[ Upstream commit 43bd09d5b7 ]
If device max_mr_size bits in the range [mr_page_shift+31:mr_page_shift]
are zero, the `min3` function will set clt_path::max_pages_per_mr to
zero.
`alloc_path_reqs` will pass zero, which is invalid, as the third parameter
to `ib_alloc_mr`.
Fixes: 6a98d71dae ("RDMA/rtrs: client: main functionality")
Signed-off-by: Honggang LI <honggangli@163.com>
Link: https://patch.msgid.link/20251229025617.13241-1-honggangli@163.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3c68cf6823 ]
rxe_odp_map_range_and_lock() must release umem_odp->umem_mutex when an
error occurs, including cases where rxe_check_pagefault() fails.
Fixes: 2fae67ab63 ("RDMA/rxe: Add support for Send/Recv/Write/Read with ODP")
Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
Link: https://patch.msgid.link/20251226094112.3042583-1-lizhijian@fujitsu.com
Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 0c56693b06 ]
In get_file_all_info(), if vfs_getattr() fails, the function returns
immediately without freeing the allocated filename, leading to a memory
leak.
Fix this by freeing the filename before returning in this error case.
Fixes: 5614c8c487 ("ksmbd: replace generic_fillattr with vfs_getattr")
Signed-off-by: Zilin Guan <zilin@seu.edu.cn>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit bed2a6bd20 ]
Use READ_ONCE and WRITE_ONCE when operating on g2h_fence->done
to prevent the compiler from ignoring important modifications
to its value.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Suggested-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patch.msgid.link/20251222201957.63245-5-jonathan.cavitt@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
(cherry picked from commit b5179dbd1c)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 7fc4da6a30 ]
Implement async partition scan to avoid IO hang when reading partition
tables. Similar to nvme_partition_scan_work(), partition scanning is
deferred to a work queue to prevent deadlocks.
When partition scan happens synchronously during add_disk(), IO errors
can cause the partition scan to wait while holding ub->mutex, which
can deadlock with other operations that need the mutex.
Changes:
- Add partition_scan_work to ublk_device structure
- Implement ublk_partition_scan_work() to perform async scan
- Always suppress sync partition scan during add_disk()
- Schedule async work after add_disk() for trusted daemons
- Add flush_work() in ublk_stop_dev() before grabbing ub->mutex
Reviewed-by: Caleb Sander Mateos <csander@purestorage.com>
Reported-by: Yoav Cohen <yoav@nvidia.com>
Closes: https://lore.kernel.org/linux-block/DM4PR12MB63280C5637917C071C2F0D65A9A8A@DM4PR12MB6328.namprd12.prod.outlook.com/
Fixes: 71f28f3136 ("ublk_drv: add io_uring based userspace block driver")
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 529d4d6327 ]
Implement NUMA-friendly memory allocation for ublk driver to improve
performance on multi-socket systems.
This commit includes the following changes:
1. Rename __queues to queues, dropping the __ prefix since the field is
now accessed directly throughout the codebase rather than only through
the ublk_get_queue() helper.
2. Remove the queue_size field from struct ublk_device as it is no longer
needed.
3. Move queue allocation and deallocation into ublk_init_queue() and
ublk_deinit_queue() respectively, improving encapsulation. This
simplifies ublk_init_queues() and ublk_deinit_queues() to just
iterate and call the per-queue functions.
4. Add ublk_get_queue_numa_node() helper function to determine the
appropriate NUMA node for a queue by finding the first CPU mapped
to that queue via tag_set.map[HCTX_TYPE_DEFAULT].mq_map[] and
converting it to a NUMA node using cpu_to_node(). This function is
called internally by ublk_init_queue() to determine the allocation
node.
5. Allocate each queue structure on its local NUMA node using
kvzalloc_node() in ublk_init_queue().
6. Allocate the I/O command buffer on the same NUMA node using
alloc_pages_node().
This reduces memory access latency on multi-socket NUMA systems by
ensuring each queue's data structures are local to the CPUs that
access them.
Reviewed-by: Caleb Sander Mateos <csander@purestorage.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Stable-dep-of: 7fc4da6a30 ("ublk: scan partition in async way")
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 7ad6ef91d8 ]
The variable mddev->private is first assigned to conf and then checked:
conf = mddev->private;
if (!conf) ...
If conf is NULL, then mddev->private is also NULL. In this case,
null-pointer dereferences can occur when calling raid5_quiesce():
raid5_quiesce(mddev, true);
raid5_quiesce(mddev, false);
since mddev->private is assigned to conf again in raid5_quiesce(), and conf
is dereferenced in several places, for example:
conf->quiesce = 0;
wake_up(&conf->wait_for_quiescent);
To fix this issue, the function should unlock mddev and return before
invoking raid5_quiesce() when conf is NULL, following the existing pattern
in raid5_change_consistency_policy().
Fixes: fa1944bbe6 ("md/raid5: Wait sync io to finish before changing group cnt")
Signed-off-by: Tuo Li <islituo@gmail.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Link: https://lore.kernel.org/linux-raid/20251225130326.67780-1-islituo@gmail.com
Signed-off-by: Yu Kuai <yukuai@fnnas.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3d70e0fb0f ]
In bnxt_qplib_alloc_init_hwq(), while allocating memory for PDE table
driver incorrectly is using the "pg_size" value passed to the function.
