| Age | Commit message (Collapse) | Author |
|
kvm_riscv_vcpu_aia_imsic_put() can run from the scheduler preemption
notifier while the raw runqueue lock is held. Taking vsfile_lock, an
rwlock_t, there violates the lockdep wait-context rules.
Protect the CPU/HGEI pair with a separate raw spinlock. All updates to
that pair retain vsfile_lock and take the new lock, while scheduler-out
uses only the new lock around the HGEIE update. This keeps the wakeup
setup serialized with VS-file publication and release without bringing
wider IMSIC operations into a raw critical section.
Fixes: 4cec89db80ba ("RISC-V: KVM: Move HGEI[E|P] CSR access to IMSIC virtualization")
Signed-off-by: Andrew Jones <andrew.jones@oss.qualcomm.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Link: https://lore.kernel.org/r/20260925155116.465203-1-andrew.jones@oss.qualcomm.com
Signed-off-by: Anup Patel <anup@brainfault.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core
Pull driver core updates from Danilo Krummrich:
"Core:
- Add TAINT_FORCED_BIND to taint the kernel and driver module before
manual sysfs bind/unbind operations; make add_taint_module()
available to driver core and update taint decoding tools and
documentation
- Constify device_show/store_*() attribute parameters and adapt x86
MCE and perf users to prepare for const dev_ext_attribute instances
- Fix UAF in NUMA access node registration failure handling by
dropping the device reference with put_device() instead of directly
freeing the node
debugfs:
- Make debugfs_create_str() read-only to eliminate double-free races
between concurrent writers; reject writable modes; convert
interconnect and SoundWire users to local file operations with
mutex-protected strings
kobject:
- Fix out-of-bounds access in kobject_action_type() when parsing
action strings containing embedded NUL bytes
- Fix the kobject_set_name() reference in kset_register() kernel-doc
Rust:
- DMA:
- Tie Coherent, CoherentBox and CoherentHandle to the device's
bound lifetime to prevent retaining DMA allocations beyond
driver unbind; propagate the lifetime through nova-core and the
DMA sample driver
- I/O:
- Require explicit base types for registers and replace relative
registers with typed subregion projections; reimplement
register!() as a proc macro, support registers using existing
types without inline bitfields, and convert nova-core, Tyr and
the PCI sample; add AsRepr/AsReprMut and transmute helpers with
deferred size checks for representation conversions
- Wrap ResourceSize in a transparent newtype with explicit
conversions, including checked conversion to usize
- PCI:
- Add typed PCIe extended capability lookup and SR-IOV accessors
for the First VF Offset and decoded VF BAR addresses and widths,
keeping raw SR-IOV registers private to preserve PCI core
ownership of their state
- scatterlist:
- Honor the device's maximum segment size as well as the DMA
mapping limit when constructing SG tables
- Return u32 from SGEntry::dma_len() to match the C segment length
type
- misc:
- Validate auxiliary DeviceId name length explicitly to improve
compile-time diagnostics for oversized device ID table entries
- Drop unnecessary 'static bounds from debugfs ScopedDir file
creation methods to support data containing non-'static
references, including DMA buffers
Misc:
- Add Johan Hovold as driver core maintainer
- Simplify platform_device_alloc() by returning early on allocation
failure
- Fix typos in driver core and kernfs comments"
* tag 'driver-core-7.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core: (41 commits)
rust: pci: add extended capability and SR-IOV support
kobject: Fix the kobj_set_name reference in kset_register() kernel-doc
rust: io: convert ResourceSize into a transparent newtype
rust: scatterlist: return u32 from SGEntry::dma_len()
MAINTAINERS: add myself as driver core maintainer
debugfs: make debugfs_create_str() read-only
soundwire: debugfs: replace writable string helper
interconnect: debugfs: replace writable string helper
docs: admin-guide: Handle TAINT_FORCED_BIND when parsing /proc/sys/kernel/tainted
driver core: platform: Simplify platform_device_alloc()
kobject: fix out-of-bounds access in kobject_action_type()
rust: scatterlist: honor the device's maximum segment size
driver core: add TAINT_FORCED_BIND for when userspace manually messes with devices and drivers
module: fix up documentation for add_taint() and add_taint_module()
module: pull out add_taint_module() to be public
driver core: Constify the signature of device_show/store_*()
drivers: perf: Use __DEVICE_ATTR() macro to initialize dev_ext_attribute
x86/mce: Use __DEVICE_ATTR() macro to initialize dev_ext_attribute
rust: auxiliary: validate DeviceId name length
driver core: fix typos in comments
...
|
|
In SMB2_tcon(), unc_path is allocated with a fixed buffer size of
MAX_SHARENAME_LENGTH * 2 bytes (672 bytes). However, the user-controlled
UNC path (tree) length is passed to cifs_strtoUTF16(), which assumes the
destination buffer is large enough and writes strlen(tree) + 1 UTF-16 code
units.
Commit 62a1a439e0fd ("cifs: clean up handling of unc= option") removed the
length check on UNC mount options, allowing mount paths to exceed
MAX_SHARENAME_LENGTH while retaining the fixed-size buffer allocation in
SMB2_tcon(). Consequently, an oversized UNC string causes cifs_strtoUTF16()
to write past the end of the heap object, resulting in a slab-out-of-bounds
write. Furthermore, if a tree exceeds MAX_TREE_SIZE characters, strscpy()
truncates it in cifs_tcon::tree_name, causing subsequent reconnects or
failovers to fail.
Fix this by first verifying that the length of tree does not exceed
MAX_TREE_SIZE, returning -EINVAL if it does. Then, dynamically allocate
unc_path based on the length of tree so that it safely accommodates all
UTF-16 code units for paths within MAX_TREE_SIZE.
Fixes: 62a1a439e0fd ("cifs: clean up handling of unc= option")
Cc: stable@vger.kernel.org
Reported-by: Xin Liu <bird@lzu.edu.cn>
Closes: https://bugtracker.nebusec.ai/f/3292
Assisted-by: LLM
Suggested-by: Paulo Alcantara <pc@manguebit.org>
Signed-off-by: Luxiao Xu <rakukuip@gmail.com>
Signed-off-by: Ren Wei <n05ec@lzu.edu.cn>
[pc: remove unused MAX_SHARENAME_LENGTH]
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
Commit 3cdeaef1754a ("r8169: disable EEE on RTL8168h/8111h") disables
EEE on RTL_GIGA_MAC_VER_46 with phy_disable_eee() but keeps the version
EEE-capable, so that link up still calls rtl_enable_tx_lpi(tp, false)
and clears the MAC TX-LPI bits (ERI 0x1b0[1:0]). rtl_hw_start_8168h_1()
does not clear them, so a warm reboot from an EEE-active state could
otherwise leave TX-LPI asserted while the PHY no longer negotiates EEE.
In net-next, commit 8a3c76523e44 ("r8169: add support for phylink")
moved link handling to phylink, and rtl_mac_link_up() does not touch
the TX-LPI bits. The merge of net into net-next kept the
phy_disable_eee() quirk and its comment, but not the link-up clear it
depends on. The only remaining clear is rtl_mac_disable_tx_lpi(), which
phylink calls from phylink_deactivate_lpi() only after
phylink_activate_lpi() has enabled LPI. With EEE disabled on this chip
that never happens, so the bits are never cleared and the comment in
r8169_mdio_register() no longer holds.
Clear TX-LPI explicitly in rtl_mac_link_up() for RTL_GIGA_MAC_VER_46.
Fixes: f49defea7668 ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net")
Link: https://lore.kernel.org/all/0e0b727e94d046b19ad463e7a1a9fe68@realsil.com.cn/
Signed-off-by: Sasha Levin <sashal@kernel.org>
Link: https://patch.msgid.link/20261020161543.3712208-1-sashal@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Twenty years ago I made creating a tuntap device require CAP_NET_ADMIN
in commit ca6bb5d7ab22 ("[NET]: Require CAP_NET_ADMIN to create tuntap
devices.")
The idea was to allow the permissions on /dev/net/tun to be relaxed to
0666 and thus allow owners of tun devices assigned with TUNSETOWNER to
actually open it and get at their devices.
What we missed at the time was that persistent devices with *no* owner
were already essentially world-consumable, and anyone who could open
the device node could also attach to those.
Even then, using persistent devices with no explicitly set owner — on
the (false) presumption that CAP_NET_ADMIN would be required to open
them — wasn't an egregious userspace bug. There are cases where it does
make sense to separate the creation/configuration of the netdevice, from
the process which shovels packets into it.
Since then, TUNSETGROUP was added a year later in commit 8c644623fe7e
("[NET]: Allow group ownership of TUN/TAP devices."). Its commit message
noted that if both uid and gid are set, both have to match (in itself a
bizarre choice) — but still not that with *neither* set, no permission
at all was required.
Then in 2012 commit c8d68e6be1c3 ("tuntap: multiqueue support") allowed
new queues to be attached to a device from the moment of its creation,
with no way for userspace to close the race between creation and a
subsequent TUNSET{OWNER,GROUP}.
In 2024, commit 3ca459eaba1bf ("tun: fix group permission check") tried
to fix things to be saner: only require the user *or* the group to
match, in line with standard POSIX permissions, and require
CAP_NET_ADMIN if neither the user nor group authorised the access
(although it put the CAP_NET_ADMIN check first, and would wrongly have
reported the capability as having been used even when it was not).
That fix was reverted in commit a70c7b3cbc06 ("tun: revert fix group
permission check") due to allegedly causing a regression in an selinux
test suite which *expected* the attachment to succeed when authorised
by none of user, group, nor CAP_NET_ADMIN:
https://lore.kernel.org/netdev/CAFqZXNtkCBT4f+PwyVRmQGoT3p1eVa01fCG_aNtpt6dakXncUg@mail.gmail.com/
We should not break UAPI. But we don't have to consider an selinux test
case which deliberately exercises a security flaw to be UAPI worth
preserving either. To which end, I have fixed it in selinux-testsuite
commit b2f894f0016b ("tests/tun_tap: set an owner on the device for the
relabel tests").
