rcu: Introduce rcu_cpu_online()

[ Upstream commit 2be4686d866ad5896f2bb94d82fe892197aea9c7 ]

Export the RCU point of view as to when a CPU is considered offline
(ie: when does RCU consider that a CPU is sufficiently down in the
hotplug process to not feature any possible read side).

This will be used by RCU-tasks whose vision of an offline CPU should
reasonably match the one of RCU core.

Fixes: cff9b2332ab7 ("kernel/sched: Modify initial boot task idle setup")
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rcu: Break rcu_node_0 --> &rq->__lock order

[ Upstream commit 85d68222ddc5f4522e456d97d201166acb50f716 ]

Commit 851a723e45d1 ("sched: Always clear user_cpus_ptr in
do_set_cpus_allowed()") added a kfree() call to free any user
provided affinity mask, if present. It was changed later to use
kfree_rcu() in commit 9a5418bc48ba ("sched/core: Use kfree_rcu()
in do_set_cpus_allowed()") to avoid a circular locking dependency
problem.

It turns out that even kfree_rcu() isn't safe for avoiding
circular locking problem. As reported by kernel test robot,
the following circular locking dependency now exists:

  &rdp->nocb_lock --> rcu_node_0 --> &rq->__lock

Solve this by breaking the rcu_node_0 --> &rq->__lock chain by moving
the resched_cpu() out from under rcu_node lock.

[peterz: heavily borrowed from Waiman's Changelog]
[paulmck: applied Z qiang feedback]

Fixes: 851a723e45d1 ("sched: Always clear user_cpus_ptr in do_set_cpus_allowed()")
Reported-by: kernel test robot <oliver.sang@intel.com>
Acked-by: Waiman Long <longman@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/oe-lkp/202310302207.a25f1a30-oliver.sang@intel.com
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: thermal: Fix acpi_thermal_unregister_thermal_zone() cleanup

[ Upstream commit 4b27d5c420335dad7aea1aa6e799fe1d05c63b7e ]

The acpi_thermal_unregister_thermal_zone() is paired with
acpi_thermal_register_thermal_zone() so it should mirror it.  It should
clean up all the resources that the register function allocated and
leave the stuff that was allocated elsewhere.

Unfortunately, it doesn't call thermal_zone_device_disable().  Also it
calls kfree(tz->trip_table) when it shouldn't.  That was allocated in
acpi_thermal_add().  Putting the kfree() here leads to a double free
in the acpi_thermal_add() clean up function.

Likewise, the acpi_thermal_remove() should mirror acpi_thermal_add() so
it should have an explicit kfree(tz->trip_table) as well.

Fixes: ec23c1c462de ("ACPI: thermal: Use trip point table to register thermal zones")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/mlx5: Fix mkey cache WQ flush

[ Upstream commit a53e215f90079f617360439b1b6284820731e34c ]

The cited patch tries to ensure no pending works on the mkey cache
workqueue by disabling adding new works and call flush_workqueue().
But this workqueue also has delayed works which might still be pending
the delay time to be queued.

Add cancel_delayed_work() for the delayed works which waits to be queued
and then the flush_workqueue() will flush all works which are already
queued and running.

Fixes: 374012b00457 ("RDMA/mlx5: Fix mkey cache possible deadlock on cleanup")
Link: https://lore.kernel.org/r/b8722f14e7ed81452f791764a26d2ed4cfa11478.1698256179.git.leon@kernel.org
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: si521xx: Increase stack based print buffer size in probe

[ Upstream commit 7e52b1164a474dc7b90f68fbb40e35ccd7f7e2e2 ]

Increase the size of temporary print buffer on stack to fix the
following warnings reported by LKP.

Since all the input parameters of snprintf() are under control
of this driver, it is not possible to trigger and overflow here,
but since the print buffer is on stack and discarded once driver
probe() finishes, it is not an issue to increase it by 10 bytes
and fix the warning in the process. Make it so.

"
   drivers/clk/clk-si521xx.c: In function 'si521xx_probe':
>> drivers/clk/clk-si521xx.c:318:26: warning: '%d' directive output may be truncated writing between 1 and 10 bytes into a region of size 2 [-Wformat-truncation=]
      snprintf(name, 6, "DIFF%d", i);
                             ^~
   drivers/clk/clk-si521xx.c:318:21: note: directive argument in the range [0, 2147483647]
      snprintf(name, 6, "DIFF%d", i);
                        ^~~~~~~~
   drivers/clk/clk-si521xx.c:318:3: note: 'snprintf' output between 6 and 15 bytes into a destination of size 6
      snprintf(name, 6, "DIFF%d", i);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"

Fixes: edc12763a3a2 ("clk: si521xx: Clock driver for Skyworks Si521xx I2C PCIe clock generators")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202310260412.AGASjFN4-lkp@intel.com/
Signed-off-by: Marek Vasut <marex@denx.de>
Link: https://lore.kernel.org/r/20231027085840.30098-1-marex@denx.de
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vfio/mtty: Overhaul mtty interrupt handling

[ Upstream commit 293fbc28818135743f54d46c418ede3e4a20a742 ]

The mtty driver does not currently conform to the vfio SET_IRQS uAPI.
For example, it claims to support mask and unmask of INTx, but actually
does nothing.  It claims to support AUTOMASK for INTx, but doesn't.  It
fails to teardown eventfds under the full semantics specified by the
SET_IRQS ioctl.  It also fails to teardown eventfds when the device is
closed, leading to memory leaks.  It claims to support the request IRQ,
but doesn't.

Fix all these.

A side effect of this is that QEMU will now report a warning:

vfio <uuid>: Failed to set up UNMASK eventfd signaling for interrupt \
INTX-0: VFIO_DEVICE_SET_IRQS failure: Inappropriate ioctl for device

The fact is that the unmask eventfd was never supported but quietly
failed.  mtty never honored the AUTOMASK behavior, therefore there
was nothing to unmask.  QEMU is verbose about the failure, but
properly falls back to userspace unmasking.

Fixes: 9d1a546c53b4 ("docs: Sample driver to demonstrate how to use Mediated device framework.")
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Link: https://lore.kernel.org/r/20231016224736.2575718-2-alex.williamson@redhat.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: hisilicon/qm - fix EQ/AEQ interrupt issue

[ Upstream commit 5acab6eb592387191c1bb745ba9b815e1e076db5 ]

During hisilicon accelerator live migration operation. In order to
prevent the problem of EQ/AEQ interrupt loss. Migration driver will
trigger an EQ/AEQ doorbell at the end of the migration.

This operation may cause double interruption of EQ/AEQ events.
To ensure that the EQ/AEQ interrupt processing function is normal.
The interrupt handling functionality of EQ/AEQ needs to be updated.
Used to handle repeated interrupts event.

Fixes: b0eed085903e ("hisi_acc_vfio_pci: Add support for VFIO live migration")
Signed-off-by: Longfang Liu <liulongfang@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: qat - fix double free during reset

[ Upstream commit 01aed663e6c421aeafc9c330bda630976b50a764 ]

There is no need to free the reset_data structure if the recovery is
unsuccessful and the reset is synchronous. The function
adf_dev_aer_schedule_reset() handles the cleanup properly. Only
asynchronous resets require such structure to be freed inside the reset
worker.

Fixes: d8cba25d2c68 ("crypto: qat - Intel(R) QAT driver framework")
Signed-off-by: Svyatoslav Pankratov <svyatoslav.pankratov@intel.com>
Signed-off-by: Giovanni Cabiddu <giovanni.cabiddu@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: xts - use 'spawn' for underlying single-block cipher

[ Upstream commit bb40d32689d73c46de39a0529d551f523f21dc9b ]

Since commit adad556efcdd ("crypto: api - Fix built-in testing
dependency failures"), the following warning appears when booting an
x86_64 kernel that is configured with
CONFIG_CRYPTO_MANAGER_EXTRA_TESTS=y and CONFIG_CRYPTO_AES_NI_INTEL=y,
even when CONFIG_CRYPTO_XTS=y and CONFIG_CRYPTO_AES=y:

    alg: skcipher: skipping comparison tests for xts-aes-aesni because xts(ecb(aes-generic)) is unavailable

This is caused by an issue in the xts template where it allocates an
"aes" single-block cipher without declaring a dependency on it via the
crypto_spawn mechanism.  This issue was exposed by the above commit
because it reversed the order that the algorithms are tested in.

Specifically, when "xts(ecb(aes-generic))" is instantiated and tested
during the comparison tests for "xts-aes-aesni", the "xts" template
allocates an "aes" crypto_cipher for encrypting tweaks.  This resolves
to "aes-aesni".  (Getting "aes-aesni" instead of "aes-generic" here is a
bit weird, but it's apparently intended.)  Due to the above-mentioned
commit, the testing of "aes-aesni", and the finalization of its
registration, now happens at this point instead of before.  At the end
of that, crypto_remove_spawns() unregisters all algorithm instances that
depend on a lower-priority "aes" implementation such as "aes-generic"
but that do not depend on "aes-aesni".  However, because "xts" does not
use the crypto_spawn mechanism for its "aes", its dependency on
"aes-aesni" is not recognized by crypto_remove_spawns().  Thus,
crypto_remove_spawns() unexpectedly unregisters "xts(ecb(aes-generic))".

Fix this issue by making the "xts" template use the crypto_spawn
mechanism for its "aes" dependency, like what other templates do.

Note, this fix could be applied as far back as commit f1c131b45410
("crypto: xts - Convert to skcipher").  However, the issue only got
exposed by the much more recent changes to how the crypto API runs the
self-tests, so there should be no need to backport this to very old
kernels.  Also, an alternative fix would be to flip the list iteration
order in crypto_start_tests() to restore the original testing order.
I'm thinking we should do that too, since the original order seems more
natural, but it shouldn't be relied on for correctness.

Fixes: adad556efcdd ("crypto: api - Fix built-in testing dependency failures")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpftool: Align output skeleton ELF code

[ Upstream commit 23671f4dfd10b48b4a2fee4768886f0d8ec55b7e ]

libbpf accesses the ELF data requiring at least 8 byte alignment,
however, the data is generated into a C string that doesn't guarantee
alignment. Fix this by assigning to an aligned char array. Use sizeof
on the array, less one for the \0 terminator, rather than generating a
constant.

Fixes: a6cc6b34b93e ("bpftool: Provide a helper method for accessing skeleton's embedded ELF data")
Signed-off-by: Ian Rogers <irogers@google.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Acked-by: Quentin Monnet <quentin@isovalent.com>
Link: https://lore.kernel.org/bpf/20231007044439.25171-1-irogers@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpftool: Fix -Wcast-qual warning

[ Upstream commit ebc8484d0e6da9e6c9e8cfa1f40bf94e9c6fc512 ]

This cast was made by purpose for older libbpf where the
bpf_object_skeleton field is void * instead of const void *
to eliminate a warning (as i understand
-Wincompatible-pointer-types-discards-qualifiers) but this
cast introduces another warning (-Wcast-qual) for libbpf
where data field is const void *

It makes sense for bpftool to be in sync with libbpf from
kernel sources

Signed-off-by: Denys Zagorui <dzagorui@cisco.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Quentin Monnet <quentin@isovalent.com>
Link: https://lore.kernel.org/bpf/20230907090210.968612-1-dzagorui@cisco.com
Stable-dep-of: 23671f4dfd10 ("bpftool: Align output skeleton ELF code")
Signed-off-by: Sasha Levin <sashal@kernel.org>

tcp: derive delack_max from rto_min

[ Upstream commit bbf80d713fe75cfbecda26e7c03a9a8d22af2f4f ]

While BPF allows to set icsk->->icsk_delack_max
and/or icsk->icsk_rto_min, we have an ip route
attribute (RTAX_RTO_MIN) to be able to tune rto_min,
but nothing to consequently adjust max delayed ack,
which vary from 40ms to 200 ms (TCP_DELACK_{MIN|MAX}).

This makes RTAX_RTO_MIN of almost no practical use,
unless customers are in big trouble.

Modern days datacenter communications want to set
rto_min to ~5 ms, and the max delayed ack one jiffie
smaller to avoid spurious retransmits.

After this patch, an "rto_min 5" route attribute will
effectively lower max delayed ack timers to 4 ms.

