Since the GuC is reset during GT reset, we need to re-send the
entire SR-IOV provisioning configuration to the GuC. But since
this whole configuration is protected by the PF master mutex and
we can't avoid making allocations under this mutex (like during
LMEM provisioning), we can't do this reprovisioning from gt-reset
path if we want to be reclaim-safe. Move VFs reprovisioning to a
async worker that we will start from the gt-reset path.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Michał Winiarski <michal.winiarski@intel.com> Reviewed-by: Stuart Summers <stuart.summers@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20250125215505.720-1-michal.wajdeczko@intel.com
Stable-dep-of: 81dccec448d2 ("drm/xe/pf: Prepare to stop SR-IOV support prior GT reset") Signed-off-by: Sasha Levin <sashal@kernel.org>
Some VF accessible registers (like GuC scratch registers) must be
explicitly reset during the FLR. While this is today done by the GuC
firmware, according to the design, this should be responsibility of
the PF driver, as future platforms may require more registers to be
reset. Likewise GuC, the PF can access VFs registers by adding some
platform specific offset to the original register address.
A crash in conntrack was reported while trying to unlink the conntrack
entry from the hash bucket list:
[exception RIP: __nf_ct_delete_from_lists+172]
[..]
#7 [ff539b5a2b043aa0] nf_ct_delete at ffffffffc124d421 [nf_conntrack]
#8 [ff539b5a2b043ad0] nf_ct_gc_expired at ffffffffc124d999 [nf_conntrack]
#9 [ff539b5a2b043ae0] __nf_conntrack_find_get at ffffffffc124efbc [nf_conntrack]
[..]
The nf_conn struct is marked as allocated from slab but appears to be in
a partially initialised state:
ct hlist pointer is garbage; looks like the ct hash value
(hence crash).
ct->status is equal to IPS_CONFIRMED|IPS_DYING, which is expected
ct->timeout is 30000 (=30s), which is unexpected.
Everything else looks like normal udp conntrack entry. If we ignore
ct->status and pretend its 0, the entry matches those that are newly
allocated but not yet inserted into the hash:
- ct hlist pointers are overloaded and store/cache the raw tuple hash
- ct->timeout matches the relative time expected for a new udp flow
rather than the absolute 'jiffies' value.
If it were not for the presence of IPS_CONFIRMED,
__nf_conntrack_find_get() would have skipped the entry.
Theory is that we did hit following race:
cpu x cpu y cpu z
found entry E found entry E
E is expired <preemption>
nf_ct_delete()
return E to rcu slab
init_conntrack
E is re-inited,
ct->status set to 0
reply tuplehash hnnode.pprev
stores hash value.
cpu y found E right before it was deleted on cpu x.
E is now re-inited on cpu z. cpu y was preempted before
checking for expiry and/or confirm bit.
->refcnt set to 1
E now owned by skb
->timeout set to 30000
If cpu y were to resume now, it would observe E as
expired but would skip E due to missing CONFIRMED bit.
nf_conntrack_confirm gets called
sets: ct->status |= CONFIRMED
This is wrong: E is not yet added
to hashtable.
cpu y resumes, it observes E as expired but CONFIRMED:
<resumes>
nf_ct_expired()
-> yes (ct->timeout is 30s)
confirmed bit set.
cpu y will try to delete E from the hashtable:
nf_ct_delete() -> set DYING bit
__nf_ct_delete_from_lists
Even this scenario doesn't guarantee a crash:
cpu z still holds the table bucket lock(s) so y blocks:
wait for spinlock held by z
CONFIRMED is set but there is no
guarantee ct will be added to hash:
"chaintoolong" or "clash resolution"
logic both skip the insert step.
reply hnnode.pprev still stores the
hash value.
unlocks spinlock
return NF_DROP
<unblocks, then
crashes on hlist_nulls_del_rcu pprev>
In case CPU z does insert the entry into the hashtable, cpu y will unlink
E again right away but no crash occurs.
Without 'cpu y' race, 'garbage' hlist is of no consequence:
ct refcnt remains at 1, eventually skb will be free'd and E gets
destroyed via: nf_conntrack_put -> nf_conntrack_destroy -> nf_ct_destroy.
To resolve this, move the IPS_CONFIRMED assignment after the table
insertion but before the unlock.
Pablo points out that the confirm-bit-store could be reordered to happen
before hlist add resp. the timeout fixup, so switch to set_bit and
before_atomic memory barrier to prevent this.
It doesn't matter if other CPUs can observe a newly inserted entry right
before the CONFIRMED bit was set:
Such event cannot be distinguished from above "E is the old incarnation"
case: the entry will be skipped.
Also change nf_ct_should_gc() to first check the confirmed bit.
The gc sequence is:
1. Check if entry has expired, if not skip to next entry
2. Obtain a reference to the expired entry.
3. Call nf_ct_should_gc() to double-check step 1.
nf_ct_should_gc() is thus called only for entries that already failed an
expiry check. After this patch, once the confirmed bit check passes
ct->timeout has been altered to reflect the absolute 'best before' date
instead of a relative time. Step 3 will therefore not remove the entry.
Without this change to nf_ct_should_gc() we could still get this sequence:
1. Check if entry has expired.
2. Obtain a reference.
3. Call nf_ct_should_gc() to double-check step 1:
4 - entry is still observed as expired
5 - meanwhile, ct->timeout is corrected to absolute value on other CPU
and confirm bit gets set
6 - confirm bit is seen
7 - valid entry is removed again
First do check 6), then 4) so the gc expiry check always picks up either
confirmed bit unset (entry gets skipped) or expiry re-check failure for
re-inited conntrack objects.
This change cannot be backported to releases before 5.19. Without
commit 8a75a2c17410 ("netfilter: conntrack: remove unconfirmed list")
|= IPS_CONFIRMED line cannot be moved without further changes.
Since "net: gro: use cb instead of skb->network_header", the skb network
header is no longer set in the GRO path.
This breaks fraglist segmentation, which relies on ip_hdr()/tcp_hdr()
to check for address/port changes.
Fix this regression by selectively setting the network header for merged
segment skbs.
Fixes: 186b1ea73ad8 ("net: gro: use cb instead of skb->network_header") Signed-off-by: Felix Fietkau <nbd@nbd.name> Reviewed-by: Willem de Bruijn <willemb@google.com> Link: https://patch.msgid.link/20250705150622.10699-1-nbd@nbd.name Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
gso_size is expected by the networking stack to be the size of the
payload (thus, not including ethernet/IP/TCP-headers). However, cqe_bcnt
is the full sized frame (including the headers). Dividing cqe_bcnt by
lro_num_seg will then give incorrect results.
For example, running a bpftrace higher up in the TCP-stack
(tcp_event_data_recv), we commonly have gso_size set to 1450 or 1451 even
though in reality the payload was only 1448 bytes.
This can have unintended consequences:
- In tcp_measure_rcv_mss() len will be for example 1450, but. rcv_mss
will be 1448 (because tp->advmss is 1448). Thus, we will always
recompute scaling_ratio each time an LRO-packet is received.
- In tcp_gro_receive(), it will interfere with the decision whether or
not to flush and thus potentially result in less gro'ed packets.
So, we need to discount the protocol headers from cqe_bcnt so we can
actually divide the payload by lro_num_seg to get the real gso_size.
For GF variant of WCN6855 without board ID programmed
btusb_generate_qca_nvm_name() will chose wrong NVM
'qca/nvm_usb_00130201.bin' to download.
Fix by choosing right NVM 'qca/nvm_usb_00130201_gf.bin'.