So: fix the lack of CAP_NET_ADMIN check on devices which *haven't* been
assigned an owner or group, while leaving the weird tun ownership model
as it was (user *and* group have to match, if both are set). If someone
still wants to fix the latter (which wasn't actually the reason the
previous attempt was reverted), they can do it under separate cover.
Permit attaching to an existing device only if an owner and/or group was
set and the caller matches all that were set, or if the caller has
CAP_NET_ADMIN. The only behavioural change is that a device with no
owner or group now requires CAP_NET_ADMIN to attach — precisely the
loophole we want to close.
And put the CAP_NET_ADMIN check in the right place this time.
Fixes: ca6bb5d7ab22 ("[NET]: Require CAP_NET_ADMIN to create tuntap devices.")
Reported-by: Andrew Lutomirski <luto@kernel.org>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Cc: stable@vger.kernel.org
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/3c77f88acff5b3c2617b22d64faad15f6212619c.camel@infradead.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Jamal Hadi Salim says:
====================
net/sched: sch_cake: fixes for accounting and reconfiguration
Four independent fixes in the CAKE qdisc. Three are follow-ups to the
autorate reconfiguration work around commit 7cbfb180945c ("net/sched:
sch_cake: fix autorate reconfiguration throttling") and the 32-bit
admission-counter wraps fixed in gred/bfifo/plug by commit 4c660ee8c809
("net: sched: fix 32-bit backlog wrap in gred, bfifo and plug enqueue");
the fourth is a dead-code cleanup found while auditing the dump path.
Patch 1 serializes a 'tc qdisc change' with the datapath's autorate
update under the qdisc lock, so the netlink writer and the autorate
writer can no longer lose the configured rate. rate_bps needs no atomic
(it is only loaded and stored), so it stays a plain u64 with
READ_ONCE()/WRITE_ONCE() annotations.
Patch 2 propagates the backlog reduction of the reconfigure-time tin
purge to classful parents, so a CAKE leaf under HTB no longer leaves the
purged packets in the parent's qlen/backlog forever.
Patch 3 widens buffer_used and buffer_max_used to u64. The u32
accumulator wraps once the resident truesize passes 2^32, bypassing the
over-limit drop loop while the configured limit is a u32 attribute; the
u32 MEMORY_USED export is clamped so widening the counter does not
truncate the reported value.
Patch 4 drops an unreachable 'nla_nest_end() < 0' error check: the
helper never returns a negative value.
All four are hardening for net-next (no Fixes:, no Cc: stable): the
counter wrap, the missing parent-backlog propagation and the
reconfiguration race all predate any contract we could anchor a
regression on. Patches 1-2 were reported by Sashiko (gemini) and patch
3 comes from the gred 32-bit-wrap family; patch 4 is a static cleanup.
====================
Link: https://patch.msgid.link/QDISC-5RTM.v3.r0.20261020053711@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
cake_dump_class_stats() closes the TCA_STATS_APP nest with
if (nla_nest_end(d->skb, stats) < 0)
return -1;
nla_nest_end() never returns a negative value: it has a single
`return skb->len;` and no error path (that is why the overflow-aware
nla_nest_end_safe() helper was later added). The '< 0' test can never
be true, so the 'return -1' is unreachable dead code.
Drop the check and close the nest with the bare statement, matching the
sibling cake nests in the same file.
This issue was discovered while fixing a bug on the gate action, see:
https://lore.kernel.org/netdev/QDISC-H19Z.v1.20261001053234@mojatatu.com/
Conditions to recreate the bug:
- not runtime observable: the issue is found by inspection.
Link: https://lore.kernel.org/netdev/87v77csd5q.fsf@toke.dk/
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://patch.msgid.link/QDISC-5RTM.v3.r0.20261020053711-4@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
This is a follow-up to commit 4c660ee8c809 ("net: sched: fix 32-bit
backlog wrap in gred, bfifo and plug enqueue"), which promoted the
backlog + length admission sums in gred, bfifo and plug to u64. CAKE's
own memory accounting has the same 32-bit wrap.
The issue:
cake_sched_data.buffer_used accumulates skb->truesize and is compared
against memory_limit:
q->buffer_used += skb->truesize;
if (q->buffer_used <= q->buffer_limit)
return NET_XMIT_SUCCESS;
while (q->buffer_used > q->buffer_limit)
drop...
buffer_used is u32, so once the resident truesize passes 2^32 it wraps to
a small value, the <= test passes and the drop loop is skipped even
though far more than the configured limit is queued. memory_limit is a
u32 attribute (TCA_CAKE_MEMORY), so buffer_limit can be as large as
U32_MAX and never caps buffer_used below the wrap point; with a large
memory_limit the queue then grows until the machine is out of memory.
The fix:
Widen buffer_used and buffer_max_used to u64, the same shape as the
gred/bfifo/plug fix. buffer_limit stays u32; the comparison promotes it
to u64, so the bounded queue stops at the configured limit below 2^32 and
the counter can no longer wrap. buffer_used is subtracted on dequeue and
drop, so the widened type flows through unchanged. The ACK-filter
replacement applied the net delta as one expression,
q->buffer_used += skb->truesize - ack->truesize. Both operands are
unsigned int, so the subtraction is evaluated at u32 and a larger removed
ACK wraps the delta; the old u32 accumulator folded that back to the
correct net value, but the widened u64 accumulator would persist it as a
roughly 4 GiB over-count. Apply the replacement as two exact operations
on the widened counter instead; the queue already accounts ack->truesize,
so the subtraction cannot underflow.
Widening the counters alone is incomplete because cake_dump_stats() still
exports buffer_max_used through PUT_STAT_U32(MEMORY_USED, ...), and
tc.yaml declares memory-used as u32. buffer_max_used only ever rises, so
once the widened counter passes 2^32 the export reports peak - 2^32 for
the rest of the qdisc's life. Clamp the exported value to U32_MAX; the
counter itself is unaffected.
The dump path runs without the qdisc lock: tc_fill_qdisc() passes NULL to
gnet_stats_start_copy_compat(), so the reader races with the enqueue-side
update. READ_ONCE() is not an atomic 64-bit load on 32-bit CPUs, so a
reader can observe a torn high-water mark and report a bogus MEMORY_USED.
Publish the counter as u64_stats_t under the qdisc lock, which the enqueue
path already holds, and take a seqcount-protected snapshot in the dump.
The counter stays u64; only the exported value is bounded.
Conditions to recreate the bug:
CAP_NET_ADMIN (root or a user namespace, the qdisc attach is gated by
netlink_net_capable(CAP_NET_ADMIN)).
A TUN device is needed so the root qdisc can park packets instead of
dropping them. Open /dev/net/tun, request TUNSETIFF with
IFF_TUN | IFF_NO_PI | IFF_BACKPRESSURE and keep the fd open without
reading it (run it in the background and stop it at teardown): with
IFF_BACKPRESSURE the full tx ring calls netif_tx_stop_queue() rather than
dropping, so the queue behind the qdisc can grow past 4 GiB of resident
truesize. A plain 'ip tuntap add' attaches no fd and does not request
IFF_BACKPRESSURE, so it cannot park the queue this way.
ip link set tun0 txqueuelen 32 up
ip addr add 10.99.0.1/24 dev tun0
tc qdisc add dev tun0 root handle 1: cake memlimit 4294967295
# Then push more than 4 GiB of resident truesize through the qdisc: six
# UDP senders using large SO_SNDBUFFORCE buffers send 1400-byte
# datagrams (the reported OOM witness used 6 workers for 100 s on a
# 12 GiB guest). On the unfixed kernel buffer_used wraps, the admission
# test passes, the drop loop is bypassed and the guest runs out of
# memory; with the fix the counter passes the limit without wrapping and
# the excess is dropped.
# The ACK-filter over-count is a separate corner: it needs ack-filter
# enabled and a queued pure ACK whose truesize is larger than the
# replacement ACK. Use a fresh interface (or 'tc qdisc change') rather
# than a second root 'tc qdisc add' on an interface that already has a
# qdisc:
tc qdisc change dev tun0 root cake ack-filter
CONFIG_NET_SCH_CAKE=y and CONFIG_TUN=y are required, and the
memory-residency recipe needs a root-run sender envelope with enough
guest RAM to hold the queued data.
Reported-by: Sashiko (gemini) <sashiko-bot@kernel.org>
Closes: https://sashiko.dev/#/patchset/20260818095927.15901-1-jhs@mojatatu.com
Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260818095927.15901-1-jhs@mojatatu.com
Link: https://lore.kernel.org/netdev/20260818095927.15901-1-jhs@mojatatu.com/
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Link: https://patch.msgid.link/QDISC-5RTM.v3.r0.20261020053711-3@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The issue:
CAKE has purged unused tins on DiffServ reconfiguration since commit
83f8fd69af4f ("sch_cake: Add DiffServ handling") without propagating the
resulting backlog reduction to classful parents. The issue was noticed
while reviewing 7cbfb180945c, but predates that change.
cake_clear_tin() drops every packet from a tin with
kfree_skb_reason() but never calls qdisc_tree_reduce_backlog(), so a
CAKE qdisc under a classful parent (e.g. HTB) leaves the dropped packets
in the parent qdisc's and class's qlen/backlog accounting forever. The
normal dequeue path does reduce the parent; only the live reconfigure
purge misses it.
