopa@cumulusnetworks.com>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoamd-xgbe: Fix mdio access for non-zero ports and clause 45 PHYs
Lendacky, Thomas [Thu, 17 Jan 2019 14:20:14 +0000 (14:20 +0000)]
amd-xgbe: Fix mdio access for non-zero ports and clause 45 PHYs

[ Upstream commit 5ab3121beeb76aa6090195b67d237115860dd9ec ]

The XGBE hardware has support for performing MDIO operations using an
MDIO command request. The driver mistakenly uses the mdio port address
as the MDIO command request device address instead of the MDIO command
request port address. Additionally, the driver does not properly check
for and create a clause 45 MDIO command.

Check the supplied MDIO register to determine if the request is a clause
45 operation (MII_ADDR_C45). For a clause 45 operation, extract the device
address and register number from the supplied MDIO register and use them
to set the MDIO command request device address and register number fields.
For a clause 22 operation, the MDIO request device address is set to zero
and the MDIO command request register number is set to the supplied MDIO
register. In either case, the supplied MDIO port address is used as the
MDIO command request port address.

Fixes: 732f2ab7afb9 ("amd-xgbe: Add support for MDIO attached PHYs")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Tested-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoLinux 4.19.18 v4.19.18
Greg Kroah-Hartman [Sat, 26 Jan 2019 08:32:45 +0000 (09:32 +0100)]
Linux 4.19.18

5 years agoipmi: Don't initialize anything in the core until something uses it
Corey Minyard [Thu, 20 Dec 2018 22:50:23 +0000 (16:50 -0600)]
ipmi: Don't initialize anything in the core until something uses it

commit 913a89f009d98c85a902d718cd54bb32ab11d167 upstream.

The IPMI driver was recently modified to use SRCU, but it turns out
this uses a chunk of percpu memory, even if IPMI is never used.

So modify thing to on initialize on the first use.  There was already
code to sort of handle this for handling init races, so piggy back
on top of that, and simplify it in the process.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Reported-by: Tejun Heo <tj@kernel.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: stable@vger.kernel.org # 4.18
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoipmi:ssif: Fix handling of multi-part return messages
Corey Minyard [Fri, 16 Nov 2018 15:59:21 +0000 (09:59 -0600)]
ipmi:ssif: Fix handling of multi-part return messages

commit 7d6380cd40f7993f75c4bde5b36f6019237e8719 upstream.

The block number was not being compared right, it was off by one
when checking the response.

Some statistics wouldn't be incremented properly in some cases.

Check to see if that middle-part messages always have 31 bytes of
data.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: stable@vger.kernel.org # 4.4
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoipmi: Prevent use-after-free in deliver_response
Fred Klassen [Sat, 19 Jan 2019 22:28:18 +0000 (14:28 -0800)]
ipmi: Prevent use-after-free in deliver_response

commit 479d6b39b9e0d2de648ebf146f23a1e40962068f upstream.

Some IPMI modules (e.g. ibmpex_msg_handler()) will have ipmi_usr_hdlr
handlers that call ipmi_free_recv_msg() directly. This will essentially
kfree(msg), leading to use-after-free.

This does not happen in the ipmi_devintf module, which will queue the
message and run ipmi_free_recv_msg() later.

BUG: KASAN: use-after-free in deliver_response+0x12f/0x1b0
Read of size 8 at addr ffff888a7bf20018 by task ksoftirqd/3/27
CPU: 3 PID: 27 Comm: ksoftirqd/3 Tainted: G           O      4.19.11-amd64-ani99-debug #12.0.1.601133+pv
Hardware name: AppNeta r1000/X11SPW-TF, BIOS 2.1a-AP 09/17/2018
Call Trace:
dump_stack+0x92/0xeb
print_address_description+0x73/0x290
kasan_report+0x258/0x380
deliver_response+0x12f/0x1b0
? ipmi_free_recv_msg+0x50/0x50
deliver_local_response+0xe/0x50
handle_one_recv_msg+0x37a/0x21d0
handle_new_recv_msgs+0x1ce/0x440
...

Allocated by task 9885:
kasan_kmalloc+0xa0/0xd0
kmem_cache_alloc_trace+0x116/0x290
ipmi_alloc_recv_msg+0x28/0x70
i_ipmi_request+0xb4a/0x1640
ipmi_request_settime+0x1b8/0x1e0
...

Freed by task 27:
__kasan_slab_free+0x12e/0x180
kfree+0xe9/0x280
deliver_response+0x122/0x1b0
deliver_local_response+0xe/0x50
handle_one_recv_msg+0x37a/0x21d0
handle_new_recv_msgs+0x1ce/0x440
tasklet_action_common.isra.19+0xc4/0x250
__do_softirq+0x11f/0x51f

Fixes: e86ee2d44b44 ("ipmi: Rework locking and shutdown for hot remove")
Cc: stable@vger.kernel.org # 4.18
Signed-off-by: Fred Klassen <fklassen@appneta.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoipmi: msghandler: Fix potential Spectre v1 vulnerabilities
Gustavo A. R. Silva [Wed, 9 Jan 2019 23:39:06 +0000 (17:39 -0600)]
ipmi: msghandler: Fix potential Spectre v1 vulnerabilities

commit a7102c7461794a5bb31af24b08e9e0f50038897a upstream.

channel and addr->channel are indirectly controlled by user-space,
hence leading to a potential exploitation of the Spectre variant 1
vulnerability.

These issues were detected with the help of Smatch:

drivers/char/ipmi/ipmi_msghandler.c:1381 ipmi_set_my_address() warn: potential spectre issue 'user->intf->addrinfo' [w] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:1401 ipmi_get_my_address() warn: potential spectre issue 'user->intf->addrinfo' [r] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:1421 ipmi_set_my_LUN() warn: potential spectre issue 'user->intf->addrinfo' [w] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:1441 ipmi_get_my_LUN() warn: potential spectre issue 'user->intf->addrinfo' [r] (local cap)
drivers/char/ipmi/ipmi_msghandler.c:2260 check_addr() warn: potential spectre issue 'intf->addrinfo' [r] (local cap)

Fix this by sanitizing channel and addr->channel before using them to
index user->intf->addrinfo and intf->addrinfo, correspondingly.

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

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

Cc: stable@vger.kernel.org
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoipmi: fix use-after-free of user->release_barrier.rda
Yang Yingliang [Wed, 16 Jan 2019 05:33:22 +0000 (13:33 +0800)]
ipmi: fix use-after-free of user->release_barrier.rda

commit 77f8269606bf95fcb232ee86f6da80886f1dfae8 upstream.

When we do the following test, we got oops in ipmi_msghandler driver
while((1))
do
service ipmievd restart & service ipmievd restart
done

---------------------------------------------------------------
[  294.230186] Unable to handle kernel paging request at virtual address 0000803fea6ea008
[  294.230188] Mem abort info:
[  294.230190]   ESR = 0x96000004
[  294.230191]   Exception class = DABT (current EL), IL = 32 bits
[  294.230193]   SET = 0, FnV = 0
[  294.230194]   EA = 0, S1PTW = 0
[  294.230195] Data abort info:
[  294.230196]   ISV = 0, ISS = 0x00000004
[  294.230197]   CM = 0, WnR = 0
[  294.230199] user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000a1c1b75a
[  294.230201] [0000803fea6ea008] pgd=0000000000000000
[  294.230204] Internal error: Oops: 96000004 [#1] SMP
[  294.235211] Modules linked in: nls_utf8 isofs rpcrdma ib_iser ib_srpt target_core_mod ib_srp scsi_transport_srp ib_ipoib rdma_ucm ib_umad rdma_cm ib_cm iw_cm dm_mirror dm_region_hash dm_log dm_mod aes_ce_blk crypto_simd cryptd aes_ce_cipher ghash_ce sha2_ce ses sha256_arm64 sha1_ce hibmc_drm hisi_sas_v2_hw enclosure sg hisi_sas_main sbsa_gwdt ip_tables mlx5_ib ib_uverbs marvell ib_core mlx5_core ixgbe ipmi_si mdio hns_dsaf ipmi_devintf ipmi_msghandler hns_enet_drv hns_mdio
[  294.277745] CPU: 3 PID: 0 Comm: swapper/3 Kdump: loaded Not tainted 5.0.0-rc2+ #113
[  294.285511] Hardware name: Huawei TaiShan 2280 /BC11SPCD, BIOS 1.37 11/21/2017
[  294.292835] pstate: 80000005 (Nzcv daif -PAN -UAO)
[  294.297695] pc : __srcu_read_lock+0x38/0x58
[  294.301940] lr : acquire_ipmi_user+0x2c/0x70 [ipmi_msghandler]
[  294.307853] sp : ffff00001001bc80
[  294.311208] x29: ffff00001001bc80 x28: ffff0000117e5000
[  294.316594] x27: 0000000000000000 x26: dead000000000100
[  294.321980] x25: dead000000000200 x24: ffff803f6bd06800
[  294.327366] x23: 0000000000000000 x22: 0000000000000000
[  294.332752] x21: ffff00001001bd04 x20: ffff80df33d19018
[  294.338137] x19: ffff80df33d19018 x18: 0000000000000000
[  294.343523] x17: 0000000000000000 x16: 0000000000000000
[  294.348908] x15: 0000000000000000 x14: 0000000000000002
[  294.354293] x13: 0000000000000000 x12: 0000000000000000
[  294.359679] x11: 0000000000000000 x10: 0000000000100000
[  294.365065] x9 : 0000000000000000 x8 : 0000000000000004
[  294.370451] x7 : 0000000000000000 x6 : ffff80df34558678
[  294.375836] x5 : 000000000000000c x4 : 0000000000000000
[  294.381221] x3 : 0000000000000001 x2 : 0000803fea6ea000
[  294.386607] x1 : 0000803fea6ea008 x0 : 0000000000000001
[  294.391994] Process swapper/3 (pid: 0, stack limit = 0x0000000083087293)
[  294.398791] Call trace:
[  294.401266]  __srcu_read_lock+0x38/0x58
[  294.405154]  acquire_ipmi_user+0x2c/0x70 [ipmi_msghandler]
[  294.410716]  deliver_response+0x80/0xf8 [ipmi_msghandler]
[  294.416189]  deliver_local_response+0x28/0x68 [ipmi_msghandler]
[  294.422193]  handle_one_recv_msg+0x158/0xcf8 [ipmi_msghandler]
[  294.432050]  handle_new_recv_msgs+0xc0/0x210 [ipmi_msghandler]
[  294.441984]  smi_recv_tasklet+0x8c/0x158 [ipmi_msghandler]
[  294.451618]  tasklet_action_common.isra.5+0x88/0x138
[  294.460661]  tasklet_action+0x2c/0x38
[  294.468191]  __do_softirq+0x120/0x2f8
[  294.475561]  irq_exit+0x134/0x140
[  294.482445]  __handle_domain_irq+0x6c/0xc0
[  294.489954]  gic_handle_irq+0xb8/0x178
[  294.497037]  el1_irq+0xb0/0x140
[  294.503381]  arch_cpu_idle+0x34/0x1a8
[  294.510096]  do_idle+0x1d4/0x290
[  294.516322]  cpu_startup_entry+0x28/0x30
[  294.523230]  secondary_start_kernel+0x184/0x1d0
[  294.530657] Code: d538d082 d2800023 8b010c81 8b020021 (c85f7c25)
[  294.539746] ---[ end trace 8a7a880dee570b29 ]---
[  294.547341] Kernel panic - not syncing: Fatal exception in interrupt
[  294.556837] SMP: stopping secondary CPUs
[  294.563996] Kernel Offset: disabled
[  294.570515] CPU features: 0x002,21006008
[  294.577638] Memory Limit: none
[  294.587178] Starting crashdump kernel...
[  294.594314] Bye!

Because the user->release_barrier.rda is freed in ipmi_destroy_user(), but
the refcount is not zero, when acquire_ipmi_user() uses user->release_barrier.rda
in __srcu_read_lock(), it causes oops.
Fix this by calling cleanup_srcu_struct() when the refcount is zero.

Fixes: e86ee2d44b44 ("ipmi: Rework locking and shutdown for hot remove")
Cc: stable@vger.kernel.org # 4.18
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoBluetooth: Fix unnecessary error message for HCI request completion
Johan Hedberg [Tue, 27 Nov 2018 09:37:46 +0000 (11:37 +0200)]
Bluetooth: Fix unnecessary error message for HCI request completion

commit 1629db9c75342325868243d6bca5853017d91cf8 upstream.

In case a command which completes in Command Status was sent using the
hci_cmd_send-family of APIs there would be a misleading error in the
hci_get_cmd_complete function, since the code would be trying to fetch
the Command Complete parameters when there are none.

Avoid the misleading error and silently bail out from the function in
case the received event is a command status.

Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Acked-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Tested-by Adam Ford <aford173@gmail.com> #4.19.16
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoiwlwifi: mvm: Send LQ command as async when necessary
Avraham Stern [Thu, 3 May 2018 12:02:16 +0000 (15:02 +0300)]
iwlwifi: mvm: Send LQ command as async when necessary

commit 3baf7528d6f832b28622d1ddadd2e47f6c2b5e08 upstream.

The parameter that indicated whether the LQ command should be sent
as sync or async was removed, causing the LQ command to be sent as
sync from interrupt context (e.g. from the RX path). This resulted
in a kernel warning: "scheduling while atomic" and failing to send
the LQ command, which ultimately leads to a queue hang.

Fix it by adding back the required parameter to send the command as
sync only when it is allowed.

Fixes: d94c5a820d10 ("iwlwifi: mvm: open BA session only when sta is authorized")
Signed-off-by: Avraham Stern <avraham.stern@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agomm, proc: be more verbose about unstable VMA flags in /proc/<pid>/smaps
Michal Hocko [Fri, 28 Dec 2018 08:38:17 +0000 (00:38 -0800)]
mm, proc: be more verbose about unstable VMA flags in /proc/<pid>/smaps

[ Upstream commit 7550c6079846a24f30d15ac75a941c8515dbedfb ]

Patch series "THP eligibility reporting via proc".

This series of three patches aims at making THP eligibility reporting much
more robust and long term sustainable.  The trigger for the change is a
regression report [2] and the long follow up discussion.  In short the
specific application didn't have good API to query whether a particular
mapping can be backed by THP so it has used VMA flags to workaround that.
These flags represent a deep internal state of VMAs and as such they
should be used by userspace with a great deal of caution.

A similar has happened for [3] when users complained that VM_MIXEDMAP is
no longer set on DAX mappings.  Again a lack of a proper API led to an
abuse.

The first patch in the series tries to emphasise that that the semantic of
flags might change and any application consuming those should be really
careful.

The remaining two patches provide a more suitable interface to address [2]
and provide a consistent API to query the THP status both for each VMA and
process wide as well.  [1]

http://lkml.kernel.org/r/20181120103515.25280-1-mhocko@kernel.org [2]
http://lkml.kernel.org/r/http://lkml.kernel.org/r/alpine.DEB.2.21.1809241054050.224429@chino.kir.corp.google.com
[3] http://lkml.kernel.org/r/20181002100531.GC4135@quack2.suse.cz

This patch (of 3):

Even though vma flags exported via /proc/<pid>/smaps are explicitly
documented to be not guaranteed for future compatibility the warning
doesn't go far enough because it doesn't mention semantic changes to those
flags.  And they are important as well because these flags are a deep
implementation internal to the MM code and the semantic might change at
any time.

Let's consider two recent examples:
http://lkml.kernel.org/r/20181002100531.GC4135@quack2.suse.cz
: commit e1fb4a086495 "dax: remove VM_MIXEDMAP for fsdax and device dax" has
: removed VM_MIXEDMAP flag from DAX VMAs. Now our testing shows that in the
: mean time certain customer of ours started poking into /proc/<pid>/smaps
: and looks at VMA flags there and if VM_MIXEDMAP is missing among the VMA
: flags, the application just fails to start complaining that DAX support is
: missing in the kernel.

http://lkml.kernel.org/r/alpine.DEB.2.21.1809241054050.224429@chino.kir.corp.google.com
: Commit 1860033237d4 ("mm: make PR_SET_THP_DISABLE immediately active")
: introduced a regression in that userspace cannot always determine the set
: of vmas where thp is ineligible.
: Userspace relies on the "nh" flag being emitted as part of /proc/pid/smaps
: to determine if a vma is eligible to be backed by hugepages.
: Previous to this commit, prctl(PR_SET_THP_DISABLE, 1) would cause thp to
: be disabled and emit "nh" as a flag for the corresponding vmas as part of
: /proc/pid/smaps.  After the commit, thp is disabled by means of an mm
: flag and "nh" is not emitted.
: This causes smaps parsing libraries to assume a vma is eligible for thp
: and ends up puzzling the user on why its memory is not backed by thp.

In both cases userspace was relying on a semantic of a specific VMA flag.
The primary reason why that happened is a lack of a proper interface.
While this has been worked on and it will be fixed properly, it seems that
our wording could see some refinement and be more vocal about semantic
aspect of these flags as well.

Link: http://lkml.kernel.org/r/20181211143641.3503-2-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Jan Kara <jack@suse.cz>
Acked-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Paul Oppenheimer <bepvte@gmail.com>
Cc: William Kucharski <william.kucharski@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agouserfaultfd: clear flag if remap event not enabled
Peter Xu [Fri, 28 Dec 2018 08:38:47 +0000 (00:38 -0800)]
userfaultfd: clear flag if remap event not enabled

[ Upstream commit 3cfd22be0ad663248fadfc8f6ffa3e255c394552 ]

When the process being tracked does mremap() without
UFFD_FEATURE_EVENT_REMAP on the corresponding tracking uffd file handle,
we should not generate the remap event, and at the same time we should
clear all the uffd flags on the new VMA.  Without this patch, we can still
have the VM_UFFD_MISSING|VM_UFFD_WP flags on the new VMA even the fault
handling process does not even know the existance of the VMA.

Link: http://lkml.kernel.org/r/20181211053409.20317-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Reviewed-by: William Kucharski <william.kucharski@oracle.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Kirill A. Shutemov <kirill@shutemov.name>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Cc: Pravin Shedge <pravin.shedge4linux@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agomm/swap: use nr_node_ids for avail_lists in swap_info_struct
Aaron Lu [Fri, 28 Dec 2018 08:34:39 +0000 (00:34 -0800)]
mm/swap: use nr_node_ids for avail_lists in swap_info_struct

[ Upstream commit 66f71da9dd38af17dc17209cdde7987d4679a699 ]

Since a2468cc9bfdf ("swap: choose swap device according to numa node"),
avail_lists field of swap_info_struct is changed to an array with
MAX_NUMNODES elements.  This made swap_info_struct size increased to 40KiB
and needs an order-4 page to hold it.

This is not optimal in that:
1 Most systems have way less than MAX_NUMNODES(1024) nodes so it
  is a waste of memory;
2 It could cause swapon failure if the swap device is swapped on
  after system has been running for a while, due to no order-4
  page is available as pointed out by Vasily Averin.

Solve the above two issues by using nr_node_ids(which is the actual
possible node number the running system has) for avail_lists instead of
MAX_NUMNODES.

nr_node_ids is unknown at compile time so can't be directly used when
declaring this array.  What I did here is to declare avail_lists as zero
element array and allocate space for it when allocating space for
swap_info_struct.  The reason why keep using array but not pointer is
plist_for_each_entry needs the field to be part of the struct, so pointer
will not work.

This patch is on top of Vasily Averin's fix commit.  I think the use of
kvzalloc for swap_info_struct is still needed in case nr_node_ids is
really big on some systems.

Link: http://lkml.kernel.org/r/20181115083847.GA11129@intel.com
Signed-off-by: Aaron Lu <aaron.lu@intel.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Vasily Averin <vvs@virtuozzo.com>
Cc: Huang Ying <ying.huang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agomm/page-writeback.c: don't break integrity writeback on ->writepage() error
Brian Foster [Fri, 28 Dec 2018 08:37:20 +0000 (00:37 -0800)]
mm/page-writeback.c: don't break integrity writeback on ->writepage() error

[ Upstream commit 3fa750dcf29e8606e3969d13d8e188cc1c0f511d ]

write_cache_pages() is used in both background and integrity writeback
scenarios by various filesystems.  Background writeback is mostly
concerned with cleaning a certain number of dirty pages based on various
mm heuristics.  It may not write the full set of dirty pages or wait for
I/O to complete.  Integrity writeback is responsible for persisting a set
of dirty pages before the writeback job completes.  For example, an
fsync() call must perform integrity writeback to ensure data is on disk
before the call returns.

write_cache_pages() unconditionally breaks out of its processing loop in
the event of a ->writepage() error.  This is fine for background
writeback, which had no strict requirements and will eventually come
around again.  This can cause problems for integrity writeback on
filesystems that might need to clean up state associated with failed page
writeouts.  For example, XFS performs internal delayed allocation
accounting before returning a ->writepage() error, where applicable.  If
the current writeback happens to be associated with an unmount and
write_cache_pages() completes the writeback prematurely due to error, the
filesystem is unmounted in an inconsistent state if dirty+delalloc pages
still exist.

To handle this problem, update write_cache_pages() to always process the
full set of pages for integrity writeback regardless of ->writepage()
errors.  Save the first encountered error and return it to the caller once
complete.  This facilitates XFS (or any other fs that expects integrity
writeback to process the entire set of dirty pages) to clean up its
internal state completely in the event of persistent mapping errors.
Background writeback continues to exit on the first error encountered.

[akpm@linux-foundation.org: fix typo in comment]
Link: http://lkml.kernel.org/r/20181116134304.32440-1-bfoster@redhat.com
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoocfs2: fix panic due to unrecovered local alloc
Junxiao Bi [Fri, 28 Dec 2018 08:32:50 +0000 (00:32 -0800)]
ocfs2: fix panic due to unrecovered local alloc

[ Upstream commit 532e1e54c8140188e192348c790317921cb2dc1c ]

mount.ocfs2 ignore the inconsistent error that journal is clean but
local alloc is unrecovered.  After mount, local alloc not empty, then
reserver cluster didn't alloc a new local alloc window, reserveration
map is empty(ocfs2_reservation_map.m_bitmap_len = 0), that triggered the
following panic.

This issue was reported at

  https://oss.oracle.com/pipermail/ocfs2-devel/2015-May/010854.html

and was advised to fixed during mount.  But this is a very unusual
inconsistent state, usually journal dirty flag should be cleared at the
last stage of umount until every other things go right.  We may need do
further debug to check that.  Any way to avoid possible futher
corruption, mount should be abort and fsck should be run.

  (mount.ocfs2,1765,1):ocfs2_load_local_alloc:353 ERROR: Local alloc hasn't been recovered!
  found = 6518, set = 6518, taken = 8192, off = 15912372
  ocfs2: Mounting device (202,64) on (node 0, slot 3) with ordered data mode.
  o2dlm: Joining domain 89CEAC63CC4F4D03AC185B44E0EE0F3F ( 0 1 2 3 4 5 6 8 ) 8 nodes
  ocfs2: Mounting device (202,80) on (node 0, slot 3) with ordered data mode.
  o2hb: Region 89CEAC63CC4F4D03AC185B44E0EE0F3F (xvdf) is now a quorum device
  o2net: Accepted connection from node yvwsoa17p (num 7) at 172.22.77.88:7777
  o2dlm: Node 7 joins domain 64FE421C8C984E6D96ED12C55FEE2435 ( 0 1 2 3 4 5 6 7 8 ) 9 nodes
  o2dlm: Node 7 joins domain 89CEAC63CC4F4D03AC185B44E0EE0F3F ( 0 1 2 3 4 5 6 7 8 ) 9 nodes
  ------------[ cut here ]------------
  kernel BUG at fs/ocfs2/reservations.c:507!
  invalid opcode: 0000 [#1] SMP
  Modules linked in: ocfs2 rpcsec_gss_krb5 auth_rpcgss nfsv4 nfs fscache lockd grace ocfs2_dlmfs ocfs2_stack_o2cb ocfs2_dlm ocfs2_nodemanager ocfs2_stackglue configfs sunrpc ipt_REJECT nf_reject_ipv4 nf_conntrack_ipv4 nf_defrag_ipv4 iptable_filter ip_tables ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables ib_ipoib rdma_ucm ib_ucm ib_uverbs ib_umad rdma_cm ib_cm iw_cm ib_sa ib_mad ib_core ib_addr ipv6 ovmapi ppdev parport_pc parport xen_netfront fb_sys_fops sysimgblt sysfillrect syscopyarea acpi_cpufreq pcspkr i2c_piix4 i2c_core sg ext4 jbd2 mbcache2 sr_mod cdrom xen_blkfront pata_acpi ata_generic ata_piix floppy dm_mirror dm_region_hash dm_log dm_mod
  CPU: 0 PID: 4349 Comm: startWebLogic.s Not tainted 4.1.12-124.19.2.el6uek.x86_64 #2
  Hardware name: Xen HVM domU, BIOS 4.4.4OVM 09/06/2018
  task: ffff8803fb04e200 ti: ffff8800ea4d8000 task.ti: ffff8800ea4d8000
  RIP: 0010:[<ffffffffa05e96a8>]  [<ffffffffa05e96a8>] __ocfs2_resv_find_window+0x498/0x760 [ocfs2]
  Call Trace:
    ocfs2_resmap_resv_bits+0x10d/0x400 [ocfs2]
    ocfs2_claim_local_alloc_bits+0xd0/0x640 [ocfs2]
    __ocfs2_claim_clusters+0x178/0x360 [ocfs2]
    ocfs2_claim_clusters+0x1f/0x30 [ocfs2]
    ocfs2_convert_inline_data_to_extents+0x634/0xa60 [ocfs2]
    ocfs2_write_begin_nolock+0x1c6/0x1da0 [ocfs2]
    ocfs2_write_begin+0x13e/0x230 [ocfs2]
    generic_perform_write+0xbf/0x1c0
    __generic_file_write_iter+0x19c/0x1d0
    ocfs2_file_write_iter+0x589/0x1360 [ocfs2]
    __vfs_write+0xb8/0x110
    vfs_write+0xa9/0x1b0
    SyS_write+0x46/0xb0
    system_call_fastpath+0x18/0xd7
  Code: ff ff 8b 75 b8 39 75 b0 8b 45 c8 89 45 98 0f 84 e5 fe ff ff 45 8b 74 24 18 41 8b 54 24 1c e9 56 fc ff ff 85 c0 0f 85 48 ff ff ff <0f> 0b 48 8b 05 cf c3 de ff 48 ba 00 00 00 00 00 00 00 10 48 85
  RIP   __ocfs2_resv_find_window+0x498/0x760 [ocfs2]
   RSP <ffff8800ea4db668>
  ---[ end trace 566f07529f2edf3c ]---
  Kernel panic - not syncing: Fatal exception
  Kernel Offset: disabled

Link: http://lkml.kernel.org/r/20181121020023.3034-2-junxiao.bi@oracle.com
Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com>
Reviewed-by: Yiwen Jiang <jiangyiwen@huawei.com>
Acked-by: Joseph Qi <jiangqi903@gmail.com>
Cc: Jun Piao <piaojun@huawei.com>
Cc: Mark Fasheh <mfasheh@versity.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Changwei Ge <ge.changwei@h3c.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoiomap: don't search past page end in iomap_is_partially_uptodate
Eric Sandeen [Fri, 21 Dec 2018 16:42:50 +0000 (08:42 -0800)]
iomap: don't search past page end in iomap_is_partially_uptodate

[ Upstream commit 3cc31fa65d85610574c0f6a474e89f4c419923d5 ]

iomap_is_partially_uptodate() is intended to check wither blocks within
the selected range of a not-uptodate page are uptodate; if the range we
care about is up to date, it's an optimization.

