flow_dissector: fix TTL and TOS dissection on IPv4 fragments

[ Upstream commit d2126838050ccd1dadf310ffb78b2204f3b032b9 ]

the following command:

 # tc filter add dev $h2 ingress protocol ip pref 1 handle 101 flower \
   $tcflags dst_ip 192.0.2.2 ip_ttl 63 action drop

doesn't drop all IPv4 packets that match the configured TTL / destination
address. In particular, if "fragment offset" or "more fragments" have non
zero value in the IPv4 header, setting of FLOW_DISSECTOR_KEY_IP is simply
ignored. Fix this dissecting IPv4 TTL and TOS before fragment info; while
at it, add a selftest for tc flower's match on 'ip_ttl' that verifies the
correct behavior.

Fixes: 518d8a2e9bad ("net/flow_dissector: add support for dissection of misc ip header fields")
Reported-by: Shuang Li <shuali@redhat.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: mvpp2: fix interrupt mask/unmask skip condition

[ Upstream commit 7867299cde34e9c2d2c676f2a384a9d5853b914d ]

The condition should be skipped if CPU ID equal to nthreads.
The patch doesn't fix any actual issue since
nthreads = min_t(unsigned int, num_present_cpus(), MVPP2_MAX_THREADS).
On all current Armada platforms, the number of CPU's is
less than MVPP2_MAX_THREADS.

Fixes: e531f76757eb ("net: mvpp2: handle cases where more CPUs are available than s/w threads")
Reported-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Stefan Chulski <stefanc@marvell.com>
Reviewed-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

io_uring: call req_set_fail_links() on short send[msg]()/recv[msg]() with MSG_WAITALL

[ Upstream commit 0031275d119efe16711cd93519b595e6f9b4b330 ]

Without that it's not safe to use them in a linked combination with
others.

Now combinations like IORING_OP_SENDMSG followed by IORING_OP_SPLICE
should be possible.

We already handle short reads and writes for the following opcodes:

- IORING_OP_READV
- IORING_OP_READ_FIXED
- IORING_OP_READ
- IORING_OP_WRITEV
- IORING_OP_WRITE_FIXED
- IORING_OP_WRITE
- IORING_OP_SPLICE
- IORING_OP_TEE

Now we have it for these as well:

- IORING_OP_SENDMSG
- IORING_OP_SEND
- IORING_OP_RECVMSG
- IORING_OP_RECV

For IORING_OP_RECVMSG we also check for the MSG_TRUNC and MSG_CTRUNC
flags in order to call req_set_fail_links().

There might be applications arround depending on the behavior
that even short send[msg]()/recv[msg]() retuns continue an
IOSQE_IO_LINK chain.

It's very unlikely that such applications pass in MSG_WAITALL,
which is only defined in 'man 2 recvmsg', but not in 'man 2 sendmsg'.

It's expected that the low level sock_sendmsg() call just ignores
MSG_WAITALL, as MSG_ZEROCOPY is also ignored without explicitly set
SO_ZEROCOPY.

We also expect the caller to know about the implicit truncation to
MAX_RW_COUNT, which we don't detect.

cc: netdev@vger.kernel.org
Link: https://lore.kernel.org/r/c4e1a4cc0d905314f4d5dc567e65a7b09621aab3.1615908477.git.metze@samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: do not iput inode under running transaction in ext4_rename()

[ Upstream commit 5dccdc5a1916d4266edd251f20bbbb113a5c495f ]

In ext4_rename(), when RENAME_WHITEOUT failed to add new entry into
directory, it ends up dropping new created whiteout inode under the
running transaction. After commit <9b88f9fb0d2> ("ext4: Do not iput inode
under running transaction"), we follow the assumptions that evict() does
not get called from a transaction context but in ext4_rename() it breaks
this suggestion. Although it's not a real problem, better to obey it, so
this patch add inode to orphan list and stop transaction before final
iput().

Signed-off-by: zhangyi (F) <yi.zhang@huawei.com>
Link: https://lore.kernel.org/r/20210303131703.330415-2-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

static_call: Align static_call_is_init() patching condition

[ Upstream commit 698bacefe993ad2922c9d3b1380591ad489355e9 ]

The intent is to avoid writing init code after init (because the text
might have been freed). The code is needlessly different between
jump_label and static_call and not obviously correct.

The existing code relies on the fact that the module loader clears the
init layout, such that within_module_init() always fails, while
jump_label relies on the module state which is more obvious and
matches the kernel logic.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Sumit Garg <sumit.garg@linaro.org>
Link: https://lkml.kernel.org/r/20210318113610.636651340@infradead.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

io_uring: imply MSG_NOSIGNAL for send[msg]()/recv[msg]() calls

[ Upstream commit 76cd979f4f38a27df22efb5773a0d567181a9392 ]

We never want to generate any SIGPIPE, -EPIPE only is much better.

Signed-off-by: Stefan Metzmacher <metze@samba.org>
Link: https://lore.kernel.org/r/38961085c3ec49fd21550c7788f214d1ff02d2d4.1615908477.git.metze@samba.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nvmet-tcp: fix kmap leak when data digest in use

[ Upstream commit bac04454ef9fada009f0572576837548b190bf94 ]

When data digest is enabled we should unmap pdu iovec before handling
the data digest pdu.

Signed-off-by: Elad Grupi <elad.grupi@dell.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

locking/ww_mutex: Fix acquire/release imbalance in ww_acquire_init()/ww_acquire_fini()

[ Upstream commit bee645788e07eea63055d261d2884ea45c2ba857 ]

In ww_acquire_init(), mutex_acquire() is gated by CONFIG_DEBUG_LOCK_ALLOC.
The dep_map in the ww_acquire_ctx structure is also gated by the
same config. However mutex_release() in ww_acquire_fini() is gated by
CONFIG_DEBUG_MUTEXES. It is possible to set CONFIG_DEBUG_MUTEXES without
setting CONFIG_DEBUG_LOCK_ALLOC though it is an unlikely configuration.
That may cause a compilation error as dep_map isn't defined in this
case. Fix this potential problem by enclosing mutex_release() inside
CONFIG_DEBUG_LOCK_ALLOC.

Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20210316153119.13802-3-longman@redhat.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

locking/ww_mutex: Simplify use_ww_ctx & ww_ctx handling

[ Upstream commit 5de2055d31ea88fd9ae9709ac95c372a505a60fa ]

The use_ww_ctx flag is passed to mutex_optimistic_spin(), but the
function doesn't use it. The frequent use of the (use_ww_ctx && ww_ctx)
combination is repetitive.

In fact, ww_ctx should not be used at all if !use_ww_ctx.  Simplify
ww_mutex code by dropping use_ww_ctx from mutex_optimistic_spin() an
clear ww_ctx if !use_ww_ctx. In this way, we can replace (use_ww_ctx &&
ww_ctx) by just (ww_ctx).

Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Davidlohr Bueso <dbueso@suse.de>
Link: https://lore.kernel.org/r/20210316153119.13802-2-longman@redhat.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

thermal/core: Add NULL pointer check before using cooling device stats

[ Upstream commit 2046a24ae121cd107929655a6aaf3b8c5beea01f ]

There is a possible chance that some cooling device stats buffer
allocation fails due to very high cooling device max state value.
Later cooling device update sysfs can try to access stats data
for the same cooling device. It will lead to NULL pointer
dereference issue.

Add a NULL pointer check before accessing thermal cooling device
stats data. It fixes the following bug

[ 26.812833] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000004
[ 27.122960] Call trace:
[ 27.122963] do_raw_spin_lock+0x18/0xe8
[ 27.122966] _raw_spin_lock+0x24/0x30
[ 27.128157] thermal_cooling_device_stats_update+0x24/0x98
[ 27.128162] cur_state_store+0x88/0xb8
[ 27.128166] dev_attr_store+0x40/0x58
[ 27.128169] sysfs_kf_write+0x50/0x68
[ 27.133358] kernfs_fop_write+0x12c/0x1c8
[ 27.133362] __vfs_write+0x54/0x160
[ 27.152297] vfs_write+0xcc/0x188
[ 27.157132] ksys_write+0x78/0x108
[ 27.162050] ksys_write+0xf8/0x108
[ 27.166968] __arm_smccc_hvc+0x158/0x4b0
[ 27.166973] __arm_smccc_hvc+0x9c/0x4b0
[ 27.186005] el0_svc+0x8/0xc

Signed-off-by: Manaf Meethalavalappu Pallikunhi <manafm@codeaurora.org>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://lore.kernel.org/r/1607367181-24589-1-git-send-email-manafm@codeaurora.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: rt711: add snd_soc_component remove callback

[ Upstream commit 899b12542b0897f92de9ba30944937c39ebb246d ]

We do some IO operations in the snd_soc_component_set_jack callback
function and snd_soc_component_set_jack() will be called when soc
component is removed. However, we should not access SoundWire registers
when the bus is suspended.
So set regcache_cache_only(regmap, true) to avoid accessing in the
soc component removal process.

Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Link: https://lore.kernel.org/r/20210316005254.29699-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: rt5659: Update MCLK rate in set_sysclk()

[ Upstream commit dbf54a9534350d6aebbb34f5c1c606b81a4f35dd ]

Simple-card/audio-graph-card drivers do not handle MCLK clock when it
is specified in the codec device node. The expectation here is that,
the codec should actually own up the MCLK clock and do necessary setup
in the driver.

Suggested-by: Mark Brown <broonie@kernel.org>
Suggested-by: Michael Walle <michael@walle.cc>
Signed-off-by: Sameer Pujar <spujar@nvidia.com>
Link: https://lore.kernel.org/r/1615829492-8972-3-git-send-email-spujar@nvidia.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

staging: comedi: cb_pcidas64: fix request_irq() warn

[ Upstream commit d2d106fe3badfc3bf0dd3899d1c3f210c7203eab ]

request_irq() wont accept a name which contains slash so we need to
repalce it with something else -- otherwise it will trigger a warning
and the entry in /proc/irq/ will not be created
since the .name might be used by userspace and we don't want to break
userspace, so we are changing the parameters passed to request_irq()

[    1.565966] name 'pci-das6402/16'
[    1.566149] WARNING: CPU: 0 PID: 184 at fs/proc/generic.c:180 __xlate_proc_name+0x93/0xb0
[    1.568923] RIP: 0010:__xlate_proc_name+0x93/0xb0
[    1.574200] Call Trace:
[    1.574722]  proc_mkdir+0x18/0x20
[    1.576629]  request_threaded_irq+0xfe/0x160
[    1.576859]  auto_attach+0x60a/0xc40 [cb_pcidas64]

Suggested-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Link: https://lore.kernel.org/r/20210315195814.4692-1-ztong0001@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

staging: comedi: cb_pcidas: fix request_irq() warn

[ Upstream commit 2e5848a3d86f03024ae096478bdb892ab3d79131 ]

request_irq() wont accept a name which contains slash so we need to
repalce it with something else -- otherwise it will trigger a warning
and the entry in /proc/irq/ will not be created
since the .name might be used by userspace and we don't want to break
userspace, so we are changing the parameters passed to request_irq()

[    1.630764] name 'pci-das1602/16'
[    1.630950] WARNING: CPU: 0 PID: 181 at fs/proc/generic.c:180 __xlate_proc_name+0x93/0xb0
[    1.634009] RIP: 0010:__xlate_proc_name+0x93/0xb0
[    1.639441] Call Trace:
[    1.639976]  proc_mkdir+0x18/0x20
[    1.641946]  request_threaded_irq+0xfe/0x160
[    1.642186]  cb_pcidas_auto_attach+0xf4/0x610 [cb_pcidas]

Suggested-by: Ian Abbott <abbotti@mev.co.uk>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Link: https://lore.kernel.org/r/20210315195914.4801-1-ztong0001@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: Fix broken #endif placement

[ Upstream commit 5999b9e5b1f8a2f5417b755130919b3ac96f5550 ]

Only half of the file is under include guard because terminating #endif
is placed too early.