Note in the following ss output, "rto:6 ... ato:4"

$ ss -temoi dst XXXXXX
State Recv-Q Send-Q           Local Address:Port       Peer Address:Port  Process
ESTAB 0      0        [2002:a05:6608:295::]:52950   [2002:a05:6608:297::]:41597
     ino:255134 sk:1001 <->
         skmem:(r0,rb1707063,t872,tb262144,f0,w0,o0,bl0,d0) ts sack
 cubic wscale:8,8 rto:6 rtt:0.02/0.002 ato:4 mss:4096 pmtu:4500
 rcvmss:536 advmss:4096 cwnd:10 bytes_sent:54823160 bytes_acked:54823121
 bytes_received:54823120 segs_out:1370582 segs_in:1370580
 data_segs_out:1370579 data_segs_in:1370578 send 16.4Gbps
 pacing_rate 32.6Gbps delivery_rate 1.72Gbps delivered:1370579
 busy:26920ms unacked:1 rcv_rtt:34.615 rcv_space:65920
 rcv_ssthresh:65535 minrtt:0.015 snd_wnd:65536

While we could argue this patch fixes a bug with RTAX_RTO_MIN,
I do not add a Fixes: tag, so that we can soak it a bit before
asking backports to stable branches.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: qcom: camss: Fix genpd cleanup

[ Upstream commit f69791c39745e64621216fe8919cb73c0065002b ]

Right now we never release the power-domains properly on the error path.
Add a routine to be reused for this purpose and appropriate jumps in
probe() to run that routine where necessary.

Fixes: 2f6f8af67203 ("media: camss: Refactor VFE power domain toggling")
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: qcom: camss: Fix V4L2 async notifier error path

[ Upstream commit b278080a89f452063915beda0ade6b3ed5ee4271 ]

Previously the jump label err_cleanup was used higher in the probe()
function to release the async notifier however the async notifier
registration was moved later in the code rendering the previous four jumps
redundant.

Rename the label from err_cleanup to err_v4l2_device_unregister to capture
what the jump does.

Fixes: 51397a4ec75d ("media: qcom: Initialise V4L2 async notifier later")
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
[hverkuil: fix old name in commit log: err_v4l2_device_register -> err_v4l2_device_unregister]
Signed-off-by: Sasha Levin <sashal@kernel.org>

xsk: add multi-buffer support for sockets sharing umem

[ Upstream commit d609f3d228a8efe991f44f11f24146e2a5209755 ]

Userspace applications indicate their multi-buffer capability to xsk
using XSK_USE_SG socket bind flag. For sockets using shared umem the
bind flag may contain XSK_USE_SG only for the first socket. For any
subsequent socket the only option supported is XDP_SHARED_UMEM.

Add option XDP_UMEM_SG_FLAG in umem config flags to store the
multi-buffer handling capability when indicated by XSK_USE_SG option in
bing flag by the first socket. Use this to derive multi-buffer capability
for subsequent sockets in xsk core.

Signed-off-by: Tirthendu Sarkar <tirthendu.sarkar@intel.com>
Fixes: 81470b5c3c66 ("xsk: introduce XSK_USE_SG bind flag for xsk socket")
Acked-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Link: https://lore.kernel.org/r/20230907035032.2627879-1-tirthendu.sarkar@intel.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mm/memory-failure: pass the folio and the page to collect_procs()

[ Upstream commit 376907f3a0b34a17e80417825f8cc1c40fcba81b ]

Patch series "Three memory-failure fixes".

I've been looking at the memory-failure code and I believe I have found
three bugs that need fixing -- one going all the way back to 2010!  I'll
have more patches later to use folios more extensively but didn't want
these bugfixes to get caught up in that.

This patch (of 3):

Both collect_procs_anon() and collect_procs_file() iterate over the VMA
interval trees looking for a single pgoff, so it is wrong to look for the
pgoff of the head page as is currently done.  However, it is also wrong to
look at page->mapping of the precise page as this is invalid for tail
pages.  Clear up the confusion by passing both the folio and the precise
page to collect_procs().

Link: https://lkml.kernel.org/r/20231218135837.3310403-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20231218135837.3310403-2-willy@infradead.org
Fixes: 415c64c1453a ("mm/memory-failure: split thp earlier in memory error handling")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mm: convert DAX lock/unlock page to lock/unlock folio

[ Upstream commit 91e79d22be75fec88ae58d274a7c9e49d6215099 ]

The one caller of DAX lock/unlock page already calls compound_head(), so
use page_folio() instead, then use a folio throughout the DAX code to
remove uses of page->mapping and page->index.

[jane.chu@oracle.com: add comment to mf_generic_kill_procss(), simplify mf_generic_kill_procs:folio initialization]
Link: https://lkml.kernel.org/r/20230908222336.186313-1-jane.chu@oracle.com
Link: https://lkml.kernel.org/r/20230822231314.349200-1-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Jane Chu <jane.chu@oracle.com>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Jane Chu <jane.chu@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Stable-dep-of: 376907f3a0b3 ("mm/memory-failure: pass the folio and the page to collect_procs()")
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: Implement missing SO_TIMESTAMPING_NEW cmsg support

[ Upstream commit 382a32018b74f407008615e0e831d05ed28e81cd ]

Commit 9718475e6908 ("socket: Add SO_TIMESTAMPING_NEW") added the new
socket option SO_TIMESTAMPING_NEW. However, it was never implemented in
__sock_cmsg_send thus breaking SO_TIMESTAMPING cmsg for platforms using
SO_TIMESTAMPING_NEW.

Fixes: 9718475e6908 ("socket: Add SO_TIMESTAMPING_NEW")
Link: https://lore.kernel.org/netdev/6a7281bf-bc4a-4f75-bb88-7011908ae471@app.fastmail.com/
Signed-off-by: Thomas Lange <thomas@corelatus.se>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/r/20240104085744.49164-1-thomas@corelatus.se
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bnxt_en: Remove mis-applied code from bnxt_cfg_ntp_filters()

[ Upstream commit e009b2efb7a8850498796b360043ac25c8d3d28f ]

The 2 lines to check for the BNXT_HWRM_PF_UNLOAD_SP_EVENT bit was
mis-applied to bnxt_cfg_ntp_filters() and should have been applied to
bnxt_sp_task().

Fixes: 19241368443f ("bnxt_en: Send PF driver unload notification to all VFs.")
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: ravb: Wait for operating mode to be applied

[ Upstream commit 9039cd4c61635b2d541009a7cd5e2cc052402f28 ]

CSR.OPS bits specify the current operating mode and (according to
documentation) they are updated by HW when the operating mode change
request is processed. To comply with this check CSR.OPS before proceeding.

Commit introduces ravb_set_opmode() that does all the necessities for
setting the operating mode (set CCC.OPC (and CCC.GAC, CCC.CSEL, if any) and
wait for CSR.OPS) and call it where needed. This should comply with all the
HW manuals requirements as different manual variants specify that different
modes need to be checked in CSR.OPS when setting CCC.OPC.

If gPTP active in config mode is supported and it needs to be enabled, the
CCC.GAC and CCC.CSEL needs to be configured along with CCC.OPC in the same
write access. For this, ravb_set_opmode() allows passing GAC and CSEL as
part of opmode and the function updates accordingly CCC register.

Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
Signed-off-by: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com>
Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

asix: Add check for usbnet_get_endpoints

[ Upstream commit eaac6a2d26b65511e164772bec6918fcbc61938e ]

Add check for usbnet_get_endpoints() and return the error if it fails
in order to transfer the error.

Fixes: 16626b0cc3d5 ("asix: Add a new driver for the AX88172A")
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

octeontx2-af: Re-enable MAC TX in otx2_stop processing

[ Upstream commit 818ed8933bd17bc91a9fa8b94a898189c546fc1a ]

During QoS scheduling testing with multiple strict priority flows, the
netdev tx watchdog timeout routine is invoked when a low priority QoS
queue doesn't get a chance to transmit the packets because other high
priority flows are completely subscribing the transmit link. The netdev
tx watchdog timeout routine will stop MAC RX and TX functionality in
otx2_stop() routine before cleanup of HW TX queues which results in SMQ
flush errors because the packets belonging to low priority queues will
never gets flushed since MAC TX is disabled. This patch fixes the issue
by re-enabling MAC TX to ensure the packets in HW pipeline gets flushed
properly.

Fixes: a7faa68b4e7f ("octeontx2-af: Start/Stop traffic in CGX along with NPC")
Signed-off-by: Naveen Mamindlapalli <naveenm@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

octeontx2-af: Always configure NIX TX link credits based on max frame size

[ Upstream commit a0d9528f6daf7fe8de217fa80a94d2989d2a57a7 ]

Currently the NIX TX link credits are initialized based on the max frame
size that can be transmitted on a link but when the MTU is changed, the
NIX TX link credits are reprogrammed by the SW based on the new MTU value.
Since SMQ max packet length is programmed to max frame size by default,
there is a chance that NIX TX may stall while sending a max frame sized
packet on the link with insufficient credits to send the packet all at
once. This patch avoids stall issue by not changing the link credits
dynamically when the MTU is changed.

Fixes: 1c74b89171c3 ("octeontx2-af: Wait for TX link idle for credits change")
Signed-off-by: Naveen Mamindlapalli <naveenm@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: Nithin Kumar Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/smc: fix invalid link access in dumping SMC-R connections

[ Upstream commit 9dbe086c69b8902c85cece394760ac212e9e4ccc ]

A crash was found when dumping SMC-R connections. It can be reproduced
by following steps:

- environment: two RNICs on both sides.
- run SMC-R between two sides, now a SMC_LGR_SYMMETRIC type link group
  will be created.
- set the first RNIC down on either side and link group will turn to
  SMC_LGR_ASYMMETRIC_LOCAL then.
- run 'smcss -R' and the crash will be triggered.

 BUG: kernel NULL pointer dereference, address: 0000000000000010
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 8000000101fdd067 P4D 8000000101fdd067 PUD 10ce46067 PMD 0
 Oops: 0000 [#1] PREEMPT SMP PTI
 CPU: 3 PID: 1810 Comm: smcss Kdump: loaded Tainted: G W   E      6.7.0-rc6+ #51
 RIP: 0010:__smc_diag_dump.constprop.0+0x36e/0x620 [smc_diag]
 Call Trace:
  <TASK>
  ? __die+0x24/0x70
  ? page_fault_oops+0x66/0x150
  ? exc_page_fault+0x69/0x140
  ? asm_exc_page_fault+0x26/0x30
  ? __smc_diag_dump.constprop.0+0x36e/0x620 [smc_diag]
  smc_diag_dump_proto+0xd0/0xf0 [smc_diag]
  smc_diag_dump+0x26/0x60 [smc_diag]
  netlink_dump+0x19f/0x320
  __netlink_dump_start+0x1dc/0x300
  smc_diag_handler_dump+0x6a/0x80 [smc_diag]
  ? __pfx_smc_diag_dump+0x10/0x10 [smc_diag]
  sock_diag_rcv_msg+0x121/0x140
  ? __pfx_sock_diag_rcv_msg+0x10/0x10
  netlink_rcv_skb+0x5a/0x110
  sock_diag_rcv+0x28/0x40
  netlink_unicast+0x22a/0x330
  netlink_sendmsg+0x240/0x4a0
  __sock_sendmsg+0xb0/0xc0
  ____sys_sendmsg+0x24e/0x300
  ? copy_msghdr_from_user+0x62/0x80
  ___sys_sendmsg+0x7c/0xd0
  ? __do_fault+0x34/0x1a0
  ? do_read_fault+0x5f/0x100
  ? do_fault+0xb0/0x110
  __sys_sendmsg+0x4d/0x80
  do_syscall_64+0x45/0xf0
  entry_SYSCALL_64_after_hwframe+0x6e/0x76

When the first RNIC is set down, the lgr->lnk[0] will be cleared and an
asymmetric link will be allocated in lgr->link[SMC_LINKS_PER_LGR_MAX - 1]
by smc_llc_alloc_alt_link(). Then when we try to dump SMC-R connections
in __smc_diag_dump(), the invalid lgr->lnk[0] will be accessed, resulting
in this issue. So fix it by accessing the right link.

Fixes: f16a7dd5cf27 ("smc: netlink interface for SMC sockets")
Reported-by: henaumars <henaumars@sina.com>
Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=7616
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Reviewed-by: Tony Lu <tonylu@linux.alibaba.com>
Link: https://lore.kernel.org/r/1703662835-53416-1-git-send-email-guwen@linux.alibaba.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/qla3xxx: fix potential memleak in ql_alloc_buffer_queues

[ Upstream commit 89f45c30172c80e55c887f32f1af8e184124577b ]

When dma_alloc_coherent() fails, we should free qdev->lrg_buf
to prevent potential memleak.

