mm: migrate: annotate data-race in migrate_folio_unmap()

commit 8001070cfbec5cd4ea00b8b48ea51df91122f265 upstream.

I found a report from syzbot [1]

This report shows that the value can be changed, but in reality, the
value of __folio_set_movable() cannot be changed because it holds the
folio refcount.

Therefore, it is appropriate to add an annotate to make KCSAN
ignore that data-race.

[1]

==================================================================
BUG: KCSAN: data-race in __filemap_remove_folio / migrate_pages_batch

write to 0xffffea0004b81dd8 of 8 bytes by task 6348 on cpu 0:
 page_cache_delete mm/filemap.c:153 [inline]
 __filemap_remove_folio+0x1ac/0x2c0 mm/filemap.c:233
 filemap_remove_folio+0x6b/0x1f0 mm/filemap.c:265
 truncate_inode_folio+0x42/0x50 mm/truncate.c:178
 shmem_undo_range+0x25b/0xa70 mm/shmem.c:1028
 shmem_truncate_range mm/shmem.c:1144 [inline]
 shmem_evict_inode+0x14d/0x530 mm/shmem.c:1272
 evict+0x2f0/0x580 fs/inode.c:731
 iput_final fs/inode.c:1883 [inline]
 iput+0x42a/0x5b0 fs/inode.c:1909
 dentry_unlink_inode+0x24f/0x260 fs/dcache.c:412
 __dentry_kill+0x18b/0x4c0 fs/dcache.c:615
 dput+0x5c/0xd0 fs/dcache.c:857
 __fput+0x3fb/0x6d0 fs/file_table.c:439
 ____fput+0x1c/0x30 fs/file_table.c:459
 task_work_run+0x13a/0x1a0 kernel/task_work.c:228
 resume_user_mode_work include/linux/resume_user_mode.h:50 [inline]
 exit_to_user_mode_loop kernel/entry/common.c:114 [inline]
 exit_to_user_mode_prepare include/linux/entry-common.h:328 [inline]
 __syscall_exit_to_user_mode_work kernel/entry/common.c:207 [inline]
 syscall_exit_to_user_mode+0xbe/0x130 kernel/entry/common.c:218
 do_syscall_64+0xd6/0x1c0 arch/x86/entry/common.c:89
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

read to 0xffffea0004b81dd8 of 8 bytes by task 6342 on cpu 1:
 __folio_test_movable include/linux/page-flags.h:699 [inline]
 migrate_folio_unmap mm/migrate.c:1199 [inline]
 migrate_pages_batch+0x24c/0x1940 mm/migrate.c:1797
 migrate_pages_sync mm/migrate.c:1963 [inline]
 migrate_pages+0xff1/0x1820 mm/migrate.c:2072
 do_mbind mm/mempolicy.c:1390 [inline]
 kernel_mbind mm/mempolicy.c:1533 [inline]
 __do_sys_mbind mm/mempolicy.c:1607 [inline]
 __se_sys_mbind+0xf76/0x1160 mm/mempolicy.c:1603
 __x64_sys_mbind+0x78/0x90 mm/mempolicy.c:1603
 x64_sys_call+0x2b4d/0x2d60 arch/x86/include/generated/asm/syscalls_64.h:238
 do_syscall_x64 arch/x86/entry/common.c:52 [inline]
 do_syscall_64+0xc9/0x1c0 arch/x86/entry/common.c:83
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

value changed: 0xffff888127601078 -> 0x0000000000000000

Link: https://lkml.kernel.org/r/20240924130053.107490-1-aha310510@gmail.com
Fixes: 7e2a5e5ab217 ("mm: migrate: use __folio_test_movable()")
Signed-off-by: Jeongjun Park <aha310510@gmail.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Zi Yan <ziy@nvidia.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fuse: use exclusive lock when FUSE_I_CACHE_IO_MODE is set

commit 2f3d8ff457982f4055fe8f7bf19d3821ba22c376 upstream.

This may be a typo. The comment has said shared locks are
not allowed when this bit is set. If using shared lock, the
wait in `fuse_file_cached_io_open` may be forever.

Fixes: 205c1d802683 ("fuse: allow parallel dio writes with FUSE_DIRECT_IO_ALLOW_MMAP")
CC: stable@vger.kernel.org # v6.9
Signed-off-by: yangyun <yangyun50@huawei.com>
Reviewed-by: Bernd Schubert <bschubert@ddn.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring: check for presence of task_work rather than TIF_NOTIFY_SIGNAL

commit 04beb6e0e08c30c6f845f50afb7d7953603d7a6f upstream.

If some part of the kernel adds task_work that needs executing, in terms
of signaling it'll generally use TWA_SIGNAL or TWA_RESUME. Those two
directly translate to TIF_NOTIFY_SIGNAL or TIF_NOTIFY_RESUME, and can
be used for a variety of use case outside of task_work.

However, io_cqring_wait_schedule() only tests explicitly for
TIF_NOTIFY_SIGNAL. This means it can miss if task_work got added for
the task, but used a different kind of signaling mechanism (or none at
all). Normally this doesn't matter as any task_work will be run once
the task exits to userspace, except if:

1) The ring is setup with DEFER_TASKRUN
2) The local work item may generate normal task_work

For condition 2, this can happen when closing a file and it's the final
put of that file, for example. This can cause stalls where a task is
waiting to make progress inside io_cqring_wait(), but there's nothing else
that will wake it up. Hence change the "should we schedule or loop around"
check to check for the presence of task_work explicitly, rather than just
TIF_NOTIFY_SIGNAL as the mechanism. While in there, also change the
ordering of what type of task_work first in terms of ordering, to both
make it consistent with other task_work runs in io_uring, but also to
better handle the case of defer task_work generating normal task_work,
like in the above example.

Reported-by: Jan Hendrik Farr <kernel@jfarr.cc>
Link: https://github.com/axboe/liburing/issues/1235
Cc: stable@vger.kernel.org
Fixes: 846072f16eed ("io_uring: mimimise io_cqring_wait_schedule")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring/rw: treat -EOPNOTSUPP for IOCB_NOWAIT like -EAGAIN

commit c0a9d496e0fece67db777bd48550376cf2960c47 upstream.

Some file systems, ocfs2 in this case, will return -EOPNOTSUPP for
an IOCB_NOWAIT read/write attempt. While this can be argued to be
correct, the usual return value for something that requires blocking
issue is -EAGAIN.

A refactoring io_uring commit dropped calling kiocb_done() for
negative return values, which is otherwise where we already do that
transformation. To ensure we catch it in both spots, check it in
__io_read() itself as well.

Reported-by: Robert Sander <r.sander@heinlein-support.de>
Link: https://fosstodon.org/@gurubert@mastodon.gurubert.de/113112431889638440
Cc: stable@vger.kernel.org
Fixes: a08d195b586a ("io_uring/rw: split io_read() into a helper")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring/sqpoll: do not allow pinning outside of cpuset

commit f011c9cf04c06f16b24f583d313d3c012e589e50 upstream.

The submit queue polling threads are userland threads that just never
exit to the userland. When creating the thread with IORING_SETUP_SQ_AFF,
the affinity of the poller thread is set to the cpu specified in
sq_thread_cpu. However, this CPU can be outside of the cpuset defined
by the cgroup cpuset controller. This violates the rules defined by the
cpuset controller and is a potential issue for realtime applications.

In b7ed6d8ffd6 we fixed the default affinity of the poller thread, in
case no explicit pinning is required by inheriting the one of the
creating task. In case of explicit pinning, the check is more
complicated, as also a cpu outside of the parent cpumask is allowed.
We implemented this by using cpuset_cpus_allowed (that has support for
cgroup cpusets) and testing if the requested cpu is in the set.

Fixes: 37d1e2e3642e ("io_uring: move SQPOLL thread io-wq forked worker")
Cc: stable@vger.kernel.org # 6.1+
Signed-off-by: Felix Moessbauer <felix.moessbauer@siemens.com>
Link: https://lore.kernel.org/r/20240909150036.55921-1-felix.moessbauer@siemens.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests: netfilter: Avoid hanging ipvs.sh

[ Upstream commit fc786304ad9803e8bb86b8599bc64d1c1746c75f ]

If the client can't reach the server, the latter remains listening
forever. Kill it after 5s of waiting.

Fixes: 867d2190799a ("selftests: netfilter: add ipvs test script")
Signed-off-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_tables: missing objects with no memcg accounting

[ Upstream commit 69e687cea79fc99a17dfb0116c8644b9391b915e ]

Several ruleset objects are still not using GFP_KERNEL_ACCOUNT for
memory accounting, update them. This includes:

- catchall elements
- compat match large info area
- log prefix
- meta secctx
- numgen counters
- pipapo set backend datastructure
- tunnel private objects

Fixes: 33758c891479 ("memcg: enable accounting for nft objects")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_tables: use rcu chain hook list iterator from netlink dump path

[ Upstream commit 4ffcf5ca81c3b83180473eb0d3c010a1a7c6c4de ]

Lockless iteration over hook list is possible from netlink dump path,
use rcu variant to iterate over the hook list as is done with flowtable
hooks.

Fixes: b9703ed44ffb ("netfilter: nf_tables: support for adding new devices to an existing netdev chain")
Reported-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: ctnetlink: compile ctnetlink_label_size with CONFIG_NF_CONNTRACK_EVENTS

[ Upstream commit e1f1ee0e9ad8cbe660f5c104e791c5f1a7cf4c31 ]

Only provide ctnetlink_label_size when it is used,
which is when CONFIG_NF_CONNTRACK_EVENTS is configured.

Flagged by clang-18 W=1 builds as:

.../nf_conntrack_netlink.c:385:19: warning: unused function 'ctnetlink_label_size' [-Wunused-function]
  385 | static inline int ctnetlink_label_size(const struct nf_conn *ct)
      |                   ^~~~~~~~~~~~~~~~~~~~

The condition on CONFIG_NF_CONNTRACK_LABELS being removed by
this patch guards compilation of non-trivial implementations
of ctnetlink_dump_labels() and ctnetlink_label_size().

However, this is not necessary as each of these functions
will always return 0 if CONFIG_NF_CONNTRACK_LABELS is not defined
as each function starts with the equivalent of:

struct nf_conn_labels *labels = nf_ct_labels_find(ct);

if (!labels)
return 0;

And nf_ct_labels_find always returns NULL if CONFIG_NF_CONNTRACK_LABELS
is not enabled.  So I believe that the compiler optimises the code away
in such cases anyway.

Found by inspection.
Compile tested only.

Originally splitted in two patches, Pablo Neira Ayuso collapsed them and
added Fixes: tag.

Fixes: 0ceabd83875b ("netfilter: ctnetlink: deliver labels to userspace")
Link: https://lore.kernel.org/netfilter-devel/20240909151712.GZ2097826@kernel.org/
Signed-off-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_tables: Keep deleted flowtable hooks until after RCU

[ Upstream commit 642c89c475419b4d0c0d90e29d9c1a0e4351f379 ]

Documentation of list_del_rcu() warns callers to not immediately free
the deleted list item. While it seems not necessary to use the
RCU-variant of list_del() here in the first place, doing so seems to
require calling kfree_rcu() on the deleted item as well.

Fixes: 3f0465a9ef02 ("netfilter: nf_tables: dynamically allocate hooks per net_device in flowtables")
Signed-off-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: stmmac: set PP_FLAG_DMA_SYNC_DEV only if XDP is enabled

[ Upstream commit b514c47ebf41a6536551ed28a05758036e6eca7c ]

Commit 5fabb01207a2 ("net: stmmac: Add initial XDP support") sets
PP_FLAG_DMA_SYNC_DEV flag for page_pool unconditionally,
page_pool_recycle_direct() will call page_pool_dma_sync_for_device()
on every page even the page is not going to be reused by XDP program.

When XDP is not enabled, the page which holds the received buffer
will be recycled once the buffer is copied into new SKB by
skb_copy_to_linear_data(), then the MAC core will never reuse this
page any longer. Always setting PP_FLAG_DMA_SYNC_DEV wastes CPU cycles
on unnecessary calling of page_pool_dma_sync_for_device().

After this patch, up to 9% noticeable performance improvement was observed
on certain platforms.

Fixes: 5fabb01207a2 ("net: stmmac: Add initial XDP support")
Signed-off-by: Furong Xu <0x1207@gmail.com>
Link: https://patch.msgid.link/20240919121028.1348023-1-0x1207@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

virtio_net: Fix mismatched buf address when unmapping for small packets

[ Upstream commit c11a49d58ad229a1be1ebe08a2b68fedf83db6c8 ]

Currently, the virtio-net driver will perform a pre-dma-mapping for
small or mergeable RX buffer. But for small packets, a mismatched address
without VIRTNET_RX_PAD and xdp_headroom is used for unmapping.

That will result in unsynchronized buffers when SWIOTLB is enabled, for
example, when running as a TDX guest.

This patch unifies the address passed to the virtio core as the address of
the virtnet header and fixes the mismatched buffer address.

Changes from v2: unify the buf that passed to the virtio core in small
and merge mode.
Changes from v1: Use ctx to get xdp_headroom.

Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers")
Signed-off-by: Wenbo Li <liwenbo.martin@bytedance.com>
Signed-off-by: Jiahui Cen <cenjiahui@bytedance.com>
Signed-off-by: Ying Fang <fangying.tommy@bytedance.com>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20240919081351.51772-1-liwenbo.martin@bytedance.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bonding: Fix unnecessary warnings and logs from bond_xdp_get_xmit_slave()

[ Upstream commit 0cbfd45fbcf0cb26d85c981b91c62fe73cdee01c ]

syzbot reported a WARNING in bond_xdp_get_xmit_slave. To reproduce
this[1], one bond device (bond1) has xdpdrv, which increases
bpf_master_redirect_enabled_key. Another bond device (bond0) which is
unsupported by XDP but its slave (veth3) has xdpgeneric that returns
XDP_TX. This triggers WARN_ON_ONCE() from the xdp_master_redirect().
To reduce unnecessary warnings and improve log management, we need to
delete the WARN_ON_ONCE() and add ratelimit to the netdev_err().

[1] Steps to reproduce:
    # Needs tx_xdp with return XDP_TX;
    ip l add veth0 type veth peer veth1
    ip l add veth3 type veth peer veth4
    ip l add bond0 type bond mode 6 # BOND_MODE_ALB, unsupported by XDP
    ip l add bond1 type bond # BOND_MODE_ROUNDROBIN by default
    ip l set veth0 master bond1
    ip l set bond1 up
    # Increases bpf_master_redirect_enabled_key
    ip l set dev bond1 xdpdrv object tx_xdp.o section xdp_tx
    ip l set veth3 master bond0
    ip l set bond0 up
    ip l set veth4 up
    # Triggers WARN_ON_ONCE() from the xdp_master_redirect()
    ip l set veth3 xdpgeneric object tx_xdp.o section xdp_tx

Reported-by: syzbot+c187823a52ed505b2257@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=c187823a52ed505b2257
Fixes: 9e2ee5c7e7c3 ("net, bonding: Add XDP support to the bonding driver")
Signed-off-by: Jiwon Kim <jiwonaid0@gmail.com>
Signed-off-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/20240918140602.18644-1-jiwonaid0@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: ravb: Fix R-Car RX frame size limit

[ Upstream commit ec8234717db8589078d08b17efa528a235c61f4f ]

The RX frame size limit should not be based on the current MTU setting.
Instead it should be based on the hardware capabilities.

While we're here, improve the description of the receive frame length
setting as suggested by Niklas.

Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: qrtr: Update packets cloning when broadcasting

[ Upstream commit f011b313e8ebd5b7abd8521b5119aecef403de45 ]

When broadcasting data to multiple nodes via MHI, using skb_clone()
causes all nodes to receive the same header data. This can result in
packets being discarded by endpoints, leading to lost data.

This issue occurs when a socket is closed, and a QRTR_TYPE_DEL_CLIENT
packet is broadcasted. All nodes receive the same destination node ID,
causing the node connected to the client to discard the packet and
remain unaware of the client's deletion.

Replace skb_clone() with pskb_copy(), to create a separate copy of
the header for each sk_buff.

