| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| In LibreChat 0.8.1-rc2, a logged-in user obtains a JWT for both the LibreChat API and the RAG API. |
| A low-privileged remote attacker may be able to replace the boot application of the CODESYS Control runtime system, enabling unauthorized code execution. |
| In the Linux kernel, the following vulnerability has been resolved:
btrfs: do not strictly require dirty metadata threshold for metadata writepages
[BUG]
There is an internal report that over 1000 processes are
waiting at the io_schedule_timeout() of balance_dirty_pages(), causing
a system hang and trigger a kernel coredump.
The kernel is v6.4 kernel based, but the root problem still applies to
any upstream kernel before v6.18.
[CAUSE]
From Jan Kara for his wisdom on the dirty page balance behavior first.
This cgroup dirty limit was what was actually playing the role here
because the cgroup had only a small amount of memory and so the dirty
limit for it was something like 16MB.
Dirty throttling is responsible for enforcing that nobody can dirty
(significantly) more dirty memory than there's dirty limit. Thus when
a task is dirtying pages it periodically enters into balance_dirty_pages()
and we let it sleep there to slow down the dirtying.
When the system is over dirty limit already (either globally or within
a cgroup of the running task), we will not let the task exit from
balance_dirty_pages() until the number of dirty pages drops below the
limit.
So in this particular case, as I already mentioned, there was a cgroup
with relatively small amount of memory and as a result with dirty limit
set at 16MB. A task from that cgroup has dirtied about 28MB worth of
pages in btrfs btree inode and these were practically the only dirty
pages in that cgroup.
So that means the only way to reduce the dirty pages of that cgroup is
to writeback the dirty pages of btrfs btree inode, and only after that
those processes can exit balance_dirty_pages().
Now back to the btrfs part, btree_writepages() is responsible for
writing back dirty btree inode pages.
The problem here is, there is a btrfs internal threshold that if the
btree inode's dirty bytes are below the 32M threshold, it will not
do any writeback.
This behavior is to batch as much metadata as possible so we won't write
back those tree blocks and then later re-COW them again for another
modification.
This internal 32MiB is higher than the existing dirty page size (28MiB),
meaning no writeback will happen, causing a deadlock between btrfs and
cgroup:
- Btrfs doesn't want to write back btree inode until more dirty pages
- Cgroup/MM doesn't want more dirty pages for btrfs btree inode
Thus any process touching that btree inode is put into sleep until
the number of dirty pages is reduced.
Thanks Jan Kara a lot for the analysis of the root cause.
[ENHANCEMENT]
Since kernel commit b55102826d7d ("btrfs: set AS_KERNEL_FILE on the
btree_inode"), btrfs btree inode pages will only be charged to the root
cgroup which should have a much larger limit than btrfs' 32MiB
threshold.
So it should not affect newer kernels.
But for all current LTS kernels, they are all affected by this problem,
and backporting the whole AS_KERNEL_FILE may not be a good idea.
Even for newer kernels I still think it's a good idea to get
rid of the internal threshold at btree_writepages(), since for most cases
cgroup/MM has a better view of full system memory usage than btrfs' fixed
threshold.
For internal callers using btrfs_btree_balance_dirty() since that
function is already doing internal threshold check, we don't need to
bother them.
But for external callers of btree_writepages(), just respect their
requests and write back whatever they want, ignoring the internal
btrfs threshold to avoid such deadlock on btree inode dirty page
balancing. |
| In the Linux kernel, the following vulnerability has been resolved:
net: stmmac: fix TSO DMA API usage causing oops
Commit 66600fac7a98 ("net: stmmac: TSO: Fix unbalanced DMA map/unmap
for non-paged SKB data") moved the assignment of tx_skbuff_dma[]'s
members to be later in stmmac_tso_xmit().