Fixes: 1357bfcf7106 ("qla3xxx: Dynamically size the rx buffer queue based on the MTU.")
Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
Link: https://lore.kernel.org/r/20231227070227.10527-1-dinghao.liu@zju.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

virtio_net: fix missing dma unmap for resize

[ Upstream commit 2311e06b9bf3d44e15f9175af177a782806f688f ]

For rq, we have three cases getting buffers from virtio core:

1. virtqueue_get_buf{,_ctx}
2. virtqueue_detach_unused_buf
3. callback for virtqueue_resize

But in commit 295525e29a5b("virtio_net: merge dma operations when
filling mergeable buffers"), I missed the dma unmap for the #3 case.

That will leak some memory, because I did not release the pages referred
by the unused buffers.

If we do such script, we will make the system OOM.

    while true
    do
            ethtool -G ens4 rx 128
            ethtool -G ens4 rx 256
            free -m
    done

Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Link: https://lore.kernel.org/r/20231226094333.47740-1-xuanzhuo@linux.alibaba.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

virtio_net: avoid data-races on dev->stats fields

[ Upstream commit d12a26b74fb77434b73fe39022266c4b00907219 ]

Use DEV_STATS_INC() and DEV_STATS_READ() which provide
atomicity on paths that can be used concurrently.

Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable-dep-of: 2311e06b9bf3 ("virtio_net: fix missing dma unmap for resize")
Signed-off-by: Sasha Levin <sashal@kernel.org>

apparmor: Fix move_mount mediation by detecting if source is detached

[ Upstream commit 8026e40608b4d552216d2a818ca7080a4264bb44 ]

Prevent move_mount from applying the attach_disconnected flag
to move_mount(). This prevents detached mounts from appearing
as / when applying mount mediation, which is not only incorrect
but could result in bad policy being generated.

Basic mount rules like
  allow mount,
  allow mount options=(move) -> /target/,

will allow detached mounts, allowing older policy to continue
to function. New policy gains the ability to specify `detached` as
a source option
  allow mount detached -> /target/,

In addition make sure support of move_mount is advertised as
a feature to userspace so that applications that generate policy
can respond to the addition.

Note: this fixes mediation of move_mount when a detached mount is used,
      it does not fix the broader regression of apparmor mediation of
      mounts under the new mount api.

Link: https://lore.kernel.org/all/68c166b8-5b4d-4612-8042-1dee3334385b@leemhuis.info/T/#mb35fdde37f999f08f0b02d58dc1bf4e6b65b8da2
Fixes: 157a3537d6bc ("apparmor: Fix regression in mount mediation")
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

igc: Fix hicredit calculation

[ Upstream commit 947dfc8138dfaeb6e966e2d661de89eb203e3064 ]

According to the Intel Software Manual for I225, Section 7.5.2.7,
hicredit should be multiplied by the constant link-rate value, 0x7736.

Currently, the old constant link-rate value, 0x7735, from the boards
supported on igb are being used, most likely due to a copy'n'paste, as
the rest of the logic is the same for both drivers.

Update hicredit accordingly.

Fixes: 1ab011b0bf07 ("igc: Add support for CBS offloading")
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de>
Signed-off-by: Rodrigo Cataldo <rodrigo.cadore@l-acoustics.com>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i40e: Restore VF MSI-X state during PCI reset

[ Upstream commit 371e576ff3e8580d91d49026e5d5faebf5565558 ]

During a PCI FLR the MSI-X Enable flag in the VF PCI MSI-X capability
register will be cleared. This can lead to issues when a VF is
assigned to a VM because in these cases the VF driver receives no
indication of the PF PCI error/reset and additionally it is incapable
of restoring the cleared flag in the hypervisor configuration space
without fully reinitializing the driver interrupt functionality.

Since the VF driver is unable to easily resolve this condition on its own,
restore the VF MSI-X flag during the PF PCI reset handling.

Fixes: 19b7960b2da1 ("i40e: implement split PCI error reset handler")
Co-developed-by: Karen Ostrowska <karen.ostrowska@intel.com>
Signed-off-by: Karen Ostrowska <karen.ostrowska@intel.com>
Co-developed-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Andrii Staikov <andrii.staikov@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: meson: g12a-tohdmitx: Fix event generation for S/PDIF mux

[ Upstream commit b036d8ef3120b996751495ce25994eea58032a98 ]

When a control changes value the return value from _put() should be 1 so
we get events generated to userspace notifying applications of the change.
While the I2S mux gets this right the S/PDIF mux does not, fix the return
value.

Fixes: c8609f3870f7 ("ASoC: meson: add g12a tohdmitx control")
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20240103-meson-enum-val-v1-4-424af7a8fb91@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: meson: g12a-toacodec: Fix event generation

[ Upstream commit 172c88244b5f2d3375403ebb504d407be0fded59 ]

When a control changes value the return value from _put() should be 1 so
we get events generated to userspace notifying applications of the change.
We are checking if there has been a change and exiting early if not but we
are not providing the correct return value in the latter case, fix this.

Fixes: af2618a2eee8 ("ASoC: meson: g12a: add internal DAC glue driver")
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20240103-meson-enum-val-v1-3-424af7a8fb91@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: meson: g12a-tohdmitx: Validate written enum values

[ Upstream commit 1e001206804be3f3d21f4a1cf16e5d059d75643f ]

When writing to an enum we need to verify that the value written is valid
for the enumeration, the helper function snd_soc_item_enum_to_val() doesn't
do it since it needs to return an unsigned (and in any case we'd need to
check the return value).

Fixes: c8609f3870f7 ("ASoC: meson: add g12a tohdmitx control")
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20240103-meson-enum-val-v1-2-424af7a8fb91@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: meson: g12a-toacodec: Validate written enum values

[ Upstream commit 3150b70e944ead909260285dfb5707d0bedcf87b ]

When writing to an enum we need to verify that the value written is valid
for the enumeration, the helper function snd_soc_item_enum_to_val() doesn't
do it since it needs to return an unsigned (and in any case we'd need to
check the return value).

Fixes: af2618a2eee8 ("ASoC: meson: g12a: add internal DAC glue driver")
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20240103-meson-enum-val-v1-1-424af7a8fb91@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i40e: fix use-after-free in i40e_aqc_add_filters()

[ Upstream commit 6a15584e99db8918b60e507539c7446375dcf366 ]

Commit 3116f59c12bd ("i40e: fix use-after-free in
i40e_sync_filters_subtask()") avoided use-after-free issues,
by increasing refcount during update the VSI filter list to
the HW. However, it missed the unicast situation.

When deleting an unicast FDB entry, the i40e driver will release
the mac_filter, and i40e_service_task will concurrently request
firmware to add the mac_filter, which will lead to the following
use-after-free issue.

Fix again for both netdev->uc and netdev->mc.

BUG: KASAN: use-after-free in i40e_aqc_add_filters+0x55c/0x5b0 [i40e]
Read of size 2 at addr ffff888eb3452d60 by task kworker/8:7/6379

CPU: 8 PID: 6379 Comm: kworker/8:7 Kdump: loaded Tainted: G
Workqueue: i40e i40e_service_task [i40e]
Call Trace:
 dump_stack+0x71/0xab
 print_address_description+0x6b/0x290
 kasan_report+0x14a/0x2b0
 i40e_aqc_add_filters+0x55c/0x5b0 [i40e]
 i40e_sync_vsi_filters+0x1676/0x39c0 [i40e]
 i40e_service_task+0x1397/0x2bb0 [i40e]
 process_one_work+0x56a/0x11f0
 worker_thread+0x8f/0xf40
 kthread+0x2a0/0x390
 ret_from_fork+0x1f/0x40

Allocated by task 21948:
 kasan_kmalloc+0xa6/0xd0
 kmem_cache_alloc_trace+0xdb/0x1c0
 i40e_add_filter+0x11e/0x520 [i40e]
 i40e_addr_sync+0x37/0x60 [i40e]
 __hw_addr_sync_dev+0x1f5/0x2f0
 i40e_set_rx_mode+0x61/0x1e0 [i40e]
 dev_uc_add_excl+0x137/0x190
 i40e_ndo_fdb_add+0x161/0x260 [i40e]
 rtnl_fdb_add+0x567/0x950
 rtnetlink_rcv_msg+0x5db/0x880
 netlink_rcv_skb+0x254/0x380
 netlink_unicast+0x454/0x610
 netlink_sendmsg+0x747/0xb00
 sock_sendmsg+0xe2/0x120
 __sys_sendto+0x1ae/0x290
 __x64_sys_sendto+0xdd/0x1b0
 do_syscall_64+0xa0/0x370
 entry_SYSCALL_64_after_hwframe+0x65/0xca

Freed by task 21948:
 __kasan_slab_free+0x137/0x190
 kfree+0x8b/0x1b0
 __i40e_del_filter+0x116/0x1e0 [i40e]
 i40e_del_mac_filter+0x16c/0x300 [i40e]
 i40e_addr_unsync+0x134/0x1b0 [i40e]
 __hw_addr_sync_dev+0xff/0x2f0
 i40e_set_rx_mode+0x61/0x1e0 [i40e]
 dev_uc_del+0x77/0x90
 rtnl_fdb_del+0x6a5/0x860
 rtnetlink_rcv_msg+0x5db/0x880
 netlink_rcv_skb+0x254/0x380
 netlink_unicast+0x454/0x610
 netlink_sendmsg+0x747/0xb00
 sock_sendmsg+0xe2/0x120
 __sys_sendto+0x1ae/0x290
 __x64_sys_sendto+0xdd/0x1b0
 do_syscall_64+0xa0/0x370
 entry_SYSCALL_64_after_hwframe+0x65/0xca

Fixes: 3116f59c12bd ("i40e: fix use-after-free in i40e_sync_filters_subtask()")
Fixes: 41c445ff0f48 ("i40e: main driver core")
Signed-off-by: Ke Xiao <xiaoke@sangfor.com.cn>
Signed-off-by: Ding Hui <dinghui@sangfor.com.cn>
Cc: Di Zhu <zhudi2@huawei.com>
Reviewed-by: Jan Sokolowski <jan.sokolowski@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: Save and restore msg_namelen in sock_sendmsg

[ Upstream commit 01b2885d9415152bcb12ff1f7788f500a74ea0ed ]

Commit 86a7e0b69bd5 ("net: prevent rewrite of msg_name in
sock_sendmsg()") made sock_sendmsg save the incoming msg_name pointer
and restore it before returning, to insulate the caller against
msg_name being changed by the called code.  If the address length
was also changed however, we may return with an inconsistent structure
where the length doesn't match the address, and attempts to reuse it may
lead to lost packets.

For example, a kernel that doesn't have commit 1c5950fc6fe9 ("udp6: fix
potential access to stale information") will replace a v4 mapped address
with its ipv4 equivalent, and shorten namelen accordingly from 28 to 16.
If the caller attempts to reuse the resulting msg structure, it will have
the original ipv6 (v4 mapped) address but an incorrect v4 length.

Fixes: 86a7e0b69bd5 ("net: prevent rewrite of msg_name in sock_sendmsg()")
Signed-off-by: Marc Dionne <marc.dionne@auristor.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nft_immediate: drop chain reference counter on error

[ Upstream commit b29be0ca8e816119ccdf95cc7d7c7be9bde005f1 ]

In the init path, nft_data_init() bumps the chain reference counter,
decrement it on error by following the error path which calls
nft_data_release() to restore it.

Fixes: 4bedf9eee016 ("netfilter: nf_tables: fix chain binding transaction logic")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_nat: fix action not being set for all ct states

[ Upstream commit e6345d2824a3f58aab82428d11645e0da861ac13 ]

This fixes openvswitch's handling of nat packets in the related state.

In nf_ct_nat_execute(), which is called from nf_ct_nat(), ICMP/ICMPv6
packets in the IP_CT_RELATED or IP_CT_RELATED_REPLY state, which have
not been dropped, will follow the goto, however the placement of the
goto label means that updating the action bit field will be bypassed.

This causes ovs_nat_update_key() to not be called from ovs_ct_nat()
which means the openvswitch match key for the ICMP/ICMPv6 packet is not
updated and the pre-nat value will be retained for the key, which will
result in the wrong openflow rule being matched for that packet.

Move the goto label above where the action bit field is being set so
that it is updated in all cases where the packet is accepted.