Fixed to use the right value 4K. Also, fixed the allocation size for
PBL table.
Fixes: 0c4dcd6028 ("RDMA/bnxt_re: Refactor hardware queue memory allocation")
Signed-off-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com>
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Link: https://patch.msgid.link/20251223131855.145955-1-kalesh-anakkur.purayil@broadcom.com
Reviewed-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c31f4aa8fe ]
The kunit_run_irq_test() helper allows a function to be run in hardirq
and softirq contexts (in addition to the task context). It does this by
running the user-provided function concurrently in the three contexts,
until either a timeout has expired or a number of iterations have
completed in the normal task context.
However, on setups where the initialisation of the hardirq and softirq
contexts (or, indeed, the scheduling of those tasks) is significantly
slower than the function execution, it's possible for that number of
iterations to be exceeded before any runs in irq contexts actually
occur. This occurs with the polyval.test_polyval_preparekey_in_irqs
test, which runs 20000 iterations of the relatively fast preparekey
function, and therefore fails often under many UML, 32-bit arm, m68k and
other environments.
Instead, ensure that the max_iterations limit counts executions in all
three contexts, and requires at least one of each. This will cause the
test to continue iterating until at least the irq contexts have been
tested, or the 1s wall-clock limit has been exceeded. This causes the
test to pass in all of my environments.
In so doing, we also update the task counters to atomic ints, to better
match both the 'int' max_iterations input, and to ensure they are
correctly updated across contexts.
Finally, we also fix a few potential assertion messages to be
less-specific to the original crypto usecases.
Fixes: 950a81224e ("lib/crypto: tests: Add hash-test-template.h and gen-hash-testvecs.py")
Signed-off-by: David Gow <davidgow@google.com>
Link: https://lore.kernel.org/r/20251219085259.1163048-1-davidgow@google.com
Signed-off-by: Eric Biggers <ebiggers@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 9b68a1cc96 ]
Commit ef56081d18 ("RDMA/bnxt_re: RoCE related hardware counters
update") added three new counters and placed them after
BNXT_RE_OUT_OF_SEQ_ERR.
BNXT_RE_OUT_OF_SEQ_ERR acts as a boundary marker for allocating hardware
statistics with different num_counters values on chip_gen_p5_p7 devices.
As a result, BNXT_RE_NUM_STD_COUNTERS are used when allocating
hw_stats, which leads to an out-of-bounds write in
bnxt_re_copy_err_stats().
The counters BNXT_RE_REQ_CQE_ERROR, BNXT_RE_RESP_CQE_ERROR, and
BNXT_RE_RESP_REMOTE_ACCESS_ERRS are applicable to generic hardware, not
only p5/p7 devices.
Fix this by moving these counters before BNXT_RE_OUT_OF_SEQ_ERR so they
are included in the generic counter set.
Fixes: ef56081d18 ("RDMA/bnxt_re: RoCE related hardware counters update")
Reported-by: Yingying Zheng <zhengyingying@sangfor.com.cn>
Signed-off-by: Ding Hui <dinghui@sangfor.com.cn>
Link: https://patch.msgid.link/20251208072110.28874-1-dinghui@sangfor.com.cn
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Tested-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f01765a236 ]
The bnxt_re SEND path checks wr->send_flags to enable features such as
IP checksum offload. However, send_flags is a bitmask and may contain
multiple flags (e.g. IB_SEND_SIGNALED | IB_SEND_IP_CSUM), while the
existing code uses a switch() statement that only matches when
send_flags is exactly IB_SEND_IP_CSUM.
As a result, checksum offload is not enabled when additional SEND
flags are present.
Replace the switch() with a bitmask test:
if (wr->send_flags & IB_SEND_IP_CSUM)
This ensures IP checksum offload is enabled correctly when multiple
SEND flags are used.
Fixes: 1ac5a40479 ("RDMA/bnxt_re: Add bnxt_re RoCE driver")
Signed-off-by: Alok Tiwari <alok.a.tiwari@oracle.com>
Link: https://patch.msgid.link/20251219093308.2415620-1-alok.a.tiwari@oracle.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>
[ Upstream commit 3fbd97618f ]
Replace the bogus "GPL v2" with "GPL" as MODULE_LICNSE() string. The
value does not declare the module's exact license, but only lets the
module loader test whether the module is Free Software or not.
See commit bf7fbeeae6 ("module: Cure the MODULE_LICENSE "GPL" vs.
"GPL v2" bogosity") in the details of the issue. The fix is to use
"GPL" for all modules under any variant of the GPL.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Fixes: 4b2b5e142f ("drm: Move GEM memory managers into modules")
Link: https://patch.msgid.link/20251209140141.94407-3-tzimmermann@suse.de
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 145a417a39 ]
RCFW_COMM_CONS_PCI_BAR_REGION is defined as BAR 2, so checking
!creq_db->reg.bar_id is incorrect and always false.
pci_resource_start() returns the BAR base address, and a value of 0
indicates that the BAR is unassigned. Update the condition to test
bar_base == 0 instead.
This ensures the driver detects and logs an error for an unassigned
RCFW communication BAR.
Fixes: cee0c7bba4 ("RDMA/bnxt_re: Refactor command queue management code")
Signed-off-by: Alok Tiwari <alok.a.tiwari@oracle.com>
Link: https://patch.msgid.link/20251217100158.752504-1-alok.a.tiwari@oracle.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>
[ Upstream commit 8aaa848ead ]
Fix missing comparison operator for RDMA_NETWORK_ROCE_V1 in the
conditional statement. The constant was used directly instead of
being compared with net_type, causing the condition to always
evaluate to true.