The buf (dma cookie) and len stored in this structure are passed to
dma_unmap_single() by stmmac_tx_clean(). The DMA API requires that
the dma cookie passed to dma_unmap_single() is the same as the value
returned from dma_map_single(). However, by moving the assignment
later, this is not the case when priv->dma_cap.addr64 > 32 as "des"
is offset by proto_hdr_len.
This causes problems such as:
dwc-eth-dwmac 2490000.ethernet eth0: Tx DMA map failed
and with DMA_API_DEBUG enabled:
DMA-API: dwc-eth-dwmac 2490000.ethernet: device driver tries to +free DMA memory it has not allocated [device address=0x000000ffffcf65c0] [size=66 bytes]
Fix this by maintaining "des" as the original DMA cookie, and use
tso_des to pass the offset DMA cookie to stmmac_tso_allocator().
Full details of the crashes can be found at:
https://lore.kernel.org/all/d8112193-0386-4e14-b516-37c2d838171a@nvidia.com/
https://lore.kernel.org/all/klkzp5yn5kq5efgtrow6wbvnc46bcqfxs65nz3qy77ujr5turc@bwwhelz2l4dw/ |
| Apache Airflow versions 3.1.0 through 3.1.7 session token (_token) in cookies is set to path=/ regardless of the configured [webserver] base_url or [api] base_url.
This allows any application co-hosted under the same domain to capture valid Airflow session tokens from HTTP request headers, allowing full session takeover without attacking Airflow itself.
Users are recommended to upgrade to Apache Airflow 3.1.8 or later, which resolves this issue. |
| In the Linux kernel, the following vulnerability has been resolved:
netfilter: ipset: Rework long task execution when adding/deleting entries
When adding/deleting large number of elements in one step in ipset, it can
take a reasonable amount of time and can result in soft lockup errors. The
patch 5f7b51bf09ba ("netfilter: ipset: Limit the maximal range of
consecutive elements to add/delete") tried to fix it by limiting the max
elements to process at all. However it was not enough, it is still possible
that we get hung tasks. Lowering the limit is not reasonable, so the
approach in this patch is as follows: rely on the method used at resizing
sets and save the state when we reach a smaller internal batch limit,
unlock/lock and proceed from the saved state. Thus we can avoid long
continuous tasks and at the same time removed the limit to add/delete large
number of elements in one step.
The nfnl mutex is held during the whole operation which prevents one to
issue other ipset commands in parallel. |
| In the Linux kernel, the following vulnerability has been resolved:
USB: Gadget: core: Help prevent panic during UVC unconfigure
Avichal Rakesh reported a kernel panic that occurred when the UVC
gadget driver was removed from a gadget's configuration. The panic
involves a somewhat complicated interaction between the kernel driver
and a userspace component (as described in the Link tag below), but
the analysis did make one thing clear: The Gadget core should
accomodate gadget drivers calling usb_gadget_deactivate() as part of
their unbind procedure.
Currently this doesn't work. gadget_unbind_driver() calls
driver->unbind() while holding the udc->connect_lock mutex, and
usb_gadget_deactivate() attempts to acquire that mutex, which will
result in a deadlock.
The simple fix is for gadget_unbind_driver() to release the mutex when
invoking the ->unbind() callback. There is no particular reason for
it to be holding the mutex at that time, and the mutex isn't held
while the ->bind() callback is invoked. So we'll drop the mutex
before performing the unbind callback and reacquire it afterward.