Fixes: ebddb1404900 ("net: move the nat function to nf_nat_ovs for ovs and tc")
Signed-off-by: Brad Cowie <brad@faucet.nz>
Reviewed-by: Simon Horman <horms@kernel.org>
Acked-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Aaron Conole <aconole@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: bcmgenet: Fix FCS generation for fragmented skbuffs

[ Upstream commit e584f2ff1e6cc9b1d99e8a6b0f3415940d1b3eb3 ]

The flag DMA_TX_APPEND_CRC was only written to the first DMA descriptor
in the TX path, where each descriptor corresponds to a single skbuff
fragment (or the skbuff head). This led to packets with no FCS appearing
on the wire if the kernel allocated the packet in fragments, which would
always happen when using PACKET_MMAP/TPACKET (cf. tpacket_fill_skb() in
net/af_packet.c).

Fixes: 1c1008c793fa ("net: bcmgenet: add main driver file")
Signed-off-by: Adrian Cinal <adriancinal1@gmail.com>
Acked-by: Doug Berger <opendmb@gmail.com>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Link: https://lore.kernel.org/r/20231228135638.1339245-1-adriancinal1@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

sfc: fix a double-free bug in efx_probe_filters

[ Upstream commit d5a306aedba34e640b11d7026dbbafb78ee3a5f6 ]

In efx_probe_filters, the channel->rps_flow_id is freed in a
efx_for_each_channel marco  when success equals to 0.
However, after the following call chain:

ef100_net_open
  |-> efx_probe_filters
  |-> ef100_net_stop
        |-> efx_remove_filters

The channel->rps_flow_id is freed again in the efx_for_each_channel of
efx_remove_filters, triggering a double-free bug.

Fixes: a9dc3d5612ce ("sfc_ef100: RX filter table management and related gubbins")
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Edward Cree <ecree.xilinx@gmail.com>
Signed-off-by: Zhipeng Lu <alexious@zju.edu.cn>
Link: https://lore.kernel.org/r/20231225112915.3544581-1-alexious@zju.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: sun9i: smp: Fix array-index-out-of-bounds read in sunxi_mc_smp_init

[ Upstream commit 72ad3b772b6d393701df58ba1359b0bb346a19ed ]

Running a multi-arch kernel (multi_v7_defconfig) on a Raspberry Pi 3B+
with enabled CONFIG_UBSAN triggers the following warning:

 UBSAN: array-index-out-of-bounds in arch/arm/mach-sunxi/mc_smp.c:810:29
 index 2 is out of range for type 'sunxi_mc_smp_data [2]'
 CPU: 0 PID: 1 Comm: swapper/0 Not tainted 6.7.0-rc6-00248-g5254c0cbc92d
 Hardware name: BCM2835
  unwind_backtrace from show_stack+0x10/0x14
  show_stack from dump_stack_lvl+0x40/0x4c
  dump_stack_lvl from ubsan_epilogue+0x8/0x34
  ubsan_epilogue from __ubsan_handle_out_of_bounds+0x78/0x80
  __ubsan_handle_out_of_bounds from sunxi_mc_smp_init+0xe4/0x4cc
  sunxi_mc_smp_init from do_one_initcall+0xa0/0x2fc
  do_one_initcall from kernel_init_freeable+0xf4/0x2f4
  kernel_init_freeable from kernel_init+0x18/0x158
  kernel_init from ret_from_fork+0x14/0x28

Since the enabled method couldn't match with any entry from
sunxi_mc_smp_data, the value of the index shouldn't be used right after
the loop. So move it after the check of ret in order to have a valid
index.

Fixes: 1631090e34f5 ("ARM: sun9i: smp: Add is_a83t field")
Signed-off-by: Stefan Wahren <wahrenst@gmx.net>
Link: https://lore.kernel.org/r/20231228193903.9078-1-wahrenst@gmx.net
Reviewed-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests: bonding: do not set port down when adding to bond

[ Upstream commit 61fa2493ca76fd7bb74e13f0205274f4ab0aa696 ]

Similar to commit be809424659c ("selftests: bonding: do not set port down
before adding to bond"). The bond-arp-interval-causes-panic test failed
after commit a4abfa627c38 ("net: rtnetlink: Enslave device before bringing
it up") as the kernel will set the port down _after_ adding to bond if setting
port down specifically.

Fix it by removing the link down operation when adding to bond.

Fixes: 2ffd57327ff1 ("selftests: bonding: cause oops in bond_rr_gen_slave_id")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Tested-by: Benjamin Poirier <benjamin.poirier@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: Implement missing getsockopt(SO_TIMESTAMPING_NEW)

[ Upstream commit 7f6ca95d16b96567ce4cf458a2790ff17fa620c3 ]

Commit 9718475e6908 ("socket: Add SO_TIMESTAMPING_NEW") added the new
socket option SO_TIMESTAMPING_NEW. Setting the option is handled in
sk_setsockopt(), querying it was not handled in sk_getsockopt(), though.

Following remarks on an earlier submission of this patch, keep the old
behavior of getsockopt(SO_TIMESTAMPING_OLD) which returns the active
flags even if they actually have been set through SO_TIMESTAMPING_NEW.

The new getsockopt(SO_TIMESTAMPING_NEW) is stricter, returning flags
only if they have been set through the same option.

Fixes: 9718475e6908 ("socket: Add SO_TIMESTAMPING_NEW")
Link: https://lore.kernel.org/lkml/20230703175048.151683-1-jthinz@mailbox.tu-berlin.de/
Link: https://lore.kernel.org/netdev/0d7cddc9-03fa-43db-a579-14f3e822615b@app.fastmail.com/
Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

r8169: Fix PCI error on system resume

[ Upstream commit 9c476269bff2908a20930c58085bf0b05ebd569a ]

Some r8168 NICs stop working upon system resume:

[  688.051096] r8169 0000:02:00.1 enp2s0f1: rtl_ep_ocp_read_cond == 0 (loop: 10, delay: 10000).
[  688.175131] r8169 0000:02:00.1 enp2s0f1: Link is Down
...
[  691.534611] r8169 0000:02:00.1 enp2s0f1: PCI error (cmd = 0x0407, status_errs = 0x0000)

Not sure if it's related, but those NICs have a BMC device at function
0:
02:00.0 Unassigned class [ff00]: Realtek Semiconductor Co., Ltd. Realtek RealManage BMC [10ec:816e] (rev 1a)

Trial and error shows that increase the loop wait on
rtl_ep_ocp_read_cond to 30 can eliminate the issue, so let
rtl8168ep_driver_start() to wait a bit longer.

Fixes: e6d6ca6e1204 ("r8169: Add support for another RTL8168FP")
Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Reviewed-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: sched: em_text: fix possible memory leak in em_text_destroy()

[ Upstream commit 8fcb0382af6f1ef50936f1be05b8149eb2f88496 ]

m->data needs to be freed when em_text_destroy is called.

Fixes: d675c989ed2d ("[PKT_SCHED]: Packet classification based on textsearch (ematch)")
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mlxbf_gige: fix receive packet race condition

[ Upstream commit dcea1bd45e6d111cc8fc1aaefa7e31694089bda3 ]

Under heavy traffic, the BlueField Gigabit interface can
become unresponsive. This is due to a possible race condition
in the mlxbf_gige_rx_packet function, where the function exits
with producer and consumer indices equal but there are remaining
packet(s) to be processed. In order to prevent this situation,
read receive consumer index *before* the HW replenish so that
the mlxbf_gige_rx_packet function returns an accurate return
value even if a packet is received into just-replenished buffer
prior to exiting this routine. If the just-replenished buffer
is received and occupies the last RX ring entry, the interface
would not recover and instead would encounter RX packet drops
related to internal buffer shortages since the driver RX logic
is not being triggered to drain the RX ring. This patch will
address and prevent this "ring full" condition.

Fixes: f92e1869d74e ("Add Mellanox BlueField Gigabit Ethernet driver")
Reviewed-by: Asmaa Mnebhi <asmaa@nvidia.com>
Signed-off-by: David Thompson <davthompson@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: mediatek: mt8186: fix AUD_PAD_TOP register and offset

[ Upstream commit 38744c3fa00109c51076121c2deb4f02e2f09194 ]

AUD_PAD_TOP widget's correct register is AFE_AUD_PAD_TOP , and not zero.
Having a zero as register, it would mean that the `snd_soc_dapm_new_widgets`
would try to read the register at offset zero when trying to get the power
status of this widget, which is incorrect.

Fixes: b65c466220b3 ("ASoC: mediatek: mt8186: support adda in platform driver")
Signed-off-by: Eugen Hristev <eugen.hristev@collabora.com>
Link: https://lore.kernel.org/r/20231229114342.195867-1-eugen.hristev@collabora.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: fsl_rpmsg: Fix error handler with pm_runtime_enable

[ Upstream commit f9d378fc68c43fd41b35133edec9cd902ec334ec ]

There is error message when defer probe happens:

fsl_rpmsg rpmsg_audio: Unbalanced pm_runtime_enable!

Fix the error handler with pm_runtime_enable.

Fixes: b73d9e6225e8 ("ASoC: fsl_rpmsg: Add CPU DAI driver for audio base on rpmsg")
Signed-off-by: Chancel Liu <chancel.liu@nxp.com>
Acked-by: Shengjiu Wang <shengjiu.wang@gmail.com>
Link: https://lore.kernel.org/r/20231225080608.967953-1-chancel.liu@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

igc: Check VLAN EtherType mask

[ Upstream commit 7afd49a38e73afd57ff62c8d1cf5af760c4d49c0 ]

Currently the driver accepts VLAN EtherType steering rules regardless of
the configured mask. And things might fail silently or with confusing error
messages to the user. The VLAN EtherType can only be matched by full
mask. Therefore, add a check for that.

For instance the following rule is invalid, but the driver accepts it and
ignores the user specified mask:
|root@host:~# ethtool -N enp3s0 flow-type ether vlan-etype 0x8100 \
|             m 0x00ff action 0
|Added rule with ID 63
|root@host:~# ethtool --show-ntuple enp3s0
|4 RX rings available
|Total 1 rules
|
|Filter: 63
|        Flow Type: Raw Ethernet
|        Src MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Dest MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Ethertype: 0x0 mask: 0xFFFF
|        VLAN EtherType: 0x8100 mask: 0x0
|        VLAN: 0x0 mask: 0xffff
|        User-defined: 0x0 mask: 0xffffffffffffffff
|        Action: Direct to queue 0

After:
|root@host:~# ethtool -N enp3s0 flow-type ether vlan-etype 0x8100 \
|             m 0x00ff action 0
|rmgr: Cannot insert RX class rule: Operation not supported

Fixes: 2b477d057e33 ("igc: Integrate flex filter into ethtool ops")
Suggested-by: Suman Ghosh <sumang@marvell.com>
Signed-off-by: Kurt Kanzenbach <kurt@linutronix.de>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

igc: Check VLAN TCI mask

[ Upstream commit b5063cbe148b829e8eb97672c2cbccc058835476 ]

Currently the driver accepts VLAN TCI steering rules regardless of the
configured mask. And things might fail silently or with confusing error
messages to the user.

There are two ways to handle the VLAN TCI mask:

 1. Match on the PCP field using a VLAN prio filter
 2. Match on complete TCI field using a flex filter

Therefore, add checks and code for that.

For instance the following rule is invalid and will be converted into a
VLAN prio rule which is not correct:
|root@host:~# ethtool -N enp3s0 flow-type ether vlan 0x0001 m 0xf000 \
|             action 1
|Added rule with ID 61
|root@host:~# ethtool --show-ntuple enp3s0
|4 RX rings available
|Total 1 rules
|
|Filter: 61
|        Flow Type: Raw Ethernet
|        Src MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Dest MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Ethertype: 0x0 mask: 0xFFFF
|        VLAN EtherType: 0x0 mask: 0xffff
|        VLAN: 0x1 mask: 0x1fff
|        User-defined: 0x0 mask: 0xffffffffffffffff
|        Action: Direct to queue 1

After:
|root@host:~# ethtool -N enp3s0 flow-type ether vlan 0x0001 m 0xf000 \
|             action 1
|rmgr: Cannot insert RX class rule: Operation not supported

Fixes: 7991487ecb2d ("igc: Allow for Flex Filters to be installed")
Signed-off-by: Kurt Kanzenbach <kurt@linutronix.de>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

igc: Report VLAN EtherType matching back to user

[ Upstream commit 088464abd48cf3735aee91f9e211b32da9d81117 ]

Currently the driver allows to configure matching by VLAN EtherType.
However, the retrieval function does not report it back to the user. Add
it.