However, the iomap implementation continues to check all blocks up to
from+count, which is beyond the page, and can even be well beyond the
iop->uptodate bitmap.

I think the worst that will happen is that we may eventually find a zero
bit and return "not partially uptodate" when it would have otherwise
returned true, and skip the optimization.  Still, it's clearly an invalid
memory access that must be fixed.

So: fix this by limiting the search to within the page as is done in the
non-iomap variant, block_is_partially_uptodate().

Zorro noticed thiswhen KASAN went off for 512 byte blocks on a 64k
page system:

 BUG: KASAN: slab-out-of-bounds in iomap_is_partially_uptodate+0x1a0/0x1e0
 Read of size 8 at addr ffff800120c3a318 by task fsstress/22337

Reported-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: megaraid: fix out-of-bound array accesses
Qian Cai [Thu, 13 Dec 2018 13:27:27 +0000 (08:27 -0500)]
scsi: megaraid: fix out-of-bound array accesses

[ Upstream commit c7a082e4242fd8cd21a441071e622f87c16bdacc ]

UBSAN reported those with MegaRAID SAS-3 3108,

[   77.467308] UBSAN: Undefined behaviour in drivers/scsi/megaraid/megaraid_sas_fp.c:117:32
[   77.475402] index 255 is out of range for type 'MR_LD_SPAN_MAP [1]'
[   77.481677] CPU: 16 PID: 333 Comm: kworker/16:1 Not tainted 4.20.0-rc5+ #1
[   77.488556] Hardware name: Huawei TaiShan 2280 /BC11SPCD, BIOS 1.50 06/01/2018
[   77.495791] Workqueue: events work_for_cpu_fn
[   77.500154] Call trace:
[   77.502610]  dump_backtrace+0x0/0x2c8
[   77.506279]  show_stack+0x24/0x30
[   77.509604]  dump_stack+0x118/0x19c
[   77.513098]  ubsan_epilogue+0x14/0x60
[   77.516765]  __ubsan_handle_out_of_bounds+0xfc/0x13c
[   77.521767]  mr_update_load_balance_params+0x150/0x158 [megaraid_sas]
[   77.528230]  MR_ValidateMapInfo+0x2cc/0x10d0 [megaraid_sas]
[   77.533825]  megasas_get_map_info+0x244/0x2f0 [megaraid_sas]
[   77.539505]  megasas_init_adapter_fusion+0x9b0/0xf48 [megaraid_sas]
[   77.545794]  megasas_init_fw+0x1ab4/0x3518 [megaraid_sas]
[   77.551212]  megasas_probe_one+0x2c4/0xbe0 [megaraid_sas]
[   77.556614]  local_pci_probe+0x7c/0xf0
[   77.560365]  work_for_cpu_fn+0x34/0x50
[   77.564118]  process_one_work+0x61c/0xf08
[   77.568129]  worker_thread+0x534/0xa70
[   77.571882]  kthread+0x1c8/0x1d0
[   77.575114]  ret_from_fork+0x10/0x1c

[   89.240332] UBSAN: Undefined behaviour in drivers/scsi/megaraid/megaraid_sas_fp.c:117:32
[   89.248426] index 255 is out of range for type 'MR_LD_SPAN_MAP [1]'
[   89.254700] CPU: 16 PID: 95 Comm: kworker/u130:0 Not tainted 4.20.0-rc5+ #1
[   89.261665] Hardware name: Huawei TaiShan 2280 /BC11SPCD, BIOS 1.50 06/01/2018
[   89.268903] Workqueue: events_unbound async_run_entry_fn
[   89.274222] Call trace:
[   89.276680]  dump_backtrace+0x0/0x2c8
[   89.280348]  show_stack+0x24/0x30
[   89.283671]  dump_stack+0x118/0x19c
[   89.287167]  ubsan_epilogue+0x14/0x60
[   89.290835]  __ubsan_handle_out_of_bounds+0xfc/0x13c
[   89.295828]  MR_LdRaidGet+0x50/0x58 [megaraid_sas]
[   89.300638]  megasas_build_io_fusion+0xbb8/0xd90 [megaraid_sas]
[   89.306576]  megasas_build_and_issue_cmd_fusion+0x138/0x460 [megaraid_sas]
[   89.313468]  megasas_queue_command+0x398/0x3d0 [megaraid_sas]
[   89.319222]  scsi_dispatch_cmd+0x1dc/0x8a8
[   89.323321]  scsi_request_fn+0x8e8/0xdd0
[   89.327249]  __blk_run_queue+0xc4/0x158
[   89.331090]  blk_execute_rq_nowait+0xf4/0x158
[   89.335449]  blk_execute_rq+0xdc/0x158
[   89.339202]  __scsi_execute+0x130/0x258
[   89.343041]  scsi_probe_and_add_lun+0x2fc/0x1488
[   89.347661]  __scsi_scan_target+0x1cc/0x8c8
[   89.351848]  scsi_scan_channel.part.3+0x8c/0xc0
[   89.356382]  scsi_scan_host_selected+0x130/0x1f0
[   89.361002]  do_scsi_scan_host+0xd8/0xf0
[   89.364927]  do_scan_async+0x9c/0x320
[   89.368594]  async_run_entry_fn+0x138/0x420
[   89.372780]  process_one_work+0x61c/0xf08
[   89.376793]  worker_thread+0x13c/0xa70
[   89.380546]  kthread+0x1c8/0x1d0
[   89.383778]  ret_from_fork+0x10/0x1c

This is because when populating Driver Map using firmware raid map, all
non-existing VDs set their ldTgtIdToLd to 0xff, so it can be skipped later.

From drivers/scsi/megaraid/megaraid_sas_base.c ,
memset(instance->ld_ids, 0xff, MEGASAS_MAX_LD_IDS);

From drivers/scsi/megaraid/megaraid_sas_fp.c ,
/* For non existing VDs, iterate to next VD*/
if (ld >= (MAX_LOGICAL_DRIVES_EXT - 1))
continue;

However, there are a few places that failed to skip those non-existing VDs
due to off-by-one errors. Then, those 0xff leaked into MR_LdRaidGet(0xff,
map) and triggered the out-of-bound accesses.

Fixes: 51087a8617fe ("megaraid_sas : Extended VD support")
Signed-off-by: Qian Cai <cai@lca.pw>
Acked-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: smartpqi: call pqi_free_interrupts() in pqi_shutdown()
Yanjiang Jin [Thu, 20 Dec 2018 08:32:35 +0000 (16:32 +0800)]
scsi: smartpqi: call pqi_free_interrupts() in pqi_shutdown()

[ Upstream commit e57b2945aa654e48f85a41e8917793c64ecb9de8 ]

We must free all irqs during shutdown, else kexec's 2nd kernel would hang
in pqi_wait_for_completion_io() as below:

Call trace:

 pqi_wait_for_completion_io
 pqi_submit_raid_request_synchronous.constprop.78+0x23c/0x310 [smartpqi]
 pqi_configure_events+0xec/0x1f8 [smartpqi]
 pqi_ctrl_init+0x814/0xca0 [smartpqi]
 pqi_pci_probe+0x400/0x46c [smartpqi]
 local_pci_probe+0x48/0xb0
 pci_device_probe+0x14c/0x1b0
 really_probe+0x218/0x3fc
 driver_probe_device+0x70/0x140
 __driver_attach+0x11c/0x134
 bus_for_each_dev+0x70/0xc8
 driver_attach+0x30/0x38
 bus_add_driver+0x1f0/0x294
 driver_register+0x74/0x12c
 __pci_register_driver+0x64/0x70
 pqi_init+0xd0/0x10000 [smartpqi]
 do_one_initcall+0x60/0x1d8
 do_init_module+0x64/0x1f8
 load_module+0x10ec/0x1350
 __se_sys_finit_module+0xd4/0x100
 __arm64_sys_finit_module+0x28/0x34
 el0_svc_handler+0x104/0x160
 el0_svc+0x8/0xc

This happens only in the following combinations:

1. smartpqi is built as module, not built-in;
2. We have a disk connected to smartpqi card;
3. Both kexec's 1st and 2nd kernels use this disk as Rootfs' mount point.

Signed-off-by: Yanjiang Jin <yanjiang.jin@hxt-semitech.com>
Acked-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoath10k: fix peer stats null pointer dereference
Zhi Chen [Thu, 20 Dec 2018 12:24:43 +0000 (14:24 +0200)]
ath10k: fix peer stats null pointer dereference

[ Upstream commit 2d3b55853b123c177037cf534c5aaa2650310094 ]

There was a race condition in SMP that an ath10k_peer was created but its
member sta was null. Following are procedures of ath10k_peer creation and
member sta access in peer statistics path.

    1. Peer creation:
        ath10k_peer_create()
            =>ath10k_wmi_peer_create()
                =>ath10k_wait_for_peer_created()
                ...

        # another kernel path, RX from firmware
        ath10k_htt_t2h_msg_handler()
        =>ath10k_peer_map_event()
                =>wake_up()
                # ar->peer_map[id] = peer //add peer to map

        #wake up original path from waiting
                ...
                # peer->sta = sta //sta assignment

    2.  RX path of statistics
        ath10k_htt_t2h_msg_handler()
            =>ath10k_update_per_peer_tx_stats()
                =>ath10k_htt_fetch_peer_stats()
                # peer->sta //sta accessing

Any access of peer->sta after peer was added to peer_map but before sta was
assigned could cause a null pointer issue. And because these two steps are
asynchronous, no proper lock can protect them. So both peer and sta need to
be checked before access.

Tested: QCA9984 with firmware ver 10.4-3.9.0.1-00005
Signed-off-by: Zhi Chen <zhichen@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: smartpqi: correct lun reset issues
Kevin Barnett [Fri, 7 Dec 2018 22:29:51 +0000 (16:29 -0600)]
scsi: smartpqi: correct lun reset issues

[ Upstream commit 2ba55c9851d74eb015a554ef69ddf2ef061d5780 ]

Problem:
The Linux kernel takes a logical volume offline after a LUN reset.  This is
generally accompanied by this message in the dmesg output:

Device offlined - not ready after error recovery

Root Cause:
The root cause is a "quirk" in the timeout handling in the Linux SCSI
layer. The Linux kernel places a 30-second timeout on most media access
commands (reads and writes) that it send to device drivers.  When a media
access command times out, the Linux kernel goes into error recovery mode
for the LUN that was the target of the command that timed out. Every
command that timed out is kept on a list inside of the Linux kernel to be
retried later. The kernel attempts to recover the command(s) that timed out
by issuing a LUN reset followed by a TEST UNIT READY. If the LUN reset and
TEST UNIT READY commands are successful, the kernel retries the command(s)
that timed out.

Each SCSI command issued by the kernel has a result field associated with
it. This field indicates the final result of the command (success or
error). When a command times out, the kernel places a value in this result
field indicating that the command timed out.

The "quirk" is that after the LUN reset and TEST UNIT READY commands are
completed, the kernel checks each command on the timed-out command list
before retrying it. If the result field is still "timed out", the kernel
treats that command as not having been successfully recovered for a
retry. If the number of commands that are in this state are greater than
two, the kernel takes the LUN offline.

Fix:
When our RAIDStack receives a LUN reset, it simply waits until all
outstanding commands complete. Generally, all of these outstanding commands
complete successfully. Therefore, the fix in the smartpqi driver is to
always set the command result field to indicate success when a request
completes successfully. This normally isn’t necessary because the result
field is always initialized to success when the command is submitted to the
driver. So when the command completes successfully, the result field is
left untouched. But in this case, the kernel changes the result field
behind the driver’s back and then expects the field to be changed by the
driver as the commands that timed-out complete.

Reviewed-by: Dave Carroll <david.carroll@microsemi.com>
Reviewed-by: Scott Teel <scott.teel@microsemi.com>
Signed-off-by: Kevin Barnett <kevin.barnett@microsemi.com>
Signed-off-by: Don Brace <don.brace@microsemi.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: mpt3sas: fix memory ordering on 64bit writes
Stephan Günther [Sun, 16 Dec 2018 12:08:21 +0000 (13:08 +0100)]
scsi: mpt3sas: fix memory ordering on 64bit writes

[ Upstream commit 23c3828aa2f84edec7020c7397a22931e7a879e1 ]

With commit 09c2f95ad404 ("scsi: mpt3sas: Swap I/O memory read value back
to cpu endianness"), 64bit writes in _base_writeq() were rewritten to use
__raw_writeq() instad of writeq().

This introduced a bug apparent on powerpc64 systems such as the Raptor
Talos II that causes the HBA to drop from the PCIe bus under heavy load and
being reinitialized after a couple of seconds.

It can easily be triggered on affacted systems by using something like

  fio --name=random-write --iodepth=4 --rw=randwrite --bs=4k --direct=0 \
    --size=128M --numjobs=64 --end_fsync=1
  fio --name=random-write --iodepth=4 --rw=randwrite --bs=64k --direct=0 \
    --size=128M --numjobs=64 --end_fsync=1

a couple of times. In my case I tested it on both a ZFS raidz2 and a btrfs
raid6 using LSI 9300-8i and 9400-8i controllers.

The fix consists in resembling the write ordering of writeq() by adding a
mandatory write memory barrier before device access and a compiler barrier
afterwards. The additional MMIO barrier is superfluous.

Signed-off-by: Stephan Günther <moepi@moepi.net>
Reported-by: Matt Corallo <linux@bluematt.me>
Acked-by: Sreekanth Reddy <Sreekanth.Reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoIB/usnic: Fix potential deadlock
Parvi Kaustubhi [Tue, 11 Dec 2018 22:15:42 +0000 (14:15 -0800)]
IB/usnic: Fix potential deadlock

[ Upstream commit 8036e90f92aae2784b855a0007ae2d8154d28b3c ]

Acquiring the rtnl lock while holding usdev_lock could result in a
deadlock.

For example:

usnic_ib_query_port()
| mutex_lock(&us_ibdev->usdev_lock)
 | ib_get_eth_speed()
  | rtnl_lock()

rtnl_lock()
| usnic_ib_netdevice_event()
 | mutex_lock(&us_ibdev->usdev_lock)

This commit moves the usdev_lock acquisition after the rtnl lock has been
released.

This is safe to do because usdev_lock is not protecting anything being
accessed in ib_get_eth_speed(). Hence, the correct order of holding locks
(rtnl -> usdev_lock) is not violated.

Signed-off-by: Parvi Kaustubhi <pkaustub@cisco.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agosysfs: Disable lockdep for driver bind/unbind files
Daniel Vetter [Wed, 19 Dec 2018 12:39:09 +0000 (13:39 +0100)]
sysfs: Disable lockdep for driver bind/unbind files

[ Upstream commit 4f4b374332ec0ae9c738ff8ec9bed5cd97ff9adc ]

This is the much more correct fix for my earlier attempt at:

https://lkml.org/lkml/2018/12/10/118

Short recap:

- There's not actually a locking issue, it's just lockdep being a bit
  too eager to complain about a possible deadlock.

- Contrary to what I claimed the real problem is recursion on
  kn->count. Greg pointed me at sysfs_break_active_protection(), used
  by the scsi subsystem to allow a sysfs file to unbind itself. That
  would be a real deadlock, which isn't what's happening here. Also,
  breaking the active protection means we'd need to manually handle
  all the lifetime fun.

- With Rafael we discussed the task_work approach, which kinda works,
  but has two downsides: It's a functional change for a lockdep
  annotation issue, and it won't work for the bind file (which needs
  to get the errno from the driver load function back to userspace).

- Greg also asked why this never showed up: To hit this you need to
  unregister a 2nd driver from the unload code of your first driver. I
  guess only gpus do that. The bug has always been there, but only
  with a recent patch series did we add more locks so that lockdep
  built a chain from unbinding the snd-hda driver to the
  acpi_video_unregister call.

Full lockdep splat:

[12301.898799] ============================================
[12301.898805] WARNING: possible recursive locking detected
[12301.898811] 4.20.0-rc7+ #84 Not tainted
[12301.898815] --------------------------------------------
[12301.898821] bash/5297 is trying to acquire lock:
[12301.898826] 00000000f61c6093 (kn->count#39){++++}, at: kernfs_remove_by_name_ns+0x3b/0x80
[12301.898841] but task is already holding lock:
[12301.898847] 000000005f634021 (kn->count#39){++++}, at: kernfs_fop_write+0xdc/0x190
[12301.898856] other info that might help us debug this:
[12301.898862]  Possible unsafe locking scenario:
[12301.898867]        CPU0
[12301.898870]        ----
[12301.898874]   lock(kn->count#39);
[12301.898879]   lock(kn->count#39);
[12301.898883] *** DEADLOCK ***
[12301.898891]  May be due to missing lock nesting notation
[12301.898899] 5 locks held by bash/5297:
[12301.898903]  #0: 00000000cd800e54 (sb_writers#4){.+.+}, at: vfs_write+0x17f/0x1b0
[12301.898915]  #1: 000000000465e7c2 (&of->mutex){+.+.}, at: kernfs_fop_write+0xd3/0x190
[12301.898925]  #2: 000000005f634021 (kn->count#39){++++}, at: kernfs_fop_write+0xdc/0x190
[12301.898936]  #3: 00000000414ef7ac (&dev->mutex){....}, at: device_release_driver_internal+0x34/0x240
[12301.898950]  #4: 000000003218fbdf (register_count_mutex){+.+.}, at: acpi_video_unregister+0xe/0x40
[12301.898960] stack backtrace:
[12301.898968] CPU: 1 PID: 5297 Comm: bash Not tainted 4.20.0-rc7+ #84
[12301.898974] Hardware name: Hewlett-Packard HP EliteBook 8460p/161C, BIOS 68SCF Ver. F.01 03/11/2011
[12301.898982] Call Trace:
[12301.898989]  dump_stack+0x67/0x9b
[12301.898997]  __lock_acquire+0x6ad/0x1410
[12301.899003]  ? kernfs_remove_by_name_ns+0x3b/0x80
[12301.899010]  ? find_held_lock+0x2d/0x90
[12301.899017]  ? mutex_spin_on_owner+0xe4/0x150
[12301.899023]  ? find_held_lock+0x2d/0x90
[12301.899030]  ? lock_acquire+0x90/0x180
[12301.899036]  lock_acquire+0x90/0x180
[12301.899042]  ? kernfs_remove_by_name_ns+0x3b/0x80
[12301.899049]  __kernfs_remove+0x296/0x310
[12301.899055]  ? kernfs_remove_by_name_ns+0x3b/0x80
[12301.899060]  ? kernfs_name_hash+0xd/0x80
[12301.899066]  ? kernfs_find_ns+0x6c/0x100
[12301.899073]  kernfs_remove_by_name_ns+0x3b/0x80
[12301.899080]  bus_remove_driver+0x92/0xa0
[12301.899085]  acpi_video_unregister+0x24/0x40
[12301.899127]  i915_driver_unload+0x42/0x130 [i915]
[12301.899160]  i915_pci_remove+0x19/0x30 [i915]
[12301.899169]  pci_device_remove+0x36/0xb0
[12301.899176]  device_release_driver_internal+0x185/0x240
[12301.899183]  unbind_store+0xaf/0x180
[12301.899189]  kernfs_fop_write+0x104/0x190
[12301.899195]  __vfs_write+0x31/0x180
[12301.899203]  ? rcu_read_lock_sched_held+0x6f/0x80
[12301.899209]  ? rcu_sync_lockdep_assert+0x29/0x50
[12301.899216]  ? __sb_start_write+0x13c/0x1a0
[12301.899221]  ? vfs_write+0x17f/0x1b0
[12301.899227]  vfs_write+0xb9/0x1b0
[12301.899233]  ksys_write+0x50/0xc0
[12301.899239]  do_syscall_64+0x4b/0x180
[12301.899247]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
[12301.899253] RIP: 0033:0x7f452ac7f7a4
[12301.899259] Code: 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 80 00 00 00 00 8b 05 aa f0 2c 00 48 63 ff 85 c0 75 13 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 54 f3 c3 66 90 55 53 48 89 d5 48 89 f3 48 83
[12301.899273] RSP: 002b:00007ffceafa6918 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
[12301.899282] RAX: ffffffffffffffda RBX: 000000000000000d RCX: 00007f452ac7f7a4
[12301.899288] RDX: 000000000000000d RSI: 00005612a1abf7c0 RDI: 0000000000000001
[12301.899295] RBP: 00005612a1abf7c0 R08: 000000000000000a R09: 00005612a1c46730
[12301.899301] R10: 000000000000000a R11: 0000000000000246 R12: 000000000000000d
[12301.899308] R13: 0000000000000001 R14: 00007f452af4a740 R15: 000000000000000d

Looking around I've noticed that usb and i2c already handle similar
recursion problems, where a sysfs file can unbind the same type of
sysfs somewhere else in the hierarchy. Relevant commits are:

commit 356c05d58af05d582e634b54b40050c73609617b
Author: Alan Stern <stern@rowland.harvard.edu>
Date:   Mon May 14 13:30:03 2012 -0400

    sysfs: get rid of some lockdep false positives

commit e9b526fe704812364bca07edd15eadeba163ebfb
Author: Alexander Sverdlin <alexander.sverdlin@nsn.com>
Date:   Fri May 17 14:56:35 2013 +0200

    i2c: suppress lockdep warning on delete_device

Implement the same trick for driver bind/unbind.

v2: Put the macro into bus.c (Greg).

Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Ramalingam C <ramalingam.c@intel.com>
Cc: Arend van Spriel <aspriel@gmail.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: Bartosz Golaszewski <brgl@bgdev.pl>
Cc: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Cc: Vivek Gautam <vivek.gautam@codeaurora.org>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoALSA: bebob: fix model-id of unit for Apogee Ensemble
Takashi Sakamoto [Wed, 19 Dec 2018 11:00:42 +0000 (20:00 +0900)]
ALSA: bebob: fix model-id of unit for Apogee Ensemble

[ Upstream commit 644b2e97405b0b74845e1d3c2b4fe4c34858062b ]

This commit fixes hard-coded model-id for an unit of Apogee Ensemble with
a correct value. This unit uses DM1500 ASIC produced ArchWave AG (formerly
known as BridgeCo AG).

I note that this model supports three modes in the number of data channels
in tx/rx streams; 8 ch pairs, 10 ch pairs, 18 ch pairs. The mode is
switched by Vendor-dependent AV/C command, like:

$ cd linux-firewire-utils
$ ./firewire-request /dev/fw1 fcp 0x00ff000003dbeb0600000000 (8ch pairs)
$ ./firewire-request /dev/fw1 fcp 0x00ff000003dbeb0601000000 (10ch pairs)
$ ./firewire-request /dev/fw1 fcp 0x00ff000003dbeb0602000000 (18ch pairs)

When switching between different mode, the unit disappears from IEEE 1394
bus, then appears on the bus with different combination of stream formats.
In a mode of 18 ch pairs, available sampling rate is up to 96.0 kHz, else
up to 192.0 kHz.

$ ./hinawa-config-rom-printer /dev/fw1
{ 'bus-info': { 'adj': False,
                'bmc': True,
                'chip_ID': 21474898341,
                'cmc': True,
                'cyc_clk_acc': 100,
                'generation': 2,
                'imc': True,
                'isc': True,
                'link_spd': 2,
                'max_ROM': 1,
                'max_rec': 512,
                'name': '1394',
                'node_vendor_ID': 987,
                'pmc': False},
  'root-directory': [ ['HARDWARE_VERSION', 19],
                      [ 'NODE_CAPABILITIES',
                        { 'addressing': {'64': True, 'fix': True, 'prv': False},
                          'misc': {'int': False, 'ms': False, 'spt': True},
                          'state': { 'atn': False,
                                     'ded': False,
                                     'drq': True,
                                     'elo': False,
                                     'init': False,
                                     'lst': True,
                                     'off': False},
                          'testing': {'bas': False, 'ext': False}}],
                      ['VENDOR', 987],
                      ['DESCRIPTOR', 'Apogee Electronics'],
                      ['MODEL', 126702],
                      ['DESCRIPTOR', 'Ensemble'],
                      ['VERSION', 5297],
                      [ 'UNIT',
                        [ ['SPECIFIER_ID', 41005],
                          ['VERSION', 65537],
                          ['MODEL', 126702],
                          ['DESCRIPTOR', 'Ensemble']]],
                      [ 'DEPENDENT_INFO',
                        [ ['SPECIFIER_ID', 2037],
                          ['VERSION', 1],
                          [(58, 'IMMEDIATE'), 16777159],
                          [(59, 'IMMEDIATE'), 1048576],
                          [(60, 'IMMEDIATE'), 16777159],
                          [(61, 'IMMEDIATE'), 6291456]]]]}

Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoBluetooth: btusb: Add support for Intel bluetooth device 8087:0029
Raghuram Hegde [Wed, 19 Dec 2018 06:12:18 +0000 (11:42 +0530)]
Bluetooth: btusb: Add support for Intel bluetooth device 8087:0029

[ Upstream commit 2da711bcebe81209a9f2f90e145600eb1bae2b71 ]

Include the new USB product ID for Intel Bluetooth device 22260
family(CcPeak)

The /sys/kernel/debug/usb/devices portion for this device is:

T:  Bus=01 Lev=01 Prnt=01 Port=02 Cnt=02 Dev#=  2 Spd=12   MxCh= 0
D:  Ver= 2.00 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=8087 ProdID=0029 Rev= 0.01
C:* #Ifs= 2 Cfg#= 1 Atr=e0 MxPwr=100mA
I:* If#= 0 Alt= 0 #EPs= 3 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=81(I) Atr=03(Int.) MxPS=  64 Ivl=1ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=82(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   0 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   0 Ivl=1ms
I:  If#= 1 Alt= 1 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   9 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   9 Ivl=1ms
I:  If#= 1 Alt= 2 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  17 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  17 Ivl=1ms
I:  If#= 1 Alt= 3 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  25 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  25 Ivl=1ms
I:  If#= 1 Alt= 4 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  33 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  33 Ivl=1ms
I:  If#= 1 Alt= 5 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  49 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  49 Ivl=1ms
I:  If#= 1 Alt= 6 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  63 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  63 Ivl=1ms

Signed-off-by: Raghuram Hegde <raghuram.hegde@intel.com>
Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agodm: Check for device sector overflow if CONFIG_LBDAF is not set
Milan Broz [Wed, 7 Nov 2018 21:24:55 +0000 (22:24 +0100)]
dm: Check for device sector overflow if CONFIG_LBDAF is not set

[ Upstream commit ef87bfc24f9b8da82c89aff493df20f078bc9cb1 ]

Reference to a device in device-mapper table contains offset in sectors.

If the sector_t is 32bit integer (CONFIG_LBDAF is not set), then
several device-mapper targets can overflow this offset and validity
check is then performed on a wrong offset and a wrong table is activated.