Also simplify NVM choice logic of btusb_generate_qca_nvm_name().
Fixes: d6cba4e6d0e2 ("Bluetooth: btusb: Add support using different nvm for variant WCN6855 controller") Signed-off-by: Zijun Hu <zijun.hu@oss.qualcomm.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
This replaces the usage of HCI_ERROR_REMOTE_USER_TERM, which as the name
suggest is to indicate a regular disconnection initiated by an user,
with HCI_ERROR_AUTH_FAILURE to indicate the session has timeout thus any
pairing shall be considered as failed.
Fixes: 1e91c29eb60c ("Bluetooth: Use hci_disconnect for immediate disconnection from SMP") Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
If a command is received while a bonding is ongoing consider it a
pairing failure so the session is cleanup properly and the device is
disconnected immediately instead of continuing with other commands that
may result in the session to get stuck without ever completing such as
the case bellow:
> ACL Data RX: Handle 2048 flags 0x02 dlen 21
SMP: Identity Information (0x08) len 16
Identity resolving key[16]: d7e08edef97d3e62cd2331f82d8073b0
> ACL Data RX: Handle 2048 flags 0x02 dlen 21
SMP: Signing Information (0x0a) len 16
Signature key[16]: 1716c536f94e843a9aea8b13ffde477d
Bluetooth: hci0: unexpected SMP command 0x0a from XX:XX:XX:XX:XX:XX
> ACL Data RX: Handle 2048 flags 0x02 dlen 12
SMP: Identity Address Information (0x09) len 7
Address: XX:XX:XX:XX:XX:XX (Intel Corporate)
While accourding to core spec 6.1 the expected order is always BD_ADDR
first first then CSRK:
When using LE legacy pairing, the keys shall be distributed in the
following order:
LTK by the Peripheral
EDIV and Rand by the Peripheral
IRK by the Peripheral
BD_ADDR by the Peripheral
CSRK by the Peripheral
LTK by the Central
EDIV and Rand by the Central
IRK by the Central
BD_ADDR by the Central
CSRK by the Central
When using LE Secure Connections, the keys shall be distributed in the
following order:
IRK by the Peripheral
BD_ADDR by the Peripheral
CSRK by the Peripheral
IRK by the Central
BD_ADDR by the Central
CSRK by the Central
According to the Core 6.1 for commands used for key distribution "Key
Rejected" can be used:
'3.6.1. Key distribution and generation
A device may reject a distributed key by sending the Pairing Failed command
with the reason set to "Key Rejected".
Fixes: b28b4943660f ("Bluetooth: Add strict checks for allowed SMP PDUs") Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently, the connectable flag used by the setup of an extended
advertising instance drives whether we require privacy when trying to pass
a random address to the advertising parameters (Own Address).
If privacy is not required, then it automatically falls back to using the
controller's public address. This can cause problems when using controllers
that do not have a public address set, but instead use a static random
address.
e.g. Assume a BLE controller that does not have a public address set.
The controller upon powering is set with a random static address by default
by the kernel.
< HCI Command: LE Set Random Address (0x08|0x0005) plen 6
Address: E4:AF:26:D8:3E:3A (Static)
> HCI Event: Command Complete (0x0e) plen 4
LE Set Random Address (0x08|0x0005) ncmd 1
Status: Success (0x00)
Setting non-connectable extended advertisement parameters in bluetoothctl
mgmt
add-ext-adv-params -r 0x801 -x 0x802 -P 2M -g 1
correctly sets Own address type as Random
< HCI Command: LE Set Extended Advertising Parameters (0x08|0x0036)
plen 25
...
Own address type: Random (0x01)
Setting connectable extended advertisement parameters in bluetoothctl mgmt
mistakenly sets Own address type to Public (which causes to use Public
Address 00:00:00:00:00:00)
< HCI Command: LE Set Extended Advertising Parameters (0x08|0x0036)
plen 25
...
Own address type: Public (0x00)
This causes either the controller to emit an Invalid Parameters error or to
mishandle the advertising.
This patch makes sure that we use the already set static random address
when requesting a connectable extended advertising when we don't require
privacy and our public address is not set (00:00:00:00:00:00).
Fixes: 3fe318ee72c5 ("Bluetooth: move hci_get_random_address() to hci_sync") Signed-off-by: Alessandro Gasbarroni <alex.gasbarroni@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
syzbot reported null-ptr-deref in l2cap_sock_resume_cb(). [0]
l2cap_sock_resume_cb() has a similar problem that was fixed by commit 1bff51ea59a9 ("Bluetooth: fix use-after-free error in lock_sock_nested()").
Since both l2cap_sock_kill() and l2cap_sock_resume_cb() are executed
under l2cap_sock_resume_cb(), we can avoid the issue simply by checking
if chan->data is NULL.
Let's not access to the killed socket in l2cap_sock_resume_cb().
[0]:
BUG: KASAN: null-ptr-deref in instrument_atomic_write include/linux/instrumented.h:82 [inline]
BUG: KASAN: null-ptr-deref in clear_bit include/asm-generic/bitops/instrumented-atomic.h:41 [inline]
BUG: KASAN: null-ptr-deref in l2cap_sock_resume_cb+0xb4/0x17c net/bluetooth/l2cap_sock.c:1711
Write of size 8 at addr 0000000000000570 by task kworker/u9:0/52
force_sig_fault() takes a spinlock, which is a sleeping lock with
CONFIG_PREEMPT_RT=y. However, exception handling calls force_sig_fault()
with interrupt disabled, causing a sleeping in atomic context warning.
This can be reproduced using userspace programs such as:
int main() { asm ("ebreak"); }
or
int main() { asm ("unimp"); }
There is no reason that interrupt must be disabled while handling
exceptions from userspace.
Enable interrupt while handling user exceptions. This also has the added
benefit of avoiding unnecessary delays in interrupt handling.
Fixes: f0bddf50586d ("riscv: entry: Convert to generic entry") Suggested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Nam Cao <namcao@linutronix.de> Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com> Link: https://lore.kernel.org/r/20250625085630.3649485-1-namcao@linutronix.de Signed-off-by: Palmer Dabbelt <palmer@dabbelt.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The lockdep tool can report a circular lock dependency warning in the loop
driver's AIO read/write path:
```
[ 6540.587728] kworker/u96:5/72779 is trying to acquire lock:
[ 6540.593856] ff110001b5968440 (sb_writers#9){.+.+}-{0:0}, at: loop_process_work+0x11a/0xf70 [loop]
[ 6540.603786]
[ 6540.603786] but task is already holding lock:
[ 6540.610291] ff110001b5968440 (sb_writers#9){.+.+}-{0:0}, at: loop_process_work+0x11a/0xf70 [loop]
[ 6540.620210]
[ 6540.620210] other info that might help us debug this:
[ 6540.627499] Possible unsafe locking scenario:
[ 6540.627499]
[ 6540.634110] CPU0
[ 6540.636841] ----
[ 6540.639574] lock(sb_writers#9);
[ 6540.643281] lock(sb_writers#9);
[ 6540.646988]
[ 6540.646988] *** DEADLOCK ***
```
This patch fixes the issue by using the AIO-specific helpers
`kiocb_start_write()` and `kiocb_end_write()`. These functions are
designed to be used with a `kiocb` and manage write sequencing
correctly for asynchronous I/O without introducing the problematic
lock dependency.
The `kiocb` is already part of the `loop_cmd` struct, so this change
also simplifies the completion function `lo_rw_aio_do_completion()` by
using the `iocb` from the `cmd` struct directly, instead of retrieving
the loop device from the request queue.