Fixes: bdabad3e363d ("net: Add Qualcomm IPC router")
Signed-off-by: Youssef Samir <quic_yabdulra@quicinc.com>
Reviewed-by: Jeffery Hugo <quic_jhugo@quicinc.com>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Reviewed-by: Chris Lew <quic_clew@quicinc.com>
Link: https://patch.msgid.link/20240916170858.2382247-1-quic_yabdulra@quicinc.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tcp: check skb is non-NULL in tcp_rto_delta_us()

[ Upstream commit c8770db2d54437a5f49417ae7b46f7de23d14db6 ]

We have some machines running stock Ubuntu 20.04.6 which is their 5.4.0-174-generic
kernel that are running ceph and recently hit a null ptr dereference in
tcp_rearm_rto(). Initially hitting it from the TLP path, but then later we also
saw it getting hit from the RACK case as well. Here are examples of the oops
messages we saw in each of those cases:

Jul 26 15:05:02 rx [11061395.780353] BUG: kernel NULL pointer dereference, address: 0000000000000020
Jul 26 15:05:02 rx [11061395.787572] #PF: supervisor read access in kernel mode
Jul 26 15:05:02 rx [11061395.792971] #PF: error_code(0x0000) - not-present page
Jul 26 15:05:02 rx [11061395.798362] PGD 0 P4D 0
Jul 26 15:05:02 rx [11061395.801164] Oops: 0000 [#1] SMP NOPTI
Jul 26 15:05:02 rx [11061395.805091] CPU: 0 PID: 9180 Comm: msgr-worker-1 Tainted: G W 5.4.0-174-generic #193-Ubuntu
Jul 26 15:05:02 rx [11061395.814996] Hardware name: Supermicro SMC 2x26 os-gen8 64C NVME-Y 256G/H12SSW-NTR, BIOS 2.5.V1.2U.NVMe.UEFI 05/09/2023
Jul 26 15:05:02 rx [11061395.825952] RIP: 0010:tcp_rearm_rto+0xe4/0x160
Jul 26 15:05:02 rx [11061395.830656] Code: 87 ca 04 00 00 00 5b 41 5c 41 5d 5d c3 c3 49 8b bc 24 40 06 00 00 eb 8d 48 bb cf f7 53 e3 a5 9b c4 20 4c 89 ef e8 0c fe 0e 00 <48> 8b 78 20 48 c1 ef 03 48 89 f8 41 8b bc 24 80 04 00 00 48 f7 e3
Jul 26 15:05:02 rx [11061395.849665] RSP: 0018:ffffb75d40003e08 EFLAGS: 00010246
Jul 26 15:05:02 rx [11061395.855149] RAX: 0000000000000000 RBX: 20c49ba5e353f7cf RCX: 0000000000000000
Jul 26 15:05:02 rx [11061395.862542] RDX: 0000000062177c30 RSI: 000000000000231c RDI: ffff9874ad283a60
Jul 26 15:05:02 rx [11061395.869933] RBP: ffffb75d40003e20 R08: 0000000000000000 R09: ffff987605e20aa8
Jul 26 15:05:02 rx [11061395.877318] R10: ffffb75d40003f00 R11: ffffb75d4460f740 R12: ffff9874ad283900
Jul 26 15:05:02 rx [11061395.884710] R13: ffff9874ad283a60 R14: ffff9874ad283980 R15: ffff9874ad283d30
Jul 26 15:05:02 rx [11061395.892095] FS: 00007f1ef4a2e700(0000) GS:ffff987605e00000(0000) knlGS:0000000000000000
Jul 26 15:05:02 rx [11061395.900438] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jul 26 15:05:02 rx [11061395.906435] CR2: 0000000000000020 CR3: 0000003e450ba003 CR4: 0000000000760ef0
Jul 26 15:05:02 rx [11061395.913822] PKRU: 55555554
Jul 26 15:05:02 rx [11061395.916786] Call Trace:
Jul 26 15:05:02 rx [11061395.919488]
Jul 26 15:05:02 rx [11061395.921765] ? show_regs.cold+0x1a/0x1f
Jul 26 15:05:02 rx [11061395.925859] ? __die+0x90/0xd9
Jul 26 15:05:02 rx [11061395.929169] ? no_context+0x196/0x380
Jul 26 15:05:02 rx [11061395.933088] ? ip6_protocol_deliver_rcu+0x4e0/0x4e0
Jul 26 15:05:02 rx [11061395.938216] ? ip6_sublist_rcv_finish+0x3d/0x50
Jul 26 15:05:02 rx [11061395.943000] ? __bad_area_nosemaphore+0x50/0x1a0
Jul 26 15:05:02 rx [11061395.947873] ? bad_area_nosemaphore+0x16/0x20
Jul 26 15:05:02 rx [11061395.952486] ? do_user_addr_fault+0x267/0x450
Jul 26 15:05:02 rx [11061395.957104] ? ipv6_list_rcv+0x112/0x140
Jul 26 15:05:02 rx [11061395.961279] ? __do_page_fault+0x58/0x90
Jul 26 15:05:02 rx [11061395.965458] ? do_page_fault+0x2c/0xe0
Jul 26 15:05:02 rx [11061395.969465] ? page_fault+0x34/0x40
Jul 26 15:05:02 rx [11061395.973217] ? tcp_rearm_rto+0xe4/0x160
Jul 26 15:05:02 rx [11061395.977313] ? tcp_rearm_rto+0xe4/0x160
Jul 26 15:05:02 rx [11061395.981408] tcp_send_loss_probe+0x10b/0x220
Jul 26 15:05:02 rx [11061395.985937] tcp_write_timer_handler+0x1b4/0x240
Jul 26 15:05:02 rx [11061395.990809] tcp_write_timer+0x9e/0xe0
Jul 26 15:05:02 rx [11061395.994814] ? tcp_write_timer_handler+0x240/0x240
Jul 26 15:05:02 rx [11061395.999866] call_timer_fn+0x32/0x130
Jul 26 15:05:02 rx [11061396.003782] __run_timers.part.0+0x180/0x280
Jul 26 15:05:02 rx [11061396.008309] ? recalibrate_cpu_khz+0x10/0x10
Jul 26 15:05:02 rx [11061396.012841] ? native_x2apic_icr_write+0x30/0x30
Jul 26 15:05:02 rx [11061396.017718] ? lapic_next_event+0x21/0x30
Jul 26 15:05:02 rx [11061396.021984] ? clockevents_program_event+0x8f/0xe0
Jul 26 15:05:02 rx [11061396.027035] run_timer_softirq+0x2a/0x50
Jul 26 15:05:02 rx [11061396.031212] __do_softirq+0xd1/0x2c1
Jul 26 15:05:02 rx [11061396.035044] do_softirq_own_stack+0x2a/0x40
Jul 26 15:05:02 rx [11061396.039480]
Jul 26 15:05:02 rx [11061396.041840] do_softirq.part.0+0x46/0x50
Jul 26 15:05:02 rx [11061396.046022] __local_bh_enable_ip+0x50/0x60
Jul 26 15:05:02 rx [11061396.050460] _raw_spin_unlock_bh+0x1e/0x20
Jul 26 15:05:02 rx [11061396.054817] nf_conntrack_tcp_packet+0x29e/0xbe0 [nf_conntrack]
Jul 26 15:05:02 rx [11061396.060994] ? get_l4proto+0xe7/0x190 [nf_conntrack]
Jul 26 15:05:02 rx [11061396.066220] nf_conntrack_in+0xe9/0x670 [nf_conntrack]
Jul 26 15:05:02 rx [11061396.071618] ipv6_conntrack_local+0x14/0x20 [nf_conntrack]
Jul 26 15:05:02 rx [11061396.077356] nf_hook_slow+0x45/0xb0
Jul 26 15:05:02 rx [11061396.081098] ip6_xmit+0x3f0/0x5d0
Jul 26 15:05:02 rx [11061396.084670] ? ipv6_anycast_cleanup+0x50/0x50
Jul 26 15:05:02 rx [11061396.089282] ? __sk_dst_check+0x38/0x70
Jul 26 15:05:02 rx [11061396.093381] ? inet6_csk_route_socket+0x13b/0x200
Jul 26 15:05:02 rx [11061396.098346] inet6_csk_xmit+0xa7/0xf0
Jul 26 15:05:02 rx [11061396.102263] __tcp_transmit_skb+0x550/0xb30
Jul 26 15:05:02 rx [11061396.106701] tcp_write_xmit+0x3c6/0xc20
Jul 26 15:05:02 rx [11061396.110792] ? __alloc_skb+0x98/0x1d0
Jul 26 15:05:02 rx [11061396.114708] __tcp_push_pending_frames+0x37/0x100
Jul 26 15:05:02 rx [11061396.119667] tcp_push+0xfd/0x100
Jul 26 15:05:02 rx [11061396.123150] tcp_sendmsg_locked+0xc70/0xdd0
Jul 26 15:05:02 rx [11061396.127588] tcp_sendmsg+0x2d/0x50
Jul 26 15:05:02 rx [11061396.131245] inet6_sendmsg+0x43/0x70
Jul 26 15:05:02 rx [11061396.135075] __sock_sendmsg+0x48/0x70
Jul 26 15:05:02 rx [11061396.138994] ____sys_sendmsg+0x212/0x280
Jul 26 15:05:02 rx [11061396.143172] ___sys_sendmsg+0x88/0xd0
Jul 26 15:05:02 rx [11061396.147098] ? __seccomp_filter+0x7e/0x6b0
Jul 26 15:05:02 rx [11061396.151446] ? __switch_to+0x39c/0x460
Jul 26 15:05:02 rx [11061396.155453] ? __switch_to_asm+0x42/0x80
Jul 26 15:05:02 rx [11061396.159636] ? __switch_to_asm+0x5a/0x80
Jul 26 15:05:02 rx [11061396.163816] __sys_sendmsg+0x5c/0xa0
Jul 26 15:05:02 rx [11061396.167647] __x64_sys_sendmsg+0x1f/0x30
Jul 26 15:05:02 rx [11061396.171832] do_syscall_64+0x57/0x190
Jul 26 15:05:02 rx [11061396.175748] entry_SYSCALL_64_after_hwframe+0x5c/0xc1
Jul 26 15:05:02 rx [11061396.181055] RIP: 0033:0x7f1ef692618d
Jul 26 15:05:02 rx [11061396.184893] Code: 28 89 54 24 1c 48 89 74 24 10 89 7c 24 08 e8 ca ee ff ff 8b 54 24 1c 48 8b 74 24 10 41 89 c0 8b 7c 24 08 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 2f 44 89 c7 48 89 44 24 08 e8 fe ee ff ff 48
Jul 26 15:05:02 rx [11061396.203889] RSP: 002b:00007f1ef4a26aa0 EFLAGS: 00000293 ORIG_RAX: 000000000000002e
Jul 26 15:05:02 rx [11061396.211708] RAX: ffffffffffffffda RBX: 000000000000084b RCX: 00007f1ef692618d
Jul 26 15:05:02 rx [11061396.219091] RDX: 0000000000004000 RSI: 00007f1ef4a26b10 RDI: 0000000000000275
Jul 26 15:05:02 rx [11061396.226475] RBP: 0000000000004000 R08: 0000000000000000 R09: 0000000000000020
Jul 26 15:05:02 rx [11061396.233859] R10: 0000000000000000 R11: 0000000000000293 R12: 000000000000084b
Jul 26 15:05:02 rx [11061396.241243] R13: 00007f1ef4a26b10 R14: 0000000000000275 R15: 000055592030f1e8
Jul 26 15:05:02 rx [11061396.248628] Modules linked in: vrf bridge stp llc vxlan ip6_udp_tunnel udp_tunnel nls_iso8859_1 amd64_edac_mod edac_mce_amd kvm_amd kvm crct10dif_pclmul ghash_clmulni_intel aesni_intel crypto_simd cryptd glue_helper wmi_bmof ipmi_ssif input_leds joydev rndis_host cdc_ether usbnet mii ast drm_vram_helper ttm drm_kms_helper i2c_algo_bit fb_sys_fops syscopyarea sysfillrect sysimgblt ccp mac_hid ipmi_si ipmi_devintf ipmi_msghandler nft_ct sch_fq_codel nf_tables_set nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 nf_tables nfnetlink ramoops reed_solomon efi_pstore drm ip_tables x_tables autofs4 raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid0 multipath linear mlx5_ib ib_uverbs ib_core raid1 mlx5_core hid_generic pci_hyperv_intf crc32_pclmul tls usbhid ahci mlxfw bnxt_en libahci hid nvme i2c_piix4 nvme_core wmi
Jul 26 15:05:02 rx [11061396.324334] CR2: 0000000000000020
Jul 26 15:05:02 rx [11061396.327944] ---[ end trace 68a2b679d1cfb4f1 ]---
Jul 26 15:05:02 rx [11061396.433435] RIP: 0010:tcp_rearm_rto+0xe4/0x160
Jul 26 15:05:02 rx [11061396.438137] Code: 87 ca 04 00 00 00 5b 41 5c 41 5d 5d c3 c3 49 8b bc 24 40 06 00 00 eb 8d 48 bb cf f7 53 e3 a5 9b c4 20 4c 89 ef e8 0c fe 0e 00 <48> 8b 78 20 48 c1 ef 03 48 89 f8 41 8b bc 24 80 04 00 00 48 f7 e3
Jul 26 15:05:02 rx [11061396.457144] RSP: 0018:ffffb75d40003e08 EFLAGS: 00010246
Jul 26 15:05:02 rx [11061396.462629] RAX: 0000000000000000 RBX: 20c49ba5e353f7cf RCX: 0000000000000000
Jul 26 15:05:02 rx [11061396.470012] RDX: 0000000062177c30 RSI: 000000000000231c RDI: ffff9874ad283a60
Jul 26 15:05:02 rx [11061396.477396] RBP: ffffb75d40003e20 R08: 0000000000000000 R09: ffff987605e20aa8
Jul 26 15:05:02 rx [11061396.484779] R10: ffffb75d40003f00 R11: ffffb75d4460f740 R12: ffff9874ad283900
Jul 26 15:05:02 rx [11061396.492164] R13: ffff9874ad283a60 R14: ffff9874ad283980 R15: ffff9874ad283d30
Jul 26 15:05:02 rx [11061396.499547] FS: 00007f1ef4a2e700(0000) GS:ffff987605e00000(0000) knlGS:0000000000000000
Jul 26 15:05:02 rx [11061396.507886] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jul 26 15:05:02 rx [11061396.513884] CR2: 0000000000000020 CR3: 0000003e450ba003 CR4: 0000000000760ef0
Jul 26 15:05:02 rx [11061396.521267] PKRU: 55555554
Jul 26 15:05:02 rx [11061396.524230] Kernel panic - not syncing: Fatal exception in interrupt
Jul 26 15:05:02 rx [11061396.530885] Kernel Offset: 0x1b200000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
Jul 26 15:05:03 rx [11061396.660181] ---[ end Kernel panic - not syncing: Fatal
 exception in interrupt ]---

After we hit this we disabled TLP by setting tcp_early_retrans to 0 and then hit the crash in the RACK case:

Aug 7 07:26:16 rx [1006006.265582] BUG: kernel NULL pointer dereference, address: 0000000000000020
Aug 7 07:26:16 rx [1006006.272719] #PF: supervisor read access in kernel mode
Aug 7 07:26:16 rx [1006006.278030] #PF: error_code(0x0000) - not-present page
Aug 7 07:26:16 rx [1006006.283343] PGD 0 P4D 0
Aug 7 07:26:16 rx [1006006.286057] Oops: 0000 [#1] SMP NOPTI
Aug 7 07:26:16 rx [1006006.289896] CPU: 5 PID: 0 Comm: swapper/5 Tainted: G W 5.4.0-174-generic #193-Ubuntu
Aug 7 07:26:16 rx [1006006.299107] Hardware name: Supermicro SMC 2x26 os-gen8 64C NVME-Y 256G/H12SSW-NTR, BIOS 2.5.V1.2U.NVMe.UEFI 05/09/2023
Aug 7 07:26:16 rx [1006006.309970] RIP: 0010:tcp_rearm_rto+0xe4/0x160
Aug 7 07:26:16 rx [1006006.314584] Code: 87 ca 04 00 00 00 5b 41 5c 41 5d 5d c3 c3 49 8b bc 24 40 06 00 00 eb 8d 48 bb cf f7 53 e3 a5 9b c4 20 4c 89 ef e8 0c fe 0e 00 <48> 8b 78 20 48 c1 ef 03 48 89 f8 41 8b bc 24 80 04 00 00 48 f7 e3
Aug 7 07:26:16 rx [1006006.333499] RSP: 0018:ffffb42600a50960 EFLAGS: 00010246
Aug 7 07:26:16 rx [1006006.338895] RAX: 0000000000000000 RBX: 20c49ba5e353f7cf RCX: 0000000000000000
Aug 7 07:26:16 rx [1006006.346193] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff92d687ed8160
Aug 7 07:26:16 rx [1006006.353489] RBP: ffffb42600a50978 R08: 0000000000000000 R09: 00000000cd896dcc
Aug 7 07:26:16 rx [1006006.360786] R10: ffff92dc3404f400 R11: 0000000000000001 R12: ffff92d687ed8000
Aug 7 07:26:16 rx [1006006.368084] R13: ffff92d687ed8160 R14: 00000000cd896dcc R15: 00000000cd8fca81
Aug 7 07:26:16 rx [1006006.375381] FS: 0000000000000000(0000) GS:ffff93158ad40000(0000) knlGS:0000000000000000
Aug 7 07:26:16 rx [1006006.383632] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Aug 7 07:26:16 rx [1006006.389544] CR2: 0000000000000020 CR3: 0000003e775ce006 CR4: 0000000000760ee0
Aug 7 07:26:16 rx [1006006.396839] PKRU: 55555554
Aug 7 07:26:16 rx [1006006.399717] Call Trace:
Aug 7 07:26:16 rx [1006006.402335]
Aug 7 07:26:16 rx [1006006.404525] ? show_regs.cold+0x1a/0x1f
Aug 7 07:26:16 rx [1006006.408532] ? __die+0x90/0xd9
Aug 7 07:26:16 rx [1006006.411760] ? no_context+0x196/0x380
Aug 7 07:26:16 rx [1006006.415599] ? __bad_area_nosemaphore+0x50/0x1a0
Aug 7 07:26:16 rx [1006006.420392] ? _raw_spin_lock+0x1e/0x30
Aug 7 07:26:16 rx [1006006.424401] ? bad_area_nosemaphore+0x16/0x20
Aug 7 07:26:16 rx [1006006.428927] ? do_user_addr_fault+0x267/0x450
Aug 7 07:26:16 rx [1006006.433450] ? __do_page_fault+0x58/0x90
Aug 7 07:26:16 rx [1006006.437542] ? do_page_fault+0x2c/0xe0
Aug 7 07:26:16 rx [1006006.441470] ? page_fault+0x34/0x40
Aug 7 07:26:16 rx [1006006.445134] ? tcp_rearm_rto+0xe4/0x160
Aug 7 07:26:16 rx [1006006.449145] tcp_ack+0xa32/0xb30
Aug 7 07:26:16 rx [1006006.452542] tcp_rcv_established+0x13c/0x670
Aug 7 07:26:16 rx [1006006.456981] ? sk_filter_trim_cap+0x48/0x220
Aug 7 07:26:16 rx [1006006.461419] tcp_v6_do_rcv+0xdb/0x450
Aug 7 07:26:16 rx [1006006.465257] tcp_v6_rcv+0xc2b/0xd10
Aug 7 07:26:16 rx [1006006.468918] ip6_protocol_deliver_rcu+0xd3/0x4e0
Aug 7 07:26:16 rx [1006006.473706] ip6_input_finish+0x15/0x20
Aug 7 07:26:16 rx [1006006.477710] ip6_input+0xa2/0xb0
Aug 7 07:26:16 rx [1006006.481109] ? ip6_protocol_deliver_rcu+0x4e0/0x4e0
Aug 7 07:26:16 rx [1006006.486151] ip6_sublist_rcv_finish+0x3d/0x50
Aug 7 07:26:16 rx [1006006.490679] ip6_sublist_rcv+0x1aa/0x250
Aug 7 07:26:16 rx [1006006.494779] ? ip6_rcv_finish_core.isra.0+0xa0/0xa0
Aug 7 07:26:16 rx [1006006.499828] ipv6_list_rcv+0x112/0x140
Aug 7 07:26:16 rx [1006006.503748] __netif_receive_skb_list_core+0x1a4/0x250
Aug 7 07:26:16 rx [1006006.509057] netif_receive_skb_list_internal+0x1a1/0x2b0
Aug 7 07:26:16 rx [1006006.514538] gro_normal_list.part.0+0x1e/0x40
Aug 7 07:26:16 rx [1006006.519068] napi_complete_done+0x91/0x130
Aug 7 07:26:16 rx [1006006.523352] mlx5e_napi_poll+0x18e/0x610 [mlx5_core]
Aug 7 07:26:16 rx [1006006.528481] net_rx_action+0x142/0x390
Aug 7 07:26:16 rx [1006006.532398] __do_softirq+0xd1/0x2c1
Aug 7 07:26:16 rx [1006006.536142] irq_exit+0xae/0xb0
Aug 7 07:26:16 rx [1006006.539452] do_IRQ+0x5a/0xf0
Aug 7 07:26:16 rx [1006006.542590] common_interrupt+0xf/0xf
Aug 7 07:26:16 rx [1006006.546421]
Aug 7 07:26:16 rx [1006006.548695] RIP: 0010:native_safe_halt+0xe/0x10
Aug 7 07:26:16 rx [1006006.553399] Code: 7b ff ff ff eb bd 90 90 90 90 90 90 e9 07 00 00 00 0f 00 2d 36 2c 50 00 f4 c3 66 90 e9 07 00 00 00 0f 00 2d 26 2c 50 00 fb f4 90 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 53 e8 dd 5e 61 ff 65
Aug 7 07:26:16 rx [1006006.572309] RSP: 0018:ffffb42600177e70 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffc2
Aug 7 07:26:16 rx [1006006.580040] RAX: ffffffff8ed08b20 RBX: 0000000000000005 RCX: 0000000000000001
Aug 7 07:26:16 rx [1006006.587337] RDX: 00000000f48eeca2 RSI: 0000000000000082 RDI: 0000000000000082
Aug 7 07:26:16 rx [1006006.594635] RBP: ffffb42600177e90 R08: 0000000000000000 R09: 000000000000020f
Aug 7 07:26:16 rx [1006006.601931] R10: 0000000000100000 R11: 0000000000000000 R12: 0000000000000005
Aug 7 07:26:16 rx [1006006.609229] R13: ffff93157deb5f00 R14: 0000000000000000 R15: 0000000000000000
Aug 7 07:26:16 rx [1006006.616530] ? __cpuidle_text_start+0x8/0x8
Aug 7 07:26:16 rx [1006006.620886] ? default_idle+0x20/0x140
Aug 7 07:26:16 rx [1006006.624804] arch_cpu_idle+0x15/0x20
Aug 7 07:26:16 rx [1006006.628545] default_idle_call+0x23/0x30
Aug 7 07:26:16 rx [1006006.632640] do_idle+0x1fb/0x270
Aug 7 07:26:16 rx [1006006.636035] cpu_startup_entry+0x20/0x30
Aug 7 07:26:16 rx [1006006.640126] start_secondary+0x178/0x1d0
Aug 7 07:26:16 rx [1006006.644218] secondary_startup_64+0xa4/0xb0
Aug 7 07:26:17 rx [1006006.648568] Modules linked in: vrf bridge stp llc vxlan ip6_udp_tunnel udp_tunnel nls_iso8859_1 nft_ct amd64_edac_mod edac_mce_amd kvm_amd kvm crct10dif_pclmul ghash_clmulni_intel aesni_intel crypto_simd cryptd glue_helper wmi_bmof ipmi_ssif input_leds joydev rndis_host cdc_ether usbnet ast mii drm_vram_helper ttm drm_kms_helper i2c_algo_bit fb_sys_fops syscopyarea sysfillrect sysimgblt ccp mac_hid ipmi_si ipmi_devintf ipmi_msghandler sch_fq_codel nf_tables_set nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 nf_tables nfnetlink ramoops reed_solomon efi_pstore drm ip_tables x_tables autofs4 raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid0 multipath linear mlx5_ib ib_uverbs ib_core raid1 hid_generic mlx5_core pci_hyperv_intf crc32_pclmul usbhid ahci tls mlxfw bnxt_en hid libahci nvme i2c_piix4 nvme_core wmi [last unloaded: cpuid]
Aug 7 07:26:17 rx [1006006.726180] CR2: 0000000000000020
Aug 7 07:26:17 rx [1006006.729718] ---[ end trace e0e2e37e4e612984 ]---