See for example (on 32bit without CONFIG_LBDAF) this overflow:

  # dmsetup create test --table "0 2048 linear /dev/sdg 4294967297"
  # dmsetup table test
  0 2048 linear 8:96 1

This patch adds explicit check for overflow if the offset is sector_t type.

Signed-off-by: Milan Broz <gmazyland@gmail.com>
Reviewed-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoclocksource/drivers/integrator-ap: Add missing of_node_put()
Yangtao Li [Sun, 25 Nov 2018 05:00:49 +0000 (00:00 -0500)]
clocksource/drivers/integrator-ap: Add missing of_node_put()

[ Upstream commit 5eb73c831171115d3b4347e1e7124a5a35d8086c ]

The function of_find_node_by_path() acquires a reference to the node
returned by it and that reference needs to be dropped by its caller.

integrator_ap_timer_init_of() doesn't do that.  The pri_node and the
sec_node are used as an identifier to compare against the current
node, so we can directly drop the refcount after getting the node from
the path as it is not used as pointer.

By dropping the refcount right after getting it, a single variable is
needed instead of two.

Fix this by use a single variable and drop the refcount right after
of_find_node_by_path().

Signed-off-by: Yangtao Li <tiny.windzz@gmail.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoquota: Lock s_umount in exclusive mode for Q_XQUOTA{ON,OFF} quotactls.
Javier Barrio [Thu, 13 Dec 2018 00:06:29 +0000 (01:06 +0100)]
quota: Lock s_umount in exclusive mode for Q_XQUOTA{ON,OFF} quotactls.

[ Upstream commit 41c4f85cdac280d356df1f483000ecec4a8868be ]

Commit 1fa5efe3622db58cb8c7b9a50665e9eb9a6c7e97 (ext4: Use generic helpers for quotaon
and quotaoff) made possible to call quotactl(Q_XQUOTAON/OFF) on ext4 filesystems
with sysfile quota support. This leads to calling dquot_enable/disable without s_umount
held in excl. mode, because quotactl_cmd_onoff checks only for Q_QUOTAON/OFF.

The following WARN_ON_ONCE triggers (in this case for dquot_enable, ext4, latest Linus' tree):

[  117.807056] EXT4-fs (dm-0): mounted filesystem with ordered data mode. Opts: quota,prjquota

[...]

[  155.036847] WARNING: CPU: 0 PID: 2343 at fs/quota/dquot.c:2469 dquot_enable+0x34/0xb9
[  155.036851] Modules linked in: quota_v2 quota_tree ipv6 af_packet joydev mousedev psmouse serio_raw pcspkr i2c_piix4 intel_agp intel_gtt e1000 ttm drm_kms_helper drm agpgart fb_sys_fops syscopyarea sysfillrect sysimgblt i2c_core input_leds kvm_intel kvm irqbypass qemu_fw_cfg floppy evdev parport_pc parport button crc32c_generic dm_mod ata_generic pata_acpi ata_piix libata loop ext4 crc16 mbcache jbd2 usb_storage usbcore sd_mod scsi_mod
[  155.036901] CPU: 0 PID: 2343 Comm: qctl Not tainted 4.20.0-rc6-00025-gf5d582777bcb #9
[  155.036903] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
[  155.036911] RIP: 0010:dquot_enable+0x34/0xb9
[  155.036915] Code: 41 56 41 55 41 54 55 53 4c 8b 6f 28 74 02 0f 0b 4d 8d 7d 70 49 89 fc 89 cb 41 89 d6 89 f5 4c 89 ff e8 23 09 ea ff 85 c0 74 0a <0f> 0b 4c 89 ff e8 8b 09 ea ff 85 db 74 6a 41 8b b5 f8 00 00 00 0f
[  155.036918] RSP: 0018:ffffb09b00493e08 EFLAGS: 00010202
[  155.036922] RAX: 0000000000000001 RBX: 0000000000000008 RCX: 0000000000000008
[  155.036924] RDX: 0000000000000001 RSI: 0000000000000002 RDI: ffff9781b67cd870
[  155.036926] RBP: 0000000000000002 R08: 0000000000000000 R09: 61c8864680b583eb
[  155.036929] R10: ffffb09b00493e48 R11: ffffffffff7ce7d4 R12: ffff9781b7ee8d78
[  155.036932] R13: ffff9781b67cd800 R14: 0000000000000004 R15: ffff9781b67cd870
[  155.036936] FS:  00007fd813250b88(0000) GS:ffff9781ba000000(0000) knlGS:0000000000000000
[  155.036939] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  155.036942] CR2: 00007fd812ff61d6 CR3: 000000007c882000 CR4: 00000000000006b0
[  155.036951] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  155.036953] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  155.036955] Call Trace:
[  155.037004]  dquot_quota_enable+0x8b/0xd0
[  155.037011]  kernel_quotactl+0x628/0x74e
[  155.037027]  ? do_mprotect_pkey+0x2a6/0x2cd
[  155.037034]  __x64_sys_quotactl+0x1a/0x1d
[  155.037041]  do_syscall_64+0x55/0xe4
[  155.037078]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  155.037105] RIP: 0033:0x7fd812fe1198
[  155.037109] Code: 02 77 0d 48 89 c1 48 c1 e9 3f 75 04 48 8b 04 24 48 83 c4 50 5b c3 48 83 ec 08 49 89 ca 48 63 d2 48 63 ff b8 b3 00 00 00 0f 05 <48> 89 c7 e8 c1 eb ff ff 5a c3 48 63 ff b8 bb 00 00 00 0f 05 48 89
[  155.037112] RSP: 002b:00007ffe8cd7b050 EFLAGS: 00000206 ORIG_RAX: 00000000000000b3
[  155.037116] RAX: ffffffffffffffda RBX: 00007ffe8cd7b148 RCX: 00007fd812fe1198
[  155.037119] RDX: 0000000000000000 RSI: 00007ffe8cd7cea9 RDI: 0000000000580102
[  155.037121] RBP: 00007ffe8cd7b0f0 R08: 000055fc8eba8a9d R09: 0000000000000000
[  155.037124] R10: 00007ffe8cd7b074 R11: 0000000000000206 R12: 00007ffe8cd7b168
[  155.037126] R13: 000055fc8eba8897 R14: 0000000000000000 R15: 0000000000000000
[  155.037131] ---[ end trace 210f864257175c51 ]---

and then the syscall proceeds without s_umount locking.

This patch locks the superblock ->s_umount sem. in exclusive mode for all Q_XQUOTAON/OFF
quotactls too in addition to Q_QUOTAON/OFF.

AFAICT, other than ext4, only xfs and ocfs2 are affected by this change.
The VFS will now call in xfs_quota_* functions with s_umount held, which wasn't the case
before. This looks good to me but I can not say for sure. Ext4 and ocfs2 where already
beeing called with s_umount exclusive via quota_quotaon/off which is basically the same.

Signed-off-by: Javier Barrio <javier.barrio.mart@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf tools: Add missing open_memstream() prototype for systems lacking it
Arnaldo Carvalho de Melo [Tue, 11 Dec 2018 19:31:19 +0000 (16:31 -0300)]
perf tools: Add missing open_memstream() prototype for systems lacking it

[ Upstream commit d7a8c4a6a055097a67ccfa3ca7c9ff1b64603a70 ]

There are systems such as the Android NDK API level 24 has the
open_memstream() function but doesn't provide a prototype, adding noise
to the build:

  builtin-timechart.c: In function 'cat_backtrace':
  builtin-timechart.c:486:2: warning: implicit declaration of function 'open_memstream' [-Wimplicit-function-declaration]
    FILE *f = open_memstream(&p, &p_len);
    ^
  builtin-timechart.c:486:2: warning: nested extern declaration of 'open_memstream' [-Wnested-externs]
  builtin-timechart.c:486:12: warning: initialization makes pointer from integer without a cast
    FILE *f = open_memstream(&p, &p_len);
              ^

Define a LACKS_OPEN_MEMSTREAM_PROTOTYPE define so that code needing that
can get a prototype.

Checked in the bionic git repo to be available since level 23:

https://android.googlesource.com/platform/bionic/+/master/libc/include/stdio.h#241

  FILE* open_memstream(char** __ptr, size_t* __size_ptr) __INTRODUCED_IN(23);

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-343ashae97e5bq6vizusyfno@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf tools: Add missing sigqueue() prototype for systems lacking it
Arnaldo Carvalho de Melo [Tue, 11 Dec 2018 18:48:47 +0000 (15:48 -0300)]
perf tools: Add missing sigqueue() prototype for systems lacking it

[ Upstream commit 748fe0889c1ff12d378946bd5326e8ee8eacf5cf ]

There are systems such as the Android NDK API level 24 has the
sigqueue() function but doesn't provide a prototype, adding noise to the
build:

  util/evlist.c: In function 'perf_evlist__prepare_workload':
  util/evlist.c:1494:4: warning: implicit declaration of function 'sigqueue' [-Wimplicit-function-declaration]
      if (sigqueue(getppid(), SIGUSR1, val))
      ^
  util/evlist.c:1494:4: warning: nested extern declaration of 'sigqueue' [-Wnested-externs]

Define a LACKS_SIGQUEUE_PROTOTYPE define so that code needing that can
get a prototype.

Checked in the bionic git repo to be available since level 23:

https://android.googlesource.com/platform/bionic/+/master/libc/include/signal.h#123

  int sigqueue(pid_t __pid, int __signal, const union sigval __value) __INTRODUCED_IN(23);

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-lmhpev1uni9kdrv7j29glyov@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf cs-etm: Correct packets swapping in cs_etm__flush()
Leo Yan [Tue, 11 Dec 2018 07:38:21 +0000 (15:38 +0800)]
perf cs-etm: Correct packets swapping in cs_etm__flush()

[ Upstream commit 43fd56669c28cd354e9228bdb58e4bca1c1a8b66 ]

The structure cs_etm_queue uses 'prev_packet' to point to previous
packet, this can be used to combine with new coming packet to generate
samples.

In function cs_etm__flush() it swaps packets only when the flag
'etm->synth_opts.last_branch' is true, this means that it will not swap
packets if without option '--itrace=il' to generate last branch entries;
thus for this case the 'prev_packet' doesn't point to the correct
previous packet and the stale packet still will be used to generate
sequential sample.  Thus if dump trace with 'perf script' command we can
see the incorrect flow with the stale packet's address info.

This patch corrects packets swapping in cs_etm__flush(); except using
the flag 'etm->synth_opts.last_branch' it also checks the another flag
'etm->sample_branches', if any flag is true then it swaps packets so can
save correct content to 'prev_packet'.  Finally this can fix the wrong
program flow dumping issue.

The patch has a minor refactoring to use 'etm->synth_opts.last_branch'
instead of 'etmq->etm->synth_opts.last_branch' for condition checking,
this is consistent with that is done in cs_etm__sample().

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Robert Walker <robert.walker@arm.com>
Cc: coresight@lists.linaro.org
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lkml.kernel.org/r/1544513908-16805-2-git-send-email-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agodm snapshot: Fix excessive memory usage and workqueue stalls
Nikos Tsironis [Wed, 31 Oct 2018 21:53:08 +0000 (17:53 -0400)]
dm snapshot: Fix excessive memory usage and workqueue stalls

[ Upstream commit 721b1d98fb517ae99ab3b757021cf81db41e67be ]

kcopyd has no upper limit to the number of jobs one can allocate and
issue. Under certain workloads this can lead to excessive memory usage
and workqueue stalls. For example, when creating multiple dm-snapshot
targets with a 4K chunk size and then writing to the origin through the
page cache. Syncing the page cache causes a large number of BIOs to be
issued to the dm-snapshot origin target, which itself issues an even
larger (because of the BIO splitting taking place) number of kcopyd
jobs.

Running the following test, from the device mapper test suite [1],

  dmtest run --suite snapshot -n many_snapshots_of_same_volume_N

, with 8 active snapshots, results in the kcopyd job slab cache growing
to 10G. Depending on the available system RAM this can lead to the OOM
killer killing user processes:

[463.492878] kthreadd invoked oom-killer: gfp_mask=0x6040c0(GFP_KERNEL|__GFP_COMP),
              nodemask=(null), order=1, oom_score_adj=0
[463.492894] kthreadd cpuset=/ mems_allowed=0
[463.492948] CPU: 7 PID: 2 Comm: kthreadd Not tainted 4.19.0-rc7 #3
[463.492950] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
[463.492952] Call Trace:
[463.492964]  dump_stack+0x7d/0xbb
[463.492973]  dump_header+0x6b/0x2fc
[463.492987]  ? lockdep_hardirqs_on+0xee/0x190
[463.493012]  oom_kill_process+0x302/0x370
[463.493021]  out_of_memory+0x113/0x560
[463.493030]  __alloc_pages_slowpath+0xf40/0x1020
[463.493055]  __alloc_pages_nodemask+0x348/0x3c0
[463.493067]  cache_grow_begin+0x81/0x8b0
[463.493072]  ? cache_grow_begin+0x874/0x8b0
[463.493078]  fallback_alloc+0x1e4/0x280
[463.493092]  kmem_cache_alloc_node+0xd6/0x370
[463.493098]  ? copy_process.part.31+0x1c5/0x20d0
[463.493105]  copy_process.part.31+0x1c5/0x20d0
[463.493115]  ? __lock_acquire+0x3cc/0x1550
[463.493121]  ? __switch_to_asm+0x34/0x70
[463.493129]  ? kthread_create_worker_on_cpu+0x70/0x70
[463.493135]  ? finish_task_switch+0x90/0x280
[463.493165]  _do_fork+0xe0/0x6d0
[463.493191]  ? kthreadd+0x19f/0x220
[463.493233]  kernel_thread+0x25/0x30
[463.493235]  kthreadd+0x1bf/0x220
[463.493242]  ? kthread_create_on_cpu+0x90/0x90
[463.493248]  ret_from_fork+0x3a/0x50
[463.493279] Mem-Info:
[463.493285] active_anon:20631 inactive_anon:4831 isolated_anon:0
[463.493285]  active_file:80216 inactive_file:80107 isolated_file:435
[463.493285]  unevictable:0 dirty:51266 writeback:109372 unstable:0
[463.493285]  slab_reclaimable:31191 slab_unreclaimable:3483521
[463.493285]  mapped:526 shmem:4903 pagetables:1759 bounce:0
[463.493285]  free:33623 free_pcp:2392 free_cma:0
...
[463.493489] Unreclaimable slab info:
[463.493513] Name                      Used          Total
[463.493522] bio-6                   1028KB       1028KB
[463.493525] bio-5                   1028KB       1028KB
[463.493528] dm_snap_pending_exception     236783KB     243789KB
[463.493531] dm_exception              41KB         42KB
[463.493534] bio-4                   1216KB       1216KB
[463.493537] bio-3                 439396KB     439396KB
[463.493539] kcopyd_job           6973427KB    6973427KB
...
[463.494340] Out of memory: Kill process 1298 (ruby2.3) score 1 or sacrifice child
[463.494673] Killed process 1298 (ruby2.3) total-vm:435740kB, anon-rss:20180kB, file-rss:4kB, shmem-rss:0kB
[463.506437] oom_reaper: reaped process 1298 (ruby2.3), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB

Moreover, issuing a large number of kcopyd jobs results in kcopyd
hogging the CPU, while processing them. As a result, processing of work
items, queued for execution on the same CPU as the currently running
kcopyd thread, is stalled for long periods of time, hurting performance.
Running the aforementioned test we get, in dmesg, messages like the
following:

[67501.194592] BUG: workqueue lockup - pool cpus=4 node=0 flags=0x0 nice=0 stuck for 27s!
[67501.195586] Showing busy workqueues and worker pools:
[67501.195591] workqueue events: flags=0x0
[67501.195597]   pwq 8: cpus=4 node=0 flags=0x0 nice=0 active=1/256
[67501.195611]     pending: cache_reap
[67501.195641] workqueue mm_percpu_wq: flags=0x8
[67501.195645]   pwq 8: cpus=4 node=0 flags=0x0 nice=0 active=1/256
[67501.195656]     pending: vmstat_update
[67501.195682] workqueue kblockd: flags=0x18
[67501.195687]   pwq 5: cpus=2 node=0 flags=0x0 nice=-20 active=1/256
[67501.195698]     pending: blk_timeout_work
[67501.195753] workqueue kcopyd: flags=0x8
[67501.195757]   pwq 8: cpus=4 node=0 flags=0x0 nice=0 active=1/256
[67501.195768]     pending: do_work [dm_mod]
[67501.195802] workqueue kcopyd: flags=0x8
[67501.195806]   pwq 8: cpus=4 node=0 flags=0x0 nice=0 active=1/256
[67501.195817]     pending: do_work [dm_mod]
[67501.195834] workqueue kcopyd: flags=0x8
[67501.195838]   pwq 8: cpus=4 node=0 flags=0x0 nice=0 active=1/256
[67501.195848]     pending: do_work [dm_mod]
[67501.195881] workqueue kcopyd: flags=0x8
[67501.195885]   pwq 8: cpus=4 node=0 flags=0x0 nice=0 active=1/256
[67501.195896]     pending: do_work [dm_mod]
[67501.195920] workqueue kcopyd: flags=0x8
[67501.195924]   pwq 8: cpus=4 node=0 flags=0x0 nice=0 active=2/256
[67501.195935]     in-flight: 67:do_work [dm_mod]
[67501.195945]     pending: do_work [dm_mod]
[67501.195961] pool 8: cpus=4 node=0 flags=0x0 nice=0 hung=27s workers=3 idle: 129 23765

The root cause for these issues is the way dm-snapshot uses kcopyd. In
particular, the lack of an explicit or implicit limit to the maximum
number of in-flight COW jobs. The merging path is not affected because
it implicitly limits the in-flight kcopyd jobs to one.

Fix these issues by using a semaphore to limit the maximum number of
in-flight kcopyd jobs. We grab the semaphore before allocating a new
kcopyd job in start_copy() and start_full_bio() and release it after the
job finishes in copy_callback().

The initial semaphore value is configurable through a module parameter,
to allow fine tuning the maximum number of in-flight COW jobs. Setting
this parameter to zero initializes the semaphore to INT_MAX.

A default value of 2048 maximum in-flight kcopyd jobs was chosen. This
value was decided experimentally as a trade-off between memory
consumption, stalling the kernel's workqueues and maintaining a high
enough throughput.

Re-running the aforementioned test:

  * Workqueue stalls are eliminated
  * kcopyd's job slab cache uses a maximum of 130MB
  * The time taken by the test to write to the snapshot-origin target is
    reduced from 05m20.48s to 03m26.38s

[1] https://github.com/jthornber/device-mapper-test-suite

Signed-off-by: Nikos Tsironis <ntsironis@arrikto.com>
Signed-off-by: Ilias Tsitsimpis <iliastsi@arrikto.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agotools lib subcmd: Don't add the kernel sources to the include path
Arnaldo Carvalho de Melo [Tue, 11 Dec 2018 18:00:52 +0000 (15:00 -0300)]
tools lib subcmd: Don't add the kernel sources to the include path

[ Upstream commit ece9804985b57e1ccd83b1fb6288520955a29d51 ]

At some point we decided not to directly include kernel sources files
when building tools/perf/, but when tools/lib/subcmd/ was forked from
tools/perf it somehow ended up adding it via these two lines in its
Makefile:

  CFLAGS += -I$(srctree)/include/uapi
  CFLAGS += -I$(srctree)/include

As $(srctree) points to the kernel sources.

Removing those lines and keeping just:

  CFLAGS += -I$(srctree)/tools/include/

Is enough to build tools/perf and tools/objtool.

This fixes the build when building from the sources in environments such
as the Android NDK crossbuilding from a fedora:26 system:

  subcmd-util.h:11:15: error: expected ',' or ';' before 'void'
   static inline void report(const char *prefix, const char *err, va_list params)
                 ^
  In file included from /git/perf/include/uapi/linux/stddef.h:2:0,
                   from /git/perf/include/uapi/linux/posix_types.h:5,
                   from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/sys/types.h:36,
                   from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/unistd.h:33,
                   from run-command.c:2:
  subcmd-util.h:18:17: error: '__no_instrument_function__' attribute applies only to functions

The /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/sys/types.h
file that includes linux/posix_types.h ends up getting the one in the kernel
sources causing the breakage. Fix it.

Test built tools/objtool/ too.

Reported-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Jiri Olsa <jolsa@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Fixes: 4b6ab94eabe4 ("perf subcmd: Create subcmd library")
Link: https://lkml.kernel.org/n/tip-5lhaoecrj12t0bqwvpiu14sm@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf stat: Avoid segfaults caused by negated options
Michael Petlan [Mon, 10 Dec 2018 16:00:04 +0000 (11:00 -0500)]
perf stat: Avoid segfaults caused by negated options

[ Upstream commit 51433ead1460fb3f46e1c34f68bb22fd2dd0f5d0 ]

Some 'perf stat' options do not make sense to be negated (event,
cgroup), some do not have negated path implemented (metrics). Due to
that, it is better to disable the "no-" prefix for them, since
otherwise, the later opt-parsing segfaults.

Before:

  $ perf stat --no-metrics -- ls
  Segmentation fault (core dumped)

After:

  $ perf stat --no-metrics -- ls
   Error: option `no-metrics' isn't available
   Usage: perf stat [<options>] [<command>]

Signed-off-by: Michael Petlan <mpetlan@redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
LPU-Reference: 1485912065.62416880.1544457604340.JavaMail.zimbra@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agodm kcopyd: Fix bug causing workqueue stalls
Nikos Tsironis [Wed, 31 Oct 2018 21:53:09 +0000 (17:53 -0400)]
dm kcopyd: Fix bug causing workqueue stalls

[ Upstream commit d7e6b8dfc7bcb3f4f3a18313581f67486a725b52 ]

When using kcopyd to run callbacks through dm_kcopyd_do_callback() or
submitting copy jobs with a source size of 0, the jobs are pushed
directly to the complete_jobs list, which could be under processing by
the kcopyd thread. As a result, the kcopyd thread can continue running
completed jobs indefinitely, without releasing the CPU, as long as
someone keeps submitting new completed jobs through the aforementioned
paths. Processing of work items, queued for execution on the same CPU as
the currently running kcopyd thread, is thus stalled for excessive
amounts of time, hurting performance.

Running the following test, from the device mapper test suite [1],

  dmtest run --suite snapshot -n parallel_io_to_many_snaps_N

, with 8 active snapshots, we get, in dmesg, messages like the
following:

[68899.948523] BUG: workqueue lockup - pool cpus=0 node=0 flags=0x0 nice=0 stuck for 95s!
[68899.949282] Showing busy workqueues and worker pools:
[68899.949288] workqueue events: flags=0x0
[68899.949295]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=2/256
[68899.949306]     pending: vmstat_shepherd, cache_reap
[68899.949331] workqueue mm_percpu_wq: flags=0x8
[68899.949337]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256
[68899.949345]     pending: vmstat_update
[68899.949387] workqueue dm_bufio_cache: flags=0x8
[68899.949392]   pwq 4: cpus=2 node=0 flags=0x0 nice=0 active=1/256
[68899.949400]     pending: work_fn [dm_bufio]
[68899.949423] workqueue kcopyd: flags=0x8
[68899.949429]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256
[68899.949437]     pending: do_work [dm_mod]
[68899.949452] workqueue kcopyd: flags=0x8
[68899.949458]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=2/256
[68899.949466]     in-flight: 13:do_work [dm_mod]
[68899.949474]     pending: do_work [dm_mod]
[68899.949487] workqueue kcopyd: flags=0x8
[68899.949493]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256
[68899.949501]     pending: do_work [dm_mod]
[68899.949515] workqueue kcopyd: flags=0x8
[68899.949521]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256
[68899.949529]     pending: do_work [dm_mod]
[68899.949541] workqueue kcopyd: flags=0x8
[68899.949547]   pwq 0: cpus=0 node=0 flags=0x0 nice=0 active=1/256
[68899.949555]     pending: do_work [dm_mod]
[68899.949568] pool 0: cpus=0 node=0 flags=0x0 nice=0 hung=95s workers=4 idle: 27130 27223 1084

Fix this by splitting the complete_jobs list into two parts: A user
facing part, named callback_jobs, and one used internally by kcopyd,
retaining the name complete_jobs. dm_kcopyd_do_callback() and
dispatch_job() now push their jobs to the callback_jobs list, which is
spliced to the complete_jobs list once, every time the kcopyd thread
wakes up. This prevents kcopyd from hogging the CPU indefinitely and
causing workqueue stalls.

Re-running the aforementioned test:

  * Workqueue stalls are eliminated
  * The maximum writing time among all targets is reduced from 09m37.10s
    to 06m04.85s and the total run time of the test is reduced from
    10m43.591s to 7m19.199s

[1] https://github.com/jthornber/device-mapper-test-suite

Signed-off-by: Nikos Tsironis <ntsironis@arrikto.com>
Signed-off-by: Ilias Tsitsimpis <iliastsi@arrikto.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agodm crypt: use u64 instead of sector_t to store iv_offset
AliOS system security [Mon, 5 Nov 2018 07:31:42 +0000 (15:31 +0800)]
dm crypt: use u64 instead of sector_t to store iv_offset

[ Upstream commit 8d683dcd65c037efc9fb38c696ec9b65b306e573 ]

The iv_offset in the mapping table of crypt target is a 64bit number when
IV algorithm is plain64, plain64be, essiv or benbi. It will be assigned to
iv_offset of struct crypt_config, cc_sector of struct convert_context and
iv_sector of struct dm_crypt_request. These structures members are defined
as a sector_t. But sector_t is 32bit when CONFIG_LBDAF is not set in 32bit
kernel. In this situation sector_t is not big enough to store the 64bit
iv_offset.

Here is a reproducer.
Prepare test image and device (loop is automatically allocated by cryptsetup):

  # dd if=/dev/zero of=tst.img bs=1M count=1
  # echo "tst"|cryptsetup open --type plain -c aes-xts-plain64 \
  --skip 500000000000000000 tst.img test

On 32bit system (use IV offset value that overflows to 64bit; CONFIG_LBDAF if off)
and device checksum is wrong:

  # dmsetup table test --showkeys
  0 2048 crypt aes-xts-plain64 dfa7cfe3c481f2239155739c42e539ae8f2d38f304dcc89d20b26f69daaf0933 3551657984 7:0 0

  # sha256sum /dev/mapper/test
  533e25c09176632b3794f35303488c4a8f3f965dffffa6ec2df347c168cb6c19 /dev/mapper/test

On 64bit system (and on 32bit system with the patch), table and checksum is now correct:

  # dmsetup table test --showkeys
  0 2048 crypt aes-xts-plain64 dfa7cfe3c481f2239155739c42e539ae8f2d38f304dcc89d20b26f69daaf0933 500000000000000000 7:0 0

  # sha256sum /dev/mapper/test
  5d16160f9d5f8c33d8051e65fdb4f003cc31cd652b5abb08f03aa6fce0df75fc /dev/mapper/test

Signed-off-by: AliOS system security <alios_sys_security@linux.alibaba.com>
Tested-and-Reviewed-by: Milan Broz <gmazyland@gmail.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agox86/topology: Use total_cpus for max logical packages calculation
Hui Wang [Wed, 7 Nov 2018 02:36:43 +0000 (10:36 +0800)]
x86/topology: Use total_cpus for max logical packages calculation

[ Upstream commit aa02ef099cff042c2a9109782ec2bf1bffc955d4 ]

nr_cpu_ids can be limited on the command line via nr_cpus=. This can break the
logical package management because it results in a smaller number of packages
while in kdump kernel.