The fix:
Tins are cleared from cake_reset() and, on reconfigure, from
cake_configure_rates(). The qdisc core already owns the parent delta for
the reset path: qdisc_purge_queue() snapshots qlen/backlog, runs
qdisc_reset(), then reduces the tree, so propagating the reduction from
cake_clear_tin() itself would subtract the same backlog twice. Snapshot
CAKE's qlen/backlog around only the reconfiguration purge in
cake_configure_rates() and reduce the parent once afterwards, matching
the delta pattern already used by cake_enqueue().
Conditions to recreate the bug:
put CAKE (diffserv8) under an HTB class, flood DSCP-classified traffic
into a high tin so it backlogs, then change CAKE to diffserv3. The HTB
parent keeps the purged packets in its qlen and backlog; a kernel with
the fix shows the parent at the leaf's backlog (426b/5p) instead of
retaining the purged packets.
Reported-by: Sashiko (gemini) <sashiko-bot@kernel.org>
Link: https://sashiko.dev/#/patchset/20260816012109.2865223-1-ooonea@gmail.com
Link: https://lore.kernel.org/netdev/20260816012109.2865223-1-ooonea@gmail.com/
Link: https://lore.kernel.org/netdev/8733ugtry8.fsf@toke.dk/
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://patch.msgid.link/QDISC-5RTM.v3.r0.20261020053711-2@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
This is a follow-up to commit 7cbfb180945c ("net/sched: sch_cake: fix
autorate reconfiguration throttling"). That change only stored
last_reconfig_time; it did not address the concurrency between the
autorate rate update in cake_enqueue() and the netlink reconfiguration
path, which the original review flagged.
The issue:
cake_change() -> cake_config_change() updates the fields of the shared
struct cake_sched_config under RTNL but not under the qdisc lock: it only
took sch_tree_lock() later, around cake_reconfigure(). The datapath in
turn serializes cake_enqueue() with itself under the qdisc lock, and its
autorate-ingress path stores the estimated rate into
q->config->rate_bps. Two writers therefore update the same rate with no
common lock:
CPU0: cake_enqueue() sees CAKE_FLAG_AUTORATE_INGRESS, computes an
estimate, and is about to store it.
CPU1: 'tc qdisc change ... bandwidth X' clears autorate in its local
rate_flags, writes rate_bps = X, publishes the cleared flag, and
waits in sch_tree_lock().
CPU0: stores the estimate, calls cake_reconfigure(), and unlocks.
CPU1: reconfigures from the estimate and returns.
The qdisc is left with autorate disabled but with the estimate installed
instead of X.
The fix:
Commit the parsed configuration and reconfigure while holding the same
qdisc lock that serializes the datapath, so the netlink writer and the
autorate writer can no longer interleave. rate_bps needs no atomic: it is
only ever loaded and stored, so plain READ_ONCE()/WRITE_ONCE() annotations
are enough for the accesses that run outside the lock. The lockless
paths read the rate once into a local and reuse that value, so a single
coherent snapshot is used per operation.
The annotations do not make a 64-bit access single-copy atomic on 32-bit:
READ_ONCE()/WRITE_ONCE() are still two 32-bit accesses there. The lockless
readers -- cake_config_dump() and the cake_mq children, which observe the
shared config under a different child's lock -- can therefore still see a
torn rate on 32-bit. That rate is an advisory shaper value and is accepted
as best-effort; the state that must be consistent (the configured rate and
the autorate flag) is written under the qdisc lock.
Conditions to recreate the bug:
The lost update (a) needs no sanitizer. Put cake in autorate-ingress on a
device, drive bursty traffic with gaps so the 250 ms autorate reconfiguration
fires, and concurrently loop 'tc qdisc change ... cake bandwidth <N>kbit'
without autorate-ingress (specifying a bandwidth clears the flag).
If the autorate store lands after the change wrote its rate but before the
change reconfigures, the qdisc ends up with autorate disabled and the estimate
installed instead of <N>; reading the rate back then shows the estimate, which
an unfixed kernel can do and a fixed one cannot.
(b) The unsynchronized access is reported by KCSAN: build with CONFIG_KCSAN=y
and run the same trigger with the change loop keeping autorate-ingress set;
KCSAN reports a data race in cake_config_change / cake_enqueue.
The non-atomic 64-bit store only shows up as tearing on a 32-bit machine;
on 64-bit the store is atomic and KCSAN flags the access discipline instead.
Reported-by: Sashiko (gemini) <sashiko-bot@kernel.org>
Link: https://sashiko.dev/#/patchset/20260816012109.2865223-1-ooonea@gmail.com
Link: https://lore.kernel.org/netdev/20260816012109.2865223-1-ooonea@gmail.com/
Link: https://lore.kernel.org/netdev/875wzctsky.fsf@toke.dk/
Link: https://lore.kernel.org/netdev/87ld7t3kl3.fsf@all.your.base.are.belong.to.us/
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Link: https://patch.msgid.link/QDISC-5RTM.v3.r0.20261020053711-1@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The Datto SIRIS 5 1U Server is a Dell OEMR XL R640 (Dell's OEM-ready
variant of the PowerEdge R640) sold under Datto's identity. Its DMI
strings report sys_vendor "datto" and product_name "SIRIS 5 1U Server",
so it does not match the existing "PowerEdge R640" entry in
tg3_restart_aer_quirk_table.
As on the PowerEdge R640, a system restart triggers a fatal PCIe AER
on the BCM5720 during ACPI _PTS for S5, which the iDRAC logs as
PCI1318 and which hangs the reboot. Power off is unaffected.
Clearing the same PCI_EXP_DEVCTL error-reporting bits on both BCM5720
functions from userspace immediately before restart allows the system
to reboot normally, confirming the quirk applies to this platform.
Fixes: e0efe83ed325 ("tg3: Disable tg3 PCIe AER on system reboot")
Signed-off-by: Travis Howard <thoward@lafayetteinstrument.com>
Link: https://patch.msgid.link/20261019-tg3-datto-aer-quirk-v1-1-f0eccfb2ab5d@lafayetteinstrument.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
mlx5e_close_channels() has an ASSERT_RTNL() from the time when RTNL was
required for closing channels.
But the driver has been switched to using the netdev lock a while back,
and this fired from a flow which was forgotten during the conversion:
mlx5e_tx_timeout_work()
-> mlx5e_reporter_tx_timeout()
-> mlx5e_health_report()
-> mlx5e_tx_reporter_timeout_recover()
-> mlx5e_safe_reopen_channels()
-> mlx5e_safe_switch_params()
-> mlx5e_close_channels()
The RTNL is not acquired here, but the netdev lock is, and it is enough.
Rather than switching to netdev_assert_locked(), simply drop the assert.
Upper layers (networking core, ethtool core) provide locking and lower
layers (NAPI helpers) assert that the netdev lock is held.
Fixes: 8f7b00307bf1 ("net/mlx5e: Convert mlx5 netdevs to instance locking")
Signed-off-by: Cosmin Ratiu <cratiu@nvidia.com>
Reviewed-by: Carolina Jubran <cjubran@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20261020153659.3440902-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull thermal control updates from Rafael Wysocki:
"These are mostly updates of thermal drivers including a new Exynos
ACPM thermal driver for Google GS101, new Renesas RZ/G3L thermal
sensor support, new Qualcomm MSM8952 TSENS support, new Qualcomm
PMAU0102 Master BandGap temperature monitor support in DT bindings,
and new rugged Panther Lake support in intel_tcc_cooling. Moreover,
the AN7583 driver is extended to register separate thermal zones for
bandgap, GbE and CPU (its DT bindings are updated accordingly).
There are also multiple fixes and assorted cleanups, as usual.