We'll also add a couple of comments to usb_gadget_activate() and
usb_gadget_deactivate(). Because they run in process context they
must not be called from a gadget driver's ->disconnect() callback,
which (according to the kerneldoc for struct usb_gadget_driver in
include/linux/usb/gadget.h) may run in interrupt context. This may
help prevent similar bugs from arising in the future. |
| OpenPLC Runtime v3 contains an input validation flaw in the /upload-program-action endpoint: the epoch_time field supplied during program uploads is not validated and can be crafted to induce corruption of the programs database. After a successful malformed upload the runtime continues to operate until a restart; on restart the runtime can fail to start because of corrupted database entries, resulting in persistent denial of service requiring complete rebase of the product to recover. This vulnerability was remediated by commit 095ee09. |
| telnet in GNU inetutils through 2.7 allows servers to read arbitrary environment variables from clients via NEW_ENVIRON SEND USERVAR. |
| A memory corruption issue was addressed with improved lock state checking. This issue is fixed in watchOS 26.1, iOS 18.7.2 and iPadOS 18.7.2, macOS Tahoe 26.1, visionOS 26.1, tvOS 26.1, macOS Sonoma 14.8.2, macOS Sequoia 15.7.2, iOS 26.1 and iPadOS 26.1. A malicious application may cause unexpected changes in memory shared between processes. |
| In the Linux kernel, the following vulnerability has been resolved:
net/mlx5e: Fix deadlock in tc route query code
Cited commit causes ABBA deadlock[0] when peer flows are created while
holding the devcom rw semaphore. Due to peer flows offload implementation
the lock is taken much higher up the call chain and there is no obvious way
to easily fix the deadlock. Instead, since tc route query code needs the
peer eswitch structure only to perform a lookup in xarray and doesn't
perform any sleeping operations with it, refactor the code for lockless
execution in following ways:
- RCUify the devcom 'data' pointer. When resetting the pointer
synchronously wait for RCU grace period before returning. This is fine
since devcom is currently only used for synchronization of
pairing/unpairing of eswitches which is rare and already expensive as-is.
- Wrap all usages of 'paired' boolean in {READ|WRITE}_ONCE(). The flag has
already been used in some unlocked contexts without proper
annotations (e.g. users of mlx5_devcom_is_paired() function), but it wasn't
an issue since all relevant code paths checked it again after obtaining the
devcom semaphore. Now it is also used by mlx5_devcom_get_peer_data_rcu() as
"best effort" check to return NULL when devcom is being unpaired. Note that
while RCU read lock doesn't prevent the unpaired flag from being changed
concurrently it still guarantees that reader can continue to use 'data'.
- Refactor mlx5e_tc_query_route_vport() function to use new
mlx5_devcom_get_peer_data_rcu() API which fixes the deadlock.
[0]:
[ 164.599612] ======================================================
[ 164.600142] WARNING: possible circular locking dependency detected
[ 164.600667] 6.3.0-rc3+ #1 Not tainted
[ 164.601021] ------------------------------------------------------
[ 164.601557] handler1/3456 is trying to acquire lock:
[ 164.601998] ffff88811f1714b0 (&esw->offloads.encap_tbl_lock){+.+.}-{3:3}, at: mlx5e_attach_encap+0xd8/0x8b0 [mlx5_core]
[ 164.603078]
but task is already holding lock:
[ 164.603617] ffff88810137fc98 (&comp->sem){++++}-{3:3}, at: mlx5_devcom_get_peer_data+0x37/0x80 [mlx5_core]
[ 164.604459]
which lock already depends on the new lock.