Prior to seeing the first crash and on other machines we also see the warning in
tcp_send_loss_probe() where packets_out is non-zero, but both transmit and retrans
queues are empty so we know the box is seeing some accounting issue in this area:

Jul 26 09:15:27 kernel: ------------[ cut here ]------------
Jul 26 09:15:27 kernel: invalid inflight: 2 state 1 cwnd 68 mss 8988
Jul 26 09:15:27 kernel: WARNING: CPU: 16 PID: 0 at net/ipv4/tcp_output.c:2605 tcp_send_loss_probe+0x214/0x220
Jul 26 09:15:27 kernel: Modules linked in: vrf bridge stp llc vxlan ip6_udp_tunnel udp_tunnel nls_iso8859_1 nft_ct amd64_edac_mod edac_mce_amd kvm_amd kvm crct10dif_pclmul ghash_clmulni_intel aesni_intel crypto_simd cryptd glue_helper wmi_bmof ipmi_ssif joydev input_leds rndis_host cdc_ether usbnet mii ast drm_vram_helper ttm drm_kms_he>
Jul 26 09:15:27 kernel: CPU: 16 PID: 0 Comm: swapper/16 Not tainted 5.4.0-174-generic #193-Ubuntu
Jul 26 09:15:27 kernel: Hardware name: Supermicro SMC 2x26 os-gen8 64C NVME-Y 256G/H12SSW-NTR, BIOS 2.5.V1.2U.NVMe.UEFI 05/09/2023
Jul 26 09:15:27 kernel: RIP: 0010:tcp_send_loss_probe+0x214/0x220
Jul 26 09:15:27 kernel: Code: 08 26 01 00 75 e2 41 0f b6 54 24 12 41 8b 8c 24 c0 06 00 00 45 89 f0 48 c7 c7 e0 b4 20 a7 c6 05 8d 08 26 01 01 e8 4a c0 0f 00 <0f> 0b eb ba 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 89 e5 41
Jul 26 09:15:27 kernel: RSP: 0018:ffffb7838088ce00 EFLAGS: 00010286
Jul 26 09:15:27 kernel: RAX: 0000000000000000 RBX: ffff9b84b5630430 RCX: 0000000000000006
Jul 26 09:15:27 kernel: RDX: 0000000000000007 RSI: 0000000000000096 RDI: ffff9b8e4621c8c0
Jul 26 09:15:27 kernel: RBP: ffffb7838088ce18 R08: 0000000000000927 R09: 0000000000000004
Jul 26 09:15:27 kernel: R10: 0000000000000000 R11: 0000000000000001 R12: ffff9b84b5630000
Jul 26 09:15:27 kernel: R13: 0000000000000000 R14: 000000000000231c R15: ffff9b84b5630430
Jul 26 09:15:27 kernel: FS: 0000000000000000(0000) GS:ffff9b8e46200000(0000) knlGS:0000000000000000
Jul 26 09:15:27 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jul 26 09:15:27 kernel: CR2: 000056238cec2380 CR3: 0000003e49ede005 CR4: 0000000000760ee0
Jul 26 09:15:27 kernel: PKRU: 55555554
Jul 26 09:15:27 kernel: Call Trace:
Jul 26 09:15:27 kernel: <IRQ>
Jul 26 09:15:27 kernel: ? show_regs.cold+0x1a/0x1f
Jul 26 09:15:27 kernel: ? __warn+0x98/0xe0
Jul 26 09:15:27 kernel: ? tcp_send_loss_probe+0x214/0x220
Jul 26 09:15:27 kernel: ? report_bug+0xd1/0x100
Jul 26 09:15:27 kernel: ? do_error_trap+0x9b/0xc0
Jul 26 09:15:27 kernel: ? do_invalid_op+0x3c/0x50
Jul 26 09:15:27 kernel: ? tcp_send_loss_probe+0x214/0x220
Jul 26 09:15:27 kernel: ? invalid_op+0x1e/0x30
Jul 26 09:15:27 kernel: ? tcp_send_loss_probe+0x214/0x220
Jul 26 09:15:27 kernel: tcp_write_timer_handler+0x1b4/0x240
Jul 26 09:15:27 kernel: tcp_write_timer+0x9e/0xe0
Jul 26 09:15:27 kernel: ? tcp_write_timer_handler+0x240/0x240
Jul 26 09:15:27 kernel: call_timer_fn+0x32/0x130
Jul 26 09:15:27 kernel: __run_timers.part.0+0x180/0x280
Jul 26 09:15:27 kernel: ? timerqueue_add+0x9b/0xb0
Jul 26 09:15:27 kernel: ? enqueue_hrtimer+0x3d/0x90
Jul 26 09:15:27 kernel: ? do_error_trap+0x9b/0xc0
Jul 26 09:15:27 kernel: ? do_invalid_op+0x3c/0x50
Jul 26 09:15:27 kernel: ? tcp_send_loss_probe+0x214/0x220
Jul 26 09:15:27 kernel: ? invalid_op+0x1e/0x30
Jul 26 09:15:27 kernel: ? tcp_send_loss_probe+0x214/0x220
Jul 26 09:15:27 kernel: tcp_write_timer_handler+0x1b4/0x240
Jul 26 09:15:27 kernel: tcp_write_timer+0x9e/0xe0
Jul 26 09:15:27 kernel: ? tcp_write_timer_handler+0x240/0x240
Jul 26 09:15:27 kernel: call_timer_fn+0x32/0x130
Jul 26 09:15:27 kernel: __run_timers.part.0+0x180/0x280
Jul 26 09:15:27 kernel: ? timerqueue_add+0x9b/0xb0
Jul 26 09:15:27 kernel: ? enqueue_hrtimer+0x3d/0x90
Jul 26 09:15:27 kernel: ? recalibrate_cpu_khz+0x10/0x10
Jul 26 09:15:27 kernel: ? ktime_get+0x3e/0xa0
Jul 26 09:15:27 kernel: ? native_x2apic_icr_write+0x30/0x30
Jul 26 09:15:27 kernel: run_timer_softirq+0x2a/0x50
Jul 26 09:15:27 kernel: __do_softirq+0xd1/0x2c1
Jul 26 09:15:27 kernel: irq_exit+0xae/0xb0
Jul 26 09:15:27 kernel: smp_apic_timer_interrupt+0x7b/0x140
Jul 26 09:15:27 kernel: apic_timer_interrupt+0xf/0x20
Jul 26 09:15:27 kernel: </IRQ>
Jul 26 09:15:27 kernel: RIP: 0010:native_safe_halt+0xe/0x10
Jul 26 09:15:27 kernel: Code: 7b ff ff ff eb bd 90 90 90 90 90 90 e9 07 00 00 00 0f 00 2d 36 2c 50 00 f4 c3 66 90 e9 07 00 00 00 0f 00 2d 26 2c 50 00 fb f4 <c3> 90 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 53 e8 dd 5e 61 ff 65
Jul 26 09:15:27 kernel: RSP: 0018:ffffb783801cfe70 EFLAGS: 00000246 ORIG_RAX: ffffffffffffff13
Jul 26 09:15:27 kernel: RAX: ffffffffa6908b20 RBX: 0000000000000010 RCX: 0000000000000001
Jul 26 09:15:27 kernel: RDX: 000000006fc0c97e RSI: 0000000000000082 RDI: 0000000000000082
Jul 26 09:15:27 kernel: RBP: ffffb783801cfe90 R08: 0000000000000000 R09: 0000000000000225
Jul 26 09:15:27 kernel: R10: 0000000000100000 R11: 0000000000000000 R12: 0000000000000010
Jul 26 09:15:27 kernel: R13: ffff9b8e390b0000 R14: 0000000000000000 R15: 0000000000000000
Jul 26 09:15:27 kernel: ? __cpuidle_text_start+0x8/0x8
Jul 26 09:15:27 kernel: ? default_idle+0x20/0x140
Jul 26 09:15:27 kernel: arch_cpu_idle+0x15/0x20
Jul 26 09:15:27 kernel: default_idle_call+0x23/0x30
Jul 26 09:15:27 kernel: do_idle+0x1fb/0x270
Jul 26 09:15:27 kernel: cpu_startup_entry+0x20/0x30
Jul 26 09:15:27 kernel: start_secondary+0x178/0x1d0
Jul 26 09:15:27 kernel: secondary_startup_64+0xa4/0xb0
Jul 26 09:15:27 kernel: ---[ end trace e7ac822987e33be1 ]---

The NULL ptr deref is coming from tcp_rto_delta_us() attempting to pull an skb
off the head of the retransmit queue and then dereferencing that skb to get the
skb_mstamp_ns value via tcp_skb_timestamp_us(skb).

The crash is the same one that was reported a # of years ago here:
https://lore.kernel.org/netdev/86c0f836-9a7c-438b-d81a-839be45f1f58@gmail.com/T/#t

and the kernel we're running has the fix which was added to resolve this issue.

Unfortunately we've been unsuccessful so far in reproducing this problem in the
lab and do not have the luxury of pushing out a new kernel to try and test if
newer kernels resolve this issue at the moment. I realize this is a report
against both an Ubuntu kernel and also an older 5.4 kernel. I have reported this
issue to Ubuntu here: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2077657
however I feel like since this issue has possibly cropped up again it makes
sense to build in some protection in this path (even on the latest kernel
versions) since the code in question just blindly assumes there's a valid skb
without testing if it's NULL b/f it looks at the timestamp.

Given we have seen crashes in this path before and now this case it seems like
we should protect ourselves for when packets_out accounting is incorrect.
While we should fix that root cause we should also just make sure the skb
is not NULL before dereferencing it. Also add a warn once here to capture
some information if/when the problem case is hit again.

Fixes: e1a10ef7fa87 ("tcp: introduce tcp_rto_delta_us() helper for xmit timer fix")
Signed-off-by: Josh Hunt <johunt@akamai.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>

net: ipv6: select DST_CACHE from IPV6_RPL_LWTUNNEL

[ Upstream commit 93c21077bb9ba08807c459982d440dbbee4c7af3 ]

The rpl sr tunnel code contains calls to dst_cache_*() which are
only present when the dst cache is built.
Select DST_CACHE to build the dst cache, similar to other kconfig
options in the same file.
Compiling the rpl sr tunnel without DST_CACHE will lead to linker
errors.

Fixes: a7a29f9c361f ("net: ipv6: add rpl sr tunnel")
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Simon Horman <horms@kernel.org> # build-tested
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: seeq: Fix use after free vulnerability in ether3 Driver Due to Race Condition

[ Upstream commit b5109b60ee4fcb2f2bb24f589575e10cc5283ad4 ]

In the ether3_probe function, a timer is initialized with a callback
function ether3_ledoff, bound to &prev(dev)->timer. Once the timer is
started, there is a risk of a race condition if the module or device
is removed, triggering the ether3_remove function to perform cleanup.
The sequence of operations that may lead to a UAF bug is as follows:

CPU0                                    CPU1

                      |  ether3_ledoff
ether3_remove         |
  free_netdev(dev);   |
  put_devic           |
  kfree(dev);         |
 |  ether3_outw(priv(dev)->regs.config2 |= CFG2_CTRLO, REG_CONFIG2);
                      | // use dev

Fix it by ensuring that the timer is canceled before proceeding with
the cleanup in ether3_remove.

Fixes: 6fd9c53f7186 ("net: seeq: Convert timers to use timer_setup()")
Signed-off-by: Kaixin Wang <kxwang23@m.fudan.edu.cn>
Link: https://patch.msgid.link/20240915144045.451-1-kxwang23@m.fudan.edu.cn
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_reject_ipv6: fix nf_reject_ip6_tcphdr_put()

[ Upstream commit 9c778fe48d20ef362047e3376dee56d77f8500d4 ]

syzbot reported that nf_reject_ip6_tcphdr_put() was possibly sending
garbage on the four reserved tcp bits (th->res1)

Use skb_put_zero() to clear the whole TCP header,
as done in nf_reject_ip_tcphdr_put()

BUG: KMSAN: uninit-value in nf_reject_ip6_tcphdr_put+0x688/0x6c0 net/ipv6/netfilter/nf_reject_ipv6.c:255
  nf_reject_ip6_tcphdr_put+0x688/0x6c0 net/ipv6/netfilter/nf_reject_ipv6.c:255
  nf_send_reset6+0xd84/0x15b0 net/ipv6/netfilter/nf_reject_ipv6.c:344
  nft_reject_inet_eval+0x3c1/0x880 net/netfilter/nft_reject_inet.c:48
  expr_call_ops_eval net/netfilter/nf_tables_core.c:240 [inline]
  nft_do_chain+0x438/0x22a0 net/netfilter/nf_tables_core.c:288
  nft_do_chain_inet+0x41a/0x4f0 net/netfilter/nft_chain_filter.c:161
  nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline]
  nf_hook_slow+0xf4/0x400 net/netfilter/core.c:626
  nf_hook include/linux/netfilter.h:269 [inline]
  NF_HOOK include/linux/netfilter.h:312 [inline]
  ipv6_rcv+0x29b/0x390 net/ipv6/ip6_input.c:310
  __netif_receive_skb_one_core net/core/dev.c:5661 [inline]
  __netif_receive_skb+0x1da/0xa00 net/core/dev.c:5775
  process_backlog+0x4ad/0xa50 net/core/dev.c:6108
  __napi_poll+0xe7/0x980 net/core/dev.c:6772
  napi_poll net/core/dev.c:6841 [inline]
  net_rx_action+0xa5a/0x19b0 net/core/dev.c:6963
  handle_softirqs+0x1ce/0x800 kernel/softirq.c:554
  __do_softirq+0x14/0x1a kernel/softirq.c:588
  do_softirq+0x9a/0x100 kernel/softirq.c:455
  __local_bh_enable_ip+0x9f/0xb0 kernel/softirq.c:382
  local_bh_enable include/linux/bottom_half.h:33 [inline]
  rcu_read_unlock_bh include/linux/rcupdate.h:908 [inline]
  __dev_queue_xmit+0x2692/0x5610 net/core/dev.c:4450
  dev_queue_xmit include/linux/netdevice.h:3105 [inline]
  neigh_resolve_output+0x9ca/0xae0 net/core/neighbour.c:1565
  neigh_output include/net/neighbour.h:542 [inline]
  ip6_finish_output2+0x2347/0x2ba0 net/ipv6/ip6_output.c:141
  __ip6_finish_output net/ipv6/ip6_output.c:215 [inline]
  ip6_finish_output+0xbb8/0x14b0 net/ipv6/ip6_output.c:226
  NF_HOOK_COND include/linux/netfilter.h:303 [inline]
  ip6_output+0x356/0x620 net/ipv6/ip6_output.c:247
  dst_output include/net/dst.h:450 [inline]
  NF_HOOK include/linux/netfilter.h:314 [inline]
  ip6_xmit+0x1ba6/0x25d0 net/ipv6/ip6_output.c:366
  inet6_csk_xmit+0x442/0x530 net/ipv6/inet6_connection_sock.c:135
  __tcp_transmit_skb+0x3b07/0x4880 net/ipv4/tcp_output.c:1466
  tcp_transmit_skb net/ipv4/tcp_output.c:1484 [inline]
  tcp_connect+0x35b6/0x7130 net/ipv4/tcp_output.c:4143
  tcp_v6_connect+0x1bcc/0x1e40 net/ipv6/tcp_ipv6.c:333
  __inet_stream_connect+0x2ef/0x1730 net/ipv4/af_inet.c:679
  inet_stream_connect+0x6a/0xd0 net/ipv4/af_inet.c:750
  __sys_connect_file net/socket.c:2061 [inline]
  __sys_connect+0x606/0x690 net/socket.c:2078
  __do_sys_connect net/socket.c:2088 [inline]
  __se_sys_connect net/socket.c:2085 [inline]
  __x64_sys_connect+0x91/0xe0 net/socket.c:2085
  x64_sys_call+0x27a5/0x3ba0 arch/x86/include/generated/asm/syscalls_64.h:43
  do_syscall_x64 arch/x86/entry/common.c:52 [inline]
  do_syscall_64+0xcd/0x1e0 arch/x86/entry/common.c:83
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