Specifics:
- Make the Intel powerclamp thermal driver reject invalid window_size
values, drop redundant clamp() from duration_set() in it, and
simplify 4 functions in it (Thorsten Blum)
- Add rugged Panther Lake support to the intel_tcc_cooling thermal
driver (Srinivas Pandruvada)
- Fix temperature selection around 0 C in the int340x thermal driver,
drop redundant else from proc_thermal_add() in it, and simplify
slider_def_*_set() in it (Thorsten Blum)
- Drop unbalanced paren from cpu-idle-cooling.rst (Manuel Ebner)
- Use %pe to print error pointers symbolically in the thermal control
code and fix assorted typos in comments in it (Sumeet Pawnikar and
Hemanth Selam)
- Add the Exynos ACPM thermal driver for Google GS101, along with DT
bindings and a MAINTAINERS entry (Tudor Ambarus)
- Add Renesas RZ/G3L thermal sensor support and DT bindings, and move
default calibration values to per-SoC match data (Biju Das)
- Add Qualcomm MSM8952 TSENS support and the corresponding DT binding
compatible (Muzaffer Kadir)
- Document the Qualcomm PMAU0102 Master BandGap temperature monitor
compatible in the DT bindings (Dhruvin Rajpura)
- Extend the QCom TSENS v2 trip temperature limits to the hardware
range of -204°C to +204°C to prevent interrupt storms when
operating beyond the previous software limits (Priyansh Jain)
- Fix temperature threshold map indexing in the QCom SPMI temperature
alarm driver to prevent out-of-bounds reads (Georgi Djakov)
- Describe the additional bus clock required by the Amlogic T7
thermal sensor and switch the driver to the clk_bulk API to manage
multiple clocks (Ronald Claveau)
- Request the Airoha EN7581 IRQ after thermal zone registration to
prevent the interrupt handler from accessing an unregistered or
already freed thermal zone (Vitaliy Sochnev)
- Fix the Airoha AN7583 ADC mux field mapping, serialize access to
the shared ADC, and verify mux selection to avoid invalid
temperature readings and spurious critical shutdowns (Vitaliy
Sochnev)
- Register separate thermal zones for the AN7583 bandgap, GbE and CPU
sensors, and update the DT bindings to use one thermal sensor cell
to select the sensor (Vitaliy Sochnev)
- Fix signed integer overflow in the Renesas RZ/G3E
temperature-to-code conversion, preserve compare interrupt state
across runtime PM, and prevent overlapping temperature conversions
(Ovidiu Panait)
- Remove redundant trip window validation and runtime PM
pm_runtime_mark_last_busy() calls from the Renesas RZ/G3E driver
(Ovidiu Panait)
- Propagate reset assertion failures from Renesas RZ/G3E suspend and
attempt to restore power if the sensor was active (Pengpeng Hou)
- Stop the TI bandgap alert IRQ before removing sensor interfaces and
cancel pending alert work to prevent use-after-free on driver
removal (Weigang He)
- Validate Spreadtrum sensor IDs read from the device tree before
using them for register offsets and array indexing (Weigang He)
- Convert the TI OMAP bandgap DT bindings to schema and document the
DRA752 compatible and thermal sensor cells (Bhargav Joshi)
- Use DT property helpers matching the signed temperature and
unsigned hysteresis types defined by the thermal zone bindings (Rob
Herring)
- Reuse common error handling code in the power allocator governor
bind callback (Markus Elfring)
- Constify the thermal zone device operations in the i.MX9 driver
(Christophe JAILLET)
- Fix kernel-doc parameter names and firmware node references in the
cooling device registration helpers (Aryan Sinha)
- Correct the documented return value of ti_bandgap_get_sensor_data()
to describe validation errors and potentially NULL stored data
(Karl Mehltretter)
- Ensure NUL-termination when copying the cooling device command-line
argument in tmon and fix the associated string truncation warning
(Florian Fainelli)
- Fix undefined behavior caused by a signed shift in the tmon thermal
zone binding display (Liu Jing)
- Fix wording in tmon comments (Hemanth Selam)
- Simplify cpufreq_set_cur_state() in the cpufreq_cooling thermal
control driver (Thorsten Blum)"
* tag 'thermal-7.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (51 commits)
thermal/drivers/rzg3e: Propagate reset assertion failure from suspend
thermal: ti-soc-thermal: Fix the return value in ti_bandgap_get_sensor_data() kernel-doc
thermal/fw: Fix kernel-doc parameter names
thermal/qcom/tsens: Add support for MSM8952 tsens
dt-bindings: thermal: tsens: Add MSM8952
thermal/drivers/amlogic: Switch to clk_bulk API to support multiple clocks
dt-bindings: thermal: amlogic: describe bus clock for t7-thermal
tools/thermal/tmon: Fix string truncation warning in optarg copy
thermal/drivers/airoha: Register a thermal zone per AN7583 sensor
dt-bindings: arm: airoha: Use one thermal sensor cell for AN7583
thermal/drivers/airoha: Serialise access to the shared ADC
thermal/drivers/airoha: Fix AN7583 ADC mux field mapping
thermal/drivers/sprd: validate the sensor id from the device tree
thermal/drivers/ti-soc-thermal: Cancel pending alert work on remove
thermal/drivers/qcom-spmi-temp-alarm: Fix temp_map indexing
thermal: gov_power_allocator: Use common error handling code in power_allocator_bind()
tools/thermal/tmon: Fix signed shift UB in binding display
MAINTAINERS: Add entry for Samsung Exynos ACPM thermal driver
thermal/drivers/samsung: Add Exynos ACPM TMU driver GS101
dt-bindings: thermal: Add Google GS101 TMU
...
|
|
skb_shinfo(skb)->dataref is split in two 16-bit halves: the lower
one counts all references, the upper one the payload-only (nohdr)
references.
__skb_clone() increments dataref without bound, and skb_cloned()
only checks the low 16 bits. With 65,537 references the low half
reads 1, the head looks private, and pskb_expand_head() (or
skb_unclone(), skb_cow(), ...) frees it while clones still point
to it.
An unprivileged user can reach this from a user and network namespace:
about 32,770 AF_PACKET sockets on lo bring dataref to 0x10001, then
any pskb_expand_head() on the receive path (here a pull of a large
IPv6 Hop-by-Hop header) frees the shared head:
BUG: KASAN: slab-use-after-free in _copy_to_iter+0x187/0x1330
Read of size 1590 at addr ff110001104a9002 by task dataref_wrap/87
skb_copy_datagram_iter+0x60/0x190
packet_recvmsg+0x1f5/0x12b0
__sys_recvfrom+0x18d/0x290
Freed by task 87:
kfree+0xf3/0x3b0
pskb_expand_head+0x344/0xeb0
__pskb_pull_tail+0xf9/0x1bf0
ipv6_parse_hopopts+0x377/0x4f0
ip6_rcv_core+0xc67/0x1620
The upper half never holds more than 1. It is only set by
__skb_header_release(), with atomic_set() on an skb owning its head
(TCP/SCTP output, and GRO which never sees cloned skbs).
__skb_clone() clears nohdr, and __copy_skb_header() does not copy it.
Move SKB_DATAREF_SHIFT to 30, so that the reference count gets 30 bits,
and dataref stays positive. Wrapping it would need 2^30 live clones of
a single skb, which socket buffer limits and memory make unreachable.
This issue was found by Anthropic using Claude, and validated by
Ada Logics.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reported-by: Stephan Lipp <Stephan@adalogics.com>
Signed-off-by: Eric Dumazet <edumazet@kernel.org>
Link: https://patch.msgid.link/20261020151954.22953-1-edumazet@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
ipc_protocol_free_msg_get() reads msg_head from shared memory and
uses it directly as an index into msg_ring[]. If the head is out
of range, the subsequent memset() writes beyond the message ring.
Validate head against IPC_MEM_MSG_ENTRIES before using it as an
index. The ring is a circular buffer, so a value outside the valid
range indicates a corrupted head pointer; report the error and fail
the allocation instead of writing out of bounds.
Fixes: 64516f633bfd ("net: iosm: protocol operations")
Cc: stable@vger.kernel.org
Reviewed-by: Fang Xieyan <fangxy@xiaopeng.com>
Signed-off-by: Liu Chao <liuc63@xiaopeng.com>
Reviewed-by: Loic Poulain <loic.poulain@oss.qualcomm.com>
Link: https://patch.msgid.link/20261020094254.4017726-1-liuc63@xiaopeng.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
When the RX ring is full the RX FIFO overflows. With RX overflow
continue enabled, the DMA engine then writes one frame with a bogus
length (always 0x45c2 on a BCM4716), which the driver drops with
"Found oversized packet at slot %d, DMA issue!".
Under sustained overload this happens on every NAPI poll. On an Asus
RT-N16 a 950 Mbit/s UDP stream to the router printed about 1450 such
lines in 20 seconds. The messages are printed from NAPI context, so
with a serial console the router spends most of its time printing.
The frames are already accounted in rx_length_errors and rx_errors.
Rate limit both RX error messages. With this change the same flood
printed 60 lines while rx_length_errors counted 14297 dropped frames.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Reviewed-by: Daniel Machon <daniel.machon@microchip.com>
Link: https://patch.msgid.link/20261018231629.39734-1-rosenp@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
'veth-schedule-the-peer-s-napi-after-xdp_tx-when-only-gro-is-enabled'
Tianyi Gao says:
====================
veth: schedule the peer's NAPI after XDP_TX when only GRO is enabled
When an XDP program on a veth device returns XDP_TX and the peer has
GRO enabled but no XDP program, the frames are queued on the peer's
ring, but the peer's NAPI is never scheduled to process them. Once the
ring is full, every XDP_TX frame is dropped.
Commit 0e672f306a28 ("veth: check for NAPI instead of xdp_prog before
xmit of XDP frame") made such a peer work as an XDP_REDIRECT target,
where veth_xdp_xmit() both queues the frames and schedules the peer's
NAPI. XDP_TX queues its frames through the same function but schedules
the NAPI in veth_xdp_flush(), which still checks xdp_prog.
The usual workaround is to load a dummy XDP program on the peer [1].
That turns off TSO on the device, and every skb it sends then goes
through the peer's XDP path, where most of them are copied. In our
tests, TCP sent from a namespace dropped from 22.1 to 14.8 Gbit/s with
the dummy program. With this fix the program is no longer needed.
Patch 1 makes veth_xdp_flush() check the peer's napi pointer, as
veth_xdp_xmit() does. Patch 2 adds a test for this case to veth.sh.
====================
Link: https://patch.msgid.link/20261016214217.16942-1-tianyi@cloudflare.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Test that XDP_TX frames reach a peer that has GRO enabled and no XDP
program. GRO enables NAPI on the peer, so the frames are queued on the
peer's ring, and the peer's NAPI must be scheduled to process them.
Attach a program that returns XDP_TX for every frame, send 5 echo
requests and check that all of them were sent back and that the peer's
NAPI received them, as counted in its xdp_packets statistic. IPv6 is
disabled on both devices, so that only the echo requests are counted.
Add the program to xdp_dummy.bpf.c in a section of its own, so that the
tests that load "section xdp" still get XDP_PASS.
Signed-off-by: Tianyi Gao <tianyi@cloudflare.com>
Link: https://patch.msgid.link/20261016214217.16942-3-tianyi@cloudflare.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
veth_xdp_xmit() queues XDP_TX frames on the peer's ring if the peer has
NAPI. A peer with GRO enabled and no XDP program also has NAPI.
veth_xdp_flush() then schedules the peer's NAPI, but only if the peer's
xdp_prog is set, so for such a peer the frames are never processed.
Nothing else schedules that NAPI by default.
Once the ring is full, every XDP_TX frame is dropped, and so are the
skbs that veth_xmit() queues on the same ring.
Commit 0e672f306a28 ("veth: check for NAPI instead of xdp_prog before
xmit of XDP frame") changed the check in veth_xdp_xmit() but not the one
in veth_xdp_flush(). Make veth_xdp_flush() check the napi pointer
instead of xdp_prog, as veth_xdp_xmit() does.