Link: https://lore.kernel.org/r/YE4snvoW1SuwcXAn@localhost.localdomain
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: st: Fix a use after free in st_open()

[ Upstream commit c8c165dea4c8f5ad67b1240861e4f6c5395fa4ac ]

In st_open(), if STp->in_use is true, STp will be freed by
scsi_tape_put(). However, STp is still used by DEBC_printk() after. It is
better to DEBC_printk() before scsi_tape_put().

Link: https://lore.kernel.org/r/20210311064636.10522-1-lyl2019@mail.ustc.edu.cn
Acked-by: Kai Mäkisara <kai.makisara@kolumbus.fi>
Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

io_uring: fix ->flags races by linked timeouts

[ Upstream commit efe814a471e0e58f28f1efaf430c8784a4f36626 ]

It's racy to modify req->flags from a not owning context, e.g. linked
timeout calling req_set_fail_links() for the master request might race
with that request setting/clearing flags while being executed
concurrently. Just remove req_set_fail_links(prev) from
io_link_timeout_fn(), io_async_find_and_cancel() and functions down the
line take care of setting the fail bit.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vhost: Fix vhost_vq_reset()

[ Upstream commit beb691e69f4dec7bfe8b81b509848acfd1f0dbf9 ]

vhost_reset_is_le() is vhost_init_is_le(), and in the case of
cross-endian legacy, vhost_init_is_le() depends on vq->user_be.

vq->user_be is set by vhost_disable_cross_endian().

But in vhost_vq_reset(), we have:

    vhost_reset_is_le(vq);
    vhost_disable_cross_endian(vq);

And so user_be is used before being set.

To fix that, reverse the lines order as there is no other dependency
between them.

Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Link: https://lore.kernel.org/r/20210312140913.788592-1-lvivier@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

kernel: freezer should treat PF_IO_WORKER like PF_KTHREAD for freezing

[ Upstream commit 15b2219facadec583c24523eed40fa45865f859f ]

Don't send fake signals to PF_IO_WORKER threads, they don't accept
signals. Just treat them like kthreads in this regard, all they need
is a wakeup as no forced kernel/user transition is needed.

Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

NFSD: fix error handling in NFSv4.0 callbacks

[ Upstream commit b4250dd868d1b42c0a65de11ef3afbee67ba5d2f ]

When the server tries to do a callback and a client fails it due to
authentication problems, we need the server to set callback down
flag in RENEW so that client can recover.

Suggested-by: Bruce Fields <bfields@redhat.com>
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Tested-by: Benjamin Coddington <bcodding@redhat.com>
Link: https://lore.kernel.org/linux-nfs/FB84E90A-1A03-48B3-8BF7-D9D10AC2C9FE@oracle.com/T/#t
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: cs42l42: Always wait at least 3ms after reset

[ Upstream commit 19325cfea04446bc79b36bffd4978af15f46a00e ]

This delay is part of the power-up sequence defined in the datasheet.
A runtime_resume is a power-up so must also include the delay.

Signed-off-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20210305173442.195740-6-tanureal@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: cs42l42: Fix mixer volume control

[ Upstream commit 72d904763ae6a8576e7ad034f9da4f0e3c44bf24 ]

The minimum value is 0x3f (-63dB), which also is mute

Signed-off-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20210305173442.195740-4-tanureal@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: cs42l42: Fix channel width support

[ Upstream commit 2bdc4f5c6838f7c3feb4fe68e4edbeea158ec0a2 ]

Remove the hard coded 32 bits width and replace with the correct width
calculated by params_width.

Signed-off-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20210305173442.195740-3-tanureal@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: cs42l42: Fix Bitclock polarity inversion

[ Upstream commit e793c965519b8b7f2fea51a48398405e2a501729 ]

The driver was setting bit clock polarity opposite to intended polarity.
Also simplify the code by grouping ADC and DAC clock configurations into
a single field.

Signed-off-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20210305173442.195740-2-tanureal@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: soc-core: Prevent warning if no DMI table is present

[ Upstream commit 7de14d581dbed57c2b3c6afffa2c3fdc6955a3cd ]

Many systems do not use ACPI and hence do not provide a DMI table. On
non-ACPI systems a warning, such as the following, is printed on boot.

 WARNING KERN tegra-audio-graph-card sound: ASoC: no DMI vendor name!

The variable 'dmi_available' is not exported and so currently cannot be
used by kernel modules without adding an accessor. However, it is
possible to use the function is_acpi_device_node() to determine if the
sound card is an ACPI device and hence indicate if we expect a DMI table
to be present. Therefore, call is_acpi_device_node() to see if we are
using ACPI and only parse the DMI table if we are booting with ACPI.

Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://lore.kernel.org/r/20210303115526.419458-1-jonathanh@nvidia.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: es8316: Simplify adc_pga_gain_tlv table

[ Upstream commit bb18c678754ce1514100fb4c0bf6113b5af36c48 ]

Most steps in this table are steps of 3dB (300 centi-dB), so we can
simplify the table.

This not only reduces the amount of space it takes inside the kernel,
this also makes alsa-lib's mixer code actually accept the table, where
as before this change alsa-lib saw the "ADC PGA Gain" control as a
control without a dB scale.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20210228160441.241110-1-hdegoede@redhat.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: sgtl5000: set DAP_AVC_CTRL register to correct default value on probe

[ Upstream commit f86f58e3594fb0ab1993d833d3b9a2496f3c928c ]

According to the SGTL5000 datasheet [1], the DAP_AVC_CTRL register has
the following bit field definitions:

| BITS  | FIELD       | RW | RESET | DEFINITION                        |
| 15    | RSVD        | RO | 0x0   | Reserved                          |
| 14    | RSVD        | RW | 0x1   | Reserved                          |
| 13:12 | MAX_GAIN    | RW | 0x1   | Max Gain of AVC in expander mode  |
| 11:10 | RSVD        | RO | 0x0   | Reserved                          |
| 9:8   | LBI_RESP    | RW | 0x1   | Integrator Response               |
| 7:6   | RSVD        | RO | 0x0   | Reserved                          |
| 5     | HARD_LMT_EN | RW | 0x0   | Enable hard limiter mode          |
| 4:1   | RSVD        | RO | 0x0   | Reserved                          |
| 0     | EN          | RW | 0x0   | Enable/Disable AVC                |

The original default value written to the DAP_AVC_CTRL register during
sgtl5000_i2c_probe() was 0x0510.  This would incorrectly write values to
bits 4 and 10, which are defined as RESERVED.  It would also not set
bits 12 and 14 to their correct RESET values of 0x1, and instead set
them to 0x0.  While the DAP_AVC module is effectively disabled because
the EN bit is 0, this default value is still writing invalid values to
registers that are marked as read-only and RESERVED as well as not
setting bits 12 and 14 to their correct default values as defined by the
datasheet.

The correct value that should be written to the DAP_AVC_CTRL register is
0x5100, which configures the register bits to the default values defined
by the datasheet, and prevents any writes to bits defined as
'read-only'.  Generally speaking, it is best practice to NOT attempt to
write values to registers/bits defined as RESERVED, as it generally
produces unwanted/undefined behavior, or errors.

Also, all credit for this patch should go to my colleague Dan MacDonald
<dmacdonald@curbellmedical.com> for finding this error in the first
place.

[1] https://www.nxp.com/docs/en/data-sheet/SGTL5000.pdf

Signed-off-by: Benjamin Rood <benjaminjrood@gmail.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
Link: https://lore.kernel.org/r/20210219183308.GA2117@ubuntu-dev
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: rt5651: Fix dac- and adc- vol-tlv values being off by a factor of 10

[ Upstream commit eee51df776bd6cac10a76b2779a9fdee3f622b2b ]

The adc_vol_tlv volume-control has a range from -17.625 dB to +30 dB,
not -176.25 dB to + 300 dB. This wrong scale is esp. a problem in userspace
apps which translate the dB scale to a linear scale. With the logarithmic
dB scale being of by a factor of 10 we loose all precision in the lower
area of the range when apps translate things to a linear scale.

E.g. the 0 dB default, which corresponds with a value of 47 of the
0 - 127 range for the control, would be shown as 0/100 in alsa-mixer.

Since the centi-dB values used in the TLV struct cannot represent the
0.375 dB step size used by these controls, change the TLV definition
for them to specify a min and max value instead of min + stepsize.

Note this mirrors commit 3f31f7d9b540 ("ASoC: rt5670: Fix dac- and adc-
vol-tlv values being off by a factor of 10") which made the exact same
change to the rt5670 codec driver.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20210226143817.84287-3-hdegoede@redhat.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: rt5640: Fix dac- and adc- vol-tlv values being off by a factor of 10

[ Upstream commit cfa26ed1f9f885c2fd8f53ca492989d1e16d0199 ]

The adc_vol_tlv volume-control has a range from -17.625 dB to +30 dB,
not -176.25 dB to + 300 dB. This wrong scale is esp. a problem in userspace
apps which translate the dB scale to a linear scale. With the logarithmic
dB scale being of by a factor of 10 we loose all precision in the lower
area of the range when apps translate things to a linear scale.

E.g. the 0 dB default, which corresponds with a value of 47 of the
0 - 127 range for the control, would be shown as 0/100 in alsa-mixer.

Since the centi-dB values used in the TLV struct cannot represent the
0.375 dB step size used by these controls, change the TLV definition
for them to specify a min and max value instead of min + stepsize.

Note this mirrors commit 3f31f7d9b540 ("ASoC: rt5670: Fix dac- and adc-
vol-tlv values being off by a factor of 10") which made the exact same
change to the rt5670 codec driver.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20210226143817.84287-2-hdegoede@redhat.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: rt1015: fix i2c communication error

[ Upstream commit 9e0bdaa9fcb8c64efc1487a7fba07722e7bc515e ]

Remove 0x100 cache re-sync to solve i2c communication error.

Signed-off-by: Jack Yu <jack.yu@realtek.com>
Link: https://lore.kernel.org/r/20210222090057.29532-1-jack.yu@realtek.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iomap: Fix negative assignment to unsigned sis->pages in iomap_swapfile_activate

[ Upstream commit 5808fecc572391867fcd929662b29c12e6d08d81 ]

In case if isi.nr_pages is 0, we are making sis->pages (which is
unsigned int) a huge value in iomap_swapfile_activate() by assigning -1.
This could cause a kernel crash in kernel v4.18 (with below signature).
Or could lead to unknown issues on latest kernel if the fake big swap gets
used.

Fix this issue by returning -EINVAL in case of nr_pages is 0, since it
is anyway a invalid swapfile. Looks like this issue will be hit when
we have pagesize < blocksize type of configuration.

I was able to hit the issue in case of a tiny swap file with below
test script.
https://raw.githubusercontent.com/riteshharjani/LinuxStudy/master/scripts/swap-issue.sh

kernel crash analysis on v4.18
==============================
On v4.18 kernel, it causes a kernel panic, since sis->pages becomes
a huge value and isi.nr_extents is 0. When 0 is returned it is
considered as a swapfile over NFS and SWP_FILE is set (sis->flags |= SWP_FILE).
Then when swapoff was getting called it was calling a_ops->swap_deactivate()
if (sis->flags & SWP_FILE) is true. Since a_ops->swap_deactivate() is
NULL in case of XFS, it causes below panic.