Check below case:
There is a two sockets system, each socket has 8 cores, which has 16 logical
cpus while HT was turn on.

 0  1  2  3  4  5  6  7     |    16 17 18 19 20 21 22 23
 cores on socket 0               threads on socket 0
 8  9 10 11 12 13 14 15     |    24 25 26 27 28 29 30 31
 cores on socket 1               threads on socket 1

While starting the kdump kernel with command line option nr_cpus=16 panic
was triggered on one of the cpus 24-31 eg. 26, then online cpu will be
1-15, 26(cpu 0 was disabled in kdump), ncpus will be 16 and
__max_logical_packages will be 1, but actually two packages were booted on.

This issue can reproduced by set kdump option nr_cpus=<real physical core
numbers>, and then trigger panic on last socket's thread, for example:

taskset -c 26 echo c > /proc/sysrq-trigger

Use total_cpus which will not be limited by nr_cpus command line to calculate
the value of __max_logical_packages.

Signed-off-by: Hui Wang <john.wanghui@huawei.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: <guijianfeng@huawei.com>
Cc: <wencongyang2@huawei.com>
Cc: <douliyang1@huawei.com>
Cc: <qiaonuohan@huawei.com>
Link: https://lkml.kernel.org/r/20181107023643.22174-1-john.wanghui@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agonetfilter: ipt_CLUSTERIP: fix deadlock in netns exit routine
Taehee Yoo [Mon, 5 Nov 2018 09:22:44 +0000 (18:22 +0900)]
netfilter: ipt_CLUSTERIP: fix deadlock in netns exit routine

[ Upstream commit 5a86d68bcf02f2d1e9a5897dd482079fd5f75e7f ]

When network namespace is destroyed, cleanup_net() is called.
cleanup_net() holds pernet_ops_rwsem then calls each ->exit callback.
So that clusterip_tg_destroy() is called by cleanup_net().
And clusterip_tg_destroy() calls unregister_netdevice_notifier().

But both cleanup_net() and clusterip_tg_destroy() hold same
lock(pernet_ops_rwsem). hence deadlock occurrs.

After this patch, only 1 notifier is registered when module is inserted.
And all of configs are added to per-net list.

test commands:
   %ip netns add vm1
   %ip netns exec vm1 bash
   %ip link set lo up
   %iptables -A INPUT -p tcp -i lo -d 192.168.0.5 --dport 80 \
-j CLUSTERIP --new --hashmode sourceip \
--clustermac 01:00:5e:00:00:20 --total-nodes 2 --local-node 1
   %exit
   %ip netns del vm1

splat looks like:
[  341.809674] ============================================
[  341.809674] WARNING: possible recursive locking detected
[  341.809674] 4.19.0-rc5+ #16 Tainted: G        W
[  341.809674] --------------------------------------------
[  341.809674] kworker/u4:2/87 is trying to acquire lock:
[  341.809674] 000000005da2d519 (pernet_ops_rwsem){++++}, at: unregister_netdevice_notifier+0x8c/0x460
[  341.809674]
[  341.809674] but task is already holding lock:
[  341.809674] 000000005da2d519 (pernet_ops_rwsem){++++}, at: cleanup_net+0x119/0x900
[  341.809674]
[  341.809674] other info that might help us debug this:
[  341.809674]  Possible unsafe locking scenario:
[  341.809674]
[  341.809674]        CPU0
[  341.809674]        ----
[  341.809674]   lock(pernet_ops_rwsem);
[  341.809674]   lock(pernet_ops_rwsem);
[  341.809674]
[  341.809674]  *** DEADLOCK ***
[  341.809674]
[  341.809674]  May be due to missing lock nesting notation
[  341.809674]
[  341.809674] 3 locks held by kworker/u4:2/87:
[  341.809674]  #0: 00000000d9df6c92 ((wq_completion)"%s""netns"){+.+.}, at: process_one_work+0xafe/0x1de0
[  341.809674]  #1: 00000000c2cbcee2 (net_cleanup_work){+.+.}, at: process_one_work+0xb60/0x1de0
[  341.809674]  #2: 000000005da2d519 (pernet_ops_rwsem){++++}, at: cleanup_net+0x119/0x900
[  341.809674]
[  341.809674] stack backtrace:
[  341.809674] CPU: 1 PID: 87 Comm: kworker/u4:2 Tainted: G        W         4.19.0-rc5+ #16
[  341.809674] Workqueue: netns cleanup_net
[  341.809674] Call Trace:
[ ... ]
[  342.070196]  down_write+0x93/0x160
[  342.070196]  ? unregister_netdevice_notifier+0x8c/0x460
[  342.070196]  ? down_read+0x1e0/0x1e0
[  342.070196]  ? sched_clock_cpu+0x126/0x170
[  342.070196]  ? find_held_lock+0x39/0x1c0
[  342.070196]  unregister_netdevice_notifier+0x8c/0x460
[  342.070196]  ? register_netdevice_notifier+0x790/0x790
[  342.070196]  ? __local_bh_enable_ip+0xe9/0x1b0
[  342.070196]  ? __local_bh_enable_ip+0xe9/0x1b0
[  342.070196]  ? clusterip_tg_destroy+0x372/0x650 [ipt_CLUSTERIP]
[  342.070196]  ? trace_hardirqs_on+0x93/0x210
[  342.070196]  ? __bpf_trace_preemptirq_template+0x10/0x10
[  342.070196]  ? clusterip_tg_destroy+0x372/0x650 [ipt_CLUSTERIP]
[  342.123094]  clusterip_tg_destroy+0x3ad/0x650 [ipt_CLUSTERIP]
[  342.123094]  ? clusterip_net_init+0x3d0/0x3d0 [ipt_CLUSTERIP]
[  342.123094]  ? cleanup_match+0x17d/0x200 [ip_tables]
[  342.123094]  ? xt_unregister_table+0x215/0x300 [x_tables]
[  342.123094]  ? kfree+0xe2/0x2a0
[  342.123094]  cleanup_entry+0x1d5/0x2f0 [ip_tables]
[  342.123094]  ? cleanup_match+0x200/0x200 [ip_tables]
[  342.123094]  __ipt_unregister_table+0x9b/0x1a0 [ip_tables]
[  342.123094]  iptable_filter_net_exit+0x43/0x80 [iptable_filter]
[  342.123094]  ops_exit_list.isra.10+0x94/0x140
[  342.123094]  cleanup_net+0x45b/0x900
[ ... ]

Fixes: 202f59afd441 ("netfilter: ipt_CLUSTERIP: do not hold dev")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agonetfilter: ipt_CLUSTERIP: remove wrong WARN_ON_ONCE in netns exit routine
Taehee Yoo [Mon, 5 Nov 2018 09:22:55 +0000 (18:22 +0900)]
netfilter: ipt_CLUSTERIP: remove wrong WARN_ON_ONCE in netns exit routine

[ Upstream commit b12f7bad5ad3724d19754390a3e80928525c0769 ]

When network namespace is destroyed, both clusterip_tg_destroy() and
clusterip_net_exit() are called. and clusterip_net_exit() is called
before clusterip_tg_destroy().
Hence cleanup check code in clusterip_net_exit() doesn't make sense.

test commands:
   %ip netns add vm1
   %ip netns exec vm1 bash
   %ip link set lo up
   %iptables -A INPUT -p tcp -i lo -d 192.168.0.5 --dport 80 \
-j CLUSTERIP --new --hashmode sourceip \
--clustermac 01:00:5e:00:00:20 --total-nodes 2 --local-node 1
   %exit
   %ip netns del vm1

splat looks like:
[  341.184508] WARNING: CPU: 1 PID: 87 at net/ipv4/netfilter/ipt_CLUSTERIP.c:840 clusterip_net_exit+0x319/0x380 [ipt_CLUSTERIP]
[  341.184850] Modules linked in: ipt_CLUSTERIP nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 xt_tcpudp iptable_filter bpfilter ip_tables x_tables
[  341.184850] CPU: 1 PID: 87 Comm: kworker/u4:2 Not tainted 4.19.0-rc5+ #16
[  341.227509] Workqueue: netns cleanup_net
[  341.227509] RIP: 0010:clusterip_net_exit+0x319/0x380 [ipt_CLUSTERIP]
[  341.227509] Code: 0f 85 7f fe ff ff 48 c7 c2 80 64 2c c0 be a8 02 00 00 48 c7 c7 a0 63 2c c0 c6 05 18 6e 00 00 01 e8 bc 38 ff f5 e9 5b fe ff ff <0f> 0b e9 33 ff ff ff e8 4b 90 50 f6 e9 2d fe ff ff 48 89 df e8 de
[  341.227509] RSP: 0018:ffff88011086f408 EFLAGS: 00010202
[  341.227509] RAX: dffffc0000000000 RBX: 1ffff1002210de85 RCX: 0000000000000000
[  341.227509] RDX: 1ffff1002210de85 RSI: ffff880110813be8 RDI: ffffed002210de58
[  341.227509] RBP: ffff88011086f4d0 R08: 0000000000000000 R09: 0000000000000000
[  341.227509] R10: 0000000000000000 R11: 0000000000000000 R12: 1ffff1002210de81
[  341.227509] R13: ffff880110625a48 R14: ffff880114cec8c8 R15: 0000000000000014
[  341.227509] FS:  0000000000000000(0000) GS:ffff880116600000(0000) knlGS:0000000000000000
[  341.227509] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  341.227509] CR2: 00007f11fd38e000 CR3: 000000013ca16000 CR4: 00000000001006e0
[  341.227509] Call Trace:
[  341.227509]  ? __clusterip_config_find+0x460/0x460 [ipt_CLUSTERIP]
[  341.227509]  ? default_device_exit+0x1ca/0x270
[  341.227509]  ? remove_proc_entry+0x1cd/0x390
[  341.227509]  ? dev_change_net_namespace+0xd00/0xd00
[  341.227509]  ? __init_waitqueue_head+0x130/0x130
[  341.227509]  ops_exit_list.isra.10+0x94/0x140
[  341.227509]  cleanup_net+0x45b/0x900
[ ... ]

Fixes: 613d0776d3fe ("netfilter: exit_net cleanup check added")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agonetfilter: ipt_CLUSTERIP: check MAC address when duplicate config is set
Taehee Yoo [Mon, 5 Nov 2018 09:23:25 +0000 (18:23 +0900)]
netfilter: ipt_CLUSTERIP: check MAC address when duplicate config is set

[ Upstream commit 06aa151ad1fc74a49b45336672515774a678d78d ]

If same destination IP address config is already existing, that config is
just used. MAC address also should be same.
However, there is no MAC address checking routine.
So that MAC address checking routine is added.

test commands:
   %iptables -A INPUT -p tcp -i lo -d 192.168.0.5 --dport 80 \
   -j CLUSTERIP --new --hashmode sourceip \
   --clustermac 01:00:5e:00:00:20 --total-nodes 2 --local-node 1
   %iptables -A INPUT -p tcp -i lo -d 192.168.0.5 --dport 80 \
   -j CLUSTERIP --new --hashmode sourceip \
   --clustermac 01:00:5e:00:00:21 --total-nodes 2 --local-node 1

After this patch, above commands are disallowed.

Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf vendor events intel: Fix Load_Miss_Real_Latency on SKL/SKX
Andi Kleen [Tue, 20 Nov 2018 05:06:35 +0000 (21:06 -0800)]
perf vendor events intel: Fix Load_Miss_Real_Latency on SKL/SKX

[ Upstream commit 91b2b97025097ce7ca7536bc87eba2bf14760fb4 ]

Fix incorrect event names for the Load_Miss_Real_Latency metric for
Skylake and Skylake Server.

Fixes https://github.com/andikleen/pmu-tools/issues/158

Before:

  % perf stat -M Load_Miss_Real_Latency true
  event syntax error: '..ss.pending,mem_load_retired.l1_miss_ps,mem_load_retired.fb_hit_ps}:W'
                                    \___ parser error

   Usage: perf stat [<options>] [<command>]

      -M, --metrics <metric/metric group list>
                            monitor specified metrics or metric groups (separated by ,)

After:

  % perf stat -M Load_Miss_Real_Latency true

   Performance counter stats for 'true':

             279,204      l1d_pend_miss.pending     #     14.0 Load_Miss_Real_Latency
               4,784      mem_load_uops_retired.l1_miss
              15,188      mem_load_uops_retired.hit_lfb

         0.000899640 seconds time elapsed

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Link: http://lkml.kernel.org/r/20181120050635.4215-1-andi@firstfloor.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf parse-events: Fix unchecked usage of strncpy()
Arnaldo Carvalho de Melo [Thu, 6 Dec 2018 16:52:13 +0000 (13:52 -0300)]
perf parse-events: Fix unchecked usage of strncpy()

[ Upstream commit bd8d57fb7e25e9fcf67a9eef5fa13aabe2016e07 ]

The strncpy() function may leave the destination string buffer
unterminated, better use strlcpy() that we have a __weak fallback
implementation for systems without it.

This fixes this warning on an Alpine Linux Edge system with gcc 8.2:

  util/parse-events.c: In function 'print_symbol_events':
  util/parse-events.c:2465:4: error: 'strncpy' specified bound 100 equals destination size [-Werror=stringop-truncation]
      strncpy(name, syms->symbol, MAX_NAME_LEN);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  In function 'print_symbol_events.constprop',
      inlined from 'print_events' at util/parse-events.c:2508:2:
  util/parse-events.c:2465:4: error: 'strncpy' specified bound 100 equals destination size [-Werror=stringop-truncation]
      strncpy(name, syms->symbol, MAX_NAME_LEN);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  In function 'print_symbol_events.constprop',
      inlined from 'print_events' at util/parse-events.c:2511:2:
  util/parse-events.c:2465:4: error: 'strncpy' specified bound 100 equals destination size [-Werror=stringop-truncation]
      strncpy(name, syms->symbol, MAX_NAME_LEN);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  cc1: all warnings being treated as errors

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Fixes: 947b4ad1d198 ("perf list: Fix max event string size")
Link: https://lkml.kernel.org/n/tip-b663e33bm6x8hrkie4uxh7u2@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf svghelper: Fix unchecked usage of strncpy()
Arnaldo Carvalho de Melo [Thu, 6 Dec 2018 14:29:48 +0000 (11:29 -0300)]
perf svghelper: Fix unchecked usage of strncpy()

[ Upstream commit 2f5302533f306d5ee87bd375aef9ca35b91762cb ]

The strncpy() function may leave the destination string buffer
unterminated, better use strlcpy() that we have a __weak fallback
implementation for systems without it.

In this specific case this would only happen if fgets() was buggy, as
its man page states that it should read one less byte than the size of
the destination buffer, so that it can put the nul byte at the end of
it, so it would never copy 255 non-nul chars, as fgets reads into the
orig buffer at most 254 non-nul chars and terminates it. But lets just
switch to strlcpy to keep the original intent and silence the gcc 8.2
warning.

This fixes this warning on an Alpine Linux Edge system with gcc 8.2:

  In function 'cpu_model',
      inlined from 'svg_cpu_box' at util/svghelper.c:378:2:
  util/svghelper.c:337:5: error: 'strncpy' output may be truncated copying 255 bytes from a string of length 255 [-Werror=stringop-truncation]
       strncpy(cpu_m, &buf[13], 255);
       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Fixes: f48d55ce7871 ("perf: Add a SVG helper library file")
Link: https://lkml.kernel.org/n/tip-xzkoo0gyr56gej39ltivuh9g@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf tests ARM: Disable breakpoint tests 32-bit
Florian Fainelli [Mon, 3 Dec 2018 19:11:36 +0000 (11:11 -0800)]
perf tests ARM: Disable breakpoint tests 32-bit

[ Upstream commit 24f967337f6d6bce931425769c0f5ff5cf2d212e ]

The breakpoint tests on the ARM 32-bit kernel are broken in several
ways.

The breakpoint length requested does not necessarily match whether the
function address has the Thumb bit (bit 0) set or not, and this does
matter to the ARM kernel hw_breakpoint infrastructure. See [1] for
background.

[1]: https://lkml.org/lkml/2018/11/15/205

As Will indicated, the overflow handling would require single-stepping
which is not supported at the moment. Just disable those tests for the
ARM 32-bit platforms and update the comment above to explain these
limitations.

Co-developed-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20181203191138.2419-1-f.fainelli@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoperf intel-pt: Fix error with config term "pt=0"
Adrian Hunter [Mon, 26 Nov 2018 12:12:52 +0000 (14:12 +0200)]
perf intel-pt: Fix error with config term "pt=0"

[ Upstream commit 1c6f709b9f96366cc47af23c05ecec9b8c0c392d ]

Users should never use 'pt=0', but if they do it may give a meaningless
error:

$ perf record -e intel_pt/pt=0/u uname
Error:
The sys_perf_event_open() syscall returned with 22 (Invalid argument) for
event (intel_pt/pt=0/u).

Fix that by forcing 'pt=1'.

Committer testing:

  # perf record -e intel_pt/pt=0/u uname
  Error:
  The sys_perf_event_open() syscall returned with 22 (Invalid argument) for event (intel_pt/pt=0/u).
  /bin/dmesg | grep -i perf may provide additional information.

  # perf record -e intel_pt/pt=0/u uname
  pt=0 doesn't make sense, forcing pt=1
  Linux
  [ perf record: Woken up 1 times to write data ]
  [ perf record: Captured and wrote 0.020 MB perf.data ]
  #

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lkml.kernel.org/r/b7c5b4e5-9497-10e5-fd43-5f3e4a0fe51d@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agotty/serial: do not free trasnmit buffer page under port lock
Sergey Senozhatsky [Thu, 13 Dec 2018 04:58:39 +0000 (13:58 +0900)]
tty/serial: do not free trasnmit buffer page under port lock

[ Upstream commit d72402145ace0697a6a9e8e75a3de5bf3375f78d ]

LKP has hit yet another circular locking dependency between uart
console drivers and debugobjects [1]:

     CPU0                                    CPU1

                                            rhltable_init()
                                             __init_work()
                                              debug_object_init
     uart_shutdown()                          /* db->lock */
      /* uart_port->lock */                    debug_print_object()
       free_page()                              printk()
                                                 call_console_drivers()
        debug_check_no_obj_freed()                /* uart_port->lock */
         /* db->lock */
          debug_print_object()

So there are two dependency chains:
uart_port->lock -> db->lock
And
db->lock -> uart_port->lock

This particular circular locking dependency can be addressed in several
ways:

a) One way would be to move debug_print_object() out of db->lock scope
   and, thus, break the db->lock -> uart_port->lock chain.
b) Another one would be to free() transmit buffer page out of db->lock
   in UART code; which is what this patch does.

It makes sense to apply a) and b) independently: there are too many things
going on behind free(), none of which depend on uart_port->lock.

The patch fixes transmit buffer page free() in uart_shutdown() and,
additionally, in uart_port_startup() (as was suggested by Dmitry Safonov).

[1] https://lore.kernel.org/lkml/20181211091154.GL23332@shao2-debian/T/#u
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jiri Slaby <jslaby@suse.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Waiman Long <longman@redhat.com>
Cc: Dmitry Safonov <dima@arista.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agobtrfs: improve error handling of btrfs_add_link
Johannes Thumshirn [Wed, 12 Dec 2018 14:14:17 +0000 (15:14 +0100)]
btrfs: improve error handling of btrfs_add_link

[ Upstream commit 1690dd41e0cb1dade80850ed8a3eb0121b96d22f ]

In the error handling block, err holds the return value of either
btrfs_del_root_ref() or btrfs_del_inode_ref() but it hasn't been checked
since it's introduction with commit fe66a05a0679 (Btrfs: improve error
handling for btrfs_insert_dir_item callers) in 2012.

If the error handling in the error handling fails, there's not much left
to do and the abort either happened earlier in the callees or is
necessary here.

So if one of btrfs_del_root_ref() or btrfs_del_inode_ref() failed, abort
the transaction, but still return the original code of the failure
stored in 'ret' as this will be reported to the user.

Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agobtrfs: fix use-after-free due to race between replace start and cancel
Anand Jain [Wed, 14 Nov 2018 05:50:26 +0000 (13:50 +0800)]
btrfs: fix use-after-free due to race between replace start and cancel

[ Upstream commit d189dd70e2556181732598956d808ea53cc8774e ]

The device replace cancel thread can race with the replace start thread
and if fs_info::scrubs_running is not yet set, btrfs_scrub_cancel() will
fail to stop the scrub thread.

The scrub thread continues with the scrub for replace which then will
try to write to the target device and which is already freed by the
cancel thread.

scrub_setup_ctx() warns as tgtdev is NULL.

  struct scrub_ctx *scrub_setup_ctx(struct btrfs_device *dev, int is_dev_replace)
  {
  ...
  if (is_dev_replace) {
  WARN_ON(!fs_info->dev_replace.tgtdev);  <===
  sctx->pages_per_wr_bio = SCRUB_PAGES_PER_WR_BIO;
  sctx->wr_tgtdev = fs_info->dev_replace.tgtdev;
  sctx->flush_all_writes = false;
  }

  [ 6724.497655] BTRFS info (device sdb): dev_replace from /dev/sdb (devid 1) to /dev/sdc started
  [ 6753.945017] BTRFS info (device sdb): dev_replace from /dev/sdb (devid 1) to /dev/sdc canceled
  [ 6852.426700] WARNING: CPU: 0 PID: 4494 at fs/btrfs/scrub.c:622 scrub_setup_ctx.isra.19+0x220/0x230 [btrfs]
  ...
  [ 6852.428928] RIP: 0010:scrub_setup_ctx.isra.19+0x220/0x230 [btrfs]
  ...
  [ 6852.432970] Call Trace:
  [ 6852.433202]  btrfs_scrub_dev+0x19b/0x5c0 [btrfs]
  [ 6852.433471]  btrfs_dev_replace_start+0x48c/0x6a0 [btrfs]
  [ 6852.433800]  btrfs_dev_replace_by_ioctl+0x3a/0x60 [btrfs]
  [ 6852.434097]  btrfs_ioctl+0x2476/0x2d20 [btrfs]
  [ 6852.434365]  ? do_sigaction+0x7d/0x1e0
  [ 6852.434623]  do_vfs_ioctl+0xa9/0x6c0
  [ 6852.434865]  ? syscall_trace_enter+0x1c8/0x310
  [ 6852.435124]  ? syscall_trace_enter+0x1c8/0x310
  [ 6852.435387]  ksys_ioctl+0x60/0x90
  [ 6852.435663]  __x64_sys_ioctl+0x16/0x20
  [ 6852.435907]  do_syscall_64+0x50/0x180
  [ 6852.436150]  entry_SYSCALL_64_after_hwframe+0x49/0xbe

Further, as the replace thread enters scrub_write_page_to_dev_replace()
without the target device it panics:

  static int scrub_add_page_to_wr_bio(struct scrub_ctx *sctx,
      struct scrub_page *spage)
  {
  ...
bio_set_dev(bio, sbio->dev->bdev); <======

  [ 6929.715145] BUG: unable to handle kernel NULL pointer dereference at 00000000000000a0
  ..
  [ 6929.717106] Workqueue: btrfs-scrub btrfs_scrub_helper [btrfs]
  [ 6929.717420] RIP: 0010:scrub_write_page_to_dev_replace+0xb4/0x260
  [btrfs]
  ..
  [ 6929.721430] Call Trace:
  [ 6929.721663]  scrub_write_block_to_dev_replace+0x3f/0x60 [btrfs]
  [ 6929.721975]  scrub_bio_end_io_worker+0x1af/0x490 [btrfs]
  [ 6929.722277]  normal_work_helper+0xf0/0x4c0 [btrfs]
  [ 6929.722552]  process_one_work+0x1f4/0x520
  [ 6929.722805]  ? process_one_work+0x16e/0x520
  [ 6929.723063]  worker_thread+0x46/0x3d0
  [ 6929.723313]  kthread+0xf8/0x130
  [ 6929.723544]  ? process_one_work+0x520/0x520
  [ 6929.723800]  ? kthread_delayed_work_timer_fn+0x80/0x80
  [ 6929.724081]  ret_from_fork+0x3a/0x50

Fix this by letting the btrfs_dev_replace_finishing() to do the job of
cleaning after the cancel, including freeing of the target device.
btrfs_dev_replace_finishing() is called when btrfs_scub_dev() returns
along with the scrub return status.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agobtrfs: alloc_chunk: fix more DUP stripe size handling
Hans van Kranenburg [Thu, 4 Oct 2018 21:24:40 +0000 (23:24 +0200)]
btrfs: alloc_chunk: fix more DUP stripe size handling

[ Upstream commit baf92114c7e6dd6124aa3d506e4bc4b694da3bc3 ]

Commit 92e222df7b "btrfs: alloc_chunk: fix DUP stripe size handling"
fixed calculating the stripe_size for a new DUP chunk.

However, the same calculation reappears a bit later, and that one was
not changed yet. The resulting bug that is exposed is that the newly
allocated device extents ('stripes') can have a few MiB overlap with the
next thing stored after them, which is another device extent or the end
of the disk.

The scenario in which this can happen is:
* The block device for the filesystem is less than 10GiB in size.
* The amount of contiguous free unallocated disk space chosen to use for
  chunk allocation is 20% of the total device size, or a few MiB more or
  less.

An example:
- The filesystem device is 7880MiB (max_chunk_size gets set to 788MiB)
- There's 1578MiB unallocated raw disk space left in one contiguous
  piece.

In this case stripe_size is first calculated as 789MiB, (half of
1578MiB).

Since 789MiB (stripe_size * data_stripes) > 788MiB (max_chunk_size), we
enter the if block. Now stripe_size value is immediately overwritten
while calculating an adjusted value based on max_chunk_size, which ends
up as 788MiB.

Next, the value is rounded up to a 16MiB boundary, 800MiB, which is
actually more than the value we had before. However, the last comparison
fails to detect this, because it's comparing the value with the total
amount of free space, which is about twice the size of stripe_size.

In the example above, this means that the resulting raw disk space being
allocated is 1600MiB, while only a gap of 1578MiB has been found. The
second device extent object for this DUP chunk will overlap for 22MiB
with whatever comes next.

The underlying problem here is that the stripe_size is reused all the
time for different things. So, when entering the code in the if block,
stripe_size is immediately overwritten with something else. If later we
decide we want to have the previous value back, then the logic to
compute it was copy pasted in again.

With this change, the value in stripe_size is not unnecessarily
destroyed, so the duplicated calculation is not needed any more.

Signed-off-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agobtrfs: volumes: Make sure there is no overlap of dev extents at mount time
Qu Wenruo [Fri, 5 Oct 2018 09:45:54 +0000 (17:45 +0800)]
btrfs: volumes: Make sure there is no overlap of dev extents at mount time

[ Upstream commit 5eb193812a42dc49331f25137a38dfef9612d3e4 ]

Enhance btrfs_verify_dev_extents() to remember previous checked dev
extents, so it can verify no dev extents can overlap.

Analysis from Hans:

"Imagine allocating a DATA|DUP chunk.