Fixes: 39d86db34e41 ("loop: add file_start_write() and file_end_write()") Cc: Changhui Zhong <czhong@redhat.com> Signed-off-by: Ming Lei <ming.lei@redhat.com> Link: https://lore.kernel.org/r/20250716114808.3159657-1-ming.lei@redhat.com Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Sasha Levin <sashal@kernel.org>
The driver checks for having three endpoints and
having bulk in and out endpoints, but not that
the third endpoint is interrupt input.
Rectify the omission.
The function ice_lag_is_switchdev_running() is being called from outside of
the LAG event handler code. This results in the lag->upper_netdev being
NULL sometimes. To avoid a NULL-pointer dereference, there needs to be a
check before it is dereferenced.
Fixes: 776fe19953b0 ("ice: block default rule setting on LAG interface") Signed-off-by: Dave Ertman <david.m.ertman@intel.com> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Tested-by: Sujai Buvaneswaran <sujai.buvaneswaran@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The mentioned test is not very stable when running on top of
debug kernel build. Increase the inter-packet timeout to allow
more slack in such environments.
Fixes reset GPIO usage during probe by ensuring we retrieve the GPIO and
take the device out of reset (if it defaults to being in reset) before
we attempt to communicate with the device. This is achieved by moving
the call to tcan4x5x_get_gpios() before tcan4x5x_find_version() and
avoiding any device communication while getting the GPIOs. Once we
determine the version, we can then take the knowledge of which GPIOs we
obtained and use it to decide whether we need to disable the wake or
state pin functions within the device.
This change is necessary in a situation where the reset GPIO is pulled
high externally before the CPU takes control of it, meaning we need to
explicitly bring the device out of reset before we can start
communicating with it at all.
This also has the effect of fixing an issue where a reset of the device
would occur after having called tcan4x5x_disable_wake(), making the
original behavior not actually disable the wake. This patch should now
disable wake or state pin functions well after the reset occurs.
Signed-off-by: Brett Werling <brett.werling@garmin.com> Link: https://patch.msgid.link/20250711141728.1826073-1-brett.werling@garmin.com Cc: Markus Schneider-Pargmann <msp@baylibre.com> Fixes: 142c6dc6d9d7 ("can: tcan4x5x: Add support for tcan4552/4553") Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
The nWKRQ pin supports an output voltage of either the internal reference
voltage (3.6V) or the reference voltage of
the digital interface 0-6V (VIO).
Add the devicetree option ti,nwkrq-voltage-vio to set it to VIO.
If this property is omitted the reset default, the internal reference
voltage, is used.
Signed-off-by: Sean Nyekjaer <sean@geanix.com> Reviewed-by: Marc Kleine-Budde <mkl@pengutronix.de> Reviewed-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr> Link: https://patch.msgid.link/20241114-tcan-wkrqv-v5-2-a2d50833ed71@geanix.com
[mkl: remove unused variable in tcan4x5x_get_dt_data()] Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Stable-dep-of: 0f97a7588db7 ("can: tcan4x5x: fix reset gpio usage during probe") Signed-off-by: Sasha Levin <sashal@kernel.org>
This reverts commit e3eac9f32ec0 ("wifi: cfg80211: Annotate struct
cfg80211_scan_request with __counted_by").
This really has been a completely failed experiment. There were
no actual bugs found, and yet at this point we already have four
"fixes" to it, with nothing to show for but code churn, and it
never even made the code any safer.
In all of the cases that ended up getting "fixed", the structure
is also internally inconsistent after the n_channels setting as
the channel list isn't actually filled yet. You cannot scan with
such a structure, that's just wrong. In mac80211, the struct is
also reused multiple times, so initializing it once is no good.
Some previous "fixes" (e.g. one in brcm80211) are also just setting
n_channels before accessing the array, under the assumption that the
code is correct and the array can be accessed, further showing that
the whole thing is just pointless when the allocation count and use
count are not separate.
If we really wanted to fix it, we'd need to separately track the
number of channels allocated and the number of channels currently
used, but given that no bugs were found despite the numerous syzbot
reports, that'd just be a waste of time.
Remove the __counted_by() annotation. We really should also remove
a number of the n_channels settings that are setting up a structure
that's inconsistent, but that can wait.
When restoring the default socket callbacks during a TLS handshake, we
need to acquire a write lock on sk_callback_lock. Previously, a read
lock was used, which is insufficient for modifying sk_user_data and
sk_data_ready.
1) initialize nvme_request and clear flags;
2) set NVME_MPATH_IO_STATS and increase inflight counter when IO
started;
3) check NVME_MPATH_IO_STATS and decrease inflight counter when IO is
done;
However, for the case nvme_fail_nonready_command(), both step 1) and 2)
are skipped, and if old nvme_request set NVME_MPATH_IO_STATS and then
request is reused, step 3) will still be executed, causing inflight I/O
counter to be negative.
Fix the problem by clearing nvme_request in nvme_fail_nonready_command().
Fixes: ea5e5f42cd2c ("nvme-fabrics: avoid double completions in nvmf_fail_nonready_command") Reported-by: Yi Zhang <yi.zhang@redhat.com> Closes: https://lore.kernel.org/all/CAHj4cs_+dauobyYyP805t33WMJVzOWj=7+51p4_j9rA63D9sog@mail.gmail.com/ Signed-off-by: Yu Kuai <yukuai3@huawei.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
If a PHY has no driver, the genphy driver is probed/removed directly in
phy_attach/detach. If the PHY's ofnode has an "leds" subnode, then the
LEDs will be (un)registered when probing/removing the genphy driver.
This could occur if the leds are for a non-generic driver that isn't
loaded for whatever reason. Synchronously removing the PHY device in
phy_detach leads to the following deadlock:
There is a corresponding deadlock on the open/register side of things
(and that one is reported by lockdep), but it requires a race while this
one is deterministic.
Generic PHYs do not support LEDs anyway, so don't bother registering
them.
Fixes: 01e5b728e9e4 ("net: phy: Add a binding for PHY LEDs") Signed-off-by: Sean Anderson <sean.anderson@linux.dev> Link: https://patch.msgid.link/20250707195803.666097-1-sean.anderson@linux.dev Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
syzbot reported weird splats [0][1] in cipso_v4_sock_setattr() while
freeing inet_sk(sk)->inet_opt.
The address was freed multiple times even though it was read-only memory.
cipso_v4_sock_setattr() did nothing wrong, and the root cause was type
confusion.
The cited commit made it possible to create smc_sock as an INET socket.
The issue is that struct smc_sock does not have struct inet_sock as the
first member but hijacks AF_INET and AF_INET6 sk_family, which confuses
various places.
In this case, inet_sock.inet_opt was actually smc_sock.clcsk_data_ready(),
which is an address of a function in the text segment.
When inserting a namespace into the controller's namespace list, the
function uses list_add_rcu() when the namespace is inserted in the middle
of the list, but falls back to a regular list_add() when adding at the
head of the list.
This inconsistency could lead to race conditions during concurrent
access, as users might observe a partially updated list. Fix this by
consistently using list_add_rcu() in both code paths to ensure proper
RCU protection throughout the entire function.
Fixes: be647e2c76b2 ("nvme: use srcu for iterating namespace list") Signed-off-by: Zheng Qixing <zhengqixing@huawei.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Lifetime of new_dn_mark is controlled by that of its ->fsn_mark,
pointed to by new_fsn_mark. Unfortunately, a failure exit had
been inserted between the allocation of new_dn_mark and the
call of fsnotify_init_mark(), ending up with a leak.