Panic signature on v4.18 kernel:
=======================================
root@qemu:/home/qemu# [ 8291.723351] XFS (loop2): Unmounting Filesystem
[ 8292.123104] XFS (loop2): Mounting V5 Filesystem
[ 8292.132451] XFS (loop2): Ending clean mount
[ 8292.263362] Adding 4294967232k swap on /mnt1/test/swapfile.  Priority:-2 extents:1 across:274877906880k
[ 8292.277834] Unable to handle kernel paging request for instruction fetch
[ 8292.278677] Faulting instruction address: 0x00000000
cpu 0x19: Vector: 400 (Instruction Access) at [c0000009dd5b7ad0]
    pc: 0000000000000000
    lr: c0000000003eb9dc: destroy_swap_extents+0xfc/0x120
    sp: c0000009dd5b7d50
   msr: 8000000040009033
  current = 0xc0000009b6710080
  paca    = 0xc00000003ffcb280   irqmask: 0x03   irq_happened: 0x01
    pid   = 5604, comm = swapoff
Linux version 4.18.0 (riteshh@xxxxxxx) (gcc version 8.4.0 (Ubuntu 8.4.0-1ubuntu1~18.04)) #57 SMP Wed Mar 3 01:33:04 CST 2021
enter ? for help
[link register   ] c0000000003eb9dc destroy_swap_extents+0xfc/0x120
[c0000009dd5b7d50] c0000000025a7058 proc_poll_event+0x0/0x4 (unreliable)
[c0000009dd5b7da0] c0000000003f0498 sys_swapoff+0x3f8/0x910
[c0000009dd5b7e30] c00000000000bbe4 system_call+0x5c/0x70
Exception: c01 (System Call) at 00007ffff7d208d8

Signed-off-by: Ritesh Harjani <riteshh@linux.ibm.com>
[djwong: rework the comment to provide more details]
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rpc: fix NULL dereference on kmalloc failure

[ Upstream commit 0ddc942394013f08992fc379ca04cffacbbe3dae ]

I think this is unlikely but possible:

svc_authenticate sets rq_authop and calls svcauth_gss_accept.  The
kmalloc(sizeof(*svcdata), GFP_KERNEL) fails, leaving rq_auth_data NULL,
and returning SVC_DENIED.

This causes svc_process_common to go to err_bad_auth, and eventually
call svc_authorise.  That calls ->release == svcauth_gss_release, which
tries to dereference rq_auth_data.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Link: https://lore.kernel.org/linux-nfs/3F1B347F-B809-478F-A1E9-0BE98E22B0F0@oracle.com/T/#t
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

fs: nfsd: fix kconfig dependency warning for NFSD_V4

[ Upstream commit 7005227369079963d25fb2d5d736d0feb2c44cf6 ]

When NFSD_V4 is enabled and CRYPTO is disabled,
Kbuild gives the following warning:

WARNING: unmet direct dependencies detected for CRYPTO_SHA256
  Depends on [n]: CRYPTO [=n]
  Selected by [y]:
  - NFSD_V4 [=y] && NETWORK_FILESYSTEMS [=y] && NFSD [=y] && PROC_FS [=y]

WARNING: unmet direct dependencies detected for CRYPTO_MD5
  Depends on [n]: CRYPTO [=n]
  Selected by [y]:
  - NFSD_V4 [=y] && NETWORK_FILESYSTEMS [=y] && NFSD [=y] && PROC_FS [=y]

This is because NFSD_V4 selects CRYPTO_MD5 and CRYPTO_SHA256,
without depending on or selecting CRYPTO, despite those config options
being subordinate to CRYPTO.

Signed-off-by: Julian Braha <julianbraha@gmail.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: fix bh ref count on error paths

[ Upstream commit c915fb80eaa6194fa9bd0a4487705cd5b0dda2f1 ]

__ext4_journalled_writepage should drop bhs' ref count on error paths

Signed-off-by: Zhaolong Zhang <zhangzl2013@126.com>
Link: https://lore.kernel.org/r/1614678151-70481-1-git-send-email-zhangzl2013@126.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: shrink race window in ext4_should_retry_alloc()

[ Upstream commit efc61345274d6c7a46a0570efbc916fcbe3e927b ]

When generic/371 is run on kvm-xfstests using 5.10 and 5.11 kernels, it
fails at significant rates on the two test scenarios that disable
delayed allocation (ext3conv and data_journal) and force actual block
allocation for the fallocate and pwrite functions in the test.  The
failure rate on 5.10 for both ext3conv and data_journal on one test
system typically runs about 85%.  On 5.11, the failure rate on ext3conv
sometimes drops to as low as 1% while the rate on data_journal
increases to nearly 100%.

The observed failures are largely due to ext4_should_retry_alloc()
cutting off block allocation retries when s_mb_free_pending (used to
indicate that a transaction in progress will free blocks) is 0.
However, free space is usually available when this occurs during runs
of generic/371.  It appears that a thread attempting to allocate
blocks is just missing transaction commits in other threads that
increase the free cluster count and reset s_mb_free_pending while
the allocating thread isn't running.  Explicitly testing for free space
availability avoids this race.

The current code uses a post-increment operator in the conditional
expression that determines whether the retry limit has been exceeded.
This means that the conditional expression uses the value of the
retry counter before it's increased, resulting in an extra retry cycle.
The current code actually retries twice before hitting its retry limit
rather than once.

Increasing the retry limit to 3 from the current actual maximum retry
count of 2 in combination with the change described above reduces the
observed failure rate to less that 0.1% on both ext3conv and
data_journal with what should be limited impact on users sensitive to
the overhead caused by retries.

A per filesystem percpu counter exported via sysfs is added to allow
users or developers to track the number of times the retry limit is
exceeded without resorting to debugging methods.  This should provide
some insight into worst case retry behavior.

Signed-off-by: Eric Whitney <enwlinux@gmail.com>
Link: https://lore.kernel.org/r/20210218151132.19678-1-enwlinux@gmail.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

virtiofs: Fail dax mount if device does not support it

[ Upstream commit 3f9b9efd82a84f27e95d0414f852caf1fa839e83 ]

Right now "mount -t virtiofs -o dax myfs /mnt/virtiofs" succeeds even
if filesystem deivce does not have a cache window and hence DAX can't
be supported.

This gives a false sense to user that they are using DAX with virtiofs
but fact of the matter is that they are not.

Fix this by returning error if dax can't be supported and user has asked
for it.

Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Fix fexit trampoline.

[ Upstream commit e21aa341785c679dd409c8cb71f864c00fe6c463 ]

The fexit/fmod_ret programs can be attached to kernel functions that can sleep.
The synchronize_rcu_tasks() will not wait for such tasks to complete.
In such case the trampoline image will be freed and when the task
wakes up the return IP will point to freed memory causing the crash.
Solve this by adding percpu_ref_get/put for the duration of trampoline
and separate trampoline vs its image life times.
The "half page" optimization has to be removed, since
first_half->second_half->first_half transition cannot be guaranteed to
complete in deterministic time. Every trampoline update becomes a new image.
The image with fmod_ret or fexit progs will be freed via percpu_ref_kill and
call_rcu_tasks. Together they will wait for the original function and
trampoline asm to complete. The trampoline is patched from nop to jmp to skip
fexit progs. They are freed independently from the trampoline. The image with
fentry progs only will be freed via call_rcu_tasks_trace+call_rcu_tasks which
will wait for both sleepable and non-sleepable progs to complete.

Fixes: fec56f5890d9 ("bpf: Introduce BPF trampoline")
Reported-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Paul E. McKenney <paulmck@kernel.org> # for RCU
Link: https://lore.kernel.org/bpf/20210316210007.38949-1-alexei.starovoitov@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: mm: correct the inside linear map range during hotplug check

[ Upstream commit ee7febce051945be28ad86d16a15886f878204de ]

Memory hotplug may fail on systems with CONFIG_RANDOMIZE_BASE because the
linear map range is not checked correctly.

The start physical address that linear map covers can be actually at the
end of the range because of randomization. Check that and if so reduce it
to 0.

This can be verified on QEMU with setting kaslr-seed to ~0ul:

memstart_offset_seed = 0xffff
START: __pa(_PAGE_OFFSET(vabits_actual)) = ffff9000c0000000
END:   __pa(PAGE_END - 1) =  1000bfffffff

Signed-off-by: Pavel Tatashin <pasha.tatashin@soleen.com>
Fixes: 58284a901b42 ("arm64/mm: Validate hotplug range before creating linear mapping")
Tested-by: Tyler Hicks <tyhicks@linux.microsoft.com>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Link: https://lore.kernel.org/r/20210216150351.129018-2-pasha.tatashin@soleen.com
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Linux 5.10.27

Tested-by: Andrei Rabusov <a.rabusov@tum.de>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Jason Self <jason@bluehome.net>
Tested-by: Hulk Robot <hulkrobot@huawei.com>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://lore.kernel.org/r/20210329101340.196712908@linuxfoundation.org
Link: https://lore.kernel.org/r/20210329075629.172032742@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

xen-blkback: don't leak persistent grants from xen_blkbk_map()

commit a846738f8c3788d846ed1f587270d2f2e3d32432 upstream.

The fix for XSA-365 zapped too many of the ->persistent_gnt[] entries.
Ones successfully obtained should not be overwritten, but instead left
for xen_blkbk_unmap_prepare() to pick up and put.

This is XSA-371.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Cc: stable@vger.kernel.org
Reviewed-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Wei Liu <wl@xen.org>
Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

can: peak_usb: Revert "can: peak_usb: add forgotten supported devices"

commit 5d7047ed6b7214fbabc16d8712a822e256b1aa44 upstream.

In commit 6417f03132a6 ("module: remove never implemented
MODULE_SUPPORTED_DEVICE") the MODULE_SUPPORTED_DEVICE macro was
removed from the kerne entirely. Shortly before this patch was applied
mainline the commit 59ec7b89ed3e ("can: peak_usb: add forgotten
supported devices") was added to net/master. As this would result in a
merge conflict, let's revert this patch.

Fixes: 59ec7b89ed3e ("can: peak_usb: add forgotten supported devices")
Link: https://lore.kernel.org/r/20210320192649.341832-1-mkl@pengutronix.de
Suggested-by: Leon Romanovsky <leon@kernel.org>
Cc: Stephane Grosjean <s.grosjean@peak-system.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

nvme: fix the nsid value to print in nvme_validate_or_alloc_ns

commit f4f9fc29e56b6fa9d7fa65ec51d3c82aff99c99b upstream.

ns can be NULL at this point, and my move of the check from
the original patch by Chaitanya broke this.

Fixes: 0ec84df4953b ("nvme-core: check ctrl css before setting up zns")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "net: bonding: fix error return code of bond_neigh_init()"

commit 080bfa1e6d928a5d1f185cc44e5f3c251df06df5 upstream.

This reverts commit 2055a99da8a253a357bdfd359b3338ef3375a26c.

This change rejects legitimate configurations.

A slave doesn't need to exist nor implement ndo_slave_setup.

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

Revert "xen: fix p2m size in dom0 for disabled memory hotplug case"

commit af44a387e743ab7aa39d3fb5e29c0a973cf91bdc upstream.

This partially reverts commit 882213990d32 ("xen: fix p2m size in dom0
for disabled memory hotplug case")

There's no need to special case XEN_UNPOPULATED_ALLOC anymore in order
to correctly size the p2m. The generic memory hotplug option has
already been tied together with the Xen hotplug limit, so enabling
memory hotplug should already trigger a properly sized p2m on Xen PV.

Note that XEN_UNPOPULATED_ALLOC depends on ZONE_DEVICE which pulls in
MEMORY_HOTPLUG.

Leave the check added to __set_phys_to_machine and the adjusted
comment about EXTRA_MEM_RATIO.

Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Link: https://lore.kernel.org/r/20210324122424.58685-3-roger.pau@citrix.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[boris: fixed formatting issues]
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>

fs/ext4: fix integer overflow in s_log_groups_per_flex

commit f91436d55a279f045987e8b8c1385585dca54be9 upstream.

syzbot found UBSAN: shift-out-of-bounds in ext4_mb_init [1], when
1 << sbi->s_es->s_log_groups_per_flex is bigger than UINT_MAX,
where sbi->s_mb_prefetch is unsigned integer type.

32 is the maximum allowed power of s_log_groups_per_flex. Following if
check will also trigger UBSAN shift-out-of-bound:

if (1 << sbi->s_es->s_log_groups_per_flex >= UINT_MAX) {

So I'm checking it against the raw number, perhaps there is another way
to calculate UINT_MAX max power. Also use min_t as to make sure it's
uint type.