[ 164.605190]
the existing dependency chain (in reverse order) is:
[ 164.605848]
-> #1 (&comp->sem){++++}-{3:3}:
[ 164.606380] down_read+0x39/0x50
[ 164.606772] mlx5_devcom_get_peer_data+0x37/0x80 [mlx5_core]
[ 164.607336] mlx5e_tc_query_route_vport+0x86/0xc0 [mlx5_core]
[ 164.607914] mlx5e_tc_tun_route_lookup+0x1a4/0x1d0 [mlx5_core]
[ 164.608495] mlx5e_attach_decap_route+0xc6/0x1e0 [mlx5_core]
[ 164.609063] mlx5e_tc_add_fdb_flow+0x1ea/0x360 [mlx5_core]
[ 164.609627] __mlx5e_add_fdb_flow+0x2d2/0x430 [mlx5_core]
[ 164.610175] mlx5e_configure_flower+0x952/0x1a20 [mlx5_core]
[ 164.610741] tc_setup_cb_add+0xd4/0x200
[ 164.611146] fl_hw_replace_filter+0x14c/0x1f0 [cls_flower]
[ 164.611661] fl_change+0xc95/0x18a0 [cls_flower]
[ 164.612116] tc_new_tfilter+0x3fc/0xd20
[ 164.612516] rtnetlink_rcv_msg+0x418/0x5b0
[ 164.612936] netlink_rcv_skb+0x54/0x100
[ 164.613339] netlink_unicast+0x190/0x250
[ 164.613746] netlink_sendmsg+0x245/0x4a0
[ 164.614150] sock_sendmsg+0x38/0x60
[ 164.614522] ____sys_sendmsg+0x1d0/0x1e0
[ 164.614934] ___sys_sendmsg+0x80/0xc0
[ 164.615320] __sys_sendmsg+0x51/0x90
[ 164.615701] do_syscall_64+0x3d/0x90
[ 164.616083] entry_SYSCALL_64_after_hwframe+0x46/0xb0
[ 164.616568]
-> #0 (&esw->offloads.encap_tbl_lock){+.+.}-{3:3}:
[ 164.617210] __lock_acquire+0x159e/0x26e0
[ 164.617638] lock_acquire+0xc2/0x2a0
[ 164.618018] __mutex_lock+0x92/0xcd0
[ 164.618401] mlx5e_attach_encap+0xd8/0x8b0 [mlx5_core]
[ 164.618943] post_process_attr+0x153/0x2d0 [
---truncated--- |
| In the Linux kernel, the following vulnerability has been resolved:
sctp: add a refcnt in sctp_stream_priorities to avoid a nested loop
With this refcnt added in sctp_stream_priorities, we don't need to
traverse all streams to check if the prio is used by other streams
when freeing one stream's prio in sctp_sched_prio_free_sid(). This
can avoid a nested loop (up to 65535 * 65535), which may cause a
stuck as Ying reported:
watchdog: BUG: soft lockup - CPU#23 stuck for 26s! [ksoftirqd/23:136]
Call Trace:
<TASK>
sctp_sched_prio_free_sid+0xab/0x100 [sctp]
sctp_stream_free_ext+0x64/0xa0 [sctp]
sctp_stream_free+0x31/0x50 [sctp]
sctp_association_free+0xa5/0x200 [sctp]
Note that it doesn't need to use refcount_t type for this counter,
as its accessing is always protected under the sock lock.
v1->v2:
- add a check in sctp_sched_prio_set to avoid the possible prio_head
refcnt overflow. |
| A potential improper initialization vulnerability was reported in the BIOS of some ThinkPads that could allow a local privileged user to modify data and execute arbitrary code. |
| Spring MVC and WebFlux applications are vulnerable to stream corruption when using Server-Sent Events (SSE). This issue affects Spring Foundation: from 7.0.0 through 7.0.5, from 6.2.0 through 6.2.16, from 6.1.0 through 6.1.25, from 5.3.0 through 5.3.46. |
| In the Linux kernel, the following vulnerability has been resolved:
ipvlan: Make the addrs_lock be per port
Make the addrs_lock be per port, not per ipvlan dev.
Initial code seems to be written in the assumption,
that any address change must occur under RTNL.
But it is not so for the case of IPv6. So
1) Introduce per-port addrs_lock.
2) It was needed to fix places where it was forgotten
to take lock (ipvlan_open/ipvlan_close)
This appears to be a very minor problem though.
Since it's highly unlikely that ipvlan_add_addr() will
be called on 2 CPU simultaneously. But nevertheless,
this could cause:
1) False-negative of ipvlan_addr_busy(): one interface
iterated through all port->ipvlans + ipvlan->addrs
under some ipvlan spinlock, and another added IP
under its own lock. Though this is only possible
for IPv6, since looks like only ipvlan_addr6_event() can be
called without rtnl_lock.