Before:
|root@host:~# ethtool -N enp3s0 flow-type ether vlan-etype 0x8100 action 0
|Added rule with ID 63
|root@host:~# ethtool --show-ntuple enp3s0
|4 RX rings available
|Total 1 rules
|
|Filter: 63
|        Flow Type: Raw Ethernet
|        Src MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Dest MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Ethertype: 0x0 mask: 0xFFFF
|        Action: Direct to queue 0

After:
|root@host:~# ethtool -N enp3s0 flow-type ether vlan-etype 0x8100 action 0
|Added rule with ID 63
|root@host:~# ethtool --show-ntuple enp3s0
|4 RX rings available
|Total 1 rules
|
|Filter: 63
|        Flow Type: Raw Ethernet
|        Src MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Dest MAC addr: 00:00:00:00:00:00 mask: FF:FF:FF:FF:FF:FF
|        Ethertype: 0x0 mask: 0xFFFF
|        VLAN EtherType: 0x8100 mask: 0x0
|        VLAN: 0x0 mask: 0xffff
|        User-defined: 0x0 mask: 0xffffffffffffffff
|        Action: Direct to queue 0

Fixes: 2b477d057e33 ("igc: Integrate flex filter into ethtool ops")
Signed-off-by: Kurt Kanzenbach <kurt@linutronix.de>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i40e: Fix filter input checks to prevent config with invalid values

[ Upstream commit 3e48041d9820c17e0a51599d12e66c6e12a8d08d ]

Prevent VF from configuring filters with unsupported actions or use
REDIRECT action with invalid tc number. Current checks could cause
out of bounds access on PF side.

Fixes: e284fc280473 ("i40e: Add and delete cloud filter")
Reviewed-by: Andrii Staikov <andrii.staikov@intel.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ice: Shut down VSI with "link-down-on-close" enabled

[ Upstream commit 6d05ff55ef4f4954d28551236239f297bd52ea48 ]

Disabling netdev with ethtool private flag "link-down-on-close" enabled
can cause NULL pointer dereference bug. Shut down VSI regardless of
"link-down-on-close" state.

Fixes: 8ac7132704f3 ("ice: Fix interface being down after reset with link-down-on-close flag on")
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Ngai-Mint Kwan <ngai-mint.kwan@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ice: Fix link_down_on_close message

[ Upstream commit 6a8d8bb55e7001de2d50920381cc858f3a3e9fb7 ]

The driver should not report an error message when for a medialess port
the link_down_on_close flag is enabled and the physical link cannot be
set down.

Fixes: 8ac7132704f3 ("ice: Fix interface being down after reset with link-down-on-close flag on")
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Katarzyna Wieczerzycka <katarzyna.wieczerzycka@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/i915/perf: Update handling of MMIO triggered reports

[ Upstream commit ee11d2d37f5c05bd7bf5ccc820a58f48423d032b ]

On XEHP platforms user is not able to find MMIO triggered reports in the
OA buffer since i915 squashes the context ID fields. These context ID
fields hold the MMIO trigger markers.

Update logic to not squash the context ID fields of MMIO triggered
reports.

Fixes: cba94bbcff08 ("drm/i915/perf: Determine context valid in OA reports")
Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231219000543.1087706-1-umesh.nerlige.ramappa@intel.com
(cherry picked from commit 0c68132df6e66244acec1bb5b9e19b0751414389)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/i915/dp: Fix passing the correct DPCD_REV for drm_dp_set_phy_test_pattern

[ Upstream commit 2bd7a06a1208aaacb4e7a2a5436c23bce8d70801 ]

Using link_status to get DPCD_REV fails when disabling/defaulting
phy pattern. Use intel_dp->dpcd to access DPCD_REV correctly.

Fixes: 8cdf72711928 ("drm/i915/dp: Program vswing, pre-emphasis, test-pattern")
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Cc: Lee Shawn C <shawn.c.lee@intel.com>
Signed-off-by: Khaled Almahallawy <khaled.almahallawy@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231213211542.3585105-3-khaled.almahallawy@intel.com
(cherry picked from commit 3ee302ec22d6e1d7d1e6d381b0d507ee80f2135c)
Signed-off-by: Sasha Levin <sashal@kernel.org>

octeontx2-af: Fix marking couple of structure as __packed

[ Upstream commit 0ee2384a5a0f3b4eeac8d10bb01a0609d245a4d1 ]

Couple of structures was not marked as __packed. This patch
fixes the same and mark them as __packed.

Fixes: 42006910b5ea ("octeontx2-af: cleanup KPU config data")
Signed-off-by: Suman Ghosh <sumang@marvell.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfc: llcp_core: Hold a ref to llcp_local->dev when holding a ref to llcp_local

[ Upstream commit c95f919567d6f1914f13350af61a1b044ac85014 ]

llcp_sock_sendmsg() calls nfc_llcp_send_ui_frame() which in turn calls
nfc_alloc_send_skb(), which accesses the nfc_dev from the llcp_sock for
getting the headroom and tailroom needed for skb allocation.

Parallelly the nfc_dev can be freed, as the refcount is decreased via
nfc_free_device(), leading to a UAF reported by Syzkaller, which can
be summarized as follows:

(1) llcp_sock_sendmsg() -> nfc_llcp_send_ui_frame()
-> nfc_alloc_send_skb() -> Dereference *nfc_dev
(2) virtual_ncidev_close() -> nci_free_device() -> nfc_free_device()
-> put_device() -> nfc_release() -> Free *nfc_dev

When a reference to llcp_local is acquired, we do not acquire the same
for the nfc_dev. This leads to freeing even when the llcp_local is in
use, and this is the case with the UAF described above too.

Thus, when we acquire a reference to llcp_local, we should acquire a
reference to nfc_dev, and release the references appropriately later.

References for llcp_local is initialized in nfc_llcp_register_device()
(which is called by nfc_register_device()). Thus, we should acquire a
reference to nfc_dev there.

nfc_unregister_device() calls nfc_llcp_unregister_device() which in
turn calls nfc_llcp_local_put(). Thus, the reference to nfc_dev is
appropriately released later.

Reported-and-tested-by: syzbot+bbe84a4010eeea00982d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=bbe84a4010eeea00982d
Fixes: c7aa12252f51 ("NFC: Take a reference on the LLCP local pointer when creating a socket")
Reviewed-by: Suman Ghosh <sumang@marvell.com>
Signed-off-by: Siddh Raman Pant <code@siddh.me>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_tables: set transport offset from mac header for netdev/egress

[ Upstream commit 0ae8e4cca78781401b17721bfb72718fdf7b4912 ]

Before this patch, transport offset (pkt->thoff) provides an offset
relative to the network header. This is fine for the inet families
because skb->data points to the network header in such case. However,
from netdev/egress, skb->data points to the mac header (if available),
thus, pkt->thoff is missing the mac header length.

Add skb_network_offset() to the transport offset (pkt->thoff) for
netdev, so transport header mangling works as expected. Adjust payload
fast eval function to use skb->data now that pkt->thoff provides an
absolute offset. This explains why users report that matching on
egress/netdev works but payload mangling does not.

This patch implicitly fixes payload mangling for IPv4 packets in
netdev/egress given skb_store_bits() requires an offset from skb->data
to reach the transport header.

I suspect that nft_exthdr and the trace infra were also broken from
netdev/egress because they also take skb->data as start, and pkt->thoff
was not correct.

Note that IPv6 is fine because ipv6_find_hdr() already provides a
transport offset starting from skb->data, which includes
skb_network_offset().

The bridge family also uses nft_set_pktinfo_ipv4_validate(), but there
skb_network_offset() is zero, so the update in this patch does not alter
the existing behaviour.

Fixes: 42df6e1d221d ("netfilter: Introduce egress hook")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/bridge: ps8640: Fix size mismatch warning w/ len

[ Upstream commit 35ba6bd582cf926a082296b7e9a876ec81136cb1 ]

After commit 26195af57798 ("drm/bridge: ps8640: Drop the ability of
ps8640 to fetch the EDID"), I got an error compiling:

  error: comparison of distinct pointer types
  ('typeof (len) *' (aka 'unsigned int *') and
   'typeof (msg->size) *' (aka 'unsigned long *'))
  [-Werror,-Wcompare-distinct-pointer-types]

Fix it by declaring the `len` as size_t.

The above error only shows up on downstream kernels without commit
d03eba99f5bf ("minmax: allow min()/max()/clamp() if the arguments have
the same signedness."), but since commit 26195af57798 ("drm/bridge:
ps8640: Drop the ability of ps8640 to fetch the EDID") is a "Fix" that
will likely be backported it seems nice to make it easy. ...plus it's
more correct to declare `len` as size_t anyway.

Fixes: 26195af57798 ("drm/bridge: ps8640: Drop the ability of ps8640 to fetch the EDID")
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231218090454.1.I5c6eb80b2f746439c4b58efab788e00701d08759@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/bridge: ti-sn65dsi86: Never store more than msg->size bytes in AUX xfer

[ Upstream commit aca58eac52b88138ab98c814afb389a381725cd7 ]

For aux reads, the value `msg->size` indicates the size of the buffer
provided by `msg->buffer`. We should never in any circumstances write
more bytes to the buffer since it may overflow the buffer.

In the ti-sn65dsi86 driver there is one code path that reads the
transfer length from hardware. Even though it's never been seen to be
a problem, we should make extra sure that the hardware isn't
increasing the length since doing so would cause us to overrun the
buffer.

Fixes: 982f589bde7a ("drm/bridge: ti-sn65dsi86: Update reply on aux failures")
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231214123752.v3.2.I7b83c0f31aeedc6b1dc98c7c741d3e1f94f040f8@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/bridge: parade-ps8640: Never store more than msg->size bytes in AUX xfer

[ Upstream commit 3164c8a70073d43629b4e11e083d3d2798f7750f ]

While testing, I happened to notice a random crash that looked like:

  Kernel panic - not syncing: stack-protector:
  Kernel stack is corrupted in: drm_dp_dpcd_probe+0x120/0x120

Analysis of drm_dp_dpcd_probe() shows that we pass in a 1-byte buffer
(allocated on the stack) to the aux->transfer() function. Presumably
if the aux->transfer() writes more than one byte to this buffer then
we're in a bad shape.

Dropping into kgdb, I noticed that "aux->transfer" pointed at
ps8640_aux_transfer().

Reading through ps8640_aux_transfer(), I can see that there are cases
where it could write more bytes to msg->buffer than were specified by
msg->size. This could happen if the hardware reported back something
bogus to us. Let's fix this so we never write more than msg->size
bytes. We'll still read all the bytes from the hardware just in case
the hardware requires it since the aux transfer data comes through an
auto-incrementing register.

NOTE: I have no actual way to reproduce this issue but it seems likely
this is what was happening in the crash I looked at.

Fixes: 13afcdd7277e ("drm/bridge: parade-ps8640: Add support for AUX channel")
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20231214123752.v3.1.I9d1afcaad76a3e2c0ca046dc4adbc2b632c22eda@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: iwlwifi: pcie: don't synchronize IRQs from IRQ

[ Upstream commit 400f6ebbc175286576c7f7fddf3c347d09d12310 ]

On older devices (before unified image!) we can end up calling
stop_device from an rfkill interrupt. However, in stop_device
we attempt to synchronize IRQs, which then of course deadlocks.

Avoid this by checking the context, if running from the IRQ
thread then don't synchronize. This wouldn't be correct on a
new device since RSS is supported, but older devices only have
a single interrupt/queue.

Fixes: 37fb29bd1f90 ("wifi: iwlwifi: pcie: synchronize IRQs before NAPI")
Reviewed-by: Miri Korenblit <miriam.rachel.korenblit@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://msgid.link/20231215111335.59aab00baed7.Iadfe154d6248e7f9dfd69522e5429dbbd72925d7@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

accel/qaic: Implement quirk for SOC_HW_VERSION

[ Upstream commit 4c8874c2a6512b9fe7285cab1a6910d9211a6cfb ]

The SOC_HW_VERSION register in the BHI space is not correctly initialized
by the device and in many cases contains uninitialized data. The register
could contain 0xFFFFFFFF which is a special value to indicate a link
error in PCIe, therefore if observed, we could incorrectly think the
device is down.

Intercept reads for this register, and provide the correct value - every
production instance would read 0x60110200 if the device was operating as
intended.

Fixes: a36bf7af868b ("accel/qaic: Add MHI controller")
Signed-off-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Reviewed-by: Pranjal Ramajor Asha Kanojiya <quic_pkanojiy@quicinc.com>
Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231208163101.1295769-3-quic_jhugo@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

accel/qaic: Fix GEM import path code

[ Upstream commit c8b6f4ad2ff9c6d88cdeb9acf16d0c4a323dd499 ]

Do not modify the size of dmabuf as it is immutable.