[1] UBSAN: shift-out-of-bounds in fs/ext4/mballoc.c:2713:24
shift exponent 60 is too large for 32-bit type 'int'
Call Trace:
 __dump_stack lib/dump_stack.c:79 [inline]
 dump_stack+0x137/0x1be lib/dump_stack.c:120
 ubsan_epilogue lib/ubsan.c:148 [inline]
 __ubsan_handle_shift_out_of_bounds+0x432/0x4d0 lib/ubsan.c:395
 ext4_mb_init_backend fs/ext4/mballoc.c:2713 [inline]
 ext4_mb_init+0x19bc/0x19f0 fs/ext4/mballoc.c:2898
 ext4_fill_super+0xc2ec/0xfbe0 fs/ext4/super.c:4983

Reported-by: syzbot+a8b4b0c60155e87e9484@syzkaller.appspotmail.com
Signed-off-by: Sabyrzhan Tasbolatov <snovitoll@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210224095800.3350002-1-snovitoll@gmail.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: add reclaim checks to xattr code

commit 163f0ec1df33cf468509ff38cbcbb5eb0d7fac60 upstream.

Syzbot is reporting that ext4 can enter fs reclaim from kvmalloc() while
the transaction is started like:

  fs_reclaim_acquire+0x117/0x150 mm/page_alloc.c:4340
  might_alloc include/linux/sched/mm.h:193 [inline]
  slab_pre_alloc_hook mm/slab.h:493 [inline]
  slab_alloc_node mm/slub.c:2817 [inline]
  __kmalloc_node+0x5f/0x430 mm/slub.c:4015
  kmalloc_node include/linux/slab.h:575 [inline]
  kvmalloc_node+0x61/0xf0 mm/util.c:587
  kvmalloc include/linux/mm.h:781 [inline]
  ext4_xattr_inode_cache_find fs/ext4/xattr.c:1465 [inline]
  ext4_xattr_inode_lookup_create fs/ext4/xattr.c:1508 [inline]
  ext4_xattr_set_entry+0x1ce6/0x3780 fs/ext4/xattr.c:1649
  ext4_xattr_ibody_set+0x78/0x2b0 fs/ext4/xattr.c:2224
  ext4_xattr_set_handle+0x8f4/0x13e0 fs/ext4/xattr.c:2380
  ext4_xattr_set+0x13a/0x340 fs/ext4/xattr.c:2493

This should be impossible since transaction start sets PF_MEMALLOC_NOFS.
Add some assertions to the code to catch if something isn't working as
expected early.

Link: https://lore.kernel.org/linux-ext4/000000000000563a0205bafb7970@google.com/
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210222171626.21884-1-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mac80211: fix double free in ibss_leave

commit 3bd801b14e0c5d29eeddc7336558beb3344efaa3 upstream.

Clear beacon ie pointer and ie length after free
in order to prevent double free.

==================================================================
BUG: KASAN: double-free or invalid-free \
in ieee80211_ibss_leave+0x83/0xe0 net/mac80211/ibss.c:1876

CPU: 0 PID: 8472 Comm: syz-executor100 Not tainted 5.11.0-rc6-syzkaller #0
Call Trace:
 __dump_stack lib/dump_stack.c:79 [inline]
 dump_stack+0x107/0x163 lib/dump_stack.c:120
 print_address_description.constprop.0.cold+0x5b/0x2c6 mm/kasan/report.c:230
 kasan_report_invalid_free+0x51/0x80 mm/kasan/report.c:355
 ____kasan_slab_free+0xcc/0xe0 mm/kasan/common.c:341
 kasan_slab_free include/linux/kasan.h:192 [inline]
 __cache_free mm/slab.c:3424 [inline]
 kfree+0xed/0x270 mm/slab.c:3760
 ieee80211_ibss_leave+0x83/0xe0 net/mac80211/ibss.c:1876
 rdev_leave_ibss net/wireless/rdev-ops.h:545 [inline]
 __cfg80211_leave_ibss+0x19a/0x4c0 net/wireless/ibss.c:212
 __cfg80211_leave+0x327/0x430 net/wireless/core.c:1172
 cfg80211_leave net/wireless/core.c:1221 [inline]
 cfg80211_netdev_notifier_call+0x9e8/0x12c0 net/wireless/core.c:1335
 notifier_call_chain+0xb5/0x200 kernel/notifier.c:83
 call_netdevice_notifiers_info+0xb5/0x130 net/core/dev.c:2040
 call_netdevice_notifiers_extack net/core/dev.c:2052 [inline]
 call_netdevice_notifiers net/core/dev.c:2066 [inline]
 __dev_close_many+0xee/0x2e0 net/core/dev.c:1586
 __dev_close net/core/dev.c:1624 [inline]
 __dev_change_flags+0x2cb/0x730 net/core/dev.c:8476
 dev_change_flags+0x8a/0x160 net/core/dev.c:8549
 dev_ifsioc+0x210/0xa70 net/core/dev_ioctl.c:265
 dev_ioctl+0x1b1/0xc40 net/core/dev_ioctl.c:511
 sock_do_ioctl+0x148/0x2d0 net/socket.c:1060
 sock_ioctl+0x477/0x6a0 net/socket.c:1177
 vfs_ioctl fs/ioctl.c:48 [inline]
 __do_sys_ioctl fs/ioctl.c:753 [inline]
 __se_sys_ioctl fs/ioctl.c:739 [inline]
 __x64_sys_ioctl+0x193/0x200 fs/ioctl.c:739
 do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Reported-by: syzbot+93976391bf299d425f44@syzkaller.appspotmail.com
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
Link: https://lore.kernel.org/r/20210213133653.367130-1-markus.theil@tu-ilmenau.de
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: dsa: b53: VLAN filtering is global to all users

commit d45c36bafb94e72fdb6dee437279b61b6d97e706 upstream.

The bcm_sf2 driver uses the b53 driver as a library but does not make
usre of the b53_setup() function, this made it fail to inherit the
vlan_filtering_is_global attribute. Fix this by moving the assignment to
b53_switch_alloc() which is used by bcm_sf2.

Fixes: 7228b23e68f7 ("net: dsa: b53: Let DSA handle mismatched VLAN filtering settings")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

r8169: fix DMA being used after buffer free if WoL is enabled

commit f658b90977d2e79822a558e48116e059a7e75dec upstream.

IOMMU errors have been reported if WoL is enabled and interface is
brought down. It turned out that the network chip triggers DMA
transfers after the DMA buffers have been freed. For WoL to work we
need to leave rx enabled, therefore simply stop the chip from being
a DMA busmaster.

Fixes: 567ca57faa62 ("r8169: add rtl8169_up")
Tested-by: Paul Blazejowski <paulb@blazebox.homeip.net>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

can: dev: Move device back to init netns on owning netns delete

commit 3a5ca857079ea022e0b1b17fc154f7ad7dbc150f upstream.

When a non-initial netns is destroyed, the usual policy is to delete
all virtual network interfaces contained, but move physical interfaces
back to the initial netns. This keeps the physical interface visible
on the system.

CAN devices are somewhat special, as they define rtnl_link_ops even
if they are physical devices. If a CAN interface is moved into a
non-initial netns, destroying that netns lets the interface vanish
instead of moving it back to the initial netns. default_device_exit()
skips CAN interfaces due to having rtnl_link_ops set. Reproducer:

  ip netns add foo
  ip link set can0 netns foo
  ip netns delete foo

WARNING: CPU: 1 PID: 84 at net/core/dev.c:11030 ops_exit_list+0x38/0x60
CPU: 1 PID: 84 Comm: kworker/u4:2 Not tainted 5.10.19 #1
Workqueue: netns cleanup_net
[<c010e700>] (unwind_backtrace) from [<c010a1d8>] (show_stack+0x10/0x14)
[<c010a1d8>] (show_stack) from [<c086dc10>] (dump_stack+0x94/0xa8)
[<c086dc10>] (dump_stack) from [<c086b938>] (__warn+0xb8/0x114)
[<c086b938>] (__warn) from [<c086ba10>] (warn_slowpath_fmt+0x7c/0xac)
[<c086ba10>] (warn_slowpath_fmt) from [<c0629f20>] (ops_exit_list+0x38/0x60)
[<c0629f20>] (ops_exit_list) from [<c062a5c4>] (cleanup_net+0x230/0x380)
[<c062a5c4>] (cleanup_net) from [<c0142c20>] (process_one_work+0x1d8/0x438)
[<c0142c20>] (process_one_work) from [<c0142ee4>] (worker_thread+0x64/0x5a8)
[<c0142ee4>] (worker_thread) from [<c0148a98>] (kthread+0x148/0x14c)
[<c0148a98>] (kthread) from [<c0100148>] (ret_from_fork+0x14/0x2c)

To properly restore physical CAN devices to the initial netns on owning
netns exit, introduce a flag on rtnl_link_ops that can be set by drivers.
For CAN devices setting this flag, default_device_exit() considers them
non-virtual, applying the usual namespace move.

The issue was introduced in the commit mentioned below, as at that time
CAN devices did not have a dellink() operation.

Fixes: e008b5fc8dc7 ("net: Simplfy default_device_exit and improve batching.")
Link: https://lore.kernel.org/r/20210302122423.872326-1-martin@strongswan.org
Signed-off-by: Martin Willi <martin@strongswan.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ch_ktls: fix enum-conversion warning

commit 6f235a69e59484e382dc31952025b0308efedc17 upstream.

gcc points out an incorrect enum assignment:

drivers/net/ethernet/chelsio/inline_crypto/ch_ktls/chcr_ktls.c: In function 'chcr_ktls_cpl_set_tcb_rpl':
drivers/net/ethernet/chelsio/inline_crypto/ch_ktls/chcr_ktls.c:684:22: warning: implicit conversion from 'enum <anonymous>' to 'enum ch_ktls_open_state' [-Wenum-conversion]

This appears harmless, and should apparently use 'CH_KTLS_OPEN_SUCCESS'
instead of 'false', with the same value '0'.

Fixes: efca3878a5fb ("ch_ktls: Issue if connection offload fails")
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fs/cachefiles: Remove wait_bit_key layout dependency

commit 39f985c8f667c80a3d1eb19d31138032fa36b09e upstream.

Cachefiles was relying on wait_page_key and wait_bit_key being the
same layout, which is fragile.  Now that wait_page_key is exposed in
the pagemap.h header, we can remove that fragility

A comment on the need to maintain structure layout equivalence was added by
Linus[1] and that is no longer applicable.

Fixes: 62906027091f ("mm: add PageWaiters indicating tasks are waiting for a page bit")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: kafs-testing@auristor.com
cc: linux-cachefs@redhat.com
cc: linux-mm@kvack.org
Link: https://lore.kernel.org/r/20210320054104.1300774-2-willy@infradead.org/
Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=3510ca20ece0150af6b10c77a74ff1b5c198e3e2
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/memcg: fix 5.10 backport of splitting page memcg

The straight backport of 5.12's e1baddf8475b ("mm/memcg: set memcg when
splitting page") works fine in 5.11, but turned out to be wrong for 5.10:
because that relies on a separate flag, which must also be set for the
memcg to be recognized and uncharged and cleared when freeing. Fix that.

Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/mem_encrypt: Correct physical address calculation in __set_clr_pte_enc()

commit 8249d17d3194eac064a8ca5bc5ca0abc86feecde upstream.

The pfn variable contains the page frame number as returned by the
pXX_pfn() functions, shifted to the right by PAGE_SHIFT to remove the
page bits. After page protection computations are done to it, it gets
shifted back to the physical address using page_level_shift().

That is wrong, of course, because that function determines the shift
length based on the level of the page in the page table but in all the
cases, it was shifted by PAGE_SHIFT before.

Therefore, shift it back using PAGE_SHIFT to get the correct physical
address.

 [ bp: Rewrite commit message. ]

Fixes: dfaaec9033b8 ("x86: Add support for changing memory encryption attribute in early boot")
Signed-off-by: Isaku Yamahata <isaku.yamahata@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com>
Cc: <stable@vger.kernel.org>
Link: https://lkml.kernel.org/r/81abbae1657053eccc535c16151f63cd049dcb97.1616098294.git.isaku.yamahata@intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

locking/mutex: Fix non debug version of mutex_lock_io_nested()

commit 291da9d4a9eb3a1cb0610b7f4480f5b52b1825e7 upstream.