Fixes: 1c15b4f2a4 ("RDMA/core: Modify enum ib_gid_type and enum rdma_network_type")
Signed-off-by: Jang Ingyu <ingyujang25@korea.ac.kr>
Link: https://patch.msgid.link/20251219041508.1725947-1-ingyujang25@korea.ac.kr
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 85463eb6a4 ]
The page size used for device might in some cases be smaller than
PAGE_SIZE what results in a negative shift when calculating the number of
host pages in PAGE_SIZE for a debug log. Remove the debug line together
with the calculation.
Fixes: 40909f664d ("RDMA/efa: Add EFA verbs implementation")
Link: https://patch.msgid.link/r/20251210173656.8180-1-mrgolin@amazon.com
Reviewed-by: Tom Sela <tomsela@amazon.com>
Reviewed-by: Yonatan Nachum <ynachum@amazon.com>
Signed-off-by: Michael Margolin <mrgolin@amazon.com>
Reviewed-by: Gal Pressman <gal.pressman@linux.dev>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6f05611728 ]
irdma_net_event() should not dereference anything from "neigh" (alias
"ptr") until it has checked that the event is NETEVENT_NEIGH_UPDATE.
Other events come with different structures pointed to by "ptr" and they
may be smaller than struct neighbour.
Move the read of neigh->dev under the NETEVENT_NEIGH_UPDATE case.
The bug is mostly harmless, but it triggers KASAN on debug kernels:
BUG: KASAN: stack-out-of-bounds in irdma_net_event+0x32e/0x3b0 [irdma]
Read of size 8 at addr ffffc900075e07f0 by task kworker/27:2/542554
CPU: 27 PID: 542554 Comm: kworker/27:2 Kdump: loaded Not tainted 5.14.0-630.el9.x86_64+debug #1
Hardware name: [...]
Workqueue: events rt6_probe_deferred
Call Trace:
<IRQ>
dump_stack_lvl+0x60/0xb0
print_address_description.constprop.0+0x2c/0x3f0
print_report+0xb4/0x270
kasan_report+0x92/0xc0
irdma_net_event+0x32e/0x3b0 [irdma]
notifier_call_chain+0x9e/0x180
atomic_notifier_call_chain+0x5c/0x110
rt6_do_redirect+0xb91/0x1080
tcp_v6_err+0xe9b/0x13e0
icmpv6_notify+0x2b2/0x630
ndisc_redirect_rcv+0x328/0x530
icmpv6_rcv+0xc16/0x1360
ip6_protocol_deliver_rcu+0xb84/0x12e0
ip6_input_finish+0x117/0x240
ip6_input+0xc4/0x370
ipv6_rcv+0x420/0x7d0
__netif_receive_skb_one_core+0x118/0x1b0
process_backlog+0xd1/0x5d0
__napi_poll.constprop.0+0xa3/0x440
net_rx_action+0x78a/0xba0
handle_softirqs+0x2d4/0x9c0
do_softirq+0xad/0xe0
</IRQ>
Fixes: 915cc7ac0f ("RDMA/irdma: Add miscellaneous utility definitions")
Link: https://patch.msgid.link/r/20251127143150.121099-1-mschmidt@redhat.com
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d95e99a74e ]
A recent commit modified struct irdma_alloc_ucontext_resp by adding a
member with implicit padding in front of it, though this does not change
the offset of the data members other than m68k. Reported by
scripts/check-uapi.sh:
==== ABI differences detected in include/rdma/irdma-abi.h from 1dd7bde2e91c -> HEAD ====
[C] 'struct irdma_alloc_ucontext_resp' changed:
type size changed from 704 to 640 (in bits)
1 data member deletion:
'__u8 rsvd3[2]', at offset 640 (in bits) at irdma-abi.h:61:1
1 data member insertion:
'__u8 revd3[2]', at offset 592 (in bits) at irdma-abi.h:60:1
Change the size back to the previous version, and remove the implicit
padding by making it explicit and matching what x86-64 would do by placing
max_hw_srq_quanta member into a naturally aligned location.
Fixes: 563e1feb5f ("RDMA/irdma: Add SRQ support")
Link: https://patch.msgid.link/r/20251208133849.315451-1-arnd@kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Geert Uytterhoeven <geert@linux-m68k.org>
Tested-by: Jacob Moroni <jmoroni@google.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 2dc675f614 ]
On a few 32-bit architectures, the newly added ib_user_service_rec
structure is not 64-bit aligned the way it is on most regular ones.
Add explicit padding into the rdma_ucm_query_ib_service_resp and
rdma_ucm_resolve_ib_service structures that embed it, so that the layout
is compatible across all of them.
This is an ABI change on i386, aligning it with x86_64 and the other
64-bit architectures to avoid having to use a compat ioctl handler.
Fixes: 810f874eda ("RDMA/ucma: Support query resolved service records")
Link: https://patch.msgid.link/r/20251208133311.313977-1-arnd@kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 1adaea51c6 ]
On PREEMPT_RT kernels, after rt6_get_pcpu_route() returns NULL, the
current task can be preempted. Another task running on the same CPU
may then execute rt6_make_pcpu_route() and successfully install a
pcpu_rt entry. When the first task resumes execution, its cmpxchg()
in rt6_make_pcpu_route() will fail because rt6i_pcpu is no longer
NULL, triggering the BUG_ON(prev). It's easy to reproduce it by adding
mdelay() after rt6_get_pcpu_route().