Uninit was stored to memory at:
  nf_reject_ip6_tcphdr_put+0x60c/0x6c0 net/ipv6/netfilter/nf_reject_ipv6.c:249
  nf_send_reset6+0xd84/0x15b0 net/ipv6/netfilter/nf_reject_ipv6.c:344
  nft_reject_inet_eval+0x3c1/0x880 net/netfilter/nft_reject_inet.c:48
  expr_call_ops_eval net/netfilter/nf_tables_core.c:240 [inline]
  nft_do_chain+0x438/0x22a0 net/netfilter/nf_tables_core.c:288
  nft_do_chain_inet+0x41a/0x4f0 net/netfilter/nft_chain_filter.c:161
  nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline]
  nf_hook_slow+0xf4/0x400 net/netfilter/core.c:626
  nf_hook include/linux/netfilter.h:269 [inline]
  NF_HOOK include/linux/netfilter.h:312 [inline]
  ipv6_rcv+0x29b/0x390 net/ipv6/ip6_input.c:310
  __netif_receive_skb_one_core net/core/dev.c:5661 [inline]
  __netif_receive_skb+0x1da/0xa00 net/core/dev.c:5775
  process_backlog+0x4ad/0xa50 net/core/dev.c:6108
  __napi_poll+0xe7/0x980 net/core/dev.c:6772
  napi_poll net/core/dev.c:6841 [inline]
  net_rx_action+0xa5a/0x19b0 net/core/dev.c:6963
  handle_softirqs+0x1ce/0x800 kernel/softirq.c:554
  __do_softirq+0x14/0x1a kernel/softirq.c:588

Uninit was stored to memory at:
  nf_reject_ip6_tcphdr_put+0x2ca/0x6c0 net/ipv6/netfilter/nf_reject_ipv6.c:231
  nf_send_reset6+0xd84/0x15b0 net/ipv6/netfilter/nf_reject_ipv6.c:344
  nft_reject_inet_eval+0x3c1/0x880 net/netfilter/nft_reject_inet.c:48
  expr_call_ops_eval net/netfilter/nf_tables_core.c:240 [inline]
  nft_do_chain+0x438/0x22a0 net/netfilter/nf_tables_core.c:288
  nft_do_chain_inet+0x41a/0x4f0 net/netfilter/nft_chain_filter.c:161
  nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline]
  nf_hook_slow+0xf4/0x400 net/netfilter/core.c:626
  nf_hook include/linux/netfilter.h:269 [inline]
  NF_HOOK include/linux/netfilter.h:312 [inline]
  ipv6_rcv+0x29b/0x390 net/ipv6/ip6_input.c:310
  __netif_receive_skb_one_core net/core/dev.c:5661 [inline]
  __netif_receive_skb+0x1da/0xa00 net/core/dev.c:5775
  process_backlog+0x4ad/0xa50 net/core/dev.c:6108
  __napi_poll+0xe7/0x980 net/core/dev.c:6772
  napi_poll net/core/dev.c:6841 [inline]
  net_rx_action+0xa5a/0x19b0 net/core/dev.c:6963
  handle_softirqs+0x1ce/0x800 kernel/softirq.c:554
  __do_softirq+0x14/0x1a kernel/softirq.c:588

Uninit was created at:
  slab_post_alloc_hook mm/slub.c:3998 [inline]
  slab_alloc_node mm/slub.c:4041 [inline]
  kmem_cache_alloc_node_noprof+0x6bf/0xb80 mm/slub.c:4084
  kmalloc_reserve+0x13d/0x4a0 net/core/skbuff.c:583
  __alloc_skb+0x363/0x7b0 net/core/skbuff.c:674
  alloc_skb include/linux/skbuff.h:1320 [inline]
  nf_send_reset6+0x98d/0x15b0 net/ipv6/netfilter/nf_reject_ipv6.c:327
  nft_reject_inet_eval+0x3c1/0x880 net/netfilter/nft_reject_inet.c:48
  expr_call_ops_eval net/netfilter/nf_tables_core.c:240 [inline]
  nft_do_chain+0x438/0x22a0 net/netfilter/nf_tables_core.c:288
  nft_do_chain_inet+0x41a/0x4f0 net/netfilter/nft_chain_filter.c:161
  nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline]
  nf_hook_slow+0xf4/0x400 net/netfilter/core.c:626
  nf_hook include/linux/netfilter.h:269 [inline]
  NF_HOOK include/linux/netfilter.h:312 [inline]
  ipv6_rcv+0x29b/0x390 net/ipv6/ip6_input.c:310
  __netif_receive_skb_one_core net/core/dev.c:5661 [inline]
  __netif_receive_skb+0x1da/0xa00 net/core/dev.c:5775
  process_backlog+0x4ad/0xa50 net/core/dev.c:6108
  __napi_poll+0xe7/0x980 net/core/dev.c:6772
  napi_poll net/core/dev.c:6841 [inline]
  net_rx_action+0xa5a/0x19b0 net/core/dev.c:6963
  handle_softirqs+0x1ce/0x800 kernel/softirq.c:554
  __do_softirq+0x14/0x1a kernel/softirq.c:588

Fixes: c8d7b98bec43 ("netfilter: move nf_send_resetX() code to nf_reject_ipvX modules")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Pablo Neira Ayuso <pablo@netfilter.org>
Link: https://patch.msgid.link/20240913170615.3670897-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: xilinx: axienet: Fix packet counting

[ Upstream commit 5a6caa2cfabb559309b5ce29ee7c8e9ce1a9a9df ]

axienet_free_tx_chain returns the number of DMA descriptors it's
handled. However, axienet_tx_poll treats the return as the number of
packets. When scatter-gather SKBs are enabled, a single packet may use
multiple DMA descriptors, which causes incorrect packet counts. Fix this
by explicitly keepting track of the number of packets processed as
separate from the DMA descriptors.

Budget does not affect the number of Tx completions we can process for
NAPI, so we use the ring size as the limit instead of budget. As we no
longer return the number of descriptors processed to axienet_tx_poll, we
now update tx_bd_ci in axienet_free_tx_chain.

Fixes: 8a3b7a252dca ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
Signed-off-by: Sean Anderson <sean.anderson@linux.dev>
Link: https://patch.msgid.link/20240913145156.2283067-1-sean.anderson@linux.dev
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: xilinx: axienet: Schedule NAPI in two steps

[ Upstream commit ba0da2dc934ec5ac32bbeecbd0670da16ba03565 ]

As advised by Documentation/networking/napi.rst, masking IRQs after
calling napi_schedule can be racy. Avoid this by only masking/scheduling
if napi_schedule_prep returns true.

Fixes: 9e2bc267e780 ("net: axienet: Use NAPI for TX completion path")
Fixes: cc37610caaf8 ("net: axienet: implement NAPI and GRO receive")
Signed-off-by: Sean Anderson <sean.anderson@linux.dev>
Reviewed-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20240913145711.2284295-1-sean.anderson@linux.dev
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Revert "dm: requeue IO if mapping table not yet available"

[ Upstream commit c8691cd0fc11197515ed148de0780d927bfca38b ]

This reverts commit fa247089de9936a46e290d4724cb5f0b845600f5.

The following sequence of commands causes a livelock - there will be
workqueue process looping and consuming 100% CPU:

dmsetup create --notable test
truncate -s 1MiB testdata
losetup /dev/loop0 testdata
dmsetup load test --table '0 2048 linear /dev/loop0 0'
dd if=/dev/zero of=/dev/dm-0 bs=16k count=1 conv=fdatasync

The livelock is caused by the commit fa247089de99. The commit claims that
it fixes a race condition, however, it is unknown what the actual race
condition is and what program is involved in the race condition.

When the inactive table is loaded, the nodes /dev/dm-0 and
/sys/block/dm-0 are created. /dev/dm-0 has zero size at this point. When
the device is suspended and resumed, the nodes /dev/mapper/test and
/dev/disk/* are created.

If some program opens a block device before it is created by dmsetup or
lvm, the program is buggy, so dm could just report an error as it used to
do before.

Reported-by: Zdenek Kabelac <zkabelac@redhat.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Fixes: fa247089de99 ("dm: requeue IO if mapping table not yet available")
Signed-off-by: Sasha Levin <sashal@kernel.org>

ep93xx: clock: Fix off by one in ep93xx_div_recalc_rate()

[ Upstream commit c7f06284a6427475e3df742215535ec3f6cd9662 ]

The psc->div[] array has psc->num_div elements.  These values come from
when we call clk_hw_register_div().  It's adc_divisors and
ARRAY_SIZE(adc_divisors)) and so on.  So this condition needs to be >=
instead of > to prevent an out of bounds read.

Fixes: 9645ccc7bd7a ("ep93xx: clock: convert in-place to COMMON_CLK")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Acked-by: Alexander Sverdlin <alexander.sverdlin@gmail.com>
Reviewed-by: Nikita Shubin <nikita.shubin@maquefel.me>
Signed-off-by: Alexander Sverdlin <alexander.sverdlin@gmail.com>
Link: https://lore.kernel.org/r/1caf01ad4c0a8069535813c26c7f0b8ea011155e.camel@linaro.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vhost_vdpa: assign irq bypass producer token correctly

[ Upstream commit 02e9e9366fefe461719da5d173385b6685f70319 ]

We used to call irq_bypass_unregister_producer() in
vhost_vdpa_setup_vq_irq() which is problematic as we don't know if the
token pointer is still valid or not.

Actually, we use the eventfd_ctx as the token so the life cycle of the
token should be bound to the VHOST_SET_VRING_CALL instead of
vhost_vdpa_setup_vq_irq() which could be called by set_status().

Fixing this by setting up irq bypass producer's token when handling
VHOST_SET_VRING_CALL and un-registering the producer before calling
vhost_vring_ioctl() to prevent a possible use after free as eventfd
could have been released in vhost_vring_ioctl(). And such registering
and unregistering will only be done if DRIVER_OK is set.

Reported-by: Dragos Tatulea <dtatulea@nvidia.com>
Tested-by: Dragos Tatulea <dtatulea@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Fixes: 2cf1ba9a4d15 ("vhost_vdpa: implement IRQ offloading in vhost_vdpa")
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20240816031900.18013-1-jasowang@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vdpa/mlx5: Fix invalid mr resource destroy

[ Upstream commit dc12502905b7a3de9097ea6b98870470c2921e09 ]

Certain error paths from mlx5_vdpa_dev_add() can end up releasing mr
resources which never got initialized in the first place.

This patch adds the missing check in mlx5_vdpa_destroy_mr_resources()
to block releasing non-initialized mr resources.

Reference trace:

  mlx5_core 0000:08:00.2: mlx5_vdpa_dev_add:3274:(pid 2700) warning: No mac address provisioned?
  BUG: kernel NULL pointer dereference, address: 0000000000000000
  #PF: supervisor read access in kernel mode
  #PF: error_code(0x0000) - not-present page
  PGD 140216067 P4D 0
  Oops: 0000 [#1] PREEMPT SMP NOPTI
  CPU: 8 PID: 2700 Comm: vdpa Kdump: loaded Not tainted 5.14.0-496.el9.x86_64 #1
  Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
  RIP: 0010:vhost_iotlb_del_range+0xf/0xe0 [vhost_iotlb]
  Code: [...]
  RSP: 0018:ff1c823ac23077f0 EFLAGS: 00010246
  RAX: ffffffffc1a21a60 RBX: ffffffff899567a0 RCX: 0000000000000000
  RDX: ffffffffffffffff RSI: 0000000000000000 RDI: 0000000000000000
  RBP: ff1bda1f7c21e800 R08: 0000000000000000 R09: ff1c823ac2307670
  R10: ff1c823ac2307668 R11: ffffffff8a9e7b68 R12: 0000000000000000
  R13: 0000000000000000 R14: ff1bda1f43e341a0 R15: 00000000ffffffea
  FS:  00007f56eba7c740(0000) GS:ff1bda269f800000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000000000000 CR3: 0000000104d90001 CR4: 0000000000771ef0
  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
  PKRU: 55555554
  Call Trace:

   ? show_trace_log_lvl+0x1c4/0x2df
   ? show_trace_log_lvl+0x1c4/0x2df
   ? mlx5_vdpa_free+0x3d/0x150 [mlx5_vdpa]
   ? __die_body.cold+0x8/0xd
   ? page_fault_oops+0x134/0x170
   ? __irq_work_queue_local+0x2b/0xc0
   ? irq_work_queue+0x2c/0x50
   ? exc_page_fault+0x62/0x150
   ? asm_exc_page_fault+0x22/0x30
   ? __pfx_mlx5_vdpa_free+0x10/0x10 [mlx5_vdpa]
   ? vhost_iotlb_del_range+0xf/0xe0 [vhost_iotlb]
   mlx5_vdpa_free+0x3d/0x150 [mlx5_vdpa]
   vdpa_release_dev+0x1e/0x50 [vdpa]
   device_release+0x31/0x90
   kobject_cleanup+0x37/0x130
   mlx5_vdpa_dev_add+0x2d2/0x7a0 [mlx5_vdpa]
   vdpa_nl_cmd_dev_add_set_doit+0x277/0x4c0 [vdpa]
   genl_family_rcv_msg_doit+0xd9/0x130
   genl_family_rcv_msg+0x14d/0x220
   ? __pfx_vdpa_nl_cmd_dev_add_set_doit+0x10/0x10 [vdpa]
   ? _copy_to_user+0x1a/0x30
   ? move_addr_to_user+0x4b/0xe0
   genl_rcv_msg+0x47/0xa0
   ? __import_iovec+0x46/0x150
   ? __pfx_genl_rcv_msg+0x10/0x10
   netlink_rcv_skb+0x54/0x100
   genl_rcv+0x24/0x40
   netlink_unicast+0x245/0x370
   netlink_sendmsg+0x206/0x440
   __sys_sendto+0x1dc/0x1f0
   ? do_read_fault+0x10c/0x1d0
   ? do_pte_missing+0x10d/0x190
   __x64_sys_sendto+0x20/0x30
   do_syscall_64+0x5c/0xf0
   ? __count_memcg_events+0x4f/0xb0
   ? mm_account_fault+0x6c/0x100
   ? handle_mm_fault+0x116/0x270
   ? do_user_addr_fault+0x1d6/0x6a0
   ? do_syscall_64+0x6b/0xf0
   ? clear_bhb_loop+0x25/0x80
   ? clear_bhb_loop+0x25/0x80
   ? clear_bhb_loop+0x25/0x80
   ? clear_bhb_loop+0x25/0x80
   ? clear_bhb_loop+0x25/0x80
   entry_SYSCALL_64_after_hwframe+0x78/0x80

Fixes: 512c0cdd80c1 ("vdpa/mlx5: Decouple cvq iotlb handling from hw mapping code")
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com>
Message-Id: <20240827160808.2448017-2-dtatulea@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Shannon Nelson <shannon.nelson@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cxl/pci: Fix to record only non-zero ranges

[ Upstream commit 55e268694e8b07026c88191f9b6949b6887d9ce3 ]

The function cxl_dvsec_rr_decode() retrieves and records DVSEC ranges
into info->dvsec_range[], regardless of whether it is non-zero range,
and the variable info->ranges indicates the number of non-zero ranges.
However, in cxl_hdm_decode_init(), the validation for
info->dvsec_range[] occurs in a for loop that iterates based on
info->ranges. It may result in zero range to be validated but non-zero
range not be validated, in turn, the number of allowed ranges is to be
0. Address it by only record non-zero ranges.

This fix is not urgent as it requires a configuration that zeroes out
the first dvsec range while populating the second. This has not been
observed, but it is theoretically possible. If this gets picked up for
-stable, no harm done, but there is no urgency to backport.

Fixes: 560f78559006 ("cxl/pci: Retrieve CXL DVSEC memory info")
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Yanfei Xu <yanfei.xu@intel.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Link: https://patch.msgid.link/20240828084231.1378789-2-yanfei.xu@intel.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

interconnect: qcom: sm8250: Enable sync_state

[ Upstream commit d3681b30214eb5885092ce4586f07237dc3c522f ]

Enable the generic icc sync_state callback to ensure interconnect votes
are actually taken into account, instead of being forced to the maximum
value.

Fixes: b95b668eaaa2 ("interconnect: qcom: icc-rpmh: Add BCMs to commit list in pre_aggregate")
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20240804-sm8350-fixes-v1-8-1149dd8399fe@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

interconnect: icc-clk: Add missed num_nodes initialization

[ Upstream commit c801ed86840ec38b2a9bcafeee3d7c9e14c743f3 ]

With the new __counted_by annotation, the "num_nodes" struct member must
be set before accessing the "nodes" array. This initialization was done
in other places where a new struct icc_onecell_data is allocated, but this
case in icc_clk_register() was missed. Set "num_nodes" after allocation.