If CONFIG_DEBUG_LOCK_ALLOC=n then mutex_lock_io_nested() maps to
mutex_lock() which is clearly wrong because mutex_lock() lacks the
io_schedule_prepare()/finish() invocations.

Map it to mutex_lock_io().

Fixes: f21860bac05b ("locking/mutex, sched/wait: Fix the mutex_lock_io_nested() define")
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/878s6fshii.fsf@nanos.tec.linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: Adjust key sizes and key generation routines for AES256 encryption

commit 45a4546c6167a2da348a31ca439d8a8ff773b6ea upstream.

For AES256 encryption (GCM and CCM), we need to adjust the size of a few
fields to 32 bytes instead of 16 to accommodate the larger keys.

Also, the L value supplied to the key generator needs to be changed from
to 256 when these algorithms are used.

Keeping the ioctl struct for dumping keys of the same size for now.
Will send out a different patch for that one.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
CC: <stable@vger.kernel.org> # v5.10+
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

smb3: fix cached file size problems in duplicate extents (reflink)

commit cfc63fc8126a93cbf95379bc4cad79a7b15b6ece upstream.

There were two problems (one of which could cause data corruption)
that were noticed with duplicate extents (ie reflink)
when debugging why various xfstests were being incorrectly skipped
(e.g. generic/138, generic/140, generic/142). First, we were not
updating the file size locally in the cache when extending a
file due to reflink (it would refresh after actimeo expires)
but xfstest was checking the size immediately which was still
0 so caused the test to be skipped.  Second, we were setting
the target file size (which could shrink the file) in all cases
to the end of the reflinked range rather than only setting the
target file size when reflink would extend the file.

CC: <stable@vger.kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: mpt3sas: Fix error return code of mpt3sas_base_attach()

[ Upstream commit 3401ecf7fc1b9458a19d42c0e26a228f18ac7dda ]

When kzalloc() returns NULL, no error return code of mpt3sas_base_attach()
is assigned. To fix this bug, r is assigned with -ENOMEM in this case.

Link: https://lore.kernel.org/r/20210308035241.3288-1-baijiaju1990@gmail.com
Fixes: c696f7b83ede ("scsi: mpt3sas: Implement device_remove_in_progress check in IOCTL path")
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qedi: Fix error return code of qedi_alloc_global_queues()

[ Upstream commit f69953837ca5d98aa983a138dc0b90a411e9c763 ]

When kzalloc() returns NULL to qedi->global_queues[i], no error return code
of qedi_alloc_global_queues() is assigned.  To fix this bug, status is
assigned with -ENOMEM in this case.

Link: https://lore.kernel.org/r/20210308033024.27147-1-baijiaju1990@gmail.com
Fixes: ace7f46ba5fd ("scsi: qedi: Add QLogic FastLinQ offload iSCSI driver framework.")
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Acked-by: Manish Rangankar <mrangankar@marvell.com>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: Revert "qla2xxx: Make sure that aborted commands are freed"

[ Upstream commit 39c0c8553bfb5a3d108aa47f1256076d507605e3 ]

Calling vha->hw->tgt.tgt_ops->free_cmd() from qlt_xmit_response() is wrong
since the command for which a response is sent must remain valid until the
SCSI target core calls .release_cmd(). It has been observed that the
following scenario triggers a kernel crash:

 - qlt_xmit_response() calls qlt_check_reserve_free_req()

 - qlt_check_reserve_free_req() returns -EAGAIN

 - qlt_xmit_response() calls vha->hw->tgt.tgt_ops->free_cmd(cmd)

 - transport_handle_queue_full() tries to retransmit the response

Fix this crash by reverting the patch that introduced it.

Link: https://lore.kernel.org/r/20210320232359.941-2-bvanassche@acm.org
Fixes: 0dcec41acb85 ("scsi: qla2xxx: Make sure that aborted commands are freed")
Cc: Quinn Tran <qutran@marvell.com>
Cc: Mike Christie <michael.christie@oracle.com>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
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>

block: recalculate segment count for multi-segment discards correctly

[ Upstream commit a958937ff166fc60d1c3a721036f6ff41bfa2821 ]

When a stacked block device inserts a request into another block device
using blk_insert_cloned_request, the request's nr_phys_segments field gets
recalculated by a call to blk_recalc_rq_segments in
blk_cloned_rq_check_limits. But blk_recalc_rq_segments does not know how to
handle multi-segment discards. For disk types which can handle
multi-segment discards like nvme, this results in discard requests which
claim a single segment when it should report several, triggering a warning
in nvme and causing nvme to fail the discard from the invalid state.

 WARNING: CPU: 5 PID: 191 at drivers/nvme/host/core.c:700 nvme_setup_discard+0x170/0x1e0 [nvme_core]
 ...
 nvme_setup_cmd+0x217/0x270 [nvme_core]
 nvme_loop_queue_rq+0x51/0x1b0 [nvme_loop]
 __blk_mq_try_issue_directly+0xe7/0x1b0
 blk_mq_request_issue_directly+0x41/0x70
 ? blk_account_io_start+0x40/0x50
 dm_mq_queue_rq+0x200/0x3e0
 blk_mq_dispatch_rq_list+0x10a/0x7d0
 ? __sbitmap_queue_get+0x25/0x90
 ? elv_rb_del+0x1f/0x30
 ? deadline_remove_request+0x55/0xb0
 ? dd_dispatch_request+0x181/0x210
 __blk_mq_do_dispatch_sched+0x144/0x290
 ? bio_attempt_discard_merge+0x134/0x1f0
 __blk_mq_sched_dispatch_requests+0x129/0x180
 blk_mq_sched_dispatch_requests+0x30/0x60
 __blk_mq_run_hw_queue+0x47/0xe0
 __blk_mq_delay_run_hw_queue+0x15b/0x170
 blk_mq_sched_insert_requests+0x68/0xe0
 blk_mq_flush_plug_list+0xf0/0x170
 blk_finish_plug+0x36/0x50
 xlog_cil_committed+0x19f/0x290 [xfs]
 xlog_cil_process_committed+0x57/0x80 [xfs]
 xlog_state_do_callback+0x1e0/0x2a0 [xfs]
 xlog_ioend_work+0x2f/0x80 [xfs]
 process_one_work+0x1b6/0x350
 worker_thread+0x53/0x3e0
 ? process_one_work+0x350/0x350
 kthread+0x11b/0x140
 ? __kthread_bind_mask+0x60/0x60
 ret_from_fork+0x22/0x30

This patch fixes blk_recalc_rq_segments to be aware of devices which can
have multi-segment discards. It calculates the correct discard segment
count by counting the number of bio as each discard bio is considered its
own segment.

Fixes: 1e739730c5b9 ("block: optionally merge discontiguous discard bios into a single request")
Signed-off-by: David Jeffery <djeffery@redhat.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Laurence Oberman <loberman@redhat.com>
Link: https://lore.kernel.org/r/20210211143807.GA115624@redhat
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

io_uring: fix provide_buffers sign extension

[ Upstream commit d81269fecb8ce16eb07efafc9ff5520b2a31c486 ]

io_provide_buffers_prep()'s "p->len * p->nbufs" to sign extension
problems. Not a huge problem as it's only used for access_ok() and
increases the checked length, but better to keep typing right.

Reported-by: Colin Ian King <colin.king@canonical.com>
Fixes: efe68c1ca8f49 ("io_uring: validate the full range of provided buffers for access")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Reviewed-by: Colin Ian King <colin.king@canonical.com>
Link: https://lore.kernel.org/r/562376a39509e260d8532186a06226e56eb1f594.1616149233.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf synthetic events: Avoid write of uninitialized memory when generating PERF_RECORD_MMAP* records

[ Upstream commit 2a76f6de07906f0bb5f2a13fb02845db1695cc29 ]

Account for alignment bytes in the zero-ing memset.

Fixes: 1a853e36871b533c ("perf record: Allow specifying a pid to record")
Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lore.kernel.org/lkml/20210309234945.419254-1-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf auxtrace: Fix auxtrace queue conflict

[ Upstream commit b410ed2a8572d41c68bd9208555610e4b07d0703 ]

The only requirement of an auxtrace queue is that the buffers are in
time order.  That is achieved by making separate queues for separate
perf buffer or AUX area buffer mmaps.

That generally means a separate queue per cpu for per-cpu contexts, and
a separate queue per thread for per-task contexts.

When buffers are added to a queue, perf checks that the buffer cpu and
thread id (tid) match the queue cpu and thread id.

However, generally, that need not be true, and perf will queue buffers
correctly anyway, so the check is not needed.

In addition, the check gets erroneously hit when using sample mode to
trace multiple threads.

Consequently, fix that case by removing the check.

Fixes: e502789302a6 ("perf auxtrace: Add helpers for queuing AUX area tracing data")
Reported-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lore.kernel.org/lkml/20210308151143.18338-1-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: scan: Use unique number for instance_no

[ Upstream commit eb50aaf960e3bedfef79063411ffd670da94b84b ]

The decrementation of acpi_device_bus_id->instance_no
in acpi_device_del() is incorrect, because it may cause
a duplicate instance number to be allocated next time
a device with the same acpi_device_bus_id is added.

Replace above mentioned approach by using IDA framework.

While at it, define the instance range to be [0, 4096).

Fixes: e49bd2dd5a50 ("ACPI: use PNPID:instance_no as bus_id of ACPI device")
Fixes: ca9dc8d42b30 ("ACPI / scan: Fix acpi_bus_id_list bookkeeping")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: 4.10+ <stable@vger.kernel.org> # 4.10+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: scan: Rearrange memory allocation in acpi_device_add()

[ Upstream commit c1013ff7a5472db637c56bb6237f8343398c03a7 ]

The upfront allocation of new_bus_id is done to avoid allocating
memory under acpi_device_lock, but it doesn't really help,
because (1) it leads to many unnecessary memory allocations for
_ADR devices, (2) kstrdup_const() is run under that lock anyway and
(3) it complicates the code.

Rearrange acpi_device_add() to allocate memory for a new struct
acpi_device_bus_id instance only when necessary, eliminate a redundant
local variable from it and reduce the number of labels in there.

No intentional functional impact.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Revert "netfilter: x_tables: Update remaining dereference to RCU"

[ Upstream commit abe7034b9a8d57737e80cc16d60ed3666990bdbf ]

This reverts commit 443d6e86f821a165fae3fc3fc13086d27ac140b1.

This (and the following) patch basically re-implemented the RCU
mechanisms of patch 784544739a25. That patch was replaced because of the
performance problems that it created when replacing tables. Now, we have
the same issue: the call to synchronize_rcu() makes replacing tables
slower by as much as an order of magnitude.

Revert these patches and fix the issue in a different way.

Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mm/mmu_notifiers: ensure range_end() is paired with range_start()

[ Upstream commit c2655835fd8cabdfe7dab737253de3ffb88da126 ]

If one or more notifiers fails .invalidate_range_start(), invoke
.invalidate_range_end() for "all" notifiers.  If there are multiple
notifiers, those that did not fail are expecting _start() and _end() to
be paired, e.g.  KVM's mmu_notifier_count would become imbalanced.
Disallow notifiers that can fail _start() from implementing _end() so
that it's unnecessary to either track which notifiers rejected _start(),
or had already succeeded prior to a failed _start().

Note, the existing behavior of calling _start() on all notifiers even
after a previous notifier failed _start() was an unintented "feature".
Make it canon now that the behavior is depended on for correctness.

As of today, the bug is likely benign:

  1. The only caller of the non-blocking notifier is OOM kill.
  2. The only notifiers that can fail _start() are the i915 and Nouveau
     drivers.
  3. The only notifiers that utilize _end() are the SGI UV GRU driver
     and KVM.
  4. The GRU driver will never coincide with the i195/Nouveau drivers.
  5. An imbalanced kvm->mmu_notifier_count only causes soft lockup in the
     _guest_, and the guest is already doomed due to being an OOM victim.