Fixes: 1934b212615d "file: reclaim 24 bytes from f_owner" Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Link: https://lore.kernel.org/20250712171843.GB1880847@ZenIV Signed-off-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The issue occurs when umount has already released its reference to the
superblock. When _cifsFileInfo_put() calls cifs_sb_deactive(), this
releases the last reference, triggering the immediate cleanup of all
inodes under RCU. However, cifs_oplock_break() continues to access the
cinode after this point, resulting in use-after-free.
Fix this by holding an extra reference to the superblock during the
entire oplock break operation. This ensures that the superblock and
its inodes remain valid until the oplock break completes.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=220309 Fixes: b98749cac4a6 ("CIFS: keep FileInfo handle live during oplock break") Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.org> Signed-off-by: Wang Zhaolong <wangzhaolong@huaweicloud.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The buggy address belongs to the object at ffff888122bf96c0
which belongs to the cache skbuff_small_head of size 704
The buggy address is located 24 bytes inside of
freed 704-byte region [ffff888122bf96c0, ffff888122bf9980)
Memory state around the buggy address: ffff888122bf9580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff888122bf9600: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
>ffff888122bf9680: fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb fb
^ ffff888122bf9700: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff888122bf9780: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
Fixes: a7a29f9c361f8 ("net: ipv6: add rpl sr tunnel") Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
A race condition can occur when 'agg' is modified in qfq_change_agg
(called during qfq_enqueue) while other threads access it
concurrently. For example, qfq_dump_class may trigger a NULL
dereference, and qfq_delete_class may cause a use-after-free.
This patch addresses the issue by:
1. Moved qfq_destroy_class into the critical section.
2. Added sch_tree_lock protection to qfq_dump_class and
qfq_dump_class_stats.
Fixes: 462dbc9101ac ("pkt_sched: QFQ Plus: fair-queueing service at DRR cost") Signed-off-by: Xiang Mei <xmei5@asu.edu> Reviewed-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The kobject for the queue, `disk->queue_kobj`, is initialized with a
reference count of 1 via `kobject_init()` in `blk_register_queue()`.
While `kobject_del()` is called during the unregister path to remove
the kobject from sysfs, the initial reference is never released.
Add a call to `kobject_put()` in `blk_unregister_queue()` to properly
decrement the reference count and fix the leak.
Fixes: 2bd85221a625 ("block: untangle request_queue refcounting from sysfs") Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Ming Lei <ming.lei@redhat.com> Link: https://lore.kernel.org/r/20250711083009.2574432-1-ming.lei@redhat.com Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Sasha Levin <sashal@kernel.org>
Add missing post-increment operators for byte pointers in the
loop that copies remaining bytes in xemaclite_aligned_read().
Without the increment, the same byte was written repeatedly
to the destination.
This update aligns with xemaclite_aligned_write()
In __cachefiles_write(), if the return value of the write operation > 0, it
is set to 0. This makes it impossible to distinguish scenarios where a
partial write has occurred, and will affect the outer calling functions:
1) cachefiles_write_complete() will call "term_func" such as
netfs_write_subrequest_terminated(). When "ret" in __cachefiles_write()
is used as the "transferred_or_error" of this function, it can not
distinguish the amount of data written, makes the WARN meaningless.
2) cachefiles_ondemand_fd_write_iter() can only assume all writes were
successful by default when "ret" is 0, and unconditionally return the full
length specified by user space.
Fix it by modifying "ret" to reflect the actual number of bytes written.
Furthermore, returning a value greater than 0 from __cachefiles_write()
does not affect other call paths, such as cachefiles_issue_write() and
fscache_write().
The above BPF program isn't rejected and causes a kernel warning at
runtime:
Please remove unsupported %\x00 in format string
WARNING: CPU: 1 PID: 7244 at lib/vsprintf.c:2680 format_decode+0x49c/0x5d0
This happens because bpf_bprintf_prepare skips over the second %,
detected as punctuation, while processing %p. This patch fixes it by
not skipping over punctuation. %\x00 is then processed in the next
iteration and rejected.
Reported-by: syzbot+e2c932aec5c8a6e1d31c@syzkaller.appspotmail.com Fixes: 48cac3f4a96d ("bpf: Implement formatted output helpers with bstr_printf") Acked-by: Yonghong Song <yonghong.song@linux.dev> Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com> Link: https://lore.kernel.org/r/a0e06cc479faec9e802ae51ba5d66420523251ee.1751395489.git.paul.chaignon@gmail.com Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Sometimes, its observed that during system level suspend callback
execution, after link is down, handling pending slave status workqueue
results in mipi register access failures as shown below.
Fixes: 3f5d336d64d6 ("arm64: dts: rockchip: Add support for rk3588s based board Cool Pi 4B") Signed-off-by: Andy Yan <andyshrk@163.com> Link: https://lore.kernel.org/r/20250524064223.5741-2-andyshrk@163.com Signed-off-by: Heiko Stuebner <heiko@sntech.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fixes: 791c154c3982 ("arm64: dts: rockchip: Add support for rk3588 based board Cool Pi CM5 EVB") Signed-off-by: Andy Yan <andyshrk@163.com> Link: https://lore.kernel.org/r/20250524064223.5741-1-andyshrk@163.com Signed-off-by: Heiko Stuebner <heiko@sntech.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some Comedi subdevice instruction handlers are known to access
instruction data elements beyond the first `insn->n` elements in some
cases. The `do_insn_ioctl()` and `do_insnlist_ioctl()` functions
allocate at least `MIN_SAMPLES` (16) data elements to deal with this,
but they do not initialize all of that. For Comedi instruction codes
that write to the subdevice, the first `insn->n` data elements are
copied from user-space, but the remaining elements are left
uninitialized. That could be a problem if the subdevice instruction
handler reads the uninitialized data. Ensure that the first
`MIN_SAMPLES` elements are initialized before calling these instruction
handlers, filling the uncopied elements with 0. For
`do_insnlist_ioctl()`, the same data buffer elements are used for
handling a list of instructions, so ensure the first `MIN_SAMPLES`
elements are initialized for each instruction that writes to the
subdevice.
For Comedi `INSN_READ` and `INSN_WRITE` instructions on "digital"
subdevices (subdevice types `COMEDI_SUBD_DI`, `COMEDI_SUBD_DO`, and
`COMEDI_SUBD_DIO`), it is common for the subdevice driver not to have
`insn_read` and `insn_write` handler functions, but to have an
`insn_bits` handler function for handling Comedi `INSN_BITS`
instructions. In that case, the subdevice's `insn_read` and/or
`insn_write` function handler pointers are set to point to the
`insn_rw_emulate_bits()` function by `__comedi_device_postconfig()`.
For `INSN_WRITE`, `insn_rw_emulate_bits()` currently assumes that the
supplied `data[0]` value is a valid copy from user memory. It will at
least exist because `do_insnlist_ioctl()` and `do_insn_ioctl()` in
"comedi_fops.c" ensure at lease `MIN_SAMPLES` (16) elements are
allocated. However, if `insn->n` is 0 (which is allowable for
`INSN_READ` and `INSN_WRITE` instructions, then `data[0]` may contain
uninitialized data, and certainly contains invalid data, possibly from a
different instruction in the array of instructions handled by
`do_insnlist_ioctl()`. This will result in an incorrect value being
written to the digital output channel (or to the digital input/output
channel if configured as an output), and may be reflected in the
internal saved state of the channel.