Fixes: dd4904f3b924 ("interconnect: qcom: Annotate struct icc_onecell_data with __counted_by")
Signed-off-by: Kees Cook <kees@kernel.org>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Link: https://lore.kernel.org/r/20240716214819.work.328-kees@kernel.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

coresight: tmc: sg: Do not leak sg_table

[ Upstream commit c58dc5a1f886f2fcc1133746d0cbaa1fe7fd44ff ]

Running perf with cs_etm on Juno triggers the following kmemleak warning !

:~# cat /sys/kernel/debug/kmemleak
 unreferenced object 0xffffff8806b6d720 (size 96):
 comm "perf", pid 562, jiffies 4297810960
 hex dump (first 32 bytes):
 38 d8 13 07 88 ff ff ff 00 d0 9e 85 c0 ff ff ff  8...............
 00 10 00 88 c0 ff ff ff 00 f0 ff f7 ff 00 00 00  ................
 backtrace (crc 1dbf6e00):
 [<ffffffc08107381c>] kmemleak_alloc+0xbc/0xd8
 [<ffffffc0802f9798>] kmalloc_trace_noprof+0x220/0x2e8
 [<ffffffc07bb71948>] tmc_alloc_sg_table+0x48/0x208 [coresight_tmc]
 [<ffffffc07bb71cbc>] tmc_etr_alloc_sg_buf+0xac/0x240 [coresight_tmc]
 [<ffffffc07bb72538>] tmc_alloc_etr_buf.constprop.0+0x1f0/0x260 [coresight_tmc]
 [<ffffffc07bb7280c>] alloc_etr_buf.constprop.0.isra.0+0x74/0xa8 [coresight_tmc]
 [<ffffffc07bb72950>] tmc_alloc_etr_buffer+0x110/0x260 [coresight_tmc]
 [<ffffffc07bb38afc>] etm_setup_aux+0x204/0x3b0 [coresight]
 [<ffffffc08025837c>] rb_alloc_aux+0x20c/0x318
 [<ffffffc08024dd84>] perf_mmap+0x2e4/0x7a0
 [<ffffffc0802cceb0>] mmap_region+0x3b0/0xa08
 [<ffffffc0802cd8a8>] do_mmap+0x3a0/0x500
 [<ffffffc080295328>] vm_mmap_pgoff+0x100/0x1d0
 [<ffffffc0802cadf8>] ksys_mmap_pgoff+0xb8/0x110
 [<ffffffc080020688>] __arm64_sys_mmap+0x38/0x58
 [<ffffffc080028fc0>] invoke_syscall.constprop.0+0x58/0x100

This due to the fact that we do not free the "sg_table" itself while
freeing up  the SG table and data pages. Fix this by freeing the sg_table
in tmc_free_sg_table().

Fixes: 99443ea19e8b ("coresight: Add generic TMC sg table framework")
Cc: Mike Leach <mike.leach@linaro.org>
Cc: James Clark <james.clark@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Link: https://lore.kernel.org/r/20240702132846.1677261-1-suzuki.poulose@arm.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

Coresight: Set correct cs_mode for dummy source to fix disable issue

[ Upstream commit e6b64cda393efd84709ab3df2e42d36d36d7553e ]

The coresight_disable_source_sysfs function should verify the
mode of the coresight device before disabling the source.
However, the mode for the dummy source device is always set to
CS_MODE_DISABLED, resulting in the check consistently failing.
As a result, dummy source cannot be properly disabled.

Configure CS_MODE_SYSFS/CS_MODE_PERF during the enablement.
Configure CS_MODE_DISABLED during the disablement.

Fixes: 9d3ba0b6c056 ("Coresight: Add coresight dummy driver")
Signed-off-by: Jie Gan <quic_jiegan@quicinc.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Link: https://lore.kernel.org/r/20240812042844.2890115-1-quic_jiegan@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

Coresight: Set correct cs_mode for TPDM to fix disable issue

[ Upstream commit 14f5fa9b5fcbe2b3d5098893aba6ad62254d2ef6 ]

The coresight_disable_source_sysfs function should verify the
mode of the coresight device before disabling the source.

However, the mode for the TPDM device is always set to
CS_MODE_DISABLED, resulting in the check consistently failing.
As a result, TPDM cannot be properly disabled.

Configure CS_MODE_SYSFS/CS_MODE_PERF during the enablement.
Configure CS_MODE_DISABLED during the disablement.

Fixes: b3c71626a933 ("Coresight: Add coresight TPDM source driver")
Signed-off-by: Jie Gan <quic_jiegan@quicinc.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Link: https://lore.kernel.org/r/20240812043043.2890694-1-quic_jiegan@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

serial: 8250: omap: Cleanup on error in request_irq

[ Upstream commit 35e648a16018b747897be2ccc3ce95ff23237bb5 ]

If devm_request_irq fails, the code does not cleanup many things that
were setup before. Instead of directly returning ret we should jump to
err.

Fixes: fef4f600319e ("serial: 8250: omap: Fix life cycle issues for interrupt handlers")
Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Link: https://lore.kernel.org/r/20240807141227.1093006-4-msp@baylibre.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

driver core: Fix a potential null-ptr-deref in module_add_driver()

[ Upstream commit 18ec12c97b39ff6aa15beb8d2b25d15cd44b87d8 ]

Inject fault while probing of-fpga-region, if kasprintf() fails in
module_add_driver(), the second sysfs_remove_link() in exit path will cause
null-ptr-deref as below because kernfs_name_hash() will call strlen() with
NULL driver_name.

Fix it by releasing resources based on the exit path sequence.

 KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
 Mem abort info:
   ESR = 0x0000000096000005
   EC = 0x25: DABT (current EL), IL = 32 bits
   SET = 0, FnV = 0
   EA = 0, S1PTW = 0
   FSC = 0x05: level 1 translation fault
 Data abort info:
   ISV = 0, ISS = 0x00000005, ISS2 = 0x00000000
   CM = 0, WnR = 0, TnD = 0, TagAccess = 0
   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
 [dfffffc000000000] address between user and kernel address ranges
 Internal error: Oops: 0000000096000005 [#1] PREEMPT SMP
 Dumping ftrace buffer:
    (ftrace buffer empty)
 Modules linked in: of_fpga_region(+) fpga_region fpga_bridge cfg80211 rfkill 8021q garp mrp stp llc ipv6 [last unloaded: of_fpga_region]
 CPU: 2 UID: 0 PID: 2036 Comm: modprobe Not tainted 6.11.0-rc2-g6a0e38264012 #295
 Hardware name: linux,dummy-virt (DT)
 pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
 pc : strlen+0x24/0xb0
 lr : kernfs_name_hash+0x1c/0xc4
 sp : ffffffc081f97380
 x29: ffffffc081f97380 x28: ffffffc081f97b90 x27: ffffff80c821c2a0
 x26: ffffffedac0be418 x25: 0000000000000000 x24: ffffff80c09d2000
 x23: 0000000000000000 x22: 0000000000000000 x21: 0000000000000000
 x20: 0000000000000000 x19: 0000000000000000 x18: 0000000000001840
 x17: 0000000000000000 x16: 0000000000000000 x15: 1ffffff8103f2e42
 x14: 00000000f1f1f1f1 x13: 0000000000000004 x12: ffffffb01812d61d
 x11: 1ffffff01812d61c x10: ffffffb01812d61c x9 : dfffffc000000000
 x8 : 0000004fe7ed29e4 x7 : ffffff80c096b0e7 x6 : 0000000000000001
 x5 : ffffff80c096b0e0 x4 : 1ffffffdb990efa2 x3 : 0000000000000000
 x2 : 0000000000000000 x1 : dfffffc000000000 x0 : 0000000000000000
 Call trace:
  strlen+0x24/0xb0
  kernfs_name_hash+0x1c/0xc4
  kernfs_find_ns+0x118/0x2e8
  kernfs_remove_by_name_ns+0x80/0x100
  sysfs_remove_link+0x74/0xa8
  module_add_driver+0x278/0x394
  bus_add_driver+0x1f0/0x43c
  driver_register+0xf4/0x3c0
  __platform_driver_register+0x60/0x88
  of_fpga_region_init+0x20/0x1000 [of_fpga_region]
  do_one_initcall+0x110/0x788
  do_init_module+0x1dc/0x5c8
  load_module+0x3c38/0x4cac
  init_module_from_file+0xd4/0x128
  idempotent_init_module+0x2cc/0x528
  __arm64_sys_finit_module+0xac/0x100
  invoke_syscall+0x6c/0x258
  el0_svc_common.constprop.0+0x160/0x22c
  do_el0_svc+0x44/0x5c
  el0_svc+0x48/0xb8
  el0t_64_sync_handler+0x13c/0x158
  el0t_64_sync+0x190/0x194
 Code: f2fbffe1 a90157f4 12000802 aa0003f5 (38e16861)
 ---[ end trace 0000000000000000 ]---
 Kernel panic - not syncing: Oops: Fatal exception

Fixes: 85d2b0aa1703 ("module: don't ignore sysfs_create_link() failures")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Link: https://lore.kernel.org/r/20240812080658.2791982-1-ruanjinjie@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

dt-bindings: iio: asahi-kasei,ak8975: drop incorrect AK09116 compatible

[ Upstream commit c7668ac67bc21aebdd8e2d7f839bfffba31b7713 ]

All compatibles in this binding without prefixes were deprecated, so
adding a new deprecated one after some time is not allowed, because it
defies the core logic of deprecating things.

Drop the AK09916 vendorless compatible.

Fixes: 76e28aa97fa0 ("iio: magnetometer: ak8975: add AK09116 support")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://patch.msgid.link/20240806053016.6401-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: magnetometer: ak8975: drop incorrect AK09116 compatible

[ Upstream commit da6e3160df230692bbd48a6d52318035f19595e2 ]

All compatibles in this binding without prefixes were deprecated, so
adding a new deprecated one after some time is not allowed, because it
defies the core logic of deprecating things.

Drop the AK09916 vendorless compatible.

Fixes: 76e28aa97fa0 ("iio: magnetometer: ak8975: add AK09116 support")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240806053016.6401-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: chemical: bme680: Fix read/write ops to device by adding mutexes

[ Upstream commit 77641e5a477d428335cd094b88ac54e09ccb70f4 ]

Add mutexes in the {read/write}_raw() functions of the device to guard the
read/write of data from/to the device. This is necessary because for any
operation other than temperature, multiple reads need to take place from
the device. Even though regmap has a locking by itself, it won't protect us
from multiple applications trying to read at the same time temperature and
pressure since the pressure reading includes an internal temperature
reading and there is nothing to ensure that this temperature+pressure
reading will happen sequentially without any other operation interfering
in the meantime.

Fixes: 1b3bd8592780 ("iio: chemical: Add support for Bosch BME680 sensor")
Signed-off-by: Vasileios Amoiridis <vassilisamir@gmail.com>
Link: https://patch.msgid.link/20240609233826.330516-2-vassilisamir@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ABI: testing: fix admv8818 attr description

[ Upstream commit 7d34b4ad8cd2867b130b5b8d7d76d0d6092bd019 ]

Fix description of the filter_mode_available attribute by pointing to
the correct name of the attribute that can be written with valid values.

Fixes: bf92d87d7c67 ("iio:filter:admv8818: Add sysfs ABI documentation")
Signed-off-by: Antoniu Miclaus <antoniu.miclaus@analog.com>
Link: https://patch.msgid.link/20240702081851.4663-1-antoniu.miclaus@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

driver core: Fix error handling in driver API device_rename()

[ Upstream commit 6d8249ac29bc23260dfa9747eb398ce76012d73c ]

For class-device, device_rename() failure maybe cause unexpected link name
within its class folder as explained below:

/sys/class/.../old_name -> /sys/devices/.../old_name
device_rename(..., new_name) and failed
/sys/class/.../new_name -> /sys/devices/.../old_name

Fixed by undoing renaming link if renaming kobject failed.

Fixes: f349cf34731c ("driver core: Implement ns directory support for device classes.")
Signed-off-by: Zijun Hu <quic_zijuhu@quicinc.com>
Link: https://lore.kernel.org/r/20240722-device_rename_fix-v2-1-77de1a6c6495@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: adc: ad7606: fix standby gpio state to match the documentation

[ Upstream commit 059fe4f8bbdf5cad212e1aeeb3e8968c80b9ff3b ]

The binding's documentation specifies that "As the line is active low, it
should be marked GPIO_ACTIVE_LOW". However, in the driver, it was handled
the opposite way. This commit sets the driver's behaviour in sync with the
documentation

Fixes: 722407a4e8c0 ("staging:iio:ad7606: Use GPIO descriptor API")
Signed-off-by: Guillaume Stols <gstols@baylibre.com>
Reviewed-by: Nuno Sa <nuno.sa@analog.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: adc: ad7606: fix oversampling gpio array

[ Upstream commit 8dc4594b54dbaaba40dc8884ad3d42083de39434 ]

gpiod_set_array_value was misused here: the implementation relied on the
assumption that an unsigned long was required for each gpio, while the
function expects a bit array stored in "as much unsigned long as needed
for storing one bit per GPIO", i.e it is using a bit field.

This leaded to incorrect parameter passed to gpiod_set_array_value, that
would set 1 value instead of 3.
It also prevents to select the software mode correctly for the AD7606B.

Fixes: d2a415c86c6b ("iio: adc: ad7606: Add support for AD7606B ADC")
Fixes: 41f71e5e7daf ("staging: iio: adc: ad7606: Use find_closest() macro")
Signed-off-by: Guillaume Stols <gstols@baylibre.com>
Reviewed-by: Nuno Sa <nuno.sa@analog.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nvme-multipath: system fails to create generic nvme device

[ Upstream commit 63bcf9014e95a7d279d10d8e2caa5d88db2b1855 ]

NVME_NSHEAD_DISK_LIVE is a flag for struct nvme_ns_head, not nvme_ns.
The current code has a typo causing NVME_NSHEAD_DISK_LIVE never to
be cleared once device_add_disk_fails, causing the system never to
create the 'generic' character device. Even several rescan attempts
will change the situation and the system has to be rebooted to fix
the issue.

Fixes: 11384580e332 ("nvme-multipath: add error handling support for add_disk()")
Signed-off-by: Hannes Reinecke <hare@kernel.org>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: atmel-quadspi: Avoid overwriting delay register settings

[ Upstream commit 329ca3eed4a9a161515a8714be6ba182321385c7 ]

Previously the MR and SCR registers were just set with the supposedly
required values, from cached register values (cached reg content
initialized to zero).

All parts fixed here did not consider the current register (cache)
content, which would make future support of cs_setup, cs_hold, and
cs_inactive impossible.

Setting SCBR in atmel_qspi_setup() erases a possible DLYBS setting from
atmel_qspi_set_cs_timing().  The DLYBS setting is applied by ORing over
the current setting, without resetting the bits first.  All writes to MR
did not consider possible settings of DLYCS and DLYBCT.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Fixes: f732646d0ccd ("spi: atmel-quadspi: Add support for configuring CS timing")
Link: https://patch.msgid.link/20240918082744.379610-2-ada@thorsis.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: airoha: remove read cache in airoha_snand_dirmap_read()

[ Upstream commit fffca269e4f31c3633c6d810833ba1b184407915 ]

Current upstream driver reports errors running mtd_oobtest kernel module
test:

root@OpenWrt:/# insmod mtd_test.ko
root@OpenWrt:/# insmod mtd_oobtest.ko dev=5
[ 7023.730584] =================================================
[ 7023.736399] mtd_oobtest: MTD device: 5
[ 7023.740160] mtd_oobtest: MTD device size 3670016, eraseblock size 131072, page size 2048, count of eraseblocks 28, pages per eraseblock 64, OOB size 128
[ 7023.753837] mtd_test: scanning for bad eraseblocks
[ 7023.758636] mtd_test: scanned 28 eraseblocks, 0 are bad
[ 7023.763861] mtd_oobtest: test 1 of 5
[ 7024.042076] mtd_oobtest: writing OOBs of whole device
[ 7024.682069] mtd_oobtest: written up to eraseblock 0
[ 7041.962077] mtd_oobtest: written 28 eraseblocks
[ 7041.966626] mtd_oobtest: verifying all eraseblocks
[ 7041.972276] mtd_oobtest: error @addr[0x0:0x0] 0xff -> 0xe diff 0xf1
[ 7041.978550] mtd_oobtest: error @addr[0x0:0x1] 0xff -> 0x10 diff 0xef
[ 7041.984932] mtd_oobtest: error @addr[0x0:0x2] 0xff -> 0x82 diff 0x7d
[ 7041.991293] mtd_oobtest: error @addr[0x0:0x3] 0xff -> 0x10 diff 0xef
[ 7041.997659] mtd_oobtest: error @addr[0x0:0x4] 0xff -> 0x0 diff 0xff
[ 7042.003942] mtd_oobtest: error @addr[0x0:0x5] 0xff -> 0x8a diff 0x75
[ 7042.010294] mtd_oobtest: error @addr[0x0:0x6] 0xff -> 0x20 diff 0xdf
[ 7042.016659] mtd_oobtest: error @addr[0x0:0x7] 0xff -> 0x1 diff 0xfe
[ 7042.022935] mtd_oobtest: error @addr[0x0:0x8] 0xff -> 0x2e diff 0xd1
[ 7042.029295] mtd_oobtest: error @addr[0x0:0x9] 0xff -> 0x40 diff 0xbf
[ 7042.035661] mtd_oobtest: error @addr[0x0:0xa] 0xff -> 0x0 diff 0xff
[ 7042.041935] mtd_oobtest: error @addr[0x0:0xb] 0xff -> 0x89 diff 0x76
[ 7042.048300] mtd_oobtest: error @addr[0x0:0xc] 0xff -> 0x82 diff 0x7d
[ 7042.054662] mtd_oobtest: error @addr[0x0:0xd] 0xff -> 0x15 diff 0xea
[ 7042.061014] mtd_oobtest: error @addr[0x0:0xe] 0xff -> 0x90 diff 0x6f
[ 7042.067380] mtd_oobtest: error @addr[0x0:0xf] 0xff -> 0x0 diff 0xff
....
[ 7432.421369] mtd_oobtest: error @addr[0x237800:0x36] 0xff -> 0x5f diff 0xa0
[ 7432.428242] mtd_oobtest: error @addr[0x237800:0x37] 0xff -> 0x21 diff 0xde
[ 7432.435118] mtd_oobtest: error: verify failed at 0x237800
[ 7432.440510] mtd_oobtest: error: too many errors
[ 7432.445053] mtd_oobtest: error -1 occurred

The above errors are due to the buggy logic in the 'read cache' available
in airoha_snand_dirmap_read() routine since there are some corner cases
where we are missing data updates. Since we do not get any read/write speed
improvement using the cache (according to the mtd_speedtest kernel
module test), in order to fix the mtd_oobtest test, remove the 'read cache'
in airoha_snand_dirmap_read routine. Now the driver is passing all the
tests available in mtd_test suite.