Fixes: ff13be830333 ("accel/qaic: Add datapath")
Signed-off-by: Pranjal Ramajor Asha Kanojiya <quic_pkanojiy@quicinc.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Signed-off-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231208163101.1295769-2-quic_jhugo@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: x86/pmu: fix masking logic for MSR_CORE_PERF_GLOBAL_CTRL

commit 971079464001c6856186ca137778e534d983174a upstream.

When commit c59a1f106f5c ("KVM: x86/pmu: Add IA32_PEBS_ENABLE
MSR emulation for extended PEBS") switched the initialization of
cpuc->guest_switch_msrs to use compound literals, it screwed up
the boolean logic:

+ u64 pebs_mask = cpuc->pebs_enabled & x86_pmu.pebs_capable;
...
- arr[0].guest = intel_ctrl & ~cpuc->intel_ctrl_host_mask;
- arr[0].guest &= ~(cpuc->pebs_enabled & x86_pmu.pebs_capable);
+               .guest = intel_ctrl & (~cpuc->intel_ctrl_host_mask | ~pebs_mask),

Before the patch, the value of arr[0].guest would have been intel_ctrl &
~cpuc->intel_ctrl_host_mask & ~pebs_mask.  The intent is to always treat
PEBS events as host-only because, while the guest runs, there is no way
to tell the processor about the virtual address where to put PEBS records
intended for the host.

Unfortunately, the new expression can be expanded to

(intel_ctrl & ~cpuc->intel_ctrl_host_mask) | (intel_ctrl & ~pebs_mask)

which makes no sense; it includes any bit that isn't *both* marked as
exclude_guest and using PEBS.  So, reinstate the old logic.  Another
way to write it could be "intel_ctrl & ~(cpuc->intel_ctrl_host_mask |
pebs_mask)", presumably the intention of the author of the faulty.
However, I personally find the repeated application of A AND NOT B to
be a bit more readable.

This shows up as guest failures when running concurrent long-running
perf workloads on the host, and was reported to happen with rcutorture.
All guests on a given host would die simultaneously with something like an
instruction fault or a segmentation violation.

Reported-by: Paul E. McKenney <paulmck@kernel.org>
Analyzed-by: Sean Christopherson <seanjc@google.com>
Tested-by: Paul E. McKenney <paulmck@kernel.org>
Cc: stable@vger.kernel.org
Fixes: c59a1f106f5c ("KVM: x86/pmu: Add IA32_PEBS_ENABLE MSR emulation for extended PEBS")
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: do not depend on release_iface for maintaining iface_list

commit 09eeb0723f219fbd96d8865bf9b935e03ee2ec22 upstream.

parse_server_interfaces should be in complete charge of maintaining
the iface_list linked list. Today, iface entries are removed
from the list only when the last refcount is dropped.
i.e. in release_iface. However, this can result in undercounting
of refcount if the server stops advertising interfaces (which
Azure SMB server does).

This change puts parse_server_interfaces in full charge of
maintaining the iface_list. So if an empty list is returned
by the server, the entries in the list will immediately be
removed. This way, a following call to the same function will
not find entries in the list.

Fixes: aa45dadd34e4 ("cifs: change iface_list from array to sorted linked list")
Cc: stable@vger.kernel.org
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: cifs_chan_is_iface_active should be called with chan_lock held

commit 7257bcf3bdc785eabc4eef1f329a59815b032508 upstream.

cifs_chan_is_iface_active checks the channels of a session to see
if the associated iface is active. This should always happen
with chan_lock held. However, these two callers of this function
were missing this locking.

This change makes sure the function calls are protected with
proper locking.

Fixes: b54034a73baf ("cifs: during reconnect, update interface if necessary")
Fixes: fa1d0508bdd4 ("cifs: account for primary channel in the interface list")
Cc: stable@vger.kernel.org
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/mgag200: Fix gamma lut not initialized for G200ER, G200EV, G200SE

commit 11f9eb899ecc8c02b769cf8d2532ba12786a7af7 upstream.

When mgag200 switched from simple KMS to regular atomic helpers,
the initialization of the gamma settings was lost.
This leads to a black screen, if the bios/uefi doesn't use the same
pixel color depth.
This has been fixed with commit ad81e23426a6 ("drm/mgag200: Fix gamma
lut not initialized.") for most G200, but G200ER, G200EV, G200SE use
their own version of crtc_helper_atomic_enable() and need to be fixed
too.

Fixes: 1baf9127c482 ("drm/mgag200: Replace simple-KMS with regular atomic helpers")
Cc: <stable@vger.kernel.org> #v6.1+
Reported-by: Roger Sewell <roger.sewell@cantab.net>
Suggested-by: Roger Sewell <roger.sewell@cantab.net>
Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20231214163849.359691-1-jfalempe@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "PCI/ASPM: Remove pcie_aspm_pm_state_change()"

commit f93e71aea6c60ebff8adbd8941e678302d377869 upstream.

This reverts commit 08d0cc5f34265d1a1e3031f319f594bd1970976c.

Michael reported that when attempting to resume from suspend to RAM on ASUS
mini PC PN51-BB757MDE1 (DMI model: MINIPC PN51-E1), 08d0cc5f3426
("PCI/ASPM: Remove pcie_aspm_pm_state_change()") caused a 12-second delay
with no output, followed by a reboot.

Workarounds include:

  - Reverting 08d0cc5f3426 ("PCI/ASPM: Remove pcie_aspm_pm_state_change()")
  - Booting with "pcie_aspm=off"
  - Booting with "pcie_aspm.policy=performance"
  - "echo 0 | sudo tee /sys/bus/pci/devices/0000:03:00.0/link/l1_aspm"
    before suspending
  - Connecting a USB flash drive

Link: https://lore.kernel.org/r/20240102232550.1751655-1-helgaas@kernel.org
Fixes: 08d0cc5f3426 ("PCI/ASPM: Remove pcie_aspm_pm_state_change()")
Reported-by: Michael Schaller <michael@5challer.de>
Link: https://lore.kernel.org/r/76c61361-b8b4-435f-a9f1-32b716763d62@5challer.de
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mptcp: prevent tcp diag from closing listener subflows

commit 4c0288299fd09ee7c6fbe2f57421f314d8c981db upstream.

The MPTCP protocol does not expect that any other entity could change
the first subflow status when such socket is listening.
Unfortunately the TCP diag interface allows aborting any TCP socket,
including MPTCP listeners subflows. As reported by syzbot, that trigger
a WARN() and could lead to later bigger trouble.

The MPTCP protocol needs to do some MPTCP-level cleanup actions to
properly shutdown the listener. To keep the fix simple, prevent
entirely the diag interface from stopping such listeners.

We could refine the diag callback in a later, larger patch targeting
net-next.

Fixes: 57fc0f1ceaa4 ("mptcp: ensure listener is unhashed before updating the sk status")
Cc: stable@vger.kernel.org
Reported-by: <syzbot+5a01c3a666e726bc8752@syzkaller.appspotmail.com>
Closes: https://lore.kernel.org/netdev/0000000000004f4579060c68431b@google.com/
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Matthieu Baerts <matttbe@kernel.org>
Link: https://lore.kernel.org/r/20231226-upstream-net-20231226-mptcp-prevent-warn-v1-2-1404dcc431ea@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amd/display: pbn_div need be updated for hotplug event

commit 9cdef4f720376ef0fb0febce1ed2377c19e531f9 upstream.

link_rate sometime will be changed when DP MST connector hotplug, so
pbn_div also need be updated; otherwise, it will mismatch with
link_rate, causes no output in external monitor.

This is a backport to 6.7 and older.

Cc: stable@vger.kernel.org
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Jerry Zuo <jerry.zuo@amd.com>
Acked-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Wade Wang <wade.wang@hp.com>
Signed-off-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: hda/realtek: Fix mute and mic-mute LEDs for HP ProBook 440 G6

commit b6ce6e6c79e4ec650887f1fe391a70e54972001a upstream.

LEDs in 'HP ProBook 440 G6' laptop are controlled by ALC236 codec.
Enable already existing quirk 'ALC236_FIXUP_HP_MUTE_LED_MICMUTE_VREF'
to fix mute and mic-mute LEDs.

Signed-off-by: Siddhesh Dharme <siddheshdharme18@gmail.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20240104060736.5149-1-siddheshdharme18@gmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: hda/realtek: fix mute/micmute LEDs for a HP ZBook

commit 18a434f32fa61b3fda8ddcd9a63d5274569c6a41 upstream.

There is a HP ZBook which using ALC236 codec and need the
ALC236_FIXUP_HP_MUTE_LED_MICMUTE_VREF quirk to make mute LED
and micmute LED work.

[ confirmed that the new entries are for new models that have no
  proper name, so the strings are left as "HP" which will be updated
  eventually later -- tiwai ]

Signed-off-by: Andy Chi <andy.chi@canonical.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20240102024916.19093-1-andy.chi@canonical.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: hda/realtek: enable SND_PCI_QUIRK for hp pavilion 14-ec1xxx series

commit 13a5b21197587a3d9cac9e1a00de9b91526a55e4 upstream.

The HP Pavilion 14 ec1xxx series uses the HP mainboard 8A0F with the
ALC287 codec.
The mute led can be enabled using the already existing
ALC287_FIXUP_HP_GPIO_LED quirk.
Tested on an HP Pavilion ec1003AU

Signed-off-by: Aabish Malik <aabishmalik3337@gmail.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20231229170352.742261-3-aabishmalik3337@gmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: hda/tas2781: remove sound controls in unbind

commit 4e7914eb1dae377b8e6de59c96b0653aacb47646 upstream.

Remove sound controls in hda_unbind to make
module loadable after module unload.

Add a driver specific struct (tas2781_hda) to store
the controls.

This patch depends on patch:
ALSA: hda/tas2781: do not use regcache

Fixes: 5be27f1e3ec9 ("ALSA: hda/tas2781: Add tas2781 HDA driver")
CC: stable@vger.kernel.org
Signed-off-by: Gergo Koteles <soyer@irl.hu>
Link: https://lore.kernel.org/r/362aa3e2f81b9259a3e5222f576bec5debfc5e88.1703204848.git.soyer@irl.hu
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: hda/tas2781: move set_drv_data outside tasdevice_init

commit e7aa105657f7f62f54a493480588895cc8a9a1a7 upstream.

allow driver specific driver data in tas2781-hda-i2c and tas2781-i2c

Fixes: ef3bcde75d06 ("ASoC: tas2781: Add tas2781 driver")
CC: stable@vger.kernel.org
Signed-off-by: Gergo Koteles <soyer@irl.hu>
Link: https://lore.kernel.org/r/1398bd8bf3e935b1595a99128320e4a1913e210a.1703204848.git.soyer@irl.hu
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: hda/tas2781: do not use regcache

commit 6dad45f4d28977bd1948973107cf325d431e5b7e upstream.

There are two problems with using regcache in this module.

The amplifier has 3 addressing levels (BOOK, PAGE, REG). The firmware
contains blocks that must be written to BOOK 0x8C. The regcache doesn't
know anything about BOOK, so regcache_sync writes invalid values to the
actual BOOK.

The module handles 2 or more separate amplifiers. The amplifiers have
different register values, and the module uses only one regmap/regcache
for all the amplifiers. The regcache_sync only writes the last amplifier
used.

The module successfully restores all the written register values (RC
profile, program, configuration, calibration) without regcache.

Remove regcache functions and set regmap cache_type to REGCACHE_NONE.

Link: https://lore.kernel.org/r/21a183b5a08cb23b193af78d4b1114cc59419272.1701906455.git.soyer@irl.hu/
Fixes: 5be27f1e3ec9 ("ALSA: hda/tas2781: Add tas2781 HDA driver")
Acked-by: Mark Brown <broonie@kernel.org>
CC: stable@vger.kernel.org
Signed-off-by: Gergo Koteles <soyer@irl.hu>
Link: https://lore.kernel.org/r/491aeed0e2eecc3b704ec856f815db21bad3ba0e.1703202126.git.soyer@irl.hu
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

keys, dns: Fix missing size check of V1 server-list header

commit 1997b3cb4217b09e49659b634c94da47f0340409 upstream.

The dns_resolver_preparse() function has a check on the size of the
payload for the basic header of the binary-style payload, but is missing
a check for the size of the V1 server-list payload header after
determining that's what we've been given.

Fix this by getting rid of the the pointer to the basic header and just
assuming that we have a V1 server-list payload and moving the V1 server
list pointer inside the if-statement.  Dealing with other types and
versions can be left for when such have been defined.