Fix it by returning 0 early if `insn->n` is 0, before reaching the code
that accesses `data[0]`. Previously, the function always returned 1 on
success, but it is supposed to be the number of data samples actually
read or written up to `insn->n`, which is 0 in this case.
Correct some left shifts of the signed integer constant 1 by some
unsigned number less than 32. Change the constant to 1U to avoid
shifting a 1 into the sign bit.
The corrected functions are comedi_dio_insn_config(),
comedi_dio_update_state(), and __comedi_device_postconfig().
The handling of the `COMEDI_INSNLIST` ioctl allocates a kernel buffer to
hold the array of `struct comedi_insn`, getting the length from the
`n_insns` member of the `struct comedi_insnlist` supplied by the user.
The allocation will fail with a WARNING and a stack dump if it is too
large.
Avoid that by failing with an `-EINVAL` error if the supplied `n_insns`
value is unreasonable.
Define the limit on the `n_insns` value in the `MAX_INSNS` macro. Set
this to the same value as `MAX_SAMPLES` (65536), which is the maximum
allowed sum of the values of the member `n` in the array of `struct
comedi_insn`, and sensible comedi instructions will have an `n` of at
least 1.
When checking for a supported IRQ number, the following test is used:
/* IRQs 2,3,5,6,7, 10,11,15 are valid for "enhanced" mode */
if ((1 << it->options[1]) & 0x8cec) {
However, `it->options[i]` is an unchecked `int` value from userspace, so
the shift amount could be negative or out of bounds. Fix the test by
requiring `it->options[1]` to be within bounds before proceeding with
the original test. Valid `it->options[1]` values that select the IRQ
will be in the range [1,15]. The value 0 explicitly disables the use of
interrupts.
When checking for a supported IRQ number, the following test is used:
/* only irqs 2, 3, 4, 5, 6, 7, 10, 11, 12, 14, and 15 are valid */
if ((1 << it->options[1]) & 0xdcfc) {
However, `it->options[i]` is an unchecked `int` value from userspace, so
the shift amount could be negative or out of bounds. Fix the test by
requiring `it->options[1]` to be within bounds before proceeding with
the original test.
Reported-by: syzbot+c52293513298e0fd9a94@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=c52293513298e0fd9a94 Fixes: 729988507680 ("staging: comedi: das16m1: tidy up the irq support in das16m1_attach()") Tested-by: syzbot+c52293513298e0fd9a94@syzkaller.appspotmail.com Suggested-by: "Enju, Kohei" <enjuk@amazon.co.jp> Cc: stable@vger.kernel.org # 5.13+ Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Link: https://lore.kernel.org/r/20250707130908.70758-1-abbotti@mev.co.uk Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When checking for a supported IRQ number, the following test is used:
if ((1 << it->options[1]) & 0xdcfc) {
However, `it->options[i]` is an unchecked `int` value from userspace, so
the shift amount could be negative or out of bounds. Fix the test by
requiring `it->options[1]` to be within bounds before proceeding with
the original test. Valid `it->options[1]` values that select the IRQ
will be in the range [1,15]. The value 0 explicitly disables the use of
interrupts.
Fixes: ad7a370c8be4 ("staging: comedi: aio_iiro_16: add command support for change of state detection") Cc: stable@vger.kernel.org # 5.13+ Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Link: https://lore.kernel.org/r/20250707134622.75403-1-abbotti@mev.co.uk Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When checking for a supported IRQ number, the following test is used:
if ((1 << it->options[1]) & board->irq_bits) {
However, `it->options[i]` is an unchecked `int` value from userspace, so
the shift amount could be negative or out of bounds. Fix the test by
requiring `it->options[1]` to be within bounds before proceeding with
the original test. Valid `it->options[1]` values that select the IRQ
will be in the range [1,15]. The value 0 explicitly disables the use of
interrupts.
Throughout the various probe functions &indio_dev->dev is used before it
is initialized. This caused a kernel panic in st_sensors_power_enable()
when the call to devm_regulator_bulk_get_enable() fails and then calls
dev_err_probe() with the uninitialized device.
This seems to only cause a panic with dev_err_probe(), dev_err(),
dev_warn() and dev_info() don't seem to cause a panic, but are fixed
as well.
The buffer is set to 80 character. If a caller write more characters,
count is truncated to the max available space in "simple_write_to_buffer".
But afterwards a string terminator is written to the buffer at offset count
without boundary check. The zero termination is written OUT-OF-BOUND.
Add a check that the given buffer is smaller then the buffer to prevent.
Fixes: 035b4989211d ("iio: backend: make sure to NULL terminate stack buffer") Signed-off-by: Markus Burri <markus.burri@mt.com> Reviewed-by: Nuno Sá <nuno.sa@analog.com> Link: https://patch.msgid.link/20250508130612.82270-2-markus.burri@mt.com Cc: <Stable@vger.kernel.org> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix a race where a pending interrupt could be received and the handler
called before the handler's data has been setup, by converting to
irq_set_chained_handler_and_data().
Fixes: 1add69880240 ("iio: adc: Add support for STM32 ADC core") Signed-off-by: Chen Ni <nichen@iscas.ac.cn> Reviewed-by: Nuno Sá <nuno.sa@analog.com> Tested-by: Fabrice Gasnier <fabrice.gasnier@foss.st.com> Reviewed-by: Fabrice Gasnier <fabrice.gasnier@foss.st.com> Link: https://patch.msgid.link/20250515083101.3811350-1-nichen@iscas.ac.cn Cc: <Stable@vger.kernel.org> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The IIO core issues warnings when a scan mask is a subset of a previous
entry in the available_scan_masks array.
On a board using a MAX11601, the following warning is observed:
max1363 1-0064: available_scan_mask 7 subset of 6. Never used
This occurs because the entries in the max11607_mode_list[] array are not
ordered correctly. To fix this, reorder the entries so that no scan mask is
a subset of an earlier one.
While at it, reorder the mode_list[] arrays for other supported chips as
well, to prevent similar warnings on different variants.
Note fixes tag dropped as these were introduced over many commits a long
time back and the side effect until recently was a reduction in sampling
rate due to reading too many channels when only a few were desired.
Now we have a sanity check that reports this error but that is not
where the issue was introduced.
Cc: stable@vger.kernel.org Signed-off-by: Fabio Estevam <festevam@denx.de> Acked-by: Matti Vaittinen <mazziesaccount@gmail.com> Link: https://patch.msgid.link/20250516173900.677821-2-festevam@gmail.com Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit 2718f15403fb ("iio: sanity check available_scan_masks array"),
booting a board populated with a MAX11601 results in a flood of warnings:
max1363 1-0064: available_scan_mask 8 subset of 0. Never used
max1363 1-0064: available_scan_mask 9 subset of 0. Never used
max1363 1-0064: available_scan_mask 10 subset of 0. Never used
max1363 1-0064: available_scan_mask 11 subset of 0. Never used
max1363 1-0064: available_scan_mask 12 subset of 0. Never used
max1363 1-0064: available_scan_mask 13 subset of 0. Never used
...
These warnings are caused by incorrect offsets used for differential
channels in the MAX1363_4X_CHANS() and MAX1363_8X_CHANS() macros.
The max1363_mode_table[] defines the differential channel mappings as
follows:
fxls8962af_fifo_flush() uses indio_dev->active_scan_mask (with
iio_for_each_active_channel()) without making sure the indio_dev
stays in buffer mode.
There is a race if indio_dev exits buffer mode in the middle of the
interrupt that flushes the fifo. Fix this by calling
synchronize_irq() to ensure that no interrupt is currently running when
disabling buffer mode.