Fixes: 0e672f306a28 ("veth: check for NAPI instead of xdp_prog before xmit of XDP frame")
Signed-off-by: Tianyi Gao <tianyi@cloudflare.com>
Reviewed-by: Mohsin Bashir <hmohsin@meta.com>
Acked-by: Lorenzo Bianconi <lorenzo@kernel.org>
Acked-by: Jesper Dangaard Brouer <hawk@kernel.org>
Link: https://patch.msgid.link/20261016214217.16942-2-tianyi@cloudflare.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management updates from Rafael Wysocki:
"This time, the most visible part are the ACPI CPPC updates from
Christian Loehle, including CPPC cpufreq driver changes, even though
updates of cpufreq drivers proper are still the largest topic group.
However, the updates of cpufreq drivers depend on the ACPI CPPC
updates, so they should really be counted together and that is well
over 1/2 of the total commit count.
Interestingly enough, almost all of the updates included here are
fixes or cleanups, with some new feature support in the amd-pstate
cpufreq driver and IPQ9650 support in the qcom-nvmem cpufreq driver.
The runtime PM documentation improvements from Brian Norris can also
be mostly regarded as fixes, but a new kunit test goes along with
them.
Specifics:
- Fix multiple issues related to input validation, synchronization,
access safety, memory layout and register field validation, and
error handling in the ACPI CPPC library (Christian Loehle and Sumit
Gupta)
- Select the frequency-invariance callback per CPU in the CPPC
cpufreq driver and create the FIE worker before enabling PCC
callbacks in it to avoid queueing work through a missing worker and
dereferencing a NULL pointer (Christian Loehle)
- Update the CPPC cpufreq driver entry in MAINTAINERS and add a new
entry for the ACPI CPPC library (Christian Loehle)
- Minor fixes in various cpufreq drivers (Alexander Koskovich, Daniel
Schwierzeck, Frank Zhang, Haotian Zhang, Kathiravan Thirumoorthy,
Sumeet Pawnikar, and Xueqin Luo)
- Fix wrong behaviors on multi-node servers in the Loongson cpufreq
driver (Huacai Chen)
- Reject NULL from cpufreq_cpu_get() in the Rust code (Mehmet
Koseoglu)
- Add suspend OPP to mediatek cpufreq driver and DT (Ryan Brue)
- Fixups for a few bugs identified by Sashiko in the amd-pstate
cpufreq driver (Mario Limoncielloi, Wentao Guan)
- Initial support for maximum frequency capacity handling on Zen6
client in amd-pstate (Mario Limonciello)
- Fix for amd-pstate-ut with 'Requested CPU Min frequency' BIOS
option (K Prateek Nayak)
- New Zen6 tuning values for EPP in amd-pstate (Mario Limonciello,
Vishal Badole)
- amd-pstate documentation fix (Karl Mehltretter)
- Restrict deferred downscaling to low load for shared policies in
the conservative cpufreq governor to avoid undesirable CPU
frequency reduction in some cases (Shengming Hu)
- Serialize sugov_start() against rate limit updates in the schedutil
cpufreq governor to avoid using a stale rate limit value (Hui Su)
- Prevent cpuidle governors menu and teo from returning idle state 0
after it has been disabled (Xueqin Luo)
- Replace min_t() with the better alternatives in the intel_idle
driver and make it use 2-argument strscpy() (Andy Shevchenko)
- Fix two repeated phrase instances in cpuidle (Hemanth Selam,
Syazwina Fazrul)
- Add rugged Panther Lake support to the intel_rapl power capping
driver and make it use %pe for printing error pointers (Srinivas
Pandruvada and Sumeet Pawnikar)
- Update and improve runtime PM documentation and add a kunit test
for supplier idle/suspend (Brian Norris)
- Add a pm_runtime_dont_use_autosuspend() call to pm_runtime_reinit()
to ensure autosuspend disabling on driver teardown (Joshua Crofts)
- Ignore page-zero RAM in E820 checksum during hibernation on x86 to
avoid rejecting images created after kexec() (Matthias Goergens)
- Add resume event mapping for PMSG_POWEROFF to the PM core to allow
it to carry out device shutdown in the future (Mario Limonciello)
- Add DPM watchdog to prepare/late/early/noirq/complete phases of
device suspend and resume (Mayank Rungta)
- Update swsusp.txt references to swsusp.rst in hibernation
documentation (Li Jun)
- Fix use-after-free of perf domain in netlink doit handlers for
energy models (Fan Wu)
- Add synchronize_rcu() back to cpu_pm_unregister_notifier() after it
was removed inadvertently in previous development cycles to avoid
use-after-free (Bradley Morgan)
- Unregister CPU PM notifier outside the timer lock in the
timer-ti-dm clocksource driver (Bradley Morgan)
- Fix typo "sucessfully" in a comment in pm_clk_add_clk() kerneldoc
comment and typo "hierachy" in a comment in pm-graph (Hemanth
Selam)
- Fix the return value in dev_pm_opp_get_of_node() kernel-doc (Karl
Mehltretter)"
* tag 'pm-7.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (89 commits)
cpufreq: loongson3: Replace IOCSR read/write with MMIO ones
cpufreq: loongson3: Use global physical CPU ID in get/target callbacks
cpufreq: loongson3: Replace per-package mutex with per-node
cpufreq: loongson3: Adjust the width of id and val in smc_message
cpufreq: loongson3: Make this driver depend on MACH_LOONGSON64
PM: runtime: Expand introduction with core concepts and structure
cpufreq: schedutil: Serialize sugov_start() against rate limit updates
MAINTAINERS: ACPI: Add a separate CPPC entry
MAINTAINERS: cpufreq: Replace CPPC reviewer
arm64: dts: mediatek: mt8173: Mark the lowest CPU OPPs for suspend
cpufreq: mediatek: Enter system sleep at the suspend OPP
cpufreq: qcom-cpufreq-hw: fix device node refcount leak in phandle parsing
cpufreq: qcom-nvmem: Add IPQ9650 support
cpufreq: Add Milos to cpufreq-dt-platdev blocklist
cpufreq: Use %pe to print error pointers symbolically
rust: rcpufreq_dt: add module alias
cpufreq: sti: avoid NULL dereference in dev_err()
cpufreq: tegra194: fix double-pointer error in get_cpu_ndiv
cpufreq: sparc-us2e: fix frequency table index copy-paste error
rust: cpufreq: reject NULL from cpufreq_cpu_get()
...
|
|
pppoe_connect() copies the requested PPPoE address to a socket before
attempting to insert the socket into the session hash table. If the
tuple already exists, __set_item() returns -EALREADY, but the copied
address and ifindex remain set on the failed socket.
Closing that socket then deletes the existing socket's hash entry
through pppoe_release(), preventing it from receiving further session
traffic. An unprivileged process in the same network namespace can
trigger this through a duplicate connection attempt.
To avoid the accidental delete, pass the pppox_sock to __delete_item(),
and unlink only an entry whose pointer matches the requested socket.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Cc: stable@vger.kernel.org
Signed-off-by: Qingfang Deng <qingfang.deng@linux.dev>
Link: https://patch.msgid.link/20261020092118.47416-1-qingfang.deng@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
validate_ntransact() assumes that the response uses the fixed 18-word NT
transaction layout. A different WordCount makes get_bcc() locate the byte
count at a different offset from pSMBr->ByteCount, allowing the subsequent
bounds checks to use inconsistent response boundaries.
The function also turns the 32-bit ParameterOffset and DataOffset fields
into pointers before checking their bounds. On a 32-bit client, a large
offset can wrap the pointer below end_of_smb and pass the one-sided pointer
comparison. The caller then dereferences the parameter pointer or copies
from the data pointer.
Require the expected WordCount before reading the byte count. Check each
offset and count against the received SMB length as integers before forming
either pointer. Use subtraction after validating the offset so that the
range checks cannot overflow.
Reported-by: Ratnaker Gautam <ratnaker.gautam@kuleuven.be>
Fixes: 0a4b92c05ed0 ("[CIFS] Add worker function for Get ACL cifs style")
Cc: stable@vger.kernel.org
Signed-off-by: Ratnaker Gautam <ratnaker.gautam@kuleuven.be>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
lan969x_fdma_tx_clear_buf() runs from the NAPI poll and frees the skbs
of completed tx DCBs with napi_consume_skb(), under sparx5->tx_lock.
Before this patch, the lock is taken with spin_lock_irqsave(), so the
skbs are freed with hardirqs disabled. RX skbs are marked for page_pool
recycling, so when such an skb is forwarded back out and its tx
completion frees it here, the free path reaches __page_pool_put_page(),
which warns via lockdep_assert_no_hardirq().
Take tx_lock with a plain spin_lock(), as the only other tx_lock user,
sparx5_port_xmit_impl(), already does. tx_clear_buf() is only called
from softirq context, so disabling hardirqs is not needed.
Found with lockdep enabled (CONFIG_PROVE_LOCKING), which the assert
depends on. Reproduced on lan969x by redirecting received frames out of
another port with tc mirred.
Fixes: d84ad2c0d80c ("net: lan969x: add FDMA implementation")
Signed-off-by: Daniel Machon <daniel.machon@microchip.com>
Link: https://patch.msgid.link/20261020-lan969x-fdma-tx-clear-irq-v1-1-b9f9d1eda317@microchip.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
sparx5_psfp_init() enables the periodic stream gate cycle timer update,
which keeps the cycle start time of each stream gate within one cycle
time of the current time. SG_CT_CLKS sets the update period, in system
clock cycles.
Before this patch, the update is enabled with a full register write of
SG_CT_UPDATE_ENA only, which also sets SG_CT_CLKS to zero. That is below
the documented minimum, the number of stream gates (1024 on sparx5 and
256 on lan969x).