This can be tested by doing the following with KASAN enabled:

    echo -n -e '\x0\x0\x1\x2' | keyctl padd dns_resolver foo @p

and produces an oops like the following:

    BUG: KASAN: slab-out-of-bounds in dns_resolver_preparse+0xc9f/0xd60 net/dns_resolver/dns_key.c:127
    Read of size 1 at addr ffff888028894084 by task syz-executor265/5069
    ...
    Call Trace:
      dns_resolver_preparse+0xc9f/0xd60 net/dns_resolver/dns_key.c:127
      __key_create_or_update+0x453/0xdf0 security/keys/key.c:842
      key_create_or_update+0x42/0x50 security/keys/key.c:1007
      __do_sys_add_key+0x29c/0x450 security/keys/keyctl.c:134
      do_syscall_x64 arch/x86/entry/common.c:52 [inline]
      do_syscall_64+0x40/0x110 arch/x86/entry/common.c:83
      entry_SYSCALL_64_after_hwframe+0x62/0x6a

This patch was originally by Edward Adam Davis, but was modified by
Linus.

Fixes: b946001d3bb1 ("keys, dns: Allow key types (eg. DNS) to be reclaimed immediately on expiry")
Reported-and-tested-by: syzbot+94bbb75204a05da3d89f@syzkaller.appspotmail.com
Link: https://lore.kernel.org/r/0000000000009b39bc060c73e209@google.com/
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Edward Adam Davis <eadavis@qq.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: David Howells <dhowells@redhat.com>
Cc: Edward Adam Davis <eadavis@qq.com>
Cc: Jarkko Sakkinen <jarkko@kernel.org>
Cc: Jeffrey E Altman <jaltman@auristor.com>
Cc: Wang Lei <wang840925@gmail.com>
Cc: Jeff Layton <jlayton@redhat.com>
Cc: Steve French <sfrench@us.ibm.com>
Cc: Marc Dionne <marc.dionne@auristor.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Jeffrey E Altman <jaltman@auristor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Linux 6.6.10

Link: https://lore.kernel.org/r/20240103164834.970234661@linuxfoundation.org
Tested-by: Nam Cao <namcao@linutronix.de>
Tested-by: Ronald Warsow <rwarsow@gmx.de>
Tested-by: SeongJae Park <sj@kernel.org>
Tested-by: Salvatore Bonaccorso <carnil@debian.org>
Tested-by: Florian Fainelli <florian.fainelli@broadcom.com>
Tested-by: Kelsey Steele <kelseysteele@linux.microsoft.com>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Takeshi Ogasawara <takeshi.ogasawara@futuring-girl.com>
Tested-by: Bagas Sanjaya <bagasdotme@gmail.com>
Tested-by: Ron Economos <re@w6rz.net>
Tested-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Tested-by: Allen Pais <apais@linux.microsoft.com>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "platform/x86: p2sb: Allow p2sb_bar() calls during PCI device probe"

commit b20712e853305cbd04673f02b7e52ba5b12c11a9 upstream.

This reverts commit b28ff7a7c3245d7f62acc20f15b4361292fe4117.

The commit introduced P2SB device scan and resource cache during the
boot process to avoid deadlock. But it caused detection failure of
IDE controllers on old systems [1]. The IDE controllers on old systems
and P2SB devices on newer systems have same PCI DEVFN. It is suspected
the confusion between those two is the failure cause. Revert the change
at this moment until the proper solution gets ready.

Link: https://lore.kernel.org/platform-driver-x86/CABq1_vjfyp_B-f4LAL6pg394bP6nDFyvg110TOLHHb0x4aCPeg@mail.gmail.com/T/#m07b30468d9676fc5e3bb2122371121e4559bb383
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Link: https://lore.kernel.org/r/20240104114050.3142690-1-shinichiro.kawasaki@wdc.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

netfilter: nf_tables: skip set commit for deleted/destroyed sets

commit 7315dc1e122c85ffdfc8defffbb8f8b616c2eb1a upstream.

NFT_MSG_DELSET deactivates all elements in the set, skip
set->ops->commit() to avoid the unnecessary clone (for the pipapo case)
as well as the sync GC cycle, which could deactivate again expired
elements in such set.

Fixes: 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Reported-by: Kevin Rich <kevinrich1337@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

wifi: nl80211: fix deadlock in nl80211_set_cqm_rssi (6.6.x)

Commit 008afb9f3d57 ("wifi: cfg80211: fix CQM for non-range use"
backported to 6.6.x) causes nl80211_set_cqm_rssi not to release the
wdev lock in some of the error paths.

Of course, the ensuing deadlock causes userland network managers to
break pretty badly, and on typical systems this also causes lockups on
on suspend, poweroff and reboot. See [1], [2], [3] for example reports.

The upstream commit 7e7efdda6adb ("wifi: cfg80211: fix CQM for non-range
use"), committed in November 2023, is completely fine because there was
another commit in August 2023 that removed the wdev lock:
see commit 076fc8775daf ("wifi: cfg80211: remove wdev mutex").

The reason things broke in 6.6.5 is that commit 4338058f6009 was applied
without also applying 076fc8775daf.

Commit 076fc8775daf ("wifi: cfg80211: remove wdev mutex") is a rather
large commit; adjusting the error handling (which is what this commit does)
yields a much simpler patch and was tested to work properly.

Fix the deadlock by releasing the lock before returning.

[1] https://bugzilla.kernel.org/show_bug.cgi?id=218247
[2] https://bbs.archlinux.org/viewtopic.php?id=290976
[3] https://lore.kernel.org/all/87sf4belmm.fsf@turtle.gmx.de/

Link: https://lore.kernel.org/stable/e374bb16-5b13-44cc-b11a-2f4eefb1ecf5@manjaro.org/
Fixes: 008afb9f3d57 ("wifi: cfg80211: fix CQM for non-range use")
Tested-by: "Léo Lam" <leo@leolam.fr>
Tested-by: Philip Müller <philm@manjaro.org>
Cc: stable@vger.kernel.org
Cc: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: "Léo Lam" <leo@leolam.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

wifi: cfg80211: fix CQM for non-range use

commit 7e7efdda6adb385fbdfd6f819d76bc68c923c394 upstream.

[note: this is commit 4a7e92551618f3737b305f62451353ee05662f57 reapplied;
that commit had been reverted in 6.6.6 because it caused regressions, see
https://lore.kernel.org/stable/2023121450-habitual-transpose-68a1@gregkh/
for details]

My prior race fix here broke CQM when ranges aren't used, as
the reporting worker now requires the cqm_config to be set in
the wdev, but isn't set when there's no range configured.

Rather than continuing to special-case the range version, set
the cqm_config always and configure accordingly, also tracking
if range was used or not to be able to clear the configuration
appropriately with the same API, which was actually not right
if both were implemented by a driver for some reason, as is
the case with mac80211 (though there the implementations are
equivalent so it doesn't matter.)

Also, the original multiple-RSSI commit lost checking for the
callback, so might have potentially crashed if a driver had
neither implementation, and userspace tried to use it despite
not being advertised as supported.

Cc: stable@vger.kernel.org
Fixes: 4a4b8169501b ("cfg80211: Accept multiple RSSI thresholds for CQM")
Fixes: 37c20b2effe9 ("wifi: cfg80211: fix cqm_config access race")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: "Léo Lam" <leo@leolam.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Fix blocked reader of snapshot buffer

commit 39a7dc23a1ed0fe81141792a09449d124c5953bd upstream.

If an application blocks on the snapshot or snapshot_raw files, expecting
to be woken up when a snapshot occurs, it will not happen. Or it may
happen with an unexpected result.

That result is that the application will be reading the main buffer
instead of the snapshot buffer. That is because when the snapshot occurs,
the main and snapshot buffers are swapped. But the reader has a descriptor
still pointing to the buffer that it originally connected to.

This is fine for the main buffer readers, as they may be blocked waiting
for a watermark to be hit, and when a snapshot occurs, the data that the
main readers want is now on the snapshot buffer.

But for waiters of the snapshot buffer, they are waiting for an event to
occur that will trigger the snapshot and they can then consume it quickly
to save the snapshot before the next snapshot occurs. But to do this, they
need to read the new snapshot buffer, not the old one that is now
receiving new data.

Also, it does not make sense to have a watermark "buffer_percent" on the
snapshot buffer, as the snapshot buffer is static and does not receive new
data except all at once.

Link: https://lore.kernel.org/linux-trace-kernel/20231228095149.77f5b45d@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Fixes: debdd57f5145f ("tracing: Make a snapshot feature available from userspace")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ftrace: Fix modification of direct_function hash while in use

commit d05cb470663a2a1879277e544f69e660208f08f2 upstream.

Masami Hiramatsu reported a memory leak in register_ftrace_direct() where
if the number of new entries are added is large enough to cause two
allocations in the loop:

        for (i = 0; i < size; i++) {
                hlist_for_each_entry(entry, &hash->buckets[i], hlist) {
                        new = ftrace_add_rec_direct(entry->ip, addr, &free_hash);
                        if (!new)
                                goto out_remove;
                        entry->direct = addr;
                }
        }

Where ftrace_add_rec_direct() has:

        if (ftrace_hash_empty(direct_functions) ||
            direct_functions->count > 2 * (1 << direct_functions->size_bits)) {
                struct ftrace_hash *new_hash;
                int size = ftrace_hash_empty(direct_functions) ? 0 :
                        direct_functions->count + 1;

                if (size < 32)
                        size = 32;

                new_hash = dup_hash(direct_functions, size);
                if (!new_hash)
                        return NULL;

                *free_hash = direct_functions;
                direct_functions = new_hash;
        }

The "*free_hash = direct_functions;" can happen twice, losing the previous
allocation of direct_functions.

But this also exposed a more serious bug.

The modification of direct_functions above is not safe. As
direct_functions can be referenced at any time to find what direct caller
it should call, the time between:

                new_hash = dup_hash(direct_functions, size);
 and
                direct_functions = new_hash;

can have a race with another CPU (or even this one if it gets interrupted),
and the entries being moved to the new hash are not referenced.

That's because the "dup_hash()" is really misnamed and is really a
"move_hash()". It moves the entries from the old hash to the new one.

Now even if that was changed, this code is not proper as direct_functions
should not be updated until the end. That is the best way to handle
function reference changes, and is the way other parts of ftrace handles
this.

The following is done:

 1. Change add_hash_entry() to return the entry it created and inserted
    into the hash, and not just return success or not.

 2. Replace ftrace_add_rec_direct() with add_hash_entry(), and remove
    the former.

 3. Allocate a "new_hash" at the start that is made for holding both the
    new hash entries as well as the existing entries in direct_functions.

 4. Copy (not move) the direct_function entries over to the new_hash.

 5. Copy the entries of the added hash to the new_hash.

 6. If everything succeeds, then use rcu_pointer_assign() to update the
    direct_functions with the new_hash.

This simplifies the code and fixes both the memory leak as well as the
race condition mentioned above.

Link: https://lore.kernel.org/all/170368070504.42064.8960569647118388081.stgit@devnote2/
Link: https://lore.kernel.org/linux-trace-kernel/20231229115134.08dd5174@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Fixes: 763e34e74bb7d ("ftrace: Add register_ftrace_direct()")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ring-buffer: Fix wake ups when buffer_percent is set to 100

commit 623b1f896fa8a669a277ee5a258307a16c7377a3 upstream.

The tracefs file "buffer_percent" is to allow user space to set a
water-mark on how much of the tracing ring buffer needs to be filled in
order to wake up a blocked reader.

 0 - is to wait until any data is in the buffer
 1 - is to wait for 1% of the sub buffers to be filled
 50 - would be half of the sub buffers are filled with data
 100 - is not to wake the waiter until the ring buffer is completely full

Unfortunately the test for being full was:

dirty = ring_buffer_nr_dirty_pages(buffer, cpu);
return (dirty * 100) > (full * nr_pages);

Where "full" is the value for "buffer_percent".

There is two issues with the above when full == 100.

1. dirty * 100 > 100 * nr_pages will never be true
   That is, the above is basically saying that if the user sets
   buffer_percent to 100, more pages need to be dirty than exist in the
   ring buffer!

2. The page that the writer is on is never considered dirty, as dirty
   pages are only those that are full. When the writer goes to a new
   sub-buffer, it clears the contents of that sub-buffer.

That is, even if the check was ">=" it would still not be equal as the
most pages that can be considered "dirty" is nr_pages - 1.

To fix this, add one to dirty and use ">=" in the compare.