Unable to handle kernel NULL pointer dereference at virtual address 00000000 when read
[...]
_find_first_bit_le from fxls8962af_fifo_flush+0x17c/0x290
fxls8962af_fifo_flush from fxls8962af_interrupt+0x80/0x178
fxls8962af_interrupt from irq_thread_fn+0x1c/0x7c
irq_thread_fn from irq_thread+0x110/0x1f4
irq_thread from kthread+0xe0/0xfc
kthread from ret_from_fork+0x14/0x2c
Fixes: 79e3a5bdd9ef ("iio: accel: fxls8962af: add hw buffered sampling") Cc: stable@vger.kernel.org Suggested-by: David Lechner <dlechner@baylibre.com> Signed-off-by: Sean Nyekjaer <sean@geanix.com> Link: https://patch.msgid.link/20250603-fxlsrace-v2-1-5381b36ba1db@geanix.com Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Free the kfifo after unregistering the miscdev in
aspeed_lpc_disable_snoop() as the kfifo is initialised before the
miscdev in aspeed_lpc_enable_snoop().
Fixes: 3772e5da4454 ("drivers/misc: Aspeed LPC snoop output using misc chardev") Cc: stable@vger.kernel.org Cc: Jean Delvare <jdelvare@suse.de> Acked-by: Jean Delvare <jdelvare@suse.de> Link: https://patch.msgid.link/20250616-aspeed-lpc-snoop-fixes-v2-1-3cdd59c934d3@codeconstruct.com.au Signed-off-by: Andrew Jeffery <andrew@codeconstruct.com.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The CVE-2024-50047 fix removed asynchronous crypto handling from
crypt_message(), assuming all crypto operations are synchronous.
However, when hardware crypto accelerators are used, this can cause
use-after-free crashes:
crypt_message()
// Allocate the creq buffer containing the req
creq = smb2_get_aead_req(..., &req);
// Free creq while async operation is still in progress
kvfree_sensitive(creq, ...);
Hardware crypto modules often implement async AEAD operations for
performance. When crypto_aead_encrypt/decrypt() returns -EINPROGRESS,
the operation completes asynchronously. Without crypto_wait_req(),
the function immediately frees the request buffer, leading to crashes
when the driver later accesses the freed memory.
This results in a use-after-free condition when the hardware crypto
driver later accesses the freed request structure, leading to kernel
crashes with NULL pointer dereferences.
The issue occurs because crypto_alloc_aead() with mask=0 doesn't
guarantee synchronous operation. Even without CRYPTO_ALG_ASYNC in
the mask, async implementations can be selected.
Fix by restoring the async crypto handling:
- DECLARE_CRYPTO_WAIT(wait) for completion tracking
- aead_request_set_callback() for async completion notification
- crypto_wait_req() to wait for operation completion
This ensures the request buffer isn't freed until the crypto operation
completes, whether synchronous or asynchronous, while preserving the
CVE-2024-50047 fix.
Fixes: b0abcd65ec54 ("smb: client: fix UAF in async decryption") Link: https://lore.kernel.org/all/8b784a13-87b0-4131-9ff9-7a8993538749@huaweicloud.com/ Cc: stable@vger.kernel.org Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.org> Signed-off-by: Wang Zhaolong <wangzhaolong@huaweicloud.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 7ded842b356d ("s390/bpf: Fix bpf_plt pointer arithmetic") has
accidentally removed the critical piece of commit c730fce7c70c
("s390/bpf: Fix bpf_arch_text_poke() with new_addr == NULL"), causing
intermittent kernel panics in e.g. perf's on_switch() prog to reappear.
pm_domain_cpu_gov is selecting a cluster idle state but does not consider
latency tolerance of child CPUs. This results in deeper cluster idle state
whose latency does not meet latency tolerance requirement.
Select deeper idle state only if global and device latency tolerance of all
child CPUs meet.
Test results on SM8750 with 300 usec PM-QoS on CPU0 which is less than
domain idle state entry (2150) + exit (1983) usec latency mentioned in
devicetree, demonstrate the issue.
Before: (Usage is incrementing)
======
# cat /sys/kernel/debug/pm_genpd/power-domain-cluster0/idle_states
State Time Spent(ms) Usage Rejected Above Below
S0 29817 537 8 270 0
# cat /sys/kernel/debug/pm_genpd/power-domain-cluster0/idle_states
State Time Spent(ms) Usage Rejected Above Below
S0 30348 542 8 271 0
After: (Usage is not incrementing due to latency tolerance)
======
# cat /sys/kernel/debug/pm_genpd/power-domain-cluster0/idle_states
State Time Spent(ms) Usage Rejected Above Below
S0 39319 626 14 307 0
# cat /sys/kernel/debug/pm_genpd/power-domain-cluster0/idle_states
State Time Spent(ms) Usage Rejected Above Below
S0 39319 626 14 307 0
When device reset is triggered by feature changes such as toggling Rx
VLAN offload, wx->do_reset() is called to reinitialize Rx rings. The
hardware descriptor ring may retain stale values from previous sessions.
And only set the length to 0 in rx_desc[0] would result in building
malformed SKBs. Fix it to ensure a clean slate after device reset.
The wx_rx_buffer structure contained two DMA address fields: 'dma' and
'page_dma'. However, only 'page_dma' was actually initialized and used
to program the Rx descriptor. But 'dma' was uninitialized and used in
some paths.
This could lead to undefined behavior, including DMA errors or
use-after-free, if the uninitialized 'dma' was used. Althrough such
error has not yet occurred, it is worth fixing in the code.
Fixes: 3c47e8ae113a ("net: libwx: Support to receive packets in NAPI") Cc: stable@vger.kernel.org Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20250714024755.17512-3-jiawenwu@trustnetic.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
page_pool_put_full_page() should only be invoked when freeing Rx buffers
or building a skb if the size is too short. At other times, the pages
need to be reused. So remove the redundant page put. In the original
code, double free pages cause kernel panic:
Errata i2312 [0] for K3 silicon mentions the maximum obtainable
timeout through MMC host controller is 700ms. And for commands taking
longer than 700ms, hardware timeout should be disabled and software
timeout should be used.
The workaround for Errata i2312 can be achieved by adding
SDHCI_QUIRK2_DISABLE_HW_TIMEOUT quirk in sdhci_am654.
Disable command queuing on Intel GLK-based Positivo models.
Without this quirk, CQE (Command Queuing Engine) causes instability
or I/O errors during operation. Disabling it ensures stable
operation on affected devices.
A new warning in clang [1] points out that id_reg is uninitialized then
passed to memstick_init_req() as a const pointer:
drivers/memstick/core/memstick.c:330:59: error: variable 'id_reg' is uninitialized when passed as a const pointer argument here [-Werror,-Wuninitialized-const-pointer]
330 | memstick_init_req(&card->current_mrq, MS_TPC_READ_REG, &id_reg,
| ^~~~~~
Commit de182cc8e882 ("drivers/memstick/core/memstick.c: avoid -Wnonnull
warning") intentionally passed this variable uninitialized to avoid an
-Wnonnull warning from a NULL value that was previously there because
id_reg is never read from the call to memstick_init_req() in
h_memstick_read_dev_id(). Just zero initialize id_reg to avoid the
warning, which is likely happening in the majority of builds using
modern compilers that support '-ftrivial-auto-var-init=zero'.
The nbpf->chan[] array is allocated earlier in the nbpf_probe() function
and it has "num_channels" elements. These three loops iterate one
element farther than they should and corrupt memory.