Using preempt_disable/enable is not appropriate here because
ip6_rt_pcpu_alloc() may sleep.
Fix this by handling the cmpxchg() failure gracefully on PREEMPT_RT:
free our allocation and return the existing pcpu_rt installed by
another task. The BUG_ON is replaced by WARN_ON_ONCE for non-PREEMPT_RT
kernels where such races should not occur.
Link: https://syzkaller.appspot.com/bug?extid=9b35e9bc0951140d13e6
Fixes: d2d6422f8b ("x86: Allow to enable PREEMPT_RT.")
Reported-by: syzbot+9b35e9bc0951140d13e6@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/6918cd88.050a0220.1c914e.0045.GAE@google.com/T/
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Link: https://patch.msgid.link/20251223051413.124687-1-jiayuan.chen@linux.dev
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6595beb40f ]
rose_kill_by_device() collects sockets into a local array[] and then
iterates over them to disconnect sockets bound to a device being brought
down.
The loop mistakenly indexes array[cnt] instead of array[i]. For cnt <
ARRAY_SIZE(array), this reads an uninitialized entry; for cnt ==
ARRAY_SIZE(array), it is an out-of-bounds read. Either case can lead to
an invalid socket pointer dereference and also leaks references taken
via sock_hold().
Fix the index to use i.
Fixes: 64b8bc7d5f ("net/rose: fix races in rose_kill_by_device()")
Co-developed-by: Fatma Alwasmi <falwasmi@purdue.edu>
Signed-off-by: Fatma Alwasmi <falwasmi@purdue.edu>
Signed-off-by: Pwnverse <stanksal@purdue.edu>
Link: https://patch.msgid.link/20251222212227.4116041-1-ritviktanksalkar@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6e17474aa9 ]
Preference of nexthop with source address broke ECMP for packets with
source addresses which are not in the broadcast domain, but rather added
to loopback/dummy interfaces. Original behaviour was to balance over
nexthops while now it uses the latest nexthop from the group. To fix the
issue introduce next hop scoring system where next hops with source
address equal to requested will always have higher priority.
For the case with 198.51.100.1/32 assigned to dummy0 and routed using
192.0.2.0/24 and 203.0.113.0/24 networks:
2: dummy0: <BROADCAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default qlen 1000
link/ether d6:54:8a:ff:78:f5 brd ff:ff:ff:ff:ff:ff
inet 198.51.100.1/32 scope global dummy0
valid_lft forever preferred_lft forever
7: veth1@if6: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether 06:ed:98:87:6d:8a brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 192.0.2.2/24 scope global veth1
valid_lft forever preferred_lft forever
inet6 fe80::4ed:98ff:fe87:6d8a/64 scope link proto kernel_ll
valid_lft forever preferred_lft forever
9: veth3@if8: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether ae:75:23:38:a0:d2 brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 203.0.113.2/24 scope global veth3
valid_lft forever preferred_lft forever
inet6 fe80::ac75:23ff:fe38:a0d2/64 scope link proto kernel_ll
valid_lft forever preferred_lft forever
~ ip ro list:
default
nexthop via 192.0.2.1 dev veth1 weight 1
nexthop via 203.0.113.1 dev veth3 weight 1
192.0.2.0/24 dev veth1 proto kernel scope link src 192.0.2.2
203.0.113.0/24 dev veth3 proto kernel scope link src 203.0.113.2
before:
for i in {1..255} ; do ip ro get 10.0.0.$i; done | grep veth | awk ' {print $(NF-2)}' | sort | uniq -c:
255 veth3
after:
for i in {1..255} ; do ip ro get 10.0.0.$i; done | grep veth | awk ' {print $(NF-2)}' | sort | uniq -c:
122 veth1
133 veth3
Fixes: 32607a332c ("ipv4: prefer multipath nexthop that matches source address")
Signed-off-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20251221192639.3911901-1-vadim.fedorenko@linux.dev
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit ac782f4e3b ]
When a nexthop object is deleted, it is marked as dead and then
fib_table_flush() is called to flush all the routes that are using the
dead nexthop.
The current logic in fib_table_flush() is to only flush error routes
(e.g., blackhole) when it is called as part of network namespace
dismantle (i.e., with flush_all=true). Therefore, error routes are not
flushed when their nexthop object is deleted:
# ip link add name dummy1 up type dummy
# ip nexthop add id 1 dev dummy1
# ip route add 198.51.100.1/32 nhid 1
# ip route add blackhole 198.51.100.2/32 nhid 1
# ip nexthop del id 1
# ip route show
blackhole 198.51.100.2 nhid 1 dev dummy1
As such, they keep holding a reference on the nexthop object which in
turn holds a reference on the nexthop device, resulting in a reference
count leak:
# ip link del dev dummy1
[ 70.516258] unregister_netdevice: waiting for dummy1 to become free. Usage count = 2
Fix by flushing error routes when their nexthop is marked as dead.
IPv6 does not suffer from this problem.