 In the chunk allocator, we first set...
   max_stripe_size = SZ_1G;
   max_chunk_size = BTRFS_MAX_DATA_CHUNK_SIZE
 ... which is 10GiB.

 Then...
   /* we don't want a chunk larger than 10% of writeable space */
   max_chunk_size = min(div_factor(fs_devices->total_rw_bytes, 1),
         max_chunk_size);

 Imagine we only have one 7880MiB block device in this filesystem. Now
 max_chunk_size is down to 788MiB.

 The next step in the code is to search for max_stripe_size * dev_stripes
 amount of free space on the device, which is in our example 1GiB * 2 =
 2GiB. Imagine the device has exactly 1578MiB free in one contiguous
 piece. This amount of bytes will be put in devices_info[ndevs - 1].max_avail

 Next we recalculate the stripe_size (which is actually the device extent
 length), based on the actual maximum amount of available raw disk space:
   stripe_size = div_u64(devices_info[ndevs - 1].max_avail, dev_stripes);

 stripe_size is now 789MiB

 Next we do...
   data_stripes = num_stripes / ncopies
 ...where data_stripes ends up as 1, because num_stripes is 2 (the amount
 of device extents we're going to have), and DUP has ncopies 2.

 Next there's a check...
   if (stripe_size * data_stripes > max_chunk_size)
 ...which matches because 789MiB * 1 > 788MiB.

 We go into the if code, and next is...
   stripe_size = div_u64(max_chunk_size, data_stripes);
 ...which resets stripe_size to max_chunk_size: 788MiB

 Next is a fun one...
   /* bump the answer up to a 16MB boundary */
   stripe_size = round_up(stripe_size, SZ_16M);
 ...which changes stripe_size from 788MiB to 800MiB.

 We're not done changing stripe_size yet...
   /* But don't go higher than the limits we found while searching
    * for free extents
    */
   stripe_size = min(devices_info[ndevs - 1].max_avail,
              stripe_size);

 This is bad. max_avail is twice the stripe_size (we need to fit 2 device
 extents on the same device for DUP).

 The result here is that 800MiB < 1578MiB, so it's unchanged. However,
 the resulting DUP chunk will need 1600MiB disk space, which isn't there,
 and the second dev_extent might extend into the next thing (next
 dev_extent? end of device?) for 22MiB.

 The last shown line of code relies on a situation where there's twice
 the value of stripe_size present as value for the variable stripe_size
 when it's DUP. This was actually the case before commit 92e222df7b
 "btrfs: alloc_chunk: fix DUP stripe size handling", from which I quote:
   "[...] in the meantime there's a check to see if the stripe_size does
 not exceed max_chunk_size. Since during this check stripe_size is twice
 the amount as intended, the check will reduce the stripe_size to
 max_chunk_size if the actual correct to be used stripe_size is more than
 half the amount of max_chunk_size."

 In the previous version of the code, the 16MiB alignment (why is this
 done, by the way?) would result in a 50% chance that it would actually
 do an 8MiB alignment for the individual dev_extents, since it was
 operating on double the size. Does this matter?

 Does it matter that stripe_size can be set to anything which is not
 16MiB aligned because of the amount of remaining available disk space
 which is just taken?

 What is the main purpose of this round_up?

 The most straightforward thing to do seems something like...
   stripe_size = min(
       div_u64(devices_info[ndevs - 1].max_avail, dev_stripes),
       stripe_size
   )
 ..just putting half of the max_avail into stripe_size."

Link: https://lore.kernel.org/linux-btrfs/b3461a38-e5f8-f41d-c67c-2efac8129054@mendix.com/
Reported-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
[ add analysis from report ]
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agommc: atmel-mci: do not assume idle after atmci_request_end
Jonas Danielsson [Fri, 19 Oct 2018 14:40:05 +0000 (16:40 +0200)]
mmc: atmel-mci: do not assume idle after atmci_request_end

[ Upstream commit ae460c115b7aa50c9a36cf78fced07b27962c9d0 ]

On our AT91SAM9260 board we use the same sdio bus for wifi and for the
sd card slot. This caused the atmel-mci to give the following splat on
the serial console:

  ------------[ cut here ]------------
  WARNING: CPU: 0 PID: 538 at drivers/mmc/host/atmel-mci.c:859 atmci_send_command+0x24/0x44
  Modules linked in:
  CPU: 0 PID: 538 Comm: mmcqd/0 Not tainted 4.14.76 #14
  Hardware name: Atmel AT91SAM9
  [<c000fccc>] (unwind_backtrace) from [<c000d3dc>] (show_stack+0x10/0x14)
  [<c000d3dc>] (show_stack) from [<c0017644>] (__warn+0xd8/0xf4)
  [<c0017644>] (__warn) from [<c0017704>] (warn_slowpath_null+0x1c/0x24)
  [<c0017704>] (warn_slowpath_null) from [<c033bb9c>] (atmci_send_command+0x24/0x44)
  [<c033bb9c>] (atmci_send_command) from [<c033e984>] (atmci_start_request+0x1f4/0x2dc)
  [<c033e984>] (atmci_start_request) from [<c033f3b4>] (atmci_request+0xf0/0x164)
  [<c033f3b4>] (atmci_request) from [<c0327108>] (mmc_start_request+0x280/0x2d0)
  [<c0327108>] (mmc_start_request) from [<c032800c>] (mmc_start_areq+0x230/0x330)
  [<c032800c>] (mmc_start_areq) from [<c03366f8>] (mmc_blk_issue_rw_rq+0xc4/0x310)
  [<c03366f8>] (mmc_blk_issue_rw_rq) from [<c03372c4>] (mmc_blk_issue_rq+0x118/0x5ac)
  [<c03372c4>] (mmc_blk_issue_rq) from [<c033781c>] (mmc_queue_thread+0xc4/0x118)
  [<c033781c>] (mmc_queue_thread) from [<c002daf8>] (kthread+0x100/0x118)
  [<c002daf8>] (kthread) from [<c000a580>] (ret_from_fork+0x14/0x34)
  ---[ end trace 594371ddfa284bd6 ]---

This is:
  WARN_ON(host->cmd);

This was fixed on our board by letting atmci_request_end determine what
state we are in. Instead of unconditionally setting it to STATE_IDLE on
STATE_END_REQUEST.

Signed-off-by: Jonas Danielsson <jonas@orbital-systems.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agokconfig: fix memory leak when EOF is encountered in quotation
Masahiro Yamada [Tue, 11 Dec 2018 11:00:45 +0000 (20:00 +0900)]
kconfig: fix memory leak when EOF is encountered in quotation

[ Upstream commit fbac5977d81cb2b2b7e37b11c459055d9585273c ]

An unterminated string literal followed by new line is passed to the
parser (with "multi-line strings not supported" warning shown), then
handled properly there.

On the other hand, an unterminated string literal at end of file is
never passed to the parser, then results in memory leak.

[Test Code]

  ----------(Kconfig begin)----------
  source "Kconfig.inc"

  config A
          bool "a"
  -----------(Kconfig end)-----------

  --------(Kconfig.inc begin)--------
  config B
          bool "b\No new line at end of file
  ---------(Kconfig.inc end)---------

[Summary from Valgrind]

  Before the fix:

    LEAK SUMMARY:
       definitely lost: 16 bytes in 1 blocks
       ...

  After the fix:

    LEAK SUMMARY:
       definitely lost: 0 bytes in 0 blocks
       ...

Eliminate the memory leak path by handling this case. Of course, such
a Kconfig file is wrong already, so I will add an error message later.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agokconfig: fix file name and line number of warn_ignored_character()
Masahiro Yamada [Tue, 11 Dec 2018 11:00:44 +0000 (20:00 +0900)]
kconfig: fix file name and line number of warn_ignored_character()

[ Upstream commit 77c1c0fa8b1477c5799bdad65026ea5ff676da44 ]

Currently, warn_ignore_character() displays invalid file name and
line number.

The lexer should use current_file->name and yylineno, while the parser
should use zconf_curname() and zconf_lineno().

This difference comes from that the lexer is always going ahead
of the parser. The parser needs to look ahead one token to make a
shift/reduce decision, so the lexer is requested to scan more text
from the input file.

This commit fixes the warning message from warn_ignored_character().

[Test Code]

  ----(Kconfig begin)----
  /
  -----(Kconfig end)-----

[Output]

  Before the fix:

  <none>:0:warning: ignoring unsupported character '/'

  After the fix:

  Kconfig:1:warning: ignoring unsupported character '/'

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agobpf: relax verifier restriction on BPF_MOV | BPF_ALU
Jiong Wang [Fri, 7 Dec 2018 17:16:18 +0000 (12:16 -0500)]
bpf: relax verifier restriction on BPF_MOV | BPF_ALU

[ Upstream commit e434b8cdf788568ba65a0a0fd9f3cb41f3ca1803 ]

Currently, the destination register is marked as unknown for 32-bit
sub-register move (BPF_MOV | BPF_ALU) whenever the source register type is
SCALAR_VALUE.

This is too conservative that some valid cases will be rejected.
Especially, this may turn a constant scalar value into unknown value that
could break some assumptions of verifier.

For example, test_l4lb_noinline.c has the following C code:

    struct real_definition *dst

1:  if (!get_packet_dst(&dst, &pckt, vip_info, is_ipv6))
2:    return TC_ACT_SHOT;
3:
4:  if (dst->flags & F_IPV6) {

get_packet_dst is responsible for initializing "dst" into valid pointer and
return true (1), otherwise return false (0). The compiled instruction
sequence using alu32 will be:

  412: (54) (u32) r7 &= (u32) 1
  413: (bc) (u32) r0 = (u32) r7
  414: (95) exit

insn 413, a BPF_MOV | BPF_ALU, however will turn r0 into unknown value even
r7 contains SCALAR_VALUE 1.

This causes trouble when verifier is walking the code path that hasn't
initialized "dst" inside get_packet_dst, for which case 0 is returned and
we would then expect verifier concluding line 1 in the above C code pass
the "if" check, therefore would skip fall through path starting at line 4.
Now, because r0 returned from callee has became unknown value, so verifier
won't skip analyzing path starting at line 4 and "dst->flags" requires
dereferencing the pointer "dst" which actually hasn't be initialized for
this path.

This patch relaxed the code marking sub-register move destination. For a
SCALAR_VALUE, it is safe to just copy the value from source then truncate
it into 32-bit.

A unit test also included to demonstrate this issue. This test will fail
before this patch.

This relaxation could let verifier skipping more paths for conditional
comparison against immediate. It also let verifier recording a more
accurate/strict value for one register at one state, if this state end up
with going through exit without rejection and it is used for state
comparison later, then it is possible an inaccurate/permissive value is
better. So the real impact on verifier processed insn number is complex.
But in all, without this fix, valid program could be rejected.

>From real benchmarking on kernel selftests and Cilium bpf tests, there is
no impact on processed instruction number when tests ares compiled with
default compilation options. There is slightly improvements when they are
compiled with -mattr=+alu32 after this patch.

Also, test_xdp_noinline/-mattr=+alu32 now passed verification. It is
rejected before this fix.

Insn processed before/after this patch:

                        default     -mattr=+alu32

Kernel selftest

===
test_xdp.o              371/371      369/369
test_l4lb.o             6345/6345    5623/5623
test_xdp_noinline.o     2971/2971    rejected/2727
test_tcp_estates.o      429/429      430/430

Cilium bpf
===
bpf_lb-DLB_L3.o:        2085/2085     1685/1687
bpf_lb-DLB_L4.o:        2287/2287     1986/1982
bpf_lb-DUNKNOWN.o:      690/690       622/622
bpf_lxc.o:              95033/95033   N/A
bpf_netdev.o:           7245/7245     N/A
bpf_overlay.o:          2898/2898     3085/2947

NOTE:
  - bpf_lxc.o and bpf_netdev.o compiled by -mattr=+alu32 are rejected by
    verifier due to another issue inside verifier on supporting alu32
    binary.
  - Each cilium bpf program could generate several processed insn number,
    above number is sum of them.

v1->v2:
 - Restrict the change on SCALAR_VALUE.
 - Update benchmark numbers on Cilium bpf tests.

Signed-off-by: Jiong Wang <jiong.wang@netronome.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoarm64: Fix minor issues with the dcache_by_line_op macro
Will Deacon [Mon, 10 Dec 2018 13:39:48 +0000 (13:39 +0000)]
arm64: Fix minor issues with the dcache_by_line_op macro

[ Upstream commit 33309ecda0070506c49182530abe7728850ebe78 ]

The dcache_by_line_op macro suffers from a couple of small problems:

First, the GAS directives that are currently being used rely on
assembler behavior that is not documented, and probably not guaranteed
to produce the correct behavior going forward. As a result, we end up
with some undefined symbols in cache.o:

$ nm arch/arm64/mm/cache.o
         ...
         U civac
         ...
         U cvac
         U cvap
         U cvau

This is due to the fact that the comparisons used to select the
operation type in the dcache_by_line_op macro are comparing symbols
not strings, and even though it seems that GAS is doing the right
thing here (undefined symbols by the same name are equal to each
other), it seems unwise to rely on this.

Second, when patching in a DC CVAP instruction on CPUs that support it,
the fallback path consists of a DC CVAU instruction which may be
affected by CPU errata that require ARM64_WORKAROUND_CLEAN_CACHE.

Solve these issues by unrolling the various maintenance routines and
using the conditional directives that are documented as operating on
strings. To avoid the complexity of nested alternatives, we move the
DC CVAP patching to __clean_dcache_area_pop, falling back to a branch
to __clean_dcache_area_poc if DCPOP is not supported by the CPU.

Reported-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Suggested-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoclk: imx6q: reset exclusive gates on init
Lucas Stach [Thu, 15 Nov 2018 14:30:26 +0000 (15:30 +0100)]
clk: imx6q: reset exclusive gates on init

[ Upstream commit f7542d817733f461258fd3a47d77da35b2d9fc81 ]

The exclusive gates may be set up in the wrong way by software running
before the clock driver comes up. In that case the exclusive setup is
locked in its initial state, as the complementary function can't be
activated without disabling the initial setup first.

To avoid this lock situation, reset the exclusive gates to the off
state and allow the kernel to provide the proper setup.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Dong Aisheng <Aisheng.dong@nxp.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoarm64: kasan: Increase stack size for KASAN_EXTRA
Qian Cai [Fri, 7 Dec 2018 22:34:49 +0000 (17:34 -0500)]
arm64: kasan: Increase stack size for KASAN_EXTRA

[ Upstream commit 6e8830674ea77f57d57a33cca09083b117a71f41 ]

If the kernel is configured with KASAN_EXTRA, the stack size is
increased significantly due to setting the GCC -fstack-reuse option to
"none" [1]. As a result, it can trigger a stack overrun quite often with
32k stack size compiled using GCC 8. For example, this reproducer

  https://github.com/linux-test-project/ltp/blob/master/testcases/kernel/syscalls/madvise/madvise06.c

can trigger a "corrupted stack end detected inside scheduler" very
reliably with CONFIG_SCHED_STACK_END_CHECK enabled. There are other
reports at:

  https://lore.kernel.org/lkml/1542144497.12945.29.camel@gmx.us/
  https://lore.kernel.org/lkml/721E7B42-2D55-4866-9C1A-3E8D64F33F9C@gmx.us/

There are just too many functions that could have a large stack with
KASAN_EXTRA due to large local variables that have been called over and
over again without being able to reuse the stacks. Some noticiable ones
are,

size
7536 shrink_inactive_list
7440 shrink_page_list
6560 fscache_stats_show
3920 jbd2_journal_commit_transaction
3216 try_to_unmap_one
3072 migrate_page_move_mapping
3584 migrate_misplaced_transhuge_page
3920 ip_vs_lblcr_schedule
4304 lpfc_nvme_info_show
3888 lpfc_debugfs_nvmestat_data.constprop

There are other 49 functions over 2k in size while compiling kernel with
"-Wframe-larger-than=" on this machine. Hence, it is too much work to
change Makefiles for each object to compile without
-fsanitize-address-use-after-scope individually.

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81715#c23

Signed-off-by: Qian Cai <cai@lca.pw>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoselftests: do not macro-expand failed assertion expressions
Dmitry V. Levin [Sun, 9 Dec 2018 23:00:47 +0000 (02:00 +0300)]
selftests: do not macro-expand failed assertion expressions

[ Upstream commit b708a3cc9600390ccaa2b68a88087dd265154b2b ]

I've stumbled over the current macro-expand behaviour of the test
harness:

$ gcc -Wall -xc - <<'__EOF__'
TEST(macro) {
int status = 0;
ASSERT_TRUE(WIFSIGNALED(status));
}
TEST_HARNESS_MAIN
__EOF__
$ ./a.out
[==========] Running 1 tests from 1 test cases.
[ RUN      ] global.macro
<stdin>:4:global.macro:Expected 0 (0) != (((signed char) (((status) & 0x7f) + 1) >> 1) > 0) (0)
global.macro: Test terminated by assertion
[     FAIL ] global.macro
[==========] 0 / 1 tests passed.
[  FAILED  ]

With this change the output of the same test looks much more
comprehensible:

[==========] Running 1 tests from 1 test cases.
[ RUN      ] global.macro
<stdin>:4:global.macro:Expected 0 (0) != WIFSIGNALED(status) (0)
global.macro: Test terminated by assertion
[     FAIL ] global.macro
[==========] 0 / 1 tests passed.
[  FAILED  ]

The issue is very similar to the bug fixed in glibc assert(3)
three years ago:
https://sourceware.org/bugzilla/show_bug.cgi?id=18604

Cc: Shuah Khan <shuah@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Will Drewry <wad@chromium.org>
Cc: linux-kselftest@vger.kernel.org
Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Shuah Khan <shuah@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: target/core: Make sure that target_wait_for_sess_cmds() waits long enough
Bart Van Assche [Tue, 27 Nov 2018 23:51:58 +0000 (15:51 -0800)]
scsi: target/core: Make sure that target_wait_for_sess_cmds() waits long enough

[ Upstream commit ad669505c4e9db9af9faeb5c51aa399326a80d91 ]

A session must only be released after all code that accesses the session
structure has finished. Make sure that this is the case by introducing a
new command counter per session that is only decremented after the
.release_cmd() callback has finished. This patch fixes the following crash:

BUG: KASAN: use-after-free in do_raw_spin_lock+0x1c/0x130
Read of size 4 at addr ffff8801534b16e4 by task rmdir/14805
CPU: 16 PID: 14805 Comm: rmdir Not tainted 4.18.0-rc2-dbg+ #5
Call Trace:
dump_stack+0xa4/0xf5
print_address_description+0x6f/0x270
kasan_report+0x241/0x360
__asan_load4+0x78/0x80
do_raw_spin_lock+0x1c/0x130
_raw_spin_lock_irqsave+0x52/0x60
srpt_set_ch_state+0x27/0x70 [ib_srpt]
srpt_disconnect_ch+0x1b/0xc0 [ib_srpt]
srpt_close_session+0xa8/0x260 [ib_srpt]
target_shutdown_sessions+0x170/0x180 [target_core_mod]
core_tpg_del_initiator_node_acl+0xf3/0x200 [target_core_mod]
target_fabric_nacl_base_release+0x25/0x30 [target_core_mod]
config_item_release+0x9c/0x110 [configfs]
config_item_put+0x26/0x30 [configfs]
configfs_rmdir+0x3b8/0x510 [configfs]
vfs_rmdir+0xb3/0x1e0
do_rmdir+0x262/0x2c0
do_syscall_64+0x77/0x230
entry_SYSCALL_64_after_hwframe+0x49/0xbe

Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Cc: Mike Christie <mchristi@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Disseldorp <ddiss@suse.de>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: target: use consistent left-aligned ASCII INQUIRY data
David Disseldorp [Wed, 5 Dec 2018 12:18:34 +0000 (13:18 +0100)]
scsi: target: use consistent left-aligned ASCII INQUIRY data

[ Upstream commit 0de263577de5d5e052be5f4f93334e63cc8a7f0b ]

spc5r17.pdf specifies:

  4.3.1 ASCII data field requirements
  ASCII data fields shall contain only ASCII printable characters (i.e.,
  code values 20h to 7Eh) and may be terminated with one or more ASCII null
  (00h) characters.  ASCII data fields described as being left-aligned
  shall have any unused bytes at the end of the field (i.e., highest
  offset) and the unused bytes shall be filled with ASCII space characters
  (20h).

LIO currently space-pads the T10 VENDOR IDENTIFICATION and PRODUCT
IDENTIFICATION fields in the standard INQUIRY data. However, the PRODUCT
REVISION LEVEL field in the standard INQUIRY data as well as the T10 VENDOR
IDENTIFICATION field in the INQUIRY Device Identification VPD Page are
zero-terminated/zero-padded.

Fix this inconsistency by using space-padding for all of the above fields.

Signed-off-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bryant G. Ly <bly@catalogicsoftware.com>
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agonet: call sk_dst_reset when set SO_DONTROUTE
yupeng [Thu, 6 Dec 2018 02:56:28 +0000 (18:56 -0800)]
net: call sk_dst_reset when set SO_DONTROUTE

[ Upstream commit 0fbe82e628c817e292ff588cd5847fc935e025f2 ]

after set SO_DONTROUTE to 1, the IP layer should not route packets if
the dest IP address is not in link scope. But if the socket has cached
the dst_entry, such packets would be routed until the sk_dst_cache
expires. So we should clean the sk_dst_cache when a user set
SO_DONTROUTE option. Below are server/client python scripts which
could reprodue this issue:

server side code:

==========================================================================
import socket
import struct
import time

s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.bind(('0.0.0.0', 9000))
s.listen(1)
sock, addr = s.accept()
sock.setsockopt(socket.SOL_SOCKET, socket.SO_DONTROUTE, struct.pack('i', 1))
while True:
    sock.send(b'foo')
    time.sleep(1)
==========================================================================

client side code:
==========================================================================
import socket
import time

s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.connect(('server_address', 9000))
while True:
    data = s.recv(1024)
    print(data)
==========================================================================

Signed-off-by: yupeng <yupeng0921@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agostaging: erofs: fix use-after-free of on-stack `z_erofs_vle_unzip_io'
Gao Xiang [Fri, 7 Dec 2018 16:19:12 +0000 (00:19 +0800)]
staging: erofs: fix use-after-free of on-stack `z_erofs_vle_unzip_io'

[ Upstream commit 848bd9acdcd00c164b42b14aacec242949ecd471 ]

The root cause is the race as follows:
 Thread #0                         Thread #1

 z_erofs_vle_unzip_kickoff         z_erofs_submit_and_unzip

                                    struct z_erofs_vle_unzip_io io[]
   atomic_add_return()
                                    wait_event()
                                    [end of function]
   wake_up()

Fix it by taking the waitqueue lock between atomic_add_return and
wake_up to close such the race.

kernel message:

Unable to handle kernel paging request at virtual address 97f7052caa1303dc
...
Workqueue: kverityd verity_work
task: ffffffe32bcb8000 task.stack: ffffffe3298a0000
PC is at __wake_up_common+0x48/0xa8
LR is at __wake_up+0x3c/0x58
...
Call trace:
...
[<ffffff94a08ff648>] __wake_up_common+0x48/0xa8
[<ffffff94a08ff8b8>] __wake_up+0x3c/0x58
[<ffffff94a0c11b60>] z_erofs_vle_unzip_kickoff+0x40/0x64
[<ffffff94a0c118e4>] z_erofs_vle_read_endio+0x94/0x134
[<ffffff94a0c83c9c>] bio_endio+0xe4/0xf8
[<ffffff94a1076540>] dec_pending+0x134/0x32c
[<ffffff94a1076f28>] clone_endio+0x90/0xf4
[<ffffff94a0c83c9c>] bio_endio+0xe4/0xf8
[<ffffff94a1095024>] verity_work+0x210/0x368
[<ffffff94a08c4150>] process_one_work+0x188/0x4b4
[<ffffff94a08c45bc>] worker_thread+0x140/0x458
[<ffffff94a08cad48>] kthread+0xec/0x108
[<ffffff94a0883ab4>] ret_from_fork+0x10/0x1c
Code: d1006273 54000260 f9400804 b9400019 (b85fc081)
---[ end trace be9dde154f677cd1 ]---

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agomedia: venus: core: Set dma maximum segment size
Vivek Gautam [Wed, 5 Dec 2018 08:31:51 +0000 (03:31 -0500)]
media: venus: core: Set dma maximum segment size

[ Upstream commit de2563bce7a157f5296bab94f3843d7d64fb14b4 ]

Turning on CONFIG_DMA_API_DEBUG_SG results in the following error:

[  460.308650] ------------[ cut here ]------------
[  460.313490] qcom-venus aa00000.video-codec: DMA-API: mapping sg segment longer than device claims to support [len=4194304] [max=65536]
[  460.326017] WARNING: CPU: 3 PID: 3555 at src/kernel/dma/debug.c:1301 debug_dma_map_sg+0x174/0x254
[  460.338888] Modules linked in: venus_dec venus_enc videobuf2_dma_sg videobuf2_memops hci_uart btqca bluetooth venus_core v4l2_mem2mem videobuf2_v4l2 videobuf2_common ath10k_snoc ath10k_core ath lzo lzo_compress zramjoydev
[  460.375811] CPU: 3 PID: 3555 Comm: V4L2DecoderThre Tainted: G        W         4.19.1 #82
[  460.384223] Hardware name: Google Cheza (rev1) (DT)
[  460.389251] pstate: 60400009 (nZCv daif +PAN -UAO)
[  460.394191] pc : debug_dma_map_sg+0x174/0x254
[  460.398680] lr : debug_dma_map_sg+0x174/0x254
[  460.403162] sp : ffffff80200c37d0
[  460.406583] x29: ffffff80200c3830 x28: 0000000000010000
[  460.412056] x27: 00000000ffffffff x26: ffffffc0f785ea80
[  460.417532] x25: 0000000000000000 x24: ffffffc0f4ea1290
[  460.423001] x23: ffffffc09e700300 x22: ffffffc0f4ea1290
[  460.428470] x21: ffffff8009037000 x20: 0000000000000001
[  460.433936] x19: ffffff80091b0000 x18: 0000000000000000
[  460.439411] x17: 0000000000000000 x16: 000000000000f251
[  460.444885] x15: 0000000000000006 x14: 0720072007200720
[  460.450354] x13: ffffff800af536e0 x12: 0000000000000000
[  460.455822] x11: 0000000000000000 x10: 0000000000000000
[  460.461288] x9 : 537944d9c6c48d00 x8 : 537944d9c6c48d00
[  460.466758] x7 : 0000000000000000 x6 : ffffffc0f8d98f80
[  460.472230] x5 : 0000000000000000 x4 : 0000000000000000
[  460.477703] x3 : 000000000000008a x2 : ffffffc0fdb13948
[  460.483170] x1 : ffffffc0fdb0b0b0 x0 : 000000000000007a
[  460.488640] Call trace:
[  460.491165]  debug_dma_map_sg+0x174/0x254
[  460.495307]  vb2_dma_sg_alloc+0x260/0x2dc [videobuf2_dma_sg]
[  460.501150]  __vb2_queue_alloc+0x164/0x374 [videobuf2_common]
[  460.507076]  vb2_core_reqbufs+0xfc/0x23c [videobuf2_common]
[  460.512815]  vb2_reqbufs+0x44/0x5c [videobuf2_v4l2]
[  460.517853]  v4l2_m2m_reqbufs+0x44/0x78 [v4l2_mem2mem]
[  460.523144]  v4l2_m2m_ioctl_reqbufs+0x1c/0x28 [v4l2_mem2mem]
[  460.528976]  v4l_reqbufs+0x30/0x40
[  460.532480]  __video_do_ioctl+0x36c/0x454
[  460.536610]  video_usercopy+0x25c/0x51c
[  460.540572]  video_ioctl2+0x38/0x48
[  460.544176]  v4l2_ioctl+0x60/0x74
[  460.547602]  do_video_ioctl+0x948/0x3520
[  460.551648]  v4l2_compat_ioctl32+0x60/0x98
[  460.555872]  __arm64_compat_sys_ioctl+0x134/0x20c
[  460.560718]  el0_svc_common+0x9c/0xe4
[  460.564498]  el0_svc_compat_handler+0x2c/0x38
[  460.568982]  el0_svc_compat+0x8/0x18
[  460.572672] ---[ end trace ce209b87b2f3af88 ]---

>From above warning one would deduce that the sg segment will overflow
the device's capacity. In reality, the hardware can accommodate larger
sg segments.
So, initialize the max segment size properly to weed out this warning.