Fixes: a403997c1201 ("spi: airoha: add SPI-NAND Flash controller driver")
Tested-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20240919-airoha-spi-fixes-v2-1-cb0f0ed9920a@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

lib/sbitmap: define swap_lock as raw_spinlock_t

[ Upstream commit 65f666c6203600053478ce8e34a1db269a8701c9 ]

When called from sbitmap_queue_get(), sbitmap_deferred_clear() may be run
with preempt disabled. In RT kernel, spin_lock() can sleep, then warning
of "BUG: sleeping function called from invalid context" can be triggered.

Fix it by replacing it with raw_spin_lock.

Cc: Yang Yang <yang.yang@vivo.com>
Fixes: 72d04bdcf3f7 ("sbitmap: fix io hung due to race on sbitmap_word::cleared")
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Yang Yang <yang.yang@vivo.com>
Link: https://lore.kernel.org/r/20240919021709.511329-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: spi-fsl-lpspi: Undo runtime PM changes at driver exit time

[ Upstream commit 3b577de206d52dbde9428664b6d823d35a803d75 ]

It's important to undo pm_runtime_use_autosuspend() with
pm_runtime_dont_use_autosuspend() at driver exit time unless driver
initially enabled pm_runtime with devm_pm_runtime_enable()
(which handles it for you).

Hence, call pm_runtime_dont_use_autosuspend() at driver exit time
to fix it.

Fixes: 944c01a889d9 ("spi: lpspi: enable runtime pm for lpspi")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Link: https://patch.msgid.link/20240906021251.610462-1-ruanjinjie@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: atmel-quadspi: Undo runtime PM changes at driver exit time

[ Upstream commit 438efb23f9581659495b85f1f6c7d5946200660c ]

It's important to undo pm_runtime_use_autosuspend() with
pm_runtime_dont_use_autosuspend() at driver exit time unless driver
initially enabled pm_runtime with devm_pm_runtime_enable()
(which handles it for you).

Hence, call pm_runtime_dont_use_autosuspend() at driver exit time
to fix it.

Fixes: 4a2f83b7f780 ("spi: atmel-quadspi: add runtime pm support")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Link: https://patch.msgid.link/20240906023956.1004440-1-ruanjinjie@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: airoha: fix airoha_snand_{write,read}_data data_len estimation

[ Upstream commit 0e58637eb968c636725dcd6c7055249b4e5326fb ]

Fix data length written and read in airoha_snand_write_data and
airoha_snand_read_data routines respectively if it is bigger than
SPI_MAX_TRANSFER_SIZE.

Fixes: a403997c1201 ("spi: airoha: add SPI-NAND Flash controller driver")
Tested-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20240913-airoha-spi-fixes-v1-2-de2e74ed4664@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: airoha: fix dirmap_{read,write} operations

[ Upstream commit 2e6bbfe7b0c0607001b784082c2685b134174fac ]

SPI_NFI_READ_FROM_CACHE_DONE bit must be written at the end of
dirmap_read operation even if it is already set.
In the same way, SPI_NFI_LOAD_TO_CACHE_DONE bit must be written at the
end of dirmap_write operation even if it is already set.
For this reason use regmap_write_bits() instead of regmap_set_bits().
This patch fixes mtd_pagetest kernel module test.

Fixes: a403997c1201 ("spi: airoha: add SPI-NAND Flash controller driver")
Tested-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20240913-airoha-spi-fixes-v1-1-de2e74ed4664@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: fix to don't set SB_RDONLY in f2fs_handle_critical_error()

[ Upstream commit 930c6ab93492c4b15436524e704950b364b2930c ]

syzbot reports a f2fs bug as below:

------------[ cut here ]------------
WARNING: CPU: 1 PID: 58 at kernel/rcu/sync.c:177 rcu_sync_dtor+0xcd/0x180 kernel/rcu/sync.c:177
CPU: 1 UID: 0 PID: 58 Comm: kworker/1:2 Not tainted 6.10.0-syzkaller-12562-g1722389b0d86 #0
Workqueue: events destroy_super_work
RIP: 0010:rcu_sync_dtor+0xcd/0x180 kernel/rcu/sync.c:177
Call Trace:
 percpu_free_rwsem+0x41/0x80 kernel/locking/percpu-rwsem.c:42
 destroy_super_work+0xec/0x130 fs/super.c:282
 process_one_work kernel/workqueue.c:3231 [inline]
 process_scheduled_works+0xa2c/0x1830 kernel/workqueue.c:3312
 worker_thread+0x86d/0xd40 kernel/workqueue.c:3390
 kthread+0x2f0/0x390 kernel/kthread.c:389
 ret_from_fork+0x4b/0x80 arch/x86/kernel/process.c:147
 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:244

As Christian Brauner pointed out [1]: the root cause is f2fs sets
SB_RDONLY flag in internal function, rather than setting the flag
covered w/ sb->s_umount semaphore via remount procedure, then below
race condition causes this bug:

- freeze_super()
 - sb_wait_write(sb, SB_FREEZE_WRITE)
 - sb_wait_write(sb, SB_FREEZE_PAGEFAULT)
 - sb_wait_write(sb, SB_FREEZE_FS)
- f2fs_handle_critical_error
 - sb->s_flags |= SB_RDONLY
- thaw_super
 - thaw_super_locked
  - sb_rdonly() is true, so it skips
    sb_freeze_unlock(sb, SB_FREEZE_FS)
  - deactivate_locked_super

Since f2fs has almost the same logic as ext4 [2] when handling critical
error in filesystem if it mounts w/ errors=remount-ro option:
- set CP_ERROR_FLAG flag which indicates filesystem is stopped
- record errors to superblock
- set SB_RDONLY falg
Once we set CP_ERROR_FLAG flag, all writable interfaces can detect the
flag and stop any further updates on filesystem. So, it is safe to not
set SB_RDONLY flag, let's remove the logic and keep in line w/ ext4 [3].

[1] https://lore.kernel.org/all/20240729-himbeeren-funknetz-96e62f9c7aee@brauner
[2] https://lore.kernel.org/all/20240729132721.hxih6ehigadqf7wx@quack3
[3] https://lore.kernel.org/linux-ext4/20240805201241.27286-1-jack@suse.cz

Fixes: b62e71be2110 ("f2fs: support errors=remount-ro|continue|panic mountoption")
Reported-by: syzbot+20d7e439f76bbbd863a7@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/000000000000b90a8e061e21d12f@google.com/
Cc: Jan Kara <jack@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Chao Yu <chao@kernel.org>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: get rid of online repaire on corrupted directory

[ Upstream commit 884ee6dc85b959bc152f15bca80c30f06069e6c4 ]

syzbot reports a f2fs bug as below:

kernel BUG at fs/f2fs/inode.c:896!
RIP: 0010:f2fs_evict_inode+0x1598/0x15c0 fs/f2fs/inode.c:896
Call Trace:
 evict+0x532/0x950 fs/inode.c:704
 dispose_list fs/inode.c:747 [inline]
 evict_inodes+0x5f9/0x690 fs/inode.c:797
 generic_shutdown_super+0x9d/0x2d0 fs/super.c:627
 kill_block_super+0x44/0x90 fs/super.c:1696
 kill_f2fs_super+0x344/0x690 fs/f2fs/super.c:4898
 deactivate_locked_super+0xc4/0x130 fs/super.c:473
 cleanup_mnt+0x41f/0x4b0 fs/namespace.c:1373
 task_work_run+0x24f/0x310 kernel/task_work.c:228
 ptrace_notify+0x2d2/0x380 kernel/signal.c:2402
 ptrace_report_syscall include/linux/ptrace.h:415 [inline]
 ptrace_report_syscall_exit include/linux/ptrace.h:477 [inline]
 syscall_exit_work+0xc6/0x190 kernel/entry/common.c:173
 syscall_exit_to_user_mode_prepare kernel/entry/common.c:200 [inline]
 __syscall_exit_to_user_mode_work kernel/entry/common.c:205 [inline]
 syscall_exit_to_user_mode+0x279/0x370 kernel/entry/common.c:218
 do_syscall_64+0x100/0x230 arch/x86/entry/common.c:89
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0010:f2fs_evict_inode+0x1598/0x15c0 fs/f2fs/inode.c:896

Online repaire on corrupted directory in f2fs_lookup() can generate
dirty data/meta while racing w/ readonly remount, it may leave dirty
inode after filesystem becomes readonly, however, checkpoint() will
skips flushing dirty inode in a state of readonly mode, result in
above panic.

Let's get rid of online repaire in f2fs_lookup(), and leave the work
to fsck.f2fs.

Fixes: 510022a85839 ("f2fs: add F2FS_INLINE_DOTS to recover missing dot dentries")
Reported-by: syzbot+ebea2790904673d7c618@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/000000000000a7b20f061ff2d56a@google.com
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: prevent atomic file from being dirtied before commit

[ Upstream commit fccaa81de87e80b1809906f7e438e5766fbdc172 ]

Keep atomic file clean while updating and make it dirtied during commit
in order to avoid unnecessary and excessive inode updates in the previous
fix.

Fixes: 4bf78322346f ("f2fs: mark inode dirty for FI_ATOMIC_COMMITTED flag")
Signed-off-by: Daeho Jeong <daehojeong@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: compress: don't redirty sparse cluster during {,de}compress

[ Upstream commit f785cec298c95d00058560c0715233294a04b8f3 ]

In f2fs_do_write_data_page, when the data block is NULL_ADDR, it skips
writepage considering that it has been already truncated.
This results in an infinite loop as the PAGECACHE_TAG_TOWRITE tag is not
cleared during the writeback process for a compressed file including
NULL_ADDR in compress_mode=user.

This is the reproduction process:

1. dd if=/dev/zero bs=4096 count=1024 seek=1024 of=testfile
2. f2fs_io compress testfile
3. dd if=/dev/zero bs=4096 count=1 conv=notrunc of=testfile
4. f2fs_io decompress testfile

To prevent the problem, let's check whether the cluster is fully
allocated before redirty its pages.

Fixes: 5fdb322ff2c2 ("f2fs: add F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE")
Reviewed-by: Sungjong Seo <sj1557.seo@samsung.com>
Reviewed-by: Sunmin Jeong <s_min.jeong@samsung.com>
Tested-by: Jaewook Kim <jw5454.kim@samsung.com>
Signed-off-by: Yeongjin Gil <youngjin.gil@samsung.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: fix to avoid use-after-free in f2fs_stop_gc_thread()

[ Upstream commit c7f114d864ac91515bb07ac271e9824a20f5ed95 ]

syzbot reports a f2fs bug as below:

 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0x241/0x360 lib/dump_stack.c:114
 print_report+0xe8/0x550 mm/kasan/report.c:491
 kasan_report+0x143/0x180 mm/kasan/report.c:601
 kasan_check_range+0x282/0x290 mm/kasan/generic.c:189
 instrument_atomic_read_write include/linux/instrumented.h:96 [inline]
 atomic_fetch_add_relaxed include/linux/atomic/atomic-instrumented.h:252 [inline]
 __refcount_add include/linux/refcount.h:184 [inline]
 __refcount_inc include/linux/refcount.h:241 [inline]
 refcount_inc include/linux/refcount.h:258 [inline]
 get_task_struct include/linux/sched/task.h:118 [inline]
 kthread_stop+0xca/0x630 kernel/kthread.c:704
 f2fs_stop_gc_thread+0x65/0xb0 fs/f2fs/gc.c:210
 f2fs_do_shutdown+0x192/0x540 fs/f2fs/file.c:2283
 f2fs_ioc_shutdown fs/f2fs/file.c:2325 [inline]
 __f2fs_ioctl+0x443a/0xbe60 fs/f2fs/file.c:4325
 vfs_ioctl fs/ioctl.c:51 [inline]
 __do_sys_ioctl fs/ioctl.c:907 [inline]
 __se_sys_ioctl+0xfc/0x170 fs/ioctl.c:893
 do_syscall_x64 arch/x86/entry/common.c:52 [inline]
 do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

The root cause is below race condition, it may cause use-after-free
issue in sbi->gc_th pointer.

- remount
 - f2fs_remount
  - f2fs_stop_gc_thread
   - kfree(gc_th)
- f2fs_ioc_shutdown
 - f2fs_do_shutdown
  - f2fs_stop_gc_thread
   - kthread_stop(gc_th->f2fs_gc_task)
   : sbi->gc_thread = NULL;

We will call f2fs_do_shutdown() in two paths:
- for f2fs_ioc_shutdown() path, we should grab sb->s_umount semaphore
for fixing.
- for f2fs_shutdown() path, it's safe since caller has already grabbed
sb->s_umount semaphore.

Reported-by: syzbot+1a8e2b31f2ac9bd3d148@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/linux-f2fs-devel/0000000000005c7ccb061e032b9b@google.com
Fixes: 7950e9ac638e ("f2fs: stop gc/discard thread after fs shutdown")
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: atomic: fix to truncate pagecache before on-disk metadata truncation

[ Upstream commit ebd3309aec6271c4616573b0cb83ea25e623070a ]

We should always truncate pagecache while truncating on-disk data.

Fixes: a46bebd502fe ("f2fs: synchronize atomic write aborts")
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: fix to wait page writeback before setting gcing flag

[ Upstream commit a4d7f2b3238fd5f76b9e6434a0bd5d2e29049cff ]

Soft IRQ Thread
- f2fs_write_end_io
- f2fs_defragment_range
 - set_page_private_gcing
 - type = WB_DATA_TYPE(page, false);
 : assign type w/ F2FS_WB_CP_DATA
 due to page_private_gcing() is true
  - dec_page_count() w/ wrong type
  - end_page_writeback()

Value of F2FS_WB_CP_DATA reference count may become negative under above
race condition, the root cause is we missed to wait page writeback before
setting gcing page private flag, let's fix it.

Fixes: 2d1fe8a86bf5 ("f2fs: fix to tag gcing flag on page during file defragment")
Fixes: 4961acdd65c9 ("f2fs: fix to tag gcing flag on page during block migration")
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: Create COW inode from parent dentry for atomic write

[ Upstream commit 8c1b787938fd86bab27a1492fa887408c75fec2b ]

The i_pino in f2fs_inode_info has the previous parent's i_ino when inode
was renamed, which may cause f2fs_ioc_start_atomic_write to fail.
If file_wrong_pino is true and i_nlink is 1, then to find a valid pino,
we should refer to the dentry from inode.

To resolve this issue, let's get parent inode using parent dentry
directly.

Fixes: 3db1de0e582c ("f2fs: change the current atomic write way")
Reviewed-by: Sungjong Seo <sj1557.seo@samsung.com>
Reviewed-by: Sunmin Jeong <s_min.jeong@samsung.com>
Signed-off-by: Yeongjin Gil <youngjin.gil@samsung.com>
Reviewed-by: Daeho Jeong <daehojeong@google.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: fix to avoid racing in between read and OPU dio write

[ Upstream commit 0cac51185e65dc2a20686184e02f3cafc99eb202 ]

If lfs mode is on, buffered read may race w/ OPU dio write as below,
it may cause buffered read hits unwritten data unexpectly, and for
dio read, the race condition exists as well.

Thread A Thread B
- f2fs_file_write_iter
 - f2fs_dio_write_iter
  - __iomap_dio_rw
   - f2fs_iomap_begin
    - f2fs_map_blocks
     - __allocate_data_block
      - allocated blkaddr #x
       - iomap_dio_submit_bio
- f2fs_file_read_iter
 - filemap_read
  - f2fs_read_data_folio
   - f2fs_mpage_readpages
    - f2fs_map_blocks
     : get blkaddr #x
    - f2fs_submit_read_bio
IRQ
- f2fs_read_end_io
 : read IO on blkaddr #x complete
IRQ
- iomap_dio_bio_end_io
 : direct write IO on blkaddr #x complete

In LFS mode, if there is inflight dio, let's wait for its completion,
this policy won't cover all race cases, however it is a tradeoff which
avoids abusing lock around IO paths.

Fixes: f847c699cff3 ("f2fs: allow out-place-update for direct IO in LFS mode")
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: reduce expensive checkpoint trigger frequency

[ Upstream commit aaf8c0b9ae042494cb4585883b15c1332de77840 ]

We may trigger high frequent checkpoint for below case:
1. mkdir /mnt/dir1; set dir1 encrypted
2. touch /mnt/file1; fsync /mnt/file1
3. mkdir /mnt/dir2; set dir2 encrypted
4. touch /mnt/file2; fsync /mnt/file2
...

Although, newly created dir and file are not related, due to
commit bbf156f7afa7 ("f2fs: fix lost xattrs of directories"), we will
trigger checkpoint whenever fsync() comes after a new encrypted dir
created.

In order to avoid such performance regression issue, let's record an
entry including directory's ino in global cache whenever we update
directory's xattr data, and then triggerring checkpoint() only if
xattr metadata of target file's parent was updated.