Fixes: 493ced1ac4 ("ipv4: Allow routes to use nexthop objects")
Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Closes: https://lore.kernel.org/netdev/d943f806-4da6-4970-ac28-b9373b0e63ac@I-love.SAKURA.ne.jp/
Reported-by: syzbot+881d65229ca4f9ae8c84@syzkaller.appspotmail.com
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://patch.msgid.link/20251221144829.197694-1-idosch@nvidia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 58fc7342b5 ]
There exists a kernel oops caused by a BUG_ON(nhead < 0) at
net/core/skbuff.c:2232 in pskb_expand_head().
This bug is triggered as part of the calipso_skbuff_setattr()
routine when skb_cow() is passed headroom > INT_MAX
(i.e. (int)(skb_headroom(skb) + len_delta) < 0).
The root cause of the bug is due to an implicit integer cast in
__skb_cow(). The check (headroom > skb_headroom(skb)) is meant to ensure
that delta = headroom - skb_headroom(skb) is never negative, otherwise
we will trigger a BUG_ON in pskb_expand_head(). However, if
headroom > INT_MAX and delta <= -NET_SKB_PAD, the check passes, delta
becomes negative, and pskb_expand_head() is passed a negative value for
nhead.
Fix the trigger condition in calipso_skbuff_setattr(). Avoid passing
"negative" headroom sizes to skb_cow() within calipso_skbuff_setattr()
by only using skb_cow() to grow headroom.
PoC:
Using `netlabelctl` tool:
netlabelctl map del default
netlabelctl calipso add pass doi:7
netlabelctl map add default address:0::1/128 protocol:calipso,7
Then run the following PoC:
int fd = socket(AF_INET6, SOCK_DGRAM, IPPROTO_UDP);
// setup msghdr
int cmsg_size = 2;
int cmsg_len = 0x60;
struct msghdr msg;
struct sockaddr_in6 dest_addr;
struct cmsghdr * cmsg = (struct cmsghdr *) calloc(1,
sizeof(struct cmsghdr) + cmsg_len);
msg.msg_name = &dest_addr;
msg.msg_namelen = sizeof(dest_addr);
msg.msg_iov = NULL;
msg.msg_iovlen = 0;
msg.msg_control = cmsg;
msg.msg_controllen = cmsg_len;
msg.msg_flags = 0;
// setup sockaddr
dest_addr.sin6_family = AF_INET6;
dest_addr.sin6_port = htons(31337);
dest_addr.sin6_flowinfo = htonl(31337);
dest_addr.sin6_addr = in6addr_loopback;
dest_addr.sin6_scope_id = 31337;
// setup cmsghdr
cmsg->cmsg_len = cmsg_len;
cmsg->cmsg_level = IPPROTO_IPV6;
cmsg->cmsg_type = IPV6_HOPOPTS;
char * hop_hdr = (char *)cmsg + sizeof(struct cmsghdr);
hop_hdr[1] = 0x9; //set hop size - (0x9 + 1) * 8 = 80
sendmsg(fd, &msg, 0);
Fixes: 2917f57b6b ("calipso: Allow the lsm to label the skbuff directly.")
Suggested-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Will Rosenberg <whrosenb@asu.edu>
Acked-by: Paul Moore <paul@paul-moore.com>
Link: https://patch.msgid.link/20251219173637.797418-1-whrosenb@asu.edu
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a48e232210 ]
There is a crash issue when running zero copy XDP_TX action, the crash
log is shown below.
[ 216.122464] Unable to handle kernel paging request at virtual address fffeffff80000000
[ 216.187524] Internal error: Oops: 0000000096000144 [#1] SMP
[ 216.301694] Call trace:
[ 216.304130] dcache_clean_poc+0x20/0x38 (P)
[ 216.308308] __dma_sync_single_for_device+0x1bc/0x1e0
[ 216.313351] stmmac_xdp_xmit_xdpf+0x354/0x400
[ 216.317701] __stmmac_xdp_run_prog+0x164/0x368
[ 216.322139] stmmac_napi_poll_rxtx+0xba8/0xf00
[ 216.326576] __napi_poll+0x40/0x218
[ 216.408054] Kernel panic - not syncing: Oops: Fatal exception in interrupt
For XDP_TX action, the xdp_buff is converted to xdp_frame by
xdp_convert_buff_to_frame(). The memory type of the resulting xdp_frame
depends on the memory type of the xdp_buff. For page pool based xdp_buff
it produces xdp_frame with memory type MEM_TYPE_PAGE_POOL. For zero copy
XSK pool based xdp_buff it produces xdp_frame with memory type
MEM_TYPE_PAGE_ORDER0. However, stmmac_xdp_xmit_back() does not check the
memory type and always uses the page pool type, this leads to invalid
mappings and causes the crash. Therefore, check the xdp_buff memory type
in stmmac_xdp_xmit_back() to fix this issue.
Fixes: bba2556efa ("net: stmmac: Enable RX via AF_XDP zero-copy")
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Hariprasad Kelam <hkelam@marvell.com>
Link: https://patch.msgid.link/20251204071332.1907111-1-wei.fang@nxp.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 85f4b0c650 ]
This patch ensures that the RX ring size (rx_pending) is not
set below the permitted length. This avoids UBSAN
shift-out-of-bounds errors when users passes small or zero
ring sizes via ethtool -G.
Fixes: d45d897984 ("octeontx2-pf: Add basic ethtool support")
Signed-off-by: Anshumali Gaur <agaur@marvell.com>
Link: https://patch.msgid.link/20251219062226.524844-1-agaur@marvell.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 66e245db16 ]
The PMT feature probe creates a child device with device_create().