2) Race since ipvlan_ht_addr_add(port) is called under
different ipvlan->addrs_lock locks
This should not affect performance, since add/remove IP
is a rare situation and spinlock is not taken on fast
paths. |
| In the Linux kernel, the following vulnerability has been resolved:
net: usb: r8152: fix resume reset deadlock
rtl8152 can trigger device reset during reset which
potentially can result in a deadlock:
**** DPM device timeout after 10 seconds; 15 seconds until panic ****
Call Trace:
<TASK>
schedule+0x483/0x1370
schedule_preempt_disabled+0x15/0x30
__mutex_lock_common+0x1fd/0x470
__rtl8152_set_mac_address+0x80/0x1f0
dev_set_mac_address+0x7f/0x150
rtl8152_post_reset+0x72/0x150
usb_reset_device+0x1d0/0x220
rtl8152_resume+0x99/0xc0
usb_resume_interface+0x3e/0xc0
usb_resume_both+0x104/0x150
usb_resume+0x22/0x110
The problem is that rtl8152 resume calls reset under
tp->control mutex while reset basically re-enters rtl8152
and attempts to acquire the same tp->control lock once
again.
Reset INACCESSIBLE device outside of tp->control mutex
scope to avoid recursive mutex_lock() deadlock. |
| In the Linux kernel, the following vulnerability has been resolved:
procfs: avoid fetching build ID while holding VMA lock
Fix PROCMAP_QUERY to fetch optional build ID only after dropping mmap_lock
or per-VMA lock, whichever was used to lock VMA under question, to avoid
deadlock reported by syzbot:
-> #1 (&mm->mmap_lock){++++}-{4:4}:
__might_fault+0xed/0x170
_copy_to_iter+0x118/0x1720
copy_page_to_iter+0x12d/0x1e0
filemap_read+0x720/0x10a0
blkdev_read_iter+0x2b5/0x4e0
vfs_read+0x7f4/0xae0
ksys_read+0x12a/0x250
do_syscall_64+0xcb/0xf80
entry_SYSCALL_64_after_hwframe+0x77/0x7f
-> #0 (&sb->s_type->i_mutex_key#8){++++}-{4:4}:
__lock_acquire+0x1509/0x26d0
lock_acquire+0x185/0x340
down_read+0x98/0x490
blkdev_read_iter+0x2a7/0x4e0
__kernel_read+0x39a/0xa90
freader_fetch+0x1d5/0xa80
__build_id_parse.isra.0+0xea/0x6a0
do_procmap_query+0xd75/0x1050
procfs_procmap_ioctl+0x7a/0xb0
__x64_sys_ioctl+0x18e/0x210
do_syscall_64+0xcb/0xf80
entry_SYSCALL_64_after_hwframe+0x77/0x7f
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
rlock(&mm->mmap_lock);
lock(&sb->s_type->i_mutex_key#8);
lock(&mm->mmap_lock);
rlock(&sb->s_type->i_mutex_key#8);
*** DEADLOCK ***
This seems to be exacerbated (as we haven't seen these syzbot reports
before that) by the recent:
777a8560fd29 ("lib/buildid: use __kernel_read() for sleepable context")
To make this safe, we need to grab file refcount while VMA is still locked, but
other than that everything is pretty straightforward. Internal build_id_parse()
API assumes VMA is passed, but it only needs the underlying file reference, so
just add another variant build_id_parse_file() that expects file passed
directly.
[akpm@linux-foundation.org: fix up kerneldoc] |
| In the Linux kernel, the following vulnerability has been resolved:
riscv: trace: fix snapshot deadlock with sbi ecall
If sbi_ecall.c's functions are traceable,
echo "__sbi_ecall:snapshot" > /sys/kernel/tracing/set_ftrace_filter
may get the kernel into a deadlock.
(Functions in sbi_ecall.c are excluded from tracing if
CONFIG_RISCV_ALTERNATIVE_EARLY is set.)