Based on a similar patch sent by Sean Paul for mdss:
https://patchwork.kernel.org/patch/10671457/

Signed-off-by: Vivek Gautam <vivek.gautam@codeaurora.org>
Acked-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoASoC: use dma_ops of parent device for acp_audio_dma
Yu Zhao [Tue, 4 Dec 2018 22:42:53 +0000 (15:42 -0700)]
ASoC: use dma_ops of parent device for acp_audio_dma

[ Upstream commit 23aa128bb28d9da69bb1bdb2b70e50128857884a ]

AMD platform device acp_audio_dma can only be created by parent PCI
device driver (drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c). Pass struct
device of the parent to snd_pcm_lib_preallocate_pages() so
dma_alloc_coherent() can use correct dma_ops. Otherwise, it will
use default dma_ops which is nommu_dma_ops on x86_64 even when
IOMMU is enabled and set to non passthrough mode.

Though platform device inherits some dma related fields during its
creation in mfd_add_device(), we can't simply pass its struct device
to snd_pcm_lib_preallocate_pages() because dma_ops is not among the
inherited fields. Even it were, drivers/iommu/amd_iommu.c would
ignore it because get_device_id() doesn't handle platform device.

This change shouldn't give us any trouble even struct device of the
parent becomes null or represents some non PCI device in the future,
because get_dma_ops() correctly handles null struct device or uses
the default dma_ops if struct device doesn't have it set.

Signed-off-by: Yu Zhao <yuzhao@google.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agomedia: firewire: Fix app_info parameter type in avc_ca{,_app}_info
Nathan Chancellor [Thu, 18 Oct 2018 20:03:06 +0000 (16:03 -0400)]
media: firewire: Fix app_info parameter type in avc_ca{,_app}_info

[ Upstream commit b2e9a4eda11fd2cb1e6714e9ad3f455c402568ff ]

Clang warns:

drivers/media/firewire/firedtv-avc.c:999:45: warning: implicit
conversion from 'int' to 'char' changes value from 159 to -97
[-Wconstant-conversion]
        app_info[0] = (EN50221_TAG_APP_INFO >> 16) & 0xff;
                    ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
drivers/media/firewire/firedtv-avc.c:1000:45: warning: implicit
conversion from 'int' to 'char' changes value from 128 to -128
[-Wconstant-conversion]
        app_info[1] = (EN50221_TAG_APP_INFO >>  8) & 0xff;
                    ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
drivers/media/firewire/firedtv-avc.c:1040:44: warning: implicit
conversion from 'int' to 'char' changes value from 159 to -97
[-Wconstant-conversion]
        app_info[0] = (EN50221_TAG_CA_INFO >> 16) & 0xff;
                    ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
drivers/media/firewire/firedtv-avc.c:1041:44: warning: implicit
conversion from 'int' to 'char' changes value from 128 to -128
[-Wconstant-conversion]
        app_info[1] = (EN50221_TAG_CA_INFO >>  8) & 0xff;
                    ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
4 warnings generated.

Change app_info's type to unsigned char to match the type of the
member msg in struct ca_msg, which is the only thing passed into the
app_info parameter in this function.

Link: https://github.com/ClangBuiltLinux/linux/issues/105
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agopowerpc/pseries/cpuidle: Fix preempt warning
Breno Leitao [Fri, 23 Nov 2018 16:30:11 +0000 (14:30 -0200)]
powerpc/pseries/cpuidle: Fix preempt warning

[ Upstream commit 2b038cbc5fcf12a7ee1cc9bfd5da1e46dacdee87 ]

When booting a pseries kernel with PREEMPT enabled, it dumps the
following warning:

   BUG: using smp_processor_id() in preemptible [00000000] code: swapper/0/1
   caller is pseries_processor_idle_init+0x5c/0x22c
   CPU: 13 PID: 1 Comm: swapper/0 Not tainted 4.20.0-rc3-00090-g12201a0128bc-dirty #828
   Call Trace:
   [c000000429437ab0] [c0000000009c8878] dump_stack+0xec/0x164 (unreliable)
   [c000000429437b00] [c0000000005f2f24] check_preemption_disabled+0x154/0x160
   [c000000429437b90] [c000000000cab8e8] pseries_processor_idle_init+0x5c/0x22c
   [c000000429437c10] [c000000000010ed4] do_one_initcall+0x64/0x300
   [c000000429437ce0] [c000000000c54500] kernel_init_freeable+0x3f0/0x500
   [c000000429437db0] [c0000000000112dc] kernel_init+0x2c/0x160
   [c000000429437e20] [c00000000000c1d0] ret_from_kernel_thread+0x5c/0x6c

This happens because the code calls get_lppaca() which calls
get_paca() and it checks if preemption is disabled through
check_preemption_disabled().

Preemption should be disabled because the per CPU variable may make no
sense if there is a preemption (and a CPU switch) after it reads the
per CPU data and when it is used.

In this device driver specifically, it is not a problem, because this
code just needs to have access to one lppaca struct, and it does not
matter if it is the current per CPU lppaca struct or not (i.e. when
there is a preemption and a CPU migration).

That said, the most appropriate fix seems to be related to avoiding
the debug_smp_processor_id() call at get_paca(), instead of calling
preempt_disable() before get_paca().

Signed-off-by: Breno Leitao <leitao@debian.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agopowerpc/xmon: Fix invocation inside lock region
Breno Leitao [Thu, 8 Nov 2018 17:12:42 +0000 (15:12 -0200)]
powerpc/xmon: Fix invocation inside lock region

[ Upstream commit 8d4a862276a9c30a269d368d324fb56529e6d5fd ]

Currently xmon needs to get devtree_lock (through rtas_token()) during its
invocation (at crash time). If there is a crash while devtree_lock is being
held, then xmon tries to get the lock but spins forever and never get into
the interactive debugger, as in the following case:

int *ptr = NULL;
raw_spin_lock_irqsave(&devtree_lock, flags);
*ptr = 0xdeadbeef;

This patch avoids calling rtas_token(), thus trying to get the same lock,
at crash time. This new mechanism proposes getting the token at
initialization time (xmon_init()) and just consuming it at crash time.

This would allow xmon to be possible invoked independent of devtree_lock
being held or not.

Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agomedia: uvcvideo: Refactor teardown of uvc on USB disconnect
Daniel Axtens [Sun, 23 Apr 2017 04:53:49 +0000 (00:53 -0400)]
media: uvcvideo: Refactor teardown of uvc on USB disconnect

[ Upstream commit 10e1fdb95809ed21406f53b5b4f064673a1b9ceb ]

Currently, disconnecting a USB webcam while it is in use prints out a
number of warnings, such as:

WARNING: CPU: 2 PID: 3118 at /build/linux-ezBi1T/linux-4.8.0/fs/sysfs/group.c:237 sysfs_remove_group+0x8b/0x90
sysfs group ffffffffa7cd0780 not found for kobject 'event13'

This has been noticed before. [0]

This is because of the order in which things are torn down.

If there are no streams active during a USB disconnect:

 - uvc_disconnect() is invoked via device_del() through the bus
   notifier mechanism.

 - this calls uvc_unregister_video().

 - uvc_unregister_video() unregisters the video device for each
   stream,

 - because there are no streams open, it calls uvc_delete()

 - uvc_delete() calls uvc_status_cleanup(), which cleans up the status
   input device.

 - uvc_delete() calls media_device_unregister(), which cleans up the
   media device

 - uvc_delete(), uvc_unregister_video() and uvc_disconnect() all
   return, and we end up back in device_del().

 - device_del() then cleans up the sysfs folder for the camera with
   dpm_sysfs_remove(). Because uvc_status_cleanup() and
   media_device_unregister() have already been called, this all works
   nicely.

If, on the other hand, there *are* streams active during a USB disconnect:

 - uvc_disconnect() is invoked

 - this calls uvc_unregister_video()

 - uvc_unregister_video() unregisters the video device for each
   stream,

 - uvc_unregister_video() and uvc_disconnect() return, and we end up
   back in device_del().

 - device_del() then cleans up the sysfs folder for the camera with
   dpm_sysfs_remove(). Because the status input device and the media
   device are children of the USB device, this also deletes their
   sysfs folders.

 - Sometime later, the final stream is closed, invoking uvc_release().

 - uvc_release() calls uvc_delete()

 - uvc_delete() calls uvc_status_cleanup(), which cleans up the status
   input device. Because the sysfs directory has already been removed,
   this causes a WARNing.

 - uvc_delete() calls media_device_unregister(), which cleans up the
   media device. Because the sysfs directory has already been removed,
   this causes another WARNing.

To fix this, we need to make sure the devices are always unregistered
before the end of uvc_disconnect(). To this, move the unregistration
into the disconnect path:

 - split uvc_status_cleanup() into two parts, one on disconnect that
   unregisters and one on delete that frees.

 - move v4l2_device_unregister() and media_device_unregister() into
   the disconnect path.

[0]: https://lkml.org/lkml/2016/12/8/657

[Renamed uvc_input_cleanup() to uvc_input_unregister()]

Signed-off-by: Daniel Axtens <dja@axtens.net>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agopstore/ram: Do not treat empty buffers as valid
Joel Fernandes (Google) [Sat, 3 Nov 2018 23:38:18 +0000 (16:38 -0700)]
pstore/ram: Do not treat empty buffers as valid

[ Upstream commit 30696378f68a9e3dad6bfe55938b112e72af00c2 ]

The ramoops backend currently calls persistent_ram_save_old() even
if a buffer is empty. While this appears to work, it is does not seem
like the right thing to do and could lead to future bugs so lets avoid
that. It also prevents misleading prints in the logs which claim the
buffer is valid.

I got something like:

found existing buffer, size 0, start 0

When I was expecting:

no valid data in buffer (sig = ...)

This bails out early (and reports with pr_debug()), since it's an
acceptable state.

Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Co-developed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoclk: imx: make mux parent strings const
A.s. Dong [Wed, 14 Nov 2018 13:02:00 +0000 (13:02 +0000)]
clk: imx: make mux parent strings const

[ Upstream commit 9e5ef7a57ca75a1b9411c46caeeb6881124284a3 ]

As the commit 2893c379461a ("clk: make strings in parent name arrays
const"), let's make the parent strings const, otherwise we may meet
the following warning when compiling:

drivers/clk/imx/clk-imx7ulp.c: In function 'imx7ulp_clocks_init':
drivers/clk/imx/clk-imx7ulp.c:73:35: warning: passing argument 5 of
'imx_clk_mux_flags' discards 'const' qualifier from pointer target type

  clks[IMX7ULP_CLK_APLL_PRE_SEL] = imx_clk_mux_flags("apll_pre_sel", base + 0x508, 0,
1, pll_pre_sels, ARRAY_SIZE(pll_pre_sels), CLK_SET_PARENT_GATE);
                                   ^
In file included from drivers/clk/imx/clk-imx7ulp.c:23:0:
drivers/clk/imx/clk.h:200:27: note: expected 'const char **' but argument is
 of type 'const char * const*'
...

Cc: Stephen Boyd <sboyd@codeaurora.org>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Shawn Guo <shawnguo@kernel.org>
Signed-off-by: Dong Aisheng <aisheng.dong@nxp.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agojffs2: Fix use of uninitialized delayed_work, lockdep breakage
Daniel Santos [Fri, 19 Oct 2018 08:30:20 +0000 (03:30 -0500)]
jffs2: Fix use of uninitialized delayed_work, lockdep breakage

[ Upstream commit a788c5272769ddbcdbab297cf386413eeac04463 ]

jffs2_sync_fs makes the assumption that if CONFIG_JFFS2_FS_WRITEBUFFER
is defined then a write buffer is available and has been initialized.
However, this does is not the case when the mtd device has no
out-of-band buffer:

int jffs2_nand_flash_setup(struct jffs2_sb_info *c)
{
        if (!c->mtd->oobsize)
                return 0;
...

The resulting call to cancel_delayed_work_sync passing a uninitialized
(but zeroed) delayed_work struct forces lockdep to become disabled.

[   90.050639] overlayfs: upper fs does not support tmpfile.
[   90.652264] INFO: trying to register non-static key.
[   90.662171] the code is fine but needs lockdep annotation.
[   90.673090] turning off the locking correctness validator.
[   90.684021] CPU: 0 PID: 1762 Comm: mount_root Not tainted 4.14.63 #0
[   90.696672] Stack : 00000000 00000000 80d8f6a2 00000038 805f0000 80444600 8fe364f4 805dfbe7
[   90.713349]         80563a30 000006e2 8068370c 00000001 00000000 00000001 8e2fdc48 ffffffff
[   90.730020]         00000000 00000000 80d90000 00000000 00000106 00000000 6465746e 312e3420
[   90.746690]         6b636f6c 03bf0000 f8000000 20676e69 00000000 80000000 00000000 8e2c2a90
[   90.763362]         80d90000 00000001 00000000 8e2c2a90 00000003 80260dc0 08052098 80680000
[   90.780033]         ...
[   90.784902] Call Trace:
[   90.789793] [<8000f0d8>] show_stack+0xb8/0x148
[   90.798659] [<8005a000>] register_lock_class+0x270/0x55c
[   90.809247] [<8005cb64>] __lock_acquire+0x13c/0xf7c
[   90.818964] [<8005e314>] lock_acquire+0x194/0x1dc
[   90.828345] [<8003f27c>] flush_work+0x200/0x24c
[   90.837374] [<80041dfc>] __cancel_work_timer+0x158/0x210
[   90.847958] [<801a8770>] jffs2_sync_fs+0x20/0x54
[   90.857173] [<80125cf4>] iterate_supers+0xf4/0x120
[   90.866729] [<80158fc4>] sys_sync+0x44/0x9c
[   90.875067] [<80014424>] syscall_common+0x34/0x58

Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Reviewed-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoefi/libstub: Disable some warnings for x86{,_64}
Nathan Chancellor [Thu, 29 Nov 2018 17:12:26 +0000 (18:12 +0100)]
efi/libstub: Disable some warnings for x86{,_64}

[ Upstream commit 3db5e0ba8b8f4aee631d7ee04b7a11c56cfdc213 ]

When building the kernel with Clang, some disabled warnings appear
because this Makefile overrides KBUILD_CFLAGS for x86{,_64}. Add them to
this list so that the build is clean again.

-Wpointer-sign was disabled for the whole kernel before the beginning of Git history.

-Waddress-of-packed-member was disabled for the whole kernel and for
the early boot code in these commits:

  bfb38988c51e ("kbuild: clang: Disable 'address-of-packed-member' warning")
  20c6c1890455 ("x86/boot: Disable the address-of-packed-member compiler warning").

-Wgnu was disabled for the whole kernel and for the early boot code in
these commits:

  61163efae020 ("kbuild: LLVMLinux: Add Kbuild support for building kernel with Clang")
  6c3b56b19730 ("x86/boot: Disable Clang warnings about GNU extensions").

 [ mingo: Made the changelog more readable. ]

Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Sedat Dilek <sedat.dilek@gmail.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arend van Spriel <arend.vanspriel@broadcom.com>
Cc: Bhupesh Sharma <bhsharma@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Eric Snowberg <eric.snowberg@oracle.com>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Joe Perches <joe@perches.com>
Cc: Jon Hunter <jonathanh@nvidia.com>
Cc: Julien Thierry <julien.thierry@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: YiFei Zhu <zhuyifei1999@gmail.com>
Cc: linux-efi@vger.kernel.org
Link: http://lkml.kernel.org/r/20181129171230.18699-8-ard.biesheuvel@linaro.org
Link: https://github.com/ClangBuiltLinux/linux/issues/112
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agorxe: IB_WR_REG_MR does not capture MR's iova field
Chuck Lever [Sun, 25 Nov 2018 22:13:08 +0000 (17:13 -0500)]
rxe: IB_WR_REG_MR does not capture MR's iova field

[ Upstream commit b024dd0eba6e6d568f69d63c5e3153aba94c23e3 ]

FRWR memory registration is done with a series of calls and WRs.
1. ULP invokes ib_dma_map_sg()
2. ULP invokes ib_map_mr_sg()
3. ULP posts an IB_WR_REG_MR on the Send queue

Step 2 generates an iova. It is permissible for ULPs to change this
iova (with certain restrictions) between steps 2 and 3.

rxe_map_mr_sg captures the MR's iova but later when rxe processes the
REG_MR WR, it ignores the MR's iova field. If a ULP alters the MR's iova
after step 2 but before step 3, rxe never captures that change.

When the remote sends an RDMA Read targeting that MR, rxe looks up the
R_key, but the altered iova does not match the iova stored in the MR,
causing the RDMA Read request to fail.

Reported-by: Anna Schumaker <schumaker.anna@gmail.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agodrm/amdgpu: Reorder uvd ring init before uvd resume
Chris Wilson [Fri, 16 Nov 2018 09:34:24 +0000 (09:34 +0000)]
drm/amdgpu: Reorder uvd ring init before uvd resume

[ Upstream commit 3b34c14fd50c302db091f020f26dd00ede902c80 ]

As amd_uvd_resume() accesses the uvd ring, it must be initialised first
or else we trigger errors like:

[    5.595963] [drm] Found UVD firmware Version: 1.87 Family ID: 17
[    5.595969] [drm] PSP loading UVD firmware
[    5.596266] ------------[ cut here ]------------
[    5.596268] ODEBUG: assert_init not available (active state 0) object type: timer_list hint:           (null)
[    5.596285] WARNING: CPU: 0 PID: 507 at lib/debugobjects.c:329 debug_print_object+0x6a/0x80
[    5.596286] Modules linked in: amdgpu(+) hid_logitech_hidpp(+) chash gpu_sched amd_iommu_v2 ttm drm_kms_helper crc32c_intel drm hid_sony ff_memless igb hid_logitech_dj nvme dca i2c_algo_bit nvme_core wmi pinctrl_amd uas usb_storage
[    5.596299] CPU: 0 PID: 507 Comm: systemd-udevd Tainted: G        W         4.20.0-0.rc1.git4.1.fc30.x86_64 #1
[    5.596301] Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 0901 07/23/2018
[    5.596303] RIP: 0010:debug_print_object+0x6a/0x80
[    5.596305] Code: 8b 43 10 83 c2 01 8b 4b 14 4c 89 e6 89 15 e6 82 b0 02 4c 8b 45 00 48 c7 c7 60 fd 34 a6 48 8b 14 c5 a0 da 08 a6 e8 6a 6a b8 ff <0f> 0b 5b 83 05 d0 45 3e 01 01 5d 41 5c c3 83 05 c5 45 3e 01 01 c3
[    5.596306] RSP: 0018:ffffa02ac863f8c0 EFLAGS: 00010282
[    5.596307] RAX: 0000000000000000 RBX: ffffa02ac863f8e0 RCX: 0000000000000006
[    5.596308] RDX: 0000000000000007 RSI: ffff9160e9a7bfe8 RDI: ffff9160f91d6c60
[    5.596310] RBP: ffffffffa6742740 R08: 0000000000000002 R09: 0000000000000000
[    5.596311] R10: 0000000000000000 R11: 0000000000000000 R12: ffffffffa634ff69
[    5.596312] R13: 00000000000b79d0 R14: ffffffffa80f76d8 R15: 0000000000266000
[    5.596313] FS:  00007f762abf7940(0000) GS:ffff9160f9000000(0000) knlGS:0000000000000000
[    5.596314] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[    5.596315] CR2: 000055fdc593f000 CR3: 00000007e999c000 CR4: 00000000003406f0
[    5.596317] Call Trace:
[    5.596321]  debug_object_assert_init+0x14a/0x180
[    5.596327]  del_timer+0x2e/0x90
[    5.596383]  amdgpu_fence_process+0x47/0x100 [amdgpu]
[    5.596430]  amdgpu_uvd_resume+0xf6/0x120 [amdgpu]
[    5.596475]  uvd_v7_0_sw_init+0xe0/0x280 [amdgpu]
[    5.596523]  amdgpu_device_init.cold.30+0xf97/0x14b6 [amdgpu]
[    5.596563]  ? amdgpu_driver_load_kms+0x53/0x330 [amdgpu]
[    5.596604]  amdgpu_driver_load_kms+0x86/0x330 [amdgpu]
[    5.596614]  drm_dev_register+0x115/0x150 [drm]
[    5.596654]  amdgpu_pci_probe+0xbd/0x120 [amdgpu]
[    5.596658]  local_pci_probe+0x41/0x90
[    5.596661]  pci_device_probe+0x188/0x1a0
[    5.596666]  really_probe+0xf8/0x3b0
[    5.596669]  driver_probe_device+0xb3/0xf0
[    5.596672]  __driver_attach+0xe1/0x110
[    5.596674]  ? driver_probe_device+0xf0/0xf0
[    5.596676]  bus_for_each_dev+0x79/0xc0
[    5.596679]  bus_add_driver+0x155/0x230
[    5.596681]  ? 0xffffffffc07d9000
[    5.596683]  driver_register+0x6b/0xb0
[    5.596685]  ? 0xffffffffc07d9000
[    5.596688]  do_one_initcall+0x5d/0x2be
[    5.596691]  ? rcu_read_lock_sched_held+0x79/0x80
[    5.596693]  ? kmem_cache_alloc_trace+0x264/0x290
[    5.596695]  ? do_init_module+0x22/0x210
[    5.596698]  do_init_module+0x5a/0x210
[    5.596701]  load_module+0x2137/0x2430
[    5.596703]  ? lockdep_hardirqs_on+0xed/0x180
[    5.596714]  ? __do_sys_init_module+0x150/0x1a0
[    5.596715]  __do_sys_init_module+0x150/0x1a0
[    5.596722]  do_syscall_64+0x60/0x1f0
[    5.596725]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
[    5.596726] RIP: 0033:0x7f762b877dee
[    5.596728] Code: 48 8b 0d 9d 20 0c 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 49 89 ca b8 af 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 6a 20 0c 00 f7 d8 64 89 01 48
[    5.596729] RSP: 002b:00007ffc777b8558 EFLAGS: 00000246 ORIG_RAX: 00000000000000af
[    5.596730] RAX: ffffffffffffffda RBX: 000055fdc48da320 RCX: 00007f762b877dee
[    5.596731] RDX: 00007f762b9f284d RSI: 00000000006c5fc6 RDI: 000055fdc527a060
[    5.596732] RBP: 00007f762b9f284d R08: 0000000000000003 R09: 0000000000000002
[    5.596733] R10: 000055fdc48ad010 R11: 0000000000000246 R12: 000055fdc527a060
[    5.596734] R13: 000055fdc48dca20 R14: 0000000000020000 R15: 0000000000000000
[    5.596740] irq event stamp: 134618
[    5.596743] hardirqs last  enabled at (134617): [<ffffffffa513d52e>] console_unlock+0x45e/0x610
[    5.596744] hardirqs last disabled at (134618): [<ffffffffa50037e8>] trace_hardirqs_off_thunk+0x1a/0x1c
[    5.596746] softirqs last  enabled at (133146): [<ffffffffa5e00365>] __do_softirq+0x365/0x47c
[    5.596748] softirqs last disabled at (133139): [<ffffffffa50c64f9>] irq_exit+0x119/0x120
[    5.596749] ---[ end trace eaee508abfebccdc ]---

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108709
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoscsi: qedi: Check for session online before getting iSCSI TLV data.
Manish Rangankar [Wed, 21 Nov 2018 09:25:19 +0000 (01:25 -0800)]
scsi: qedi: Check for session online before getting iSCSI TLV data.

[ Upstream commit d5632b11f0a17efa6356311e535ae135d178438d ]

The kernel panic was observed after switch side perturbation,

BUG: unable to handle kernel NULL pointer dereference at (null)
     IP: [<ffffffff8132b5a0>] strcmp+0x20/0x40
     PGD 0 Oops: 0000 [#1] SMP
CPU: 8 PID: 647 Comm: kworker/8:1 Tainted: G        W  OE  ------------   3.10.0-693.el7.x86_64 #1
Hardware name: HPE ProLiant DL380 Gen10/ProLiant DL380 Gen10, BIOS U30 06/20/2018
Workqueue: slowpath-13:00. qed_slowpath_task [qed]
task: ffff880429eb8fd0 ti: ffff880429190000 task.ti: ffff880429190000
RIP: 0010:[<ffffffff8132b5a0>]  [<ffffffff8132b5a0>] strcmp+0x20/0x40
RSP: 0018:ffff880429193c68  EFLAGS: 00010202
RAX: 000000000000000a RBX: 0000000000000002 RCX: 0000000000000000
RDX: 0000000000000001 RSI: 0000000000000001 RDI: ffff88042bda7a41
RBP: ffff880429193c68 R08: 000000000000ffff R09: 000000000000ffff
R10: 0000000000000007 R11: ffff88042b3af338 R12: ffff880420b007a0
R13: ffff88081aa56af8 R14: 0000000000000001 R15: ffff88081aa50410
FS:  0000000000000000(0000) GS:ffff88042fe00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000000 CR3: 00000000019f2000 CR4: 00000000003407e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Stack:
ffff880429193d20 ffffffffc02a0c90 ffffc90004b32000 ffff8803fd3ec600
ffff88042bda7800 ffff88042bda7a00 ffff88042bda7840 ffff88042bda7a40
0000000129193d10 2e3836312e323931 ff000a342e363232 ffffffffc01ad99d
Call Trace:
[<ffffffffc02a0c90>] qedi_get_protocol_tlv_data+0x270/0x470 [qedi]
[<ffffffffc01ad99d>] ? qed_mfw_process_tlv_req+0x24d/0xbf0 [qed]
[<ffffffffc01653ae>] qed_mfw_fill_tlv_data+0x5e/0xd0 [qed]
[<ffffffffc01ad9b9>] qed_mfw_process_tlv_req+0x269/0xbf0 [qed]

Fix kernel NULL pointer deref by checking for session is online before
getting iSCSI TLV data.