If device creation fail, the code pass priv->dev (which is an ERR_PTR)
to dev_err_probe(), which is not a valid device pointer.
This patch change the dev_err_probe() call to use the parent auxiliary
device (&auxdev->dev) and update the error message to reference the
parent device name. It ensure correct error reporting and avoid
passing an invalid device pointer.
Fixes: d9a0788093 ("platform/x86/intel/pmt: Add PMT Discovery driver")
Signed-off-by: Alok Tiwari <alok.a.tiwari@oracle.com>
Link: https://patch.msgid.link/20251224095133.115678-1-alok.a.tiwari@oracle.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e44c42c830 ]
The hp_populate_*_elements_from_package() functions in the hp-bioscfg
driver contain out-of-bounds array access vulnerabilities.
These functions parse ACPI packages into internal data structures using
a for loop with index variable 'elem' that iterates through
enum_obj/integer_obj/order_obj/password_obj/string_obj arrays.
When processing multi-element fields like PREREQUISITES and
ENUM_POSSIBLE_VALUES, these functions read multiple consecutive array
elements using expressions like 'enum_obj[elem + reqs]' and
'enum_obj[elem + pos_values]' within nested loops.
The bug is that the bounds check only validated elem, but did not consider
the additional offset when accessing elem + reqs or elem + pos_values.
The fix changes the bounds check to validate the actual accessed index.
Reported-by: Yuhao Jiang <danisjiang@gmail.com>
Reported-by: Junrui Luo <moonafterrain@outlook.com>
Fixes: e6c7b3e155 ("platform/x86: hp-bioscfg: string-attributes")
Signed-off-by: Junrui Luo <moonafterrain@outlook.com>
Link: https://patch.msgid.link/SYBPR01MB788173D7DD4EA2CB6383683DAFB0A@SYBPR01MB7881.ausprd01.prod.outlook.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 665077d78d ]
pds_vfio_dirty_enable() allocates memory for region_info. If
interval_tree_iter_first() returns NULL, the function returns -EINVAL
immediately without freeing the allocated memory, causing a memory leak.
Fix this by jumping to the out_free_region_info label to ensure
region_info is freed.
Fixes: 2e7c6feb4e ("vfio/pds: Add multi-region support")
Signed-off-by: Zilin Guan <zilin@seu.edu.cn>
Link: https://lore.kernel.org/r/20251225143150.1117366-1-zilin@seu.edu.cn
Signed-off-by: Alex Williamson <alex@shazbot.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a4f800c448 ]
Currently in the return path, fwnode_handle_put calls are missing. Just use
_scoped to avoid the issue.
Fixes: 24e31e4747 ("net: mdio: Add RTL9300 MDIO driver")
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://patch.msgid.link/20251217210153.14641-1-rosenp@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 1f4ea4838b ]
mcb bus is not prepared to autoload client drivers with the data defined on
the drivers' MODULE_DEVICE_TABLE. modpost cannot access to mcb_table_id
inside MODULE_DEVICE_TABLE so the data declared inside is ignored.
Add modpost build support for accessing to the mcb_table_id coded on device
drivers' MODULE_DEVICE_TABLE.
Fixes: 3764e82e51 ("drivers: Introduce MEN Chameleon Bus")
Reviewed-by: Jorge Sanjuan Garcia <dev-jorge.sanjuangarcia@duagon.com>
Signed-off-by: Jose Javier Rodriguez Barbarin <dev-josejavier.rodriguez@duagon.com>
Acked-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com>
Link: https://patch.msgid.link/20251202084200.10410-1-dev-josejavier.rodriguez@duagon.com
Signed-off-by: Nicolas Schier <nsc@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b08fc4d0ec ]
Since commit e7e2941300 ("kbuild: split device tree build rules into
scripts/Makefile.dtbs"), it is no longer possible to compile a device tree
blob that is not specified in a make rule
like:
dtb-$(CONFIG_FOO) += foo.dtb
Before the mentioned commit, one could copy a dts file to e.g.
arch/arm64/boot/dts/ (or a new subdirectory) and then convert it to a dtb
file using:
make ARCH=arm64 foo.dtb
In this scenario, both 'dtb-y' and 'dtb-' are empty, and the inclusion of
scripts/Makefile.dtbs relies on 'targets' to contain the MAKECMDGOALS. The
value of 'targets', however, is only final later in the code.
Move the conditional include of scripts/Makefile.dtbs down to where the
value of 'targets' is final. Since Makefile.dtbs updates 'always-y' which is
used as a prerequisite in the build rule, the build rule also needs to move
down.
Fixes: e7e2941300 ("kbuild: split device tree build rules into scripts/Makefile.dtbs")
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Tested-by: Nathan Chancellor <nathan@kernel.org>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patch.msgid.link/20251126100017.1162330-1-thomas.de_schampheleire@nokia.com
Signed-off-by: Nicolas Schier <nsc@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d42bce414d ]
port_fdb_dump() is supposed to only add fdb entries, but we iterate over
the full ARL table, which also includes multicast entries.
So check if the entry is a multicast entry before passing it on to the
callback().
Additionally, the port of those entries is a bitmask, not a port number,
so any included entries would have even be for the wrong port.