__sbi_ecall triggers a snapshot of the ringbuffer. The snapshot code
raises an IPI interrupt, which results in another call to __sbi_ecall
and another snapshot...
All it takes to get into this endless loop is one initial __sbi_ecall.
On RISC-V systems without SSTC extension, the clock events in
timer-riscv.c issue periodic sbi ecalls, making the problem easy to
trigger.
Always exclude the sbi_ecall.c functions from tracing to fix the
potential deadlock.
sbi ecalls can easiliy be logged via trace events, excluding ecall
functions from function tracing is not a big limitation. |
| Dark Reader is an accessibility browser extension that makes web pages colors dark. The dynamic dark mode feature of the extension works by analyzing the colors of web pages found in CSS style sheet files. In order to analyze cross-origin style sheets (stored on websites different from the original web page), Dark Reader requests such files via a background worker, ensuring the request is performed with no credentials and that the content type of the response is a CSS file. Prior to Dark Reader 4.9.117, this style content was assigned to an HTML Style Element in order to parse and loop through style declarations, and also stored in page's Session Storage for performance gains. This could allow a website author to request a style sheet from a locally running web server, for example by having a link pointing to `http[:]//localhost[:]8080/style[.]css`. The brute force of the host name, port and file name would be unlikely due to performance impact, that would cause the browser tab to hang shortly, but it could be possible to request a style sheet if the full URL was known in advance. As per December 18, 2025 there is no known exploit of the issue. The problem has been fixed in version 4.9.117 on December 3, 2025. The style sheets are now parsed using modern Constructed Style Sheets API and the contents of cross-origin style sheets is no longer stored in page's Session Storage. Version 4.9.118 (December 8, 2025) restricts cross-origin requests to localhost aliases, IP addresses, hosts with ports and non-HTTPS resources. The absolute majority of users have received an update 4.1.117 or 4.9.118 automatically within a week. However users must ensure their automatic updates are not blocked and they are using the latest version of the extension by going to chrome://extensions or about:addons pages in browser settings. Users utilizing manual builds must upgrade to version 4.9.118 and above. Developers using `darkreader` NPM package for their own websites are likely not affected, but must ensure the function passed to `setFetchMethod()` for performing cross-origin requests works within the intended scope. Developers using custom forks of earlier versions of Dark Reader to build other extensions or integrating into their apps or browsers must ensure they perform cross-origin requests safely and the responses are not accessible outside of the app or extension. |
| In the Linux kernel, the following vulnerability has been resolved:
hwmon: (acpi_power_meter) Fix deadlocks related to acpi_power_meter_notify()
The acpi_power_meter driver's .notify() callback function,
acpi_power_meter_notify(), calls hwmon_device_unregister() under a lock
that is also acquired by callbacks in sysfs attributes of the device
being unregistered which is prone to deadlocks between sysfs access and
device removal.
Address this by moving the hwmon device removal in
acpi_power_meter_notify() outside the lock in question, but notice
that doing it alone is not sufficient because two concurrent
METER_NOTIFY_CONFIG notifications may be attempting to remove the
same device at the same time. To prevent that from happening, add a
new lock serializing the execution of the switch () statement in
acpi_power_meter_notify(). For simplicity, it is a static mutex
which should not be a problem from the performance perspective.
The new lock also allows the hwmon_device_register_with_info()
in acpi_power_meter_notify() to be called outside the inner lock
because it prevents the other notifications handled by that function
from manipulating the "resource" object while the hwmon device based
on it is being registered. The sending of ACPI netlink messages from
acpi_power_meter_notify() is serialized by the new lock too which
generally helps to ensure that the order of handling firmware
notifications is the same as the order of sending netlink messages
related to them.
In addition, notice that hwmon_device_register_with_info() may fail
in which case resource->hwmon_dev will become an error pointer,
so add checks to avoid attempting to unregister the hwmon device
pointer to by it in that case to acpi_power_meter_notify() and
acpi_power_meter_remove(). |