The changes to the second loop are more involved. In this case, we're
copying data from the irqbuf[] array into the nbpf->chan[] array. If
the data in irqbuf[i] is the error IRQ then we skip it, so the iterators
are not in sync. I added a check to ensure that we don't go beyond the
end of the irqbuf[] array. I'm pretty sure this can't happen, but it
seemed harmless to add a check.
On the other hand, after the loop has ended there is a check to ensure
that the "chan" iterator is where we expect it to be. In the original
code we went one element beyond the end of the array so the iterator
wasn't in the correct place and it would always return -EINVAL. However,
now it will always be in the correct place. I deleted the check since
we know the result.
Cc: stable@vger.kernel.org Fixes: b45b262cefd5 ("dmaengine: add a driver for AMBA AXI NBPF DMAC IP cores") Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org> Link: https://lore.kernel.org/r/b13c5225-7eff-448c-badc-a2c98e9bcaca@sabinyo.mountain Signed-off-by: Vinod Koul <vkoul@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently cpu hotplug with the PREEMPT_RT option set in the kernel is
not supported because the underlying generic power domain functions
used in the cpu hotplug callbacks are incompatible from a lock point
of view. This situation prevents the suspend to idle to reach the
deepest idle state for the "cluster" as identified in the
undermentioned commit.
Use the compatible ones when PREEMPT_RT is enabled and remove the
boolean disabling the hotplug callbacks with this option.
With this change the platform can reach the deepest idle state
allowing at suspend time to consume less power.
Due to what seem to be a bug with variant version returned by some
firmwares the code may set hdev->classify_pkt_type with
btintel_classify_pkt_type when in fact the controller doesn't even
support ISO channels feature but may use the handle range expected from
a controllers that does causing the packets to be reclassified as ISO
causing several bugs.
To fix the above btintel_classify_pkt_type will attempt to check if the
controller really supports ISO channels and in case it doesn't don't
reclassify even if the handle range is considered to be ISO, this is
considered safer than trying to fix the specific controller/firmware
version as that could change over time and causing similar problems in
the future.
When MSG_DONTWAIT is not set, the tpacket_snd operation will wait for
pending_refcnt to decrement to zero before returning. The pending_refcnt
is decremented by 1 when the skb->destructor function is called,
indicating that the skb has been successfully sent and needs to be
destroyed.
If an error occurs during this process, the tpacket_snd() function will
exit and return error, but pending_refcnt may not yet have decremented to
zero. Assuming the next send operation is executed immediately, but there
are no available frames to be sent in tx_ring (i.e., packet_current_frame
returns NULL), and skb is also NULL, the function will not execute
wait_for_completion_interruptible_timeout() to yield the CPU. Instead, it
will enter a do-while loop, waiting for pending_refcnt to be zero. Even
if the previous skb has completed transmission, the skb->destructor
function can only be invoked in the ksoftirqd thread (assuming NAPI
threading is enabled). When both the ksoftirqd thread and the tpacket_snd
operation happen to run on the same CPU, and the CPU trapped in the
do-while loop without yielding, the ksoftirqd thread will not get
scheduled to run. As a result, pending_refcnt will never be reduced to
zero, and the do-while loop cannot exit, eventually leading to a CPU soft
lockup issue.
In fact, skb is true for all but the first iterations of that loop, and
as long as pending_refcnt is not zero, even if incremented by a previous
call, wait_for_completion_interruptible_timeout() should be executed to
yield the CPU, allowing the ksoftirqd thread to be scheduled. Therefore,
the execution condition of this function should be modified to check if
pending_refcnt is not zero, instead of check skb.
- if (need_wait && skb) {
+ if (need_wait && packet_read_pending(&po->tx_ring)) {
As a result, the judgment conditions are duplicated with the end code of
the while loop, and packet_read_pending() is a very expensive function.
Actually, this loop can only exit when ph is NULL, so the loop condition
can be changed to while (1), and in the "ph = NULL" branch, if the
subsequent condition of if is not met, the loop can break directly. Now,
the loop logic remains the same as origin but is clearer and more obvious.
Fixes: 89ed5b519004 ("af_packet: Block execution of tasks waiting for transmit to complete in AF_PACKET") Cc: stable@kernel.org Suggested-by: LongJun Tang <tanglongjun@kylinos.cn> Signed-off-by: Yun Lu <luyun@kylinos.cn> Reviewed-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Due to the changes in commit 581073f626e3 ("af_packet: do not call
packet_read_pending() from tpacket_destruct_skb()"), every time
tpacket_destruct_skb() is executed, the skb_completion is marked as
completed. When wait_for_completion_interruptible_timeout() returns
completed, the pending_refcnt has not yet been reduced to zero.
Therefore, when ph is NULL, the wait function may need to be called
multiple times until packet_read_pending() finally returns zero.
We should call sock_sndtimeo() only once, otherwise the SO_SNDTIMEO
constraint could be way off.
Fixes: 581073f626e3 ("af_packet: do not call packet_read_pending() from tpacket_destruct_skb()") Cc: stable@kernel.org Suggested-by: Eric Dumazet <edumazet@google.com> Signed-off-by: Yun Lu <luyun@kylinos.cn> Reviewed-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Hardware CS has a very slow rise time of about 6us,
causing transmission errors when CS does not reach
high between transaction.
It looks like it's not driven actively when transitioning
from low to high but switched to input, so only the CPU
pull-up pulls it high, slowly. Transitions from high to low
are fast. On the oscilloscope, CS looks like an irregular sawtooth
pattern like this:
_____
^ / |
^ /| / |
/| / | / |
/ | / | / |
___/ |___/ |_____/ |___
With cs-gpios we have a CS rise time of about 20ns, as it should be,
and CS looks rectangular.
This fixes the data errors when running a flashcp loop against a
m25p40 spi flash.
With the Rockchip 6.1 kernel we see the same slow rise time, but
for some reason CS is always high for long enough to reach a solid
high.
The RK3399 and RK3588 SoCs use the same SPI driver, so we also
checked our "Puma" (RK3399) and "Tiger" (RK3588) boards.
They do not have this problem. Hardware CS rise time is good.
The IMX8MPDS Table 37 [1] shows that the max SPI master read frequency
depends on the pins the interface is muxed behind with ECSPI2
muxed behind ECSPI2 supporting up to 25MHz.
Adjust the spi-max-frequency based on these findings.
The IMX8MPDS Table 37 [1] shows that the max SPI master read frequency
depends on the pins the interface is muxed behind with ECSPI2
muxed behind ECSPI2 supporting up to 25MHz.
Adjust the spi-max-frequency based on these findings.
The IMX8MPDS Table 37 [1] shows that the max SPI master read frequency
depends on the pins the interface is muxed behind with ECSPI2
muxed behind ECSPI2 supporting up to 25MHz.
Adjust the spi-max-frequency based on these findings.
LDO5 regulator is used to power the i.MX8MM NVCC_SD2 I/O supply, that is
used for the SD2 card interface and also for some GPIOs.
When the SD card interface is not enabled the regulator subsystem could
turn off this supply, since it is not used anywhere else, however this
will also remove the power to some other GPIOs, for example one I/O that
is used to power the ethernet phy, leading to a non working ethernet
interface.
[ 31.820515] On-module +V3.3_1.8_SD (LDO5): disabling
[ 31.821761] PMIC_USDHC_VSELECT: disabling
[ 32.764949] fec 30be0000.ethernet end0: Link is Down
Fix this keeping the LDO5 supply always on.