Fixes: 1da6df85c6 ("net: dsa: b53: Implement ARL add/del/dump operations")
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Link: https://patch.msgid.link/20251217205756.172123-1-jonas.gorski@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c48c0fd0e1 ]
It looks like the buffer allocated and mapped in add_card() is done
with size RCV_BUFFER_SIZE which is 16 KB and 4KB.
Fixes: 286468210d ("firewire: new driver: nosy - IEEE 1394 traffic sniffer")
Co-developed-by: Thomas Fourier <fourier.thomas@gmail.com>
Signed-off-by: Thomas Fourier <fourier.thomas@gmail.com>
Co-developed-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/20251216165420.38355-2-fourier.thomas@gmail.com
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 00c22b1e84 ]
When kobject_init_and_add() fails in pmt_features_discovery(), the
function returns without calling kobject_put(). This violates the
kobject API contract where kobject_put() must be called even on
initialization failure to properly release allocated resources.
Fixes: d9a0788093 ("platform/x86/intel/pmt: Add PMT Discovery driver")
Signed-off-by: Kaushlendra Kumar <kaushlendra.kumar@intel.com>
Link: https://patch.msgid.link/20251223084041.3832933-1-kaushlendra.kumar@intel.com
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b94f11af9d ]
The LIBWX library code is what calls into phylink, so any user of
it has to select CONFIG_PHYLINK at the moment, with NGBEVF missing this:
x86_64-linux-ld: drivers/net/ethernet/wangxun/libwx/wx_ethtool.o: in function `wx_nway_reset':
wx_ethtool.c:(.text+0x613): undefined reference to `phylink_ethtool_nway_reset'
x86_64-linux-ld: drivers/net/ethernet/wangxun/libwx/wx_ethtool.o: in function `wx_get_link_ksettings':
wx_ethtool.c:(.text+0x62b): undefined reference to `phylink_ethtool_ksettings_get'
x86_64-linux-ld: drivers/net/ethernet/wangxun/libwx/wx_ethtool.o: in function `wx_set_link_ksettings':
wx_ethtool.c:(.text+0x643): undefined reference to `phylink_ethtool_ksettings_set'
x86_64-linux-ld: drivers/net/ethernet/wangxun/libwx/wx_ethtool.o: in function `wx_get_pauseparam':
wx_ethtool.c:(.text+0x65b): undefined reference to `phylink_ethtool_get_pauseparam'
x86_64-linux-ld: drivers/net/ethernet/wangxun/libwx/wx_ethtool.o: in function `wx_set_pauseparam':
wx_ethtool.c:(.text+0x677): undefined reference to `phylink_ethtool_set_pauseparam'
Add the 'select PHYLINK' line in the libwx option directly so this will
always be enabled for all current and future wangxun drivers, and remove
the now duplicate lines.
Fixes: a0008a3658 ("net: wangxun: add ngbevf build")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Link: https://patch.msgid.link/20251216213547.115026-1-arnd@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 472c5dd6b9 ]
When the selftest 'tap.c' is compiled with '-D_FORTIFY_SOURCE=3',
the strcpy() in rtattr_add_strsz() is replaced with a checked
version which causes the test to consistently fail when compiled
with toolchains for which this option is enabled by default.
TAP version 13
1..3
# Starting 3 tests from 1 test cases.
# RUN tap.test_packet_valid_udp_gso ...
*** buffer overflow detected ***: terminated
# test_packet_valid_udp_gso: Test terminated by assertion
# FAIL tap.test_packet_valid_udp_gso
not ok 1 tap.test_packet_valid_udp_gso
# RUN tap.test_packet_valid_udp_csum ...
*** buffer overflow detected ***: terminated
# test_packet_valid_udp_csum: Test terminated by assertion
# FAIL tap.test_packet_valid_udp_csum
not ok 2 tap.test_packet_valid_udp_csum
# RUN tap.test_packet_crash_tap_invalid_eth_proto ...
*** buffer overflow detected ***: terminated
# test_packet_crash_tap_invalid_eth_proto: Test terminated by assertion
# FAIL tap.test_packet_crash_tap_invalid_eth_proto
not ok 3 tap.test_packet_crash_tap_invalid_eth_proto
# FAILED: 0 / 3 tests passed.
# Totals: pass:0 fail:3 xfail:0 xpass:0 skip:0 error:0
A buffer overflow is detected by the fortified glibc __strcpy_chk()
since the __builtin_object_size() of `RTA_DATA(rta)` is incorrectly
reported as 1, even though there is ample space in its bounding
buffer `req`.
Additionally, given that IFLA_IFNAME also expects a null-terminated
string, callers of rtaddr_add_str{,sz}() could simply use the
rtaddr_add_strsz() variant. (which has been renamed to remove the
trailing `sz`) memset() has been used for this function since it
is unchecked and thus circumvents the issue discussed in the
previous paragraph.
Fixes: 2e64fe4624 ("selftests: add few test cases for tap driver")
Signed-off-by: Alice C. Munduruca <alice.munduruca@canonical.com>
Reviewed-by: Cengiz Can <cengiz.can@canonical.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20251216170641.250494-1-alice.munduruca@canonical.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 12cab1191d ]
In async_set_registers(), when usb_submit_urb() fails, the allocated
async_req structure and URB are not freed, causing a memory leak.
The completion callback async_set_reg_cb() is responsible for freeing
these allocations, but it is only called after the URB is successfully
submitted and completes (successfully or with error). If submission
fails, the callback never runs and the memory is leaked.