This patch updates to cover below no encryption case as well:
1) parent is checkpointed
2) set_xattr(dir) w/ new xnid
3) create(file)
4) fsync(file)

Fixes: bbf156f7afa7 ("f2fs: fix lost xattrs of directories")
Reported-by: wangzijie <wangzijie1@honor.com>
Reported-by: Zhiguo Niu <zhiguo.niu@unisoc.com>
Tested-by: Zhiguo Niu <zhiguo.niu@unisoc.com>
Reported-by: Yunlei He <heyunlei@hihonor.com>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: atomic: fix to avoid racing w/ GC

[ Upstream commit 1a0bd289a5db1df8df8fab949633a0b8d3f235ee ]

Case #1:
SQLite App GC Thread Kworker Shrinker
- f2fs_ioc_start_atomic_write

- f2fs_ioc_commit_atomic_write
 - f2fs_commit_atomic_write
  - filemap_write_and_wait_range
  : write atomic_file's data to cow_inode
echo 3 > drop_caches
to drop atomic_file's
cache.
- f2fs_gc
 - gc_data_segment
  - move_data_page
   - set_page_dirty

- writepages
 - f2fs_do_write_data_page
 : overwrite atomic_file's data
   to cow_inode
  - f2fs_down_write(&fi->i_gc_rwsem[WRITE])
  - __f2fs_commit_atomic_write
  - f2fs_up_write(&fi->i_gc_rwsem[WRITE])

Case #2:
SQLite App GC Thread Kworker
- f2fs_ioc_start_atomic_write

- __writeback_single_inode
 - do_writepages
  - f2fs_write_cache_pages
   - f2fs_write_single_data_page
    - f2fs_do_write_data_page
    : write atomic_file's data to cow_inode
- f2fs_gc
 - gc_data_segment
  - move_data_page
   - set_page_dirty

- writepages
 - f2fs_do_write_data_page
 : overwrite atomic_file's data to cow_inode
- f2fs_ioc_commit_atomic_write

In above cases racing in between atomic_write and GC, previous
data in atomic_file may be overwrited to cow_file, result in
data corruption.

This patch introduces PAGE_PRIVATE_ATOMIC_WRITE bit flag in page.private,
and use it to indicate that there is last dirty data in atomic file,
and the data should be writebacked into cow_file, if the flag is not
tagged in page, we should never write data across files.

Fixes: 3db1de0e582c ("f2fs: change the current atomic write way")
Cc: Daeho Jeong <daehojeong@google.com>
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: powerpc/p10-aes-gcm - Disable CRYPTO_AES_GCM_P10

[ Upstream commit 44ac4625ea002deecd0c227336c95b724206c698 ]

Data mismatch found when testing ipsec tunnel with AES/GCM crypto.
Disabling CRYPTO_AES_GCM_P10 in Kconfig for this feature.

Fixes: fd0e9b3e2ee6 ("crypto: p10-aes-gcm - An accelerated AES/GCM stitched implementation")
Fixes: cdcecfd9991f ("crypto: p10-aes-gcm - Glue code for AES/GCM stitched implementation")
Fixes: 45a4672b9a6e2 ("crypto: p10-aes-gcm - Update Kconfig and Makefile")
Signed-off-by: Danny Tsen <dtsen@linux.ibm.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: caam - Pad SG length when allocating hash edesc

[ Upstream commit 5124bc96162667766f6120b19f57a640c2eccb2a ]

Because hardware will read in multiples of 4 SG entries, ensure
the allocated length is always padded.  This was already done
by some callers of ahash_edesc_alloc, but ahash_digest was conspicuously
missing.

In any case, doing it in the allocation function ensures that the
memory is always there.

Reported-by: Guangwu Zhang <guazhang@redhat.com>
Fixes: a5e5c13398f3 ("crypto: caam - fix S/G table passing page boundary")
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: fix initial getattr on write delegation

[ Upstream commit bf92e5008b17f935a6de8b708551e02c2294121c ]

At this point in compound processing, currentfh refers to the parent of
the file, not the file itself. Get the correct dentry from the delegation
stateid instead.

Fixes: c5967721e106 ("NFSD: handle GETATTR conflict with write delegation")
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: untangle code in nfsd4_deleg_getattr_conflict()

[ Upstream commit a078a7dc0eaa9db288ae45319f7f7503968af546 ]

The code in nfsd4_deleg_getattr_conflict() is convoluted and buggy.

With this patch we:
 - properly handle non-nfsd leases.  We must not assume flc_owner is a
    delegation unless fl_lmops == &nfsd_lease_mng_ops
 - move the main code out of the for loop
 - have a single exit which calls nfs4_put_stid()
   (and other exits which don't need to call that)

[ jlayton: refactored on top of Neil's other patch: nfsd: fix
   nfsd4_deleg_getattr_conflict in presence of third party lease ]

Fixes: c5967721e106 ("NFSD: handle GETATTR conflict with write delegation")
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: return -EINVAL when namelen is 0

[ Upstream commit 22451a16b7ab7debefce660672566be887db1637 ]

When we have a corrupted main.sqlite in /var/lib/nfs/nfsdcld/, it may
result in namelen being 0, which will cause memdup_user() to return
ZERO_SIZE_PTR.
When we access the name.data that has been assigned the value of
ZERO_SIZE_PTR in nfs4_client_to_reclaim(), null pointer dereference is
triggered.

[ T1205] ==================================================================
[ T1205] BUG: KASAN: null-ptr-deref in nfs4_client_to_reclaim+0xe9/0x260
[ T1205] Read of size 1 at addr 0000000000000010 by task nfsdcld/1205
[ T1205]
[ T1205] CPU: 11 PID: 1205 Comm: nfsdcld Not tainted 5.10.0-00003-g2c1423731b8d #406
[ T1205] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS ?-20190727_073836-buildvm-ppc64le-16.ppc.fedoraproject.org-3.fc31 04/01/2014
[ T1205] Call Trace:
[ T1205]  dump_stack+0x9a/0xd0
[ T1205]  ? nfs4_client_to_reclaim+0xe9/0x260
[ T1205]  __kasan_report.cold+0x34/0x84
[ T1205]  ? nfs4_client_to_reclaim+0xe9/0x260
[ T1205]  kasan_report+0x3a/0x50
[ T1205]  nfs4_client_to_reclaim+0xe9/0x260
[ T1205]  ? nfsd4_release_lockowner+0x410/0x410
[ T1205]  cld_pipe_downcall+0x5ca/0x760
[ T1205]  ? nfsd4_cld_tracking_exit+0x1d0/0x1d0
[ T1205]  ? down_write_killable_nested+0x170/0x170
[ T1205]  ? avc_policy_seqno+0x28/0x40
[ T1205]  ? selinux_file_permission+0x1b4/0x1e0
[ T1205]  rpc_pipe_write+0x84/0xb0
[ T1205]  vfs_write+0x143/0x520
[ T1205]  ksys_write+0xc9/0x170
[ T1205]  ? __ia32_sys_read+0x50/0x50
[ T1205]  ? ktime_get_coarse_real_ts64+0xfe/0x110
[ T1205]  ? ktime_get_coarse_real_ts64+0xa2/0x110
[ T1205]  do_syscall_64+0x33/0x40
[ T1205]  entry_SYSCALL_64_after_hwframe+0x67/0xd1
[ T1205] RIP: 0033:0x7fdbdb761bc7
[ T1205] Code: 0f 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 514
[ T1205] RSP: 002b:00007fff8c4b7248 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
[ T1205] RAX: ffffffffffffffda RBX: 000000000000042b RCX: 00007fdbdb761bc7
[ T1205] RDX: 000000000000042b RSI: 00007fff8c4b75f0 RDI: 0000000000000008
[ T1205] RBP: 00007fdbdb761bb0 R08: 0000000000000000 R09: 0000000000000001
[ T1205] R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000042b
[ T1205] R13: 0000000000000008 R14: 00007fff8c4b75f0 R15: 0000000000000000
[ T1205] ==================================================================

Fix it by checking namelen.

Signed-off-by: Li Lingfeng <lilingfeng3@huawei.com>
Fixes: 74725959c33c ("nfsd: un-deprecate nfsdcld")
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Scott Mayhew <smayhew@redhat.com>
Tested-by: Scott Mayhew <smayhew@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: call cache_put if xdr_reserve_space returns NULL

[ Upstream commit d078cbf5c38de83bc31f83c47dcd2184c04a50c7 ]

If not enough buffer space available, but idmap_lookup has triggered
lookup_fn which calls cache_get and returns successfully. Then we
missed to call cache_put here which pairs with cache_get.

Fixes: ddd1ea563672 ("nfsd4: use xdr_reserve_space in attribute encoding")
Signed-off-by: Guoqing Jiang <guoqing.jiang@linux.dev>
Reviwed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ntb: Force physically contiguous allocation of rx ring buffers

[ Upstream commit 061a785a114f159e990ea8ed8d1b7dca4b41120f ]

Physical addresses under IOVA on x86 platform are mapped contiguously
as a side effect before the patch that removed CONFIG_DMA_REMAP. The
NTB rx buffer ring is a single chunk DMA buffer that is allocated
against the NTB PCI device. If the receive side is using a DMA device,
then the buffers are remapped against the DMA device before being
submitted via the dmaengine API. This scheme becomes a problem when
the physical memory is discontiguous. When dma_map_page() is called
on the kernel virtual address from the dma_alloc_coherent() call, the
new IOVA mapping no longer points to all the physical memory allocated
due to being discontiguous. Change dma_alloc_coherent() to dma_alloc_attrs()
in order to force DMA_ATTR_FORCE_CONTIGUOUS attribute. This is the best
fix for the circumstance. A potential future solution may be having the DMA
mapping API providing a way to alias an existing IOVA mapping to a new
device perhaps.

This fix is not to fix the patch pointed to by the fixes tag, but to fix
the issue arised in the ntb_transport driver on x86 platforms after the
said patch is applied.

Reported-by: Jerry Dai <jerry.dai@intel.com>
Fixes: f5ff79fddf0e ("dma-mapping: remove CONFIG_DMA_REMAP")
Tested-by: Jerry Dai <jerry.dai@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ntb_perf: Fix printk format

[ Upstream commit 1501ae7479c8d0f66efdbfdc9ae8d6136cefbd37 ]

The correct printk format is %pa or %pap, but not %pa[p].

Fixes: 99a06056124d ("NTB: ntb_perf: Fix address err in perf_copy_chunk")
Signed-off-by: Max Hawking <maxahawking@sonnenkinder.org>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ntb: intel: Fix the NULL vs IS_ERR() bug for debugfs_create_dir()

[ Upstream commit e229897d373a87ee09ec5cc4ecd4bb2f895fc16b ]

The debugfs_create_dir() function returns error pointers.
It never returns NULL. So use IS_ERR() to check it.

Fixes: e26a5843f7f5 ("NTB: Split ntb_hw_intel and ntb_transport drivers")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Jon Mason <jdmason@kudzu.us>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/irdma: fix error message in irdma_modify_qp_roce()

[ Upstream commit 9f0eafe86ea0a589676209d0cff1a1ed49a037d3 ]

Use a correct field max_dest_rd_atomic instead of max_rd_atomic for the
error output.

Found by Linux Verification Center (linuxtesting.org) with Svace.

Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs")
Signed-off-by: Vitaliy Shevtsov <v.shevtsov@maxima.ru>
Link: https://lore.kernel.org/stable/20240916165817.14691-1-v.shevtsov%40maxima.ru
Link: https://patch.msgid.link/20240916165817.14691-1-v.shevtsov@maxima.ru
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/cxgb4: Added NULL check for lookup_atid

[ Upstream commit e766e6a92410ca269161de059fff0843b8ddd65f ]

The lookup_atid() function can return NULL if the ATID is
invalid or does not exist in the identifier table, which
could lead to dereferencing a null pointer without a
check in the `act_establish()` and `act_open_rpl()` functions.
Add a NULL check to prevent null pointer dereferencing.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: cfdda9d76436 ("RDMA/cxgb4: Add driver for Chelsio T4 RNIC")
Signed-off-by: Mikhail Lobanov <m.lobanov@rosalinux.ru>
Link: https://patch.msgid.link/20240912145844.77516-1-m.lobanov@rosalinux.ru
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Fix ah error counter in sw stat not increasing

[ Upstream commit 39c047d4047a1242aeefa87513174b56a91080ab ]

There are several error cases where hns_roce_create_ah() returns
directly without jumping to sw stat path, thus leading to a problem
that the ah error counter does not increase.

Fixes: ee20cc17e9d8 ("RDMA/hns: Support DSCP")
Fixes: eb7854d63db5 ("RDMA/hns: Support SW stats with debugfs")
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240912115700.2016443-1-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

riscv: Fix fp alignment bug in perf_callchain_user()

[ Upstream commit 22ab08955ea13be04a8efd20cc30890e0afaa49c ]

The standard RISC-V calling convention said:
"The stack grows downward and the stack pointer is always
kept 16-byte aligned".

So perf_callchain_user() should check whether 16-byte aligned for fp.

Link: https://riscv.org/wp-content/uploads/2015/01/riscv-calling.pdf
Fixes: dbeb90b0c1eb ("riscv: Add perf callchain support")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Cc: Björn Töpel <bjorn@kernel.org>
Link: https://lore.kernel.org/r/20240708032847.2998158-2-ruanjinjie@huawei.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: qcom-ep: Enable controller resources like PHY only after refclk is available

[ Upstream commit d3745e3ae6c0eec517d431be926742b6e8b9b64a ]

qcom_pcie_enable_resources() is called by qcom_pcie_ep_probe() and it
enables the controller resources like clocks, regulator, PHY. On one of the
new unreleased Qcom SoC, PHY enablement depends on the active refclk. And
on all of the supported Qcom endpoint SoCs, refclk comes from the host
(RC). So calling qcom_pcie_enable_resources() without refclk causes the
NoC (Network On Chip) error in the endpoint SoC and in turn results in a
whole SoC crash and rebooting into EDL (Emergency Download) mode which is
an unrecoverable state.

But qcom_pcie_enable_resources() is already called by
qcom_pcie_perst_deassert() when PERST# is deasserted, and refclk is
available at that time.

Hence, remove the unnecessary call to qcom_pcie_enable_resources() from
qcom_pcie_ep_probe() to prevent the above mentioned crash.

It should be noted that this commit prevents the crash only under normal
working condition (booting endpoint before host), but the crash may also
occur if PERST# assert happens at the wrong time. For avoiding the crash
completely, it is recommended to use SRIS mode which allows the endpoint
SoC to generate its own refclk. The driver is not supporting SRIS mode
currently, but will be added in the future.

Fixes: 869bc5253406 ("PCI: dwc: ep: Fix DBI access failure for drivers requiring refclk from host")
Link: https://lore.kernel.org/linux-pci/20240830082319.51387-1-manivannan.sadhasivam@linaro.org
Tested-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/mlx5: Obtain upper net device only when needed

[ Upstream commit 3ed7f9e239938a0cfaf3689e2f545229ecabec06 ]

Report the upper device's state as the RDMA port state only in RoCE LAG or
switchdev LAG.

Fixes: 27f9e0ccb6da ("net/mlx5: Lag, Add single RDMA device in multiport mode")
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Michael Guralnik <michaelgur@nvidia.com>
Link: https://patch.msgid.link/20240909173025.30422-3-michaelgur@nvidia.com
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Fix restricted __le16 degrades to integer issue

[ Upstream commit f4ccc0a2a0c5977540f519588636b5bc81aae2db ]

Fix sparse warnings: restricted __le16 degrades to integer.

Fixes: 5a87279591a1 ("RDMA/hns: Support hns HW stats")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202409080508.g4mNSLwy-lkp@intel.com/
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240909065331.3950268-1-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Optimize hem allocation performance

[ Upstream commit fe51f6254d81f5a69c31df16353d6539b2b51630 ]

When allocating MTT hem, for each hop level of each hem that is being
allocated, the driver iterates the hem list to find out whether the
bt page has been allocated in this hop level. If not, allocate a new
one and splice it to the list. The time complexity is O(n^2) in worst
cases.

Currently the allocation for-loop uses 'unit' as the step size. This
actually has taken into account the reuse of last-hop-level MTT bt
pages by multiple buffer pages. Thus pages of last hop level will
never have been allocated, so there is no need to iterate the hem list
in last hop level.

Removing this unnecessary iteration can reduce the time complexity to
O(n).

Fixes: 38389eaa4db1 ("RDMA/hns: Add mtr support for mixed multihop addressing")
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240906093444.3571619-9-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Fix 1bit-ECC recovery address in non-4K OS

[ Upstream commit ce196f6297c7f3ab7780795e40efd6c521f60c8b ]

The 1bit-ECC recovery address read from HW only contain bits 64:12, so
it should be fixed left-shifted 12 bits when used.

Currently, the driver will shift the address left by PAGE_SHIFT when
used, which is wrong in non-4K OS.

Fixes: 2de949abd6a5 ("RDMA/hns: Recover 1bit-ECC error of RAM on chip")
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240906093444.3571619-8-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Fix VF triggering PF reset in abnormal interrupt handler

[ Upstream commit 4321feefa5501a746ebf6a7d8b59e6b955ae1860 ]

In abnormal interrupt handler, a PF reset will be triggered even if
the device is a VF. It should be a VF reset.

Fixes: 2b9acb9a97fe ("RDMA/hns: Add the process of AEQ overflow for hip08")
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240906093444.3571619-7-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Fix spin_unlock_irqrestore() called with IRQs enabled

[ Upstream commit 74d315b5af180220d561684d15897730135733a6 ]

Fix missuse of spin_lock_irq()/spin_unlock_irq() when
spin_lock_irqsave()/spin_lock_irqrestore() was hold.

This was discovered through the lock debugging, and the corresponding
log is as follows:

raw_local_irq_restore() called with IRQs enabled
WARNING: CPU: 96 PID: 2074 at kernel/locking/irqflag-debug.c:10 warn_bogus_irq_restore+0x30/0x40
...
Call trace:
 warn_bogus_irq_restore+0x30/0x40
 _raw_spin_unlock_irqrestore+0x84/0xc8
 add_qp_to_list+0x11c/0x148 [hns_roce_hw_v2]
 hns_roce_create_qp_common.constprop.0+0x240/0x780 [hns_roce_hw_v2]
 hns_roce_create_qp+0x98/0x160 [hns_roce_hw_v2]
 create_qp+0x138/0x258
 ib_create_qp_kernel+0x50/0xe8
 create_mad_qp+0xa8/0x128
 ib_mad_port_open+0x218/0x448
 ib_mad_init_device+0x70/0x1f8
 add_client_context+0xfc/0x220
 enable_device_and_get+0xd0/0x140
 ib_register_device.part.0+0xf4/0x1c8
 ib_register_device+0x34/0x50
 hns_roce_register_device+0x174/0x3d0 [hns_roce_hw_v2]
 hns_roce_init+0xfc/0x2c0 [hns_roce_hw_v2]
 __hns_roce_hw_v2_init_instance+0x7c/0x1d0 [hns_roce_hw_v2]
 hns_roce_hw_v2_init_instance+0x9c/0x180 [hns_roce_hw_v2]

Fixes: 9a4435375cd1 ("IB/hns: Add driver files for hns RoCE driver")
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240906093444.3571619-6-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Fix the overflow risk of hem_list_calc_ba_range()

[ Upstream commit d586628b169d14bbf36be64d2b3ec9d9d2fe0432 ]

The max value of 'unit' and 'hop_num' is 2^24 and 2, so the value of
'step' may exceed the range of u32. Change the type of 'step' to u64.