Cc: stable@vger.kernel.org Fixes: 6a57f224f734 ("arm64: dts: freescale: add initial support for verdin imx8m mini") Fixes: f5aab0438ef1 ("regulator: pca9450: Fix enable register for LDO5") Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com> Reviewed-by: Frank Li <Frank.Li@nxp.com> Signed-off-by: Shawn Guo <shawnguo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Watchdog doesn't work on NXP ls1046ardb board because in commit 7c8ffc5555cb("arm64: dts: layerscape: remove big-endian for mmc nodes"),
it intended to remove the big-endian from mmc node, but the big-endian of
watchdog node is also removed by accident. So, add watchdog big-endian
property back.
In addition, add compatible string fsl,ls1046a-wdt, which allow big-endian
property.
Fixes: 7c8ffc5555cb ("arm64: dts: layerscape: remove big-endian for mmc nodes") Cc: stable@vger.kernel.org Signed-off-by: Meng Li <Meng.Li@windriver.com> Reviewed-by: Ahmad Fatoum <a.fatoum@pengutronix.de> Signed-off-by: Shawn Guo <shawnguo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The IMX8MPDS Table 37 [1] shows that the max SPI master read frequency
depends on the pins the interface is muxed behind with ECSPI2
muxed behind ECSPI2 supporting up to 25MHz.
Adjust the spi-max-frequency based on these findings.
A new warning in clang [1] points out a place in pep_sock_accept() where
dst is uninitialized then passed as a const pointer to pep_find_pipe():
net/phonet/pep.c:829:37: error: variable 'dst' is uninitialized when passed as a const pointer argument here [-Werror,-Wuninitialized-const-pointer]
829 | newsk = pep_find_pipe(&pn->hlist, &dst, pipe_handle);
| ^~~:
Move the call to pn_skb_get_dst_sockaddr(), which initializes dst, to
before the call to pep_find_pipe(), so that dst is consistently used
initialized throughout the function.
mptcp_disconnect() clears the fallback bit unconditionally, without
touching the associated flags.
The bit clear is safe, as no fallback operation can race with that --
all subflow are already in TCP_CLOSE status thanks to the previous
FASTCLOSE -- but we need to consistently reset all the fallback related
status.
Also acquire the relevant lock, to avoid fouling static analyzers.
We have races similar to the one addressed by the previous patch between
subflow failing and additional subflow creation. They are just harder to
trigger.
The solution is similar. Use a separate flag to track the condition
'socket state prevent any additional subflow creation' protected by the
fallback lock.
The socket fallback makes such flag true, and also receiving or sending
an MP_FAIL option.
The field 'allow_infinite_fallback' is now always touched under the
relevant lock, we can drop the ONCE annotation on write.
Fixes: 478d770008b0 ("mptcp: send out MP_FAIL when data checksum fails") Cc: stable@vger.kernel.org Signed-off-by: Paolo Abeni <pabeni@redhat.com> Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Link: https://patch.msgid.link/20250714-net-mptcp-fallback-races-v1-2-391aff963322@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since we need to track the 'fallback is possible' condition and the
fallback status separately, there are a few possible races open between
the check and the actual fallback action.
Add a spinlock to protect the fallback related information and use it
close all the possible related races. While at it also remove the
too-early clearing of allow_infinite_fallback in __mptcp_subflow_connect():
the field will be correctly cleared by subflow_finish_connect() if/when
the connection will complete successfully.
If fallback is not possible, as per RFC, reset the current subflow.
Since the fallback operation can now fail and return value should be
checked, rename the helper accordingly.
8c8492ca64e7 ("io_uring/net: don't retry connect operation on EPOLLERR")
is a little dirty hack that
1) wrongfully assumes that POLLERR equals to a failed request, which
breaks all POLLERR users, e.g. all error queue recv interfaces.
2) deviates the connection request behaviour from connect(2), and
3) racy and solved at a wrong level.
Nothing can be done with 2) now, and 3) is beyond the scope of the
patch. At least solve 1) by moving the hack out of generic poll handling
into io_connect().
In DCN401 pre-blending degamma LUT isn't affecting cursor as in previous
DCN version. As this is not the behavior close to what is expected for
CRTC degamma LUT, disable CRTC degamma LUT property in this HW.
Increment the reset counter only if soft recovery succeeded. This is
consistent with a ring hard reset behaviour where counter gets
incremented only if hard reset succeeded.
Signed-off-by: Lijo Lazar <lijo.lazar@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 25c314aa3ec3d30e4ee282540e2096b5c66a2437) Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 42cdf6f687da ("drm/amdgpu/gfx8: always restore kcq MQDs") made the
ring pointer always to be reset on resume from suspend. This caused compute
rings to fail since the reset was done without also resetting it for the
firmware. Reset wptr on the GPU to avoid a disconnect between the driver
and firmware wptr.
Starting with Rust 1.89.0 (expected 2025-08-07), under
`CONFIG_RUST_DEBUG_ASSERTIONS=y`, `objtool` may report:
rust/kernel.o: warning: objtool: _R..._6kernel4pageNtB5_4Page8read_raw()
falls through to next function _R..._6kernel4pageNtB5_4Page9write_raw()
(and many others) due to calls to the `noreturn` symbol:
core::panicking::panic_nounwind_fmt
Thus add the mangled one to the list so that `objtool` knows it is
actually `noreturn`.
See commit 56d680dd23c3 ("objtool/rust: list `noreturn` Rust functions")
for more details.
Cc: stable@vger.kernel.org # Needed in 6.12.y and later (Rust is pinned in older LTSs). Cc: Josh Poimboeuf <jpoimboe@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lore.kernel.org/r/20250712160103.1244945-2-ojeda@kernel.org Signed-off-by: Miguel Ojeda <ojeda@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit e7186af7fb26 ("tracing: Add back FORTIFY_SOURCE logic to
kernel_stack event structure"), struct stack_entry marks its caller
field with __counted_by(size). At the time of the memcpy, entry->size
contains garbage from the ringbuffer, which under some circumstances is
zero, triggering a kernel panic by buffer overflow.
Populate the size field before the memcpy so that the out-of-bounds
check knows the correct size. This is analogous to
__ftrace_trace_stack().
Cc: stable@vger.kernel.org Cc: John Kacur <jkacur@redhat.com> Cc: Luis Goncalves <lgoncalv@redhat.com> Cc: Attila Fazekas <afazekas@redhat.com> Link: https://lore.kernel.org/20250716143601.7313-1-tglozar@redhat.com Fixes: e7186af7fb26 ("tracing: Add back FORTIFY_SOURCE logic to kernel_stack event structure") Signed-off-by: Tomas Glozar <tglozar@redhat.com> Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When a module is loaded, it adds trace events defined by the module. It
may also need to modify the modules trace printk formats to replace enum
names with their values.
If two modules are loaded at the same time, the adding of the event to the
ftrace_events list can corrupt the walking of the list in the code that is
modifying the printk format strings and crash the kernel.
The addition of the event should take the trace_event_sem for write while
it adds the new event.
Also add a lockdep_assert_held() on that semaphore in
__trace_add_event_dirs() as it iterates the list.
After a recent change in clang to strengthen uninitialized warnings [1],
it points out that in one of the error paths in parse_btf_arg(), params
is used uninitialized:
kernel/trace/trace_probe.c:660:19: warning: variable 'params' is uninitialized when used here [-Wuninitialized]
660 | return PTR_ERR(params);
| ^~~~~~
Match many other NO_BTF_ENTRY error cases and return -ENOENT, clearing
up the warning.
projects / thirdparty / kernel / stable.git / log