Fix this by freeing both the URB and the request structure in the error
path when usb_submit_urb() fails.
Reported-by: syzbot+8dd915c7cb0490fc8c52@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=8dd915c7cb0490fc8c52
Fixes: 4d12997a9b ("drivers: net: usb: rtl8150: concurrent URB bugfix")
Signed-off-by: Deepakkumar Karn <dkarn@redhat.com>
Link: https://patch.msgid.link/20251216151304.59865-2-dkarn@redhat.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f0e5126f5e ]
test_case will only take on the formatted name after being
called. This does not work with the way ksft_run() currently
works. Assign the name after the test_case is created.
Fixes: 81236c74db ("selftests: drv-net: psp: add test for auto-adjusting TCP MSS")
Signed-off-by: Daniel Zahka <daniel.zahka@gmail.com>
Link: https://patch.msgid.link/20251216-psp-test-fix-v1-2-3b5a6dde186f@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d52668cac3 ]
test_case will only take on its formatted name after it is called by
the test runner. Move the assignment to test_case.__name__ to when the
test_case is constructed, not called.
Fixes: 8f90dc6e41 ("selftests: drv-net: psp: add basic data transfer and key rotation tests")
Signed-off-by: Daniel Zahka <daniel.zahka@gmail.com>
Link: https://patch.msgid.link/20251216-psp-test-fix-v1-1-3b5a6dde186f@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit df60c332ca ]
During the stress tests, early RX adaptation handshakes can fail, such
as missing the RX_ADAPT ACK or not receiving a coefficient update before
block lock is established. Continuing to retry RX adaptation in this
state is often ineffective if the current mode selection is not viable.
Resetting the RX adaptation retry counter when an RX_ADAPT request fails
to receive ACK or a coefficient update prior to block lock, and clearing
mode_set so the next bring-up performs a fresh mode selection rather
than looping on a likely invalid configuration.
Fixes: 4f3b20bfbb ("amd-xgbe: add support for rx-adaptation")
Signed-off-by: Raju Rangoju <Raju.Rangoju@amd.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
Link: https://patch.msgid.link/20251215151728.311713-1-Raju.Rangoju@amd.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit a9f96dc59b ]
of_find_net_device_by_node() searches net devices by their /sys/class/net/,
entry. It is documented in its kernel-doc that:
* If successful, returns a pointer to the net_device with the embedded
* struct device refcount incremented by one, or NULL on failure. The
* refcount must be dropped when done with the net_device.
We are missing a put_device(&conduit->dev) which we could place at the
end of dsa_tree_find_first_conduit(). But to explain why calling
put_device() right away is safe is the same as to explain why the chosen
solution is different.
The code is very poorly split: dsa_tree_find_first_conduit() was first
introduced in commit 95f510d0b7 ("net: dsa: allow the DSA master to be
seen and changed through rtnetlink") but was first used several commits
later, in commit acc43b7bf5 ("net: dsa: allow masters to join a LAG").
Assume there is a switch with 2 CPU ports and 2 conduits, eno2 and eno3.
When we create a LAG (bonding or team device) and place eno2 and eno3
beneath it, we create a 3rd conduit (the LAG device itself), but this is
slightly different than the first two.
Namely, the cpu_dp->conduit pointer of the CPU ports does not change,
and remains pointing towards the physical Ethernet controllers which are
now LAG ports. Only 2 things change:
- the LAG device has a dev->dsa_ptr which marks it as a DSA conduit
- dsa_port_to_conduit(user port) finds the LAG and not the physical
conduit, because of the dp->cpu_port_in_lag bit being set.
When the LAG device is destroyed, dsa_tree_migrate_ports_from_lag_conduit()
is called and this is where dsa_tree_find_first_conduit() kicks in.
This is the logical mistake and the reason why introducing code in one
patch and using it from another is bad practice. I didn't realize that I
don't have to call of_find_net_device_by_node() again; the cpu_dp->conduit
association was never undone, and is still available for direct (re)use.
There's only one concern - maybe the conduit disappeared in the
meantime, but the netdev_hold() call we made during dsa_port_parse_cpu()
(see previous change) ensures that this was not the case.
Therefore, fixing the code means reimplementing it in the simplest way.
I am blaming the time of use, since this is what "git blame" would show
if we were to monitor for the conduit's kobject's refcount remaining
elevated instead of being freed.
Tested on the NXP LS1028A, using the steps from
Documentation/networking/dsa/configuration.rst section "Affinity of user
ports to CPU ports", followed by (extra prints added by me):
$ ip link del bond0
mscc_felix 0000:00:00.5 swp3: Link is Down
bond0 (unregistering): (slave eno2): Releasing backup interface
fsl_enetc 0000:00:00.2 eno2: Link is Down
mscc_felix 0000:00:00.5 swp0: bond0 disappeared, migrating to eno2
mscc_felix 0000:00:00.5 swp1: bond0 disappeared, migrating to eno2
mscc_felix 0000:00:00.5 swp2: bond0 disappeared, migrating to eno2
mscc_felix 0000:00:00.5 swp3: bond0 disappeared, migrating to eno2
Fixes: acc43b7bf5 ("net: dsa: allow masters to join a LAG")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://patch.msgid.link/20251215150236.3931670-2-vladimir.oltean@nxp.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