Signed-off-by: Manish Rangankar <manish.rangankar@cavium.com>
Reviewed-by: Lee Duncan <lduncan@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoASoC: pcm3168a: Don't disable pcm3168a when CONFIG_PM defined
Jiada Wang [Wed, 28 Nov 2018 12:26:12 +0000 (21:26 +0900)]
ASoC: pcm3168a: Don't disable pcm3168a when CONFIG_PM defined

[ Upstream commit 489db5d941500249583ec6b49fa70e006bd8f632 ]

pcm3168 codec support runtime_[resume|suspend], whenever it
is not active, it enters suspend mode, and it's clock and regulators
will be disabled. so there is no need to disable them again in
remove callback.  Otherwise we got following kernel warnings,
when unload pcm3168a driver

[  222.257514] unbalanced disables for amp-en-regulator
[  222.262526] ------------[ cut here ]------------
[  222.267158] WARNING: CPU: 0 PID: 2423 at drivers/regulator/core.c:2264 _regulator_disable+0x28/0x108
[  222.276291] Modules linked in:
[  222.279343]  snd_soc_pcm3168a_i2c(-)
[  222.282916]  snd_aloop
[  222.285272]  arc4
[  222.287194]  wl18xx
[  222.289289]  wlcore
[  222.291385]  mac80211
[  222.293654]  cfg80211
[  222.295923]  aes_ce_blk
[  222.298366]  crypto_simd
[  222.300896]  cryptd
[  222.302992]  aes_ce_cipher
[  222.305696]  crc32_ce
[  222.307965]  ghash_ce
[  222.310234]  aes_arm64
[  222.312590]  gf128mul
[  222.314860]  snd_soc_rcar
[  222.317476]  sha2_ce
[  222.319658]  xhci_plat_hcd
[  222.322362]  sha256_arm64
[  222.324978]  xhci_hcd
[  222.327247]  sha1_ce
[  222.329430]  renesas_usbhs
[  222.332133]  evdev
[  222.334142]  sha1_generic
[  222.336758]  rcar_gen3_thermal
[  222.339810]  cpufreq_dt
[  222.342253]  ravb_streaming(C)
[  222.345304]  wlcore_sdio
[  222.347834]  thermal_sys
[  222.350363]  udc_core
[  222.352632]  mch_core(C)
[  222.355161]  usb_dmac
[  222.357430]  snd_soc_pcm3168a
[  222.360394]  snd_soc_ak4613
[  222.363184]  gpio_keys
[  222.365540]  virt_dma
[  222.367809]  nfsd
[  222.369730]  ipv6
[  222.371652]  autofs4
[  222.373834]  [last unloaded: snd_soc_pcm3168a_i2c]
[  222.378629] CPU: 0 PID: 2423 Comm: rmmod Tainted: G        WC      4.14.63-04798-gd456126e4a42-dirty #457
[  222.388196] Hardware name: Renesas H3ULCB Kingfisher board based on r8a7795 ES2.0+ (DT)
[  222.396199] task: ffff8006fa8c6200 task.stack: ffff00000a0a0000
[  222.402117] PC is at _regulator_disable+0x28/0x108
[  222.406906] LR is at _regulator_disable+0x28/0x108
[  222.411695] pc : [<ffff0000083bd89c>] lr : [<ffff0000083bd89c>] pstate: 00000145
[  222.419089] sp : ffff00000a0a3c80
[  222.422401] x29: ffff00000a0a3c80
[  222.425799] x28: ffff8006fa8c6200
[  222.429199] x27: ffff0000086f1000
[  222.432597] x26: 000000000000006a
[  222.435997] x25: 0000000000000124
[  222.439395] x24: 0000000000000018
[  222.442795] x23: 0000000000000006
[  222.446193] x22: ffff8006f925d490
[  222.449592] x21: ffff8006f9ac2068
[  222.452991] x20: ffff8006f9ac2000
[  222.456390] x19: 0000000000000005
[  222.459787] x18: 000000000000000a
[  222.463186] x17: 0000000000000000
[  222.466584] x16: 0000000000000000
[  222.469984] x15: 000000000d3f616a
[  222.473382] x14: 0720072007200720
[  222.476781] x13: 0720072007200720
[  222.480179] x12: 0720072007200720
[  222.483578] x11: 0720072007200720
[  222.486975] x10: 0720072007200720
[  222.490375] x9 : 0720072007200720
[  222.493773] x8 : 07200772076f0774
[  222.497172] x7 : 0000000000000000
[  222.500570] x6 : 0000000000000007
[  222.503969] x5 : 0000000000000000
[  222.507367] x4 : 0000000000000000
[  222.510766] x3 : 0000000000000000
[  222.514164] x2 : c790b852091e2600
[  222.517563] x1 : 0000000000000000
[  222.520961] x0 : 0000000000000028
[  222.524361] Call trace:
[  222.526805] Exception stack(0xffff00000a0a3b40 to 0xffff00000a0a3c80)
[  222.533245] 3b40: 0000000000000028 0000000000000000 c790b852091e2600 0000000000000000
[  222.541075] 3b60: 0000000000000000 0000000000000000 0000000000000007 0000000000000000
[  222.548905] 3b80: 07200772076f0774 0720072007200720 0720072007200720 0720072007200720
[  222.556735] 3ba0: 0720072007200720 0720072007200720 0720072007200720 000000000d3f616a
[  222.564564] 3bc0: 0000000000000000 0000000000000000 000000000000000a 0000000000000005
[  222.572394] 3be0: ffff8006f9ac2000 ffff8006f9ac2068 ffff8006f925d490 0000000000000006
[  222.580224] 3c00: 0000000000000018 0000000000000124 000000000000006a ffff0000086f1000
[  222.588053] 3c20: ffff8006fa8c6200 ffff00000a0a3c80 ffff0000083bd89c ffff00000a0a3c80
[  222.595883] 3c40: ffff0000083bd89c 0000000000000145 0000000000000000 0000000000000000
[  222.603713] 3c60: 0000ffffffffffff ffff00000a0a3c30 ffff00000a0a3c80 ffff0000083bd89c
[  222.611543] [<ffff0000083bd89c>] _regulator_disable+0x28/0x108
[  222.617375] [<ffff0000083bd9c4>] regulator_disable+0x48/0x68
[  222.623033] [<ffff0000083be8e4>] regulator_bulk_disable+0x58/0xc0
[  222.629134] [<ffff0000007d831c>] pcm3168a_remove+0x30/0x50 [snd_soc_pcm3168a]
[  222.636270] [<ffff0000007e5010>] pcm3168a_i2c_remove+0x10/0x1c [snd_soc_pcm3168a_i2c]
[  222.644106] [<ffff0000084b9d9c>] i2c_device_remove+0x38/0x70
[  222.649766] [<ffff00000843cd5c>] device_release_driver_internal+0xd0/0x1c0
[  222.656640] [<ffff00000843ced8>] driver_detach+0x70/0x7c
[  222.661951] [<ffff00000843bf68>] bus_remove_driver+0x74/0xa0
[  222.667609] [<ffff00000843d7e4>] driver_unregister+0x48/0x4c
[  222.673268] [<ffff0000084ba8dc>] i2c_del_driver+0x24/0x30
[  222.678666] [<ffff0000007e5078>] pcm3168a_i2c_driver_exit+0x10/0xf98 [snd_soc_pcm3168a_i2c]
[  222.687019] [<ffff00000811bd28>] SyS_delete_module+0x198/0x1d4
[  222.692850] Exception stack(0xffff00000a0a3ec0 to 0xffff00000a0a4000)
[  222.699289] 3ec0: 0000aaaafeb4b268 0000000000000800 14453f6470497100 0000fffffaa520d8
[  222.707119] 3ee0: 0000fffffaa520d9 000000000000000a 1999999999999999 0000000000000000
[  222.714948] 3f00: 000000000000006a 0000ffffa8f7d1d8 000000000000000a 0000000000000005
[  222.722778] 3f20: 0000000000000000 0000000000000000 000000000000002d 0000000000000000
[  222.730607] 3f40: 0000aaaae19b9f68 0000ffffa8f411f0 0000000000000000 0000aaaae19b9000
[  222.738436] 3f60: 0000fffffaa533b8 0000fffffaa531f0 0000000000000000 0000000000000001
[  222.746266] 3f80: 0000fffffaa53ec6 0000000000000000 0000aaaafeb4b200 0000aaaafeb4a010
[  222.754096] 3fa0: 0000000000000000 0000fffffaa53130 0000aaaae199f36c 0000fffffaa53130
[  222.761926] 3fc0: 0000ffffa8f411f8 0000000000000000 0000aaaafeb4b268 000000000000006a
[  222.769755] 3fe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  222.777589] [<ffff0000080832c0>] el0_svc_naked+0x34/0x38
[  222.782899] ---[ end trace eaf8939a3698b1a8 ]---
[  222.787609] Failed to disable VCCDA2: -5
[  222.791649] ------------[ cut here ]------------
[  222.796283] WARNING: CPU: 0 PID: 2423 at drivers/clk/clk.c:595 clk_core_disable+0xc/0x1d8
[  222.804460] Modules linked in:
[  222.807511]  snd_soc_pcm3168a_i2c(-)
[  222.811083]  snd_aloop
[  222.813439]  arc4
[  222.815360]  wl18xx
[  222.817456]  wlcore
[  222.819551]  mac80211
[  222.821820]  cfg80211
[  222.824088]  aes_ce_blk
[  222.826531]  crypto_simd
[  222.829060]  cryptd
[  222.831155]  aes_ce_cipher
[  222.833859]  crc32_ce
[  222.836127]  ghash_ce
[  222.838396]  aes_arm64
[  222.840752]  gf128mul
[  222.843020]  snd_soc_rcar
[  222.845637]  sha2_ce
[  222.847818]  xhci_plat_hcd
[  222.850522]  sha256_arm64
[  222.853138]  xhci_hcd
[  222.855407]  sha1_ce
[  222.857589]  renesas_usbhs
[  222.860292]  evdev
[  222.862300]  sha1_generic
[  222.864917]  rcar_gen3_thermal
[  222.867968]  cpufreq_dt
[  222.870410]  ravb_streaming(C)
[  222.873461]  wlcore_sdio
[  222.875991]  thermal_sys
[  222.878520]  udc_core
[  222.880789]  mch_core(C)
[  222.883318]  usb_dmac
[  222.885587]  snd_soc_pcm3168a
[  222.888551]  snd_soc_ak4613
[  222.891341]  gpio_keys
[  222.893696]  virt_dma
[  222.895965]  nfsd
[  222.897886]  ipv6
[  222.899808]  autofs4
[  222.901990]  [last unloaded: snd_soc_pcm3168a_i2c]
[  222.906783] CPU: 0 PID: 2423 Comm: rmmod Tainted: G        WC      4.14.63-04798-gd456126e4a42-dirty #457
[  222.916349] Hardware name: Renesas H3ULCB Kingfisher board based on r8a7795 ES2.0+ (DT)
[  222.924351] task: ffff8006fa8c6200 task.stack: ffff00000a0a0000
[  222.930270] PC is at clk_core_disable+0xc/0x1d8
[  222.934799] LR is at clk_core_disable_lock+0x20/0x34
[  222.939761] pc : [<ffff0000083ab9b8>] lr : [<ffff0000083acd28>] pstate: 800001c5
[  222.947154] sp : ffff00000a0a3cf0
[  222.950466] x29: ffff00000a0a3cf0
[  222.953864] x28: ffff8006fa8c6200
[  222.957263] x27: ffff0000086f1000
[  222.960661] x26: 000000000000006a
[  222.964061] x25: 0000000000000124
[  222.967458] x24: 0000000000000015
[  222.970858] x23: ffff8006f9ffa8d0
[  222.974256] x22: ffff8006faf16480
[  222.977655] x21: ffff0000007e7040
[  222.981053] x20: ffff8006faadd100
[  222.984452] x19: 0000000000000140
[  222.987850] x18: 000000000000000a
[  222.991249] x17: 0000000000000000
[  222.994647] x16: 0000000000000000
[  222.998046] x15: 000000000d477819
[  223.001444] x14: 0720072007200720
[  223.004843] x13: 0720072007200720
[  223.008242] x12: 0720072007200720
[  223.011641] x11: 0720072007200720
[  223.015039] x10: 0720072007200720
[  223.018438] x9 : 0720072007200720
[  223.021837] x8 : 0720072007200720
[  223.025236] x7 : 0000000000000000
[  223.028634] x6 : 0000000000000007
[  223.032034] x5 : 0000000000000000
[  223.035432] x4 : 0000000000000000
[  223.038831] x3 : 0000000000000000
[  223.042229] x2 : 0000000004720471
[  223.045628] x1 : 0000000000000000
[  223.049026] x0 : ffff8006faadd100
[  223.052426] Call trace:
[  223.054870] Exception stack(0xffff00000a0a3bb0 to 0xffff00000a0a3cf0)
[  223.061309] 3ba0:                                   ffff8006faadd100 0000000000000000
[  223.069139] 3bc0: 0000000004720471 0000000000000000 0000000000000000 0000000000000000
[  223.076969] 3be0: 0000000000000007 0000000000000000 0720072007200720 0720072007200720
[  223.084798] 3c00: 0720072007200720 0720072007200720 0720072007200720 0720072007200720
[  223.092628] 3c20: 0720072007200720 000000000d477819 0000000000000000 0000000000000000
[  223.100458] 3c40: 000000000000000a 0000000000000140 ffff8006faadd100 ffff0000007e7040
[  223.108287] 3c60: ffff8006faf16480 ffff8006f9ffa8d0 0000000000000015 0000000000000124
[  223.116117] 3c80: 000000000000006a ffff0000086f1000 ffff8006fa8c6200 ffff00000a0a3cf0
[  223.123947] 3ca0: ffff0000083acd28 ffff00000a0a3cf0 ffff0000083ab9b8 00000000800001c5
[  223.131777] 3cc0: ffff00000a0a3cf0 ffff0000083acd1c 0000ffffffffffff ffff8006faadd100
[  223.139606] 3ce0: ffff00000a0a3cf0 ffff0000083ab9b8
[  223.144483] [<ffff0000083ab9b8>] clk_core_disable+0xc/0x1d8
[  223.150054] [<ffff0000083acd58>] clk_disable+0x1c/0x28
[  223.155198] [<ffff0000007d8328>] pcm3168a_remove+0x3c/0x50 [snd_soc_pcm3168a]
[  223.162334] [<ffff0000007e5010>] pcm3168a_i2c_remove+0x10/0x1c [snd_soc_pcm3168a_i2c]
[  223.170167] [<ffff0000084b9d9c>] i2c_device_remove+0x38/0x70
[  223.175826] [<ffff00000843cd5c>] device_release_driver_internal+0xd0/0x1c0
[  223.182700] [<ffff00000843ced8>] driver_detach+0x70/0x7c
[  223.188012] [<ffff00000843bf68>] bus_remove_driver+0x74/0xa0
[  223.193669] [<ffff00000843d7e4>] driver_unregister+0x48/0x4c
[  223.199329] [<ffff0000084ba8dc>] i2c_del_driver+0x24/0x30
[  223.204726] [<ffff0000007e5078>] pcm3168a_i2c_driver_exit+0x10/0xf98 [snd_soc_pcm3168a_i2c]
[  223.213079] [<ffff00000811bd28>] SyS_delete_module+0x198/0x1d4
[  223.218909] Exception stack(0xffff00000a0a3ec0 to 0xffff00000a0a4000)
[  223.225349] 3ec0: 0000aaaafeb4b268 0000000000000800 14453f6470497100 0000fffffaa520d8
[  223.233179] 3ee0: 0000fffffaa520d9 000000000000000a 1999999999999999 0000000000000000
[  223.241008] 3f00: 000000000000006a 0000ffffa8f7d1d8 000000000000000a 0000000000000005
[  223.248838] 3f20: 0000000000000000 0000000000000000 000000000000002d 0000000000000000
[  223.256668] 3f40: 0000aaaae19b9f68 0000ffffa8f411f0 0000000000000000 0000aaaae19b9000
[  223.264497] 3f60: 0000fffffaa533b8 0000fffffaa531f0 0000000000000000 0000000000000001
[  223.272327] 3f80: 0000fffffaa53ec6 0000000000000000 0000aaaafeb4b200 0000aaaafeb4a010
[  223.280157] 3fa0: 0000000000000000 0000fffffaa53130 0000aaaae199f36c 0000fffffaa53130
[  223.287986] 3fc0: 0000ffffa8f411f8 0000000000000000 0000aaaafeb4b268 000000000000006a
[  223.295816] 3fe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  223.303648] [<ffff0000080832c0>] el0_svc_naked+0x34/0x38
[  223.308958] ---[ end trace eaf8939a3698b1a9 ]---
[  223.313752] ------------[ cut here ]------------
[  223.318383] WARNING: CPU: 0 PID: 2423 at drivers/clk/clk.c:477 clk_core_unprepare+0xc/0x1ac
[  223.326733] Modules linked in:
[  223.329784]  snd_soc_pcm3168a_i2c(-)
[  223.333356]  snd_aloop
[  223.335712]  arc4
[  223.337633]  wl18xx
[  223.339728]  wlcore
[  223.341823]  mac80211
[  223.344092]  cfg80211
[  223.346360]  aes_ce_blk
[  223.348803]  crypto_simd
[  223.351332]  cryptd
[  223.353428]  aes_ce_cipher
[  223.356131]  crc32_ce
[  223.358400]  ghash_ce
[  223.360668]  aes_arm64
[  223.363024]  gf128mul
[  223.365293]  snd_soc_rcar
[  223.367909]  sha2_ce
[  223.370091]  xhci_plat_hcd
[  223.372794]  sha256_arm64
[  223.375410]  xhci_hcd
[  223.377679]  sha1_ce
[  223.379861]  renesas_usbhs
[  223.382564]  evdev
[  223.384572]  sha1_generic
[  223.387188]  rcar_gen3_thermal
[  223.390239]  cpufreq_dt
[  223.392682]  ravb_streaming(C)
[  223.395732]  wlcore_sdio
[  223.398261]  thermal_sys
[  223.400790]  udc_core
[  223.403059]  mch_core(C)
[  223.405588]  usb_dmac
[  223.407856]  snd_soc_pcm3168a
[  223.410820]  snd_soc_ak4613
[  223.413609]  gpio_keys
[  223.415965]  virt_dma
[  223.418234]  nfsd
[  223.420155]  ipv6
[  223.422076]  autofs4
[  223.424258]  [last unloaded: snd_soc_pcm3168a_i2c]
[  223.429050] CPU: 0 PID: 2423 Comm: rmmod Tainted: G        WC      4.14.63-04798-gd456126e4a42-dirty #457
[  223.438616] Hardware name: Renesas H3ULCB Kingfisher board based on r8a7795 ES2.0+ (DT)
[  223.446618] task: ffff8006fa8c6200 task.stack: ffff00000a0a0000
[  223.452536] PC is at clk_core_unprepare+0xc/0x1ac
[  223.457239] LR is at clk_unprepare+0x28/0x3c
[  223.461506] pc : [<ffff0000083ab5a4>] lr : [<ffff0000083ace4c>] pstate: 60000145
[  223.468900] sp : ffff00000a0a3d00
[  223.472211] x29: ffff00000a0a3d00
[  223.475609] x28: ffff8006fa8c6200
[  223.479009] x27: ffff0000086f1000
[  223.482407] x26: 000000000000006a
[  223.485807] x25: 0000000000000124
[  223.489205] x24: 0000000000000015
[  223.492604] x23: ffff8006f9ffa8d0
[  223.496003] x22: ffff8006faf16480
[  223.499402] x21: ffff0000007e7040
[  223.502800] x20: ffff8006faf16420
[  223.506199] x19: ffff8006faadd100
[  223.509597] x18: 000000000000000a
[  223.512997] x17: 0000000000000000
[  223.516395] x16: 0000000000000000
[  223.519794] x15: 0000000000000000
[  223.523192] x14: 00000033fe89076c
[  223.526591] x13: 0000000000000400
[  223.529989] x12: 0000000000000400
[  223.533388] x11: 0000000000000000
[  223.536786] x10: 00000000000009e0
[  223.540185] x9 : ffff00000a0a3be0
[  223.543583] x8 : ffff8006fa8c6c40
[  223.546982] x7 : ffff8006fa8c6400
[  223.550380] x6 : 0000000000000001
[  223.553780] x5 : 0000000000000000
[  223.557178] x4 : ffff8006fa8c6200
[  223.560577] x3 : 0000000000000000
[  223.563975] x2 : ffff8006fa8c6200
[  223.567374] x1 : 0000000000000000
[  223.570772] x0 : ffff8006faadd100
[  223.574170] Call trace:
[  223.576615] Exception stack(0xffff00000a0a3bc0 to 0xffff00000a0a3d00)
[  223.583054] 3bc0: ffff8006faadd100 0000000000000000 ffff8006fa8c6200 0000000000000000
[  223.590884] 3be0: ffff8006fa8c6200 0000000000000000 0000000000000001 ffff8006fa8c6400
[  223.598714] 3c00: ffff8006fa8c6c40 ffff00000a0a3be0 00000000000009e0 0000000000000000
[  223.606544] 3c20: 0000000000000400 0000000000000400 00000033fe89076c 0000000000000000
[  223.614374] 3c40: 0000000000000000 0000000000000000 000000000000000a ffff8006faadd100
[  223.622204] 3c60: ffff8006faf16420 ffff0000007e7040 ffff8006faf16480 ffff8006f9ffa8d0
[  223.630033] 3c80: 0000000000000015 0000000000000124 000000000000006a ffff0000086f1000
[  223.637863] 3ca0: ffff8006fa8c6200 ffff00000a0a3d00 ffff0000083ace4c ffff00000a0a3d00
[  223.645693] 3cc0: ffff0000083ab5a4 0000000060000145 0000000000000140 ffff8006faadd100
[  223.653523] 3ce0: 0000ffffffffffff ffff0000083ace44 ffff00000a0a3d00 ffff0000083ab5a4
[  223.661353] [<ffff0000083ab5a4>] clk_core_unprepare+0xc/0x1ac
[  223.667103] [<ffff0000007d8330>] pcm3168a_remove+0x44/0x50 [snd_soc_pcm3168a]
[  223.674239] [<ffff0000007e5010>] pcm3168a_i2c_remove+0x10/0x1c [snd_soc_pcm3168a_i2c]
[  223.682070] [<ffff0000084b9d9c>] i2c_device_remove+0x38/0x70
[  223.687731] [<ffff00000843cd5c>] device_release_driver_internal+0xd0/0x1c0
[  223.694604] [<ffff00000843ced8>] driver_detach+0x70/0x7c
[  223.699915] [<ffff00000843bf68>] bus_remove_driver+0x74/0xa0
[  223.705572] [<ffff00000843d7e4>] driver_unregister+0x48/0x4c
[  223.711230] [<ffff0000084ba8dc>] i2c_del_driver+0x24/0x30
[  223.716628] [<ffff0000007e5078>] pcm3168a_i2c_driver_exit+0x10/0xf98 [snd_soc_pcm3168a_i2c]
[  223.724980] [<ffff00000811bd28>] SyS_delete_module+0x198/0x1d4
[  223.730811] Exception stack(0xffff00000a0a3ec0 to 0xffff00000a0a4000)
[  223.737250] 3ec0: 0000aaaafeb4b268 0000000000000800 14453f6470497100 0000fffffaa520d8
[  223.745079] 3ee0: 0000fffffaa520d9 000000000000000a 1999999999999999 0000000000000000
[  223.752909] 3f00: 000000000000006a 0000ffffa8f7d1d8 000000000000000a 0000000000000005
[  223.760739] 3f20: 0000000000000000 0000000000000000 000000000000002d 0000000000000000
[  223.768568] 3f40: 0000aaaae19b9f68 0000ffffa8f411f0 0000000000000000 0000aaaae19b9000
[  223.776398] 3f60: 0000fffffaa533b8 0000fffffaa531f0 0000000000000000 0000000000000001
[  223.784227] 3f80: 0000fffffaa53ec6 0000000000000000 0000aaaafeb4b200 0000aaaafeb4a010
[  223.792057] 3fa0: 0000000000000000 0000fffffaa53130 0000aaaae199f36c 0000fffffaa53130
[  223.799886] 3fc0: 0000ffffa8f411f8 0000000000000000 0000aaaafeb4b268 000000000000006a
[  223.807715] 3fe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  223.815546] [<ffff0000080832c0>] el0_svc_naked+0x34/0x38
[  223.820855] ---[ end trace eaf8939a3698b1aa ]---

Fix this issue by only disable clock and regulators in remove callback
when CONFIG_PM isn't defined

Signed-off-by: Jiada Wang <jiada_wang@mentor.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoselinux: always allow mounting submounts
Ondrej Mosnacek [Fri, 16 Nov 2018 13:12:02 +0000 (14:12 +0100)]
selinux: always allow mounting submounts

[ Upstream commit 2cbdcb882f97a45f7475c67ac6257bbc16277dfe ]

If a superblock has the MS_SUBMOUNT flag set, we should always allow
mounting it. These mounts are done automatically by the kernel either as
part of mounting some parent mount (e.g. debugfs always mounts tracefs
under "tracing" for compatibility) or they are mounted automatically as
needed on subdirectory accesses (e.g. NFS crossmnt mounts). Since such
automounts are either an implicit consequence of the parent mount (which
is already checked) or they can happen during regular accesses (where it
doesn't make sense to check against the current task's context), the
mount permission check should be skipped for them.

Without this patch, attempts to access contents of an automounted
directory can cause unexpected SELinux denials.

In the current kernel tree, the MS_SUBMOUNT flag is set only via
vfs_submount(), which is called only from the following places:
 - AFS, when automounting special "symlinks" referencing other cells
 - CIFS, when automounting "referrals"
 - NFS, when automounting subtrees
 - debugfs, when automounting tracefs

In all cases the submounts are meant to be transparent to the user and
it makes sense that if mounting the master is allowed, then so should be
the automounts. Note that CAP_SYS_ADMIN capability checking is already
skipped for (SB_KERNMOUNT|SB_SUBMOUNT) in:
 - sget_userns() in fs/super.c:
if (!(flags & (SB_KERNMOUNT|SB_SUBMOUNT)) &&
    !(type->fs_flags & FS_USERNS_MOUNT) &&
    !capable(CAP_SYS_ADMIN))
return ERR_PTR(-EPERM);
 - sget() in fs/super.c:
        /* Ensure the requestor has permissions over the target filesystem */
        if (!(flags & (SB_KERNMOUNT|SB_SUBMOUNT)) && !ns_capable(user_ns, CAP_SYS_ADMIN))
                return ERR_PTR(-EPERM);

Verified internally on patched RHEL 7.6 with a reproducer using
NFS+httpd and selinux-tesuite.