Write both fields, and set the period to the recommended 10 us,
computed from the core clock since it differs per target (6250 cycles
at 625 MHz, 3280 at 328 MHz).
Found by an AI code review agent, and confirmed on lan969x and sparx5
by reading the register back as 0x80000000.
Fixes: e116b19db202 ("net: microchip: sparx5: initialize PSFP")
Signed-off-by: Daniel Machon <daniel.machon@microchip.com>
Link: https://patch.msgid.link/20261020-sparx5-psfp-cycle-timer-v1-1-00f2908f096a@microchip.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
lan969x_dsm_calendar_calc() computes the DSM calendar schedule of one
taxi. sparx5_config_dsm_calendar() calls it for each taxi, reusing the
same sparx5_calendar_data.
Before this patch, the schedule is only cleared after the n_devs == 0
early return. For a taxi with no active ports, the function writes
SPX5_DSM_CAL_EMPTY only to schedule[0] and leaves the remaining slots
at whatever they held before. For the first taxi that is zero from
kzalloc(), so sparx5_dsm_cal_len() returns a bogus length and
sparx5_dsm_calendar_check() fails, which fails the probe:
Port 0: distance 20 above limit 0
A later empty taxi inherits the previous taxi's schedule instead, which
passes the check, so only an empty taxi 0 triggers the failure.
Clear the full schedule up front on every path, and on the empty path
place a single idle marker (SPX5_DSM_CAL_MAX_DEVS_PER_TAXI) in slot 0,
so the update step sees a calendar length of one and programs one idle
slot into the hardware, matching what sparx5_dsm_calendar_calc() does.
Reported by a customer whose DT enables none of the ports in a taxi.
Reproduced on lan969x by disabling all taxi 0 ports in the DT.
Fixes: 5d2ba3941016 ("net: lan969x: add function for calculating the DSM calendar")
Signed-off-by: Daniel Machon <daniel.machon@microchip.com>
Link: https://patch.msgid.link/20261020-lan969x-dsm-empty-taxi-v1-1-14790f1e7430@microchip.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
In big packets mode with guest GSO, add_recvbuf_big() posts a chain of
MAX_SKB_FRAGS + 1 pages. page_to_skb() copies only ETH_HLEN of the
first page into the linear area and adds the rest of that page as a
frag, then one frag per remaining page. A device that fills the entire
buffer therefore causes MAX_SKB_FRAGS + 1 frags, writing one
skb_frag_t past the end of skb_shinfo(skb)->frags[].
Post one page less. The buffer still fits a maximal GSO packet. A well
behaved device does not trigger this, as GSO packets are at most 64KB.
Also check the frag count in page_to_skb() before adding any frag as
defense in depth.
Fixes: 9ab86bbcf8be ("virtio_net: Defer skb allocation in receive path Date: Wed, 13 Jan 2010 12:53:38 -0800")
Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Link: https://patch.msgid.link/20261020032108.81879-1-willemdebruijn.kernel@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The driver assigns netdev_ops after hbg_init(), but hbg_init() calls
hbg_mdio_init() -> hbg_phy_connect() -> phy_connect_direct() ->
phy_attach_direct() -> phy_link_topo_add_phy(), which invokes
netdev_need_ops_lock(). With CONFIG_NET_SHAPER enabled, this function
dereferences dev->netdev_ops->net_shaper_ops. Since netdev_ops is still
NULL at this point, a NULL pointer dereference occurs.
Move the netdev_ops assignment before hbg_init() so that
netdev_need_ops_lock() can safely access netdev_ops during PHY
attachment.
The issue was hit and verified on an evaluation board.
Fixes: ff4edac6e9bd ("net: hibmcge: Implement some .ndo functions")
Signed-off-by: Jijie Shao <shaojijie@huawei.com>
Reviewed-by: Daniel Machon <daniel.machon@microchip.com>
Link: https://patch.msgid.link/20261016110421.2288670-1-shaojijie@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Eric Joyner says:
====================
ionic: fix port_info lifetime bugs
Two fixes for the lifetime of the ionic port_info DMA buffer.
Patch 1 fixes a use-after-free. ionic_port_init() allocates port_info
only when the pointer is NULL, but its error path frees the buffer
unconditionally. On the firmware recovery paths,
ionic_lif_handle_fw_up() and ionic_reset_done(), the buffer belongs to
an earlier successful init, so a failed PORT_INIT frees it under any
ethtool op that is already reading it. The error path now frees only
what the same call allocated.
Patch 2 (v1 patch 3) fixes a leak of the same buffer when probe fails
after the port has been set up.
Issues were discovered by a Sashiko. I tested it on an actual DSC3
"Pollara" with a hacked driver and test kernel to cause both issues.
====================
Link: https://patch.msgid.link/20261015-ionic-port-info-lifetime-v2-0-57232013fd79@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
ionic_probe() allocates the port_info DMA buffer via ionic_setup_one()
-> ionic_port_init(). If probe then fails anywhere after that, it
unwinds through err_out_pci, which never calls ionic_port_reset();
ionic_dev_teardown() and ionic_clear_pci() do not touch port_info, and
so the buffer is leaked. The call was lost when the setup code was
moved into ionic_setup_one().
Since ionic_remove() already frees the port_info DMA buffer with
ionic_port_reset(), call that on the probe error path too. Every path
to err_out_pci runs after ionic_setup_one() succeeded, so the BARs are
still mapped and the firmware gets a PORT_RESET before the buffer is
freed, the same order as in ionic_remove().
Fixes: 0de38d9f1dba ("ionic: extract common bits from ionic_probe")
Signed-off-by: Eric Joyner <eric.joyner@amd.com>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Link: https://patch.msgid.link/20261015-ionic-port-info-lifetime-v2-2-57232013fd79@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
ionic_port_init() allocates port_info when the pointer is NULL, but its
error path frees it without checking whether this call allocated it.
Before commit ddc5911b9bd2 ("ionic: call ionic_port_init after
fw-upgrade") the function returned early when the buffer already
existed, so the two were symmetric.
Since then a failed PORT_INIT on either recovery path - fw_up or
reset_done - frees a buffer an earlier init owns, while the netdev is
registered and the ethtool ops are reading it.
Free only what this call allocated.
Fixes: ddc5911b9bd2 ("ionic: call ionic_port_init after fw-upgrade")
Signed-off-by: Nikhil P. Rao <nikhil.rao@amd.com>
Signed-off-by: Eric Joyner <eric.joyner@amd.com>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Link: https://patch.msgid.link/20261015-ionic-port-info-lifetime-v2-1-57232013fd79@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
CIFSSMBPosixLock() receives a TRANS2 response whose server-provided
DataOffset and DataCount are not required by validate_t2() to describe an
interval inside the response. A server can therefore make the client read a
cifs_posix_lock beyond the response buffer and return those kernel bytes
through F_GETLK.
Use smbCalcSize() to obtain the full response length after any multipart
TRANS2 reassembly, then reject data areas that extend past it before
dereferencing them.
Fixes: fc94cdb94462 ("[CIFS] Fix new POSIX Locking for setting lock_type correctly on unlock")
Reported-by: Ratnaker Gautam <ratnaker.gautam@kuleuven.be>
Cc: stable@vger.kernel.org
Signed-off-by: Ratnaker Gautam <ratnaker.gautam@kuleuven.be>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
The legacy CPSW driver registers its netdevs before requesting IRQs. If
a later IRQ request fails in dual-EMAC mode, the probe error path
unregisters only the primary netdev. Devres then calls free_netdev() on
the still-registered secondary netdev, triggering the reg_state BUG_ON.
Once registered, either interface can queue rx_mode_work through
dev_set_rx_mode(). The error path does not drain the work before devres
releases the netdev and its private data, allowing the worker to access
freed memory.
Use the same per-port cleanup for probe failure and remove. Unregister
each netdev whose reg_state is NETREG_REGISTERED, then drain its
rx_mode_work. The state check is needed because cpsw_probe_dual_emac()
stores the secondary pointer before register_netdev() succeeds.
The workqueue lifetime issue was found during an LLM-assisted review of
the 6.12.112-rc1 stable series. Inspecting the same path then exposed
the dual-EMAC cleanup bug.
An instrumented KASAN ARM kernel and a QEMU CPSW probe stub reproduced
both failures by forcing the TX IRQ request to return -EBUSY. Not tested
on real hardware.
Cc: stable+noautosel@kernel.org # LLM report + LLM fix, untested
Fixes: 070f9c658a59 ("net: ethernet: ti: cpsw: Push the request_irq function to the end of probe")
Fixes: 0b8c878d1173 ("net: cpsw: Execute ndo_set_rx_mode callback in a work queue")
Signed-off-by: Karl Mehltretter <kmehltretter@gmail.com>
Reviewed-by: Alexander Sverdlin <alexander.sverdlin@siemens.com>
Reviewed-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Link: https://patch.msgid.link/20261019060430.33254-1-kmehltretter@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Mina Almasry says:
====================
netmem: rename non-pp ref helpers to *_inc/dec_ref()
Rename the net_iov and netmem non-page_pool reference helpers from
get/put_net_iov(), net_devmem_get/put_net_iov(), and get/put_netmem() to
net_iov_inc/dec_ref(), net_devmem_net_iov_inc/dec_ref(), and
netmem_inc/dec_ref() to follow the object_action() naming convention.
Use *_inc_ref() and *_dec_ref() rather than netmem_ref() because
netmem_ref is already the typedef for the generic network memory handle.
====================
Link: https://patch.msgid.link/20261021015558.4124825-1-almasrymina@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Follow the object_action() naming convention for netmem non-page_pool
reference helpers by renaming get_netmem() and put_netmem() to
netmem_inc_ref() and netmem_dec_ref().