Fix the bug now to play nice with future usage, e.g.  KVM has a
potential use case for blocking memslot updates in KVM while an
invalidation is in-progress, and failure to unblock would result in said
updates being blocked indefinitely and hanging.

Found by inspection.  Verified by adding a second notifier in KVM that
periodically returns -EAGAIN on non-blockable ranges, triggering OOM,
and observing that KVM exits with an elevated notifier count.

Link: https://lkml.kernel.org/r/20210311180057.1582638-1-seanjc@google.com
Fixes: 93065ac753e4 ("mm, oom: distinguish blockable mode for mmu notifiers")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Suggested-by: Jason Gunthorpe <jgg@ziepe.ca>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ben Gardon <bgardon@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Jérôme Glisse" <jglisse@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Dimitri Sivanich <dimitri.sivanich@hpe.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>

dm table: Fix zoned model check and zone sectors check

[ Upstream commit 2d669ceb69c276f7637cf760287ca4187add082e ]

Commit 24f6b6036c9e ("dm table: fix zoned iterate_devices based device
capability checks") triggered dm table load failure when dm-zoned device
is set up for zoned block devices and a regular device for cache.

The commit inverted logic of two callback functions for iterate_devices:
device_is_zoned_model() and device_matches_zone_sectors(). The logic of
device_is_zoned_model() was inverted then all destination devices of all
targets in dm table are required to have the expected zoned model. This
is fine for dm-linear, dm-flakey and dm-crypt on zoned block devices
since each target has only one destination device. However, this results
in failure for dm-zoned with regular cache device since that target has
both regular block device and zoned block devices.

As for device_matches_zone_sectors(), the commit inverted the logic to
require all zoned block devices in each target have the specified
zone_sectors. This check also fails for regular block device which does
not have zones.

To avoid the check failures, fix the zone model check and the zone
sectors check. For zone model check, introduce the new feature flag
DM_TARGET_MIXED_ZONED_MODEL, and set it to dm-zoned target. When the
target has this flag, allow it to have destination devices with any
zoned model. For zone sectors check, skip the check if the destination
device is not a zoned block device. Also add comments and improve an
error message to clarify expectations to the two checks.

Fixes: 24f6b6036c9e ("dm table: fix zoned iterate_devices based device capability checks")
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: x_tables: Use correct memory barriers.

[ Upstream commit 175e476b8cdf2a4de7432583b49c871345e4f8a1 ]

When a new table value was assigned, it was followed by a write memory
barrier. This ensured that all writes before this point would complete
before any writes after this point. However, to determine whether the
rules are unused, the sequence counter is read. To ensure that all
writes have been done before these reads, a full memory barrier is
needed, not just a write memory barrier. The same argument applies when
incrementing the counter, before the rules are read.

Changing to using smp_mb() instead of smp_wmb() fixes the kernel panic
reported in cc00bcaa5899 (which is still present), while still
maintaining the same speed of replacing tables.

The smb_mb() barriers potentially slow the packet path, however testing
has shown no measurable change in performance on a 4-core MIPS64
platform.

Fixes: 7f5c6d4f665b ("netfilter: get rid of atomic ops in fast path")
Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Revert "netfilter: x_tables: Switch synchronization to RCU"

[ Upstream commit d3d40f237480abf3268956daf18cdc56edd32834 ]

This reverts commit cc00bcaa589914096edef7fb87ca5cee4a166b5c.

This (and the preceding) patch basically re-implemented the RCU
mechanisms of patch 784544739a25. That patch was replaced because of the
performance problems that it created when replacing tables. Now, we have
the same issue: the call to synchronize_rcu() makes replacing tables
slower by as much as an order of magnitude.

Prior to using RCU a script calling "iptables" approx. 200 times was
taking 1.16s. With RCU this increased to 11.59s.

Revert these patches and fix the issue in a different way.

Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: phy: broadcom: Fix RGMII delays for BCM50160 and BCM50610M

[ Upstream commit b1dd9bf688b0dcc5a34dca660de46c7570bd9243 ]

The PHY driver entry for BCM50160 and BCM50610M calls
bcm54xx_config_init() but does not call bcm54xx_config_clock_delay() in
order to configuration appropriate clock delays on the PHY, fix that.

Fixes: 733336262b28 ("net: phy: Allow BCM5481x PHYs to setup internal TX/RX clock delay")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: phy: broadcom: Set proper 1000BaseX/SGMII interface mode for BCM54616S

[ Upstream commit 3afd0218992a8d1398e9791d6c2edd4c948ae7ee ]

The default configuration for the BCM54616S PHY may not match the desired
mode when using 1000BaseX or SGMII interface modes, such as when it is on
an SFP module. Add code to explicitly set the correct mode using
programming sequences provided by Bel-Fuse:

https://www.belfuse.com/resources/datasheets/powersolutions/ds-bps-sfp-1gbt-05-series.pdf
https://www.belfuse.com/resources/datasheets/powersolutions/ds-bps-sfp-1gbt-06-series.pdf

Signed-off-by: Robert Hancock <robert.hancock@calian.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: phy: broadcom: Avoid forward for bcm54xx_config_clock_delay()

[ Upstream commit 133bf7b4fbbe58cff5492e37e95e75c88161f1b8 ]

Avoid a forward declaration by moving the callers of
bcm54xx_config_clock_delay() below its body.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: phy: introduce phydev->port

[ Upstream commit 4217a64e18a1647a0dbc68cb3169a5a06f054ec8 ]

At the moment, PORT_MII is reported in the ethtool ops. This is odd
because it is an interface between the MAC and the PHY and no external
port. Some network card drivers will overwrite the port to twisted pair
or fiber, though. Even worse, the MDI/MDIX setting is only used by
ethtool if the port is twisted pair.

Set the port to PORT_TP by default because most PHY drivers are copper
ones. If there is fibre support and it is enabled, the PHY driver will
set it to PORT_FIBRE.

This will change reporting PORT_MII to either PORT_TP or PORT_FIBRE;
except for the genphy fallback driver.

Suggested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: axienet: Fix probe error cleanup

[ Upstream commit 59cd4f19267a0aab87a8c07e4426eb7187ee548d ]

The driver did not always clean up all allocated resources when probe
failed. Fix the probe cleanup path to clean up everything that was
allocated.

Fixes: 57baf8cc70ea ("net: axienet: Handle deferred probe on clock properly")
Signed-off-by: Robert Hancock <robert.hancock@calian.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: axienet: Properly handle PCS/PMA PHY for 1000BaseX mode

[ Upstream commit 1a02556086fc0eb16e0a0d09043e9ffb0e31c7db ]

Update the axienet driver to properly support the Xilinx PCS/PMA PHY
component which is used for 1000BaseX and SGMII modes, including
properly configuring the auto-negotiation mode of the PHY and reading
the negotiated state from the PHY.

Signed-off-by: Robert Hancock <robert.hancock@calian.com>
Reviewed-by: Radhey Shyam Pandey <radhey.shyam.pandey@xilinx.com>
Link: https://lore.kernel.org/r/20201028171429.1699922-1-robert.hancock@calian.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

igb: avoid premature Rx buffer reuse

[ Upstream commit 98dfb02aa22280bd8833836d1b00ab0488fa951f ]

Igb needs a similar fix as commit 75aab4e10ae6a ("i40e: avoid
premature Rx buffer reuse")

The page recycle code, incorrectly, relied on that a page fragment
could not be freed inside xdp_do_redirect(). This assumption leads to
that page fragments that are used by the stack/XDP redirect can be
reused and overwritten.

To avoid this, store the page count prior invoking xdp_do_redirect().

Longer explanation:

Intel NICs have a recycle mechanism. The main idea is that a page is
split into two parts. One part is owned by the driver, one part might
be owned by someone else, such as the stack.

t0: Page is allocated, and put on the Rx ring
              +---------------
used by NIC ->| upper buffer
(rx_buffer)   +---------------
              | lower buffer
              +---------------
  page count  == USHRT_MAX
  rx_buffer->pagecnt_bias == USHRT_MAX

t1: Buffer is received, and passed to the stack (e.g.)
              +---------------
              | upper buff (skb)
              +---------------
used by NIC ->| lower buffer
(rx_buffer)   +---------------
  page count  == USHRT_MAX
  rx_buffer->pagecnt_bias == USHRT_MAX - 1

t2: Buffer is received, and redirected
              +---------------
              | upper buff (skb)
              +---------------
used by NIC ->| lower buffer
(rx_buffer)   +---------------

Now, prior calling xdp_do_redirect():
  page count  == USHRT_MAX
  rx_buffer->pagecnt_bias == USHRT_MAX - 2

This means that buffer *cannot* be flipped/reused, because the skb is
still using it.

The problem arises when xdp_do_redirect() actually frees the
segment. Then we get:
  page count  == USHRT_MAX - 1
  rx_buffer->pagecnt_bias == USHRT_MAX - 2

From a recycle perspective, the buffer can be flipped and reused,
which means that the skb data area is passed to the Rx HW ring!

To work around this, the page count is stored prior calling
xdp_do_redirect().

Fixes: 9cbc948b5a20 ("igb: add XDP support")
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>
Tested-by: Vishakha Jambekar <vishakha.jambekar@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net, bpf: Fix ip6ip6 crash with collect_md populated skbs

[ Upstream commit a188bb5638d41aa99090ebf2f85d3505ab13fba5 ]

I ran into a crash where setting up a ip6ip6 tunnel device which was /not/
set to collect_md mode was receiving collect_md populated skbs for xmit.

The BPF prog was populating the skb via bpf_skb_set_tunnel_key() which is
assigning special metadata dst entry and then redirecting the skb to the
device, taking ip6_tnl_start_xmit() -> ipxip6_tnl_xmit() -> ip6_tnl_xmit()
and in the latter it performs a neigh lookup based on skb_dst(skb) where
we trigger a NULL pointer dereference on dst->ops->neigh_lookup() since
the md_dst_ops do not populate neigh_lookup callback with a fake handler.

Transform the md_dst_ops into generic dst_blackhole_ops that can also be
reused elsewhere when needed, and use them for the metadata dst entries as
callback ops.

Also, remove the dst_md_discard{,_out}() ops and rely on dst_discard{,_out}()
from dst_init() which free the skb the same way modulo the splat. Given we
will be able to recover just fine from there, avoid any potential splats
iff this gets ever triggered in future (or worse, panic on warns when set).

Fixes: f38a9eb1f77b ("dst: Metadata destinations")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: Consolidate common blackhole dst ops

[ Upstream commit c4c877b2732466b4c63217baad05c96f775912c7 ]

Move generic blackhole dst ops to the core and use them from both
ipv4_dst_blackhole_ops and ip6_dst_blackhole_ops where possible. No
functional change otherwise. We need these also in other locations
and having to define them over and over again is not great.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Don't do bpf_cgroup_storage_set() for kuprobe/tp programs

[ Upstream commit 05a68ce5fa51a83c360381630f823545c5757aa2 ]

For kuprobe and tracepoint bpf programs, kernel calls
trace_call_bpf() which calls BPF_PROG_RUN_ARRAY_CHECK()
to run the program array. Currently, BPF_PROG_RUN_ARRAY_CHECK()
also calls bpf_cgroup_storage_set() to set percpu
cgroup local storage with NULL value. This is
due to Commit 394e40a29788 ("bpf: extend bpf_prog_array to store
pointers to the cgroup storage") which modified
__BPF_PROG_RUN_ARRAY() to call bpf_cgroup_storage_set()
and this macro is also used by BPF_PROG_RUN_ARRAY_CHECK().

kuprobe and tracepoint programs are not allowed to call
bpf_get_local_storage() helper hence does not
access percpu cgroup local storage. Let us
change BPF_PROG_RUN_ARRAY_CHECK() not to
modify percpu cgroup local storage.