Fixes: 38389eaa4db1 ("RDMA/hns: Add mtr support for mixed multihop addressing")
Signed-off-by: wenglianfa <wenglianfa@huawei.com>
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240906093444.3571619-5-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Fix Use-After-Free of rsv_qp on HIP08

[ Upstream commit fd8489294dd2beefb70f12ec4f6132aeec61a4d0 ]

Currently rsv_qp is freed before ib_unregister_device() is called
on HIP08. During the time interval, users can still dereg MR and
rsv_qp will be used in this process, leading to a UAF. Move the
release of rsv_qp after calling ib_unregister_device() to fix it.

Fixes: 70f92521584f ("RDMA/hns: Use the reserved loopback QPs to free MR before destroying MPT")
Signed-off-by: wenglianfa <wenglianfa@huawei.com>
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240906093444.3571619-3-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/hns: Don't modify rq next block addr in HIP09 QPC

[ Upstream commit 6928d264e328e0cb5ee7663003a6e46e4cba0a7e ]

The field 'rq next block addr' in QPC can be updated by driver only
on HIP08. On HIP09 HW updates this field while driver is not allowed.

Fixes: 926a01dc000d ("RDMA/hns: Add QP operations support for hip08 SoC")
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20240906093444.3571619-2-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

watchdog: imx_sc_wdt: Don't disable WDT in suspend

[ Upstream commit 2d9d6d300fb0a4ae4431bb308027ac9385746d42 ]

Parts of the suspend and resume chain is left unprotected if we disable
the WDT here.

>From experiments we can see that the SCU disables and re-enables the WDT
when we enter and leave suspend to ram. By not touching the WDT here we
are protected by the WDT all the way to the SCU.

Signed-off-by: Jonas Blixt <jonas.blixt@actia.se>
CC: Anson Huang <anson.huang@nxp.com>
Fixes: 986857acbc9a ("watchdog: imx_sc: Add i.MX system controller watchdog support")
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20240801121845.1465765-1-jonas.blixt@actia.se
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@linux-watchdog.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/mlx5: Fix MR cache temp entries cleanup

[ Upstream commit 7ebb00cea49db641b458edef0ede389f7004821d ]

Fix the cleanup of the temp cache entries that are dynamically created
in the MR cache.

The cleanup of the temp cache entries is currently scheduled only when a
new entry is created. Since in the cleanup of the entries only the mkeys
are destroyed and the cache entry stays in the cache, subsequent
registrations might reuse the entry and it will eventually be filled with
new mkeys without cleanup ever getting scheduled again.

On workloads that register and deregister MRs with a wide range of
properties we see the cache ends up holding many cache entries, each
holding the max number of mkeys that were ever used through it.

Additionally, as the cleanup work is scheduled to run over the whole
cache, any mkey that is returned to the cache after the cleanup was
scheduled will be held for less than the intended 30 seconds timeout.

Solve both issues by dropping the existing remove_ent_work and reusing
the existing per-entry work to also handle the temp entries cleanup.

Schedule the work to run with a 30 seconds delay every time we push an
mkey to a clean temp entry.
This ensures the cleanup runs on each entry only 30 seconds after the
first mkey was pushed to an empty entry.

As we have already been distinguishing between persistent and temp entries
when scheduling the cache_work_func, it is not being scheduled in any
other flows for the temp entries.

Another benefit from moving to a per-entry cleanup is we now not
required to hold the rb_tree mutex, thus enabling other flow to run
concurrently.

Fixes: dd1b913fb0d0 ("RDMA/mlx5: Cache all user cacheable mkeys on dereg MR flow")
Signed-off-by: Michael Guralnik <michaelgur@nvidia.com>
Link: https://patch.msgid.link/e4fa4bb03bebf20dceae320f26816cd2dde23a26.1725362530.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/mlx5: Drop redundant work canceling from clean_keys()

[ Upstream commit 30e6bd8d3b5639f8f4261e5e6c0917ce264b8dc2 ]

The canceling of dealyed work in clean_keys() is a leftover from years
back and was added to prevent races in the cleanup process of MR cache.
The cleanup process was rewritten a few years ago and the canceling of
delayed work and flushing of workqueue was added before the call to
clean_keys().

Signed-off-by: Michael Guralnik <michaelgur@nvidia.com>
Link: https://patch.msgid.link/943d21f5a9dba7b98a3e1d531e3561ffe9745d71.1725362530.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Stable-dep-of: 7ebb00cea49d ("RDMA/mlx5: Fix MR cache temp entries cleanup")
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/mlx5: Limit usage of over-sized mkeys from the MR cache

[ Upstream commit ee6d57a2e13d11ce9050cfc3e3b69ef707a44a63 ]

When searching the MR cache for suitable cache entries, don't use mkeys
larger than twice the size required for the MR.
This should ensure the usage of mkeys closer to the minimal required size
and reduce memory waste.

On driver init we create entries for mkeys with clear attributes and
powers of 2 sizes from 4 to the max supported size.
This solves the issue for anyone using mkeys that fit these
requirements.

In the use case where an MR is registered with different attributes,
like an access flag we can't UMR, we'll create a new cache entry to store
it upon dereg.
Without this fix, any later registration with same attributes and smaller
size will use the newly created cache entry and it's mkeys, disregarding
the memory waste of using mkeys larger than required.

For example, one worst-case scenario can be when registering and
deregistering a 1GB mkey with ATS enabled which will cause the creation of
a new cache entry to hold those type of mkeys. A user registering a 4k MR
with ATS will end up using the new cache entry and an mkey that can
support a 1GB MR, thus wasting x250k memory than actually needed in the HW.

Additionally, allow all small registration to use the smallest size
cache entry that is initialized on driver load even if size is larger
than twice the required size.

Fixes: 73d09b2fe833 ("RDMA/mlx5: Introduce mlx5r_cache_rb_key")
Signed-off-by: Michael Guralnik <michaelgur@nvidia.com>
Link: https://patch.msgid.link/8ba3a6e3748aace2026de8b83da03aba084f78f4.1725362530.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/mlx5: Fix counter update on MR cache mkey creation

[ Upstream commit 6f5cd6ac9a4201e4ba6f10b76a9da8044d6e38b0 ]

After an mkey is created, update the counter for pending mkeys before
reshceduling the work that is filling the cache.

Rescheduling the work with a full MR cache entry and a wrong 'pending'
counter will cause us to miss disabling the fill_to_high_water flag.
Thus leaving the cache full but with an indication that it's still
needs to be filled up to it's full size (2 * limit).
Next time an mkey will be taken from the cache, we'll unnecessarily
continue the process of filling the cache to it's full size.

Fixes: 57e7071683ef ("RDMA/mlx5: Implement mkeys management via LIFO queue")
Signed-off-by: Michael Guralnik <michaelgur@nvidia.com>
Link: https://patch.msgid.link/0f44f462ba22e45f72cb3d0ec6a748634086b8d0.1725362530.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/erdma: Return QP state in erdma_query_qp

[ Upstream commit e77127ff6416b17e0b3e630ac46ee5c9a6570f57 ]

Fix qp_state and cur_qp_state to return correct values in
struct ib_qp_attr.

Fixes: 155055771704 ("RDMA/erdma: Add verbs implementation")
Signed-off-by: Cheng Xu <chengyou@linux.alibaba.com>
Link: https://patch.msgid.link/20240902112920.58749-4-chengyou@linux.alibaba.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: kirin: Fix buffer overflow in kirin_pcie_parse_port()

[ Upstream commit c500a86693a126c9393e602741e348f80f1b0fc5 ]

Within kirin_pcie_parse_port(), the pcie->num_slots is compared to
pcie->gpio_id_reset size (MAX_PCI_SLOTS) which is correct and would lead
to an overflow.

Thus, fix condition to pcie->num_slots + 1 >= MAX_PCI_SLOTS and move
pcie->num_slots increment below the if-statement to avoid out-of-bounds
array access.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: b22dbbb24571 ("PCI: kirin: Support PERST# GPIOs for HiKey970 external PEX 8606 bridge")
Link: https://lore.kernel.org/linux-pci/20240903115823.30647-1-adiupina@astralinux.ru
Signed-off-by: Alexandra Diupina <adiupina@astralinux.ru>
[kwilczynski: commit log]
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iommufd: Check the domain owner of the parent before creating a nesting domain

[ Upstream commit 73183ad6ea51029d04b098286dcee98d715015f1 ]

This check was missed, before we can pass a struct iommu_domain to a
driver callback we need to validate that the domain was created by that
driver.

Fixes: bd529dbb661d ("iommufd: Add a nested HW pagetable object")
Link: https://patch.msgid.link/r/0-v1-c8770519edde+1a-iommufd_nesting_ops_jgg@nvidia.com
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

dt-bindings: PCI: layerscape-pci: Replace fsl,lx2160a-pcie with fsl,lx2160ar2-pcie

[ Upstream commit 1a1bf58897d20fc38b454dfecd2771e142d36966 ]

The fsl,lx2160a-pcie compatible is used for mobivel according to the
Documentation/devicetree/bindings/pci/layerscape-pcie-gen4.txt file.

Whereas the fsl,layerscape-pcie is used for DesignWare PCIe controller binding.

So change it to fsl,lx2160ar2-pcie and allow a fall back to fsl,ls2088a-pcie.

While at it, sort compatible string.

Fixes: 24cd7ecb3886 ("dt-bindings: PCI: layerscape-pci: Convert to YAML format")
Link: https://lore.kernel.org/linux-pci/20240826-2160r2-v1-1-106340d538d6@nxp.com
Signed-off-by: Frank Li <Frank.Li@nxp.com>
[kwilczynski: commit log]
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

IB/core: Fix ib_cache_setup_one error flow cleanup

[ Upstream commit 1403c8b14765eab805377dd3b75e96ace8747aed ]

When ib_cache_update return an error, we exit ib_cache_setup_one
instantly with no proper cleanup, even though before this we had
already successfully done gid_table_setup_one, that results in
the kernel WARN below.

Do proper cleanup using gid_table_cleanup_one before returning
the err in order to fix the issue.

WARNING: CPU: 4 PID: 922 at drivers/infiniband/core/cache.c:806 gid_table_release_one+0x181/0x1a0
Modules linked in:
CPU: 4 UID: 0 PID: 922 Comm: c_repro Not tainted 6.11.0-rc1+ #3
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
RIP: 0010:gid_table_release_one+0x181/0x1a0
Code: 44 8b 38 75 0c e8 2f cb 34 ff 4d 8b b5 28 05 00 00 e8 23 cb 34 ff 44 89 f9 89 da 4c 89 f6 48 c7 c7 d0 58 14 83 e8 4f de 21 ff <0f> 0b 4c 8b 75 30 e9 54 ff ff ff 48 8    3 c4 10 5b 5d 41 5c 41 5d 41
RSP: 0018:ffffc90002b835b0 EFLAGS: 00010286
RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffffff811c8527
RDX: 0000000000000000 RSI: ffffffff811c8534 RDI: 0000000000000001
RBP: ffff8881011b3d00 R08: ffff88810b3abe00 R09: 205d303839303631
R10: 666572207972746e R11: 72746e6520444947 R12: 0000000000000001
R13: ffff888106390000 R14: ffff8881011f2110 R15: 0000000000000001
FS:  00007fecc3b70800(0000) GS:ffff88813bd00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000020000340 CR3: 000000010435a001 CR4: 00000000003706b0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
 ? show_regs+0x94/0xa0
 ? __warn+0x9e/0x1c0
 ? gid_table_release_one+0x181/0x1a0
 ? report_bug+0x1f9/0x340
 ? gid_table_release_one+0x181/0x1a0
 ? handle_bug+0xa2/0x110
 ? exc_invalid_op+0x31/0xa0
 ? asm_exc_invalid_op+0x16/0x20
 ? __warn_printk+0xc7/0x180
 ? __warn_printk+0xd4/0x180
 ? gid_table_release_one+0x181/0x1a0
 ib_device_release+0x71/0xe0
 ? __pfx_ib_device_release+0x10/0x10
 device_release+0x44/0xd0
 kobject_put+0x135/0x3d0
 put_device+0x20/0x30
 rxe_net_add+0x7d/0xa0
 rxe_newlink+0xd7/0x190
 nldev_newlink+0x1b0/0x2a0
 ? __pfx_nldev_newlink+0x10/0x10
 rdma_nl_rcv_msg+0x1ad/0x2e0
 rdma_nl_rcv_skb.constprop.0+0x176/0x210
 netlink_unicast+0x2de/0x400
 netlink_sendmsg+0x306/0x660
 __sock_sendmsg+0x110/0x120
 ____sys_sendmsg+0x30e/0x390
 ___sys_sendmsg+0x9b/0xf0
 ? kstrtouint+0x6e/0xa0
 ? kstrtouint_from_user+0x7c/0xb0
 ? get_pid_task+0xb0/0xd0
 ? proc_fail_nth_write+0x5b/0x140
 ? __fget_light+0x9a/0x200
 ? preempt_count_add+0x47/0xa0
 __sys_sendmsg+0x61/0xd0
 do_syscall_64+0x50/0x110
 entry_SYSCALL_64_after_hwframe+0x76/0x7e

Fixes: 1901b91f9982 ("IB/core: Fix potential NULL pointer dereference in pkey cache")
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Maher Sanalla <msanalla@nvidia.com>
Link: https://patch.msgid.link/79137687d829899b0b1c9835fcb4b258004c439a.1725273354.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pinctrl: mvebu: Fix devinit_dove_pinctrl_probe function

[ Upstream commit c25478419f6fd3f74c324a21ec007cf14f2688d7 ]

When an error occurs during the execution of the function
__devinit_dove_pinctrl_probe, the clk is not properly disabled.

Fix this by calling clk_disable_unprepare before return.

Fixes: ba607b6238a1 ("pinctrl: mvebu: make pdma clock on dove mandatory")
Signed-off-by: Wang Jianzheng <wangjianzheng@vivo.com>
Link: https://lore.kernel.org/20240829064823.19808-1-wangjianzheng@vivo.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: fix refcount leak when file is unhashed after being found

[ Upstream commit 8a7926176378460e0d91e02b03f0ff20a8709a60 ]

If we wait_for_construction and find that the file is no longer hashed,
and we're going to retry the open, the old nfsd_file reference is
currently leaked. Put the reference before retrying.

Fixes: c6593366c0bf ("nfsd: don't kill nfsd_files because of lease break error")
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Tested-by: Youzhong Yang <youzhong@gmail.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: remove unneeded EEXIST error check in nfsd_do_file_acquire

[ Upstream commit 81a95c2b1d605743220f28db04b8da13a65c4059 ]

Given that we do the search and insertion while holding the i_lock, I
don't think it's possible for us to get EEXIST here. Remove this case.

Fixes: c6593366c0bf ("nfsd: don't kill nfsd_files because of lease break error")
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Tested-by: Youzhong Yang <youzhong@gmail.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: rockchip: rk3588: Fix 32k clock name for pmu_24m_32k_100m_src_p

[ Upstream commit 0d02e8d284a45bfa8997ebe8764437b8eb6b108b ]

The 32kHz input clock is named "xin32k" in the driver,
so the name "32k" appears to be a typo in this case. Lets fix this.

Signed-off-by: Alexander Shiyan <eagle.alexander923@gmail.com>
Reviewed-by: Dragan Simic <dsimic@manjaro.org>
Fixes: f1c506d152ff ("clk: rockchip: add clock controller for the RK3588")
Link: https://lore.kernel.org/r/20240829052820.3604-1-eagle.alexander923@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: starfive: Use pm_runtime_resume_and_get to fix pm_runtime_get_sync() usage

[ Upstream commit 55c312c1b2be6d43e39c280ad6ab4b711e545b89 ]

We need to call pm_runtime_put_noidle() when pm_runtime_get_sync()
fails, so use pm_runtime_resume_and_get() instead. this function
will handle this.

Fixes: dae5448a327ed ("clk: starfive: Add StarFive JH7110 Video-Output clock driver")
Signed-off-by: Yuntao Liu <liuyuntao12@huawei.com>
Link: https://lore.kernel.org/r/20240815093853.757487-1-liuyuntao12@huawei.com
Reviewed-by: Xingyu Wu <xingyu.wu@starfivetech.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: ti: dra7-atl: Fix leak of of_nodes

[ Upstream commit 9d6e9f10e2e031fb7bfb3030a7d1afc561a28fea ]

This fix leaking the of_node references in of_dra7_atl_clk_probe().

The docs for of_parse_phandle_with_args() say that the caller must call
of_node_put() on the returned node. This adds the missing of_node_put()
to fix the leak.

Fixes: 9ac33b0ce81f ("CLK: TI: Driver for DRA7 ATL (Audio Tracking Logic)")
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://lore.kernel.org/r/20240826-clk-fix-leak-v1-1-f55418a13aa6@baylibre.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/rtrs-clt: Reset cid to con_num - 1 to stay in bounds

[ Upstream commit 3e4289b29e216a55d08a89e126bc0b37cbad9f38 ]

In the function init_conns(), after the create_con() and create_cm() for
loop if something fails. In the cleanup for loop after the destroy tag, we
access out of bound memory because cid is set to clt_path->s.con_num.

This commits resets the cid to clt_path->s.con_num - 1, to stay in bounds
in the cleanup loop later.

Fixes: 6a98d71daea1 ("RDMA/rtrs: client: main functionality")
Signed-off-by: Md Haris Iqbal <haris.iqbal@ionos.com>
Signed-off-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com>
Link: https://patch.msgid.link/20240821112217.41827-7-haris.iqbal@ionos.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/rtrs: Reset hb_missed_cnt after receiving other traffic from peer

[ Upstream commit 3258cbbd86deaa2675e1799bc3d18bd1ef472641 ]

Reset hb_missed_cnt after receiving traffic from other peer, so
hb is more robust again high load on host or network.

Fixes: 6a98d71daea1 ("RDMA/rtrs: client: main functionality")
Signed-off-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Md Haris Iqbal <haris.iqbal@ionos.com>
Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com>
Link: https://patch.msgid.link/20240821112217.41827-5-haris.iqbal@ionos.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>