Link: https://lore.kernel.org/linux-trace-kernel/20231226125902.4a057f1d@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Fixes: 03329f9939781 ("tracing: Add tracefs file buffer_percentage")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "nvme-fc: fix race between error recovery and creating association"

commit d3e8b1858734bf46cda495be4165787b9a3981a6 upstream.

The commit was identified to might sleep in invalid context and is
blocking regression testing.

This reverts commit ee6fdc5055e916b1dd497f11260d4901c4c1e55e.

Link: https://lore.kernel.org/linux-nvme/hkhl56n665uvc6t5d6h3wtx7utkcorw4xlwi7d2t2bnonavhe6@xaan6pu43ap6/
Link: https://lists.infradead.org/pipermail/linux-nvme/2023-December/043756.html
Reported-by: Daniel Wagner <dwagner@suse.de>
Reported-by: Maurizio Lombardi <mlombard@redhat.com>
Cc: Michael Liang <mliang@purestorage.com>
Tested-by: Daniel Wagner <dwagner@suse.de>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/memory-failure: check the mapcount of the precise page

commit c79c5a0a00a9457718056b588f312baadf44e471 upstream.

A process may map only some of the pages in a folio, and might be missed
if it maps the poisoned page but not the head page.  Or it might be
unnecessarily hit if it maps the head page, but not the poisoned page.

Link: https://lkml.kernel.org/r/20231218135837.3310403-3-willy@infradead.org
Fixes: 7af446a841a2 ("HWPOISON, hugetlb: enable error handling path for hugepage")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/memory-failure: cast index to loff_t before shifting it

commit 39ebd6dce62d8cfe3864e16148927a139f11bc9a upstream.

On 32-bit systems, we'll lose the top bits of index because arithmetic
will be performed in unsigned long instead of unsigned long long.  This
affects files over 4GB in size.

Link: https://lkml.kernel.org/r/20231218135837.3310403-4-willy@infradead.org
Fixes: 6100e34b2526 ("mm, memory_failure: Teach memory_failure() about dev_pagemap pages")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm: migrate high-order folios in swap cache correctly

commit fc346d0a70a13d52fe1c4bc49516d83a42cd7c4c upstream.

Large folios occupy N consecutive entries in the swap cache instead of
using multi-index entries like the page cache.  However, if a large folio
is re-added to the LRU list, it can be migrated.  The migration code was
not aware of the difference between the swap cache and the page cache and
assumed that a single xas_store() would be sufficient.

This leaves potentially many stale pointers to the now-migrated folio in
the swap cache, which can lead to almost arbitrary data corruption in the
future.  This can also manifest as infinite loops with the RCU read lock
held.

[willy@infradead.org: modifications to the changelog & tweaked the fix]
Fixes: 3417013e0d18 ("mm/migrate: Add folio_migrate_mapping()")
Link: https://lkml.kernel.org/r/20231214045841.961776-1-willy@infradead.org
Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reported-by: Charan Teja Kalla <quic_charante@quicinc.com>
Closes: https://lkml.kernel.org/r/1700569840-17327-1-git-send-email-quic_charante@quicinc.com
Cc: David Hildenbrand <david@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/filemap: avoid buffered read/write race to read inconsistent data

commit e2c27b803bb664748e090d99042ac128b3f88d92 upstream.

The following concurrency may cause the data read to be inconsistent with
the data on disk:

             cpu1                           cpu2
------------------------------|------------------------------
                               // Buffered write 2048 from 0
                               ext4_buffered_write_iter
                                generic_perform_write
                                 copy_page_from_iter_atomic
                                 ext4_da_write_end
                                  ext4_da_do_write_end
                                   block_write_end
                                    __block_commit_write
                                     folio_mark_uptodate
// Buffered read 4096 from 0          smp_wmb()
ext4_file_read_iter                   set_bit(PG_uptodate, folio_flags)
 generic_file_read_iter            i_size_write // 2048
  filemap_read                     unlock_page(page)
   filemap_get_pages
    filemap_get_read_batch
    folio_test_uptodate(folio)
     ret = test_bit(PG_uptodate, folio_flags)
     if (ret)
      smp_rmb();
      // Ensure that the data in page 0-2048 is up-to-date.

                               // New buffered write 2048 from 2048
                               ext4_buffered_write_iter
                                generic_perform_write
                                 copy_page_from_iter_atomic
                                 ext4_da_write_end
                                  ext4_da_do_write_end
                                   block_write_end
                                    __block_commit_write
                                     folio_mark_uptodate
                                      smp_wmb()
                                      set_bit(PG_uptodate, folio_flags)
                                   i_size_write // 4096
                                   unlock_page(page)

   isize = i_size_read(inode) // 4096
   // Read the latest isize 4096, but without smp_rmb(), there may be
   // Load-Load disorder resulting in the data in the 2048-4096 range
   // in the page is not up-to-date.
   copy_page_to_iter
   // copyout 4096

In the concurrency above, we read the updated i_size, but there is no read
barrier to ensure that the data in the page is the same as the i_size at
this point, so we may copy the unsynchronized page out.  Hence adding the
missing read memory barrier to fix this.

This is a Load-Load reordering issue, which only occurs on some weak
mem-ordering architectures (e.g.  ARM64, ALPHA), but not on strong
mem-ordering architectures (e.g.  X86).  And theoretically the problem
doesn't only happen on ext4, filesystems that call filemap_read() but
don't hold inode lock (e.g.  btrfs, f2fs, ubifs ...) will have this
problem, while filesystems with inode lock (e.g.  xfs, nfs) won't have
this problem.

Link: https://lkml.kernel.org/r/20231213062324.739009-1-libaokun1@huawei.com
Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: yangerkun <yangerkun@huawei.com>
Cc: Yu Kuai <yukuai3@huawei.com>
Cc: Zhang Yi <yi.zhang@huawei.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests: secretmem: floor the memory size to the multiple of page_size

commit 0aac13add26d546ac74c89d2883b3a5f0fbea039 upstream.

The "locked-in-memory size" limit per process can be non-multiple of
page_size.  The mmap() fails if we try to allocate locked-in-memory with
same size as the allowed limit if it isn't multiple of the page_size
because mmap() rounds off the memory size to be allocated to next multiple
of page_size.

Fix this by flooring the length to be allocated with mmap() to the
previous multiple of the page_size.

This was getting triggered on KernelCI regularly because of different
ulimit settings which wasn't multiple of the page_size.  Find logs
here: https://linux.kernelci.org/test/plan/id/657654bd8e81e654fae13532/
The bug in was present from the time test was first added.

Link: https://lkml.kernel.org/r/20231214101931.1155586-1-usama.anjum@collabora.com
Fixes: 76fe17ef588a ("secretmem: test: add basic selftest for memfd_secret(2)")
Signed-off-by: Muhammad Usama Anjum <usama.anjum@collabora.com>
Reported-by: "kernelci.org bot" <bot@kernelci.org>
Closes: https://linux.kernelci.org/test/plan/id/657654bd8e81e654fae13532/
Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

maple_tree: do not preallocate nodes for slot stores

commit 4249f13c11be8b8b7bf93204185e150c3bdc968d upstream.

mas_preallocate() defaults to requesting 1 node for preallocation and then
,depending on the type of store, will update the request variable.  There
isn't a check for a slot store type, so slot stores are preallocating the
default 1 node.  Slot stores do not require any additional nodes, so add a
check for the slot store case that will bypass node_count_gfp().  Update
the tests to reflect that slot stores do not require allocations.

User visible effects of this bug include increased memory usage from the
unneeded node that was allocated.

Link: https://lkml.kernel.org/r/20231213205058.386589-1-sidhartha.kumar@oracle.com
Fixes: 0b8bb544b1a7 ("maple_tree: update mas_preallocate() testing")
Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Peng Zhang <zhangpeng.00@bytedance.com>
Cc: <stable@vger.kernel.org> [6.6+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

platform/x86: p2sb: Allow p2sb_bar() calls during PCI device probe

commit b28ff7a7c3245d7f62acc20f15b4361292fe4117 upstream.

p2sb_bar() unhides P2SB device to get resources from the device. It
guards the operation by locking pci_rescan_remove_lock so that parallel
rescans do not find the P2SB device. However, this lock causes deadlock
when PCI bus rescan is triggered by /sys/bus/pci/rescan. The rescan
locks pci_rescan_remove_lock and probes PCI devices. When PCI devices
call p2sb_bar() during probe, it locks pci_rescan_remove_lock again.
Hence the deadlock.

To avoid the deadlock, do not lock pci_rescan_remove_lock in p2sb_bar().
Instead, do the lock at fs_initcall. Introduce p2sb_cache_resources()
for fs_initcall which gets and caches the P2SB resources. At p2sb_bar(),
refer the cache and return to the caller.

Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Fixes: 9745fb07474f ("platform/x86/intel: Add Primary to Sideband (P2SB) bridge support")
Cc: stable@vger.kernel.org
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/linux-pci/6xb24fjmptxxn5js2fjrrddjae6twex5bjaftwqsuawuqqqydx@7cl3uik5ef6j/
Link: https://lore.kernel.org/r/20231229063912.2517922-2-shinichiro.kawasaki@wdc.com
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ksmbd: fix slab-out-of-bounds in smb_strndup_from_utf16()

commit d10c77873ba1e9e6b91905018e29e196fd5f863d upstream.

If ->NameOffset/Length is bigger than ->CreateContextsOffset/Length,
ksmbd_check_message doesn't validate request buffer it correctly.
So slab-out-of-bounds warning from calling smb_strndup_from_utf16()
in smb2_open() could happen. If ->NameLength is non-zero, Set the larger
of the two sums (Name and CreateContext size) as the offset and length of
the data area.

Reported-by: Yang Chaoming <lometsj@live.com>
Cc: stable@vger.kernel.org
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

platform/x86/intel/pmc: Move GBE LTR ignore to suspend callback

[ Upstream commit 70681aa0746ae61d7668b9f651221fad5e30c71e ]

Commit 804951203aa5 ("platform/x86:intel/pmc: Combine core_init() and
core_configure()") caused a network performance regression due to the GBE
LTR ignore that it added at probe. This was needed in order to allow the
SoC to enter the deepest Package C state. To fix the regression and at
least support PC10 during suspend, move the LTR ignore from probe to the
suspend callback, and enable it again on resume. This solution will allow
PC10 during suspend but restrict Package C entry at runtime to no deeper
than PC8/9 while a network cable it attach to the PCH LAN.

Fixes: 804951203aa5 ("platform/x86:intel/pmc: Combine core_init() and core_configure()")
Signed-off-by: "David E. Box" <david.e.box@linux.intel.com>
Link: https://lore.kernel.org/r/20231223032548.1680738-6-david.e.box@linux.intel.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/intel/pmc: Allow reenabling LTRs

[ Upstream commit 6f9cc5c1f94daa98846b2073733d03ced709704b ]

Commit 804951203aa5 ("platform/x86:intel/pmc: Combine core_init() and
core_configure()") caused a network performance regression due to the GBE
LTR ignore that it added during probe. The fix will move the ignore to
occur at suspend-time (so as to not affect suspend power). This will
require the ability to enable the LTR again on resume. Modify
pmc_core_send_ltr_ignore() to allow enabling an LTR.

Fixes: 804951203aa5 ("platform/x86:intel/pmc: Combine core_init() and core_configure()")
Signed-off-by: "David E. Box" <david.e.box@linux.intel.com>
Link: https://lore.kernel.org/r/20231223032548.1680738-5-david.e.box@linux.intel.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/intel/pmc: Add suspend callback

[ Upstream commit 7c13f365aee68b01e7e68ee293a71fdc7571c111 ]

Add a suspend callback to struct pmc for performing platform specific tasks
before device suspend. This is needed in order to perform GBE LTR ignore on
certain platforms at suspend-time instead of at probe-time and replace the
GBE LTR ignore removal that was done in order to fix a bug introduced by
commit 804951203aa5 ("platform/x86:intel/pmc: Combine core_init() and
core_configure()").

Fixes: 804951203aa5 ("platform/x86:intel/pmc: Combine core_init() and core_configure()")
Signed-off-by: "David E. Box" <david.e.box@linux.intel.com>
Link: https://lore.kernel.org/r/20231223032548.1680738-4-david.e.box@linux.intel.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

block: renumber QUEUE_FLAG_HW_WC

[ Upstream commit 02d374f3418df577c850f0cd45c3da9245ead547 ]

For the QUEUE_FLAG_HW_WC to actually work, it needs to have a separate
number from QUEUE_FLAG_FUA, doh.

Fixes: 43c9835b144c ("block: don't allow enabling a cache on devices that don't support it")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20231226081524.180289-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>