The issue is observed when I tried to debug [1] where
percpu data is overwritten due to
  preempt_disable -> migration_disable
change. This patch does not completely fix the above issue,
which will be addressed separately, e.g., multiple cgroup
prog runs may preempt each other. But it does fix
any potential issue caused by tracing program
overwriting percpu cgroup storage:
 - in a busy system, a tracing program is to run between
   bpf_cgroup_storage_set() and the cgroup prog run.
 - a kprobe program is triggered by a helper in cgroup prog
   before bpf_get_local_storage() is called.

 [1] https://lore.kernel.org/bpf/CAKH8qBuXCfUz=w8L+Fj74OaUpbosO29niYwTki7e3Ag044_aww@mail.gmail.com/T

Fixes: 394e40a29788 ("bpf: extend bpf_prog_array to store pointers to the cgroup storage")
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Roman Gushchin <guro@fb.com>
Link: https://lore.kernel.org/bpf/20210309185028.3763817-1-yhs@fb.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/cxgb4: Fix adapter LE hash errors while destroying ipv6 listening server

[ Upstream commit 3408be145a5d6418ff955fe5badde652be90e700 ]

Not setting the ipv6 bit while destroying ipv6 listening servers may
result in potential fatal adapter errors due to lookup engine memory hash
errors. Therefore always set ipv6 field while destroying ipv6 listening
servers.

Fixes: 830662f6f032 ("RDMA/cxgb4: Add support for active and passive open connection with IPv6 address")
Link: https://lore.kernel.org/r/20210324190453.8171-1-bharat@chelsio.com
Signed-off-by: Potnuri Bharat Teja <bharat@chelsio.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

xen/x86: make XEN_BALLOON_MEMORY_HOTPLUG_LIMIT depend on MEMORY_HOTPLUG

[ Upstream commit 2b514ec72706a31bea0c3b97e622b81535b5323a ]

The Xen memory hotplug limit should depend on the memory hotplug
generic option, rather than the Xen balloon configuration. It's
possible to have a kernel with generic memory hotplug enabled, but
without Xen balloon enabled, at which point memory hotplug won't work
correctly due to the size limitation of the p2m.

Rename the option to XEN_MEMORY_HOTPLUG_LIMIT since it's no longer
tied to ballooning.

Fixes: 9e2369c06c8a18 ("xen: add helpers to allocate unpopulated memory")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Link: https://lore.kernel.org/r/20210324122424.58685-2-roger.pau@citrix.com
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

octeontx2-af: Fix memory leak of object buf

[ Upstream commit 9e0a537d06fc36861e4f78d0a7df1fe2b3592714 ]

Currently the error return path when lfs fails to allocate is not free'ing
the memory allocated to buf. Fix this by adding the missing kfree.

Addresses-Coverity: ("Resource leak")
Fixes: f7884097141b ("octeontx2-af: Formatting debugfs entry rsrc_alloc.")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: bridge: don't notify switchdev for local FDB addresses

[ Upstream commit 6ab4c3117aec4e08007d9e971fa4133e1de1082d ]

As explained in this discussion:
https://lore.kernel.org/netdev/20210117193009.io3nungdwuzmo5f7@skbuf/

the switchdev notifiers for FDB entries managed to have a zero-day bug.
The bridge would not say that this entry is local:

ip link add br0 type bridge
ip link set swp0 master br0
bridge fdb add dev swp0 00:01:02:03:04:05 master local

and the switchdev driver would be more than happy to offload it as a
normal static FDB entry. This is despite the fact that 'local' and
non-'local' entries have completely opposite directions: a local entry
is locally terminated and not forwarded, whereas a static entry is
forwarded and not locally terminated. So, for example, DSA would install
this entry on swp0 instead of installing it on the CPU port as it should.

There is an even sadder part, which is that the 'local' flag is implicit
if 'static' is not specified, meaning that this command produces the
same result of adding a 'local' entry:

bridge fdb add dev swp0 00:01:02:03:04:05 master

I've updated the man pages for 'bridge', and after reading it now, it
should be pretty clear to any user that the commands above were broken
and should have never resulted in the 00:01:02:03:04:05 address being
forwarded (this behavior is coherent with non-switchdev interfaces):
https://patchwork.kernel.org/project/netdevbpf/cover/20210211104502.2081443-1-olteanv@gmail.com/
If you're a user reading this and this is what you want, just use:

bridge fdb add dev swp0 00:01:02:03:04:05 master static

Because switchdev should have given drivers the means from day one to
classify FDB entries as local/non-local, but didn't, it means that all
drivers are currently broken. So we can just as well omit the switchdev
notifications for local FDB entries, which is exactly what this patch
does to close the bug in stable trees. For further development work
where drivers might want to trap the local FDB entries to the host, we
can add a 'bool is_local' to br_switchdev_fdb_call_notifiers(), and
selectively make drivers act upon that bit, while all the others ignore
those entries if the 'is_local' bit is set.

Fixes: 6b26b51b1d13 ("net: bridge: Add support for notifying devices about FDB add/del")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PM: EM: postpone creating the debugfs dir till fs_initcall

[ Upstream commit fb9d62b27ab1e07d625591549c314b7d406d21df ]

The debugfs directory '/sys/kernel/debug/energy_model' is needed before
the Energy Model registration can happen. With the recent change in
debugfs subsystem it's not allowed to create this directory at early
stage (core_initcall). Thus creating this directory would fail.

Postpone the creation of the EM debug dir to later stage: fs_initcall.

It should be safe since all clients: CPUFreq drivers, Devfreq drivers
will be initialized in later stages.

The custom debug log below prints the time of creation the EM debug dir
at fs_initcall and successful registration of EMs at later stages.

[    1.505717] energy_model: creating rootdir
[    3.698307] cpu cpu0: EM: created perf domain
[    3.709022] cpu cpu1: EM: created perf domain

Fixes: 56348560d495 ("debugfs: do not attempt to create a new file before the filesystem is initalized")
Reported-by: Ionela Voinescu <ionela.voinescu@arm.com>
Signed-off-by: Lukasz Luba <lukasz.luba@arm.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/mlx5e: Fix error path for ethtool set-priv-flag

[ Upstream commit 4eacfe72e3e037e3fc019113df32c39a705148c2 ]

Expose error value when failing to comply to command:
$ ethtool --set-priv-flags eth2 rx_cqe_compress [on/off]

Fixes: be7e87f92b58 ("net/mlx5e: Fail safe cqe compressing/moderation mode setting")
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/mlx5e: Offload tuple rewrite for non-CT flows

[ Upstream commit 96b5b4585843e3c83fb1930e5dfbefd0fb889c55 ]

Setting connection tracking OVS flows and then setting non-CT flows that
use tuple rewrite action (e.g. mod_tp_dst), causes the latter flows not
being offloaded.

Fix by using a stricter condition in modify_header_match_supported() to
check tuple rewrite support only for flows with CT action. The check is
factored out into standalone modify_tuple_supported() function to aid
readability.

Fixes: 7e36feeb0467 ("net/mlx5e: CT: Don't offload tuple rewrites for established tuples")
Signed-off-by: Dima Chumak <dchumak@nvidia.com>
Reviewed-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/mlx5e: Allow to match on MPLS parameters only for MPLS over UDP

[ Upstream commit 7d6c86e3ccb5ceea767df5c7a9a17cdfccd3df9a ]

Currently, we support hardware offload only for MPLS over UDP.
However, rules matching on MPLS parameters are now wrongly offloaded
for regular MPLS, without actually taking the parameters into
consideration when doing the offload.
Fix it by rejecting such unsupported rules.

Fixes: 72046a91d134 ("net/mlx5e: Allow to match on mpls parameters")
Signed-off-by: Alaa Hleihel <alaa@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/mlx5: Add back multicast stats for uplink representor

[ Upstream commit a07231084da2207629b42244380ae2f1e10bd9b4 ]

The multicast counter got removed from uplink representor due to the
cited patch.

Fixes: 47c97e6b10a1 ("net/mlx5e: Fix multicast counter not up-to-date in "ip -s"")
Signed-off-by: Huy Nguyen <huyn@nvidia.com>
Reviewed-by: Daniel Jurgens <danielj@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PM: runtime: Defer suspending suppliers

[ Upstream commit 5244f5e2d801259af877ee759e8c22364c607072 ]

Because the PM-runtime status of the device is not updated in
__rpm_callback(), attempts to suspend the suppliers of the given
device triggered by the rpm_put_suppliers() call in there may
cause a supplier to be suspended completely before the status of
the consumer is updated to RPM_SUSPENDED, which is confusing.

To avoid that (1) modify __rpm_callback() to only decrease the
PM-runtime usage counter of each supplier and (2) make rpm_suspend()
try to suspend the suppliers after changing the consumer's status to
RPM_SUSPENDED, in analogy with the device's parent.

Link: https://lore.kernel.org/linux-pm/CAPDyKFqm06KDw_p8WXsM4dijDbho4bb6T4k50UqqvR1_COsp8g@mail.gmail.com/
Fixes: 21d5c57b3726 ("PM / runtime: Use device links")
Reported-by: elaine.zhang <zhangqing@rock-chips.com>
Diagnosed-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: kdump: update ppos when reading elfcorehdr

[ Upstream commit 141f8202cfa4192c3af79b6cbd68e7760bb01b5a ]

The ppos points to a position in the old kernel memory (and in case of
arm64 in the crash kernel since elfcorehdr is passed as a segment). The
function should update the ppos by the amount that was read. This bug is
not exposed by accident, but other platforms update this value properly.
So, fix it in ARM64 version of elfcorehdr_read() as well.

Signed-off-by: Pavel Tatashin <pasha.tatashin@soleen.com>
Fixes: e62aaeac426a ("arm64: kdump: provide /proc/vmcore file")
Reviewed-by: Tyler Hicks <tyhicks@linux.microsoft.com>
Link: https://lore.kernel.org/r/20210319205054.743368-1-pasha.tatashin@soleen.com
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/msm: Fix suspend/resume on i.MX5

[ Upstream commit a9748134ea4aad989e52a6a91479e0acfd306e5b ]

When putting iMX5 into suspend, the following flow is
observed:

[   70.023427] [<c07755f0>] (msm_atomic_commit_tail) from [<c06e7218>]
(commit_tail+0x9c/0x18c)
[   70.031890] [<c06e7218>] (commit_tail) from [<c0e2920c>]
(drm_atomic_helper_commit+0x1a0/0x1d4)
[   70.040627] [<c0e2920c>] (drm_atomic_helper_commit) from
[<c06e74d4>] (drm_atomic_helper_disable_all+0x1c4/0x1d4)
[   70.050913] [<c06e74d4>] (drm_atomic_helper_disable_all) from
[<c0e2943c>] (drm_atomic_helper_suspend+0xb8/0x170)
[   70.061198] [<c0e2943c>] (drm_atomic_helper_suspend) from
[<c06e84bc>] (drm_mode_config_helper_suspend+0x24/0x58)

In the i.MX5 case, priv->kms is not populated (as i.MX5 does not use any
of the Qualcomm display controllers), causing a NULL pointer
dereference in msm_atomic_commit_tail():

[   24.268964] 8<--- cut here ---
[   24.274602] Unable to handle kernel NULL pointer dereference at
virtual address 00000000
[   24.283434] pgd = (ptrval)
[   24.286387] [00000000] *pgd=ca212831
[   24.290788] Internal error: Oops: 17 [#1] SMP ARM
[   24.295609] Modules linked in:
[   24.298777] CPU: 0 PID: 197 Comm: init Not tainted 5.11.0-rc2-next-20210111 #333
[   24.306276] Hardware name: Freescale i.MX53 (Device Tree Support)
[   24.312442] PC is at msm_atomic_commit_tail+0x54/0xb9c
[   24.317743] LR is at commit_tail+0xa4/0x1b0

Fix the problem by calling drm_mode_config_helper_suspend/resume()
only when priv->kms is available.