Use netmem_inc_ref() and netmem_dec_ref() rather than netmem_ref() and
netmem_unref() because netmem_ref is already a typedef (typedef unsigned
long __bitwise netmem_ref).
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Mina Almasry <almasrymina@google.com>
Link: https://patch.msgid.link/20261021015558.4124825-3-almasrymina@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Follow the object_action() naming convention for net_iov non-page_pool
reference helpers by renaming get_net_iov() and put_net_iov() to
net_iov_inc_ref() and net_iov_dec_ref(), and renaming
net_devmem_get_net_iov() and net_devmem_put_net_iov() to
net_devmem_net_iov_inc_ref() and net_devmem_net_iov_dec_ref().
Use *_inc_ref() and *_dec_ref() rather than *_ref() and *_unref() to
pair symmetrically with netmem_inc_ref() and netmem_dec_ref() (since
netmem_ref is already the typedef for the generic network memory
handle).
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Mina Almasry <almasrymina@google.com>
Link: https://patch.msgid.link/20261021015558.4124825-2-almasrymina@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Use a designated initializer for the acpi_device_id fields which makes the
code more readable and consistent with how lists are initialized in the
rest of the kernel code base. Also drop explicitly setting fields to 0
where it is redundant.
Unify the list terminator to have a single space between the brackets and
no trailing comma.
Signed-off-by: Pawel Zalewski <pzalewski@thegoodpenguin.co.uk>
Signed-off-by: Jassi Brar <jassisinghbrar@gmail.com>
|
|
Mailbox hardware registers are 64 bit wide, but the interrupt handler
currently reads only 32 bits with readl() at offset 0x4 (upper 32
bits). This returns an incorrect value, while reading using readq()
returns the expected value.
Use readq() to access the 64 bit register and pass the expected value
to the mailbox client.
Fixes: 0e2a9a03106c ("mailbox: Add support for QTI CPUCP mailbox controller")
Signed-off-by: Sivansh Gupta <sivansh.gupta@oss.qualcomm.com>
Reviewed-by: Sibi Sankar <sibi.sankar@oss.qualcomm.com>
Signed-off-by: Jassi Brar <jassisinghbrar@gmail.com>
|
|
Coia Prant says:
====================
net: pcs: add basic support for RK3568 XPCS
This series adds proper SGMII support for the Rockchip RK3568 SoC
using the integrated Synopsys DesignWare XPCS, along with necessary
fixes and refactoring in the stmmac core and XPCS driver.
Motivation
==========
The RK3568 integrates a DW XPCS accessed via APB3 and connected to
a Naneng Combo SerDes PHY. Several boards (e.g., Ariaboard
Photonicat) use this interface for Gigabit Ethernet. However, the
current upstream stmmac driver does not support this configuration,
and the XPCS driver has issues in SGMII poll mode that cause the
link to be reported incorrectly.
This series addresses these issues by:
- Adding ANRESTART support for SGMII link recovery
- Adding a Rockchip XPCS platform glue driver and wiring it up in
dwmac-rk
Series overview
===============
RK3568 XPCS/SGMII:
Patch 1: add ANRESTART support for SGMII link recovery
Patch 2: DT binding for Rockchip RK3568 XPCS
Patch 3: implement the Rockchip XPCS platform glue driver
Patch 4: DT binding for Rockchip DWMAC PCS
Patch 5: wire up SGMII support in dwmac-rk
Patch 6: update MAINTAINERS
Dependencies
============
- Generic PM domain (genpd) changes:
https://lore.kernel.org/all/20260925041751.495818-1-coiaprant@gmail.com/
- PHY binding+driver (Naneng Combo PHY SGMII MAC selection):
https://lore.kernel.org/all/20261006162807.2843858-1-coiaprant@gmail.com/
The DTS changes using the PHY property are sent separately to
Heiko/linux-rockchip. There is no hard build dependency; the pieces
can converge during the merge window.
Notes
=====
- When testing out-band mode with set_clk_tx_rate, only 1000Mbps
works on both TX/RX; 10/100Mbps only works on TX side.
Acknowledgments
===============
This work was inspired by and builds upon the excellent work of others:
- Serge Semin's Synopsys DesignWare XPCS platform driver (pcs-xpcs-plat.c)
- Clément Léger's Renesas MIIC driver (pcs-rzn1-miic.c)
- The Rockchip TRM and downstream OEM drivers
====================
Link: https://patch.msgid.link/20261017054406.1828061-1-coiaprant@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Add a MAINTAINERS entry for the Rockchip RK3568 XPCS platform driver
and its device tree binding.
Include the relevant mailing lists (netdev and linux-rockchip) so that
future patches are properly distributed.
Signed-off-by: Coia Prant <coiaprant@gmail.com>
Link: https://patch.msgid.link/20261017054406.1828061-7-coiaprant@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The RK3568 SoC integrates a Synopsys DesignWare XPCS that can be
connected to GMAC0 or GMAC1 in SGMII mode. Add the necessary glue
logic to support this configuration.
The current dwmac-rk driver does not support SGMII mode. SGMII
requires a PCS to handle auto-negotiation and link state reporting,
but the existing driver only supports RGMII and RMII.
Add a set_to_sgmii() callback to configure the GMAC GRF register for
SGMII mode (bit 7 set, interface selection bits 4:6 ignored when set).
Also add a set_to_rmii() callback for rk3568 to explicitly clear bit 7,
since the new SGMII path leaves it set and the RMII branch previously
relied on the SoC reset value.
Provide pcs_init/pcs_exit callbacks to create/destroy the XPCS via
xpcs_rk_create() from the Rockchip XPCS platform driver, and a
select_pcs callback to return the XPCS to phylink. The MII port passed
to xpcs_rk_create() is the GMAC id (bsp_priv->id): GMAC0 maps to port 0,
GMAC1 maps to port 1, matching the hardware routing.
SGMII is not added to rk_get_interfaces(): it comes from the XPCS's own
supported_interfaces, merged by stmmac_phylink_setup().
The SerDes PHY and the PD_PIPE power domain are owned by the XPCS
driver rather than managed through the stmmac
serdes_poweron/serdes_poweroff callbacks, which are legacy and meant
for single-MAC platforms. On RK3568 the XPCS is the natural owner of
the shared SerDes.
DWMAC_ROCKCHIP selects PCS_XPCS_ROCKCHIP. PCS_XPCS itself is already
selected by STMMAC_ETH, and PM is selected by ARCH_ROCKCHIP, so no
further selects are needed.
Reorder rk_gmac_powerup() so that gmac_clk_enable() is called before
the SGMII check. The SGMII path skips rk_get_phy_intf_sel(), so the
clock must be enabled earlier to cover all register accesses in that
path. While at it, unify the error unwinding into a single
clk_disable label and add error handling for the default (unhandled
interface) case.
SGMII In-band vs Out-of-band
============================
On RK3568, the MAC clock is fixed at 125 MHz and cannot be changed per
speed. Out-of-band mode requires the MAC to change its clock to
125/25/2.5 MHz for 1000/100/10 Mbps respectively, which the hardware
does not support here. The driver therefore clears rk_set_clk_tx_rate()
and the clock stays at 125 MHz.
The PCS is still told the speed in out-of-band mode (through a BMCR
write in xpcs_link_up_sgmii_1000basex()), but the MAC clock does not
follow. At 10/100 Mbps this gives working TX and failing RX.
Enable default_an_inband for SGMII and clear rk_set_clk_tx_rate(), so
phylink uses in-band mode, where the PCS is responsible for
speed/duplex negotiation.
Note that default_an_inband can be overridden by a fixed-link node, and
phylink may also fall back to out-of-band if the PHY does not support
in-band signalling. Out-of-band SGMII is not supported by this driver.
The only out-of-band configuration that would work is 1000BASE-X with a
fixed-link, where the MAC clock stays at 125 MHz; 10/100 Mbps gives
working TX and failing RX. Boards must use in-band mode
(managed = "in-band-status" or an in-band capable PHY).
1000BASE-X is not enabled by this series; it may be added later if
there is a user for it.
Link: https://dl.radxa.com/rock3/docs/hw/datasheet/Rockchip%20RK3568%20TRM%20Part1%20V1.1-20210301.pdf (Page 386)
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Coia Prant <coiaprant@gmail.com>
Link: https://patch.msgid.link/20261017054406.1828061-6-coiaprant@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The Rockchip GMAC binding needs to describe the PCS reference used by
the SGMII support added later in this series. The property will be
parsed by rk_pcs_init(), and a missing phandle fails the probe. Add it
and require it when phy-mode is "sgmii" on rockchip,rk3568-gmac, the
only SoC in this binding that has SGMII support.
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Signed-off-by: Coia Prant <coiaprant@gmail.com>
Link: https://patch.msgid.link/20261017054406.1828061-5-coiaprant@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The RK3568 SoC integrates a Synopsys DesignWare XPCS that is accessed
via APB3 memory-mapped registers. This driver provides the glue logic
to make the XPCS accessible to the generic pcs-xpcs core.
The XPCS block contains four MII ports (0..3). On RK3568 the MII port
used by SGMII follows the MAC it is routed to: GMAC0 maps to port 0 and
GMAC1 maps to port 1. The glue driver receives the port index from the
MAC side and passes it to xpcs_create_mdiodev().
The hardware maps these ports to different MMDs:
- port 0: MMD 7 (ROCKCHIP_MMD_MII)
- port 1: MMD 2 (ROCKCHIP_MMD_MII1)
- port 2: MMD 3 (ROCKCHIP_MMD_MII2)
- port 3: MMD 4 (ROCKCHIP_MMD_MII3)
This driver creates a virtual MDIO bus that translates MDIO operations
to APB3 register accesses, with proper address remapping for each port.
The generic xpcs driver then creates a phylink_pcs instance on top of
this bus, allowing the MAC to use the PCS via the standard phylink API.