Fixes: 93faccbbfa95 ("fs: Better permission checking for submounts")
Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
Signed-off-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agofpga: altera-cvp: fix probing for multiple FPGAs on the bus
Anatolij Gustschin [Mon, 26 Nov 2018 17:35:27 +0000 (11:35 -0600)]
fpga: altera-cvp: fix probing for multiple FPGAs on the bus

[ Upstream commit 30522a951f9d02f261d0697c35cb42205b1fae17 ]

Currently registering CvP managers works only for first probed CvP
device, for all other devices it is refused due to duplicated chkcfg
sysfs entry:

fpga_manager fpga3: Altera CvP FPGA Manager @0000:0c:00.0 registered
sysfs: cannot create duplicate filename '/bus/pci/drivers/altera-cvp/chkcfg'
CPU: 0 PID: 3808 Comm: bash Tainted: G           O      4.19.0-custom+ #5
Call Trace:
  dump_stack+0x46/0x5b
  sysfs_warn_dup+0x53/0x60
  sysfs_add_file_mode_ns+0x16d/0x180
  sysfs_create_file_ns+0x51/0x60
  altera_cvp_probe+0x16f/0x2a0 [altera_cvp]
  local_pci_probe+0x3f/0xa0
  ? pci_match_device+0xb1/0xf0
  pci_device_probe+0x116/0x170
  really_probe+0x21b/0x2c0
  driver_probe_device+0x4b/0xe0
  bind_store+0xcb/0x130
  kernfs_fop_write+0xfd/0x180
  __vfs_write+0x21/0x150
  ? selinux_file_permission+0xdc/0x130
  vfs_write+0xa8/0x1a0
  ? find_vma+0xd/0x60
  ksys_write+0x3d/0x90
  do_syscall_64+0x44/0xf0
  entry_SYSCALL_64_after_hwframe+0x44/0xa9
  ...
 altera-cvp 0000:0c:00.0: Can't create sysfs chkcfg file
 fpga_manager fpga3: fpga_mgr_unregister Altera CvP FPGA Manager @0000:0c:00.0

Move chkcfg creation to module init as suggested by Alan.

Signed-off-by: Anatolij Gustschin <agust@denx.de>
Acked-by: Alan Tull <atull@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agousb: gadget: udc: renesas_usb3: add a safety connection way for forced_b_device
Yoshihiro Shimoda [Fri, 9 Nov 2018 11:44:36 +0000 (20:44 +0900)]
usb: gadget: udc: renesas_usb3: add a safety connection way for forced_b_device

[ Upstream commit ceb94bc52c437463f0903e61060a94a2226fb672 ]

This patch adds a safety connection way for "forced_b_device" with
"workaround_for_vbus" like below:

< Example for R-Car E3 Ebisu >
 # modprobe <any usb gadget driver>
 # echo 1 > /sys/kernel/debug/ee020000.usb/b_device
 (connect a usb cable to host side.)
 # echo 2 > /sys/kernel/debug/ee020000.usb/b_device

Previous code should have connected a usb cable before the "b_device"
is set to 1 on the Ebisu board. However, if xHCI driver on the board
is probed, it causes some troubles:
 - Conflicts USB VBUS/signals between the board and another host.
 - "Cannot enable. Maybe the USB cable is bad?" might happen on
   both the board and another host with a usb hub.
 - Cannot enumerate a usb gadget correctly because an interruption
   of VBUS change happens unexpectedly.

Reported-by: Kazuya Mizuguchi <kazuya.mizuguchi.ks@renesas.com>
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agosamples: bpf: fix: error handling regarding kprobe_events
Daniel T. Lee [Thu, 22 Nov 2018 22:14:32 +0000 (07:14 +0900)]
samples: bpf: fix: error handling regarding kprobe_events

[ Upstream commit 5a863813216ce79e16a8c1503b2543c528b778b6 ]

Currently, kprobe_events failure won't be handled properly.
Due to calling system() indirectly to write to kprobe_events,
it can't be identified whether an error is derived from kprobe or system.

    // buf = "echo '%c:%s %s' >> /s/k/d/t/kprobe_events"
    err = system(buf);
    if (err < 0) {
        printf("failed to create kprobe ..");
        return -1;
    }

For example, running ./tracex7 sample in ext4 partition,
"echo p:open_ctree open_ctree >> /s/k/d/t/kprobe_events"
gets 256 error code system() failure.
=> The error comes from kprobe, but it's not handled correctly.

According to man of system(3), it's return value
just passes the termination status of the child shell
rather than treating the error as -1. (don't care success)

Which means, currently it's not working as desired.
(According to the upper code snippet)

    ex) running ./tracex7 with ext4 env.
    # Current Output
    sh: echo: I/O error
    failed to open event open_ctree

    # Desired Output
    failed to create kprobe 'open_ctree' error 'No such file or directory'

The problem is, error can't be verified whether from child ps
or system. But using write() directly can verify the command
failure, and it will treat all error as -1. So I suggest using
write() directly to 'kprobe_events' rather than calling system().

Signed-off-by: Daniel T. Lee <danieltimlee@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoclk: meson: meson8b: fix incorrect divider mapping in cpu_scale_table
Martin Blumenstingl [Thu, 27 Sep 2018 08:59:20 +0000 (10:59 +0200)]
clk: meson: meson8b: fix incorrect divider mapping in cpu_scale_table

[ Upstream commit ad9b2b8e53af61375322e3c7d624acf3a3ef53b0 ]

The public S805 datasheet only mentions that
HHI_SYS_CPU_CLK_CNTL1[20:29] contains a divider called "cpu_scale_div".
Unfortunately it does not mention how to use the register contents.

The Amlogic 3.10 GPL kernel sources are using the following code to
calculate the CPU clock based on that register (taken from
arch/arm/mach-meson8/clock.c in the 3.10 Amlogic kernel, shortened to
make it easier to read):
N = (aml_read_reg32(P_HHI_SYS_CPU_CLK_CNTL1) >> 20) & 0x3FF;
if (sel == 3) /* use cpu_scale_div */
  div = 2 * N;
else
  div = ... /* not relevant for this example */
cpu_clk = parent_clk / div;

This suggests that the formula is: parent_rate / 2 * register_value
However, running perf (which can measure the CPU clock rate thanks to
the ARM PMU) shows that this formula is not correct.
This can be reproduced with the following steps:
1. boot into u-boot
2. let the CPU clock run off the XTAL clock:
   mw.l 0xC110419C 0x30 1
3. set the cpu_scale_div register:
   to value 0x1: mw.l 0xC110415C 0x801016A2 1
   to value 0x2: mw.l 0xC110415C 0x802016A2 1
   to value 0x5: mw.l 0xC110415C 0x805016A2 1
4. let the CPU clock run off cpu_scale_div:
   mw.l 0xC110419C 0xbd 1
5. boot Linux
6. run: perf stat -aB stress --cpu 4 --timeout 10
7. check the "cycles" value

I get the following results depending on the cpu_scale_div value:
- (cpu_in_sel - this is the input clock for cpu_scale_div - runs at
   1.2GHz)
- 0x1 = 300MHz
- 0x2 = 200MHz
- 0x5 = 100MHz

This means that the actual formula to calculate the output of the
cpu_scale_div clock is: parent_rate / 2 * (register value + 1).

The register value 0x0 is reserved. When letting the CPU clock run off
the cpu_scale_div while the value is 0x0 the whole board hangs (even in
u-boot).

I also verified this with the TWD timer: when adding this to the .dts
without specifying it's clock it will auto-detect the PERIPH (which is
the input clock of the TWD) clock rate (and the result is shown in the
kernel log). On Meson8, Meson8b and Meson8m2 the PERIPH clock is CPUCLK
divided by 4. This also matched for all three test-cases from above (in
all cases the TWD timer clock rate was approx. one fourth of the CPU
clock rate).

A small note regarding the "fixes" tag: the original issue seems to
exist virtually since forever. Even commit 28b9fcd016126e ("clk:
meson8b: Add support for Meson8b clocks") seems to handle this wrong. I
still decided to use commit 251b6fd38bcb9c ("clk: meson: rework meson8b
cpu clock") because this is the first commit which gets the CPU hiearchy
correct and thus it's the first commit where the cpu_scale_div register
is used correctly (apart from the bug in the cpu_scale_table).

Fixes: 251b6fd38bcb9c ("clk: meson: rework meson8b cpu clock")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://lkml.kernel.org/r/20180927085921.24627-2-martin.blumenstingl@googlemail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agodrm/atomic-helper: Complete fake_commit->flip_done potentially earlier
Ville Syrjälä [Thu, 22 Nov 2018 14:34:11 +0000 (16:34 +0200)]
drm/atomic-helper: Complete fake_commit->flip_done potentially earlier

[ Upstream commit 2de42f79bb21a412f40ade8831eb6fc445cb78a4 ]

Consider the following scenario:
1. nonblocking enable crtc
2. wait for the event
3. nonblocking disable crtc

On i915 this can lead to a spurious -EBUSY from step 3 on
account of non-enabled planes getting the fake_commit in step 1
and we don't complete the fake_commit-> flip_done until
drm_atomic_helper_commit_hw_done() which can happen a long
time after the flip event was sent out.

This will become somewhat easy to hit on SKL+ once we start
to add all the planes for the crtc to every modeset commit
for the purposes of forcing a watermark register programming
[1].

To make the race a little less pronounced let's complete
fake_commit->flip_done after drm_atomic_helper_wait_for_flip_done().
For the single crtc case this should make the race quite
theoretical, assuming drm_atomic_helper_wait_for_flip_done()
actually has to wait for the real commit flip_done. In case
the real commit flip_done gets completed singificantly before
drm_atomic_helper_wait_for_flip_done(), or we are dealing with
multiple crtcs whose vblanks don't line up nicely the race still
exists.

[1] https://patchwork.freedesktop.org/patch/262670/

Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Fixes: 080de2e5be2d ("drm/atomic: Check for busy planes/connectors before setting the commit")
Testcase: igt/kms_cursor_legacy/*nonblocking-modeset-vs-cursor-atomic
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20181122143412.11655-1-ville.syrjala@linux.intel.com
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoarm64: perf: set suppress_bind_attrs flag to true
Anders Roxell [Wed, 17 Oct 2018 15:26:22 +0000 (17:26 +0200)]
arm64: perf: set suppress_bind_attrs flag to true

[ Upstream commit 81e9fa8bab381f8b6eb04df7cdf0f71994099bd4 ]

The armv8_pmuv3 driver doesn't have a remove function, and when the test
'CONFIG_DEBUG_TEST_DRIVER_REMOVE=y' is enabled, the following Call trace
can be seen.

[    1.424287] Failed to register pmu: armv8_pmuv3, reason -17
[    1.424870] WARNING: CPU: 0 PID: 1 at ../kernel/events/core.c:11771 perf_event_sysfs_init+0x98/0xdc
[    1.425220] Modules linked in:
[    1.425531] CPU: 0 PID: 1 Comm: swapper/0 Tainted: G        W         4.19.0-rc7-next-20181012-00003-ge7a97b1ad77b-dirty #35
[    1.425951] Hardware name: linux,dummy-virt (DT)
[    1.426212] pstate: 80000005 (Nzcv daif -PAN -UAO)
[    1.426458] pc : perf_event_sysfs_init+0x98/0xdc
[    1.426720] lr : perf_event_sysfs_init+0x98/0xdc
[    1.426908] sp : ffff00000804bd50
[    1.427077] x29: ffff00000804bd50 x28: ffff00000934e078
[    1.427429] x27: ffff000009546000 x26: 0000000000000007
[    1.427757] x25: ffff000009280710 x24: 00000000ffffffef
[    1.428086] x23: ffff000009408000 x22: 0000000000000000
[    1.428415] x21: ffff000009136008 x20: ffff000009408730
[    1.428744] x19: ffff80007b20b400 x18: 000000000000000a
[    1.429075] x17: 0000000000000000 x16: 0000000000000000
[    1.429418] x15: 0000000000000400 x14: 2e79726f74636572
[    1.429748] x13: 696420656d617320 x12: 656874206e692065
[    1.430060] x11: 6d616e20656d6173 x10: 2065687420687469
[    1.430335] x9 : ffff00000804bd50 x8 : 206e6f7361657220
[    1.430610] x7 : 2c3376756d705f38 x6 : ffff00000954d7ce
[    1.430880] x5 : 0000000000000000 x4 : 0000000000000000
[    1.431226] x3 : 0000000000000000 x2 : ffffffffffffffff
[    1.431554] x1 : 4d151327adc50b00 x0 : 0000000000000000
[    1.431868] Call trace:
[    1.432102]  perf_event_sysfs_init+0x98/0xdc
[    1.432382]  do_one_initcall+0x6c/0x1a8
[    1.432637]  kernel_init_freeable+0x1bc/0x280
[    1.432905]  kernel_init+0x18/0x160
[    1.433115]  ret_from_fork+0x10/0x18
[    1.433297] ---[ end trace 27fd415390eb9883 ]---

Rework to set suppress_bind_attrs flag to avoid removing the device when
CONFIG_DEBUG_TEST_DRIVER_REMOVE=y, since there's no real reason to
remove the armv8_pmuv3 driver.

Cc: Arnd Bergmann <arnd@arndb.de>
Co-developed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agocrypto: ecc - regularize scalar for scalar multiplication
Vitaly Chikunov [Sun, 11 Nov 2018 17:40:02 +0000 (20:40 +0300)]
crypto: ecc - regularize scalar for scalar multiplication

[ Upstream commit 3da2c1dfdb802b184eea0653d1e589515b52d74b ]

ecc_point_mult is supposed to be used with a regularized scalar,
otherwise, it's possible to deduce the position of the top bit of the
scalar with timing attack. This is important when the scalar is a
private key.

ecc_point_mult is already using a regular algorithm (i.e. having an
operation flow independent of the input scalar) but regularization step
is not implemented.

Arrange scalar to always have fixed top bit by adding a multiple of the
curve order (n).

References:
The constant time regularization step is based on micro-ecc by Kenneth
MacKay and also referenced in the literature (Bernstein, D. J., & Lange,
T. (2017). Montgomery curves and the Montgomery ladder. (Cryptology
ePrint Archive; Vol. 2017/293). s.l.: IACR. Chapter 4.6.2.)

Signed-off-by: Vitaly Chikunov <vt@altlinux.org>
Cc: kernel-hardening@lists.openwall.com
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoMIPS: SiByte: Enable swiotlb for SWARM, LittleSur and BigSur
Maciej W. Rozycki [Tue, 13 Nov 2018 22:42:44 +0000 (22:42 +0000)]
MIPS: SiByte: Enable swiotlb for SWARM, LittleSur and BigSur

[ Upstream commit e4849aff1e169b86c561738daf8ff020e9de1011 ]

The Broadcom SiByte BCM1250, BCM1125, and BCM1125H SOCs have an onchip
DRAM controller that supports memory amounts of up to 16GiB, and due to
how the address decoder has been wired in the SOC any memory beyond 1GiB
is actually mapped starting from 4GiB physical up, that is beyond the
32-bit addressable limit[1].  Consequently if the maximum amount of
memory has been installed, then it will span up to 19GiB.

Many of the evaluation boards we support that are based on one of these
SOCs have their memory soldered and the amount present fits in the
32-bit address range.  The BCM91250A SWARM board however has actual DIMM
slots and accepts, depending on the peripherals revision of the SOC, up
to 4GiB or 8GiB of memory in commercially available JEDEC modules[2].
I believe this is also the case with the BCM91250C2 LittleSur board.
This means that up to either 3GiB or 7GiB of memory requires 64-bit
addressing to access.

I believe the BCM91480B BigSur board, which has the BCM1480 SOC instead,
accepts at least as much memory, although I have no documentation or
actual hardware available to verify that.

Both systems have PCI slots installed for use by any PCI option boards,
including ones that only support 32-bit addressing (additionally the
32-bit PCI host bridge of the BCM1250, BCM1125, and BCM1125H SOCs limits
addressing to 32-bits), and there is no IOMMU available.  Therefore for
PCI DMA to work in the presence of memory beyond enable swiotlb for the
affected systems.

All the other SOC onchip DMA devices use 40-bit addressing and therefore
can address the whole memory, so only enable swiotlb if PCI support and
support for DMA beyond 4GiB have been both enabled in the configuration
of the kernel.

This shows up as follows:

Broadcom SiByte BCM1250 B2 @ 800 MHz (SB1 rev 2)
Board type: SiByte BCM91250A (SWARM)
Determined physical RAM map:
 memory: 000000000fe7fe00 @ 0000000000000000 (usable)
 memory: 000000001ffffe00 @ 0000000080000000 (usable)
 memory: 000000000ffffe00 @ 00000000c0000000 (usable)
 memory: 0000000087fffe00 @ 0000000100000000 (usable)
software IO TLB: mapped [mem 0xcbffc000-0xcfffc000] (64MB)

in the bootstrap log and removes failures like these:

defxx 0000:02:00.0: dma_direct_map_page: overflow 0x0000000185bc6080+4608 of device mask ffffffff bus mask 0
fddi0: Receive buffer allocation failed
fddi0: Adapter open failed!
IP-Config: Failed to open fddi0
defxx 0000:09:08.0: dma_direct_map_page: overflow 0x0000000185bc6080+4608 of device mask ffffffff bus mask 0
fddi1: Receive buffer allocation failed
fddi1: Adapter open failed!
IP-Config: Failed to open fddi1

when memory beyond 4GiB is handed out to devices that can only do 32-bit
addressing.

This updates commit cce335ae47e2 ("[MIPS] 64-bit Sibyte kernels need
DMA32.").

References:

[1] "BCM1250/BCM1125/BCM1125H User Manual", Revision 1250_1125-UM100-R,
    Broadcom Corporation, 21 Oct 2002, Section 3: "System Overview",
    "Memory Map", pp. 34-38

[2] "BCM91250A User Manual", Revision 91250A-UM100-R, Broadcom
    Corporation, 18 May 2004, Section 3: "Physical Description",
    "Supported DRAM", p. 23

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
[paul.burton@mips.com: Remove GPL text from dma.c; SPDX tag covers it]
Signed-off-by: Paul Burton <paul.burton@mips.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Patchwork: https://patchwork.linux-mips.org/patch/21108/
References: cce335ae47e2 ("[MIPS] 64-bit Sibyte kernels need DMA32.")
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agox86/mce: Fix -Wmissing-prototypes warnings
Borislav Petkov [Fri, 9 Nov 2018 22:13:13 +0000 (23:13 +0100)]
x86/mce: Fix -Wmissing-prototypes warnings

[ Upstream commit 68b5e4326e4b8ac9080835005d8254fed0fb3c56 ]

Add the proper includes and make smca_get_name() static.

Fix an actual bug too which the warning triggered:

  arch/x86/kernel/cpu/mcheck/therm_throt.c:395:39: error: conflicting \
  types for ‘smp_thermal_interrupt’
   asmlinkage __visible void __irq_entry smp_thermal_interrupt(struct pt_regs *r)
                                         ^~~~~~~~~~~~~~~~~~~~~
  In file included from arch/x86/kernel/cpu/mcheck/therm_throt.c:29:
  ./arch/x86/include/asm/traps.h:107:17: note: previous declaration of \
  ‘smp_thermal_interrupt’ was here
   asmlinkage void smp_thermal_interrupt(void);

Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Yi Wang <wang.yi59@zte.com.cn>
Cc: Michael Matz <matz@suse.de>
Cc: x86@kernel.org
Link: https://lkml.kernel.org/r/alpine.DEB.2.21.1811081633160.1549@nanos.tec.linutronix.de
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoALSA: oxfw: add support for APOGEE duet FireWire
Takashi Sakamoto [Tue, 13 Nov 2018 03:01:30 +0000 (12:01 +0900)]
ALSA: oxfw: add support for APOGEE duet FireWire

[ Upstream commit fba43f454cdf9caa3185219d116bd2a6e6354552 ]

This commit adds support for APOGEE duet FireWire, launched 2007, already
discontinued. This model uses Oxford Semiconductor FW971 as its
communication engine. Below is information on Configuration ROM of this
unit. The unit supports some AV/C commands defined by Audio subunit
specification and vendor dependent commands.

$ ./hinawa-config-rom-printer /dev/fw1
{ 'bus-info': { 'adj': False,
                'bmc': False,
                'chip_ID': 42949742248,
                'cmc': False,
                'cyc_clk_acc': 255,
                'generation': 0,
                'imc': False,
                'isc': True,
                'link_spd': 3,
                'max_ROM': 0,
                'max_rec': 64,
                'name': '1394',
                'node_vendor_ID': 987,
                'pmc': False},
  'root-directory': [ ['VENDOR', 987],
                      ['DESCRIPTOR', 'Apogee Electronics'],
                      ['MODEL', 122333],
                      ['DESCRIPTOR', 'Duet'],
                      [ 'NODE_CAPABILITIES',
                        { 'addressing': {'64': True, 'fix': True, 'prv': False},
                          'misc': {'int': False, 'ms': False, 'spt': True},
                          'state': { 'atn': False,
                                     'ded': False,
                                     'drq': True,
                                     'elo': False,
                                     'init': False,
                                     'lst': True,
                                     'off': False},
                          'testing': {'bas': False, 'ext': False}}],
                      [ 'UNIT',
                        [ ['SPECIFIER_ID', 41005],
                          ['VERSION', 65537],
                          ['MODEL', 122333],
                          ['DESCRIPTOR', 'Duet']]]]}

Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agobpf: Allow narrow loads with offset > 0
Andrey Ignatov [Sun, 11 Nov 2018 06:15:13 +0000 (22:15 -0800)]
bpf: Allow narrow loads with offset > 0

[ Upstream commit 46f53a65d2de3e1591636c22b626b09d8684fd71 ]

Currently BPF verifier allows narrow loads for a context field only with
offset zero. E.g. if there is a __u32 field then only the following
loads are permitted:
  * off=0, size=1 (narrow);
  * off=0, size=2 (narrow);
  * off=0, size=4 (full).

On the other hand LLVM can generate a load with offset different than
zero that make sense from program logic point of view, but verifier
doesn't accept it.

E.g. tools/testing/selftests/bpf/sendmsg4_prog.c has code:

  #define DST_IP4 0xC0A801FEU /* 192.168.1.254 */
  ...
   if ((ctx->user_ip4 >> 24) == (bpf_htonl(DST_IP4) >> 24) &&

where ctx is struct bpf_sock_addr.

Some versions of LLVM can produce the following byte code for it:

       8:       71 12 07 00 00 00 00 00         r2 = *(u8 *)(r1 + 7)
       9:       67 02 00 00 18 00 00 00         r2 <<= 24
      10:       18 03 00 00 00 00 00 fe 00 00 00 00 00 00 00 00         r3 = 4261412864 ll
      12:       5d 32 07 00 00 00 00 00         if r2 != r3 goto +7 <LBB0_6>

where `*(u8 *)(r1 + 7)` means narrow load for ctx->user_ip4 with size=1
and offset=3 (7 - sizeof(ctx->user_family) = 3). This load is currently
rejected by verifier.

Verifier code that rejects such loads is in bpf_ctx_narrow_access_ok()
what means any is_valid_access implementation, that uses the function,
works this way, e.g. bpf_skb_is_valid_access() for __sk_buff or
sock_addr_is_valid_access() for bpf_sock_addr.

The patch makes such loads supported. Offset can be in [0; size_default)
but has to be multiple of load size. E.g. for __u32 field the following
loads are supported now:
  * off=0, size=1 (narrow);
  * off=1, size=1 (narrow);
  * off=2, size=1 (narrow);
  * off=3, size=1 (narrow);
  * off=0, size=2 (narrow);
  * off=2, size=2 (narrow);
  * off=0, size=4 (full).

Reported-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
5 years agoserial: set suppress_bind_attrs flag only if builtin
Anders Roxell [Tue, 30 Oct 2018 11:35:44 +0000 (12:35 +0100)]
serial: set suppress_bind_attrs flag only if builtin

[ Upstream commit 646097940ad35aa2c1f2012af932d55976a9f255 ]

When the test 'CONFIG_DEBUG_TEST_DRIVER_REMOVE=y' is enabled,
arch_initcall(pl011_init) came before subsys_initcall(default_bdi_init).
devtmpfs gets killed because we try to remove a file and decrement the
wb reference count before the noop_backing_device_info gets initialized.

[    0.332075] Serial: AMBA PL011 UART driver
[    0.485276] 9000000.pl011: ttyAMA0 at MMIO 0x9000000 (irq = 39, base_baud = 0) is a PL011 rev1
[    0.502382] console [ttyAMA0] enabled
[    0.515710] Unable to handle kernel paging request at virtual address 0000800074c12000
[    0.516053] Mem abort info:
[    0.516222]   ESR = 0x96000004
[    0.516417]   Exception class = DABT (current EL), IL = 32 bits
[    0.516641]   SET = 0, FnV = 0
[    0.516826]   EA = 0, S1PTW = 0
[    0.516984] Data abort info:
[    0.517149]   ISV = 0, ISS = 0x00000004
[    0.517339]   CM = 0, WnR = 0
[    0.517553] [0000800074c12000] user address but active_mm is swapper
[    0.517928] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[    0.518305] Modules linked in:
[    0.518839] CPU: 0 PID: 13 Comm: kdevtmpfs Not tainted 4.19.0-rc5-next-20180928-00002-g2ba39ab0cd01-dirty #82
[    0.519307] Hardware name: linux,dummy-virt (DT)
[    0.519681] pstate: 80000005 (Nzcv daif -PAN -UAO)
[    0.519959] pc : __destroy_inode+0x94/0x2a8
[    0.520212] lr : __destroy_inode+0x78/0x2a8
[    0.520401] sp : ffff0000098c3b20
[    0.520590] x29: ffff0000098c3b20 x28: 00000000087a3714
[    0.520904] x27: 0000000000002000 x26: 0000000000002000
[    0.521179] x25: ffff000009583000 x24: 0000000000000000
[    0.521467] x23: ffff80007bb52000 x22: ffff80007bbaa7c0
[    0.521737] x21: ffff0000093f9338 x20: 0000000000000000
[    0.522033] x19: ffff80007bbb05d8 x18: 0000000000000400
[    0.522376] x17: 0000000000000000 x16: 0000000000000000
[    0.522727] x15: 0000000000000400 x14: 0000000000000400
[    0.523068] x13: 0000000000000001 x12: 0000000000000001
[    0.523421] x11: 0000000000000000 x10: 0000000000000970
[    0.523749] x9 : ffff0000098c3a60 x8 : ffff80007bbab190
[    0.524017] x7 : ffff80007bbaa880 x6 : 0000000000000c88
[    0.524305] x5 : ffff0000093d96c8 x4 : 61c8864680b583eb
[    0.524567] x3 : ffff0000093d6180 x2 : ffffffffffffffff
[    0.524872] x1 : 0000800074c12000 x0 : 0000800074c12000
[    0.525207] Process kdevtmpfs (pid: 13, stack limit = 0x(____ptrval____))
[    0.525529] Call trace:
[    0.525806]  __destroy_inode+0x94/0x2a8
[    0.526108]  destroy_inode+0x34/0x88
[    0.526370]  evict+0x144/0x1c8
[    0.526636]  iput+0x184/0x230
[    0.526871]  dentry_unlink_inode+0x118/0x130
[    0.527152]  d_delete+0xd8/0xe0
[    0.527420]  vfs_unlink+0x240/0x270
[    0.527665]  handle_remove+0x1d8/0x330
[    0.527875]  devtmpfsd+0x138/0x1c8
[    0.528085]  kthread+0x14c/0x158
[    0.528291]  ret_from_fork+0x10/0x18
[    0.528720] Code: 92800002 aa1403e0 d538d081 8b010000 (c85f7c04)
[    0.529367] ---[ end trace 5a3dee47727f877c ]---

Rework to set suppress_bind_attrs flag to avoid removing the device when
CONFIG_DEBUG_TEST_DRIVER_REMOVE=y. This applies for pic32_uart and
xilinx_uartps as well.

Co-developed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>