Fixes: ca8199f13498 ("drm/msm/dpu: ensure device suspend happens during PM sleep")
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/msm: fix shutdown hook in case GPU components failed to bind

[ Upstream commit 623f279c77811475ac8fd5635cc4e4451aa71291 ]

If GPU components have failed to bind, shutdown callback would fail with
the following backtrace. Add safeguard check to stop that oops from
happening and allow the board to reboot.

[   66.617046] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
[   66.626066] Mem abort info:
[   66.628939]   ESR = 0x96000006
[   66.632088]   EC = 0x25: DABT (current EL), IL = 32 bits
[   66.637542]   SET = 0, FnV = 0
[   66.640688]   EA = 0, S1PTW = 0
[   66.643924] Data abort info:
[   66.646889]   ISV = 0, ISS = 0x00000006
[   66.650832]   CM = 0, WnR = 0
[   66.653890] user pgtable: 4k pages, 48-bit VAs, pgdp=0000000107f81000
[   66.660505] [0000000000000000] pgd=0000000100bb2003, p4d=0000000100bb2003, pud=0000000100897003, pmd=0000000000000000
[   66.671398] Internal error: Oops: 96000006 [#1] PREEMPT SMP
[   66.677115] Modules linked in:
[   66.680261] CPU: 6 PID: 352 Comm: reboot Not tainted 5.11.0-rc2-00309-g79e3faa756b2 #38
[   66.688473] Hardware name: Qualcomm Technologies, Inc. Robotics RB5 (DT)
[   66.695347] pstate: 60400005 (nZCv daif +PAN -UAO -TCO BTYPE=--)
[   66.701507] pc : msm_atomic_commit_tail+0x78/0x4e0
[   66.706437] lr : commit_tail+0xa4/0x184
[   66.710381] sp : ffff8000108f3af0
[   66.713791] x29: ffff8000108f3af0 x28: ffff418c44337000
[   66.719242] x27: 0000000000000000 x26: ffff418c40a24490
[   66.724693] x25: ffffd3a842a4f1a0 x24: 0000000000000008
[   66.730146] x23: ffffd3a84313f030 x22: ffff418c444ce000
[   66.735598] x21: ffff418c408a4980 x20: 0000000000000000
[   66.741049] x19: 0000000000000000 x18: ffff800010710fbc
[   66.746500] x17: 000000000000000c x16: 0000000000000001
[   66.751954] x15: 0000000000010008 x14: 0000000000000068
[   66.757405] x13: 0000000000000001 x12: 0000000000000000
[   66.762855] x11: 0000000000000001 x10: 00000000000009b0
[   66.768306] x9 : ffffd3a843192000 x8 : ffff418c44337000
[   66.773757] x7 : 0000000000000000 x6 : 00000000a401b34e
[   66.779210] x5 : 00ffffffffffffff x4 : 0000000000000000
[   66.784660] x3 : 0000000000000000 x2 : ffff418c444ce000
[   66.790111] x1 : ffffd3a841dce530 x0 : ffff418c444cf000
[   66.795563] Call trace:
[   66.798075]  msm_atomic_commit_tail+0x78/0x4e0
[   66.802633]  commit_tail+0xa4/0x184
[   66.806217]  drm_atomic_helper_commit+0x160/0x390
[   66.811051]  drm_atomic_commit+0x4c/0x60
[   66.815082]  drm_atomic_helper_disable_all+0x1f4/0x210
[   66.820355]  drm_atomic_helper_shutdown+0x80/0x130
[   66.825276]  msm_pdev_shutdown+0x14/0x20
[   66.829303]  platform_shutdown+0x28/0x40
[   66.833330]  device_shutdown+0x158/0x330
[   66.837357]  kernel_restart+0x40/0xa0
[   66.841122]  __do_sys_reboot+0x228/0x250
[   66.845148]  __arm64_sys_reboot+0x28/0x34
[   66.849264]  el0_svc_common.constprop.0+0x74/0x190
[   66.854187]  do_el0_svc+0x24/0x90
[   66.857595]  el0_svc+0x14/0x20
[   66.860739]  el0_sync_handler+0x1a4/0x1b0
[   66.864858]  el0_sync+0x174/0x180
[   66.868269] Code: 1ac020a0 2a000273 eb02007f 54ffff01 (f9400285)
[   66.874525] ---[ end trace 20dedb2a3229fec8 ]---

Fixes: 9d5cbf5fe46e ("drm/msm: add shutdown support for display platform_driver")
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

can: isotp: tx-path: zero initialize outgoing CAN frames

[ Upstream commit b5f020f82a8e41201c6ede20fa00389d6980b223 ]

Commit d4eb538e1f48 ("can: isotp: TX-path: ensure that CAN frame flags are
initialized") ensured the TX flags to be properly set for outgoing CAN
frames.

In fact the root cause of the issue results from a missing initialization
of outgoing CAN frames created by isotp. This is no problem on the CAN bus
as the CAN driver only picks the correctly defined content from the struct
can(fd)_frame. But when the outgoing frames are monitored (e.g. with
candump) we potentially leak some bytes in the unused content of
struct can(fd)_frame.

Fixes: e057dd3fc20f ("can: add ISO 15765-2:2016 transport protocol")
Cc: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://lore.kernel.org/r/20210319100619.10858-1-socketcan@hartkopp.net
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Fix umd memory leak in copy_process()

[ Upstream commit f60a85cad677c4f9bb4cadd764f1d106c38c7cf8 ]

The syzbot reported a memleak as follows:

BUG: memory leak
unreferenced object 0xffff888101b41d00 (size 120):
  comm "kworker/u4:0", pid 8, jiffies 4294944270 (age 12.780s)
  backtrace:
    [<ffffffff8125dc56>] alloc_pid+0x66/0x560
    [<ffffffff81226405>] copy_process+0x1465/0x25e0
    [<ffffffff81227943>] kernel_clone+0xf3/0x670
    [<ffffffff812281a1>] kernel_thread+0x61/0x80
    [<ffffffff81253464>] call_usermodehelper_exec_work
    [<ffffffff81253464>] call_usermodehelper_exec_work+0xc4/0x120
    [<ffffffff812591c9>] process_one_work+0x2c9/0x600
    [<ffffffff81259ab9>] worker_thread+0x59/0x5d0
    [<ffffffff812611c8>] kthread+0x178/0x1b0
    [<ffffffff8100227f>] ret_from_fork+0x1f/0x30

unreferenced object 0xffff888110ef5c00 (size 232):
  comm "kworker/u4:0", pid 8414, jiffies 4294944270 (age 12.780s)
  backtrace:
    [<ffffffff8154a0cf>] kmem_cache_zalloc
    [<ffffffff8154a0cf>] __alloc_file+0x1f/0xf0
    [<ffffffff8154a809>] alloc_empty_file+0x69/0x120
    [<ffffffff8154a8f3>] alloc_file+0x33/0x1b0
    [<ffffffff8154ab22>] alloc_file_pseudo+0xb2/0x140
    [<ffffffff81559218>] create_pipe_files+0x138/0x2e0
    [<ffffffff8126c793>] umd_setup+0x33/0x220
    [<ffffffff81253574>] call_usermodehelper_exec_async+0xb4/0x1b0
    [<ffffffff8100227f>] ret_from_fork+0x1f/0x30

After the UMD process exits, the pipe_to_umh/pipe_from_umh and
tgid need to be released.

Fixes: d71fa5c9763c ("bpf: Add kernel module with user mode driver that populates bpffs.")
Reported-by: syzbot+44908bb56d2bfe56b28e@syzkaller.appspotmail.com
Signed-off-by: Zqiang <qiang.zhang@windriver.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210317030915.2865-1-qiang.zhang@windriver.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

libbpf: Fix BTF dump of pointer-to-array-of-struct

[ Upstream commit 901ee1d750f29a335423eeb9463c3ca461ca18c2 ]

The vmlinux.h generated from BTF is invalid when building
drivers/phy/ti/phy-gmii-sel.c with clang:

vmlinux.h:61702:27: error: array type has incomplete element type ‘struct reg_field’
61702 |  const struct reg_field (*regfields)[3];
      |                           ^~~~~~~~~

bpftool generates a forward declaration for this struct regfield, which
compilers aren't happy about. Here's a simplified reproducer:

struct inner {
int val;
};
struct outer {
struct inner (*ptr_to_array)[2];
} A;

After build with clang -> bpftool btf dump c -> clang/gcc:
./def-clang.h:11:23: error: array has incomplete element type 'struct inner'
        struct inner (*ptr_to_array)[2];

Member ptr_to_array of struct outer is a pointer to an array of struct
inner. In the DWARF generated by clang, struct outer appears before
struct inner, so when converting BTF of struct outer into C, bpftool
issues a forward declaration to struct inner. With GCC the DWARF info is
reversed so struct inner gets fully defined.

That forward declaration is not sufficient when compilers handle an
array of the struct, even when it's only used through a pointer. Note
that we can trigger the same issue with an intermediate typedef:

struct inner {
        int val;
};
typedef struct inner inner2_t[2];
struct outer {
        inner2_t *ptr_to_array;
} A;

Becomes:

struct inner;
typedef struct inner inner2_t[2];

And causes:

./def-clang.h:10:30: error: array has incomplete element type 'struct inner'
typedef struct inner inner2_t[2];

To fix this, clear through_ptr whenever we encounter an intermediate
array, to make the inner struct part of a strong link and force full
declaration.

Fixes: 351131b51c7a ("libbpf: add btf_dump API for BTF-to-C conversion")
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210319112554.794552-2-jean-philippe@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests: forwarding: vxlan_bridge_1d: Fix vxlan ecn decapsulate value

[ Upstream commit 5aa3c334a449bab24519c4967f5ac2b3304c8dcf ]

The ECN bit defines ECT(1) = 1, ECT(0) = 2. So inner 0x02 + outer 0x01
should be inner ECT(0) + outer ECT(1). Based on the description of
__INET_ECN_decapsulate, the final decapsulate value should be
ECT(1). So fix the test expect value to 0x01.

Before the fix:
TEST: VXLAN: ECN decap: 01/02->0x02                                 [FAIL]
        Expected to capture 10 packets, got 0.

After the fix:
TEST: VXLAN: ECN decap: 01/02->0x01                                 [ OK ]

Fixes: a0b61f3d8ebf ("selftests: forwarding: vxlan_bridge_1d: Add an ECN decap test")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selinux: vsock: Set SID for socket returned by accept()

[ Upstream commit 1f935e8e72ec28dddb2dc0650b3b6626a293d94b ]

For AF_VSOCK, accept() currently returns sockets that are unlabelled.
Other socket families derive the child's SID from the SID of the parent
and the SID of the incoming packet. This is typically done as the
connected socket is placed in the queue that accept() removes from.

Reuse the existing 'security_sk_clone' hook to copy the SID from the
parent (server) socket to the child. There is no packet SID in this
case.

Fixes: d021c344051a ("VSOCK: Introduce VM Sockets")
Signed-off-by: David Brazdil <dbrazdil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: stmmac: dwmac-sun8i: Provide TX and RX fifo sizes

[ Upstream commit 014dfa26ce1c647af09bf506285ef67e0e3f0a6b ]

MTU cannot be changed on dwmac-sun8i. (ip link set eth0 mtu xxx returning EINVAL)
This is due to tx_fifo_size being 0, since this value is used to compute valid
MTU range.
Like dwmac-sunxi (with commit 806fd188ce2a ("net: stmmac: dwmac-sunxi: Provide TX and RX fifo sizes"))
dwmac-sun8i need to have tx and rx fifo sizes set.
I have used values from datasheets.
After this patch, setting a non-default MTU (like 1000) value works and network is still useable.

Tested-on: sun8i-h3-orangepi-pc
Tested-on: sun8i-r40-bananapi-m2-ultra
Tested-on: sun50i-a64-bananapi-m64
Tested-on: sun50i-h5-nanopi-neo-plus2
Tested-on: sun50i-h6-pine-h64
Fixes: 9f93ac8d408 ("net-next: stmmac: Add dwmac-sun8i")
Reported-by: Belisko Marek <marek.belisko@gmail.com>
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>