The generic XPCS platform glue (pcs-xpcs-plat.o) is split out of the
pcs_xpcs composite object into its own module, gated behind the new
PCS_XPCS_PLATFORM symbol. There is no in-tree user of the generic
snps,dw-xpcs glue that would need it selected by default, so the symbol
has no default and platforms that need it select it explicitly.
PCS_XPCS_ROCKCHIP selects GENERIC_PHY and PM_GENERIC_DOMAINS.
ARCH_ROCKCHIP already selects PM, so the dependency of PM_GENERIC_DOMAINS
on PM is satisfied on the target platform.
The EEE multiplier is derived at runtime from the EEE clock rate instead
of being hardcoded, because the clock is muxed between gpll200 (200 MHz)
and cpll125 (125 MHz) and can be changed by the board or firmware. A
64-bit intermediate avoids overflow on 32-bit builds, and the result is
clamped to the 4-bit DW_VR_MII_EEE_MULT_FACT_100NS field.
Power management
================
The XPCS, its SerDes PHY and the SATA/PCIe controllers share the PD_PIPE
power domain on RK3568. genpd powers the domain down during system
suspend once every consumer is suspended, which also kills the SerDes
and the XPCS.
Boards that rely on MAC WoL need the PCS and SerDes alive to receive
magic packets, so PD_PIPE must stay powered across system suspend. Keep
it on by marking the XPCS as part of the wakeup path via
device_set_wakeup_path(). genpd then leaves the domain powered, because
the Rockchip power domain driver sets GENPD_FLAG_ACTIVE_WAKEUP on
PD_PIPE, which makes genpd check the wakeup path of its consumers during
system suspend.
This is unconditional because the XPCS core currently has no platform
callback through which a consumer could convey the MAC WoL state, and
the state itself is dynamic: WoL is toggled at runtime via ethtool, so
a static DT property cannot express it either. The affected hardware
is limited to RK3568 boards using SGMII, which are always-on routers
where system suspend is not a realistic use case; the extra power draw
is therefore acceptable.
Runtime PM is unaffected: the runtime callbacks only gate the CSR clock,
and dev_pm_genpd_rpm_always_on() already keeps PD_PIPE on at runtime.
Link: https://dl.radxa.com/rock3/docs/hw/datasheet/Rockchip%20RK3568%20TRM%20Part1%20V1.1-20210301.pdf (Page 59, CRU_CLKSEL_CON29)
Link: https://dl.radxa.com/rock3/docs/hw/datasheet/Rockchip%20RK3568%20TRM%20Part2%20V1.1-20210301.pdf (Page 2078)
Signed-off-by: Coia Prant <coiaprant@gmail.com>
Link: https://patch.msgid.link/20261017054406.1828061-4-coiaprant@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Add device tree binding documentation for the Synopsys DesignWare
XPCS integrated on the Rockchip RK3568 SoC.
The XPCS is accessed over the APB3 bus and internally connected to
a Naneng Combo SerDes PHY. It supports 1000BASE-X, SGMII, and
QSGMII modes, with four MII ports.
The MII port used by the SGMII link is derived from the MAC it is
routed to: GMAC0 uses port 0, GMAC1 uses port 1. The port index is
not an independent board-level configuration, so the binding does
not describe the ports. If a future SoC needs per-port configuration
that cannot be derived, pcs-handle argument cells can be added then.
The CRU reset lines (SRST_XPCS*) are intentionally not described: no
in-tree user requests them, and bring-up relies on the PD_PIPE power
domain, the SerDes PHY and the in-IP soft reset. They can be added
later as optional without breaking ABI.
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Signed-off-by: Coia Prant <coiaprant@gmail.com>
Link: https://patch.msgid.link/20261017054406.1828061-3-coiaprant@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
On some hardware using the DesignWare XPCS IP (e.g., RK3568 MAC side
SGMII), the PCS does not automatically restart auto-negotiation when the
link goes down and comes back up. Without an explicit ANRESTART, the link
stays down forever.
Add BMCR_ANRESTART in two places:
1. In xpcs_config_aneg_c37_sgmii(), when starting AN, set ANRESTART
alongside ANENABLE to initiate a fresh negotiation.
2. In xpcs_get_state_c37_sgmii(), when link is down and AN completion is
detected, clear the interrupt and trigger ANRESTART to restart the
negotiation process. Propagate the return value of the restart so
errors are not silently ignored.
The restart cannot go through the .pcs_an_restart op: phylink only
calls it for 802.3z interfaces, and SGMII is not one. Changing hardware
state from pcs_get_state() is already done elsewhere in this driver
(xpcs_get_state_c73() calls xpcs_soft_reset() and xpcs_do_config()),
so the same pattern is used here.
The latch is cleared before issuing the restart, not after: clearing it
afterwards would discard a freshly latched ANCMPLT from the new
negotiation. If an MDIO access fails at this point, it indicates an
unrecoverable hardware condition until reset.
Also clear DW_VR_MII_AN_INTR_STS in xpcs_config_aneg_c37_sgmii() before
starting AN, matching what xpcs_config_aneg_c37_1000basex() already does.
On the non-inband path the function now returns the result of that write
instead of the DIG_CTRL1 modify.
Update the comment in xpcs_config_aneg_c37_sgmii() to note that although
the DesignWare databook says AN restart is not needed for MAC side SGMII,
some implementations (e.g. Rockchip RK3568) require it to recover the
link after a disconnect.
This is not a fix for an existing mainline platform: the affected
platform (RK3568 XPCS) is introduced later in the same series.
Tested-by: Jiawen Wu <jiawenwu@trustnetic.com>
Tested-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Coia Prant <coiaprant@gmail.com>
Link: https://patch.msgid.link/20261017054406.1828061-2-coiaprant@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Maxime Chevallier says:
====================
net: phy_port: SFP modules representation and phy_port listing
Thanks to Christophe for reviewing and testing, this is much
appreciated.
This work extends on the recent addition of phy_port representation to enable
listing the front-facing ports of an interface. For now, we don't control
these ports, we merely list their presence and their capabilities.
As the most common use-case of multi-port interfaces is combo-ports that
provide both RJ45 and SFP connectors on a single MAC, there's a lot of
SFP stuff in this series.
This series is in 2 main parts. The first one aims at representing the
SFP cages and modules using phy_port, as combo-ports with RJ45 + SFP are
by far the most common cases for multi-connector setups.
The second part is the netlink interface to list those ports, now that
most use-cases are covered.
Let's see what we can do with some examples of the new ethtool API :
- Get MII interfaces supported by an empty SFP cage :
Port for eth3:
Port id: 1
Supported MII interfaces : sgmii, 1000base-x, 2500base-x
Port type: sfp
- Get Combo-ports supported modes, on each port :
Port for eth1:
Port id: 1
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
10000baseT/Full
2500baseT/Full
5000baseT/Full
Port type: mdi
Port for eth1:
Port id: 2
Supported MII interfaces : 10gbase-r
Port type: sfp
- Get Achievable linkmodes on a SFP module (combo port with a DAC in the
SFP cage)
Port for eth1:
Port id: 1
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
10000baseT/Full
2500baseT/Full
5000baseT/Full
Port type: mdi
Port for eth1:
Port id: 2
Supported MII interfaces : 10gbase-r
Port type: sfp
Port for eth1:
Port id: 3
Upstream id: 2
Supported link modes: 10000baseCR/Full
Port type: mdi
Note that here, we have 3 ports :
- The Copper port
- The SFP Cage itself,
- The SFP module
This series builds on top of phy_port and phy_link_topology to allow
tracking the ports of an interface. We maintain a list of supported
linkmodes/interfaces on each port, which allows for fine-grained
reporting of each port's capability.
What this series doesn't do :
- We don't support selecting which port is active. This is the next step.
- We only support PHY-driven combo ports. The end-goal of this whole
journey that started with phy_link_topology is to get support for MII
muxes, such as the one we have on the Turris Omnia. This will eventually
be upstreamed as well.
====================
Link: https://patch.msgid.link/20261016133107.54054-1-maxime.chevallier@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
With SFP now properly supported with phy_port, add some details in the
documentation. Fix a typo along the way (driver -> driven).
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20261016133107.54054-11-maxime.chevallier@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Expose the phy_port information to userspace, so that we can know how
many ports are available on a given interface, as well as their
capabilities. For MDI ports, we report the list of supported linkmodes
based on what the PHY that drives this port says.
For MII ports, i.e. empty SFP cages, we report the MII linkmodes that we
can output on this port.
Tested-by: Aleksei Sviridkin <f@lex.la>
Reviewed-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Tested-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20261016133107.54054-10-maxime.chevallier@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Ethernet network interfaces may have more than one front-facing port.
The phy_port infrastructure was introduced to keep track of
these ports, and allow userspace to know about the presence and
capability of these ports. Add a ethnl netlink message to report this
information.
Reviewed-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20261016133107.54054-9-maxime.chevallier@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
In order to allow netlink access to phy_ports, let's add a helper to
retrieve them. When handling a port coming from phy_link_topology, the
caller must hold rtnl until it's done with it.
We don't really support PHY and phy_port operation on ops-locked devices
yet, warn when we try to do so on such an interface.
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Aleksei Sviridkin <f@lex.la>
Reviewed-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Tested-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20261016133107.54054-8-maxime.chevallier@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
MII phy_ports are not meant to be connected directly to a link partner.
They are meant to feed into some media converter devices that will
expose an MDI phy_port, so far we only support SFP modules for that.
In the case an MDI phy_port is backed by an MII port (e.g. a SFP
module's port, backed by the SFP cage port), let's keep track of the
port id of the MII port backing it.
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Aleksei Sviridkin <f@lex.la>
Reviewed-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Tested-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20261016133107.54054-7-maxime.chevallier@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|